On the trail of PCC for the 8086

Posted on November 4, 2022 by neozeed

While on discord the topic came up of why there is no good/free C compiler for MS-DOS. Oh sure there is OpenWatcom but the 2 heavy hitters of the era, Microsoft C & Borland C are not open in the slightest.

There is DeSmet C, although it’s source is full of unnamed structs meaning that building it with anything sane would require a ‘lot of work ™’ which of course is not what I’m all that about. Instead, I remembered a directory up on TUHS /Applications/Portable_CC with a zip file 8086.zip Although this is a zip file, you’ll want to unzip on something Unix-y as there is a lot of case duplicate files. That said this is a PCC port to the 8086, which includes a libc, 8087 support, and is all expected to be built on a VAX-11/780 running 4.1BSD. Now this ended up being a stumbling block because I tried a *LOT* of things thinking that they were upwards compatible with 4.1, and the answer is USE 4.1!

So to effectively get going you’ll need a SIMH VAX780 and just follow my old steps on Installing 4.1BSD. As far as the zip file, I used Linux but had to create a tar file specifying the Unix v7 format with:

tar --format=v7 -cf pcc.tar .

And of course, convert the tar file to a simh tap file. Or if you are like me, just download a tap file here: PCC-Machines.tap.bz2.

With that said it’s a very strange setup as it relies on the 4.1BSD Vax environment so much that there is assembly injected into the linker.

asm("movc3 r8,(r11),(r7)");

So this will not cleanly run. Just as it depends on many system a.out specifics on building for MS-DOS. It’s not so much a MS-DOS tool chain, rather it outputs to vax a.out and uses a slightly modified vax linker. The MS-DOS magic happens in the conversion of the final a.out into a com file.

That is right it’s a VAX specific cross compiler that only build’s COM files.

I’ve managed to build some trivial stuff, and they work. Sadly my attempt at building that InfoTaskforce of ’87 failed.

I haven’t dug that much further into the linker although I have to wonder if a GNU cross linker to make a.out could make something that the conversion program would be happy with. The assembler of course doesn’t work, perhaps it’s something with packing structs?

As always, the simple stuff looks trivial but it was a fair bit involved.

Since there is no real ‘cc’ it’s a script but the vauge steps are:

/lib/cpp -I/usr/src/pcc/Machines/8086/lib86/include hi.c hi.i
/usr/src/pcc/Machines/8086/c86/c86 < hi.i > hi.a86
/usr/src/pcc/Machines/8086/a86/a86 hi
/usr/src/pcc/Machines/8086/a86/ld86 -X -N -r -o hi.out /usr/src/pcc/Machines/8086/lib86/crt0.b hi.b /usr/src/pcc/Machines/8086/lib86/libc.a 
/usr/src/pcc/Machines/8086/a86/cvt86 hi.out hi.com

It kind of makes sense.

Seems like somehow a lost opportunity in of itself back in the day

Confessions of a paranoid DEC Engineer: Robert Supnik talks about the great Dungeon heist!

Posted on August 5, 2019 by neozeed

What an incredible adventure!

Apparently this was all recorded in 2017, and just now released.

It’s very long, but I would still highly recommend watching the full thing.

Bob goes into detail about the rise of the integrated circuit versions of the PDP-11 & VAX processors, the challenges of how Digital was spiraling out of control, and how he was the one that not only championed the Alpha, but had to make the difficult decisions that if the Alpha succeeded that many people were now out of a job, and many directions had to be closed off.

He goes into great detail how the Alpha was basically out maneuvered politically and how the PC business had not only dragged them down by management not embracing the Alpha but how trying to pull a quick one on Intel led to their demise.

Also of interest was his time in research witnessing the untapped possibilities of AltaVista, and how Compaq had bogged it down, and ceded the market to the upstart Google, the inability to launch a portable MP3 player (Although to be fair the iPod wasn’t first to market by a long shot, it was the best user experience by far).

What was also interesting was his last job, working at Unisys and getting them out of the legacy mainframe hardware business and into emulation on x86, along with the lesson that if you can run your engine in primary CPU cache it’s insanely fast (in GCC land -Os is better than -O9).

