From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755966AbYBHVxU (ORCPT ); Fri, 8 Feb 2008 16:53:20 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751437AbYBHVxK (ORCPT ); Fri, 8 Feb 2008 16:53:10 -0500 Received: from smtp2.linux-foundation.org ([207.189.120.14]:59374 "EHLO smtp2.linux-foundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751314AbYBHVxH (ORCPT ); Fri, 8 Feb 2008 16:53:07 -0500 Date: Fri, 8 Feb 2008 13:51:54 -0800 From: Andrew Morton To: Harvey Harrison Cc: torvalds@linux-foundation.org, viro@ZenIV.linux.org.uk, linux-kernel@vger.kernel.org, mingo@elte.hu Subject: Re: [PATCH] fix sparse warning from include/linux/mmzone.h Message-Id: <20080208135154.ccdac0c1.akpm@linux-foundation.org> In-Reply-To: <1202417543.31361.5.camel@brick> References: <1202417543.31361.5.camel@brick> X-Mailer: Sylpheed version 2.2.4 (GTK+ 2.8.20; i486-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, 07 Feb 2008 12:52:23 -0800 Harvey Harrison wrote: > include/linux/mmzone.h:640:22: warning: potentially expensive pointer subtraction > > The code in question was doing a pointer subtraction to find the index of > the zone argument in the node_zones array. This essentially boils down to: > > ((unsigned long)zone - (unsigned long)zone->zone_pgdat->node_zones)/sizeof(struct zone) > > Which can be expensive if struct zone is not a power of two. Instead, just > calculate the offsets rather than the index in the array. > > Signed-off-by: Harvey Harrison > --- > This shows up > 30 times in an x86 allyesconfig, would be nice to get > rid of it and makes the code a little more efficient. I believe this > is correct, unless there is a subtlety I have missed. > > include/linux/mmzone.h | 9 ++++++--- > 1 files changed, 6 insertions(+), 3 deletions(-) > > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h > index 8d8d197..cb07758 100644 > --- a/include/linux/mmzone.h > +++ b/include/linux/mmzone.h > @@ -637,9 +637,12 @@ static inline int is_normal_idx(enum zone_type idx) > static inline int is_highmem(struct zone *zone) > { > #ifdef CONFIG_HIGHMEM > - int zone_idx = zone - zone->zone_pgdat->node_zones; > - return zone_idx == ZONE_HIGHMEM || > - (zone_idx == ZONE_MOVABLE && zone_movable_is_highmem()); > + const int highmem_off = ZONE_HIGHMEM * sizeof(*zone); > + const int movable_off = ZONE_MOVABLE * sizeof(*zone); > + > + int zone_off = (unsigned long)zone - (unsigned long)zone->zone_pgdat->node_zones; > + return zone_off == highmem_off || > + (zone_off == movable_off && zone_movable_is_highmem()); > #else > return 0; > #endif > -- hrm. For my i386 build (and that's where CONFIG_HIGHMEM really matters) this patch makes mm/page_alloc.o six bytes larger, and I don't think the change did much for readability.