Subject: it's alive!
To: NetBSD/VAX Mailing List <port-vax@NetBSD.ORG>
From: Brian D Chase <brianc@carpediem.com>
List: port-vax
Date: 07/26/1997 19:31:03
vaxzilla lives!!!

I kept diligently working on my MicroVAX II problems and it paid off.
I'm still not sure why my system would hang during boot up in my
legitimate 5Meg configuration.

---
For history's sake and educational value for the other Q-bus clueless:

Attempt #1

My initial attempt was fruitless.  The machine would HALT while booting
from the TK50 (and other media as well) with the following error:

 ?06 HLT INST
     PC = A0050419

Attempt #2

Simplify simplify simplify.  I started pulling boards until I had nothing
left but the TK50 controller, the disk controller, ethernet, the memory
boards, and CPU.  Still no luck.  I pulled one of the memory boards which
left me with a 5Meg system. That seemed to make things work.  I was able
to perform all the basic installation steps for 1.1A on my RD54.  I was
even able to boot from the miniroot and could perform some minor tasks,
but the kernel would shortly panic with memory errors.  Grrr.  I had
noticed that NetBSD detected 9Megs of RAM as opposed to the actual 5Megs
installed.  This stewed in the back of my head for an hour or so, that's
when I became suspect of my arrangement of Q-bus modules. 

I've read most of the VAX related info I could get my hands on, so buried
in some latent memory pocket of my brain I knew I'd done something wrong.
Checking my sources again, I found some material on Q-bus module
arrangement in the MicroVAX/VAXstaion FAQ.  Sure enough, I'd stuck double
length cards in the C-D halves of slots 1-4 in my BA123 enclosure.  I
rearranged things to reflect my interpretation of what I'd read (later
verified as correctly arranged by David Evans).  It turns out that the
system thought those cards I'd placed in the C-D slot halves were another
4Megs of RAM.

Attempt #3

Ack, I still had problems.  The kernel would start booting only to error
out with:

 realmem = 5241856
 availmem = 3296256
 Using 255 buffers containing 261120 bytes of memory.
 backplane0 (root)
 cpu0 at backplane0: MicroVAX 78032/78132
 sbi0 at backplane0panic: Microvax not supported
 Stopped at      0x800809ae:     brb     0x800809c5

At least the amount of memory was detected correctly.  I'm still not
certain why this configuration didn't work, but I've since assumed that
perhaps 5Megs was a bit too tight for a MicroVAX II.

When in doubt or when all other reasonable alternatives are exhausted,
reseat everything and check all your cables.  So I did this.


Attempt #4

Hmmmmm... this time the machine boots with that same error from attempt #1

 ?06 HLT INST
     PC = A0050419

Something's weird with that.  I lucked out and happened to immediately
re-check the short segment of ribbon cable for connecting teh CPU with the
memory boards -- and there it was.  The top right pin was slightly bent on
the CPU board where the memory cable connects to it.  (Hard to describe 
without a picture but) The cable's connector had a slight gap in it which
the pin was sliding into instead of the proper pin hole.

Attempt #5

Bending it back down and reconnecting made the ?06 HLT INST go away, and
then it dawned on me that this was probably the cause of same problem in
my first attempt. I just happened to be fortunate to reconnect the cable
properly after pulling the memory board after attempt #1.

So I plugged the second memory board back into the MicroVAX. 


Atempt #6

Wow, free memory!  The second board, a National Seminconductor clone
board, was billed as a 4Meg board when I bought the used MicroVAX a few
months back. I couldn't find any reference to the particular board on the
net so I just accepted it as being a 4Meg board.  But *ding* it shows up
as having 8Megs under NetBSD.  That brings the machine to a dizzying
13Megs.

So once I got the Q-bus arrangement straightened out, and once I got the
memory cable attached right -- everything was good.   The newly dubbed
`vaxzilla' boots NetBSD 1.1A just perfectly.

Now onto the challenge of 1.2G...

-brian.
---------------------------------------------------------------------------
Brian D. Chase         Systems Coordinator        brian.chase@carpediem.com
-- Compression, Inc. - 13765 Alton Pkwy, Suite B - Irvine, CA 92618, USA --