From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Michael S. Tsirkin" Subject: Re: [PATCH] vhost-user: enable virtio 1.0 Date: Fri, 16 Oct 2015 10:56:35 +0300 Message-ID: <20151016105013-mutt-send-email-mst@redhat.com> References: <1444907319-26348-1-git-send-email-marcel@redhat.com> <20151015161150-mutt-send-email-mst@redhat.com> <20151016022438.GH3115@yliu-dev.sh.intel.com> <20151016091327-mutt-send-email-mst@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Marcel Apfelbaum , dev@dpdk.org To: Andriy Berestovskyy Return-path: Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) by dpdk.org (Postfix) with ESMTP id C6773FE5 for ; Fri, 16 Oct 2015 09:56:39 +0200 (CEST) Content-Disposition: inline In-Reply-To: List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On Fri, Oct 16, 2015 at 09:43:09AM +0200, Andriy Berestovskyy wrote: > Hi guys, > Just a minor note: ARM is bi-endian in fact. For instance, there are > both endians tool chains available on Linaro. > > Andriy Yea. BE support is around for legacy stuff. So I'm not sure it's all that important to support that mode in NFV/DPDK applications. If it becomes important, for virtio 0.9 you won't be able to support transparently in dpdk anyway - you'll need a new vhost-user message to get the endian-ness of the guest, and do byteswaps conditionally, adding overhead. Maybe going virtio 1 only is the sane thing to do for these applications. -- MST