From: Valentin Schneider <vschneid@redhat.com>
To: Chengming Zhou <zhouchengming@bytedance.com>,
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
Cc: linux-kernel@vger.kernel.org
Subject: Re: [PATCH] sched/core: Minor optimize ttwu_runnable()
Date: Mon, 07 Nov 2022 12:56:55 +0000 [thread overview]
Message-ID: <xhsmhwn86ncrs.mognet@vschneid.remote.csb> (raw)
In-Reply-To: <7b8bd69a-39cb-a1f0-6ccd-7b0810ef0db7@bytedance.com>
On 05/11/22 09:34, Chengming Zhou wrote:
> On 2022/11/5 02:19, Valentin Schneider wrote:
>> On 02/11/22 18:23, Chengming Zhou wrote:
>>> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
>>> index 87c9cdf37a26..3785418de127 100644
>>> --- a/kernel/sched/core.c
>>> +++ b/kernel/sched/core.c
>>> @@ -3718,9 +3718,8 @@ static int ttwu_runnable(struct task_struct *p, int wake_flags)
>>>
>>> rq = __task_rq_lock(p, &rf);
>>> if (task_on_rq_queued(p)) {
>>> - /* check_preempt_curr() may use rq clock */
>>> - update_rq_clock(rq);
>>> - ttwu_do_wakeup(rq, p, wake_flags, &rf);
>>> + WRITE_ONCE(p->__state, TASK_RUNNING);
>>> + trace_sched_wakeup(p);
>>
>> This also loses the class->task_woken() call, AFAICT we could have
>> !p->on_cpu here (e.g. an IRQ hit before the task got to schedule()), but
>> then again if there is a need to push we should have done that at the IRQ
>> preempt via set_next_task_{rt, dl}()... So I'm starting to think this is
>> OK, but that needs elaborating in the changelog.
>
> Sorry, I don't get why we could have !p->on_cpu here?
>
> I thought if we have task_on_rq_queued(p) here, it means p hasn't got to
> __schedule() to deactivate_task(), so p should still be on_cpu?
>
> Thanks.
>
So p is still on the rq for sure, but it may not be the currently running
task. Consider, on a CONFIG_PREEMPT kernel, task p0 in a wait loop:
0 for (;;) {
1 set_current_state(TASK_UNINTERRUPTIBLE);
2
3 if (CONDITION)
4 break;
5
6 schedule();
7 }
8 __set_current_state(TASK_RUNNING);
Now further consider p0 executes line 1, but then gets interrupted by an
IRQ, at the end of which preempt_schedule_irq() happens. We enter
__schedule(SM_PREEMPT), so p0 isn't dequeued due to its non-zero task
state, but we *can* end up switching the CPU's current task.
ISTR that's why Peter renamed that function ttwu_*runnable*() and not
ttwu_*running*().
>>
>>
>>> ret = 1;
>>> }
>>> __task_rq_unlock(rq, &rf);
>>> --
>>> 2.37.2
>>
next prev parent reply other threads:[~2022-11-07 12:58 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-11-02 10:23 Chengming Zhou
2022-11-04 18:19 ` Valentin Schneider
2022-11-05 1:34 ` Chengming Zhou
2022-11-07 12:56 ` Valentin Schneider [this message]
2022-11-07 13:19 ` Chengming Zhou
2022-11-07 15:54 ` Valentin Schneider
2022-11-08 7:38 ` Chengming Zhou
2022-11-08 9:11 ` Peter Zijlstra
2022-11-08 10:08 ` Peter Zijlstra
2022-11-08 11:17 ` Chengming Zhou
2022-11-08 9:20 ` Peter Zijlstra
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=xhsmhwn86ncrs.mognet@vschneid.remote.csb \
--to=vschneid@redhat.com \
--cc=bristot@redhat.com \
--cc=bsegall@google.com \
--cc=dietmar.eggemann@arm.com \
--cc=juri.lelli@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mgorman@suse.de \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=vincent.guittot@linaro.org \
--cc=zhouchengming@bytedance.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®