From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932067Ab3LCXqq (ORCPT ); Tue, 3 Dec 2013 18:46:46 -0500 Received: from cantor2.suse.de ([195.135.220.15]:53785 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755682Ab3LCXql (ORCPT ); Tue, 3 Dec 2013 18:46:41 -0500 Date: Tue, 3 Dec 2013 23:46:37 +0000 From: Mel Gorman To: Rik van Riel Cc: Alex Thorlton , Linux-MM , LKML Subject: Re: [PATCH 14/15] mm: numa: Flush TLB if NUMA hinting faults race with PTE scan update Message-ID: <20131203234637.GS11295@suse.de> References: <1386060721-3794-1-git-send-email-mgorman@suse.de> <1386060721-3794-15-git-send-email-mgorman@suse.de> <529E641A.7040804@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Disposition: inline In-Reply-To: <529E641A.7040804@redhat.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 Tue, Dec 03, 2013 at 06:07:06PM -0500, Rik van Riel wrote: > On 12/03/2013 03:52 AM, Mel Gorman wrote: > > NUMA PTE updates and NUMA PTE hinting faults can race against each other. The > > setting of the NUMA bit defers the TLB flush to reduce overhead. NUMA > > hinting faults do not flush the TLB as X86 at least does not cache TLB > > entries for !present PTEs. However, in the event that the two race a NUMA > > hinting fault may return with the TLB in an inconsistent state between > > different processors. This patch detects potential for races between the > > NUMA PTE scanner and fault handler and will flush the TLB for the affected > > range if there is a race. > > > > Signed-off-by: Mel Gorman > > > diff --git a/mm/migrate.c b/mm/migrate.c > > index 5dfd552..ccc814b 100644 > > --- a/mm/migrate.c > > +++ b/mm/migrate.c > > @@ -1662,6 +1662,39 @@ void wait_migrate_huge_page(struct anon_vma *anon_vma, pmd_t *pmd) > > smp_rmb(); > > } > > > > +unsigned long numa_fault_prepare(struct mm_struct *mm) > > +{ > > + /* Paired with task_numa_work */ > > + smp_rmb(); > > + return mm->numa_next_reset; > > +} > > The patch that introduces mm->numa_next_reset, and the > patch that increments it, seem to be missing from your > series... > Damn. s/numa_next_reset/numa_next_scan/ in that patch -- Mel Gorman SUSE Labs