> > FWIW, my 68LC040-based Quadra 610 had *exactly* the same problem with fsck.
> > This doesn't guarantee that the problem is related to the FPE bug, but it
> > shows that at least two different 68LC040 systems have had exactly the same
> > problems when booting up.
> 
> Sounds more like a scsi driver bug, at least that'd be my guess.  The
> ncrscsi96 driver is partially derived from the ncrscsi driver, but with
> several changed commands, right?  Would that happen to be a quantum drive,
> less than 3 years old?  Because that's basically the same symptom that
> some people got with the ncrscsi driver on some pre ...c96 with certain
> drives.  That's why a lot of ppl use the sbc driver.  If there's some
> shared code, it would then seem somewhat likely that the problem would be
> in the code that's held in common between the two drivers, and in
> something that's not in the sbc driver.  (Of course, it could be two
> separate problems, simply w/ the same basic symptoms, i.e.  hard
> read/write errors.) 

Actually the ncrscsi96 driver we use is based on the esp driver from (I 
think) the sparc port.  So, I doubt that it would have quite the same 
trouble as the older Mac ncrscsi driver.  However, I do believe that fsck 
makes use of floating point operations (I seem to remember broken fsck 
being a symptom of the earlier FPE problems), so it is quite probably 
related to the FPE bug.  Of course, I could be totally wrong ;-)

Later.

-- 
Colin Wood                                      ender@is.rice.edu
Consultant                                        Rice University
Information Technology Services                       Houston, TX