Messages in this thread | | | Date | Wed, 15 Sep 2004 12:21:17 +0200 | From | Ingo Molnar <> | Subject | Re: [patch] sched: fix scheduling latencies for !PREEMPT kernels |
| |
* William Lee Irwin III <wli@holomorphy.com> wrote:
> I don't see deterministic execution time of hardirqs/softirqs > happening on stock hardware and without serious driver work, and I > don't see much hard RT ever happening on SMP due to lock contention. > But maybe that just means it's difficult.
actually, check out what i did to SMP latencies via the preempt-smp.patch. This patch brings SMP worst-case latencies to UP levels. E.g. on a dual system the worst-case should be roughly twice the UP worst-case latency: if both CPUs are trying to execute the same critical section. But there is no nondeterministic spinning anymore. You obviously cannot get better than that (other than increasing parallelism and reducing the number of critical sections).
(not all cases are covered - if some code is not using spinlocks or rwlocks but some static flag and is emulating spinlocks then it could spin uncontrollably. Those places show up in the tracer quite easily - i had to fix only one so far.)
Ingo - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |