The idle_balance() code is treated special in the scheduler for no real apparent reason. There's an unlikely check for !rq->nr_running in the hot path of the scheduler, for just the idle load balancing.
As sched classes can have their own post schedule routine, and the idle task has its own scheduler class, we can move the idle_balance() into the idle task's post schedule routine, and out of the hot path of the scheduler (no sepecial check). An added benefit for doing this is that this enables us to enable interrupts in the idle balance code. Shrinking the time interrupts are disabled while doing load balancing. These patches have been thoroughly tested (on a 40 core box as well as a 4 core and UP config), and are prime for inclusion (for 3.9). But I'm posted them now as an RFC for comments. Thanks! -- Steve -- 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/
