From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <netdev-owner@vger.kernel.org>
Received: from www62.your-server.de ([213.133.104.62]:51982 "EHLO
        www62.your-server.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1752297AbeCCU5W (ORCPT
        <rfc822;netdev@vger.kernel.org>); Sat, 3 Mar 2018 15:57:22 -0500
Subject: Re: [PATCH bpf] bpf: fix bpf_skb_adjust_net/bpf_skb_proto_xlat to
 deal with gso sctp skbs
To: Daniel Axtens <dja@axtens.net>, ast@kernel.org
Cc: netdev@vger.kernel.org,
        Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
References: <20180303020346.9574-1-daniel@iogearbox.net>
 <87y3j9cflf.fsf@linkitivity.dja.id.au>
From: Daniel Borkmann <daniel@iogearbox.net>
Message-ID: <98ceaf0e-1f53-36f4-3310-3ceae1a9ff2d@iogearbox.net>
Date: Sat, 3 Mar 2018 21:57:15 +0100
MIME-Version: 1.0
In-Reply-To: <87y3j9cflf.fsf@linkitivity.dja.id.au>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On 03/03/2018 05:02 PM, Daniel Axtens wrote:
>> From: Daniel Axtens <dja@axtens.net>
>>
>> SCTP GSO skbs have a gso_size of GSO_BY_FRAGS, so any sort of
>> unconditionally mangling of that will result in nonsense value
>> and would corrupt the skb later on.
>>
>> Therefore, i) add two helpers skb_increase_gso_size() and
>> skb_decrease_gso_size() that would throw a one time warning and
>> bail out for such skbs and ii) refuse and return early with an
>> error in those BPF helpers that are affected. We do need to bail
>> out as early as possible from there before any changes on the
>> skb have been performed.
>>
>> Fixes: 6578171a7ff0 ("bpf: add bpf_skb_change_proto helper")
>> Co-authored-by: Daniel Borkmann <daniel@iogearbox.net>
>> Signed-off-by: Daniel Axtens <dja@axtens.net>
>> Cc: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
>> Acked-by: Alexei Starovoitov <ast@kernel.org>
> 
> I've looked over your changes and they all look good to me.
> 
>> +/* Note: Should be called only if skb_is_gso(skb) is true */
>> +static inline bool skb_is_gso_sctp(const struct sk_buff *skb)
>> +{
>> +	return skb_shinfo(skb)->gso_type & SKB_GSO_SCTP;
>> +}
>> +
> 
> This helper is a fantastic idea and I will send a docs update to
> highlight it.

Sounds good. There are infact several places in the code that
could make use of this right away. If you have a chance, this
could be done in net-next along with the doc update or so.

Thanks,
Daniel