Received-SPF: pass (google.com: domain of linux-kernel+bounces-9485-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Precedence: bulk
MIME-Version: 1.0
References: <20231221150238.856960-1-wudaemon@163.com>
In-Reply-To: <20231221150238.856960-1-wudaemon@163.com>
From: Yafang Shao <laoar.shao@gmail.com>
Date: Fri, 22 Dec 2023 16:16:37 +0800
Message-ID: <CALOAHbDXWNbxeQEOhGW5m6bd3cLW_jnE2q6XgSpRBHzxt1GOeg@mail.gmail.com>
Subject: Re: [PATCH v2] sched/rt: Fix rt task's sched latency statistics error
 in sched_stat_wait trace_point
To: Junwen Wu <wudaemon@163.com>
Cc: mingo@redhat.com, peterz@infradead.org, juri.lelli@redhat.com, 
	vincent.guittot@linaro.org, dietmar.eggemann@arm.com, rostedt@goodmis.org, 
	bsegall@google.com, mgorman@suse.de, bristot@redhat.com, vschneid@redhat.com, 
	linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Dec 21, 2023 at 11:03=E2=80=AFPM Junwen Wu <wudaemon@163.com> wrote=
:
>
> When enable sched_stat_wait trace_point, some rt tasks sched latency
> so long, like this:
> sched_stat_wait: comm=3Drcu_preempt pid=3D14 delay=3D4936139545261 [ns]
> Rt task has low latency, it must have a bug. When rt task balance off
> source cpu, dequeue operation not update the sched_statistics, so follow
> update_stats_wait_end_fair update method.
>
> Signed-off-by: Junwen Wu <wudaemon@163.com>
> ---
> Changes since v1:
> https://lore.kernel.org/all/20231218150322.788382-1-wudaemon@163.com/
>
>  kernel/sched/rt.c | 7 +++++++
>  1 file changed, 7 insertions(+)
>
> diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
> index 6aaf0a3d6081..6a2600213991 100644
> --- a/kernel/sched/rt.c
> +++ b/kernel/sched/rt.c
> @@ -1360,12 +1360,19 @@ update_stats_dequeue_rt(struct rt_rq *rt_rq, stru=
ct sched_rt_entity *rt_se,
>                         int flags)
>  {
>         struct task_struct *p =3D NULL;
> +       struct rq *rq =3D rq_of_rt_se(rt_se);
>
>         if (!schedstat_enabled())
>                 return;
>
>         if (rt_entity_is_task(rt_se))
>                 p =3D rt_task_of(rt_se);
> +        /*
> +         * Mark the end of the wait period
> +         * if dequeueing a waiting task.
> +         */
> +       if (p && (p !=3D rq->curr))
> +               update_stats_wait_end_rt(rt_rq, rt_se);

It seems DL has the same issue. Pls. also fix it in update_stats_dequeue_dl=
().
And add the Fixes tag in the commit log:
Fixes: 57a5c2dafca8 ("sched/rt: Support schedstats for RT sched class")
Fixes: b5eb4a5f6521 ("sched/dl: Support schedstats for deadline sched class=
")

>
>         if ((flags & DEQUEUE_SLEEP) && p) {
>                 unsigned int state;
> --
> 2.34.1
>


--=20
Regards
Yafang