Protocol Fuzzer Community Edition

7 posts · Aug 2019 – May 2023

What this trail covers

GitLab is integrating fuzz testing capabilities, acquired from Peach Tech and Fuzzit, into its platform. This aims to make fuzz testing more accessible by embedding it within the existing GitLab workflow, presenting results inline with other CI/CD outputs. The offering will include both coverage-guided and behavioral fuzz testing methods. Initially, it will support web API fuzzing (behavioral) and Go applications (coverage-guided), with plans to expand to other languages and use cases, including automated instrumented fuzzing with AFL.

All GitLab trails

May 2023Building GitLab with GitLab: Web API Fuzz TestingThis post details the process of integrating GitLab's Web API Fuzz Testing analyzer with the company's own API. It covers auto-generating an OpenAPI specification from Grape annotations, handling edge cases like file uploads and wildcard parameters, performance tuning for large APIs by splitting jobs and using dedicated runners, and triaging findings by repeating requests and debugging local instances. Lessons learned include the difficulty of automated API documentation generation and the benefits of using GitLab features like Review Apps and dynamically generated child pipelines. The post also mentions the development of a HAR Recorder tool and the high compute cost of fuzzing, suggesting scheduled pipelines for this workflow.
Mar 2021We're open sourcing Protocol Fuzzer Community Edition!This post announces the open-sourcing of the core protocol fuzz testing engine of Peach Fuzzer as GitLab Protocol Fuzzer Community Edition. It details the capabilities of the engine, including running and orchestrating fuzz tests and defining custom protocols. It also provides background on fuzz testing and Peach Fuzzer, discusses the benefits and challenges of fuzz testing, outlines best practices, and explains how to use fuzz testing in GitLab CI/CD. Future plans for fuzz testing in GitLab are also mentioned.
Dec 2020How to fuzz Go code with go-fuzz continuouslyThis post details how to use go-fuzz for continuous fuzzing of Go code, including building and running fuzz tests locally and integrating them into GitLab CI/CD pipelines. It provides a practical example of finding an off-by-one bug in a Go function and outlines the steps for setting up fuzzing as a blocking pipeline stage or as a long-running asynchronous job.
Dec 2020How to fuzz Rust code continuouslyThis post details how to use `cargo-fuzz` for fuzzing Rust code, explaining its relationship with libFuzzer and its coverage-guided algorithm. It provides a step-by-step guide to building and running a fuzz target using a sample Rust function with an off-by-one bug, demonstrating how the fuzzer finds the bug and generates a crash artifact. It also outlines how to integrate `cargo-fuzz` into GitLab CI/CD pipelines using the `.gitlab-ci.yml` file and the `gitlab-cov-fuzz` CLI for both blocking pipeline execution and long-running asynchronous fuzzing sessions.
Oct 2020Find Bugs with Coverage-Guided Fuzz TestingThis post introduces GitLab's built-in coverage-guided fuzz testing capability, explaining how it helps discover bugs and security issues by sending random inputs to instrumented applications. It also provides a link to a video tutorial on configuring fuzz testing on GitLab.
Advertisement
Aug 2020How Arctic Engine uses GitLab's fuzz testingThis post details how Arctic Engine integrated GitLab's fuzz testing capabilities into its CI/CD pipeline to test its sound loader, CSV, and TGA file parsers. It describes the process of creating a test harness with `LLVMFuzzerTestOneInput`, configuring CMakeLists.txt with `-fsanitize=fuzzer`, and adding a few lines to `.gitlab-ci.yml`. The author highlights the discovery of several bugs, including signed integer overflow, division by zero, buffer overruns, and out-of-memory errors, and expresses plans to apply fuzz testing to future network code.
Jul 2020How recent acquisitions introduce fuzz testing to GitLabThis post announces the acquisition of Peach Tech and Fuzzit, bringing their fuzz testing technologies into GitLab. It introduces the concepts of coverage-guided and behavioral fuzz testing, explaining how GitLab will integrate these into its CI/CD workflow for easier adoption and inline results presentation. Several fuzz testing engines and Peach Tech Community Edition are being open-sourced. The initial release will focus on web APIs and Go applications, with future plans for protocol fuzz testing.
Aug 2019GitLab automates instrumented fuzzing via American Fuzzy LopThis post details the automation of instrumented fuzzing with American Fuzzy Lop (AFL) within a GitLab CI/CD pipeline. It provides a practical example using Docker, a sample vulnerable C program, and a Python script to manage AFL execution, artifact collection, and potential issue creation. The post also discusses considerations for performance, distributed fuzzing, and advanced output triage.