SUMMARY: Help booting a Sparc 1 - EEPROM params incorrect?

From: Syed Zaeem Hosain (szh@zcon.com)
Date: Mon Jan 24 1994 - 10:46:09 CST


Hi, all! Here is a quick summary, since I found a solution after walking
away from the machine to get something to eat, and sitting down and
thinking about the problem and what was affecting the system.

The reported problem:
____________________

> At one point I inadvertently (without thinking too clearly - affected
> by antihistamines), entered the following at the boot prompt:
>
> set-defaults
> setenv sbus-probe-list 0123456
> reset
>
> and then the machine will not boot (note that a Sparc 1 only has Sbus
> slots 1, 2, and 3 - 0 being the motherboard, I guess). The keyboard
> lights flash once, and then nothing. After I thought about what I had
> done and kicked myself thoroughly, I tried to do things like hold down
> L1-N while booting (see later), etc. Also tried to blindly enter
> "set-defaults" again after an "n" (to get to new-mode), etc. Nothing
> worked.

> In desperation, in an attempt to cure the mis-recognition of the GX
> board in the system (which is what I thought it was), I connected up a
> serial terminal to ttya. Ahha! Some stuff at 9800,N,1 ... what it
> says at power-on (out the serial port) is:
>
> PROM Checksum test
> Context Reg Test
> Segment Map Test
> Sizing Memory
> Page Map Test
> Page Map Test
> Memory Test
> Unknown device in Sbus slot#4
> Memory address not aligned
> ok
>
> and then comes back with an "ok " prompt. Anything I enter at the "ok "
> prompt on this serial terminal then gets the following response and
> nothing happens:
>
> Data Access Exception

The response:
____________

The one respondee so far, Casper Dik (casper@fwi.uva.nl), pointed that
out that the L1-N key sequence is normally done by holding down the L1
key and pressing N repeatedly. Well, I had also tried that without any
luck.

During the powerup test sequence, at least in this older prom, nothing
can break the main powerup and test cycle! Which is not too surprising
really - the machine has to get to *some* basic state before it can
accept any interrupts (to prevent lockups in strange modes). The
problem is that once the PROM is read, the improperly set values of
sbus-probe-list mungs it before it can proceed!

Casper also asked what F1-N and F1-A were supposed to do.

Well, nothing, really! I was trying desperate measures, and thought I
remembered reading somewhere that these keystrokes might do something.
But they do not - I was mistaken.

The solution I found:
____________________

Anyway, over dinner, I thought about what was causing me the problems
(incorrect variable stored in the eeprom) and asked myself what would
happen if the eeprom were not in the machine? I did not have an good
answer to this question, so I tried it!

I removed the hostid PROM (which also contains all the env variable
assignments) and powered up. This causes the machine to come up without
any hostid, without an ethernet address, etc., but it does come up! It
reports that it cannot find the hostid PROM and sets the system env
variables to default values. Well, changing the sbus-probe-list at
this point would not have solved anything, since powering down and
putting the prom back in merely gets the prom values back again (I did
try it - once - for the sake of convincing myself that I was right in
thinking this, and I was).

So the solution was to let the machine power up without the prom, hot
plug the prom back in while the machine was powered up (!), and change
the sbus-probe-list before another reboot. And it worked!!! Whew! Now
the machine boots totally normally. I have reset the sbus-probe-list
(to what I needed) and other parameters (like the serial port values)
and all is normal!

                                                Sigh of relief!

                                                                Z

-------------------------------------------------------------------------
| Syed Zaeem Hosain P. O. Box 610097 (408) 441-7021 |
| Z Consulting Group San Jose, CA 95161 szh@zcon.com |
-------------------------------------------------------------------------



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:55 CDT