1 files changed, 446 insertions, 98 deletions

diff --git a/kernel/time/tick-broadcast.c b/kernel/time/tick-broadcast.c
index 48b2761b566..64c5990fd50 100644
--- a/kernel/time/tick-broadcast.c
+++ b/kernel/time/tick-broadcast.c
@@ -18,7 +18,8 @@
 #include <linux/percpu.h>
 #include <linux/profile.h>
 #include <linux/sched.h>
-#include <linux/tick.h>
+#include <linux/smp.h>
+#include <linux/module.h>
 
 #include "tick-internal.h"
 
@@ -28,9 +29,9 @@
  */
 
 static struct tick_device tick_broadcast_device;
-/* FIXME: Use cpumask_var_t. */
-static DECLARE_BITMAP(tick_broadcast_mask, NR_CPUS);
-static DECLARE_BITMAP(tmpmask, NR_CPUS);
+static cpumask_var_t tick_broadcast_mask;
+static cpumask_var_t tick_broadcast_on;
+static cpumask_var_t tmpmask;
 static DEFINE_RAW_SPINLOCK(tick_broadcast_lock);
 static int tick_broadcast_force;
 
@@ -50,7 +51,7 @@ struct tick_device *tick_get_broadcast_device(void)
 
 struct cpumask *tick_get_broadcast_mask(void)
 {
-	return to_cpumask(tick_broadcast_mask);
+	return tick_broadcast_mask;
 }
 
 /*
@@ -65,18 +66,50 @@ static void tick_broadcast_start_periodic(struct clock_event_device *bc)
 /*
  * Check, if the device can be utilized as broadcast device:
  */
-int tick_check_broadcast_device(struct clock_event_device *dev)
+static bool tick_check_broadcast_device(struct clock_event_device *curdev,
+					struct clock_event_device *newdev)
 {
-	if ((tick_broadcast_device.evtdev &&
-	     tick_broadcast_device.evtdev->rating >= dev->rating) ||
-	     (dev->features & CLOCK_EVT_FEAT_C3STOP))
-		return 0;
+	if ((newdev->features & CLOCK_EVT_FEAT_DUMMY) ||
+	    (newdev->features & CLOCK_EVT_FEAT_PERCPU) ||
+	    (newdev->features & CLOCK_EVT_FEAT_C3STOP))
+		return false;
+
+	if (tick_broadcast_device.mode == TICKDEV_MODE_ONESHOT &&
+	    !(newdev->features & CLOCK_EVT_FEAT_ONESHOT))
+		return false;
+
+	return !curdev || newdev->rating > curdev->rating;
+}
+
+/*
+ * Conditionally install/replace broadcast device
+ */
+void tick_install_broadcast_device(struct clock_event_device *dev)
+{
+	struct clock_event_device *cur = tick_broadcast_device.evtdev;
+
+	if (!tick_check_broadcast_device(cur, dev))
+		return;
 
-	clockevents_exchange_device(NULL, dev);
+	if (!try_module_get(dev->owner))
+		return;
+
+	clockevents_exchange_device(cur, dev);
+	if (cur)
+		cur->event_handler = clockevents_handle_noop;
 	tick_broadcast_device.evtdev = dev;
-	if (!cpumask_empty(tick_get_broadcast_mask()))
+	if (!cpumask_empty(tick_broadcast_mask))
 		tick_broadcast_start_periodic(dev);
-	return 1;
+	/*
+	 * Inform all cpus about this. We might be in a situation
+	 * where we did not switch to oneshot mode because the per cpu
+	 * devices are affected by CLOCK_EVT_FEAT_C3STOP and the lack
+	 * of a oneshot capable broadcast device. Without that
+	 * notification the systems stays stuck in periodic mode
+	 * forever.
+	 */
+	if (dev->features & CLOCK_EVT_FEAT_ONESHOT)
+		tick_clock_notify();
 }
 
 /*
@@ -87,14 +120,44 @@ int tick_is_broadcast_device(struct clock_event_device *dev)
 	return (dev && tick_broadcast_device.evtdev == dev);
 }
 
