watchdog.c 14.6 KB
Newer Older
1 2 3 4 5
/*
 * Detect hard and soft lockups on a system
 *
 * started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
 *
6 7 8
 * Note: Most of this code is borrowed heavily from the original softlockup
 * detector, so thanks to Ingo for the initial implementation.
 * Some chunks also taken from the old x86-specific nmi watchdog code, thanks
9 10 11
 * to those contributors as well.
 */

12 13
#define pr_fmt(fmt) "NMI watchdog: " fmt

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
#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>

29
int watchdog_enabled = 1;
30
int __read_mostly watchdog_thresh = 10;
31 32 33 34 35 36

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);
37
#ifdef CONFIG_HARDLOCKUP_DETECTOR
38 39
static DEFINE_PER_CPU(bool, hard_watchdog_warn);
static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
40 41 42 43 44 45 46 47 48
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

/* boot commands */
/*
 * Should we panic when a soft-lockup or hard-lockup occurs:
 */
49
#ifdef CONFIG_HARDLOCKUP_DETECTOR
50 51
static int hardlockup_panic =
			CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE;
52 53 54 55 56

static int __init hardlockup_panic_setup(char *str)
{
	if (!strncmp(str, "panic", 5))
		hardlockup_panic = 1;
57 58
	else if (!strncmp(str, "nopanic", 7))
		hardlockup_panic = 0;
59
	else if (!strncmp(str, "0", 1))
60
		watchdog_enabled = 0;
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78
	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)
{
79
	watchdog_enabled = 0;
80 81 82 83 84 85 86
	return 1;
}
__setup("nowatchdog", nowatchdog_setup);

/* deprecated */
static int __init nosoftlockup_setup(char *str)
{
87
	watchdog_enabled = 0;
88 89 90 91 92
	return 1;
}
__setup("nosoftlockup", nosoftlockup_setup);
/*  */

93 94 95 96 97 98 99
/*
 * Hard-lockup warnings should be triggered after just a few seconds. Soft-
 * lockups can have false positives under extreme conditions. So we generally
 * want a higher threshold for soft lockups than for hard lockups. So we couple
 * the thresholds with a factor: we make the soft threshold twice the amount of
 * time the hard threshold is.
 */
100
static int get_softlockup_thresh(void)
101 102 103
{
	return watchdog_thresh * 2;
}
104 105 106 107 108 109 110 111 112 113 114 115 116 117

/*
 * 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)
{
	/*
118
	 * convert watchdog_thresh from seconds to ns
119 120 121 122
	 * the divide by 5 is to give hrtimer several chances (two
	 * or three with the current relation between the soft
	 * and hard thresholds) to increment before the
	 * hardlockup detector generates a warning
123
	 */
124
	return get_softlockup_thresh() * (NSEC_PER_SEC / 5);
125 126 127 128 129
}

/* Commands for resetting the watchdog */
static void __touch_watchdog(void)
{
130
	int this_cpu = smp_processor_id();
131

132
	__this_cpu_write(watchdog_touch_ts, get_timestamp(this_cpu));
133 134
}

135
void touch_softlockup_watchdog(void)
136
{
137
	__this_cpu_write(watchdog_touch_ts, 0);
138
}
139
EXPORT_SYMBOL(touch_softlockup_watchdog);
140

141
void touch_all_softlockup_watchdogs(void)
142 143 144 145 146 147 148 149 150 151 152 153
{
	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;
}

154
#ifdef CONFIG_HARDLOCKUP_DETECTOR
155 156
void touch_nmi_watchdog(void)
{
157 158 159 160 161 162 163 164
	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;
		}
	}
165
	touch_softlockup_watchdog();
166 167 168
}
EXPORT_SYMBOL(touch_nmi_watchdog);

169 170
#endif

171 172 173 174 175 176
void touch_softlockup_watchdog_sync(void)
{
	__raw_get_cpu_var(softlockup_touch_sync) = true;
	__raw_get_cpu_var(watchdog_touch_ts) = 0;
}

177
#ifdef CONFIG_HARDLOCKUP_DETECTOR
178
/* watchdog detector functions */
179
static int is_hardlockup(void)
180
{
181
	unsigned long hrint = __this_cpu_read(hrtimer_interrupts);
182

183
	if (__this_cpu_read(hrtimer_interrupts_saved) == hrint)
184 185
		return 1;

186
	__this_cpu_write(hrtimer_interrupts_saved, hrint);
187 188 189 190
	return 0;
}
#endif

