Windows 9x Video Minidriver HD+

Posted on June 6, 2022 by Michal Necasek

The OS/2 Museum has made available the first version of a display driver disk for Windows 9x running on VirtualBox. The driver uses a linear framebuffer and supports 8/16/24/32bpp modes with resolutions up to 1920×1200 pixels (see more below). The driver is not accelerated but tends to be very speedy on modern hardware.

Windows 95 in a usable resolution

I’d like to say that it was easy to adapt the existing Windows NT video miniport driver for Windows 9x… but of course it wasn’t. The Windows 9x display driver model is completely different and has nothing in common with NT. The Windows 9x display driver has much more in common with Windows 3.1 (and 3.0 and 2.x) drivers, and it has clearly directly evolved from those older drivers.

So what makes it a “minidriver”? A Windows 2.x/3.x display driver has to implement a very significant chunk of GDI. Bit blits, lines, text output. There is a lot of cases to handle and a great deal of complexity. To give some sense of the complexity, the Windows 3.1 DDK sample driver for Video 7 cards is about 1.6 MB (circa 60,000 lines) of assembler source code. And that’s just for 8bpp displays.

Continue reading →
Posted in Development, VirtualBox, Watcom, Windows 95 | 66 Comments

Compaq EGA Technical Reference Guide

Posted on June 1, 2022 by Michal Necasek

A rater nice Compaq EGA Tech Ref recently turned up on archive.org, under the title COMPAQ Enhanced Color Graphics Board Technical Reference Guide. It’s from December 1986, relatively late in EGA’s life (given that it was about to be obsoleted by the VGA in just a few months), and it is rather good.

The Compaq Tech Ref is significantly more detailed and in-depth than the IBM Tech Ref. On the other hand, IBM supplied a BIOS listing which is very valuable, both when interfacing with the BIOS itself and when trying to understand certain details of the hardware. Compaq documented the BIOS well and provided nice examples, but that’s not the same as the actual source code. As is often the case, looking at both references is best.

Posted in Compaq, Documentation, PC history | Leave a comment

NT video miniport UHD

Posted on May 27, 2022 by Michal Necasek

I’ve finally managed to update the previously released NT video miniport. The reason for the update was indirect, hacking up the video “hardware” access code to support other environments. That led me to separate the generic mode set code from the mode table oriented logic. Since the NT miniport does not need the mode tables, the resulting boxvideo.sys is now about 1 KB smaller without losing any functionality.

VirtualBox NT video miniport driver, now at version 1.6

As before, the miniport should run on any x86 version of NT from 3.1 up to 7, with the caveat that for NT 3.x, the user has to supply FRAMEBUF.DLL from the NT installation media (see included README file). Also for NT 3.1 only, the installation and mode selection process is different from later versions.

At some previous point, so long ago that I don’t even remember, I expanded the list of supported modes to go up to 5,120×2,880 pixels. This is not extensively tested because I don’t have such a large display, but at first glance it does not appear to be completely broken.

Get the updated driver here. The source code can be viewed here.

Posted in NT, VirtualBox | 5 Comments

IDENTIFY ESDI DRIVE

Posted on May 25, 2022 by Michal Necasek

As previously mentioned on this site, the IDENTIFY DRIVE command in the ATA specification almost certainly first appeared in ESDI controllers supplied to Compaq by Western Digital.

Since I have now finally secured a working ESDI hard disk, I could do some probing. Unfortunately I don’t have access to a WD1005 ESDI controller that should be extremely close to what Compaq used circa 1986, but I have two of its successors, WD1007A and WD1007V. A spare WD1005 anyone?

My WD1007A (1987) clearly came out of a Compaq machine and it is a rather interesting piece of hardware. It is a hard disk controller only, with no floppy support. It also has no BIOS (though other models did, and the PCB clearly has room for it).

The newer WD1007V-SE2 (1989) might be a retail model; it includes a floppy controller and a BIOS, although the BIOS can be disabled since it wasn’t necessary in many PC/AT compatibles.

