On Thu, Jan 14, 2016 at 3:38 PM, Michal Hocko [off-list ref] wrote:

On Thu 14-01-16 14:33:52, Martijn Coenen wrote:

quoted

A spare array holding mem cgroup threshold events is kept around
to make sure we can always safely deregister an event and have an
array to store the new set of events in.

In the scenario where we're going from 1 to 0 registered events, the
pointer to the primary array containing 1 event is copied to the spare
slot, and then the spare slot is freed because no events are left.
However, it is freed before calling synchronize_rcu(), which means
readers may still be accessing threshold->primary after it is freed.

Have you seen this triggering in the real life?

It was pretty easy to reproduce in a stress test setup, where we spawn a
process, put it in a mem cgroup and setup the threshold, have it allocate a
lot of memory quickly (crossing the threshold), unregister the event, kill
and repeat. Usually within 30 mins.

quoted

Fixed by only freeing after synchronize_rcu().

Fixes: 8c7577637ca3 ("memcg: free spare array to avoid memory leak")

quoted

Signed-off-by: Martijn Coenen <redacted>

Cc: stable

Acked-by: Michal Hocko <mhocko@suse.com>

Thanks!

quoted

---
 mm/memcontrol.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 14cb1db..73228b6 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c

@@ -3522,16 +3522,17 @@ static void

__mem_cgroup_usage_unregister_event(struct mem_cgroup *memcg,
 swap_buffers:
      /* Swap primary and spare array */
      thresholds->spare = thresholds->primary;
-     /* If all events are unregistered, free the spare array */
-     if (!new) {
-             kfree(thresholds->spare);
-             thresholds->spare = NULL;
-     }

      rcu_assign_pointer(thresholds->primary, new);

      /* To be sure that nobody uses thresholds */
      synchronize_rcu();
+
+     /* If all events are unregistered, free the spare array */
+     if (!new) {
+             kfree(thresholds->spare);
+             thresholds->spare = NULL;
+     }
 unlock:
      mutex_unlock(&memcg->thresholds_lock);
 }
--
2.6.0.rc2.230.g3dd15c0

--
Michal Hocko
SUSE Labs