191
static int is_softlockup(unsigned long touch_ts)
192
{
193
	unsigned long now = get_timestamp(smp_processor_id());
194 195

	/* Warn about unreasonable delays: */
196
	if (time_after(now, touch_ts + get_softlockup_thresh()))
197 198 199 200 201
		return now - touch_ts;

	return 0;
}

202
#ifdef CONFIG_HARDLOCKUP_DETECTOR
203

204 205 206 207 208 209 210 211 212
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 */
213
static void watchdog_overflow_callback(struct perf_event *event,
214 215 216
		 struct perf_sample_data *data,
		 struct pt_regs *regs)
{
217 218 219
	/* Ensure the watchdog never gets throttled */
	event->hw.interrupts = 0;

220 221
	if (__this_cpu_read(watchdog_nmi_touch) == true) {
		__this_cpu_write(watchdog_nmi_touch, false);
222 223 224 225 226 227 228 229 230
		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
	 */
231 232 233
	if (is_hardlockup()) {
		int this_cpu = smp_processor_id();

234
		/* only print hardlockups once */
235
		if (__this_cpu_read(hard_watchdog_warn) == true)
236 237 238 239 240 241 242
			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);

243
		__this_cpu_write(hard_watchdog_warn, true);
244 245 246
		return;
	}

247
	__this_cpu_write(hard_watchdog_warn, false);
248 249 250 251
	return;
}
static void watchdog_interrupt_count(void)
{
252
	__this_cpu_inc(hrtimer_interrupts);
253 254 255
}
#else
static inline void watchdog_interrupt_count(void) { return; }
256
#endif /* CONFIG_HARDLOCKUP_DETECTOR */
257 258 259 260

/* watchdog kicker functions */
static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
{
261
	unsigned long touch_ts = __this_cpu_read(watchdog_touch_ts);
262 263 264 265 266 267 268
	struct pt_regs *regs = get_irq_regs();
	int duration;

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

	/* kick the softlockup detector */
269
	wake_up_process(__this_cpu_read(softlockup_watchdog));
270 271 272 273 274

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

	if (touch_ts == 0) {
275
		if (unlikely(__this_cpu_read(softlockup_touch_sync))) {
276 277 278 279
			/*
			 * If the time stamp was touched atomically
			 * make sure the scheduler tick is up to date.
			 */
280
			__this_cpu_write(softlockup_touch_sync, false);
281 282 283 284 285 286 287 288 289 290 291 292
			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
	 */
293
	duration = is_softlockup(touch_ts);
294 295
	if (unlikely(duration)) {
		/* only warn once */
296
		if (__this_cpu_read(soft_watchdog_warn) == true)
297 298
			return HRTIMER_RESTART;

299
		printk(KERN_EMERG "BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
300
			smp_processor_id(), duration,
301 302 303 304 305 306 307 308 309 310
			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");
311
		__this_cpu_write(soft_watchdog_warn, true);
312
	} else
313
		__this_cpu_write(soft_watchdog_warn, false);
314 315 316 317 318 319 320 321

	return HRTIMER_RESTART;
}


/*
 * The watchdog thread - touches the timestamp.
 */
322
static int watchdog(void *unused)
323
{
324
	struct sched_param param = { .sched_priority = 0 };
325
	struct hrtimer *hrtimer = &__raw_get_cpu_var(watchdog_hrtimer);
326 327 328 329 330 331 332 333 334 335 336

	/* 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);
	/*
337 338 339 340 341
	 * Run briefly (kicked by the hrtimer callback function) once every
	 * get_sample_period() seconds (4 seconds by default) to reset the
	 * softlockup timestamp. If this gets delayed for more than
	 * 2*watchdog_thresh seconds then the debug-printout triggers in
	 * watchdog_timer_fn().
342 343 344 345 346 347 348 349 350 351
	 */
	while (!kthread_should_stop()) {
		__touch_watchdog();
		schedule();

		if (kthread_should_stop())
			break;

		set_current_state(TASK_INTERRUPTIBLE);
	}
352 353 354 355
	/*
	 * Drop the policy/priority elevation during thread exit to avoid a
	 * scheduling latency spike.
	 */
356
	__set_current_state(TASK_RUNNING);
357
	sched_setscheduler(current, SCHED_NORMAL, &param);
358 359 360 361
	return 0;
}


362
#ifdef CONFIG_HARDLOCKUP_DETECTOR
363 364 365 366 367 368 369 370 371 372 373 374 375 376
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;

	wd_attr = &wd_hw_attr;
377
	wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh);
378 379

