kakra
commented
3 years ago journal is flooded with "consumer controls not detected" message between 3 and 27 per second
Yeah, I was already suspecting that something like this might happen. I've tested two different models and both seemed to work correctly. I think you're using the new Xbox X|S Controller model?
This seems minor but as I don't know what "consumer controls" are, I really cannot tell.
Consumer controls is a category in the HID specification. This includes media buttons (like pause, play, etc) which control consumer devices and do not map properly into keyboard controls (127 buttons) or other input devices (joysticks, mice, touchpads, etc). The gamepad has some of such buttons, that is the Guide button which has no proper equivalent in the keyboard mappings, and also the new Share button which neither has any proper mapping in one of the standard HID categories.
The older models use HID over Bluetooth, and thus split their mappings into up to three categories (gamepad controls, consumer controls, and - with the advent of the Elite controller - also keyboard controls). I suspect that the new controller model isn't HID because it seems to transfer its Bluetooth data in a very different format and it's a BLE (Bluetooth low energy) device. So the kernel somehow maps that to HID but doesn't categorize the buttons.
Apparently, I don't own that controller yet so I cannot test this: It might be a problem in the kernel where it doesn't categorize the mappings properly when it should, or this controller just doesn't present itself as split subdevices. Whatever that is, I still need to figure out how to create missing subdevices on the fly from the driver anyways because we will need another category for mouse controls as part of the mouse emulation which is in the making.
Currently, while parsing the HID descriptor, the kernel will assign sub devices to each category, and when we discover that in the driver, we will store that device handler in our own structures so we'll be able to emit input events on them. But that also means: If a category never shows up when the kernel initializes the lower layers of the device, we will never get a device handle i.e. for consumer controls but still see a button later which belongs there and we cannot emit it - thus the message.
The current situation was somewhat expected (although I was hoping that would never happen for consumer controls). I pushed the driver changes out to get more test coverage. It is somewhat bad that the message spams multiple times per input frame and per second, I should look at a way to fix that. But in short term, I'll not be able to fix the underlying problem which means the consumer control buttons are probably dead for you currently. Maybe it's time to back-port all the stable fixes to the v0.9 branch and release a v0.9.1 version while the v0.10 branch stays in this slightly fragile state for a while - at least for your control model.
TheEpeeist
Version of xpadneo
v0.9-23-g6162dbc
Severity / Impact
Minor
Describe the bug
After connection is established, journal is flooded with "consumer controls not detected" message between 3 and 27 per second.
Steps to Reproduce
Connect the controller through bluetooth; touch a joystick.
Expected behavior
A single message in the log, or at least a limited number.
Screenshots/Gifs
System information
My kernel is 5.11.4 with only the L2CAP patch applied.
Controller and Bluetooth information
xpadneo-btmon.txt xpadneo-dmesg.txt xpadneo-lsusb.txt
Additional context
This seems minor but as I don't know what "consumer controls" are, I really cannot tell. Like many others I have had great trouble getting the controller to connect and even with the L2CAP patch it still seems to refuse to connect at random. It has never paired with my onboard Intel AX200 bluetooth.