From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:55021) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1V7JJX-000600-Cm for qemu-devel@nongnu.org; Thu, 08 Aug 2013 02:03:17 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1V7JJR-0008SV-DN for qemu-devel@nongnu.org; Thu, 08 Aug 2013 02:03:11 -0400 Received: from mx1.redhat.com ([209.132.183.28]:17049) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1V7JJR-0008SO-5M for qemu-devel@nongnu.org; Thu, 08 Aug 2013 02:03:05 -0400 Date: Thu, 8 Aug 2013 09:04:27 +0300 From: "Michael S. Tsirkin" Message-ID: <20130808060427.GB18031@redhat.com> References: <1375167638-4325-1-git-send-email-imammedo@redhat.com> <20130731055959.GA31017@redhat.com> <20130731081459.77eba7bb@nial.usersys.redhat.com> <51FFCCF4.706@redhat.com> <20130805181618.GB4244@redhat.com> <911613672.9763982.1375729901921.JavaMail.root@redhat.com> <20130806143901.GA17072@redhat.com> <1243962588.10286037.1375807384073.JavaMail.root@redhat.com> <20130806165820.GB20305@redhat.com> <5201F763.3030507@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <5201F763.3030507@redhat.com> Subject: Re: [Qemu-devel] [SeaBIOS] [PATCH] acpi: hide 64-bit PCI hole for Windows XP List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gerd Hoffmann Cc: Paolo Bonzini , kevin@koconnor.net, seabios@seabios.org, qemu-devel@nongnu.org, Anthony Liguori On Wed, Aug 07, 2013 at 09:29:39AM +0200, Gerd Hoffmann wrote: > We'll need some way for seabios to fill in the pci window information > into the qemu-provided tables. Easiest way to do that would be to > extend the COMMAND_ADD_POINTER bios linker script command. This idea certainly has an advantage: the two patch-sets (to control PCI hole from QEMU, and to pass ACPI tables from QEMU) would become independent. One difficulty would be coming up with a sane interface that's not tied to specific AML code: unlike table pointers which have a specific fixed-width format, we are talking about generic AML code here. Patching that works (we do it today with the ACPI_EXTRACT code) but requires that you code AML in a specific way, for example, number encoding is variable-length so we pad values ahead of the time making sure the AML encoding is large enough to hold the maximum possible value. Also, in the past both Anthony and Kevin indicated preference to the pci-info solution that we have in QEMU today. -- MST