[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20251128134427.GS4068168@noisy.programming.kicks-ass.net>
Date: Fri, 28 Nov 2025 14:44:27 +0100
From: Peter Zijlstra <peterz@...radead.org>
To: Kuba Piecuch <jpiecuch@...gle.com>
Cc: mingo@...nel.org, vincent.guittot@...aro.org,
linux-kernel@...r.kernel.org, juri.lelli@...hat.com,
dietmar.eggemann@....com, rostedt@...dmis.org, bsegall@...gle.com,
mgorman@...e.de, vschneid@...hat.com, tj@...nel.org,
void@...ifault.com, arighi@...dia.com, changwoo@...lia.com,
sched-ext@...ts.linux.dev
Subject: Re: [PATCH 5/5] sched: Rework sched_class::wakeup_preempt() and
rq_modified_*()
On Fri, Nov 28, 2025 at 02:36:38PM +0100, Peter Zijlstra wrote:
> On Fri, Nov 28, 2025 at 01:26:30PM +0000, Kuba Piecuch wrote:
> > Hi Peter,
> >
> > On Thu Nov 27, 2025 at 3:39 PM UTC, Peter Zijlstra wrote:
> > > Additionally have set_next_task() re-set the value to the current class.
> >
> > I don't see this part reflected in the patch. Is something missing?
>
> Hmm, that does appear to have gone walk-about :/
Aah, here:
@@ -6797,6 +6799,7 @@ static void __sched notrace __schedule(i
pick_again:
next = pick_next_task(rq, rq->donor, &rf);
rq_set_donor(rq, next);
+ rq->next_class = next->sched_class;
if (unlikely(task_is_blocked(next))) {
next = find_proxy_task(rq, next, &rf);
if (!next)
Will fix changelog. Had to do the above instead of set_next_task()
because if proxy stuff.
Powered by blists - more mailing lists