summaryrefslogtreecommitdiff
path: root/mm
diff options
context:
space:
mode:
authorMichal Hocko <mhocko@suse.cz>2014-01-23 15:53:35 -0800
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>2014-02-13 13:50:11 -0800
commitb0a1f4eca939c37031ca7f91b7e4ef12f6c55e2a (patch)
treefcc658d46cebc1da95616bb209d8aa7b4732383a /mm
parent414f6b9f9fa13c636c7dc136f958cd1911fa62d0 (diff)
memcg: fix endless loop caused by mem_cgroup_iter
commit ecc736fc3c71c411a9d201d8588c9e7e049e5d8c upstream. Hugh has reported an endless loop when the hardlimit reclaim sees the same group all the time. This might happen when the reclaim races with the memcg removal. shrink_zone [rmdir root] mem_cgroup_iter(root, NULL, reclaim) // prev = NULL rcu_read_lock() mem_cgroup_iter_load last_visited = iter->last_visited // gets root || NULL css_tryget(last_visited) // failed last_visited = NULL [1] memcg = root = __mem_cgroup_iter_next(root, NULL) mem_cgroup_iter_update iter->last_visited = root; reclaim->generation = iter->generation mem_cgroup_iter(root, root, reclaim) // prev = root rcu_read_lock mem_cgroup_iter_load last_visited = iter->last_visited // gets root css_tryget(last_visited) // failed [1] The issue seemed to be introduced by commit 5f5781619718 ("memcg: relax memcg iter caching") which has replaced unconditional css_get/css_put by css_tryget/css_put for the cached iterator. This patch fixes the issue by skipping css_tryget on the root of the tree walk in mem_cgroup_iter_load and symmetrically doesn't release it in mem_cgroup_iter_update. Signed-off-by: Michal Hocko <mhocko@suse.cz> Reported-by: Hugh Dickins <hughd@google.com> Tested-by: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Greg Thelen <gthelen@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Diffstat (limited to 'mm')
-rw-r--r--mm/memcontrol.c17
1 files changed, 14 insertions, 3 deletions
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index b55eb75cdf4d..17daed7106ac 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1120,7 +1120,15 @@ mem_cgroup_iter_load(struct mem_cgroup_reclaim_iter *iter,
if (iter->last_dead_count == *sequence) {
smp_rmb();
position = iter->last_visited;
- if (position && !css_tryget(&position->css))
+
+ /*
+ * We cannot take a reference to root because we might race
+ * with root removal and returning NULL would end up in
+ * an endless loop on the iterator user level when root
+ * would be returned all the time.
+ */
+ if (position && position != root &&
+ !css_tryget(&position->css))
position = NULL;
}
return position;
@@ -1129,9 +1137,11 @@ mem_cgroup_iter_load(struct mem_cgroup_reclaim_iter *iter,
static void mem_cgroup_iter_update(struct mem_cgroup_reclaim_iter *iter,
struct mem_cgroup *last_visited,
struct mem_cgroup *new_position,
+ struct mem_cgroup *root,
int sequence)
{
- if (last_visited)
+ /* root reference counting symmetric to mem_cgroup_iter_load */
+ if (last_visited && last_visited != root)
css_put(&last_visited->css);
/*
* We store the sequence count from the time @last_visited was
@@ -1206,7 +1216,8 @@ struct mem_cgroup *mem_cgroup_iter(struct mem_cgroup *root,
memcg = __mem_cgroup_iter_next(root, last_visited);
if (reclaim) {
- mem_cgroup_iter_update(iter, last_visited, memcg, seq);
+ mem_cgroup_iter_update(iter, last_visited, memcg, root,
+ seq);
if (!memcg)
iter->generation++;