Hi Juha, On Thu, Dec 22, 2016 at 7:46 AM, Juha Kuikka <juha.kuikka@xxxxxxxxx> wrote: > Hi, > > I believe I have identified an issue with BLE HoG service. > > It is possible for the HoG service to get probed before the Device > Information Service has finished reading the VID&PID of the > peripheral. > > This causes the uHID device to get created with zero VID&PID. For most > devices this is not an issue but if there is a need to have a quirk > driver or use of hidraw this will make identifying the device > impossible. > > This only seems to affect the first connection, on subsequent > connections the cached VID&PID are used. > > Here is the output from a slightly instrumented bluetoothd 5.43. > bluetoothd[2479]: profiles/gap/gas.c:gap_probe() GAP profile probe > (AA:BB:CC:44:16:AF) > bluetoothd[2479]: src/service.c:change_state() 0x22aef10: device > AA:BB:CC:44:16:AF profile gap-profile state changed: unavailable -> > disconnected (0) > bluetoothd[2479]: profiles/gap/gas.c:gap_accept() GAP profile accept > (AA:BB:CC:44:16:AF) > bluetoothd[2479]: profiles/gap/gas.c:handle_characteristic() > Unsupported characteristic: 00002a04-0000-1000-8000-00805f9b34fb > bluetoothd[2479]: src/service.c:change_state() 0x22aef10: device > AA:BB:CC:44:16:AF profile gap-profile state changed: disconnected -> > connected (0) > bluetoothd[2479]: src/device.c:device_probe_profiles() Probing > profiles for device AA:BB:CC:44:16:AF > bluetoothd[2479]: src/device.c:device_probe_profiles() Probing > profiles for device AA:BB:CC:44:16:AF > bluetoothd[2479]: src/service.c:change_state() 0x22b4180: device > AA:BB:CC:44:16:AF profile deviceinfo state changed: unavailable -> > disconnected (0) > bluetoothd[2479]: profiles/deviceinfo/deviceinfo.c:deviceinfo_accept() > deviceinfo profile accept (AA:BB:CC:44:16:AF) > bluetoothd[2479]: > profiles/deviceinfo/deviceinfo.c:handle_characteristic() Unsupported > characteristic: 00002a29-0000-1000-8000-00805f9b34fb > bluetoothd[2479]: src/service.c:change_state() 0x22b4180: device > AA:BB:CC:44:16:AF profile deviceinfo state changed: disconnected -> > connected (0) > bluetoothd[2479]: src/device.c:device_probe_profiles() Probing > profiles for device AA:BB:CC:44:16:AF > bluetoothd[2479]: profiles/input/hog.c:hog_probe() path > /org/bluez/hci0/dev_AA_BB_CC_44_16_AF > > Here VID and PID are zeroes when the HoG device is created. > > bluetoothd[2479]: profiles/input/hog.c:hog_device_new() name=MyDevice > vendor=0x0, product=0x0, version=0x0 > bluetoothd[2479]: src/service.c:change_state() 0x22c8220: device > AA:BB:CC:44:16:AF profile input-hog state changed: unavailable -> > disconnected (0) > bluetoothd[2479]: src/device.c:device_set_auto_connect() > AA:BB:CC:44:16:AF auto connect: 1 > bluetoothd[2479]: src/device.c:device_set_auto_connect() Already connected > bluetoothd[2479]: attrib/gattrib.c:g_attrib_ref() 0x22aecf0: g_attrib_ref=2 > bluetoothd[2479]: profiles/input/hog-lib.c:bt_hog_attach() HoG > discovering characteristics > bluetoothd[2479]: attrib/gattrib.c:g_attrib_ref() 0x22aecf0: g_attrib_ref=3 > bluetoothd[2479]: src/service.c:change_state() 0x22c8220: device > AA:BB:CC:44:16:AF profile input-hog state changed: disconnected -> > connected (0) > bluetoothd[2479]: src/service.c:btd_service_ref() 0x22c8220: ref=2 > bluetoothd[2479]: plugins/policy.c:service_cb() Added input-hog reconnect 0 > bluetoothd[2479]: src/gatt-client.c:btd_gatt_client_ready() GATT client ready > bluetoothd[2479]: src/gatt-client.c:create_services() Exporting > objects for GATT services: AA:BB:CC:44:16:AF > bluetoothd[2479]: src/gatt-client.c:service_create() Exported GATT > service: /org/bluez/hci0/dev_AA_BB_CC_44_16_AF/service0008 > bluetoothd[2479]: src/gatt-client.c:service_create() Exported GATT > service: /org/bluez/hci0/dev_AA_BB_CC_44_16_AF/service0009 > bluetoothd[2479]: src/gatt-client.c:characteristic_create() Exported > GATT characteristic: > /org/bluez/hci0/dev_AA_BB_CC_44_16_AF/service0009/char000a > bluetoothd[2479]: src/gatt-client.c:characteristic_create() Exported > GATT characteristic: > /org/bluez/hci0/dev_AA_BB_CC_44_16_AF/service0009/char000c > bluetoothd[2479]: src/device.c:device_svc_resolved() > /org/bluez/hci0/dev_AA_BB_CC_44_16_AF err 0 > bluetoothd[2479]: src/adapter.c:add_device_complete() > AA:BB:CC:44:16:AF (2) added to kernel connect list > bluetoothd[2479]: profiles/gap/gas.c:read_device_name_cb() GAP Device > Name: MyDevice > bluetoothd[2479]: profiles/gap/gas.c:read_appearance_cb() GAP Appearance: 0x03c2 > bluetoothd[2479]: src/adapter.c:new_conn_param() hci0 > AA:BB:CC:44:16:AF (2) min 0x0006 max 0x0008 latency 0x000a timeout > 0x012c > bluetoothd[2479]: src/adapter.c:store_conn_param() > bluetoothd[2479]: src/adapter.c:new_irk_callback() hci0 new IRK for > AA:BB:CC:44:16:AF RPA 00:00:00:00:00:00 > > And they are set here. > > bluetoothd[2479]: src/device.c:btd_device_set_pnpid() source=2 > vendor=1234 product=5678 version=1 > bluetoothd[2479]: src/adapter.c:new_long_term_key_callback() hci0 new > LTK for AA:BB:CC:44:16:AF type 0 enc_size 16 > bluetoothd[2479]: src/device.c:device_set_bonded() > bluetoothd[2479]: src/device.c:device_bonding_complete() bonding (nil) > status 0x00 > bluetoothd[2479]: src/adapter.c:resume_discovery() > > What would be a good approach to defer the HoG probe until the DIS is > ready? Or should I look into a mechanism to indicate I wouldn't create a dependency of different profiles as this could delay the HoG discovery making it too slow and we would also have to make a mechanism to have the core to pass the values in an async manner, so it seems simpler to make HoG itself to check if there PID, VID, etc is not set attempt to read them by itself. Note that we have yet to port bt_hog to use gatt_db and bt_gatt_client, but Im planning to work on these items in the next couple of weeks. -- Luiz Augusto von Dentz -- To unsubscribe from this list: send the line "unsubscribe linux-bluetooth" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
