From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-3.8 required=3.0 tests=DKIM_SIGNED,DKIM_VALID, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI,SPF_HELO_NONE, SPF_PASS autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id A91B7C2BA83 for ; Wed, 12 Feb 2020 18:18:39 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id 1F54720848 for ; Wed, 12 Feb 2020 18:18:39 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=cmpxchg-org.20150623.gappssmtp.com header.i=@cmpxchg-org.20150623.gappssmtp.com header.b="y2gv+7sT" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 1F54720848 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=cmpxchg.org Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id CA6F76B0468; Wed, 12 Feb 2020 13:18:38 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id C7E226B0478; Wed, 12 Feb 2020 13:18:38 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id B95576B047A; Wed, 12 Feb 2020 13:18:38 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0144.hostedemail.com [216.40.44.144]) by kanga.kvack.org (Postfix) with ESMTP id A1BB66B0468 for ; Wed, 12 Feb 2020 13:18:38 -0500 (EST) Received: from smtpin11.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay04.hostedemail.com (Postfix) with ESMTP id 606A6248D for ; Wed, 12 Feb 2020 18:18:38 +0000 (UTC) X-FDA: 76482285516.11.coach13_4b5207eea5c1e X-HE-Tag: coach13_4b5207eea5c1e X-Filterd-Recvd-Size: 7382 Received: from mail-qt1-f195.google.com (mail-qt1-f195.google.com [209.85.160.195]) by imf05.hostedemail.com (Postfix) with ESMTP for ; Wed, 12 Feb 2020 18:18:37 +0000 (UTC) Received: by mail-qt1-f195.google.com with SMTP id d18so2308261qtj.10 for ; Wed, 12 Feb 2020 10:18:37 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cmpxchg-org.20150623.gappssmtp.com; s=20150623; h=date:from:to:cc:subject:message-id:references:mime-version :content-disposition:in-reply-to; bh=OWGcBq+RmtovTTracPqR5yMhReLnPEQNRqBqlsGYoqQ=; b=y2gv+7sTsHniWkrRXHbMycwPBOZkSVEUBxS0FbQa9VB6OPJaYJ1Ny9M6xr/G2DVX+A 2y3zu+4t8K+YEOEZiGZ16SmwaFBkHE7wfdPJ73oAGMEcRLyA7SdYtUq2rlqAIOJ7AlQ+ UybJYIDz3qsZGxZfUM1AtJQB0pELtShzFC9k3aLs7g4W9U5IbFUlcjQfKEHGvhzfclm+ JveHX0RcnHpMZXSqGvlbSX7VBuxQcKiVaV6YvUnzx7SOAGPdgFUsRDnB4h3byEIPdOR4 Deg6SO3d+i3NoTjYxf140EY5fK1CDhyhjBV3mFZrOFF2taWoChENIoAjBrb9azCLuXXz mScA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:from:to:cc:subject:message-id:references :mime-version:content-disposition:in-reply-to; bh=OWGcBq+RmtovTTracPqR5yMhReLnPEQNRqBqlsGYoqQ=; b=CyH02cnsLSbpuYZvvbjXF9jGGlGfypYTWKHn7chhsBQWN9eO2STrLilDGtVVPSEreV kDZxP9aQIvfChNe2GRD5Na9w3O0uu1FBi5Byjuq3uEbnNenzcRveBMPpeH8Er4gTpx+N X9kgmDkvIdu/wuF3ALvUXoqh+NsqsSzPSdhUsO8MQxIdR+vAMjVBXZ5Xxznl4xWHJQJ3 cd7Z9vURV35TWGlv0n59dWq58bgPCstYMOqgIQR1ZiWDVpEYgr3OZyUKDXYTg993EwGH pSJBG1PRU+u+HUSMNoFFkBIiD1D78s36Ly9UZ3OY/qzverhfb0xLxMfjtgVg4YMjnAGa nI/A== X-Gm-Message-State: APjAAAWc7orIl8i9srAhzhgTAThWz5fGW9Wfc4OCQT1c0qK6b0UxkkYn Fl4kqe0CjCnXasoaU1158CvzbA== X-Google-Smtp-Source: APXvYqwp6yb7UZLL9gr229Rnz2uuFRz38SvQqwJmNe7+ZVm7dUDCROR1mDfi5DonZ9jY0+5tL1WvCA== X-Received: by 2002:ac8:6f27:: with SMTP id i7mr8195494qtv.253.1581531516578; Wed, 12 Feb 2020 10:18:36 -0800 (PST) Received: from localhost ([2620:10d:c091:500::5a94]) by smtp.gmail.com with ESMTPSA id b84sm619400qkc.73.2020.02.12.10.18.34 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 12 Feb 2020 10:18:35 -0800 (PST) Date: Wed, 12 Feb 2020 13:18:34 -0500 From: Johannes Weiner To: Joonsoo Kim Cc: Andrew Morton , Andrey Ryabinin , Suren Baghdasaryan , Shakeel Butt , Rik van Riel , Michal Hocko , linux-mm@kvack.org, cgroups@vger.kernel.org, linux-kernel@vger.kernel.org, kernel-team@fb.com Subject: Re: [PATCH 2/3] mm: vmscan: detect file thrashing at the reclaim root Message-ID: <20200212181834.GD180867@cmpxchg.org> References: <20191107205334.158354-1-hannes@cmpxchg.org> <20191107205334.158354-3-hannes@cmpxchg.org> <20200212102817.GA18107@js1304-desktop> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20200212102817.GA18107@js1304-desktop> X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: