From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail203.messagelabs.com (mail203.messagelabs.com [216.82.254.243]) by kanga.kvack.org (Postfix) with ESMTP id 58DCE6B01DA for ; Thu, 14 May 2009 12:24:19 -0400 (EDT) Date: Thu, 14 May 2009 18:22:01 +0200 From: Johannes Weiner Subject: Re: [PATCH] mmtom: Prevent shrinking of active anon lru list in case of no swap space V3 Message-ID: <20090514162201.GA2361@cmpxchg.org> References: <20090514231555.f52c81eb.minchan.kim@gmail.com> <2f11576a0905140727j5ba02b07t94826f57dd99839c@mail.gmail.com> <44c63dc40905140739n271d3d2w2e0cc364c0012d71@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <44c63dc40905140739n271d3d2w2e0cc364c0012d71@mail.gmail.com> Sender: owner-linux-mm@kvack.org To: Minchan Kim Cc: KOSAKI Motohiro , MinChan Kim , Andrew Morton , LKML , linux-mm , Rik van Riel List-ID: On Thu, May 14, 2009 at 11:39:49PM +0900, Minchan Kim wrote: > On Thu, May 14, 2009 at 11:27 PM, KOSAKI Motohiro > wrote: > >> A mm/vmscan.c | A A 2 +- > >> A 1 files changed, 1 insertions(+), 1 deletions(-) > >> > >> diff --git a/mm/vmscan.c b/mm/vmscan.c > >> index 2f9d555..621708f 100644 > >> --- a/mm/vmscan.c > >> +++ b/mm/vmscan.c > >> @@ -1577,7 +1577,7 @@ static void shrink_zone(int priority, struct zone *zone, > >> A A A A * Even if we did not try to evict anon pages at all, we want to > >> A A A A * rebalance the anon lru active/inactive ratio. > >> A A A A */ > >> - A A A if (inactive_anon_is_low(zone, sc)) > >> + A A A if (inactive_anon_is_low(zone, sc) && nr_swap_pages > 0) > >> A A A A A A A A shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0); > > > > > > A A A if (nr_swap_pages > 0 && inactive_anon_is_low(zone, sc)) > > > > is better? > > compiler can't swap evaluate order around &&. > > If GCC optimizes away that branch with CONFIG_SWAP=n as Rik mentioned, > we don't have a concern. It can only optimize it away when the condition is a compile time constant. But inactive_anon_is_low() contains atomic operations which the compiler is not allowed to drop and so the && semantics lead to atomic_read() && 0 emitting the read while still knowing the whole expression is 0 at compile-time, optimizing away only the branch itself but leaving the read in place! Compared to 0 && atomic_read() where the && short-circuitry leads to atomic_read() not being executed. And since the 0 is a compile time constant, no code has to be emitted for the read. So KOSAKI-san's is right. Your version results in bigger object code. Hannes -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org