Zwane Mwaikambo wrote:
>To: BlaisorBlade
>Cc: [email protected]; [email protected];
Brown,
>Len
>Subject: [ACPI] Re: [PATCH] Oops and panic while unloading holder of
>pm_idle
>
>On Sun, 5 Sep 2004, BlaisorBlade wrote:
>
>> > There aren't many users of pm_idle
>> > outside of arch/*/kernel/process.c
>> Both APM and ACPI set pm_idle, and both can be modular. It seems,
however,
>> they are the only such ones. And since they APM and ACPI refuse to be
>both
>> loaded, we cannot have (actually) two modules which override pm_idle.
So
>> you're right.
>
>There are a few other issues with pm_idle, preempt and modular drivers
>which someone else is looking at, we'll see how things go from there.
Yes, preempt will cause oops in pc_idle. Attached patch should close the
final race corner.
Thanks,
Shaohua
Hello Shaohua,
On Wed, 8 Sep 2004, Li, Shaohua wrote:
> Yes, preempt will cause oops in pc_idle. Attached patch should close the
> final race corner.
Please try and inline your patches in future, i know it may be hard with
your mail client but it makes commenting on patches easier.
while (!need_resched()) {
+ void (*idle)(void) = NULL;
+
You don't need to initialise variable `idle` here. Otherwise the rest
looks fine with me.
Thanks,
Zwane