Blame view
kernel/watchdog.c
29.6 KB
58687acba
|
1 2 3 4 5 |
/* * Detect hard and soft lockups on a system * * started by Don Zickus, Copyright (C) 2010 Red Hat, Inc. * |
86f5e6a7b
|
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 |
58687acba
|
9 10 |
* to those contributors as well. */ |
4501980aa
|
11 |
#define pr_fmt(fmt) "NMI watchdog: " fmt |
58687acba
|
12 13 14 15 |
#include <linux/mm.h> #include <linux/cpu.h> #include <linux/nmi.h> #include <linux/init.h> |
58687acba
|
16 17 |
#include <linux/module.h> #include <linux/sysctl.h> |
bcd951cf1
|
18 |
#include <linux/smpboot.h> |
8bd75c77b
|
19 |
#include <linux/sched/rt.h> |
fe4ba3c34
|
20 |
#include <linux/tick.h> |
82607adcf
|
21 |
#include <linux/workqueue.h> |
58687acba
|
22 23 |
#include <asm/irq_regs.h> |
5d1c0f4a8
|
24 |
#include <linux/kvm_para.h> |
58687acba
|
25 |
#include <linux/perf_event.h> |
81a4beef9
|
26 |
#include <linux/kthread.h> |
58687acba
|
27 |
|
84d56e66b
|
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
/* * 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) |
ab992dc38
|
44 |
static DEFINE_MUTEX(watchdog_proc_mutex); |
84d56e66b
|
45 46 47 48 49 50 51 52 |
#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; |
4eec42f39
|
53 |
int __read_mostly watchdog_thresh = 10; |
84d56e66b
|
54 |
|
ed235875e
|
55 56 |
#ifdef CONFIG_SMP int __read_mostly sysctl_softlockup_all_cpu_backtrace; |
55537871e
|
57 |
int __read_mostly sysctl_hardlockup_all_cpu_backtrace; |
ed235875e
|
58 59 |
#else #define sysctl_softlockup_all_cpu_backtrace 0 |
55537871e
|
60 |
#define sysctl_hardlockup_all_cpu_backtrace 0 |
ed235875e
|
61 |
#endif |
fe4ba3c34
|
62 63 64 65 66 67 |
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) |
ed235875e
|
68 |
|
ec6a90661
|
69 70 71 72 73 |
/* * 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. */ |
3c00ea82c
|
74 |
static int __read_mostly watchdog_running; |
ec6a90661
|
75 76 77 78 79 80 81 82 83 84 85 86 87 |
/* * 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; |
0f34c4009
|
88 |
static u64 __read_mostly sample_period; |
58687acba
|
89 90 91 92 93 |
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); |
58687acba
|
94 |
static DEFINE_PER_CPU(bool, soft_watchdog_warn); |
bcd951cf1
|
95 96 |
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts); static DEFINE_PER_CPU(unsigned long, soft_lockup_hrtimer_cnt); |
b1a8de1f5
|
97 |
static DEFINE_PER_CPU(struct task_struct *, softlockup_task_ptr_saved); |
23637d477
|
98 |
#ifdef CONFIG_HARDLOCKUP_DETECTOR |
cafcd80d2
|
99 100 |
static DEFINE_PER_CPU(bool, hard_watchdog_warn); static DEFINE_PER_CPU(bool, watchdog_nmi_touch); |
58687acba
|
101 102 103 |
static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved); static DEFINE_PER_CPU(struct perf_event *, watchdog_ev); #endif |
ed235875e
|
104 |
static unsigned long soft_lockup_nmi_warn; |
58687acba
|
105 |
|
58687acba
|
106 107 108 109 |
/* boot commands */ /* * Should we panic when a soft-lockup or hard-lockup occurs: */ |
23637d477
|
110 |
#ifdef CONFIG_HARDLOCKUP_DETECTOR |
ac1f59124
|
111 |
unsigned int __read_mostly hardlockup_panic = |
fef2c9bc1
|
112 |
CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE; |
55537871e
|
113 |
static unsigned long hardlockup_allcpu_dumped; |
