Hi Philippe, On 3/20/20 12:26 PM, Philippe Mathieu-Daudé wrote: > On 3/20/20 11:41 AM, Paolo Bonzini wrote: >> The virtio-iommu device attaches itself to a PCI bus, so it makes >> no sense to include it unless PCI is supported---and in fact >> compilation fails without this change. >> >> Reported-by: Gerd Hoffmann <[email protected]> >> Signed-off-by: Paolo Bonzini <[email protected]> >> --- >> hw/virtio/Kconfig | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/hw/virtio/Kconfig b/hw/virtio/Kconfig >> index d29525b36f..83122424fa 100644 >> --- a/hw/virtio/Kconfig >> +++ b/hw/virtio/Kconfig >> @@ -12,7 +12,7 @@ config VIRTIO_RNG >> config VIRTIO_IOMMU >> bool >> default y >> - depends on VIRTIO >> + depends on PCI && VIRTIO >> config VIRTIO_PCI >> bool >> > > Reviewed-by: Philippe Mathieu-Daudé <[email protected]> > > I see hw/virtio/virtio-iommu-pci.c apart, what is the difference? That's the virtio-iommu pci proxy whereas TYPE_VIRTIO_IOMMU inherits from TYPE_VIRTIO_DEVICE.
The virtio-iommu device could also be instantiated directly as a TYPE_VIRTIO_IOMMU device but that's not the choice eventually made for machvirt. Thanks Eric >
