Subject: Re: Sun3/E use
To: None <>
From: Ian Wells <>
List: port-sun3
Date: 10/19/1999 12:28:39
> Peter Koch on Germany and I both have Sun3/E systems.
> Take a look at

I've already talked to Peter Koch about things (after reading his
website) although not since I got it to do anything useful.

> The newest operating systems that will run on it are SunOS
> 4.1.1_U1 and NetBSD 1.4.1. The DMA for the SCSI under NetBSD
> has not been completed, but it otherwise runs just fine.

Possibly, but the standard install kernel doesn't

> The spurious interrupts could be from not having the jumper settings
> for IACK on the backplane set correctly.

I have no idea whatsoever when it comes to VME.  However, I've not
changed anything since I got the system, which used to run SunOS 3.5
OK, other than removing some boards to see if it helped.

> Not enumerating the VME boards also sounds like a BGNT jumper
> configuration problem.

Well, it enumerates them OK but doesn't identify them or associate
them with a driver - which is the sort of behaviour I'd expect of a
card without a driver in the kernel.  Hence, I wondered if there was a 
3/E SCSI driver in the distribution kernel.

Again, since it used to work I would be surprised if the backplane was
misconfigured.  However, it's had a bit of rough handling (it weighs
~200kgs...) so there's a possibility I managed to dislodge a jumper
banging it about or investigating it afterwards.

> How many slot backplane do you have and what boards are in what slots?

Er, well.  The machine itself has a 2x21 slot backplane.  The rear
backplane is wholly devoted to a custom-built graphics processing unit
(about 18 cards all full of TTL DIL chips; quite impressive,
actually).  It has its own connection to a special interface board on
the front unit.  From memory, the front unit has (left to right, slots
1 to 21):

1. Sun 3/E processor
2. Board labelled SIO (serial?) with connections from P2 on the rear
3. empty slot with P2 connection on the rear, presumably for another SIO
6. Interphase SMD controller
7. Xylogics 1/2in. tape streamer controller (I have the tape drive for this, 
   too, in working order, and 170 1/2in tapes, so getting this to work
   would be nice!)
8. empty slot
9, 10, 11. 3 custom boards with crosslinks between them at the front
12. empty slot
13. Interface board to rear VME cage
14. Sun SCSI/Ethernet card
15. 68000 processor (controls a custom framebuffer and framegrabber
16-21 Several boards with various connections across the front,
   together forming the custom framebuffer and framegrabber judging by 
   the RGB inputs and outputs to it

This is from memory and may be wrong in order and even number of
cards.  With the custom boards I've guessed at the function.  There's
a few more connectors between boards on the back, but I couldn't tell
you what they were without my notes.

I had removed the boards in 2, 4 and 5 before any boot attempt.  I've
since removed boards 6, 7 and 15 while trying to boot.  The boards
which are tied together have small PCBs across the front, which makes
it inconvenient to remove them...  I suspect some of the custom boards 
are not actually VME devices and just sit passively on the backplane.

When I booted initially, I got 7 devices enumerated.  Among these I
suspect there's the SMD, tape, SCSI/Ethernet and 68000, but I don't
know what the others might be.  I may have the session log somewhere
at home, though.

Oh, and I've had the occasional whinge about a bad date in the clock,
too - can't remember what.  But it remembers its ethernet hardware
address, which I always understood was a symptom of the backup battery
dying, so I'm not quite sure what to make of this.