Integer Range Analysis in LLVM

This repository contains the code for the Integer Range Analysis passes written for LLVM. The project contains two passes: ConstantRange.cpp and BranchRange.cpp.

Installation

The following instructions explain how to install and build the pass inside LLVM:

• Create file example.c (source code to analize) inside ~/Public/project/code
• Open directory ~/Public/project/llvm-project/llvm/lib/Transforms
• Create directory ConstantRange
• Update CMakeLists.txt inside ~/Public/project/llvm-project/llvm/lib/Transforms by adding the newly created directory:

...
add_subdirectory(ConstantRange)

• Open ConstantRange folder
• Upload ConstantRange.cpp file
• Create CMakeLists.txt file
• Add make code to CMakeLists.txt:

add_llvm_loadable_module( LLVMConstantRange
  ConstantRange.cpp

  PLUGIN_TOOL
  opt
  )

• Open directory ~/Public/project/llvm-project/build
• Run command make -j4 to build the pass

Benchmarks

The benchmarks sources are taken from the following repositories:

• https://github.com/TheAlgorithms/C
• https://github.com/xtaci/algorithms
• https://github.com/AllAlgorithms/c

Info

The passes have been tested on some example files. The code is not guaranteed to function in all cases. The passes can be expanded to encompass more code statements. See src/branch-range/example and src/constant-range/example to view the test cases and their results.

License

Created by Maglione Sandro, MIT license