+int tick_broadcast_update_freq(struct clock_event_device *dev, u32 freq)
+{
+	int ret = -ENODEV;
+
+	if (tick_is_broadcast_device(dev)) {
+		raw_spin_lock(&tick_broadcast_lock);
+		ret = __clockevents_update_freq(dev, freq);
+		raw_spin_unlock(&tick_broadcast_lock);
+	}
+	return ret;
+}
+
+
+static void err_broadcast(const struct cpumask *mask)
+{
+	pr_crit_once("Failed to broadcast timer tick. Some CPUs may be unresponsive.\n");
+}
+
+static void tick_device_setup_broadcast_func(struct clock_event_device *dev)
+{
+	if (!dev->broadcast)
+		dev->broadcast = tick_broadcast;
+	if (!dev->broadcast) {
+		pr_warn_once("%s depends on broadcast, but no broadcast function available\n",
+			     dev->name);
+		dev->broadcast = err_broadcast;
+	}
+}
+
 /*
  * Check, if the device is disfunctional and a place holder, which
  * needs to be handled by the broadcast device.
  */
 int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu)
 {
+	struct clock_event_device *bc = tick_broadcast_device.evtdev;
 	unsigned long flags;
-	int ret = 0;
+	int ret;
 
 	raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
 
@@ -106,26 +169,87 @@ int tick_device_uses_broadcast(struct clock_event_device *dev, int cpu)
 	 */
 	if (!tick_device_is_functional(dev)) {
 		dev->event_handler = tick_handle_periodic;
-		cpumask_set_cpu(cpu, tick_get_broadcast_mask());
-		tick_broadcast_start_periodic(tick_broadcast_device.evtdev);
+		tick_device_setup_broadcast_func(dev);
+		cpumask_set_cpu(cpu, tick_broadcast_mask);
+		if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC)
+			tick_broadcast_start_periodic(bc);
+		else
+			tick_broadcast_setup_oneshot(bc);
 		ret = 1;
 	} else {
 		/*
-		 * When the new device is not affected by the stop
-		 * feature and the cpu is marked in the broadcast mask
-		 * then clear the broadcast bit.
+		 * Clear the broadcast bit for this cpu if the
+		 * device is not power state affected.
 		 */
-		if (!(dev->features & CLOCK_EVT_FEAT_C3STOP)) {
-			int cpu = smp_processor_id();
+		if (!(dev->features & CLOCK_EVT_FEAT_C3STOP))
+			cpumask_clear_cpu(cpu, tick_broadcast_mask);
+		else
+			tick_device_setup_broadcast_func(dev);
 
-			cpumask_clear_cpu(cpu, tick_get_broadcast_mask());
+		/*
+		 * Clear the broadcast bit if the CPU is not in
+		 * periodic broadcast on state.
+		 */
+		if (!cpumask_test_cpu(cpu, tick_broadcast_on))
+			cpumask_clear_cpu(cpu, tick_broadcast_mask);
+
+		switch (tick_broadcast_device.mode) {
+		case TICKDEV_MODE_ONESHOT:
+			/*
+			 * If the system is in oneshot mode we can
+			 * unconditionally clear the oneshot mask bit,
+			 * because the CPU is running and therefore
+			 * not in an idle state which causes the power
+			 * state affected device to stop. Let the
+			 * caller initialize the device.
+			 */
 			tick_broadcast_clear_oneshot(cpu);
+			ret = 0;
+			break;
+
+		case TICKDEV_MODE_PERIODIC:
+			/*
+			 * If the system is in periodic mode, check
+			 * whether the broadcast device can be
+			 * switched off now.
+			 */
+			if (cpumask_empty(tick_broadcast_mask) && bc)
+				clockevents_shutdown(bc);
+			/*
+			 * If we kept the cpu in the broadcast mask,
+			 * tell the caller to leave the per cpu device
+			 * in shutdown state. The periodic interrupt
+			 * is delivered by the broadcast device.
+			 */
+			ret = cpumask_test_cpu(cpu, tick_broadcast_mask);
+			break;
+		default:
+			/* Nothing to do */
+			ret = 0;
+			break;
 		}
 	}
 	raw_spin_unlock_irqrestore(&tick_broadcast_lock, flags);
 	return ret;
 }
 
