ADC hardware timing framework considerations

[zbas]

I am part way through implementing a hardware timed adc driver and think there are some issues with the existing framework that need to be addressed.

Adc_context.h says drivers should implement adc_context_enable_timer and adc_context_disable_timer but these functions only pass an adc_context which has no reference to hardware. A driver can ignore these and keep the timer enabled but this is wasteful and defeats the intention.

As a regular C function it can also only be implemented once, either a driver implementation or the existing kernel implementation controlled by ADC_CONTEXT_USES_KERNEL_TIMER. It seems to me that having some adc controlled by a hardware timer and some other controlled by a kernel timer is quite a natural use case. Reimplementing the kernel timer in the driver seems quite clunky and wouldn’t respect the configuration logic.

I don’t know why the framework is the way it is, but could these issues be resolved? Adc_context_on_sampling_done passes a const struct device so perhaps other adc_context functions could, including adc_context_enable_timer and adc_context_disable_timer, or something could be added to adc_context.

[github-actions]

Hi @zbas! We appreciate you submitting your first issue for our open-source project. 🌟

Even though I'm a bot, I can assure you that the whole community is genuinely grateful for your time and effort. 🤖💙

[zbas]

I have seen that the first issue (no reference to hardware) is solved by the CONTAINER_OF macro. I would still appreciate advice/input on implementation of hardware and kernel timers together.