From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932267Ab3DBOo7 (ORCPT ); Tue, 2 Apr 2013 10:44:59 -0400 Received: from mga03.intel.com ([143.182.124.21]:36365 "EHLO mga03.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760166Ab3DBOo5 (ORCPT ); Tue, 2 Apr 2013 10:44:57 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.87,393,1363158000"; d="scan'208";a="221900036" From: "Kirill A. Shutemov" To: Dave Hansen Cc: "Kirill A. Shutemov" , Andrea Arcangeli , Andrew Morton , Al Viro , Hugh Dickins , Wu Fengguang , Jan Kara , Mel Gorman , linux-mm@kvack.org, Andi Kleen , Matthew Wilcox , "Kirill A. Shutemov" , Hillf Danton , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org In-Reply-To: <514C6E68.4050203@sr71.net> References: <1363283435-7666-1-git-send-email-kirill.shutemov@linux.intel.com> <1363283435-7666-8-git-send-email-kirill.shutemov@linux.intel.com> <514B336C.6070404@sr71.net> <20130322101211.34A5EE0085@blue.fi.intel.com> <514C6E68.4050203@sr71.net> Subject: Re: [PATCHv2, RFC 07/30] thp, mm: introduce mapping_can_have_hugepages() predicate Content-Transfer-Encoding: 7bit Message-Id: <20130402144648.47240E0085@blue.fi.intel.com> Date: Tue, 2 Apr 2013 17:46:48 +0300 (EEST) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Dave Hansen wrote: > On 03/22/2013 03:12 AM, Kirill A. Shutemov wrote: > > Dave Hansen wrote: > >> On 03/14/2013 10:50 AM, Kirill A. Shutemov wrote: > >>> +static inline bool mapping_can_have_hugepages(struct address_space *m) > >>> +{ > >>> + if (IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE)) { > >>> + gfp_t gfp_mask = mapping_gfp_mask(m); > >>> + return !!(gfp_mask & __GFP_COMP); > >>> + } > >>> + > >>> + return false; > >>> +} > >> > >> I did a quick search in all your patches and don't see __GFP_COMP > >> getting _set_ anywhere. Am I missing something? > > > > __GFP_COMP is part of GFP_TRANSHUGE. We set it for ramfs in patch 20/30. > > That's a bit non-obvious. For a casual observer, it _seems_ like you > should just be setting and checking GFP_TRANSHUGE directly. It looks > like you were having some problems with __GFP_MOVABLE and masked it out > of GFP_TRANSHUGE and that has cascaded over to _this_ check. Checking GFP_TRANSHUGE directly is not right way. File systems can clear GFP bits or set additional for its own reason. We should not limit file systems here. So the only way robust way is to check __GFP_COMP. I'll add comment. -- Kirill A. Shutemov From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Kirill A. Shutemov" Subject: Re: [PATCHv2, RFC 07/30] thp, mm: introduce mapping_can_have_hugepages() predicate Date: Tue, 2 Apr 2013 17:46:48 +0300 (EEST) Message-ID: <20130402144648.47240E0085@blue.fi.intel.com> References: <1363283435-7666-1-git-send-email-kirill.shutemov@linux.intel.com> <1363283435-7666-8-git-send-email-kirill.shutemov@linux.intel.com> <514B336C.6070404@sr71.net> <20130322101211.34A5EE0085@blue.fi.intel.com> <514C6E68.4050203@sr71.net> Content-Transfer-Encoding: 7bit Cc: "Kirill A. Shutemov" , Andrea Arcangeli , Andrew Morton , Al Viro , Hugh Dickins , Wu Fengguang , Jan Kara , Mel Gorman , linux-mm@kvack.org, Andi Kleen , Matthew Wilcox , "Kirill A. Shutemov" , Hillf Danton , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org To: Dave Hansen Return-path: In-Reply-To: <514C6E68.4050203@sr71.net> Sender: owner-linux-mm@kvack.org List-Id: linux-fsdevel.vger.kernel.org Dave Hansen wrote: > On 03/22/2013 03:12 AM, Kirill A. Shutemov wrote: > > Dave Hansen wrote: > >> On 03/14/2013 10:50 AM, Kirill A. Shutemov wrote: > >>> +static inline bool mapping_can_have_hugepages(struct address_space *m) > >>> +{ > >>> + if (IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE)) { > >>> + gfp_t gfp_mask = mapping_gfp_mask(m); > >>> + return !!(gfp_mask & __GFP_COMP); > >>> + } > >>> + > >>> + return false; > >>> +} > >> > >> I did a quick search in all your patches and don't see __GFP_COMP > >> getting _set_ anywhere. Am I missing something? > > > > __GFP_COMP is part of GFP_TRANSHUGE. We set it for ramfs in patch 20/30. > > That's a bit non-obvious. For a casual observer, it _seems_ like you > should just be setting and checking GFP_TRANSHUGE directly. It looks > like you were having some problems with __GFP_MOVABLE and masked it out > of GFP_TRANSHUGE and that has cascaded over to _this_ check. Checking GFP_TRANSHUGE directly is not right way. File systems can clear GFP bits or set additional for its own reason. We should not limit file systems here. So the only way robust way is to check __GFP_COMP. I'll add comment. -- Kirill A. Shutemov -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org