Re: mm, oom_reaper: How to handle race with oom_killer_disable() ?

From	Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Newsgroups	linux.kernel
Subject	Re: mm, oom_reaper: How to handle race with oom_killer_disable() ?
Date	2016-06-21 23:50 +0200
Message-ID	<rMBIu-2Rm-17@gated-at.bofh.it> (permalink)
References	<rMslR-5hj-55@gated-at.bofh.it> <rMslR-5hj-53@gated-at.bofh.it> <rMtUC-6nL-7@gated-at.bofh.it> <rMvWq-7DO-23@gated-at.bofh.it> <rMy7T-xD-3@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

Michal Hocko wrote:
> On Wed 22-06-16 00:32:29, Tetsuo Handa wrote:
> > Michal Hocko wrote:
> [...]
> > > Hmm, what about the following instead. It is rather a workaround than a
> > > full flaged fix but it seems much more easier and shouldn't introduce
> > > new issues.
> > 
> > Yes, I think that will work. But I think below patch (marking signal_struct
> > to ignore TIF_MEMDIE instead of clearing TIF_MEMDIE from task_struct) on top of
> > current linux.git will implement no-lockup requirement. No race is possible unlike
> > "[PATCH 10/10] mm, oom: hide mm which is shared with kthread or global init".
> 
> Not really. Because without the exit_oom_victim from oom_reaper you have
> no guarantee that the oom_killer_disable will ever return. I have
> mentioned that in the changelog. There is simply no guarantee the oom
> victim will ever reach exit_mm->exit_oom_victim.

Why? Since any allocation after setting oom_killer_disabled = true will be
forced to fail, nobody will be blocked on waiting for memory allocation. Thus,
the TIF_MEMDIE tasks will eventually reach exit_mm->exit_oom_victim, won't it?

The only possibility that the TIF_MEMDIE tasks won't reach exit_mm->exit_oom_victim
is __GFP_NOFAIL allocations failing to make forward progress even after
ALLOC_NO_WATERMARKS is used. But that is a different problem which I think
we can call panic() when __GFP_NOFAIL allocations failed after setting
oom_killer_disabled = true.

Thread

Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-13 13:20 +0200
  Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-21 10:40 +0200
    Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-21 13:50 +0200
      Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-21 15:30 +0200
        Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2016-06-21 17:40 +0200
          Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-21 20:00 +0200
            Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2016-06-21 23:50 +0200
              Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-22 08:50 +0200
                Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-22 09:00 +0200
                Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2016-06-22 13:00 +0200
                Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-22 14:10 +0200
                Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Michal Hocko <mhocko@kernel.org> - 2016-06-22 14:20 +0200
    Re: mm, oom_reaper: How to handle race with oom_killer_disable() ? Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2016-06-21 14:00 +0200

csiph-web