On Mon, 10 Aug 2015, Kirill A. Shutemov wrote: > I don't see anything actionable here. Your wish list doesn't cope with > reality. Compound pages are mapped with PTEs for almost ten years and I > don't see why we should stop the practice. Well they have to if they are smaller than huge pages. Treating each PTE as each having their own state instead of having the whole compound mapped completely causes the problem. Refcounting in tail pages is not necessary if the whole compound is either mapped or not mapped at all by a process. Refcounting in tail pages is only necessary if you allow 4k slices to be mapped. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>