tick-sched.c 20.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11
/*
 *  linux/kernel/time/tick-sched.c
 *
 *  Copyright(C) 2005-2006, Thomas Gleixner <tglx@linutronix.de>
 *  Copyright(C) 2005-2007, Red Hat, Inc., Ingo Molnar
 *  Copyright(C) 2006-2007  Timesys Corp., Thomas Gleixner
 *
 *  No idle tick implementation for low and high resolution timers
 *
 *  Started by: Thomas Gleixner and Ingo Molnar
 *
P
Pavel Machek 已提交
12
 *  Distribute under GPLv2.
13 14 15 16 17 18 19 20 21 22
 */
#include <linux/cpu.h>
#include <linux/err.h>
#include <linux/hrtimer.h>
#include <linux/interrupt.h>
#include <linux/kernel_stat.h>
#include <linux/percpu.h>
#include <linux/profile.h>
#include <linux/sched.h>
#include <linux/tick.h>
23
#include <linux/module.h>
24

25 26
#include <asm/irq_regs.h>

27 28 29 30 31 32 33 34 35 36 37 38
#include "tick-internal.h"

/*
 * Per cpu nohz control structure
 */
static DEFINE_PER_CPU(struct tick_sched, tick_cpu_sched);

/*
 * The time, when the last jiffy update happened. Protected by xtime_lock.
 */
static ktime_t last_jiffies_update;

39 40 41 42 43
struct tick_sched *tick_get_tick_sched(int cpu)
{
	return &per_cpu(tick_cpu_sched, cpu);
}

44 45 46 47 48 49 50 51
/*
 * Must be called with interrupts disabled !
 */
static void tick_do_update_jiffies64(ktime_t now)
{
	unsigned long ticks = 0;
	ktime_t delta;

52 53 54 55 56 57 58
	/*
	 * Do a quick check without holding xtime_lock:
	 */
	delta = ktime_sub(now, last_jiffies_update);
	if (delta.tv64 < tick_period.tv64)
		return;

59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78
	/* Reevalute with xtime_lock held */
	write_seqlock(&xtime_lock);

	delta = ktime_sub(now, last_jiffies_update);
	if (delta.tv64 >= tick_period.tv64) {

		delta = ktime_sub(delta, tick_period);
		last_jiffies_update = ktime_add(last_jiffies_update,
						tick_period);

		/* Slow path for long timeouts */
		if (unlikely(delta.tv64 >= tick_period.tv64)) {
			s64 incr = ktime_to_ns(tick_period);

			ticks = ktime_divns(delta, incr);

			last_jiffies_update = ktime_add_ns(last_jiffies_update,
							   incr * ticks);
		}
		do_timer(++ticks);
79 80 81

		/* Keep the tick_next_period variable up to date */
		tick_next_period = ktime_add(last_jiffies_update, tick_period);
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136
	}
	write_sequnlock(&xtime_lock);
}

/*
 * Initialize and return retrieve the jiffies update.
 */
static ktime_t tick_init_jiffy_update(void)
{
	ktime_t period;

	write_seqlock(&xtime_lock);
	/* Did we start the jiffies update yet ? */
	if (last_jiffies_update.tv64 == 0)
		last_jiffies_update = tick_next_period;
	period = last_jiffies_update;
	write_sequnlock(&xtime_lock);
	return period;
}

/*
 * NOHZ - aka dynamic tick functionality
 */
#ifdef CONFIG_NO_HZ
/*
 * NO HZ enabled ?
 */
static int tick_nohz_enabled __read_mostly  = 1;

/*
 * Enable / Disable tickless mode
 */
static int __init setup_tick_nohz(char *str)
{
	if (!strcmp(str, "off"))
		tick_nohz_enabled = 0;
	else if (!strcmp(str, "on"))
		tick_nohz_enabled = 1;
	else
		return 0;
	return 1;
}

__setup("nohz=", setup_tick_nohz);

/**
 * tick_nohz_update_jiffies - update jiffies when idle was interrupted
 *
 * Called from interrupt entry when the CPU was idle
 *
 * In case the sched_tick was stopped on this CPU, we have to check if jiffies
 * must be updated. Otherwise an interrupt handler could use a stale jiffy
 * value. We do this unconditionally on any cpu, as we don't know whether the
 * cpu, which has the update task assigned is in a long sleep.
 */
137
static void tick_nohz_update_jiffies(ktime_t now)
138 139 140 141 142
{
	int cpu = smp_processor_id();
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
	unsigned long flags;

