Okay, I applied the patch and build a GENERIC kernel from it. When I 
re-enabled the USB ports and booted the new kernel, "systat vmstat" was 
showing 200000 pin 11 interrupts per sample period, and over 60% CPU 
time spent in interrupt. I tried extracting pkgsrc.tar.gz as a speed 
test, and the machine locked hard.

I then disabled the USB ports and rebooted the new kernel. The SATA 
controller is definitely capable of much higher throughput with the 
patch applied, which I appreciate. The machine is speedy, even while 
it's rewriting its RAID parity.

However, I'm still in essentially the same position: I can either have 
my USB controller enabled, and the machine is barely usable; or I can 
have good performance but no console keyboard (because USB is disabled). 
Any thoughts on what I can do to diagnose the USB problems? Would 
"options ACPI" and friends make any difference? Should I try enabling 
the ehci?

Thanks again for the help.

Chris

Blair Sadewitz wrote:
> You don't have bus master DMA enabled, presumably because there's no
> entry in pcidevs for it.  This will force use of PIO mode.  Could you
> try adding a line to pcidevs beneath the entry for 0x024a for 0x024b,
> otherwise identical except, obviously, for the PCI device name?
>
> Regards,
>
> --Blair
>