---

yaml
---
r: 110103
b: refs/heads/master
c: d3a47e8
h: refs/heads/master
i:
  110101: 257ec36
  110099: 452e942
  110095: 45ea3ea
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: c459ce8b5a7d933a3bcf6915ab17ac1e036e2ac4
+refs/heads/master: d3a47e82b6bc3724dd60f3ee4e84fe4479104382

trunk/arch/powerpc/boot/dts/holly.dts

@@ -133,61 +133,61 @@
 			reg = <0x00007400 0x00000400>;
 			big-endian;
 		};
+	};
 
-		pci@1000 {
-			device_type = "pci";
-			compatible = "tsi109-pci", "tsi108-pci";
-			#interrupt-cells = <1>;
-			#size-cells = <2>;
-			#address-cells = <3>;
-			reg = <0x00001000 0x00001000>;
-			bus-range = <0x0 0x0>;
-			/*----------------------------------------------------+
-			| PCI memory range.
-			| 01 denotes I/O space
-			| 02 denotes 32-bit memory space
-			+----------------------------------------------------*/
-			ranges = <0x02000000 0x00000000 0x40000000 0x40000000 0x00000000 0x10000000
-				  0x01000000 0x00000000 0x00000000 0x7e000000 0x00000000 0x00010000>;
-			clock-frequency = <133333332>;
-			interrupt-parent = <&MPIC>;
+	pci@c0001000 {
+		device_type = "pci";
+		compatible = "tsi109-pci", "tsi108-pci";
+		#interrupt-cells = <1>;
+		#size-cells = <2>;
+		#address-cells = <3>;
+		reg = <0xc0001000 0x00001000>;
+		bus-range = <0x0 0x0>;
+		/*----------------------------------------------------+
+		| PCI memory range.
+		| 01 denotes I/O space
+		| 02 denotes 32-bit memory space
+		+----------------------------------------------------*/
+		ranges = <0x02000000 0x00000000 0x40000000 0x40000000 0x00000000 0x10000000
+			  0x01000000 0x00000000 0x00000000 0x7e000000 0x00000000 0x00010000>;
+		clock-frequency = <133333332>;
+		interrupt-parent = <&MPIC>;
+		interrupts = <0x17 0x2>;
+		interrupt-map-mask = <0xf800 0x0 0x0 0x7>;
+		/*----------------------------------------------------+
+		| The INTA, INTB, INTC, INTD are shared.
+		+----------------------------------------------------*/
+		interrupt-map = <
+			0x800 0x0 0x0 0x1 &RT0 0x24 0x0
+			0x800 0x0 0x0 0x2 &RT0 0x25 0x0
+			0x800 0x0 0x0 0x3 &RT0 0x26 0x0
+			0x800 0x0 0x0 0x4 &RT0 0x27 0x0
+
+			0x1000 0x0 0x0 0x1 &RT0 0x25 0x0
+			0x1000 0x0 0x0 0x2 &RT0 0x26 0x0
+			0x1000 0x0 0x0 0x3 &RT0 0x27 0x0
+			0x1000 0x0 0x0 0x4 &RT0 0x24 0x0
+
+			0x1800 0x0 0x0 0x1 &RT0 0x26 0x0
+			0x1800 0x0 0x0 0x2 &RT0 0x27 0x0
+			0x1800 0x0 0x0 0x3 &RT0 0x24 0x0
+			0x1800 0x0 0x0 0x4 &RT0 0x25 0x0
+
+			0x2000 0x0 0x0 0x1 &RT0 0x27 0x0
+			0x2000 0x0 0x0 0x2 &RT0 0x24 0x0
+			0x2000 0x0 0x0 0x3 &RT0 0x25 0x0
+			0x2000 0x0 0x0 0x4 &RT0 0x26 0x0
+			>;
+
+		RT0: router@1180 {
+			device_type = "pic-router";
+			interrupt-controller;
+			big-endian;
+			clock-frequency = <0>;
+			#address-cells = <0>;
+			#interrupt-cells = <2>;
 			interrupts = <0x17 0x2>;
-			interrupt-map-mask = <0xf800 0x0 0x0 0x7>;
-			/*----------------------------------------------------+
-			| The INTA, INTB, INTC, INTD are shared.
-			+----------------------------------------------------*/
-			interrupt-map = <
-				0x800 0x0 0x0 0x1 &RT0 0x24 0x0
-				0x800 0x0 0x0 0x2 &RT0 0x25 0x0
-				0x800 0x0 0x0 0x3 &RT0 0x26 0x0
-				0x800 0x0 0x0 0x4 &RT0 0x27 0x0
-
-				0x1000 0x0 0x0 0x1 &RT0 0x25 0x0
-				0x1000 0x0 0x0 0x2 &RT0 0x26 0x0
-				0x1000 0x0 0x0 0x3 &RT0 0x27 0x0
-				0x1000 0x0 0x0 0x4 &RT0 0x24 0x0
-
-				0x1800 0x0 0x0 0x1 &RT0 0x26 0x0
-				0x1800 0x0 0x0 0x2 &RT0 0x27 0x0
-				0x1800 0x0 0x0 0x3 &RT0 0x24 0x0
-				0x1800 0x0 0x0 0x4 &RT0 0x25 0x0
-
-				0x2000 0x0 0x0 0x1 &RT0 0x27 0x0
-				0x2000 0x0 0x0 0x2 &RT0 0x24 0x0
-				0x2000 0x0 0x0 0x3 &RT0 0x25 0x0
-				0x2000 0x0 0x0 0x4 &RT0 0x26 0x0
-				>;
-
-			RT0: router@1180 {
- 				device_type = "pic-router";
- 				interrupt-controller;
- 				big-endian;
- 				clock-frequency = <0>;
- 				#address-cells = <0>;
- 				#interrupt-cells = <2>;
- 				interrupts = <0x17 0x2>;
-				interrupt-parent = <&MPIC>;
-			};
+			interrupt-parent = <&MPIC>;
 		};
 	};
 

