On Wed, Jun 18, 2025 at 08:19:58AM -0500, Rob Herring wrote:
> On Wed, Jun 18, 2025 at 11:13:25AM +0100, Igor Korotin wrote:
> > From: Danilo Krummrich <dakr@xxxxxxxxxx>
> >
> > Only call Self::properties_parse() when the device is compatible with
> > "test,rust-device".
> >
> > Once we add ACPI support, we don't want the ACPI device to fail probing
> > in Self::properties_parse().
> >
> > Signed-off-by: Danilo Krummrich <dakr@xxxxxxxxxx>
>
> This needs your S-o-b as well since you sent the patch.
>
> > ---
> > samples/rust/rust_driver_platform.rs | 7 ++++++-
> > 1 file changed, 6 insertions(+), 1 deletion(-)
> >
> > diff --git a/samples/rust/rust_driver_platform.rs b/samples/rust/rust_driver_platform.rs
> > index 000bb915af60..036dd0b899b0 100644
> > --- a/samples/rust/rust_driver_platform.rs
> > +++ b/samples/rust/rust_driver_platform.rs
> > @@ -40,7 +40,12 @@ fn probe(
> > dev_info!(dev, "Probed with info: '{}'.\n", info.0);
> > }
> >
> > - Self::properties_parse(dev)?;
> > + if dev
> > + .fwnode()
> > + .is_some_and(|node| node.is_compatible(c_str!("test,rust-device")))
>
> I think you should be checking just is this ACPI or DT rather than
> compatible. It's kind of an anti-pattern to test compatible in probe.
> The reason is we've already matched to a compatible and have match data
> to use, so we don't need to do it again. It becomes quite messy when
> there are numerous possible compatibles.
Yeah, that was my first approach; here's the patch from a few days ago [1].
The reason why I decided against this, was that all the properties we check in
Self::properties_parse() in a fallible way *only* apply to the device with this
compatible string.
But I don't mind if we replace it with [1] either.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/dakr/linux.git/commit/?h=rust/is_of_node
