watchdog.c 26.4 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
#include <linux/mm.h>
#include <linux/cpu.h>
#include <linux/nmi.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/sysctl.h>
20
#include <linux/smpboot.h>
21
#include <linux/sched/rt.h>
22
#include <linux/tick.h>
23 24

#include <asm/irq_regs.h>
25
#include <linux/kvm_para.h>
26
#include <linux/perf_event.h>
27
#include <linux/kthread.h>
28

29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
/*
 * The run state of the lockup detectors is controlled by the content of the
 * 'watchdog_enabled' variable. Each lockup detector has its dedicated bit -
 * bit 0 for the hard lockup detector and bit 1 for the soft lockup detector.
 *
 * 'watchdog_user_enabled', 'nmi_watchdog_enabled' and 'soft_watchdog_enabled'
 * are variables that are only used as an 'interface' between the parameters
 * in /proc/sys/kernel and the internal state bits in 'watchdog_enabled'. The
 * 'watchdog_thresh' variable is handled differently because its value is not
 * boolean, and the lockup detectors are 'suspended' while 'watchdog_thresh'
 * is equal zero.
 */
#define NMI_WATCHDOG_ENABLED_BIT   0
#define SOFT_WATCHDOG_ENABLED_BIT  1
#define NMI_WATCHDOG_ENABLED      (1 << NMI_WATCHDOG_ENABLED_BIT)
#define SOFT_WATCHDOG_ENABLED     (1 << SOFT_WATCHDOG_ENABLED_BIT)

46 47
static DEFINE_MUTEX(watchdog_proc_mutex);

48 49 50 51 52 53 54 55
#ifdef CONFIG_HARDLOCKUP_DETECTOR
static unsigned long __read_mostly watchdog_enabled = SOFT_WATCHDOG_ENABLED|NMI_WATCHDOG_ENABLED;
#else
static unsigned long __read_mostly watchdog_enabled = SOFT_WATCHDOG_ENABLED;
#endif
int __read_mostly nmi_watchdog_enabled;
int __read_mostly soft_watchdog_enabled;
int __read_mostly watchdog_user_enabled;
56
int __read_mostly watchdog_thresh = 10;
57

58 59 60 61 62
#ifdef CONFIG_SMP
int __read_mostly sysctl_softlockup_all_cpu_backtrace;
#else
#define sysctl_softlockup_all_cpu_backtrace 0
#endif
63 64 65 66 67 68
static struct cpumask watchdog_cpumask __read_mostly;
unsigned long *watchdog_cpumask_bits = cpumask_bits(&watchdog_cpumask);

/* Helper for online, unparked cpus. */
#define for_each_watchdog_cpu(cpu) \
	for_each_cpu_and((cpu), cpu_online_mask, &watchdog_cpumask)
69

70
static int __read_mostly watchdog_running;
71
static u64 __read_mostly sample_period;
72 73 74 75 76 77

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);
78 79
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
static DEFINE_PER_CPU(unsigned long, soft_lockup_hrtimer_cnt);
80
static DEFINE_PER_CPU(struct task_struct *, softlockup_task_ptr_saved);
81
#ifdef CONFIG_HARDLOCKUP_DETECTOR
82 83
static DEFINE_PER_CPU(bool, hard_watchdog_warn);
static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
84 85 86
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
static DEFINE_PER_CPU(struct perf_event *, watchdog_ev);
#endif
87
static unsigned long soft_lockup_nmi_warn;
88 89 90 91 92

/* boot commands */
/*
 * Should we panic when a soft-lockup or hard-lockup occurs:
 */
93
#ifdef CONFIG_HARDLOCKUP_DETECTOR
94 95
static int hardlockup_panic =
			CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE;
96 97 98 99 100 101 102 103
/*
 * We may not want to enable hard lockup detection by default in all cases,
 * for example when running the kernel as a guest on a hypervisor. In these
 * cases this function can be called to disable hard lockup detection. This
 * function should only be executed once by the boot processor before the
 * kernel command line parameters are parsed, because otherwise it is not
 * possible to override this in hardlockup_panic_setup().
 */
104
void hardlockup_detector_disable(void)
105
{
106
	watchdog_enabled &= ~NMI_WATCHDOG_ENABLED;
107 108
}

