From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:60807) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1RlKUI-0000KU-BX for qemu-devel@nongnu.org; Thu, 12 Jan 2012 08:14:39 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1RlKUC-0006PS-DW for qemu-devel@nongnu.org; Thu, 12 Jan 2012 08:14:38 -0500 Received: from mx1.redhat.com ([209.132.183.28]:19225) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1RlKUC-0006PN-4F for qemu-devel@nongnu.org; Thu, 12 Jan 2012 08:14:32 -0500 Message-ID: <4F0EDD6A.30804@redhat.com> Date: Thu, 12 Jan 2012 14:17:30 +0100 From: Kevin Wolf MIME-Version: 1.0 References: <1325858501-25741-1-git-send-email-stefanha@linux.vnet.ibm.com> <1325858501-25741-5-git-send-email-stefanha@linux.vnet.ibm.com> <4F0EBD12.7080702@redhat.com> <4F0ED7B3.2000904@redhat.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH v4 04/15] block: add image streaming block job List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Stefan Hajnoczi Cc: Luiz Capitulino , Marcelo Tosatti , Stefan Hajnoczi , qemu-devel@nongnu.org Am 12.01.2012 14:05, schrieb Stefan Hajnoczi: > On Thu, Jan 12, 2012 at 12:53 PM, Kevin Wolf wrote: >> Am 12.01.2012 12:39, schrieb Stefan Hajnoczi: >>> On Thu, Jan 12, 2012 at 10:59 AM, Kevin Wolf wrote: >>>> Am 06.01.2012 15:01, schrieb Stefan Hajnoczi: >>>>> + buf = qemu_blockalign(bs, STREAM_BUFFER_SIZE); >>>>> + s->common.len = bdrv_getlength(bs); >>>> >>>> No error check? >>> >>> Will fix. >>> >>>>> + bdrv_get_geometry(bs, (uint64_t *)&end); >>>> >>>> Why call bdrv_getlength() twice? end = s->common.len >> BDRV_SECTOR_BITS >>>> should be the same. >>> >>> Okay, I'll change it. I got sick of BDRV_SECTOR_* and called twice instead. >> >> Well, you can try and change everything in the streaming code to bytes >> instead of sectors. We should probably do this sooner or later anyway. >> Sectors of 512 bytes are a completely arbitrary unit that doesn't make >> much sense generally. > > That doesn't work because block layer interfaces use nb_sectors. We > still need to convert. Sure, somewhere you'll have the conversion. You can only push it a bit closer to the invocation of the block drivers if you like. Everything else would be a major refactoring (but eventually I think we'll do it). Kevin