5 Jan
2002
5 Jan
'02
11:07 a.m.
On Jan 5, 23:41, The Wanderer wrote:
It's an odd sort of address. I'd expect it to stop at an address in the
ROM. If you had to halt it manually, it must have been running in a loop.
Why that address, I have no idea, because according to the M9312 manual,
it only runs code from ROM until it has booted some device. According to
the M9312 manual, it runs various CPU instruction tests, then tests memory
Pete Turnbull wrote:
I swapped the Unibus map and the cache control board,
and the machine
behaves
apparently much better.
Hmm.. Lots of DEC CPU backplanes have dedicated slots for various things,
and the position/order of the boards is very important. I don't know if
the 11/70 is like that, but I'd not be at all surprised if it were. Have
you checked that everything is in exactly the right place? If you've
removed any optional boards, have you checked in case any jumpers need
replaced?
After pressing the bootstrap key (someone made a
small
switch between tabs 1 & 2 of the M9312) and it goes into a sort of loop
after
a second or so.
Do you mean that someone swapped the wires from their normal places, or
that someone added a switch that wouldn't normally be there?
When pressing halt, the address display ends in 344,
while the
data display shows 116.
Also memory location 400000 through 477777 are accessible via the
console and
I can dump data and read from it.
That's an unusual address, and it's only 32K bytes (16KW). You said you
had two 64KW boards. What type are they? They probably have switches to
set their base addresses. Are these set correctly? It would be worth
trying just one, set to address zero, and see if that improves things.
Start with the minimum in the system, and build it up as you get things
working.
> That sounds encouraging. At least it seems to be
running the
diagnostic
> code. I think, though, that the address it halts
at is what tells you
what
> (if anything) failed the test. If you can tell
where it halted, it may
be
able to look
that up.
At 344 (addr) and 116 in the data display. No idea if this a valid value
though... from 1000 to as high as it can go (up to 28KW), then
tests the cache. It
will halt if it gets an error in any of the memory or cache
tests, but you
can press CONTINUE to ignore the fault and carry on.
> Do you have the manual for the M9312? There are
several ROMs that
might
Yes, the 9312 has the 233 type rom (11/70 diagnostic,
so this is ok).
All DEC ROM numbers begin 23-<something>. All DEC diagnostic PROM numbers
end in F1. The correct bootstrap for a M9312 in an 11/70 is 23-616F1. If
that's not what you've got, it's unlikely to work correctly.
Has the location of the available ram in my case
anything to do with the
high address
limit register? According to the usermanual it is the high imit of the
memory, not
it starting address?
Well, you ought to set it to the correct value, obviously, but I've no idea
what would happen if you set it too high or too low. Yes, it's the high
limit.
On the other hand, the starting address of 400000 can
maybe also the
result of some
initial values in the unibus map registers?
I think the bus init is supposed to set them to zeros. Of course, the map
may be faulty. Normally each memory board has a switch pack to set its
starting address. What type of board(s) do you have?
--
Pete Peter Turnbull
Network Manager
University of York