From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964831AbWDZPpy (ORCPT ); Wed, 26 Apr 2006 11:45:54 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964832AbWDZPpy (ORCPT ); Wed, 26 Apr 2006 11:45:54 -0400 Received: from gold.veritas.com ([143.127.12.110]:7075 "EHLO gold.veritas.com") by vger.kernel.org with ESMTP id S964831AbWDZPpx (ORCPT ); Wed, 26 Apr 2006 11:45:53 -0400 X-IronPort-AV: i="4.04,158,1144047600"; d="scan'208"; a="58969643:sNHT33862192" Date: Wed, 26 Apr 2006 16:45:47 +0100 (BST) From: Hugh Dickins X-X-Sender: hugh@blonde.wat.veritas.com To: Keir Fraser cc: Jan Beulich , Zachary Amsden , linux-kernel@vger.kernel.org Subject: Re: [PATCH] i386: PAE entries must have their low word cleared first In-Reply-To: <946b367619cfd3dcd3ba547e216e494b@cl.cam.ac.uk> Message-ID: References: <444F95D8.76E4.0078.0@novell.com> <946b367619cfd3dcd3ba547e216e494b@cl.cam.ac.uk> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-OriginalArrivalTime: 26 Apr 2006 15:45:51.0869 (UTC) FILETIME=[7F2A7AD0:01C66948] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 26 Apr 2006, Keir Fraser wrote: > > We cannot use pte_clear() unless we redefine it for PAE. Currently it reduces > to set_pte() which explicitly uses the wrong ordering (sets high *then* low, > because it's normally used to introduce a mapping). I overlooked that reversal completely. What a very good point. I think that actually pte_clear() _does_ need to be redefined for PAE, to reverse that ordering as you point out. Take a look at its use in mm/highmem.c (where a comment states it's safe against speculative execution, but a comment can't guarantee that!): what do you think? Hugh