[net-next-2.6.git] / kernel / watchdog.c

/*
 * Detect hard and soft lockups on a system
 *
 * started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
 *
 * this code detects hard lockups: incidents in where on a CPU
 * the kernel does not respond to anything except NMI.
 *
 * Note: Most of this code is borrowed heavily from softlockup.c,
 * so thanks to Ingo for the initial implementation.
 * Some chunks also taken from arch/x86/kernel/apic/nmi.c, thanks
 * to those contributors as well.
 */

#include <linux/mm.h>
#include <linux/cpu.h>
#include <linux/nmi.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/freezer.h>
#include <linux/kthread.h>
#include <linux/lockdep.h>
#include <linux/notifier.h>
#include <linux/module.h>
#include <linux/sysctl.h>

#include <asm/irq_regs.h>
#include <linux/perf_event.h>

int watchdog_enabled;
int __read_mostly softlockup_thresh = 60;

static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts);
static DEFINE_PER_CPU(struct task_struct *, softlockup_watchdog);
static DEFINE_PER_CPU(struct hrtimer, watchdog_hrtimer);
static DEFINE_PER_CPU(bool, softlockup_touch_sync);
static DEFINE_PER_CPU(bool, soft_watchdog_warn);
#ifdef CONFIG_HARDLOCKUP_DETECTOR
static DEFINE_PER_CPU(bool, hard_watchdog_warn);
static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
static DEFINE_PER_CPU(struct perf_event *, watchdog_ev);
#endif

static int __read_mostly did_panic;
static int __initdata no_watchdog;


/* boot commands */
/*
 * Should we panic when a soft-lockup or hard-lockup occurs:
 */
#ifdef CONFIG_HARDLOCKUP_DETECTOR
static int hardlockup_panic;

static int __init hardlockup_panic_setup(char *str)
{
	if (!strncmp(str, "panic", 5))
		hardlockup_panic = 1;
	return 1;
}
__setup("nmi_watchdog=", hardlockup_panic_setup);
#endif

unsigned int __read_mostly softlockup_panic =
			CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC_VALUE;

static int __init softlockup_panic_setup(char *str)
{
	softlockup_panic = simple_strtoul(str, NULL, 0);

	return 1;
}
__setup("softlockup_panic=", softlockup_panic_setup);

static int __init nowatchdog_setup(char *str)
{
	no_watchdog = 1;
	return 1;
}
__setup("nowatchdog", nowatchdog_setup);

/* deprecated */
static int __init nosoftlockup_setup(char *str)
{
	no_watchdog = 1;
	return 1;
}
__setup("nosoftlockup", nosoftlockup_setup);
/*  */


/*
 * Returns seconds, approximately.  We don't need nanosecond
 * resolution, and we don't need to waste time with a big divide when
 * 2^30ns == 1.074s.
 */
static unsigned long get_timestamp(int this_cpu)
{
	return cpu_clock(this_cpu) >> 30LL;  /* 2^30 ~= 10^9 */
}

static unsigned long get_sample_period(void)
{
	/*
	 * convert softlockup_thresh from seconds to ns
	 * the divide by 5 is to give hrtimer 5 chances to
	 * increment before the hardlockup detector generates
	 * a warning
	 */
	return softlockup_thresh / 5 * NSEC_PER_SEC;
}

/* Commands for resetting the watchdog */
static void __touch_watchdog(void)
{
	int this_cpu = smp_processor_id();

	__get_cpu_var(watchdog_touch_ts) = get_timestamp(this_cpu);
}

void touch_softlockup_watchdog(void)
{
	__raw_get_cpu_var(watchdog_touch_ts) = 0;
}
EXPORT_SYMBOL(touch_softlockup_watchdog);

void touch_all_softlockup_watchdogs(void)
{
	int cpu;

	/*
	 * this is done lockless
	 * do we care if a 0 races with a timestamp?
	 * all it means is the softlock check starts one cycle later
	 */
	for_each_online_cpu(cpu)
		per_cpu(watchdog_touch_ts, cpu) = 0;
}

#ifdef CONFIG_HARDLOCKUP_DETECTOR
void touch_nmi_watchdog(void)
{
	if (watchdog_enabled) {
		unsigned cpu;

		for_each_present_cpu(cpu) {
			if (per_cpu(watchdog_nmi_touch, cpu) != true)
				per_cpu(watchdog_nmi_touch, cpu) = true;
		}
	}
	touch_softlockup_watchdog();
}
EXPORT_SYMBOL(touch_nmi_watchdog);

#endif

void touch_softlockup_watchdog_sync(void)
{
	__raw_get_cpu_var(softlockup_touch_sync) = true;
	__raw_get_cpu_var(watchdog_touch_ts) = 0;
}

