Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752072AbdDIBpI (ORCPT ); Sat, 8 Apr 2017 21:45:08 -0400 Received: from mail-pf0-f195.google.com ([209.85.192.195]:36228 "EHLO mail-pf0-f195.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751066AbdDIBpA (ORCPT ); Sat, 8 Apr 2017 21:45:00 -0400 Date: Sun, 9 Apr 2017 09:44:57 +0800 From: Wei Yang To: Michal Hocko Cc: Wei Yang , mgorman@techsingularity.net, jiang.liu@linux.intel.com, akpm@linux-foundation.org, tj@kernel.org, mingo@kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [RFC] calc_memmap_size() isn't accurate and one suggestion to improve Message-ID: <20170409014457.GA24681@WeideMBP.lan> Reply-To: Wei Yang References: <20170328011137.GA8655@WeideMacBook-Pro.local> <20170403091818.GI24661@dhcp22.suse.cz> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="x+6KMIRAuhnl3hBn" Content-Disposition: inline In-Reply-To: <20170403091818.GI24661@dhcp22.suse.cz> User-Agent: Mutt/1.7.2 (2016-11-26) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2195 Lines: 74 --x+6KMIRAuhnl3hBn Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Mon, Apr 03, 2017 at 11:18:19AM +0200, Michal Hocko wrote: >On Tue 28-03-17 09:11:37, Wei Yang wrote: >> Hi, masters, >>=20 >> # What I found >>=20 >> I found the function calc_memmap_size() may not be that accurate to get = the >> pages for memmap. >>=20 >> The reason is: >>=20 >> > memmap is allocated on a node base, >> > while the calculation is on a zone base >>=20 >> This applies both to SPARSEMEM and FLATMEM. >>=20 >> For example, on my laptop with 6G memory, all the memmap space is alloca= ted >> from ZONE_NORMAL. > >Please try to be more specific. Why is this a problem? Are you trying to >fix some bad behavior or you want to make it more optimal? > >I am sorry I didn't look closer into your proposal but I am quite busy >and other people are probably in a similar situation. If you want to get >a proper feedback please try to state the problem and be explicit if it >is user observable. Michal Glad to hear from you. Sure, let me do more investigation on this and try some experiment to see whether this change is observable. Have a nice day~ >--=20 >Michal Hocko >SUSE Labs --=20 Wei Yang Help you, Help me --x+6KMIRAuhnl3hBn Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJY6ZIZAAoJEKcLNpZP5cTd+IQP+wWiO+ewscdHWxBJkLaoaL7w NWym36GE1vARSpDgTUU9k6WcTWVFz0XnNDB5h3jSPmbD25R8vQao1MEt3eWxbiUB jRQevU6S/l5ng+aJvjjSyuRNSFzJr/iDWZpjJaLemVNDhFlNNFcVgS5ilVNOxtfr OgT4qB0olfoHwQMjq+4xSxDgpzWTeA8TAQPlKjvQLrYEBUu9yAJ3ZwqYks70a0Th 27/gWGso+O5MJ2xSPjl0sbpGWy3UDPf/zr1xHqmiMxU8WJtAXP1IYr98KCKZ+PQA SDudRvRa5CHZmIKM2dY4TAeg30lgO+aqKtXWpD9S8L+UiL6wSbgDuB0UcqVPg49e uQy4fqDoZasHjOhZ41QtOwOgNlMDGyL9Fo2kMX5Jn42hawdxsL8/LWt3ZQ7yLsSN Yz5luQ7zyT6gM8vItdgWH7qc9j4cazZjLnujiegD2c42GI3gkl/RQd+2kJl9CXwU uTJHPIBwKV89ow38PtfiGgmPnb7x+RNw79saQRfMGBxs6Jyz4CMAWFfb8X5tBl+V hlZBM4Su4G6SQ9/emEfgHA5hp0rSOMf6nCfVtHH9/38za6ev6g4jPmMyxLeQIZ2R kyiziwvXF9DuhcJmlugkrQdY+2xauzuMTGYhg+VOqwvu/oLWE8RYONbhREB63vzY wEfttHQB1hazhK1TpeNw =b96m -----END PGP SIGNATURE----- --x+6KMIRAuhnl3hBn--