memcg: fix possible use-after-free in memcg_kmem_get_cache()

Suppose task @t that belongs to a memory cgroup @memcg is going to
allocate an object from a kmem cache @c.  The copy of @c corresponding to
@memcg, @mc, is empty.  Then if kmem_cache_alloc races with the memory
cgroup destruction we can access the memory cgroup's copy of the cache
after it was destroyed:

CPU0				CPU1
----				----
[ current=@t
  @mc->memcg_params->nr_pages=0 ]

kmem_cache_alloc(@c):
  call memcg_kmem_get_cache(@c);
  proceed to allocation from @mc:
    alloc a page for @mc:
      ...

				move @t from @memcg
				destroy @memcg:
				  mem_cgroup_css_offline(@memcg):
				    memcg_unregister_all_caches(@memcg):
				      kmem_cache_destroy(@mc)

    add page to @mc

We could fix this issue by taking a reference to a per-memcg cache, but
that would require adding a per-cpu reference counter to per-memcg caches,
which would look cumbersome.

Instead, let's take a reference to a memory cgroup, which already has a
per-cpu reference counter, in the beginning of kmem_cache_alloc to be
dropped in the end, and move per memcg caches destruction from css offline
to css free.  As a side effect, per-memcg caches will be destroyed not one
by one, but all at once when the last page accounted to the memory cgroup
is freed.  This doesn't sound as a high price for code readability though.

Note, this patch does add some overhead to the kmem_cache_alloc hot path,
but it is pretty negligible - it's just a function call plus a per cpu
counter decrement, which is comparable to what we already have in
memcg_kmem_get_cache.  Besides, it's only relevant if there are memory
cgroups with kmem accounting enabled.  I don't think we can find a way to
handle this race w/o it, because alloc_page called from kmem_cache_alloc
may sleep so we can't flush all pending kmallocs w/o reference counting.

Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

3 files changed, 27 insertions, 40 deletions

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index dac81b975996..05e1584750ac 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2635,7 +2635,6 @@ static void memcg_register_cache(struct mem_cgroup *memcg,
         if (!cachep)
                 return;
 
-        css_get(&memcg->css);
         list_add(&cachep->memcg_params->list, &memcg->memcg_slab_caches);
 
         /*
@@ -2669,9 +2668,6 @@ static void memcg_unregister_cache(struct kmem_cache *cachep)
         list_del(&cachep->memcg_params->list);
 
         kmem_cache_destroy(cachep);
-
-        /* drop the reference taken in memcg_register_cache */
-        css_put(&memcg->css);
 }
 
 int __memcg_cleanup_cache_params(struct kmem_cache *s)
@@ -2705,9 +2701,7 @@ static void memcg_unregister_all_caches(struct mem_cgroup *memcg)
         mutex_lock(&memcg_slab_mutex);
         list_for_each_entry_safe(params, tmp, &memcg->memcg_slab_caches, list) {
                 cachep = memcg_params_to_cache(params);
-                kmem_cache_shrink(cachep);
-                if (atomic_read(&cachep->memcg_params->nr_pages) == 0)
-                        memcg_unregister_cache(cachep);
+                memcg_unregister_cache(cachep);
         }
         mutex_unlock(&memcg_slab_mutex);
 }
@@ -2742,10 +2736,10 @@ static void __memcg_schedule_register_cache(struct mem_cgroup *memcg,
         struct memcg_register_cache_work *cw;
 
         cw = kmalloc(sizeof(*cw), GFP_NOWAIT);
-        if (cw == NULL) {
-                css_put(&memcg->css);
+        if (!cw)
                 return;
-        }
+
+        css_get(&memcg->css);
 
         cw->memcg = memcg;
         cw->cachep = cachep;
@@ -2776,12 +2770,8 @@ static void memcg_schedule_register_cache(struct mem_cgroup *memcg,
 int __memcg_charge_slab(struct kmem_cache *cachep, gfp_t gfp, int order)
 {
         unsigned int nr_pages = 1 << order;
-        int res;
 
-        res = memcg_charge_kmem(cachep->memcg_params->memcg, gfp, nr_pages);
-        if (!res)
-                atomic_add(nr_pages, &cachep->memcg_params->nr_pages);
-        return res;
+        return memcg_charge_kmem(cachep->memcg_params->memcg, gfp, nr_pages);
 }
 
 void __memcg_uncharge_slab(struct kmem_cache *cachep, int order)
@@ -2789,7 +2779,6 @@ void __memcg_uncharge_slab(struct kmem_cache *cachep, int order)
         unsigned int nr_pages = 1 << order;
 
         memcg_uncharge_kmem(cachep->memcg_params->memcg, nr_pages);
-        atomic_sub(nr_pages, &cachep->memcg_params->nr_pages);
 }
 
 /*
@@ -2816,22 +2805,13 @@ struct kmem_cache *__memcg_kmem_get_cache(struct kmem_cache *cachep)
         if (current->memcg_kmem_skip_account)
                 return cachep;
 
