CP/M

Gary Kildall originally developed CP/M during 1974,[5][6] as an operating system to run on an Intel Intellec-8 development system, equipped with a Shugart Associates 8-inch floppy-disk drive interfaced via a custom floppy-disk controller.[12] It was written in Kildall's own PL/M (Programming Language for Microcomputers).[13] Various aspects of CP/M were influenced by the TOPS-10 operating system of the DECsystem-10 mainframe computer, which Kildall had used as a development environment.[14][15][16] An early outside licensee of CP/M was Gnat Computers, an early microcomputer developer out of San Diego, California. In 1977, the company was granted the license to use CP/M 1.0 for any micro they desired for $90. Within the year, demand for CP/M was so high that Digital Research was able to increase the license to tens of thousands of dollars.[17]

Under Kildall's direction, the development of CP/M 2.0 was mostly carried out by John Pierce in 1978. Kathryn Strutynski, a friend of Kildall from Naval Postgraduate School (NPS), became the fourth employee of Digital Research Inc. in early 1979. She started by debugging CP/M 2.0, and later became influential as key developer for CP/M 2.2 and CP/M Plus. Other early developers of the CP/M base included Robert "Bob" Silberstein and David "Dave" K. Brown.[18][19]

CP/M originally stood for "Control Program/Monitor",[3] a name which implies a resident monitor—a primitive precursor to the operating system. However, during the conversion of CP/M to a commercial product, trademark registration documents filed in November 1977 gave the product's name as "Control Program for Microcomputers".[6] The CP/M name follows a prevailing naming scheme of the time, as in Kildall's PL/M language, and Prime Computer's PL/P (Programming Language for Prime), both suggesting IBM's PL/I; and IBM's CP/CMS operating system, which Kildall had used when working at the NPS. This renaming of CP/M was part of a larger effort by Kildall and his wife with business partner, Dorothy McEwen[4] to convert Kildall's personal project of CP/M and the Intel-contracted PL/M compiler into a commercial enterprise. The Kildalls intended to establish the Digital Research brand and its product lines as synonymous with "microcomputer" in the consumer's mind, similar to what IBM and Microsoft together later successfully accomplished in making "personal computer" synonymous with their product offerings. Intergalactic Digital Research, Inc. was later renamed via a corporation change-of-name filing to Digital Research, Inc.[4]

Apple CP/M Card with manual

By September 1981, Digital Research had sold more than 260,000 CP/M licenses; InfoWorld stated that the actual market was likely larger because of sublicenses. Many different companies produced CP/M-based computers for many different markets; the magazine stated that "CP/M is well on its way to establishing itself as the small-computer operating system".[20] The companies chose to support CP/M because of its large library of software. The Xerox 820 ran the operating system because "where there are literally thousands of programs written for it, it would be unwise not to take advantage of it", Xerox said.[21] (Xerox included a Howard W. Sams CP/M manual as compensation for Digital Research's documentation, which InfoWorld in 1982 described as atrocious.[22]) By 1984, Columbia University used the same source code to build Kermit binaries for more than a dozen different CP/M systems, plus a generic version.[23] The operating system was described as a "software bus",[24][25] allowing multiple programs to interact with different hardware in a standardized way.[26] Programs written for CP/M were typically portable among different machines, usually requiring only the specification of the escape sequences for control of the screen and printer. This portability made CP/M popular, and much more software was written for CP/M than for operating systems that ran on only one brand of hardware. One restriction on portability was that certain programs used the extended instruction set of the Z80 processor and would not operate on an 8080 or 8085 processor. Another was graphics routines, especially in games and graphics programs, which were generally machine-specific as they used direct hardware access for speed, bypassing the OS and BIOS (this was also a common problem in early DOS machines).

