From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Michael S. Tsirkin" Subject: Re: [PATCH qemu-kvm] Add raw(af_packet) network backend to qemu Date: Wed, 27 Jan 2010 11:24:52 +0200 Message-ID: <20100127092451.GC3476@redhat.com> References: <1264538423.24933.144.camel@w-sridhar.beaverton.ibm.com> <4B5F54E8.3080507@codemonkey.ws> <4B5F5594.6080006@codemonkey.ws> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Sridhar Samudrala , avi@redhat.com, markmc@redhat.com, ogerlitz@voltaire.com, kvm@vger.kernel.org, qemu-devel@vger.kernel.org To: Anthony Liguori Return-path: Received: from mx1.redhat.com ([209.132.183.28]:17379 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753436Ab0A0J2F (ORCPT ); Wed, 27 Jan 2010 04:28:05 -0500 Content-Disposition: inline In-Reply-To: <4B5F5594.6080006@codemonkey.ws> Sender: kvm-owner@vger.kernel.org List-ID: On Tue, Jan 26, 2010 at 02:50:28PM -0600, Anthony Liguori wrote: > On 01/26/2010 02:47 PM, Anthony Liguori wrote: >> On 01/26/2010 02:40 PM, Sridhar Samudrala wrote: >>> This patch adds raw socket backend to qemu and is based on Or Gerlitz's >>> patch re-factored and ported to the latest qemu-kvm git tree. >>> It also includes support for vnet_hdr option that enables gso/checksum >>> offload with raw backend. You can find the linux kernel patch to support >>> this feature here. >>> http://thread.gmane.org/gmane.linux.network/150308 >>> >>> Signed-off-by: Sridhar Samudrala >> >> See the previous discussion about the raw backend from Or's original >> patch. There's no obvious reason why we should have this in addition >> to a tun/tap backend. I thought this was cleared already: vepa support is the requirement here. Existing tap solution requires management of host linux networking which some users would rather avoid. >> The only use-case I know of is macvlan but macvtap addresses this >> functionality while not introduce the rather nasty security problems >> associated with a raw backend. I am not sure I agree with this sentiment. The main issue being that macvtap doesn't exist on all kernels :). macvlan also requires hardware support, packet socket can work with any network card in promisc mode. > Not to mention that from a user perspective, raw makes almost no sense > as it's an obscure socket protocol family. > > A user wants to do useful things like bridged networking or direct VF > assignment. We should have -net backends that reflect things that make > sense to a user. > > Regards, > > Anthony Liguori I agree to that. People don't even seem to agree whether it's a raw socket or a packet socket :) We need a better name for this option: what it really does is rely on an external device to loopback a packet to us, so how about -net loopback or -net extbridge? -- MST