[059/200] mm: memcg/slab: fix use after free in obj_cgroup_charge

Commit Message

Andrew Morton Dec. 15, 2020, 3:06 a.m. UTC

From: Muchun Song <[e-mail]>
Subject: mm: memcg/slab: fix use after free in obj_cgroup_charge

The rcu_read_lock/unlock only can guarantee that the memcg will not be
freed, but it cannot guarantee the success of css_get to memcg.

If the whole process of a cgroup offlining is completed between reading a
objcg->memcg pointer and bumping the css reference on another CPU, and
there are exactly 0 external references to this memory cgroup (how we get
to the obj_cgroup_charge() then?), css_get() can change the ref counter
from 0 back to 1.

Link: https://lkml.kernel.org/r/[e-mail]
Fixes: bf4f059954dc ("mm: memcg/slab: obj_cgroup API")
Signed-off-by: Muchun Song <[e-mail]>
Acked-by: Roman Gushchin <[e-mail]>
Reviewed-by: Shakeel Butt <[e-mail]>
Cc: Johannes Weiner <[e-mail]>
Cc: Michal Hocko <[e-mail]>
Cc: Vladimir Davydov <[e-mail]>
Cc: Joonsoo Kim <[e-mail]>
Cc: Yafang Shao <[e-mail]>
Cc: Chris Down <[e-mail]>
Cc: Christian Brauner <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
---

 mm/memcontrol.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Patch

--- a/mm/memcontrol.c~mm-memcg-slab-fix-use-after-free-in-obj_cgroup_charge
+++ a/mm/memcontrol.c
@@ -3235,8 +3235,10 @@  int obj_cgroup_charge(struct obj_cgroup
 	 * independently later.
 	 */
 	rcu_read_lock();
+retry:
 	memcg = obj_cgroup_memcg(objcg);
-	css_get(&memcg->css);
+	if (unlikely(!css_tryget(&memcg->css)))
+		goto retry;
 	rcu_read_unlock();
 
 	nr_pages = size >> PAGE_SHIFT;