netdev - Re: [PATCH 2.6.27] ixgbe: initialize interrupt throttle rate

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: <1219877294.30033.4.camel@jbrandeb-ich9b.jf.intel.com>
Date:	Wed, 27 Aug 2008 15:48:14 -0700
From:	Jesse Brandeburg <jesse.brandeburg@...el.com>
To:	Andy Gospodarek <andy@...yhouse.net>
Cc:	netdev@...r.kernel.org, stable@...nel.org,
	jeffrey.t.kirsher@...el.com, bruce.w.allan@...el.com
Subject: Re: [PATCH 2.6.27] ixgbe: initialize interrupt throttle rate

On Wed, 2008-08-27 at 17:50 -0400, Andy Gospodarek wrote:
> This commit dropped the setting of the default interrupt throttle rate.
> 
> commit 021230d40ae0e6508d6c717b6e0d6d81cd77ac25
> Author: Ayyappan Veeraiyan <ayyappan.veeraiyan@...el.com>
> Date:   Mon Mar 3 15:03:45 2008 -0800
>     
>     ixgbe: Introduce MSI-X queue vector code
>     
> The following patch adds it back.  Without this the default value of 0
> causes the performance of this card to be awful.  Restoring these to the
> default values yields much better performance.
> 
> This regression has been around since 2.6.25.
> 
> Signed-off-by: Andy Gospodarek <andy@...yhouse.net>
> CC: stable@...nel.org [2.6.25 and later]

We hadn't heard about this, but more specifically, this bug effects UDP
transmit performance because the interrupt rate on the tx vectors was so
low.  Here is what I got in testing against 2.6.27-rc3, there was no
appreciable difference testing with TCP on a back to back setup.

before
======
[root linux-2.6]# netperf -H lh2-10 -t UDP_STREAM -C -c
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
lh2-10 (10.0.0.1) port 0 AF_INET
Socket  Message  Elapsed      Messages                   CPU
Service
Size    Size     Time         Okay Errors   Throughput   Util     Demand
bytes   bytes    secs            #      #   10^6bits/sec % SS     us/KB

112640   65507   10.00        1964      0      102.9     26.35    90.483
122880           10.00        1821              95.4     1.35     4.643 

after patch
===========
[root linux-2.6]# netperf -H lh2-10 -t UDP_STREAM -C -c
UDP UNIDIRECTIONAL SEND TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
lh2-10 (10.0.0.1) port 0 AF_INET
Socket  Message  Elapsed      Messages                   CPU
Service
Size    Size     Time         Okay Errors   Throughput   Util     Demand
bytes   bytes    secs            #      #   10^6bits/sec % SS     us/KB

112640   65507   10.00       51641      0     2706.2     32.63    5.016 
122880           10.00       40675            2131.5     30.95    4.758 

Acked-by: Jesse Brandeburg <jesse.brandeburg@...el.com>

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html