On 2017年03月20日 19:38, Paolo Bonzini wrote: > Fair enough, though I'd be worried about increasing the attack surface > of the hypervisor. For KVM, for example, IOMMU emulation requires using > the "split irqchip" feature to move the PIC and IOAPIC out of the kernel > and back to QEMU.
Yes, just like Roger mentioned we also need to support no-qemu mode on Xen and this is tradeoff result. > > Also, I think this series is missing changes to support IOMMU > translation in the vIOMMU device model. Yes, this series just enabled vIOMMU's irq remapping function and we need to pass virtual device's DMA request to Xen hypervisor for translation when enable DMA translation. -- Best regards Tianyu Lan
