On Thu, Sep 30, 2021 at 10:46:20AM -0600, Alex Williamson wrote: > I'm only aware that the PF driver enables basic SR-IOV configuration of > VFs, ie. the number of enabled VFs. This is quite common in the netdev world, for instance you use the PF driver to set the MAC addresses, QOS and other details on the VF devices. > only management of the number of child devices, but the flavor of each > child, for example the non-homogeneous slice of resources allocated per > child device. Since the devices are PCI VFs they should be able to be used, with configuration, any place a PCI VF is usable. EG vfio-pci, a Kernel driver, etc. This is why the PF needs to provide the configuration to support all the use cases. > I'm not aware of any standard mechanism for a PF driver to apply a > configuration per VF. devlink is the standard way these days. It can model the PCI devices and puts a control point in the PF driver. I'd defer to Jiri, Leon and others to explain the details of this though :) > create these sorts device flavors. For example, we might expose NIC VFs > and administrative configuration should restrict VF1 to a 1Gbit > interface while VF2 gets 10Gbit. This is all being done already today through the PF driver using either netink or devlink interfaces Jason