From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from out-172.mta1.migadu.com (out-172.mta1.migadu.com [95.215.58.172]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id CAF433DC860 for ; Fri, 26 Jun 2026 07:05:08 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=95.215.58.172 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1782457510; cv=none; b=VaGeGDj2PotT1TpqoJEmKsSjWAV0uI+vlXqCfFCKmWGBA/XZeWHC8XPqRVQgQreiPoIUeRgccXqM4VkqXEt7ZF4JsruguHfcSeQ9QMZATZec2TDIu4o2kwjGZjRn7Yr72rksDYXgI6SwC/lUsMlWg4Xs3ll77dDBhLHs6aADbT8= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1782457510; c=relaxed/simple; bh=H6m4BWksqjGuBsh/huVSQ7T+H0s2iPd+SaoiMUt6NOY=; h=Message-ID:Date:MIME-Version:Subject:To:Cc:References:From: In-Reply-To:Content-Type; b=kJ9T1O8R0jrsa8EIdIO8sHxXzPuxJ1VkY2b+9C72y5wejtqalTAecgk4kWC41XrhHdfAAYgIEH5dYERM1ILdwGSGyDjxnxVzRfJAIqTghnfDfCKoWeGuzKTH0fNIVEDWpZA99cpdpYSWbijKMb/osb/dp4K4p7xSowq4n6xXEMY= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev; spf=pass smtp.mailfrom=linux.dev; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b=b8c9dZcU; arc=none smtp.client-ip=95.215.58.172 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.dev Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.dev Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.dev header.i=@linux.dev header.b="b8c9dZcU" Message-ID: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1782457506; h=from:from: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; bh=z+bLG5UfpgPjMKqrgsZ8bW9KeIBxng26HYMlgjgCAW8=; b=b8c9dZcUJ81dwTVzshi/HpGfP2I4P05U8G/nHRbW1IMykFToKsM4HP5YTqv8SYOgHWNRlo py1nnHBBimpqvdb1M374AOU511u8l2FGtbmXIXTytfR6NgpZecQavFQSkOvnbf8To8MWhh Pkwt4WyjLFQlCmpa+M34ZHTpXOuepAs= Date: Fri, 26 Jun 2026 15:04:17 +0800 Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Subject: Re: [PATCH v3] mm: mglru: fix stale batch updates after memcg reparenting To: Harry Yoo , Johannes Weiner Cc: akpm@linux-foundation.org, david@kernel.org, kasong@tencent.com, shakeel.butt@linux.dev, baohua@kernel.org, axelrasmussen@google.com, yuanchu@google.com, weixugc@google.com, muchun.song@linux.dev, peiyang_he@smail.nju.edu.cn, mhocko@kernel.org, roman.gushchin@linux.dev, ljs@kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org, Qi Zheng , stable@vger.kernel.org References: <20260625151554.55105-1-qi.zheng@linux.dev> <4c7b0c46-14f0-4a62-893e-e50714e09b74@linux.dev> <46ac28bf-5be1-4600-b522-0a1aa76c28e6@kernel.org> <08cf8972-6cfc-4452-9a3c-88e0368dbbf9@linux.dev> <90fd5300-1016-42e7-abad-08ad85fb62b4@linux.dev> <5a0c6597-6b96-4781-a71b-fd1298b2b7bb@kernel.org> X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. From: Qi Zheng In-Reply-To: <5a0c6597-6b96-4781-a71b-fd1298b2b7bb@kernel.org> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-Migadu-Flow: FLOW_OUT On 6/26/26 2:48 PM, Harry Yoo wrote: > > > On 6/26/26 3:24 PM, Qi Zheng wrote: >> >> >> On 6/26/26 12:59 PM, Harry Yoo wrote: >>> >>> >>> On 6/26/26 1:48 PM, Qi Zheng wrote: >>>> >>>> >>>> On 6/26/26 12:43 PM, Harry Yoo wrote: >>>>> >>>>> >>>>> On 6/26/26 11:27 AM, Qi Zheng wrote: >>>>>> Hi Johannes, >>>>>> >>>>>> On 6/26/26 2:41 AM, Johannes Weiner wrote: >>>>>>> On Thu, Jun 25, 2026 at 11:15:54PM +0800, Qi Zheng wrote: >>>>>>>> From: Qi Zheng >>>>>>>> >>>>>>>> The mglru page table walker batches per-generation size deltas in >>>>>>>> walk->nr_pages while walking page tables without holding the lruvec >>>>>>>> lock. >>>>>>>> The reset_batch_size() later folds those deltas into walk->lruvec >>>>>>>> under >>>>>>>> the lruvec lock. >>>>>>>> >>>>>>>> The page table walker can run concurrently with the memcg >>>>>>>> reparenting >>>>>>>> path >>>>>>>> as follows: >>>>>>>> >>>>>>>> CPU0                           CPU1 >>>>>>>> ====                           ==== >>>>>>>> >>>>>>>> walk_mm >>>>>>>> --> walk_page_range >>>>>>>>        --> update_batch_size >>>>>>>>            --> walk->nr_pages += delta >>>>>>>> >>>>>>>>                                  mem_cgroup_css_offline >>>>>>>>                                  --> memcg_reparent_objcgs >>>>>>>>                                      --> lock lruvec >>>>>>>>                                          lru_gen_reparent_memcg >>>>>>>>                                          --> reparent child >>>>>>>> folios to >>>>>>>> parent >>>>>>>>                                          unlock lruvec >>>>>>>> >>>>>>>>        lock lruvec >>>>>>>>        reset_batch_size >>>>>>>>        --> child lrugen->nr_pages += delta >>>>>>>> >>>>>>>> This will trigger the following warning in lru_gen_exit_memcg(): >>>>>>>> >>>>>>>>       VM_WARN_ON_ONCE(memchr_inv(lruvec->lrugen.nr_pages, 0, >>>>>>>>                      sizeof(lruvec->lrugen.nr_pages))); >>>>>>>> >>>>>>>> And the user-visible impact of underestimated nr_pages in MGLRU was >>>>>>>> premature OOMs because MGLRU does not try to reclaim memory when >>>>>>>> nr_pages >>>>>>>> reaches zero, but there are still more pages. >>>>>>>> >>>>>>>> To fix it, make reset_batch_size() check CSS_DYING under RCU before >>>>>>>> flushing the pending batch. A non-dying memcg keeps the original >>>>>>>> lruvec >>>>>>>> stable against RCU-delayed offlining; a dying memcg redirects the >>>>>>>> deltas >>>>>>>> to the first non-dying ancestor. >>>>>>>> >>>>>>>> Reported-by: Peiyang He >>>>>>>> Closes: https://lore.kernel.org/all/5A9E929D82717101+12fcf643- >>>>>>>> efb8-4b9a-a53a-1e28cc894f0b@smail.nju.edu.cn >>>>>>>> Fixes: f304652609ea ("mm: vmscan: prepare for reparenting MGLRU >>>>>>>> folios") >>>>>>>> Cc: >>>>>>>> Signed-off-by: Qi Zheng >>>>>>>> --- >>>>>>>> Changes in v3: >>>>>>>>     - re-implement lock_batch_lruvec() by checking CSS_DYING >>>>>>>> under the >>>>>>>> RCU lock >>>>>>>>       (suggested by Harry) >>>>>>>>     - update the commit message (suggested by Harry) >>>>>>>>     - temporarily drop the previous Reviewed-by tags >>>>>>>>       (since the sync method has changed) >>>>>>>>     - rebase onto the next-20260624 >>>>>>>> >>>>>>>> Changes in v2: >>>>>>>>     - update the commit message (pointed by Barry) >>>>>>>>     - collect Reviewed-by >>>>>>>> >>>>>>>>     mm/vmscan.c | 45 ++++++++++++++++++++++++++++++++++++++------- >>>>>>>>     