[$] Long-term get_user_pages() and truncate(): solved at last?

Technologies like RDMA benefit from the ability to map file-backed pagesinto memory. This benefit extends to persistent-memory devices, where thebacking store for the file can be mapped directly without the need to gothrough the kernel's page cache. There is a fundamental conflict, though,between mapping a file's backing store directly and letting the filesystemcode modify that file's on-disk layout, especially when the mapping is heldin place for a long time (as RDMA is wont to do). The problem seemsintractable, but there may yet be a solution in the form of thispatch set (marked "V1,000,002") from Ira Weiny.