linux-kernel - Re: [PATCH net] Drop packets with invalid headers to prevent KMSAN infoleak

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: <5c8fb835-b0cb-428b-ab07-e20f905eb19f@linux.dev>
Date: Tue, 22 Oct 2024 11:14:06 -0700
From: Martin KaFai Lau <martin.lau@...ux.dev>
To: Daniel Yang <danielyangkang@...il.com>
Cc: Daniel Borkmann <daniel@...earbox.net>,
 John Fastabend <john.fastabend@...il.com>,
 Alexei Starovoitov <ast@...nel.org>, Andrii Nakryiko <andrii@...nel.org>,
 Eduard Zingerman <eddyz87@...il.com>, Song Liu <song@...nel.org>,
 Yonghong Song <yonghong.song@...ux.dev>, KP Singh <kpsingh@...nel.org>,
 Stanislav Fomichev <sdf@...ichev.me>, Hao Luo <haoluo@...gle.com>,
 Jiri Olsa <jolsa@...nel.org>, "David S. Miller" <davem@...emloft.net>,
 Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>,
 Paolo Abeni <pabeni@...hat.com>,
 "open list:BPF [NETWORKING] (tcx & tc BPF, sock_addr)"
 <bpf@...r.kernel.org>,
 "open list:BPF [NETWORKING] (tcx & tc BPF, sock_addr)"
 <netdev@...r.kernel.org>, open list <linux-kernel@...r.kernel.org>,
 syzbot+346474e3bf0b26bd3090@...kaller.appspotmail.com
Subject: Re: [PATCH net] Drop packets with invalid headers to prevent KMSAN
 infoleak

On 10/21/24 6:37 PM, Daniel Yang wrote:
>> A test in selftests/bpf is needed to reproduce and better understand this.
> I don't know much about self tests but I've just been using the syzbot
> repro and #syz test at the link in the patch:
> https://syzkaller.appspot.com/bug?extid=346474e3bf0b26bd3090. Testing
> the patch showed that the uninitialized memory was not getting written
> to memory.
> 
>> Only bpf_clone_redirect() is needed to reproduce or other bpf_skb_*() helpers calls
>> are needed to reproduce?

If only bpf_clone_redirect() is needed, it should be simple to write a selftest 
to reproduce it. It also helps to catch future regression.

Please tag the next respin as "bpf" also.

> 
>  From what I can see in the crash report here:
> https://syzkaller.appspot.com/text?tag=CrashReport&x=10ba3ca9980000,
> only bpf_clone_redirect() is needed to trigger this issue. The issue
> seems to be that bpf_try_make_head_writable clones the skb and creates
> uninitialized memory but __bpf_tx_skb() gets called and the ethernet
> header never got written, resulting in the skb having a data section
> without a proper mac header. Current check:
> 
> if (unlikely(skb->mac_header >= skb->network_header || skb->len == 0))
> {
> **drop packet**
> }
> 
> in __bpf_redirect_common() is insufficient since it only checks if the
> mac header is misordered or if the data length is 0. So, any packet
> with a malformed MAC header that is not 14 bytes but is not 0 doesn't
> get dropped. Adding bounds checks for mac header size should fix this.
> And from what I see in the syz test of this patch, it does.
> 
> Are there any possible unexpected issues that can be caused by this?