cifuzz

cifuzz

IMPORTANT: This project is under active development. Be aware that the behavior of the commands or the configuration can change.

cifuzz is a CLI tool that helps you to integrate and run fuzzing based tests into your project.

Getting started

If you are new to the world of fuzzing, we recommend you to take a look at our Glossary.

Installation

Prerequisites

• CMake >= 3.16
• LLVM >= 11

Ubuntu / Debian

sudo apt install cmake clang llvm

Arch

sudo pacman -S cmake clang llvm

MacOS

brew install cmake llvm

Windows

choco install cmake llvm

Get the latest release binaries from the release page. Run the installer by executing:

curl -L -o cifuzz_installer https://github.com/CodeIntelligenceTesting/cifuzz/releases/latest/download/cifuzz_installer_linux # or _darwin for MacOS
chmod +x cifuzz_installer && ./cifuzz_installer

By default, cifuzz gets installed in your home directory under ~/cifuzz. You can customize the installation directory with ./cifuzz_installer -i /target/dir.

Do not forget to add the installation directory to your $PATH.

Setup / Create your first fuzz test

cifuzz commands will interactively guide you through the needed options and show next steps. You can find a complete list of the available commands with all supported options and parameters by calling cifuzz command --help or here.

1. To initialize your project with cifuzz just execute cifuzz init in the root directory of your project. This will create a file named cifuzz.yaml containing the needed configuration.

2. The next step is to create a fuzz test. Execute cifuzz create and follow the instructions given by the command. This will create a stub for your fuzz test, lets say it is called my_fuzz_test.cpp.

3. Edit my_fuzz_test.cpp so it actually calls the function you want to test with the input generated by the fuzzer. To learn more about writing fuzz tests you can take a look at our Tutorial or one of the example projects.

4. Start the fuzzing by executing cifuzz run my_fuzz_test. cifuzz now tries to build the fuzz test and starts a fuzzing run.

Generate coverage report

Once you executed a fuzz test, you can generate a coverage report which shows the line by line coverage of the fuzzed code:

cifuzz coverage my_fuzz_test

Regression testing

Important: In general there are two ways to run your fuzz test:

1. An actual fuzzing run by calling: cifuzz run my_fuzz_test. The fuzzer will rapidly generate new inputs and feed them into your fuzz test. Any input that covers new parts of the fuzzed project will be added to the generated corpus. cifuzz will run until a crash occurs and report detailed information about the finding.

2. As a regression test, by invoking it through your IDE/editor or by directly executing the replayer binary (see here on how to build that binary). This will use the replayer to apply existing input data from the seed corpus, which has to be stored in the directory <fuzz-test-name>_seed_corpus beside your fuzz test. Note that this directory has to be created manually. In this case the fuzz test will stop immediately after applying all input or earlier if a regression occurs.

Sandboxing

On Linux, cifuzz runs the fuzz tests in a sandbox by default, to avoid the fuzz test accidentally harming the system, for example by deleting files or killing processes. It uses Minijail for that.

If you experience problems when running fuzz tests via cifuzz and you don't expect your fuzz tests to do any harm to the system (or you're already running cifuzz in a container), you might want to disable the sandbox via the --use-sandbox=false flag or the use-sandbox: false config file setting.

Contributing

Want to help improve cifuzz? Check out our contributing documentation. There you will find instructions for building the tool locally.

If you find an issue, please report it on the issue tracker.