Make this code the same as b1_6 (mostly whitespace removal and comment updates)
for ease of comparison.  The only remaining difference is the optimization to
skip getting lq_rw_sem if we are using RR allocation.