109 110 111 112
static int __init hardlockup_panic_setup(char *str)
{
	if (!strncmp(str, "panic", 5))
		hardlockup_panic = 1;
113 114
	else if (!strncmp(str, "nopanic", 7))
		hardlockup_panic = 0;
115
	else if (!strncmp(str, "0", 1))
116 117 118
		watchdog_enabled &= ~NMI_WATCHDOG_ENABLED;
	else if (!strncmp(str, "1", 1))
		watchdog_enabled |= NMI_WATCHDOG_ENABLED;
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136
	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)
{
137
	watchdog_enabled = 0;
138 139 140 141 142 143
	return 1;
}
__setup("nowatchdog", nowatchdog_setup);

static int __init nosoftlockup_setup(char *str)
{
144
	watchdog_enabled &= ~SOFT_WATCHDOG_ENABLED;
145 146 147
	return 1;
}
__setup("nosoftlockup", nosoftlockup_setup);
148

149 150 151 152 153 154 155 156 157
#ifdef CONFIG_SMP
static int __init softlockup_all_cpu_backtrace_setup(char *str)
{
	sysctl_softlockup_all_cpu_backtrace =
		!!simple_strtol(str, NULL, 0);
	return 1;
}
__setup("softlockup_all_cpu_backtrace=", softlockup_all_cpu_backtrace_setup);
#endif
158

159 160 161 162 163 164 165
/*
 * 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.
 */
166
static int get_softlockup_thresh(void)
167 168 169
{
	return watchdog_thresh * 2;
}
170 171 172 173 174 175

/*
 * 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.
 */
176
static unsigned long get_timestamp(void)
177
{
178
	return running_clock() >> 30LL;  /* 2^30 ~= 10^9 */
179 180
}

181
static void set_sample_period(void)
182 183
{
	/*
184
	 * convert watchdog_thresh from seconds to ns
185 186 187 188
	 * 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
189
	 */
190
	sample_period = get_softlockup_thresh() * ((u64)NSEC_PER_SEC / 5);
191 192 193 194 195
}

/* Commands for resetting the watchdog */
static void __touch_watchdog(void)
{
196
	__this_cpu_write(watchdog_touch_ts, get_timestamp());
197 198
}

199
void touch_softlockup_watchdog(void)
200
{
201 202 203 204 205
	/*
	 * Preemption can be enabled.  It doesn't matter which CPU's timestamp
	 * gets zeroed here, so use the raw_ operation.
	 */
	raw_cpu_write(watchdog_touch_ts, 0);
206
}
207
EXPORT_SYMBOL(touch_softlockup_watchdog);
208

209
void touch_all_softlockup_watchdogs(void)
210 211 212 213 214 215 216 217
{
	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
	 */
218
	for_each_watchdog_cpu(cpu)
219 220 221
		per_cpu(watchdog_touch_ts, cpu) = 0;
}

222
#ifdef CONFIG_HARDLOCKUP_DETECTOR
223 224
void touch_nmi_watchdog(void)
{
225 226 227 228 229 230 231
	/*
	 * Using __raw here because some code paths have
	 * preemption enabled.  If preemption is enabled
	 * then interrupts should be enabled too, in which
	 * case we shouldn't have to worry about the watchdog
	 * going off.
	 */
232
	raw_cpu_write(watchdog_nmi_touch, true);
233
	touch_softlockup_watchdog();
234 235 236
}
EXPORT_SYMBOL(touch_nmi_watchdog);

237 238
#endif

239 240
void touch_softlockup_watchdog_sync(void)
{
241 242
	__this_cpu_write(softlockup_touch_sync, true);
	__this_cpu_write(watchdog_touch_ts, 0);
243 244
}