On Wed, Feb 12, 2020 at 07:28:19PM +0900, Joonsoo Kim wrote: > Hello, Johannes. > > When I tested my patchset on v5.5, I found that my patchset doesn't > work as intended. I tracked down the issue and this patch would be the > reason of unintended work. I don't fully understand the patchset so I > could be wrong. Please let me ask some questions. > > On Thu, Nov 07, 2019 at 12:53:33PM -0800, Johannes Weiner wrote: > ...snip... > > -static void snapshot_refaults(struct mem_cgroup *root_memcg, pg_data_t *pgdat) > > +static void snapshot_refaults(struct mem_cgroup *target_memcg, pg_data_t *pgdat) > > { > > - struct mem_cgroup *memcg; > > - > > - memcg = mem_cgroup_iter(root_memcg, NULL, NULL); > > - do { > > - unsigned long refaults; > > - struct lruvec *lruvec; > > + struct lruvec *target_lruvec; > > + unsigned long refaults; > > > > - lruvec = mem_cgroup_lruvec(memcg, pgdat); > > - refaults = lruvec_page_state_local(lruvec, WORKINGSET_ACTIVATE); > > - lruvec->refaults = refaults; > > - } while ((memcg = mem_cgroup_iter(root_memcg, memcg, NULL))); > > + target_lruvec = mem_cgroup_lruvec(target_memcg, pgdat); > > + refaults = lruvec_page_state(target_lruvec, WORKINGSET_ACTIVATE); > > + target_lruvec->refaults = refaults; > > Is it correct to just snapshot the refault for the target memcg? I > think that we need to snapshot the refault for all the child memcgs > since we have traversed all the child memcgs with the refault count > that is aggregration of all the child memcgs. If next reclaim happens > from the child memcg, workingset transition that is already considered > could be considered again. Good catch, you're right! We have to update all cgroups in the tree, like we used to. However, we need to use lruvec_page_state() instead of _local, because we do recursive comparisons in shrink_node()! So it's not a clean revert of that hunk. Does this patch here fix the problem you are seeing? diff --git a/mm/vmscan.c b/mm/vmscan.c index c82e9831003f..e7431518db13 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -2993,12 +2993,17 @@ static void shrink_zones(struct zonelist *zonelist, struct scan_control *sc) static void snapshot_refaults(struct mem_cgroup *target_memcg, pg_data_t *pgdat) { - struct lruvec *target_lruvec; - unsigned long refaults; + struct mem_cgroup *memcg; - target_lruvec = mem_cgroup_lruvec(target_memcg, pgdat); - refaults = lruvec_page_state(target_lruvec, WORKINGSET_ACTIVATE); - target_lruvec->refaults = refaults; + memcg = mem_cgroup_iter(target_memcg, NULL, NULL); + do { + unsigned long refaults; + struct lruvec *lruvec; + + lruvec = mem_cgroup_lruvec(memcg, pgdat); + refaults = lruvec_page_state(lruvec, WORKINGSET_ACTIVATE); + lruvec->refaults = refaults; + } while ((memcg = mem_cgroup_iter(target_memcg, memcg, NULL))); } /* > > @@ -277,12 +305,12 @@ void workingset_refault(struct page *page, void *shadow) > > * would be better if the root_mem_cgroup existed in all > > * configurations instead. > > */ > > - memcg = mem_cgroup_from_id(memcgid); > > - if (!mem_cgroup_disabled() && !memcg) > > + eviction_memcg = mem_cgroup_from_id(memcgid); > > + if (!mem_cgroup_disabled() && !eviction_memcg) > > goto out; > > - lruvec = mem_cgroup_lruvec(memcg, pgdat); > > - refault = atomic_long_read(&lruvec->inactive_age); > > - active_file = lruvec_lru_size(lruvec, LRU_ACTIVE_FILE, MAX_NR_ZONES); > > + eviction_lruvec = mem_cgroup_lruvec(eviction_memcg, pgdat); > > + refault = atomic_long_read(&eviction_lruvec->inactive_age); > > + active_file = lruvec_page_state(eviction_lruvec, NR_ACTIVE_FILE); > > Do we need to use the aggregation LRU count of all the child memcgs? > AFAIU, refault here is the aggregation counter of all the related > memcgs. Without using the aggregation count for LRU, active_file could > be so small than the refault distance and refault cannot happen > correctly. lruvec_page_state() *is* aggregated for all child memcgs (as opposed to lruvec_page_state_local()), so that comparison looks correct to me.