debugging hard faults in NCS

[Farhang]

Hi Nordic Community,

I have been experimenting with samples in NCS and making modifications.

Sometimes adding some code causes a hardfault during runtime. I can often find the root cause, e.g de-referencing a struct in memory that doesn't exist.

This often causes z_arm_exc_spurious  . SES (Segger Embedded Studio) doesn't show where the error was first initiated - the only info is that the firmware has reached the z_arm_secure_fault()

Are there any debugging guides (methodical approach) for the Zephyr spurious error?

Are there extra hard fault handlers I can enable to catch things earlier on that show the root cause?

Comparing Nordic NCS/SES to the legacy Nordic SDK and IAR embedded - It seems I'm in much less control of fault handlers and debugging info. 

PS. A separate issue is I can't get thread-level debugging (e.g. threads running, paused, stack usage etc) info from SES. I read somewhere that is coming in v5.40 of SES, is that true?

[Simon]

First, try setting CONFIG_DEBUG_OPTIMIZATIONS=y in the prj.conf file.

Could you try to press continue (green arrow) and see if you end up in secure_fault() and/or esf_dump()?

Also, could you try to enable this in prj.conf:

CONFIG_RESET_ON_FATAL_ERROR=n
CONFIG_LOG=y

Then you should be able to see more information about the hard fault. Open a serial terminal like Termite to see the log.

Best regards,

Simon