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=-1.0 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 64DCEC43441 for ; Thu, 15 Nov 2018 00:50:26 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 328192086A for ; Thu, 15 Nov 2018 00:50:26 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 328192086A Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.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 S1727059AbeKOK4A (ORCPT ); Thu, 15 Nov 2018 05:56:00 -0500 Received: from mga02.intel.com ([134.134.136.20]:14100 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1725895AbeKOKz7 (ORCPT ); Thu, 15 Nov 2018 05:55:59 -0500 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga003.jf.intel.com ([10.7.209.27]) by orsmga101.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 14 Nov 2018 16:50:23 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.56,234,1539673200"; d="scan'208";a="100330509" Received: from ahduyck-mobl.amr.corp.intel.com (HELO [10.251.21.216]) ([10.251.21.216]) by orsmga003.jf.intel.com with ESMTP; 14 Nov 2018 16:50:23 -0800 Subject: Re: [mm PATCH v5 0/7] Deferred page init improvements To: Michal Hocko Cc: akpm@linux-foundation.org, linux-mm@kvack.org, sparclinux@vger.kernel.org, linux-kernel@vger.kernel.org, linux-nvdimm@lists.01.org, davem@davemloft.net, pavel.tatashin@microsoft.com, mingo@kernel.org, kirill.shutemov@linux.intel.com, dan.j.williams@intel.com, dave.jiang@intel.com, rppt@linux.vnet.ibm.com, willy@infradead.org, vbabka@suse.cz, khalid.aziz@oracle.com, ldufour@linux.vnet.ibm.com, mgorman@techsingularity.net, yi.z.zhang@linux.intel.com References: <154145268025.30046.11742652345962594283.stgit@ahduyck-desk1.jf.intel.com> <20181114150742.GZ23419@dhcp22.suse.cz> From: Alexander Duyck Message-ID: <9e8218eb-80bf-fc02-ae56-42ccfddb572e@linux.intel.com> Date: Wed, 14 Nov 2018 16:50:23 -0800 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101 Thunderbird/60.3.0 MIME-Version: 1.0 In-Reply-To: <20181114150742.GZ23419@dhcp22.suse.cz> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 11/14/2018 7:07 AM, Michal Hocko wrote: > On Mon 05-11-18 13:19:25, Alexander Duyck wrote: >> This patchset is essentially a refactor of the page initialization logic >> that is meant to provide for better code reuse while providing a >> significant improvement in deferred page initialization performance. >> >> In my testing on an x86_64 system with 384GB of RAM and 3TB of persistent >> memory per node I have seen the following. In the case of regular memory >> initialization the deferred init time was decreased from 3.75s to 1.06s on >> average. For the persistent memory the initialization time dropped from >> 24.17s to 19.12s on average. This amounts to a 253% improvement for the >> deferred memory initialization performance, and a 26% improvement in the >> persistent memory initialization performance. >> >> I have called out the improvement observed with each patch. > > I have only glanced through the code (there is a lot of the code to look > at here). And I do not like the code duplication and the way how you > make the hotplug special. There shouldn't be any real reason for that > IMHO (e.g. why do we init pfn-at-a-time in early init while we do > pageblock-at-a-time for hotplug). I might be wrong here and the code > reuse might be really hard to achieve though. Actually it isn't so much that hotplug is special. The issue is more that the non-hotplug case is special in that you have to perform a number of extra checks for things that just aren't necessary for the hotplug case. If anything I would probably need a new iterator that would be able to take into account all the checks for the non-hotplug case and then provide ranges of PFNs to initialize. > I am also not impressed by new iterators because this api is quite > complex already. But this is mostly a detail. Yeah, the iterators were mostly an attempt at hiding some of the complexity. Being able to break a loop down to just an iterator provding the start of the range and the number of elements to initialize is pretty easy to visualize, or at least I thought so. > Thing I do not like is that you keep microptimizing PageReserved part > while there shouldn't be anything fundamental about it. We should just > remove it rather than make the code more complex. I fell more and more > guilty to add there actually. I plan to remove it, but don't think I can get to it in this patch set. I was planning to submit one more iteration of this patch set early next week, and then start focusing more on the removal of the PageReserved bit for hotplug. I figure it is probably going to be a full patch set onto itself and as you pointed out at the start of this email there is already enough code to review without adding that.