netdev - Re: [PATCH net-next RFC 0/5] xen-netback: TX grant mapping instead of copy

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]

Date:	Mon, 16 Dec 2013 10:14:27 +0000
From:	Ian Campbell <Ian.Campbell@...rix.com>
To:	Zoltan Kiss <zoltan.kiss@...rix.com>
CC:	<wei.liu2@...rix.com>, <xen-devel@...ts.xenproject.org>,
	<netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
	<jonathan.davies@...rix.com>
Subject: Re: [PATCH net-next RFC 0/5] xen-netback: TX grant mapping instead
 of copy

On Thu, 2013-12-12 at 22:08 +0000, Zoltan Kiss wrote:
> On 28/11/13 17:43, Ian Campbell wrote:
> > On Thu, 2013-11-28 at 17:37 +0000, Zoltan Kiss wrote:
> > Routing/firewalling domUs is as valid as bridging. There is nothing in
> > the slightest bit suboptimal about it.
> >
> > If this use case regresses with this approach then I'm afraid that
> > either needs to be addressed or a different approach considered.
> >
> >> Anyway, I will try this out, and see if it really copies everything, and
> >> get some numbers as well.
> >
> > Thanks.
> 
> Now I managed to try it out. As I expected, Dom0 does copy the mapped 
> page. The peak throuhput I could get was 6.6 Gbps, however it could keep 
> that only for short periods, I guess when the unmapping was ideally 
> batched. The average was 5.53.
> On the same machine the same 10 min iperf session, without my patches 
> made the peak 5.9 while the average was 5.65. Do you think it is an 
> acceptable regression?

Well, it would of course be preferable to avoid it. I'm quite reluctant
to see this scenario become a second class citizen.

> I used 3.12 Dom0 and guest kernel, the guest transmitted though a 10Gb 
> card to a bare metal box.
> I plan to look further if we can avoid somehow this:
> 
> https://lkml.org/lkml/2012/7/20/363
> 
> So then this scenario can benefit from grant mapping.
> 
> Regards,
> 
> Zoli


--
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