milne@ICS.UCI.EDU (Alastair Milne) (02/26/91)
I hope somebody can suggest something here, because this problem
is both elusive and *very* distruptive.
I have a Mac II that I bought 2 or 3 years ago. Rather than
an Apple hard disc, I bought a Seagate, which lasted a couple of
years then died a painful death, and which I replaced with a
QuantumPro 105.
The problem: after a successful initial use of the QuantumPro,
the machine now has great trouble at cold boot realising there is
anything at its position on the SCSI bus, # 0 (there are no other SCSI
devices on the machine). Most strangely, if it can just be
tricked into seeing the disc, it then mounts it and boots from it
with no trouble at all. (The trickery has usually been some combination
of actions that include resetting the SCSI bus).
Clearly, the bus can be restored to working
order, and the disc itself seems to have no problems -- fast, quiet,
and the media seems to verify all right. But if the machine gets switched
off (as it did again this weekend when I was away) getting it to realise
there's a disc there is something between difficult and impossible.
The specs (reported by MacEnvy 2.1):
- SCSI chip: NCR 5380
- ROM: 256K (version 120), rev. 1
- ROM checksum: $97851DB6 (old II ROM)
- no SCSI-related inits that I know of
An application called SCSI Commander, that claims to be from the
vendor who sold me the QuantumPro, allows some useful exploration
of the bus/disc, and here are some of its reactions to my attempts:
- a table of SCSI device entries lists vendor name, description,
and firmware version for each entry. None of the SCSI ID's show
any device being present (each number is a radio button,
all are disabled).
- when I use its SPECIALS menu to reset the SCSI bus, ID Search
will then enable address 0's button, but still nothing appears
in its entries. (This is worse than on previous occasions,
where partial entries would show; when the drive is properly
accessible, all entries show completely).
- clicking on MOUNT causes a series of 4 or 5 error dialogues
reporting SCSI errors 2 and 5. Inside Mac says these indicate
missing device or terminators, or SCSI protocol error; and
disagreement about the bus phase or what is being requested
of the device. The mount attempt fails.
- other examination operations also report SCSI errors 2 and 5.
Not all of them fail, though: retrieving the driver descriptor map
shows it seems to be alright; the signature is $4552, anyway,
which seems to agree with Inside Mac's specification
- reading a sector is impossible, and reading mode sense reports
failure.
- switching off the SCSI device (including stopping the drive's spinning)
apparently works -- I can hear the drive stopping. Yet SCSI commander
reports that it fails, and the drive restarts!
- attempting to run diagnostics encounters too many errors
(2 and 5 again) right from the start, and gives up.
A separate application called SCSI Evaluator 1.00 reports, among
other things, the phase of the bus. When I look at it before resetting
the bus, I see a variety of phases reported. 'Message in' and 'data in'
are not uncommon. After the reset, it's usually 'data out'. I have not
recorded the settings of the individual signals, though they're available
if need be.
I might also mention that every time this condition recurs,
the cdev SCSIProbe 2.01 complains the SCSI bus isn't terminated.
Yet without touching the terminators (first thing I asked to have
checked) it was induced to work, and never failed until it was
turned off.
Such things as Disk First Aid or the ForceMount init make no visible
difference at all.
I have rebooted the machine often with Restart (when it had finally
been booted to the hard disc), and there is no problem. But if I do
a cold boot, there definitely is!
This is a recurring problem. The first time, it turned out a chip
on the drive's board was bad, and it was replaced. Since then,
I've tried such things as alternating ribbon cables, or turning them around.
Sometimes this actually seems to work. But nothing fixes it reliably.
Does this sound familiar to anybody? Bad or obsolete ROM routines?
SCSI Manager or Utilities bug? Internal SCSI connector going bad?
SCSI address 0 known to go flaky?
Any suggestions would be much appreciated. I have had the machine itself
diagnosed once, but I didn't see the results personally -- was just
assured they were OK.
Alastair Milne
- ------- End of Failed Message
==================================================
Sender: mmdf@ICS.UCI.EDU
------- End of Forwarded Message