On Wed, 2014-10-01 at 15:38 +0200, Andre Richter wrote:
> If a PCIe device bound to vfio-pci happens to be SR-IOV capabale,
> there is no possibility to bring up/shutdown the device's VFs.
>
> This patch adds a generic callback for the sysfs sriov_numvfs attribute.
> The attribute will only show up for SR-IOV devices. Additionally,
> each utilized pci_* function checks if the device is a PF.
>
> Signed-off-by: Andre Richter <andre.o.richter@xxxxxxxxx>
> ---
> drivers/vfio/pci/vfio_pci.c | 25 ++++++++++++++++++++-----
> 1 file changed, 20 insertions(+), 5 deletions(-)
I don't understand why you'd want to do this. The PF is typically
managed by a host driver. By allowing vfio-pci to manage SR-IOV, don't
we potentially have assign-able PFs managing assign-able VFs? That
seems like a bad idea. What use case do you have in mind? Thanks,
Alex
> diff --git a/drivers/vfio/pci/vfio_pci.c b/drivers/vfio/pci/vfio_pci.c
> index f782533..2f6dfb3 100644
> --- a/drivers/vfio/pci/vfio_pci.c
> +++ b/drivers/vfio/pci/vfio_pci.c
> @@ -919,12 +919,27 @@ static struct pci_error_handlers vfio_err_handlers = {
> .error_detected = vfio_pci_aer_err_detected,
> };
>
> +static int vfio_pci_sriov_configure(struct pci_dev *pdev, int num_vfs)
> +{
> + int ret = 0;
> +
> + if (num_vfs > 0) {
> + ret = pci_enable_sriov(pdev, num_vfs);
> + if (!ret)
> + ret = pci_num_vf(pdev);
> + } else
> + pci_disable_sriov(pdev);
> +
> + return ret;
> +}
> +
> static struct pci_driver vfio_pci_driver = {
> - .name = "vfio-pci",
> - .id_table = NULL, /* only dynamic ids */
> - .probe = vfio_pci_probe,
> - .remove = vfio_pci_remove,
> - .err_handler = &vfio_err_handlers,
> + .name = "vfio-pci",
> + .id_table = NULL, /* only dynamic ids */
> + .probe = vfio_pci_probe,
> + .remove = vfio_pci_remove,
> + .sriov_configure = vfio_pci_sriov_configure,
> + .err_handler = &vfio_err_handlers,
> };
>
> /*
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
