Add smarter fuzzing strategies.

[personnumber3377]

Hi!

There exists a fuzzer for the tcpsigner testing binary, but it just uses the default mutation engine which afl uses. The interesting APDU messages sent to the program are highly structured and random mutations of these binary messages are unlikely to yield any interesting results. I have begun developing a custom mutator specifically for this program which you can find here: https://github.com/personnumber3377/apdu_custom_mutator which aims to provide smarter fuzzing strategies for these messages. It for example takes into account the input format for the data to the program. It deserializes the input data into message objects, then mutates those message objects (such as swapping their places, deleting messages and multiplying them) and then serializes them back to binary. This way the deeper program logic gets fuzzed, not just the shallow parsing logic. It is still work in progress, so feel free to suggest changes to it. I have written a (badly written) blog post here: https://personnumber3377.github.io/projects/implementing_a_custom_mutator_for_apdu.html about the development of this mutator.

Thanks in advance for your response!