Re: [PATCH v2 3/3] mm/slub: Actually fix freelist pointer vs redzoning

From: Vlastimil Babka
Date: Tue Oct 13 2020 - 12:44:20 EST


On 10/9/20 9:54 PM, Kees Cook wrote:
It turns out that SLUB redzoning ("slub_debug=Z") checks from
s->object_size rather than from s->inuse (which is normally bumped to
make room for the freelist pointer), so a cache created with an object
size less than 24 would have their freelist pointer written beyond
s->object_size, causing the redzone to corrupt the freelist pointer.

Is this accurate? Seems to me that redzone is (re)initialized only when freepointer is not active. So it is actually freelist pointer corrupting the redzone...

This was very visible with "slub_debug=ZF":

... as this report shows :)


BUG test (Tainted: G B ): Right Redzone overwritten
-----------------------------------------------------------------------------

INFO: 0xffff957ead1c05de-0xffff957ead1c05df @offset=1502. First byte 0x1a instead of 0xbb
INFO: Slab 0xffffef3950b47000 objects=170 used=170 fp=0x0000000000000000 flags=0x8000000000000200
INFO: Object 0xffff957ead1c05d8 @offset=1496 fp=0xffff957ead1c0620

Redzone (____ptrval____): bb bb bb bb bb bb bb bb ........
Object (____ptrval____): 00 00 00 00 00 f6 f4 a5 ........
Redzone (____ptrval____): 40 1d e8 1a aa @....
Padding (____ptrval____): 00 00 00 00 00 00 00 00 ........

Adjust the offset to stay within s->object_size.

Reported-by: Marco Elver <elver@xxxxxxxxxx>
Link: https://lore.kernel.org/linux-mm/20200807160627.GA1420741@xxxxxxxxxxxxxxxx/
Fixes: 89b83f282d8b (slub: avoid redzone when choosing freepointer location)
Tested-by: Marco Elver <elver@xxxxxxxxxx>
Link: https://lore.kernel.org/lkml/CANpmjNOwZ5VpKQn+SYWovTkFB4VsT-RPwyENBmaK0dLcpqStkA@xxxxxxxxxxxxxx
Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>

Acked-by: Vlastimil Babka <vbabka@xxxxxxx>

This struggle to get it right perhaps calls for some selftests of all combinations of flags that affect object layout, e.g. of redzone/poison/store_user, on sizes from sizeof(void *) to e.g. 3*sizeof(void *), with sanity_checks enabled. Shouldn't be too many tests...

---
mm/slub.c | 14 +++-----------
1 file changed, 3 insertions(+), 11 deletions(-)

diff --git a/mm/slub.c b/mm/slub.c
index 752fad36522c..6f115e56c5d0 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -3637,7 +3637,6 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order)
{
slab_flags_t flags = s->flags;
unsigned int size = s->object_size;
- unsigned int freepointer_area;
unsigned int order;
/*
@@ -3646,13 +3645,6 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order)
* the possible location of the free pointer.
*/
size = ALIGN(size, sizeof(void *));
- /*
- * This is the area of the object where a freepointer can be
- * safely written. If redzoning adds more to the inuse size, we
- * can't use that portion for writing the freepointer, so
- * s->offset must be limited within this for the general case.
- */
- freepointer_area = size;
#ifdef CONFIG_SLUB_DEBUG
/*
@@ -3678,7 +3670,7 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order)
/*
* With that we have determined the number of bytes in actual use
- * by the object. This is the potential offset to the free pointer.
+ * by the object and redzoning.
*/
s->inuse = size;
@@ -3701,13 +3693,13 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order)
*/
s->offset = size;
size += sizeof(void *);
- } else if (freepointer_area > sizeof(void *)) {
+ } else {
/*
* Store freelist pointer near middle of object to keep
* it away from the edges of the object to avoid small
* sized over/underflows from neighboring allocations.
*/
- s->offset = ALIGN(freepointer_area / 2, sizeof(void *));
+ s->offset = ALIGN_DOWN(s->object_size / 2, sizeof(void *));
}
#ifdef CONFIG_SLUB_DEBUG