From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-m15579.qiye.163.com (mail-m15579.qiye.163.com [101.71.155.79]) (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 B0A2132C8B; Tue, 1 Sep 2026 06:49:57 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=101.71.155.79 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1788245402; cv=none; b=MwwNij/PcUbPOXMmZLMb9lNPbSbfwcznqI+6iEkySzwqpcCgUeo7jhl8bjWNZVFtOcaHOsuVubX2kFzar7yzkzOePGdjfjJQQqg2rTYrczacbaOhQLLVGtB+XmxjI5Dzar42LMMwY5wcPyCsv4jFwuoTmDVtjcRsnAgmoGND2WA= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1788245402; c=relaxed/simple; bh=NHNxEZJXYgAHKb+TI8KOG1lDALWcsz0dQQPADFKjH2U=; h=Message-ID:Date:MIME-Version:Subject:To:Cc:References:From: In-Reply-To:Content-Type; b=R6RsRqXEycg+77UDSVY9AWxL4838ZDztqCLxP9B6G+xxaVGinrHtu8v8J06+srWfOAzgZ6uqYQW/pgFyyzdzqRukJJim/8gJ+abWQ0QX9HbvsFfONONVwHxQuFfTokritWCE11uDd+VcSWGvFteHKQ/0K/wl1P5XOBhswlQPZ1k= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=easystack.cn; spf=pass smtp.mailfrom=easystack.cn; arc=none smtp.client-ip=101.71.155.79 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=easystack.cn Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=easystack.cn Received: from [192.168.0.59] (unknown [218.94.118.90]) by smtp.qiye.163.com (Hmail) with ESMTP id 1e8c82612; Tue, 1 Sep 2026 14:34:34 +0800 (GMT+08:00) Message-ID: <9fce3210-3548-4df3-ae00-730b666f5d57@easystack.cn> Date: Tue, 1 Sep 2026 14:34:34 +0800 Precedence: bulk X-Mailing-List: linux-doc@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH 0/8] mm/page_owner: Add PID/TGID/COMM and cgroup filtering To: Andrew Morton , "Lorenzo Stoakes (ARM)" Cc: David Hildenbrand , "Liam R . Howlett" , Vlastimil Babka , Mike Rapoport , Suren Baghdasaryan , Michal Hocko , Jonathan Corbet , Shuah Khan , Randy Dunlap , Brendan Jackman , Johannes Weiner , Zi Yan , linux-mm@kvack.org, linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org References: <20260828031339.1270699-1-zhen.ni@easystack.cn> <20260828113654.d01abd8baed3f73e87506eb3@linux-foundation.org> From: "zhen.ni" In-Reply-To: <20260828113654.d01abd8baed3f73e87506eb3@linux-foundation.org> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-HM-Tid: 0aa05bad3a510229kunm61780f1f274d70 X-HM-MType: 1 X-HM-Spam-Status: e1kfGhgUHx5ZQUpXWQgPGg8OCBgUHx5ZQUlOS1dZFg8aDwILHllBWSg2Ly tZV1koWUFJQjdXWRgWCB1ZQUpXWS1ZQUlXWQ8JGhUIEh9ZQVlDQhpMVkhLTEIYGkNKGE1OSlYVFA kWGhdVGRETFhoSFyQUDg9ZV1kYEgtZQVlJSkNVQk9VSkpDVUJLWVdZFhoPEhUdFFlBWU9LSFVKS0 hKSkJMVUpLS1VKQktLWQY+ 在 2026/8/29 02:36, Andrew Morton 写道: > On Fri, 28 Aug 2026 08:03:38 +0100 "Lorenzo Stoakes (ARM)" wrote: > >> On Fri, Aug 28, 2026 at 11:13:31AM +0800, Zhen Ni wrote: >>> This patch series adds process and memory cgroup filtering support to >>> page_owner. Following the previous series that introduced print_mode and >>> NUMA node filters: >>> https://lore.kernel.org/linux-mm/20260707115411.1714314-1-zhen.ni@easystack.cn/ >>> >>> This series adds filtering capabilities to page_owner, allowing users to >>> filter output by specific processes and memory cgroups. Users can now >>> filter page_owner output by PID, TGID, COMM (with wildcard support), and >>> memory cgroup path. This makes page_owner debugging more focused and >>> efficient for tracking memory allocations in specific contexts. >> >> The majority of this cover letter feels like it should have been >> documentation put somewhere :) > > yes please. > > The only thing longer than the cover letter is the Sashiko report ;) > > https://sashiko.dev/#/patchset/20260828031339.1270699-1-zhen.ni@easystack.cn > > Hi Andrew, Lorenzo, Thanks for the review. I have analyzed all the Sashiko report findings. Some will be fixed in the next version, and for the rest I propose not to fix them, with reasons below. If there are no objections I will send v2 accordingly. Will be fixed in the next version: - mm/page_owner.c: drop the kstrdup() copy in parse_pid_t_list(); the token will be parsed in place. This also fixes a leak on the success path and a kfree() of an advanced (interior) pointer on the error path. cmp_int() will replace plain subtraction in cmp_pid_t(), and pid values exceeding PID_MAX_LIMIT will be rejected. - mm/page_owner.c: PAGE_OWNER will select GLOB so glob_match() is always linked in; parse_comm_list() will drop its kstrdup() copy the same way. - mm/page_owner.c: the cgroup path buffer will be allocated once per read() outside the page_ext RCU read-side critical section instead of per page inside get_page_memcg_info() (GFP_KERNEL allocations must not sleep there). The memcg= parsing branch will be guarded by CONFIG_MEMCG so kernels built without memcg reject the command instead of silently enabling a filter that never matches. - tools/mm/page_owner_filter.c: user-visible input errors (empty -p/-t/-c/-g arguments) will print error messages instead of exiting silently. - Documentation: the wildcard pattern in the -c example will be quoted to prevent shell glob expansion. Proposed not to fix, by design: 1. Shared-fd concurrent read/write races (READ_ONCE around the pid passed to bsearch, torn reads of pid/tgid lists, glob_match() racing comm rewrites, concurrent write() leading to state->memcg_path double-allocation): multi-threaded sharing of one page_owner fd is not a designed use of this interface. The filters are per-fd state meant to be configured once and then read, which is what the page_owner_filter tool does. Adding locking to the read path would put overhead into the per-page scan loop for no designed benefit. This matches the semantics of the original filter introduction. 2. No "clear filter" support (empty pid=/tgid= lists partially clearing proc filters, empty memcg=/nid= being rejected): write commands are incremental -- "keep the unmentioned filters" -- and there is no clear operation by design. To start over, close the fd and open a fresh one; the page_owner_filter tool already works this way. This also matches the semantics of the original filter introduction. 3. kcalloc() vs kmalloc_array() for new_comm_list: no consumer of the list reads past strscpy()'s NUL terminator, so uninitialized bytes are unreachable. 4. char cgroup_path[512] in validate_cgroup_path(): acknowledged that the kernel side accepts paths up to PATH_MAX (4096). The userspace check truncates an over-long path with snprintf() and then fails access(), so it is rejected, never silently accepted. Bumping the buffer to PATH_MAX would only serve pathological paths; realistic cgroup paths are well under 100 bytes, so 512 wastes nothing in practice. If this plan looks reasonable I will send v2. Thanks, Zhen Ni