NRF52 Bootloader hanging

[Chris]

Hi,

I've been using these parts for a few years (the 51) and recently did some work getting the boot loader to be joined together with the application into a single burnable file.

My application is working fine with the SD, and previously I had the boot loader working fine as well but for whatever reason it is now hanging in

sd_softdevice_enable

I've checked on two seperate boards and same problem. If I flash the app + SD it works fine. I use a common .h file between them so the crystal reference is the same.

    // Low frequency clock source to be used by the SoftDevice
#define NRF_CLOCK_LFCLKSRC      {.source        = NRF_CLOCK_LF_SRC_XTAL,            \
.rc_ctiv       = 0,                                \
.rc_temp_ctiv  = 0,                                \
.xtal_accuracy = NRF_CLOCK_LF_XTAL_ACCURACY_20_PPM}

I've spent a day on this so far and am stumped.

I've tried the default boot loaders for the PCA10040 (as I have the latest release parts on my custom boards) and they don't hang but they don't advertise either.

My own code is a duplicate of the boot loader but had a little bit of extra stuff like UART output etc etc. So I adjusted the .ld file to make it a bit bigger and also changed the starting offset in the dfu_types.h (plus the check during startup isn't complaining).

But this all worked fine a few months ago (yes, check the change logs I know but nothing showing).

I know not the most obvious. It's not throwing an error (I build with -DDEBUG and put in an error handler to dump the error code).

The SD is the s132_nrf52_2.0.1_softdevice.hex version.

Basically it never returns from sd_softdevice_enable which would imply some Osc issue (i.e.: we set it to expect an external osc which isn't there.. but this exact same setting when I run my application code works fine)... I have two oscillators on the boards..

I am on SDK 11 recently updated so wondering if something there caused me some grief.. but I had done a merge to check on diffs and nothing exciting.

Any clues? Guesses?

[Hung Bui]

@Chris: Please let me know which chip you have on your board ? nRF52 or nRF51 ? And please read the laser marking on top of the chip. Please let me know what do you mean by "two oscillators on the boards" but "an external osc which isn't there" , I don't get it.

If you simply test S132 softdevice + default bootloader for nRF52 (PCA10040) what happens ?

[Chris]

Ok, this isn't exactly an answer as on earlier SDKs this didn't cause a problem but now it seems to. And why on earth it then causes an issue far after the following code and during the init of the SD I don't understand either.

But I tracked it down to this

#define BUTTONS_NUMBER 0
#define BUTTON_START   0
#define BUTTON_1       0
#define BUTTON_2       0
#define BUTTON_STOP    0



/**@brief Function for initializing the button module.
 */
static void buttons_init(void)
{

    nrf_gpio_cfg_sense_input(BOOTLOADER_BUTTON,
                             BUTTON_PULL,
                             NRF_GPIO_PIN_SENSE_LOW);
    
    nrf_gpio_cfg_sense_input(DIAGS_BUTTON,
                             BUTTON_PULL,
                             NRF_GPIO_PIN_SENSE_LOW);
    
}

Yes, I know, why define them if we're not using them. But this is a common file I use across a bunch of boards.

Anyway, it caused all the grief with the problem described above.

Don't init my lack of buttons. All working fine.

No idea why this has now caused a problem but it has... bit of a wild goose chase.

[Chris]

FYI this was the nRF52, bit hard to read here at the moment.

[Hung Bui]

@Chris: You can always reconfigure the bootloader so that you can run it in debug mode and add a breakpoint at the app_error_handler to track back on why it causes an error. Have a look here to know how to configure the bootloader for debugging.