hwe@lanl.ARPA (Skip Egdorf) (04/08/86)
A lot of mis-information seems to be floating around about the DEC
J-11 CPU (11/73, etc.)
I have ported 2.9BSD to the 11/73 (passed on to CSS, naturally)
and am currently running a few of the little beasts...
The 11/73 is a nice fast PDP-11 that can make use of big fast disks.
I use one with an Emulex SC03/CDC-9766 and another with a Sigma RQDX look-alike
and a Fuji 2243. So... the following is from at least one horse's mouth.
The 11/73 looks like an 11/{44,70} with NO unibus map.
The modifications to allow UN*X V7 (2.9 BSD ...) are really simple.
/boot turns on memory management and must be modified to notice the 11/73.
This is done as follows:
1) After the mem mgmt regs are probed to see if split I/D is there,
the code differentiates between a 45 and a {44,70} by checking
for a Unibus map. No map means 45. Add a check for the 11/73
cache control register to differentiate between the 45 and the 73.
The initialization can proceed as for a 70. (make sure to turn on
sep I/D, 22-bit addressing).
Without this, boot thinks the thing is a 45, and only turns on
18-bit addressing. Everything works without modification as long as
this is all you need...
It is nice to see:
73boot
:
2) In UN*X itself, the startup code should initialize the cache
control register as per the {44,70}.
At the point where memory is sized, comment out the code that
does not allow >256K with no Unibus Map. The check is not needed(??).
3) Modify device drivers to handle 22-bit addressing.
For RQDX style disks, and TS11s, this just means to load all the
bits from the extended memory slot in the buffer. All 22 bits
are there to support RH70s anyway...
For things like the RLV controller, a little bit more is needed,
but is not hard.
These mods also allow 23s and 23+s to run with 22-bit backplanes. I have
tried one of my very old (REV B or C) 11/23s, and they run just fine (well
slowly fine).
If you really need 11/73 support, contact keith@seismo.ARPA for info on the
2.9 CSS distribution. It's the best game in town.
Skip Egdorf
egdorf@lanl.{ARPA,UUCP}
#include <disclaimer.h>