Update tinyusb to fix gamepad;add HID OUT interface descriptor

[dhalbert]

Fixes #1696.

• Needed to update tinyusb to bring in HID fixes necessary for gamepad to work on XAC controller.
• Added HID OUT interface descriptor to match many existing devices. Not yet supported by CircuitPython libraries.

To use the XAC-compatible gamepad descriptor, add a line like this to, say, ports/atmel-samd/boards/circuitplayground_express/mpconfigboard.mk:

USB_HID_DEVICES=XAC_COMPATIBLE_GAMEPAD

[tannewt]

Code looks ok. CI failed because STM files got shuffled around in the TinyUSB update.

[dhalbert]

@hierophect Note the change in 463415a. Now tinyusb distinguishes which STM32 USB device to support based on the origin (IP) of the USB device. In this case, synopsys.

[dhalbert]

@tannewt I fixed stm builds.

[tannewt]

Great! Thank you!

[hierophect]

@dhalbert @tannewt Sadly, this has broken USB on the STM32F412, by excessively restricting the FS endpoints on that device. Also testing F411 which may suffer from the same issue. There are 6 FS endpoints on the F412, should be enough to support all circuitpython devices. Specific error:

0x0804a8aa in dcd_edpt_open (rhport=0 '\000', desc_edpt=0x8068862 <usb_desc_cfg+194>) at ../../lib/tinyusb/src/portable/st/synopsys/dcd_synopsys.c:273
273	  TU_ASSERT(epnum < EP_MAX);

[hierophect]

Has something to do with the endpoint count. Could you loop me in as a reviewer next time you push an ST update? I'd like to help out with testing to help prevent issues like this. I didn't realize from the mention that this changed anything other than file names.

[dhalbert]

@hierophect Sorry about this. Is this due to the OUT endpoint for HID? I thought this would not be an issue because it's the OUT half of an endpoint pair.

[hierophect]

@dhalbert if you gave it a different index, probably - that would be similar to the doubled CDC index causing the same issue in the past. you didn't, though. Endpoint pairs should not count. So I think this is a tinyusb update error relating to the incrementation of epnum.

[hierophect]

@dhalbert My fix at the moment, by the way, is simply increasing the number of endpoints allowed by 1, and everything is working fine.