245
#ifdef CONFIG_HARDLOCKUP_DETECTOR
246
/* watchdog detector functions */
247
static int is_hardlockup(void)
248
{
249
	unsigned long hrint = __this_cpu_read(hrtimer_interrupts);
250

251
	if (__this_cpu_read(hrtimer_interrupts_saved) == hrint)
252 253
		return 1;

254
	__this_cpu_write(hrtimer_interrupts_saved, hrint);
255 256 257 258
	return 0;
}
#endif

259
static int is_softlockup(unsigned long touch_ts)
260
{
261
	unsigned long now = get_timestamp();
262

263 264 265 266 267
	if (watchdog_enabled & SOFT_WATCHDOG_ENABLED) {
		/* Warn about unreasonable delays. */
		if (time_after(now, touch_ts + get_softlockup_thresh()))
			return now - touch_ts;
	}
268 269 270
	return 0;
}

271
#ifdef CONFIG_HARDLOCKUP_DETECTOR
272

273 274 275 276 277 278 279 280 281
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 */
282
static void watchdog_overflow_callback(struct perf_event *event,
283 284 285
		 struct perf_sample_data *data,
		 struct pt_regs *regs)
{
286 287 288
	/* Ensure the watchdog never gets throttled */
	event->hw.interrupts = 0;

289 290
	if (__this_cpu_read(watchdog_nmi_touch) == true) {
		__this_cpu_write(watchdog_nmi_touch, false);
291 292 293 294 295 296 297 298 299
		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
	 */
300 301 302
	if (is_hardlockup()) {
		int this_cpu = smp_processor_id();

303
		/* only print hardlockups once */
304
		if (__this_cpu_read(hard_watchdog_warn) == true)
305 306 307
			return;

		if (hardlockup_panic)
308 309
			panic("Watchdog detected hard LOCKUP on cpu %d",
			      this_cpu);
310
		else
311 312
			WARN(1, "Watchdog detected hard LOCKUP on cpu %d",
			     this_cpu);
313

314
		__this_cpu_write(hard_watchdog_warn, true);
315 316 317
		return;
	}

318
	__this_cpu_write(hard_watchdog_warn, false);
319 320
	return;
}
321 322
#endif /* CONFIG_HARDLOCKUP_DETECTOR */

323 324
static void watchdog_interrupt_count(void)
{
325
	__this_cpu_inc(hrtimer_interrupts);
326
}
327 328 329

static int watchdog_nmi_enable(unsigned int cpu);
static void watchdog_nmi_disable(unsigned int cpu);
330 331 332 333

/* watchdog kicker functions */
static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
{
334
	unsigned long touch_ts = __this_cpu_read(watchdog_touch_ts);
335 336
	struct pt_regs *regs = get_irq_regs();
	int duration;
337
	int softlockup_all_cpu_backtrace = sysctl_softlockup_all_cpu_backtrace;
338 339 340 341 342

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

	/* kick the softlockup detector */
343
	wake_up_process(__this_cpu_read(softlockup_watchdog));
344 345

	/* .. and repeat */
346
	hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period));
347 348

	if (touch_ts == 0) {
349
		if (unlikely(__this_cpu_read(softlockup_touch_sync))) {
350 351 352 353
			/*
			 * If the time stamp was touched atomically
			 * make sure the scheduler tick is up to date.
			 */
354
			__this_cpu_write(softlockup_touch_sync, false);
355 356
			sched_clock_tick();
		}
357 358 359

		/* Clear the guest paused flag on watchdog reset */
		kvm_check_and_clear_guest_paused();
360 361 362 363 364 365 366 367 368 369
		__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
	 */
370
	duration = is_softlockup(touch_ts);
371
	if (unlikely(duration)) {
372 373 374 375 376 377 378 379
		/*
		 * If a virtual machine is stopped by the host it can look to
		 * the watchdog like a soft lockup, check to see if the host
		 * stopped the vm before we issue the warning
		 */
		if (kvm_check_and_clear_guest_paused())
			return HRTIMER_RESTART;

