Re: [PATCH] CPU C-state breakage with PM Qos change

[Jean Pihet <jean.pihet@newoldbits.com>]

On Mon, Feb 6, 2012 at 9:16 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> On Monday, February 06, 2012, Jean Pihet wrote:
>> Rafael,
>>
>> On Mon, Feb 6, 2012 at 11:18 AM, Jean Pihet <jean.pihet@newoldbits.com> wrote:
>> > Hi Rafael, Mark,
>> >
>> > On Sun, Feb 5, 2012 at 12:04 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
>> ...
>>
>> >> In kernel/power/Makefile:
>> >>
>> >> obj-$(CONFIG_PM)                += main.o qos.o
>> >>
>> >> I guess that explains things. :-)
>> > Initially I thought we should have a way of disabling the feature on
>> > some (minimal) kernels and so thought CONFIG_PM was the option to use.
>> >
>> >> It's quite easy to make qos.o be independent of CONFIG_PM, in which case the
>> >> code added by Venki can be removed, so patches welcome (for 3.4, though).
>> > I am working on it, more to come soon.
>>
>> I have a couple of patches ready, to be applied on 3.3-rc1 (so without
>> Venki's patch applied).
>> The first one is on PM QoS, the second one on per-device PM QoS. Is
>> the latter needed?
>
> I'm not sure without looking. :-)
That makes sense ;p

Just sent out the patch set as '[PATCH 0/2] PM / QoS: unconditionally
build the feature'.
It has been compile tested only using the i386 defconfig, with and
without CONFIG_PM.

It requires some more testing on x86.
Venki, can you check it out if this fixes the initial problem?

>
> Thanks,
> Rafael

Regards,
Jean