[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YpexCpVBW76GCN2X@carbon>
Date: Wed, 1 Jun 2022 11:33:46 -0700
From: Roman Gushchin <roman.gushchin@...ux.dev>
To: Michal Koutný <mkoutny@...e.com>
Cc: Muchun Song <songmuchun@...edance.com>, hannes@...xchg.org,
mhocko@...nel.org, shakeelb@...gle.com, akpm@...ux-foundation.org,
cgroups@...r.kernel.org, linux-mm@...ck.org,
linux-kernel@...r.kernel.org, duanxiongchun@...edance.com,
longman@...hat.com
Subject: Re: [PATCH v5 03/11] mm: memcontrol: prepare objcg API for non-kmem
usage
On Wed, Jun 01, 2022 at 07:34:34PM +0200, Michal Koutny wrote:
> Hello.
>
> On Mon, May 30, 2022 at 03:49:11PM +0800, Muchun Song <songmuchun@...edance.com> wrote:
> > So we also allocate an object cgroup for the root_mem_cgroup.
>
> This change made me wary that this patch also kmem charging in the
> root_mem_cgroup. Fortunately, get_obj_cgroup_from_current won't return
> this objcg so all is fine.
Yes, I had the same experience here :)
Overall I feel like the handling of the root memcg and objcg are begging
for a cleanup, but it's really far from being trivial.
Maybe starting with documenting how it works now is a good idea...
Powered by blists - more mailing lists