watchdog.c 28.7 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
#include <linux/workqueue.h>
24 25

#include <asm/irq_regs.h>
26
#include <linux/kvm_para.h>
27
#include <linux/perf_event.h>
28
#include <linux/kthread.h>
29

30 31
static DEFINE_MUTEX(watchdog_proc_mutex);

32 33
#if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
unsigned long __read_mostly watchdog_enabled = SOFT_WATCHDOG_ENABLED|NMI_WATCHDOG_ENABLED;
34
#else
35
unsigned long __read_mostly watchdog_enabled = SOFT_WATCHDOG_ENABLED;
36 37 38 39
#endif
int __read_mostly nmi_watchdog_enabled;
int __read_mostly soft_watchdog_enabled;
int __read_mostly watchdog_user_enabled;
40
int __read_mostly watchdog_thresh = 10;
41

42 43
#ifdef CONFIG_SMP
int __read_mostly sysctl_softlockup_all_cpu_backtrace;
44
int __read_mostly sysctl_hardlockup_all_cpu_backtrace;
45
#endif
46 47 48 49 50 51
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)
52

53 54 55 56 57
/*
 * The 'watchdog_running' variable is set to 1 when the watchdog threads
 * are registered/started and is set to 0 when the watchdog threads are
 * unregistered/stopped, so it is an indicator whether the threads exist.
 */
58
static int __read_mostly watchdog_running;
59 60 61 62 63 64 65 66 67 68 69 70 71 72
/*
 * If a subsystem has a need to deactivate the watchdog temporarily, it
 * can use the suspend/resume interface to achieve this. The content of
 * the 'watchdog_suspended' variable reflects this state. Existing threads
 * are parked/unparked by the lockup_detector_{suspend|resume} functions
 * (see comment blocks pertaining to those functions for further details).
 *
 * 'watchdog_suspended' also prevents threads from being registered/started
 * or unregistered/stopped via parameters in /proc/sys/kernel, so the state
 * of 'watchdog_running' cannot change while the watchdog is deactivated
 * temporarily (see related code in 'proc' handlers).
 */
static int __read_mostly watchdog_suspended;

73
static u64 __read_mostly sample_period;
74 75 76 77 78 79

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);
80 81
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts);
static DEFINE_PER_CPU(unsigned long, soft_lockup_hrtimer_cnt);
82
static DEFINE_PER_CPU(struct task_struct *, softlockup_task_ptr_saved);
83
#ifdef CONFIG_HARDLOCKUP_DETECTOR
84 85
static DEFINE_PER_CPU(bool, hard_watchdog_warn);
static DEFINE_PER_CPU(bool, watchdog_nmi_touch);
86 87 88
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved);
static DEFINE_PER_CPU(struct perf_event *, watchdog_ev);
#endif
89
static unsigned long soft_lockup_nmi_warn;
90 91 92 93 94

/* boot commands */
/*
 * Should we panic when a soft-lockup or hard-lockup occurs:
 */
95
#ifdef CONFIG_HARDLOCKUP_DETECTOR
96
unsigned int __read_mostly hardlockup_panic =
97
			CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE;
98
static unsigned long hardlockup_allcpu_dumped;
99 100 101 102 103 104 105 106
/*
 * 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().
 */
107
void hardlockup_detector_disable(void)
108
{
109
	watchdog_enabled &= ~NMI_WATCHDOG_ENABLED;
110 111
}

112 113 114 115
static int __init hardlockup_panic_setup(char *str)
{
	if (!strncmp(str, "panic", 5))
		hardlockup_panic = 1;
116 117
	else if (!strncmp(str, "nopanic", 7))
		hardlockup_panic = 0;
118
	else if (!strncmp(str, "0", 1))
119 120 121
		watchdog_enabled &= ~NMI_WATCHDOG_ENABLED;
	else if (!strncmp(str, "1", 1))
		watchdog_enabled |= NMI_WATCHDOG_ENABLED;
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
	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)
{
140
	watchdog_enabled = 0;
141 142 143 144 145 146
	return 1;
}
__setup("nowatchdog", nowatchdog_setup);

static int __init nosoftlockup_setup(char *str)
{
147
	watchdog_enabled &= ~SOFT_WATCHDOG_ENABLED;
148 149 150
	return 1;
}
__setup("nosoftlockup", nosoftlockup_setup);
151

