slub: fix off by one in number of slab tests
authorJoonsoo Kim <iamjoonsoo.kim@lge.com>
Wed, 2 Jul 2014 22:22:35 +0000 (15:22 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 3 Jul 2014 16:21:53 +0000 (09:21 -0700)
min_partial means minimum number of slab cached in node partial list.
So, if nr_partial is less than it, we keep newly empty slab on node
partial list rather than freeing it.  But if nr_partial is equal or
greater than it, it means that we have enough partial slabs so should
free newly empty slab.  Current implementation missed the equal case so
if we set min_partial is 0, then, at least one slab could be cached.
This is critical problem to kmemcg destroying logic because it doesn't
works properly if some slabs is cached.  This patch fixes this problem.

Fixes 91cb69620284 ("slub: make dead memcg caches discard free slabs
immediately").

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/slub.c

index b2b047327d761a46802377b407d992ace14358bc..73004808537ea841e05c85e0b68312ea645eb114 100644 (file)
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1881,7 +1881,7 @@ redo:
 
        new.frozen = 0;
 
-       if (!new.inuse && n->nr_partial > s->min_partial)
+       if (!new.inuse && n->nr_partial >= s->min_partial)
                m = M_FREE;
        else if (new.freelist) {
                m = M_PARTIAL;
@@ -1992,7 +1992,7 @@ static void unfreeze_partials(struct kmem_cache *s,
                                new.freelist, new.counters,
                                "unfreezing slab"));
 
-               if (unlikely(!new.inuse && n->nr_partial > s->min_partial)) {
+               if (unlikely(!new.inuse && n->nr_partial >= s->min_partial)) {
                        page->next = discard_page;
                        discard_page = page;
                } else {
@@ -2620,7 +2620,7 @@ static void __slab_free(struct kmem_cache *s, struct page *page,
                 return;
         }
 
-       if (unlikely(!new.inuse && n->nr_partial > s->min_partial))
+       if (unlikely(!new.inuse && n->nr_partial >= s->min_partial))
                goto slab_empty;
 
        /*