From owner-freebsd-hackers  Sun May  5 16:03:51 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id QAA17149
          for hackers-outgoing; Sun, 5 May 1996 16:03:51 -0700 (PDT)
Received: from eac.iafrica.com (h196-7-192-150.iafrica.com [196.7.192.150])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id QAA17143
          for <freebsd-hackers@FreeBSD.ORG>; Sun, 5 May 1996 16:03:46 -0700 (PDT)
Received: (from rnordier@localhost) by eac.iafrica.com (8.6.12/8.6.12) id BAA00250; Mon, 6 May 1996 01:00:47 +0200
From: Robert Nordier <rnordier@iafrica.com>
Message-Id: <199605052300.BAA00250@eac.iafrica.com>
Subject: Re: dosfsck anyone?
To: terry@lambert.org (Terry Lambert)
Date: Mon, 6 May 1996 01:00:46 +0200 (SAT)
Cc: joerg_wunsch@uriah.heep.sax.de, freebsd-hackers@FreeBSD.ORG,
        rnordier@iafrica.com
In-Reply-To: <199605052105.OAA20101@phaeton.artisoft.com> from "Terry Lambert" at May 5, 96 02:05:15 pm
X-Mailer: ELM [version 2.4 PL24 ME8a]
Content-Type: text
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

On Sun, 5 May 1996, Terry Lambert wrote:
> 
> J"org writes:
> > As Robert Nordier wrote:
> > 
> > > A preen option is a Good Thing.  'fsck' itself has code to parse
> > > /etc/fstab, skipping non-ufs filesystems.  One solution would be
> > > to incorporate equivalent code in 'dosfsck'.  Possibly a more
> > > elegant approach (which may be what you had in mind) would be to
> > > handle the /etc/fstab parsing in a generic front-end.
> > 
> > I rather thought of it the other way round: similar to mount(8), keep
> > fsck(8) being the generic front-end that does the fstab parsing and
> > dispatching.  Much like mount(8), it could have builtin knowledge
> > about some file system types (a builtin ufs checker, and the wisdom
> > that procfs, swap, and cd9660 don't require checking at all), while it
> > will call {/usr/sbin,/sbin}/fsck_${fstype} for all other file system
> > types.
> 
> Foo.
> 
> man getfsent
> 
> There are already parsing routines for that file; don't write your
> own code to do it.

I missed 'getfsent'.  Comes of browsing the 'fsck' code too
unquestioningly.

> The things should be calling "fstyp" anyway, which should call an
> ioctl on the device to return FS type from the list of recognized
> types.
> 
> The mount code for each FS needs this information anyway, so it can
> tell if it should try mounting it to avoid a panic when you try to
> mount an inappropriate device.  The code should be broken out for
> an iterative call "fstyp" interface.

Presumably this sorts out the problem of doing a mount_msdos on (say)
an HPFS and actually getting away with it, as has happened.

I guess, for mount_msdos, DOS compatibility can't be pursued to the
extent of not requiring a boot sector BPB at all, and just deriving
all FS parameters.

-- 
Robert Nordier