This is a first try at making some sense out of what i can see in the usb logs.
Following those logs, i could push the driver a little bit further than what's in CVS does : the first mgmt frames are exchanged (it never worked for me, but maybe it worked for them ; see initialization for a possible reason : with the current code you must have the very same firmware as the authors to get the driver to work).
Please send comments to jean-baptiste dot note at wanadoo dot fr.
The following describes the sequence of events and tries to abstract functions from them.
You'll find C declarations for all those protocol elements in the file protocol.h file once you've patched the CVS.
Interrupt handling : what are the actions taken in the interrupt handler (urb callback of the irq pipe) – this is an abstraction of what happens in the windows driver.