problems with bt_id_create

[frankbuss]

The documentation here:

https://docs.zephyrproject.org/apidoc/latest/group__bt__gap.html#gae11eb8ad254418c38a0e8689df25a159

says, I can call l bt_id_create() before bt_enable(). I tried it like this:

uint8_t irk[16] = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 };

int err = bt_id_create(BT_ADDR_LE_ANY, irk);
if (err < 0) {
  LOG_ERROR("bt_id_create failed (err %d)", err);
}

and I get a -22 as the error code. Do I call the function with the right arguments, and what does error -22 mean? I guess it is EINVAL, but I don't see any invalid argument.

So I tried to call it after bt_enable(). This resulted in a log message like this:

[00:00:11.336,578] <inf> bt_hci_core: Identity: 00:00:00:00:00:00 (public)    

and obviously any call to other functions, like start advertise, doesn't work anymore, because it is an invalid BLE address.

So finally I called bt_id_create() in the ble_ready callback, before the settings_load() call. This seems to work, I get a random BLE address.

But "bt show-id" shows always the same address (after reset). I set CONFIG_BT_RPA_TIMEOUT to 60, looks like it is ignored. But after reset, I get a new BLE address, so seems to work in general, just the automatic change doesn't work. How can I fix this?

[frankbuss]

PS: if I try to call bt_id_create later after some time, to manually create a new ID, it returns error -12. Is this ENOMEM? Sounds strange, since there is still lots of memory free.

[Vidar Berg]

Hello,

and I get a -22 as the error code. Do I call the function with the right arguments, and what does error -22 mean? I guess it is EINVAL, but I don't see any invalid argument.

So I tried to call it after bt_enable().

The static address value must be specified if you are calling this function before bt_enable(). From the API documentation:

"Generating random static address or random IRK is not supported when calling this function before bt_enable()."

But after reset, I get a new BLE address, so seems to work in general, just the automatic change doesn't work. How can I fix this?

Are you looking at the address included in the advertisment packet? This should include the private resolvable address generated from the IRK you made. It is not the same as the random static address generated by bt_id_create() 

frankbuss said:

if I try to call bt_id_create later after some time, to manually create a new ID, it returns error -12. Is this ENOMEM? Sounds strange, since there is still lots of memory free.

It will try to create a new ID when you call it after bt_enable() and settings load, which means you need to allocate buffer for additional IDs through the CONFIG_BT_ID_MAX symbol.

Best regards,

Vidar

[frankbuss]

Thanks, I guess I mixed up ID and Bluetooth address. What I expected was that the Bluetooth address changes, when I set an IRK, every 15 minutes by default. But the Bluetooth address doesn't change. I debugged it a bit and looks like in bt_id_set_private_addr it is only changed once after reset, but not every time the CONFIG_BT_RPA_TIMEOUT interval elapses.

Not sure, if I have to do more to trigger the change. The goal is that the Bluetooth address changes regularly, so that a user with the device is not trackable by devices which are not bonded, which is the also main reason for the IRK concept as I understand it.

[Vidar Berg]

It should be sufficient to build with CONFIG_BT_PRIVACY=y and CONFIG_BT_SMP=y to enable address rotation as far as I know. The advertisement address should then be updated after every CONFIG_BT_RPA_TIMEOUT seconds.

Please double check that both of the symbols above are enabled in your build.

[frankbuss]

Yes, both symbols are set to Y, and for testing I tried CONFIG_BT_RPA_TIMEOUT=5. I added a few BT_INFO messages in bt_id_set_private_addr in id.c and can see that it is called in 5 s intervals, but looks like BT_DEV_RPA_VALID is only after reset 0, so the message BT_INFO("RPA: %s", bt_addr_str(&rpa)); at the end is only printed once. Tested with sd-nrf v2.1.0. But I guess this didn't change much, the current master version of the function looks the same:
https://github.com/intel/zephyr/blob/main/subsys/bluetooth/host/id.c#L213

I can try to verify it with one of the sample apps, if you like. I have a nRF5340 DK. Which sample app would be useful for testing the Bluetooth address rotation?