nrf51822 + gdb + printf

[L.B.]

Hello everyone,

I'm trying to set up entirely open source project for my NRF51822 chip. Everything works fine so far but something strange happens during debugging.

I use nRF5_SDK_11.0.0_89a8197 and I've taken a template with a softdevice. I'm flashing the s130_nrf51_2.0.0_softdevice.hex softdevice, then my application nrf51422_xxac_s130.hex and I've done an example with printf over UART. I'm using a Tayo Yuden demo board and I have a virtual COM port that outputs the expected result.

However if I try to run gdb here is what happens:

Breakpoint 1, main () at main.c:593
593        timers_init();
(gdb) next
halted: PC: 0x0002051c
594        buttons_leds_init(&erase_bonds);
(gdb) next
halted: PC: 0x00020bbc
halted: PC: 0x00020bbe
halted: PC: 0x00020b40
595        uart_init();
(gdb) next
halted: PC: 0x00024880
596        ble_stack_init();
(gdb) next
halted: PC: 0x0002082c
597        device_manager_init(erase_bonds);
(gdb) next
halted: PC: 0x00020bcc
halted: PC: 0x00020bce
halted: PC: 0x00020bd0
halted: PC: 0x000209c4
598        gap_params_init();
(gdb) next
halted: PC: 0x00020550
599        advertising_init();
(gdb) next
halted: PC: 0x00020ab4
600        services_init();
(gdb) next
halted: PC: 0x000205f4
601        conn_params_init();
(gdb) next
halted: PC: 0x00020658
604        application_timers_start();
(gdb) next
halted: PC: 0x000206cc
605        err_code = ble_advertising_start(BLE_ADV_MODE_FAST);
(gdb) next
halted: PC: 0x00020bea
halted: PC: 0x00021e48
halted: PC: 0x00020bf0
halted: PC: 0x00020bf2
606        APP_ERROR_CHECK(err_code);
(gdb) next
halted: PC: 0x00020bf4
halted: PC: 0x00020bf6
halted: PC: 0x00020bf8
halted: PC: 0x00020bfa
halted: PC: 0x00020c04
608        printf("Starting application!\n\r");
(gdb) next
halted: PC: 0x00020c06
halted: PC: 0x00020c08
halted: PC: 0x00024c4c

Breakpoint 1, main () at main.c:593
593        timers_init();
(gdb)

i.e. as soon as I try to read address 0x00024c4c my PC goes back to the beginning of main. I do not see this behavior if I just flash the chip and let it run freely.

What could be the cause of this behaviour? I tried to disable watchdog, the MPU is not on during debug ...

Regards,

L. Bogdanov

[Hakon]

Hey, try removing all the breakpoints and see if it still resets.

[L.B.]

Unfortunately it still resets but this time stops at the line when I was previously on. I can't even explain this behahiour :-D 

[Hakon]

Try to set the DEBUG preprocessor symbol and run the code again.

[L.B.]

YESSS! I've included -DDEBUG as you said and then the printf would go into the function app_error_save_and_stop( ) and hang. I followed the comments in it -

// If printing is disrupted, remove the irq calls, or set the loop variable to 0 in the debugger.

and changed (at compile time, not with the debugger)

volatile bool loop = true; ------> to ------> volatile bool loop = false;

in that same function and yay - it works now!

Thanks for the help!

Really appreciate it!