[REGRESSION] b8b2c7d845 breaks suspend/resume in qemu

[REGRESSION] b8b2c7d845 breaks suspend/resume in qemu

[John Stultz]

Hey Uwe,

So I recently noticed that my alarmtimer suspend/resume tests
(selftests/timers/alartimer-suspend.c) were getting stuck testing w/
4.4-rc kernels when running under qemu (x86_64).

I've bisected this back to: b8b2c7d845 (base/platform: assert that
dev_pm_domain callbacks are called unconditionally).

I've not yet spent done any further analysis (bisecting it down, after
some user-error, ate up my morning). But I wanted to see if there were
any initial suggestions for solving this.

thanks
-john

Re: [REGRESSION] b8b2c7d845 breaks suspend/resume in qemu

[John Stultz]

On Wed, Dec 2, 2015 at 12:39 PM, John Stultz <john.stultz@linaro.org> wrote:
> Hey Uwe,
>
> So I recently noticed that my alarmtimer suspend/resume tests
> (selftests/timers/alartimer-suspend.c) were getting stuck testing w/
> 4.4-rc kernels when running under qemu (x86_64).
>
> I've bisected this back to: b8b2c7d845 (base/platform: assert that
> dev_pm_domain callbacks are called unconditionally).
>
> I've not yet spent done any further analysis (bisecting it down, after
> some user-error, ate up my morning). But I wanted to see if there were
> any initial suggestions for solving this.

Just one other quick note: Reverting this commit against 4.4-rc3 seems
to work as well.

thanks
-john

Re: [REGRESSION] b8b2c7d845 breaks suspend/resume in qemu

[Greg Kroah-Hartman]

On Wed, Dec 02, 2015 at 12:46:10PM -0800, John Stultz wrote:
> On Wed, Dec 2, 2015 at 12:39 PM, John Stultz <john.stultz@linaro.org> wrote:
> > Hey Uwe,
> >
> > So I recently noticed that my alarmtimer suspend/resume tests
> > (selftests/timers/alartimer-suspend.c) were getting stuck testing w/
> > 4.4-rc kernels when running under qemu (x86_64).
> >
> > I've bisected this back to: b8b2c7d845 (base/platform: assert that
> > dev_pm_domain callbacks are called unconditionally).
> >
> > I've not yet spent done any further analysis (bisecting it down, after
> > some user-error, ate up my morning). But I wanted to see if there were
> > any initial suggestions for solving this.
> 
> Just one other quick note: Reverting this commit against 4.4-rc3 seems
> to work as well.

There's a proposed patch to resolve this on the mailing list, "Subject:
[PATCH] base/platform: fix panic when probe function is NULL" from
Martin, but I think he's going to send me an updated version before I
forward it on to Linus.

It should resolve your issue, can you verify that?

thanks,

greg k-h

Re: [REGRESSION] b8b2c7d845 breaks suspend/resume in qemu

[John Stultz]

On Wed, Dec 2, 2015 at 12:57 PM, Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
> On Wed, Dec 02, 2015 at 12:46:10PM -0800, John Stultz wrote:
>> On Wed, Dec 2, 2015 at 12:39 PM, John Stultz <john.stultz@linaro.org> wrote:
>> > Hey Uwe,
>> >
>> > So I recently noticed that my alarmtimer suspend/resume tests
>> > (selftests/timers/alartimer-suspend.c) were getting stuck testing w/
>> > 4.4-rc kernels when running under qemu (x86_64).
>> >
>> > I've bisected this back to: b8b2c7d845 (base/platform: assert that
>> > dev_pm_domain callbacks are called unconditionally).
>> >
>> > I've not yet spent done any further analysis (bisecting it down, after
>> > some user-error, ate up my morning). But I wanted to see if there were
>> > any initial suggestions for solving this.
>>
>> Just one other quick note: Reverting this commit against 4.4-rc3 seems
>> to work as well.
>
> There's a proposed patch to resolve this on the mailing list, "Subject:
> [PATCH] base/platform: fix panic when probe function is NULL" from
> Martin, but I think he's going to send me an updated version before I
> forward it on to Linus.
>
> It should resolve your issue, can you verify that?

Yep. That fix seems to avoid the problem for me.

thanks
-john