[linux.git] / kernel / softlockup.c

/*
 * Detect Soft Lockups
 *
 * started by Ingo Molnar, Copyright (C) 2005, 2006 Red Hat, Inc.
 *
 * this code detects soft lockups: incidents in where on a CPU
 * the kernel does not reschedule for 10 seconds or more.
 */
#include <linux/mm.h>
#include <linux/cpu.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/kthread.h>
#include <linux/notifier.h>
#include <linux/module.h>

static DEFINE_SPINLOCK(print_lock);

static DEFINE_PER_CPU(unsigned long, touch_timestamp);
static DEFINE_PER_CPU(unsigned long, print_timestamp);
static DEFINE_PER_CPU(struct task_struct *, watchdog_task);

static int did_panic = 0;

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

	return NOTIFY_DONE;
}

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

void touch_softlockup_watchdog(void)
{
	__raw_get_cpu_var(touch_timestamp) = jiffies;
}
EXPORT_SYMBOL(touch_softlockup_watchdog);

/*
 * This callback runs from the timer interrupt, and checks
 * whether the watchdog thread has hung or not:
 */
void softlockup_tick(void)
{
	int this_cpu = smp_processor_id();
	unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu);

	/* prevent double reports: */
	if (per_cpu(print_timestamp, this_cpu) == touch_timestamp ||
		did_panic ||
			!per_cpu(watchdog_task, this_cpu))
		return;

	/* do not print during early bootup: */
	if (unlikely(system_state != SYSTEM_RUNNING)) {
		touch_softlockup_watchdog();
		return;
	}

	/* Wake up the high-prio watchdog task every second: */
	if (time_after(jiffies, touch_timestamp + HZ))
		wake_up_process(per_cpu(watchdog_task, this_cpu));

	/* Warn about unreasonable 10+ seconds delays: */
	if (time_after(jiffies, touch_timestamp + 10*HZ)) {
		per_cpu(print_timestamp, this_cpu) = touch_timestamp;

		spin_lock(&print_lock);
		printk(KERN_ERR "BUG: soft lockup detected on CPU#%d!\n",
			this_cpu);
		dump_stack();
		spin_unlock(&print_lock);
	}
}

/*
 * The watchdog thread - runs every second and touches the timestamp.
 */
static int watchdog(void * __bind_cpu)
{
	struct sched_param param = { .sched_priority = 99 };

	sched_setscheduler(current, SCHED_FIFO, &param);
	current->flags |= PF_NOFREEZE;

	/*
	 * Run briefly once per second to reset the softlockup timestamp.
	 * If this gets delayed for more than 10 seconds then the
	 * debug-printout triggers in softlockup_tick().
	 */
	while (!kthread_should_stop()) {
		set_current_state(TASK_INTERRUPTIBLE);
		touch_softlockup_watchdog();
		schedule();
	}

	return 0;
}

/*
 * 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;
	struct task_struct *p;

	switch (action) {
	case CPU_UP_PREPARE:
		BUG_ON(per_cpu(watchdog_task, hotcpu));
		p = kthread_create(watchdog, hcpu, "watchdog/%d", hotcpu);
		if (IS_ERR(p)) {
			printk("watchdog for %i failed\n", hotcpu);
			return NOTIFY_BAD;
		}
  		per_cpu(touch_timestamp, hotcpu) = jiffies;
  		per_cpu(watchdog_task, hotcpu) = p;
		kthread_bind(p, hotcpu);
 		break;
	case CPU_ONLINE:
		wake_up_process(per_cpu(watchdog_task, hotcpu));
		break;
#ifdef CONFIG_HOTPLUG_CPU
	case CPU_UP_CANCELED:
		if (!per_cpu(watchdog_task, hotcpu))
			break;
		/* Unbind so it can run.  Fall thru. */
		kthread_bind(per_cpu(watchdog_task, hotcpu),
			     any_online_cpu(cpu_online_map));
	case CPU_DEAD:
		p = per_cpu(watchdog_task, hotcpu);
		per_cpu(watchdog_task, hotcpu) = NULL;
		kthread_stop(p);
		break;
#endif /* CONFIG_HOTPLUG_CPU */
 	}
	return NOTIFY_OK;
}

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

__init void spawn_softlockup_task(void)
{
	void *cpu = (void *)(long)smp_processor_id();
	int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);

