Efficient Self-Invalidation/Self-Downgrade for Critical Sections with Relaxed Semantics | Publicación