Softdevice is always busy on write

RE: Softdevice is always busy on write

Kenneth — Wed, 14 Jun 2023 18:40:59 GMT

[quote user="RVA"]Do you think that FIRST_CONN_PARAMS_UPDATE_DELAY is too late in order to negotiate strait after CONNECTED event ?[/quote]

It's up to you, there are no requirements by BLE that impose any specific delays here.

Kenneth

RE: Softdevice is always busy on write

RVA — Wed, 14 Jun 2023 15:17:42 GMT

Hello Keneth,

Eventually the solution I found is to I ignore the request of the peripheral, and managed to set the initiator connection parameters that are OK with the power consumption requirements.

Moreover,

FIRST_CONN_PARAMS_UPDATE_DELAY explained:

/*Time from initiating event (connect or start of notification) to first time sd_ble_gap_conn_param_update is called (5 seconds). */

means that I need to wait at least 5 seconds in order set the initiator connection parameters?

Do you think that FIRST_CONN_PARAMS_UPDATE_DELAY is too late in order to negotiate strait after CONNECTED event ?

-RVA

RE: Softdevice is always busy on write

RVA — Wed, 14 Jun 2023 10:49:41 GMT

Hello,

Unfortunately I don't have currently dev kits.

Regarding option 1. I will try to use S140 v7.3.0 in my client application and see what happens

RE: Softdevice is always busy on write

Kenneth — Tue, 30 May 2023 13:49:28 GMT

Hello,

We can't think of any reason why this should occur. I have two suggestions moving forward:

1. Try to update to S140 v7.3.0 in case this have been covered by a fix that we can't understand at the moment.

2. Try to provide us with two projects with can run on nRF52-DK's to recreate this here.

Kenneth

RE: Softdevice is always busy on write

RVA — Sun, 28 May 2023 11:32:31 GMT

Hi,

I disabled the line that you mentioned on my peripheral. unfortunately nothing has changed.

I think that the device is keep being busy after the first call to ble_nus_c_tx_notif_enable(),

but in fact the write request to handle 0x13 never happen.

Log:

devzone.nordicsemi.com/.../on_5F00_connect_5F00_evt_5F00_disabled.pcapng

Regards,

- RVA

RE: Softdevice is always busy on write

Kenneth — Thu, 25 May 2023 13:24:14 GMT

Hi,

If you have the same sdk for both projects, you may need to hack this (at least for test). But try to go inot nrf_ble_gatt.c, and in nrf_ble_gatt_on_ble_evt() comment out the on_connected_evt() in BLE_GAP_EVT_CONNECTED for the peripheral project.

See if this make a difference.

Kenneth

RE: Softdevice is always busy on write

RVA — Thu, 25 May 2023 13:05:45 GMT

Hello,

On both client and peripheral gatt init sets the mtu size by

void gatt_init(void)

{

ret_code_t err_code;

err_code = nrf_ble_gatt_init(&m_gatt, gatt_evt_handler);

APP_ERROR_CHECK(err_code);

//err_code = nrf_ble_gatt_att_mtu_periph_set(&m_gatt, NRF_SDH_BLE_GATT_MAX_MTU_SIZE);

//APP_ERROR_CHECK(err_code);

}

and

/**@brief Function for initializing the GATT library. */

void gatt_init(void)

{

ret_code_t err_code;

err_code = nrf_ble_gatt_init(&m_gatt, gatt_evt_handler);

APP_ERROR_CHECK(err_code);

err_code = nrf_ble_gatt_att_mtu_central_set(&m_gatt, NRF_SDH_BLE_GATT_MAX_MTU_SIZE);

APP_ERROR_CHECK(err_code);

}

I don't have any other place where I call sd_ble_gattc_exchange_mtu_request().

As here,

1. I commented nrf_ble_gatt_att_mtu_periph_set() call

2. Added a counter in gatt_evt_handler() to see how many times the update occurs. I found that the update happens twice after CONNECTED.

2. called the conn_params_update from main after 500ms. when this counter equal to two. and the result is the same. here is the log:

devzone.nordicsemi.com/.../update_5F00_params_5F00_later_5F00_failes.pcapng

Questions:

1. Does my issue related to this exchange?

2. How can I control mtu exchange here?

3. both client and peripheral are based on the nrf examples. no additional configuration request was added for the mtu size. where could the requests come from? maybe I miss something.

Please help!

RVA

RE: Softdevice is always busy on write

Kenneth — Thu, 25 May 2023 11:01:30 GMT

I notice there is a lot of MTU exchances here, likely partially because it's initiated by both central and client at connection event. I can see that in the failing case it seems to never end. I suggest to look into why you are seemingly calling sd_ble_gattc_exchange_mtu_request() and sd_ble_gatts_exchange_mtu_reply() in your application. It's not unlikely this will cause your gattc write to return busy.

Kenneth

RE: Softdevice is always busy on write

RVA — Thu, 25 May 2023 08:29:51 GMT

Hello,

I think this is it :

devzone.nordicsemi.com/.../update_5F00_params_5F00_issue_5F00_wireshark.pcapng devzone.nordicsemi.com/.../without_5F00_update_5F00_comm_5F00_ok_5F00_wireshark.pcapng

- RVA

RE: Softdevice is always busy on write

Kenneth — Thu, 25 May 2023 07:20:11 GMT

Hello,

Good to hear that you have made the sniffer working, but I need two files saved as .pcapng format (one working and one failing). Can you do that?

Kenneth

RE: Softdevice is always busy on write

RVA — Wed, 24 May 2023 13:23:43 GMT

Hi Kenneth,

Here I recorded repeated session of Wireshark sniffing output.

peripheral mac: d2:b3:0e:a4:68:f8

Initiator mac: db:c6:a2:70:72:98

devzone.nordicsemi.com/.../wireshark_5F00_log_5F00_68_5F00_f8.txt

Please advise !

Best regards,

RVA

RE: Softdevice is always busy on write

RVA — Mon, 22 May 2023 14:28:31 GMT

Hello Kenneth,

Thanks for your quick response.

I will try to setup the system with a sniffer and return back to you with the logs..

RE: Softdevice is always busy on write

Kenneth — Mon, 22 May 2023 12:43:25 GMT

Hello,

Yes, this is indeed strange.

Do you by any chance have the possibility to make two on-air sniffer logs of the above? e.g. using nrf sniffer for ble, ellisys or frontline.

Kenneth