From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756621AbXIENWc (ORCPT ); Wed, 5 Sep 2007 09:22:32 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754137AbXIENWZ (ORCPT ); Wed, 5 Sep 2007 09:22:25 -0400 Received: from 207.47.60.147.static.nextweb.net ([207.47.60.147]:31415 "EHLO rpc.xensource.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753219AbXIENWY (ORCPT ); Wed, 5 Sep 2007 09:22:24 -0400 X-Greylist: delayed 974 seconds by postgrey-1.27 at vger.kernel.org; Wed, 05 Sep 2007 09:22:24 EDT Subject: Re: New x86-Setup code breaks HVM-XEN boot From: Ian Campbell To: Christian Ehrhardt Cc: Jeremy Fitzhardinge , linux-kernel@vger.kernel.org, hpa@zytor.com In-Reply-To: <20070831075434.GF20195@genua.de> References: <20070830141528.GA307@lisa.in-ulm.de> <46D714A9.4060103@goop.org> <20070831075434.GF20195@genua.de> Content-Type: text/plain Date: Wed, 05 Sep 2007 14:06:05 +0100 Message-Id: <1188997566.11625.16.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.10.3 Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 05 Sep 2007 13:06:09.0125 (UTC) FILETIME=[86B58950:01C7EFBD] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, 2007-08-31 at 09:54 +0200, Christian Ehrhardt wrote: > (XEN) (GUEST: 353) Detected Xen v3.0.3-1 I was unable to reproduce this problem on Xen 3.1.0 although I can't immediately see a particular changeset which obviously fixed it. There have been plenty of fixes to the emulator since 3.0.3 though so it's worth trying a newer version. In a later mail: > The problem is with the lgdt instruction. Apparently XEN does not keep > the contents of the 48-bit gdt_48 data structure that is passed to > lgdt in the XEN machine state. Instead it appears to save the > _address_ of the 48-bit descriptor somewhere. Unfortunately this data > happens to reside on the stack and is probably no longer availiable at > the time of the actual protected mode jump. The emulation of lgdt (in tools/firmware/vmxassist/vm86.c) looks sane enough on first glance (i.e. it saves the base and length not the pointer) although it isn't an area of the code I'm particularly familiar with. Ian.