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=-15.5 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,NICE_REPLY_A,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=unavailable 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 510ADC433E0 for ; Wed, 6 Jan 2021 11:08:17 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 2246F23106 for ; Wed, 6 Jan 2021 11:08:17 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726641AbhAFLHs (ORCPT ); Wed, 6 Jan 2021 06:07:48 -0500 Received: from relay.sw.ru ([185.231.240.75]:45968 "EHLO relay3.sw.ru" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1726211AbhAFLHr (ORCPT ); Wed, 6 Jan 2021 06:07:47 -0500 Received: from [192.168.15.143] by relay3.sw.ru with esmtp (Exim 4.94) (envelope-from ) id 1kx6dY-00Fd3a-PP; Wed, 06 Jan 2021 14:05:56 +0300 Subject: Re: [v3 PATCH 07/11] mm: vmscan: add per memcg shrinker nr_deferred To: Yang Shi , guro@fb.com, shakeelb@google.com, david@fromorbit.com, hannes@cmpxchg.org, mhocko@suse.com, akpm@linux-foundation.org Cc: linux-mm@kvack.org, linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org References: <20210105225817.1036378-1-shy828301@gmail.com> <20210105225817.1036378-8-shy828301@gmail.com> From: Kirill Tkhai Message-ID: Date: Wed, 6 Jan 2021 14:06:06 +0300 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.6.0 MIME-Version: 1.0 In-Reply-To: <20210105225817.1036378-8-shy828301@gmail.com> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 06.01.2021 01:58, Yang Shi wrote: > Currently the number of deferred objects are per shrinker, but some slabs, for example, > vfs inode/dentry cache are per memcg, this would result in poor isolation among memcgs. > > The deferred objects typically are generated by __GFP_NOFS allocations, one memcg with > excessive __GFP_NOFS allocations may blow up deferred objects, then other innocent memcgs > may suffer from over shrink, excessive reclaim latency, etc. > > For example, two workloads run in memcgA and memcgB respectively, workload in B is vfs > heavy workload. Workload in A generates excessive deferred objects, then B's vfs cache > might be hit heavily (drop half of caches) by B's limit reclaim or global reclaim. > > We observed this hit in our production environment which was running vfs heavy workload > shown as the below tracing log: > > <...>-409454 [016] .... 28286961.747146: mm_shrink_slab_start: super_cache_scan+0x0/0x1a0 ffff9a83046f3458: > nid: 1 objects to shrink 3641681686040 gfp_flags GFP_HIGHUSER_MOVABLE|__GFP_ZERO pgs_scanned 1 lru_pgs 15721 > cache items 246404277 delta 31345 total_scan 123202138 > <...>-409454 [022] .... 28287105.928018: mm_shrink_slab_end: super_cache_scan+0x0/0x1a0 ffff9a83046f3458: > nid: 1 unused scan count 3641681686040 new scan count 3641798379189 total_scan 602 > last shrinker return val 123186855 > > The vfs cache and page cache ration was 10:1 on this machine, and half of caches were dropped. > This also resulted in significant amount of page caches were dropped due to inodes eviction. > > Make nr_deferred per memcg for memcg aware shrinkers would solve the unfairness and bring > better isolation. > > When memcg is not enabled (!CONFIG_MEMCG or memcg disabled), the shrinker's nr_deferred > would be used. And non memcg aware shrinkers use shrinker's nr_deferred all the time. > > Signed-off-by: Yang Shi > --- > include/linux/memcontrol.h | 7 +++--- > mm/vmscan.c | 49 +++++++++++++++++++++++++------------- > 2 files changed, 37 insertions(+), 19 deletions(-) > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h > index e05bbe8277cc..5599082df623 100644 > --- a/include/linux/memcontrol.h > +++ b/include/linux/memcontrol.h > @@ -93,12 +93,13 @@ struct lruvec_stat { > }; > > /* > - * Bitmap of shrinker::id corresponding to memcg-aware shrinkers, > - * which have elements charged to this memcg. > + * Bitmap and deferred work of shrinker::id corresponding to memcg-aware > + * shrinkers, which have elements charged to this memcg. > */ > struct memcg_shrinker_info { > struct rcu_head rcu; > - unsigned long map[]; > + unsigned long *map; > + atomic_long_t *nr_deferred; > }; > > /* > diff --git a/mm/vmscan.c b/mm/vmscan.c > index 0033659abf9e..72259253e414 