It looks like I'm having massive interrupt routing problems on a Tyan S2882-D
board running 4.0beta2/amd64.

Everything worked fine as long as I was only using the first SCSI channel (ahd0)
but then I connected eight disks to the second channel (ahd1) and, under high
I/O load, got lots of "timedout scb is alrady complete. Interrupts may not be
functioning" messages.

Suspecting ACPI problems, I updated my BIOS.
The good news is that the problems on ahd1 disappeared.
The bad news is I'm now getting exactly the same problems on ahd0.
Also, I get "block failed to stop" messages on bge0, and dmesg says that bge0
and ahd0 share the same interrupt (I hadn't be using bge1, which shares a line
with ahd1).

Any ideas how to debug this? Anything like acpilint to check if the BIOS's
interrupt routing information is sane?

Thanks for any help. The problem is I'm supposed to put that machine into
production as a file server by the end of the month.