Exabyte Problem

[Dave Olson]

In <1991May3.043324.26592 at iscnvx.uucp> cunning at iscnvx.uucp (Dave Cunningham) writes:

| Once the system is up and running, everything seems to work.  I can read
| and write tapes, position to eom, etc. with no indication of problems
| in SYSLOG.  I get a failure every time the Iris runs its power-up diags, 
| however.
| 
| The messages are something like
| 
| 	SC0: Unexpected Transfer Phase, State 4B Phase 31
| 	scsi(0,7,0) transfer aborted (hardware error)
| 	Device 7 failed DMA test
|     
|     Press return to continue...

This must be on a 4D20 or 25, since the other systems
don't do much in terms of powerup scsi diags.  The Exabyte
drive doesn't implement the SCSI senddiagnostic command.
Not only that, but after the very first byte, they just
KNOW they don't implement it, so they try to go to a message
in phase to tell us about it.  The PROMs aren't designed
to handle that, so you get the error.  The currently shipping
Exabytes have firmware that accepts the whole senddiag command,
and THEN returns illegal request, which we know how to handle,
and which is much more 'standard' for unsupported command.

Anyway, setting the bootmode to C, rather than c, in the 
PROM monitor (setenv bootmode C) suppresses most of the scsi
diags, so your system will autoboot.  For the 4D35, we learned
from this, and also wanted to speed up the bootup procedure,
so we suppress many of the power on diags, unless bootmode is
set to d (for diagnostic).

Some of the early 4D20 machines don't support the bootmode C,
for those you just won't be able to autoboot, but you CAN
safely ignore the error.
--

	Dave Olson

Life would be so much easier if we could just look at the source code.