Colin Wood wrote:
> David A. Gatwood wrote:
> > 
> > On Wed, 22 Oct 1997, Bill Studenmund wrote:
> > 
> > > > That works fine.  I wasn't aware of umount -a working for the root fs.
> > > > Actually, I've thought of one more factor... as far back as I can
> > > > remember, I've had problems when using shutdown -r now.  I don't recall
> > > > whether shutdown -h now causes the problem or not, now that I think about
> > > > it -- which makes sense, since -r does reboot, while -h basically
> > > > approximates "it is now safe to shut of your macintosh"....
> > > > 
> > > > I'm trying a shutdown -h now from multi-user, rebooting from its prompt,
> > > > booting NetBSD....  No problems whatsoever....  Therefore, it seems to
> > > > only happen with a shutdown -r now....
> > > 
> > > Does your drive go into some sort of sleep mode automatically? It could be
> > > that the on-disk cache is getting the sectors to write but not getting to
> > > send them out before MacOS resets the SCSI bus (and possably the drive).
> > 
> > It spins down during a reboot, if that's what you mean.
> 
> David-
> 
> Do you know what kind of hard drive you're using?  I think that what Bill
> is getting at here is that some hard drives support a sleep mode where
> they spin down, and it might be that your hard drive is managing to spin
> down _before_ the sync is performed by a 'shutdown -r'.  If this is the
> case, I think that fixing it should be relatively simple, although I don't
> know for sure.  Most Quantum drives for example have a jumper setting that
> keeps them from spinning down...if you're only going to be running NetBSD
> on the PB, that might be an option.  Otherwise, there may be some kind of
> quirk that needs to be set for your hard drive (although I don't know
> enough about quirks to really tell you on that one :-)

Just to add my 5 cents worth, I have always felt that the time bewteen the
last disk access and the power being cut on my IIsi when shutting down is
a little to short for comfort. I agree that this corruption sounds alot
like a disk-based cache not getting flushed before reset is sent. Reading
through some of the Mac news archives, some PowerMacs had the same problem
in MacOS - the ROM routines were cutting power before the disk had time
to flush its cache.

Perhaps there is a need for a simple spinwait when shutting down and
rebooting?

Cheers,
--
Paul Ripke
BHP Information Technology
Open VMS, AXP & UNIX (AIX, HP/UX, DG/UX, SCO, SGI, Digital, SunOS...) Sysadmin
Computer Centre,    Five Islands Rd,    Port Kembla,    NSW 2505,    AUSTRALIA
  ripke.paul.pr@bhp.com.au     weripp@itwol.bhp.com.au     pjr02@uow.edu.au
            Anyone wishing to lay claim to the opinions expressed
                      herein, do so at their own risk.