AES-128 and Integer overflow detected in keyspace of mask

[Chick3nman]

Yeah, there's no advantage against attacks like this where the "password" is not human created. That said, there's not really any disadvantage either, as you will still cover the full keyspace. Switching the markov model off, as you've done, will return it to "expected order", there should be no difference in performance as using the model to order the keyspace does not incur any penalty.