[PATCH 3/3] ARM: SAMSUNG: Add idle support

[ben-linux@fluff.org (Ben Dooks)]

On Tue, Jan 12, 2010 at 10:12:26AM +0000, Russell King - ARM Linux wrote:
> On Tue, Jan 12, 2010 at 12:06:06AM +0000, Ben Dooks wrote:
> > On Mon, Jan 11, 2010 at 10:02:54AM +0000, Russell King - ARM Linux wrote:
> > > On Mon, Jan 11, 2010 at 06:44:32PM +0900, Kukjin Kim wrote:
> > > > This patch adds common idle functionality for all Samsung SoC's.
> > > 
> > > NAK.  Please use the existing support already provided.  There's no point
> > > in double-indirection - it just makes the code harder to follow.
> > > 
> > > Just set pm_idle to your idle function.
> > 
> > Hmm, the default_idle() function that pm_idle is set to at init time does
> > more than what arch_idle() does. We'd end up having to copy the logic in
> > default_idle() into each of our idle routines. How difficult would it be
> > to change the callers of pm_idle() to check whether it actually needs
> > calling and to re-enable the interrupt state after calling it?
> 
> Not possible without changing all the other architectures - pm_idle() is
> the standard interface for hooking into the idle code, and is used by
> things like cpuidle, APM, ACPI, etc.
> 
> Please use the standard cross-architecture pm_idle() hook if you want to
> dynamically override the default idle code.

So basically i've got to copy the default_idle() and the necessary cpu code
for each of the cpus that are supported? I'd much rather just do it once
and have a pointer to the bits that change thanks.

-- 
Ben

Q:      What's a light-year?
A:      One-third less calories than a regular year.