#ifdef CONFIG_HARDLOCKUP_DETECTOR
/* watchdog detector functions */
static int is_hardlockup(void)
{
	unsigned long hrint = __get_cpu_var(hrtimer_interrupts);

	if (__get_cpu_var(hrtimer_interrupts_saved) == hrint)
		return 1;

	__get_cpu_var(hrtimer_interrupts_saved) = hrint;
	return 0;
}
#endif

static int is_softlockup(unsigned long touch_ts)
{
	unsigned long now = get_timestamp(smp_processor_id());

	/* Warn about unreasonable delays: */
	if (time_after(now, touch_ts + softlockup_thresh))
		return now - touch_ts;

	return 0;
}

static int
watchdog_panic(struct notifier_block *this, unsigned long event, void *ptr)
{
	did_panic = 1;

	return NOTIFY_DONE;
}

static struct notifier_block panic_block = {
	.notifier_call = watchdog_panic,
};

#ifdef CONFIG_HARDLOCKUP_DETECTOR
static struct perf_event_attr wd_hw_attr = {
	.type		= PERF_TYPE_HARDWARE,
	.config		= PERF_COUNT_HW_CPU_CYCLES,
	.size		= sizeof(struct perf_event_attr),
	.pinned		= 1,
	.disabled	= 1,
};

/* Callback function for perf event subsystem */
void watchdog_overflow_callback(struct perf_event *event, int nmi,
		 struct perf_sample_data *data,
		 struct pt_regs *regs)
{
	/* Ensure the watchdog never gets throttled */
	event->hw.interrupts = 0;

	if (__get_cpu_var(watchdog_nmi_touch) == true) {
		__get_cpu_var(watchdog_nmi_touch) = false;
		return;
	}

	/* check for a hardlockup
	 * This is done by making sure our timer interrupt
	 * is incrementing.  The timer interrupt should have
	 * fired multiple times before we overflow'd.  If it hasn't
	 * then this is a good indication the cpu is stuck
	 */
	if (is_hardlockup()) {
		int this_cpu = smp_processor_id();

		/* only print hardlockups once */
		if (__get_cpu_var(hard_watchdog_warn) == true)
			return;

		if (hardlockup_panic)
			panic("Watchdog detected hard LOCKUP on cpu %d", this_cpu);
		else
			WARN(1, "Watchdog detected hard LOCKUP on cpu %d", this_cpu);

		__get_cpu_var(hard_watchdog_warn) = true;
		return;
	}

	__get_cpu_var(hard_watchdog_warn) = false;
	return;
}
static void watchdog_interrupt_count(void)
{
	__get_cpu_var(hrtimer_interrupts)++;
}
#else
static inline void watchdog_interrupt_count(void) { return; }
#endif /* CONFIG_HARDLOCKUP_DETECTOR */

/* watchdog kicker functions */
static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
{
	unsigned long touch_ts = __get_cpu_var(watchdog_touch_ts);
	struct pt_regs *regs = get_irq_regs();
	int duration;

	/* kick the hardlockup detector */
	watchdog_interrupt_count();

	/* kick the softlockup detector */
	wake_up_process(__get_cpu_var(softlockup_watchdog));

	/* .. and repeat */
	hrtimer_forward_now(hrtimer, ns_to_ktime(get_sample_period()));

	if (touch_ts == 0) {
		if (unlikely(__get_cpu_var(softlockup_touch_sync))) {
			/*
			 * If the time stamp was touched atomically
			 * make sure the scheduler tick is up to date.
			 */
			__get_cpu_var(softlockup_touch_sync) = false;
			sched_clock_tick();
		}
		__touch_watchdog();
		return HRTIMER_RESTART;
	}

	/* check for a softlockup
	 * This is done by making sure a high priority task is
	 * being scheduled.  The task touches the watchdog to
	 * indicate it is getting cpu time.  If it hasn't then
	 * this is a good indication some task is hogging the cpu
	 */
	duration = is_softlockup(touch_ts);
	if (unlikely(duration)) {
		/* only warn once */
		if (__get_cpu_var(soft_watchdog_warn) == true)
			return HRTIMER_RESTART;

		printk(KERN_ERR "BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
			smp_processor_id(), duration,
			current->comm, task_pid_nr(current));
		print_modules();
		print_irqtrace_events(current);
		if (regs)
			show_regs(regs);
		else
			dump_stack();

		if (softlockup_panic)
			panic("softlockup: hung tasks");
		__get_cpu_var(soft_watchdog_warn) = true;
	} else
		__get_cpu_var(soft_watchdog_warn) = false;

	return HRTIMER_RESTART;
}


/*
 * The watchdog thread - touches the timestamp.
 */
static int watchdog(void *unused)
{
	struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
	struct hrtimer *hrtimer = &__raw_get_cpu_var(watchdog_hrtimer);

	sched_setscheduler(current, SCHED_FIFO, &param);

