In message <199812140910.BAA200790@pdxcs205.pdx.intel.com>,Colin Wood writes:
>something that got corrupted via interaction with the scsi code and the
>pool allocator perhaps?  there's definitely been something slightly screwy 
>going on in that area for several months now given the reports that i've

i think its actually worse than that really.  even on a 1.3.2 system i 
have trouble with the following sequence (it doesnt matter which macine
i am using):

[repeat 10 times]
	mount /dev/sd1<blah> /mnt
	umount /mnt

eventually, the kernel will hang solid and even an nmi doesnt get a reponse.