On 11/17/26 8:32 PM, Alice Ryhl wrote: > From: Sami Tolvanen <[email protected]> > > Tyr's VM_BIND remap path rebuilds page-table mappings for the fragments > that survive a split, and those fragments must keep the protection flags > of the original mapping. Add GpuVa::data_ref() so the remap step can > read the VaData of an existing mapping. > > Signed-off-by: Sami Tolvanen <[email protected]> > [aliceryhl: split into separate gpuvm and iommu commits] > Signed-off-by: Alice Ryhl <[email protected]>
One question below, otherwise: Acked-by: Danilo Krummrich <[email protected]> > --- > rust/kernel/drm/gpuvm/va.rs | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/rust/kernel/drm/gpuvm/va.rs b/rust/kernel/drm/gpuvm/va.rs > index 46493f0ec5a6..f81302faba61 100644 > --- a/rust/kernel/drm/gpuvm/va.rs > +++ b/rust/kernel/drm/gpuvm/va.rs > @@ -87,6 +87,12 @@ pub fn obj(&self) -> &T::Object { > unsafe { <T::Object as > IntoGEMObject>::from_raw((*self.as_raw()).gem.obj) } > } > > + /// Access the driver-private [`VaData`](DriverGpuVm::VaData) associated > with this mapping. > + #[inline] > + pub fn data_ref(&self) -> &T::VaData { > + &self.data > + } > + Why data_ref() instead of just data() or a Deref impl?
