Error Logging - core/crash dump

[Luke_123]

Hi,

In my system, the BLE part is handled by nRF52832 and there is another master MCU which takes care of other tasks in the system. The Master MCU communicates with nRF52832 over UART. 

nRF52832 uses SDK 17.1.0.

We are trying to design a logic which will capture all the necessary information in the case of a fault for nRF52832 (eg:- hardfault, stackoverflow, device hung). If a customer returns the product saying not working, we should be able to retrieve this information and find the root cause.

1. What information (should be stored) will be useful to find out the root cause ?
2. Does copying an entire RAM to flash in the event of a fault will help ?
3. How is this handled in a general embedded system ?

[Sigurd]

Hi,

The logger module has a crashlog mode you can use to store crashlogs to flash. See this link: https://infocenter.nordicsemi.com/topic/sdk_nrf5_v17.1.0/lib_nrf_log.html?cp=9_1_3_27_5_0_2#nrf_log_backend_flash_crashlog

[Luke_123]

Thank you for the link. This is very helpful for the user logs.

How about the CPU register details ? If an exception happens, is it possible to know where in the code has happened it ? Is there someway to store the call stack and retrieve it ?

[Sigurd]

Luke_123 said:

How about the CPU register details ? If an exception happens, is it possible to know where in the code has happened it ? Is there someway to store the call stack and retrieve it ?

Yes, it's possible. 

See this post here for details:  Saving the location of Watchdog timeout