On 27/06/18 20:59, Michael S. Tsirkin wrote: >> Another reason to keep the MMIO transport option is that one >> virtio-iommu can manage DMA from endpoints on multiple PCI domains at >> the same time, as well as platform devices. Some VMMs might want that, >> in which case the IOMMU would be a separate platform device. > > Which buses are managed by the IOMMU is separate from the bus > on which it's programming interface resides. Sorry I didn't get this. We probably don't want to instantiate a PCI root complex just for the IOMMU, so it needs to be in the same PCI segment as managed endpoints. For example in my VM the AMD IOMMU is presented as 00:02.0, between other devices on PCI bus 00. In any case, I have a solution for virtio-pci that works with DT and ACPI, and isn't excessively awful. I'll probably send it as part of the next version. Thanks, Jean -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
