From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964853AbaCTVeJ (ORCPT ); Thu, 20 Mar 2014 17:34:09 -0400 Received: from cantor2.suse.de ([195.135.220.15]:59980 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932328AbaCTVeH (ORCPT ); Thu, 20 Mar 2014 17:34:07 -0400 Date: Thu, 20 Mar 2014 21:33:58 +0000 From: Mel Gorman To: Johannes Weiner Cc: Andrew Morton , Rik van Riel , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [patch] mm: page_alloc: spill to remote nodes before waking kswapd Message-ID: <20140320213358.GO25087@suse.de> References: <1395348816-4733-1-git-send-email-hannes@cmpxchg.org> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Disposition: inline In-Reply-To: <1395348816-4733-1-git-send-email-hannes@cmpxchg.org> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Mar 20, 2014 at 04:53:36PM -0400, Johannes Weiner wrote: > On NUMA systems, a node may start thrashing cache or even swap > anonymous pages while there are still free pages on remote nodes. > > This is a result of 81c0a2bb515f ("mm: page_alloc: fair zone allocator > policy") and fff4068cba48 ("mm: page_alloc: revert NUMA aspect of fair > allocation policy"). Before those changes, the allocator would first > try all allowed zones, including those on remote nodes, before waking > any kswapds. But now, the allocator fastpath doubles as the fairness > pass, which in turn can only consider the local node to prevent remote > spilling based on exhausted fairness batches alone. Remote nodes are > only considered in the slowpath, after the kswapds are woken up. But > if remote nodes still have free memory, kswapd should not be woken to > rebalance the local node or it may thrash cash or swap prematurely. > > Fix this by adding one more unfair pass over the zonelist that is > allowed to spill to remote nodes after the local fairness pass fails > but before entering the slowpath and waking the kswapds. > > This also gets rid of the GFP_THISNODE exemption from the fairness > protocol because the unfair pass is no longer tied to kswapd, which > GFP_THISNODE is not allowed to wake up. > > However, because remote spills can be more frequent now - we prefer > them over local kswapd reclaim - the allocation batches on remote > nodes could underflow more heavily. When resetting the batches, use > atomic_long_read() directly instead of zone_page_state() to calculate > the delta as the latter filters negative counter values. > > Signed-off-by: Johannes Weiner > Cc: [3.12+] Acked-by: Mel Gorman -- Mel Gorman SUSE Labs