Rather than an "iointr" routine that decomposes a vector into an
IRQ, we maintain a vector table directly, hooking up each "iointr"
routine at the correct vector. This also allows us to hook device
interrupts up to specific vectors (c.f. Jensen).
We can shave even more cycles off, here, and I will, but it requires
some changes to the alpha_shared_intr stuff.
(Lynx), written from scratch by me over a year ago, but never committed
to the tree because there was a bug I could never quite find. I have
fixed a few problems in the code, but still don't know if that bug is
quite fixed. Since I don't have access to the hardware directly, I'll
have to call for testers again.
