tests: boards: nrf: hwinfo: Add test for reset cause on nrf54*

[nordic-segl]

Add test that checks reset cause reported by HWINFO.

Check that hwinfo_get_supported_reset_cause() returns expected value.

Check that hwinfo_get_reset_cause() correctly detects:

• RESET_PIN,
• RESET_SOFTWARE,
• RESET_WATCHDOG.

Check that reset cause can be cleared with hwinfo_clear_reset_cause().

[nordic-segl]

FYI: @nordic-piks , @kl-cruz.

[nordic-segl]

Build failure on nrf54l15dk (undefined reference to `sys_reboot') will be analysed by driver developers.
I will remove this target from platform_allow until build failure is fixed.

[alexanderwachter]

To be honest, I'm not in favor of this tests.
This test are very hw dependent. For example, I don't think that there is a guarantee that data is retained after a reset in noinit sections.
Second, you should make use of the testing framework for maintainability.

[nordic-segl]

To be honest, I'm not in favor of this tests. This test are very hw dependent. For example, I don't think that there is a guarantee that data is retained after a reset in noinit sections. Second, you should make use of the testing framework for maintainability.

Indeed, this test is "complicated". However, this is because there are resets in the middle of each test case:

• read reset causes and confirm it was RESET_PIN (first assumption - there is pin-reset after flashing HEX file);
• print supported reset causes;
• trigger software rest;
• after boot, confirm that reset cause is RESET_SOFTWARE;
• configure watchdog and wait until it expires / resets CPU;
• after boot, confirm that reset cause is RESET_WATCHDOG

This test are very hw dependent.

Therefore, I put platform_allow in the testcase.yaml

For example, I don't think that there is a guarantee that data is retained after a reset in noinit sections.

I have to store somehow information about test suite state. Solution where test state is stored in variable at .noinit section was copied from following watchdog driver test:
https://github.com/zephyrproject-rtos/zephyr/blob/main/tests/drivers/watchdog/wdt_basic_api/src/test_wdt.c#L384

Second, you should make use of the testing framework for maintainability.

ZTEST test framework doesn't support reset in the middle of test. After boot, it will always start with the first test case.
Above mentioned watchdog test has single ZTEST for all test cases. It uses single variable and multiple if statements to detect which test case was executed and to determine test result. What I don't like there is that test result PASS doesn't inform which particular test cases were executed.
I've decided to use State Machine design pattern instead.

@alexanderwachter : What about moving it to a different directory?
https://github.com/zephyrproject-rtos/zephyr/tree/main/tests/boards/nrf

[nordic-segl]

I moved test to zephyr/tests/boards/nrf/hwinfo/reset_cause/