arvidn / libtorrent-fuzz

fuzzing infrastructure for libtorrent

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

libtorrent fuzzing

Fuzzing of various libtorrent APIs (both internal and external), inspired by Kostya Serebryany's cppcon 2017 presentation

This project requires:


A very recent version of clang that supports libFuzzer. clang-5.0 may not be recent enough, you may have to build head from source.


Also known as b2. To configure boost build with your fresh clang build, create a ~/user-config.jam with something like this in it (example for macOS):

using darwin : 6.0 : ~/Documents/dev/clang/build/bin/clang++ : <cxxflags>-std=c++11 ;


The corpus is the set of inputs that has been built by libFuzzer. It's the seed for testing more mutations. The corpus is not checked into the repository, before running the fuzzer it is advised to download and unzip the corpus associated with the latest release on github.

Uzip the corpus in the root directory:

tar xzf corpus.tar.gz


To run the fuzzer:

b2 darwin-6.0 link=static logging=off -j4

By default, 900000 iterations are run per API. This can be adjusted in the Jamfile.


Please consider contributing back any updated corpuses (amended by more seed inputs) or fuzzers for more APIs in libtorrent.

ezoic increase your site revenue


fuzzing infrastructure for libtorrent

License:BSD 3-Clause "New" or "Revised" License


Language:C++ 98.4%Language:Shell 1.6%