On linear over-actuated regulation using input allocation