+#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST
+int tick_receive_broadcast(void)
+{
+	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);
+	struct clock_event_device *evt = td->evtdev;
+
+	if (!evt)
+		return -ENODEV;
+
+	if (!evt->event_handler)
+		return -EINVAL;
+
+	evt->event_handler(evt);
+	return 0;
+}
+#endif
+
 /*
  * Broadcast the event to the cpus, which are set in the mask (mangled).
  */
@@ -161,13 +285,8 @@ static void tick_do_broadcast(struct cpumask *mask)
  */
 static void tick_do_periodic_broadcast(void)
 {
-	raw_spin_lock(&tick_broadcast_lock);
-
-	cpumask_and(to_cpumask(tmpmask),
-		    cpu_online_mask, tick_get_broadcast_mask());
-	tick_do_broadcast(to_cpumask(tmpmask));
-
-	raw_spin_unlock(&tick_broadcast_lock);
+	cpumask_and(tmpmask, cpu_online_mask, tick_broadcast_mask);
+	tick_do_broadcast(tmpmask);
 }
 
 /*
@@ -177,13 +296,15 @@ static void tick_handle_periodic_broadcast(struct clock_event_device *dev)
 {
 	ktime_t next;
 
+	raw_spin_lock(&tick_broadcast_lock);
+
 	tick_do_periodic_broadcast();
 
 	/*
 	 * The device is in periodic mode. No reprogramming necessary:
 	 */
 	if (dev->mode == CLOCK_EVT_MODE_PERIODIC)
-		return;
+		goto unlock;
 
 	/*
 	 * Setup the next period for devices, which do not have
@@ -195,10 +316,12 @@ static void tick_handle_periodic_broadcast(struct clock_event_device *dev)
 	for (next = dev->next_event; ;) {
 		next = ktime_add(next, tick_period);
 
-		if (!clockevents_program_event(dev, next, ktime_get()))
-			return;
+		if (!clockevents_program_event(dev, next, false))
+			goto unlock;
 		tick_do_periodic_broadcast();
 	}
+unlock:
+	raw_spin_unlock(&tick_broadcast_lock);
 }
 
 /*
@@ -228,13 +351,13 @@ static void tick_do_broadcast_on_off(unsigned long *reason)
 	if (!tick_device_is_functional(dev))
 		goto out;
 
-	bc_stopped = cpumask_empty(tick_get_broadcast_mask());
+	bc_stopped = cpumask_empty(tick_broadcast_mask);
 
 	switch (*reason) {
 	case CLOCK_EVT_NOTIFY_BROADCAST_ON:
 	case CLOCK_EVT_NOTIFY_BROADCAST_FORCE:
-		if (!cpumask_test_cpu(cpu, tick_get_broadcast_mask())) {
-			cpumask_set_cpu(cpu, tick_get_broadcast_mask());
+		cpumask_set_cpu(cpu, tick_broadcast_on);
+		if (!cpumask_test_and_set_cpu(cpu, tick_broadcast_mask)) {
 			if (tick_broadcast_device.mode ==
 			    TICKDEV_MODE_PERIODIC)
 				clockevents_shutdown(dev);
@@ -243,9 +366,12 @@ static void tick_do_broadcast_on_off(unsigned long *reason)
 			tick_broadcast_force = 1;
 		break;
 	case CLOCK_EVT_NOTIFY_BROADCAST_OFF:
-		if (!tick_broadcast_force &&
-		    cpumask_test_cpu(cpu, tick_get_broadcast_mask())) {
-			cpumask_clear_cpu(cpu, tick_get_broadcast_mask());
+		if (tick_broadcast_force)
+			break;
+		cpumask_clear_cpu(cpu, tick_broadcast_on);
+		if (!tick_device_is_functional(dev))
+			break;
+		if (cpumask_test_and_clear_cpu(cpu, tick_broadcast_mask)) {
 			if (tick_broadcast_device.mode ==
 			    TICKDEV_MODE_PERIODIC)
 				tick_setup_periodic(dev, 0);
@@ -253,7 +379,7 @@ static void tick_do_broadcast_on_off(unsigned long *reason)
 		break;
 	}
 
