On 04/25/2013 11:57 AM, Laine Stump wrote:
The device option for vfio-pci is nearly identical to that for pci-assign - only the configfd parameter isn't supported (or needed).
Checking for presence of the bootindex parameter is done separately from constructing the commandline, similar to how it is done for pci-assign. --- src/qemu/qemu_command.c | 48 ++++++++++++++++++++++++++++++++++++++---------- src/qemu/qemu_hotplug.c | 13 ++++++++++++- 2 files changed, 50 insertions(+), 11 deletions(-)
@@ -7850,12 +7855,23 @@ qemuBuildCommandLine(virConnectPtr conn, " supported for PCI and USB devices")); goto error; } else { - if (hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI && - !virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_BOOTINDEX)) { - virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", - _("booting from assigned PCI devices is not" - " supported with this version of qemu")); - goto error; + if (hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI) { + if (hostdev->source.subsys.u.pci.backend + == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_TYPE_VFIO) { + if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VFIO_PCI_BOOTINDEX)) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("booting from PCI devices assigned with VFIO " + "is not supported with this version of qemu"));
Line break after space...
+ goto error; + } + } else { + if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_BOOTINDEX)) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("booting from assigned PCI devices is not" + " supported with this version of qemu"));
...line break before space. Looks a bit inconsistent, but the end result is the same. ACK. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org