netdev - Re: [PATCH bpf-next v3 09/15] bpf: add bpf_skc_to_udp6

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]

Date:   Tue, 23 Jun 2020 10:03:53 -0700
From:   Yonghong Song <yhs@...com>
To:     Eric Dumazet <eric.dumazet@...il.com>, <bpf@...r.kernel.org>,
        <netdev@...r.kernel.org>
CC:     Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>, <kernel-team@...com>,
        Martin KaFai Lau <kafai@...com>
Subject: Re: [PATCH bpf-next v3 09/15] bpf: add bpf_skc_to_udp6_sock() helper



On 6/23/20 9:27 AM, Eric Dumazet wrote:
> 
> 
> On 6/22/20 7:22 PM, Yonghong Song wrote:
>>
>>
>> On 6/22/20 6:47 PM, Eric Dumazet wrote:
> &
>>>
>>> Why is the sk_fullsock(sk) needed ?
>>
>> The parameter 'sk' could be a sock_common. That is why the
>> helper name bpf_skc_to_udp6_sock implies. The sock_common cannot
>> access sk_protocol, hence we requires sk_fullsock(sk) here.
>> Did I miss anything?
> 
> OK, if arbitrary sockets can land here, you need also to check sk_type

The current check is:
         if (sk_fullsock(sk) && sk->sk_protocol == IPPROTO_UDP &&
             sk->sk_family == AF_INET6)
                 return (unsigned long)sk;
it checks to ensure it is full socket, it is a ipv6 socket and then 
check protocol.

Are you suggesting to add the following check?
   sk->sk_type == SOCK_DGRAM

Not a networking expert. Maybe you can explain when we could have
protocol is IPPROTO_UDP and sk_type not SOCK_DGRAM?