Remove extranious memory barriers, and correct the placement of a few others.
This provides a 30% reduction in system time and a 6% reduction in wallclock time
for a make buildworld on my xp1000 (one 21264).

FWIW, I've been running this for nearly 2 months without problems.

Portions submitted by: ticso, jhb
Tested by: jhb (ds20 dual 21264)