Category Archives: dos extenders

A weird EBay interlude on the way to Extending DOS.

Posted on by
22

I’ve always been fascinated with DOS Extenders, as they are such an ingenious method of breaking the rules of an environment by cheating the system. The first one I, like many others I imagine was using was DOSX that is the heart of Windows 3.0 on the 286. An incredibly tiny program but it let Windows run in protected mode, unlocking the potential of the machine. Great!

But as an aspiring programmer things were a quite a bit different. While QuickC for Windows did give you Microsoft C 5 hosted inside of Windows, effectively making the $99 for Windows and $99 for QuickC for Windows the cheapest DOS Extender dev kit of the time, it was of course limited to the 286. The 386 stuff although being the genesis of the DOS Extender was far more expensive.

But then with the advent of EMX for OS/2 and DJGPP for MS-DOS, both of them included DOS Extenders, and even better they were gratis! But of course they were GCC focused, and me being a kid saw the incredible 1MB++ size of CC1.EXE was thinking it’s insanely bloated, and of course slow. I don’t think many people were using 80386sx-16’s but I was.

Then there was this incredible offer from Watcom, who’d just become an internet (yes on usenet!) darling for being the compiler used to port DooM from NeXT to MS-DOS and it was available to students for a whopping $99. Neat!

Watcom C/C++ 10.0 retail packaging

Being in collage this was great. Another big plus of Watcom is that it was able to host on MS-DOS, Windows NT, and OS/2. And target far more!

So much included in the box!

Back then there was such incredible platform diversity in the PC space. Servers could be NetWare, Unix, OS/2 and that new fangled and often delayed Windows NT. Watcom was uniquely positioned to support so many of them, all from the same compiler set.

And of course most famously was DOS/4GW included in the box!

Despite having this, I’d always wanted Phar Lap TNT, as it brought insane features of Windows NT to MS-DOS, like DLL’s, and threads! But it was expensive, and I couldn’t justify it.

And this is where the story gets weird as I was also looking for their 286 product, and I found a copy on Ebay, and even better I easily won!

I can’t get a higher resolution image as the auction has expired out. But take note of the thumbnail.

PBI

And this happened. I was all happy expecting it to arrive when out of the blue the order was cancelled and refunded. I contacted the seller as I was totally wanting to get this, and to see if there had been some screwup in shipping, and they had told me that they didn’t know either has they had been refunded the sale price+shipping. So there we both were, me without my 286 Extender, and the seller with no disks to try to sell back again. And then this happened:

Phar Lap 286 magically reappears!

The disks showed up again for sale. We were both dumbfounded, how did this happen? Somehow someone intercepted the sale, and refunded us both, and then is trying to sell it on their own? Interestingly they won’t ship it to me, almost as if they are trying to block me from it.

Sketchy as hell as they say they are reselling from 3rd parties. Is there some kind of hustle going on where someone at a combined overseas shipping centre grabs random items, they can issue a refund for both parties so we don’t care, and then use a proxy sales site to sell them again?

Interestingly they know the disks read fine, as I know that too as I’d asked the seller to image them for me incase anything happened in shipping. I just never imagined it’d have been intercepted and resold. Maybe it’s common place, but I’ve been using Ebay since 1998 or so and well it’s the first time I’d ever seen something like this. Naturally bringing it up with Ebay is a total waste of time.

I’ll have to continue this with a deeper look into Phar Lap 286|DOS-Extender 2.5 as this is already far too much of a diversion from where I was going. But I thought I should point out this thing so buyer beware about the current listing.

**EDIT

Well seems I had the opposite effect I was looking for:

SOLD!

Free386 (386|DOS-Extender – RUN386 compatible)

Posted on by
Reply

I just found this late last night. The world can always use another DOS Extender, and here we go, out of Japan with Free386!

  • Version: 0.61
  • Date: 2016/12/28
  • Author: nabe@abk
  • Machine: PC/AT(DOS/V)
  • Machine in Japanese: FM-TOWNS, PC-9801/PC-9821
  • Compatible: MS-DOS and XMS and VCPI (with HIMEM.SYS and EMM386.EXE)
  • Language: NASM (Full assembly language)
  • Licence: PDS (Free386.com and Free386’s source files)

You can find it here: https://github.com/nabe-abk/free386

Poorly machine translated readme as follows:

From the futon, I thought i’d publish the “Free386” of dos-extender that I had made before to GitHub.

