Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
Date:   Mon, 12 Dec 2022 02:05:40 +0000
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
From:   "Yajun Deng" <yajun.deng@linux.dev>
Message-ID: <7687e119c7b6beab06c4da41a6b4dcca@linux.dev>
Subject: Re: [PATCH] sched/rt: Add pid and comm when RT throttling
 activated
To:     "Steven Rostedt" <rostedt@goodmis.org>
Cc:     mingo@redhat.com, peterz@infradead.org, juri.lelli@redhat.com,
        vincent.guittot@linaro.org, dietmar.eggemann@arm.com,
        bsegall@google.com, mgorman@suse.de, bristot@redhat.com,
        vschneid@redhat.com, linux-kernel@vger.kernel.org
In-Reply-To: <20221209163606.53a2370e@gandalf.local.home>
References: <20221209163606.53a2370e@gandalf.local.home>
 <20221206083641.103832-1-yajun.deng@linux.dev>
Precedence: bulk

December 10, 2022 5:36 AM, "Steven Rostedt" <rostedt@goodmis.org> wrote:=
=0A=0A> On Tue, 6 Dec 2022 16:36:41 +0800=0A> Yajun Deng <yajun.deng@linu=
x.dev> wrote:=0A> =0A>> It is difficult to know which process consumes ti=
me when RT throttling=0A>> activated.=0A>> =0A>> So add pid and comm for =
this case.=0A> =0A> Unfortunately, that only shows the process that "brok=
e the camel's=0A> back"[1]. The things is, it is the amount of time *all*=
 RT tasks use to=0A> trigger throttling. You are only printing the task t=
hat the throttling=0A> occurred on.=0A> =0A> Try running https://rostedt.=
org/code/migrate.c . It will create a bunch of=0A> RT tasks, and they all=
 will throttle once it hits 950ms ever second.=0A> =0A> trace-cmd record =
-e sched_switch ./migrate=0A> kernelshark=0A> =0A> Then plot all the migr=
ate tasks.=0A> =0A> The above will demonstrate it very well.=0A=0AThanks =
for your patient answer, I'll test it.=0A=0A> =0A> -- Steve=0A> =0A> [1] =
https://en.wikipedia.org/wiki/Straw_that_broke_the_camel's_back=0A> =0A>>=
 Signed-off-by: Yajun Deng <yajun.deng@linux.dev>=0A>> ---=0A>> kernel/sc=
hed/rt.c | 4 +++-=0A>> 1 file changed, 3 insertions(+), 1 deletion(-)=0A>=
> =0A>> diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c=0A>> index ed2=
a47e4ddae..1f267ab1b59a 100644=0A>> --- a/kernel/sched/rt.c=0A>> +++ b/ke=
rnel/sched/rt.c=0A>> @@ -1012,6 +1012,7 @@ static int sched_rt_runtime_ex=
ceeded(struct rt_rq *rt_rq)=0A>> =0A>> if (rt_rq->rt_time > runtime) {=0A=
>> struct rt_bandwidth *rt_b =3D sched_rt_bandwidth(rt_rq);=0A>> + struct=
 task_struct *curr =3D rq_of_rt_rq(rt_rq)->curr;=0A>> =0A>> /*=0A>> * Don=
't actually throttle groups that have no runtime assigned=0A>> @@ -1019,7=
 +1020,8 @@ static int sched_rt_runtime_exceeded(struct rt_rq *rt_rq)=0A>=
> */=0A>> if (likely(rt_b->rt_runtime)) {=0A>> rt_rq->rt_throttled =3D 1;=
=0A>> - printk_deferred_once("sched: RT throttling activated\n");=0A>> + =
printk_deferred_once("pid: %d, comm: %s, sched: RT throttling activated\n=
",=0A>> + curr->pid, curr->comm);=0A>> } else {=0A>> /*=0A>> * In case we=
 did anyway, make it go away,