Re: [PATCH v2] events/core: fix acoount failure for event's total_enable_time

[Yeoreum Yun]

> > It works. but what about this?
> >
> > diff --git a/kernel/events/core.c b/kernel/events/core.c
> > index 065f9188b44a..71ed8f847b04 100644
> > --- a/kernel/events/core.c
> > +++ b/kernel/events/core.c
> > @@ -2432,6 +2432,7 @@ __perf_remove_from_context(struct perf_event *event,
> >         if (flags & DETACH_DEAD)
> >                 event->pending_disable = 1;
> >         event_sched_out(event, ctx);
> > +       perf_event_update_time(event);
> >         if (flags & DETACH_GROUP)
> >                 perf_group_detach(event);
> >         if (flags & DETACH_CHILD)
>
> Well, the rule is that timekeeping is tied to state change. And as you
> can see __perf_remove_from_context() violates that.
>
> Probably because someone (probably me) figured that the event would not
> be observed again after detach or something like that.
>

Ah. Thanks to let me know to think this rule.
I agree with your last suggestion.

Could I send patch with it again?

Thanks