On Fri, Apr 21, 2023 at 02:58:01PM -0300, Jason Gunthorpe wrote: > When the VM writes it totally-a-lie MSI address to the PCI MSI-X > registers the hypervisor traps it and subsitutes, what it valiantly > hopes, is the right address for the ITS in the VM's S1 IOMMU table > based on the ACPI where it nicely asked the guest to keep this > specific IOVA mapped. > > I'm not sure how the data bit works on ARM.. Not sure if I follow everything here correctly, yet the data seems to be the hwirq idx in its_alloc_device_irq()? So, not only the MSI in the guest is fake, the data also? Thanks Nicolin
