> On 17 Sep, Wulf, Bernhard B DS-RHR-SI-APC wrote:
> 
> > Then I tried out my freshly burned NETBSD CD-Rom on that VAX.
> [...]
> > After that, the VAX is a bit confused now...
> > 
> > The POST will give an error like "?? 000 8 sys 0512" and the 
> > autostart didn't work anymore ( the error is fatal ).
> This error is knowen to us. I have the same trouble on my -90.
> Unfortunately nobody knows what error it is and what caused it. And

I don't know if this is the problem, but I once found out the hard way 
that the flash on the 4000/90 is configured to be written.

I was doing some development for a raw NVAX system, for which I developed
my own FORTH. I did the original base FORTH system on a 4000/60 (booted
by Ethernet from my trusty 3100), then tried a 4000/90 for the NVAX
specific words. I accidentally toasted my 4000/90.

The I/O space used for the console on the 4000/60 is occupied by flash
on the 4000/90. I had forgotten to assemble my code with the 4000/90
consoel support, so I was using the 4000/60 console. Running the
console involves fun things like setting bits in one of the registers.
It just so happened that the address for one of the console DZ
registers on the 4000/60 is at the location of some sort of console
message in the flash of a 4000/90, which just happens to have a bunch
of ASCII spaces there. When my 4000/60 console code tried to work
with the 4000/60 console, it read those spaces from the flash, set
a bit, and wrote them back.

The flash on a 4000/90 is made by AMD. Coincidentally, the Erase Flash
command for the particular chips they used is 0x20; i.e., an ASCII
space! I managed to accidentally erase the flash in my 4000/90 by
inadvertantly attempting to manipulate the 4000/60's console hardware
on the machine.

What really got me in trouble, though, was when I fixed it. I wrote
a program to suck the flash from another machine, pulled the chips on
my dead 4000/90, and re-blasted them. This would have been OK, except
that the chips are really, really close tot he power connector. So I
put sockets in because I couldn't solder to one side of the chips. The
next time field circus looked at the machine I got in trouble.

Roger Ivie
ivie@cc.usu.edu