From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:37691)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <benh@kernel.crashing.org>) id 1V8oQC-00064m-Ol
	for qemu-devel@nongnu.org; Mon, 12 Aug 2013 05:28:22 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <benh@kernel.crashing.org>) id 1V8oQ6-0005Zr-Q1
	for qemu-devel@nongnu.org; Mon, 12 Aug 2013 05:28:16 -0400
Received: from gate.crashing.org ([63.228.1.57]:33170)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <benh@kernel.crashing.org>) id 1V8oQ6-0005ZZ-H5
	for qemu-devel@nongnu.org; Mon, 12 Aug 2013 05:28:10 -0400
Message-ID: <1376299687.32100.159.camel@pasglop>
From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: Mon, 12 Aug 2013 19:28:07 +1000
In-Reply-To: <1376294363-4650-2-git-send-email-rusty@rustcorp.com.au>
References: <1376294363-4650-1-git-send-email-rusty@rustcorp.com.au>
	<1376294363-4650-2-git-send-email-rusty@rustcorp.com.au>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCH 1/8] virtio_get_byteswap: function for
 endian-ambivalent targets using virtio.
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: Rusty Russell <rusty@rustcorp.com.au>
Cc: qemu-devel@nongnu.org

On Mon, 2013-08-12 at 17:29 +0930, Rusty Russell wrote:
> virtio data structures are defined as "target endian", which assumes
> that's a fixed value.  In fact, that actually means it's
> platform-specific.
> 
> Hopefully the OASIS virtio 1.0 spec will fix this.  Meanwhile, create
> a hook for little endian ppc.

Ok, sorry if I missed a previous debate on that one but why do you do a
call-out to architecture specific stuff (that is not even inline) on
every access ?

If we consider that virtio byte order is global, can't you make it a
global that is *set* by the architecture rather than *polled* by
virtio ?

We have explicit knowledge of when our endianness change (we get the
hcall or a write to some SPR), we can call virtio *then* to adjust the
endianness rather than having a call-out to the platform on every
access.

Ben.