Bill Gates claimed that the Apple II series with a Z-80 SoftCard was the single most-popular CP/M hardware platform.[27] Many different brands of machines ran the operating system, some notable examples being the Altair 8800, the IMSAI 8080, the Osborne 1 and Kaypro luggables, and MSX computers. The best-selling CP/M-capable system of all time was probably the Amstrad PCW. In the UK, CP/M was also available on Research Machines educational computers (with the CP/M source code published as an educational resource), and for the BBC Micro when equipped with a Z80 co-processor. Furthermore, it was available for the Amstrad CPC series, the Commodore 128, TRS-80, and later models of the ZX Spectrum. CP/M 3 was also used on the NIAT, a custom handheld computer designed for A. C. Nielsen's internal use with 1 MB of SSD memory.

In 1979, a multi-user compatible derivative of CP/M was released. MP/M allowed multiple users to connect to a single computer, using multiple terminals to provide each user with a screen and keyboard. Later versions ran on 16-bit processors.

CP/M Plus (CP/M 3) System Guide

The last 8-bit version of CP/M was version 3, often called CP/M Plus, released in 1983.[18] Its BDOS was designed by Brown.[18] It incorporated the bank switching memory management of MP/M in a single-user single-task operating system compatible with CP/M 2.2 applications. CP/M 3 could therefore use more than 64 KB of memory on an 8080 or Z80 processor. The system could be configured to support date stamping of files.[18] The operating system distribution software also included a relocating assembler and linker.[2] CP/M 3 was available for the last generation of 8-bit computers, notably the Amstrad PCW, the Amstrad CPC, the ZX Spectrum +3, the Commodore 128, MSX machines and the Radio Shack TRS-80 Model 4.[28]

DEC PRO-CP/M-80 floppy-disk distribution for the Z80-A co-processor in a DEC Professional 3xx series

There were versions of CP/M for some 16-bit CPUs as well.

The first version in the 16-bit family was CP/M-86 for the Intel 8086 in November 1981.[29] Kathryn Strutynski was the project manager for the evolving CP/M-86 line of operating systems.[18][19] At this point, the original 8-bit CP/M became known by the retronym CP/M-80 to avoid confusion.[29]

CP/M-86 was expected to be the standard operating system of the new IBM PCs, but DRI and IBM were unable to negotiate development and licensing terms. IBM turned to Microsoft instead, and Microsoft delivered PC DOS based on 86-DOS. Although CP/M-86 became an option for the IBM PC after DRI threatened legal action, it never overtook Microsoft's system. Most customers were repelled by the significantly greater price IBM charged for CP/M-86 over PC DOS (US$240 and US$40, respectively).[30]

When Digital Equipment Corporation (DEC) put out the Rainbow 100 to compete with IBM, it came with CP/M-80 using a Z80 chip, CP/M-86 or MS-DOS using an 8088 microprocessor, or CP/M-86/80 using both. The Z80 and 8088 CPUs ran concurrently.[31][32] A benefit of the Rainbow was that it could continue to run 8-bit CP/M software, preserving a user's possibly sizable investment as they moved into the 16-bit world of MS-DOS.[31] A similar dual-processor adaption for the CompuPro System 816 was named CP/M 8-16. The CP/M-86 adaptation for the 8085/8088-based Zenith Z-100 also supported running programs for both of its CPUs.

Soon following CP/M-86, another 16-bit version of CP/M was CP/M-68K for the Motorola 68000. The original version of CP/M-68K in 1982 was written in Pascal/MT+68k, but it was ported to C later on. CP/M-68K, already running on the Motorola EXORmacs systems, was initially to be used in the Atari ST computer, but Atari decided to go with a newer disk operating system called GEMDOS. CP/M-68K was also used on the SORD M68 and M68MX computers.[33]

In 1982, there was also a port from CP/M-68K to the 16-bit Zilog Z8000 for the Olivetti M20, written in C, named CP/M-8000.[34][35]

These 16-bit versions of CP/M required application programs to be re-compiled for the new CPUs. Some programs written in assembly language could be automatically translated for a new processor. One tool for this was Digital Research's XLT86, which translated .ASM source code for the Intel 8080 processor into .A86 source code for the Intel 8086. The translator would also optimize the output for code size and take care of calling conventions, so that CP/M-80 and MP/M-80 programs could be ported to the CP/M-86 and MP/M-86 platforms automatically. XLT86 itself was written in PL/I-80 and was available for CP/M-80 platforms as well as for VAX/VMS.[36]

