Attaching physical PCI devices to guests

updated: 2019-08-23 19:37

Attaching physical PCI devices to guests

Attaching physical PCI devices to guests¶

The PCI passthrough feature in OpenStack allows full access and direct control of a physical PCI device in guests. This mechanism is generic for any kind of PCI devices (for example: Network Interface Card (NIC), Graphics Processing Unit (GPU) or any other devices that can be attached to a PCI bus). Correct driver installation is the only requirement for the guest to properly use the devices.

PCI devices, if supported by the hardware, can provide Single Root I/O Virtualization and Sharing (SR-IOV) functionality. In this case, a physical device is virtualized and appears as multiple PCI devices. Virtual PCI devices are assigned to the same or different guests. In the case of PCI passthrough, the full physical device is assigned to only one guest and cannot be shared.

Note

For Attaching virtual SR-IOV devices to guests, refer to the Networking Guide.

To enable PCI passthrough, follow the steps below:

Configure nova-scheduler (Controller)
Configure nova-api (Controller)**
Configure a flavor (Controller)
Enable PCI passthrough (Compute)
Configure PCI devices in nova-compute (Compute)

Note

The PCI device with address 0000:41:00.0 is as an example. Expect to change this according to your actual environment.

Configure nova-scheduler (Controller)¶

Configure nova-scheduler as specified in Configure nova-scheduler.
Restart the nova-scheduler service.

Configure nova-api (Controller)¶

Specify the PCI alias for the device.

Configure a PCI alias a1 to request a PCI device with a vendor_id of 0x8086 and a product_id of 0x154d. The vendor_id and product_id correspond the PCI device with address 0000:41:00.0.

Edit /etc/nova/nova.conf:
```
[default]
pci_alias = { "vendor_id":"8086", "product_id":"154d", "device_type":"type-PF", "name":"a1" }
```
For more information about the syntax of pci_alias, refer to nova.conf configuration options.
Restart the nova-api service.

Configure a flavor (Controller)¶

Configure a flavor to request two PCI devices, each with vendor_id as 0x8086 and product_id as 0x154d.

# openstack flavor set m1.large --property "pci_passthrough:alias"="a1:2"

For more information about the syntax for pci_passthrough:alias, refer to flavor.

Enable PCI passthrough (Compute)¶

Enable VT-d and IOMMU. For more information, refer to steps one and two in Create Virtual Functions.

Configure PCI devices `nova-compute` (Compute)¶

Configure nova-compute to allow the PCI device to be passed through to VMs. Edit /etc/nova/nova.conf:
```
[default]
pci_passthrough_whitelist = { "address": "0000:41:00.0" }
```
Alternatively specify multiple PCI devices using whitelisting:
```
[default]
pci_passthrough_whitelist = { "vendor_id": "8086", "product_id": "10fb" }
```
All PCI devices matching the vendor_id and product_id are added to the pool of PCI devices available for passthrough to VMs.

For more information about the syntax of pci_passthrough_whitelist, refer to nova.conf configuration options.
Specify the PCI alias for the device.

From the Newton release, to resize guest with PCI device, configure the PCI alias on the compute node as well.

Configure a PCI alias a1 to request a PCI device with a vendor_id of 0x8086 and a product_id of 0x154d. The vendor_id and product_id correspond the PCI device with address 0000:41:00.0.

Edit /etc/nova/nova.conf:
```
[default]
pci_alias = { "vendor_id":"8086", "product_id":"154d", "device_type":"type-PF", "name":"a1" }
```
For more information about the syntax of pci_alias, refer to nova.conf configuration options.
Restart the nova-compute service.

Create instances with PCI passthrough devices¶

The nova-scheduler selects a destination host that has PCI devices available with the specified vendor_id and product_id that matches the pci_alias from the flavor.

# openstack server create --flavor m1.large --image cirros-0.3.4-x86_64-uec --wait test-pci

updated: 2019-08-23 19:37

Except where otherwise noted, this document is licensed under Creative Commons Attribution 3.0 License. See all OpenStack Legal Documents.