From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266496AbUGPHFi (ORCPT ); Fri, 16 Jul 2004 03:05:38 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266495AbUGPHFi (ORCPT ); Fri, 16 Jul 2004 03:05:38 -0400 Received: from parcelfarce.linux.theplanet.co.uk ([195.92.249.252]:19387 "EHLO www.linux.org.uk") by vger.kernel.org with ESMTP id S266496AbUGPHFd (ORCPT ); Fri, 16 Jul 2004 03:05:33 -0400 Date: Fri, 16 Jul 2004 08:05:31 +0100 From: viro@parcelfarce.linux.theplanet.co.uk To: Ron House Cc: linux-kernel@vger.kernel.org Subject: Re: Bug or Feature? Multiply mounted device Message-ID: <20040716070531.GA12308@parcelfarce.linux.theplanet.co.uk> References: <40F75738.4040601@usq.edu.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <40F75738.4040601@usq.edu.au> User-Agent: Mutt/1.4.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Jul 16, 2004 at 02:19:04PM +1000, Ron House wrote: > Hi all, apologies for posting here, but I need to ask the guys who know > for sure. In my fstab, I have two entries for my DOS HD so I can mount > it with or without CRLF mangling, as follows (sorry for the line splitting): > > /dev/hda5 /d vfat > rw,suid,dev,async,users,umask=033,uid=1500,gid=1001,conv=b 0 0 > /dev/hda5 /dc vfat > rw,suid,dev,async,users,umask=033,uid=1500,gid=1001,noauto,conv=a 0 0 > > I just noticed with my new Linux setup (kernel 2.4.20-8smp - don't know > what all that means) that it will let me mount both at once, so I can > see the partition as either /d or /dc at the same time. I may be going > mad, but I seem to recall with my previous kernel that I couldn't do this. > > My question: Bug or feature? If I write all sorts of stuff to both > logical devices, will it correctly sort it all out on the same physical > device, or will I wreck the partition? All help much appreciated! You will get both mountpoints as aliases to the tree. Note that all fs-specific flags will be ignored on subsequent mounts right now. Eventually different flags will give you -EBUSY. BTW, conv= is silently ignored by vfat, so your conv=a is a no-op - that stuff simply doesn't belong in kernel and CRLF mangling had been removed quite a while ago (try to make it play nice with mmap(2) and see how soon you'll run away screaming).