On Fri 24-06-16 01:24:46, Tetsuo Handa wrote:
> I missed that victim != p case needs to use get_task_struct(). Patch updated.
> ----------------------------------------
> >From 1819ec63b27df2d544f66482439e754d084cebed Mon Sep 17 00:00:00 2001
> From: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
> Date: Fri, 24 Jun 2016 01:16:02 +0900
> Subject: [PATCH v2] mm, oom: don't set TIF_MEMDIE on a mm-less thread.
>
> Patch "mm, oom: fortify task_will_free_mem" removed p->mm != NULL test for
> shortcut path in oom_kill_process(). But since commit f44666b04605d1c7
> ("mm,oom: speed up select_bad_process() loop") changed to iterate using
> thread group leaders, the possibility of p->mm == NULL has increased
> compared to when commit 83363b917a2982dd ("oom: make sure that TIF_MEMDIE
> is set under task_lock") was proposed. On CONFIG_MMU=n kernels, nothing
> will clear TIF_MEMDIE and the system can OOM livelock if TIF_MEMDIE was
> by error set to a mm-less thread group leader.
>
> Let's do steps for regular path except printing OOM killer messages and
> sending SIGKILL.
I fully agree with Oleg. It would be much better to encapsulate this
into mark_oom_victim and guard it by ifdef NOMMU as this is nommu
specific with a big fat warning why we need it.
--
Michal Hocko
SUSE Labs
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>