On 6/8/25 22:44, Andrea Righi via Devel wrote:
Introduce apci-generic-initiator device to the domain XML.
Example definition:
<acpi-generic-initiator> <pci-dev>dev0</pci-dev> <numa-node>1</numa-node> </acpi-generic-initiator>
This enables partitioning of PCI resources into multiple isolated instances, each requiring a dedicated NUMA node definition, that can be represented by the acpi-generic-initiator object.
Signed-off-by: Andrea Righi <arighi@nvidia.com> --- src/ch/ch_domain.c | 1 + src/conf/domain_conf.c | 122 ++++++++++++++++++++++++++++-- src/conf/domain_conf.h | 5 +- src/conf/domain_postparse.c | 1 + src/conf/domain_validate.c | 37 +++++++++ src/conf/schemas/domaincommon.rng | 7 +- src/hyperv/hyperv_driver.c | 1 + src/libxl/libxl_driver.c | 6 ++ src/lxc/lxc_driver.c | 6 ++ src/qemu/qemu_alias.c | 11 +++ src/qemu/qemu_command.c | 1 + src/qemu/qemu_domain.c | 2 + src/qemu/qemu_domain_address.c | 4 + src/qemu/qemu_driver.c | 3 + src/qemu/qemu_hotplug.c | 5 ++ src/qemu/qemu_postparse.c | 1 + src/qemu/qemu_validate.c | 1 + src/test/test_driver.c | 4 + 18 files changed, 205 insertions(+), 13 deletions(-)
diff --git a/src/qemu/qemu_domain_address.c b/src/qemu/qemu_domain_address.c index e89cdee487..573bbc282c 100644 --- a/src/qemu/qemu_domain_address.c +++ b/src/qemu/qemu_domain_address.c @@ -471,6 +471,7 @@ qemuDomainDeviceSupportZPCI(virDomainDeviceDef *device) case VIR_DOMAIN_DEVICE_AUDIO: case VIR_DOMAIN_DEVICE_CRYPTO: case VIR_DOMAIN_DEVICE_PSTORE: + case VIR_DOMAIN_DEVICE_ACPI_INITIATOR: break;
case VIR_DOMAIN_DEVICE_NONE: @@ -819,6 +820,9 @@ qemuDomainDeviceCalculatePCIConnectFlags(virDomainDeviceDef *dev, return pciFlags; }
+ case VIR_DOMAIN_DEVICE_ACPI_INITIATOR: + return pciFlags; +
This doesn't make much sense to me. In QEMU this is just an object, not a device. IOW, it's not 'guest visible' and thus doesn't connect to any bus really.
case VIR_DOMAIN_DEVICE_MEMBALLOON: switch (dev->data.memballoon->model) { case VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO_TRANSITIONAL:
Michal