380
		/* only warn once */
381 382 383 384 385 386 387 388 389 390 391 392 393 394
		if (__this_cpu_read(soft_watchdog_warn) == true) {
			/*
			 * When multiple processes are causing softlockups the
			 * softlockup detector only warns on the first one
			 * because the code relies on a full quiet cycle to
			 * re-arm.  The second process prevents the quiet cycle
			 * and never gets reported.  Use task pointers to detect
			 * this.
			 */
			if (__this_cpu_read(softlockup_task_ptr_saved) !=
			    current) {
				__this_cpu_write(soft_watchdog_warn, false);
				__touch_watchdog();
			}
395
			return HRTIMER_RESTART;
396
		}
397

398 399 400 401 402 403 404 405 406 407 408
		if (softlockup_all_cpu_backtrace) {
			/* Prevent multiple soft-lockup reports if one cpu is already
			 * engaged in dumping cpu back traces
			 */
			if (test_and_set_bit(0, &soft_lockup_nmi_warn)) {
				/* Someone else will report us. Let's give up */
				__this_cpu_write(soft_watchdog_warn, true);
				return HRTIMER_RESTART;
			}
		}

409
		pr_emerg("BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
410
			smp_processor_id(), duration,
411
			current->comm, task_pid_nr(current));
412
		__this_cpu_write(softlockup_task_ptr_saved, current);
413 414 415 416 417 418 419
		print_modules();
		print_irqtrace_events(current);
		if (regs)
			show_regs(regs);
		else
			dump_stack();

420 421 422 423 424 425 426 427 428 429 430
		if (softlockup_all_cpu_backtrace) {
			/* Avoid generating two back traces for current
			 * given that one is already made above
			 */
			trigger_allbutself_cpu_backtrace();

			clear_bit(0, &soft_lockup_nmi_warn);
			/* Barrier to sync with other cpus */
			smp_mb__after_atomic();
		}

Josh Hunt's avatar
Josh Hunt committed
431
		add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
432 433
		if (softlockup_panic)
			panic("softlockup: hung tasks");
434
		__this_cpu_write(soft_watchdog_warn, true);
435
	} else
436
		__this_cpu_write(soft_watchdog_warn, false);
437 438 439 440

	return HRTIMER_RESTART;
}

441 442 443
static void watchdog_set_prio(unsigned int policy, unsigned int prio)
{
	struct sched_param param = { .sched_priority = prio };
444

445 446 447 448
	sched_setscheduler(current, policy, &param);
}

static void watchdog_enable(unsigned int cpu)
449
{
450
	struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer);
451

452 453 454 455
	/* kick off the timer for the hardlockup detector */
	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	hrtimer->function = watchdog_timer_fn;

456 457
	/* Enable the perf event */
	watchdog_nmi_enable(cpu);
458 459

	/* done here because hrtimer_start can only pin to smp_processor_id() */
460
	hrtimer_start(hrtimer, ns_to_ktime(sample_period),
461 462
		      HRTIMER_MODE_REL_PINNED);

463 464 465 466
	/* initialize timestamp */
	watchdog_set_prio(SCHED_FIFO, MAX_RT_PRIO - 1);
	__touch_watchdog();
}
467

468 469
static void watchdog_disable(unsigned int cpu)
{
470
	struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer);
471

472 473 474 475
	watchdog_set_prio(SCHED_NORMAL, 0);
	hrtimer_cancel(hrtimer);
	/* disable the perf event */
	watchdog_nmi_disable(cpu);
476 477
}

478 479 480 481 482
static void watchdog_cleanup(unsigned int cpu, bool online)
{
	watchdog_disable(cpu);
}

483 484 485 486 487 488 489 490 491
static int watchdog_should_run(unsigned int cpu)
{
	return __this_cpu_read(hrtimer_interrupts) !=
		__this_cpu_read(soft_lockup_hrtimer_cnt);
}

