Il 27/09/2012 11:16, Avi Kivity ha scritto: > needs to be a little more complicated. After translation, we need to > look up the address again in a different phys_map. So a MemoryRegion > that is an iommu needs to hold its own phys_map pointer for the lookup. > > But let's ignore the problem for now, we have too much on our plate. > With a recursive big lock, there is no problem with iommus, yes? So as > long as there is no intersection between converted devices and platforms > with iommus, we're safe.
pSeries has both PCI and iommu, but I think no one should be using e1000 on a pSeries. virtio-net is per-spec not going through the iommu. Paolo