>Actually, there *was* a driver (unencumbered by the author signing NDAs)
>almost ready to go over a year ago. Jason specifically asked me to not commit
>it. I've been (perhasp overly) respectful of that since then. In the mean
>time, the author has lost interest in working on this h/w, but others should
>feel free to swipe the current instantiation from OpenBSD and DTRT if their
>timer interrupt says "Intel releasing Jason has taken too long".

>It's also possibly true that the current 'lead' driver resident in FreeBSD
>(which is done by an Intel engineer, actually: Prafulla Deuskar
><pdeuskar@FreeBSD.ORG>), could be ported for NetBSD. The trouble with this
>driver is that it appears to be a port from Linux and cleaning it up to be
>acceptable for the different platforms that NetBSD has might require more time
>investment than he or Intel is willing to make.

And then there's the *third* FreeBSD driver, if_gx.c by Jonathan
Lemon. That driver addresses most of the design issues Jason raised
with your driver (some of which I know have been fixed since).  The
rxeof and tx routines in if_gx.c look awfully like some of Bill Paul's
FreeBSD drivers, which have already been bus-dma-ified and imported
into NetBSD.

Intel's stance on licensing drivers for their gig-E cards looks just
plain silly, and it can only be hurting them in the marketplace.
Whether they see it that way or not is another story, though :).