A Note on a Penalty Function Approach for Solving Bilevel Linear Programs