linux-kernel - Re: [PATCH] mm/slab_common.c: fix possible spectre-v1 in kmalloc

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20190530062418.GB6703@dhcp22.suse.cz>
Date:   Thu, 30 May 2019 08:24:18 +0200
From:   Michal Hocko <mhocko@...nel.org>
To:     Dianzhang Chen <dianzhangchen0@...il.com>
Cc:     cl@...ux.com, penberg@...nel.org, rientjes@...gle.com,
        iamjoonsoo.kim@....com, akpm@...ux-foundation.org,
        linux-mm@...ck.org, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] mm/slab_common.c: fix possible spectre-v1 in
 kmalloc_slab()

[Please do not top-post]

On Thu 30-05-19 13:20:01, Dianzhang Chen wrote:
> It is possible that a CPU mis-predicts the conditional branch, and
> speculatively loads size_index[size_index_elem(size)], even if size >192.
> Although this value will subsequently be discarded,
> but it can not drop all the effects of speculative execution,
> such as the presence or absence of data in caches. Such effects may
> form side-channels which can be
> observed to extract secret information.

I understand the general mechanism of spectre v1. What I was asking for
is an example of where userspace directly controls the allocation size
as this is usually bounded to an in kernel object size. I can see how
and N * sizeof(object) where N is controlled by the userspace could be
the target. But calling that out explicitly would be appreciated.
 
> As for "why this particular path a needs special treatment while other
> size branches are ok",
> i think the other size branches need to treatment as well at first place,
> but in code `index = fls(size - 1)` the function `fls` will make the
> index at specific range,
> so it can not use `kmalloc_caches[kmalloc_type(flags)][index]` to load
> arbitury data.
> But, still it may load some date that it shouldn't, if necessary, i
> think can add array_index_nospec as well.

Please mention that in the changelog as well.
 
> On Thu, May 30, 2019 at 1:49 AM Michal Hocko <mhocko@...nel.org> wrote:
> >
> > On Thu 30-05-19 00:39:53, Dianzhang Chen wrote:
> > > It's come from `192+1`.
> > >
> > >
> > > The more code fragment is:
> > >
> > >
> > > if (size <= 192) {
> > >
> > >     if (!size)
> > >
> > >         return ZERO_SIZE_PTR;
> > >
> > >     size = array_index_nospec(size, 193);
> > >
> > >     index = size_index[size_index_elem(size)];
> > >
> > > }
> >
> > OK I see, I could have looked into the code, my bad. But I am still not
> > sure what is the potential exploit scenario and why this particular path
> > a needs special treatment while other size branches are ok. Could you be
> > more specific please?
> > --
> > Michal Hocko
> > SUSE Labs

-- 
Michal Hocko
SUSE Labs