If you want to publish it anyway, NASM and alink also included together and if there is a DOS environment, i thought that anyone can assemble it is out of luck. I found a bug in alink when generating flat mode.exe/.com file. It’s around here that i started to go crazy in a lot of ways(laughs)

Patching alink was done on Linux. I then used TOWNS-gcc to generate alink.exp, but i used the MP header format that TOWNS-gcc generates. We found a bug that the EXP file cannot run on its own. If this is not corrected, it is not possible to distribute including the development environment because it does not usually have the EXP execution environment. When I checked, there was a bug in how to allocate memory, and when the memory capacity started to exceed 8MB, i was allocating memory space that does not exist in the back.

In fact, Free386 at the time was a lot of files that didn’t work properly, and i was worried because it became unstable, it was a mistake in the allocation of memory that is not. However, to examine this, i created a tool to dump memory maps and paging (i.e., it’s included), it was quite a bit of a hassle.

Now, when the memory allocation bug is fixed, almost all DOS generic EXP files and many TOWNS software now work. However, towns-OS’s biggest mystery system is the CoCo/NSD driver around the moss, and the software written in F-BASIC386 does not start. When you come this far, you want to move it.

So we start editing the CoCo/NSD driver. After a little research, I immediately found out the following.

  • CoCo.EXE resides in DOS memory (real memory).
  • NSDD resides in extended memory.

This means that CoCo is presumed to load nsd files into extended memory and manage that information. Now the question is how to get that management information. Is there information in coco memory that resides like SYSINIT? I thought.

For now, to check the area, Free386, i attached the ability to dump the register status before and after the int service was executed by hooking up the interrupt. We analyzed \hcopy\deldrv.exp, which has the ability to remove the specified NSD driver, as “we need to find the NSD driver and the structure seems simpler” in the mechanism.

------------------------------------------------------------------
Int = 0000_008E  CS:EIP = 000C:0000_1ADC   SS:ESP = 0014:0001_0B88
 DS = 0014        ES = 0060        FS = 0014        GS = 0014
EAX = 0000_C003  EBX = 0000_0001  ECX = 0000_0000  EDX = 0000_66EC
ESI = 0000_0246  EDI = 2074_6E00  EBP = 0001_0C48  FLG = 0000_0046
CR0 = 8000_0021  CR2 = 0000_0000  CR3 = 0002_9000    D0 S1 P1 C0  
------------------------------------------------------------------
*Ret:*
 DS = 0014        ES = 0014        FS = 0014        GS = 0014
EAX = 0000_0003  EBX = 0000_0010  ECX = 0000_0000  EDX = 0001_0C18
ESI = 0000_0246  EDI = 2074_6E00  EBP = 0001_0C48  FLG = 0000_0006
CR0 = 8000_0021  CR2 = 0000_0000  CR3 = 0002_9000    D0 S0 P1 C0  
------------------------------------------------------------------

Information like this comes out a lot in turn. If you look at the changes in coco’s residency and other changes in behavior, you can see that int 8eh/AX=Cx0x is a CoCo service. At the same time, log int 8eh and make a resident.com file (included) run386. I also looked at the behavior of the EXE and explored the commonalities of both of them, and i thought, “How would I design the mechanism if I were you?” We looked up coco services from the perspective of “**.

Then we traced to a service that provides driver resident information called int 8eh/AX=C103h. Using this information, the NSD driver in extended memory could be correctly pasted into memory and implemented on the selector. To verify, I ran deldrv.exp using Free386 and was able to uninstall the NSD driver correctly.

Great. End.

…… I wish I had solved it in that way.

TOWNS-OS is an OS of a mysterious structure, and even though there is a BIOS (TBIOS) of 32bit Native mode for graphicprocessing, some services such as timers use the BIOS of FM-R compatible 16-bit operation as it is. It has an incomprehensible structure to use it from the 32bit program side while managing resources, such as a 16-bit timer BIOS.

In terrible cases, each time the processing and interrupt of real-mode resources such as timers and keyboards, switch the CPU to real mode, if during those real-mode BIOS processing, interrupt the PROCESSING of the BIOS, such as FM sound source or VSYNC occurs, it seems to return to protected mode once.

NSD driver called forRBIOS (for Real BIOS) is the intermediary for this incomprehensible structure. Just as DOS-Extender acts as an intermediary for 32-bit programs and MS-DOS, it acts as a real-mode BIOS and a 32bit program intermediary.

