Reset on SVCALL BLE enable

[Tostikoning]

Hi all,

I am trying to get the ble template of SDK12.2 working in eclipse using a managed makefile. The softdevice I use is s132 v3.0.0. The board is an evaluation kit from Taiyo Yuden (EKSHCNZXZ) which uses the NRF52832 chip.

I programmed the softdevice using nRFgo Studio.

I use ARM GNU build tools and compiler.

I can compile the project and even debug the project using J-Link, however when I try to enable the bluetooth stack it resets and start at the beginning of main.

I am using the main file from ble_peripherals/ble_app_template

The SDK I added manually in Eclipse which seems to work fine.

The following code is used (in main.c):

int main(void)
{
    uint32_t err_code;
    bool     erase_bonds;

    // Initialize.
    err_code = NRF_LOG_INIT(NULL);
    APP_ERROR_CHECK(err_code);

    timers_init();
    buttons_leds_init(&erase_bonds);
    ble_stack_init();

From there we go to (in main.c):

static void ble_stack_init(void)
{
    uint32_t err_code;

    nrf_clock_lf_cfg_t clock_lf_cfg = NRF_CLOCK_LFCLKSRC;

    // Initialize the SoftDevice handler module.
    SOFTDEVICE_HANDLER_INIT(&clock_lf_cfg, NULL);

    ble_enable_params_t ble_enable_params;
    err_code = softdevice_enable_get_default_config(CENTRAL_LINK_COUNT,
                                                    PERIPHERAL_LINK_COUNT,
                                                    &ble_enable_params);
    APP_ERROR_CHECK(err_code);

    // Check the ram settings against the used number of links
    CHECK_RAM_START_ADDR(CENTRAL_LINK_COUNT, PERIPHERAL_LINK_COUNT);

    // Enable BLE stack.
    ble_enable_params.gatt_enable_params.att_mtu = NRF_BLE_MAX_MTU_SIZE;
    err_code = softdevice_enable(&ble_enable_params);

And then to (in softdevice_handler.c):

uint32_t softdevice_enable(ble_enable_params_t * p_ble_enable_params)
{
    uint32_t err_code;
    uint32_t app_ram_base;
    extern uint32_t __data_start__;
    volatile uint32_t ram_start = (uint32_t) &__data_start__;

    app_ram_base = ram_start;
    NRF_LOG_DEBUG("sd_ble_enable: RAM start at 0x%x\r\n",
                    app_ram_base);
    err_code = sd_ble_enable(p_ble_enable_params, &app_ram_base);

Which launches (in ble.h):

SVCALL(SD_BLE_ENABLE, uint32_t, sd_ble_enable(ble_enable_params_t * p_ble_enable_params, uint32_t * p_app_ram_base))

After this call it resets and start at the beginning of main.

Do any of you have an idea what I might be missing.

EDIT: I found some more info about the problem I am experiencing. Investigating some more i found the sd_ble_enable function returned an error code 0x04 which represents:

 * @retval ::NRF_ERROR_NO_MEM         The amount of memory assigned to the SoftDevice by *p_app_ram_base is not
 *                                    large enough to fit this configuration's memory requirement. Check *p_app_ram_base
 *                                    and set the start address of the application RAM region accordingly.

I'm going to check why this is happening.

[Vidar Berg]

[sd_ble_enable()](infocenter.nordicsemi.com/.../group___b_l_e___c_o_m_m_o_n___f_u_n_c_t_i_o_n_s.html cp=2_3_0_1_1_2_0_2_2_0#ga812ea2c6258c1cf705c7c114b6297aab) returns the lowest RAM start address you can use based on your softdevice configuration (ble_enable_params_t). Make sure that this corresponds to the RAM start address set in your linker file (*.ld). Background: devzone.nordicsemi.com/.../

[Tostikoning]

Thanks for the response Vidar! The linker scripts I use are from the SDK. I use the gcc_nrf52.ld and included the nrf52_common.ld which sets the data_start. I do not see how I have control over this value. In the code the p_app_ram_base is 0x20002128. This is over the minimum and the same I see when I try the template in KEIL. The code compiled in KEIL works. I can see the board advertising with the KEIL compiled file. I am going to dig some more in the memory allocation of the softdevice and the ble params.

[Vidar Berg]

You only need to edit the project specific linker file (gcc_nrf52.ld in this case) to change the memory layout. Do you get the same assertion (NRF_NO_MEM) if you try with the following values:

MEMORY
{
  FLASH (rx) : ORIGIN = 0x1f000, LENGTH = 0x61000
  RAM (rwx) :  ORIGIN = 0x20004000, LENGTH = 0xC000
}

[Tostikoning]

The error seems to be gone. app_ram_base is still 0x20002128. I'm not sure if this is still ok? I do not see it advertising yet, but we are moving forward. I do not see any other error code pop up in the debugger.

It does not reset anymore, and the error_code is 0.

EDIT:

I have the board advertising! Thank you Vidar!

[Vidar Berg]

Glad to hear that it's working now. It's not a problem to set the RAM base 'too' high, it just means that you have an unused area that could have been used by the application (0x4000-2128 = ~7KB)