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 Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 7E5A0CCA47C for ; Tue, 14 Jun 2022 07:18:15 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1353886AbiFNHSO (ORCPT ); Tue, 14 Jun 2022 03:18:14 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:56618 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1352539AbiFNHSI (ORCPT ); Tue, 14 Jun 2022 03:18:08 -0400 Received: from mail-io1-xd49.google.com (mail-io1-xd49.google.com [IPv6:2607:f8b0:4864:20::d49]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 64CD622BD3 for ; Tue, 14 Jun 2022 00:17:42 -0700 (PDT) Received: by mail-io1-xd49.google.com with SMTP id i126-20020a6bb884000000b006691e030971so3905237iof.15 for ; Tue, 14 Jun 2022 00:17:42 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20210112; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc:content-transfer-encoding; bh=hWQ/pnH1IXBczpgDhNdPRVcRKkX01kpJeoSQmJTZ/E8=; b=NP0JFeVh5/LPbnW/y3636zUCWVda9EYqywiDnvF21ZXsgIBShq1h7BvC4Rrl603A8T +Ace/YsvlFsB0aaxLzuvXzyIhd58XHpVrU//G+O7x2dqEMj1C6LeIb36fPHxh1LZc5zi yub5O7gzo5kn+xhaS8slwkOTTjt6wpwZI0pb5jESWxPZrEIaXq9Rlu/n4JVK2KQoxT5P d3uX+1JCbWJz+sWb9DsMV9q9PLbaZ2v3wHc50k2QgsqR4rfcyPU1cm1TuOqghHT8ZEWV 2UxQspqmHz7pUvNuMLC+39LM+xSbuaOWvvsqA9n0TOWQ4ZxUjAmF27Oi8jQabEtD73Pm tCzA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc:content-transfer-encoding; bh=hWQ/pnH1IXBczpgDhNdPRVcRKkX01kpJeoSQmJTZ/E8=; b=c6jcZxnPhpi3JbipoOw967urkiFFmUK1LXuYPmpmv+Vt2VSpT8QELBPESJJLED2wAb yZzh8XeKBnWetdiKctILD6ifGGyI67shFiyGK34UqSRYobHHkWWBzDK6Aq/Dmw1a5oZP UMbXjxn5ctNcXX7jJhzvAT5F3+Wk3u65SB1EobLmFlAwrT+7+RmEA4WRgDlZqjx8RSea 53Ti4gp8/kzsWrzBlI6COWkmwIyHr4ZE/W04EHOItbDefFg51Lq0Q9PAQVigRqeYJi/X mbbr1uSLR5Koftc5bZrP5qK4apRHNZWfVsGgZXujG2W264wn6nVtxACT0ISZw+w6zz6D civQ== X-Gm-Message-State: AOAM530Zgn8E4SRgjcZMErVR4e06Y+iMgLcoeMK/Lb6JJgfLeZsQIbP1 /HTG1Ssu8Vj8q2d6o/pVSRzgWxchOss= X-Google-Smtp-Source: ABdhPJw5kkiLi1VUjJRTA9t18NMo5X41YB0/ctzW7svU9pT/nobxcfccImde+K9fDSALv1ESSLJTpNZDXas= X-Received: from yuzhao.bld.corp.google.com ([2620:15c:183:200:eaa7:1f3f:e74a:2a26]) (user=yuzhao job=sendgmr) by 2002:a5d:96d0:0:b0:654:91d3:97b7 with SMTP id r16-20020a5d96d0000000b0065491d397b7mr1829135iol.164.1655191062117; Tue, 14 Jun 2022 00:17:42 -0700 (PDT) Date: Tue, 14 Jun 2022 01:16:48 -0600 In-Reply-To: <20220614071650.206064-1-yuzhao@google.com> Message-Id: <20220614071650.206064-12-yuzhao@google.com> Mime-Version: 1.0 References: <20220614071650.206064-1-yuzhao@google.com> X-Mailer: git-send-email 2.36.1.476.g0c4daa206d-goog Subject: [PATCH v12 11/14] mm: multi-gen LRU: thrashing prevention From: Yu Zhao To: Andrew Morton Cc: Andi Kleen , Aneesh Kumar , Catalin Marinas , Dave Hansen , Hillf Danton , Jens Axboe , Johannes Weiner , Jonathan Corbet , Linus Torvalds , Matthew Wilcox , Mel Gorman , Michael Larabel , Michal Hocko , Mike Rapoport , Peter Zijlstra , Tejun Heo , Vlastimil Babka , Will Deacon , linux-arm-kernel@lists.infradead.org, linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, x86@kernel.org, page-reclaim@google.com, Yu Zhao , Brian Geffon , Jan Alexander Steffens , Oleksandr Natalenko , Steven Barrett , Suleiman Souhlal , Daniel Byrne , Donald Carr , "=?UTF-8?q?Holger=20Hoffst=C3=A4tte?