On Tue, 8 Jan 2013, Andrea Arcangeli wrote: > > Looking at this, one thing that isn't clear is where the page_count is > checked in migrate_misplaced_transhuge_page. Ok that it's unable to > migrate anon transhuge COW shared pages so it doesn't need to mess > with rmap (the mapcount check makes it safe), but it shouldn't be > allowed to migrate memory that has gup direct-IO in flight (and that > can only be detected with a page_count vs mapcount check). Real > migrate does page_freeze_refs to be safe. Mel comments? Yes, I protested to Mel about that before the holidays, and he quickly provided a patch, now in akpm's tree; but checking it again today, I believe it's still not quite right yet - see other mail. Hugh -- 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>