spurrious interrupt problem

[Andrew Gallatin]

I have a quad processor amd64 box running RELENG_5 where we're seeing 
some spurrious interrupts from the usb ohci controller.

The people on site claim there is nothing plugged in to the USB ports,
yet we get tons of ohci interrupts.  Note the 1:1 relationship
with interrupts for bge0 -- irq 19 and irq 31 seem to be "wired"
together.  Eg, for every interrupt on irq 31 (bge), I get a spurrious
interrupt on irq 19 (ohci).

% vmstat -i
interrupt                          total       rate
irq0: clk                        2662019       1023
irq4: sio0                           192          0
irq6: fdc0                             5          0
irq8: rtc                         332638        127
irq14: ata0                           47          0
irq19: ohci0 ohci1                 15019          5
irq29: mpt0                           16          0
irq30: mpt1                         1746          0
irq31: bge0                        15011          5
Total                            3026693       1163


I've left verbose boot, mptable and acpidump output at
http://people.freebsd.org/~gallatin/quartet

The box is ~3000 miles away and getting it power-cycled is a pain, so
I haven't tried removing device acpi, as I'm not sure that's supported
on AMD64.


Thanks,

Drew