=" , Konstantin Kharlamov , Shuang Zhai , Sofia Trinh , Vaibhav Jain Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-doc@vger.kernel.org Add /sys/kernel/mm/lru_gen/min_ttl_ms for thrashing prevention, as requested by many desktop users [1]. When set to value N, it prevents the working set of N milliseconds from getting evicted. The OOM killer is triggered if this working set cannot be kept in memory. Based on the average human detectable lag (~100ms), N=3D1000 usually eliminates intolerable lags due to thrashing. Larger values like N=3D3000 make lags less noticeable at the risk of premature OOM kills. Compared with the size-based approach [2], this time-based approach has the following advantages: 1. It is easier to configure because it is agnostic to applications and memory sizes. 2. It is more reliable because it is directly wired to the OOM killer. [1] https://lore.kernel.org/r/Ydza%2FzXKY9ATRoh6@google.com/ [2] https://lore.kernel.org/r/20101028191523.GA14972@google.com/ Signed-off-by: Yu Zhao Acked-by: Brian Geffon Acked-by: Jan Alexander Steffens (heftig) Acked-by: Oleksandr Natalenko Acked-by: Steven Barrett Acked-by: Suleiman Souhlal Tested-by: Daniel Byrne Tested-by: Donald Carr Tested-by: Holger Hoffst=C3=A4tte Tested-by: Konstantin Kharlamov Tested-by: Shuang Zhai Tested-by: Sofia Trinh Tested-by: Vaibhav Jain --- include/linux/mmzone.h | 2 ++ mm/vmscan.c | 71 +++++++++++++++++++++++++++++++++++++++--- 2 files changed, 69 insertions(+), 4 deletions(-) diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 840b7ca8b91f..472bd5335517 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -419,6 +419,8 @@ struct lru_gen_struct { unsigned long max_seq; /* the eviction increments the oldest generation numbers */ unsigned long min_seq[ANON_AND_FILE]; + /* the birth time of each generation in jiffies */ + unsigned long timestamps[MAX_NR_GENS]; /* the multi-gen LRU lists, lazily sorted on eviction */ struct list_head lists[MAX_NR_GENS][ANON_AND_FILE][MAX_NR_ZONES]; /* the multi-gen LRU sizes, eventually consistent */ diff --git a/mm/vmscan.c b/mm/vmscan.c index 2b6de7e38aa7..075525a9526a 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -4231,6 +4231,7 @@ static void inc_max_seq(struct lruvec *lruvec, bool c= an_swap) for (type =3D 0; type < ANON_AND_FILE; type++) reset_ctrl_pos(lruvec, type, false); =20 + WRITE_ONCE(lrugen->timestamps[next], jiffies); /* make sure preceding modifications appear */ smp_store_release(&lrugen->max_seq, lrugen->max_seq + 1); =20 @@ -4357,7 +4358,7 @@ static unsigned long get_nr_evictable(struct lruvec *= lruvec, unsigned long max_s return total; } =20 -static void age_lruvec(struct lruvec *lruvec, struct scan_control *sc) +static bool age_lruvec(struct lruvec *lruvec, struct scan_control *sc, uns= igned long min_ttl) { bool need_aging; unsigned long nr_to_scan; @@ -4371,21 +4372,40 @@ static void age_lruvec(struct lruvec *lruvec, struc= t scan_control *sc) mem_cgroup_calculate_protection(NULL, memcg); =20 if (mem_cgroup_below_min(memcg)) - return; + return false; =20 nr_to_scan =3D get_nr_evictable(lruvec, max_seq, min_seq, swappiness, &ne= ed_aging); if (!nr_to_scan) - return; + return false; =20 nr_to_scan >>=3D mem_cgroup_online(memcg) ? sc->priority : 0; =20 + if (min_ttl) { + int gen =3D lru_gen_from_seq(min_seq[LRU_GEN_FILE]); + unsigned long birth =3D READ_ONCE(lruvec->lrugen.timestamps[gen]); + + if (time_is_after_jiffies(birth + min_ttl)) + return false; + + /* the size is likely too small to be helpful */ + if (!nr_to_scan && sc->priority !