From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752630Ab0CQDlN (ORCPT ); Tue, 16 Mar 2010 23:41:13 -0400 Received: from hera.kernel.org ([140.211.167.34]:51285 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751924Ab0CQDlM (ORCPT ); Tue, 16 Mar 2010 23:41:12 -0400 Message-ID: <4BA05004.3040405@kernel.org> Date: Wed, 17 Mar 2010 12:44:04 +0900 From: Tejun Heo User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.8) Gecko/20100228 SUSE/3.0.3-1.1.1 Thunderbird/3.0.3 MIME-Version: 1.0 To: Kevin Easton CC: James Bottomley , linux-kernel@vger.kernel.org Subject: Re: ATA 4 KiB sector issues. References: <20100317025101.GA13339@guarana.org> In-Reply-To: <20100317025101.GA13339@guarana.org> X-Enigmail-Version: 1.0.1 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (hera.kernel.org [127.0.0.1]); Wed, 17 Mar 2010 03:41:00 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hello, On 03/17/2010 11:51 AM, Kevin Easton wrote: > Can't we fix the problem by defaulting to aligning partitions to > start on an LBA that is a multiple of 64260 ? > > Such partitions will always be 4KiB-aligned, *and* start-of-cylinder > aligned (assuming 255/63, as seems to be the norm). > > Sure, that reduces your partition granularity to almost-32-MiB, but > that's pretty small potatoes these days (and it's only a *default*, so > you could always override that if you really cared, and didn't need > the compatibility). The only thing we can gain by that is possible compatibility w/ very old operating systems (<=w2k, BTW, it would be great if someone can actually test it). Plus, breaking the first cylinder assumption might not be always safe to begin with. I personally don't think it's something worth departing from the behavior most vendors would assume from now on (1MiB alignment). It should be enough and safer to provide a mechanism to choose legacy alignment if someone is trying to put something which is older than a decade there. Thanks. -- tejun