Hi Daniel,
Top level libvirt device representation in XML is based on the device
*class*, not the specific device impl. Adding a device
type XML element in libvirt is totally inappropriate. Any configuration
must be done beneath the element.
Would keeping track of PXB <=> host SMMU nodes b
Add a pcie-expander-bus controller to the VM definition for each "nestedSmmuv3"
device that is generated when the "nestedSmmuv3" IOMMU model is parsed from the
VM definition. Assign each "nestedSmmuv3" device to one PXB controller, and
route any unmanaged "hostdev" VFIO devices with associated
Add support for parsing "nestedSmmuv3" devices from the VM definition. When a
"nestedSmmuv3" IOMMU model is parsed from the VM definition, add a
"nestedSmmuv3" device to the VM definition for each host SMMU node detected from
sysfs. Specify the associated host SMMU sysfs filename as the "name" attr
Implement a sub-test in qemuxmlconftest that mocks the scanning of sysfs to
determine how many "nestedSmmuv3" devices to assign to the VM when
"nestedSmmuv3" IOMMU model is parsed from the VM definition. Add a VM
definition example with an associated qemu command line that exercises the
detection o
Hi,
This is a draft solution for supporting multiple vSMMU instances in a qemu VM.
Based on discussions/suggestions received for a previous RFC by Nicolin here[0],
the association of vSMMUs to VFIO devices in VM PCIe topology should be moved
out of qemu into libvirt. In addition, the nested SMMU
Add a pcie-expander-bus controller to the VM definition for each "nestedSmmuv3"
device that is generated when the "nestedSmmuv3" IOMMU model is parsed from the
VM definition. Assign each "nestedSmmuv3" device to one PXB controller, and
route any unmanaged "hostdev" VFIO devices with associated host
Update the PXB controller busNrs to account for devices we attached in the
previous commit, ensuring there are enough VM bus numbers to be assigned for
each device attached downstream from each PXB controller.
Signed-off-by: Nathan Chen
---
src/qemu/qemu_domain_address.c | 52 +++
Add support for specifying "nestedSmmuv3" as the IOMMU model. In the following
commits, when the "nestedSmmuv3" IOMMU model is parsed from the VM definition,
PXB controllers and "nestedSmmuv3" devices will be auto-added to the VM
definition and VFIO devices will be routed to PXB controllers based o
Hi Daniel,
>Top level libvirt device representation in XML is based on the device
>*class*, not the specific device impl. Adding a device
>type XML element in libvirt is totally inappropriate. Any configuration
>must be done beneath the element.
Would keeping track of PXB <=> host SMMU nodes b
Hi Daniel,
Top level libvirt device representation in XML is based on the device
*class*, not the specific device impl. Adding a device
type XML element in libvirt is totally inappropriate. Any configuration
must be done beneath the element.
Would keeping track of PXB <=> host SMMU nodes be
Hi Daniel,
Top level libvirt device representation in XML is based on the device
*class*, not the specific device impl. Adding a device
type XML element in libvirt is totally inappropriate. Any configuration
must be done beneath the element.
1. Adding a attribute for PXB controller.
2. Havin
On 5/16/2025 3:19 AM, Shameerali Kolothum Thodi wrote:
Hi,
This is a follow up to the first RFC patchset [0] for supporting multiple
vSMMU instances in a qemu VM. This patchset also introduces support for
using iommufd to propagate DMA mappings to kernel for assigned devices.
This patchset i
On 5/20/2025 4:20 AM, Daniel P. Berrangé wrote:
Add support for "smmuv3Dev" IOMMU model, and add
support for parsing multiple IOMMU devices from
the VM definition when "smmuv3Dev" is the IOMMU model.
Enable plugging smmuv3Dev into pcie-root and
pcie-expander-bus.
Signed-off-by: Nathan Chen
--
On 5/20/2025 6:01 AM, Daniel P. Berrangé wrote:
Add support for an "iommufd" virDomainIOMMUDef member that will be
translated to a qemu "-object iommufd" command line argument. This
iommufd struct has an "id" member to specify the iommufd object id
that can be associated with a passthrough dev
Hi Daniel,
On 5/20/2025 5:51 AM, Daniel P. Berrangé wrote:
Hi,
This is a follow up to the first RFC patchset [0] for supporting multiple
vSMMU instances in a qemu VM. This patchset also introduces support for
using iommufd to propagate DMA mappings to kernel for assigned devices.
This patchset
On 5/20/2025 6:05 AM, Daniel P. Berrangé wrote:
Implement "iommufdId" and "iommufdFd" attributes for "hostdev" devices that
can be used to specify associated iommufd object and fd for externally opening
the /dev/iommu and VFIO cdev, when launching a qemu VM.
What does it mean if we enable iom
Hi,
This is a follow up to the first RFC patchset [0] for supporting multiple
vSMMU instances in a qemu VM. This patchset also introduces support for
using iommufd to propagate DMA mappings to kernel for assigned devices.
This patchset implements support for specifying multiple devices
within th
Add support for an "iommufd" virDomainIOMMUDef member that will be
translated to a qemu "-object iommufd" command line argument. This
iommufd struct has an "id" member to specify the iommufd object id
that can be associated with a passthrough device, and an "fd"
member to specify the fd for externa
Implement "iommufdId" and "iommufdFd" attributes for "hostdev" devices that
can be used to specify associated iommufd object and fd for externally opening
the /dev/iommu and VFIO cdev, when launching a qemu VM.
Signed-off-by: Nathan Chen
---
src/conf/domain_conf.c| 31 +++
Add support for "smmuv3Dev" IOMMU model, and add
support for parsing multiple IOMMU devices from
the VM definition when "smmuv3Dev" is the IOMMU model.
Enable plugging smmuv3Dev into pcie-root and
pcie-expander-bus.
Signed-off-by: Nathan Chen
---
docs/formatdomain.rst | 5 +-
src/c
Allow access to /dev/iommu and /dev/vfio/devices/vfio* when launching a qemu
VM with iommufd feature enabled.
Signed-off-by: Nathan Chen
---
src/qemu/qemu_cgroup.c| 47 +++
src/qemu/qemu_cgroup.h| 1 +
src/qemu/qemu_namespace.c | 36 ++
Implement a sub-test in qemuxmlconftest that takes an iommufd attribute
from the iommu device stanza, and iommufdId and iommufdFd hostdev
attributes in a VM definition. Translate the VM definition to a qemu
command line that associates the hostdev withan iommufd object, as well
as a fd for external
On 5/30/2025 8:05 AM, Daniel P. Berrangé wrote:
Hi Daniel,
On 5/20/2025 5:51 AM, Daniel P. Berrangé wrote:
Hi,
This is a follow up to the first RFC patchset [0] for supporting multiple
vSMMU instances in a qemu VM. This patchset also introduces support for
using iommufd to propagate DMA map
On 5/30/2025 6:47 AM, Daniel P. Berrangé wrote:
On 5/20/2025 6:01 AM, Daniel P. Berrangé wrote:
Add support for an "iommufd" virDomainIOMMUDef member that will be
translated to a qemu "-object iommufd" command line argument. This
iommufd struct has an "id" member to specify the iommufd objec
24 matches
Mail list logo
