L1. Introduction to assembly and hybrid programming

Assembly languages

Language instructions corresponding to processor instructions
Symbolic (alphanumeric) notation
instructions, arguments readable for humans
machine language uses binary words to express instructions
Every processors' family has their own assembly language
For some architectures, there is more than one assembly language

Assembly mnemonics

Old style assemblers - every machine instruction had its unique mnemonics
Modern assemblers - mnemonics describe function of an instruction, several diff erent machine instructions may perform the same operation with different arguments

Assembly notation

The program consists of directives (instructions for a compiler) and actual processor instructions

Every instruction or directive occupies a single line of text

Elements of assembly notation (in order):

• Label/name
• Instruction/directive with arguments
  Directive names usually start with a dot
• Comment

Labels and other names
Labels is a symbolic representation of an address

Label definition associates the label with an address of an object defined immediately after the label

In many assembly languages label definition must be followed by a colon

Directives
Assembler commands not being symbolic names of processor's machine instructions

Used for:

• Defining and declaring data
• Selecting memory sections in which the following data and instructions will be places
• Enforcing address values
• Declaring public and external symbols

Assembly notation

Label/name/symbol | Proc instruction/as directive | Arguments | Comment

Assembly directives

Assembler commands not being symbolic names of processor's machine instructions

• Defining and declaring data
• Selecting the memory sections for the data
• Enforcing address values
• Declaring public and external symbols
• Aligning data or instructions

Applications of assembly programming

Programming microcontrollers with limited resources or architectures not suited for HLLs
Time-critical routines (computational kernels)
Access to resources not visible to HLLs

• special registers
• instructions not usable for HLL compilers (BCD arithmetic)
• vector units
  Exception processing
  OS kernel routines
  Code optimization by hand

Hybrid Programming

Using procedures written in different languages in a single program, ex. C and assembler
Not time-critical routines programmed in HLL
Control passing mechanisms for HLL
Agreed on control passing mechanisms for interfacing HLL code to assembly code

Application program in an OS env

OS and startup module (being a part of an application) are responsible for preparing program working env:

• Program loading
• Memory allocation
• Stack initialization
• I/O
• File system
  The programmer only creates the body of the application
  The above refers to program running on a g.p. computer
  C standard refers to it as 'hosted environment'

Standalone program

Program running on a bare computer, without an OS or with minimal OS
It must be able to setup its working enviro by itself - Memory allocation, stack
Examples:

• OS kernel
• OS loader
• Startup module of a program running in an OS env

Phases of program creation

Compilation

• The compiler translates a HLL program into assembly program or directly into object file
• Every source file is translated into a separate output (assembly or object)

Assembly

• Assembler source file written by programmer or temporary assemble file created by compiler is translated by assembler into an object file
• Each assemble source file is translated into a separate object file

Linking

• Linker joins object files and library files into executable file which is later loaded into computer's memory

Object file

Contains binary executable code with some references marked as undefined

Usually executable program consists of many linked modules
Module may reference variables and functions defined in other modules

Object file contains the description of external references - symbols used but not defined in the module and globals
The symbols are linked using their names

There can be only a single definition of the module/symbol in all of the object files

Usual object file names:

• Windows file.obj
• Unix file.o

External symbols

Symbols referenced in a given module, defined in other modules

The symbols must be declared as external in a way depending on the programming language used:

• C:
  • extern keyword
  • functions not defined in the module are treated by external by default
• Assembler:
  • Explicit declaration always required - extern/extrn

Public symbols

Defined in a given module which may be used by other modules

Must be declared public:

• C:
  • Every symbol defined on external level is considered public if not declared with static symbol
• Assembler:
  • Explicit declaration required - keyword public, global, globl

Single module program

Short and simple programs written in assembly may consist of only one module with no external references

Linking of such program means simple conversion from object to exec form

Some assemblers are capable of generating exec form directly (ex. NASM in DOS env)

Libraries

Library is a collection of object modules which may be linked to other modules written by programmer

Lib file contains archived .obj files created from several source modules

Libs containing standard library functions for a given HLL are usually supplied with a compiler

Programmer may create his own libraries which may be subsequently used by many programs

Library files

• Unix/Linux - .a extension
• Windows - .lib extension