2022-02-24 15:22:25

by Marcelo Tosatti

Subject: Re: [RFC PATCH] tick/sched: Ensure quiet_vmstat() is called when the idle tick was stopped too

On Thu, Feb 24, 2022 at 01:28:16PM +0000, Aaron Tomlin wrote:
> On Thu 2022-02-24 10:14 -0300, Marcelo Tosatti wrote:
> > If the per-CPU vmstat_update is limited to happen once per second, that
> > shouldnt be a significant performance impact?
>
> Perhaps not. Albeit, is the interrupt worth it?

Its a requirement for correctness right?

> Then again it could indeed
> be a long time before the idle task is selected and a return to idle code
> were we'd check for any remaining differentials with the aforementioned
> patch.

Or a long time before userspace returns to kernel.

2022-02-24 16:23:33

by Aaron Tomlin

Subject: Re: [RFC PATCH] tick/sched: Ensure quiet_vmstat() is called when the idle tick was stopped too

On Thu 2022-02-24 10:40 -0300, Marcelo Tosatti wrote:
> On Thu, Feb 24, 2022 at 01:28:16PM +0000, Aaron Tomlin wrote:
> > On Thu 2022-02-24 10:14 -0300, Marcelo Tosatti wrote:
> > > If the per-CPU vmstat_update is limited to happen once per second, that
> > > shouldnt be a significant performance impact?
> >
> > Perhaps not. Albeit, is the interrupt worth it?
>
> Its a requirement for correctness right?

Yes, this is true.

> > Then again it could indeed
> > be a long time before the idle task is selected and a return to idle code
> > were we'd check for any remaining differentials with the aforementioned
> > patch.
>
> Or a long time before userspace returns to kernel.

Indeed. I'll put together a patch for comment.


Kind regards,

--
Aaron Tomlin