kakoune-fuzzers

This repository contains some fuzzers based on libFuzzer for the Kakoune code editor.

Dependencies

The code should be compiled with clang++, under a version that has libFuzzer available without any extra installation necessary. According to the documentation, that would be version 6.0 and up.

The version of Clang used should also be supported by upstream Kakoune. Make sure to check the project's build requirements for further details.

Setup

Clone this repository, then place a copy of Kakoune's source code in it:

$ git clone https://github.com/lenormf/kakoune-fuzzers.git
$ cd kakoune-fuzzers
$ git clone https://github.com/mawww/kakoune.git

Note that you can download any version of Kakoune that you want to build the fuzzers against, using Git or not. The code doesn't need to be within kakoune-fuzzers either, it's just more convenient that way.

Compiling

Compile all the fuzzers at once with the following command:

$ make KAKOUNE="${PWD}/kakoune"

The above command will visit all the directories that contain fuzzers, and compile them. You have to pass the absolute path to the Kakoune code to compile against using the KAKOUNE variable.

If you only want to compile the fuzzers for a given category, for example regex, then cd into that category's directory, and run the same command above (adjust the path to Kakoune if needed).

Running a fuzzer

Read the list of all available options using the following command:

$ ./fuzzer -help=1

Once you've decided which flag you're going to use to run the fuzzer, run it, and wait for results. For example:

$ ./fuzzer -only_ascii=1 -detect_leaks=0

Some fuzzers come with a corpus populated with relevant data (for example, regular expressions used at runtime by the editor), feel free to use them by passing their path to the program:

$ ./fuzzer [options] /path/to/corpus

Using a corpus, whether it is provided by this repository or written by yourself, is highly recommended to make fuzzing more effective.

If you want to run several fuzzers that rely on the same initial corpus, create a new corpus directory and perform a merge:

$ mkdir corpus_fuzzer
$ ./fuzzer -merge=1 corpus_fuzzer corpus
$ ./fuzzer [options] corpus_fuzzer