[pups] bootstrapping an RT-11 image via VTserver

[pups] bootstrapping an RT-11 image via VTserver

[Jonathan Engdahl]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2718 bytes --]

I built an 11/23 with 256K RAM, a UDC11 disk controller, one 80 meg MFM hard
drive. The hard drive formatter runs under RT-11. The assumption is that you
have a working floppy from which you can boot RT-11. I want to devise a
method to run the formatter via VTserver.

I figured out how to do this for the RQDX3 and XXDP. You can run XXDP under
E11, load the utility you want to run, then stop E11 and dump the entire 28K
word memory image to a disk file. By hacking a header onto this memory
image, you can turn it into a standalone that can be bootstrapped via
VTserver, just like the disklabel, mkfs, and restor standlones. This is
easy, because XXDP tells you what the restart address is when you load a
program.

The question: is is possible to restart RT11SJ in the same manner as XXDP? I
read some of the manuals, and tried using ODT to restart RT-11 at various
points pointed to by the vector table and fixed area. By starting at the
trap 4 address I can get it to restart and live. However, this seems to make
RT-11 forget that I had done a "GET" of the utility that I wanted to run. I
tried restarting at the RMON address, but that crashes.

One other experiment I've tried is to GET the utility then "START 1000", but
that crashes too. If I GET then just type "START" it lives. What am I doing
wrong?

One assumption is that the utility only uses memory and TT: system calls --
no disk accesses or swapping. Is it possible to abuse RT-11 in this manner?

The utilities I'm trying to run are the UDC11 OCT and UDCT utilities. I want
to make it possible to reformat the hard drive on this single drive system
without tearing it apart and moving pieces to another system. It is also
possible to mess up the NVRAM on the UDC11 so that you cannot boot. If this
happens, and you are lucky enough to have another running system (and I am),
you can rejumper the CSR of the stuck board and fix it on the other system.
Otherwise, you are in big trouble.

I wonder if it would be possible to bootstrap a VM0: image into high RAM and
boot from it?

I realize there are other solutions. The standard answer to questions like
this is "get more hardware". The reason for doing this is I want to invent a
method that will work for a very minimal pile of hardware. And the whole
point of that is to make it possible for people to get a PDP-11 running with
minimal investment.


--
Jonathan Engdahl                 Rockwell Automation
Principal Research Engineer      1 Allen-Bradley Drive
Advanced Technology              Mayfield Heights, OH 44124, USA
Mayfield Heights Labs            engdahl at safeaccess.com  440-646-7326

http://users.safeaccess.com/engdahl/PDP-11.htm

[pups] bootstrapping an RT-11 image via VTserver

[Tim Shoppa]

> One assumption is that the utility only uses memory and TT: system calls --
> no disk accesses or swapping. Is it possible to abuse RT-11 in this manner?

RT-11 does support magtape installation via a special version of DUP
called MDUP.  If you look on a full RT-11 V5.x distribution, you will
find "MDUP.MT", "MDUP.MS", and (depending on version number) "MDUP.AI"
and "MDUP.MU".  These are "snapshots" of RT-11 systems with a single
tape driver (MT, MS, or MU) and a number of different disk drivers
installed (so they don't have to be FETCHed) and running the special
install-only build of DUP.

I think there is some MDUP info in the doc set, though it is
heavily oriented (of course) towards magtape installtion, not
serial port installation.  But the idea is sound.

> I wonder if it would be possible to bootstrap a VM0: image into high RAM and
> boot from it?

Yes, this works, and is even officially supported in the later 5.x
releases (where it is used as part of the AI installation procedure.)

Tim.