6e7458a6f
|
114 115 116 117 118 119 120 121 |
/* * 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(). */ |
692297d8f
|
122 |
void hardlockup_detector_disable(void) |
6e7458a6f
|
123 |
{ |
692297d8f
|
124 |
watchdog_enabled &= ~NMI_WATCHDOG_ENABLED; |
6e7458a6f
|
125 |
} |
58687acba
|
126 127 128 129 |
static int __init hardlockup_panic_setup(char *str) { if (!strncmp(str, "panic", 5)) hardlockup_panic = 1; |
fef2c9bc1
|
130 131 |
else if (!strncmp(str, "nopanic", 7)) hardlockup_panic = 0; |
5dc305587
|
132 |
else if (!strncmp(str, "0", 1)) |
195daf665
|
133 134 135 |
watchdog_enabled &= ~NMI_WATCHDOG_ENABLED; else if (!strncmp(str, "1", 1)) watchdog_enabled |= NMI_WATCHDOG_ENABLED; |
58687acba
|
136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 |
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) { |
195daf665
|
154 |
watchdog_enabled = 0; |
58687acba
|
155 156 157 |
return 1; } __setup("nowatchdog", nowatchdog_setup); |
58687acba
|
158 159 |
static int __init nosoftlockup_setup(char *str) { |
195daf665
|
160 |
watchdog_enabled &= ~SOFT_WATCHDOG_ENABLED; |
58687acba
|
161 162 163 |
return 1; } __setup("nosoftlockup", nosoftlockup_setup); |
195daf665
|
164 |
|
ed235875e
|
165 166 167 168 169 170 171 172 |
#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); |
55537871e
|
173 174 175 176 177 178 179 |
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); |
ed235875e
|
180 |
#endif |
58687acba
|
181 |
|
4eec42f39
|
182 183 184 185 186 187 188 |
/* * 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. */ |
6e9101aee
|
189 |
static int get_softlockup_thresh(void) |
4eec42f39
|
190 191 192 |
{ return watchdog_thresh * 2; } |
58687acba
|
193 194 195 196 197 198 |
/* * 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. */ |
c06b4f194
|
199 |
static unsigned long get_timestamp(void) |
58687acba
|
200 |
{ |
545a2bf74
|
201 |
return running_clock() >> 30LL; /* 2^30 ~= 10^9 */ |
58687acba
|
202 |
} |
0f34c4009
|
203 |
static void set_sample_period(void) |
58687acba
|
204 205 |
{ /* |
586692a5a
|
206 |
* convert watchdog_thresh from seconds to ns |
86f5e6a7b
|
207 208 209 210 |
* 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 |
58687acba
|
211 |
*/ |
0f34c4009
|
212 |
sample_period = get_softlockup_thresh() * ((u64)NSEC_PER_SEC / 5); |
58687acba
|
213 214 215 216 217 |
} /* Commands for resetting the watchdog */ static void __touch_watchdog(void) { |
c06b4f194
|
218 |
__this_cpu_write(watchdog_touch_ts, get_timestamp()); |
58687acba
|
219 |
} |
03e0d4610
|
220 221 222 223 224 225 226 227 228 |
/** * 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) |
58687acba
|
229 |
{ |
7861144b8
|
230 231 232 233 234 |
/* * 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); |
58687acba
|
235 |
} |
03e0d4610
|
236 237 238 239 |
void touch_softlockup_watchdog(void) { touch_softlockup_watchdog_sched(); |
82607adcf
|
240 |
wq_watchdog_touch(raw_smp_processor_id()); |
03e0d4610
|
241 |
} |
0167c7819
|
242 |
EXPORT_SYMBOL(touch_softlockup_watchdog); |
58687acba
|
243 |
|
332fbdbca
|
244 |
void touch_all_softlockup_watchdogs(void) |
58687acba
|
245 246 247 248 249 250 251 252 |
{ 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 */ |
fe4ba3c34
