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 031B3C433F5 for ; Fri, 5 Nov 2021 02:51:40 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id CA93E611EE for ; Fri, 5 Nov 2021 02:51:39 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232040AbhKECyR (ORCPT ); Thu, 4 Nov 2021 22:54:17 -0400 Received: from mga12.intel.com ([192.55.52.136]:23036 "EHLO mga12.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230176AbhKECyQ (ORCPT ); Thu, 4 Nov 2021 22:54:16 -0400 X-IronPort-AV: E=McAfee;i="6200,9189,10158"; a="211885833" X-IronPort-AV: E=Sophos;i="5.87,210,1631602800"; d="scan'208";a="211885833" Received: from orsmga008.jf.intel.com ([10.7.209.65]) by fmsmga106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Nov 2021 19:51:36 -0700 X-IronPort-AV: E=Sophos;i="5.87,210,1631602800"; d="scan'208";a="501791069" Received: from yhuang6-desk2.sh.intel.com (HELO yhuang6-desk2.ccr.corp.intel.com) ([10.239.159.101]) by orsmga008-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Nov 2021 19:51:33 -0700 From: "Huang, Ying" To: Dave Hansen , Baolin Wang Cc: , , , , , , Subject: Re: [RFC PATCH] mm: migrate: Add new node demotion strategy References: <665cb882-6dbc-335f-1435-e52659d7ee58@intel.com> Date: Fri, 05 Nov 2021 10:51:31 +0800 In-Reply-To: <665cb882-6dbc-335f-1435-e52659d7ee58@intel.com> (Dave Hansen's message of "Thu, 4 Nov 2021 08:18:02 -0700") Message-ID: <87tugrxqks.fsf@yhuang6-desk2.ccr.corp.intel.com> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=ascii Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Dave Hansen writes: > On 11/4/21 2:13 AM, Baolin Wang wrote: >> +What: /sys/kernel/mm/numa/demotion_mode >> +Date: November 2021 >> +Contact: Linux memory management mailing list >> +Description: Set the demotion mode when enabling demoting pages during reclaim > > I don't think we need a tunable for this. The existing behavior is just > stupid for your hardware and can be replaced. Yes. I think so too. I don't think DIRECT_DEMOTION is reasonable for your system. > Let's also try to do it with the existing node_demotion[] data > structure before we go adding more. To avoid cache ping-pong, I guess some kind of per-CPU data structure may be more suitable for interleaving among multiple nodes. Best Regards, Huang, Ying