trunk/arch/powerpc/kernel/idle.c

@@ -34,11 +34,7 @@
 #include <asm/smp.h>
 
 #ifdef CONFIG_HOTPLUG_CPU
-/* this is used for software suspend, and that shuts down
- * CPUs even while the system is still booting... */
-#define cpu_should_die()	(cpu_is_offline(smp_processor_id()) && \
-				   (system_state == SYSTEM_RUNNING     \
-				 || system_state == SYSTEM_BOOTING))
+#define cpu_should_die()	cpu_is_offline(smp_processor_id())
 #else
 #define cpu_should_die()	0
 #endif

trunk/drivers/mfd/Kconfig

@@ -21,15 +21,15 @@ config MFD_SM501
 
 config MFD_SM501_GPIO
 	bool "Export GPIO via GPIO layer"
-	depends on MFD_SM501 && HAVE_GPIO_LIB
+	depends on MFD_SM501 && GPIOLIB
 	 ---help---
 	 This option uses the gpio library layer to export the 64 GPIO
 	 lines on the SM501. The platform data is used to supply the
 	 base number for the first GPIO line to register.
 
 config MFD_ASIC3
 	bool "Support for Compaq ASIC3"
-	depends on GENERIC_HARDIRQS && HAVE_GPIO_LIB && ARM
+	depends on GENERIC_HARDIRQS && GPIOLIB && ARM
 	 ---help---
 	  This driver supports the ASIC3 multifunction chip found on many
 	  PDAs (mainly iPAQ and HTC based ones)

trunk/drivers/mfd/asic3.c

@@ -312,7 +312,6 @@ static int __init asic3_irq_probe(struct platform_device *pdev)
 	struct asic3 *asic = platform_get_drvdata(pdev);
 	unsigned long clksel = 0;
 	unsigned int irq, irq_base;
-	int map_size;
 	int ret;
 
 	ret = platform_get_irq(pdev, 0);
@@ -534,6 +533,7 @@ static int __init asic3_probe(struct platform_device *pdev)
 	struct asic3 *asic;
 	struct resource *mem;
 	unsigned long clksel;
+	int map_size;
 	int ret = 0;
 
 	asic = kzalloc(sizeof(struct asic3), GFP_KERNEL);

trunk/include/linux/hrtimer.h

@@ -47,14 +47,22 @@ enum hrtimer_restart {
  *	HRTIMER_CB_IRQSAFE:		Callback may run in hardirq context
  *	HRTIMER_CB_IRQSAFE_NO_RESTART:	Callback may run in hardirq context and
  *					does not restart the timer
- *	HRTIMER_CB_IRQSAFE_NO_SOFTIRQ:	Callback must run in hardirq context
- *					Special mode for tick emultation
+ *	HRTIMER_CB_IRQSAFE_PERCPU:	Callback must run in hardirq context
+ *					Special mode for tick emulation and
+ *					scheduler timer. Such timers are per
+ *					cpu and not allowed to be migrated on
+ *					cpu unplug.
+ *	HRTIMER_CB_IRQSAFE_UNLOCKED:	Callback should run in hardirq context
+ *					with timer->base lock unlocked
+ *					used for timers which call wakeup to
+ *					avoid lock order problems with rq->lock
  */
 enum hrtimer_cb_mode {
 	HRTIMER_CB_SOFTIRQ,
 	HRTIMER_CB_IRQSAFE,
 	HRTIMER_CB_IRQSAFE_NO_RESTART,
-	HRTIMER_CB_IRQSAFE_NO_SOFTIRQ,
+	HRTIMER_CB_IRQSAFE_PERCPU,
+	HRTIMER_CB_IRQSAFE_UNLOCKED,
 };
 