-	if (cpumask_empty(tick_get_broadcast_mask())) {
+	if (cpumask_empty(tick_broadcast_mask)) {
 		if (!bc_stopped)
 			clockevents_shutdown(bc);
 	} else if (bc_stopped) {
@@ -302,10 +428,11 @@ void tick_shutdown_broadcast(unsigned int *cpup)
 	raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
 
 	bc = tick_broadcast_device.evtdev;
-	cpumask_clear_cpu(cpu, tick_get_broadcast_mask());
+	cpumask_clear_cpu(cpu, tick_broadcast_mask);
+	cpumask_clear_cpu(cpu, tick_broadcast_on);
 
 	if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) {
-		if (bc && cpumask_empty(tick_get_broadcast_mask()))
+		if (bc && cpumask_empty(tick_broadcast_mask))
 			clockevents_shutdown(bc);
 	}
 
@@ -341,13 +468,14 @@ int tick_resume_broadcast(void)
 
 		switch (tick_broadcast_device.mode) {
 		case TICKDEV_MODE_PERIODIC:
-			if (!cpumask_empty(tick_get_broadcast_mask()))
+			if (!cpumask_empty(tick_broadcast_mask))
 				tick_broadcast_start_periodic(bc);
 			broadcast = cpumask_test_cpu(smp_processor_id(),
-						     tick_get_broadcast_mask());
+						     tick_broadcast_mask);
 			break;
 		case TICKDEV_MODE_ONESHOT:
-			broadcast = tick_resume_broadcast_oneshot(bc);
+			if (!cpumask_empty(tick_broadcast_mask))
+				broadcast = tick_resume_broadcast_oneshot(bc);
 			break;
 		}
 	}
@@ -359,22 +487,58 @@ int tick_resume_broadcast(void)
 
 #ifdef CONFIG_TICK_ONESHOT
 
-/* FIXME: use cpumask_var_t. */
-static DECLARE_BITMAP(tick_broadcast_oneshot_mask, NR_CPUS);
+static cpumask_var_t tick_broadcast_oneshot_mask;
+static cpumask_var_t tick_broadcast_pending_mask;
+static cpumask_var_t tick_broadcast_force_mask;
 
 /*
  * Exposed for debugging: see timer_list.c
  */
 struct cpumask *tick_get_broadcast_oneshot_mask(void)
 {
-	return to_cpumask(tick_broadcast_oneshot_mask);
+	return tick_broadcast_oneshot_mask;
 }
 
-static int tick_broadcast_set_event(ktime_t expires, int force)
+/*
+ * Called before going idle with interrupts disabled. Checks whether a
+ * broadcast event from the other core is about to happen. We detected
+ * that in tick_broadcast_oneshot_control(). The callsite can use this
+ * to avoid a deep idle transition as we are about to get the
+ * broadcast IPI right away.
+ */
+int tick_check_broadcast_expired(void)
 {
-	struct clock_event_device *bc = tick_broadcast_device.evtdev;
+	return cpumask_test_cpu(smp_processor_id(), tick_broadcast_force_mask);
+}
+
+/*
+ * Set broadcast interrupt affinity
+ */
+static void tick_broadcast_set_affinity(struct clock_event_device *bc,
+					const struct cpumask *cpumask)
+{
+	if (!(bc->features & CLOCK_EVT_FEAT_DYNIRQ))
+		return;
+
+	if (cpumask_equal(bc->cpumask, cpumask))
+		return;
+
+	bc->cpumask = cpumask;
+	irq_set_affinity(bc->irq, bc->cpumask);
+}
+
+static int tick_broadcast_set_event(struct clock_event_device *bc, int cpu,
+				    ktime_t expires, int force)
+{
+	int ret;
+
+	if (bc->mode != CLOCK_EVT_MODE_ONESHOT)
+		clockevents_set_mode(bc, CLOCK_EVT_MODE_ONESHOT);
 
-	return tick_dev_program_event(bc, expires, force);
+	ret = clockevents_program_event(bc, expires, force);
+	if (!ret)
+		tick_broadcast_set_affinity(bc, cpumask_of(cpu));
+	return ret;
 }
 
 int tick_resume_broadcast_oneshot(struct clock_event_device *bc)
@@ -387,12 +551,20 @@ int tick_resume_broadcast_oneshot(struct clock_event_device *bc)
  * Called from irq_enter() when idle was interrupted to reenable the
  * per cpu device.
  */
-void tick_check_oneshot_broadcast(int cpu)
+void tick_check_oneshot_broadcast_this_cpu(void)
 {
-	if (cpumask_test_cpu(cpu, to_cpumask(tick_broadcast_oneshot_mask))) {
-		struct tick_device *td = &per_cpu(tick_cpu_device, cpu);
+	if (cpumask_test_cpu(smp_processor_id(), tick_broadcast_oneshot_mask)) {
+		struct tick_device *td = &__get_cpu_var(tick_cpu_device);
 
