12% improvement in fork load from mm3 to mm4

autoconf build test creates about 1.2 million processes in 35 minutes. On quad xeon, there was an improvement of ~ 12% between 2.5.70-mm3 and 2.5.70-mm4.

Avg of 3 runs

2.5.69-mm5    830.3 seconds
2.5.70        732.8
2.5.70-mjb1   719.5
2.5.70-mm3    829.0
2.5.70-mm4    737.9

Randy Hron
http://home.earthlink.net/~rwhron/kernel/bigbox.html