Groups keyboard shortcuts have been updated
Dismiss
See shortcuts

i286 Multibus UNIX SVR2 - help needed

46 views
Skip to first unread message

Vladislav D.

unread,
Mar 15, 2025, 8:14:47 AMMar 15
to intel-devsys
Hello.

Quite some time ago I managed to recompile SVR2 for x86 when I rolled Microport UNIX for the IBM PC (took me 2 nights of sitting at my machine, figuring out the innards of SVR2's build system and debugging all kinds of random issues to get something that actually compiles, and even then the build system somehow managed to fuck up my emulated VAX's /etc so the build box is now half broken).

However, the stock SVR2 included in the Microport source leak is in fact for the Intel Multibus devkits, not for the IBM PC. While I have been able to roll most of the actual distribution disks from an IBM for use with Microport, the thing that prevents me from having a full SVR2 for Multibus is the two bootdisks, which AFAIK use FM on the first track and MFM on all the others, a disk format that I cannot roll using an IBM running Microport.

The only Multibus UNIX I was able to find in binary form is Intel XENIX, which AFAIK is based on System III, and has device nodes different enough for the SVR2 disk generation scripts to not work, and I am definitely not smart enough to figure out what are the XENIX device nodes I need to roll the bootdisks on there :)

My other idea was rolling SVR4v4.0 for Multibus, but I'm unsure if I can crosscompile Multibus-specific code from an IBM, and even if that's possible, I have no way of making an actual distribution image (the Source Code Provision Build Instructions explicitly specify you need a tape for that, and AFAIK no PC emulators have tape support, I don't have anything physical with a tape drive, and the tape format is likely different between IBMs and Multibus systems anyways).

I have originally posted about this on VCF forums, and someone told me that this group may be interested in my work, so here I am :)

This is a 7z file containing the entirety of my work, transferred from a SIMH VAX onto way too many 1.2MB disks for transfer onto a Microport box for final touches. It's been quite a while, I don't remember quite well what the subfolders are supposed to mean, but IIRC kernel/ contains the IBM-specific Microport stuff (which I doubt people here will be interested in), proto/ contains the compiled version of /usr/src/proto which is required to actually make the disks, and the top-level folder is a backup of /usr/src with everything compiled.


IIRC the disk generation shell scripts in /usr/src/proto have some sort of a trivial bug that causes them to fall into an infinite loop, but I've long since forgotten where that bug is or how I fixed it.

Really hope someone here can finish my work. Thanks in advance.

William Beech

unread,
Mar 15, 2025, 9:19:02 AMMar 15
to intel-...@googlegroups.com
Vladislav,

I just got an email today about helping get an Intel 310 80286 working.  I am pursuing that as I have had one of my Intel 310 operating.  I also have some Intel 320 80386 systems.  These are both multibus systems that did run the Intel Xenix V3.0 and 3.5.  Xenix is based on Unix 7, System III and V.

Have you ported the C compiler and Assembler from the "src" to build these files?  Or are you using an IBM C compiler.  If so, what version?  I think you would have to build the Intel Multibus stuff on an Intel Multibus System.

If I read the code correctly, this is all for a NS 32000 system? The 1.2 MB disk images are the results of your work.  Could you share your entire source tree with us including your tools?

Bill
--
You received this message because you are subscribed to the Google Groups "intel-devsys" group.
To unsubscribe from this group and stop receiving emails from it, send an email to intel-devsys...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/intel-devsys/f2d67df2-51d2-4c60-982d-f3b1639c459cn%40googlegroups.com.

Message has been deleted
Message has been deleted

Email Tester

unread,
Mar 15, 2025, 9:53:30 AMMar 15
to intel-...@googlegroups.com
Something keeps eating my messages. The build platform was a SIMH VAX, not an IBM or any other micro. I used the stock AT&T compiler from usr/src. This is not for the NS 32000. The source tree is in there, see usr/src in the main tarball.

