How to enable ETM trace in nRF52 PCA10040 board

RE: How to enable ETM trace in nRF52 PCA10040 board

Michael Dietz — Fri, 22 Apr 2016 13:29:53 GMT

Blog post explaining how to do this step by step: devzone.nordicsemi.com/.../.

RE: How to enable ETM trace in nRF52 PCA10040 board

Juba Belgaid — Tue, 15 Mar 2016 15:53:12 GMT

Overview of my Ble application:

.
.
.
BleInit();
uart_init();
printf("%s",start_string);
DisplayInit();
NorFlashInit();
//scheduler init
APP_SCHED_INIT(SCHED_MAX_EVENT_DATA_SIZE, SCHED_QUEUE_SIZE);
		
//create Display timer
err_code = app_timer_create(&DisplayTimerId,	APP_TIMER_MODE_REPEATED,	DisplayEventHandler);
APP_ERROR_CHECK(err_code);
		
while (1)
{
	app_sched_execute();
	power_manage();

}

when proc is idle mode i measure a consumption current of~900uA, run mode ~6mA. Hardfault occurs randomly (on connect event, display event ...etc) when it occurs i measure a consumption current of 10mA, and ble stops advertising.
To be sure that is a hardfault, i replaced

HardFault_Handler\
                PROC
                EXPORT  HardFault_Handler         [WEAK]
                B       .
                ENDP

on arm_startup_nrf52.s by Reset_Handler code

HardFault_Handler\
                PROC
                EXPORT  Reset_Handler             [WEAK]
                IMPORT  SystemInit
                IMPORT  __main
                
                LDR     R0, =SystemInit
                BLX     R0
                LDR     R0, =__main
                BX      R0
                ENDP

then when the hardfault occurs, the board resets

RE: How to enable ETM trace in nRF52 PCA10040 board

Vidar Berg — Tue, 15 Mar 2016 15:46:30 GMT

Strange. how did you conclude it was a hardfault earlier? Halting the core with nrfjprog/jlink, or did you have a LED or similar in the event handler?

RE: How to enable ETM trace in nRF52 PCA10040 board

Juba Belgaid — Tue, 15 Mar 2016 15:40:52 GMT

i am unable to reproduce the hardfault with debug session !!!

RE: How to enable ETM trace in nRF52 PCA10040 board

Juba Belgaid — Tue, 15 Mar 2016 12:46:30 GMT

Thank you for your reply. How can i display call stack? I can't insert breakpoints since the hardfault is random.

RE: How to enable ETM trace in nRF52 PCA10040 board

Vidar Berg — Tue, 15 Mar 2016 11:42:27 GMT

Unfortunately, the j-link plus does not appear to support trace. Does the call stack provide any clues as to what function caused the hardfault? I.e, are there any function calls on the stack prior to the hardfault? You can also get the PC value before the exception occurred by reading the memory at the current stackpointer pluss an offset of 24 bytes (SP+24).

Also you can just insert breakpoints and determine how far the application gets before the exception occurs. With this you should be able to locate the code that triggers it.

EDIT - Attached screen dump for illustration

The function call prior to the exception will generally be shown if it occurred in the application code. Also see that the address of the instruction that caused the hardfault is loaded to R0, but not sure if that is always the case.

To illustrate this I triggered a hardfault exception by accessing an invalid memory area in the battery_level_update function found in the hrs example.

dummy pointer to 0x8000000 which will cause hardfault when the cpu tries to fetch the content in the battery_level_update function.
SP value pushed to stack prior to entering the event handler.
problem function
PC counter
Same address to function call as shown in the memory view (4).

You can download the image to enlarge it.

RE: How to enable ETM trace in nRF52 PCA10040 board

Juba Belgaid — Tue, 15 Mar 2016 10:16:21 GMT

I have segger J-link Plus www.segger.com/j-link-plus.html

RE: How to enable ETM trace in nRF52 PCA10040 board

Vidar Berg — Tue, 15 Mar 2016 10:10:01 GMT

Do you have a J-link trace or similar? Note that the on-board J-link does not support trace.