[PATCH v15 0/7] Rust Abstractions for PWM subsystem with TH1520 PWM driver

[Michal Wilczynski]

On 10/13/25 18:48, Uwe Kleine-König wrote:
> Hello,
> 
> my diff on top of your changes looks as follows:
> 
> diff --git a/drivers/pwm/Kconfig b/drivers/pwm/Kconfig
> index dd6db01832ee..e7f770ecfe84 100644
> --- a/drivers/pwm/Kconfig
> +++ b/drivers/pwm/Kconfig
> @@ -812,9 +812,8 @@ config PWM_XILINX
>  	  will be called pwm-xilinx.
>  
>   config RUST_PWM_ABSTRACTIONS
> -	bool "Rust PWM abstractions support"
> +	bool
>  	depends on RUST
> -	depends on PWM=y
>  	help
>  	  This option enables the safe Rust abstraction layer for the PWM
>  	  subsystem. It provides idiomatic wrappers and traits necessary for
> 
> i.e. make RUST_PWM_ABSTRACTIONS invisible, it is only supposed to be
> selected and there is little (or even no?) use to enable it without a
> selector.
> 
> diff --git a/drivers/pwm/pwm_th1520.rs b/drivers/pwm/pwm_th1520.rs
> index c9fd1d8d17bc..a5666052b7ce 100644
> --- a/drivers/pwm/pwm_th1520.rs
> +++ b/drivers/pwm/pwm_th1520.rs
> @@ -121,6 +121,7 @@ fn round_waveform_tohw(
>          wf: &pwm::Waveform,
>      ) -> Result<pwm::RoundedWaveform<Self::WfHw>> {
>          let data = chip.drvdata();
> +        let status = 0;
>  
>          if wf.period_length_ns == 0 {
>              dev_dbg!(chip.device(), "Requested period is 0, disabling PWM.\n");
> @@ -141,18 +142,13 @@ fn round_waveform_tohw(
>          if period_cycles == 0 {
>              dev_dbg!(
>                  chip.device(),
> -                "Requested period {} ns is too small for clock rate {} Hz, disabling PWM.\n",
> +                "Requested period {} ns is too small for clock rate {} Hz, rounding up.\n",
>                  wf.period_length_ns,
>                  rate_hz
>              );
>  
> -            return Ok(pwm::RoundedWaveform {
> -                status: 0,
> -                hardware_waveform: Th1520WfHw {
> -                    enabled: false,
> -                    ..Default::default()
> -                },
> -            });
> +            period_cycles = 1;
> +            status = 1;
>          }
>  
>          let mut duty_cycles = ns_to_cycles(wf.duty_length_ns, rate_hz).min(u64::from(u32::MAX));
> 
> i.e. round up for too small period requests ...
> 
> @@ -189,7 +185,7 @@ fn round_waveform_tohw(
>          );
>  
>          Ok(pwm::RoundedWaveform {
> -            status: 0,
> +            status: status,
>              hardware_waveform: wfhw,
>          })
>      }
> 
> ... and return 1 then
> 
> @@ -355,7 +351,7 @@ fn probe(
>                  "Clock rate {} Hz is too high, not supported.\n",
>                  rate_hz
>              );
> -            return Err(ERANGE);
> +            return Err(EINVAL);
>          }
>  
>          let chip = pwm::Chip::new(
> 
> at least pwm-stm32 uses EINVAL. Having said that, I wonder if this check
> is sensible here at all. pwm-stm32 does it to ensure that
> mul_u64_u64_div_u64() does the right thing, but this exact issue doesn't
> exist here.

I think your changes are correct. For the existence of check I think it
is helpful as it fails early with a helpful message, but isn't strictly
necessary.

I will incorporate these changes and send another revision.

> 
> Otherwise looks fine I think.
> 
> Best regards
> Uwe

Best regards,
-- 
Michal Wilczynski <m.wilczynski at samsung.com>