* Steven Rostedt | 2013-03-19 12:50:34 [-0400]:
>@@ -1054,19 +1032,39 @@ int __hrtimer_start_range_ns(struct hrtimer *timer,
>ktime_t tim,
> *
> * XXX send_remote_softirq() ?
> */
>- if (leftmost && new_base->cpu_base == &__get_cpu_var(hrtimer_bases)) {
>- ret = hrtimer_enqueue_reprogram(timer, new_base, wakeup);
>- if (ret) {
>+ if (leftmost && new_base->cpu_base == &__get_cpu_var(hrtimer_bases)
>+ && hrtimer_enqueue_reprogram(timer, new_base)) {
>+
>+ if (wakeup
>+#ifdef CONFIG_PREEMPT_RT_BASE
>+ /*
>+ * Move softirq based timers away from the rbtree in
>+ * case it expired already. Otherwise we would have a
>+ * stale base->first entry until the softirq runs.
>+ */
>+ && hrtimer_rt_defer(timer)
>+#endif
>+ ) {
> /*
>- * In case we failed to reprogram the timer (mostly
>- * because out current timer is already elapsed),
>- * remove it again and report a failure. This avoids
>- * stale base->first entries.
>+ * We need to drop cpu_base->lock to avoid a
>+ * lock ordering issue vs. rq->lock.
> */
>- debug_deactivate(timer);
>- __remove_hrtimer(timer, new_base,
>- timer->state & HRTIMER_STATE_CALLBACK,
>0);
>+ raw_spin_unlock(&new_base->cpu_base->lock);
>+ raise_softirq_irqoff(HRTIMER_SOFTIRQ);
>+ local_irq_restore(flags);
>+ return ret;
Hmmm. Your ret here might be 0 or 1. RT did return here always zero
unless it returned -ETIME. This is what I did for v3.8 but I don't claim
to be correct here. Need to re-check this later…
Sebastian
--
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/
