LLMs Can Also Do Well! Breaking Barriers in Semantic Role Labeling via Large Language Models