Here are the tools and benchmarks of the APSEC2022 paper "NumFuzz: A Floating-Point Format Aware Fuzzer for Numerical Programs".
-
All experiments were done on an "Ubuntu20" machine. This project has already contained the corresponding llvm+clang (with LLVM version 6.0.1).
cd NumFuzz
-
Under the "tool" directory, install NumFuzz by:
sudo sh install_NumFuzz.sh
install covFPFUZZ by:
sudo sh install_covFPFUZZ.sh
Other tool are similar.
-
Under the "test" directory, run test of NumFuzz by:
sh run_test4_NumFuzz.sh
Other tools are similar.