The PC floppy subsystem

Posted on April 5, 2011 by Michal Necasek

The PC floppy subsystem, ubiquitous and indispensable until the early 21st century, suffered the typical fate of many “legacy” subsystems: The initial design was adequate, but did not adapt to newer and more complex hardware.

With the original IBM PC, things were simple. The only choices were single- or double-sided 5¼” double-density drive (the single-sided drives were soon obsoleted) and how many. The number of disks was configured via switches, and the number of sides was something the operating system could easily take care of when formatting a new disk. The NEC μPD765A floppy controller was used, setting the standard for years to come. Continue reading →

Posted in PC architecture | Leave a comment

HIMEM.SYS, unreal mode, and LOADALL

Posted on March 18, 2011 by Michal Necasek

The previous post talked about real mode on 286+ processors which behaves more like a slightly modified variant of protected mode rather than the old 8088/8086 processors. Real mode with non-compatible selector bases or limits is usually called unreal mode or big real mode. Even though Intel never clearly documented the behavior in detail, it is more or less set in stone thanks to our dear friend, backwards compatibility.

That’s because unreal mode had some rather prominent users—one of them was Microsoft’s HIMEM.SYS, used on nearly every PC in the 1990s. While disassembling HIMEM.SYS might be instructive, it would be a waste of time in this case, since Microsoft published the full source code to several HIMEM.SYS 2.x versions. That way, the evolution of the code can be easily traced. Continue reading →

Posted in DOS, x86 | 6 Comments

Will the real Real Mode please stand up?

Posted on March 11, 2011 by Michal Necasek

Every programmer familiar with the x86 architecture understands the difference between real and protected address mode of the processor. It is well known that the real mode is compatible with Intel’s old 16-bit 8088/8086 CPUs, while protected mode was a new feature introduced in the Intel 80286 and extended to 32-bit in the 80386. The 80386 further muddled things by introducing the Virtual-8086 mode (V86 mode) that was something of both.

Not every programmer knows exactly what the difference is between real and protected mode on processors that support both. Intel’s documentation, frankly, does its best to obscure and confuse the issue (a situation not unfamiliar to readers of Intel’s processor documentation) without actually lying outright. Intel suggests that real mode implies 16-bit segments with 64KB limits, 8086 style addressing, limited instruction set, no paging, no privilege levels, and no memory protection. Continue reading →

Posted in x86 | 6 Comments

Display Drivers, NT and NeXTSTEP

Posted on January 21, 2011 by Michal Necasek

It is instructive to compare the OS/2 and 16-bit Windows display driver model with other operating systems. Why NT and NeXTSTEP? NT because it was Microsoft’s third take on a (mostly) PC operating system, and NeXTSTEP because it was an OS with a very different heritage, ported to the PC relatively late in its life, and eventually spawning Mac OS X.

The general NT driver model is unusually complex and over-engineered. However, NT display drivers use a significantly different and much simpler model. Drivers are split into two separate components, a video miniport and the display driver proper. The video miniport always runs in kernel context, while the display driver is a DLL which runs in user or kernel context, depending on the NT version.

The miniport is primarily responsible for hardware initialization and setting modes. If applicable, it also manages the hardware cursor and DAC palettes. The display driver implements accelerated drawing operations and corresponds to the old GDI drivers in its responsibilities, but not in implementation. Both the miniport and the display driver are typically written in C, and assembly code is almost unheard of. Continue reading →

Posted in NeXTSTEP, NT | 2 Comments

Display Drivers, OS/2 and 16-bit Windows

Posted on January 19, 2011 by Michal Necasek

Not surprisingly, the display driver model of Windows 1.x/2.x/3.x and OS/2 1.x/2.x was quite similar. This was in sharp contrast to the drivers for just about every other device; while disks or network adapters already had existing drivers in DOS, the drivers for Windows GDI had to be written from scratch. When the Presentation Manager was created, Microsoft recycled much of the existing Windows driver model.

Both in OS/2 and Windows, displays and printers both used the same basic driver model. The driver model was designed with a C-callable interface, but traditionally only printer drivers were written in C, while display drivers were written in assembler.

Viewed from a modern perspective, the Windows and OS/2 driver model was byzantine, overly complex, with drivers being difficult to write and nearly impossible to fully debug. The drivers could implement nearly the entire graphics engine, and even the most basic driver had to implement significant functionality. This included rendering memory bitmaps (device contexts), which seems bizarre in hindsight. Continue reading →

Posted in OS/2, Windows | Leave a comment

Windows Presentation Manager

Posted on January 17, 2011 by Michal Necasek

Any programmer even passingly familiar with the Windows and Presentation Manager APIs is aware that there are significant similarities. When other windowing APIs, such as X11 or Mac OS, are taken into consideration, Windows and the Presentation Manager look like siblings, almost twins.

The similarity is not coincidental. The OS/2 GUI was even referred to as the “Windows Presentation Manager” in the early days of OS/2. The OS/2 Presentation Manager and Windows 2.0 were developed in parallel, and Microsoft was stressing the similarities.