143
	cpumask_clear_cpu(cpu, nohz_cpu_mask);
144
	ts->idle_waketime = now;
145 146 147 148

	local_irq_save(flags);
	tick_do_update_jiffies64(now);
	local_irq_restore(flags);
149 150

	touch_softlockup_watchdog();
151 152
}

153
static void tick_nohz_stop_idle(int cpu, ktime_t now)
154 155
{
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
156
	ktime_t delta;
157

158 159 160 161
	delta = ktime_sub(now, ts->idle_entrytime);
	ts->idle_lastupdate = now;
	ts->idle_sleeptime = ktime_add(ts->idle_sleeptime, delta);
	ts->idle_active = 0;
162

163
	sched_clock_idle_wakeup_event(0);
164 165
}

166
static ktime_t tick_nohz_start_idle(struct tick_sched *ts)
167 168 169 170 171 172 173 174 175 176 177
{
	ktime_t now, delta;

	now = ktime_get();
	if (ts->idle_active) {
		delta = ktime_sub(now, ts->idle_entrytime);
		ts->idle_lastupdate = now;
		ts->idle_sleeptime = ktime_add(ts->idle_sleeptime, delta);
	}
	ts->idle_entrytime = now;
	ts->idle_active = 1;
178
	sched_clock_idle_sleep_event();
179 180 181 182 183 184 185
	return now;
}

u64 get_cpu_idle_time_us(int cpu, u64 *last_update_time)
{
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);

186 187 188 189 190 191 192 193
	if (!tick_nohz_enabled)
		return -1;

	if (ts->idle_active)
		*last_update_time = ktime_to_us(ts->idle_lastupdate);
	else
		*last_update_time = ktime_to_us(ktime_get());

194 195
	return ktime_to_us(ts->idle_sleeptime);
}
196
EXPORT_SYMBOL_GPL(get_cpu_idle_time_us);
197

198 199 200 201 202 203 204
/**
 * tick_nohz_stop_sched_tick - stop the idle tick from the idle task
 *
 * When the next event is more than a tick into the future, stop the idle tick
 * Called either from the idle loop or from irq_exit() when an idle period was
 * just interrupted by an interrupt which did not cause a reschedule.
 */
205
void tick_nohz_stop_sched_tick(int inidle)
206 207 208
{
	unsigned long seq, last_jiffies, next_jiffies, delta_jiffies, flags;
	struct tick_sched *ts;
209
	ktime_t last_update, expires, now;
210
	struct clock_event_device *dev = __get_cpu_var(tick_cpu_device).evtdev;
211
	u64 time_delta;
212 213 214 215 216 217
	int cpu;

	local_irq_save(flags);

	cpu = smp_processor_id();
	ts = &per_cpu(tick_cpu_sched, cpu);
218 219 220 221 222 223 224 225 226

	/*
	 * Call to tick_nohz_start_idle stops the last_update_time from being
	 * updated. Thus, it must not be called in the event we are called from
	 * irq_exit() with the prior state different than idle.
	 */
	if (!inidle && !ts->inidle)
		goto end;

227 228 229 230 231 232 233
	/*
	 * Set ts->inidle unconditionally. Even if the system did not
	 * switch to NOHZ mode the cpu frequency governers rely on the
	 * update of the idle time accounting in tick_nohz_start_idle().
	 */
	ts->inidle = 1;

234
	now = tick_nohz_start_idle(ts);
235

236 237 238 239 240 241 242 243 244
	/*
	 * If this cpu is offline and it is the one which updates
	 * jiffies, then give up the assignment and let it be taken by
	 * the cpu which runs the tick timer next. If we don't drop
	 * this here the jiffies might be stale and do_timer() never
	 * invoked.
	 */
	if (unlikely(!cpu_online(cpu))) {
		if (cpu == tick_do_timer_cpu)
245
			tick_do_timer_cpu = TICK_DO_TIMER_NONE;
246 247
	}

