On Wed, 2010-05-19 at 14:55 +0200, Michal Simek wrote: > ==> /debugfs/tracing/latency_hist/preemptirqsoff/CPU0 <== > #Minimum latency: 12 microseconds > #Average latency: 1991 microseconds > #Maximum latency: 4294954 microseconds > #Total samples: 76773 > #There are 0 samples lower than 0 microseconds. > #There are 35 samples greater or equal than 10240 microseconds. > #usecs samples > 0 0 > 1 0 > 2 0 Note, do you disable the irqsoff tracer on idle? In x86 we have in process_*.c we have: stop_critical_timings(); pm_idle(); start_critical_timings(); This causes the irqsoff() tracer to not care about the time we are in idle. The interrupts are disabled when calling pm_idle() and enabled in assembly, so we miss it. But really, we don't care here because, if interrupts were disabled and not react, then the CPU would be dead ;-) This is just one theory, those numbers could be because of something else. But this is the most common case when the preemptoff and irqsoff tracers are ported to another arch. -- Steve -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
