From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alex Williamson Subject: [RFC PATCH v3 0/3] IGD assignment support for QEMU/vfio Date: Fri, 12 Feb 2016 17:23:01 -0700 Message-ID: <20160213001835.18456.46422.stgit@gimli.home> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Cc: alex.williamson@redhat.com, allen.m.kay@intel.com, qemu-devel@nongnu.org, kvm@vger.kernel.org To: seabios@seabios.org Return-path: Received: from mx1.redhat.com ([209.132.183.28]:55483 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751320AbcBMAXC (ORCPT ); Fri, 12 Feb 2016 19:23:02 -0500 Sender: kvm-owner@vger.kernel.org List-ID: Turns out that maybe we weren't done with IGD by simply reading a new fw_cfg, copying it into a new reserved memory buffer and writing back to the IGD device. We need another reserved memory buffer for stolen memory. The hardware minimum size is 1MB, naturally aligned, which means we need to be able to actually be able to allocate that. We write the address of the buffer back to the BDSM register on IGD so that QEMU can make use of it when doing fixups on vBIOS config of the device. Previous comments on updating the size of the high memory area suggest that any remaining free memory is returned, so it appears there's no affect of increasing this limit for systems without IGD. Thanks, Alex --- Alex Williamson (3): fw/pci: Add support for mapping Intel IGD OpRegion via QEMU Further increase maximum size of permanent high memory area. fw/pci: Allocate IGD stolen memory src/config.h | 2 +- src/fw/pciinit.c | 41 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 42 insertions(+), 1 deletion(-)