From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752877AbdCMMAw (ORCPT ); Mon, 13 Mar 2017 08:00:52 -0400 Received: from szxga03-in.huawei.com ([45.249.212.189]:4405 "EHLO dggrg03-dlp.huawei.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1751545AbdCMMAl (ORCPT ); Mon, 13 Mar 2017 08:00:41 -0400 Subject: Re: [PATCH v3 RFC] mm/vmscan: more restrictive condition for retry of shrink_zones To: Michal Hocko , Yisheng Xie References: <1489316770-25362-1-git-send-email-ysxie@foxmail.com> <20170313083314.GA31518@dhcp22.suse.cz> CC: , , , , , , , , , From: Yisheng Xie Message-ID: <8de4c31e-cfa3-25c4-52f9-f341a826534d@huawei.com> Date: Mon, 13 Mar 2017 20:00:02 +0800 User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.1.0 MIME-Version: 1.0 In-Reply-To: <20170313083314.GA31518@dhcp22.suse.cz> Content-Type: text/plain; charset="windows-1252" Content-Transfer-Encoding: 7bit X-Originating-IP: [10.177.29.40] X-CFilter-Loop: Reflected X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090205.58C689CB.00F5,ss=1,re=0.000,recu=0.000,reip=0.000,cl=1,cld=1,fgs=0, ip=0.0.0.0, so=2014-11-16 11:51:01, dmn=2013-03-21 17:37:32 X-Mirapoint-Loop-Id: 5c689025f59849fb4492746a2fa1c273 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Michal, Thanks for reviewing. On 2017/3/13 16:33, Michal Hocko wrote: > Please do not post new version after a single feedback and try to wait > for more review to accumulate. This is in the 3rd version and it is not > clear why it is still an RFC. Get it, thanks for pointing out these. > > On Sun 12-03-17 19:06:10, Yisheng Xie wrote: >> From: Yisheng Xie >> >> When we enter do_try_to_free_pages, the may_thrash is always clear, and >> it will retry shrink zones to tap cgroup's reserves memory by setting >> may_thrash when the former shrink_zones reclaim nothing. >> >> However, when memcg is disabled or on legacy hierarchy, it should not do >> this useless retry at all, for we do not have any cgroup's reserves >> memory to tap, and we have already done hard work but made no progress. >> >> To avoid this time costly and useless retrying, add a stub function >> mem_cgroup_thrashed() and return true when memcg is disabled or on >> legacy hierarchy. > > Have you actually seen this as a bad behavior? On which workload? Or > have spotted this by the code review? Sorry, this is just spotted by code review. I will point it out changelog. > > Please note that more than _what_ it is more interesting _why_ the patch > has been prepared. Get it. > > I agree the current additional round of reclaim is just lame because we > are trying hard to control the retry logic from the page allocator which > is a sufficient justification to fix this IMO. Right. But I really hate the > name. At this point we do not have any idea that the memcg is trashing > as the name of the function suggests. > > All of them simply might not have any reclaimable pages. So I would > suggest either a better name e.g. memcg_allow_lowmem_reclaim() or, > preferably, fix this properly. E.g. something like the following. hmm, it is pretty a good idea than just rename the function, I will check it's logical carefully, and then send out a new version based on this, if you do not mind. Thanks Yisheng Xie. > --- > diff --git a/mm/vmscan.c b/mm/vmscan.c > index bae698484e8e..989ba9761921 100644 > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -99,6 +99,9 @@ struct scan_control { > /* Can cgroups be reclaimed below their normal consumption range? */ > unsigned int may_thrash:1; > > + /* Did we have any memcg protected by the low limit */ > + unsigned int memcg_low_protection:1; > + > unsigned int hibernation_mode:1; > > /* One of the zones is ready for compaction */ > @@ -2513,6 +2516,7 @@ static bool shrink_node(pg_data_t *pgdat, struct scan_control *sc) > if (mem_cgroup_low(root, memcg)) { > if (!sc->may_thrash) > continue; > + sc->memcg_low_protection = true; > mem_cgroup_events(memcg, MEMCG_LOW, 1); > } > > @@ -2774,7 +2778,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist, > return 1; > > /* Untapped cgroup reserves? Don't OOM, retry. */ > - if (!sc->may_thrash) { > + if ( sc->memcg_low_protection && !sc->may_thrash) { > sc->priority = initial_priority; > sc->may_thrash = 1; > goto retry; >