hello Nick. I've been doing this work under NetBSD-5 and the ehci(4) driver definitely doesn't behave well when sent a zero-length packet under that branch. I also checked out your USB branch, as well as the -current branch from cvs, and looking through the cvs logs, I don't see specific notes about where the ehci(4) , uhci(4) or ohci(4) drivers were taught to handle zero length packets. Could you point me at the specific commits that you think address this issue? I plan to port these changes to the -current tree when I'm done, but I want to get things working with some confidence before I do that.
-thanks -Brian