Incorrect assert in pwr_mgmt lib?

[jtrovato]

I am debugging my applications transitions to SYSTEM_OFF and back again. I am connected to the debugger so I am presumably testing with "emulated system off" which is just a loop. Things seems to be working well but I was thrown off by an error reported by the pwr_mgmt lib when entering system off. 

[00:00:00.000,000] <error> app: ERROR 3735928559 [Unknown error code] at ...components/libraries/pwr_mgmt/nrf_pwr_mgmt.c:418

This line has contains the following: ASSERT((ret_code == NRF_SUCCESS) || (ret_code == NRF_ERROR_SOFTDEVICE_NOT_ENABLED)); and each time I enter SYSTEM_OFF this assert fails which then raises a 0xDEADBEEF error, which is printed to my debugger.

I believe this assert is incorrect because the SDK documentation states that sd_power_system_off() should return NRF_ERROR_SOC_POWER_OFF_SHOULD_NOT_RETURN not NRF_SUCCESS. I am assuming that this function will not actually return if I am not in debugging mode which would mask this issue, but it is hard to tell if the SD is throwing an error or if this is simply a mistake in the SDK. changing NRF_SUCCESS to NRF_ERROR_SOC_POWER_OFF_SHOULD_NOT_RETURN seems to solve the issue for me. 

[Susheel Nuguru]

Hi jtrovato,

I have had this discussion with the team when we first discovered the "emulated SYSTEM OFF" effect in the debug mode with this API. You are right, I think the condition should not have included a check for NRF_SUCCESS in debug mode.  I think you can safely remove ret_code ==NRF_SUCCESS check at that line also make sure that you have a while(1); after this line, since in debugger mode, the code reaches the end of the function, and you would like the CPU to loop in that emulated behavior.

[jtrovato]

Thanks for verifying. I have replaced NRF_SUCCESS with NRF_ERROR_SOC_POWER_OFF_SHOULD_NOT_RETURN and that is not throwing an error in debug mode. Maybe consider changing or removing this assert case in future releases of the SDK.