	/* initialize timestamp */
	__touch_watchdog();

	/* kick off the timer for the hardlockup detector */
	/* done here because hrtimer_start can only pin to smp_processor_id() */
	hrtimer_start(hrtimer, ns_to_ktime(get_sample_period()),
		      HRTIMER_MODE_REL_PINNED);

	set_current_state(TASK_INTERRUPTIBLE);
	/*
	 * Run briefly once per second to reset the softlockup timestamp.
	 * If this gets delayed for more than 60 seconds then the
	 * debug-printout triggers in watchdog_timer_fn().
	 */
	while (!kthread_should_stop()) {
		__touch_watchdog();
		schedule();

		if (kthread_should_stop())
			break;

		set_current_state(TASK_INTERRUPTIBLE);
	}
	__set_current_state(TASK_RUNNING);

	return 0;
}


#ifdef CONFIG_HARDLOCKUP_DETECTOR
static int watchdog_nmi_enable(int cpu)
{
	struct perf_event_attr *wd_attr;
	struct perf_event *event = per_cpu(watchdog_ev, cpu);

	/* is it already setup and enabled? */
	if (event && event->state > PERF_EVENT_STATE_OFF)
		goto out;

	/* it is setup but not enabled */
	if (event != NULL)
		goto out_enable;

	/* Try to register using hardware perf events */
	wd_attr = &wd_hw_attr;
	wd_attr->sample_period = hw_nmi_get_sample_period();
	event = perf_event_create_kernel_counter(wd_attr, cpu, -1, watchdog_overflow_callback);
	if (!IS_ERR(event)) {
		printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n");
		goto out_save;
	}

	printk(KERN_ERR "NMI watchdog failed to create perf event on cpu%i: %p\n", cpu, event);
	return -1;

	/* success path */
out_save:
	per_cpu(watchdog_ev, cpu) = event;
out_enable:
	perf_event_enable(per_cpu(watchdog_ev, cpu));
out:
	return 0;
}

static void watchdog_nmi_disable(int cpu)
{
	struct perf_event *event = per_cpu(watchdog_ev, cpu);

	if (event) {
		perf_event_disable(event);
		per_cpu(watchdog_ev, cpu) = NULL;

		/* should be in cleanup, but blocks oprofile */
		perf_event_release_kernel(event);
	}
	return;
}
#else
static int watchdog_nmi_enable(int cpu) { return 0; }
static void watchdog_nmi_disable(int cpu) { return; }
#endif /* CONFIG_HARDLOCKUP_DETECTOR */

/* prepare/enable/disable routines */
static int watchdog_prepare_cpu(int cpu)
{
	struct hrtimer *hrtimer = &per_cpu(watchdog_hrtimer, cpu);

	WARN_ON(per_cpu(softlockup_watchdog, cpu));
	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	hrtimer->function = watchdog_timer_fn;

	return 0;
}

static int watchdog_enable(int cpu)
{
	struct task_struct *p = per_cpu(softlockup_watchdog, cpu);

	/* enable the perf event */
	if (watchdog_nmi_enable(cpu) != 0)
		return -1;

	/* create the watchdog thread */
	if (!p) {
		p = kthread_create(watchdog, (void *)(unsigned long)cpu, "watchdog/%d", cpu);
		if (IS_ERR(p)) {
			printk(KERN_ERR "softlockup watchdog for %i failed\n", cpu);
			return -1;
		}
		kthread_bind(p, cpu);
		per_cpu(watchdog_touch_ts, cpu) = 0;
		per_cpu(softlockup_watchdog, cpu) = p;
		wake_up_process(p);
	}

	/* if any cpu succeeds, watchdog is considered enabled for the system */
	watchdog_enabled = 1;

	return 0;
}

static void watchdog_disable(int cpu)
{
	struct task_struct *p = per_cpu(softlockup_watchdog, cpu);
	struct hrtimer *hrtimer = &per_cpu(watchdog_hrtimer, cpu);

	/*
	 * cancel the timer first to stop incrementing the stats
	 * and waking up the kthread
	 */
	hrtimer_cancel(hrtimer);

	/* disable the perf event */
	watchdog_nmi_disable(cpu);

	/* stop the watchdog thread */
	if (p) {
		per_cpu(softlockup_watchdog, cpu) = NULL;
		kthread_stop(p);
	}
}

static void watchdog_enable_all_cpus(void)
{
	int cpu;
	int result = 0;

	for_each_online_cpu(cpu)
		result += watchdog_enable(cpu);

	if (result)
		printk(KERN_ERR "watchdog: failed to be enabled on some cpus\n");

}

static void watchdog_disable_all_cpus(void)
{
	int cpu;

	for_each_online_cpu(cpu)
		watchdog_disable(cpu);