	BUG_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);
}
Commit	Line	Data
8446f1d3 IM	1	/*
	2	* Detect Soft Lockups
	3	*
6687a97d	4	* started by Ingo Molnar, Copyright (C) 2005, 2006 Red Hat, Inc.
8446f1d3 IM	5	*
	6	* this code detects soft lockups: incidents in where on a CPU
	7	* the kernel does not reschedule for 10 seconds or more.
	8	*/
8446f1d3 IM	9	#include <linux/mm.h>
	10	#include <linux/cpu.h>
	11	#include <linux/init.h>
	12	#include <linux/delay.h>
	13	#include <linux/kthread.h>
	14	#include <linux/notifier.h>
	15	#include <linux/module.h>
	16
	17	static DEFINE_SPINLOCK(print_lock);
	18
6687a97d IM	19	static DEFINE_PER_CPU(unsigned long, touch_timestamp);
6687a97d IM	20	static DEFINE_PER_CPU(unsigned long, print_timestamp);
8446f1d3 IM	21	static DEFINE_PER_CPU(struct task_struct *, watchdog_task);
	22
	23	static int did_panic = 0;
6687a97d IM	24
	25	static int
	26	softlock_panic(struct notifier_block this, unsigned long event, void ptr)
8446f1d3 IM	27	{
	28	did_panic = 1;
	29
	30	return NOTIFY_DONE;
	31	}
	32
	33	static struct notifier_block panic_block = {
	34	.notifier_call = softlock_panic,
	35	};
	36
	37	void touch_softlockup_watchdog(void)
	38	{
bfe5d834	39	__raw_get_cpu_var(touch_timestamp) = jiffies;
8446f1d3 IM	40	}
	41	EXPORT_SYMBOL(touch_softlockup_watchdog);
	42
	43	/*
	44	* This callback runs from the timer interrupt, and checks
	45	* whether the watchdog thread has hung or not:
	46	*/
6687a97d	47	void softlockup_tick(void)
8446f1d3 IM	48	{
8446f1d3 IM	49	int this_cpu = smp_processor_id();
6687a97d	50	unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu);
8446f1d3	51
6687a97d IM	52	/* prevent double reports: */
	53	if (per_cpu(print_timestamp, this_cpu) == touch_timestamp \|\|
	54	did_panic \|\|
	55	!per_cpu(watchdog_task, this_cpu))
8446f1d3 IM	56	return;
8446f1d3 IM	57
6687a97d IM	58	/* do not print during early bootup: */
	59	if (unlikely(system_state != SYSTEM_RUNNING)) {
	60	touch_softlockup_watchdog();
8446f1d3	61	return;
6687a97d	62	}
8446f1d3	63
6687a97d IM	64	/* Wake up the high-prio watchdog task every second: */
	65	if (time_after(jiffies, touch_timestamp + HZ))
	66	wake_up_process(per_cpu(watchdog_task, this_cpu));
	67
	68	/* Warn about unreasonable 10+ seconds delays: */
	69	if (time_after(jiffies, touch_timestamp + 10*HZ)) {
	70	per_cpu(print_timestamp, this_cpu) = touch_timestamp;
8446f1d3 IM	71
	72	spin_lock(&print_lock);
	73	printk(KERN_ERR "BUG: soft lockup detected on CPU#%d!\n",
	74	this_cpu);
6687a97d	75	dump_stack();
8446f1d3 IM	76	spin_unlock(&print_lock);
	77	}
	78	}
	79
	80	/*
	81	* The watchdog thread - runs every second and touches the timestamp.
	82	*/
	83	static int watchdog(void * __bind_cpu)
	84	{
	85	struct sched_param param = { .sched_priority = 99 };
8446f1d3 IM	86
	87	sched_setscheduler(current, SCHED_FIFO, &param);
	88	current->flags \|= PF_NOFREEZE;
	89
8446f1d3	90	/*
6687a97d IM	91	* Run briefly once per second to reset the softlockup timestamp.
	92	* If this gets delayed for more than 10 seconds then the
	93	* debug-printout triggers in softlockup_tick().
8446f1d3 IM	94	*/
8446f1d3 IM	95	while (!kthread_should_stop()) {
6687a97d	96	set_current_state(TASK_INTERRUPTIBLE);
8446f1d3	97	touch_softlockup_watchdog();
6687a97d	98	schedule();
8446f1d3	99	}
8446f1d3 IM	100
	101	return 0;
	102	}
	103
	104	/*
	105	* Create/destroy watchdog threads as CPUs come and go:
	106	*/
8c78f307	107	static int __cpuinit
8446f1d3 IM	108	cpu_callback(struct notifier_block nfb, unsigned long action, void hcpu)
	109	{
	110	int hotcpu = (unsigned long)hcpu;
	111	struct task_struct *p;
	112
	113	switch (action) {
	114	case CPU_UP_PREPARE:
	115	BUG_ON(per_cpu(watchdog_task, hotcpu));
	116	p = kthread_create(watchdog, hcpu, "watchdog/%d", hotcpu);
	117	if (IS_ERR(p)) {
	118	printk("watchdog for %i failed\n", hotcpu);
	119	return NOTIFY_BAD;
	120	}
185ae6d7	121	per_cpu(touch_timestamp, hotcpu) = jiffies;
8446f1d3 IM	122	per_cpu(watchdog_task, hotcpu) = p;
	123	kthread_bind(p, hotcpu);
	124	break;
	125	case CPU_ONLINE:
8446f1d3 IM	126	wake_up_process(per_cpu(watchdog_task, hotcpu));
	127	break;
	128	#ifdef CONFIG_HOTPLUG_CPU
	129	case CPU_UP_CANCELED:
fc75cdfa HC	130	if (!per_cpu(watchdog_task, hotcpu))
fc75cdfa HC	131	break;
8446f1d3	132	/* Unbind so it can run. Fall thru. */
a4c4af7c HC	133	kthread_bind(per_cpu(watchdog_task, hotcpu),
a4c4af7c HC	134	any_online_cpu(cpu_online_map));
8446f1d3 IM	135	case CPU_DEAD:
	136	p = per_cpu(watchdog_task, hotcpu);
	137	per_cpu(watchdog_task, hotcpu) = NULL;
	138	kthread_stop(p);
	139	break;
	140	#endif /* CONFIG_HOTPLUG_CPU */
	141	}
	142	return NOTIFY_OK;
	143	}
	144
8c78f307	145	static struct notifier_block __cpuinitdata cpu_nfb = {
8446f1d3 IM	146	.notifier_call = cpu_callback
	147	};
	148
	149	__init void spawn_softlockup_task(void)
	150	{
	151	void cpu = (void )(long)smp_processor_id();
07dccf33	152	int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);
8446f1d3	153
07dccf33	154	BUG_ON(err == NOTIFY_BAD);
8446f1d3 IM	155	cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
	156	register_cpu_notifier(&cpu_nfb);
	157
e041c683	158	atomic_notifier_chain_register(&panic_notifier_list, &panic_block);
8446f1d3	159	}