I'm beginning to suspect a "high kernel" problem (not
driver-specific).

Some experimental observations:

The pauses appear to correlate well with sync(2) invocations and
writing of dirty data to disk.

	- pauses are not obvious for me when I kill update
	- when i replace update with a:

		while :
		do
			echo -n x; sync
			sleep 15
		done
	pauses corrlate with the printing of "x"'s

	- pauses are significantly less severe when you turn off atime
	  updates (with mount -u -o noatime <filesystem>)

I suspect that's what going on here is a scaling problem; systems are
dirtying data at ever-faster memory speeds, and then flushing it at
disk speeds only every 30s, leading to a significant pause..

It's not immediately clear what's causing the other processes to block
during the sync; ffs_update is one likely place (for instance, if the
bread() call it makes misses in the cache..).

					- Bill