From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755741AbYHRWO5 (ORCPT ); Mon, 18 Aug 2008 18:14:57 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752928AbYHRWOr (ORCPT ); Mon, 18 Aug 2008 18:14:47 -0400 Received: from terminus.zytor.com ([198.137.202.10]:39703 "EHLO terminus.zytor.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752780AbYHRWOq (ORCPT ); Mon, 18 Aug 2008 18:14:46 -0400 Message-ID: <48A9F36B.8090509@zytor.com> Date: Mon, 18 Aug 2008 15:10:51 -0700 From: "H. Peter Anvin" User-Agent: Thunderbird 2.0.0.14 (X11/20080501) MIME-Version: 1.0 To: Pavel Machek CC: David Fries , "Maciej W. Rozycki" , Ingo Molnar , linux-kernel@vger.kernel.org, Thomas Gleixner , "Rafael J. Wysocki" Subject: Re: [PATCH] Fix i486 suspend to disk CR4 oops References: <20080818040340.GB17528@spacedout.fries.net> <20080818125803.GC17528@spacedout.fries.net> <20080818220419.GB2053@elf.ucw.cz> In-Reply-To: <20080818220419.GB2053@elf.ucw.cz> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Pavel Machek wrote: > > Okay, can it happen that that cr4 is zero legitimately? If newer 486SX > chips support cr4 but not coprocessor...? > Pavel Theoretically it can, but that means no features are enabled, so there is no need to enable the features. The real question is if the following can happen: can it be such that we want CR4 to be zero in a situation where CR4 is nonzero to start out with? The main bit in CR4 that could be set that we wouldn't want set would be CR4.PAE, so this could happen if there is a CPU with CR4.PAE but none of the other CR4 bits that we would normally set unconditionally. I'm pretty sure this can't happen on any physical CPUs, since all physical CPUs supporting PAE would also support DE, MCE, and PGE. It could possibly happen on a virtual CPU, although it is of course extremely unlikely we'd get there with CR4 not zero to start out with. Still, it is at least theoretically wrong. -hpa