	/* if all watchdogs are disabled, then they are disabled for the system */
	watchdog_enabled = 0;
}


/* sysctl functions */
#ifdef CONFIG_SYSCTL
/*
 * proc handler for /proc/sys/kernel/nmi_watchdog
 */

int proc_dowatchdog_enabled(struct ctl_table *table, int write,
		     void __user *buffer, size_t *length, loff_t *ppos)
{
	proc_dointvec(table, write, buffer, length, ppos);

	if (watchdog_enabled)
		watchdog_enable_all_cpus();
	else
		watchdog_disable_all_cpus();
	return 0;
}

int proc_dowatchdog_thresh(struct ctl_table *table, int write,
			     void __user *buffer,
			     size_t *lenp, loff_t *ppos)
{
	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
}
#endif /* CONFIG_SYSCTL */


/*
 * Create/destroy watchdog threads as CPUs come and go:
 */
static int __cpuinit
cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu)
{
	int hotcpu = (unsigned long)hcpu;

	switch (action) {
	case CPU_UP_PREPARE:
	case CPU_UP_PREPARE_FROZEN:
		if (watchdog_prepare_cpu(hotcpu))
			return NOTIFY_BAD;
		break;
	case CPU_ONLINE:
	case CPU_ONLINE_FROZEN:
		if (watchdog_enable(hotcpu))
			return NOTIFY_BAD;
		break;
#ifdef CONFIG_HOTPLUG_CPU
	case CPU_UP_CANCELED:
	case CPU_UP_CANCELED_FROZEN:
		watchdog_disable(hotcpu);
		break;
	case CPU_DEAD:
	case CPU_DEAD_FROZEN:
		watchdog_disable(hotcpu);
		break;
#endif /* CONFIG_HOTPLUG_CPU */
	}
	return NOTIFY_OK;
}

static struct notifier_block __cpuinitdata cpu_nfb = {
	.notifier_call = cpu_callback
};

static int __init spawn_watchdog_task(void)
{
	void *cpu = (void *)(long)smp_processor_id();
	int err;

	if (no_watchdog)
		return 0;

	err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);
	WARN_ON(err == NOTIFY_BAD);

	cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
	register_cpu_notifier(&cpu_nfb);

	atomic_notifier_chain_register(&panic_notifier_list, &panic_block);

