netdev - Re: Big performance loss from 3.4.63 to 3.10.13 when routing ipv4

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: <2007014.RQqn1EBPtM@h2o.as.studentenwerk.mhn.de>
Date:	Wed, 23 Oct 2013 18:05:22 +0200
From:	Wolfgang Walter <linux@...m.de>
To:	Steffen Klassert <steffen.klassert@...unet.com>
Cc:	David Miller <davem@...emloft.net>, hannes@...essinduktion.org,
	netdev@...r.kernel.org, klassert@...hematik.tu-chemnitz.de
Subject: Re: Big performance loss from 3.4.63 to 3.10.13 when routing ipv4

Am Mittwoch, 23. Oktober 2013, 14:04:34 schrieb Steffen Klassert:
> On Wed, Oct 23, 2013 at 01:33:14PM +0200, Wolfgang Walter wrote:
> > Am Mittwoch, 23. Oktober 2013, 10:12:55 schrieb Steffen Klassert:
> > > On Tue, Oct 22, 2013 at 03:46:38PM -0400, David Miller wrote:
> > > > I think we should resolve this soon, even bumping it to 2048 or 4096
> > > > and leaving it at that would be I think acceptable.
> > > 
> > > Yes, of course. Let's use 4096 as the default for ipv4 and ipv6.
> > > I'll take care of it next week.
> > 
> > I don't know what this value actually means. But on 3.4.x it is much
> > higher. On a machine with 512MB ram it is 32768, on a machine with 1GB
> > ram it is 262144 and with 16GB ram it is 4194304.
> 
> Before  we removed the routing cache, the gc threshold was scaled along
> with the maximum routing cache size (ip_rt_max_size). With the routing
> cache removal, we lost the possibility to scale with ip_rt_max_size
> and we had to choose a static default. Maybe we can try to tweak the
> gc threshold again with the available memory somehow later. But to fix
> it now, we need to find a reasonable default value. Would a default of
> 4096 meet your requirements?

Can't say that at the moment. I can test it tonight. I assume that I don't 
need more than 262144 even on the large router as it worked on older hardware 
with less memory (with 3.4.x kernels). No router of us had a value less then 
32768 with 3.4.x.

Regards,
-- 
Wolfgang Walter
Studentenwerk München
Anstalt des öffentlichen Rechts
--
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