Turns out the tctrl0 interrupt handler wasn't masking out the undefined
bits in the microcontroller status register. On my particular sparcbook,
those do not all read as zeros.

So tctrl_intr() looped forever, falsely believing that it was being
swamped by interrupts.

The fix is a two-liner, I've already committed it to -current.

Now to figure out how to do the pull-up bit so I can use 1.5_ALPHA
on this machine...

Todd Whitesel
toddpw @ best.com