The detection code looks simple enough in this initial commit. Maybe the lmp_subver is higher than what they check for with this: le16_to_cpu(rp->lmp_subver) <= 0x22bb or it’s not the expected bluetooth version.
edit:
I have checked the current linux/drivers/bluetooth/btusb.c at master · torvalds/linux · GitHub they have already added a device that beats the initial checks, probably because it has a lower bt version than expected. So they just check bcdDevice == 0x0134 and for the devices bt version.
I’m not sure why did they do it this way, it might be easier to check if it’s not fake. There could be hundreds of fake devices, while it seems there are only some old bt 1.1 devices that are not fake. Or I’m missing something…
I think a quick dirty patch for @ReticentRobot would be to just set is_fake = true; It’s unlikely he has a real CSR device. A proper patch I suppose needs a debugger or print to get the values from the device. I have never done kernel debug before.