=3D DEF_PRIORITY) + return false; + } + if (nr_to_scan && need_aging) try_to_inc_max_seq(lruvec, max_seq, sc, swappiness); + + return true; } =20 +/* to protect the working set of the last N jiffies */ +static unsigned long lru_gen_min_ttl __read_mostly; + static void lru_gen_age_node(struct pglist_data *pgdat, struct scan_contro= l *sc) { struct mem_cgroup *memcg; + bool success =3D false; + unsigned long min_ttl =3D READ_ONCE(lru_gen_min_ttl); =20 VM_WARN_ON_ONCE(!current_is_kswapd()); =20 @@ -4411,12 +4431,28 @@ static void lru_gen_age_node(struct pglist_data *pg= dat, struct scan_control *sc) do { struct lruvec *lruvec =3D mem_cgroup_lruvec(memcg, pgdat); =20 - age_lruvec(lruvec, sc); + if (age_lruvec(lruvec, sc, min_ttl)) + success =3D true; =20 cond_resched(); } while ((memcg =3D mem_cgroup_iter(NULL, memcg, NULL))); =20 clear_mm_walk(); + + /* + * The main goal is to OOM kill if every generation from all memcgs is + * younger than min_ttl. However, another theoretical possibility is all + * memcgs are either below min or empty. + */ + if (!success && !sc->order && mutex_trylock(&oom_lock)) { + struct oom_control oc =3D { + .gfp_mask =3D sc->gfp_mask, + }; + + out_of_memory(&oc); + + mutex_unlock(&oom_lock); + } } =20 /* @@ -5128,6 +5164,28 @@ static void lru_gen_change_state(bool enabled) * sysfs interface *************************************************************************= *****/ =20 +static ssize_t show_min_ttl(struct kobject *kobj, struct kobj_attribute *a= ttr, char *buf) +{ + return sprintf(buf, "%u\n", jiffies_to_msecs(READ_ONCE(lru_gen_min_ttl)))= ; +} + +static ssize_t store_min_ttl(struct kobject *kobj, struct kobj_attribute *= attr, + const char *buf, size_t len) +{ + unsigned int msecs; + + if (kstrtouint(buf, 0, &msecs)) + return -EINVAL; + + WRITE_ONCE(lru_gen_min_ttl, msecs_to_jiffies(msecs)); + + return len; +} + +static struct kobj_attribute lru_gen_min_ttl_attr =3D __ATTR( + min_ttl_ms, 0644, show_min_ttl, store_min_ttl +); + static ssize_t show_enabled(struct kobject *kobj, struct kobj_attribute *a= ttr, char *buf) { unsigned int caps =3D 0; @@ -5176,6 +5234,7 @@ static struct kobj_attribute lru_gen_enabled_attr =3D= __ATTR( ); =20 static struct attribute *lru_gen_attrs[] =3D { + &lru_gen_min_ttl_attr.attr, &lru_gen_enabled_attr.attr, NULL }; @@ -5191,12 +5250,16 @@ static struct attribute_group lru_gen_attr_group = =3D { =20 void lru_gen_init_lruvec(struct lruvec *lruvec) { + int i; int gen, type, zone; struct lru_gen_struct *lrugen =3D &lruvec->lrugen; =20 lrugen->max_seq =3D MIN_NR_GENS + 1; lrugen->enabled =3D lru_gen_enabled(); =20 + for (i =3D 0; i <=3D MIN_NR_GENS + 1; i++) + lrugen->timestamps[i] =3D jiffies; + for_each_gen_type_zone(gen, type, zone) INIT_LIST_HEAD(&lrugen->lists[gen][type][zone]); =20 --=20 2.36.1.476.g0c4daa206d-goog