[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <a6a3961b-19a9-2476-effa-33bee33dd57b@samba.org>
Date: Thu, 1 Apr 2021 16:40:31 +0200
From: Stefan Metzmacher <metze@...ba.org>
To: Jens Axboe <axboe@...nel.dk>,
Linus Torvalds <torvalds@...ux-foundation.org>,
"Eric W. Biederman" <ebiederm@...ssion.com>
Cc: io-uring <io-uring@...r.kernel.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Oleg Nesterov <oleg@...hat.com>
Subject: Re: [PATCH 0/2] Don't show PF_IO_WORKER in /proc/<pid>/task/
Hi Jens,
>> I know you brought this one up as part of your series, not sure I get
>> why you want it owned by root and read-only? cmdline and exe, yeah those
>> could be hidden, but is there really any point?
>>
>> Maybe I'm missing something here, if so, do clue me in!
>
> I looked through /proc and I think it's mostly similar to
> the unshare() case, if userspace wants to do stupid things
> like changing "comm" of iothreads, it gets what was asked for.
>
> But the "cmdline" hiding would be very useful.
>
> While most tools use "comm", by default.
>
> ps -eLf or 'iotop' use "cmdline".
>
> Some processes use setproctitle to change "cmdline" in order
> to identify the process better, without the 15 chars comm restriction,
> that's why I very often press 'c' in 'top' to see the cmdline,
> in that case it would be very helpful to see '[iou-wrk-1234]'
> instead of the seeing the cmdline.
>
> So I'd very much prefer if this could be applied:
> https://lore.kernel.org/io-uring/d4487f959c778d0b1d4c5738b75bcff17d21df5b.1616197787.git.metze@samba.org/T/#u
>
> If you want I can add a comment and a more verbose commit message...
I noticed that 'iotop' actually appends ' [iou-wrk-1234]' to the cmdline value,
so that leaves us with 'ps -eLf' and 'top' (with 'c').
pstree -a -t -p is also fine:
│ └─io_uring-cp,1315 /root/kernel/linux-image-5.12.0-rc2+-dbg_5.12.0-rc2+-5_amd64.deb file
│ ├─{iou-mgr-1315},1316
│ ├─{iou-wrk-1315},1317
│ ├─{iou-wrk-1315},1318
│ ├─{iou-wrk-1315},1319
│ ├─{iou-wrk-1315},1320
In the spirit of "avoid special PF_IO_WORKER checks" I guess it's ok
to leave of as is...
metze
Powered by blists - more mailing lists