Many expected that CP/M would be the standard operating system for 16-bit computers.[37] In 1980 IBM approached Digital Research, at Bill Gates' suggestion,[38] to license a forthcoming version of CP/M for its new product, the IBM Personal Computer. Upon the failure to obtain a signed non-disclosure agreement, the talks failed, and IBM instead contracted with Microsoft to provide an operating system.[39] The resulting product, MS-DOS, soon began outselling CP/M.

Many of the basic concepts and mechanisms of early versions of MS-DOS resembled those of CP/M. Internals like file-handling data structures were identical, and both referred to disk drives with a letter (A:, B:, etc.). MS-DOS's main innovation was its FAT file system. This similarity made it easier to port popular CP/M software like WordStar and dBase. However, CP/M's concept of separate user areas for files on the same disk was never ported to MS-DOS. Since MS-DOS had access to more memory (as few IBM PCs were sold with less than 64 KB of memory, while CP/M could run in 16 KB if necessary), more commands were built into the command-line shell, making MS-DOS somewhat faster and easier to use on floppy-based computers.

Although one of the first peripherals for the IBM PC was a SoftCard-like expansion card that let it run 8-bit CP/M software,[40] InfoWorld stated in 1984 that efforts to introduce CP/M to the home market had been largely unsuccessful and most CP/M software was too expensive for home users.[41] In 1986 the magazine stated that Kaypro had stopped production of 8-bit CP/M-based models to concentrate on sales of MS-DOS compatible systems, long after most other vendors had ceased production of new equipment and software for CP/M.[42] CP/M rapidly lost market share as the microcomputing market moved to the IBM-compatible platform, and it never regained its former popularity. Byte magazine, at the time one of the leading industry magazines for microcomputers, essentially ceased covering CP/M products within a few years of the introduction of the IBM PC. For example, in 1983 there were still a few advertisements for S-100 boards and articles on CP/M software, but by 1987 these were no longer found in the magazine.

Later versions of CP/M-86 made significant strides in performance and usability and were made compatible with MS-DOS. To reflect this compatibility the name was changed, and CP/M-86 became DOS Plus, which in turn became DR-DOS.

ZCPR[43] (the Z80 Command Processor Replacement) was introduced on 2 February 1982 as a drop-in replacement for the standard Digital Research console command processor (CCP) and was initially written by a group of computer hobbyists who called themselves "The CCP Group". They were Frank Wancho, Keith Petersen (the archivist behind Simtel at the time), Ron Fowler, Charlie Strom, Bob Mathias, and Richard Conn. Richard was, in fact, the driving force in this group (all of whom maintained contact through email).

ZCPR1 was released on a disk put out by SIG/M (Special Interest Group/Microcomputers), a part of the Amateur Computer Club of New Jersey.

ZCPR2 was released on 14 February 1983. It was released as a set of ten disks from SIG/M. ZCPR2 was upgraded to 2.3, and also was released in 8080 code, permitting the use of ZCPR2 on 8080 and 8085 systems.

ZCPR3[44] was released on 14 July 1984, as a set of nine disks from SIG/M. The code for ZCPR3 could also be compiled (with reduced features) for the 8080 and would run on systems that did not have the requisite Z80 microprocessor. Features of ZCPR as of version 3 included shells, aliases, I/O redirection, flow control, named directories, search paths, custom menus, passwords, and online help. In January 1987, Richard Conn stopped developing ZCPR, and Echelon asked Jay Sage (who already had a privately enhanced ZCPR 3.1) to continue work on it. Thus, ZCPR 3.3 was developed and released. ZCPR 3.3 no longer supported the 8080 series of microprocessors, and added the most features of any upgrade in the ZCPR line. ZCPR 3.3 also included a full complement of utilities with considerably extended capabilities. While enthusiastically supported by the CP/M user base of the time, ZCPR alone was insufficient to slow the demise of CP/M.

