From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:36019) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VqPMD-0002Ko-Pl for qemu-devel@nongnu.org; Tue, 10 Dec 2013 10:36:27 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VqPM7-00010k-Qb for qemu-devel@nongnu.org; Tue, 10 Dec 2013 10:36:21 -0500 Received: from mx1.redhat.com ([209.132.183.28]:29644) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VqPM7-00010V-J6 for qemu-devel@nongnu.org; Tue, 10 Dec 2013 10:36:15 -0500 Received: from int-mx10.intmail.prod.int.phx2.redhat.com (int-mx10.intmail.prod.int.phx2.redhat.com [10.5.11.23]) by mx1.redhat.com (8.14.4/8.14.4) with ESMTP id rBAFaCF6029536 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK) for ; Tue, 10 Dec 2013 10:36:14 -0500 Message-ID: <1386689767.15686.25.camel@nilsson.home.kraxel.org> From: Gerd Hoffmann Date: Tue, 10 Dec 2013 16:36:07 +0100 In-Reply-To: <52A72C24.3080209@redhat.com> References: <1385401393-14291-1-git-send-email-pbonzini@redhat.com> <20131128102652.GA24261@redhat.com> <52A714AC.3050703@redhat.com> <1386687211.15686.21.camel@nilsson.home.kraxel.org> <52A72C24.3080209@redhat.com> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH rebased for-1.8] i386: pc: align gpa<->hpa on 1GB boundary (v6) List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Marcelo Tosatti , qemu-devel@nongnu.org, "Michael S. Tsirkin" On Di, 2013-12-10 at 15:58 +0100, Paolo Bonzini wrote: > Il 10/12/2013 15:53, Gerd Hoffmann ha scritto: > > Hi, > > > >> If we could make a small guset visible change, it would be simpler to > >> always make the PCI hole 1GB in size; it is currently 256MB for i440FX > >> and 1.25GB for q35. > > > > Easy for i440fx. > > > > Tricky for q35 as the firmware knows qemu will not map ram above > > 0xb000000 and places the mmconfig bar @ 0xb0000000. Making the window > > smaller (1.25GB -> 1GB) will create a conflict there. Making it larger > > (2G) will work. It's done this way to keep 0xc0000000+ free for pci > > bars, and we can map up to 512MB-sized bars there. > > 0xc0000000-0xfebfffff is almost 1GB, so there is room for 1 512MB-size > BAR. mmconfig could move to 0xf0000000-0xf7ffffff; Reduces number of pci busses from 256 to 128. Not that this is a problem now, but something to consider to make sure things are future-proof. > firmware is not a > problem because this would be only for new machine types. Even on qemu 2.0 it is the firmware which decides where to place the mmconfig bar. qemu will look at northbridge xbar register to generate a matching mcfg acpi table. cheers, Gerd