Re: [PATCH v4 5/9] rust: device: Introduce PropertyGuard

[Rob Herring <robh@kernel.org>]

On Mon, May 05, 2025 at 05:53:33PM +0200, Remo Senekowitsch wrote:
> On Mon May 5, 2025 at 5:37 PM CEST, Rob Herring wrote:
> > On Mon, May 5, 2025 at 8:02 AM Remo Senekowitsch <remo@buenzli.dev> wrote:
> >>
> >> On Mon May 5, 2025 at 7:14 AM CEST, Dirk Behme wrote:
> >> > On 04/05/2025 19:31, Remo Senekowitsch wrote:
> >> >> This abstraction is a way to force users to specify whether a property
> >> >> is supposed to be required or not. This allows us to move error
> >> >> logging of missing required properties into core, preventing a lot of
> >> >> boilerplate in drivers.
> >> >>
> >> >> It will be used by upcoming methods for reading device properties.
> >> >>
> >> >> Signed-off-by: Remo Senekowitsch <remo@buenzli.dev>
> >> >> ---
> >> >>  rust/kernel/device/property.rs | 59 ++++++++++++++++++++++++++++++++++
> >> >>  1 file changed, 59 insertions(+)
> >> >>
> >> >> diff --git a/rust/kernel/device/property.rs b/rust/kernel/device/property.rs
> >> >> index 6ccc7947f9c31..59c61e2493831 100644
> >> >> --- a/rust/kernel/device/property.rs
> >> >> +++ b/rust/kernel/device/property.rs
> >> >> @@ -123,3 +123,62 @@ unsafe fn dec_ref(obj: ptr::NonNull<Self>) {
> >> >>          unsafe { bindings::fwnode_handle_put(obj.cast().as_ptr()) }
> >> >>      }
> >> >>  }
> >> >> +
> >> >> +/// A helper for reading device properties.
> >> >> +///
> >> >> +/// Use [`Self::required_by`] if a missing property is considered a bug and
> >> >> +/// [`Self::optional`] otherwise.
> >> >> +///
> >> >> +/// For convenience, [`Self::or`] and [`Self::or_default`] are provided.
> >> >> +pub struct PropertyGuard<'fwnode, 'name, T> {
> >> >> +    /// The result of reading the property.
> >> >> +    inner: Result<T>,
> >> >> +    /// The fwnode of the property, used for logging in the "required" case.
> >> >> +    fwnode: &'fwnode FwNode,
> >> >> +    /// The name of the property, used for logging in the "required" case.
> >> >> +    name: &'name CStr,
> >> >> +}
> >> >> +
> >> >> +impl<T> PropertyGuard<'_, '_, T> {
> >> >> +    /// Access the property, indicating it is required.
> >> >> +    ///
> >> >> +    /// If the property is not present, the error is automatically logged. If a
> >> >> +    /// missing property is not an error, use [`Self::optional`] instead. The
> >> >> +    /// device is required to associate the log with it.
> >> >> +    pub fn required_by(self, dev: &super::Device) -> Result<T> {
> >> >> +        if self.inner.is_err() {
> >> >> +            dev_err!(
> >> >> +                dev,
> >> >> +                "{}: property '{}' is missing\n",
> >> >> +                self.fwnode.display_path(),
> >> >> +                self.name
> >> >> +            );
> >> >> +        }
> >> >> +        self.inner
> >> >> +    }
> >> >
> >> > Thinking about the .required_by(dev) I wonder if there will be cases
> >> > where we do *not* have a device? I.e. where we really have a fwnode,
> >> > only. And therefore can't pass a device. If we have such cases do we
> >> > need to be able to pass e.g. Option(dev) and switch back to pr_err() in
> >> > case of None?
> >>
> >> In that case, bringing back the previous .required() method seems
> >> reasonable to me. But only if we definitely know such cases exist.
> >
> > They definitely exist. Any property in a child node of the device's
> > node when the child itself is not another device for example.
> 
> I don't think that counts, because you do have a device in that
> situation. The log should be assicated with that. So callers are
> responsible to propagate a reference to the device to wherever the call
> to .required_by(dev) is happening.

Ah, right. So it would just be cases that aren't a driver at all. That's 
limited to the OF_DECLARE cases. I agree we can worry about those later.

Rob