|
253 |
for_each_watchdog_cpu(cpu) |
58687acba
|
254 |
per_cpu(watchdog_touch_ts, cpu) = 0; |
82607adcf
|
255 |
wq_watchdog_touch(-1); |
58687acba
|
256 |
} |
cafcd80d2
|
257 |
#ifdef CONFIG_HARDLOCKUP_DETECTOR |
58687acba
|
258 259 |
void touch_nmi_watchdog(void) { |
62572e29b
|
260 261 262 263 264 265 266 |
/* * 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. */ |
f7f66b05a
|
267 |
raw_cpu_write(watchdog_nmi_touch, true); |
332fbdbca
|
268 |
touch_softlockup_watchdog(); |
58687acba
|
269 270 |
} EXPORT_SYMBOL(touch_nmi_watchdog); |
cafcd80d2
|
271 |
#endif |
58687acba
|
272 273 |
void touch_softlockup_watchdog_sync(void) { |
f7f66b05a
|
274 275 |
__this_cpu_write(softlockup_touch_sync, true); __this_cpu_write(watchdog_touch_ts, 0); |
58687acba
|
276 |
} |
23637d477
|
277 |
#ifdef CONFIG_HARDLOCKUP_DETECTOR |
58687acba
|
278 |
/* watchdog detector functions */ |
451637e45
|
279 |
static bool is_hardlockup(void) |
58687acba
|
280 |
{ |
909ea9646
|
281 |
unsigned long hrint = __this_cpu_read(hrtimer_interrupts); |
58687acba
|
282 |
|
909ea9646
|
283 |
if (__this_cpu_read(hrtimer_interrupts_saved) == hrint) |
451637e45
|
284 |
return true; |
58687acba
|
285 |
|
909ea9646
|
286 |
__this_cpu_write(hrtimer_interrupts_saved, hrint); |
451637e45
|
287 |
return false; |
58687acba
|
288 289 |
} #endif |
26e09c6ee
|
290 |
static int is_softlockup(unsigned long touch_ts) |
58687acba
|
291 |
{ |
c06b4f194
|
292 |
unsigned long now = get_timestamp(); |
58687acba
|
293 |
|
39d2da216
|
294 |
if ((watchdog_enabled & SOFT_WATCHDOG_ENABLED) && watchdog_thresh){ |
195daf665
|
295 296 297 298 |
/* Warn about unreasonable delays. */ if (time_after(now, touch_ts + get_softlockup_thresh())) return now - touch_ts; } |
58687acba
|
299 300 |
return 0; } |
23637d477
|
301 |
#ifdef CONFIG_HARDLOCKUP_DETECTOR |
1880c4ae1
|
302 |
|
58687acba
|
303 304 305 306 307 308 309 310 311 |
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 */ |
a8b0ca17b
|
312 |
static void watchdog_overflow_callback(struct perf_event *event, |
58687acba
|
313 314 315 |
struct perf_sample_data *data, struct pt_regs *regs) { |
c6db67cda
|
316 317 |
/* Ensure the watchdog never gets throttled */ event->hw.interrupts = 0; |
909ea9646
|
318 319 |
if (__this_cpu_read(watchdog_nmi_touch) == true) { __this_cpu_write(watchdog_nmi_touch, false); |
58687acba
|
320 321 322 323 324 325 326 327 328 |
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 */ |
26e09c6ee
|
329 330 |
if (is_hardlockup()) { int this_cpu = smp_processor_id(); |
58687acba
|
331 |
/* only print hardlockups once */ |
909ea9646
|
332 |
if (__this_cpu_read(hard_watchdog_warn) == true) |
58687acba
|
333 |
return; |
55537871e
|
334 335 336 337 338 |
pr_emerg("Watchdog detected hard LOCKUP on cpu %d", this_cpu); print_modules(); print_irqtrace_events(current); if (regs) show_regs(regs); |
58687acba
|
339 |
else |
55537871e
|
340 341 342 343 344 345 346 347 348 349 350 |
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) |
58c5661f2
|
351 |
nmi_panic(regs, "Hard LOCKUP"); |
58687acba
|
352 |
|
909ea9646
|
353 |
__this_cpu_write(hard_watchdog_warn, true); |
58687acba
|
354 355 |
return; } |
909ea9646
|
356 |
__this_cpu_write(hard_watchdog_warn, false); |
58687acba
