On Thu 15-04-21 11:13:16, Muchun Song wrote: > On Wed, Apr 14, 2021 at 6:15 PM Michal Hocko <mhocko@xxxxxxxx> wrote: > > > > On Wed 14-04-21 18:04:35, Muchun Song wrote: > > > On Wed, Apr 14, 2021 at 5:24 PM Michal Hocko <mhocko@xxxxxxxx> wrote: > > > > > > > > On Tue 13-04-21 14:51:48, Muchun Song wrote: > > > > > When mm is NULL, we do not need to hold rcu lock and call css_tryget for > > > > > the root memcg. And we also do not need to check !mm in every loop of > > > > > while. So bail out early when !mm. > > > > > > > > mem_cgroup_charge and other callers unconditionally drop the reference > > > > so how come this does not underflow reference count? > > > > > > For the root memcg, the CSS_NO_REF flag is set, so css_get > > > and css_put do not get or put reference. > > > > Ohh, right you are. I must have forgot about that special case. I am > > pretty sure I (and likely few more) will stumble over that in the future > > again. A small comment explaining that the reference can be safely > > ignore would be helpful. > > OK. Will do. I would go with the following if that helps /* * No need to css_get on root memcg as the reference counting is * disabled on the root level in the cgroup core. See CSS_NO_REF */ Thanks -- Michal Hocko SUSE Labs
