An assembly language is a low-level language for programming computers. It implements a symbolic representation of the numeric machine codes and other constants needed to program a particular CPU architecture. This representation is usually defined by the hardware manufacturer, and is based on abbreviations (called mnemonics) that help the programmer remember individual instructions, registers, etc. An assembly language is thus specific to a certain physical or virtual computer architecture (as opposed to most high-level languages, which are usually portable).
Assembly languages were first developed in the 1950s, when they were referred to as second generation programming languages. They eliminated much of the error-prone and time-consuming first-generation programming needed with the earliest computers, freeing the programmer from tedium such as remembering numeric codes and calculating addresses. They were once widely used for all sorts of programming. However, by the 1980s (1990s on small computers), their use had largely been supplanted by high-level languages, in the search for improved programming productivity. Today, assembly language is used primarily for direct hardware manipulation, access to specialized processor instructions, or to address critical performance issues. Typical uses are device drivers, low-level embedded systems, and real-time systems.
A utility program called an assembler is used to translate assembly language statements into the target computer's machine code. The assembler performs a more or less isomorphic translation (a one-to-one mapping) from mnemonic statements into machine instructions and data. (This is in contrast with high-level languages, in which a single statement generally results in many machine instructions. A compiler, analogous to an assembler, is used to translate high-level language statements into machine code; or an interpreter executes statements directly.)
Many sophisticated assemblers offer additional mechanisms to facilitate program development, control the assembly process, and aid debugging. In particular, most modern assemblers (although many have been available for more than 40 years already) include a macro facility (described below), and are called macro assemblers.
Assemblers are generally simpler to write than compilers for high-level languages, and have been available since the 1950s. Modern assemblers, especially for RISC based architectures, such as MIPS, Sun SPARC and HP PA-RISC, optimize instruction scheduling to exploit the CPU pipeline efficiently.
More sophisticated high-level assemblers provide language abstractions such as:
See Language design below for more details.
Note that, in normal professional usage, the term assembler is often used ambiguously: It is frequently used to refer to an assembly language itself, rather than to the assembler utility. Thus: "CP/CMS was written in S/360 assembler" as opposed to "ASM-H was a widely-used S/370 assembler."
The equivalent assembly language representation is easier to remember (example in Intel syntax, more mnemonic):
MOV AL, 61h
This instruction means:
The mnemonic "mov" represents the opcode 1011 which moves the value in the second operand into the register indicated by the first operand. The mnemonic was chosen by the instruction set designer to abbreviate "move", making it easier for the programmer to remember. A comma-separated list of arguments or parameters follows the opcode; this is a typical assembly language statement.
In practice many programmers drop the word mnemonic and, technically incorrectly, call "mov" an opcode. When they do this they are referring to the underlying binary code which it represents. To put it another way, a mnemonic such as "mov" is not an opcode, but as it symbolizes an opcode, one might refer to "the opcode mov" for example when one intends to refer to the binary opcode it symbolizes rather than to the symbol--the mnemonic--itself. As few modern programmers have need to be mindful of actually what binary patterns are the opcodes for specific instructions, the distinction has in practice become a bit blurred among programmers but not among processor designers.
Transforming assembly into machine language is accomplished by an assembler, and the reverse by a disassembler. Unlike in high-level languages, there is usually a one-to-one correspondence between simple assembly statements and machine language instructions. However, in some cases, an assembler may provide pseudoinstructions which expand into several machine language instructions to provide commonly needed functionality. For example, for a machine that lacks a "branch if greater or equal" instruction, an assembler may provide a pseudoinstruction that expands to the machine's "set if less than" and "branch if zero (on the result of the set instruction)". Most full-featured assemblers also provide a rich macro language (discussed below) which is used by vendors and programmers to generate more complex code and data sequences.
Each computer architecture and processor architecture has its own machine language. On this level, each instruction is simple enough to be executed using a relatively small number of electronic circuits. Computers differ by the number and type of operations they support. For example, a new 64-bit machine would have different circuitry from a 32-bit machine. They may also have different sizes and numbers of registers, and different representations of data types in storage. While most general-purpose computers are able to carry out essentially the same functionality, the ways they do so differ; the corresponding assembly languages reflect these differences.
