I just got reproducible UDMA errors (with downgrades following)
on a new Athlon system with a VIA8233A UDMA133 capable controller.

This time, it seems that the driver is to be blamed. There are
two identical harddisks on one channel, ant the error sticked with
the "master" whether I swapped drives physically or logically.

There is some strange assymetry in the UDMA setup in pciide.c,
different from the FreeBSD and Linux drivers. Unfortunately I
didn't find chipset docs so far.

I'll report when I find out more...

best regards
Matthias