On Thu, Jul 03, 2025 at 06:40:48AM +0000, Tian, Kevin wrote:
> > From: Jason Gunthorpe <j...@nvidia.com>
> > Sent: Wednesday, June 25, 2025 2:35 AM
> >
> > This was missed during the initial implementation. The VFIO PCI encodes
> > the vf_token inside the device name when opening the device from the
> > group
> > FD, something like:
> >
> > "0000:04:10.0 vf_token=bd8d9d2b-5a5f-4f5a-a211-f591514ba1f3"
> >
> > This is used to control access to a VF unless there is co-ordination with
> > the owner of the PF.
> >
> > Since we no longer have a device name pass the token directly though
> > VFIO_DEVICE_BIND_IOMMUFD with an optional field indicated by
> > VFIO_DEVICE_BIND_TOKEN.
>
> not a complete sentence?
Since we no longer have a device name, pass the token directly through
VFIO_DEVICE_BIND_IOMMUFD using an optional field indicated by
VFIO_DEVICE_BIND_TOKEN.
> > Only users using a PCI SRIOV VF will need to
> > provide this. This is done in the usual backwards compatible way.
>
> and PF also needs to provide it when there are in-use VFs:
>
> vfio_pci_validate_vf_token():
> * When presented with a PF which has VFs in use, the user must also
> * provide the current VF token to prove collaboration with existing
> * VF users. If VFs are not in use, the VF token provided for the PF
> * device will act to set the VF token.
Too complicated, I'll just drop that sentence.
> > @@ -1583,6 +1583,7 @@ static const struct vfio_device_ops
> > hisi_acc_vfio_pci_ops = {
> > .mmap = vfio_pci_core_mmap,
> > .request = vfio_pci_core_request,
> > .match = vfio_pci_core_match,
> > + .match_token_uuid = vfio_pci_core_match_token_uuid,
>
> this matters only when the driver supports SR-IOV. currently only
> vfio-pci does.
Hmm, sriov_pf_core_dev requires sriov_config, but the normal vf_token
happens for all vfs and there is a little debugging related to it:
pci_info_ratelimited(vdev->pdev,
"VF token incorrectly provided, PF not bound to
vfio-pci\n");
So maybe we want to keep it. Otherwise it sounds like you are
proposing to remove match from most of the drivers since they don't
support sriov_configure? Which ever direction I think match and
match_token_uuid should be together.
Jason
