Compilers Overview

Tool Chain

Preprocessor

• Takes source code and header files.
• #pragma, #define, #include are preprocessed and inserted into source code.
• cpp is the default preprocessor.

Compiler

• cc1 is the default compiler.
• Outputs assembly source code ending with .s (as it was the only source back then)

Assembler

See assembly for assembly language.

• as is the default assembler.
• Generates Object Code (.o files)

Linker

• Takes libraries such as libc.a, libm.a and object code.
• .a file is actually a bunch of .o files, in static linking, only needed functions are included.
• Dynamic libraries (libc.so)
• There is a small overhead time for dynamically linked programs — since the program needs to find the location of the shared libraries.
• Generates executables.

Within Compiler

• Input = Source Code Raw Bytes (individual characters/character stream)
• Scanner
  • Reads raw bytes into tokens
  • Regular Expression, which can be turned into NFA, then to DFA.
  • The tools used is Flex
• Parser
  • Produces an AST (= abstract syntax tree)
  • CFG
    • This can be expansive, we need to narrow down.
    • LL and LR grammar, which are converted to FA + Stack, with Bison.
• Semantic Routines
  • Make sure that the “sentence” is valid.
• Optimizer
  • Takes and generates IR (= Intermediate Representation)
• Code Generator → Target Assembly Language.