Re: "Bluetooth: hci0: HCI_REQ-0xfcf0" errors in dmesg

From: Mateusz Jończyk
Date: Mon Oct 31 2022 - 17:58:33 EST


W dniu 31.10.2022 o 22:08, Gix, Brian pisze:
> On Mon, 2022-10-31 at 13:37 -0700, Luiz Augusto von Dentz wrote:
>> On Mon, Oct 31, 2022 at 1:15 PM Mateusz Jończyk <mat.jonczyk@xxxxx>
>> wrote:
>>> On 6.1.0-rc3, I have been seeing this error message in dmesg:
>>>
>>>         Bluetooth: hci0: HCI_REQ-0xfcf0
>>>
>>> It is printed on every boot, resume from suspend and rfkill unblock
>>> of the Bluetooth device.
>>> The device seems to be functioning normally though (but I have done
>>> limited testing only).
>>>
>>> After some investigation, it turned out to be caused by
>>>         commit dd50a864ffae ("Bluetooth: Delete unreferenced
>>> hci_request code")
>>> which modified hci_req_add() in net/bluetooth/hci_request.c to
>>> always print an error message
>>> when that function is executed.
[...]
> This, I suspect, will be an ongoing issue, particularily with Vender
> Opcodes, which we don't have much direct visability and testing
> capabilities. The error gets logged because it is using a deprecated
> (but still working) mechanism to issue HCI opcodes.
>
> If I can get the source code path of the call as it is being
> contructed, I can try making you a patch Mateusz... But you will
> probably need to do the testing/verification for me.

Sure, what info would you like me to provide? The MSFT
Vendor Opcode 0xFCF0 is defined in drivers/bluetooth/btrtl.c, line 725.

I'm not sure if I have any additional Bluetooth LE devices around, unfortunately.

Additionally, I think that the error message is a bit cryptic, and perhaps could
be changed and/or demoted to a notice for kernel 6.1.

Greetings,

Mateusz