Central / Client DB Discovery

[Imperatori]

Hi,

I'm trying to discovery the content of a vendor specific GATT service. So far I'm receiving the BLE_DB_DISCOVERY_COMPLETE event and I'm able to see that I got 4 entries. All characteristics have a different UUID (in more bytes, not only 12/13).

     for (i = 0; i < p_evt->params.discovered_db.char_count; i++)
  {
     ble_gatt_db_char_t * p_characteristic;
              
     p_characteristic = &(p_evt->params.discovered_db.charateristics[i]);

     //ble_gatt_char_props_t props = p_characteristic->characteristic.char_props;
     //p_characteristic->characteristic.handle_decl
     //p_characteristic->characteristic.handle_value
     //p_characteristic->characteristic.uuid.type
     //p_characteristic->characteristic.uuid.uuid
     
     // UartPutChar(p_characteristic->cccd_handle);
 }

Each characteristic seem to have a proper handle for the declaration and value. On the other hand, the uuid struct is showing only zeros in both, type & uuid. The CCCD handle is also set to 0. In the sniffer I see the Response to the "Read by type" request, with the 'valid' UUID content.

There is something 'special' to be performed in case of a discover over 'different' UUIDs ?

Kind regards Giona

[Petter Myhre]

Have you added the UUID bases? This may also be useful.

[Imperatori]

Yes, but this causes the SD function "sd_ble_uuid_vs_add" to exit with error "NO_MEM = 4" for the second VS entry. Basically it seems to be easy to fix, by changing the parameter vs_uuid_count, but this causes the softdevice_enable() to exit with the NO_MEM error. I posted a question on this link here

[Petter Myhre]

When you increase vs_uuid_count the SoftDevice will require more RAM. So you will have to change the IRAM1 start and size used by the application.

To see how much RAM the SoftDevice requires go into softdevice_enable() and put a breakpoint after sd_ble_enable() is called. The value of app_ram_base should be used as IRAM1 start.

[Imperatori]

This fixed the error NO_MEM. I would add these information in the documentation of the sd_ble_uuid_vs_add function.

Adding all vs UUIDs fixed the topic's issue.

Thanks!