Re: [PATCH] sched/fair: Use rq->lock when checking cfs_rq list presence

[Michal Koutný]

On Mon, Oct 11, 2021 at 08:12:08PM +0100, Odin Ugedal <odin@xxxxxxx> wrote:
> To be 100% clear, this can only happen when a control group is
> throttled while it has load
> (cfs_rq_is_decayed(cfs_rq) is false); and then its unthrottling race
> with its deletion?
> Is that a correct understanding Michal?

Yes, that's my working hypothesis but Vincent found a loophole in the
proposed fix under this assumption.


> Do you agree that that will also solve the problem Michal,
> or am I missing something obvious here?

It's not easy for me to verify this with a reproducer and as suggested
by your discomfort, let's dismiss this idea for the time being.

Michal