[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <28267ba4-94a2-4a77-9c34-7273a102c183@amd.com>
Date: Fri, 14 Mar 2025 09:44:45 +0530
From: K Prateek Nayak <kprateek.nayak@....com>
To: Aaron Lu <ziqianlu@...edance.com>, Valentin Schneider
<vschneid@...hat.com>, Ben Segall <bsegall@...gle.com>, Peter Zijlstra
<peterz@...radead.org>, Josh Don <joshdon@...gle.com>, Ingo Molnar
<mingo@...hat.com>, Vincent Guittot <vincent.guittot@...aro.org>
CC: <linux-kernel@...r.kernel.org>, Juri Lelli <juri.lelli@...hat.com>,
Dietmar Eggemann <dietmar.eggemann@....com>, Steven Rostedt
<rostedt@...dmis.org>, Mel Gorman <mgorman@...e.de>, Chengming Zhou
<chengming.zhou@...ux.dev>, Chuyi Zhou <zhouchuyi@...edance.com>
Subject: Re: [RFC PATCH 6/7] sched/fair: fix tasks_rcu with task based
throttle
Hello Aaron,
On 3/13/2025 12:52 PM, Aaron Lu wrote:
> Taskes throttled on exit to user path are scheduled by cond_resched() in
> task_work_run() but that is a preempt schedule and doesn't mark a task
> rcu quiescent state.
>
> Fix this by directly calling schedule() in throttle_cfs_rq_work().
Perhaps that can be gotten around by just using set_ti_thread_flag()
resched_curr() will also call set_preempt_need_resched() which allows
cond_resched() to resched the task.
Since exit_to_user_mode_loop() will run once again seeing that
TIF_NEED_RESCHED is set, schedule() should follow soon. Thoughts?
>
> Signed-off-by: Aaron Lu <ziqianlu@...edance.com>
> ---
> kernel/sched/fair.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index f26d53ac143fe..be96f7d32998c 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -5847,6 +5847,7 @@ static void throttle_cfs_rq_work(struct
> callback_head *work)
> struct cfs_rq *cfs_rq;
> struct rq *rq;
> struct rq_flags rf;
> + bool sched = false;
>
> WARN_ON_ONCE(p != current);
> p->sched_throttle_work.next = &p->sched_throttle_work;
> @@ -5879,9 +5880,13 @@ static void throttle_cfs_rq_work(struct
> callback_head *work)
> dequeue_task_fair(rq, p, DEQUEUE_SLEEP | DEQUEUE_SPECIAL);
> list_add(&p->throttle_node, &cfs_rq->throttled_limbo_list);
> resched_curr(rq);
> + sched = true;
>
> out_unlock:
> task_rq_unlock(rq, p, &rf);
> +
> + if (sched)
> + schedule();
> }
>
> void init_cfs_throttle_work(struct task_struct *p)
--
Thanks and Regards,
Prateek
Powered by blists - more mailing lists