Why is scan_evt.params.p_not_found used when we use the whitelist

[pressure_sensor_nonconn_adv]

I am looking at the example of the HRS collector (Central) of the SDK also called ble_app_hrs_c.

In the file nrf_ble_scan.c in the function nrf_ble_scan_on_adv_report we have the following code:

    // If the whitelist is used, do not check the filters and return.
    if (is_whitelist_used(p_scan_ctx))
    {
        scan_evt.scan_evt_id        = NRF_BLE_SCAN_EVT_WHITELIST_ADV_REPORT;
        scan_evt.params.p_not_found = p_adv_report;
        p_scan_ctx->evt_handler(&scan_evt);

        UNUSED_RETURN_VALUE(sd_ble_gap_scan_start(NULL, &p_scan_ctx->scan_buffer));
        nrf_ble_scan_connect_with_target(p_scan_ctx, p_adv_report); // connects with target! bad

        return;
    }

We see that the p_adv_report is saved in scan_evt.params.p_not_found and then the handler is called.

What I don't understand:

•          Where do we even check if the address is on the whitelist? we just check if the whitelist ist used. Does this happen before calling nrf_ble_scan_on_adv_report ?
•          Why do we not use p_whitelist_adv_report instead of p_not_found ?

Thanks for your help!

[Vidar Berg]

Hello,

Where do we even check if the address is on the whitelist? we just check if the whitelist ist used. Does this happen before calling nrf_ble_scan_on_adv_report ?

 Yes, the device filtering is performed in HW using the whitelist set by the application before scan start. This means you will not recive a BLE_GAP_EVT_ADV_REPORT from a device that is not in the list.

Why do we not use p_whitelist_adv_report instead of p_not_found ?

I'm sure this must be a bug and that it should been using the p_whitelist_adv_report pointer. I will report this internally.

Best regards,

Vidar

[pressure_sensor_nonconn_adv]

Thank you for your response.

Vidar Berg said:

I'm sure this must be a bug and that it should been using the p_whitelist_adv_report pointer. I will report this internally.

Can you confirm that this is a bug? I checked the SDK v16 and it also uses p_not_found...so maybe it there is another reason for this

[Vidar Berg]

Yes, I can confirm that this is a bug. However, if you take a look at the 'scan_evt_t' structure, you can see that both p_whitelist_adv_report and p_not_found are of the same type and stored in the same union, so it doesn't make an actual difference if you use .p_not_found or .p_whitelist_adv_report. This probably explains why this has gone unnoticed for so long.