From owner-freebsd-current  Thu Aug 14 11:40:49 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id LAA04963
          for current-outgoing; Thu, 14 Aug 1997 11:40:49 -0700 (PDT)
Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50])
          by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA04947
          for <current@FreeBSD.ORG>; Thu, 14 Aug 1997 11:40:46 -0700 (PDT)
Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA17226; Thu, 14 Aug 1997 10:54:34 -0700
From: Terry Lambert <terry@lambert.org>
Message-Id: <199708141754.KAA17226@phaeton.artisoft.com>
Subject: Re: Read-only mount of union filesystem
To: kato@migmatite.eps.nagoya-u.ac.jp (KATO Takenori)
Date: Thu, 14 Aug 1997 10:54:34 -0700 (MST)
Cc: dg@root.com, terry@lambert.org, current@FreeBSD.ORG
In-Reply-To: <199708140849.RAA04702@gneiss.eps.nagoya-u.ac.jp> from "KATO Takenori" at Aug 14, 97 05:49:46 pm
X-Mailer: ELM [version 2.4 PL24]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> >    I changed the behavior in FreeBSD to what it is now (which is also the
> > same as it is in Lite-2). For information on why it is this way, see PR#782.
> > In my opinion, the current structure is correct.
> 
> I understand why vfs layer shouldn't reference v_mount.  But, IMHO, fs
> independent operation should be in upper layer.  The problem in PR#782
> may be solved by following method:
> 
> 1. Prepare special mount struct whose mnt_flag is or'ed by special
>    flag (e.g., MNT_DEAD).
> 2. When vnode is cleaned, v_mount points the special mount struct
>    instead of beeing NULL'ed.
> 3. vfs layer checks MNT_RDONLY and MNT_DEAD.

Yes.  The current arrangement of the mount code is awful; each FS
has duplicate mount code, and some have a working root mount and some
don't.  IMO, the distinction between mounting root vs. mounting
non-root, and *especially* the NFS export manipulations, is terribly
artificial.

In my ideal world, the mount would take place, and there would be a
seperate, upper-level mapping of the resulting mountpoint vnode into
the FS hierarchy afterwards.  It would be this mapping which would
do the NFS export manipulations, and handle the "root" vs. "non-root"
distinctions.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.