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=-0.8 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_PASS autolearn=ham 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 66FDAECDFB8 for ; Wed, 18 Jul 2018 09:56:45 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id F34C02075C for ; Wed, 18 Jul 2018 09:56:44 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org F34C02075C Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726774AbeGRKds (ORCPT ); Wed, 18 Jul 2018 06:33:48 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:53256 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1726064AbeGRKds (ORCPT ); Wed, 18 Jul 2018 06:33:48 -0400 Received: from smtp.corp.redhat.com (int-mx04.intmail.prod.int.rdu2.redhat.com [10.11.54.4]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id D71538315E; Wed, 18 Jul 2018 09:56:39 +0000 (UTC) Received: from [10.36.117.49] (ovpn-117-49.ams2.redhat.com [10.36.117.49]) by smtp.corp.redhat.com (Postfix) with ESMTP id AE6102026D69; Wed, 18 Jul 2018 09:56:30 +0000 (UTC) Subject: Re: [PATCH v1 00/10] mm: online/offline 4MB chunks controlled by device driver To: Michal Hocko Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org, Alexander Potapenko , Andrew Morton , Andrey Ryabinin , Balbir Singh , Baoquan He , Benjamin Herrenschmidt , Boris Ostrovsky , Dan Williams , Dave Young , Dmitry Vyukov , Greg Kroah-Hartman , Hari Bathini , Huang Ying , Hugh Dickins , Ingo Molnar , Jaewon Kim , Jan Kara , =?UTF-8?B?SsOpcsO0bWUgR2xpc3Nl?= , Joonsoo Kim , Juergen Gross , Kate Stewart , "Kirill A. Shutemov" , Matthew Wilcox , Mel Gorman , Michael Ellerman , Miles Chen , Oscar Salvador , Paul Mackerras , Pavel Tatashin , Philippe Ombredanne , Rashmica Gupta , Reza Arbab , Souptick Joarder , Tetsuo Handa , Thomas Gleixner , Vlastimil Babka References: <20180524093121.GZ20441@dhcp22.suse.cz> <20180524120341.GF20441@dhcp22.suse.cz> <1a03ac4e-9185-ce8e-a672-c747c3e40ff2@redhat.com> <20180524142241.GJ20441@dhcp22.suse.cz> <819e45c5-6ae3-1dff-3f1d-c0411b6e2e1d@redhat.com> <3748f033-f349-6d88-d189-d77c76565981@redhat.com> <20180611115641.GL13364@dhcp22.suse.cz> <71bd1b65-2a88-5de7-9789-bf4fac26507d@redhat.com> <20180716200517.GA16803@dhcp22.suse.cz> From: David Hildenbrand Openpgp: preference=signencrypt Autocrypt: addr=david@redhat.com; prefer-encrypt=mutual; keydata= xsFNBFXLn5EBEAC+zYvAFJxCBY9Tr1xZgcESmxVNI/0ffzE/ZQOiHJl6mGkmA1R7/uUpiCjJ dBrn+lhhOYjjNefFQou6478faXE6o2AhmebqT4KiQoUQFV4R7y1KMEKoSyy8hQaK1umALTdL QZLQMzNE74ap+GDK0wnacPQFpcG1AE9RMq3aeErY5tujekBS32jfC/7AnH7I0v1v1TbbK3Gp XNeiN4QroO+5qaSr0ID2sz5jtBLRb15RMre27E1ImpaIv2Jw8NJgW0k/D1RyKCwaTsgRdwuK Kx/Y91XuSBdz0uOyU/S8kM1+ag0wvsGlpBVxRR/xw/E8M7TEwuCZQArqqTCmkG6HGcXFT0V9 PXFNNgV5jXMQRwU0O/ztJIQqsE5LsUomE//bLwzj9IVsaQpKDqW6TAPjcdBDPLHvriq7kGjt WhVhdl0qEYB8lkBEU7V2Yb+SYhmhpDrti9Fq1EsmhiHSkxJcGREoMK/63r9WLZYI3+4W2rAc UucZa4OT27U5ZISjNg3Ev0rxU5UH2/pT4wJCfxwocmqaRr6UYmrtZmND89X0KigoFD/XSeVv jwBRNjPAubK9/k5NoRrYqztM9W6sJqrH8+UWZ1Idd/DdmogJh0gNC0+N42Za9yBRURfIdKSb B3JfpUqcWwE7vUaYrHG1nw54pLUoPG6sAA7Mehl3nd4pZUALHwARAQABzSREYXZpZCBIaWxk ZW5icmFuZCA8ZGF2aWRAcmVkaGF0LmNvbT7CwX4EEwECACgFAljj9eoCGwMFCQlmAYAGCwkI BwMCBhUIAgkKCwQWAgMBAh4BAheAAAoJEE3eEPcA/4Na5IIP/3T/FIQMxIfNzZshIq687qgG 8UbspuE/YSUDdv7r5szYTK6KPTlqN8NAcSfheywbuYD9A4ZeSBWD3/NAVUdrCaRP2IvFyELj