Multiple sets of mnemonics or assembly-language syntax may exist for a single instruction set, typically instantiated in different assembler programs. In these cases, the most popular one is usually that supplied by the manufacturer and used in its documentation.
Most modern assemblers also support pseudo-operations, which are directives obeyed by the assembler at assembly time instead of the CPU at run time. (For example, pseudo-ops would be used to reserve storage areas and optionally set their initial contents.) The names of pseudo-ops often start with a dot to distinguish them from machine instructions.
Some assemblers also support pseudo-instructions, which generate two or more machine instructions.
Symbolic assemblers allow programmers to associate arbitrary names (labels or symbols) with memory locations. Usually, every constant and variable is given a name so instructions can reference those locations by name, thus promoting self-documenting code. In executable code, the name of each subroutine is associated with its entry point, so any calls to a subroutine can use its name. Inside subroutines, GOTO destinations are given labels. Some assemblers support local symbols which are lexically distinct from normal symbols (e.g., the use of "10$" as a GOTO destination).
Most assemblers provide flexible symbol management, allowing programmers to manage different namespaces, automatically calculate offsets within data structures, and assign labels that refer to literal values or the result of simple computations performed by the assembler. Labels can also be used to initialize constants and variables with relocatable addresses.
Assembly languages, like most other computer languages, allow comments to be added to assembly source code that are ignored by the assembler. Good use of comments is even more important with assembly code than with higher-level languages, as the meaning of a sequence of instructions is harder to decipher from the code itself.
Wise use of these facilities can greatly simplify the problems of coding and maintaining low-level code. Raw assembly source code as generated by compilers or disassemblers — code without any comments, meaningful symbols, or data definitions — is quite difficult to read when changes must be made.
Since macros can have 'short' names but expand to several or indeed many lines of code, they can be used to make assembly language programs appear to be much shorter (require less lines of source code from the application programmer - as with a higher level language). They can also be used to add higher levels of structure to assembly programs, optionally introduce embedded de-bugging code via parameters and other similar features.
Many assemblers have built-in macros for system calls and other special code sequences.
Macro assemblers often allow macros to take parameters. Some assemblers include quite sophisticated macro languages, incorporating such high-level language elements as optional parameters, symbolic variables, conditionals, string manipulation, and arithmetic operations, all usable during the execution of a given macros, and allowing macros to save context or exchange information. Thus a macro might generate a large number of assembly language instructions or data definitions, based on the macro arguments. This could be used to generate record-style data structures or "unrolled" loops, for example, or could generate entire algorithms based on complex parameters. An organization using assembly language that has been heavily extended using such a macro suite can be considered to be working in a higher-level language, since such programmers are not working with a computer's lowest-level conceptual elements.
Macros were used to customize large scale software systems for specific customers in the mainframe era and were also used by customer personnel to satisfy their employers' needs by making specific versions of manufacturer operating systems; this was done, for example, by systems programmers working with IBM's Conversational Monitor System/Virtual Machine (CMS/VM) and with IBM's "real time transaction processing" add-on, Customer Information Control System, CICS.
It was also possible to use solely the macro processing capabilities of an assembler to generate code written in completely different languages, for example, to generate a version of a program in Cobol using a pure macro assembler program containing lines of Cobol code inside assembly time operators instructing the assembler to generate arbitrary code.
This was because, as was realized in the 1970s, the concept of "macro processing" is independent of the concept of "assembly", the former being in modern terms more word processing, text processing, than generating object code. The concept of macro processing in fact appeared in and appears in the C programming language, which supports "preprocessor instructions" to set variables, and make conditional tests on their values. Note that unlike certain previous macro processors inside assemblers, the C preprocessor was not Turing-complete because it lacked the ability to either loop or "go to", the latter allowing the programmer to loop.
Despite the power of macro processing, it fell into disuse in high level languages while remaining a perennial for assemblers.
This was because many programmers were rather confused by macro parameter substitution and did not disambiguate macro processing from assembly and execution.
