Re: [PATCH v5 9/9] vfio/pci: Add dma-buf export support for MMIO regions

[Jason Gunthorpe <jgg@nvidia.com>]

On Thu, Oct 30, 2025 at 08:48:18AM +0200, Leon Romanovsky wrote:
> > > +void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
> > > +{
> > > +       struct vfio_pci_dma_buf *priv;
> > > +       struct vfio_pci_dma_buf *tmp;
> > > +
> > > +       lockdep_assert_held_write(&vdev->memory_lock);
> > > +
> > > +       list_for_each_entry_safe(priv, tmp, &vdev->dmabufs, dmabufs_elm) {
> > > +               if (!get_file_active(&priv->dmabuf->file))
> > > +                       continue;
> > > +
> > > +               if (priv->revoked != revoked) {
> > > +                       dma_resv_lock(priv->dmabuf->resv, NULL);
> > > +                       priv->revoked = revoked;
> > > +                       dma_buf_move_notify(priv->dmabuf);
> > 
> > I think this should only be called when revoked is true, otherwise
> > this will be calling move_notify on the already revoked dmabuf
> > attachments.
> 
> This case is protected by "if (priv->revoked)" check both in
> vfio_pci_dma_buf_map and vfio_pci_dma_buf_attach. They will prevent
> DMABUF recreation if revoked is false.

The point was to call it when revoked becomes false as well, as that
gives the importing driver an opportunity to restore any mapping it
had.

Jason