netdev - Re: [PATCH net-next 4/4] net/sched: act_mirred: Implement ingress actions

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: <cdebe688-7a25-8d3c-e6b4-4e34a41e779b@mojatatu.com>
Date:   Sun, 25 Sep 2016 19:31:45 -0400
From:   Jamal Hadi Salim <jhs@...atatu.com>
To:     Daniel Borkmann <daniel@...earbox.net>,
        Shmulik Ladkani <shmulik.ladkani@...il.com>
Cc:     "David S. Miller" <davem@...emloft.net>,
        WANG Cong <xiyou.wangcong@...il.com>,
        Eric Dumazet <edumazet@...gle.com>, netdev@...r.kernel.org,
        Florian Westphal <fw@...len.de>
Subject: Re: [PATCH net-next 4/4] net/sched: act_mirred: Implement ingress
 actions

On 16-09-25 12:26 PM, Daniel Borkmann wrote:
> On 09/25/2016 03:05 PM, Jamal Hadi Salim wrote:

[..]

>> MAX_RED_LOOP (stands for "Maximum Redirect loop") still exists in
>> current code. The idea above was that we would increment the rttl
>> counter  once and if we saw it again upto MAX_RED_LOOP we would assume
>> a loop and drop the packet (at the time i didnt think it was wise to
>> let the actions be in charge of setting the RTTL; it had to be central
>> core code - but it may not be neccessary)
>>
>> Florian, when we discussed I said it was fine to reclaim those 3 bits
>> on tc verdict for RTTL at the time because i had taken out the
>> feature and never added it back. Your comment at the time was we can
>> add it back when someone shows up with the feature.
>> Shmulik is looking to add it.
>
> Why not just reuse xmit_recursion, which is what we did in tc cls_bpf
> programs f.e. see __bpf_tx_skb()? Would be a pity to waste 3 bits on
> this in the skb.

If it is going to work, I'd be happy to save those bits.
xmit_recursion is going to prevent recursing into
dev_xmit(), no?

In our case we want to preventing looping of a singular
skb.

cheers,
jamal