-		clockevents_set_mode(td->evtdev, CLOCK_EVT_MODE_ONESHOT);
+		/*
+		 * We might be in the middle of switching over from
+		 * periodic to oneshot. If the CPU has not yet
+		 * switched over, leave the device alone.
+		 */
+		if (td->mode == TICKDEV_MODE_ONESHOT) {
+			clockevents_set_mode(td->evtdev,
+					     CLOCK_EVT_MODE_ONESHOT);
+		}
 	}
 }
 
@@ -403,27 +575,52 @@ static void tick_handle_oneshot_broadcast(struct clock_event_device *dev)
 {
 	struct tick_device *td;
 	ktime_t now, next_event;
-	int cpu;
+	int cpu, next_cpu = 0;
 
 	raw_spin_lock(&tick_broadcast_lock);
 again:
 	dev->next_event.tv64 = KTIME_MAX;
 	next_event.tv64 = KTIME_MAX;
-	cpumask_clear(to_cpumask(tmpmask));
+	cpumask_clear(tmpmask);
 	now = ktime_get();
 	/* Find all expired events */
-	for_each_cpu(cpu, tick_get_broadcast_oneshot_mask()) {
+	for_each_cpu(cpu, tick_broadcast_oneshot_mask) {
 		td = &per_cpu(tick_cpu_device, cpu);
-		if (td->evtdev->next_event.tv64 <= now.tv64)
-			cpumask_set_cpu(cpu, to_cpumask(tmpmask));
-		else if (td->evtdev->next_event.tv64 < next_event.tv64)
+		if (td->evtdev->next_event.tv64 <= now.tv64) {
+			cpumask_set_cpu(cpu, tmpmask);
+			/*
+			 * Mark the remote cpu in the pending mask, so
+			 * it can avoid reprogramming the cpu local
+			 * timer in tick_broadcast_oneshot_control().
+			 */
+			cpumask_set_cpu(cpu, tick_broadcast_pending_mask);
+		} else if (td->evtdev->next_event.tv64 < next_event.tv64) {
 			next_event.tv64 = td->evtdev->next_event.tv64;
+			next_cpu = cpu;
+		}
 	}
 
 	/*
+	 * Remove the current cpu from the pending mask. The event is
+	 * delivered immediately in tick_do_broadcast() !
+	 */
+	cpumask_clear_cpu(smp_processor_id(), tick_broadcast_pending_mask);
+
+	/* Take care of enforced broadcast requests */
+	cpumask_or(tmpmask, tmpmask, tick_broadcast_force_mask);
+	cpumask_clear(tick_broadcast_force_mask);
+
+	/*
+	 * Sanity check. Catch the case where we try to broadcast to
+	 * offline cpus.
+	 */
+	if (WARN_ON_ONCE(!cpumask_subset(tmpmask, cpu_online_mask)))
+		cpumask_and(tmpmask, tmpmask, cpu_online_mask);
+
+	/*
 	 * Wakeup the cpus which have an expired event.
 	 */
-	tick_do_broadcast(to_cpumask(tmpmask));
+	tick_do_broadcast(tmpmask);
 
 	/*
 	 * Two reasons for reprogram:
@@ -440,59 +637,176 @@ again:
 		 * Rearm the broadcast device. If event expired,
 		 * repeat the above
 		 */
-		if (tick_broadcast_set_event(next_event, 0))
+		if (tick_broadcast_set_event(dev, next_cpu, next_event, 0))
 			goto again;
 	}
 	raw_spin_unlock(&tick_broadcast_lock);
 }
 