/*
 * The watchdog thread function - touches the timestamp.
 *
492
 * It only runs once every sample_period seconds (4 seconds by
493 494 495 496 497 498 499 500 501
 * 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().
 */
static void watchdog(unsigned int cpu)
{
	__this_cpu_write(soft_lockup_hrtimer_cnt,
			 __this_cpu_read(hrtimer_interrupts));
	__touch_watchdog();
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516

	/*
	 * watchdog_nmi_enable() clears the NMI_WATCHDOG_ENABLED bit in the
	 * failure path. Check for failures that can occur asynchronously -
	 * for example, when CPUs are on-lined - and shut down the hardware
	 * perf event on each CPU accordingly.
	 *
	 * The only non-obvious place this bit can be cleared is through
	 * watchdog_nmi_enable(), so a pr_info() is placed there.  Placing a
	 * pr_info here would be too noisy as it would result in a message
	 * every few seconds if the hardlockup was disabled but the softlockup
	 * enabled.
	 */
	if (!(watchdog_enabled & NMI_WATCHDOG_ENABLED))
		watchdog_nmi_disable(cpu);
517
}
518

519
#ifdef CONFIG_HARDLOCKUP_DETECTOR
520 521 522 523 524 525 526
/*
 * People like the simple clean cpu node info on boot.
 * Reduce the watchdog noise by only printing messages
 * that are different from what cpu0 displayed.
 */
static unsigned long cpu0_err;

527
static int watchdog_nmi_enable(unsigned int cpu)
528 529 530 531
{
	struct perf_event_attr *wd_attr;
	struct perf_event *event = per_cpu(watchdog_ev, cpu);

532 533 534
	/* nothing to do if the hard lockup detector is disabled */
	if (!(watchdog_enabled & NMI_WATCHDOG_ENABLED))
		goto out;
535

536 537 538 539 540 541 542 543 544
	/* 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;
545
	wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh);
546 547

	/* Try to register using hardware perf events */
548
	event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback, NULL);
549 550 551 552 553

	/* save cpu0 error for future comparision */
	if (cpu == 0 && IS_ERR(event))
		cpu0_err = PTR_ERR(event);

554
	if (!IS_ERR(event)) {
555 556 557
		/* only print for cpu0 or different than cpu0 */
		if (cpu == 0 || cpu0_err)
			pr_info("enabled on all CPUs, permanently consumes one hw-PMU counter.\n");
558 559 560
		goto out_save;
	}

561 562 563 564 565 566 567 568 569 570 571 572
	/*
	 * Disable the hard lockup detector if _any_ CPU fails to set up
	 * set up the hardware perf event. The watchdog() function checks
	 * the NMI_WATCHDOG_ENABLED bit periodically.
	 *
	 * The barriers are for syncing up watchdog_enabled across all the
	 * cpus, as clear_bit() does not use barriers.
	 */
	smp_mb__before_atomic();
	clear_bit(NMI_WATCHDOG_ENABLED_BIT, &watchdog_enabled);
	smp_mb__after_atomic();

573 574 575
	/* skip displaying the same error again */
	if (cpu > 0 && (PTR_ERR(event) == cpu0_err))
		return PTR_ERR(event);
576 577 578

	/* vary the KERN level based on the returned errno */
	if (PTR_ERR(event) == -EOPNOTSUPP)
579
		pr_info("disabled (cpu%i): not supported (no LAPIC?)\n", cpu);
580
	else if (PTR_ERR(event) == -ENOENT)
581
		pr_warn("disabled (cpu%i): hardware events not enabled\n",
582
			 cpu);
583
	else
584 585
		pr_err("disabled (cpu%i): unable to create perf event: %ld\n",
			cpu, PTR_ERR(event));
586 587 588

	pr_info("Shutting down hard lockup detector on all cpus\n");

589
	return PTR_ERR(event);
590 591 592 593 594 595 596 597 598 599

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

600
static void watchdog_nmi_disable(unsigned int cpu)
601 602 603 604 605 606 607 608 609 610
{
	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);
	}
611 612 613 614
	if (cpu == 0) {
		/* watchdog_nmi_enable() expects this to be zero initially. */
		cpu0_err = 0;
	}
615
}
616 617 618 619 620

void watchdog_nmi_enable_all(void)
{
	int cpu;

