Re: [PATCH 1/4] rust: device: support `dev_printk` on all devices

["Gary Guo" <gary@garyguo.net>]

On Tue Jan 20, 2026 at 6:56 PM GMT, Danilo Krummrich wrote:
> On Tue Jan 20, 2026 at 7:11 PM CET, Gary Guo wrote:
>> diff --git a/rust/kernel/device.rs b/rust/kernel/device.rs
>> index c7b5db9dcca1..94e0548e7687 100644
>> --- a/rust/kernel/device.rs
>> +++ b/rust/kernel/device.rs
>> @@ -601,6 +601,13 @@ impl DeviceContext for Core {}
>>  impl DeviceContext for CoreInternal {}
>>  impl DeviceContext for Normal {}
>>  
>> +impl<Ctx: DeviceContext> AsRef<Device<Ctx>> for Device<Ctx> {
>> +    #[inline]
>> +    fn as_ref(&self) -> &Device<Ctx> {
>> +        self
>> +    }
>> +}
>> +
>>  /// Convert device references to bus device references.
>>  ///
>>  /// Bus devices can implement this trait to allow abstractions to provide the bus device in
>> @@ -720,7 +727,7 @@ macro_rules! impl_device_context_into_aref {
>>  macro_rules! dev_printk {
>>      ($method:ident, $dev:expr, $($f:tt)*) => {
>>          {
>> -            ($dev).$method($crate::prelude::fmt!($($f)*));
>> +            $crate::device::Device::$method($dev.as_ref(), $crate::prelude::fmt!($($f)*))
>
> At a first glance I wasn't sure if we really want to rely on AsRef when hide it
> in the macro. But on the other hand, if someone would implement AsRef on a class
> or bus device to return some other (unrelated) &Device it would be wrong
> anyways. So I think using AsRef is fine.
>
> I think a lot of people will be very happy about this patch. :)

^ this is basically me getting annoyed by everything being `pdev.as_ref()` when
reviewing code and want to fix the pain :)

>
> I will pick up patches 1 and 2 this cycle, patch 3 and 4 can go through the
> corresponding subsystem tree next cycle.
>
>> diff --git a/rust/kernel/pci.rs b/rust/kernel/pci.rs
>> index bea76ca9c3da..3946ca919332 100644
>> --- a/rust/kernel/pci.rs
>> +++ b/rust/kernel/pci.rs
>> @@ -351,7 +351,7 @@ impl Device {
>>      ///     // Get an instance of `Vendor`.
>>      ///     let vendor = pdev.vendor_id();
>>      ///     dev_info!(
>> -    ///         pdev.as_ref(),
>> +    ///         pdev,
>>      ///         "Device: Vendor={}, Device=0x{:x}\n",
>>      ///         vendor,
>>      ///         pdev.device_id()
>> diff --git a/rust/kernel/pci/id.rs b/rust/kernel/pci/id.rs
>> index c09125946d9e..e2d9e8804347 100644
>> --- a/rust/kernel/pci/id.rs
>> +++ b/rust/kernel/pci/id.rs
>> @@ -22,7 +22,7 @@
>>  /// fn probe_device(pdev: &pci::Device<Core>) -> Result {
>>  ///     let pci_class = pdev.pci_class();
>>  ///     dev_info!(
>> -///         pdev.as_ref(),
>> +///         pdev,
>>  ///         "Detected PCI class: {}\n",
>>  ///         pci_class
>>  ///     );
>
> Can you please send a separate patch for those?

I converted them as they're the users that benefit from this behaviour in the
kernel crate (i.e. "convert the users"). Do you want them separate so you can
take it via a different tree?

Best,
Gary