At 06:50 PM 1/20/98 -0600, you wrote:
>OK, I was really thinking the 1Gig limit (just for these Quantum Fireball 
>1.(25)G drives), not the cylinder thing.
>
>I bumped the memory allotted up to 15 M and it made not one whit of 
>difference.
>
>I first made a 500 M mac and 700 M BSD partition.  Did not work.  Then I 
>made a 500, 500, 200--the middle 500 was a BSD partition, forget where 
>swap was.  Did not work.  Then made a 50 M swap, a 500 root+user, a 500 M 
>Mac and a 200 M mac.  Damn if that didn't fix it.  You're going to have 
>to convince me there isn't a 1G problem of some sort with this disk.  I 
>reformatted the thing a zillion times and let Silverlining thrash on it 
>(reading and writing sectors) for 8 continous hours.  Near as I can tell, 
>only moving the BSD partition to the front of the disk (and away from 1G 
>mark) worked.  YMMV. (and no, I didn't mess with the cabling at any time)

I have to agree.  In fact, I believe I gave you my solution to this very
problem over on the OpenBSD mailing list a week or so ago, Kevin.  Bob
Nestor and I have also had this discussion when I was going through my
installation problems back in September or October of last year as well.  I
had the EXACT same problems Kevin and Brain describe on attempting to do my
install of NetBSD and OpenBSD using Installer 1.1g and 1.1f and mkfs 1.45
on a Quantum Fireball 1.2GB drive.  Again, I have no termination issues,
although my drive happens to be internal.  I also attempted the Installer
memory increase trick and it didn't do anything for me either, Kevin.  I
too formatted, reformatted and repartitioned the darn thing a gazillion
times before I got it to work.  Now that the system is up, I have no
problems with the drive, and not a single error (though I do have to run
the SBC kernel) but having all these small partitions running around is a
pain as my /usr partition is rapidly filling with /src/sys and /X11R6 on
it.  I currently have no way to get the rest of the source code on the
machine.  Not because it lacks the disk space, but because I had to split
the drive up so much to get things to install.

>On Tue, 20 Jan 1998, Robert Nestor wrote:
>
>> Kevin Havener <havenerk@Walden.MO.NET> wrote:
>> 
>> >None of us could say what's causing the error.   I suspect that it won't 
>> >install if the partition crosses the 1023 cylinder limit.  Others report 
>> >a mismatch between the drive geometry (C/H/S) reported by mkfs and ffs.  
>> >One who got it to work got a minimal system going on a small root at the 
>> >front of the disk.  Then used *BSD to install the rest--thus he was able 
>> >to report the geometry mismatch.  The rest of us, like you, couldn't get 
>> >that far.
>> 
>> I doubt there is a problem with the 1023 cylinder limit.  We did have a 
>> problem with Mkfs and the Installer not handling drives larger than 1Gig, 
>> but that was a long time ago and both have been fixed (long before 
>> OpenBSD appeared).  In fact the fixes were tested on my 1.6Gig drive a 
>> year or so ago.  Since then I've moved up to a 9Gig drive without a 
>> problem.  Also the drive geometry on SCSI disks shouldn't be an issue.  
>> Worst case you'd probably see some performance degradation, but shouldn't 
>> see any file corruption.  However, if it realy worries you it's possible 
>> to tell Mkfs what parameters to use.   Most likely the problem is 
>> insufficent RAM for the Installer, or possibly a SCSI termination issue 
>> if it's an external drive.  Active Termination on the SCSI bus can't hurt 
>> and in many cases it sure helps.
>> 
>> -bob

There has GOT to be some sort of issue with mkfs or Installer.  There just
has to.  The ONLY other common thread that I have seen is it's always been
with Quantum Fireball drives over 1G.  Maybe there's some kind of
interaction/translation/SCSI issue with Quantum drives and the
aforementioned programs I really don't know.  I have no way to look at or
compile MacOS source code and honestly probably lack the expertise in Mac
hardware and SCSI to be able to recognize or fix the problem if I could
view Mac source code.  But there is definitely something going on here.  I
am no longer the only person reporting this problem.  At least there's some
small sense of comfort in that...




Rodney Hopkins
rhopkins@sunflower.com