In a RUN386 environment, when forRBIOS.NSD is built in, interrupt vectors such as int 8eh are rewritten so that the NSD driver gets the interrupt. **Where is this information? ** That was a mystery that was left behind. However, RUN386 is a . No matter how much the INT log is done until you run EXP, it doesn’t look like it. If you look at the memory of the coco that is resident, there is no information that seems to be it.

If you’re not going to initialize the resident NSD itself. I thought, i patched the entry of the resident forRBIOS, and when the service routine was called, i tried to use the rough business of falling into an infinite loop was bingo.

Finally, you can now run exp files generated by F-BASIC386 and so on. The analysis results are recorded in the doc. By the way, when you run a program that does not require forRBIOS (written in High-C, etc.), the whole process is slower than when you initialize forRBIOS. I really think this is the specs of TOWNS-OS (laughs)

This is the first time in more than a decade since the development was suspended in 2001, and the DOS-Extender, which is compatible with RUN386, was made.

What is with the price of retro Japanese software?

Posted on by
6

So I was lucky enough to get to Beep before it closed, and I picked out a couple of FM Towns titles (and a junker!), and I thought ‘Return to Zork’ would be a good title, something to compare the MacOS & MS-DOS versions against.

Although slightly faded, it does come in this nice box, which reminds me of the NEOGEO… which is probably an apt comparison.

The artwork has faded, although the CD-ROM inside was still sealed, never before opened. I picked this up for an eye watering Â¥3,480 but flipping the box over revealed the launch price of an astonishing Â¥12,800! I’m not sure what the exchange rate from 1994 was, but even at a generous 100:1 JPY to USD that’s half the price of the old multimedia kits which included the drive, sound card and so many came bundled with Return to Zork.

Another random title I grabbed was even more insane!

¥ 14,800 for Silent Möbius: Case: Titanic!

I need to get a RGB monitor & keyboard to see if this thing even works, meanwhile I fought with UNZ to get it running, and the mouse tracking is totally broken unless you change the DPI scaling, credit to this post in the UNZ ‘BBS’.

One thing is sure, the voice acting in the Japanese version is so terrible.

As people complain about ‘AAA’ games, and paying $60, just look at this! $134 USD for some cartoon boat game thing.. Although I’ve never heard of Silent Möbius or played it, I just saw it was available for the x68000 and PC-98. So I guess it’s one of those Lowest Common Denominator games.

One interesting thing about the FM TOWNS is that they have that ROM DOS with CD-ROM drivers, and their apparently blanket licensing for PharLap 386. Although while I was wasting time looking at cartoon rabbits, someone else scooped but the 386 BASIC kit. Darn.

But in the Return to Zork world, the ‘made.exe’ is in fact a Pharlap 386 EXP, meaning that it runs in 386 protected mode, so you don’t have to struggle with emm386, himem.sys and trying to get a ludercus 580-600kb of conventional memory. Seriously it was such a chore to get this running the manual has a big section on setting up a boot disk. It’s a shame they didn’t license a DOS extender for the US PC platform, although I can see why they chose that route on the FM Towns (and I believe PC98), as there is a RTZ9821 directory there which includes an EXP. Shame it was never relased state side as a patch, as it would have been a GREAT user change. Well that or a Win32 executable.

Links 386 Pro

Posted on by
7

Out on the course

Links 386 is one of those programs that is very easy to love to hate.  It was 1992, and PC’s were mostly being used for business, and high powered 32-bit machines were still insanely expensive.  And then Links 386 happened.  Before there was DooM, Links 386 was the ‘must have’ executive ball clacking device.  And the specs that you needed to run this game were really over the top.  At it’s heart was the Phar Lap 386 Dos extender, along with the virtual memory module.. Which most people would have to rely on.  Links 386 really needs over 8MB of RAM to run.  Yes, that is correct, in 1992 you were recommended to get 8MB (which should have been about $400-800 USD) So you can golf at your desk.  But as the name implies you also needed a 386 classed computer, although ideally you would have one of those new 486’s!  Links 386 also pushed the edge by wanting a VESA capable SVGA card that could use mode 101, 103, or 105.  Although the higher resolutions modes just ended up with logos everywhere, it really didn’t take enough advantage of the higher resolution modes.