Screenshot showing a CP/M 3.0 directory listing using the DIR command on a Commodore 128 home computer

The Console Command Processor, or CCP, accepted input from the keyboard and conveyed results to the terminal. CP/M itself would work with either a printing terminal or a video terminal. All CP/M commands had to be typed in on the command line. The console would most often display the A> prompt, to indicate the current default disk drive. When used with a video terminal, this would usually be followed by a blinking cursor supplied by the terminal. The CCP would await input from the user. A CCP internal command, of the form drive letter followed by a colon, could be used to select the default drive. For example, typing B: and pressing enter at the command prompt would change the default drive to B, and the command prompt would then become B> to indicate this change.

CP/M's command-line interface was patterned after the operating systems from Digital Equipment, such as RT-11 for the PDP-11 and OS/8 for the PDP-8. Commands took the form of a keyword followed by a list of parameters separated by spaces or special characters. Similar to a Unix shell builtin, if an internal command was recognized, it was carried out by the CCP itself. Otherwise it would attempt to find an executable file on the currently logged disk drive and (in later versions) user area, load it, and pass it any additional parameters from the command line. These were referred to as "transient" programs. On completion, CP/M would reload the part of the CCP that had been overwritten by application programs — this allowed transient programs a larger memory space.

The commands themselves could sometimes be obscure. For instance, the command to duplicate files was named PIP (Peripheral-Interchange-Program), the name of the old DEC utility used for that purpose. The format of parameters given to a program was not standardized, so that there was no single option character that differentiated options from file names. Different programs could and did use different characters.

The CP/M Console Command Processor includes DIR, ERA, REN, SAVE, TYPE, and USER as built-in commands.[51] Transient commands in CP/M include ASM, DDT, DUMP, ED, LOAD, MOVCPM, PIP, STAT, SUBMIT, and SYSGEN.[51]

CP/M Plus (CP/M Version 3) includes DIR (display list of files from a directory except those marked with the SYS attribute), DIRSYS / DIRS (list files marked with the SYS attribute in the directory), ERASE / ERA (delete a file), RENAME / REN (rename a file), TYPE / TYP (display contents of an ASCII character file), and USER / USE (change user number) as built-in commands:[52] CP/M 3 allows the user to abbreviate the built-in commands.[53] Transient commands in CP/M 3 include COPYSYS, DATE, DEVICE, DUMP, ED, GET, HELP, HEXCOM, INITDIR, LINK, MAC, PIP, PUT, RMAC, SET, SETDEF, SHOW, SID, SUBMIT, and XREF.[53]

The Basic Disk Operating System,[13][12] or BDOS,[13][12] provided access to such operations as opening a file, output to the console, or printing. Application programs would load processor registers with a function code for the operation, and addresses for parameters or memory buffers, and call a fixed address in memory. Since the address was the same independent of the amount of memory in the system, application programs would run the same way for any type or configuration of hardware.

CP/M advertisement in the 11 December 1978 issue of InfoWorld magazine

The Basic Input Output System or BIOS,[13][12] provided the lowest level functions required by the operating system.

These included reading or writing single characters to the system console and reading or writing a sector of data from the disk. The BDOS handled some of the buffering of data from the diskette, but before CP/M 3.0 it assumed a disk sector size fixed at 128 bytes, as used on single-density 8-inch floppy disks. Since most 5.25-inch disk formats used larger sectors, the blocking and deblocking and the management of a disk buffer area was handled by model-specific code in the BIOS.

Customization was required because hardware choices were not constrained by compatibility with any one popular standard. For example, some manufacturers used a separate computer terminal, while others designed a built-in integrated video display system. Serial ports for printers and modems could use different types of UART chips, and port addresses were not fixed. Some machines used memory-mapped I/O instead of the 8080 I/O address space. All of these variations in the hardware were concealed from other modules of the system by use of the BIOS, which used standard entry points for the services required to run CP/M such as character I/O or accessing a disk block. Since support for serial communication to a modem was very rudimentary in the BIOS or may have been absent altogether, it was common practice for CP/M programs that used modems to have a user-installed overlay containing all the code required to access a particular machine's serial port.

