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 mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 1A31FC433F5 for ; Thu, 18 Nov 2021 00:34:50 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id BDD9661078 for ; Thu, 18 Nov 2021 00:34:49 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.4.1 mail.kernel.org BDD9661078 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=kvack.org Received: by kanga.kvack.org (Postfix) id 423CD6B0072; Wed, 17 Nov 2021 19:34:39 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id 3D3C96B0073; Wed, 17 Nov 2021 19:34:39 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 29AFA6B0074; Wed, 17 Nov 2021 19:34:39 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0057.hostedemail.com [216.40.44.57]) by kanga.kvack.org (Postfix) with ESMTP id 1A9096B0072 for ; Wed, 17 Nov 2021 19:34:39 -0500 (EST) Received: from smtpin25.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay04.hostedemail.com (Postfix) with ESMTP id D771188CD3 for ; Thu, 18 Nov 2021 00:34:28 +0000 (UTC) X-FDA: 78820179816.25.38E839C Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by imf09.hostedemail.com (Postfix) with ESMTP id B78DC3000118 for ; Thu, 18 Nov 2021 00:34:26 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1637195667; 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: in-reply-to:in-reply-to:references:references; bh=BeNhrEKZDmwOYYueDNfboE8DCNvjuM9I1jBB9AYYqeQ=; b=DnY4TvmEUzKVH6u8JeGfRk5qOTtJ71otzWLFUXnRrbxIB9R+c+yWamWJ0TNKQldy5xZPE2 MNiu2bckJGwApk/G5uV4QOUPDsXTz76mLOGabB3R6z9N4kUSwIiZe3dBUJg/DuESMDb25+ kMybCjBwIKB90VwWzJVEPPOr2ekgP24= Received: from mail-pl1-f197.google.com (mail-pl1-f197.google.com [209.85.214.197]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-438-bigejWC4NUyWEYi2Zyt-jQ-1; Wed, 17 Nov 2021 19:34:24 -0500 X-MC-Unique: bigejWC4NUyWEYi2Zyt-jQ-1 Received: by mail-pl1-f197.google.com with SMTP id v23-20020a170902bf9700b001421d86afc4so2029731pls.9 for ; Wed, 17 Nov 2021 16:34:24 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:date:from:to:cc:subject:message-id:references :mime-version:content-disposition:in-reply-to; bh=BeNhrEKZDmwOYYueDNfboE8DCNvjuM9I1jBB9AYYqeQ=; b=UAmM9Fz7mndY80c6lUOUKV5jHyFEALCGvXY5mPnlltYLhN02/SDFzEDCjbPrTRIQ+t M+ulW8G/Rsx1T2U3MH4uypF3rbhR8jHibaMPYnM96+PLv+VoWV4SknGkmqG+X228tPpC HAg5Jqlmy+1P78xBiexgErl8uGAc3sRUe4DkactFOcuqEYUMWa4Qn6DDMbERoUyCRGXH yCUdI8beZyaEysDtfdTRLNWnt5Uvtc7ZFJ3te69zUjzXV7zVhz2KHSt7VDayTnRbwqt2 mmgbEY9HAe72pUEz6v5+VHj/uo36RZ56xljuEPJkA3x58MNEej9kG6xBgfQidNi7nV9O MnzA== X-Gm-Message-State: AOAM533v6+ZQD/uHQIMNACAR6zDs55/TRP28CO8j7b/b8acqLcQSiuR+ 5OeRrSHBdsU/S3lJ+qHZLTrnzbood3kuKM7hi2Tbimaadisz92w7zeCEbSNqnkG4oNK+Csi19/+ eZzKFW0hsWVQ= X-Received: by 2002:a17:90b:4aca:: with SMTP id mh10mr5136768pjb.234.1637195663359; Wed, 17 Nov 2021 16:34:23 -0800 (PST) X-Google-Smtp-Source: ABdhPJz0qgqZGid7zYOV0THXIIaStWGVESZVDgQ8anBVeCNdnsEnBD0O/E8F3Tq/CgNL9Ux+taUsqA== X-Received: by 2002:a17:90b:4aca:: with SMTP id mh10mr5136707pjb.234.1637195662934; Wed, 17 Nov 2021 16:34:22 -0800 (PST) Received: from xz-m1.local ([191.101.132.73]) by smtp.gmail.com with ESMTPSA id f19sm777629pfc.122.2021.11.17.16.34.17 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 17 Nov 2021 16:34:22 -0800 (PST) Date: Thu, 18 Nov 2021 08:34:14 +0800 From: Peter Xu To: Mina Almasry Cc: Jonathan Corbet , David Hildenbrand , Matthew Wilcox , "Paul E . McKenney" , Yu Zhao , Andrew Morton , Ivan Teterevkov , Florian Schmidt , linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, linux-mm@kvack.org, linux-doc@vger.kernel.org Subject: Re: [PATCH v6] mm: Add PM_THP_MAPPED to /proc/pid/pagemap Message-ID: References: <20211117194855.398455-1-almasrymina@google.com> MIME-Version: 1.0 In-Reply-To: <20211117194855.398455-1-almasrymina@google.com> X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset=utf-8 