	/* Try to register using hardware perf events */
380
	event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback, NULL);
381
	if (!IS_ERR(event)) {
382
		pr_info("enabled, takes one hw-pmu counter.\n");
383 384 385
		goto out_save;
	}

386 387 388

	/* vary the KERN level based on the returned errno */
	if (PTR_ERR(event) == -EOPNOTSUPP)
389
		pr_info("disabled (cpu%i): not supported (no LAPIC?)\n", cpu);
390
	else if (PTR_ERR(event) == -ENOENT)
391 392
		pr_warning("disabled (cpu%i): hardware events not enabled\n",
			 cpu);
393
	else
394 395
		pr_err("disabled (cpu%i): unable to create perf event: %ld\n",
			cpu, PTR_ERR(event));
396
	return PTR_ERR(event);
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422

	/* 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; }
423
#endif /* CONFIG_HARDLOCKUP_DETECTOR */
424 425

/* prepare/enable/disable routines */
426
static void watchdog_prepare_cpu(int cpu)
427 428 429 430 431 432 433 434 435 436 437
{
	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;
}

static int watchdog_enable(int cpu)
{
	struct task_struct *p = per_cpu(softlockup_watchdog, cpu);
438
	int err = 0;
439 440

	/* enable the perf event */
441
	err = watchdog_nmi_enable(cpu);
442 443

	/* Regardless of err above, fall through and start softlockup */
444 445 446

	/* create the watchdog thread */
	if (!p) {
447
		struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
448
		p = kthread_create_on_node(watchdog, NULL, cpu_to_node(cpu), "watchdog/%d", cpu);
449
		if (IS_ERR(p)) {
450
			pr_err("softlockup watchdog for %i failed\n", cpu);
451
			if (!err) {
452 453
				/* if hardlockup hasn't already set this */
				err = PTR_ERR(p);
454 455 456
				/* and disable the perf event */
				watchdog_nmi_disable(cpu);
			}
457
			goto out;
458
		}
459
		sched_setscheduler(p, SCHED_FIFO, &param);
460 461 462 463 464 465
		kthread_bind(p, cpu);
		per_cpu(watchdog_touch_ts, cpu) = 0;
		per_cpu(softlockup_watchdog, cpu) = p;
		wake_up_process(p);
	}

466 467
out:
	return err;
468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490
}

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);
	}
}

491 492
/* sysctl functions */
#ifdef CONFIG_SYSCTL
493 494 495
static void watchdog_enable_all_cpus(void)
{
	int cpu;
496 497

	watchdog_enabled = 0;
498 499

	for_each_online_cpu(cpu)
500 501 502 503
		if (!watchdog_enable(cpu))
			/* if any cpu succeeds, watchdog is considered
			   enabled for the system */
			watchdog_enabled = 1;
504

505
	if (!watchdog_enabled)
506
		pr_err("failed to be enabled on some cpus\n");
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522

}

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;
}


/*
523
 * proc handler for /proc/sys/kernel/nmi_watchdog,watchdog_thresh
524 525
 */

526 527
int proc_dowatchdog(struct ctl_table *table, int write,
		    void __user *buffer, size_t *lenp, loff_t *ppos)
528
{
529
	int ret;
530

531
	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
532 533 534
	if (ret || !write)
		goto out;

535
	if (watchdog_enabled && watchdog_thresh)
536 537 538 539 540 541
		watchdog_enable_all_cpus();
	else
		watchdog_disable_all_cpus();

out:
	return ret;
542 543 544 545 546 547 548 549 550 551 552 553 554 555 556
}
#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:
557
		watchdog_prepare_cpu(hotcpu);
558 559 560
		break;
	case CPU_ONLINE:
	case CPU_ONLINE_FROZEN:
561
		if (watchdog_enabled)
562
			watchdog_enable(hotcpu);
563 564 565 566 567 568 569 570 571 572 573 574
		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 */
	}
575 576 577 578 579 580 581

	/*
	 * hardlockup and softlockup are not important enough
	 * to block cpu bring up.  Just always succeed and
	 * rely on printk output to flag problems.
	 */
	return NOTIFY_OK;
582 583 584 585 586 587
}

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

588
void __init lockup_detector_init(void)
589 590 591 592 593
{
	void *cpu = (void *)(long)smp_processor_id();
	int err;

	err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);
594
	WARN_ON(notifier_to_errno(err));
595 596 597 598

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

599
	return;
600
}