The most significant part towards the end of course is where he ‘rewinds’ his story to go into his interest in simulations, and of course how he started SIMH when he had some idle time in the early 90’s. SIMH of course has done an incredible amount of work to preserve computing history of many early computers. He also touches on working with the Warren’s TUHS to get Unix v0 up and running on a simulated PDP-7 and what would have been a challenge in the day using an obscure Burroughs disk & controller modified from the PDP-9.

Yes it’s 6 hours long! But really it’s great!

Mach ’86 on the SIMH VAX

Posted on July 15, 2019 by neozeed

While the kernel may boot on SIMH there is certainly something going on with the SIMH emulation of the hardware that threw me for a few loops. I had a pre-installed version of 4.3BSD which was on a RA81 disk but shortly after loading the kernel various binaries wouldn’t load, filesystems wouldn’t mount and of course the inevitable file corruption.

This led me to the fun of loading up 4.3 onto RP06 disks as they are smaller and I was hoping less prone to errors. During this fun, I found this page on plover.net, which as a fun filled tangent shows how to use the Quasijarus console floppy image to run the standalone programs. With the latest version of SIMH, I can run format and it initializes the disks, so I almost think it may be possible to do some kind of ‘native install’ on the VAX-11/780 SIMH, although It’s not what I was in the mood for.

So finally with an install over several RP06 disks, I was lucky enough to figure out how to build the Mach’86 kernel, and get it to boot, and then the corruption happened again. Luckily for me I had snapshotted the disks before experimenting and noticed how even those had issues booting up. It’s after a bit of searching I found that other people may have issues with SIMH’s Async I/O code, and the best thing to do, is just to disable it with a “set noasync”.

Now my disks could boot under the Mach kernel, and I could self host!

Setting up the build involved copying files from the ‘cs’ directory to their respective homes, along with the ‘mach/bin/m*’ commands to the /bin directory. Configuring the kernel is very much like a standard BSD kernel config, however the directory needs to exist beforehand, and instead of the in path config command run the config command in the local directory.

While maybe not perfect, keep in mind I haven’t found any real instructions as of yet, so this is more of a ‘wow it booted’ kind of thing at the moment.

While this kernel does have mentions of multi processor support I haven’t quite figured out what models (if any) are supported On the VAX, and if SIMH emulates them. While oboguev.net has a very interesting looking multiprocessor VAX emulation, VAX_MP it’s a fictional model based on the microvax, which I’m pretty sure 4.3BSD/Mach’86 is far too old for.

And for those who want to play along, here is a zip file of the disk images, emulator and config file I’m using, Mach86.zip

4.2BSD TCP/IP networking

Posted on November 3, 2017 by neozeed

I got this note from Allen Garvin, that details his adventure in taking a stock 4.2BSD VAX image, getting it running on SIMH, and turning on the network stack.

Although 4.2 may have had security issues, (R Morris), and had some clear issues with scaling. Along with a whole host of other issues. Naturally if you want something more ‘robust’ on modern networks, you’ll want 4.3BSD which corrected quite a number of issues.

You can read about it over on his blog. It’s very good with step by step instructions, goes over retrieving the NIC driver, re-building the kernel, and getting it operational on our LAN.

Quasijarus integrated source available on github!

Posted on September 8, 2017 by neozeed

https://github.com/abs0/4.3BSD-Quasijarus

It also features some additional fixes!

Time to do an update on this OS at some point. Although the author doesn’t like SIMH or emulation in general, it’s the only way the majority of us will ever get to try this one out..

Research UNIX v8

Posted on March 30, 2017 by neozeed

v8 on SIMH

So what the heck is Research UNIX v8? Or even what is Research UNIX? Well a query against utzoo gave me this answer:

>I've seen people that use System V and the like refer to their Unix as
>"tenth edition" or "ninth edition", or whatever. I've always seen things as
>"System V release n", or whatever. Anyone know the difference between these
>different naming schemes ?

