caballa / clam-tutorial

Examples showing Clam functionality

Clam Tutorial

Prerequisites

Install Clam and make sure that CLAM_INSTALL_DIR/bin is in your PATH environment variable or set environment variable CLAM to CLAM_INSTALL_DIR/bin, where CLAM_INSTALL_DIR is the directory where Clam is installed.

Notes about Clam Installation

Install all the Clam's prerequisites. This is the recipe to install Clam with ldd-boxes and Elina assuming LLVM is installed already in your machine:

git clone https://github.com/seahorn/clam 
cd clam
mkdir build && cd build
cmake -DCMAKE_INSTALL_PREFIX=_CLAM_INSTALL_DIR_ -DLLVM_DIR=__PATH_TO_LLVM_CONFIG -DCRAB_USE_LDD=ON -DCRAB_USE_ELINA=ON ../
cmake --build . --target extra                 
cmake --build . --target crab && cmake ..
cmake --build . --target ldd && cmake ..
cmake --build . --target elina && cmake ..
cmake --build . --target install 

Install lit and OutputCheck

 pip3 install lit
 pip3 install OutputCheck

Run programs

To run all programs from this tutorial just type:

make all

This command will run Clam on each program and it will check (using lit and OutputCheck) that the results match the expected ones.

Build an analyzer from scratch using C++ Clam API

This tutorial includes a standalone LLVM static analyzer that uses the C++ Clam API. The code of the full analyzer is here. To compile the code, follow these two steps:

1. Modify in api-example/Makefile the variables CLAM_INSTALL, LLVM_HOME, BOOST_ROOT, GMP_DIR, and MPFR_DIR.

2. Type make.

cd api-example
./myanalyzer ./bitcode/test3.ll

The directory api-example/bitcode has several bitcode samples that you can analyze.