|
357 358 |
return; } |
bcd951cf1
|
359 |
#endif /* CONFIG_HARDLOCKUP_DETECTOR */ |
58687acba
|
360 361 |
static void watchdog_interrupt_count(void) { |
909ea9646
|
362 |
__this_cpu_inc(hrtimer_interrupts); |
58687acba
|
363 |
} |
bcd951cf1
|
364 365 366 |
static int watchdog_nmi_enable(unsigned int cpu); static void watchdog_nmi_disable(unsigned int cpu); |
58687acba
|
367 |
|
58cf690a0
|
368 369 |
static int watchdog_enable_all_cpus(void); static void watchdog_disable_all_cpus(void); |
58687acba
|
370 371 372 |
/* watchdog kicker functions */ static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer) { |
909ea9646
|
373 |
unsigned long touch_ts = __this_cpu_read(watchdog_touch_ts); |
58687acba
|
374 375 |
struct pt_regs *regs = get_irq_regs(); int duration; |
ed235875e
|
376 |
int softlockup_all_cpu_backtrace = sysctl_softlockup_all_cpu_backtrace; |
58687acba
|
377 378 379 380 381 |
/* kick the hardlockup detector */ watchdog_interrupt_count(); /* kick the softlockup detector */ |
909ea9646
|
382 |
wake_up_process(__this_cpu_read(softlockup_watchdog)); |
58687acba
|
383 384 |
/* .. and repeat */ |
0f34c4009
|
385 |
hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period)); |
58687acba
|
386 387 |
if (touch_ts == 0) { |
909ea9646
|
388 |
if (unlikely(__this_cpu_read(softlockup_touch_sync))) { |
58687acba
|
389 390 391 392 |
/* * If the time stamp was touched atomically * make sure the scheduler tick is up to date. */ |
909ea9646
|
393 |
__this_cpu_write(softlockup_touch_sync, false); |
58687acba
|
394 395 |
sched_clock_tick(); } |
5d1c0f4a8
|
396 397 398 |
/* Clear the guest paused flag on watchdog reset */ kvm_check_and_clear_guest_paused(); |
58687acba
|
399 400 401 402 403 404 405 406 407 408 |
__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 */ |
26e09c6ee
|
409 |
duration = is_softlockup(touch_ts); |
58687acba
|
410 |
if (unlikely(duration)) { |
5d1c0f4a8
|
411 412 413 414 415 416 417 |
/* * 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; |
58687acba
|
418 |
/* only warn once */ |
b1a8de1f5
|
419 420 421 422 423 424 425 426 427 428 429 430 431 432 |
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(); } |
58687acba
|
433 |
return HRTIMER_RESTART; |
b1a8de1f5
|
434 |
} |
58687acba
|
435 |
|
ed235875e
|
436 437 438 439 440 441 442 443 444 445 |
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; } } |
656c3b79f
|
446 447 |
pr_emerg("BUG: soft lockup - CPU#%d stuck for %us! [%s:%d] ", |
26e09c6ee
|
448 |
smp_processor_id(), duration, |
58687acba
|
449 |
current->comm, task_pid_nr(current)); |
b1a8de1f5
|
450 |
__this_cpu_write(softlockup_task_ptr_saved, current); |
58687acba
|
451 452 453 454 455 456 |
print_modules(); print_irqtrace_events(current); if (regs) show_regs(regs); else dump_stack(); |
ed235875e
|
457 458 459 460 461 462 463 464 465 466 |
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(); } |
69361eef9
|
467 |
add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK); |
58687acba
|
468 469 |
if (softlockup_panic) panic("softlockup: hung tasks"); |
909ea9646
|
470 |
__this_cpu_write(soft_watchdog_warn, true); |
58687acba
|
471 |
} else |
909ea9646
|
472 |
__this_cpu_write(soft_watchdog_warn, false); |
58687acba
|
473 474 475 |
