A nonmonotonic extension of KLM Preferential Logic P