Re: [RESEND v12 3/6] mm, oom: cgroup-aware OOM killer

[Michal Hocko]

On Tue 31-10-17 08:04:19, Shakeel Butt wrote:
> > +
> > +static void select_victim_memcg(struct mem_cgroup *root, struct oom_control *oc)
> > +{
> > +       struct mem_cgroup *iter;
> > +
> > +       oc->chosen_memcg = NULL;
> > +       oc->chosen_points = 0;
> > +
> > +       /*
> > +        * The oom_score is calculated for leaf memory cgroups (including
> > +        * the root memcg).
> > +        */
> > +       rcu_read_lock();
> > +       for_each_mem_cgroup_tree(iter, root) {
> > +               long score;
> > +
> > +               if (memcg_has_children(iter) && iter != root_mem_cgroup)
> > +                       continue;
> > +
> 
> Cgroup v2 does not support charge migration between memcgs. So, there
> can be intermediate nodes which may contain the major charge of the
> processes in their leave descendents. Skipping such intermediate nodes
> will kind of protect such processes from oom-killer (lower on the list
> to be killed). Is it ok to not handle such scenario? If yes, shouldn't
> we document it?

Yes, this is a real problem and the one which is not really solvable
without the charge migration. You simply have no clue _who_ owns the
memory so I assume that admins will need to setup the hierarchy which
allows subgroups to migrate tasks to be oom_group.

Or we might want to allow opt-in for charge migration in v2. To be
honest I wasn't completely happy about removing this functionality
altogether in v2 but there was a strong pushback back then that relying
on the charge migration doesn't have any sound usecase.

Anyway, I agree that documentation should be explicit about that.
-- 
Michal Hocko
SUSE Labs