152 153 154 155 156 157 158 159
#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);
160 161 162 163 164 165 166
static int __init hardlockup_all_cpu_backtrace_setup(char *str)
{
	sysctl_hardlockup_all_cpu_backtrace =
		!!simple_strtol(str, NULL, 0);
	return 1;
}
__setup("hardlockup_all_cpu_backtrace=", hardlockup_all_cpu_backtrace_setup);
167
#endif
168

169 170 171 172 173 174 175
/*
 * 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.
 */
176
static int get_softlockup_thresh(void)
177 178 179
{
	return watchdog_thresh * 2;
}
180 181 182 183 184 185

/*
 * 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.
 */
186
static unsigned long get_timestamp(void)
187
{
188
	return running_clock() >> 30LL;  /* 2^30 ~= 10^9 */
189 190
}

191
static void set_sample_period(void)
192 193
{
	/*
194
	 * convert watchdog_thresh from seconds to ns
195 196 197 198
	 * 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
199
	 */
200
	sample_period = get_softlockup_thresh() * ((u64)NSEC_PER_SEC / 5);
201 202 203 204 205
}

/* Commands for resetting the watchdog */
static void __touch_watchdog(void)
{
206
	__this_cpu_write(watchdog_touch_ts, get_timestamp());
207 208
}

209 210 211 212 213 214 215 216 217
/**
 * touch_softlockup_watchdog_sched - touch watchdog on scheduler stalls
 *
 * Call when the scheduler may have stalled for legitimate reasons
 * preventing the watchdog task from executing - e.g. the scheduler
 * entering idle state.  This should only be used for scheduler events.
 * Use touch_softlockup_watchdog() for everything else.
 */
void touch_softlockup_watchdog_sched(void)
218
{
219 220 221 222 223
	/*
	 * 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);
224
}
225 226 227 228

void touch_softlockup_watchdog(void)
{
	touch_softlockup_watchdog_sched();
229
	wq_watchdog_touch(raw_smp_processor_id());
230
}
231
EXPORT_SYMBOL(touch_softlockup_watchdog);
232

233
void touch_all_softlockup_watchdogs(void)
234 235 236 237 238 239 240 241
{
	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
	 */
242
	for_each_watchdog_cpu(cpu)
243
		per_cpu(watchdog_touch_ts, cpu) = 0;
244
	wq_watchdog_touch(-1);
245 246
}

247
#ifdef CONFIG_HARDLOCKUP_DETECTOR
248 249
void touch_nmi_watchdog(void)
{
250 251 252 253 254 255 256
	/*
	 * 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.
	 */
257
	raw_cpu_write(watchdog_nmi_touch, true);
258
	touch_softlockup_watchdog();
259 260 261
}
EXPORT_SYMBOL(touch_nmi_watchdog);

262 263
#endif

264 265
void touch_softlockup_watchdog_sync(void)
{
266 267
	__this_cpu_write(softlockup_touch_sync, true);
	__this_cpu_write(watchdog_touch_ts, 0);
268 269
}

270
#ifdef CONFIG_HARDLOCKUP_DETECTOR
271
/* watchdog detector functions */
272
bool is_hardlockup(void)
273
{
274
	unsigned long hrint = __this_cpu_read(hrtimer_interrupts);
275

276
	if (__this_cpu_read(hrtimer_interrupts_saved) == hrint)
277
		return true;
278

279
	__this_cpu_write(hrtimer_interrupts_saved, hrint);
280
	return false;
281 282 283
}
#endif

284
static int is_softlockup(unsigned long touch_ts)
285
{
286
	unsigned long now = get_timestamp();
287

288
	if ((watchdog_enabled & SOFT_WATCHDOG_ENABLED) && watchdog_thresh){
289 290 291 292
		/* Warn about unreasonable delays. */
		if (time_after(now, touch_ts + get_softlockup_thresh()))
			return now - touch_ts;
	}
293 294 295
	return 0;
}