621 622 623 624
	mutex_lock(&watchdog_proc_mutex);

	if (!(watchdog_enabled & NMI_WATCHDOG_ENABLED))
		goto unlock;
625 626

	get_online_cpus();
627
	for_each_watchdog_cpu(cpu)
628 629
		watchdog_nmi_enable(cpu);
	put_online_cpus();
630 631

unlock:
632
	mutex_unlock(&watchdog_proc_mutex);
633 634 635 636 637 638
}

void watchdog_nmi_disable_all(void)
{
	int cpu;

639 640
	mutex_lock(&watchdog_proc_mutex);

641
	if (!watchdog_running)
642
		goto unlock;
643 644

	get_online_cpus();
645
	for_each_watchdog_cpu(cpu)
646 647
		watchdog_nmi_disable(cpu);
	put_online_cpus();
648 649 650

unlock:
	mutex_unlock(&watchdog_proc_mutex);
651
}
652
#else
653 654
static int watchdog_nmi_enable(unsigned int cpu) { return 0; }
static void watchdog_nmi_disable(unsigned int cpu) { return; }
655
#endif /* CONFIG_HARDLOCKUP_DETECTOR */
656

657 658 659 660 661 662 663 664 665 666 667
static struct smp_hotplug_thread watchdog_threads = {
	.store			= &softlockup_watchdog,
	.thread_should_run	= watchdog_should_run,
	.thread_fn		= watchdog,
	.thread_comm		= "watchdog/%u",
	.setup			= watchdog_enable,
	.cleanup		= watchdog_cleanup,
	.park			= watchdog_disable,
	.unpark			= watchdog_enable,
};

668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702
/*
 * park all watchdog threads that are specified in 'watchdog_cpumask'
 */
static int watchdog_park_threads(void)
{
	int cpu, ret = 0;

	get_online_cpus();
	for_each_watchdog_cpu(cpu) {
		ret = kthread_park(per_cpu(softlockup_watchdog, cpu));
		if (ret)
			break;
	}
	if (ret) {
		for_each_watchdog_cpu(cpu)
			kthread_unpark(per_cpu(softlockup_watchdog, cpu));
	}
	put_online_cpus();

	return ret;
}

/*
 * unpark all watchdog threads that are specified in 'watchdog_cpumask'
 */
static void watchdog_unpark_threads(void)
{
	int cpu;

	get_online_cpus();
	for_each_watchdog_cpu(cpu)
		kthread_unpark(per_cpu(softlockup_watchdog, cpu));
	put_online_cpus();
}

703 704
static void restart_watchdog_hrtimer(void *info)
{
705
	struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer);
706 707 708 709 710 711 712 713 714 715 716 717 718 719
	int ret;

	/*
	 * No need to cancel and restart hrtimer if it is currently executing
	 * because it will reprogram itself with the new period now.
	 * We should never see it unqueued here because we are running per-cpu
	 * with interrupts disabled.
	 */
	ret = hrtimer_try_to_cancel(hrtimer);
	if (ret == 1)
		hrtimer_start(hrtimer, ns_to_ktime(sample_period),
				HRTIMER_MODE_REL_PINNED);
}

720
static void update_watchdog(int cpu)
721 722 723 724 725 726 727 728 729 730
{
	/*
	 * Make sure that perf event counter will adopt to a new
	 * sampling period. Updating the sampling period directly would
	 * be much nicer but we do not have an API for that now so
	 * let's use a big hammer.
	 * Hrtimer will adopt the new period on the next tick but this
	 * might be late already so we have to restart the timer as well.
	 */
	watchdog_nmi_disable(cpu);
731
	smp_call_function_single(cpu, restart_watchdog_hrtimer, NULL, 1);
732 733 734
	watchdog_nmi_enable(cpu);
}

