From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754351AbaE1HC5 (ORCPT ); Wed, 28 May 2014 03:02:57 -0400 Received: from szxga03-in.huawei.com ([119.145.14.66]:15377 "EHLO szxga03-in.huawei.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754294AbaE1HCz (ORCPT ); Wed, 28 May 2014 03:02:55 -0400 Message-ID: <53858A06.8080507@huawei.com> Date: Wed, 28 May 2014 15:02:30 +0800 From: Li Zefan User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:17.0) Gecko/20130801 Thunderbird/17.0.8 MIME-Version: 1.0 To: Marcelo Tosatti CC: , , Lai Jiangshan , Mel Gorman , Tejun Heo , Christoph Lameter , David Rientjes , Andrew Morton Subject: Re: [PATCH] page_alloc: skip cpuset enforcement for lower zone allocations (v2) References: <20140523193706.GA22854@amt.cnet> <20140526185344.GA19976@amt.cnet> In-Reply-To: <20140526185344.GA19976@amt.cnet> Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit X-Originating-IP: [10.177.18.230] X-CFilter-Loop: Reflected Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2014/5/27 2:53, Marcelo Tosatti wrote: > > Zone specific allocations, such as GFP_DMA32, should not be restricted > to cpusets allowed node list: the zones which such allocations demand > might be contained in particular nodes outside the cpuset node list. > > The alternative would be to not perform such allocations from > applications which are cpuset restricted, which is unrealistic. > > Fixes KVM's alloc_page(gfp_mask=GFP_DMA32) with cpuset as explained. > Could you add the use case that you described in a previous email to the changelog? > Signed-off-by: Marcelo Tosatti > > v2: fix slowpath as well (David Rientjes) > > diff --git a/kernel/cpuset.c b/kernel/cpuset.c > index 3d54c41..b70a336 100644 > --- a/kernel/cpuset.c > +++ b/kernel/cpuset.c > @@ -2392,6 +2392,10 @@ int __cpuset_node_allowed_softwall(int node, gfp_t gfp_mask) > Add a comment accordingly? * in_interrupt - any node ok (current task context irrelevant) * GFP_ATOMIC - any node ok * TIF_MEMDIE - any node ok * GFP_KERNEL - any node in enclosing hardwalled cpuset ok * GFP_USER - only nodes in current tasks mems allowed ok. > if (in_interrupt() || (gfp_mask & __GFP_THISNODE)) > return 1; > +#ifdef CONFIG_NUMA > + if (gfp_zone(gfp_mask) < policy_zone) > + return 1; > +#endif > might_sleep_if(!(gfp_mask & __GFP_HARDWALL)); > if (node_isset(node, current->mems_allowed)) > return 1; > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index 5dba293..dfea3dc 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -2698,6 +2698,7 @@ __alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order, > unsigned int cpuset_mems_cookie; > int alloc_flags = ALLOC_WMARK_LOW|ALLOC_CPUSET|ALLOC_FAIR; > struct mem_cgroup *memcg = NULL; > + nodemask_t *cpuset_mems_allowed = &cpuset_current_mems_allowed; > > gfp_mask &= gfp_allowed_mask; > > @@ -2726,9 +2727,14 @@ __alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order, > retry_cpuset: > cpuset_mems_cookie = read_mems_allowed_begin(); > > +#ifdef CONFIG_NUMA > + if (gfp_zone(gfp_mask) < policy_zone) > + cpuset_mems_allowed = NULL; > +#endif > + > /* The preferred zone is used for statistics later */ > first_zones_zonelist(zonelist, high_zoneidx, > - nodemask ? : &cpuset_current_mems_allowed, > + nodemask ? : cpuset_mems_allowed, > &preferred_zone); > if (!preferred_zone) > goto out; > . >