On Sun, May 01, 2016 at 04:47:44PM +0300, Michael S. Tsirkin wrote: > On Sat, Apr 30, 2016 at 01:42:39AM +0300, David Kiarie wrote: > > These series adds AMD IOMMU support to Qemu. It's currently in the 9th > > version. > > > > In this series I have (hopefully) addressed all the comments made in the > > previous version. > > changelog? > > > I have also tested and successfully passed-through PCI device 'ac97' with > > more devices to be tested. > > > A fundamental problem with AMD IOMMUs is that the VMM must write-protect > guest I/O page tables from the guest in order to intercept all guest > updates and propagate the updates to the shadow I/O page tables. > > AMD manual says as much.
Actually while it says so, it's wrong. There's an NPcache flag which makes guest to invalidations for invalid to valid transitions. > Until this is implemented, I think device assignment must be disabled > when AMD IOMMU is in use. So I take this back. I would, however, like to see how this interacts with Aviv's patches enabling VFIO support for IOMMU. > > > > > David Kiarie (4): > > hw/i386: Introduce AMD IOMMU > > hw/i386: ACPI table for AMD IOMMU > > hw/core: Add AMD IOMMU to machine properties > > hw/pci-host: Emulate AMD IOMMU > > > > hw/acpi/aml-build.c | 2 +- > > hw/acpi/core.c | 13 - > > hw/core/machine.c | 33 +- > > hw/i386/Makefile.objs | 1 + > > hw/i386/acpi-build.c | 93 ++- > > hw/i386/amd_iommu.c | 1426 > > +++++++++++++++++++++++++++++++++++++++++ > > hw/i386/amd_iommu.h | 398 ++++++++++++ > > hw/pci-host/q35.c | 25 +- > > include/hw/acpi/acpi-defs.h | 14 + > > include/hw/acpi/acpi.h | 16 + > > include/hw/acpi/aml-build.h | 1 + > > include/hw/boards.h | 7 + > > include/hw/i386/intel_iommu.h | 1 + > > include/hw/pci/pci.h | 2 + > > qemu-options.hx | 7 +- > > util/qemu-config.c | 8 +- > > 16 files changed, 2012 insertions(+), 35 deletions(-) > > create mode 100644 hw/i386/amd_iommu.c > > create mode 100644 hw/i386/amd_iommu.h > > > > -- > > 2.1.4