735
static void update_watchdog_all_cpus(void)
736 737 738 739
{
	int cpu;

	get_online_cpus();
740
	for_each_watchdog_cpu(cpu)
741
		update_watchdog(cpu);
742 743 744
	put_online_cpus();
}

745
static int watchdog_enable_all_cpus(void)
746
{
747
	int err = 0;
748

749
	if (!watchdog_running) {
750 751
		err = smpboot_register_percpu_thread_cpumask(&watchdog_threads,
							     &watchdog_cpumask);
752 753
		if (err)
			pr_err("Failed to create watchdog threads, disabled\n");
754
		else
755
			watchdog_running = 1;
756 757 758 759 760 761
	} else {
		/*
		 * Enable/disable the lockup detectors or
		 * change the sample period 'on the fly'.
		 */
		update_watchdog_all_cpus();
762
	}
763 764

	return err;
765 766
}

767 768 769
/* prepare/enable/disable routines */
/* sysctl functions */
#ifdef CONFIG_SYSCTL
770 771
static void watchdog_disable_all_cpus(void)
{
772 773
	if (watchdog_running) {
		watchdog_running = 0;
774
		smpboot_unregister_percpu_thread(&watchdog_threads);
775
	}
776 777 778
}

/*
779 780 781 782 783 784 785 786 787 788 789 790 791 792
 * Update the run state of the lockup detectors.
 */
static int proc_watchdog_update(void)
{
	int err = 0;

	/*
	 * Watchdog threads won't be started if they are already active.
	 * The 'watchdog_running' variable in watchdog_*_all_cpus() takes
	 * care of this. If those threads are already active, the sample
	 * period will be updated and the lockup detectors will be enabled
	 * or disabled 'on the fly'.
	 */
	if (watchdog_enabled && watchdog_thresh)
793
		err = watchdog_enable_all_cpus();
794 795 796 797 798 799 800
	else
		watchdog_disable_all_cpus();

	return err;

}

801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
/*
 * common function for watchdog, nmi_watchdog and soft_watchdog parameter
 *
 * caller             | table->data points to | 'which' contains the flag(s)
 * -------------------|-----------------------|-----------------------------
 * proc_watchdog      | watchdog_user_enabled | NMI_WATCHDOG_ENABLED or'ed
 *                    |                       | with SOFT_WATCHDOG_ENABLED
 * -------------------|-----------------------|-----------------------------
 * proc_nmi_watchdog  | nmi_watchdog_enabled  | NMI_WATCHDOG_ENABLED
 * -------------------|-----------------------|-----------------------------
 * proc_soft_watchdog | soft_watchdog_enabled | SOFT_WATCHDOG_ENABLED
 */
static int proc_watchdog_common(int which, struct ctl_table *table, int write,
				void __user *buffer, size_t *lenp, loff_t *ppos)
{
	int err, old, new;
	int *watchdog_param = (int *)table->data;

	mutex_lock(&watchdog_proc_mutex);

	/*
	 * If the parameter is being read return the state of the corresponding
	 * bit(s) in 'watchdog_enabled', else update 'watchdog_enabled' and the
	 * run state of the lockup detectors.
	 */
	if (!write) {
		*watchdog_param = (watchdog_enabled & which) != 0;
		err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
	} else {
		err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
		if (err)
			goto out;

		/*
		 * There is a race window between fetching the current value
		 * from 'watchdog_enabled' and storing the new value. During
		 * this race window, watchdog_nmi_enable() can sneak in and
		 * clear the NMI_WATCHDOG_ENABLED bit in 'watchdog_enabled'.
		 * The 'cmpxchg' detects this race and the loop retries.
		 */
		do {
			old = watchdog_enabled;
			/*
			 * If the parameter value is not zero set the
			 * corresponding bit(s), else clear it(them).
			 */
			if (*watchdog_param)
				new = old | which;
			else
				new = old & ~which;
		} while (cmpxchg(&watchdog_enabled, old, new) != old);

		/*
		 * Update the run state of the lockup detectors.
		 * Restore 'watchdog_enabled' on failure.
		 */
		err = proc_watchdog_update();
		if (err)
			watchdog_enabled = old;
	}
out:
	mutex_unlock(&watchdog_proc_mutex);
	return err;
}

