I'm still digging to find where "D" refers to Dupont and "M" to
Monsanto. Ralph Stamerjohn would certainly know, if he's lurking.

--

From: Lars Poulsen <l...@cmc.com>
Subject: RSX-11D
Date: 1999/11/30

RSX-11D was a whole operating system. Multitasking, multi-user.
I think it required I/D separation. RSX-11D device drivers were
full-fledged tasks; this meant that the code to service a read-
or write-operation on your fancy custom device could suddenly
decide to go write trace record to a file on disk. Very handy
for debugging.

RSX-11D tracked resources on behalf of its tasks. If a kernel
buffer was allocated to hold parameters for an inter-task message,
it was charged to the calling task, and when the system routines
sent the message to the other task, the charge was credited back
and charged to the recipient. When the work item was finally
complete, and the kernel buffer released, the charge was canceled.
Task were only allowed to exeit when they had released all their
resources. This was an admirable idea, but made debugging of
system tasks pure hell: If the task crashed with system data
resources slightly messed up, the system could not safely
release the buffers and credit the right task, so half-debugged
device drivers that crashed would tend to get locked in memory,
so you had to reboot for the next test cycle.

RSX-11D soon grew too large to fit on a 32KB (16KW) machine.
David Cutler decided that it was possible to implement the same
services with much less code, starting from scratch, and went on
to do just that. In 18 months, he basically wrote and debugged
the RSX-11M kernel. The system was released in source form, and
with conditional assembly could be configured at compile time
for any hardware from 8KW 11/05 to a full-house 11/45.
Over the years I worked on RSX-11M based system integration,
I eventually got to take a look at most of the modules of 11M,
and Dave Cutler's name was in the header of each one!