296
#ifdef CONFIG_HARDLOCKUP_DETECTOR
297

298 299 300 301 302 303 304 305 306
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 */
307
static void watchdog_overflow_callback(struct perf_event *event,
308 309 310
		 struct perf_sample_data *data,
		 struct pt_regs *regs)
{
311 312 313
	/* Ensure the watchdog never gets throttled */
	event->hw.interrupts = 0;

314 315
	if (__this_cpu_read(watchdog_nmi_touch) == true) {
		__this_cpu_write(watchdog_nmi_touch, false);
316 317 318 319 320 321 322 323 324
		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
	 */
325 326 327
	if (is_hardlockup()) {
		int this_cpu = smp_processor_id();

328
		/* only print hardlockups once */
329
		if (__this_cpu_read(hard_watchdog_warn) == true)
330 331
			return;

332 333 334 335 336
		pr_emerg("Watchdog detected hard LOCKUP on cpu %d", this_cpu);
		print_modules();
		print_irqtrace_events(current);
		if (regs)
			show_regs(regs);
337
		else
338 339 340 341 342 343 344 345 346 347 348
			dump_stack();

		/*
		 * Perform all-CPU dump only once to avoid multiple hardlockups
		 * generating interleaving traces
		 */
		if (sysctl_hardlockup_all_cpu_backtrace &&
				!test_and_set_bit(0, &hardlockup_allcpu_dumped))
			trigger_allbutself_cpu_backtrace();

		if (hardlockup_panic)
349
			nmi_panic(regs, "Hard LOCKUP");
350

351
		__this_cpu_write(hard_watchdog_warn, true);
352 353 354
		return;
	}

355
	__this_cpu_write(hard_watchdog_warn, false);
356 357
	return;
}
358 359
#endif /* CONFIG_HARDLOCKUP_DETECTOR */

360 361
static void watchdog_interrupt_count(void)
{
362
	__this_cpu_inc(hrtimer_interrupts);
363
}
364 365 366

static int watchdog_nmi_enable(unsigned int cpu);
static void watchdog_nmi_disable(unsigned int cpu);
367

368 369 370
static int watchdog_enable_all_cpus(void);
static void watchdog_disable_all_cpus(void);

371 372 373
/* watchdog kicker functions */
static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
{
374
	unsigned long touch_ts = __this_cpu_read(watchdog_touch_ts);
375 376
	struct pt_regs *regs = get_irq_regs();
	int duration;
377
	int softlockup_all_cpu_backtrace = sysctl_softlockup_all_cpu_backtrace;
378 379 380 381 382

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

	/* kick the softlockup detector */
383
	wake_up_process(__this_cpu_read(softlockup_watchdog));
384 385

	/* .. and repeat */
386
	hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period));
387 388

	if (touch_ts == 0) {
389
		if (unlikely(__this_cpu_read(softlockup_touch_sync))) {
390 391 392 393
			/*
			 * If the time stamp was touched atomically
			 * make sure the scheduler tick is up to date.
			 */
394
			__this_cpu_write(softlockup_touch_sync, false);
395 396
			sched_clock_tick();
		}
397 398 399

		/* Clear the guest paused flag on watchdog reset */
		kvm_check_and_clear_guest_paused();
400 401 402 403 404 405 406 407 408 409
		__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
	 */
410
	duration = is_softlockup(touch_ts);
411
	if (unlikely(duration)) {
412 413 414 415 416 417 418 419
		/*
		 * 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;

420
		/* only warn once */
421 422 423 424 425 426 427 428 429 430 431 432 433 434
		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();
			}
435
			return HRTIMER_RESTART;
436
		}
437

438 439 440 441 442 443 444 445 446 447 448
		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;
			}
		}

449
		pr_emerg("BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
450
			smp_processor_id(), duration,
451
			current->comm, task_pid_nr(current));
452
		__this_cpu_write(softlockup_task_ptr_saved, current);
453 454 455 456 457 458 459
		print_modules();
		print_irqtrace_events(current);
		if (regs)
			show_regs(regs);
		else
			dump_stack();

