On Tue, Jul 14, 2020 at 2:43 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote:
>
> On Tue 14-07-20 08:50:57, Tetsuo Handa wrote:
> > On 2020/07/11 12:18, Yafang Shao wrote:
> > > If the current's MMF_OOM_SKIP is set, it means that the current is exiting
> > > or dying and likely to realease its address space. So we don't need to
> > > invoke the oom killer again. Otherwise that may cause some unexpected
> > > issues, for example, bellow is the issue found in our production
> > > environment.
> >
> > What kernel version are you using?
> >
> > Commit 7775face207922ea ("memcg: killed threads should not invoke memcg OOM killer")
> > should solve this problem.
>
> I haven't really seen the oom report but this would solve only half of
> the problem. It is still possible that other tasks from the oom memcg
> hierarchy can be blocked on the oom lock and then observe the oom reaped
> tasks. The trylock or the check for the margin is needed to address the
> other half of the problem.
Agreed. Different tasks doing memcg oom in the same memcg should also
be resolved.
I will send a patch for it.
--
Thanks
Yafang
