[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAGXu5jJckTkYqWRv5AUv=Ks8_477xuZn=RB+0tiXC=sGDe1QEA@mail.gmail.com>
Date: Tue, 20 Jun 2017 15:51:18 -0700
From: Kees Cook <keescook@...omium.org>
To: Daniel Micay <danielmicay@...il.com>
Cc: "kernel-hardening@...ts.openwall.com" <kernel-hardening@...ts.openwall.com>, David Windsor <dave@...lcore.net>,
Linux-MM <linux-mm@...ck.org>, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 23/23] mm: Allow slab_nomerge to be set
at build time
On Mon, Jun 19, 2017 at 9:09 PM, Daniel Micay <danielmicay@...il.com> wrote:
> On Mon, 2017-06-19 at 16:36 -0700, Kees Cook wrote:
>> Some hardened environments want to build kernels with slab_nomerge
>> already set (so that they do not depend on remembering to set the
>> kernel
>> command line option). This is desired to reduce the risk of kernel
>> heap
>> overflows being able to overwrite objects from merged caches,
>> increasing
>> the difficulty of these attacks. By keeping caches unmerged, these
>> kinds
>> of exploits can usually only damage objects in the same cache (though
>> the
>> risk to metadata exploitation is unchanged).
>
> It also further fragments the ability to influence slab cache layout,
> i.e. primitives to do things like filling up slabs to set things up for
> an exploit might not be able to deal with the target slabs anymore. It
> doesn't need to be mentioned but it's something to think about too. In
> theory, disabling merging can make it *easier* to get the right layout
> too if there was some annoyance that's now split away. It's definitely a
> lot more good than bad for security though, but allocator changes have
> subtle impact on exploitation. This can make caches more deterministic.
Good point about changes to heap grooming; I'll adjust the commit log.
-Kees
--
Kees Cook
Pixel Security
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