xoMvfJccbq45BxzgEspg/bVahNbyuBpLBVjVWwRtFCUEXkyazksSv8pdTMAs9IucChvFmmq3 jJ2vlaz9lYt/lxN246fIVceckPMiUveimngvXZw21VOAhfQ+/sofXF8JCFv2mFcBDoa7eYob s0FLpmqFaeNRHAlzMWgSsP80qx5nWWEvRLdKWi533N2vC/EyunN3HcBwVrXH4hxRBMco3jvM m8VKLKao9wKj82qSivUnkPIwsAGNPdFoPbgghCQiBjBe6A75Z2xHFrzo7t1jg7nQfIyNC7ez MZBJ59sqA9EDMEJPlLNIeJmqslXPjmMFnE7Mby/+335WJYDulsRybN+W5rLT5aMvhC6x6POK z55fMNKrMASCzBJum2Fwjf/VnuGRYkhKCqqZ8gJ3OvmR50tInDV2jZ1DQgc3i550T5JDpToh dPBxZocIhzg+MBSRDXcJmHOx/7nQm3iQ6iLuwmXsRC6f5FbFefk9EjuTKcLMvBsEx+2DEx0E UnmJ4hVg7u1PQ+2Oy+Lh/opK/BDiqlQ8Pz2jiXv5xkECvr/3Sv59hlOCZMOaiLTTjtOIU7Tq 7ut6OL64oAq+zsFNBFXLn5EBEADn1959INH2cwYJv0tsxf5MUCghCj/CA/lc/LMthqQ773ga uB9mN+F1rE9cyyXb6jyOGn+GUjMbnq1o121Vm0+neKHUCBtHyseBfDXHA6m4B3mUTWo13nid 0e4AM71r0DS8+KYh6zvweLX/LL5kQS9GQeT+QNroXcC1NzWbitts6TZ+IrPOwT1hfB4WNC+X 2n4AzDqp3+ILiVST2DT4VBc11Gz6jijpC/KI5Al8ZDhRwG47LUiuQmt3yqrmN63V9wzaPhC+ xbwIsNZlLUvuRnmBPkTJwwrFRZvwu5GPHNndBjVpAfaSTOfppyKBTccu2AXJXWAE1Xjh6GOC 8mlFjZwLxWFqdPHR1n2aPVgoiTLk34LR/bXO+e0GpzFXT7enwyvFFFyAS0Nk1q/7EChPcbRb hJqEBpRNZemxmg55zC3GLvgLKd5A09MOM2BrMea+l0FUR+PuTenh2YmnmLRTro6eZ/qYwWkC u8FFIw4pT0OUDMyLgi+GI1aMpVogTZJ70FgV0pUAlpmrzk/bLbRkF3TwgucpyPtcpmQtTkWS gDS50QG9DR/1As3LLLcNkwJBZzBG6PWbvcOyrwMQUF1nl4SSPV0LLH63+BrrHasfJzxKXzqg rW28CTAE2x8qi7e/6M/+XXhrsMYG+uaViM7n2je3qKe7ofum3s4vq7oFCPsOgwARAQABwsFl BBgBAgAPBQJVy5+RAhsMBQkJZgGAAAoJEE3eEPcA/4NagOsP/jPoIBb/iXVbM+fmSHOjEshl KMwEl/m5iLj3iHnHPVLBUWrXPdS7iQijJA/VLxjnFknhaS60hkUNWexDMxVVP/6lbOrs4bDZ NEWDMktAeqJaFtxackPszlcpRVkAs6Msn9tu8hlvB517pyUgvuD7ZS9gGOMmYwFQDyytpepo YApVV00P0u3AaE0Cj/o71STqGJKZxcVhPaZ+LR+UCBZOyKfEyq+ZN311VpOJZ1IvTExf+S/5 lqnciDtbO3I4Wq0ArLX1gs1q1XlXLaVaA3yVqeC8E7kOchDNinD3hJS4OX0e1gdsx/e6COvy qNg5aL5n0Kl4fcVqM0LdIhsubVs4eiNCa5XMSYpXmVi3HAuFyg9dN+x8thSwI836FoMASwOl C7tHsTjnSGufB+D7F7ZBT61BffNBBIm1KdMxcxqLUVXpBQHHlGkbwI+3Ye+nE6HmZH7IwLwV W+Ajl7oYF+jeKaH4DZFtgLYGLtZ1LDwKPjX7VAsa4Yx7S5+EBAaZGxK510MjIx6SGrZWBrrV TEvdV00F2MnQoeXKzD7O4WFbL55hhyGgfWTHwZ457iN9SgYi1JLPqWkZB0JRXIEtjd4JEQcx +8Umfre0Xt4713VxMygW0PnQt5aSQdMD58jHFxTk092mU+yIHj5LeYgvwSgZN4airXk5yRXl SE+xAvmumFBY Organization: Red Hat GmbH Message-ID: Date: Wed, 18 Jul 2018 11:56:29 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.8.0 MIME-Version: 1.0 In-Reply-To: <20180716200517.GA16803@dhcp22.suse.cz> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit X-Scanned-By: MIMEDefang 2.78 on 10.11.54.4 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.1]); Wed, 18 Jul 2018 09:56:40 +0000 (UTC) X-Greylist: inspected by milter-greylist-4.5.16 (mx1.redhat.com [10.11.55.1]); Wed, 18 Jul 2018 09:56:40 +0000 (UTC) for IP:'10.11.54.4' DOMAIN:'int-mx04.intmail.prod.int.rdu2.redhat.com' HELO:'smtp.corp.redhat.com' FROM:'david@redhat.com' RCPT:'' Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 16.07.2018 22:05, Michal Hocko wrote: > On Mon 16-07-18 21:48:59, David Hildenbrand wrote: >> On 11.06.2018 14:33, David Hildenbrand wrote: >>> On 11.06.2018 13:56, Michal Hocko wrote: >>>> On Mon 11-06-18 13:53:49, David Hildenbrand wrote: >>>>> On 24.05.2018 23:07, David Hildenbrand wrote: >>>>>> On 24.05.2018 16:22, Michal Hocko wrote: >>>>>>> I will go over the rest of the email later I just wanted to make this >>>>>>> point clear because I suspect we are talking past each other. >>>>>> >>>>>> It sounds like we are now talking about how to