WD1007A and WD1007V ESDI controllers

Now, what’s very interesting about these WD ESDI controllers is that from a software perspective, they would be very difficult to distinguish from an IDE drive. They support the exact same registers and commands as a standard PC/AT controller, but additionally also implement the IDENTIFY DRIVE command. At least in the case of the WD1007V, the controller also supports READ/WRITE MULTIPLE commands, READ/WRITE BUFFER commands, and probably some form of cache control. In other words, the WD1007V even acts like a not so basic IDE drive.

Now back to IDENTIFY DRIVE. That would have been the big difference between a controller for ST506 style MFM or RLL drives and an AT-compatible ESDI controller. ST506 (or ST412, if you’re Seagate) drives simply have no mechanism to report their characteristics to the controller. But ESDI drives do. Anyway, let’s look at the details…

Continue reading →
Posted in ESDI, IDE, PC hardware, PC history, Western Digital | 20 Comments

The Strange Case of GetEnvironmentStringsA

Posted on May 22, 2022 by Michal Necasek

It was recently pointed out to me that a simple “hello world” style application built with Open Watcom C/C++ 1.9 does not run on Win32s version 1.30, even though the same executable runs just fine on Windows NT 3.51, Windows 95, or Windows 10.

More specifically, the program crashes rather early on Win32s. With the help of map files and source code, I established that the crash occurs in an internal function called __setenvp, which tries to dereference a null pointer stored in an internal variable _RWD_Envptr.

The _RWD_Envptr variable is filled in by the GetEnvironmentStrings API in the C runtime startup code. The GetEnvironmentStrings API call ends up importing GetEnvironmentStringsA from KERNEL32.DLL. And clearly GetEnvironmentStringsA is failing on Win32s, although it works just fine on NT and Win9x.

Further probing revealed that the GetEnvironmentStrings API has curious history. On Windows NT 3.1, there was only GetEnvironmentStrings (no A or W suffix). On all later Win32 implementations, starting with NT 3.5, there’s GetEnvironmentStringsA and GetEnvironmentStringsW, as well as FreeEnvironmentStringsA and FreeEnvironmentStringsW.

On NT 3.1, there was no FreeEnvironmentStrings, presumably because GetEnvironmentStrings returned a pointer to existing memory that couldn’t be freed (and would be freed at process termination anyway). On NT 3.5, GetEnvironmentStringsA converts the strings provided by GetEnvironmentStringsW and allocates memory for the converted strings, so there is something to free.

A quick experiment with Microsoft Visual Studio 4.0 showed that a test application does run on Win32s; reading MSVC 4.0 runtime source code also revealed that Microsoft calls GetEnvironmentStringsA and immediately terminates the process if GetEnvironmentStringsA fails. So… how can that work on Win32s?

Continue reading →
Posted in Development, NT, Watcom | 23 Comments

1989 Networking: OS/2 NetWare Requester 1.1

Posted on May 7, 2022 by Michal Necasek

When I wrote about the pre-release NetWare Requester for OS/2, the oldest archived officially released NetWare OS/2 Requester was version 1.2 from 1990. In the meantime, version 1.1 of the requester showed up, although I only became aware of that very recently.

Technically the 1.1 Requester does not appear to be vastly different from the 1.0 pre-release. Which is unsurprising, because the OS/2 1.1 kernel was not that different from OS/2 1.0; the big news in OS/2 1.1 was the Presentation Manager GUI, but that was not something Novell cared about at the time.

The Requester was shipped on three disks, labeled SYSTEM, PUBLIC-1, and PUBLIC-2. As the names suggest, the PUBLIC-1 and PUBLIC-2 disks were meant to be copied on the server, whie the SYSTEM disk needed to be installed on each OS/2 workstation.

Unlike the pre-release, the 1.1 Requester came with an actual installer, although what it did was a little underwhelming.

NetWare OS/2 Requester installer

The installer copies files but does not actually change CONFIG.SYS at all. Instead, Novell provided a template which can be appended to the existing CONFIG.SYS and edited to enable the desired components (SPX, named pipe support, NetBIOS).

