From owner-freebsd-hackers  Thu Aug 15 07:15:54 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id HAA09486
          for hackers-outgoing; Thu, 15 Aug 1996 07:15:54 -0700 (PDT)
Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.109.160])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id HAA09480
          for <hackers@FreeBSD.ORG>; Thu, 15 Aug 1996 07:15:50 -0700 (PDT)
Received: (from jgreco@localhost) by brasil.moneng.mei.com (8.7.Beta.1/8.7.Beta.1) id JAA13565; Thu, 15 Aug 1996 09:14:22 -0500
From: Joe Greco <jgreco@brasil.moneng.mei.com>
Message-Id: <199608151414.JAA13565@brasil.moneng.mei.com>
Subject: Re: Nightmare.
To: bde@zeta.org.au (Bruce Evans)
Date: Thu, 15 Aug 1996 09:14:22 -0500 (CDT)
Cc: jgreco@brasil.moneng.mei.com, ulf@lamb.net, hackers@FreeBSD.ORG,
        jkh@time.cdrom.com
In-Reply-To: <199608142129.HAA04653@godzilla.zeta.org.au> from "Bruce Evans" at Aug 15, 96 07:29:21 am
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> >> I would go more into the direction of checking if the dump device is a
> >> mounted file system. Easy check. 
> 
> >I tend to agree, but wonder if it would not make more sense to tackle this
> >from a different angle.
> 
> >Consider all the programs that could clobber a mounted file system.  Would
> >it make more sense if we somehow protected a mounted disk device from
> >being clobbered?
> 
> It is already protected.  Even root can't even open a mounted disk device.

Root can open (and write to) a raw disk device that has its counterpart
mounted, however.  I think the "and write to" part is much more of a 
liability than simply being able to open and read the device.

As we saw from the message that started this thread.  ;-)

> However, mounted disk devices can easiliy be clobbered using an alias.
> There are about 2^(1+5+3) = 512 possible aliases (raw/buffered, 2^5
> slices, 2^3 partitions).  This is useful for things like installing
> new bootblocks on the root partition without having to reboot with a
> different root partition.  Don't fix it.  Root knows all about this
> problem, and never makes mistakes :-).

As we saw from the message that started this thread.  ;-)

In general I don't think it's a solvable problem.  I'm willing to live
with that personally, I think that the current state of affairs is
a good compromise, but I think it's really a good idea to toss around
ideas anyways to see if there isn't a better general solution of some
sort waiting out there to be found.

... JG