From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Renninger Subject: Re: [PATCH 0/2] ACPI: Re-factor and remove ./drivers/acpi/atomicio.[ch] Date: Thu, 3 Nov 2011 10:16:52 +0100 Message-ID: <201111031016.52676.trenn@suse.de> References: <20110929215907.21126.24480.stgit@amt.stowe> <201110311133.31320.trenn@suse.de> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from cantor2.suse.de ([195.135.220.15]:37351 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755605Ab1KCJQz (ORCPT ); Thu, 3 Nov 2011 05:16:55 -0400 In-Reply-To: Sender: linux-acpi-owner@vger.kernel.org List-Id: linux-acpi@vger.kernel.org To: Bjorn Helgaas Cc: Myron Stowe , Len Brown , bondd@us.ibm.com, lenb@kernel.org, linux-acpi@vger.kernel.org, rjw@sisk.pl, ying.huang@intel.com On Monday, October 31, 2011 04:51:07 PM Bjorn Helgaas wrote: > On Mon, Oct 31, 2011 at 4:33 AM, Thomas Renninger wrote: > Seems like these are BIOS bugs. Do we know for sure that Windows > consumes this information that seems to be wrong? Have you had a > conversation with the vendor about whether the BIOS is at fault here? Such closed specifications between a major OS and specific HW vendors should be forbidden by law and I expect in some countries you'll win if you contest this contract in a high enough court... APEI is based on the Windows WHEA specification which only specific vendors can retrieve from Windows if they sign an NDA contract. I could imagine there you find details about the GAS structure usage in WHEA/APEI tables the way Windows like it. After looking at quite a lot APEI tables and their bit width, byte access and mask values, I am pretty sure bit width is ignored on Windows. Or say, if these tables are used, access width is always correct while bit width is not. > If we make Linux ignore the bit_width, that might "fix" these boxes > with broken BIOSes, but at the cost of breaking a box that uses > bit_width correctly. None of the "broken bit width" boxes I looked at should break if access width is used. Thomas