Contributing code

This page walks you through the process of contributing new code changes to FuzzBench.

Source code structure

  • analysis/ - code for analyzing fuzzer performance using statistical tests.
  • benchmarks/ - benchmarks integrated in the FuzzBench platform (e.g. OpenSSL).
  • common/ - common helper modules (e.g. logging, new process handling, etc).
  • database/ - database handling code.
  • docker/ - Dockerfiles for infra images (does not include fuzzer images).
  • docs/ - this documentation.
  • experiment/ - code for running FuzzBench experiments.
  • experiment/build - code for building benchmarks and fuzzers for experiments.
  • experiment/measurer - code for measuring coverage trial coverage.
  • fuzzers/ - fuzzers integrated in the FuzzBench platform (e.g. AFL).
  • service/ - code for the FuzzBench service run by Google.
  • test_libs/ - test helper modules.
  • third_party/ - third-party dependencies (e.g. sancov, oss-fuzz repo).

Running unit tests

You can run all checks and unit tests for the core functionality using:

make presubmit

You can also run the following to format your code or do different checks separately:

  • make format - formats source code using yapf.
  • make lint - runs the linter checks.
  • make typecheck - runs type checker using pytype.
  • make licensecheck - runs the license header checks.