> I would expect the load is much higher... Do you have a proper > sched_clock() implementation or the default fallback-to-jiffies based > sched_clock()? Load statistics might be misleading... > See attached for a preliminary patch... (using it on a sam9261 core) No, I always was using vanilla (or sometimes tried with the atmel patches from linux4sam). With 2.6.31 the ftrace behaviour does not change (but then I've to add your serial fix) It seems with your patch at least the the timing informations in the trace look better e.g. (but maybe they're still wrong, don't know yet): --- >8 --- <idle>-0 0d.h.2 34147us+: irq_exit <-_text <idle>-0 0d...3 34160us+: __do_softirq <-irq_exit <idle>-0 0d...3 34173us+: trigger_softirqs <-__do_softirq <idle>-0 0d...1 34192us!: default_idle <-cpu_idle <idle>-0 0d...2 41277us+: _text <-__irq_svc <idle>-0 0d...2 41290us+: irq_enter <-_text <idle>-0 0d...2 41302us+: idle_cpu <-irq_enter <idle>-0 0d.h.2 41315us+: tick_check_idle <-irq_enter <idle>-0 0d.h.2 41326us+: irq_to_desc <-_text --- >8 --- Hmm my idea was to quickly try out different boards and choose the best one, but it seems that I have to go deeper. Thanks once more for you patch Azraiyl -- 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
