On 31/07/2020 7:54, Cameron Gutman wrote:

On 7/30/20 8:44 AM, Ismael Ferreras Morezuelas wrote:

quoted

-	{ 0x1532, 0x0037, "Razer Sabertooth", 0, XTYPE_XBOX360 },

The list here doesn't actually affect which devices that xpad will attach to.
xpad_table[] is what determines the devices that xpad's probe() will be
called for. It does this using the VID + vendor-defined interface subclass
and interface number, not the VID+PID combos. xpad_device[] just provides a
friendly name and quirk flags for some supported gamepads. Gamepads that
aren't in the list will just show up as "Generic X-Box pad" with no quirk
flags applied.

If 1532:0037 is actually a mouse, it's unlikely we'd even reach the probe()
in the first place. The device would lack the expected vendor-defined Xbox
interface that is required for xpad to attach to the device.

All that said, I'm definitely not opposed to removing the entry if it isn't
a real gamepad. Given the difficulty you had finding information about this
device, I take it you don't have any 'lsusb -v' output for this device, right?

Regards,
Cameron

Hi, Cameron. Yeah, the main idea is to clean the list, even if as you say it hasn't
caused mouse issues and that's why probably nobody has noticed it before.

Because, for some reason, everyone assumes some kind of correctness even when most of
these are community-sourced and self-reported, everyone copies these lists from
somewhere else. Hopefully the fixes will trickle downstream.

So yeah, I know for sure that (for example) the internal list Valve uses comes right from here:
https://hg.libsdl.org/SDL/file/7d94464f10f7/src/joystick/controller_type.h#l246

(It's commented out)

Actual change referencing this kernel issue:
https://hg.libsdl.org/SDL/rev/29809f6f0271

For more information of the device, take a look here:
https://github.com/xboxdrv/xboxdrv/pull/59

You can see a lsusb dump here: https://github.com/xboxdrv/xboxdrv/files/76581/Qa6dBcrv.txt

So yeah, let me know what you guys think. This has been a weird tangent for me. :)