On Fri, Feb 12, 2010 at 06:52:38PM +0530, Amit Shah wrote: > Michael noted we don't allow disabling of MSI for the virtio-serial-pci > device. Fix that. > > Signed-off-by: Amit Shah <amit.s...@redhat.com> > CC: "Michael S. Tsirkin" <m...@redhat.com>
How about using a symbolic constant for -1? We have NIC_NVECTORS_UNSPECIFIED in net.h, generalize it to DEV_NVECTORS_UNSPECIFIED? Acked-by: Michael S. Tsirkin <m...@redhat.com> As a side note, I think directly allocating a ton of vectors before some benchmark shows it's a measureable win for serial is a mistake. But that's a subject for another patch. > --- > hw/virtio-pci.c | 6 +++--- > 1 files changed, 3 insertions(+), 3 deletions(-) > > diff --git a/hw/virtio-pci.c b/hw/virtio-pci.c > index f3373ae..bcd40f7 100644 > --- a/hw/virtio-pci.c > +++ b/hw/virtio-pci.c > @@ -500,8 +500,8 @@ static int virtio_serial_init_pci(PCIDevice *pci_dev) > if (!vdev) { > return -1; > } > - vdev->nvectors = proxy->nvectors ? proxy->nvectors > - : proxy->max_virtserial_ports + 1; > + vdev->nvectors = proxy->nvectors == -1 ? proxy->max_virtserial_ports + 1 > + : proxy->nvectors; > virtio_init_pci(proxy, vdev, > PCI_VENDOR_ID_REDHAT_QUMRANET, > PCI_DEVICE_ID_VIRTIO_CONSOLE, > @@ -585,7 +585,7 @@ static PCIDeviceInfo virtio_info[] = { > .init = virtio_serial_init_pci, > .exit = virtio_exit_pci, > .qdev.props = (Property[]) { > - DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 0), > + DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, -1), > DEFINE_PROP_HEX32("class", VirtIOPCIProxy, class_code, 0), > DEFINE_VIRTIO_COMMON_FEATURES(VirtIOPCIProxy, host_features), > DEFINE_PROP_UINT32("max_ports", VirtIOPCIProxy, > max_virtserial_ports, > -- > 1.6.2.5