The exact timeline is not easy to establish, but a few points are clear. The Presentation Manager was announced in April 1987; at that time, the functional specification already existed, but the code was very far from being finished. In mid-1987, when Microsoft started shipping the first OS/2 SDKs, the complete Presentation Manager API documentation was part of the kits, even though there was no code yet. To get developers started, the OS/2 SDK included Windows 1.04 and the corresponding SDK. Continue reading →

Posted in OS/2, Windows | 14 Comments

Installing OS/2 Warp in a Virtual Machine

Posted on November 3, 2010 by Michal Necasek

Installing OS/2 Warp in a virtual machine isn’t difficult, but there are a few pitfalls. The issues are easy to work around, but sometimes quite difficult to diagnose, especially for users not intimately familiar with OS/2. In general, the problems are the same one might encounter when installing OS/2 on a physical system.

The first issue is that OS/2 Warp (and older 2.1 and 2.11 releases) may hang immediately when the system starts booting, even before displaying the “boot blob”.  This problem was tracked by IBM as APAR PJ23092 (Faster Intel Machines fail to boot). IBM made updated OS2LDR files available in 1998, but those only reduce the incidence of the problem and do not fully fix it. At any rate, re-trying the boot usually helps.

The next problem is disk size. Using a disk larger than 2GB is not recommended unless the boot floppies are updated with newer disk drivers. However, a larger disk should not be necessary. For plain OS/2 Warp, there should be no other showstoppers—but read on. Continue reading →

Posted in OS/2, VirtualBox, Virtualization | 46 Comments

MS OS/2 Patents

Posted on October 18, 2010 by Michal Necasek

When Microsoft OS/2 1.3 boots up, it prints a prominent message “U.S. Patent No. 4779187; 4825358”. What are those patents? As it turns out, it’s really just one invention, which for reasons perhaps clear to a patent lawyer was essentially assigned two patent numbers.

The application for patent 4779287 was filed on April 10, 1985 by James Letwin of Kirkland, WA. James Letwin is better known under his middle name, Gordon. Gordon Letwin was Microsoft’s chief architect of OS/2, although in April 1985 no one had heard of OS/2 and the entire project was in very early planning stages. The patent was granted on October 18, 1988, around the time OS/2 1.1 was released.

Patent 4825358 was applied for on June 18, 1987. At that time, OS/2 1.0 had not yet been released but the project had a very firm outline, with the first betas already shipped. This later patent was granted on April 25th, 1989, several months before OS/2 1.2 was released. Patent 4825358 contains the verbatim text and all drawings of patent 4825358, with several additional claims. Continue reading →

Posted in OS/2 | 3 Comments

Installing Windows NT 3.x in VirtualBox

Posted on October 12, 2010 by Michal Necasek

All Windows NT 3.x versions can be used in VirtualBox, although they are not officially supported (unlike all following NT versions) and no Guest Additions are provided. Furthermore, for Windows NT 3.1 and 3.5, an extra step needs to be taken before installation.

In all cases, it’s best to start with a guest configured for NT 4. It is advisable to add a floppy drive and, if audio is desired, use SoundBlaster 16 emulation. USB may be turned off completely since there are no drivers, and the emulated network adapter must be left at AMD PCnet. SMP is not supported.

Unlike NT 4, the Windows NT 3.x installation CDs are not bootable. It may be easiest to start by installing DOS with CD-ROM drivers and running the DOS based installer from the CD. For NT 3.1 that is the only option due to lack of ATAPI CD-ROM drivers. For NT 3.5x it’s possible to boot from floppy images and continue installing from CD. Continue reading →

Posted in NT, VirtualBox, Virtualization | 22 Comments

Why OS/2 is Hard to Virtualize

Posted on August 5, 2010 by Michal Necasek

For a number of reasons, OS/2 is surprisingly difficult to virtualize, much more so than DOS, Windows 3.x, Windows NT, Windows 95, or many variants of UNIX. Most of these reasons have to do with OS/2’s architecture and history.

To begin with, old OS/2 versions (especially the versions older than Warp 4) are difficult to run even on physical Pentium III class hardware. There are various software timing loops which simply do not work on fast processors and cause division overflows and similar errors. This problem especially badly affects almost all versions of OS/2 1.x. Support for large IDE disks tends to be problematic, although that is not something to blame on OS/2 directly; the IDE interface was a moving target. Here virtualization in fact helps, because virtualizing, say, a 120 MB IDE disk is much easier than finding a real functioning disk with such a tiny capacity.

Then there are numerous problems specific to virtualization. Let’s start with the oldest OS/2 versions. IBM OS/2 1.0 and 1.1 uses 286-style switching from protected to real mode. This method resets the CPU and relies on the BIOS POST code to resume execution at a specified point; the reset is usually accomplished by triple faulting the CPU. Neither VMware, Virtual PC, nor VirtualBox support this method and treat the triple fault as a fatal error. Interestingly, the corresponding Microsoft OS/2 versions detect a 386 CPU and use the much more efficient method of modifying the CR0 control register. Continue reading →

Posted in OS/2, Virtualization | 6 Comments