Some remarks on state-constrained optimal control problems