[patch] memcg: disable hierarchy support if bound to the legacy cgroup hierarchy

From: Vladimir Davydov <redacted>

If the memory cgroup controller is initially mounted in the scope of the
default cgroup hierarchy and then remounted to a legacy hierarchy, it
will still have hierarchy support enabled, which is incorrect. We should
disable hierarchy support if bound to the legacy cgroup hierarchy.

Signed-off-by: Vladimir Davydov <redacted>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/memcontrol.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Andrew, could you please pick this up for 4.0?  I don't think it's
urgent enough for -stable, though.  Thanks!

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 0c86945bcc9a..68d4890fc4bd 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c

@@ -5238,7 +5238,9 @@ static void mem_cgroup_bind(struct cgroup_subsys_state *root_css)
 	 * on for the root memcg is enough.
 	 */
 	if (cgroup_on_dfl(root_css->cgroup))
-		mem_cgroup_from_css(root_css)->use_hierarchy = true;
+		root_mem_cgroup->use_hierarchy = true;
+	else
+		root_mem_cgroup->use_hierarchy = false;
 }
 
 static u64 memory_current_read(struct cgroup_subsys_state *css,

-- 
2.3.0

On Mon 02-03-15 12:38:59, Johannes Weiner wrote:

From: Vladimir Davydov <redacted>

If the memory cgroup controller is initially mounted in the scope of the
default cgroup hierarchy and then remounted to a legacy hierarchy, it
will still have hierarchy support enabled, which is incorrect. We should
disable hierarchy support if bound to the legacy cgroup hierarchy.

Signed-off-by: Vladimir Davydov <redacted>
Signed-off-by: Johannes Weiner <redacted>

I had this one on the radar but I wasn't sure about the first cgroup
part so I haven't acked it. Once Tejun picked it up this one is good as
well.

Acked-by: Michal Hocko <redacted>

-- 
Michal Hocko
SUSE Labs