The read/write memory between address 0100 hexadecimal and the lowest address of the BDOS was the Transient Program Area (TPA) available for CP/M application programs. Although all Z80 and 8080 processors could address 64 kilobytes of memory, the amount available for application programs could vary, depending on the design of the particular computer. Some computers used large parts of the address space for such things as BIOS ROMs, or video display memory. As a result, some systems had more TPA memory available than others. Bank switching was a common technique that allowed systems to have a large TPA while switching out ROM or video memory space as needed. CP/M 3.0 allowed parts of the BDOS to be in bank-switched memory as well.

CP/M came with a Dynamic Debugging Tool, nicknamed DDT (after the insecticide, i.e. a bug-killer), which allowed memory and program modules to be examined and manipulated, and allowed a program to be executed one step at a time.[60][61][62]

CP/M originally did not support the equivalent of terminate and stay resident (TSR) programs as under DOS. Programmers could write software that could intercept certain operating system calls and extend or alter their functionality. Using this capability, programmers developed and sold auxiliary desk accessory programs, such as SmartKey, a keyboard utility to assign any string of bytes to any key.[63] CP/M 3, however, added support for dynamically loadable Resident System Extensions (RSX).[52][18] A so-called null command file could be used to allow CCP to load an RSX without a transient program.[52][18] Similar solutions like RSMs (for Resident System Modules) were also retrofitted to CP/M 2.2 systems by third-parties.[64][65][66]

Although CP/M provided some hardware abstraction to standardize the interface to disk I/O or console I/O, application programs still typically required installation to make use of all the features of such equipment as printers and terminals. Often these were controlled by escape sequences which had to be altered for different devices. For example, the escape sequence to select bold face on a printer would have differed among manufacturers, and sometimes among models within a manufacturer's range. This procedure was not defined by the operating system; a user would typically run an installation program that would either allow selection from a range of devices, or else allow feature-by-feature editing of the escape sequences required to access a function. This had to be repeated for each application program, since there was no central operating system service provided for these devices.

The initialization codes for each model of printer had to be written into the application. To use a program such as Wordstar with more than one printer (say, a fast dot-matrix printer or a slower but presentation-quality daisy wheel printer), a separate version of Wordstar had to be prepared, and one had to load the Wordstar version that corresponded to the printer selected (and exiting and reloading to change printers).

File names were specified as a string of up to eight characters, followed by a period, followed by a file name extension of up to three characters ("8.3" filename format). The extension usually identified the type of the file. For example, .COM indicated an executable program file, and .TXT indicated a file containing ASCII text. Characters in filenames entered at the command prompt were converted to upper case, but this was not enforced by the operating system. Programs (MBASIC is a notable example) were able to create filenames containing lower-case letters, which then could not easily be referenced at the command line.

Each disk drive was identified by a drive letter, for example, drive A and drive B. To refer to a file on a specific drive, the drive letter was prefixed to the file name, separated by a colon, e.g., A:FILE.TXT. With no drive letter prefixed, access was to files on the current default drive.[76]

File size was specified as the number of 128-byte records (directly corresponding to disk sectors on 8-inch drives) occupied by a file on the disk. There was no generally supported way of specifying byte-exact file sizes. The current size of a file was maintained in the file's File Control Block (FCB) by the operating system. Since many application programs (such as text editors) prefer to deal with files as sequences of characters rather than as sequences of records, by convention text files were terminated with a control-Z character (ASCII SUB, hexadecimal 1A). Determining the end of a text file therefore involved examining the last record of the file to locate the terminating control-Z. This also meant that inserting a control-Z character into the middle of a file usually had the effect of truncating the text contents of the file.

