From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757325Ab1AMRmE (ORCPT ); Thu, 13 Jan 2011 12:42:04 -0500 Received: from cavan.codon.org.uk ([93.93.128.6]:53125 "EHLO cavan.codon.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756943Ab1AMRmC (ORCPT ); Thu, 13 Jan 2011 12:42:02 -0500 Date: Thu, 13 Jan 2011 17:41:39 +0000 From: Matthew Garrett To: "Eric W. Biederman" Cc: Takao Indoh , Kenji Kaneshige , horms@verge.net.au, nhorman@tuxdriver.com, kexec@lists.infradead.org, linux-kernel@vger.kernel.org, mingo@redhat.com, hpa@zytor.com, tglx@linutronix.de, vgoyal@redhat.com Subject: Re: [PATCH v2][EFI] Run EFI in physical mode Message-ID: <20110113174139.GB22823@srcf.ucam.org> References: <4D06E7FE.5000600@jp.fujitsu.com> <9CB9BB8ED64BAindou.takao@jp.fujitsu.com> <20110113163838.GA21289@srcf.ucam.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.18 (2008-05-17) X-SA-Exim-Connect-IP: X-SA-Exim-Mail-From: mjg59@cavan.codon.org.uk X-SA-Exim-Scanned: No (on cavan.codon.org.uk); SAEximRunCond expanded to false Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Jan 13, 2011 at 09:16:26AM -0800, Eric W. Biederman wrote: > Matthew Garrett writes: > > > We're seeing issues with some Dell hardware when EFI is run in physical > > mode - it seems that Windows uses virtual mode, so the physical path is > > never exercised. I'm not sure that changing the default to physical is > > going to be possible without compatibility issues. > > The bootloaders use physical mode, and there shouldn't be a separate > physical vs virtual mode. It should be the same code with or without > relocations applied. At worst we can set virtual address to the > physical address, on buggy boards. Haven't we learned that failing to mimic Windows as closely as possible in firmware interactions is a recipe for immense disaster yet? We'll never successfully blacklist every broken machine. > Honestly if the problem is the EFI code is so buggy we can't use it > in physical mode the answer is almost certainly not to use the > buggy EFI code. Which would be fine, except... > The only bits that I don't think have an alternative are the firmware > variable accesses. Which are a nice to have, but not a show stopper > anywhere. It's not typically possible to set up the bootloader without EFI variable acccess. -- Matthew Garrett | mjg59@srcf.ucam.org