248 249 250 251 252 253
	if (unlikely(ts->nohz_mode == NOHZ_MODE_INACTIVE))
		goto end;

	if (need_resched())
		goto end;

254
	if (unlikely(local_softirq_pending() && cpu_online(cpu))) {
255 256 257 258
		static int ratelimit;

		if (ratelimit < 10) {
			printk(KERN_ERR "NOHZ: local_softirq_pending %02x\n",
T
Thomas Gleixner 已提交
259
			       (unsigned int) local_softirq_pending());
260 261
			ratelimit++;
		}
262
		goto end;
263
	}
264

M
Mike Galbraith 已提交
265 266 267
	if (nohz_ratelimit(cpu))
		goto end;

268 269 270 271 272 273
	ts->idle_calls++;
	/* Read jiffies and the time when jiffies were updated last */
	do {
		seq = read_seqbegin(&xtime_lock);
		last_update = last_jiffies_update;
		last_jiffies = jiffies;
T
Thomas Gleixner 已提交
274
		time_delta = timekeeping_max_deferment();
275 276
	} while (read_seqretry(&xtime_lock, seq));

277 278 279
	if (rcu_needs_cpu(cpu) || printk_needs_cpu(cpu) ||
	    arch_needs_cpu(cpu)) {
		next_jiffies = last_jiffies + 1;
I
Ingo Molnar 已提交
280
		delta_jiffies = 1;
281 282 283 284 285
	} else {
		/* Get the next timer wheel timer */
		next_jiffies = get_next_timer_interrupt(last_jiffies);
		delta_jiffies = next_jiffies - last_jiffies;
	}
286 287 288 289
	/*
	 * Do not stop the tick, if we are only one off
	 * or if the cpu is required for rcu
	 */
I
Ingo Molnar 已提交
290
	if (!ts->tick_stopped && delta_jiffies == 1)
291 292 293 294 295
		goto out;

	/* Schedule the tick, if we are at least one jiffie off */
	if ((long)delta_jiffies >= 1) {

296 297 298 299 300 301
		/*
		 * If this cpu is the one which updates jiffies, then
		 * give up the assignment and let it be taken by the
		 * cpu which runs the tick timer next, which might be
		 * this cpu as well. If we don't drop this here the
		 * jiffies might be stale and do_timer() never
T
Thomas Gleixner 已提交
302 303 304 305 306 307
		 * invoked. Keep track of the fact that it was the one
		 * which had the do_timer() duty last. If this cpu is
		 * the one which had the do_timer() duty last, we
		 * limit the sleep time to the timekeeping
		 * max_deferement value which we retrieved
		 * above. Otherwise we can sleep as long as we want.
308
		 */
T
Thomas Gleixner 已提交
309
		if (cpu == tick_do_timer_cpu) {
310
			tick_do_timer_cpu = TICK_DO_TIMER_NONE;
T
Thomas Gleixner 已提交
311 312 313 314 315 316 317 318
			ts->do_timer_last = 1;
		} else if (tick_do_timer_cpu != TICK_DO_TIMER_NONE) {
			time_delta = KTIME_MAX;
			ts->do_timer_last = 0;
		} else if (!ts->do_timer_last) {
			time_delta = KTIME_MAX;
		}

319
		/*
320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
		 * calculate the expiry time for the next timer wheel
		 * timer. delta_jiffies >= NEXT_TIMER_MAX_DELTA signals
		 * that there is no timer pending or at least extremely
		 * far into the future (12 days for HZ=1000). In this
		 * case we set the expiry to the end of time.
		 */
		if (likely(delta_jiffies < NEXT_TIMER_MAX_DELTA)) {
			/*
			 * Calculate the time delta for the next timer event.
			 * If the time delta exceeds the maximum time delta
			 * permitted by the current clocksource then adjust
			 * the time delta accordingly to ensure the
			 * clocksource does not wrap.
			 */
			time_delta = min_t(u64, time_delta,
					   tick_period.tv64 * delta_jiffies);
		}
337

T
Thomas Gleixner 已提交
338 339 340 341
		if (time_delta < KTIME_MAX)
			expires = ktime_add_ns(last_update, time_delta);
		else
			expires.tv64 = KTIME_MAX;
342

