From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:58819) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SPbvl-0005fB-KI for qemu-devel@nongnu.org; Wed, 02 May 2012 11:57:30 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1SPbvj-0004LF-S4 for qemu-devel@nongnu.org; Wed, 02 May 2012 11:57:29 -0400 Received: from mx1.redhat.com ([209.132.183.28]:43673) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SPbvj-0004L9-Jc for qemu-devel@nongnu.org; Wed, 02 May 2012 11:57:27 -0400 Received: from int-mx01.intmail.prod.int.phx2.redhat.com (int-mx01.intmail.prod.int.phx2.redhat.com [10.5.11.11]) by mx1.redhat.com (8.14.4/8.14.4) with ESMTP id q42FvPVH007045 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK) for ; Wed, 2 May 2012 11:57:26 -0400 Message-ID: <4FA15963.7060608@redhat.com> Date: Wed, 02 May 2012 18:57:23 +0300 From: Avi Kivity MIME-Version: 1.0 References: <1335967335-30197-1-git-send-email-kraxel@redhat.com> <4FA153D4.2020102@redhat.com> <4FA158CA.8060808@redhat.com> In-Reply-To: <4FA158CA.8060808@redhat.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 0/2] pci: 64bit bits List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gerd Hoffmann Cc: qemu-devel@nongnu.org On 05/02/2012 06:54 PM, Gerd Hoffmann wrote: > On 05/02/12 17:33, Avi Kivity wrote: > > On 05/02/2012 05:02 PM, Gerd Hoffmann wrote: > >> Hi, > >> > >> seabios (master branch) just got 64bit pci support. When running out of > >> address space in the pci memory window below 4G (0xe0000000+) seabios > >> will map 64bit pci bars above 4G to make room below 4G. > >> > >> This patch series carries two little patches for qemu to adapt it to the > >> seabios changes. First patch creates a memory window for the 64bit pci > >> bars. Second patch adds a 64bit option to the ivshmem driver, which > >> allows to use huge shared memory chunks. > > > > What happens if a pre-64-bit-pci guest boots on such a setup? > > If it worked before it should continue to work just fine. Note that the > 64bit bars are only mapped above 4G in case there is not enough address > space below 4G, so existing setups should never ever see bars mapped high. I see. I guess it depends on the algorithm for selecting the BAR to be evicted, but if everything fitted before, it should fit now as well. If it didn't, then a different device may break in each of the two cases. -- error compiling committee.c: too many arguments to function