> Since we're on the subject, where is this madness coming from?:

> Oct 17 16:04:57 localhost /netbsd: fd0: write failed st0 40<abnrml> st1 10<data_
> overrun> st2 0 cyl 9 head 0 sec 13

That comes from the fact that the IBM PC floppy system was designed by
a Nazi war criminal.

The floppy disk channel has woefully inadequate buffering, which of course
means that it is placed near the bottom of the food chain in terms of DMA
priority.  Hence, SCSI disk activity is almost certain to cause data
overruns and underruns (depending on which direction the DMA is going).
I generally do a sync before accessing the floppy drive.