Re: [PATCH] sched/cpuset/pm: Fix cpuset vs suspend-resume

[Rafael J. Wysocki]

On Thursday, September 7, 2017 11:26:16 AM CEST Peter Zijlstra wrote:
> On Thu, Sep 07, 2017 at 11:13:38AM +0200, Peter Zijlstra wrote:
> > Subject: sched/cpuset/pm: Fix cpuset vs suspend-resume
> > 
> > Cpusets vs suspend-resume is _completely_ broken. And it got noticed
> > because it now resulted in non-cpuset usage breaking too.
> > 
> > On suspend cpuset_cpu_inactive() doesn't call into
> > cpuset_update_active_cpus() because it doesn't want to move tasks about,
> > there is no need, all tasks are frozen and won't run again until after
> > we've resumed everything.
> > 
> > But this means that when we finally do call into
> > cpuset_update_active_cpus() after resuming the last frozen cpu in
> > cpuset_cpu_active(), the top_cpuset will not have any difference with
> > the cpu_active_mask and this it will not in fact do _anything_.
> > 
> > So the cpuset configuration will not be restored. This was largely
> > hidden because we would unconditionally create identity domains and
> > mobile users would not in fact use cpusets much. And servers what do use
> > cpusets tend to not suspend-resume much.
> > 
> > An addition problem is that we'd not in fact wait for the cpuset work to
> > finish before resuming the tasks, allowing spurious migrations outside
> > of the specified domains.
> > 
> > Fix the rebuild by introducing cpuset_force_rebuild() and fix the
> > ordering with cpuset_wait_for_hotplug().
> > 
> > Cc: tj@xxxxxxxxxx
> > Cc: rjw@xxxxxxxxxxxxx
> > Cc: efault@xxxxxx
> > Reported-by: Andy Lutomirski <luto@xxxxxxxxxx>
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> 
> TJ, I _think_ it was commit:
> 
>   deb7aa308ea2 ("cpuset: reorganize CPU / memory hotplug handling")
> 
> That wrecked things, but there's been so much changes in this area it is
> really hard to tell. Note how before that commit it would
> unconditionally rebuild the domains, and you 'optimized' that ;-)
> 
> That commit also introduced the work to do the async rebuild and failed
> to do that flush on resume.
> 
> In any case, I think we should put a fixes tag on this commit such that
> it gets picked up into stable kernels. Not sure anybody will try and
> backport it into 4 year old kernels, but who knows.
> 

Many thanks for fixing this!