mm/slub: Add debugging to verify correct cache use on kmem_cache_free()
authorChristoph Lameter <cl@linux.com>
Tue, 4 Sep 2012 23:06:14 +0000 (23:06 +0000)
committerPekka Enberg <penberg@kernel.org>
Wed, 5 Sep 2012 09:00:35 +0000 (12:00 +0300)
Add additional debugging to check that the objects is actually from the cache
the caller claims. Doing so currently trips up some other debugging code. It
takes a lot to infer from that what was happening.

Reviewed-by: Glauber Costa <glommer@parallels.com>
Signed-off-by: Christoph Lameter <cl@linux.com>
[ penberg@kernel.org: Use pr_err() ]
Signed-off-by: Pekka Enberg <penberg@kernel.org>
mm/slub.c

index c67bd0a4a952bed22432e5d2a86d269bb9991bae..99059217434063100ed6249012bef1395b3cf81c 100644 (file)
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -2614,6 +2614,13 @@ void kmem_cache_free(struct kmem_cache *s, void *x)
 
        page = virt_to_head_page(x);
 
+       if (kmem_cache_debug(s) && page->slab != s) {
+               pr_err("kmem_cache_free: Wrong slab cache. %s but object"
+                       " is from  %s\n", page->slab->name, s->name);
+               WARN_ON_ONCE(1);
+               return;
+       }
+
        slab_free(s, page, x, _RET_IP_);
 
        trace_kmem_cache_free(_RET_IP_, x);