Abstract

The GNU Compiler Collection (GCC) is a compiler generation framework. It constructs a compiler for a given architecture by reading the machine descriptions for that architecture. It supports several dozen processors and half a dozen source languages. GCC is a very large code base: the program slocount estimates its size to be over 3.5 million LOC ! When a particular compiler is constructed using GCC , approximately 40k additional lines of source code are generated for the compiler.

The success of GCC lies in the fact that its deployment on the default parameters is easy. However, any other customization, experimentation and modification of GCC is difficult and requires a high amount of expertise and concentrated efforts.

In this talk we take a look at some useful abstractions present in GCC and discuss some key concepts that make it easy for one to walk the maze of the GCC code base.

This talk is based on the work is being carried out at GCC Resource Center at IIT Bombay (http://www.cse.iitb.ac.in/grc).