>>>>> "JT" == Jason Thorpe <thorpej@nas.nasa.gov> writes:

JT> On Wed, 24 Sep 97 2:27:59 WET DST 
JT>  Dave Huang <khym@bga.com> wrote:

>> Why's mount complaining that the filesystem isn't clean when fsck just
>> marked it as clean?

JT> Looks like fsck_ffs isn't setting the "clean" bit, so the kernel complains
JT> the next time.  Gack.


JT> Luke?  Maybe we should just back out _all_ of the fsck_ffs changes?

Nah.  I encountered a variant of this bug long ago.  See PR 1324.
Charles thought the proposed patch there was incorrect, though.

From: John Kohl <jtk@kolvir.arlington.ma.us>
To: gnats-bugs@gnats.netbsd.org
Cc: cgd@netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: bin/1324: fsck's marking of root file system as clean doesn't stick
Date: Tue, 8 Aug 1995 23:54:52 -0400

 Chris Demetriou pointed out that fsck(8) is supposed to try and remount
 the root with MNT_UNLOAD, but it fails to do so.  Here's a patch to fix
 that problem.  [If you have multiple file systems, the second and
 successive calls to blockcheck() reset the "hotroot" variable so that
 fsck erroneously believes the root FS is not hot, hence it doesn't try
 to remount it.]
 
 You should toss out the change to /etc/rc that I suggested, or at the
 very least change it to "mount -o reload,ro -u /" (otherwise it gets
 reloaded and the clean bit *stays on* while it's mounted)
 
 ==John