Refactor STM32 PWM driver using LL API

[gmarull]

The PWM driver has been refactored using the HAL LL API. Not only that, but the set pin_set function is now faster, as channel output compare is just initialized if needed.

Support for the polarity flag has also been added to the STM32 PWM driver. STM32 boards using PWM have been updated accordingly.

Fixes #23629

[zephyrbot]

All checks are passing now.

checkpatch (informational only, not a failure)

-:564: WARNING:TRAILING_SEMICOLON: macros should not use a trailing semicolon
#564: FILE: drivers/pwm/pwm_stm32.c:314:
+#define PWM_DEVICE_INIT(index)                                                 \
+	static struct pwm_stm32_data pwm_stm32_data_##index;                   \
+									       \
+	static const struct pwm_stm32_config pwm_stm32_config_##index = {      \
+		.timer = (TIM_TypeDef *)DT_REG_ADDR(                           \
+			DT_INST(index, st_stm32_timers)),                      \
+		.prescaler = DT_INST_PROP(index, st_prescaler),                \
+		.pclken = DT_INST_CLK(index, timer)                            \
+	};                                                                     \
+									       \
+	DEVICE_AND_API_INIT(pwm_stm32_##index, DT_INST_LABEL(index),           \
+			    &pwm_stm32_init, &pwm_stm32_data_##index,          \
+			    &pwm_stm32_config_##index, POST_KERNEL,            \
+			    CONFIG_KERNEL_INIT_PRIORITY_DEVICE,                \
+			    &pwm_stm32_driver_api);

- total: 0 errors, 1 warnings, 1657 lines checked

NOTE: For some of the reported defects, checkpatch may be able to
      mechanically convert to the typical style using --fix or --fix-inplace.

Your patch has style problems, please review.

NOTE: Ignored message types: AVOID_EXTERNS BRACES CONFIG_EXPERIMENTAL CONST_STRUCT DATE_TIME FILE_PATH_CHANGES MINMAX NETWORKING_BLOCK_COMMENT_STYLE PRINTK_WITHOUT_KERN_LEVEL SPDX_LICENSE_TAG SPLIT_STRING VOLATILE

NOTE: If any of the errors are false positives, please report
      them to the maintainers.

Tip: The bot edits this comment instead of posting a new one, so you can check the comment's history to see earlier messages.

[erwango]

This PR is ambitious has it encompass 2 major changes: Boards Kconfig to dts transition and conversion to LL, plus and additional rather unrelated change (polarity support).
Please split these 3 changes to individual PRs.

Regarding the Kconfig to dts transition, I think the node labeling sound like the best option:

		timers1: timers@40012c00 {
			compatible = "st,stm32-timers";
			{...}                  
			pmw1: pwm@1 {
				compatible = "st,stm32-pwm";
				status = "disabled";
				st,prescaler = <10000>;
				label = "PWM_1";
				#pwm-cells = <2>;
			};

Besides, it would help getting rid of the paths in following expressions:

		green_pwm_led: green_pwm_led {
			pwms = <&{/soc/timers@40000000/pwm} 1 4>;

[gmarull]

@erwango I have rebased this PR now that PWM uses DT. I have splitted it in two commits, the first one is the refactor and the second the addition of the polarity flag support. I can split it into 2 PRs, not sure if needed, though.

[FRASTM]

flags is either PWM_POLARITY_NORMAL or PWM_POLARITY_INVERTED , correct ?

[gmarull]

Yes, they're defined #include <dt-bindings/pwm/pwm.h>. I guess that flags could potentially be used for other purposes in the future.

[gmarull]

@erwango ping

[erwango]

Txs for this welcome change.
Some comments, but otherwise it's fine to me.

[lochej]

I'm sorry if this is the wrong section but that is my first time contributing to Zephyr.
The LL PWM implementation has a bug in the get_polarity() function always returns LL_TIM_OCPOLARITY_LOW which in turn actually inverts your PWM on STM32s :

static uint32_t get_polarity(pwm_flags_t flags)
{
	if (flags & PWM_POLARITY_NORMAL) { <-- PWM_POLARITY_NORMAL expands to 0<<0 in include/dt-bindings/pwm/pwm.h
		return LL_TIM_OCPOLARITY_HIGH;
	}

	return LL_TIM_OCPOLARITY_LOW;
}

The PWM_POLARITY_MASK set to 0<<0 forces the if statement to always be false.

My recommendation would either be to change the defines in dt-bindings/pwm/pwm.h to:

/** PWM pin normal polarity (active-high pulse). */
#define PWM_POLARITY_NORMAL	(1 << 0)
// -> High polarity should be 1

/** PWM pin inverted polarity (active-low pulse). */
#define PWM_POLARITY_INVERTED	(0 << 0)
//-> Low polarity should be 0

Doing this fix has proven correct behavior of PWM polarity on my Nucleo64 board.

[gmarull]

@lochej thanks for reporting, fix here: #26680. You can open an issue if you want so it can be referenced on the fix PR.