A well working project for nRF51822 with SDK12.2.0 fails after migration to nRF52840 with SDK16.0.0

[good_night]

The project for nRF51822 was developed on base of example ble_peripheral ble_app_hrs for pca10028/s130
with an additional vendor specific service with 128-bit UUID for an MIDI service.

The UUID were set to:

#define BLE_UUID_MIDI_BASE_UUID   { { 0x00, 0xC7, 0xC4, 0x4E, 0xE3, 0x6C, 0x51, 0xA7, 0x33, 0x4B, 0xE8, 0xED, 0x5A, 0x0E, 0xB8, 0x03 } }
#define BLE_UUID_MIDI_SERVICE     0x0E5A
#define BLE_UUID_MIDI_CHARACTERISTC_UUID  { { 0xF3, 0x6B, 0x10, 0x9D, 0x66, 0xF2, 0xA9, 0xA1, 0x12, 0x41, 0x68, 0x38, 0xDB, 0xE5, 0x72, 0x77 } }
#define BLE_UUID_MIDI_CHARACTERISTIC     0xE5DB

and the advertising to:

ble_uuid_t m_adv_uuids[] =
  {{ BLE_UUID_DEVICE_INFORMATION_SERVICE, BLE_UUID_TYPE_BLE },
   { BLE_UUID_MIDI_SERVICE, BLE_UUID_TYPE_VENDOR_BEGIN }};

with the following changes for sdk_config.h:

#define BLE_HRS_ENABLED 0    //1
#define ADC_ENABLED 1        //0
#define APP_UART_ENABLED 1    //0
#define NRF_LOG_ENABLED 1    //0
#define NRF_LOG_BACKEND_SERIAL_USES_UART 0    //1
#define NRF_LOG_BACKEND_SERIAL_USES_RTT 1    //0

services_init() was called before advertising_init() and everything was fine.

Now for nRF52840 I used the same base example ble_peripheral ble_app_hrs for pca10056/s140 with SDK16.0.0.
The same UUID settings
and the following changes for sdk_config.h (ADC is not implemented at this time):

#define BLE_HRS_ENABLED 0    //1
#define NRF_SDH_BLE_VS_UUID_COUNT 1        //0

again services_init() is called before advertising_init() 

resulting in the "Fatal error" in ble_advdata.c at

// Get GAP device name and length
err_code = sd_ble_gap_device_name_get(&p_encoded_data[(*p_offset) + AD_DATA_OFFSET], &actual_length);
VERIFY_SUCCESS(err_code);

What do I need to fix making the project run?
Thank you for any hints on this issue!

[good_night]

Thank you for the prompt reply:

As I wrote before, the nrf_log gives a

<error> app: Fatal error
<warning> app: System reset

in ble_advdata.c.

[awneil]

That just tells you that it's an error - any error.

As I wrote before, you need to find out what error, exactly, it is.

Probably the simplest way is to set a breakpoint, and examine the error code in the debugger ...

Otherwise, enable (more) debug logs - a Debug build configuration should do that.

[good_night]

The Debug build shows now an other behaviour:

according this, the error is much earlier at:

  err_code = sd_ble_gatts_characteristic_add (p_midi_service->service_handle,
                                              &char_md, &attr_char_value,
                                              &p_midi_service->char_handles);
  APP_ERROR_CHECK(err_code);

with the message:

<error> app: ERROR 7 [NRF_ERROR_INVALID_PARAM] at C:\edp\STm_FSW_BLE_7ST2\nRF5SDK16\source\midi_service.c:203
PC at: 0x00033BD9
<error> app: End of error report

[awneil]

Sadly, NRF_ERROR_INVALID_PARAM is a prime example of an unhelpful error message:

https://devzone.nordicsemi.com/f/nordic-q-a/48490/graceful-recovery-from-bluetooth-errors/192270#192270

So you're just going to have to carefully read the documentation of sd_ble_gatts_characteristic_add() to see what the criteria are for valid parameters, and try to work out which one of yours is invalid - and why.

[good_night]

I simply restricted ble_uuid_t m_adv_uuids to only one entry and everything was fine.
Interesting was the behaviour in DEBUG build: different places for the error were shown by Segger IDE - so I finally gave up to find the "real reason" caused the error.

[awneil]