USB stack issues

[Daniel Mack]

When developing the OHCI stack, I stumbled over two things in the
libpayload USB stack which I didn't want to leave unmentioned.

1. The MSC driver issues BULK messages to its devices with 0 bytes in
length. That is, in fact, a vliad operation according to the USB
standard, but mass storage devices dodn't like that and time out on such
a request. The UHCI stack bails out in this case and returns success,
and so does the OHCI stack currently. However, this is a hack. The real
fix would be to not let the MSC driver issue these empty packets.

2. Calling usb_detach_device() makes Coreboot crash.I don't see how
the OHCI stack could possibly cause that, so I blame the generic USB
stack for that :)

Daniel