 /*
@@ -67,9 +75,10 @@ enum hrtimer_cb_mode {
  * 0x02		callback function running
  * 0x04		callback pending (high resolution mode)
  *
- * Special case:
+ * Special cases:
  * 0x03		callback function running and enqueued
  *		(was requeued on another CPU)
+ * 0x09		timer was migrated on CPU hotunplug
  * The "callback function running and enqueued" status is only possible on
  * SMP. It happens for example when a posix timer expired and the callback
  * queued a signal. Between dropping the lock which protects the posix timer
@@ -87,6 +96,7 @@ enum hrtimer_cb_mode {
 #define HRTIMER_STATE_ENQUEUED	0x01
 #define HRTIMER_STATE_CALLBACK	0x02
 #define HRTIMER_STATE_PENDING	0x04
+#define HRTIMER_STATE_MIGRATE	0x08
 
 /**
  * struct hrtimer - the basic hrtimer structure

trunk/kernel/hrtimer.c

@@ -672,13 +672,14 @@ static inline int hrtimer_enqueue_reprogram(struct hrtimer *timer,
 			 */
 			BUG_ON(timer->function(timer) != HRTIMER_NORESTART);
 			return 1;
-		case HRTIMER_CB_IRQSAFE_NO_SOFTIRQ:
+		case HRTIMER_CB_IRQSAFE_PERCPU:
+		case HRTIMER_CB_IRQSAFE_UNLOCKED:
 			/*
 			 * This is solely for the sched tick emulation with
 			 * dynamic tick support to ensure that we do not
 			 * restart the tick right on the edge and end up with
 			 * the tick timer in the softirq ! The calling site
-			 * takes care of this.
+			 * takes care of this. Also used for hrtimer sleeper !
 			 */
 			debug_hrtimer_deactivate(timer);
 			return 1;
@@ -1245,7 +1246,8 @@ static void __run_hrtimer(struct hrtimer *timer)
 	timer_stats_account_hrtimer(timer);
 
 	fn = timer->function;
-	if (timer->cb_mode == HRTIMER_CB_IRQSAFE_NO_SOFTIRQ) {
+	if (timer->cb_mode == HRTIMER_CB_IRQSAFE_PERCPU ||
+	    timer->cb_mode == HRTIMER_CB_IRQSAFE_UNLOCKED) {
 		/*
 		 * Used for scheduler timers, avoid lock inversion with
 		 * rq->lock and tasklist_lock.
@@ -1452,7 +1454,7 @@ void hrtimer_init_sleeper(struct hrtimer_sleeper *sl, struct task_struct *task)
 	sl->timer.function = hrtimer_wakeup;
 	sl->task = task;
 #ifdef CONFIG_HIGH_RES_TIMERS
-	sl->timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ;
+	sl->timer.cb_mode = HRTIMER_CB_IRQSAFE_UNLOCKED;
 #endif
 }
 
@@ -1591,29 +1593,95 @@ static void __cpuinit init_hrtimers_cpu(int cpu)
 
 #ifdef CONFIG_HOTPLUG_CPU
 
-static void migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
-				struct hrtimer_clock_base *new_base)
+static int migrate_hrtimer_list(struct hrtimer_clock_base *old_base,
+				struct hrtimer_clock_base *new_base, int dcpu)
 {
 	struct hrtimer *timer;
 	struct rb_node *node;
+	int raise = 0;
 
