netdev - Re: [RFC PATCH net-next 2/3] flow_offload: restore ability to collect separate stats per action

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] [day] [month] [year] [list]

Message-ID: <7d0a0e7b-3b74-d384-75f8-6cde603f81ee@mojatatu.com>
Date:   Thu, 9 May 2019 11:23:47 -0400
From:   Jamal Hadi Salim <jhs@...atatu.com>
To:     Edward Cree <ecree@...arflare.com>,
        Jakub Kicinski <jakub.kicinski@...ronome.com>
Cc:     Jiri Pirko <jiri@...nulli.us>,
        Pablo Neira Ayuso <pablo@...filter.org>,
        David Miller <davem@...emloft.net>,
        netdev <netdev@...r.kernel.org>,
        Cong Wang <xiyou.wangcong@...il.com>,
        Andy Gospodarek <andy@...yhouse.net>,
        Anjali Singhai Jain <anjali.singhai@...el.com>,
        Or Gerlitz <gerlitz.or@...il.com>
Subject: Re: [RFC PATCH net-next 2/3] flow_offload: restore ability to collect
 separate stats per action

On 2019-05-08 1:07 p.m., Edward Cree wrote:
> On 08/05/2019 15:02, Jamal Hadi Salim wrote:
>> The lazy thing most people have done is essentially assume that
>> there is a stat per filter rule...
>> I wouldnt call it the 'the right thing'
> Yup, that's why I'm trying to not do that ;-)

Thank you ;->

> 
>> Yes, the index at tc semantics level is per-action type.
>> So "mirred index 1" and "drop index 1" are not the same stats counter.
> Ok, then that kills the design I used here that relied entirely on the
>   index to specify counters.
> I guess instead I'll have to go with the approach Pablo suggested,
>   passing an array of struct flow_stats in the callback, thus using
>   the index into that array (which corresponds to the index in
>   f->exts->actions) to identify different counters.
> Which means I will have to change all the existing drivers, which will
>   largely revert (from the drivers' perspective) the change when Pablo
>   took f->exts away from them — they will go back to calling something
>   that looks a lot like tcf_exts_stats_update().
> However, that'll mean the API has in-tree users, so it might be
>   considered mergeable(?)

I would say yes, but post the patches and lets have the stakeholders
chime in.
Would it be simpler to just restore the f->exts?


cheers,
jamal