460 461 462 463 464 465 466 467 468 469 470
		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
471
		add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK);
472 473
		if (softlockup_panic)
			panic("softlockup: hung tasks");
474
		__this_cpu_write(soft_watchdog_warn, true);
475
	} else
476
		__this_cpu_write(soft_watchdog_warn, false);
477 478 479 480

	return HRTIMER_RESTART;
}

481 482 483
static void watchdog_set_prio(unsigned int policy, unsigned int prio)
{
	struct sched_param param = { .sched_priority = prio };
484

485 486 487 488
	sched_setscheduler(current, policy, &param);
}

static void watchdog_enable(unsigned int cpu)
489
{
490
	struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer);
491

492 493 494 495
	/* kick off the timer for the hardlockup detector */
	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	hrtimer->function = watchdog_timer_fn;

496 497
	/* Enable the perf event */
	watchdog_nmi_enable(cpu);
498 499

	/* done here because hrtimer_start can only pin to smp_processor_id() */
500
	hrtimer_start(hrtimer, ns_to_ktime(sample_period),
501 502
		      HRTIMER_MODE_REL_PINNED);

503 504 505 506
	/* initialize timestamp */
	watchdog_set_prio(SCHED_FIFO, MAX_RT_PRIO - 1);
	__touch_watchdog();
}
507

508 509
static void watchdog_disable(unsigned int cpu)
{
510
	struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer);
511

512 513 514 515
	watchdog_set_prio(SCHED_NORMAL, 0);
	hrtimer_cancel(hrtimer);
	/* disable the perf event */
	watchdog_nmi_disable(cpu);
516 517
}

518 519 520 521 522
static void watchdog_cleanup(unsigned int cpu, bool online)
{
	watchdog_disable(cpu);
}

523 524 525 526 527 528 529 530 531
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.
 *
532
 * It only runs once every sample_period seconds (4 seconds by
533 534 535 536 537 538 539 540 541
 * 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();
542 543 544 545 546 547 548 549 550 551 552 553 554 555 556

	/*
	 * 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);
557
}
558

559
#ifdef CONFIG_HARDLOCKUP_DETECTOR
560 561 562 563 564 565 566
/*
 * 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;

567
static int watchdog_nmi_enable(unsigned int cpu)
568 569 570 571
{
	struct perf_event_attr *wd_attr;
	struct perf_event *event = per_cpu(watchdog_ev, cpu);

572 573 574
	/* nothing to do if the hard lockup detector is disabled */
	if (!(watchdog_enabled & NMI_WATCHDOG_ENABLED))
		goto out;
575

576 577 578 579 580 581 582 583 584
	/* 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;
585
	wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh);
586 587

	/* Try to register using hardware perf events */
588
	event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback, NULL);
589 590 591 592 593

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

594
	if (!IS_ERR(event)) {
595 596 597
		/* 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");
598 599 600
		goto out_save;
	}

601 602 603 604 605 606 607 608 609 610 611 612
	/*
	 * 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();

613 614 615
	/* skip displaying the same error again */
	if (cpu > 0 && (PTR_ERR(event) == cpu0_err))
		return PTR_ERR(event);
616 617 618

	/* vary the KERN level based on the returned errno */
	if (PTR_ERR(event) == -EOPNOTSUPP)
619
		pr_info("disabled (cpu%i): not supported (no LAPIC?)\n", cpu);
620
	else if (PTR_ERR(event) == -ENOENT)
621
		pr_warn("disabled (cpu%i): hardware events not enabled\n",
622
			 cpu);
623
	else
624 625
		pr_err("disabled (cpu%i): unable to create perf event: %ld\n",
			cpu, PTR_ERR(event));
626 627 628

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

629
	return PTR_ERR(event);
630 631 632 633 634 635 636 637 638 639

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

640
static void watchdog_nmi_disable(unsigned int cpu)
641 642 643 644 645 646 647 648 649 650
{
	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);
	}
651 652 653 654
	if (cpu == 0) {
		/* watchdog_nmi_enable() expects this to be zero initially. */
		cpu0_err = 0;
	}
655
}
656