return HRTIMER_RESTART; } |
bcd951cf1
|
476 477 478 |
static void watchdog_set_prio(unsigned int policy, unsigned int prio) { struct sched_param param = { .sched_priority = prio }; |
58687acba
|
479 |
|
bcd951cf1
|
480 481 482 483 |
sched_setscheduler(current, policy, ¶m); } static void watchdog_enable(unsigned int cpu) |
58687acba
|
484 |
{ |
f7f66b05a
|
485 |
struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer); |
58687acba
|
486 |
|
3935e8950
|
487 488 489 |
/* kick off the timer for the hardlockup detector */ hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); hrtimer->function = watchdog_timer_fn; |
bcd951cf1
|
490 491 |
/* Enable the perf event */ watchdog_nmi_enable(cpu); |
58687acba
|
492 |
|
58687acba
|
493 |
/* done here because hrtimer_start can only pin to smp_processor_id() */ |
0f34c4009
|
494 |
hrtimer_start(hrtimer, ns_to_ktime(sample_period), |
58687acba
|
495 |
HRTIMER_MODE_REL_PINNED); |
bcd951cf1
|
496 497 498 499 |
/* initialize timestamp */ watchdog_set_prio(SCHED_FIFO, MAX_RT_PRIO - 1); __touch_watchdog(); } |
58687acba
|
500 |
|
bcd951cf1
|
501 502 |
static void watchdog_disable(unsigned int cpu) { |
f7f66b05a
|
503 |
struct hrtimer *hrtimer = raw_cpu_ptr(&watchdog_hrtimer); |
58687acba
|
504 |
|
bcd951cf1
|
505 506 507 508 |
watchdog_set_prio(SCHED_NORMAL, 0); hrtimer_cancel(hrtimer); /* disable the perf event */ watchdog_nmi_disable(cpu); |
58687acba
|
509 |
} |
b8900bc02
|
510 511 512 513 |
static void watchdog_cleanup(unsigned int cpu, bool online) { watchdog_disable(cpu); } |
bcd951cf1
|
514 515 516 517 518 519 520 521 522 |
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. * |
0f34c4009
|
523 |
* It only runs once every sample_period seconds (4 seconds by |
bcd951cf1
|
524 525 526 527 528 529 530 531 532 |
* 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(); |
bcfba4f4b
|
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 |
/* * 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); |
bcd951cf1
|
548 |
} |
58687acba
|
549 |
|
23637d477
|
550 |
#ifdef CONFIG_HARDLOCKUP_DETECTOR |
a70270468
|
551 552 553 554 555 556 |
/* * 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; |
bcd951cf1
|
557 |
static int watchdog_nmi_enable(unsigned int cpu) |
58687acba
|
558 559 560 |
{ struct perf_event_attr *wd_attr; struct perf_event *event = per_cpu(watchdog_ev, cpu); |
195daf665
|
561 562 563 |
/* nothing to do if the hard lockup detector is disabled */ if (!(watchdog_enabled & NMI_WATCHDOG_ENABLED)) goto out; |
6e7458a6f
|
564 |
|
58687acba
|
565 566 567 568 569 570 571 |
/* 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; |
58687acba
|
572 |
wd_attr = &wd_hw_attr; |
4eec42f39
|
573 |
wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh); |
1880c4ae1
|
574 575 |
/* Try to register using hardware perf events */ |
4dc0da869
|
576 |
event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback, NULL); |
a70270468
|
577 578 579 580 |
/* save cpu0 error for future comparision */ if (cpu == 0 && IS_ERR(event)) cpu0_err = PTR_ERR(event); |
58687acba
|
581 |
if (!IS_ERR(event)) { |
a70270468
|
582 583 584 585 |
/* 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. "); |
58687acba
|
586 587 |
goto out_save; } |
bcfba4f4b
|