With the advent of larger removable and fixed disk drives, disk de-blocking formulas were employed which resulted in more disk blocks per logical file allocation block. While this allowed for larger file sizes, it also meant that the smallest file which could be allocated increased in size from 1 KB (on single-density drives) to 2 KB (on double-density drives) and so on, up to 32 KB for a file containing only a single byte. This made for inefficient use of disk space if the disk contained a large number of small files.

File modification time stamps were not supported in releases up to CP/M 2.2, but were an optional feature in MP/M and CP/M 3.0.[18]

CP/M 2.2 had no subdirectories in the file structure, but provided 16 numbered user areas to organize files on a disk. To change user one had to simply type "User X" at the command prompt, X being the user number. Security was non-existent and considered unnecessary on a personal computer. The user area concept was to make the single-user version of CP/M somewhat compatible with multi-user MP/M systems. A common patch for the CP/M and derivative operating systems was to make one user area accessible to the user independent of the currently set user area. A USER command allowed the user area to be changed to any area from 0 to 15. User 0 was the default. If one changed to another user, such as USER 1, the material saved on the disk for this user would only be available to USER 1; USER 2 would not be able to see it or access it. However, files stored in the USER 0 area were accessible to all other users; their location was specified with a prefatory path, since the files of USER 0 were only visible to someone logged in as USER 0. The user area feature arguably had little utility on small floppy disks, but it was useful for organizing files on machines with hard drives. The intent of the feature was to ease use of the same computer for different tasks. For example, a secretary could do data entry, then, after switching USER areas, another employee could use the machine to do billing without their files intermixing.

Some companies made official enhancements of CP/M based on Digital Research source code. An example is IMDOS for the IMSAI 8080 computer made by IMS Associates, Inc., a clone of the famous Altair 8800.

Other CP/M compatible OSes were developed independently and made no use of Digital Research code. Some contemporary examples were:

• Cromemco CDOS from Cromemco
• MSX-DOS for the MSX range of computers is CP/M-compatible and can run CP/M programs.
• The Epson QX-10 shipped with a choice of CP/M or the compatible TPM-II or TPM-III.
• The British ZX Spectrum compatible SAM Coupé had an optional CP/M-2.2 compatible OS called Pro-DOS.
• The Amstrad/Schneider CPC series 6xx (disk-based) and PCW series computers were bundled with an CP/M disk pack.
• The Husky (computer) ran a ROM-based menu-driven program loader called DEMOS which could run many CP/M applications.
• ZSDOS is a replacement BDOS for CP/M-80 2.2 written by Harold F. Bower and Cameron W. Cotrill.
• CPMish is a new FOSS CP/M 2.2-compatible operating system which originally contained no DR code. It includes ZSDOS as its BDOS and ZCPR (see earlier) as the command processor. Since Bryan Sparks, the president of DR owners Lineo, granted permission in 2022 to modify and redistribute CP/M code, developer David Given is updating CPMish with some parts of the original DR CP/M.
• LokiOS is a CP/M 2.2 compatible OS. Version 0.9 was publicly released in 2023 by David Kitson as a solo-written Operating System exercise, intended for the Open Spectrum Project and includes source code for the BIOS, BDOS and Command-line_interface as well as other supporting applications and drivers. The distribution also includes original DR Source code and a utility to allow users to hot-swap OS components (eg, BDOS, CCP) on the fly.

Some CP/M compatible operating systems extended the basic functionality so far that they far exceeded the original, for example the multi-processor capable TurboDOS.

A number of CP/M-80 derivatives existed in the former Eastern Bloc under various names, including SCP (Single User Control Program), SCP/M, CP/A,[77] CP/J, CP/KC, CP/KSOB, CP/L, CP/Z, MICRODOS, BCU880, ZOAZ, OS/M, TOS/M, ZSDOS, M/OS, COS-PSA, DOS-PSA, CSOC, CSOS, CZ-CPM and others.[78][79] There were also CP/M-86 derivatives named SCP1700, CP/K and K8918-OS.[79] They were produced by the East German VEB Robotron and others.[79][78][77]