Add support for changing hid protocol mode in embassy-usb

[matteo-meluzzi]

Related issue: #2997

When a computer boots in the BIOS, it does not read full HID reports. Instead, it relies on a much simpler mode called "Boot" see https://wiki.osdev.org/USB_Human_Interface_Devices for more details. This MR adds support to switch to using that USB protocol instead of the normal reports.

• Added the following methods to RequestHandler:

trait RequestHandler {
...
   fn set_protocol(&mut self, protocol: HidProtocolMode) -> OutResponse
...
   fn get_protocol(&self) -> HidProtocolMode 
...
}

The change does not affect old code, because the default implementation of the methods is to reject the switch to Boot protocol, as it was in the previous versions of the code.

[matteo-meluzzi]

Mouse example running in the BIOS

IMG_0019.MOV

[HaoboGu]

I think just defining HidProtocolMode in embassy-usb is fine, no need to import it from usbd-hid.

[matteo-meluzzi]

fair point, i removed the import

[HaoboGu]

Is there any particular reason that config.device_sub_class and config.device_protocol are not used here?

[matteo-meluzzi]

@HaoboGu, I believe that the config.device_sub_class and config.device_protocol you're referring to belong to the USB device config ( embassy_usb::Config), while the build() function has a HID class config (embassy_usb::class::hid::Config) as parameter.

Now that I read your comment, I realize that it would be better if the usb_subclass and usb_protocol were specified in embassy_usb::class::hid::Config. I made the necessary enum definitions and updated embassy_usb::class::hid::Config. The issue with this is that it's not backwards compatible. I hope that's fine.

[HaoboGu]

I suggest to use the name from hid spec, aka No and Boot

[matteo-meluzzi]

@HaoboGu , that's a nice idea. I updated the names as you suggested. I also updated all usb_hid_mouse and usb_hid_keyboard examples.

[HaoboGu]

lgtm!

[matteo-meluzzi]

lgtm!

@HaoboGu i synced with main, we should be good to go

[HaoboGu]

It looks really good to me, but I'm not the person that has the power to merge PRs. Maybe Dirbaio wants to have a look too.

[matteo-meluzzi]

Ah I see. @Dirbaio would you mind having a look as well?