What if the product Z1*X2 is correlated with the exogenous variable X2?

This is moving away from a question about gretl, but likely they will be correlated indeed. So what? Valid instruments can be correlated among each other (short from being perfectly collinear or close to that).

cheers

sven