Hello,
On Wed, Feb 12, 2014 at 03:03:31PM -0800, Hugh Dickins wrote:
> From: Filipe Brandenburger <filbranden@xxxxxxxxxx>
>
> Sometimes the cleanup after memcg hierarchy testing gets stuck in
> mem_cgroup_reparent_charges(), unable to bring non-kmem usage down to 0.
>
> There may turn out to be several causes, but a major cause is this: the
> workitem to offline parent can get run before workitem to offline child;
> parent's mem_cgroup_reparent_charges() circles around waiting for the
> child's pages to be reparented to its lrus, but it's holding cgroup_mutex
> which prevents the child from reaching its mem_cgroup_reparent_charges().
>
> Further testing showed that an ordered workqueue for cgroup_destroy_wq
> is not always good enough: percpu_ref_kill_and_confirm's call_rcu_sched
> stage on the way can mess up the order before reaching the workqueue.
>
> Instead, when offlining a memcg, call mem_cgroup_reparent_charges() on
> all its children (and grandchildren, in the correct order) to have their
> charges reparented first.
>
> Fixes: e5fca243abae ("cgroup: use a dedicated workqueue for cgroup destruction")
> Signed-off-by: Filipe Brandenburger <filbranden@xxxxxxxxxx>
> Signed-off-by: Hugh Dickins <hughd@xxxxxxxxxx>
> Cc: stable@xxxxxxxxxxxxxxx # v3.10+ (but will need extra care)
> ---
> Or, you may prefer my alternative cgroup.c approach in 2/2:
> there's no need for both. Please note that neither of these patches
> attempts to handle the unlikely case of racy charges made to child
> after its offline, but parent's offline coming before child's free:
> mem_cgroup_css_free()'s backstop call to mem_cgroup_reparent_charges()
> cannot help in that case, with or without these patches. Fixing that
> would have to be a separate effort - Michal's?
I've changed my mind several times now but I think it'd be a better
idea to stick to this patch, at least for now. This one is easier for
-stable backport and it looks like the requirements for ordering
->css_offline() might go away depending on how reparenting changes
work out.
Reviewed-by: Tejun Heo <tj@xxxxxxxxxx>
Michal, Johannes, can you guys please ack this one if you guys agree?
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
