Re: [PATCH v2] mm,oom: exclude oom_task_origin processes if they are OOM-unkillable.

[Michal Hocko]

On Thu 18-02-16 19:30:12, Tetsuo Handa wrote:
[...]
> Commit 9cbb78bb314360a8 changed oom_scan_process_thread() to
> always pass memcg == NULL by removing memcg argument from
> oom_scan_process_thread(). As a result, after that commit,
> we are doing test_tsk_thread_flag(p, TIF_MEMDIE) check and
> oom_task_origin(p) check between two oom_unkillable_task()
> calls of memcg OOM case. Why don't we skip these checks by
> passing memcg != NULL to first oom_unkillable_task() call?
> Was this change by error?

I am not really sure I understand your question.  The point is
that mem_cgroup_out_of_memory does for_each_mem_cgroup_tree which
means that only tasks from the given memcg hierarchy is checked and
oom_unkillable_task cares about memcg only for

        /* When mem_cgroup_out_of_memory() and p is not member of the group */
        if (memcg && !task_in_mem_cgroup(p, memcg))
                return true;

which is never true by definition. I guess we can safely remove the memcg
argument from oom_badness and oom_unkillable_task. At least from a quick
glance...
-- 
Michal Hocko
SUSE Labs