The scheduler does not consider the power states CPUs while selecting an idle one. So it might happen, that it puts load on a cpu which is in a deeper power state than another idle one, which has higher latency and power costs than pushing the load to the core which is not in a deep power saving state.
This patch series implements infrastructure which allows to consider the power state in the selection process of an idle cpu. The decision function is default off and can be enabled via a sysctl. This is just the basic infrastructure. Further modifications are necessary to hand in proper state information from the various idle implementation. Its has been tested with a static testcase plugged into the cpuidle core. Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