	return 0;
}
early_initcall(spawn_watchdog_task);
Commit	Line	Data
58687acb DZ	1	/*
	2	* Detect hard and soft lockups on a system
	3	*
	4	* started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
	5	*
	6	* this code detects hard lockups: incidents in where on a CPU
	7	* the kernel does not respond to anything except NMI.
	8	*
	9	* Note: Most of this code is borrowed heavily from softlockup.c,
	10	* so thanks to Ingo for the initial implementation.
	11	* Some chunks also taken from arch/x86/kernel/apic/nmi.c, thanks
	12	* to those contributors as well.
	13	*/
	14
	15	#include <linux/mm.h>
	16	#include <linux/cpu.h>
	17	#include <linux/nmi.h>
	18	#include <linux/init.h>
	19	#include <linux/delay.h>
	20	#include <linux/freezer.h>
	21	#include <linux/kthread.h>
	22	#include <linux/lockdep.h>
	23	#include <linux/notifier.h>
	24	#include <linux/module.h>
	25	#include <linux/sysctl.h>
	26
	27	#include <asm/irq_regs.h>
	28	#include <linux/perf_event.h>
	29
	30	int watchdog_enabled;
	31	int __read_mostly softlockup_thresh = 60;
	32
	33	static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts);
	34	static DEFINE_PER_CPU(struct task_struct *, softlockup_watchdog);
	35	static DEFINE_PER_CPU(struct hrtimer, watchdog_hrtimer);
	36	static DEFINE_PER_CPU(bool, softlockup_touch_sync);
58687acb	37	static DEFINE_PER_CPU(bool, soft_watchdog_warn);
23637d47	38	#ifdef CONFIG_HARDLOCKUP_DETECTOR
cafcd80d DZ	39	static DEFINE_PER_CPU(bool, hard_watchdog_warn);
cafcd80d DZ	40	static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
58687acb DZ	41	static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
	42	static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
	43	static DEFINE_PER_CPU(struct perf_event *, watchdog_ev);
	44	#endif
	45
	46	static int __read_mostly did_panic;
	47	static int __initdata no_watchdog;
	48
	49
	50	/* boot commands */
	51	/*
	52	* Should we panic when a soft-lockup or hard-lockup occurs:
	53	*/
23637d47	54	#ifdef CONFIG_HARDLOCKUP_DETECTOR
58687acb DZ	55	static int hardlockup_panic;
	56
	57	static int __init hardlockup_panic_setup(char *str)
	58	{
	59	if (!strncmp(str, "panic", 5))
	60	hardlockup_panic = 1;
	61	return 1;
	62	}
	63	__setup("nmi_watchdog=", hardlockup_panic_setup);
	64	#endif
	65
	66	unsigned int __read_mostly softlockup_panic =
	67	CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC_VALUE;
	68
	69	static int __init softlockup_panic_setup(char *str)
	70	{
	71	softlockup_panic = simple_strtoul(str, NULL, 0);
	72
	73	return 1;
	74	}
	75	__setup("softlockup_panic=", softlockup_panic_setup);
	76
	77	static int __init nowatchdog_setup(char *str)
	78	{
	79	no_watchdog = 1;
	80	return 1;
	81	}
	82	__setup("nowatchdog", nowatchdog_setup);
	83
	84	/* deprecated */
	85	static int __init nosoftlockup_setup(char *str)
	86	{
	87	no_watchdog = 1;
	88	return 1;
	89	}
	90	__setup("nosoftlockup", nosoftlockup_setup);
	91	/* */
	92
	93
	94	/*
	95	* Returns seconds, approximately. We don't need nanosecond
	96	* resolution, and we don't need to waste time with a big divide when
	97	* 2^30ns == 1.074s.
	98	*/
	99	static unsigned long get_timestamp(int this_cpu)
	100	{
	101	return cpu_clock(this_cpu) >> 30LL; /* 2^30 ~= 10^9 */
	102	}
	103
	104	static unsigned long get_sample_period(void)
	105	{
	106	/*
	107	* convert softlockup_thresh from seconds to ns
	108	* the divide by 5 is to give hrtimer 5 chances to
	109	* increment before the hardlockup detector generates
	110	* a warning
	111	*/
	112	return softlockup_thresh / 5 * NSEC_PER_SEC;
	113	}
	114
	115	/* Commands for resetting the watchdog */
	116	static void __touch_watchdog(void)
	117	{
26e09c6e	118	int this_cpu = smp_processor_id();
58687acb DZ	119
	120	__get_cpu_var(watchdog_touch_ts) = get_timestamp(this_cpu);
	121	}
	122
332fbdbc	123	void touch_softlockup_watchdog(void)
58687acb	124	{
68d3f1d8	125	__raw_get_cpu_var(watchdog_touch_ts) = 0;
58687acb	126	}
0167c781	127	EXPORT_SYMBOL(touch_softlockup_watchdog);
58687acb	128
332fbdbc	129	void touch_all_softlockup_watchdogs(void)
58687acb DZ	130	{
	131	int cpu;
	132
	133	/*
	134	* this is done lockless
	135	* do we care if a 0 races with a timestamp?
	136	* all it means is the softlock check starts one cycle later
