On Thu, Aug 05, 2021 at 10:01:46PM -0300, Jason Gunthorpe wrote: > On Thu, Aug 05, 2021 at 11:07:47AM -0600, Alex Williamson wrote: > > diff --git a/drivers/vfio/vfio.c b/drivers/vfio/vfio.c > > index 1e4fc69fee7d..42ca93be152a 100644 > > +++ b/drivers/vfio/vfio.c > > @@ -875,6 +875,22 @@ struct vfio_device *vfio_device_get_from_dev(struct device *dev) > > } > > EXPORT_SYMBOL_GPL(vfio_device_get_from_dev); > > > > +static const struct file_operations vfio_device_fops; > > + > > +int vfio_device_vma_to_pfn(struct vfio_device *device, > > + struct vm_area_struct *vma, unsigned long *pfn) > > A comment here describing the locking conditions the caller must meet > would be a good addition.. It looks like this can only work under the > i_mmap_lock and the returned pfn can only be taken outside that lock > if it is placed in a VMA > > Maybe this is not a great API then? Should it be 'populate vma' and > call io_remap_pfn_range under the op? Yes, I think that would be a better API.
