From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754648Ab0ESHjV (ORCPT ); Wed, 19 May 2010 03:39:21 -0400 Received: from ozlabs.org ([203.10.76.45]:56506 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753753Ab0ESHjT (ORCPT ); Wed, 19 May 2010 03:39:19 -0400 From: Rusty Russell To: Avi Kivity Subject: Re: [Qemu-devel] [PATCH RFC] virtio: put last seen used index into ring itself Date: Wed, 19 May 2010 17:09:15 +0930 User-Agent: KMail/1.13.2 (Linux/2.6.32-21-generic; KDE/4.4.2; i686; ; ) Cc: "Michael S. Tsirkin" , linux-kernel@vger.kernel.org, virtualization@lists.linux-foundation.org, kvm@vger.kernel.org, qemu-devel@nongnu.org References: <20100505205814.GA7090@redhat.com> <201005071253.53393.rusty@rustcorp.com.au> <4BE9AF9A.8080005@redhat.com> In-Reply-To: <4BE9AF9A.8080005@redhat.com> MIME-Version: 1.0 Content-Type: Text/Plain; charset="utf-8" Content-Transfer-Encoding: 7bit Message-Id: <201005191709.16401.rusty@rustcorp.com.au> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 12 May 2010 04:57:22 am Avi Kivity wrote: > On 05/07/2010 06:23 AM, Rusty Russell wrote: > > On Thu, 6 May 2010 07:30:00 pm Avi Kivity wrote: > > > >> On 05/05/2010 11:58 PM, Michael S. Tsirkin wrote: > >> > >>> + /* We publish the last-seen used index at the end of the available ring. > >>> + * It is at the end for backwards compatibility. */ > >>> + vr->last_used_idx =&(vr)->avail->ring[num]; > >>> + /* Verify that last used index does not spill over the used ring. */ > >>> + BUG_ON((void *)vr->last_used_idx + > >>> + sizeof *vr->last_used_idx> (void *)vr->used); > >>> } > >>> > >>> > >> Shouldn't this be on its own cache line? > >> > > It's next to the available ring; because that's where the guest publishes > > its data. That whole page is guest-write, host-read. > > > > Putting it on a cacheline by itself would be a slight pessimization; the host > > cpu would have to get the last_used_idx cacheline and the avail descriptor > > cacheline every time. This way, they are sometimes the same cacheline. > > If one peer writes the tail of the available ring, while the other reads > last_used_idx, it's a false bounce, no? I think we're talking about the last 2 entries of the avail ring. That means the worst case is 1 false bounce every time around the ring. I think that's why we're debating it instead of measuring it :) Note that this is a exclusive->shared->exclusive bounce only, too. Cheers, Rusty. From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [140.186.70.92] (port=40684 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1OEhEY-0003xX-2c for qemu-devel@nongnu.org; Wed, 19 May 2010 07:14:45 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.69) (envelope-from ) id 1OEhET-00073f-IC for qemu-devel@nongnu.org; Wed, 19 May 2010 07:14:41 -0400 Received: from ozlabs.org ([203.10.76.45]:48504) by eggs.gnu.org with esmtp (Exim 4.69) (envelope-from ) id 1OEdsF-0002zC-5A for qemu-devel@nongnu.org; Wed, 19 May 2010 03:39:29 -0400 From: Rusty Russell Subject: Re: [Qemu-devel] [PATCH RFC] virtio: put last seen used index into ring itself Date: Wed, 19 May 2010 17:09:15 +0930 References: <20100505205814.GA7090@redhat.com> <201005071253.53393.rusty@rustcorp.com.au> <4BE9AF9A.8080005@redhat.com> In-Reply-To: <4BE9AF9A.8080005@redhat.com> MIME-Version: 1.0 Content-Type: Text/Plain; charset="utf-8" Content-Transfer-Encoding: 7bit Message-Id: <201005191709.16401.rusty@rustcorp.com.au> List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Avi Kivity Cc: qemu-devel@nongnu.org, virtualization@lists.linux-foundation.org, linux-kernel@vger.kernel.org, kvm@vger.kernel.org, "Michael S. Tsirkin" On Wed, 12 May 2010 04:57:22 am Avi Kivity wrote: > On 05/07/2010 06:23 AM, Rusty Russell wrote: > > On Thu, 6 May 2010 07:30:00 pm Avi Kivity wrote: > > > >> On 05/05/2010 11:58 PM, Michael S. Tsirkin wrote: > >> > >>> + /* We publish the last-seen used index at the end of the available ring. > >>> + * It is at the end for backwards compatibility. */ > >>> + vr->last_used_idx =&(vr)->avail->ring[num]; > >>> + /* Verify that last used index does not spill over the used ring. */ > >>> + BUG_ON((void *)vr->last_used_idx + > >>> + sizeof *vr->last_used_idx> (void *)vr->used); > >>> } > >>> > >>> > >> Shouldn't this be on its own cache line? > >> > > It's next to the available ring; because that's where the guest publishes > > its data. That whole page is guest-write, host-read. > > > > Putting it on a cacheline by itself would be a slight pessimization; the host > > cpu would have to get the last_used_idx cacheline and the avail descriptor > > cacheline every time. This way, they are sometimes the same cacheline. > > If one peer writes the tail of the available ring, while the other reads > last_used_idx, it's a false bounce, no? I think we're talking about the last 2 entries of the avail ring. That means the worst case is 1 false bounce every time around the ring. I think that's why we're debating it instead of measuring it :) Note that this is a exclusive->shared->exclusive bounce only, too. Cheers, Rusty.