The wrong product ID is exposed for the Wireless 360 controllers

[Omar007]

The current implementation exposes the wireless receiver itself as 4 controllers, however the receiver and controllers are actually different devices with their own product ID.

The receiver has a product ID of 0x0719 and is a hub-style device that has 4 controllers attached to it.
The controllers have a different product ID; 0x02A1.

Right now you get 4 controllers with a product ID of 0x0719.
Instead, the receiver should be an unique device with 4 children; receiver (0x0719) -> 4 controllers (0x02A1).

Because the wrong PID is advertised by the drivers, some games do not automatically pick up the controller as an Xbox controller.
This was discovered through a game running in WINE [1].

NOTE1: This potentially also applies to the Xbox receiver (PID 0x0291) but I can't personally verify this.
NOTE2: Bug was originally reported (by me) on the kernel.org bugzilla [2] but received no response for 3 months.

[paroj]

does the above commit help? (x360w_id branch)

[cgutman]

FWIW, Windows also exposes wireless Xbox 360 controllers with product ID 0x02A1.

[paroj]

yeah, from the Wine Bugreport:

Windows:
TL/DR; Both the receiver and controllers are unique devices.

• Receiver
  -- Separate device, attached to a USB port, hub-style device ('Xbox 360 peripheral')
  -- Product ID 0x0719
• Controllers (x4)
  -- Separate device, attached to the receiver, game controller
  -- Product ID 0x02A1

Linux:
TL/DR; The receiver and controllers are the same device (receiver = 4 controllers).

• Receiver
  -- Attached to a USB port, seen as 4 game controllers.
  -- Product ID 0x0719.
• Controller (4x)
  -- Not handled as a separate device.
  -- Should have Product ID = 0x02A1.
  -- In the system tree the controllers are located underneath the receiver:
  /sys/bus/usb/devices/[receiver usb]/[receiver-usb]:[controller]

[Omar007]

It does indeed help :)
If there are wireless controllers with an ID other than 0x02A1 this might have to be revised at a later time to account for that, but for now this seems to do it just fine.

[cgutman]

This patch is queued for upstream so this issue can probably be closed now.