From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([140.186.70.92]:45429)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kraxel@redhat.com>) id 1S0Csg-0003m8-5B
	for qemu-devel@nongnu.org; Wed, 22 Feb 2012 09:09:25 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <kraxel@redhat.com>) id 1S0Csc-0005aO-03
	for qemu-devel@nongnu.org; Wed, 22 Feb 2012 09:09:18 -0500
Received: from mx1.redhat.com ([209.132.183.28]:50672)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kraxel@redhat.com>) id 1S0Csb-0005aI-Mj
	for qemu-devel@nongnu.org; Wed, 22 Feb 2012 09:09:13 -0500
Message-ID: <4F44F705.8050208@redhat.com>
Date: Wed, 22 Feb 2012 15:09:09 +0100
From: Gerd Hoffmann <kraxel@redhat.com>
MIME-Version: 1.0
References: <1329860377-6284-1-git-send-email-alevy@redhat.com>
	<1329860377-6284-7-git-send-email-alevy@redhat.com>
	<4F44CF1A.6080703@redhat.com>
	<20120222122835.GC607@garlic.redhat.com>
In-Reply-To: <20120222122835.GC607@garlic.redhat.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [RFC v4 6/9] qxl: remove flipped
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: elmarco@redhat.com, yhalperi@redhat.com, qemu-devel@nongnu.org, spice-devel@freedesktop.org

On 02/22/12 13:28, Alon Levy wrote:
> On Wed, Feb 22, 2012 at 12:18:50PM +0100, Gerd Hoffmann wrote:
>>   Hi,
>>
>> It's not obvious to me how the non-flipped case (qxl_stride > 0) is
>> handled now.  Have you tested this with both windows+linux guests?
> 
> It isn't handled. The simplest way is just to if on the stride and do a
> single memcpy instead of individual line memcpy.

Single memcpy works only for full scanlines.  qxl_flip can be extended
to handle both cases (and should probably also renamed then).

> This of course means we
> are doing a redundant copy,

No.  You can wrap the qxl_flip call into ...

  if (is_shared_buffer()) { ... }.

... to skip the copy if it isn't needed.

> since using our own DisplayAllocator or just
> the existing deallocate + our own allocate of ds->surface->data removes
> one copy.

I would just do

  if (qxl_stride > 0) {
     qemu_free_displaysurface
     qemu_create_displaysurface_from
  } else {
     qemu_resize_displaysurface
  }

cheers,
  Gerd