PLEASE TEST: IPI deadlock avoidance patch

[Craig Boston]

On Fri, Aug 27, 2004 at 01:00:33AM +0200, Brad Knowles wrote:
> 	Ouch.  Bad juju, bwana.  RAID-5 really, really wants all drives 
> to be 100% identical, in every way, shape, and form.  If they're not 
> and you're very, very lucky, it might just be slow.

Heh, I knew THAT was coming ;)

Don't worry, it's a test machine that doesn't do anything important
other than filter my mail through spamassassin and otherwise stress test
things.  It is slow, especially with IDE write cache disabled, but it's
never given me trouble before.  I figure that since -- *in theory* -- it
SHOULD work if everything else is working right, it's a good way to
test timing issues across disk drivers and other things of that nature.

Hey, at least I'm not trying to do RAID-3 with that setup :)

Craig