Macro parameter substitution is strictly by name: at macro processing time, the value of a parameter is textually substituted for its name. The most famous class of bugs resulting was the use of a parameter that itself was an expression and not a simple name when the macro writer expected a name. In the macro: foo: macro a load a*b the intention was that the caller would provide the name of a variable, and the "global" variable or constant b would be used to multiply "a". If foo is called with the parameter a-c, an unexpected macro expansion occurs.
To avoid this, users of macro processors learned to religiously parenthesize formal parameters inside macro definitions, and callers had to do the same to their "actual" parameters.
PL/I and C feature macros, but this facility was underused or dangerous when used because they can only manipulate text. On the other hand, homoiconic languages, such as Lisp, Prolog, and Forth, retain the power of assembly language macros because they are able to manipulate their own code as data.
A curious design was A-natural, a "stream-oriented" assembler for 8080/Z80 processors from Whitesmiths Ltd. (developers of the Unix-like Idris operating system, and what was reported to be the first commercial C compiler). The language was classified as an assembler, because it worked with raw machine elements such as opcodes, registers, and memory references; but it incorporated an expression syntax to indicate execution order. Parentheses and other special symbols, along with block-oriented structured programming constructs, controlled the sequence of the generated instructions. A-natural was built as the object language of a C compiler, rather than for hand-coding, but its logical syntax won some fans.
There has been little apparent demand for more sophisticated assemblers since the decline of large-scale assembly language development. In spite of that, they are still being developed and applied in cases where resource constraints or peculiarities in the target system's architecture prevent the effective use of higher-level languages.
Most early microcomputers relied on hand-coded assembly language, including most operating systems and large applications. This was because these systems had severe resource constraints, imposed idiosyncratic memory and display architectures, and provided limited, buggy system services. Perhaps more important was the lack of first-class high-level language compilers suitable for microcomputer use. A psychological factor may have also played a role: the first generation of microcomputer programmers retained a hobbyist, "wires and pliers" attitude.
In a more commercial context, the biggest reasons for using assembly language were size, speed, and reliability: the writers of Cardbox-Plus said simply "we use assembler because then all the bugs are ours". This held true for 8-bit versions of the program, which had no bugs at all, but ironically it turned out to be false with 16 bits: Cardbox-Plus 2.0 had to be upgraded to Cardbox-Plus 2.1 because a bug in Microsoft's macro assembler caused Cardbox-Plus to index the number "-0" differently from the number "0".
Typical examples of large assembly language programs from this time are the MS-DOS operating system, the early IBM PC spreadsheet program Lotus 1-2-3, and almost all popular games for the Atari 800 family of home computers. Even into the 1990s, most console video games were written in assembly, including most games for the Mega Drive/Genesis and the Super Nintendo Entertainment System . According to some industry insiders, the assembly language was the best computer language to use to get the best performance out the Sega Saturn, a console that was notoriously challenging to develop and program games for . The popular arcade game NBA Jam (1993) is another example. On the Commodore 64, Amiga, Atari ST, as well as ZX Spectrum home computers, assembler has long been the primary development language. This was in large part due to the fact that BASIC dialects on these systems offered insufficient execution speed, as well as insufficient facilities to take full advantage of the available hardware on these systems. Some systems, most notably Amiga, even have IDEs with highly advanced debugging and macro facilities, such as the freeware ASM-One assembler, comparable to that of Microsoft Visual Studio facilities (ASM-One predates Microsoft Visual Studio).
The Assembler for the VIC-20 was written by Don French and published by French Silk. At 1639 bytes in length, its author believes it is the smallest symbolic assembler ever written. The assembler supported the usual symbolic addressing and the definition of character strings or hex strings. It also allowed address expressions which could be combined with addition, subtraction, multiplication, division, logical AND, logical OR, and exponentiation operators.
