Bus fault detected when entering function

[sjf]

Hi,

We have been working with a Zigbee coordinator based on nRF52840 with nRF5_SDK_for_Thread_and_Zigbee_v4.0.0_dc7186b, Softdevice and FreeRTOS. 

Zigbee task defines a stack of (64+2048)+4 = 8448 bytes. It is placed in RAM from 0x2000a3ac to 0x200082ac.

After a couple of days of normal operation, an external command is received and a bus fault error is generated when a function is triggered. The sp points to 0x2000a378 (within the stack limits) and the disassembly  points the instruction at the very beginning of the implemented function zbm_get_free_row_query_table():

sub sp, sp, #8

Why is this generated if sp value is valid and pointing to a ram offset within the stack range?

The external command was successfully tested many times. We checked that the sp points to 0x2000a378 and no bus fault is generated. But a couple of days later, the bus fault is generated again on the same instruction

Please, find attached some helpful SES debug images.

Regards,

Santiago

[sjf]

Hi, find below some extra info when Bus Fault is detected:

BFSR = 0x00000004 (IMPRECISERR = 1)

[Vidar Berg]

Hi,

Could you please try to follow instructions provided here in the blog post to see if you can find the exact instruction that lead to the hardfault: https://interrupt.memfault.com/blog/cortex-m-fault-debug#imprecise-error-code-example?  Also, maybe you have tried it already, but would it make sense to increase the stack sizes in your application, or have you already profiled the stack usage?

Best regards,

Vidar

[sjf]

Thanks so much. We could identify that there was an uninitialized pointer quite close to where the pc was pointing to.

Regards

[Vidar Berg]

Thank you for the update. I'm glad to hear that you found the problem