From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S937363AbdJRKPv (ORCPT ); Wed, 18 Oct 2017 06:15:51 -0400 Received: from outbound-smtp13.blacknight.com ([46.22.139.230]:42000 "EHLO outbound-smtp13.blacknight.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933777AbdJRKPu (ORCPT ); Wed, 18 Oct 2017 06:15:50 -0400 Date: Wed, 18 Oct 2017 11:15:47 +0100 From: Mel Gorman To: Vlastimil Babka Cc: Andrew Morton , Linux-MM , Linux-FSDevel , LKML , Jan Kara , Andi Kleen , Dave Hansen , Dave Chinner Subject: Re: [PATCH 1/8] mm, page_alloc: Enable/disable IRQs once when freeing a list of pages Message-ID: <20171018101547.mjycw7zreb66jzpa@techsingularity.net> References: <20171018075952.10627-1-mgorman@techsingularity.net> <20171018075952.10627-2-mgorman@techsingularity.net> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Disposition: inline In-Reply-To: User-Agent: NeoMutt/20170421 (1.8.2) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Oct 18, 2017 at 11:02:18AM +0200, Vlastimil Babka wrote: > On 10/18/2017 09:59 AM, Mel Gorman wrote: > > Freeing a list of pages current enables/disables IRQs for each page freed. > > This patch splits freeing a list of pages into two operations -- preparing > > the pages for freeing and the actual freeing. This is a tradeoff - we're > > taking two passes of the list to free in exchange for avoiding multiple > > enable/disable of IRQs. > > There's also some overhead of storing pfn in page->private, but all that > seems negligible compared to irq disable/enable... > Exactly and it's cheaper than doing a second page to pfn lookup. > > Looks good. > > > Signed-off-by: Mel Gorman > > Acked-by: Vlastimil Babka > Thanks. > A nit below. > > > @@ -2647,11 +2663,25 @@ void free_hot_cold_page(struct page *page, bool cold) > > void free_hot_cold_page_list(struct list_head *list, bool cold) > > { > > struct page *page, *next; > > + unsigned long flags, pfn; > > + > > + /* Prepare pages for freeing */ > > + list_for_each_entry_safe(page, next, list, lru) { > > + pfn = page_to_pfn(page); > > + if (!free_hot_cold_page_prepare(page, pfn)) > > + list_del(&page->lru); > > + page->private = pfn; > > We have (set_)page_private() helpers so better to use them (makes it a > bit easier to check for all places where page->private is used to e.g. > avoid a clash)? > Agreed and it's trivial to do so ---8<--- mm, page_alloc: Enable/disable IRQs once when freeing a list of page -fix Use page_private and set_page_private helpers. Signed-off-by: Mel Gorman --- mm/page_alloc.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 167e163cf733..092973014c1e 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -2670,14 +2670,14 @@ void free_hot_cold_page_list(struct list_head *list, bool cold) pfn = page_to_pfn(page); if (!free_hot_cold_page_prepare(page, pfn)) list_del(&page->lru); - page->private = pfn; + set_page_private(page, pfn); } local_irq_save(flags); list_for_each_entry_safe(page, next, list, lru) { - unsigned long pfn = page->private; + unsigned long pfn = page_private(page); - page->private = 0; + set_page_private(page, 0); trace_mm_page_free_batched(page, cold); free_hot_cold_page_commit(page, pfn, cold); }