+static int broadcast_needs_cpu(struct clock_event_device *bc, int cpu)
+{
+	if (!(bc->features & CLOCK_EVT_FEAT_HRTIMER))
+		return 0;
+	if (bc->next_event.tv64 == KTIME_MAX)
+		return 0;
+	return bc->bound_on == cpu ? -EBUSY : 0;
+}
+
+static void broadcast_shutdown_local(struct clock_event_device *bc,
+				     struct clock_event_device *dev)
+{
+	/*
+	 * For hrtimer based broadcasting we cannot shutdown the cpu
+	 * local device if our own event is the first one to expire or
+	 * if we own the broadcast timer.
+	 */
+	if (bc->features & CLOCK_EVT_FEAT_HRTIMER) {
+		if (broadcast_needs_cpu(bc, smp_processor_id()))
+			return;
+		if (dev->next_event.tv64 < bc->next_event.tv64)
+			return;
+	}
+	clockevents_set_mode(dev, CLOCK_EVT_MODE_SHUTDOWN);
+}
+
+static void broadcast_move_bc(int deadcpu)
+{
+	struct clock_event_device *bc = tick_broadcast_device.evtdev;
+
+	if (!bc || !broadcast_needs_cpu(bc, deadcpu))
+		return;
+	/* This moves the broadcast assignment to this cpu */
+	clockevents_program_event(bc, bc->next_event, 1);
+}
+
 /*
  * Powerstate information: The system enters/leaves a state, where
  * affected devices might stop
+ * Returns 0 on success, -EBUSY if the cpu is used to broadcast wakeups.
  */
-void tick_broadcast_oneshot_control(unsigned long reason)
+int tick_broadcast_oneshot_control(unsigned long reason)
 {
 	struct clock_event_device *bc, *dev;
 	struct tick_device *td;
 	unsigned long flags;
-	int cpu;
-
-	raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
+	ktime_t now;
+	int cpu, ret = 0;
 
 	/*
 	 * Periodic mode does not care about the enter/exit of power
 	 * states
 	 */
 	if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC)
-		goto out;
+		return 0;
 
-	bc = tick_broadcast_device.evtdev;
+	/*
+	 * We are called with preemtion disabled from the depth of the
+	 * idle code, so we can't be moved away.
+	 */
 	cpu = smp_processor_id();
 	td = &per_cpu(tick_cpu_device, cpu);
 	dev = td->evtdev;
 
 	if (!(dev->features & CLOCK_EVT_FEAT_C3STOP))
-		goto out;
+		return 0;
 
