From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262798AbTJEJA1 (ORCPT ); Sun, 5 Oct 2003 05:00:27 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262827AbTJEJA1 (ORCPT ); Sun, 5 Oct 2003 05:00:27 -0400 Received: from tartutest.cyber.ee ([193.40.6.70]:28424 "EHLO tartutest.cyber.ee") by vger.kernel.org with ESMTP id S262798AbTJEJA0 (ORCPT ); Sun, 5 Oct 2003 05:00:26 -0400 From: Meelis Roos To: linux-kernel@vger.kernel.org Subject: Re: rfc: test whether a device has a partition table In-Reply-To: User-Agent: tin/1.7.1-20030907 ("Sandray") (UNIX) (Linux/2.4.23-pre5 (i686)) Message-Id: Date: Sun, 05 Oct 2003 12:00:22 +0300 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org LT> If you have unusual cases (and let's face it, they don't much happen - we LT> have traditionally had _very_ few problems with getting things partitioned) LT> then you should be able to override them from user space and have user space LT> be able to tell the kernel about special partitions. LT> LT> And hey, surprise surprise, you can do exactly that. LT> LT> Also, surprise surprise, pretty much nobody actually does it. Because the LT> defaults are so sane. Well, I have had difficulties with this twice during last year. Both times it was a partitioned CD. One was a Mac CD (IIRC), the other was an IRIX 6.2 installation CD with SGI disklabel and EFS partition. I got the first to work but never had success with the other (probably a user error). fdisk found the partitions fine but gave hints of a different blocksize (2048 vs 512). I tried to set up the partition with devmapper (dmsetup create ...) but I could never mount the resulting partitions. Perhaps it was a EFS filesystem driver blocksize bug, perhaps something with my manual setup. But I would not say the defaults have been sufficient for _me_. -- Meelis Roos (mroos@linux.ee)