There are really only a handful of situations where today's expert practitioners would choose assembly language :
Nevertheless, assembly language is still taught in most Computer Science and Electronic Engineering programs. Although few programmers today regularly work with assembly language as a tool, the underlying concepts remain very important. Such fundamental topics as binary arithmetic, memory allocation, stack processing, character set encoding, interrupt processing, and compiler design would be hard to study in detail without a grasp of how a computer operates at the hardware level. Since a computer's behavior is fundamentally defined by its instruction set, the logical way to learn such concepts is to study an assembly language. Most modern computers have similar instruction sets. Therefore, studying a single assembly language is sufficient to learn: i) The basic concepts; ii) To recognize situations where the use of assembly language might be appropriate; and iii) To see how efficient executable code can be created from high-level languages.
Many compilers render high-level languages into assembly first before fully compiling, allowing the assembly code to be viewed for debugging and optimization purposes. Relatively low-level languages, such as C, often provide special syntax to embed assembly language directly in the source code. Programs using such facilities, such as the Linux kernel, can then construct abstractions utilizing different assembly language on each hardware platform. The system's portable code can then utilize these processor-specific components through a uniform interface.
Assembly language is also valuable in reverse engineering, since many programs are distributed only in machine code form, and machine code is usually easy to translate into assembly language and carefully examine in this form, but very difficult to translate into a higher-level language. Tools such as the Interactive Disassembler make extensive use of disassembly for such a purpose.
A particular niche that makes use of assembly language is the demoscene. Certain competitions require the contestants to restrict their creations to a very small size (e.g. 256B, 1KB, 4KB or 64 KB), and assembly language is the language of choice to achieve this goal. When resources, particularly CPU-processing constrained systems, like the Amiga and the Commodore 64, are a concern, assembler coding is a must: optimized assembler code is written "by hand" and instructions are sequenced manually by the coders in an attempt to minimize the number of CPU cycles used; the CPU constraints are so great that every CPU cycle counts. However, using such techniques has enabled systems like the Commodore 64 to produce real-time 3D graphics with advanced effects, a feat which might be considered unlikely or even impossible for a system with a 0.99MHz processor.
Within processor groups, each assembler has its own dialect. Sometimes, some assemblers can read another assembler's dialect, for example, TASM can read old MASM code, but not the reverse. FASM and NASM have similar syntax, but each support different macros that could make them difficult to translate to each other. The basics are all the same, but the advanced features will differ.
Also, assembly can sometimes be portable across different operating systems on the same type of CPU. Calling conventions between operating systems often differ slightly or not at all, and with care it is possible to gain some portability in assembly language, usually by linking with a C library that does not change between operating systems.
For example, many things in libc depend on the preprocessor to do OS-specific, C-specific things to the program before compiling. In fact, some functions and symbols are not even guaranteed to exist outside of the preprocessor. Worse, the size and field order of structs, as well as the size of certain typedefs such as off_t, are entirely unavailable in assembly language without help from a configure script, and differ even between versions of Linux, making it impossible to portably call functions in libc other than ones that only take simple integers and pointers as parameters. To address this issue, FASMLIB project provides a portable assembly library for Win32 and Linux platforms, but it is yet very incomplete.
Some higher level computer languages, such as C and Borland Pascal, support inline assembly where relatively brief sections of assembly code can be embedded into the high level language code. The Forth programming language commonly contains an assembler used in CODE words.
Many people use an emulator to debug assembly-language programs.
|Address||Label||Instruction (AT&T syntax)||Object code|
|2064||be done||00000010 10000000 00000000 00000110|
|2068||addcc %r1,-4,%r1||10000010 10000000 01111111 11111100|
|2072||addcc %r1,%r2,%r4||10001000 10000000 01000000 00000010|
|2076||ld %r4,%r5||11001010 00000001 00000000 00000000|
|2080||ba loop||00010000 10111111 11111111 11111011|
|2084||addcc %r3,%r5,%r3||10000110 10000000 11000000 00000101|
|2088||done:||jmpl %r15+4,%r0||10000001 11000011 11100000 00000100|
|2092||length:||20||00000000 00000000 00000000 00010100|
|2096||address:||a_start||00000000 00000000 00001011 10111000|
Example of a selection of instructions (for a virtual computer) with the corresponding address in memory where each instruction will be placed. These addresses are not static, see memory management. Accompanying each instruction is the generated (by the assembler) object code that coincides with the virtual computer's architecture (or ISA).