The user also has to pick the right driver. Novell provided a good selection of drivers, for Ethernet (Novell NE1000, NE2000, 3Com 3C501, 3C503, 3C505) as well as IBM PC Network, Token Ring, and Novell RX Net.

Continue reading →
Posted in NetWare, Networking, OS/2, PC history, Virtualization | 11 Comments

Learn Something Old Every Day, Part V: Early IBM PS/2 Hard Disks

Posted on May 1, 2022 by Michal Necasek

So I have been (again) trying to properly archive old MS OS/2 SDKs. The version 1.02 SDK from December 1987 (corresponding to OS/2 1.0) turned out to be a bit of a poser.

The SDK came on both 3.5″ and 5.25″ media (unlike the older 1.00 and 1.01 SDKs, which as far as I can tell only existed on 5.25″ media). My Kryoflux tells me that the Install floppies were modified in both the 3.5″ and 5.25″ disk sets. Now, it’s very common for OS/2 install disks to be modified, because the installer writes an INSTALL.LOG file. But I could find no evidence of that happening.

Instead, it looked like Microsoft made some changes after mastering the disks, but exactly what those changes were is difficult to tell.

On the 5.25″ Install disk, Microsoft almost certainly updated the INSTAID.EXE file (the main installer executable). The allocation chain for INSTAID.EXE is not contiguous, which is a strong hint that it had been overwritten with a larger file. Yet the timestamp for INSTAID.EXE is the same as the other files (12/15/1987) and the entry for INSTAID.EXE is in the middle of the root directory, alphabetically sorted.

Okay, so that makes some sense. Microsoft made some last-minute installer change and had to update the already mastered disks. One might expect to find the same sort of change on both 3.5″ and 5.25″ media… but no.

Continue reading →
Posted in ESDI, IBM, PC history, PS/2 | 16 Comments

OS/2 6.304: Finally Complete

Posted on April 29, 2022 by Michal Necasek

Some time ago, I lamented that even though the OS/2 Museum had a good number of disks of the final OS/2 2.0 beta, level 6.304, there wasn’t enough to install the OS, let alone any of the development tools or add-ons (Extended Services, LAN Server).

And now, more or less exactly 30 years after build 6.304 was released (February 1992), an IBM DAP (Developer Assistance Program) CD-ROM turned up with the complete 6.304 disk set, plus Developer’s Toolkit, C Set/2 compiler, Extended Services, LAN Server 2.0 Entry/Advanced, documentation, and a couple of other goodies. IBM also sometimes referred to OS/2 level 6.304 as EEP, or Early Experience Program, and provided a helpful “Product Considerations” booklet.

There was also one surprise: It was possible to install this pre-release of OS/2 2.0 directly from CD-ROM (that was also possible with the Limited Availiability Level 6.177 CD-ROM, but with completely different storage drivers).

As far as I know, OS/2 2.0 was never available on CD-ROM, although IBM shipped OS/2 on CD-ROM with certain PS/2 Ultimedia models. It was only with OS/2 2.1 (1993) that IBM started offering OS/2 on CD-ROM, thought by the time Warp 4 was released (1996), OS/2 was available on CD-ROM only.

When OS/2 6.304 came out in early 1992, CD-ROMs were just starting to ramp up. The trouble was that the market was rather fragmented. There were SCSI CD-ROMs (and IBM sold those), but there were also several proprietary interfaces from Philips, Sony, and others. Even SCSI CD-ROMs weren’t all that standardized before the SCSI MMC (Multi-Media Commands) specification appeared.

The upshot was that even though OS/2 6.304 included CD-ROM support, it was far from certain that a user equipped with a CD-ROM would be able to use it on a non-IBM machine. Even with a “standard” SCSI CD-ROM, the user might need to supply a driver for the SCSI HBA (IBM only shipped drivers for the most common Adaptec 154x/174x HBAs with OS/2, plus a couple of Future Domain models), and that wasn’t enough either; a separate CD-ROM support driver was required, and CDROM.SYS shipped with OS/2 only supported IBM and Toshiba SCSI CD-ROMs.