+	bc = tick_broadcast_device.evtdev;
+
+	raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
 	if (reason == CLOCK_EVT_NOTIFY_BROADCAST_ENTER) {
-		if (!cpumask_test_cpu(cpu, tick_get_broadcast_oneshot_mask())) {
-			cpumask_set_cpu(cpu, tick_get_broadcast_oneshot_mask());
-			clockevents_set_mode(dev, CLOCK_EVT_MODE_SHUTDOWN);
-			if (dev->next_event.tv64 < bc->next_event.tv64)
-				tick_broadcast_set_event(dev->next_event, 1);
+		if (!cpumask_test_and_set_cpu(cpu, tick_broadcast_oneshot_mask)) {
+			WARN_ON_ONCE(cpumask_test_cpu(cpu, tick_broadcast_pending_mask));
+			broadcast_shutdown_local(bc, dev);
+			/*
+			 * We only reprogram the broadcast timer if we
+			 * did not mark ourself in the force mask and
+			 * if the cpu local event is earlier than the
+			 * broadcast event. If the current CPU is in
+			 * the force mask, then we are going to be
+			 * woken by the IPI right away.
+			 */
+			if (!cpumask_test_cpu(cpu, tick_broadcast_force_mask) &&
+			    dev->next_event.tv64 < bc->next_event.tv64)
+				tick_broadcast_set_event(bc, cpu, dev->next_event, 1);
 		}
+		/*
+		 * If the current CPU owns the hrtimer broadcast
+		 * mechanism, it cannot go deep idle and we remove the
+		 * CPU from the broadcast mask. We don't have to go
+		 * through the EXIT path as the local timer is not
+		 * shutdown.
+		 */
+		ret = broadcast_needs_cpu(bc, cpu);
+		if (ret)
+			cpumask_clear_cpu(cpu, tick_broadcast_oneshot_mask);
 	} else {
-		if (cpumask_test_cpu(cpu, tick_get_broadcast_oneshot_mask())) {
-			cpumask_clear_cpu(cpu,
-					  tick_get_broadcast_oneshot_mask());
+		if (cpumask_test_and_clear_cpu(cpu, tick_broadcast_oneshot_mask)) {
 			clockevents_set_mode(dev, CLOCK_EVT_MODE_ONESHOT);
-			if (dev->next_event.tv64 != KTIME_MAX)
-				tick_program_event(dev->next_event, 1);
+			/*
+			 * The cpu which was handling the broadcast
+			 * timer marked this cpu in the broadcast
+			 * pending mask and fired the broadcast
+			 * IPI. So we are going to handle the expired
+			 * event anyway via the broadcast IPI
+			 * handler. No need to reprogram the timer
+			 * with an already expired event.
+			 */
+			if (cpumask_test_and_clear_cpu(cpu,
+				       tick_broadcast_pending_mask))
+				goto out;
+
+			/*
+			 * Bail out if there is no next event.
+			 */
+			if (dev->next_event.tv64 == KTIME_MAX)
+				goto out;
+			/*
+			 * If the pending bit is not set, then we are
+			 * either the CPU handling the broadcast
+			 * interrupt or we got woken by something else.
+			 *
+			 * We are not longer in the broadcast mask, so
+			 * if the cpu local expiry time is already
+			 * reached, we would reprogram the cpu local
+			 * timer with an already expired event.
+			 *
+			 * This can lead to a ping-pong when we return
+			 * to idle and therefor rearm the broadcast
+			 * timer before the cpu local timer was able
+			 * to fire. This happens because the forced
+			 * reprogramming makes sure that the event
+			 * will happen in the future and depending on
+			 * the min_delta setting this might be far
+			 * enough out that the ping-pong starts.
+			 *
+			 * If the cpu local next_event has expired
+			 * then we know that the broadcast timer
+			 * next_event has expired as well and
+			 * broadcast is about to be handled. So we
+			 * avoid reprogramming and enforce that the
+			 * broadcast handler, which did not run yet,
+			 * will invoke the cpu local handler.
+			 *
+			 * We cannot call the handler directly from
+			 * here, because we might be in a NOHZ phase
+			 * and we did not go through the irq_enter()
+			 * nohz fixups.
+			 */
+			now = ktime_get();
+			if (dev->next_event.tv64 <= now.tv64) {
+				cpumask_set_cpu(cpu, tick_broadcast_force_mask);
+				goto out;
+			}
+			/*
+			 * We got woken by something else. Reprogram
+			 * the cpu local timer device.
+			 */
+			tick_program_event(dev->next_event, 1);
 		}
 	}
-
 out:
 	raw_spin_unlock_irqrestore(&tick_broadcast_lock, flags);
+	return ret;
 }
 
 /*
@@ -502,7 +816,8 @@ out:
  */
 static void tick_broadcast_clear_oneshot(int cpu)
 {
-	cpumask_clear_cpu(cpu, tick_get_broadcast_oneshot_mask());
+	cpumask_clear_cpu(cpu, tick_broadcast_oneshot_mask);
+	cpumask_clear_cpu(cpu, tick_broadcast_pending_mask);
 }
 
 static void tick_broadcast_init_next_event(struct cpumask *mask,
@@ -523,16 +838,13 @@ static void tick_broadcast_init_next_event(struct cpumask *mask,
  */
 void tick_broadcast_setup_oneshot(struct clock_event_device *bc)
 {
+	int cpu = smp_processor_id();
+
 	/* Set it up only once ! */
 	if (bc->event_handler != tick_handle_oneshot_broadcast) {
 		int was_periodic = bc->mode == CLOCK_EVT_MODE_PERIODIC;
-		int cpu = smp_processor_id();
 
 		bc->event_handler = tick_handle_oneshot_broadcast;
-		clockevents_set_mode(bc, CLOCK_EVT_MODE_ONESHOT);
-
-		/* Take the do_timer update */
-		tick_do_timer_cpu = cpu;
 
