[PATCH] mm: add noaliencache boot option to disable numa alien caches

When using numa=fake on non-NUMA hardware there is no benefit to having the
alien caches, and they consume much memory.

Add a kernel boot option to disable them.

Christoph sayeth "This is good to have even on large NUMA.  The problem is
that the alien caches grow by the square of the size of the system in terms of
nodes."

Cc: Christoph Lameter <clameter@engr.sgi.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

2 files changed, 33 insertions, 9 deletions

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 2e1898e..2ddc43e 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -1012,6 +1012,10 @@ and is between 256 and 4096 characters. It is defined in the file
 			emulation library even if a 387 maths coprocessor
 			is present.
 
+	noaliencache	[MM, NUMA] Disables the allcoation of alien caches in
+			the slab allocator.  Saves per-node memory, but will
+			impact performance on real NUMA hardware.
+
 	noalign		[KNL,ARM]
 
 	noapic		[SMP,APIC] Tells the kernel to not make use of any
diff --git a/mm/slab.c b/mm/slab.c
index 3318252..bfd654c 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -869,6 +869,22 @@ static void __slab_error(const char *function, struct kmem_cache *cachep,
 	dump_stack();
 }
 
+/*
+ * By default on NUMA we use alien caches to stage the freeing of
+ * objects allocated from other nodes. This causes massive memory
+ * inefficiencies when using fake NUMA setup to split memory into a
+ * large number of small nodes, so it can be disabled on the command
+ * line
+  */
+
+static int use_alien_caches __read_mostly = 1;
+static int __init noaliencache_setup(char *s)
+{
+	use_alien_caches = 0;
+	return 1;
+}
+__setup("noaliencache", noaliencache_setup);
+
 #ifdef CONFIG_NUMA
 /*
  * Special reaping functions for NUMA systems called from cache_reap().
@@ -1117,7 +1133,7 @@ static inline int cache_free_alien(struct kmem_cache *cachep, void *objp)
 	 * Make sure we are not freeing a object from another node to the array
 	 * cache on this cpu.
 	 */
-	if (likely(slabp->nodeid == node))
+	if (likely(slabp->nodeid == node) || unlikely(!use_alien_caches))
 		return 0;
 
 	l3 = cachep->nodelists[node];
@@ -1195,7 +1211,7 @@ static int __cpuinit cpuup_callback(struct notifier_block *nfb,
 		list_for_each_entry(cachep, &cache_chain, next) {
 			struct array_cache *nc;
 			struct array_cache *shared;
-			struct array_cache **alien;
+			struct array_cache **alien = NULL;
 
 			nc = alloc_arraycache(node, cachep->limit,
 						cachep->batchcount);
@@ -1207,9 +1223,11 @@ static int __cpuinit cpuup_callback(struct notifier_block *nfb,
 			if (!shared)
 				goto bad;
 
-			alien = alloc_alien_cache(node, cachep->limit);
-			if (!alien)
-				goto bad;
+			if (use_alien_caches) {
+                                alien = alloc_alien_cache(node, cachep->limit);
+                                if (!alien)
+                                        goto bad;
+                        }
 			cachep->array[cpu] = nc;
 			l3 = cachep->nodelists[node];
 			BUG_ON(!l3);
@@ -3590,13 +3608,15 @@ static int alloc_kmemlist(struct kmem_cache *cachep)
 	int node;
 	struct kmem_list3 *l3;
 	struct array_cache *new_shared;
-	struct array_cache **new_alien;
+	struct array_cache **new_alien = NULL;
 
 	for_each_online_node(node) {
 
-		new_alien = alloc_alien_cache(node, cachep->limit);
-		if (!new_alien)
-			goto fail;
+                if (use_alien_caches) {
+                        new_alien = alloc_alien_cache(node, cachep->limit);
+                        if (!new_alien)
+                                goto fail;
+                }
 
 		new_shared = alloc_arraycache(node,
 				cachep->shared*cachep->batchcount,