Continue reading →
Posted in CD-ROM, IBM, OS/2, PC history, Pre-release | 5 Comments

Page Too Big

Posted on April 22, 2022 by Michal Necasek

The other day I was able to look at an IBM OS/2 pre-release CD-ROM from early 1992. The CD-ROM appears to have been produced by IBM UK under the DAP (Developer Assistance Program) umbrella.

The CD-ROM contains about 250 MB of data and would have been the perfect medium for distributing the OS/2 pre-releases. The alternatives were either a huge pile of floppies or many days of downloading over a 9,600 bps modem (and filling up one or two typical hard disks in the process).

On the CD-ROM there’s not just the base OS but also a matching pre-release of LAN Server, Extended Services, Toolkit, and C Set/2 compiler. In addition, there’s all product documentation in electronic format—an alternative to many pounds of manuals.

I should add that IBM did ship these pre-releases on floppies, as seen e.g. here. But the full set would have been about 70 floppies, with no documentation. A single CD-ROM would have been vastly cheaper to produce and mail. As for printed documentation, I’m not sure if IBM even provided any; printing reams of manuals that were going to be obsolete in a month or two probably made little sense.

Obviously IBM didn’t ship the documentation as PDFs, because PDF wasn’t available until 1993. IBM also didn’t ship the documentation in PostScript format (Microsoft used that for NT pre-releases), perhaps because not everyone had a PostScript capable printer, but also because IBM already had a format which solved much the same problems as PDF, and it was called AFP.

The documentation was shipped as ZIP archives containing compressed .LIS files. The .LIS files were in LIST3820 format, a form of AFP. Also included on the CD was an IBM “internal use only” program called LP3820 by Ken Borgendale. The LP3820 utility ran under DOS or OS/2, took the AFP .LIS files as input, and printed them on HP LaserJet or PostScript printers, but could also produce plain ASCII files.

Continue reading →
Posted in Documentation, IBM, OS/2, PC history | 4 Comments

Xenix 2.2 vs. VGA

Posted on April 15, 2022 by Michal Necasek

The other day I started wondering why certain old versions of 286 and 386 XENIX look a bit weird in emulation:

XENIX 2.2 on a VGA, unable to reprogram fonts

The characters are cut off, because XENIX sets up an EGA text mode with 8×14 character matrix but uses 8×16 VGA fonts.

Investigation showed that the XENIX console driver rather closely follows the inner workings of the EGA BIOS, no doubt because finding out exactly what IBM’s EGA BIOS did was not difficult.

Specifically, text modes normally access memory in EGA/VGA plane 0 where character codes and attributes are stored. Fonts are stored in plane 2 and the video hardware needs to be reprogrammed to let the host CPU access that plane. For whatever reason, the EGA BIOS uses a rather heavy-handed but very straightforward method: It defines two “fake” modes in its internal mode table, namely mode entry 0Bh for color and 0Ch for mono modes. The EGA BIOS simply internally sets mode 0Bh or 0Ch, uploads the font bitmaps, and then sets the real desired mode.

XENIX 2.2 uses the exact same method. If the BIOS does not fill out mode table entries 0Bh/0Ch, XENIX will be unable to program the fonts and the original VGA 8×16 font will remain in place, but there won’t be other ill effects.

So obviously the BIOS could just supply the EGA compatible mode table entries 0Bh/0Ch. And indeed that helps with some XENIX versions, such as 2.2.3c, which now gets the right 8×14 font:

XENIX 2.2.3c properly reprogramming fonts

But oops! With other XENIX versions, such as 2.2.3b, the result is much, much worse:

XENIX 2.2.3b with garbled fonts

Why would that be? Finding out requires a little more digging…

Continue reading →
Posted in 286, 386, PC architecture, PC history, VGA, Xenix | 14 Comments