1 file changed, 38 insertions(+), 7 deletions(-) >>>>>>>> >>>>>>>> diff --git a/mm/vmscan.c b/mm/vmscan.c >>>>>>>> index 35c3bb15ae96..1ec8c23c72b9 100644 >>>>>>>> --- a/mm/vmscan.c >>>>>>>> +++ b/mm/vmscan.c >>>>>>>> @@ -3262,10 +3262,44 @@ static void update_batch_size(struct >>>>>>>> lru_gen_mm_walk *walk, struct folio *folio, >>>>>>>>         walk->nr_pages[new_gen][type][zone] += delta; >>>>>>>>     } >>>>>>>>     +#ifdef CONFIG_MEMCG >>>>>>>> +static struct lruvec *lock_batch_lruvec(struct lruvec *lruvec) >>>>>>>> +{ >>>>>>>> +    struct pglist_data *pgdat = lruvec_pgdat(lruvec); >>>>>>>> +    struct mem_cgroup *memcg = lruvec_memcg(lruvec); >>>>>>>> + >>>>>>>> +    rcu_read_lock(); >>>>>>> >>>>>>> Where is this unlocked? >>>>>> >>>>>> The lruvec_unlock_irq() in reset_batch_size() will handle the >>>>>> unlocking. >>>>>> >>>>>>> >>>>>>>> +    /* >>>>>>>> +     * The memcg can be NULL when the memory controller is >>>>>>>> disabled. >>>>>>>> +     * Otherwise, the caller keeps the memcg owning @lruvec alive. >>>>>>>> +     */ >>>>>>>> +    if (!memcg || !css_is_dying(&memcg->css)) >>>>>>>> +        goto lock; >>>>>>>> + >>>>>>>> +    do { >>>>>>>> +        memcg = parent_mem_cgroup(memcg); >>>>>>>> +    } while (memcg && css_is_dying(&memcg->css)); >>>>>>>> +    lruvec = mem_cgroup_lruvec(memcg, pgdat); >>>>>>> >>>>>>>       while (unlikely(memcg && css_is_dying(&memcg->css))) { >>>>>>>           memcg = parent_mem_cgroup(memcg); >>>>>>>           lruvec = mem_cgroup_lruvec(memcg, pgdat); >>>>>> >>>>>> There is no need to acquire the lruvec before finding the first >>>>>> non-dying memcg. >>>>> >>>>> struct pglist_data *pgdat = lruvec_pgdat(lruvec); >>>>> struct mem_cgroup *memcg = lruvec_memcg(lruvec); >>>>> >>>>> rcu_read_lock() >>>>> >>>>> while (unlikely(memcg_is_dying(memcg))) >>>>>           memcg = parent_mem_cgroup(memcg); >>>>> >>>>> lruvec = mem_cgroup_lruvec(memcg, pgdat); >>>> >>>> If the first memcg is already non-dying, there's no need to re-acquire >>>> the lruvec. ;) >>> >>> Oh, right :) >>> >>> Hmm but I still think Johannes' suggestion makes the code cleaner. >> >> I don't have a strong preference on which of the two coding styles is >> more readable. BTW, is there any kernel documentation I could refer to >> for this? > > I don't think there's a coding style guide that specifically > mentions this. Just thought it's cleaner because it merges if (...) goto > lock; and do-while into a single while loop. > >>> Observing a dying cgroup should be rare anyway, it's worth focusing >>> more on readability? >> >> While it's rare to encounter consecutive dying memcgs, it can still >> happen, right? > > But is worth saving a few instruction in a basic block that is > unlikely() to be executed? I don't have a strong opinion here. Hi Johannes, I'll leave the decision up to you. If necessary, I can send out the v4. > > I'm not a memcg maintainer myself, though. just my 2 cents. I'd like to express my gratitude for your reviews, and especially for your invaluable help with the earlier dying memcg work! Thanks, Qi >