linux-kernel - Re: [PATCH v2 2/3] sched/fair: Skip detach and attach new group task

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 31 May 2016 14:33:45 +0200
From:	Vincent Guittot <vincent.guittot@...aro.org>
To:	Peter Zijlstra <peterz@...radead.org>
Cc:	Yuyang Du <yuyang.du@...el.com>, Ingo Molnar <mingo@...nel.org>,
	linux-kernel <linux-kernel@...r.kernel.org>,
	Benjamin Segall <bsegall@...gle.com>,
	Paul Turner <pjt@...gle.com>,
	Morten Rasmussen <morten.rasmussen@....com>,
	Dietmar Eggemann <dietmar.eggemann@....com>
Subject: Re: [PATCH v2 2/3] sched/fair: Skip detach and attach new group task

On 31 May 2016 at 14:03, Peter Zijlstra <peterz@...radead.org> wrote:
> On Tue, May 31, 2016 at 01:55:49PM +0200, Vincent Guittot wrote:
>> Hi Yuyang,
>>
>> On 31 May 2016 at 00:32, Yuyang Du <yuyang.du@...el.com> wrote:
>> > Vincent reported that the first task to a new task group's cfs_rq will
>> > be attached in attach_task_cfs_rq() and once more when it is enqueued
>> > (see https://lkml.org/lkml/2016/5/25/388).
>> >
>> > Actually, it is worse, attach_task_cfs_rq() is invoked for new task even
>> > way before the new task is initiated in init_entity_runnable_average().
>> >
>> > Solve this by avoiding attach as well as detach new task in
>> > task_move_group_fair(). To do it, we need to know whether the task
>> > is forked or not, so we pass this info all the way from sched_move_task()
>> > to attach_task_cfs_rq().
>>
>> I have tested your patch and I can't the spurious detach
>> /attach_task_cfs_rq anymore
>> so you can add my tested-by.
>>
>
> Could someone update the Changelog to better explain the whole
> callchain.
>
> Because a quick look seems to suggest something like:
>
>         copy_process()
>           sched_fork()
>           ...
>           cgroup_post_fork()
>             ss->fork() := cpu_cgroup_fork()
>
> Which seems to suggest init_entity_runnable_average() is placed wrong
> and should live in sched_fork() ?
>
> Or am I not getting it.. ?

The sched_avg of a task is not used anymore in copy_process with
yuyang's patchset. To be fully correct, we still have a
p->se.avg.last_update_time = 0 in cpu_cgroup_fork but this is just the
side effect of factoring cpu_group_fork and cpu_group_move in
sched_move_task and it will be overwritten by
init_entity_runnable_average before being used.
Now, cpu_cgroup_fork() only sets task group and cfs_rq of the forked task

copy_process
  sched_fork
    ...
    cgroup_post_fork
      ss->fork() := cpu_cgroup_fork() which only set task group and
cfs_rq of the sched_entity of the task

wake_up_new_task
  init_entity_runnable_average
  activate_task
    enqueue_task
      attach_task