Re: [RFC][PATCH] mm/slub.c: Allow poisoning to use the fast path

From: Kees Cook
Date: Mon Aug 07 2017 - 18:23:21 EST

On Mon, Aug 7, 2017 at 3:00 PM, Laura Abbott <labbott@xxxxxxxxxx> wrote:
> On 08/07/2017 11:03 AM, Christopher Lameter wrote:
>> On Mon, 7 Aug 2017, Laura Abbott wrote:
>>>> Ok I see that the objects are initialized with poisoning and redzoning but
>>>> I do not see that there is fastpath code to actually check the values
>>>> before the object is reinitialized. Is that intentional or am
>>>> I missing something?
>>> Yes, that's intentional here. I see the validation as a separate more
>>> expensive feature. I had a crude patch to do some checks for testing
>>> and I know Daniel Micay had an out of tree patch to do some checks
>>> as well.
>> Ok then this patch does nothing? How does this help?
> The purpose of this patch is to ensure the poisoning can happen without
> too much penalty. Even if there aren't checks to abort/warn when there
> is a problem, there's still value in ensuring objects are always poisoned.

To clarify, this is desirable to kill exploitation of
exposure-after-free flaws and some classes of use-after-free flaws,
since the contents will have be wiped out after a free. (Verification
of poison is nice, but is expensive compared to the benefit against
these exploits -- and notably doesn't protect against the other
use-after-free attacks where the contents are changed after the next
allocation, which would have passed the poison verification.)


Kees Cook
Pixel Security