NUS Multiperipheral error after pm_conn_secure

[Konstantin S]

Hi, I'm working with nRF52832 DK,  SDK 15 on Windows with Keil5. I create Multiperiherial Nus with peer manager and fstorage and it works, but after some connetions I get an error in pm_conn_secure() error code 86.I dont know how many connections are needed for this error, maybe about 30. If i erase crystal in keil and programm nRF again i see that all working fine again and after some connections/disconnections error appears again.

[AndreasF]

Hi.

Could you try to see which NRF_ERROR pm_conn_secure() returns? It is much easier to know whats faulty then. There is documentation on how to debug on Keil's website here.

Best regards,

Andreas

[Konstantin S]

I see that link_secure() return NRF_ERROR_STORAGE_FULL 

[AndreasF]

Hi.

This happend because smd_link_secure(); returned NRF_ERROR_STORAGE_FULL.

/**@brief Function for initiating security on the link, with the specified parameters.
 *
 * @note If the connection is a peripheral connection, this will send a security request to the
 *       master, but the master is not obligated to initiate pairing or encryption in response.
 * @note If the connection is a central connection and a key is available, the parameters will be
 *       used to determine whether to re-pair or to encrypt using the existing key. If no key is
 *       available, pairing will be started.
 *
 * @param[in]  conn_handle      Handle of the connection to initiate pairing on.
 * @param[in]  p_sec_params     The security parameters to use for this link. As a central, this can
 *                              be NULL to reject a slave security request.
 * @param[in]  force_repairing  Whether to force a pairing procedure to happen regardless of whether
 *                              an encryption key already exists. This argument is only relevant for
 *                              the central role. Recommended value: false
 *
 * @retval NRF_SUCCESS                    Success.
 * @retval NRF_ERROR_NULL                 p_sec_params was NULL (peripheral only).
 * @retval NRF_ERROR_INVALID_STATE        A security procedure is already in progress on the link,
 *                                        or the link is disconnecting.
 * @retval NRF_ERROR_INVALID_PARAM        Invalid combination of parameters (not including conn_handle).
 * @retval NRF_ERROR_INVALID_DATA         Peer is bonded, but no LTK was found, and repairing was
 *                                        not requested.
 * @retval NRF_ERROR_BUSY                 Unable to initiate procedure at this time.
 * @retval NRF_ERROR_TIMEOUT              There has been an SMP timeout, so no more SMP operations
 *                                        can be performed on this link.
 * @retval BLE_ERROR_INVALID_CONN_HANDLE  Invalid connection handle.
 * @retval NRF_ERROR_STORAGE_FULL         No more room in flash. Fix and reattempt after the next
 *                                        FDS garbage collection procedure.
 * @retval NRF_ERROR_INTERNAL             No more available peer IDs.
 */
ret_code_t smd_link_secure(uint16_t               conn_handle,
                           ble_gap_sec_params_t * p_sec_params,
                           bool                   force_repairing);

The error is returned because the is not enough room in your flash memory. You have to retry after the next FDS garbage collection procedure.

Best regards,

Andreas

[Konstantin S]

i use  pm_peer_delete(p_gap_evt->conn_handle) in BLE_GAP_EVT_DISCONNECTED, and I thought it would remove all connection data. Now i add  

//Run garbage collection on the flash

 // Upon disconnection, reset the connection handle of the peer which disconnected, update
        // the LEDs status and start scanning again.
        case BLE_GAP_EVT_DISCONNECTED:
        {
					
			err_code = pm_peer_delete(p_gap_evt->conn_handle);
			APP_ERROR_CHECK(err_code);

            NRF_LOG_INFO("Central link 0x%x disconnected (reason: 0x%x)",
                         p_gap_evt->conn_handle,
                         p_gap_evt->params.disconnected.reason);
					
			// Run garbage collection on the flash.
				do
					{
						err_code = fds_gc();
					}while(err_code != 0);
				APP_ERROR_CHECK(err_code);
				
            if (ble_conn_state_central_conn_count() == 0)
            {
                err_code = app_button_disable();
                APP_ERROR_CHECK(err_code);

                // Turn off connection indication LED
                bsp_board_led_off(CENTRAL_CONNECTED_LED);
            }
											
            // Start scanning
            scan_start();

            // Turn on LED for indicating scanning
            bsp_board_led_on(CENTRAL_SCANNING_LED);

        } break;

and its working. This is the right solution ?

[AndreasF]

Hi.

I would use an app timer in the section  // Run garbage collection on the flash. where you call fds_gc();

/**@brief   Function for running garbage collection.
 *
 * Garbage collection reclaims the flash space that is occupied by records that have been deleted,
 * or that failed to be completely written due to, for example, a power loss.
 *
 * This function is asynchronous. Completion is reported through an event that is sent to the
 * registered event handler function.
 *
 * @retval  FDS_SUCCESS                 If the operation was queued successfully.
 * @retval  FDS_ERR_NOT_INITIALIZED     If the module is not initialized.
 * @retval  FDS_ERR_NO_SPACE_IN_QUEUES  If the operation queue is full.
 */
ret_code_t fds_gc(void);

If you at some point in time get the error FDS_ERR_NO_SPACE_IN_QUEUES, you might end up in an infinite loop since there is no time to process the queue.

Best regards,

Andreas

[Konstantin S]

Yes, my mistake.

Thank you for help.