сб, 15 мар. 2025 г. в 16:19, William Beech <nj...@nj7p.org>:

Herbert Johnson

unread,
Mar 15, 2025, 11:50:31 AMMar 15
to intel-...@googlegroups.com
I have no background to respond to this request, but I seem to be able
to find things on Web search that others overlook. Here's outcomes from
"Microport X86 system V Unix". "SVR2" means Unix System V revision 2.

https://fsck.technology/software/x86%20Unix/x86%20Unix%20Installers/Microport/

https://vetusware.com/download/Microport%20Unix%20v1.3.3%201.3.3/?id=17384

https://vetusware.com/download/Microport%20Unix%20v2.3%202.3/?id=17234

- regards Herb

> On 3/15/2025 5:14 AM, Vladislav D. wrote:
>> Hello.
>>
>> Quite some time ago I managed to recompile SVR2 for x86 when I rolled
>> Microport UNIX for the IBM PC (took me 2 nights of sitting at my
>> machine, ... my emulated
>> VAX's /etc so the build box is now half broken).
>>
>> However, the stock SVR2 included in the Microport source leak is in
>> fact for the Intel Multibus devkits, not for the IBM PC.

>>
>> This is a 7z file containing the entirety of my work, transferred from
>> a SIMH VAX onto way too many 1.2MB disks for transfer onto a Microport
>> box for final touches. It's been quite a while, I don't remember quite
>> well what the subfolders are supposed to mean, but IIRC kernel/
>> contains the IBM-specific Microport stuff (which I doubt people here
>> will be interested in), proto/ contains the compiled version of
>> /usr/src/proto which is required to actually make the disks, and the
>> top-level folder is a backup of /usr/src with everything compiled.
>>
>> https://www.mediafire.com/file/s7fhf2gva7uedvo/mpcompiled.7z/file
>>
>> IIRC the disk generation shell scripts in /usr/src/proto have some
>> sort of a trivial bug that causes them to fall into an infinite loop,
>> but I've long since forgotten where that bug is or how I fixed it.
>>
>> Really hope someone here can finish my work. Thanks in advance.


--
Herbert R. Johnson, New Jersey USA
https://www.retrotechnology.com OR .net
preserve, recover, restore 1970's computing
email: hjohnson AT retrotechnology DOT com
or try later herbjohnson AT comcast DOT net

--
Herb Johnson, New Jersey USA
http://www.retrotechnology.com or .net
preserve and restore 1970's personal computing
email: hjohnson @ retrotechnology dot com
or try later at herbjohnson @ comcast dot net

Vladislav D.

unread,
Mar 18, 2025, 9:16:51 AMMar 18
to intel-devsys
Seems like the spam filter got the previous message... sending this again:

I have not done any bringup work for the compiler. The build platform was a VAX running SVR2, which was AFAIK a common way for compiling UNIX for micros back in the day (and apparently Multibus systems too). The source tree is in usr/src relative to the root of t.tar, albeit there's not many object files left in there, IIRC the build system cleaned them.  This is in fact x86 and not NS 32000, see usr/src/uts/ for an explicit mention of "iAPX286".

суббота, 15 марта 2025 г. в 16:19:02 UTC+3, William Beech:

William Beech

unread,
Mar 21, 2025, 3:36:37 AMMar 21
to intel-...@googlegroups.com
Vladislav,

Fine on using the VAX to compile stuff for other target machines.  The Intel 310 and 320 development work may have been a cross compile to start, but Intel did most of the development on Intel 310 and 320 machines.  We worked closely with the Intel engineers while deploying those mini computers.

I find no USR directory in any of the software I downloaded.  I did find /SRC/UTS/ directory.  It contains an MK file and two directories, ns32000 and sys.  sys contains header files.

I looked at some of the executables, but they did not contain X86 code that I recognized.

Where, specifically, did you see the "iAPX286" reference?

Bill
Reply all
Reply to author
Forward
0 new messages