Received-SPF: pass (google.com: domain of linux-kernel+bounces-184371-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Precedence: bulk
MIME-Version: 1.0
References: <o89373n4-3oq5-25qr-op7n-55p9657r96o8@vanv.qr> <CAHk-=wjxdtkFMB8BPYpU3JedjAsva3XXuzwxtzKoMwQ2e8zRzw@mail.gmail.com>
 <ZkvO-h7AsWnj4gaZ@slm.duckdns.org>
In-Reply-To: <ZkvO-h7AsWnj4gaZ@slm.duckdns.org>
From: Yafang Shao <laoar.shao@gmail.com>
Date: Tue, 21 May 2024 10:33:37 +0800
Message-ID: <CALOAHbCYpV1ubO3Z3hjMWCQnSmGd9-KYARY29p9OnZxMhXKs4g@mail.gmail.com>
Subject: Re: [PATCH workqueue/for-6.10-fixes] workqueue: Refactor worker ID
 formatting and make wq_worker_comm() use full ID string
To: Tejun Heo <tj@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>, Jan Engelhardt <jengelh@inai.de>, 
	Craig Small <csmall@enc.com.au>, linux-kernel@vger.kernel.org, 
	Lai Jiangshan <jiangshanlai@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Tue, May 21, 2024 at 6:30=E2=80=AFAM Tejun Heo <tj@kernel.org> wrote:
>
> Currently, worker ID formatting is open coded in create_worker(),
> init_rescuer() and worker_thread() (for %WORKER_DIE case). The formatted =
ID
> is saved into task->comm and wq_worker_comm() uses it as the base name to
> append extra information to when generating the name to be shown as.
>
> However, TASK_COMM_LEN is only 16 leading to badly truncated names for
> rescuers. For example, the rescuer for the inet_frag_wq workqueue becomes=
:
>
>   $ ps -ef | grep '[k]worker/R-inet'
>   root         483       2  0 Apr26 ?        00:00:00 [kworker/R-inet_]
>
> Even for non-rescue workers, it's easy to run over 15 characters on
> moderately large machines.
>
> Fit it by consolidating worker ID formatting into a new helper
> format_worker_id() and calling it from wq_worker_comm() to obtain the
> untruncated worker ID.
>
>   $ ps -ef | grep '[k]worker/R-inet'
>   root          60       2  0 12:10 ?        00:00:00 [kworker/R-inet_fra=
g_wq]
>
> Signed-off-by: Tejun Heo <tj@kernel.org>
> Reported-by: Jan Engelhardt <jengelh@inai.de>
> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
> ---
> (cc'ing LKML and Lai)
>
> On Mon, May 20, 2024 at 10:17:00AM -0700, Linus Torvalds wrote:
> > Oh well. I suspect this would be trivial to fix. I think the
> > get_kthread_comm() should use the full name, and only *then* attach
> > the extra worker pool information if it exists..
> >
> > Tejun?
>
> Yeah, looks like even the unbound worker IDs are getting truncated on lar=
ger
> machines. This patch should fix it. I'll apply it to wq/for-6.10-fixes so=
on.
>
> > Also, independently of the WQ worker issue, I do think we could
> > possibly expand TASK_COMM_LEN a bit more. 16 bytes is too short for
> > user-level names too, and while it's seldom used for 'ps', it's
> > visible in things like oops messages etc where it gets truncated.
>
> That'd be great. ISTR this being discussed a while ago. Am I imagining th=
at
> we were going to raise this to 32?

We discussed extending it to 24 characters several years ago [0], but
some userspace tools might break. Therefore, we ultimately decided to
dynamically allocate the kthread's full name instead[1].

[0] https://lore.kernel.org/linux-mm/20211101060419.4682-1-laoar.shao@gmail=
com/
[1] https://lore.kernel.org/linux-mm/20211120112738.45980-1-laoar.shao@gmai=
l.com/