From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rusty Russell Subject: Re: [PATCH 1/1] Defer skb allocation for both mergeable buffers and big packets in virtio_net Date: Wed, 25 Nov 2009 10:45:02 +1030 Message-ID: <200911251045.02736.rusty@rustcorp.com.au> References: <1258697745.7416.20.camel@localhost.localdomain> <20091124113754.GB2405@redhat.com> <4B0BEF70.5070104@codemonkey.ws> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Cc: "Michael S. Tsirkin" , Shirley Ma , Eric Dumazet , Avi Kivity , netdev@vger.kernel.org, kvm@vger.kernel.org, linux-kernel@vger.kernel.org, Hollis Blanchard To: Anthony Liguori Return-path: In-Reply-To: <4B0BEF70.5070104@codemonkey.ws> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Wed, 25 Nov 2009 01:06:32 am Anthony Liguori wrote: > So does lguest. It's been that way since the beginning. Fixing this > would result in breaking older guests. Agreed, we can't "fix" it in the guests, but it's a wart. That's why I haven't bothered fixing it, but if someone else wants to I'll cheer all the way. lguest did it because I knew I could fix lguest any time; it was a bad mistake and I will now fix lguest :) > We really need to introduce a feature bit if we want to change this. I don't think it's worth it. But the spec does say that the implementation should not rely on the framing (I think there's a note that current implementations are buggy tho, so you should frame it separately anyway). That way future devices can get it right, at least. Thanks, Rusty.