>Fine. I'll accept that you have experience. Can you tell me, then, on
>which architectures other than Wintel boxes you have tried this new
>bus_dma design on, and what the results have been?

None, but you seem to believe that I have claimed something I haven't. I
have attempted to show the justification for the changes that I made to the
bus dma interfaces in bringing them to FreeBSD.  I have stated that my
*goal* is to provide a more efficient API that yields additional
flexibility to implementations of this API while still retaining all of the
architecture neutrality of the old interface. If there are portability
problems with my changes (primarily the addition of a callback, the removal
of a mandatory copy of the S/G list in the dma map object, and making dma
tag and dma map objects completely opaque) I certainly want to hear about
them.

I must say that it still amazes me how much this group lets its emotions
get in the way of progress.  Is it so inconceivable that a piece of code
originating in NetBSD could be improved upon?  Does it simply hurt your
egos too much when a third party, operating fully within the bounds of the
license on the code, modifies it, then has the audacity to submit the
changes back and claim their modifications might have made the code better?
I have tremendous respect for the work Jason and others have done in
NetBSD.  Because of that respect, I'm certain to dip from the NetBSD well
again to improve FreeBSD.  It doesn't seem that this respect is
reciprocated, which is NetBSD's loss.  It only hurts NetBSD to ignore the
many things FreeBSD has accomplished.

--
Justin