From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:36191)
	by lists.gnu.org with esmtp (Exim 4.71) (envelope-from <pl@kamp.de>)
	id 1VoFfi-0004VZ-5u
	for qemu-devel@nongnu.org; Wed, 04 Dec 2013 11:51:42 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <pl@kamp.de>) id 1VoFfZ-0005fB-On
	for qemu-devel@nongnu.org; Wed, 04 Dec 2013 11:51:34 -0500
Received: from mx.ipv6.kamp.de ([2a02:248:0:51::16]:46674 helo=mx01.kamp.de)
	by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from <pl@kamp.de>)
	id 1VoFfZ-0005ed-Cz
	for qemu-devel@nongnu.org; Wed, 04 Dec 2013 11:51:25 -0500
Message-ID: <529F5D88.2070709@kamp.de>
Date: Wed, 04 Dec 2013 17:51:20 +0100
From: Peter Lieven <pl@kamp.de>
MIME-Version: 1.0
References: <1385546829-3839-1-git-send-email-pl@kamp.de>
	<1385546829-3839-2-git-send-email-pl@kamp.de>
	<20131204164656.GF27759@stefanha-thinkpad.redhat.com>
In-Reply-To: <20131204164656.GF27759@stefanha-thinkpad.redhat.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [Qemu-devel] [PATCHv3 1.8 1/9] qemu-img: add support for
 skipping zeroes in input during convert
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: Stefan Hajnoczi <stefanha@gmail.com>
Cc: kwolf@redhat.com, pbonzini@redhat.com, qemu-devel@nongnu.org, stefanha@redhat.com

Am 04.12.2013 17:46, schrieb Stefan Hajnoczi:
> On Wed, Nov 27, 2013 at 11:07:01AM +0100, Peter Lieven wrote:
>> +                /* If the output image is being created as a copy on write
>> +                 * image, assume that sectors which are unallocated in the
>> +                 * input image are present in both the output's and input's
>> +                 * base images (no need to copy them). */
>> +                if (out_baseimg) {
>> +                    if (!(ret & BDRV_BLOCK_DATA)) {
>> +                        sector_num += n1;
>> +                        continue;
>> +                    }
>> +                    /* The next 'n1' sectors are allocated in the input image.
>> +                     * Copy only those as they may be followed by unallocated
>> +                     * sectors. */
>> +                    nb_sectors = n1;
>> +                }
>> +                /* avoid redundant callouts to get_block_status */
>> +                sector_num_next_status = sector_num + n1;
> Can you explain when we need sector_num_next_status?  It's not clear to
> me from this patch when we will loop around already knowing that blocks
> are allocated.
We call get_block_status with MIN(INT_MAX, nb_sectors). So we might
receive an allocation status for a huge area. Later we trim the request
size to MIN(iobuf_size, nb_sectors) and eventually align the request.

For example take a fully allocated image on an iSCSI san. I can easily get
that information with the first get_block_status call, but I repeat these
calls over and over and in case of the iSCSI SAN these calls are quite
expensive.

Peter