The following reply was made to PR port-evbmips/31912; it has been noted by GNATS.

From: "Garrett D'Amore" <garrett_damore@tadpole.com>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: port-evbmips/31912: Au1500 OHCI doesn't work when CPU is little
 endian...
Date: Fri, 28 Oct 2005 16:13:01 -0700

 Simon Burge says that we need to conditionalize the endian swapping, 
 because the device already does it for us.
 
 Evidence suggests that this may be true, but we are having some "mixed" 
 results.
 
 Specifically, after turning off software htole32(), USB probes seem to 
 work properly -- devices come and go properly.
 
 However, trying to use, for example, a USB ASIX ethernet adapter, does 
 not work in Big Endian mode, where it works fine in little endian mode.  
 (The MII shows up properly, but it negotiates at some 10HDX setting 
 which doesn't match the switch, and hence is totally busted.   Since I 
 cannot transfer packets without a working media interface, I don't know 
 how busted it is.)
 
 I have also run into some general "fragility" of the OHCI driver (or 
 perhaps the USB framework) trying to test with other devices, which has 
 met with very little success.  (A couple of panics, etc.)  (I don't know 
 how to test keyboard/mouse devices when the serial port is on the console!)
 
 I am going to wait a while to see if I can make any further progress on 
 this before submitting my diffs to date.  If someone else wants them, 
 please let me know.  Thanks!