[linux-pm] [PATCH 4/4] PM: cleanup: stop mandating that platforms export (pm_idle)()

+ linux-arm-kernel

Len Brown [off-list ref] writes:

From: Len Brown <redacted>

(pm_idle)() was originally used on x86 to vector bewteen
ACPI and APM.  With the advent of CPU_IDLE, that reason
for pm_idle to exist vanished.

But x86 APM still scribbled on pm_idle from a module,
so pm_idle didn't go away.  Worse, it was declared in pm.h,
and so it spread to other architectures as dead code.

But today, APM no longer scribbles on pm_idle, so
x86 no longer requires pm_idle to be visible to modules,
or global at all.

Here we remove pm_idle from pm.h, to stop the mandate
that all architectures define it.

This deletes dead code from most architectures,
while some continue using their own internal pm_idle.

At the end of the day, pm_idle...
becomes static in sh, was global
becomes static in x86, was EXPORT_SYMBOL
becomes as global in sparc, was EXPORT_SYMBOL
continues as static in m32r (no pm.h use there)
and is deleted from all other places in the kernel.

Signed-off-by: Len Brown <redacted>
Cc: x86 at kernel.org # x86
Cc: Mike Frysinger <redacted> # blackfin
Cc: Jonas Bonn <jonas@southpole.se> # openrisc
Cc: Tony Luck <tony.luck@intel.com> # ia64
Cc: David Howells <dhowells@redhat.com> # mn10300
Cc: Mikael Starvik <redacted> # cris
Cc: Michal Simek <monstr@monstr.eu> # microblaze
Cc: Paul Mundt <redacted> # sh
Cc: David S. Miller <davem@davemloft.net> # sparc
Cc: Russell King <redacted> # ARM
Cc: Rafael J. Wysocki <redacted> # PM
---
 arch/arm/kernel/process.c         | 12 +++---------

Just looking at the ARM changes, it looks good to me.  I tested it with
and without CPUidle on my ARM-based TI/OMAP platforms (3430/n900,
4430/Panda)

So for the arch/arm changes:

Reviewed-by: Kevin Hilman <redacted>
Tested-by: Kevin Hilman <redacted>

Kevin