Another interesting thing is not only does Links 386 have sound drivers (which means you need a sound card!) but it’ll do voice through the AdLib card.  Also it has a driver model, the WLZ, which I don’t know if they ever published or if people wrote additional sound drivers.

Links 386 installer

The installer is kind of cute, in that it’s flat shading is so old it’s now modern.  How’s that for crazy?

Installation is a snap, at only four diskettes.  They sold additional courses, and I only have one additional course, although oddly enough finding others online is pretty trivial.  However I had far less luck finding the program.  One nice tip to Infocom is that the courses include a score card, like the ones you would get on actual courses.  It really tied the package together.

Don’t copy that floppy!

Although for me, I really bought it for the manual.

Pebble Beach

And I have to admit it, Access Software did a great job.  Even all these years later, it looks great.  But no doubt scaling and placing all the textures is SLOW.  Incredibly slow.

Back in the 90’s I had a lowly support job, and I’d get flown all over the country to help out with issues, and it’d never fail that the regional director would have ‘issues at home’ and amazingly they’d always ask about running Links 386 Pro.  No doubt a lot of people upgraded machines, and got to brag to their buddies on how fast Links would now load.  Running at actual 386 speeds will take nearly a minute to render the screen between shots.

The DOS Extender was forever very touchy.  It took a bit of work to get around it’s issues, with the continuous conflicts with TSR’s, drivers, sound cards, video cards.. It was a nightmare of compatibility issues.  Not to mention that although Phar Lap 4.1 was DPMI compatible, it really didn’t play that nice with OS/2 or Windows.  Microsoft would later come to the rescue for this costed gamer market in the form of buying Links away from Access software, and putting out Microsoft Golf.  And much like SimCity, being able to run this under Windows make it immensely popular in the workplace, as all you needed to find were Windows drivers for your hardware, which vendors did actually support, unlike games.

It’s amazing how companies like Phar Lap, or Rational never did try to make an actual gaming platform for their extenders, leaving it all up to individuals.  My older self says that Microsoft’s rise to prominence in the 90’s was mostly due to their competitors incompetence, rather than their brilliance.

Although DOS Extenders like Phar Lap have been around since the introduction of the 80386, Links 386 Pro is the oldest one I know of.  If you like programs that try their best to bend the limits of what you can or should do, certainly check out Links 386 Pro!

Null DooM, GCC 1.39, GO32 and DPMI

Posted on by
2

phew.

DooM via DJGPP v1 GO32

Around the time of the x68000 port of DooM, I was cutting down the DooM source for a null/portable version.  I never could get it to actually run either using EMX or  DJGPP 1.03, as I couldn’t get it to link to save my life with a constant never ending battle of unresolved symbols. After a while I just used what I had towards the x68000 version and concentrated on getting it up and running, and just shelved the null/portable effort.

Later on I wanted to get it running again as part of messing with another cross compiler, as DooM isn’t a trivial application to port and verify correct operation. And in the process of trying to get the null version to build and run on Windows using TDM GCC, I wanted to make sure it at least kept compiling with GCC v1.x.

Once more again I was able to compile individual files but unable to link.  But this time, I just looked at the diffs for binutils, I thought it should be somewhat easy to get hosted on Windows.  Although versions may point to binutils 1.0, I had to use binutils-1.9.tar.gz even though the diffs are against Mar 24 1991, and the source for 1.9 is dated April 17 1991.

My first effort gave me a linker that would happily link, but go32 would either refuse to run the executable, or just crash.  I was going to give up again, but I found mention in another file that DJGPP actually uses the linker from G++, the C++ compiler which was a separate thing in the late ’80s and early’90’s.  This time it worked, and I could link a trivial hello world style application!

Now that I finally had a cross linker actually working, I didn’t want to compile under emulation, so looking at the other diffs, they didn’t look too extensive. I went ahead ,and took DJGPP v1.06 and patched up the compiler & assembler to get a full cross toolchain.  And in no time, I had a null version of DooM running on MS-DOS well at least tested on DOSBox.

This was fun, and all but I didn’t see any easy way to do fun things like hook interrupts so I could get the keyboard & clock like any good MS-DOS program.  DPMI greatly eased this kind of stuff, so looking at the DJGPP history, DJGPP v1 version 1.10 actually adds preliminary DPMI support!  And in the next version, DPMI was much more better supported, however the binary format had changed from a.out to COFF as part of the move to v1.11. I was able to take the memory, and DPMI portions from the final v1.12 libc, and manually build and run them against the v1.06 library / dev tools.

