Re: [PATCH] sched/fair: favor non-idle group in tick preemption

[Abel Wu]

Hi Josh,

On 11/1/22 6:44 AM, Josh Don wrote:
> ...nit...
> Some weirdness about this change though, is that if there is a
> non-idle current entity, and the two next entities on the cfs_rq are
> idle and non-idle respectively, we'll now take longer to preempt the
> on-cpu non-idle entity, because the non-idle entity on the cfs_rq is
> 'hidden' by the idle 'first' entity. Wakeup preemption is different
> because we're always directly comparing the current entity with the
> newly woken entity.

Indeed. The hidden non-idle entity might run longer at the cost of
delayed preemption. This behavior is not compliant to the SCHED_NORMAL
semantics. But we also can't tell from here that the non-idle entity
contains SCHED_NORMAL tasks or not. As long as the entities do not
aware of the schedule policies, this ambiguity exists.

Best,
Abel