On Wed, Mar 10, 2021 at 08:36:34AM +0100, Christoph Hellwig wrote:
> On Tue, Mar 09, 2021 at 05:38:51PM -0400, Jason Gunthorpe wrote:
> > This tidies a few confused places that think they can have a refcount on
> > the vfio_device but the device_data could be NULL, that isn't possible by
> > design.
> >
> > Most of the change falls out when struct vfio_devices is updated to just
> > store the struct vfio_pci_device itself. This wasn't possible before
> > because there was no easy way to get from the 'struct vfio_pci_device' to
> > the 'struct vfio_device' to put back the refcount.
> >
> > Signed-off-by: Jason Gunthorpe <jgg@xxxxxxxxxx>
> > drivers/vfio/pci/vfio_pci.c | 45 ++++++++++++-------------------------
> > 1 file changed, 14 insertions(+), 31 deletions(-)
> >
> > diff --git a/drivers/vfio/pci/vfio_pci.c b/drivers/vfio/pci/vfio_pci.c
> > index af5696a96a76e0..4b0d60f7602e40 100644
> > +++ b/drivers/vfio/pci/vfio_pci.c
> > @@ -534,7 +534,7 @@ static struct vfio_pci_device *get_pf_vdev(struct vfio_pci_device *vdev,
> > return NULL;
> > }
> >
> > - return vfio_device_data(*pf_dev);
> > + return container_of(*pf_dev, struct vfio_pci_device, vdev);
>
> I think it would be useful to just return the vfio_device and let
> the caller do the container_of() here, maybe as a followup.
The callers seem to need the vfio_pci_device *?
In a later series this function gets transformed into this:
device_lock(&physfn->dev);
vdev = vfio_pci_get_drvdata(physfn);
if (!vdev) {
device_unlock(&physfn->dev);
return NULL;
}
vfio_device_get(&vdev->vdev);
device_unlock(&physfn->dev);
There is no container_of here because the drvdata now points at the
struct vfio_pci_device
Jason
