Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!rpi!batcomputer!cornell!uw-beaver!zephyr.ens.tek.com!tektronix!sequent!cseaman
From: cseaman@sequent.UUCP (Chris Seaman)
Newsgroups: comp.sys.amiga
Subject: Re: The A-590/2091 FFS problem
Keywords: Location 0
Message-ID: <38886@sequent.UUCP>
Date: 17 Jul 90 16:55:23 GMT
References: <2085@gould.doc.ic.ac.uk> <6600012@okcusr.UUCP> <1449@nyx.UUCP> <1581@nyx.UUCP>
Organization: Sequent Computer Systems, Beaverton, OR
Lines: 48

bscott@nyx.UUCP (Ben Scott) writes:
< cseaman@sequent.UUCP (Chris "I'm Outta Here, Man!" Seaman) writes:
< >bscott@nyx.UUCP (Ben Scott) writes:
< >< There's a simple, clean fix for this...
< >
< >Sorry, but this doesn't fix anything...
<  
< Does too does too!  At least it worked like a charm on my A-590.  My explanation
< of the actual cause may or may not have been accurate but the fix DOES work for
< me and everyone I've heard about (until now).
< 
< >[ ... ] Whatever the problem is, it is not in the filesystem.
< 
< You may have hit on the problem... I mean, the only thing I can think of is 
< that it is NOT the 2091 that's doing this to your location zero.  Check every
< other program in your startup (even the normal C: commands) and see if it's
< one of them.  If you really are running version 1 of the 1.3.2 FFS, then I think
< it's unlikely that it's trashing location zero.
<  
< And what do you mean by "many" programs failing?  I never found any that 
< actually died - most just exhibited that weird "gdos" string showing up or 
< reported errors after they finished doing their jobs.  At any rate a program
< affected by a trashed location zero is itself buggy, usually because of trying
< to use an uninitialized pointer.  
< 
< .                           <<<<Infinite K>>>>

OK, here goes.  My setup is/was not new with the 2091.  It worked for over a
year with a 2090a, with nary a problem.  My startup sequence HAS NOT CHANGED
since adding the 2091, other than to remove the commands which transferred
control to the 2090a FFS partition.  I have booted the machine (a 2500/20)
from my generic, unmodified 1.3.2 disk, with the SAME results.  I did a
binary compare on the FastFileSystem which came with the 2091, and the one
which came with 1.3.2, and they are identical.  The file size is 12248.  I
am willing to try the fix suggested by Peter Cherna (although I have tried
it three times already), and will report my results tomorrow.

As far as which programs break, I realize that 'MANY' was probably too strong
a word, but there are programs I need to use regularly that do go straight
to the GURU when the address zero problem exists.  These include Turbo Silver
3.0, and Post 1.1.  I also realize that the fault lies with the program(s),
and not with Commodore, but that doesn't correct anything.

-- 
Chris (Insert phrase here) Seaman |  /o  -- -- --
cseaman@sequent <or>              |||    -- -- -     I'm Outta Here, Man!
...!uunet!sequent!cseaman         |vvvv/  -- -- -
The Home of the Killer Smiley     |___/  -- -- --