Re: [PATCH 2/3] sched/core: Add ENQUEUE_RQ_SELECTED to indicate whether ->select_task_rq() was called

[Peter Zijlstra]

On Fri, Oct 04, 2024 at 10:14:20AM -1000, Tejun Heo wrote:
> On Tue, Oct 01, 2024 at 10:12:53AM -1000, Tejun Heo wrote:
> > On Fri, Sep 27, 2024 at 01:46:12PM -1000, Tejun Heo wrote:
> > > During ttwu, ->select_task_rq() can be skipped if only one CPU is allowed or
> > > migration is disabled. sched_ext schedulers may perform operations such as
> > > direct dispatch from ->select_task_rq() path and it is useful for them to
> > > know whether ->select_task_rq() was skipped in the ->enqueue_task() path.
> > > 
> > > Currently, sched_ext schedulers are using ENQUEUE_WAKEUP for this purpose
> > > and end up assuming incorrectly that ->select_task_rq() was called for tasks
> > > that are bound to a single CPU or migration disabled.
> > > 
> > > Make select_task_rq() indicate whether ->select_task_rq() was called by
> > > setting WF_RQ_SELECTED in *wake_flags and make ttwu_do_activate() map that
> > > to ENQUEUE_RQ_SELECTED for ->enqueue_task().
> > > 
> > > This will be used by sched_ext to fix ->select_task_rq() skip detection.
> > 
> > Peter, ping on patches 1-2. If they look okay, I can route them through
> > sched_ext/for-6.12-fixes.
> 
> As the core changes are on the trivial side, I'll wait till early next week
> and route them through sched_ext/for-6.12-fixes. Please holler for any
> concerns.

Right, I've been busy chasing merge window fallout, and this didn't seem
that urgent. I'm not exactly liking it, but it isn't too horrible. So
yeah, you can take it I suppose.