This RFC implements a basic LED abstraction to show how this would work with rust.

Currently this just implements a sample driver, to show how to use the abstraction, which just
prints the requested state, supporting a on/off LED and an led with brightness level up to 255 and
hardware blinking. I intend to write a hardware specific driver for submitting.

The abstractions is a generic struct that holds a generic driver data on which the vtable is
implemented. Because this struct also holds the c led_classdev (include/linux/leds.h) struct this
struct is pinned and is using pin_init to create and directly register the LED.
Dropping the struct unregisteres the LED. I plan to also add devm functions later, but as device
abstractions in rust are not yet that far I opted agains that for the first iteration of the LED
abstractions.

This is currently using core::time::Duration for the blinking interval, but will likely change that
to use the Delta time type from FUJITA Tomonori [1].

This is requiring the Opaque::try_ffi_init patch by Alice Ryhl[2] which just got merged into
char-misc-testing.

[1]: https://lore.kernel.org/rust-for-linux/20241005122531.20298-3-fujita.tomonori@gmail.com/
[2]: https://lore.kernel.org/rust-for-linux/20240926-b4-miscdevice-v1-1-7349c2b2837a@google.com/

Fiona Behrens (2):
  rust: LED abstraction
  samples: rust: led sample

 rust/kernel/leds.rs      | 399 +++++++++++++++++++++++++++++++++++++++
 rust/kernel/lib.rs       |   2 +
 samples/rust/Kconfig     |  10 +
 samples/rust/Makefile    |   1 +
 samples/rust/rust_led.rs | 103 ++++++++++
 5 files changed, 515 insertions(+)
 create mode 100644 rust/kernel/leds.rs
 create mode 100644 samples/rust/rust_led.rs

-- 
2.46.0

On Wed, 09 Oct 2024, Fiona Behrens wrote:

> This RFC implements a basic LED abstraction to show how this would work with rust.
> 
> Currently this just implements a sample driver, to show how to use the abstraction, which just
> prints the requested state, supporting a on/off LED and an led with brightness level up to 255 and
> hardware blinking. I intend to write a hardware specific driver for submitting.
> 
> The abstractions is a generic struct that holds a generic driver data on which the vtable is
> implemented. Because this struct also holds the c led_classdev (include/linux/leds.h) struct this
> struct is pinned and is using pin_init to create and directly register the LED.
> Dropping the struct unregisteres the LED. I plan to also add devm functions later, but as device
> abstractions in rust are not yet that far I opted agains that for the first iteration of the LED
> abstractions.
> 
> This is currently using core::time::Duration for the blinking interval, but will likely change that
> to use the Delta time type from FUJITA Tomonori [1].
> 
> This is requiring the Opaque::try_ffi_init patch by Alice Ryhl[2] which just got merged into
> char-misc-testing.
> 
> [1]: https://lore.kernel.org/rust-for-linux/20241005122531.20298-3-fujita.tomonori@gmail.com/
> [2]: https://lore.kernel.org/rust-for-linux/20240926-b4-miscdevice-v1-1-7349c2b2837a@google.com/
> 
> Fiona Behrens (2):
>   rust: LED abstraction
>   samples: rust: led sample
> 
>  rust/kernel/leds.rs      | 399 +++++++++++++++++++++++++++++++++++++++
>  rust/kernel/lib.rs       |   2 +
>  samples/rust/Kconfig     |  10 +
>  samples/rust/Makefile    |   1 +
>  samples/rust/rust_led.rs | 103 ++++++++++
>  5 files changed, 515 insertions(+)
>  create mode 100644 rust/kernel/leds.rs
>  create mode 100644 samples/rust/rust_led.rs

FYI: I'm not ignoring this patch-set.  On the contrary.  I'm trying to
place myself into a position where I can not only review it with some
confidence, but use it to author LED drivers.

-- 
Lee Jones [李琼斯]

On 11 Nov 2024, at 10:41, Lee Jones wrote:

> On Wed, 09 Oct 2024, Fiona Behrens wrote:
>
>> This RFC implements a basic LED abstraction to show how this would work with rust.
>>
>> Currently this just implements a sample driver, to show how to use the abstraction, which just
>> prints the requested state, supporting a on/off LED and an led with brightness level up to 255 and
>> hardware blinking. I intend to write a hardware specific driver for submitting.
>>
>> The abstractions is a generic struct that holds a generic driver data on which the vtable is
>> implemented. Because this struct also holds the c led_classdev (include/linux/leds.h) struct this
>> struct is pinned and is using pin_init to create and directly register the LED.
>> Dropping the struct unregisteres the LED. I plan to also add devm functions later, but as device
>> abstractions in rust are not yet that far I opted agains that for the first iteration of the LED
>> abstractions.
>>
>> This is currently using core::time::Duration for the blinking interval, but will likely change that
>> to use the Delta time type from FUJITA Tomonori [1].
>>
>> This is requiring the Opaque::try_ffi_init patch by Alice Ryhl[2] which just got merged into
>> char-misc-testing.
>>
>> [1]: https://lore.kernel.org/rust-for-linux/20241005122531.20298-3-fujita.tomonori@gmail.com/
>> [2]: https://lore.kernel.org/rust-for-linux/20240926-b4-miscdevice-v1-1-7349c2b2837a@google.com/
>>
>> Fiona Behrens (2):
>>   rust: LED abstraction
>>   samples: rust: led sample
>>
>>  rust/kernel/leds.rs      | 399 +++++++++++++++++++++++++++++++++++++++
>>  rust/kernel/lib.rs       |   2 +
>>  samples/rust/Kconfig     |  10 +
>>  samples/rust/Makefile    |   1 +
>>  samples/rust/rust_led.rs | 103 ++++++++++
>>  5 files changed, 515 insertions(+)
>>  create mode 100644 rust/kernel/leds.rs
>>  create mode 100644 samples/rust/rust_led.rs
>
> FYI: I'm not ignoring this patch-set.  On the contrary.  I'm trying to
> place myself into a position where I can not only review it with some
> confidence, but use it to author LED drivers.
>

Ah amazing, the first RFC was to probe for a general would LED accept rust code, and this sounds like a yes. Nice!

Currently still working on a LED driver for a Lenovo board that does not (as of when I started, and think still not yet) have a C driver for the LED. Current things missing is a static dmi match table (idea is to hardcoded that for now) and a abstraction for port memory but working on that, and can otherwise just go around that with unsafe calls in the driver itself.

  - Fiona

> -- 
> Lee Jones [李琼斯]