Re: [PATCH v1 06/26] mm: memcontrol: return root object cgroup for root memory cgroup

[Qi Zheng]

On 11/19/25 3:24 PM, Harry Yoo wrote:
> On Tue, Nov 18, 2025 at 08:11:04PM +0800, Qi Zheng wrote:
> > On 11/18/25 7:28 PM, Qi Zheng wrote:
> > > Hi Harry,
> > >
> > > On 11/17/25 5:17 PM, Harry Yoo wrote:
> > > > On Tue, Oct 28, 2025 at 09:58:19PM +0800, Qi Zheng wrote:
> > > > > From: Muchun Song <songmuchun@xxxxxxxxxxxxx>
> > > > >
> > > > > Memory cgroup functions such as get_mem_cgroup_from_folio() and
> > > > > get_mem_cgroup_from_mm() return a valid memory cgroup pointer,
> > > > > even for the root memory cgroup. In contrast, the situation for
> > > > > object cgroups has been different.
> > > > >
> > > > > Previously, the root object cgroup couldn't be returned because
> > > > > it didn't exist. Now that a valid root object cgroup exists, for
> > > > > the sake of consistency, it's necessary to align the behavior of
> > > > > object-cgroup-related operations with that of memory cgroup APIs.
> > > > >
> > > > > Signed-off-by: Muchun Song <songmuchun@xxxxxxxxxxxxx>
> > > > > Signed-off-by: Qi Zheng <zhengqi.arch@xxxxxxxxxxxxx>
> > > > > ---
> > > > >    include/linux/memcontrol.h | 29 +++++++++++++++++-------
> > > > >    mm/memcontrol.c            | 45 ++++++++++++++++++++------------------
> > > > >    mm/percpu.c                |  2 +-
> > > > >    3 files changed, 46 insertions(+), 30 deletions(-)
> > > > >
> > > > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > > > > index 6185d8399a54e..9fdbd4970021d 100644
> > > > > --- a/include/linux/memcontrol.h
> > > > > +++ b/include/linux/memcontrol.h
> > > > > @@ -332,6 +332,7 @@ struct mem_cgroup {
> > > > >    #define MEMCG_CHARGE_BATCH 64U
> > > > >    extern struct mem_cgroup *root_mem_cgroup;
> > > > > +extern struct obj_cgroup *root_obj_cgroup;
> > > > >    enum page_memcg_data_flags {
> > > > >        /* page->memcg_data is a pointer to an slabobj_ext vector */
> > > > > @@ -549,6 +550,11 @@ static inline bool
> > > > > mem_cgroup_is_root(struct mem_cgroup *memcg)
> > > > >        return (memcg == root_mem_cgroup);
> > > > >    }
> > > > > +static inline bool obj_cgroup_is_root(const struct obj_cgroup *objcg)
> > > > > +{
> > > > > +    return objcg == root_obj_cgroup;
> > > > > +}
> > > >
> > > > After reparenting, an objcg may satisfy objcg->memcg == root_mem_cgroup
> > > > while objcg != root_obj_cgroup. Should they be considered as
> > > > root objcgs?
> > >
> > > Indeed, it's pointless to charge to root_mem_cgroup (objcg->memcg).
> > >
> > > So it should be:
> > >
> > > static inline bool obj_cgroup_is_root(const struct obj_cgroup *objcg)
> > > {
> > >       return (objcg == root_obj_cgroup) || (objcg->memcg ==
> > > root_mem_cgroup);
> > > }
> >
> > Oh, we can't do that because we still need to consider this objcg when
> > uncharging. Some pages may be charged before reparenting.
>
> Ouch, right. We don't know if it's charged before reparenting and so
> it can break statistics in a few places if we skip uncharging it after
> repareting.

Right.

> And I think we don't charge new pages to objcgs that satisfy
> (objcg->memcg == root_mem_cgroup) && (objcg != root_obj_cgroup)
> after they're reparented anyway...

The charge and uncharge operations must be symmetrical, so we cannot
control the charge operation independently.

Otherwise:

charge
======

if ((objcg->memcg == root_mem_cgroup))
	skip charge this page

uncharge
========

we can't decide whether to skip this page.

Thanks,
Qi


>