There are actually three designations: Versions, Editions, and
System/Releases. The proper names of the first six Unixen were
"The #th Edition". Colloquially, people called them "Version #".
The Version Sixth Edition split off several variations, one of which
became Version Seven (the Seventh Edition) and sired BSD. From
several others, System III was born, and later named System V.
Tacked onto this name were Release numbers and yes, Versions.
So you will see things line SVr3v2.

The Eighth, Ninth, and Tenth Editions seldom left Bell Labs
and are also referred to as "Research UNIX". Another system
(not UNIX) they are playing with is called "Plan 9". Every so
often, a feature, such as STREAMS, finds its way into System V.

In some ways, Research UNIX is closer to BSD than to System V.

In short, UNIX began it’s life as a research project. Until recently versions 1-6 & 32v were available to the public. However the later versions, 8,9,10 were not. However thanks to the work over at TUHS it’s available for non commercial use:

Alcatel-Lucent USA Inc has permitted usage saying "will not assert its
copyright rights with respect to any non-commercial copying, distribution,
performance, display or creation of derivative works of 
Research Unix®1 Editions 8,9, and 10."

So awesome!

The version of Research v8 is split onto 2 tape images, one for the graphical terminals, and the other for the OS install onto the VAX. The distribution is not suitable for any standalone operation, and requires a previously installed 4.1BSD machine, with a second disk to install v8 onto. Part of the installation requires you to compile your own kernel. I ran into a bit of problems as it’s not a 100% process, but after referencing this guide from David du Colombier, I had the system up and running. Naturally reading the installation manual helped a great deal too.

As always there is strange artifacts left in the backup, such as this scoreboard from rogue:

Top Ten Rogueists:
Rank Score Name
1 5545 Rog-O-Matic XIII for mjs: quit on level 17.
2 5043 ken: killed on level 23 by a dragon.
3 3858 zip: killed on level 16 by an invisible stalker.
4 3249 Rog-O-Matic VII: killed on level 16 by an invisible stalker.
5 2226 Rog-O-Matic VII: killed on level 13 by a troll.
6 2172 St. Jude: killed on level 13 by a troll.
7 1660 Rog-O-Matic VII: quit on level 11.
8 1632 Chipmunk the Jello: killed on level 10 by a centaur.
9 844 Rog-O-Matic VII: quit on level 5.
10 401 Rog-O-Matic VII: killed on level 4 by a snake.

Does this mean Ken Thompson was an avid rogue fan? Perhaps. Naturally I quickly compiled the v100 version of aclock, and had it running.

aclock on v8

I’ll have to edit this and more and more as I find out, but I’ve been busy in real life, and of course I know that in addition to v8, there is also v9 & v10 to tackle.

As always, if you want you can download my pre-installed from my site : researchv8.7z

You will have to bring your own copy of the SIMH VAX-11/780 simulator. As of 31/3/2017 ther is issues with the github version of SIMH, and you will have issues with the disks on the VAX. You need to disable the async with a simple set command in your ini file:

set noasync

And you should now be good to go! As always you’ll have to battle the 404 page for the correct link and the username & password.

sorry.

Building and using GCC 0.9 aka the first public version

Posted on December 1, 2016 by neozeed

In my quest for old software, I’ve seen this file in multiple searches, gcc-0.9.tar.bz2, which is the first version of GCC!

article from virtuallyfun.superglobalmegacorp.com

GCC 0.9 on SIMH VAX / 4.2BSD

From the original announcement:

Date: Sun, 22 Mar 87 10:56:56 EST
From: rms (Richard M. Stallman)

The GNU C compiler is now available for ftp from the file
/u2/emacs/gcc.tar on prep.ai.mit.edu. This includes machine
descriptions for vax and sun, 60 pages of documentation on writing
machine descriptions (internals.texinfo, internals.dvi and Info
file internals).

This also contains the ANSI standard (Nov 86) C preprocessor and 30
pages of reference manual for it.

