SUMMARY: diag-switch on SPARCengine stuck on

From: Jim Hickstein (jxh@attain.ICD.Teradyne.COM)
Date: Sat Jun 13 1992 - 20:18:35 CDT

[Note: I found the answer after a while, but had so carefully composed
my tirade that I thought I'd leave it in. :-) Read on.]

Several people suggested L1-A or <break> on the serial port. I thought
of that.

My SPARCengine is stuck in self-test mode, because I did a "setenv
diag_switch? true". It doesn't even get as far as recognizing the
framebuffer and talking to me that way. I have to use the serial port.
I have plugged in a keyboard as well, and L1-A, L1-N and L1-V do
nothing, even if they are held down at power-up, as I seem to remember
the SPARCstation does to clear out its EEPROM. Nothing works.

To add insult to injury, after about half an hour of testing its
memory, it discovers something on the VME bus that it doesn't like, and
complains, ending with the word "(looping)". And then it sits there
like a bump on a log. I left it doing that for several hours last
night, and it *never* got as far as the new-mode prompt. It seems like
it's inextricably wedged. Fucking FORTH bullshit. What a piece of
crap! (This is ROM version 1.6, by the way.)

My faint hope is that someone out there has the manual for this thing
(I ordered my hardware through Teradyne Boston, and they didn't forward
any such documents), and that it has a tiny note in it somewhere
documenting an obscure jumper or method for wiping out the EEPROM at
power-up. The FORTH idiots that made this thing MUST have had some way
of accomplishing this, as I can't imagine they got anything right the
first time. The question is: HOW DO I DO IT?

[hours pass...]

I have established that it's listening to the keyboard. The "open-book
test", where you lay an open book on the keyboard, found this. :-)
When I hit certain letters, certain messages come out indicating mode

        l: <Turning LOOP Mode On|Off>
        b: <Turning BURN-IN Mode On|Off>
        s: starts over, just like powerup
        <esc>: <Warning: Selftests aborted by user>

**************** STOP THE PRESSES! *****************

Eureka! Escape. Too obvious. It seems to have wiped out the EEPROM on
the way out, and after the next power cycle it was back to normal. So
that's the answer, kiddies. Hit escape. Brownian motion saves the
day, again. Four hours wasted. I wonder if this is documented, after

By the way, this all started because I was trying to get it to boot
from le() by default. It probes for sd(), fails, and sits at a
prompt. There seems to be nothing in EEPROM you can set that tells it
the actual default boot device, but there was this thing called
"boot-from-diag". *sigh* On Friday, at 5 o'clock sharp, of course.

A further update: "boot-from" starts out as "vmunix", but seems to work
if you set it to "le()vmunix", the default value of "boot-from-diag".
First thing Monday morning, I'm going to try to round up a manual for
this stupid thing.

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:43 CDT