netdev - Re: [PATCH bpf-next v2 4/5] net/flow_dissector: correctly cap nhoff and thoff in case of BPF

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: <20181204233045.GD8717@mini-arch.hsd1.ca.comcast.net>
Date:   Tue, 4 Dec 2018 15:30:45 -0800
From:   Stanislav Fomichev <sdf@...ichev.me>
To:     Song Liu <liu.song.a23@...il.com>
Cc:     Stanislav Fomichev <sdf@...gle.com>,
        Networking <netdev@...r.kernel.org>,
        "David S . Miller" <davem@...emloft.net>,
        Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        simon.horman@...ronome.com
Subject: Re: [PATCH bpf-next v2 4/5] net/flow_dissector: correctly cap nhoff
 and thoff in case of BPF

On 12/04, Song Liu wrote:
> On Mon, Dec 3, 2018 at 8:01 PM Stanislav Fomichev <sdf@...gle.com> wrote:
> >
> > We want to make sure that the following condition holds:
> > 0 <= nhoff <= thoff <= skb->len
> >
> > BPF program can set out-of-bounds nhoff and thoff, which is dangerous, see
> > recent commit d0c081b49137 ("flow_dissector: properly cap thoff field")'.
> >
> > Signed-off-by: Stanislav Fomichev <sdf@...gle.com>
> > ---
> >  net/core/flow_dissector.c | 6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/net/core/flow_dissector.c b/net/core/flow_dissector.c
> > index ac19da6f390b..bb1a54747d64 100644
> > --- a/net/core/flow_dissector.c
> > +++ b/net/core/flow_dissector.c
> > @@ -716,6 +716,10 @@ bool __skb_flow_bpf_dissect(struct bpf_prog *prog,
> >         /* Restore state */
> >         memcpy(cb, &cb_saved, sizeof(cb_saved));
> >
> > +       flow_keys->nhoff = clamp_t(u16, flow_keys->nhoff, 0, skb->len);
> > +       flow_keys->thoff = clamp_t(u16, flow_keys->thoff,
> > +                                  flow_keys->nhoff, skb->len);
> > +
> >         return result == BPF_OK;
> >  }
> >
> > @@ -808,8 +812,6 @@ bool __skb_flow_dissect(const struct sk_buff *skb,
> >                                                      &flow_keys);
> >                         __skb_flow_bpf_to_target(&flow_keys, flow_dissector,
> >                                                  target_container);
> > -                       key_control->thoff = min_t(u16, key_control->thoff,
> > -                                                  skb->len);
> >                         rcu_read_unlock();
> >                         return ret;
> >                 }
> > --
> > 2.20.0.rc1.387.gf8505762e3-goog
> >
> 
> Same question as 3/5:
> 
> Do we need this fix without this set? If yes, do we need it for bpf
> tree as well?
No, for the older versions we do this capping when copying to key_control:
	key_control->thoff = min_t(u16, key_control->thoff,
				   skb->len);

I just moved this logic to the flow_keys and made it more approachable (use
actual clamping with min/max boundary, not cryptic min).

I think my commit message might be confusing. There is no real issue
here, it's done mostly for testing (so we see the result of clamping).

> 
> Thanks,
> Song