On Tue, Jul 28, 2020 at 11:17:03AM -0400, Sasha Levin wrote:
On Tue, Jul 28, 2020 at 08:56:41PM +0800, Muchun Song wrote:
On Mon, Jul 27, 2020 at 10:12 PM Greg Kroah-Hartman gregkh@linuxfoundation.org wrote:
From: Muchun Song songmuchun@bytedance.com
commit d38a2b7a9c939e6d7329ab92b96559ccebf7b135 upstream.
If the kmem_cache refcount is greater than one, we should not mark the root kmem_cache as dying. If we mark the root kmem_cache dying incorrectly, the non-root kmem_cache can never be destroyed. It resulted in memory leak when memcg was destroyed. We can use the following steps to reproduce.
- Use kmem_cache_create() to create a new kmem_cache named A.
- Coincidentally, the kmem_cache A is an alias for kmem_cache B, so the refcount of B is just increased.
- Use kmem_cache_destroy() to destroy the kmem_cache A, just decrease the B's refcount but mark the B as dying.
- Create a new memory cgroup and alloc memory from the kmem_cache B. It leads to create a non-root kmem_cache for allocating memory.
- When destroy the memory cgroup created in the step 4), the non-root kmem_cache can never be destroyed.
If we repeat steps 4) and 5), this will cause a lot of memory leak. So only when refcount reach zero, we mark the root kmem_cache as dying.
Fixes: 92ee383f6daa ("mm: fix race between kmem_cache destroy, create and deactivate") Signed-off-by: Muchun Song songmuchun@bytedance.com Signed-off-by: Andrew Morton akpm@linux-foundation.org Reviewed-by: Shakeel Butt shakeelb@google.com Acked-by: Roman Gushchin guro@fb.com Cc: Vlastimil Babka vbabka@suse.cz Cc: Christoph Lameter cl@linux.com Cc: Pekka Enberg penberg@kernel.org Cc: David Rientjes rientjes@google.com Cc: Joonsoo Kim iamjoonsoo.kim@lge.com Cc: Shakeel Butt shakeelb@google.com Cc: stable@vger.kernel.org Link: http://lkml.kernel.org/r/20200716165103.83462-1-songmuchun@bytedance.com Signed-off-by: Linus Torvalds torvalds@linux-foundation.org Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org
mm/slab_common.c | 35 ++++++++++++++++++++++++++++------- 1 file changed, 28 insertions(+), 7 deletions(-)
--- a/mm/slab_common.c +++ b/mm/slab_common.c @@ -310,6 +310,14 @@ int slab_unmergeable(struct kmem_cache * if (s->refcount < 0) return 1;
+#ifdef CONFIG_MEMCG_KMEM
/** Skip the dying kmem_cache.*/if (s->memcg_params.dying)return 1;+#endif
return 0;}
@@ -832,12 +840,15 @@ static int shutdown_memcg_caches(struct return 0; }
-static void flush_memcg_workqueue(struct kmem_cache *s) +static void memcg_set_kmem_cache_dying(struct kmem_cache *s) { mutex_lock(&slab_mutex); s->memcg_params.dying = true; mutex_unlock(&slab_mutex);
We should remove mutex_lock/unlock(&slab_mutex) here, because we already hold the slab_mutex from kmem_cache_destroy().
Good catch! I backported 63b02ef7dc4e ("mm: memcg/slab: synchronize access to kmem_cache dying flag using a spinlock") instead of changing this patch.
Ah, much better, let me roll my change back and then push out -rc3 with this all fixed up like this.
thanks,
greg k-h