Content-Disposition: inline X-Rspamd-Server: rspam04 X-Rspamd-Queue-Id: B78DC3000118 X-Stat-Signature: w1bsbmw1tzdd4yxj65hqihz1wdrfqtoc Authentication-Results: imf09.hostedemail.com; dkim=pass header.d=redhat.com header.s=mimecast20190719 header.b=DnY4TvmE; spf=none (imf09.hostedemail.com: domain of peterx@redhat.com has no SPF policy when checking 170.10.133.124) smtp.mailfrom=peterx@redhat.com; dmarc=pass (policy=none) header.from=redhat.com X-HE-Tag: 1637195666-855872 X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: Hi, Mina, On Wed, Nov 17, 2021 at 11:48:54AM -0800, Mina Almasry wrote: > Add PM_THP_MAPPED MAPPING to allow userspace to detect whether a given virt > address is currently mapped by a transparent huge page or not. Example > use case is a process requesting THPs from the kernel (via a huge tmpfs > mount for example), for a performance critical region of memory. The > userspace may want to query whether the kernel is actually backing this > memory by hugepages or not. > > PM_THP_MAPPED bit is set if the virt address is mapped at the PMD > level and the underlying page is a transparent huge page. > > A few options were considered: > 1. Add /proc/pid/pageflags that exports the same info as > /proc/kpageflags. This is not appropriate because many kpageflags are > inappropriate to expose to userspace processes. > 2. Simply get this info from the existing /proc/pid/smaps interface. > There are a couple of issues with that: > 1. /proc/pid/smaps output is human readable and unfriendly to > programatically parse. > 2. /proc/pid/smaps is slow. The cost of reading /proc/pid/smaps into > userspace buffers is about ~800us per call, and this doesn't > include parsing the output to get the information you need. The > cost of querying 1 virt address in /proc/pid/pagemaps however is > around 5-7us. This does not seem to be fair... Should the "800us" value relevant to the process memory size being mapped? As smaps requires walking the whole memory range and provides all stat info including THP accountings. While querying 1 virt address can only account 1 single THP at most. It means if we want to do all THP accounting for the whole range from pagemap we need multiple read()s, right? The fair comparison should compare the sum of all the read()s on the virt addr we care to a single smap call. And it's hard to be fair too, IMHO, because all these depend on size of mm. Smaps is, logically, faster because of two things: - Smaps needs only one syscall for whatever memory range (so one user->kernel->user switch). Comparing to pagemap use case of yours, you'll need to read 1 virt address for each PMD, so when the PMD mapped size is huge, it could turn out that smaps is faster even counting in the parsing time of smaps output. - Smaps knows how to handle things in PMD level without looping into PTEs: see smaps_pmd_entry(). Smaps will not duplicate the PMD entry into 512 PTE entries, because smaps is doing statistic (and that's exaxtly what your use case wants!), while pagemap is defined in 4K page size even for huge mappings because the structure is defined upon the offset of the pagemap fd; that's why it needs to duplicate the 2M entry into 512 same ones; even if they're not really so meaningful. That's also why I tried to propose the interface of smaps to allow querying partial of the memory range, because IMHO it solves the exact problem you're describing and it'll also be in the most efficient way, meanwhile I think it expose all the rest smaps data too besides THP accountings so it could help more than thp accountings. So again, no objection on this simple and working patch, but perhaps rephrasing the 2nd bullet as: "smaps is slow because it must read the whole memory range rather than a small range we care"? Thanks, -- Peter Xu