drivers: rtc: Add support for RTC for stm32n6 series

[basilegrunersmile]

This PR aims to add support for RTC on stm32n6 series.
Related Feature Request : #100667
It also provides an overlay for RTC sample for Nucleo n657x0 Q

[erwango]

@basilegrunersmile Is that strictly required ?
Can't this just eb enabled via a call to clock_control_on?

Also note that clock_control_on on N6 always enable LP clock:

zephyr/drivers/clock_control/clock_stm32_ll_n6.c

Lines 210 to 212 in 6182bc0

	/* Set Low Power clock */
	sys_set_bits(DT_REG_ADDR(DT_NODELABEL(rcc)) + pclken->bus + STM32_CLOCK_LP_BUS_SHIFT,
	pclken->enr);

[basilegrunersmile]

clock_control_on\ is called here by rtc_stm32_init` :

zephyr/drivers/rtc/rtc_stm32.c

Lines 488 to 492 in 6182bc0

	/* Enable RTC bus clock */
	if (clock_control_on(clk, (clock_control_subsys_t)&cfg->pclken[0]) != 0) {
	LOG_ERR("clock op failed\n");
	return -EIO;
	}

After further investigation, it appears that this call does not affect RCC_APB4LLPENR register (rm0486) because access to backup domain is disabled.
If I modify rtc_stm32_init like this, it works for nucleo n657x0 Q :

	stm32_backup_domain_enable_access();

	/* Enable RTC bus clock */
	if (clock_control_on(clk, (clock_control_subsys_t)&cfg->pclken[0]) != 0) {
		LOG_ERR("clock op failed\n");
		return -EIO;
	}

But I don't know how it can affect other boards.

Registers values before moving stm32_backup_domain_enable_access :

(gdb) x/1xw 0x56028274
0x56028274:	0x00038000
(gdb) x/1xw 0x560282B4
0x560282b4:	0x00000000

Registers values after moving stm32_backup_domain_enable_access :

(gdb) x/1xw 0x56028274
0x56028274:	0x00038000
(gdb) x/1xw 0x560282B4
0x560282b4:	0x00010000

[erwango]

You can put it under a series specific #ifdef. Or it won't probably hurt other series anyway.

[basilegrunersmile]

Done !

[basilegrunersmile]

Force pushed to solve merge conflict.
Also made modifications to comply with #106827.

[djiatsaf-st]

LGTM

[etienne-lms]

LGTM.

To get the full RTC be tested by default on these boards, could you also at in boards/st/{nucleo_n657x0_q|stm32n6570_dk}/Kconfig.defconfig:

 endif # VIDEO
 
+if RTC_STM32
+
+configdefault RTC_ALARM
+	default y
+
+configdefault RTC_CALIBRATION
+	default y
+
+endif # RTC_STM32
+
 endif # BOARD_STM32N6570_DK

on the model of #107342?

[etienne-lms]

For sake of simplicity of the code, it would be worth to always enable backup domain here, before RTC clock is enabled, IMHO. Not a blocking comment.

[etienne-lms]

Could you add in the commit log someething like Enable by default RTC alarm and calibration on Nucleo-N657x0-Q and STM32N6570-DK when STM32 RTC is enabled?
Otherwise LGTM.

[erwango]

Otherwise LGTM.

[etienne-lms]

Could you also add rtc to the supported tags in boards/st/stm32n6570_dk/twister.yaml and boards/st/nucleo_n657x0_q/twister.yaml?

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[etienne-lms]

LGTM.
I wonder if it would be better to be split in to 2 commits: one for changes in drivers/rtc/...+dts/... and one for thoses in boards/.... Non-blocking.

[mathieuchopstm]

I would strongly recommend adding a comment explaining why this must go before clock_control_on() to ensure this isn't moved again by accident. (Even to me right now, I'm not sure I understand the rationale...)