From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1O5LRo-0001Yq-D6 for qemu-devel@nongnu.org; Fri, 23 Apr 2010 12:09:44 -0400 Received: from [140.186.70.92] (port=45320 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1O5LRn-0001Xu-38 for qemu-devel@nongnu.org; Fri, 23 Apr 2010 12:09:44 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.69) (envelope-from ) id 1O5LRl-0003VS-7O for qemu-devel@nongnu.org; Fri, 23 Apr 2010 12:09:42 -0400 Received: from mx1.redhat.com ([209.132.183.28]:24709) by eggs.gnu.org with esmtp (Exim 4.69) (envelope-from ) id 1O5LRk-0003VL-WA for qemu-devel@nongnu.org; Fri, 23 Apr 2010 12:09:41 -0400 Message-ID: <4BD1C625.50602@redhat.com> Date: Fri, 23 Apr 2010 18:09:09 +0200 From: Kevin Wolf MIME-Version: 1.0 Subject: Re: [Qemu-devel] [PATCH] block: reject O_RDWR open for read-only images References: <20100407115753.GB12452@lst.de> <4BBC7B62.7060607@redhat.com> In-Reply-To: <4BBC7B62.7060607@redhat.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Christoph Hellwig Cc: qemu-devel@nongnu.org Am 07.04.2010 14:32, schrieb Kevin Wolf: > Am 07.04.2010 13:57, schrieb Christoph Hellwig: >> Various obscure image format drivers do not allow write access. >> Instead of silently falling back to read-only access reject attempts >> to open these images for write access. >> >> Signed-off-by: Christoph Hellwig > > I'm not sure about this patch. I don't have any such images around, so I > can't test it easily, but it looks to me as if you'd need to explicitly > specify readonly=on if you want to use such formats. This in turn means > that old style shortcuts like -hda don't work any more. Is it that > important to avoid implicit fallbacks for obscure formats that we're > willing to sacrifice compatibility? > > That said, personally I don't care about those format all that much > anyway and the change would make things more consistent. If people > consider it okay to break compatibility in this way, I won't stand in > your way. > > But if we wanted to implement this, have you thought about doing it in > one place in block.c? You could check if a driver supports bdrv_write or > bdrv_aio_write, and if it doesn't support either I think you can assume > it's read-only. Christoph, what about this one? It's still in the block branch, but I have excluded it from first pull request round because these questions are still open. What should I do with it? Drop it, wait for a new version or keep it as it is? If you're not going to drop it, I'd definitely like to hear more opinions on the compatibility thing before proposing it for a merge in a pull request. Kevin