solve the problem. I like >>>>>> that :) >>>>>> >>>>> >>>>> Hi Michal, >>>>> >>>>> did you have time to think about the details of your proposed idea? >>>> >>>> Not really. Sorry about that. It's been busy time. I am planning to >>>> revisit after merge window closes. >>>> >>> >>> Sure no worries, I still have a bunch of other things to work on. But it >>> would be nice to clarify soon in which direction I have to head to get >>> this implemented and upstream (e.g. what I proposed, what you proposed >>> or maybe something different). >>> >> I would really like to make progress here. >> >> I pointed out basic problems/questions with the proposed alternative. I >> think I answered all your questions. But you also said that you are not >> going to accept the current approach. So some decision has to be made. >> >> Although it's very demotivating and frustrating (I hope not all work in >> the MM area will be like this), if there is no guidance on how to >> proceed, I'll have to switch to adding/removing/onlining/offlining whole >> segments. This is not what I want, but maybe this has a higher chance of >> getting reviews/acks. >> >> Understanding that you are busy, please if you make suggestions, follow >> up on responses. > > I plan to get back to this. It's busy time with too many things > happening both upstream and on my work table as well. Sorry about that. > I do understand your frustration but there is only that much time I > have. There are not that many people to review this code unfortunately. > > In principle though, I still maintain my position that the memory > hotplug code is way too subtle to add more on top. Maybe the code can be > reworked to be less section oriented but that will be a lot of work. > If you _really_ need a smaller granularity I do not have a better > suggestion than to emulate that on top of sections. I still have to go > back to your last emails though. > The only way I see doing the stuff on top will be using a new bit for marking pages as offline (PageOffline - Patch 1). When a section is added, all pages are initialized to PageOffline. online_pages() can be then hindered to online specific pages using the well known hook set_online_page_callback(). In my driver, I can manually "soft offline" parts, setting them to PageOffline or "soft online" them again (including clearing PageOffline). offline_pages() can then skip all pages that are already "soft offline" - PageOffline set - and effectively set the section offline. Without this new bit offline_pages() cannot know if a page is actually offline or simply reserved by some other part of the system. Imagine that all parts of a section are "soft offline". Now I want to offline the section and remove the memory. I would have to temporarily online all pages again, adding them to the buddy in order to properly offline them using offline_pages(). Prone to races as these pages must not be touched. So in summary, PageOffline would have to be used but online_pages/offline_pages would continue calling e.g. notifiers on segment basis. Boils down to patch 1 and another patch that skips pages that are already offline in offline_pages(). Once you have some spare cycles, please let me know what you think or which other alternatives you see. Thanks. -- Thanks, David / dhildenb