657
#else
658 659
static int watchdog_nmi_enable(unsigned int cpu) { return 0; }
static void watchdog_nmi_disable(unsigned int cpu) { return; }
660
#endif /* CONFIG_HARDLOCKUP_DETECTOR */
661

662 663 664 665 666 667 668 669 670 671 672
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,
};

673 674
/*
 * park all watchdog threads that are specified in 'watchdog_cpumask'
675 676 677 678 679 680
 *
 * This function returns an error if kthread_park() of a watchdog thread
 * fails. In this situation, the watchdog threads of some CPUs can already
 * be parked and the watchdog threads of other CPUs can still be runnable.
 * Callers are expected to handle this special condition as appropriate in
 * their context.
681 682 683
 *
 * This function may only be called in a context that is protected against
 * races with CPU hotplug - for example, via get_online_cpus().
684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
 */
static int watchdog_park_threads(void)
{
	int cpu, ret = 0;

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

	return ret;
}

/*
 * unpark all watchdog threads that are specified in 'watchdog_cpumask'
700 701 702
 *
 * This function may only be called in a context that is protected against
 * races with CPU hotplug - for example, via get_online_cpus().
703 704 705 706 707 708 709 710 711
 */
static void watchdog_unpark_threads(void)
{
	int cpu;

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

712 713 714
/*
 * Suspend the hard and soft lockup detector by parking the watchdog threads.
 */
715
int lockup_detector_suspend(void)
716 717 718
{
	int ret = 0;

719
	get_online_cpus();
720 721 722 723 724 725
	mutex_lock(&watchdog_proc_mutex);
	/*
	 * Multiple suspend requests can be active in parallel (counted by
	 * the 'watchdog_suspended' variable). If the watchdog threads are
	 * running, the first caller takes care that they will be parked.
	 * The state of 'watchdog_running' cannot change while a suspend
726
	 * request is active (see related code in 'proc' handlers).
727 728 729 730 731 732
	 */
	if (watchdog_running && !watchdog_suspended)
		ret = watchdog_park_threads();

	if (ret == 0)
		watchdog_suspended++;
733 734 735 736 737
	else {
		watchdog_disable_all_cpus();
		pr_err("Failed to suspend lockup detectors, disabled\n");
		watchdog_enabled = 0;
	}
738 739 740 741 742 743 744 745 746

	mutex_unlock(&watchdog_proc_mutex);

	return ret;
}

/*
 * Resume the hard and soft lockup detector by unparking the watchdog threads.
 */
747
void lockup_detector_resume(void)
748 749 750 751 752 753 754 755 756 757 758 759
{
	mutex_lock(&watchdog_proc_mutex);

	watchdog_suspended--;
	/*
	 * The watchdog threads are unparked if they were previously running
	 * and if there is no more active suspend request.
	 */
	if (watchdog_running && !watchdog_suspended)
		watchdog_unpark_threads();

	mutex_unlock(&watchdog_proc_mutex);
760
	put_online_cpus();
761 762
}

763
static int update_watchdog_all_cpus(void)
764
{
765 766 767 768 769 770
	int ret;

	ret = watchdog_park_threads();
	if (ret)
		return ret;

771
	watchdog_unpark_threads();
772 773

	return 0;
774 775
}

776
static int watchdog_enable_all_cpus(void)
777
{
778
	int err = 0;
779

780
	if (!watchdog_running) {
781 782
		err = smpboot_register_percpu_thread_cpumask(&watchdog_threads,
							     &watchdog_cpumask);
783 784
		if (err)
			pr_err("Failed to create watchdog threads, disabled\n");
785
		else
786
			watchdog_running = 1;
787 788 789 790 791
	} else {
		/*
		 * Enable/disable the lockup detectors or
		 * change the sample period 'on the fly'.
		 */
792 793 794 795 796 797
		err = update_watchdog_all_cpus();

		if (err) {
			watchdog_disable_all_cpus();
			pr_err("Failed to update lockup detectors, disabled\n");
		}
798
	}
799

800 801 802
	if (err)
		watchdog_enabled = 0;