866 867 868 869 870 871 872 873 874 875 876 877
/*
 * /proc/sys/kernel/watchdog
 */
int proc_watchdog(struct ctl_table *table, int write,
		  void __user *buffer, size_t *lenp, loff_t *ppos)
{
	return proc_watchdog_common(NMI_WATCHDOG_ENABLED|SOFT_WATCHDOG_ENABLED,
				    table, write, buffer, lenp, ppos);
}

/*
 * /proc/sys/kernel/nmi_watchdog
878
 */
879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894
int proc_nmi_watchdog(struct ctl_table *table, int write,
		      void __user *buffer, size_t *lenp, loff_t *ppos)
{
	return proc_watchdog_common(NMI_WATCHDOG_ENABLED,
				    table, write, buffer, lenp, ppos);
}

/*
 * /proc/sys/kernel/soft_watchdog
 */
int proc_soft_watchdog(struct ctl_table *table, int write,
			void __user *buffer, size_t *lenp, loff_t *ppos)
{
	return proc_watchdog_common(SOFT_WATCHDOG_ENABLED,
				    table, write, buffer, lenp, ppos);
}
895

896 897 898 899 900
/*
 * /proc/sys/kernel/watchdog_thresh
 */
int proc_watchdog_thresh(struct ctl_table *table, int write,
			 void __user *buffer, size_t *lenp, loff_t *ppos)
901
{
902
	int err, old;
903

904
	mutex_lock(&watchdog_proc_mutex);
905

906
	old = ACCESS_ONCE(watchdog_thresh);
907
	err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
908

909
	if (err || !write)
910
		goto out;
911

912
	/*
913 914
	 * Update the sample period.
	 * Restore 'watchdog_thresh' on failure.
915
	 */
916 917 918 919
	set_sample_period();
	err = proc_watchdog_update();
	if (err)
		watchdog_thresh = old;
920 921
out:
	mutex_unlock(&watchdog_proc_mutex);
922
	return err;
923
}
924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957

/*
 * The cpumask is the mask of possible cpus that the watchdog can run
 * on, not the mask of cpus it is actually running on.  This allows the
 * user to specify a mask that will include cpus that have not yet
 * been brought online, if desired.
 */
int proc_watchdog_cpumask(struct ctl_table *table, int write,
			  void __user *buffer, size_t *lenp, loff_t *ppos)
{
	int err;

	mutex_lock(&watchdog_proc_mutex);
	err = proc_do_large_bitmap(table, write, buffer, lenp, ppos);
	if (!err && write) {
		/* Remove impossible cpus to keep sysctl output cleaner. */
		cpumask_and(&watchdog_cpumask, &watchdog_cpumask,
			    cpu_possible_mask);

		if (watchdog_running) {
			/*
			 * Failure would be due to being unable to allocate
			 * a temporary cpumask, so we are likely not in a
			 * position to do much else to make things better.
			 */
			if (smpboot_update_cpumask_percpu_thread(
				    &watchdog_threads, &watchdog_cpumask) != 0)
				pr_err("cpumask update failed\n");
		}
	}
	mutex_unlock(&watchdog_proc_mutex);
	return err;
}

958 959
#endif /* CONFIG_SYSCTL */

960
void __init lockup_detector_init(void)
961
{
962
	set_sample_period();
963

964 965
#ifdef CONFIG_NO_HZ_FULL
	if (tick_nohz_full_enabled()) {
966 967
		pr_info("Disabling watchdog on nohz_full cores by default\n");
		cpumask_copy(&watchdog_cpumask, housekeeping_mask);
968 969 970 971 972 973
	} else
		cpumask_copy(&watchdog_cpumask, cpu_possible_mask);
#else
	cpumask_copy(&watchdog_cpumask, cpu_possible_mask);
#endif

974
	if (watchdog_enabled)
975
		watchdog_enable_all_cpus();
976
}