And much to my surprise, it actually worked!  At least having the wrong format didn’t have any effect on how GO32 worked for me.

So feeling lazy, I snagged some of the support code from Maraakate’s revamp of DooM, just to make sure of the timer code, and the keyboard code, and again verified that I can build with the keyboard & timer ISR and I’m able to play the v1.9 shareware & commercial levels fine.  I haven’t done a thing to clean up or update the DooM source itself against all the dozens of bugs and issues with Ultimate DooM, or other games like Chex Quest etc.

I’m sure 99% of people wouldn’t care but you can download it here:

Win32_DJGPPv1_DooM.7z
Download crossdjgppv1

Although I’m using DPMI to drive realtime events, if I looked further at the GO32 v1.06 environments I could either figure out how it operates it’s timer, or modify the extender directly to drive the PIC timer and keyboard as I need.  But overlooking that, the vintage 1991 software is more than capable of running DooM.

Wolfenstein 3D for DOS/4GW update

Posted on by
Reply

If you remember a while back, I had found the ‘missing link’ of Wolfenstein to Wolfenstein SDL, Wolf4GW.  Well Tobias has cleaned it up somewhat, and now it compiles on the latest builds of OpenWatcom 2.0c!

The first thing you’ll notice if you try to compile it, is that now it’s a single source file, that includes all the other modules.  And it compiles FAST, for me 1 second fast.

From the changes:

  • Compiles with OpenWatcom v2.
  • Keys (for Run, Shot…) are shown.
  • Hang with optimization is fixed.
  • Missing Spear of Destiny SignonScreen added.
  • Inter-procedural optimization (unity build).
  • External assembler routines re-implemented in C.
  • Better interrupt enablement /disablement.
  • Dead Code removed or #ifdefined.

So, if you want to Wolf3d, or SPOD, I’d check out Tobias’s Wolf4GW if you have a 32bit capable machine.  The maps load instantly, and it just feels all around much more smoother than the old 8086 code.

Porting Quake II to MS-DOS pt4

Posted on by
15

Bringing it all home for release day.

Bringing it all home for release day.

Since the last update we got some help in a few fields that have really fleshed out this ‘experimental’ port into a full fledged port.  First RayerR helped us with the fun of getting us onto the latest deployment of DJGPP, 2.05 (rc1).  It’s always nice to be in the latest available release.  Next in a passing comment, Ruslan Starodubov had mentioned that he had gotten a much older build of our QDOS to support the Intel HDA sound chipset via the  MPXPlay sound library.  I wrote to the author of MPXPlay, Pádár Attila asking for us to distribute his source in our project, and he granted permission.

So at this point things were looking good.  The only ‘feature’ that modern OS’s really held over us was the ability to dynamically load and unload game modules on the fly.  I had tried to use DLM, but it stripped the DPMI functionality out of the MS-DOS Extender making the port really useless.  I tried to build the newer DXE3 support but had no luck.  I suspect now my native tool chain was interfering with the build process. But Maraakate managed to get it to not only build, but to run!

Adding in DX3 support was relatively painless.  I first looked at DJGPP’s FAQ and downloaded the example code.  In the example code there was small helper functions to make unions and check the symbols.  If they didn’t exist a printf was spit out to alert you about it.  To resolve the issue you simply just add DXE_EXPORT to the other list of missing exports.

Compiling the game code was easy, again referring to the example I saw that basically they compiled it the same, but at link time you use DX3GEN and -U flag to ignore unresolved symbols.

The biggest head scratcher was the Sys_GetGameAPI failing to find GetGameAPI from the DX3.  After some piddling around I noticed that it listed GetGameAPI as _GetGameAPI inside the DX3 itself.  I added the underscore and it worked!

Other things that were relatively to easy to import was R1Q2’s HTTP downloading code.  Compiling CURL was kind of tricky because of the linking order, but thankfully neozeed figured it out quickly.

All of Yamagi’s Quake 2 updated game DLLs were all diff’d by hand using BeyondCompare to make sure I didn’t clash using some newer functions that weren’t available in DJGPP.  I also merged their Zaero code with their baseq2 code by comparing Zaeros code to the Quake 2 SDK, marking every thing that was changed.  The result is a really stable Zaero game code.  If you haven’t played Zaero check it out.  I think it’s a lot better than Rogue, but Xatrix is probably my favourite (even over stock Q2).