588 589 590 591 592 593 594 595 596 597 598 |
/* * 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(); |
a70270468
|
599 600 601 |
/* skip displaying the same error again */ if (cpu > 0 && (PTR_ERR(event) == cpu0_err)) return PTR_ERR(event); |
5651f7f47
|
602 603 604 |
/* vary the KERN level based on the returned errno */ if (PTR_ERR(event) == -EOPNOTSUPP) |
4501980aa
|
605 606 |
pr_info("disabled (cpu%i): not supported (no LAPIC?) ", cpu); |
5651f7f47
|
607 |
else if (PTR_ERR(event) == -ENOENT) |
656c3b79f
|
608 609 |
pr_warn("disabled (cpu%i): hardware events not enabled ", |
4501980aa
|
610 |
cpu); |
5651f7f47
|
611 |
else |
4501980aa
|
612 613 614 |
pr_err("disabled (cpu%i): unable to create perf event: %ld ", cpu, PTR_ERR(event)); |
bcfba4f4b
|
615 616 617 |
pr_info("Shutting down hard lockup detector on all cpus "); |
eac243355
|
618 |
return PTR_ERR(event); |
58687acba
|
619 620 621 622 623 624 625 626 627 |
/* success path */ out_save: per_cpu(watchdog_ev, cpu) = event; out_enable: perf_event_enable(per_cpu(watchdog_ev, cpu)); out: return 0; } |
bcd951cf1
|
628 |
static void watchdog_nmi_disable(unsigned int cpu) |
58687acba
|
629 630 631 632 633 634 635 636 637 638 |
{ 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); } |
df5771495
|
639 640 641 642 |
if (cpu == 0) { /* watchdog_nmi_enable() expects this to be zero initially. */ cpu0_err = 0; } |
58687acba
|
643 |
} |
b3738d293
|
644 |
|
58687acba
|
645 |
#else |
bcd951cf1
|
646 647 |
static int watchdog_nmi_enable(unsigned int cpu) { return 0; } static void watchdog_nmi_disable(unsigned int cpu) { return; } |
23637d477
|
648 |
#endif /* CONFIG_HARDLOCKUP_DETECTOR */ |
58687acba
|
649 |
|
b8900bc02
|
650 651 652 653 654 655 656 657 658 659 |
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, }; |
81a4beef9
|
660 661 |
/* * park all watchdog threads that are specified in 'watchdog_cpumask' |
ee7fed540
|
662 663 664 665 666 667 |
* * 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. |
a2a45b85e
|
668 669 670 |
* * This function may only be called in a context that is protected against * races with CPU hotplug - for example, via get_online_cpus(). |
81a4beef9
|
671 672 673 674 |
*/ static int watchdog_park_threads(void) { int cpu, ret = 0; |
81a4beef9
|
675 676 677 678 679 |
for_each_watchdog_cpu(cpu) { ret = kthread_park(per_cpu(softlockup_watchdog, cpu)); if (ret) break; } |
81a4beef9
|
680 681 682 683 684 685 |
return ret; } /* * unpark all watchdog threads that are specified in 'watchdog_cpumask' |
a2a45b85e
|
686 687 688 |
* * This function may only be called in a context that is protected against * races with CPU hotplug - for example, via get_online_cpus(). |
81a4beef9
|
689 690 691 692 |
*/ static void watchdog_unpark_threads(void) { int cpu; |
81a4beef9
|
693 694 |
for_each_watchdog_cpu(cpu) kthread_unpark(per_cpu(softlockup_watchdog, cpu)); |
81a4beef9
|
695 |
} |
8c073d27d
|
696 697 698 |
/* * Suspend the hard and soft lockup detector by parking the watchdog threads. */ |
ec6a90661
|
699 |
int lockup_detector_suspend(void) |
8c073d27d
|
700 701 |
{ int ret = 0; |
ee89e71eb
|
702 |
get_online_cpus(); |
8c073d27d
|
703 704 705 706 707 708 |
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 |
