For "IN" fuzzing the 2nd operand is meaningless

[klaus-vb]

The "in ax,dx" instruction has just one input operand. Pre-populating the value of ax is meaningless, because it will be overwritten by the instruction. Should speed up fuzzing quite a lot.

[nstarke]

I will look into this, but one of my questions would be is if the machine code differs between in ax, dx and in ax. If there is no difference, then I think the solution you have proposed is optimal, however, if they are different they probably both still need to be fuzzed, even if we don't set dx to a specific random value before hand. What do you think @klaus-vb ?