LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* Re: [PATCH] mm/slab_common.c: fix possible spectre-v1 in kmalloc_slab()
@ 2019-05-29 20:31 Alexey Dobriyan
  2019-05-30  0:04 ` Matthew Wilcox
  0 siblings, 1 reply; 11+ messages in thread
From: Alexey Dobriyan @ 2019-05-29 20:31 UTC (permalink / raw)
  To: dianzhangchen0; +Cc: linux-kernel, mhocko, willy

> I think it makes more sense to sanitize size in size_index_elem(),
> don't you?

> -	return (bytes - 1) / 8;
> +	return array_index_nospec((bytes - 1) / 8, ARRAY_SIZE(size_index));

I think it should be fixed in poll.
Literally every small variable kmalloc call is going through this function.

^ permalink raw reply	[flat|nested] 11+ messages in thread
* [PATCH] mm/slab_common.c: fix possible spectre-v1 in kmalloc_slab()
@ 2019-05-29 12:37 Dianzhang Chen
  2019-05-29 16:25 ` Michal Hocko
  2019-05-29 19:48 ` Matthew Wilcox
  0 siblings, 2 replies; 11+ messages in thread
From: Dianzhang Chen @ 2019-05-29 12:37 UTC (permalink / raw)
  To: cl
  Cc: penberg, rientjes, iamjoonsoo.kim, akpm, linux-mm, linux-kernel,
	Dianzhang Chen

The `size` in kmalloc_slab() is indirectly controlled by userspace via syscall: poll(defined in fs/select.c), hence leading to a potential exploitation of the Spectre variant 1 vulnerability.
The `size` can be controlled from: poll -> do_sys_poll -> kmalloc -> __kmalloc -> kmalloc_slab.

Fix this by sanitizing `size` before using it to index size_index.

Signed-off-by: Dianzhang Chen <dianzhangchen0@gmail.com>
---
 mm/slab_common.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 58251ba..41c7e34 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -21,6 +21,7 @@
 #include <asm/tlbflush.h>
 #include <asm/page.h>
 #include <linux/memcontrol.h>
+#include <linux/nospec.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/kmem.h>
@@ -1056,6 +1057,7 @@ struct kmem_cache *kmalloc_slab(size_t size, gfp_t flags)
 		if (!size)
 			return ZERO_SIZE_PTR;
 
+		size = array_index_nospec(size, 193);
 		index = size_index[size_index_elem(size)];
 	} else {
 		if (WARN_ON_ONCE(size > KMALLOC_MAX_CACHE_SIZE))
-- 
2.7.4


^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2019-05-30  7:01 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-05-29 20:31 [PATCH] mm/slab_common.c: fix possible spectre-v1 in kmalloc_slab() Alexey Dobriyan
2019-05-30  0:04 ` Matthew Wilcox
  -- strict thread matches above, loose matches on Subject: below --
2019-05-29 12:37 Dianzhang Chen
2019-05-29 16:25 ` Michal Hocko
2019-05-29 16:39   ` Dianzhang Chen
2019-05-29 17:49     ` Michal Hocko
2019-05-30  5:20       ` Dianzhang Chen
2019-05-30  6:24         ` Michal Hocko
2019-05-30  7:01           ` Dianzhang Chen
2019-05-29 19:48 ` Matthew Wilcox
2019-05-30  5:21   ` Dianzhang Chen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).