Re: [PATCH] sched/core: add forced idle accounting for cgroups

[Josh Don]

On Fri, May 20, 2022 at 11:43 AM Tejun Heo <tj@xxxxxxxxxx> wrote:
>
> Hello,
>
> Sorry about late reply and thanks for the ping. I missed this one.
>
> On Fri, May 13, 2022 at 12:23:16PM -0700, Josh Don wrote:
> > Yea, that's right, this doesn't require the cpu controller to be
> > enabled. Are you suggesting to add a new field to cgroup_base_stat?
>
> Yes, that's what I meant. I think it'd fit there better.

Sounds good, I'll send a second version of the patch with that change.

> > One other weird artifact of collecting forceidle time is that a cpu
> > may account it on behalf of its hyperthread sibling. Currently, the
> > core rstat code always accounts to the current cpu's percpu rstat
> > field. I can add an accounting function to support writes to a
> > different cpu's field, in order to make sure that the per-cpu totals
> > are correct (the forceidle accounting code holds rq->__lock, which
> > protects all HT siblings of a core). percpu totals aren't currently
> > exported in cgroup v2, but this is useful information that we'll
> > consume, so it would be nice to keep it accurate.
>
> Sure, as long as it doesn't incur overhead when not used.

The extra complexity actually doesn't seem to be required. Per-core
totals will be accurate, which is the important part.

Thanks,
Josh