 		/*
 		 * We must be careful here. There might be other CPUs
@@ -540,18 +852,27 @@ void tick_broadcast_setup_oneshot(struct clock_event_device *bc)
 		 * oneshot_mask bits for those and program the
 		 * broadcast device to fire.
 		 */
-		cpumask_copy(to_cpumask(tmpmask), tick_get_broadcast_mask());
-		cpumask_clear_cpu(cpu, to_cpumask(tmpmask));
-		cpumask_or(tick_get_broadcast_oneshot_mask(),
-			   tick_get_broadcast_oneshot_mask(),
-			   to_cpumask(tmpmask));
-
-		if (was_periodic && !cpumask_empty(to_cpumask(tmpmask))) {
-			tick_broadcast_init_next_event(to_cpumask(tmpmask),
+		cpumask_copy(tmpmask, tick_broadcast_mask);
+		cpumask_clear_cpu(cpu, tmpmask);
+		cpumask_or(tick_broadcast_oneshot_mask,
+			   tick_broadcast_oneshot_mask, tmpmask);
+
+		if (was_periodic && !cpumask_empty(tmpmask)) {
+			clockevents_set_mode(bc, CLOCK_EVT_MODE_ONESHOT);
+			tick_broadcast_init_next_event(tmpmask,
 						       tick_next_period);
-			tick_broadcast_set_event(tick_next_period, 1);
+			tick_broadcast_set_event(bc, cpu, tick_next_period, 1);
 		} else
 			bc->next_event.tv64 = KTIME_MAX;
+	} else {
+		/*
+		 * The first cpu which switches to oneshot mode sets
+		 * the bit for all other cpus which are in the general
+		 * (periodic) broadcast mask. So the bit is set and
+		 * would prevent the first broadcast enter after this
+		 * to program the bc device.
+		 */
+		tick_broadcast_clear_oneshot(cpu);
 	}
 }
 
@@ -569,6 +890,7 @@ void tick_broadcast_switch_to_oneshot(void)
 	bc = tick_broadcast_device.evtdev;
 	if (bc)
 		tick_broadcast_setup_oneshot(bc);
+
 	raw_spin_unlock_irqrestore(&tick_broadcast_lock, flags);
 }
 
@@ -584,10 +906,14 @@ void tick_shutdown_broadcast_oneshot(unsigned int *cpup)
 	raw_spin_lock_irqsave(&tick_broadcast_lock, flags);
 
 	/*
-	 * Clear the broadcast mask flag for the dead cpu, but do not
-	 * stop the broadcast device!
+	 * Clear the broadcast masks for the dead cpu, but do not stop
+	 * the broadcast device!
 	 */
-	cpumask_clear_cpu(cpu, tick_get_broadcast_oneshot_mask());
+	cpumask_clear_cpu(cpu, tick_broadcast_oneshot_mask);
+	cpumask_clear_cpu(cpu, tick_broadcast_pending_mask);
+	cpumask_clear_cpu(cpu, tick_broadcast_force_mask);
+
+	broadcast_move_bc(cpu);
 
 	raw_spin_unlock_irqrestore(&tick_broadcast_lock, flags);
 }
@@ -600,4 +926,26 @@ int tick_broadcast_oneshot_active(void)
 	return tick_broadcast_device.mode == TICKDEV_MODE_ONESHOT;
 }
 
+/*
+ * Check whether the broadcast device supports oneshot.
+ */
+bool tick_broadcast_oneshot_available(void)
+{
+	struct clock_event_device *bc = tick_broadcast_device.evtdev;
+
+	return bc ? bc->features & CLOCK_EVT_FEAT_ONESHOT : false;
+}
+
 #endif
+
+void __init tick_broadcast_init(void)
+{
+	zalloc_cpumask_var(&tick_broadcast_mask, GFP_NOWAIT);
+	zalloc_cpumask_var(&tick_broadcast_on, GFP_NOWAIT);
+	zalloc_cpumask_var(&tmpmask, GFP_NOWAIT);
+#ifdef CONFIG_TICK_ONESHOT
+	zalloc_cpumask_var(&tick_broadcast_oneshot_mask, GFP_NOWAIT);
+	zalloc_cpumask_var(&tick_broadcast_pending_mask, GFP_NOWAIT);
+	zalloc_cpumask_var(&tick_broadcast_force_mask, GFP_NOWAIT);
+#endif
+}