Re: cpufreq notifiers break suspend -- Re: suspend broken in next-20190704 on Thinkpad X60

[Rafael J. Wysocki]

On Mon, Jul 8, 2019 at 5:05 AM Viresh Kumar <viresh.kumar@xxxxxxxxxx> wrote:
>
> On 06-07-19, 22:30, Pavel Machek wrote:
> > Hi!
> >
> > > Anyway, if 5.2-rc7 is OK, something in this branch causes the problem
> > > to happen for you.
> > >
> > > I would try
> > >
> > > https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/commit/?h=linux-next&id=f012a132824fc870b90980540f727c76fc72e244
> > >
> > > to narrow down the scope somewhat.
>
> I couldn't find the original mail, what exactly is the problem with
> suspend in your case ?

Something unusual:

https://lore.kernel.org/linux-pm/20190706190123.GA11603@amd/T/#mca22dd7c1e8836e9253702df9f56a68ab65192a4

> > Bisect says:
> >
> > 572542c81dec533b7dd3778ea9f5949a00595f68 is the first bad commit
> > Author: Viresh Kumar <viresh.kumar@xxxxxxxxxx>
> >
> >     cpufreq: Register notifiers with the PM QoS framework
> >
> >     This registers the notifiers for min/max frequency constraints
> >     with the
> >
> >  Reviewed-by: Matthias Kaehlcke <mka@xxxxxxxxxxxx>
> >  Reviewed-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
> >  Signed-off-by: Viresh Kumar <viresh.kumar@xxxxxxxxxx>
> >
> > Unfortunately, it does not revert cleanly:
>
> I tried following on my ARM board (both single policy and multiple
> policy configurations):
>
> rtcwake --seconds 5 -v -m mem
>
> And everything worked as expected. Please make sure the top commit of
> my series in pm/linux-next is, some issues were fixed on Friday:
>
> 0a811974f3f7 cpufreq: Add QoS requests for userspace constraints

Pavel has tested the latest version of the patch series AFAICS.

The locking added by the commit in question to
refresh_frequency_limits() requires an update of
cpufreq_update_policy(), or it will deadlock in there because of the
lock acquired by cpufreq_cpu_get() if I haven't missed anything.