Skip to contentSkip to conversation

- Fix memory corruption in slbq

PATCH In review Jack Steiner 2 participants 4 messages May 1, 2009 – May 3, 2009 Memory management

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 - Fix memory corruption in slbq

Fix memory corruption caused by slqb overrunning the end of the page allocated in kmem_cache_dyn_array_alloc() for initial caches.

---
 mm/slqb.c |    5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
mm/slqb.c +2 -3
Index: linux/mm/slqb.c
===================================================================
@@ -2194,15 +2194,14 @@ static void *kmem_cache_dyn_array_alloc(
 		 * never get freed by definition so we can do it rather
 		 * simply.
 		 */
-		if (!nextmem) {
+		if (!nextmem || offset_in_page(nextmem) + size > PAGE_SIZE) {
 			nextmem = alloc_pages_exact(size, GFP_KERNEL);
 			if (!nextmem)
 				return NULL;
 		}
 		ret = nextmem;
 		nextmem = (void *)((unsigned long)ret + size);
-		if ((unsigned long)ret >> PAGE_SHIFT !=
-				(unsigned long)nextmem >> PAGE_SHIFT)
+		if (offset_in_page(ret) + size >= PAGE_SIZE)
 			nextmem = NULL;
 		memset(ret, 0, size);
 		return ret;
#1 lore raw

Hi Jack,

(Please cc me when sending patches to SL*B.)

Quoted 22 lines · Jack Steiner

Fix memory corruption caused by slqb overrunning the end of the page allocated in kmem_cache_dyn_array_alloc() for initial caches.

---
 mm/slqb.c |    5 ++---

 1 file changed, 2 insertions(+), 3 deletions(-)

Index: linux/mm/slqb.c
===================================================================

mm/slqb.c     2009-04-30 19:08:33.000000000 -0500 renamed from mm/slqb.c        2009-04-30 15:47:16.000000000 -0500 +1 -1
@@ -2194,15 +2194,14 @@ static void *kmem_cache_dyn_array_alloc(
                 * never get freed by definition so we can do it rather
                 * simply.
                 */
-               if (!nextmem) {
+               if (!nextmem || offset_in_page(nextmem) + size > PAGE_SIZE) {
                        nextmem = alloc_pages_exact(size, GFP_KERNEL);
                        if (!nextmem)

mm/slqb.c     2009-04-30 19:08:33.000000000 -0500 static void *kmem_cache_dyn_array_alloc( · line 2207

Quoted 10 lines
mm/slqb.c     2009-04-30 19:08:33.000000000 -0500 renamed from mm/slqb.c        2009-04-30 15:47:16.000000000 -0500 +1 -2
                                return NULL;
                }
                ret = nextmem;
                nextmem = (void *)((unsigned long)ret + size);
-               if ((unsigned long)ret >> PAGE_SHIFT !=
-                               (unsigned long)nextmem >> PAGE_SHIFT)
+               if (offset_in_page(ret) + size >= PAGE_SIZE)
                        nextmem = NULL;
                memset(ret, 0, size);
                return ret;

I don't think this is needed now that commit 8804e6f ("SLQB: fix dumb early allocation cache") is merged:

http://git.kernel.org/?p=linux/kernel/git/penberg/slab-2.6.git;a=commitdiff_plain;h=8804e6f0001668656d4cb2a329e1a6e69e695a0f

Or do you still see the corruption with it?

Pekka

#3 lore raw

Hi Jack,

Quoted 9 lines · Jack Steiner
Quoted 6 lines

I don't think this is needed now that commit 8804e6f ("SLQB: fix dumb early allocation cache") is merged:

http://git.kernel.org/?p=linux/kernel/git/penberg/slab-2.6.git;a=commitdiff_plain;h=8804e6f0001668656d4cb2a329e1a6e69e695a0f

Or do you still see the corruption with it?

No. The above patch is an equivalent ( & better) fix. for the problem. You can ignore the patch that I sent.

OK, thanks for the patch anyway!

Pekka