Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1947048AbdDYVq3 (ORCPT ); Tue, 25 Apr 2017 17:46:29 -0400 Received: from gum.cmpxchg.org ([85.214.110.215]:47226 "EHLO gum.cmpxchg.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1946791AbdDYVqX (ORCPT ); Tue, 25 Apr 2017 17:46:23 -0400 Date: Tue, 25 Apr 2017 17:46:18 -0400 From: Johannes Weiner To: "Huang, Ying" Cc: Andrew Morton , "Kirill A. Shutemov" , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH -mm -v10 3/3] mm, THP, swap: Enable THP swap optimization only if has compound map Message-ID: <20170425214618.GB6841@cmpxchg.org> References: <20170425125658.28684-1-ying.huang@intel.com> <20170425125658.28684-4-ying.huang@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20170425125658.28684-4-ying.huang@intel.com> User-Agent: Mutt/1.8.0 (2017-02-23) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1596 Lines: 49 On Tue, Apr 25, 2017 at 08:56:58PM +0800, Huang, Ying wrote: > From: Huang Ying > > If there is no compound map for a THP (Transparent Huge Page), it is > possible that the map count of some sub-pages of the THP is 0. So it > is better to split the THP before swapping out. In this way, the > sub-pages not mapped will be freed, and we can avoid the unnecessary > swap out operations for these sub-pages. > > Cc: Johannes Weiner > Signed-off-by: "Huang, Ying" Acked-by: Johannes Weiner CC Kirill to double check the reasoning here > --- > mm/swap_state.c | 16 +++++++++++++--- > 1 file changed, 13 insertions(+), 3 deletions(-) > > diff --git a/mm/swap_state.c b/mm/swap_state.c > index 006d91d8fc53..13f83c6bb1b4 100644 > --- a/mm/swap_state.c > +++ b/mm/swap_state.c > @@ -192,9 +192,19 @@ int add_to_swap(struct page *page, struct list_head *list) > VM_BUG_ON_PAGE(!PageLocked(page), page); > VM_BUG_ON_PAGE(!PageUptodate(page), page); > > - /* cannot split, skip it */ > - if (PageTransHuge(page) && !can_split_huge_page(page, NULL)) > - return 0; > + if (PageTransHuge(page)) { > + /* cannot split, skip it */ > + if (!can_split_huge_page(page, NULL)) > + return 0; > + /* > + * Split pages without a PMD map right away. Chances > + * are some or all of the tail pages can be freed > + * without IO. > + */ > + if (!compound_mapcount(page) && > + split_huge_page_to_list(page, list)) > + return 0; > + } > > retry: > entry = get_swap_page(page); > -- > 2.11.0