In that case, you announce a reroll, along with a published paper explaining that x^1 == x. But, assuming you use SHA-256 or higher, the chances of that happening are less than one over the number of atoms in the observable universe, so you shouldn't worry about it the same way you don't worry about hash collisions happening purely by chance.