From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([140.186.70.92]:38592)
	by lists.gnu.org with esmtp (Exim 4.71) (envelope-from <hch@lst.de>)
	id 1RXzyH-0002dK-Jz
	for qemu-devel@nongnu.org; Tue, 06 Dec 2011 13:42:30 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <hch@lst.de>) id 1RXzyG-000058-J0
	for qemu-devel@nongnu.org; Tue, 06 Dec 2011 13:42:29 -0500
Received: from verein.lst.de ([213.95.11.211]:35685 helo=newverein.lst.de)
	by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from <hch@lst.de>)
	id 1RXzyG-00004u-BD
	for qemu-devel@nongnu.org; Tue, 06 Dec 2011 13:42:28 -0500
Date: Tue, 6 Dec 2011 19:42:26 +0100
From: Christoph Hellwig <hch@lst.de>
Message-ID: <20111206184226.GA18058@lst.de>
References: <958CDBC95A3252B592B5E3C9@nimrod.local>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <958CDBC95A3252B592B5E3C9@nimrod.local>
Subject: Re: [Qemu-devel] RFC: raw device support for block device targets
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: Alex Bligh <alex@alex.org.uk>
Cc: qemu-devel@nongnu.org

On Tue, Dec 06, 2011 at 04:20:56PM +0000, Alex Bligh wrote:
> qemu-img convert appears to support block devices as input, but not
> as output. That is irritating, as when using qemu-img convert to
> convert qcow to raw on a block partition, an intermediate file has
> to be used, which slows things down and pointlessly uses disk space.
>
> The problem is that ftruncate() is being called on the output file
> in order ensure it is sufficiently large, and this fails on
> block devices.
>
> I appreciate there may be other calls that fail depending on the
> input file format, but these will presumably be error checked
> at the time.
>
> Is it therefore worth skipping the ftruncate() if the block device
> is large enough, and at least attempting to proceed further? Something
> like the following (not-even compile tested) patch?

It probably should share the code with raw_truncate.  That is factor
a new raw_truncate_common that is equivalent to raw_truncate but
takes a file descriptor instead of a block driver state and then use
it for both raw_truncate and raw_create.