Add driver for generic HID joysticks/gamepads

[valpackett]

Tested on an off-brand cheap USB gamepad with <16 buttons :D but I'm relatively confident in the lots-of-buttons (BTN_TRIGGER_HAPPY) support code.

The list of individual buttons in struct hmap_item[] sucks, would be nice to have a .usage_mask instead of .usage that would allow me to match all HUP_BUTTONs…

[wulf7]

Thanks for hgame driver.

Could you rework it without using callbacks? Although currently callbacks look better, there are plans to implement HID usage ranges and scatter-gather maps for such a drivers

[valpackett]

• Removed callback usage for regular buttons
• Added support for d-pad usages (that required a new callback though :D)
  • basically, "hat switch" is when there's one HID usage that returns a value like: 1 is up, 2 is up-right, 3 is right, …
    • Linux exposes that value directly as the X axis of a hat switch
  • "d-pad" is like 4 separate buttons
    • but Linux exposes that as two axes of a hat switch that go from -1 to 1, hence somewhat ugly callback
• Added correct button mapping for Xbox 360 controllers
  • currently matching by vendor (Microsoft), would be nice to somehow use the USB_IFACE_SUBCLASS(UISUBCLASS_XBOX360_CONTROLLER) knowledge from usbhid instead
    • interestingly, the xpad driver for Linux keeps a list of vendor/device pairs that includes 3rd party vendors. (My 3rd party gamepad just pretends to be Microsoft when it's in XInput mode lol)

[wulf7]

Merged, thanx

that required a new callback though
hence somewhat ugly callback

Callbacks were added for things like this one, They should be used if translation can not be written in declarative way.

currently matching by vendor (Microsoft), would be nice to somehow use the USB_IFACE_SUBCLASS(UISUBCLASS_XBOX360_CONTROLLER) knowledge from usbhid instead

May be create hgame.h with required constants?

[valpackett]

May be create hgame.h with required constants?

I mean how would I pass the info between drivers.. I guess via hid_device_info, which is passed as ivars

[wulf7]

Yes, hid_device_info is intended to pass information between layers. Do you want place quirks here too?

[valpackett]

Yeah, I guess. (I specifically need to pass a USB-level match to a HID driver)

[wulf7]

Other way is a global function that compares hid_device_info structure with the predefined quirk list. That is what USB's usb_test_quirk (and usb_quirk(4)) do. But I never dug in to it so I do not know if it can replace passing quirks between drivers.