SLUB: Don't pass __GFP_FAIL for the initial allocation
authorPekka Enberg <penberg@cs.helsinki.fi>
Wed, 24 Jun 2009 18:59:51 +0000 (21:59 +0300)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 24 Jun 2009 19:20:14 +0000 (12:20 -0700)
SLUB uses higher order allocations by default but falls back to small
orders under memory pressure. Make sure the GFP mask used in the initial
allocation doesn't include __GFP_NOFAIL.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/slub.c

index ce62b77..819f056 100644 (file)
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1085,11 +1085,17 @@ static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
 {
        struct page *page;
        struct kmem_cache_order_objects oo = s->oo;
+       gfp_t alloc_gfp;
 
        flags |= s->allocflags;
 
-       page = alloc_slab_page(flags | __GFP_NOWARN | __GFP_NORETRY, node,
-                                                                       oo);
+       /*
+        * Let the initial higher-order allocation fail under memory pressure
+        * so we fall-back to the minimum order allocation.
+        */
+       alloc_gfp = (flags | __GFP_NOWARN | __GFP_NORETRY) & ~__GFP_NOFAIL;
+
+       page = alloc_slab_page(alloc_gfp, node, oo);
        if (unlikely(!page)) {
                oo = s->min;
                /*