I began seeing spurts of many scale_rt_power messages after recent upgrade to linux-3.0.80-0.5.1. Typical:
scale_rt_power: clock:24084fb46ae0 age:1ff374f92900, avg:0
This is from scheduler when doing load balancing, from scale_rt_power() in sched_fair.c. As avg is 0 and sched_avg_period() yields a constant, the problem is the difference between clock and age_stamp (both from cpu's runqueue) is too large. What does this actually mean? Running tickless.