On 21/06/2016 15:50, Joerg Roedel wrote: > The code has a few style issues (thing I'd implemented differently), but > it looks functional. Anyway, before merging this the last 3 patches need > to be acked by the KVM maintainers. > > Paolo?
I think patches 9 and 10 should be squashed because the code after patch 9 is only partly functional. Likewise, I think this: + + if (AMD_IOMMU_GUEST_IR_VAPIC(amd_iommu_guest_ir)) + amd_iommu_irq_ops.capability |= (1 << IRQ_POSTING_CAP); from patch 1 should be moved to patch 6. Otherwise the KVM bits are sane. Paolo
