Hello its me again! yeah i am not sopping to ask question ;). I have read through a few papers regarding interrupt latency of the preempt-patch as this one http://lwn.net/images/conf/rtlws11/papers/proc/p19.pdf The implementation seems to suggest that interrupt latencies of hard irqs (IRQF_NODELAY) should be better for preempt-rt. On my machine however it is worse than for CONFIG_PREEMPT_NONE. The userspace latency measured with cyclictest is much better for CONFIG_PREEMPT_RT then for CONFIG_PREEMPT_NONE. The interrupt latency however (with IRQF_NODELAY) is worse than for CONFIG_PREEMPT_NONE and the worst case is about 700us. This makes no sense to me, because the max user space latency is about 650us. How can it be possible that the irq latency is worse then userspace latency? I am on the loss here, and would highly appreciate any ideas, hints or ANYTHING. Thank you JB -- 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