 	while ((node = rb_first(&old_base->active))) {
 		timer = rb_entry(node, struct hrtimer, node);
 		BUG_ON(hrtimer_callback_running(timer));
 		debug_hrtimer_deactivate(timer);
-		__remove_hrtimer(timer, old_base, HRTIMER_STATE_INACTIVE, 0);
+
+		/*
+		 * Should not happen. Per CPU timers should be
+		 * canceled _before_ the migration code is called
+		 */
+		if (timer->cb_mode == HRTIMER_CB_IRQSAFE_PERCPU) {
+			__remove_hrtimer(timer, old_base,
+					 HRTIMER_STATE_INACTIVE, 0);
+			WARN(1, "hrtimer (%p %p)active but cpu %d dead\n",
+			     timer, timer->function, dcpu);
+			continue;
+		}
+
+		/*
+		 * Mark it as STATE_MIGRATE not INACTIVE otherwise the
+		 * timer could be seen as !active and just vanish away
+		 * under us on another CPU
+		 */
+		__remove_hrtimer(timer, old_base, HRTIMER_STATE_MIGRATE, 0);
 		timer->base = new_base;
 		/*
 		 * Enqueue the timer. Allow reprogramming of the event device
 		 */
 		enqueue_hrtimer(timer, new_base, 1);
+
+#ifdef CONFIG_HIGH_RES_TIMERS
+		/*
+		 * Happens with high res enabled when the timer was
+		 * already expired and the callback mode is
+		 * HRTIMER_CB_IRQSAFE_UNLOCKED (hrtimer_sleeper). The
+		 * enqueue code does not move them to the soft irq
+		 * pending list for performance/latency reasons, but
+		 * in the migration state, we need to do that
+		 * otherwise we end up with a stale timer.
+		 */
+		if (timer->state == HRTIMER_STATE_MIGRATE) {
+			timer->state = HRTIMER_STATE_PENDING;
+			list_add_tail(&timer->cb_entry,
+				      &new_base->cpu_base->cb_pending);
+			raise = 1;
+		}
+#endif
+		/* Clear the migration state bit */
+		timer->state &= ~HRTIMER_STATE_MIGRATE;
+	}
+	return raise;
+}
+
+#ifdef CONFIG_HIGH_RES_TIMERS
+static int migrate_hrtimer_pending(struct hrtimer_cpu_base *old_base,
+				   struct hrtimer_cpu_base *new_base)
+{
+	struct hrtimer *timer;
+	int raise = 0;
+
+	while (!list_empty(&old_base->cb_pending)) {
+		timer = list_entry(old_base->cb_pending.next,
+				   struct hrtimer, cb_entry);
+
+		__remove_hrtimer(timer, timer->base, HRTIMER_STATE_PENDING, 0);
+		timer->base = &new_base->clock_base[timer->base->index];
+		list_add_tail(&timer->cb_entry, &new_base->cb_pending);
+		raise = 1;
 	}
+	return raise;
+}
+#else
+static int migrate_hrtimer_pending(struct hrtimer_cpu_base *old_base,
+				   struct hrtimer_cpu_base *new_base)
+{
+	return 0;
 }
+#endif
 
 static void migrate_hrtimers(int cpu)
 {
 	struct hrtimer_cpu_base *old_base, *new_base;
-	int i;
+	int i, raise = 0;
 
 	BUG_ON(cpu_online(cpu));
 	old_base = &per_cpu(hrtimer_bases, cpu);
@@ -1626,14 +1694,21 @@ static void migrate_hrtimers(int cpu)
 	spin_lock_nested(&old_base->lock, SINGLE_DEPTH_NESTING);
 
 	for (i = 0; i < HRTIMER_MAX_CLOCK_BASES; i++) {
-		migrate_hrtimer_list(&old_base->clock_base[i],
-				     &new_base->clock_base[i]);
+		if (migrate_hrtimer_list(&old_base->clock_base[i],
+					 &new_base->clock_base[i], cpu))
+			raise = 1;
 	}
 
+	if (migrate_hrtimer_pending(old_base, new_base))
+		raise = 1;
+
 	spin_unlock(&old_base->lock);
 	spin_unlock(&new_base->lock);
 	local_irq_enable();
 	put_cpu_var(hrtimer_bases);
+
+	if (raise)
+		hrtimer_raise_softirq();
 }
 #endif /* CONFIG_HOTPLUG_CPU */
 

trunk/kernel/sched.c

@@ -201,7 +201,7 @@ void init_rt_bandwidth(struct rt_bandwidth *rt_b, u64 period, u64 runtime)
 	hrtimer_init(&rt_b->rt_period_timer,
 			CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	rt_b->rt_period_timer.function = sched_rt_period_timer;
-	rt_b->rt_period_timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ;
+	rt_b->rt_period_timer.cb_mode = HRTIMER_CB_IRQSAFE_UNLOCKED;
 }
 
 static void start_rt_bandwidth(struct rt_bandwidth *rt_b)
@@ -1119,7 +1119,7 @@ static void init_rq_hrtick(struct rq *rq)
 
 	hrtimer_init(&rq->hrtick_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	rq->hrtick_timer.function = hrtick;
-	rq->hrtick_timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ;
+	rq->hrtick_timer.cb_mode = HRTIMER_CB_IRQSAFE_PERCPU;
 }
 #else
 static inline void hrtick_clear(struct rq *rq)

trunk/kernel/time/tick-sched.c

@@ -625,7 +625,7 @@ void tick_setup_sched_timer(void)
 	 */
 	hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
 	ts->sched_timer.function = tick_sched_timer;
-	ts->sched_timer.cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ;
+	ts->sched_timer.cb_mode = HRTIMER_CB_IRQSAFE_PERCPU;
 
 	/* Get the next period (per cpu) */
 	ts->sched_timer.expires = tick_init_jiffy_update();