From owner-freebsd-hackers  Tue Feb 11 05:51:07 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id FAA20744
          for hackers-outgoing; Tue, 11 Feb 1997 05:51:07 -0800 (PST)
Received: from dg-rtp.dg.com (dg-rtp.rtp.dg.com [128.222.1.2])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id FAA20738
          for <hackers@freebsd.org>; Tue, 11 Feb 1997 05:51:04 -0800 (PST)
Received: by dg-rtp.dg.com (5.4R3.10/dg-rtp-v02)
	id AA22048; Tue, 11 Feb 1997 08:50:03 -0500
Received: from ponds by dg-rtp.dg.com.rtp.dg.com; Tue, 11 Feb 1997 08:50 EST
Received: from lakes.water.net (lakes [10.0.0.3]) by ponds.water.net (8.8.3/8.7.3) with ESMTP id HAA23438; Tue, 11 Feb 1997 07:10:03 -0500 (EST)
Received: (from rivers@localhost) by lakes.water.net (8.8.3/8.6.9) id HAA02075; Tue, 11 Feb 1997 07:14:20 -0500 (EST)
Date: Tue, 11 Feb 1997 07:14:20 -0500 (EST)
From: Thomas David Rivers <ponds!rivers@dg-rtp.dg.com>
Message-Id: <199702111214.HAA02075@lakes.water.net>
To: ponds!critter.dk.tfs.com!phk
Subject: Re: daily panics, ffs_valloc: dup alloc - Good news!
Cc: ponds!freebsd.org!dg, ponds!freebsd.org!hackers, ponds!freebsd.org!joerg
Content-Type: text
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

Poul-Henning writes:
> 
> In message <199702101504.KAA26752@lakes.water.net>, Thomas David Rivers writes:
> >> 
> >> > I'll try getting it with a fixit floppy.
> >> 
> >> please, it's rather crucial for debugging this.
> >
> >
> > Ok - here you go.
> 
> Thanks a lot.
> 
> >	3) Then, I ifconfig'd ed0; grabbed a working dumpfs and
> >	   go this output.
> >	   I notice that the cgmask is 0xffffffff; so I _think_
> >	   this file system would have the same problems.
> >	   (ignore the time there; I haven't got the CMOS clock set right.)
> 
> I still can't see how you can possibly have reached the conclusion
> that cgmask is the cause of this (as oposed to some other magic somewhere
> in the code)...

 Well; given your statement that 0xffffffff is the correct
cgmask - I'd have to retract my assertion.   It was simply the
only "strange-looking" thing I stumbled into when I was determining
just how the disk block offset was calculated.  I'm no filesystem guru;
I do compilers :-)

 However; I'm certain that the problem is disk block calculation,
somewhere.  Since the panic occurs in ffs_alloc.c:ffs_vget(), on
a brand new file system; on the first allocation of an inode that
just happens to be the same as the # inodes/group.  We do the
malloc(), bzero() of this data; then the bread() and we're toast.

 Did the dumpfs I sent you point to any strangeness?

	- Dave Rivers -