dwcotg.4: Initial stub manual draft

[ziaee]

Summary

Nobody seems to realize this is here, or that we have a driver such that you
can connect a raspberry pi 0/3/4/5 with a single usb cable to your laptop
supplying power, ethernet, and a serial console. Therefore, write a stub
manual to create visibility in apropos and the HW Relnotes.

I can't figure out how to use this yet. Val Packett told me it has something
to do with usb_template. Hopefully someone who knows more will come forth in
the review.

MFC after: 3 days

[ziaee]

Added everyone who it seems might know how to use this from bugs.

[mhorne]

I have not succeeded in using this myself, but I have never tried too hard.

There is an entire chapter in the handbook dedicated to the OTG features: https://docs.freebsd.org/en/books/handbook/usb-device-mode/

[val_packett.cool]

The "OTG" here is just part of the name of the hardware block in the SoC that provides the USB controller, which is not guaranteed to actually support OTG (role switching) or the device role at all in any particular implementation. I don't remember how exactly it was used on which RPi board… I *thiiink* it was used as the USB 2.0 host controller pre-RPi4, then when XHCI was added, it was moved to the upstream-facing Type-C port? Something like that..??

It does not have anything to do with UART whatsoever; also Synopsys is spelled with a "y".

[ziaee]

It does not have anything to do with UART whatsoever

But, when I use it with raspbian, I can call in with cu(1) to it,
so doesn't that make it a UART?

[peterj]

dwcotg is a USB controller - functionally similar to ehci(4), ohci(4), uhci(4) or xhci(4) except that it also supports OTG functionality. When you use cu(1) to connect to the device in OTG mode, you're connecting to a device behind the controller, not the controller itself. usb_template(4) documents the OTG devices that FreeBSD supports, though the available ones depend on the SoC.

[pauamma_gundo.com]

English LGTM. I can't speak to accuracy or relevance.

[peterj]

The English might be OK but the content is completely wrong. I would respectfully suggest that having accurate content should come before language issues.

[ziaee]

Nobody is pushing this until the content is verified.