On Mon, Mar 30, 2015 at 10:28:47PM -0700, Andrew Morton wrote: > On Tue, 31 Mar 2015 13:45:25 +0900 Minchan Kim <minchan@xxxxxxxxxx> wrote: > > > > > > deactivate_page() doesn't look at or alter PageReferenced(). Should it? > > > > Absolutely true. Thanks. > > Here it goes. > > > > >From 2b2c92eb73a1cceac615b9abd4c0f5f0c3395ff5 Mon Sep 17 00:00:00 2001 > > From: Minchan Kim <minchan@xxxxxxxxxx> > > Date: Tue, 31 Mar 2015 13:38:46 +0900 > > Subject: [PATCH] mm: lru_deactivate_fn should clear PG_referenced > > > > deactivate_page aims for accelerate for reclaiming through > > moving pages from active list to inactive list so we should > > clear PG_referenced for the goal. > > > > ... > > > > --- a/mm/swap.c > > +++ b/mm/swap.c > > @@ -800,6 +800,7 @@ static void lru_deactivate_fn(struct page *page, struct lruvec *lruvec, > > > > del_page_from_lru_list(page, lruvec, lru + LRU_ACTIVE); > > ClearPageActive(page); > > + ClearPageReferenced(page); > > add_page_to_lru_list(page, lruvec, lru); > > > > __count_vm_event(PGDEACTIVATE); > > What if we have > > PageLRU(page) && !PageActive(page) && PageReferenced(page) > > if we really want to "accelerate the reclaim of @page" then we should > clear PG_referenced there too. The function's name is *deactivate*_page. IOW, I think it should work for only pages in active list, IMHO. > > (And what about page_referenced(page) :)) Yes, I considered it when you mentioned PG_referenced. Now, madvise_free clear out access bit of page table when the syscall is called so shrink_page_list could reclaim pages easily. Of course, we could clear access bit by page_referenced for general purpose, not only madvise_free but it would hurt performance for madvise_free so I'd like to leave it unless there is a need for the function. -- Kind regards, Minchan Kim -- 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>