From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761576Ab3DBJYs (ORCPT ); Tue, 2 Apr 2013 05:24:48 -0400 Received: from cantor2.suse.de ([195.135.220.15]:56385 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754319Ab3DBJYo (ORCPT ); Tue, 2 Apr 2013 05:24:44 -0400 Date: Tue, 2 Apr 2013 11:24:41 +0200 From: Michal Hocko To: Naoya Horiguchi Cc: Andrew Morton , Mel Gorman , Hugh Dickins , Rik van Riel , KOSAKI Motohiro , stable@vger.kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 2/2] hugetlbfs: add swap entry check in follow_hugetlb_page() Message-ID: <20130402092441.GE24345@dhcp22.suse.cz> References: <1364485358-8745-1-git-send-email-n-horiguchi@ah.jp.nec.com> <1364485358-8745-3-git-send-email-n-horiguchi@ah.jp.nec.com> <20130329135730.GB21879@dhcp22.suse.cz> <1364577818-615ipxeo-mutt-n-horiguchi@ah.jp.nec.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1364577818-615ipxeo-mutt-n-horiguchi@ah.jp.nec.com> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Fri 29-03-13 13:23:38, Naoya Horiguchi wrote: > Hi, > > On Fri, Mar 29, 2013 at 02:57:30PM +0100, Michal Hocko wrote: > > On Thu 28-03-13 11:42:38, Naoya Horiguchi wrote: > > [...] > > > @@ -2968,7 +2968,8 @@ long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma, > > > * first, for the page indexing below to work. > > > */ > > > pte = huge_pte_offset(mm, vaddr & huge_page_mask(h)); > > > - absent = !pte || huge_pte_none(huge_ptep_get(pte)); > > > + absent = !pte || huge_pte_none(huge_ptep_get(pte)) || > > > + is_swap_pte(huge_ptep_get(pte)); > > > > is_swap_pte doesn't seem right. Shouldn't you use is_hugetlb_entry_hwpoisoned > > instead? > > I tested only hwpoisoned hugepage, but the same can happen for hugepages > under migration. So I intended to filter out all types of swap entries. > The local variable 'absent' seems to mean whether data on the address > is immediately available, so swap type entry isn't included in it. OK, I didn't consider huge pages under migration and I was merely worried that is_hugetlb_entry_hwpoisoned sounds more appropriate than is_swap_pte. Could you add a comment which would clarify that is_swap_pte covers both migration and hwpoison pages, please? Something like: /* * is_swap_pte test covers both is_hugetlb_entry_hwpoisoned * and hugepages under migration in which case * hugetlb_fault waits for the migration and bails out * properly for HWPosined pages. */ absent = !pte || huge_pte_none(huge_ptep_get(pte)) || is_swap_pte(huge_ptep_get(pte)); Other than that feel free to add Reviewed-by: Michal Hocko Thanks. -- Michal Hocko SUSE Labs