Allow disabling CDC with -DCDC_DISABLED

[DanielGibson]

This change makes it relatively easy to disable the CDC at compiletime.
Disabling it of course means that the serial console won't work anymore, and that you can only flash directly after resetting the board, but I think that's not much of a problem.

CDC_DISABLED is also used in ArduinoCore-samd for the same purpose,
see https://github.com/arduino/ArduinoCore-samd/blob/master/cores/arduino/USB/USBDesc.h#L24-L27

Disabling CDC helps making Arduino-based USB devices work with old or otherwise problematic hardware or software.
For example, USB Boot Keyboards don't work with many KVM switches if the CDC sub-devices are there; I'm sure there's other systems with the same problem (BIOS/UEFI implementations? Old operating systems - someone mentioned assistive devices still running Windows XP). Apart from Keyboards, I read that disabling CDC for Arduino-based MIDI devices.

based on https://github.com/gdsports/usb-metamorph/tree/master/USBSerPassThruLine

See also NicoHood/HID#225 and arduino/Arduino#6387 and https://forum.arduino.cc/index.php?topic=545288.msg3717028#msg3717028

[teliot]

This rolls back facchinm/ArduinoCore-avr@a1ba49e

[DanielGibson]

Why was this compiletime-option removed in the first place?

It's very useful, after all.

[teliot]

It looks what was removed was a define to enable/disable the CDC endpoint. My assumption is the current USB capabilities were made to work awaiting further improvement...

I can see that around the same period a few changes were made such as PluggableUSB and HID which forces a CDC endpoint. This is restricting and would be nice to be able to expand USB capabilities without blowing up the size of the compiled binaries. However its been like this for 5 years now.

[DanielGibson]

I can see that around the same period a few changes were made such as PluggableUSB and HID which forces a CDC endpoint.

I'm not sure I understand - with my patch I'm using HID without CDC being enabled?

[CLAassistant]

All committers have signed the CLA.

[mdevaev]

@facchinm Could you take a look at this? The problem really exists and affects several projects that are working on boot hid devices.

[NicoHood]

I 2nd this PR. It would help a lot of people reporting issue with the HID Keyboard at bios time!

[SukkoPera]

+1

[DanielGibson]

Any chance this can get merged?
If not, what should I change?

pinging @facchinm @silvanocerza

[teliot]

EDIT: The below issue does not effect this patch and is to be ignored.

Found an issue with the change.
The caterina bootloader is not removing the USB interrupts. Without code clearing the USB interrupts once sei() is called my sparkfun pro micro unit locks up. Should clear these interrupts if CDC_DISABLED is defined.
#179
USBCON = 0;
USBINT &= ~(1 << VBUSTI);
UHWCON &= ~(1 << UVREGE);

[DanielGibson]

Interesting - I didn't have that problem with my cheap Pro Micro clone - do you get it from just disabling CDC or only if you completeley disable USB?

[DanielGibson]

Also, what changes exactly must be done?
https://github.com/arduino/ArduinoCore-avr/blob/master/bootloaders/caterina/Caterina.c doesn't really look like anything mentioned in #179 (or your comment): USBCON, USBINT and UHWCON aren't used at all

[teliot]

I was having issues when i removed the USB code and was using hardware serial. The USB interrupts were still firing away and locked up at the first sei(); until i changed those registers. I will do a proper test now using your CDC_DISABLED patch but i suspect the same result as caterina bootloader is leaving usb in a semi initialized state.

[teliot]

I tested the patch and found no issues as it does still leave USB connected to the host which is different then what I have been doing.

[SukkoPera]

I've also been using this patch for a while and it really works like a charm, please merge.

[facchinm]

Merged, sorry for the incredibly long delay

[DanielGibson]

Thanks a lot! :)