Re: Fuzzing project brainstorming

[Michal Zalewski <lcamtuf () coredump cx>]

> If it came from a mutation-based fuzzer, the original (not-mutated) sample
> can be useful too.

FWIW, in afl, crashing entries that are otherwise very distant from
the original starting corpus can be generally linked back to the
"parent", non-crashing synthesized test case that the fuzzer was
working with at the time - often producing pretty small diffs. This
can, but doesn't have to, help =)

/mz