-        rcu_read_lock();
-        memcg = mem_cgroup_from_task(rcu_dereference(current->mm->owner));
-
+        memcg = get_mem_cgroup_from_mm(current->mm);
         if (!memcg_kmem_is_active(memcg))
                 goto out;
 
         memcg_cachep = cache_from_memcg_idx(cachep, memcg_cache_id(memcg));
-        if (likely(memcg_cachep)) {
-                cachep = memcg_cachep;
-                goto out;
-        }
-
-        /* The corresponding put will be done in the workqueue. */
-        if (!css_tryget_online(&memcg->css))
-                goto out;
-        rcu_read_unlock();
+        if (likely(memcg_cachep))
+                return memcg_cachep;
 
         /*
          * If we are in a safe context (can wait, and not in interrupt
@@ -2846,12 +2826,17 @@ struct kmem_cache *__memcg_kmem_get_cache(struct kmem_cache *cachep)
          * defer everything.
          */
         memcg_schedule_register_cache(memcg, cachep);
-        return cachep;
 out:
-        rcu_read_unlock();
+        css_put(&memcg->css);
         return cachep;
 }
 
+void __memcg_kmem_put_cache(struct kmem_cache *cachep)
+{
+        if (!is_root_cache(cachep))
+                css_put(&cachep->memcg_params->memcg->css);
+}
+
 /*
  * We need to verify if the allocation against current->mm->owner's memcg is
  * possible for the given order. But the page is not allocated yet, so we'll
@@ -2914,10 +2899,6 @@ void __memcg_kmem_uncharge_pages(struct page *page, int order)
         memcg_uncharge_kmem(memcg, 1 << order);
         page->mem_cgroup = NULL;
 }
-#else
-static inline void memcg_unregister_all_caches(struct mem_cgroup *memcg)
-{
-}
 #endif /* CONFIG_MEMCG_KMEM */
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
@@ -4188,6 +4169,7 @@ static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 
 static void memcg_destroy_kmem(struct mem_cgroup *memcg)
 {
+        memcg_unregister_all_caches(memcg);
         mem_cgroup_sockets_destroy(memcg);
 }
 #else
@@ -4797,7 +4779,6 @@ static void mem_cgroup_css_offline(struct cgroup_subsys_state *css)
         }
         spin_unlock(&memcg->event_list_lock);
 
-        memcg_unregister_all_caches(memcg);
         vmpressure_cleanup(&memcg->vmpressure);
 }
 
diff --git a/mm/slab.c b/mm/slab.c
index fee275b5b6b7..6042fe57cc60 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -3182,6 +3182,7 @@ slab_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid,
                         memset(ptr, 0, cachep->object_size);
         }
 
+        memcg_kmem_put_cache(cachep);
         return ptr;
 }
 
@@ -3247,6 +3248,7 @@ slab_alloc(struct kmem_cache *cachep, gfp_t flags, unsigned long caller)
                         memset(objp, 0, cachep->object_size);
         }
 
+        memcg_kmem_put_cache(cachep);
         return objp;
 }
 
diff --git a/mm/slub.c b/mm/slub.c
index 765c5884d03d..fe4db9c17238 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1233,13 +1233,17 @@ static inline void kfree_hook(const void *x)
         kmemleak_free(x);
 }
 
-static inline int slab_pre_alloc_hook(struct kmem_cache *s, gfp_t flags)
+static inline struct kmem_cache *slab_pre_alloc_hook(struct kmem_cache *s,
+                                                     gfp_t flags)
 {
         flags &= gfp_allowed_mask;
         lockdep_trace_alloc(flags);
         might_sleep_if(flags & __GFP_WAIT);
 
-        return should_failslab(s->object_size, flags, s->flags);
+        if (should_failslab(s->object_size, flags, s->flags))
+                return NULL;
+
+        return memcg_kmem_get_cache(s, flags);
 }
 
 static inline void slab_post_alloc_hook(struct kmem_cache *s,
@@ -1248,6 +1252,7 @@ static inline void slab_post_alloc_hook(struct kmem_cache *s,
         flags &= gfp_allowed_mask;
         kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
         kmemleak_alloc_recursive(object, s->object_size, 1, s->flags, flags);
+        memcg_kmem_put_cache(s);
 }
 
 static inline void slab_free_hook(struct kmem_cache *s, void *x)
@@ -2384,10 +2389,9 @@ static __always_inline void *slab_alloc_node(struct kmem_cache *s,
         struct page *page;
         unsigned long tid;
 
-        if (slab_pre_alloc_hook(s, gfpflags))
+        s = slab_pre_alloc_hook(s, gfpflags);
+        if (!s)
                 return NULL;
-
-        s = memcg_kmem_get_cache(s, gfpflags);
 redo:
         /*
          * Must read kmem_cache cpu data via this cpu ptr. Preemption is