The problem here is that, with an error that does prove the main results false, a good paper may be forced to restart the review procedure. This will entail some severe delays. So, my suggestion is to wait and then, after the report of the referee, a resubmission of a properly revised version of the paper.