Just thought I'd mention it in case anyone else has encountered the same
problem I had been having with this combination. I reported this a couple
of months ago. After lots of trial and error I finally found that using the
pirq="" kernel option to specify the PCI irqs apparently fixes the deadlock
problem I was having. Once I set this parameter the RAID is able to survive
a good number of stress tests. Funny thing is that it doesn't matter much
how I configure the pirq option. Setting this to pretty much anything and
in any order just makes the system work. This is despite what the
documentation on pirq says in that you have to specify exactly what the
motherboard assigns to the pci slots, usually found with trial an error.
One weird thing though. The USB irq and possibly others are consistently
remapped by the kernel on bootup anyway despite the pirq assignments and
this trick still seams to work. I've attached the dmesg on this system in
case anyone wants to take a look into this weirdness.