[PATCH] slub: Potential stack overflow

[Eric Dumazet <eric.dumazet@gmail.com>]

I discovered that we can overflow stack if CONFIG_SLUB_DEBUG=y and use
slabs with many objects, since list_slab_objects() and process_slab()
use DECLARE_BITMAP(map, page->objects);

With 65535 bits, we use 8192 bytes of stack ...

A possible fix is to lower MAX_OBJS_PER_PAGE so that these bitmaps dont
use more than a third of THREAD_SIZE. I suspect plain memory allocation
in these functions is not an option.

Using non dynamic stack allocation makes the problem more obvious if
somebody runs checkstack.pl

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
diff --git a/mm/slub.c b/mm/slub.c
index b364844..adf04c1 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -167,7 +167,13 @@
 
 #define OO_SHIFT	16
 #define OO_MASK		((1 << OO_SHIFT) - 1)
+
+#ifdef CONFIG_SLUB_DEBUG
+/* We use an onstack BITMAP while debugging, make sure this wont be too big */
+#define MAX_OBJS_PER_PAGE	min_t(int, 65535, 8*(THREAD_SIZE/3))
+#else
 #define MAX_OBJS_PER_PAGE	65535 /* since page.objects is u16 */
+#endif
 
 /* Internal SLUB flags */
 #define __OBJECT_POISON		0x80000000 /* Poison object */
@@ -2426,7 +2432,7 @@ static void list_slab_objects(struct kmem_cache *s, struct page *page,
 #ifdef CONFIG_SLUB_DEBUG
 	void *addr = page_address(page);
 	void *p;
-	DECLARE_BITMAP(map, page->objects);
+	DECLARE_BITMAP(map, MAX_OBJS_PER_PAGE);
 
 	bitmap_zero(map, page->objects);
 	slab_err(s, page, "%s", text);
@@ -3651,7 +3657,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s,
 		struct page *page, enum track_item alloc)
 {
 	void *addr = page_address(page);
-	DECLARE_BITMAP(map, page->objects);
+	DECLARE_BITMAP(map, MAX_OBJS_PER_PAGE);
 	void *p;
 
 	bitmap_zero(map, page->objects);