100644 > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -193,10 +193,12 @@ static void memcg_free_shrinker_info_rcu(struct rcu_head *head) > } > > static int memcg_expand_one_shrinker_info(struct mem_cgroup *memcg, > - int size, int old_size) > + int m_size, int d_size, > + int old_m_size, int old_d_size) > { > struct memcg_shrinker_info *new, *old; > int nid; > + int size = m_size + d_size; > > for_each_node(nid) { > old = rcu_dereference_protected( > @@ -209,9 +211,18 @@ static int memcg_expand_one_shrinker_info(struct mem_cgroup *memcg, > if (!new) > return -ENOMEM; > > - /* Set all old bits, clear all new bits */ > - memset(new->map, (int)0xff, old_size); > - memset((void *)new->map + old_size, 0, size - old_size); > + new->map = (unsigned long *)((unsigned long)new + sizeof(*new)); > + new->nr_deferred = (atomic_long_t *)((unsigned long)new + > + sizeof(*new) + m_size); Can't we write this more compact? new->map = (unsigned long *)(new + 1); new->nr_deferred = (atomic_long_t)(new->map + 1); > + > + /* map: set all old bits, clear all new bits */ > + memset(new->map, (int)0xff, old_m_size); > + memset((void *)new->map + old_m_size, 0, m_size - old_m_size); > + /* nr_deferred: copy old values, clear all new values */ > + memcpy((void *)new->nr_deferred, (void *)old->nr_deferred, > + old_d_size); Why not memcpy(new->nr_deferred, old->nr_deferred, old_d_size); ? > + memset((void *)new->nr_deferred + old_d_size, 0, > + d_size - old_d_size); > > rcu_assign_pointer(memcg->nodeinfo[nid]->shrinker_info, new); > call_rcu(&old->rcu, memcg_free_shrinker_info_rcu); > @@ -226,9 +237,6 @@ void memcg_free_shrinker_info(struct mem_cgroup *memcg) > struct memcg_shrinker_info *info; > int nid; > > - if (mem_cgroup_is_root(memcg)) > - return; > - > for_each_node(nid) { > pn = mem_cgroup_nodeinfo(memcg, nid); > info = rcu_dereference_protected(pn->shrinker_info, true); > @@ -242,12 +250,13 @@ int memcg_alloc_shrinker_info(struct mem_cgroup *memcg) > { > struct memcg_shrinker_info *info; > int nid, size, ret = 0; > - > - if (mem_cgroup_is_root(memcg)) > - return 0; > + int m_size, d_size = 0; > > down_read(&shrinker_rwsem); > - size = DIV_ROUND_UP(shrinker_nr_max, BITS_PER_LONG) * sizeof(unsigned long); > + m_size = DIV_ROUND_UP(shrinker_nr_max, BITS_PER_LONG) * sizeof(unsigned long); > + d_size = shrinker_nr_max * sizeof(atomic_long_t); > + size = m_size + d_size; > + > for_each_node(nid) { > info = kvzalloc(sizeof(*info) + size, GFP_KERNEL); > if (!info) { > @@ -255,6 +264,9 @@ int memcg_alloc_shrinker_info(struct mem_cgroup *memcg) > ret = -ENOMEM; > break; > } > + info->map = (unsigned long *)((unsigned long)info + sizeof(*info)); > + info->nr_deferred = (atomic_long_t *)((unsigned long)info + > + sizeof(*info) + m_size); > rcu_assign_pointer(memcg->nodeinfo[nid]->shrinker_info, info); > } > up_read(&shrinker_rwsem); > @@ -265,10 +277,16 @@ int memcg_alloc_shrinker_info(struct mem_cgroup *memcg) > static int memcg_expand_shrinker_info(int new_id) > { > int size, old_size, ret = 0; > + int m_size, d_size = 0; > + int old_m_size, old_d_size = 0; > struct mem_cgroup *memcg; > > - size = DIV_ROUND_UP(new_id + 1, BITS_PER_LONG) * sizeof(unsigned long); > - old_size = DIV_ROUND_UP(shrinker_nr_max, BITS_PER_LONG) * sizeof(unsigned long); > + m_size = DIV_ROUND_UP(new_id + 1, BITS_PER_LONG) * sizeof(unsigned long); > + d_size = (new_id + 1) * sizeof(atomic_long_t); > + size = m_size + d_size; > + old_m_size = DIV_ROUND_UP(shrinker_nr_max, BITS_PER_LONG) * sizeof(unsigned long); > + old_d_size = shrinker_nr_max * sizeof(atomic_long_t); > + old_size = old_m_size + old_d_size; > if (size <= old_size) > return 0; This replication of patch [4/11] looks awkwardly. Please, try to incorporate the same changes to nr_deferred as I requested for shrinker_map in [4/11]. > > @@ -277,9 +295,8 @@ static int memcg_expand_shrinker_info(int new_id) > > memcg = mem_cgroup_iter(NULL, NULL, NULL); > do { > - if (mem_cgroup_is_root(memcg)) > - continue; > - ret = memcg_expand_one_shrinker_info(memcg, size, old_size); > + ret = memcg_expand_one_shrinker_info(memcg, m_size, d_size, > + old_m_size, old_d_size); > if (ret) { > mem_cgroup_iter_break(NULL, memcg); > goto out; >