nrf51822 can't sleep

[arthur]

my project can't goto sysoff mode, (desktop2)( project based on mouse demo)

m_prw_mgmt_set_sysoff_timeout(DEFAULT_BTLE_INACTIVITY_DISCONNECT_PERIOD);//can't sleep here!

my code.DEFAULT_BTLE_INACTIVITY_DISCONNECT_PERIOD //10s The default: 1hour

when timeout,system goto off,

m_coms_ble_wakeup_prepare

goto↓

m_coms_ble_disable

goto↓

/ /Save bond information to flash
    err_code = ble_bondmngr_bonded_centrals_store();
    APP_ERROR_CHECK(err_code);

goto↓

// Save System Attributes, if changed.
if(sys_attr_changed())//==1

M_cmd_queue.conut change "1" from "0"

：goto↓

pstorage_access_wait()

do
    {
             app_sched_execute();        
               err_code = pstorage_access_status_get(&count);
        if (err_code != NRF_SUCCESS)
        {
            return err_code;
        }        
    }
    while(count != 0);//***can't pass***

my steps:

1， download the hex 2， power on mouse 3， try to connect 4， not found any host
5， connect timeout 6， error

： System Attributes in RAM is different from that in flash.

is this ↑cause the proble? how to fix it?

[Håkon Alseth]

I am not sure if your problem is with the system off power management or with the pstorage module. This specific example has a prepare routine before entering systemoff, which is then done based on which protocol the example is running (proprietary or BLE).

Also, please see if you have any asserts occurring in your application by enabling the assert-handler: devzone.nordicsemi.com/.../

[arthur]

thank for your reply.i have the solutions for my proble.

[arthur]

1. The issue is caused by a global variable from Desktop 2 v2.2.1 not initialized and hence result in FW triggering a flash erase command due to sys_attr_changed() function returning incorrect status.

To fix this issue we need to add 1 line of code (!!!!!!!!!!) in ble_bondmngr_init() function @ ble_bondmngr.c file:

/*********************************************************************/ uint32_t ble_bondmngr_init(ble_bondmngr_init_t

• p_init) { pstorage_module_param_t param; uint32_t err_code;

  if (p_init->error_handler == NULL) { return NRF_ERROR_INVALID_PARAM; }

  if (BLE_BONDMNGR_MAX_BONDED_CENTRALS > MAX_BONDS_IN_FLASH) { return NRF_ERROR_DATA_SIZE; } : : :

  m_central.bond.central_handle = INVALID_CENTRAL_HANDLE; m_conn_handle = BLE_CONN_HANDLE_INVALID; m_centrals_in_db_count = 0; m_bond_info_in_flash_count = 0; m_sys_attr_in_flash_count = 0;

  m_crc_sys_attr = crc16_compute( NULL, 0, NULL );//!!!!!!!!!!

SECURITY_STATUS_RESET();

: : : } /**********************************************************************/

This will initialize the global variable “m_crc_sys_attr” and prevent “sys_attr_changed() function from returning incorrect status.

2. It is also inappropriate for customer to set the DEFAULT_BLE_INACTIVITY_DISCONNECT_PERIOD to be shorter then advertising timeout period. Customer should configure this inactivity period to expire after BLE advertising has timeout and system is really idle.

[Håkon Alseth]

I cannot seem to reproduce the issue you are seeing wrt. to bonding information. I've tested it with MasterControlPanel, and the CRC/m_crc_sys_attr is correctly calculated at my end, causing no erase-operation.

1. Are you testing with an unchanged example? If not, do you see the same behavior then?
2. What central are you connecting to? Win8/android/iOS?

[Håkon Alseth]

Could you see if this fix helps you out? devzone.nordicsemi.com/.../