The old 4.3 driver for the qdss display did make it out on 4.4BSD-Lite, but
needs to be ported to NetBSD. Since I don't have the hardware, I won't be able
to do it, but hopefully someone else will. (If by SPX you mean the monochrome
VAXstation II, the qv.c driver for the qvss vcb01 board made it out on
4.4BSD-Lite as well and also needs to be ported to NetBSD.)

The tmscp driver works on an older kernel snapshot, but hangs the current
kernel (I suspect a problem with interrupt handling, but haven't had a chance
to grab a current kernel source tree yet, to find out:-). I have the deqna
sort of talking, so hopefully these will be fixed this week.

On bootstrapping, I had the following idea, but haven't tried implementing
it yet..

- write a simple program using VMB calls (which I hear Bertram has figured out
  how to do?) that copies a miniroot-dd to disk and then boots a generic
  kernel. (sort of a standalone copy and boot combined, but gets everything
  from memory.)
- cat this "copy-boot" with a small miniroot dd image and a stripped generic
  kernel together into one file. (since you don't have to have a kernel in
  the miniroot dd, I think the whole thing could be kept to about 800K.)
- use one of several ways to get this file into memory and then the copy-boot
  program would copy the miniroot dd image to the swap area of the disk and
  then copy the kernel to low memory and start it up, running single user
  using the miniroot
- the miniroot running single user could then be used to build the root-fs,
  getting the root dump (or cpio or...) off of tape or the net or??

A few ways the "copy-boot plus miniroot plus genvmunix" file could be loaded
into memory for a microvax are:
- put it on a tk50 cartridge as 512 byte blocks and then "b mua0"
- put it on an rx33 floppy with a boot parameter block prepended as block 0
- via decmop from a net server (someone will need to build a decmop server
  for netbsd:-)
- on two rx50 floppies in a manner similar to the rx33 (the copy program
  would have to know this special case where the second half of the file
  is still on the second rx50)
- via the serial console line using the console binary download command and
  a simple server that runs on another unix box.

For other vaxen, there would need to be other ways, such as load it off of
9 track tape, but at least the program only has to be able to read from a
device to memory well enough to load about 800K. In old 4.3 days there were
programs listed to read a tape block into memory that you could toggle into
memory via the console. Maybe these could be fixed to loop for a tape file?

Anyhow, just a bunch of ideas that may be way off, but I plan on trying some
of them soon, rick