Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 24 Sep 2017 10:50:29 +0200
From:      Hans Petter Selasky <hps@selasky.org>
To:        Stephen Hurd <shurd@sasktel.net>, "Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net>
Cc:        Stephen Hurd <shurd@FreeBSD.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r323942 - head/sys/net
Message-ID:  <3601ee57-2bf5-036a-a3d1-a4795847d0ec@selasky.org>
In-Reply-To: <89d68ff8-84ed-83a6-4e77-9a321babe2fe@sasktel.net>
References:  <201709230135.v8N1ZE6S063264@repo.freebsd.org> <B28D9879-4ECF-43E4-9A58-51F616CEC4BE@lists.zabbadoz.net> <283397c7-a01e-3776-7ed3-b64d68003d0b@sasktel.net> <6F5DC92C-2CF6-4A33-9663-BFECB7DB65F2@lists.zabbadoz.net> <89d68ff8-84ed-83a6-4e77-9a321babe2fe@sasktel.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 09/24/17 01:46, Stephen Hurd wrote:
> Bjoern A. Zeeb wrote:
>> On 23 Sep 2017, at 6:32, Stephen Hurd wrote:
>>
>>> Bjoern A. Zeeb wrote:
>>>> On 23 Sep 2017, at 1:35, Stephen Hurd wrote:
>>>>
>>>>> Author: shurd
>>>>> Date: Sat Sep 23 01:35:14 2017
>>>>> New Revision: 323942
>>>>> URL: https://svnweb.freebsd.org/changeset/base/323942
>>>>>
>>>>> Log:
>>>>>    Chain mbufs before passing to if_input()
>>>>>
>>>>>    Build a list of mbufs to pass to if_input() after LRO. Results in
>>>>>    12% small packet forwarding rate improvement.
>>>> forwarding seems a confusing word here..
>>>
>>> The test was small (64 byte frames) received on one interface, then 
>>> sent out on a different one using the net.inet.ip.forwarding sysctl 
>>> (controlled via the gateway_enable setting in rc.conf).
>>
>> Then this makes no sense as we don’t do LRO if forwarding is enabled 
>> on the machine;
>> https://svnweb.freebsd.org/base/head/sys/netinet/tcp_lro.c?annotate=317390#l645 
>>
> 
> Basically, it changed from this:
> 
> foreach (mbuf in rx) {
>    if (lro && tcp_lro_rx(mbuf) == 0)
>      continue;
>    if_input(mbuf)
> }
> 
> To this:
> 
> prev_mbuf = first_mbuf = NULL;
> foreach (mbuf in rx) {
>    if (lro && tcp_lro_rx(mbuf) == 0)
>      continue;
>    if (prev_mbuf) {
>      prev_mbuf->m_nextpkt = mbuf;
>      prev_mbuf = mbuf;
>    }
>    else {
>      first_mbuf = prev_mbuf = mbuf;
>    }
> }
> 
> if (first_mbuf)
>    if_input(first_mbuf);
> 
> So while before it called if_input() for each separate mbuf that was not 
> LROed, it now builds a chain of mbufs that were not LROed, and makes a 
> single call to if_input() with the whole chain.  For cases like packet 
> forwarding where no packets are LROed, performance is better.
> 

Can such a similar logic be applied inside TCP LRO aswell?

--HPS




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3601ee57-2bf5-036a-a3d1-a4795847d0ec>