Re: [PATCH v3 10/10] perf/cgroup: Do not switch system-wide events in cgroup switch

[Liang, Kan]

On 11/14/2019 5:43 AM, Peter Zijlstra wrote:
> On Wed, Nov 13, 2019 at 04:30:42PM -0800, Ian Rogers wrote:
> > From: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>
> >
> > When counting system-wide events and cgroup events simultaneously, the
> > system-wide events are always scheduled out then back in during cgroup
> > switches, bringing extra overhead and possibly missing events. Switching
> > out system wide flexible events may be necessary if the scheduled in
> > task's cgroups have pinned events that need to be scheduled in at a higher
> > priority than the system wide flexible events.
>
> I'm thinking this patch is actively broken. groups->index 'group' wide
> and therefore across cpu/cgroup boundaries.
>
> There is no !cgroup to cgroup hierarchy as this patch seems to assume,
> specifically look at how the merge sort in visit_groups_merge() allows
> cgroup events to be picked before !cgroup events.


No, the patch intends to avoid switch !cgroup during cgroup context 
switch.

In perf_cgroup_switch(), when the cgroup is scheduled out, current 
implementation schedule out everything including !cgroup. I think it 
definitely breaks the semantics of !cgroup aka system-wide event.

The patch itself doesn't touch the merge sort in visit_groups_merge().
The perf_cgroup_skip_switch() just skips the !cgroup in schedule_in().
Because !cgroup wasn't scheduled out. We don't want to schedule !cgroup 
in again.
The cgroup events must be after !cgroup events, since !cgroup never be 
switched.

Thanks,
Kan