> I'm not familiar with PIN but I from your proposal I understand it's
> something like the valgrind framework? The idea is certainly an
> interesting one, though if my understanding is correct it sounds like
> an awful lot of work for one summer - I could be wrong though.
Thank you for your quick reply.
Yes Pin is a dynamic binary instrumentation framework like Valgrind but it provides a rich API set which allows to build powerful tools easily.
Actually, identifying all the branches of a running program, which is the main function of the "improved fuzzer", can be achieved whith a small Pin tool (~100 lines). The remaining is only algorithms.
If I understood your idea well, both can be implemented in 12 weeks for sure.
I'll take a closer look at randptk utility.
Feel free to suggest improvements or additional ideas.
Sonny