From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759637AbcISHL5 (ORCPT ); Mon, 19 Sep 2016 03:11:57 -0400 Received: from LGEAMRELO11.lge.com ([156.147.23.51]:39344 "EHLO lgeamrelo11.lge.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759187AbcISHL4 (ORCPT ); Mon, 19 Sep 2016 03:11:56 -0400 X-Original-SENDERIP: 156.147.1.126 X-Original-MAILFROM: minchan@kernel.org X-Original-SENDERIP: 10.177.223.161 X-Original-MAILFROM: minchan@kernel.org Date: Mon, 19 Sep 2016 16:11:53 +0900 From: Minchan Kim To: "Chen, Tim C" Cc: "Huang, Ying" , Andrew Morton , "Hansen, Dave" , "Kleen, Andi" , "Lu, Aaron" , "linux-mm@kvack.org" , "linux-kernel@vger.kernel.org" , Hugh Dickins , Shaohua Li , Rik van Riel , Andrea Arcangeli , "Kirill A . Shutemov" , Vladimir Davydov , Johannes Weiner , Michal Hocko Subject: Re: [PATCH -v3 00/10] THP swap: Delay splitting THP during swapping out Message-ID: <20160919071153.GB4083@bbox> References: <1473266769-2155-1-git-send-email-ying.huang@intel.com> <20160909054336.GA2114@bbox> <87sht824n3.fsf@yhuang-mobile.sh.intel.com> <20160913061349.GA4445@bbox> <87y42wgv5r.fsf@yhuang-dev.intel.com> <20160913070524.GA4973@bbox> <87vay0ji3m.fsf@yhuang-mobile.sh.intel.com> <20160913091652.GB7132@bbox> <045D8A5597B93E4EBEDDCBF1FC15F50935BF9343@fmsmsx104.amr.corp.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <045D8A5597B93E4EBEDDCBF1FC15F50935BF9343@fmsmsx104.amr.corp.intel.com> User-Agent: Mutt/1.5.24 (2015-08-30) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Tim, On Tue, Sep 13, 2016 at 11:52:27PM +0000, Chen, Tim C wrote: > >> > >> - Avoid CPU time for splitting, collapsing THP across swap out/in. > > > >Yes, if you want, please give us how bad it is. > > > > It could be pretty bad. In an experiment with THP turned on and we > enter swap, 50% of the cpu are spent in the page compaction path. It's page compaction overhead, especially, pageblock_pfn_to_page. Why is it related to overhead THP split for swapout? I don't understand. > So if we could deal with units of large page for swap, the splitting > and compaction of ordinary pages to large page overhead could be avoided. > > 51.89% 51.89% :1688 [kernel.kallsyms] [k] pageblock_pfn_to_page > | > --- pageblock_pfn_to_page > | > |--64.57%-- compaction_alloc > | migrate_pages > | compact_zone > | compact_zone_order > | try_to_compact_pages > | __alloc_pages_direct_compact > | __alloc_pages_nodemask > | alloc_pages_vma > | do_huge_pmd_anonymous_page > | handle_mm_fault > | __do_page_fault > | do_page_fault > | page_fault > | 0x401d9a > | > |--34.62%-- compact_zone > | compact_zone_order > | try_to_compact_pages > | __alloc_pages_direct_compact > | __alloc_pages_nodemask > | alloc_pages_vma > | do_huge_pmd_anonymous_page > | handle_mm_fault > | __do_page_fault > | do_page_fault > | page_fault > | 0x401d9a > --0.81%-- [...] > > Tim