	137	*/
	138	for_each_online_cpu(cpu)
	139	per_cpu(watchdog_touch_ts, cpu) = 0;
	140	}
	141
cafcd80d	142	#ifdef CONFIG_HARDLOCKUP_DETECTOR
58687acb DZ	143	void touch_nmi_watchdog(void)
58687acb DZ	144	{
68d3f1d8 DZ	145	if (watchdog_enabled) {
	146	unsigned cpu;
	147
	148	for_each_present_cpu(cpu) {
	149	if (per_cpu(watchdog_nmi_touch, cpu) != true)
	150	per_cpu(watchdog_nmi_touch, cpu) = true;
	151	}
	152	}
332fbdbc	153	touch_softlockup_watchdog();
58687acb DZ	154	}
	155	EXPORT_SYMBOL(touch_nmi_watchdog);
	156
cafcd80d DZ	157	#endif
cafcd80d DZ	158
58687acb DZ	159	void touch_softlockup_watchdog_sync(void)
	160	{
	161	__raw_get_cpu_var(softlockup_touch_sync) = true;
	162	__raw_get_cpu_var(watchdog_touch_ts) = 0;
	163	}
	164
23637d47	165	#ifdef CONFIG_HARDLOCKUP_DETECTOR
58687acb	166	/* watchdog detector functions */
26e09c6e	167	static int is_hardlockup(void)
58687acb	168	{
26e09c6e	169	unsigned long hrint = __get_cpu_var(hrtimer_interrupts);
58687acb	170
26e09c6e	171	if (__get_cpu_var(hrtimer_interrupts_saved) == hrint)
58687acb DZ	172	return 1;
58687acb DZ	173
26e09c6e	174	__get_cpu_var(hrtimer_interrupts_saved) = hrint;
58687acb DZ	175	return 0;
	176	}
	177	#endif
	178
26e09c6e	179	static int is_softlockup(unsigned long touch_ts)
58687acb	180	{
26e09c6e	181	unsigned long now = get_timestamp(smp_processor_id());
58687acb DZ	182
	183	/* Warn about unreasonable delays: */
	184	if (time_after(now, touch_ts + softlockup_thresh))
	185	return now - touch_ts;
	186
	187	return 0;
	188	}
	189
	190	static int
	191	watchdog_panic(struct notifier_block this, unsigned long event, void ptr)
	192	{
	193	did_panic = 1;
	194
	195	return NOTIFY_DONE;
	196	}
	197
	198	static struct notifier_block panic_block = {
	199	.notifier_call = watchdog_panic,
	200	};
	201
23637d47	202	#ifdef CONFIG_HARDLOCKUP_DETECTOR
58687acb DZ	203	static struct perf_event_attr wd_hw_attr = {
	204	.type = PERF_TYPE_HARDWARE,
	205	.config = PERF_COUNT_HW_CPU_CYCLES,
	206	.size = sizeof(struct perf_event_attr),
	207	.pinned = 1,
	208	.disabled = 1,
	209	};
	210
	211	/* Callback function for perf event subsystem */
	212	void watchdog_overflow_callback(struct perf_event *event, int nmi,
	213	struct perf_sample_data *data,
	214	struct pt_regs *regs)
	215	{
c6db67cd PZ	216	/* Ensure the watchdog never gets throttled */
	217	event->hw.interrupts = 0;
	218
d7c54733 DZ	219	if (__get_cpu_var(watchdog_nmi_touch) == true) {
d7c54733 DZ	220	__get_cpu_var(watchdog_nmi_touch) = false;
58687acb DZ	221	return;
	222	}
	223
	224	/* check for a hardlockup
	225	* This is done by making sure our timer interrupt
	226	* is incrementing. The timer interrupt should have
	227	* fired multiple times before we overflow'd. If it hasn't
	228	* then this is a good indication the cpu is stuck
	229	*/
26e09c6e DZ	230	if (is_hardlockup()) {
	231	int this_cpu = smp_processor_id();
	232
58687acb DZ	233	/* only print hardlockups once */
	234	if (__get_cpu_var(hard_watchdog_warn) == true)
	235	return;
	236
	237	if (hardlockup_panic)
	238	panic("Watchdog detected hard LOCKUP on cpu %d", this_cpu);
	239	else
	240	WARN(1, "Watchdog detected hard LOCKUP on cpu %d", this_cpu);
	241
	242	__get_cpu_var(hard_watchdog_warn) = true;
	243	return;
	244	}
	245
	246	__get_cpu_var(hard_watchdog_warn) = false;
	247	return;
	248	}
	249	static void watchdog_interrupt_count(void)
	250	{
	251	__get_cpu_var(hrtimer_interrupts)++;
	252	}
	253	#else
	254	static inline void watchdog_interrupt_count(void) { return; }
23637d47	255	#endif /* CONFIG_HARDLOCKUP_DETECTOR */
58687acb DZ	256
	257	/* watchdog kicker functions */
	258	static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
	259	{
58687acb DZ	260	unsigned long touch_ts = __get_cpu_var(watchdog_touch_ts);
	261	struct pt_regs *regs = get_irq_regs();
	262	int duration;
	263
	264	/* kick the hardlockup detector */
	265	watchdog_interrupt_count();
	266
	267	/* kick the softlockup detector */
	268	wake_up_process(__get_cpu_var(softlockup_watchdog));
	269
	270	/* .. and repeat */
	271	hrtimer_forward_now(hrtimer, ns_to_ktime(get_sample_period()));
	272
	273	if (touch_ts == 0) {
26e09c6e	274	if (unlikely(__get_cpu_var(softlockup_touch_sync))) {
58687acb DZ	275	/*