ec6a90661
|
709 |
* request is active (see related code in 'proc' handlers). |
8c073d27d
|
710 711 712 713 714 715 |
*/ if (watchdog_running && !watchdog_suspended) ret = watchdog_park_threads(); if (ret == 0) watchdog_suspended++; |
c993590c6
|
716 717 718 719 720 721 |
else { watchdog_disable_all_cpus(); pr_err("Failed to suspend lockup detectors, disabled "); watchdog_enabled = 0; } |
8c073d27d
|
722 723 724 725 726 727 728 729 730 |
mutex_unlock(&watchdog_proc_mutex); return ret; } /* * Resume the hard and soft lockup detector by unparking the watchdog threads. */ |
ec6a90661
|
731 |
void lockup_detector_resume(void) |
8c073d27d
|
732 733 734 735 736 737 738 739 740 741 742 743 |
{ 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); |
ee89e71eb
|
744 |
put_online_cpus(); |
8c073d27d
|
745 |
} |
b43cb43cb
|
746 |
static int update_watchdog_all_cpus(void) |
9809b18fc
|
747 |
{ |
b43cb43cb
|
748 749 750 751 752 |
int ret; ret = watchdog_park_threads(); if (ret) return ret; |
d4bdd0b21
|
753 |
watchdog_unpark_threads(); |
b43cb43cb
|
754 755 |
return 0; |
9809b18fc
|
756 |
} |
b2f57c3a0
|
757 |
static int watchdog_enable_all_cpus(void) |
58687acba
|
758 |
{ |
b8900bc02
|
759 |
int err = 0; |
58687acba
|
760 |
|
3c00ea82c
|
761 |
if (!watchdog_running) { |
230ec9390
|
762 763 |
err = smpboot_register_percpu_thread_cpumask(&watchdog_threads, &watchdog_cpumask); |
b8900bc02
|
764 765 766 |
if (err) pr_err("Failed to create watchdog threads, disabled "); |
230ec9390
|
767 |
else |
3c00ea82c
|
768 |
watchdog_running = 1; |
b2f57c3a0
|
769 770 771 772 773 |
} else { /* * Enable/disable the lockup detectors or * change the sample period 'on the fly'. */ |
b43cb43cb
|
774 775 776 777 778 779 780 |
err = update_watchdog_all_cpus(); if (err) { watchdog_disable_all_cpus(); pr_err("Failed to update lockup detectors, disabled "); } |
bcd951cf1
|
781 |
} |
b8900bc02
|
782 |
|
b43cb43cb
|
783 784 |
if (err) watchdog_enabled = 0; |
b8900bc02
|
785 |
return err; |
58687acba
|
786 787 788 789 |
} static void watchdog_disable_all_cpus(void) { |
3c00ea82c
|
790 791 |
if (watchdog_running) { watchdog_running = 0; |
b8900bc02
|
792 |
smpboot_unregister_percpu_thread(&watchdog_threads); |
bcd951cf1
|
793 |
} |
58687acba
|
794 |
} |
58cf690a0
|
795 |
#ifdef CONFIG_SYSCTL |
58687acba
|
796 |
/* |
a0c9cbb93
|
797 798 799 800 801 802 803 804 805 806 807 808 809 810 |
* 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) |
b2f57c3a0
|
811 |
err = watchdog_enable_all_cpus(); |
a0c9cbb93
|
812 813 814 815 816 817 818 819 |
else watchdog_disable_all_cpus(); return err; } /* |
ef246a216
|
820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 |
* 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; |
8614ddef8
|
836 |
get_online_cpus(); |
ef246a216
|
837 |
mutex_lock(&watchdog_proc_mutex); |
8c073d27d
|
838 839 840 841 842 |
if (watchdog_suspended) { /* no parameter changes allowed while watchdog is suspended */ err = -EAGAIN; goto out; } |
ef246a216
|
843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 |
/* * 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); /* |
b43cb43cb
|
876 877 878 879 880 |
* 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. |
ef246a216
|
881 |
*/ |
a1ee1932a
|
