Hi, I"m the one who posted the Linux MX21 USB Host Controller Driver mentionned above.
I should point out that is is _only_ the HCD component for _linux_. [Ie it is not a complete USB stack - linux already provides that fortunately, nor is it likely to be easilly portable to another (RT)OS.
I expect to post I new version soon which has most of the problems worked out and runs the usbtest suite sucessfully.
I've still got one strange problem : when multiple devices are using the bus it sometimes happens that the transfer completes (as witnessed by the completion code in the ETD being != 0xF (not accessed) and generally == 0x00 (Success)). However the ETD is not marked as done (and I don't get an interrupt). The problem never occurs with a single device using the bus.
I've worked around this in my driver by checking for this condition at each SOF but its not pretty.
Does anyone know anything about this? (can't find any mention of it in the errata).
Martin