Fuzz testing is the art of creating "random" but interesting input to a program. For example, someone wanting to test an HTML parser might feed it thousands of horribly invalid HTML files. But someone testing code that displays SVG would want to use well-formed XML, only using SVG tags and attributes in random combinations and sometimes in invalid ways.
Hundreds of bugs have been fixed as a result of fuzz-testing components of Gecko over the last 15 months. Many of these were bogus assertion failures or obscure hangs, but some were security holes.
- How the Gecko fuzzers work
- The state of Gecko (with respect to fuzzing)
- Strategies for creating new fuzzers that are effective at finding bugs and facilitate creating reduced testcases after finding bugs
- What Gecko features, components, and APIs haven't been fuzz-tested and should be?
Please add your name here if you're likely to attend this session, this will help prioritize sessions and minimize conflicts