	276	* If the time stamp was touched atomically
	277	* make sure the scheduler tick is up to date.
	278	*/
26e09c6e	279	__get_cpu_var(softlockup_touch_sync) = false;
58687acb DZ	280	sched_clock_tick();
	281	}
	282	__touch_watchdog();
	283	return HRTIMER_RESTART;
	284	}
	285
	286	/* check for a softlockup
	287	* This is done by making sure a high priority task is
	288	* being scheduled. The task touches the watchdog to
	289	* indicate it is getting cpu time. If it hasn't then
	290	* this is a good indication some task is hogging the cpu
	291	*/
26e09c6e	292	duration = is_softlockup(touch_ts);
58687acb DZ	293	if (unlikely(duration)) {
	294	/* only warn once */
	295	if (__get_cpu_var(soft_watchdog_warn) == true)
	296	return HRTIMER_RESTART;
	297
	298	printk(KERN_ERR "BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
26e09c6e	299	smp_processor_id(), duration,
58687acb DZ	300	current->comm, task_pid_nr(current));
	301	print_modules();
	302	print_irqtrace_events(current);
	303	if (regs)
	304	show_regs(regs);
	305	else
	306	dump_stack();
	307
	308	if (softlockup_panic)
	309	panic("softlockup: hung tasks");
	310	__get_cpu_var(soft_watchdog_warn) = true;
	311	} else
	312	__get_cpu_var(soft_watchdog_warn) = false;
	313
	314	return HRTIMER_RESTART;
	315	}
	316
	317
	318	/*
	319	* The watchdog thread - touches the timestamp.
	320	*/
26e09c6e	321	static int watchdog(void *unused)
58687acb DZ	322	{
58687acb DZ	323	struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
26e09c6e	324	struct hrtimer *hrtimer = &__raw_get_cpu_var(watchdog_hrtimer);
58687acb DZ	325
	326	sched_setscheduler(current, SCHED_FIFO, &param);
	327
	328	/* initialize timestamp */
	329	__touch_watchdog();
	330
	331	/* kick off the timer for the hardlockup detector */
	332	/* done here because hrtimer_start can only pin to smp_processor_id() */
	333	hrtimer_start(hrtimer, ns_to_ktime(get_sample_period()),
	334	HRTIMER_MODE_REL_PINNED);
	335
	336	set_current_state(TASK_INTERRUPTIBLE);
	337	/*
	338	* Run briefly once per second to reset the softlockup timestamp.
	339	* If this gets delayed for more than 60 seconds then the
26e09c6e	340	* debug-printout triggers in watchdog_timer_fn().
58687acb DZ	341	*/
	342	while (!kthread_should_stop()) {
	343	__touch_watchdog();
	344	schedule();
	345
	346	if (kthread_should_stop())
	347	break;
	348
	349	set_current_state(TASK_INTERRUPTIBLE);
	350	}
	351	__set_current_state(TASK_RUNNING);
	352
	353	return 0;
	354	}
	355
	356
23637d47	357	#ifdef CONFIG_HARDLOCKUP_DETECTOR
58687acb DZ	358	static int watchdog_nmi_enable(int cpu)
	359	{
	360	struct perf_event_attr *wd_attr;
	361	struct perf_event *event = per_cpu(watchdog_ev, cpu);
	362
	363	/* is it already setup and enabled? */
	364	if (event && event->state > PERF_EVENT_STATE_OFF)
	365	goto out;
	366
	367	/* it is setup but not enabled */
	368	if (event != NULL)
	369	goto out_enable;
	370
	371	/* Try to register using hardware perf events */
	372	wd_attr = &wd_hw_attr;
	373	wd_attr->sample_period = hw_nmi_get_sample_period();
	374	event = perf_event_create_kernel_counter(wd_attr, cpu, -1, watchdog_overflow_callback);
	375	if (!IS_ERR(event)) {
	376	printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n");
	377	goto out_save;
	378	}
	379
	380	printk(KERN_ERR "NMI watchdog failed to create perf event on cpu%i: %p\n", cpu, event);
	381	return -1;
	382
	383	/* success path */
	384	out_save:
	385	per_cpu(watchdog_ev, cpu) = event;
	386	out_enable:
	387	perf_event_enable(per_cpu(watchdog_ev, cpu));
	388	out:
	389	return 0;
	390	}
	391
	392	static void watchdog_nmi_disable(int cpu)
	393	{
	394	struct perf_event *event = per_cpu(watchdog_ev, cpu);
	395
	396	if (event) {
	397	perf_event_disable(event);
	398	per_cpu(watchdog_ev, cpu) = NULL;
	399
	400	/* should be in cleanup, but blocks oprofile */
	401	perf_event_release_kernel(event);
	402	}
	403	return;
	404	}
	405	#else
	406	static int watchdog_nmi_enable(int cpu) { return 0; }
	407	static void watchdog_nmi_disable(int cpu) { return; }
23637d47	408	#endif /* CONFIG_HARDLOCKUP_DETECTOR */
58687acb DZ	409
	410	/* prepare/enable/disable routines */
	411	static int watchdog_prepare_cpu(int cpu)
	412	{
	413	struct hrtimer *hrtimer = &per_cpu(watchdog_hrtimer, cpu);
