Hello,
On Fri, Aug 28, 2015 at 07:49:18PM +0300, Vladimir Davydov wrote:
> On Fri, Aug 28, 2015 at 11:25:30AM -0400, Tejun Heo wrote:
> > On the default hierarchy, all memory consumption will be accounted
> > together and controlled by the same set of limits. Always enable
> > kmemcg on the default hierarchy.
>
> IMO we should introduce a boot time knob for disabling it, because kmem
> accounting is still not perfect, besides some users might prefer to go
> w/o it for performance reasons.
Yeah, fair enough.
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index c94b686..8a5dd01 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -4362,6 +4362,13 @@ mem_cgroup_css_online(struct cgroup_subsys_state *css)
> > if (ret)
> > return ret;
> >
> > + /* kmem is always accounted together on the default hierarchy */
> > + if (cgroup_on_dfl(css->cgroup)) {
> > + ret = memcg_activate_kmem(memcg, PAGE_COUNTER_MAX);
> > + if (ret)
> > + return ret;
> > + }
> > +
>
> This is a wrong place for this. The kernel will panic on an attempt to
> create a sub memcg, because memcg_init_kmem already enables kmem
> accounting in this case. I guess we should add this hunk to
> memcg_propagate_kmem instead.
Yeap, bypassing "parent is active" test in memcg_propagate_kmem()
seems like the right thing to do.
Thanks.
--
tejun
--
To unsubscribe from this list: send the line "unsubscribe cgroups" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