882 883 |
if (old == new) goto out; |
ef246a216
|
884 |
err = proc_watchdog_update(); |
ef246a216
|
885 886 887 |
} out: mutex_unlock(&watchdog_proc_mutex); |
8614ddef8
|
888 |
put_online_cpus(); |
ef246a216
|
889 890 891 892 |
return err; } /* |
83a80a390
|
893 894 895 896 897 898 899 900 901 902 903 |
* /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 |
58687acba
|
904 |
*/ |
83a80a390
|
905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 |
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); } |
58687acba
|
921 |
|
83a80a390
|
922 923 924 925 926 |
/* * /proc/sys/kernel/watchdog_thresh */ int proc_watchdog_thresh(struct ctl_table *table, int write, void __user *buffer, size_t *lenp, loff_t *ppos) |
58687acba
|
927 |
{ |
a1ee1932a
|
928 |
int err, old, new; |
58687acba
|
929 |
|
8614ddef8
|
930 |
get_online_cpus(); |
359e6fab6
|
931 |
mutex_lock(&watchdog_proc_mutex); |
bcd951cf1
|
932 |
|
8c073d27d
|
933 934 935 936 937 |
if (watchdog_suspended) { /* no parameter changes allowed while watchdog is suspended */ err = -EAGAIN; goto out; } |
83a80a390
|
938 |
old = ACCESS_ONCE(watchdog_thresh); |
b8900bc02
|
939 |
err = proc_dointvec_minmax(table, write, buffer, lenp, ppos); |
83a80a390
|
940 |
|
b8900bc02
|
941 |
if (err || !write) |
359e6fab6
|
942 |
goto out; |
e04ab2bc4
|
943 |
|
b66a2356d
|
944 |
/* |
d283c640c
|
945 |
* Update the sample period. Restore on failure. |
b66a2356d
|
946 |
*/ |
a1ee1932a
|
947 948 949 |
new = ACCESS_ONCE(watchdog_thresh); if (old == new) goto out; |
83a80a390
|
950 951 |
set_sample_period(); err = proc_watchdog_update(); |
d283c640c
|
952 |
if (err) { |
83a80a390
|
953 |
watchdog_thresh = old; |
d283c640c
|
954 955 |
set_sample_period(); } |
359e6fab6
|
956 957 |
out: mutex_unlock(&watchdog_proc_mutex); |
8614ddef8
|
958 |
put_online_cpus(); |
b8900bc02
|
959 |
return err; |
58687acba
|
960 |
} |
fe4ba3c34
|
961 962 963 964 965 966 967 968 969 970 971 |
/* * 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; |
8614ddef8
|
972 |
get_online_cpus(); |
fe4ba3c34
|
973 |
mutex_lock(&watchdog_proc_mutex); |
8c073d27d
|
974 975 976 977 978 979 |
if (watchdog_suspended) { /* no parameter changes allowed while watchdog is suspended */ err = -EAGAIN; goto out; } |
fe4ba3c34
|
980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 |
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 "); } } |
8c073d27d
|
998 |
out: |
fe4ba3c34
|
999 |
mutex_unlock(&watchdog_proc_mutex); |
8614ddef8
|
1000 |
put_online_cpus(); |
fe4ba3c34
|
1001 1002 |
return err; } |
58687acba
|
1003 |
#endif /* CONFIG_SYSCTL */ |
004417a6d
|
1004 |
void __init lockup_detector_init(void) |
58687acba
|
1005 |
{ |
0f34c4009
|
1006 |
set_sample_period(); |
b8900bc02
|
1007 |
|
fe4ba3c34
|
1008 1009 |
#ifdef CONFIG_NO_HZ_FULL if (tick_nohz_full_enabled()) { |
314b08ff5
|
1010 1011 1012 |
pr_info("Disabling watchdog on nohz_full cores by default "); cpumask_copy(&watchdog_cpumask, housekeeping_mask); |
fe4ba3c34
|
1013 1014 1015 1016 1017 |
} else cpumask_copy(&watchdog_cpumask, cpu_possible_mask); #else cpumask_copy(&watchdog_cpumask, cpu_possible_mask); #endif |
195daf665
|
1018 |
if (watchdog_enabled) |
b2f57c3a0
|
1019 |
watchdog_enable_all_cpus(); |
58687acba
|
1020 |
} |