803
	return err;
804 805 806 807
}

static void watchdog_disable_all_cpus(void)
{
808 809
	if (watchdog_running) {
		watchdog_running = 0;
810
		smpboot_unregister_percpu_thread(&watchdog_threads);
811
	}
812 813
}

814 815
#ifdef CONFIG_SYSCTL

816
/*
817 818 819 820 821 822 823 824 825 826 827 828 829 830
 * 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)
831
		err = watchdog_enable_all_cpus();
832 833 834 835 836 837 838
	else
		watchdog_disable_all_cpus();

	return err;

}

839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856
/*
 * 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;

857
	get_online_cpus();
858 859
	mutex_lock(&watchdog_proc_mutex);

860 861 862 863 864 865
	if (watchdog_suspended) {
		/* no parameter changes allowed while watchdog is suspended */
		err = -EAGAIN;
		goto out;
	}

866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898
	/*
	 * 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);

		/*
899 900 901 902 903
		 * Update the run state of the lockup detectors. There is _no_
		 * need to check the value returned by proc_watchdog_update()
		 * and to restore the previous value of 'watchdog_enabled' as
		 * both lockup detectors are disabled if proc_watchdog_update()
		 * returns an error.
904
		 */
905 906 907
		if (old == new)
			goto out;

908 909 910 911
		err = proc_watchdog_update();
	}
out:
	mutex_unlock(&watchdog_proc_mutex);
912
	put_online_cpus();
913 914 915
	return err;
}

916 917 918 919 920 921 922 923 924 925 926 927
/*
 * /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
928
 */
929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944
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);
}
945

946 947 948 949 950
/*
 * /proc/sys/kernel/watchdog_thresh
 */
int proc_watchdog_thresh(struct ctl_table *table, int write,
			 void __user *buffer, size_t *lenp, loff_t *ppos)
951
{
952
	int err, old, new;
953

954
	get_online_cpus();
955
	mutex_lock(&watchdog_proc_mutex);
956

957 958 959 960 961 962
	if (watchdog_suspended) {
		/* no parameter changes allowed while watchdog is suspended */
		err = -EAGAIN;
		goto out;
	}

963
	old = ACCESS_ONCE(watchdog_thresh);
964
	err = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
965

966
	if (err || !write)
967
		goto out;
968

969
	/*
970
	 * Update the sample period. Restore on failure.
971
	 */
972 973 974 975
	new = ACCESS_ONCE(watchdog_thresh);
	if (old == new)
		goto out;

976 977
	set_sample_period();
	err = proc_watchdog_update();
978
	if (err) {
979
		watchdog_thresh = old;
980 981
		set_sample_period();
	}
982 983
out:
	mutex_unlock(&watchdog_proc_mutex);
984
	put_online_cpus();
985
	return err;
986
}
987 988 989 990 991 992 993 994 995 996 997 998

/*
 * 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;

999
	get_online_cpus();
1000
	mutex_lock(&watchdog_proc_mutex);
1001 1002 1003 1004 1005 1006 1007

	if (watchdog_suspended) {
		/* no parameter changes allowed while watchdog is suspended */
		err = -EAGAIN;
		goto out;
	}

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
	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");
		}
	}
1025
out:
1026
	mutex_unlock(&watchdog_proc_mutex);
1027
	put_online_cpus();
1028 1029 1030
	return err;
}

1031 1032
#endif /* CONFIG_SYSCTL */

1033
void __init lockup_detector_init(void)
1034
{
1035
	set_sample_period();
1036

1037 1038
#ifdef CONFIG_NO_HZ_FULL
	if (tick_nohz_full_enabled()) {
1039 1040
		pr_info("Disabling watchdog on nohz_full cores by default\n");
		cpumask_copy(&watchdog_cpumask, housekeeping_mask);
1041 1042 1043 1044 1045 1046
	} else
		cpumask_copy(&watchdog_cpumask, cpu_possible_mask);
#else
	cpumask_copy(&watchdog_cpumask, cpu_possible_mask);
#endif

1047
	if (watchdog_enabled)
1048
		watchdog_enable_all_cpus();
1049
}