Hardfault at Enabled/Disabled Softdevice

RE: Hardfault at Enabled/Disabled Softdevice

Vidar Berg — Thu, 27 Oct 2022 12:16:05 GMT

Hi,

Could you explain why you need to deactivate the stack in your case? While it is possible to do, it will often just lead to increased complexity in the code. Everything initialized in the Softdevice is reset when you disable it, so you have to start the BLE initialization over again when you want to re-enable BLE.

[quote user="Beldramma"]It seems also that we should not call app_timer_init each time Stack is initialized,[/quote]

The app timer module can safely remain initialized when you disable the Softdevice, but there may be timer instances that you will won't to stop before doing so.

RE: Hardfault at Enabled/Disabled Softdevice

Beldramma — Wed, 26 Oct 2022 17:47:10 GMT

Hi Vidar Berg .

It seems that I have something that works :

When I want to activate Stack and BLE modules it is what I call:

void ble_init( void )
{
    ble_stack_init();
    gap_params_init();
    gatt_init();
    services_init();     
    advertising_init();
    conn_params_init();
#if SECURITY_MODE
    if( !peer_manager_enabled )
    {
        peer_manager_init();
        peer_manager_enabled = true;
    }
#endif
    
}

When I have to deactivate the Stack, here is what I call:

ret_code_t ble_deinit( void )
{
    ret_code_t err_code;
    err_code = ble_conn_params_stop();
    if( err_code != NRF_SUCCESS ) return err_code;
    advertising_stop();
    err_code = ble_stack_stop();
    return err_code;
}

Trick is to call peer_manager_init only at first boot, and not call it after.

After disabling/enabling stack multiples times, I have tried connecting and pairing to my device ( Legacy connection with Passkey ), and there was apparently no problem.

It seems also that we should not call app_timer_init each time Stack is initialized, because it will reset all timers. Also, when deactivating Stack, I activate LFClock to permit to my timers to continue to work.

Even though it seems to work, I don't know if the deactivation of the stack is really done properly, that's why I asked for a tutorial to deactivate/reactivate the stack properly.

RE: Hardfault at Enabled/Disabled Softdevice

Vidar Berg — Wed, 26 Oct 2022 13:04:36 GMT

Beldramma , please include details about your setup along with some screenshots of what you see when you try to debug this.

RE: Hardfault at Enabled/Disabled Softdevice

Beldramma — Wed, 19 Oct 2022 15:27:52 GMT

Hi ! I have this need too, are there some updates on this subject ?

RE: Hardfault at Enabled/Disabled Softdevice

Jared — Fri, 24 Jul 2020 09:23:36 GMT

I'm sorry I see that my previous answer wasn't typed correctly. I meant to say could you try to NOT initialize it several times i.e just disable the softdevice by calling the nrf_sdh_disable_request()?

RE: Hardfault at Enabled/Disabled Softdevice

Ted — Fri, 17 Jul 2020 06:51:43 GMT

I use a timer interrupt to sample 9600 bps serial data. It isn't UART data. Its synchronous data is 0x55, 0x55, 0x01,0x01. So it needs hard timing. The softdevice will make sample time not fixed if I don't disable it.

RE: Hardfault at Enabled/Disabled Softdevice

Jared — Wed, 15 Jul 2020 13:01:38 GMT

I don't think the Peer manager module is designed to be re-initialized several times as it doesn't have a API for doing this. However, could you try to initializing it several times i.e just disable the softdevice by calling the nrf_sdh_disable_request().

Is there a specific reason for why you would like to disable the Softdevice?

RE: Hardfault at Enabled/Disabled Softdevice

Ted — Wed, 15 Jul 2020 05:44:30 GMT

Hi,

Calling ble_conn_state_init() before calling peer_manager_init(); It will solve the problem. But it has an app_fault issue at pds_init() when I enable/disable softdevice 6 times. Can I change my code to the following? But at the function of pm_init() also calling the sub-functions of pds_init(), pdb_init(), sm_init(), smd_init(), gcm_init() and gscm_init(). How to modified them? Do you have the example code of enable/disable softdevice? What is the correct method to enable/disable softdevice?

ret_code_t pds_init()
{
ret_code_t ret;
// Check for re-initialization if debugging.
//NRF_PM_DEBUG_CHECK(!m_module_initialized);
if(m_module_initialized)
return;
ret = fds_register(fds_evt_handler);
if (ret != NRF_SUCCESS)
{
...
}
}

RE: Hardfault at Enabled/Disabled Softdevice

Jared — Tue, 14 Jul 2020 13:01:10 GMT

Hi,

Could you try calling ble_conn_state_init() before you call peer_manager_init();

RE: Hardfault at Enabled/Disabled Softdevice

Ted — Mon, 13 Jul 2020 00:34:26 GMT

Hi Jared,

I am not using FreeRTOS. The softdevice is s140_nrf52_7.0.1_softdevice.hex.

RE: Hardfault at Enabled/Disabled Softdevice

Jared — Fri, 10 Jul 2020 08:23:25 GMT

Hi,

Thank your the information from debugging. Which Softdevice and version are you using? And are you using FreeRTOS?

regards

Jared

RE: Hardfault at Enabled/Disabled Softdevice

Ted — Fri, 10 Jul 2020 00:44:03 GMT

After enabling/ disabling softdevice. It causes hardfault when I re-enable the softdevice at the third time. The member of m_flag_service_changed_pending is equal to BLE_CONN_STATE_USER_FLAG_INVALID. So the function of gcm_init will return NRF_ERROR_INTERNAL. And the program goes to app_error_fault_handler. Do I need to clear connect state? And how to clear it correctly?

RE: Hardfault at Enabled/Disabled Softdevice

Jared — Thu, 09 Jul 2020 13:53:49 GMT

Hi,

Could you set a breakpoint in gcm_init() and see where it asserts inside the function?

regards

Jared