I have runned the following two tests, but both tests failed: [1] start a VM with virtio-iommu + 2 ice PFs only via qemu-kvm 8.1.5 Test result : the qemu-kvm keeps throwing the error: VFIO_MAP_DMA failed: File exists. vfio_dma_map(0x56443d20fbe0, 0xffffe000, 0x1000, 0x7fb545709000) = -17 (File exists) [2] start a VM with virtio-iommu + 2 ice PFs via libvirt-9.5 + qemu-kvm 8.1.5 Test result: the qemu-kvm core dump with ERROR:../qom/object.c:1198:object_unref: assertion failed: (obj->ref > 0). Bail out! ERROR:../qom/object.c:1198:object_unref: assertion failed: (obj->ref > 0)
After removing the 2 PF from the VM, both tests passed. Tested-by: Yanghang Liu <yangh...@redhat.com> Best Regards, YangHang Liu On Mon, Sep 4, 2023 at 4:08 PM Eric Auger <eric.au...@redhat.com> wrote: > > On x86, when assigning VFIO-PCI devices protected with virtio-iommu > we encounter the case where the guest tries to map IOVAs beyond 48b > whereas the physical VTD IOMMU only supports 48b. This ends up with > VFIO_MAP_DMA failures at qemu level because at kernel level, > vfio_iommu_iova_dma_valid() check returns false on vfio_map_do_map(). > > This is due to the fact the virtio-iommu currently unconditionally > exposes an IOVA range of 64b through its config input range fields. > > This series removes this assumption by retrieving the usable IOVA > regions through the VFIO_IOMMU_TYPE1_INFO_CAP_IOVA_RANGE UAPI when > a VFIO device is attached. This info is communicated to the > virtio-iommu memory region, transformed into the inversed info, ie. > the host reserved IOVA regions. Then those latter are combined with the > reserved IOVA regions set though the virtio-iommu reserved-regions > property. That way, the guest virtio-iommu driver, unchanged, is > able to probe the whole set of reserved regions and prevent any IOVA > belonging to those ranges from beeing used, achieving the original goal. > > Best Regards > > Eric > > This series can be found at: > https://github.com/eauger/qemu/tree/virtio-iommu_geometry_v1 > > Eric Auger (13): > memory: Let ReservedRegion use Range > memory: Introduce memory_region_iommu_set_iova_ranges > vfio: Collect container iova range info > virtio-iommu: Rename reserved_regions into prop_resv_regions > virtio-iommu: Introduce per IOMMUDevice reserved regions > range: Introduce range_inverse_array() > virtio-iommu: Implement set_iova_ranges() callback > range: Make range_compare() public > util/reserved-region: Add new ReservedRegion helpers > virtio-iommu: Consolidate host reserved regions and property set ones > test: Add some tests for range and resv-mem helpers > virtio-iommu: Resize memory region according to the max iova info > vfio: Remove 64-bit IOVA address space assumption > > include/exec/memory.h | 30 ++++- > include/hw/vfio/vfio-common.h | 2 + > include/hw/virtio/virtio-iommu.h | 7 +- > include/qemu/range.h | 9 ++ > include/qemu/reserved-region.h | 32 +++++ > hw/core/qdev-properties-system.c | 9 +- > hw/vfio/common.c | 70 ++++++++--- > hw/virtio/virtio-iommu-pci.c | 8 +- > hw/virtio/virtio-iommu.c | 85 +++++++++++-- > softmmu/memory.c | 15 +++ > tests/unit/test-resv-mem.c | 198 +++++++++++++++++++++++++++++++ > util/range.c | 41 ++++++- > util/reserved-region.c | 94 +++++++++++++++ > hw/virtio/trace-events | 1 + > tests/unit/meson.build | 1 + > util/meson.build | 1 + > 16 files changed, 562 insertions(+), 41 deletions(-) > create mode 100644 include/qemu/reserved-region.h > create mode 100644 tests/unit/test-resv-mem.c > create mode 100644 util/reserved-region.c > > -- > 2.41.0 > >
