This not exactly a kudzu bug in itself but possibly one we can deal with in how we handle i2o, and technically a bug in the firmware. BIOS rev 1.00 rel 11 and earlier of the supertrak cannot handle re-initialisation correctly. This means that rmmod i2o_core, insmod i2o_core sequences make it rather sad. In fact its best not to unload i2o with this specific combination. The newer BIOS is reported to fix it. And in part this is also filed so support folks can find it.