From: Vladimir Davydov <vdavydov@parallels.com>
Date: Thu, 26 Jun 2014 00:42:31 +0000 (+1000)
Subject: slab: do not keep free objects/slabs on dead memcg caches
X-Git-Url: https://git.karo-electronics.de/?a=commitdiff_plain;h=3a246fba6bea9dfc79cbe8581b560180fba7d210;p=karo-tx-linux.git

slab: do not keep free objects/slabs on dead memcg caches

Since a dead memcg cache is destroyed only after the last slab allocated
to it is freed, we must disable caching of free objects/slabs for such
caches, otherwise they will be hanging around forever.

For SLAB that means we must disable per cpu free object arrays and make
free_block always discard empty slabs irrespective of node's free_limit.

To disable per cpu arrays, we free them on kmem_cache_shrink (see
drain_cpu_caches -> do_drain) and make __cache_free fall back to
free_block if there is no per cpu array.  Also, we have to disable
allocation of per cpu arrays on cpu hotplug for dead caches (see
cpuup_prepare, __do_tune_cpucache).

After we disabled free objects/slabs caching, there is no need to reap
those caches periodically.  Moreover, it will only result in slowdown.  So
we also make cache_reap skip then.

Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
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>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

diff --git a/mm/slab.c b/mm/slab.c
index d60da1dd5e89..f27ad3741de8 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -1262,6 +1262,9 @@ static int cpuup_prepare(long cpu)
 		struct array_cache *shared = NULL;
 		struct array_cache **alien = NULL;
 
+		if (memcg_cache_dead(cachep))
+			continue;
+
 		nc = alloc_arraycache(node, cachep->limit,
 					cachep->batchcount, GFP_KERNEL);
 		if (!nc)
@@ -2462,11 +2465,18 @@ static void do_drain(void *arg)
 
 	check_irq_off();
 	ac = cpu_cache_get(cachep);
+	if (!ac)
+		return;
+
 	n = get_node(cachep, node);
 	spin_lock(&n->list_lock);
 	free_block(cachep, ac->entry, ac->avail, node);
 	spin_unlock(&n->list_lock);
 	ac->avail = 0;
+	if (memcg_cache_dead(cachep)) {
+		cachep->array[smp_processor_id()] = NULL;
+		kfree(ac);
+	}
 }
 
 static void drain_cpu_caches(struct kmem_cache *cachep)
@@ -3417,7 +3427,8 @@ static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects,
 
 		/* fixup slab chains */
 		if (page->active == 0) {
-			if (n->free_objects > n->free_limit) {
+			if (n->free_objects > n->free_limit ||
+			    memcg_cache_dead(cachep)) {
 				n->free_objects -= cachep->num;
 				/* No need to drop any previously held
 				 * lock here, even if we have a off-slab slab
@@ -3511,6 +3522,17 @@ static inline void __cache_free(struct kmem_cache *cachep, void *objp,
 
 	kmemcheck_slab_free(cachep, objp, cachep->object_size);
 
+#ifdef CONFIG_MEMCG_KMEM
+	if (unlikely(!ac)) {
+		int nodeid = page_to_nid(virt_to_page(objp));
+
+		spin_lock(&cachep->node[nodeid]->list_lock);
+		free_block(cachep, &objp, 1, nodeid);
+		spin_unlock(&cachep->node[nodeid]->list_lock);
+		return;
+	}
+#endif
+
 	/*
 	 * Skip calling cache_free_alien() when the platform is not numa.
 	 * This will avoid cache misses that happen while accessing slabp (which
@@ -3851,6 +3873,9 @@ static int __do_tune_cpucache(struct kmem_cache *cachep, int limit,
 	struct ccupdate_struct *new;
 	int i;
 
+	if (memcg_cache_dead(cachep))
+		return 0;
+
 	new = kzalloc(sizeof(*new) + nr_cpu_ids * sizeof(struct array_cache *),
 		      gfp);
 	if (!new)
@@ -4042,6 +4067,9 @@ static void cache_reap(struct work_struct *w)
 	list_for_each_entry(searchp, &slab_caches, list) {
 		check_irq_on();
 
+		if (memcg_cache_dead(searchp))
+			continue;
+
 		/*
 		 * We only take the node lock if absolutely necessary and we
 		 * have established with reasonable certainty that