Maximum Principle for a Stochastic Optimal Control Problem and Application to Portfolio/Consumption Choice