Devel

Download

devel@lists.libvirt.org

July 2026

43 participants
81 discussions

[PATCH v8 0/2] qemu: Support HW-accelerated vSMMUs
by Nathan Chen 01 Jul '26

01 Jul '26

Hi, This is a follow up to the seventh patchset [0] for supporting multiple hardware-accelerated vSMMU instances. For instance, specifying hostdevs associated with multiple accelerated SMMUs, configured to be routed to pcie-expander-bus controllers in a way where VFIO device to SMMUv3 associations are matched with the host: <devices> ... <controller type='pci' index='1' model='pcie-expander-bus'> <model name='pxb-pcie'/> <target busNr='252'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x01' function='0x0'/> </controller> <controller type='pci' index='2' model='pcie-expander-bus'> <model name='pxb-pcie'/> <target busNr='248'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/> </controller> ... <controller type='pci' index='21' model='pcie-root-port'> <model name='pcie-root-port'/> <target chassis='21' port='0x0'/> <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/> </controller> <controller type='pci' index='22' model='pcie-root-port'> <model name='pcie-root-port'/> <target chassis='22' port='0xa8'/> <address type='pci' domain='0x0000' bus='0x02' slot='0x00' function='0x0'/> </controller> ... <hostdev mode='subsystem' type='pci' managed='no'> <source> <address domain='0x0009' bus='0x01' slot='0x00' function='0x0'/> </source> <address type='pci' domain='0x0000' bus='0x15' slot='0x00' function='0x0'/> </hostdev> <hostdev mode='subsystem' type='pci' managed='no'> <source> <address domain='0x0019' bus='0x01' slot='0x00' function='0x0'/> </source> <address type='pci' domain='0x0000' bus='0x16' slot='0x00' function='0x0'/> </hostdev> <iommu model='smmuv3'> <driver pciBus='1' accel='on' ats='on' ril='off' ssidSize='20' oas='44'/> </iommu> <iommu model='smmuv3'> <driver pciBus='2' accel='on' ats='on' ril='off' ssidSize='20' oas='44'/> </iommu> </devices> This would get translated to a qemu command line with the arguments below: -device '{"driver":"pxb-pcie","bus_nr":252,"id":"pci.1","bus":"pcie.0","addr":"0x1"}' \ -device '{"driver":"pxb-pcie","bus_nr":248,"id":"pci.2","bus":"pcie.0","addr":"0x2"}' \ -device '{"driver":"pcie-root-port","port":0,"chassis":21,"id":"pci.21","bus":"pci.1","addr":"0x0"}' \ -device '{"driver":"pcie-root-port","port":168,"chassis":22,"id":"pci.22","bus":"pci.2","addr":"0x0"}' \ -device '{"driver":"arm-smmuv3","primary-bus":"pci.1","id":"iommu0","accel":true,"ats":true,"ril":false,"ssidsize":20,"oas":44}' \ -device '{"driver":"arm-smmuv3","primary-bus":"pci.2","id":"iommu1","accel":true,"ats":true,"ril":false,"ssidsize":20,"oas":44}' \ -device '{"driver":"vfio-pci","host":"0009:01:00.0","id":"hostdev0","bus":"pci.21","addr":"0x0"}' \ -device '{"driver":"vfio-pci","host":"0019:01:00.0","id":"hostdev1","bus":"pci.22","addr":"0x0"}' \ Two QEMU series implementing auto support for accel SMMUv3 properties are now present in the upstream master branch [1][2]. This Libvirt series leverages the auto support: users can omit SMMUv3 properties ATS, RIL, OAS, and SSIDSIZE on the domain XML, and they will also be omitted on the QEMU command line as a result. With these new QEMU series applied, the feature property values will default to 'auto' to be resolved based on what the host SMMU supports. This series is on Github: https://github.com/NathanChenNVIDIA/libvirt/tree/smmuv3-accel-v8/ Thanks, Nathan [0] https://lists.libvirt.org/archives/list/devel@lists.libvirt.org/thread/5GG7… [1] https://lore.kernel.org/all/20260323182454.1416110-1-nathanc@nvidia.com/ [2] https://lore.kernel.org/all/20260608174900.2227340-1-nathanc@nvidia.com/ Changes from v7: - Remove support for x-vpasid-cap-offset based on feedback pointing out the QEMU feature prefixed with x- is experimental and not to be supported in Libvirt - Implement support for auto SMMUv3 properties. Omitting these properties in the XML omits them on the QEMU command line as well, which results in default auto values being chosen for ATS, RIL, SSIDSIZE, and OAS - Add validation check that only allows enabling accel SMMUv3 features when accel is enabled - Add tests for accel=on but omitting accel SMMUv3 feature properties Changes from v6: - Remove smmuv3 'pasid' attribute - Add smmuv3 'ssidsize' attribute - Add support for 'vpasidCapOffset' element - Modify logic for building accel smmuv3 qemu command line to omit ATS, RIL, accel flags when not needed (enabled/disabled by default in QEMU SMMUv3) Nathan Chen (2): qemu: Add support for HW-accelerated nested SMMUv3 tests: qemuxmlconfdata: provide HW-accel smmuv3 sample XML and CLI args docs/formatdomain.rst | 29 +++++++ src/conf/domain_conf.c | 80 +++++++++++++++++++ src/conf/domain_conf.h | 5 ++ src/conf/domain_validate.c | 27 ++++++- src/conf/schemas/domaincommon.rng | 25 ++++++ src/qemu/qemu_command.c | 5 ++ ...uv3-pci-bus-accel-auto.aarch64-latest.args | 38 +++++++++ ...muv3-pci-bus-accel-auto.aarch64-latest.xml | 63 +++++++++++++++ .../iommu-smmuv3-pci-bus-accel-auto.xml | 53 ++++++++++++ ...u-smmuv3-pci-bus-accel.aarch64-latest.args | 38 +++++++++ ...mu-smmuv3-pci-bus-accel.aarch64-latest.xml | 63 +++++++++++++++ .../iommu-smmuv3-pci-bus-accel.xml | 53 ++++++++++++ tests/qemuxmlconftest.c | 2 + 13 files changed, 480 insertions(+), 1 deletion(-) create mode 100644 tests/qemuxmlconfdata/iommu-smmuv3-pci-bus-accel-auto.aarch64-latest.args create mode 100644 tests/qemuxmlconfdata/iommu-smmuv3-pci-bus-accel-auto.aarch64-latest.xml create mode 100644 tests/qemuxmlconfdata/iommu-smmuv3-pci-bus-accel-auto.xml create mode 100644 tests/qemuxmlconfdata/iommu-smmuv3-pci-bus-accel.aarch64-latest.args create mode 100644 tests/qemuxmlconfdata/iommu-smmuv3-pci-bus-accel.aarch64-latest.xml create mode 100644 tests/qemuxmlconfdata/iommu-smmuv3-pci-bus-accel.xml -- 2.43.0

1 2