plunky@rya-online.net said:
[FreeBSD]
>   Fix the silly bug that prevented most EHCI interrupt transfers from
>   ever working correctly: the code was linking the QHs together but
>   then immediately overwriting the "next" pointers

This might corrspond to NetBSD's
revision 1.92
date: 2005/04/27 02:12:20;  author: augustss;  state: Exp;  lines: +2 -3
Don't null the qh_link point in the interrupt tree after it has been
initialized.  From Dan Ellis in kern/30003.

I don't think this is the problem. I believe now that the data toggle
stuff is buggy. If I comment out the code after
	/*
	 * For a short transfer we need to update the toggle for the missing
	 * packets within the qTD.
	 */
in the interrupt handler, the adapter initializes. If I try "btconfig -vvvv",
it hangs, so it seems that transfers up to 16 bytes work which is more
or less the same effect as I had before with the pipe size of 16.
The code above looks bogus - I can't imagine how transactions which are
not executed could contribute to the state of the data toggle protocol.
To understand more I need to have a more thourough look at the specs.

best regards
Matthias