This compiler compiles itself correctly on the 68020 and did so
recently on the vax. It recently compiled Emacs correctly on the
68020, and has also compiled tex-in-C and Kyoto Common Lisp.
However, it probably still has numerous bugs that I hope you will
find for me.

I will be away for a month, so bugs reported now will not be
handled until then.

If you can’t ftp, you can order a compiler beta-test tape from the
Free Software Foundation for $150 (plus 5% sales tax in
Massachusetts, or plus $15 overseas if you want air mail).

Free Software Foundation
1000 Mass Ave
Cambridge, MA 02138

[tapes are generally in Unix tar format. If you have other needs,
write to the above address, and ask if they can be met. -len]

And indeed, the files are dated 22/03/1987 making this the first public release of GCC.

GNU CC is a fairly portable optimizing C compiler intended for
machines with 32-bit words that have several registers and address
memory in terms of 8-bit bytes. It supports full ANSI standard C, not
including libraries (which we do not consider to be part of a
compiler).
Currently we have working machine descriptions for the Vax and for
the 68000/68020 (including 68881 support).
Optimizations performed by GNU CC include:

Invariant code motion out of loops.
Common subexpression elimination.
Automatic register packing (register declarations are unnecessary
and ignored).
Constant propagation and elimination of consequent dead code.
Copy propagation.
Elimination of dead stores.
Jump optimization including cross-jumping.
Delaying of stack adjustments after function calls.
Arithmetic performed in subword types when appropriate.
Many local optimizations.

GNU CC runs about as fast as PCC.
Most of the optimizations are machine-independent or controlled by a
machine description. GNU CC takes advantage of all the 68020
addressing modes that we can see how to make the Sun assembler
assemble. Debugging output for DBX is available whether you request
optimization or not.

Seeing as 4.3BSD didn’t ship until 1988, I went ahead and set out to build this on 4.2BSD. The first stumbling block I hit is that GCC needs bison. The oldest version of bison I have is 1.25 which honestly is just too new! However in the same location as GCC is this file gnu1988.tar.bz2 which contains all of the current GNU software of 1988! And what is on that tape?

bison-1.00
gcc-1.21
gdb-2.5.1
gplusplus-1.21
libgplusplus

So this is probably as old as it is going to get, so I downloaded and went to compile bison, however getopt is a missing call! A creative search found getopt.c (local mirror) and even better PCC liked it enough to get a running bison so I could then configure GCC.

Configuring GCC is a manual process, but not too involved:

Make a symbolic link from file `config.h’ to the top-level
config file for the machine you are using. Its name should be
`config-MACHINE.h’. This file is responsible for
defining information about the host machine. It includes
`tm.h’.
Make a symbolic link from `tm.h’ to the machine-description
macro file for your machine (its name should be
`tm-MACHINE.h’).
Make a symbolic link from `md’ to the
machine description pattern file (its name should be
`MACHINE.md’)
Make a symbolic link from
`aux-output.c’ to the output-subroutine file for your machine
(its name should be `MACHINE-output.c’).Make sure the Bison parser generator is installed.Build the compiler. Just type `make’ in the compiler directory.

And in a minute I had GCC compiled. I ran it with -v and got this output:

# gcc -v
ld /lib/crt0.o -lc
Undefined:
_main

It really is nowhere near as featured as 1.21 that is for sure! So time to do a simple hello world program:

# cat hello.c
#include <stdio.h>
void main(){
printf(“GCC 0.9 in action!\n”);
}
# gcc -v hello.c -o hello
cpp -Dvax hello.c /tmp/cc002050.cpp
cc1 /tmp/cc002050.cpp -quiet -dumpbase hello.c -noreg -o /tmp/cc002050.s
as -o hello.o /tmp/cc002050.s
ld -o hello /lib/crt0.o hello.o -lc
# ./hello
GCC 0.9 in action!

And there we go!

I don’t know why, but I haven’t seen anything about anyone actually running GCC 0.9. Or even where or how they found this ‘lost’ file. Let alone anyone even building or running it in 2016.

For anyone who wants to try, SIMH tape files of the binaries are here:

And of course source tapes are here.