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 kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id D6EB4C636D6 for ; Wed, 22 Feb 2023 19:58:49 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 5DDC36B0071; Wed, 22 Feb 2023 14:58:49 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id 567486B0072; Wed, 22 Feb 2023 14:58:49 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 4082A6B0073; Wed, 22 Feb 2023 14:58:49 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0014.hostedemail.com [216.40.44.14]) by kanga.kvack.org (Postfix) with ESMTP id 2D4B46B0071 for ; Wed, 22 Feb 2023 14:58:49 -0500 (EST) Received: from smtpin19.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay04.hostedemail.com (Postfix) with ESMTP id E65881A0CEE for ; Wed, 22 Feb 2023 19:58:48 +0000 (UTC) X-FDA: 80495990736.19.63E7FBE Received: from forward502b.mail.yandex.net (forward502b.mail.yandex.net [178.154.239.146]) by imf21.hostedemail.com (Postfix) with ESMTP id 85FE21C0020 for ; Wed, 22 Feb 2023 19:58:46 +0000 (UTC) Authentication-Results: imf21.hostedemail.com; dkim=pass header.d=ya.ru header.s=mail header.b=YoNZ2eA9; spf=pass (imf21.hostedemail.com: domain of tkhai@ya.ru designates 178.154.239.146 as permitted sender) smtp.mailfrom=tkhai@ya.ru; dmarc=pass (policy=none) header.from=ya.ru ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1677095927; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references:dkim-signature; bh=YsD5Es+crBasC+jtKbzgfw9zDouhXr6PGgoj3gfM2CQ=; b=vWs62BMspEVc3l/qPvnL1dYpVYthmwWdgypdDNkNZCneKjZ0o9Ds80jGa2meU8JbqydkcW 2gl12eTQxog/b2/aqbBQJvwcJ+O62I8DKJqa/dQUAHr1G73st8a9IguBT2QVmYVqSLdV8h 6sehCeP8ymTQ1vXHV6sh+XILiKmLUQU= ARC-Authentication-Results: i=1; imf21.hostedemail.com; dkim=pass header.d=ya.ru header.s=mail header.b=YoNZ2eA9; spf=pass (imf21.hostedemail.com: domain of tkhai@ya.ru designates 178.154.239.146 as permitted sender) smtp.mailfrom=tkhai@ya.ru; dmarc=pass (policy=none) header.from=ya.ru ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1677095927; a=rsa-sha256; cv=none; b=PkhnmrV9Z1RPk6aW4VuS1Hw4FvN6BC63oa24+v5aQVkdDie3d9+hYaB02YCqWYaVYGmV+T 16LHhtIqDswon4emrnyXxPIJPzB+mEv3C9LWoAGMGa6lS3mK52+WXO9zr5bD33uA2TkOUR cCsRErMeJVKbwR7LrhhH3Fe63Imqaxk= Received: from iva3-4792ebb2806b.qloud-c.yandex.net (iva3-4792ebb2806b.qloud-c.yandex.net [IPv6:2a02:6b8:c0c:c0e:0:640:4792:ebb2]) by forward502b.mail.yandex.net (Yandex) with ESMTP id 89EBB5F417; Wed, 22 Feb 2023 22:58:43 +0300 (MSK) Received: by iva3-4792ebb2806b.qloud-c.yandex.net (smtp/Yandex) with ESMTPSA id dwUQ802ZjOs1-fqGYg3Vm; Wed, 22 Feb 2023 22:58:42 +0300 X-Yandex-Fwd: 1 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=ya.ru; s=mail; t=1677095922; bh=YsD5Es+crBasC+jtKbzgfw9zDouhXr6PGgoj3gfM2CQ=; h=From:In-Reply-To:Cc:Date:References:To:Subject:Message-ID; b=YoNZ2eA9C/nJ2O/wtM8O5ZnkFFBi7u1d1U3/nAaKHoC03xRy+OsWNBZATcwkzYPM2 lKfyQ+GSZ2YdyxiI0UabfNNjlE9cMgH2uX076IM0jt4dFuwIEHB0e8msO/ZFDDuhBX lYdDcl4ex9LKZBn5i8KYw+JzZPoI9Yrqvl/EKWUU= Message-ID: <715594a8-1eca-7f80-adc0-3655153adffa@ya.ru> Date: Wed, 22 Feb 2023 22:58:37 +0300 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101 Thunderbird/102.8.0 Subject: Re: [PATCH 2/5] mm: vmscan: make memcg slab shrink lockless To: Qi Zheng Cc: sultan@kerneltoast.com, dave@stgolabs.net, penguin-kernel@I-love.SAKURA.ne.jp, paulmck@kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org, Andrew Morton , Johannes Weiner , Shakeel Butt , Michal Hocko , Roman Gushchin , Muchun Song , David Hildenbrand , Yang Shi References: <20230220091637.64865-1-zhengqi.arch@bytedance.com> <20230220091637.64865-3-zhengqi.arch@bytedance.com> <8733cb3c-7a6a-33c4-b84e-4fb981dae765@ya.ru> <2eb276f1-0d81-f207-4cf9-e6586c5e18c8@bytedance.com> Content-Language: en-US From: Kirill Tkhai In-Reply-To: <2eb276f1-0d81-f207-4cf9-e6586c5e18c8@bytedance.com> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-Stat-Signature: ua89tnacxoy4iqaa1t5k5bg864ayqf8m X-Rspam-User: X-Rspamd-Queue-Id: 85FE21C0020 X-Rspamd-Server: rspam06 X-HE-Tag: 1677095926-942392 X-HE-Meta: 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 wcVv8xXA Hno0dQ0e37m7UT8VfPQ7eTuEBFhPGsk/TfI+r6Yc0bbxu27ypX5Kr7c/gNhoE45/Sjx+x02jebGq+IpTYpic64gmHiLXQ9oFDGT6beOXXVkLF7vjDkxvzj2RpXMBCOv1PqWfWRVwuX1sCoGBwfxAhn7Lzk/OQT2gTszwD4q95UHfSE75xIOpwFAiwAFid9ZOfRUNvXK3F+vQgyFGmpsMZfF0dm3/M9x2pbOcVXr6vgRmCkOK8S+GfxTiguu/+qgeGBakF X-Bogosity: Ham, tests=bogofilter, spamicity=0.000002, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: On 22.02.2023 10:32, Qi Zheng wrote: > > > On 2023/2/22 05:28, Kirill Tkhai wrote: >> On 20.02.2023 12:16, Qi Zheng wrote: > <...> >>>     void reparent_shrinker_deferred(struct mem_cgroup *memcg) >>>   { >>> -    int i, nid; >>> +    int i, nid, srcu_idx; >>>       long nr; >>>       struct mem_cgroup *parent; >>>       struct shrinker_info *child_info, *parent_info; >>> @@ -429,16 +443,16 @@ void reparent_shrinker_deferred(struct mem_cgroup *memcg) >>>           parent = root_mem_cgroup; >>>         /* Prevent from concurrent shrinker_info expand */ >>> -    down_read(&shrinker_rwsem); >>> +    srcu_idx = srcu_read_lock(&shrinker_srcu); >> >> Don't we still have to be protected against parallel expand_one_shrinker_info()? >> >> It looks like parent->nodeinfo[nid]->shrinker_info pointer may be changed in expand* >> right after we've dereferenced it here. > > Hi Kirill, > > Oh, indeed. We may wrongly reparent the child's nr_deferred to the old > parent's nr_deferred (it is about to be freed by call_srcu). > > The reparent_shrinker_deferred() will only be called on the offline > path (not a hotspot path), so we may be able to use shrinker_mutex > introduced later for protection. What do you think? It looks good for me. One more thing I'd analyzed is whether we want to have is two reparent_shrinker_deferred() are executing in parallel. Possible, we should leave rwsem there as it was used before.. >> >>>       for_each_node(nid) { >>> -        child_info = shrinker_info_protected(memcg, nid); >>> -        parent_info = shrinker_info_protected(parent, nid); >>> +        child_info = shrinker_info_srcu(memcg, nid); >>> +        parent_info = shrinker_info_srcu(parent, nid); >>>           for (i = 0; i < shrinker_nr_max; i++) { >>>               nr = atomic_long_read(&child_info->nr_deferred[i]); >>>               atomic_long_add(nr, &parent_info->nr_deferred[i]); >>>           } >>>       } >>> -    up_read(&shrinker_rwsem); >>> +    srcu_read_unlock(&shrinker_srcu, srcu_idx); >>>   } >>>     static bool cgroup_reclaim(struct scan_control *sc) >>> @@ -891,15 +905,14 @@ static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid, >>>   { >>>       struct shrinker_info *info; >>>       unsigned long ret, freed = 0; >>> +    int srcu_idx; >>>       int i; >>>         if (!mem_cgroup_online(memcg)) >>>           return 0; >>>   -    if (!down_read_trylock(&shrinker_rwsem)) >>> -        return 0; >>> - >>> -    info = shrinker_info_protected(memcg, nid); >>> +    srcu_idx = srcu_read_lock(&shrinker_srcu); >>> +    info = shrinker_info_srcu(memcg, nid); >>>       if (unlikely(!info)) >>>           goto unlock; >>>   @@ -949,14 +962,9 @@ static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid, >>>                   set_shrinker_bit(memcg, nid, i); >>>           } >>>           freed += ret; >>> - >>> -        if (rwsem_is_contended(&shrinker_rwsem)) { >>> -            freed = freed ? : 1; >>> -            break; >>> -        } >>>       } >>>   unlock: >>> -    up_read(&shrinker_rwsem); >>> +    srcu_read_unlock(&shrinker_srcu, srcu_idx); >>>       return freed; >>>   } >>>   #else /* CONFIG_MEMCG */ >> >