Hi all,

I am the current installer maintainer(until someone else really wants 
the job).  But, the problem is most of my coding on it has been MacOS
functionality changes(toolbox calls, SCSI read/write changes, multiple
file selection dialog, etc).  The current problem seems to be in the 
BSD file system operation code, as near as I can tell.  This code was
ported in a long time back(Pre NetBSD 1.1) and should be updated 
anyway.

The good news is that I can easily reproduce the problem.  But, the
bad news is that, as stated above, I really don't know enough about
what I'm looking at yet to debug why it's failing.  The code being
used is old enough that it could have limitations or errors that
have long since been fixed.  The proper fix would be to port the
latest file system format and functions into the installer.  I started
into this at one time late last year and then stalled out due to time 
obligations at work.  I will pick it back up if no one else seems to 
want to do it.  But, it's not entirely trivial(the new file system uses 
64 bit quantities which are not supported by Think C, among other things).

I'm off on vacation next week.  But, I might have time to take another
go at things as soon as I'm back.  If anyone is eager enough for a patch
to the current installer to want to do any coding, the source code is
available at the same places as the application itself.

The only advice I have in the interim is to either use multiple
smaller partitions(seems to work OK), or to use an existing NetBSD/Mac68k
box to format and install the larger partitions(admittedly not an option
for most users).

Steve Brown,
sbrown@best.com

p.s.  The problems fixed in the 1.1e installer were with the SCSI command
	buffer being sent(Switched from the old 6-byte command buffer
	operations to the ones supporting 10-byte command buffer).  So,
	this fixed problems with the NetBSD partition occuring towards the
	high end of the disk(Larger than 1GB addresses supported in the 
	low-level read/write operations of the installer).

> Christopher R. Bowman wrote:
> > 
> > On Wed, 17 Sep 1997, tcjam wrote:
> > 
> > I didn't think this was the off by one partition problems as I thought
> > that had been fixed in the installer that I was using which was
> > pretty recent.
> > 
> > > Seconded!  I recently reformated my 2gig quantum with three HFS
> > > partitions, NetBSD root&usr/NetBSD swap and had the installer crap out in
> > > the same way. I managed to get past it by rearranging the partition map I
> > > think but I'd like to see this resolved.
>
> Hmmmm...I thought that the off-by-1 error only pertained to Mkfs.  Rather,
> the Installer only sees what's written in the Partition Map.  So, the
> Installer might miss the last partition on the disk if there was an error
> in the Partition Map (and I believe it _should_ miss the last partition,
> else there's a bug in the Installer), but the Installer shouldn't crash or
> "crap out" because of this.  The errors that people have been seeing on
> these 2GB drives seems to be a bit different.  At the moment, the only
> thing that I can see might be having a single root&usr partition larger
> than 1G in size...perhaps there is some artificial limit on partition size
> in the Installer?  I know that the problems with disk size were fixed in
> 1.1e, but obviously there must be some other kind of difficulty
> here...either that or a 2GB drive is just a strange beast.  I've certainly
> never seen these kinds of difficulties on my 3 GB Quantum drive (of course
> I only use 1 250MB partition on it, too).
>
> Later.
>
> -- 
> Colin Wood                                 cwood@ichips.intel.com
> Component Design Engineer - MD6                 Intel Corporation
> -----------------------------------------------------------------
> I speak only on my own behalf, not for my employer.
>