	414
	415	WARN_ON(per_cpu(softlockup_watchdog, cpu));
	416	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	417	hrtimer->function = watchdog_timer_fn;
	418
	419	return 0;
	420	}
	421
	422	static int watchdog_enable(int cpu)
	423	{
	424	struct task_struct *p = per_cpu(softlockup_watchdog, cpu);
	425
	426	/* enable the perf event */
	427	if (watchdog_nmi_enable(cpu) != 0)
	428	return -1;
	429
	430	/* create the watchdog thread */
	431	if (!p) {
	432	p = kthread_create(watchdog, (void *)(unsigned long)cpu, "watchdog/%d", cpu);
	433	if (IS_ERR(p)) {
	434	printk(KERN_ERR "softlockup watchdog for %i failed\n", cpu);
	435	return -1;
	436	}
	437	kthread_bind(p, cpu);
	438	per_cpu(watchdog_touch_ts, cpu) = 0;
	439	per_cpu(softlockup_watchdog, cpu) = p;
	440	wake_up_process(p);
	441	}
	442
68d3f1d8 DZ	443	/* if any cpu succeeds, watchdog is considered enabled for the system */
	444	watchdog_enabled = 1;
	445
58687acb DZ	446	return 0;
	447	}
	448
	449	static void watchdog_disable(int cpu)
	450	{
	451	struct task_struct *p = per_cpu(softlockup_watchdog, cpu);
	452	struct hrtimer *hrtimer = &per_cpu(watchdog_hrtimer, cpu);
	453
	454	/*
	455	* cancel the timer first to stop incrementing the stats
	456	* and waking up the kthread
	457	*/
	458	hrtimer_cancel(hrtimer);
	459
	460	/* disable the perf event */
	461	watchdog_nmi_disable(cpu);
	462
	463	/* stop the watchdog thread */
	464	if (p) {
	465	per_cpu(softlockup_watchdog, cpu) = NULL;
	466	kthread_stop(p);
	467	}
58687acb DZ	468	}
	469
	470	static void watchdog_enable_all_cpus(void)
	471	{
	472	int cpu;
eb703f98	473	int result = 0;
58687acb DZ	474
	475	for_each_online_cpu(cpu)
	476	result += watchdog_enable(cpu);
	477
	478	if (result)
	479	printk(KERN_ERR "watchdog: failed to be enabled on some cpus\n");
	480
	481	}
	482
	483	static void watchdog_disable_all_cpus(void)
	484	{
	485	int cpu;
	486
	487	for_each_online_cpu(cpu)
	488	watchdog_disable(cpu);
	489
	490	/* if all watchdogs are disabled, then they are disabled for the system */
	491	watchdog_enabled = 0;
	492	}
	493
	494
	495	/* sysctl functions */
	496	#ifdef CONFIG_SYSCTL
	497	/*
	498	* proc handler for /proc/sys/kernel/nmi_watchdog
	499	*/
	500
	501	int proc_dowatchdog_enabled(struct ctl_table *table, int write,
	502	void __user buffer, size_t length, loff_t *ppos)
	503	{
	504	proc_dointvec(table, write, buffer, length, ppos);
	505
	506	if (watchdog_enabled)
	507	watchdog_enable_all_cpus();
	508	else
	509	watchdog_disable_all_cpus();
	510	return 0;
	511	}
	512
	513	int proc_dowatchdog_thresh(struct ctl_table *table, int write,
	514	void __user *buffer,
	515	size_t lenp, loff_t ppos)
	516	{
	517	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
	518	}
58687acb DZ	519	#endif /* CONFIG_SYSCTL */
	520
	521
	522	/*
	523	* Create/destroy watchdog threads as CPUs come and go:
	524	*/
	525	static int __cpuinit
	526	cpu_callback(struct notifier_block nfb, unsigned long action, void hcpu)
	527	{
	528	int hotcpu = (unsigned long)hcpu;
	529
	530	switch (action) {
	531	case CPU_UP_PREPARE:
	532	case CPU_UP_PREPARE_FROZEN:
	533	if (watchdog_prepare_cpu(hotcpu))
	534	return NOTIFY_BAD;
	535	break;
	536	case CPU_ONLINE:
	537	case CPU_ONLINE_FROZEN:
	538	if (watchdog_enable(hotcpu))
	539	return NOTIFY_BAD;
	540	break;
	541	#ifdef CONFIG_HOTPLUG_CPU
	542	case CPU_UP_CANCELED:
	543	case CPU_UP_CANCELED_FROZEN:
	544	watchdog_disable(hotcpu);
	545	break;
	546	case CPU_DEAD:
	547	case CPU_DEAD_FROZEN:
	548	watchdog_disable(hotcpu);
	549	break;
	550	#endif /* CONFIG_HOTPLUG_CPU */
	551	}
	552	return NOTIFY_OK;
	553	}
	554
	555	static struct notifier_block __cpuinitdata cpu_nfb = {
	556	.notifier_call = cpu_callback
	557	};
	558
	559	static int __init spawn_watchdog_task(void)
	560	{
	561	void cpu = (void )(long)smp_processor_id();
	562	int err;
	563
	564	if (no_watchdog)
	565	return 0;
	566
	567	err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);
	568	WARN_ON(err == NOTIFY_BAD);
	569
	570	cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
	571	register_cpu_notifier(&cpu_nfb);
	572
	573	atomic_notifier_chain_register(&panic_notifier_list, &panic_block);
	574
	575	return 0;
	576	}
	577	early_initcall(spawn_watchdog_task);