netdev - Re: [PATCH bpf-next]: add sock_ops R/W access to ipv4 tos

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 for Android: free password hash cracker in your pocket

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

Date:   Wed, 28 Mar 2018 21:09:47 +0200
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Nikita Shirokov <tehnerd@...com>, Lawrence Brakmo <brakmo@...com>,
        "ast@...nel.org" <ast@...nel.org>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Cc:     Kernel Team <Kernel-team@...com>
Subject: Re: [PATCH bpf-next]: add sock_ops R/W access to ipv4 tos

On 03/28/2018 07:41 PM, Nikita Shirokov wrote:
>>> On 03/26/2018 05:36 PM, Nikita V. Shirokov wrote:
>>>      bpf: Add sock_ops R/W access to ipv4 tos
>>>
>>>      Sample usage for tos:
>>>
>>>        bpf_getsockopt(skops, SOL_IP, IP_TOS, &v, sizeof(v))
>>>
>>>      where skops is a pointer to the ctx (struct bpf_sock_ops).
>>>
>>> Signed-off-by: Nikita V. Shirokov <tehnerd@...com>
>>> ---
>>>   net/core/filter.c | 35 +++++++++++++++++++++++++++++++++++
>>>   1 file changed, 35 insertions(+)
>>>
>>> diff --git a/net/core/filter.c b/net/core/filter.c
>>> index 00c711c..afd8255 100644
>>> --- a/net/core/filter.c
>>> +++ b/net/core/filter.c
>>> @@ -3462,6 +3462,27 @@ BPF_CALL_5(bpf_setsockopt, struct bpf_sock_ops_kern *, bpf_sock,
>>>                         ret = -EINVAL;
>>>                 }
>>>   #ifdef CONFIG_INET
>>> +     } else if (level == SOL_IP) {
>>> +             if (optlen != sizeof(int) || sk->sk_family != AF_INET)
>>> +                     return -EINVAL;
>>> +
>>> +             val = *((int *)optval);
>>> +             /* Only some options are supported */
>>> +             switch (optname) {
>>> +             case IP_TOS:
>>> +                     if (val < -1 || val > 0xff) {
>>> +                             ret = -EINVAL;
>>> +                     } else {
>>> +                             struct inet_sock *inet = inet_sk(sk);
>>> +
>>> +                             if (val == -1)
>>> +                                     val = 0;
>>> +                             inet->tos = val;
>>
>> Should this not have the exact same semantics given the helper resembles
>> the normal setsockopt? do_ip_setsockopt() does the following when setting
>> IP_TOS:
>>
>>         case IP_TOS:    /* This sets both TOS and Precedence */
>>                 if (sk->sk_type == SOCK_STREAM) {
>>                         val &= ~INET_ECN_MASK;
>>                         val |= inet->tos & INET_ECN_MASK;
>>                 }
>>                 if (inet->tos != val) {
>>                         inet->tos = val;
>>                         sk->sk_priority = rt_tos2priority(val);
>>                         sk_dst_reset(sk);
>>                 }
>>                 break;
>>
>> E.g. why we don't need to set sk->sk_priority as well or reset the dst
>> entry here?
> 
> it feels like initially (w/ commit for IP_TOS in ip_sockglue.c) there were some usecase in mind
> where reflection of tos to prio was needed + some policy based routing (thats why dst_reset).
> but e.g. for ipv6 (IPV6_TCLASS, same as TOS but in ipv6 world) we do just this - set new tclass value
> and call it the day. in my opinion this aproach is more flexible (e.g. we have separate
> bpf_setsockopt for SOL_PRIORITY) as it did only what we want (i can imagine few usecases
> where we want just to change TOS w/o changing priority)

Ok, fair point, that way the behavior is exactly the same as in v6 case.

Applied to bpf-next, thanks Nikita!