Other cool things I’m glad to get into the code was the GameSpy Browser.  It took me quite a bit of work to get it where it is, but it’s really nice to just be able to ping to a master server (a custom GameSpy emulator I wrote specifically for Q2DOS.  Source is not finalized yet, but will be available soon for those curious), pick a server and go!  All in DOS!

So here we are at the end of the journey.  Or at least safe enough for a 1.0 release.

To recap, we have:

* VGA
* SVGA (LFB modes only)
* Mouse
* Keyboard
* SoundBlaster and Gravis UltraSound Family
* CD-ROM music
* OGG music
* Networking (You need a packet driver)
* Loading/unloading game DLLs in DX3 format.
* Intel HDA support -hda
* Mouse wheel support with -mwheel

And I should add, that it works GREAT on my MSI Z87 motherboard.

You can download Quake II for MS-DOS on bitbucket.  And as always the source is available here.

Don’t forget you can always make bootable USB stick with DOS, or CD-ROMs.

Continued in Part 5!

Wolfenstein 3D for DOS/4GW!

Posted on by
22

After reading about the Blake Stone compile fixes, as it was a Wolf3d port, I came across a post on the forum Wolf3d Haven about trying to find the source code to something called wolf4gw.  Now wolf4gw is a port of the Borland C source of Wolfenstein 3d to Open Watcom C++‘s 32bit MS-DOS extender DOS/4GW, done by ‘ripper’.

The project eventually gave way to wolf4sdl, and as they say the rest is history.

Sadly it seems that just about all the source copies of wolf4gw were lost, except I did manage to find an ‘improved’ version simply refered to as wolf3dx.  From the blurb:

Tricob has released the Wolf4GW-based source code of WolfDX. Included is a text file called (Tricob).TXT.

So I have been using Watcom 10.0 for Duke Nukem 3d, however, this version relies on the _asm inline assembler which was introduced in Watcom 11.  However Watcom 11c had issues with some of the assembly forcing me to go even further to OpenWatcom 1.3.  For me the install was easy, I used CrossOver to install OpenWatcom for DOS-DOS32bit only, copied the compiler into DOSBox, and played mostly with the makefiles, and finally got a working exe!

Screen Shot 2013-07-12 at 11.57.24 AM Screen Shot 2013-07-12 at 11.58.00 AM

I know it may not look like much, but really it is running in 32bit protected mode!

Since all of this is open/freeware/shareware I can redistribute OpenWatcom, the source to wolf3dx, and the shareware levels of Wolfenstein.  Naturally I’m using DOSBox to compile and test, but you can use anything that can run MS-DOS 32bit stuff.

Download my archive here.

DJGPP 1.03 saved thanks to shovelware + cd.textfiles.com

Posted on by
23

I can’t stress enough just how awesome cd.textfiles.com is for finding ancient stuff!

I’m not sure why I started on this quest but I was looking for some old finicky DOS extender, and started hunting for Go32, the first DOS extender used by DJGPP.  And for the heck of it, I wanted to find the first version, which I pretty much had assumed was lost to the mists of time.

However the CD-ROM shareware collection called MegaROM-1 actually had a ‘full’ copy of one of the first versions of DJGPP, 1.03.

Installation is pretty straightforward, however you have to use pkunzip for all the various old ‘methods’ of storing data in zip files, I found infozip leaves things out..

Also DJGPP 1.03 uses a LOT of environment space.. which is more so a problem for people running real MS-DOS on real machines.. (there are some!)…

Hello World!

It runs in DOSBox, but there is no doubt some stack corruption as trying to run things like dos edit result in:

Packed file is corrupt

But at least we can run more than one copy, or use a native editor.

GO-32 from this era is *NOT* DPMI compliant, nor is it VCPI compliant.  And its based on GCC 1.39, which was a popular level with things like 386 BSD, although it seems early Linux used GCC 1.40 ..  The tool chain by default outputs the GNU a.out format, but relies on modifying the linker that was separately included in G++.  Later versions of GO32 included VCPI support, and near it’s end of life version 1.10 added support for DPMI which greatly simplified things like hooking IRQ’s and doing DMA.

For those who want to play, without the pkzip fun, I’ve slapped it into a single 7zip file.  It’s not even a megabyte.  But it was 1991, when 4MB of ram seemed like an incredible amount of memory!