I
Ingo Molnar 已提交
343
		if (delta_jiffies > 1)
344
			cpumask_set_cpu(cpu, nohz_cpu_mask);
345 346 347 348 349

		/* Skip reprogram of event if its not changed */
		if (ts->tick_stopped && ktime_equal(expires, dev->next_event))
			goto out;

350 351 352 353 354 355 356 357
		/*
		 * nohz_stop_sched_tick can be called several times before
		 * the nohz_restart_sched_tick is called. This happens when
		 * interrupts arrive which do not cause a reschedule. In the
		 * first call we save the current tick time, so we can restart
		 * the scheduler tick in nohz_restart_sched_tick.
		 */
		if (!ts->tick_stopped) {
358 359 360 361
			if (select_nohz_load_balancer(1)) {
				/*
				 * sched tick not stopped!
				 */
362
				cpumask_clear_cpu(cpu, nohz_cpu_mask);
363 364 365
				goto out;
			}

366
			ts->idle_tick = hrtimer_get_expires(&ts->sched_timer);
367 368
			ts->tick_stopped = 1;
			ts->idle_jiffies = last_jiffies;
369
			rcu_enter_nohz();
370
		}
371

372 373
		ts->idle_sleeps++;

374 375 376
		/* Mark expires */
		ts->idle_expires = expires;

377
		/*
378 379
		 * If the expiration time == KTIME_MAX, then
		 * in this case we simply stop the tick timer.
380
		 */
381
		 if (unlikely(expires.tv64 == KTIME_MAX)) {
382 383 384 385 386
			if (ts->nohz_mode == NOHZ_MODE_HIGHRES)
				hrtimer_cancel(&ts->sched_timer);
			goto out;
		}

387 388
		if (ts->nohz_mode == NOHZ_MODE_HIGHRES) {
			hrtimer_start(&ts->sched_timer, expires,
389
				      HRTIMER_MODE_ABS_PINNED);
390 391 392
			/* Check, if the timer was already in the past */
			if (hrtimer_active(&ts->sched_timer))
				goto out;
P
Pavel Machek 已提交
393
		} else if (!tick_program_event(expires, 0))
394 395 396 397 398 399 400
				goto out;
		/*
		 * We are past the event already. So we crossed a
		 * jiffie boundary. Update jiffies and raise the
		 * softirq.
		 */
		tick_do_update_jiffies64(ktime_get());
401
		cpumask_clear_cpu(cpu, nohz_cpu_mask);
402 403 404 405 406
	}
	raise_softirq_irqoff(TIMER_SOFTIRQ);
out:
	ts->next_jiffies = next_jiffies;
	ts->last_jiffies = last_jiffies;
407
	ts->sleep_length = ktime_sub(dev->next_event, now);
408 409 410 411
end:
	local_irq_restore(flags);
}

412 413 414 415 416 417 418 419 420 421 422 423
/**
 * tick_nohz_get_sleep_length - return the length of the current sleep
 *
 * Called from power state control code with interrupts disabled
 */
ktime_t tick_nohz_get_sleep_length(void)
{
	struct tick_sched *ts = &__get_cpu_var(tick_cpu_sched);

	return ts->sleep_length;
}

424 425 426
static void tick_nohz_restart(struct tick_sched *ts, ktime_t now)
{
	hrtimer_cancel(&ts->sched_timer);
427
	hrtimer_set_expires(&ts->sched_timer, ts->idle_tick);
428 429 430 431 432 433

	while (1) {
		/* Forward the time to expire in the future */
		hrtimer_forward(&ts->sched_timer, now, tick_period);

		if (ts->nohz_mode == NOHZ_MODE_HIGHRES) {
434
			hrtimer_start_expires(&ts->sched_timer,
435
					      HRTIMER_MODE_ABS_PINNED);
436 437 438 439
			/* Check, if the timer was already in the past */
			if (hrtimer_active(&ts->sched_timer))
				break;
		} else {
440 441
			if (!tick_program_event(
				hrtimer_get_expires(&ts->sched_timer), 0))
442 443 444 445 446 447 448 449
				break;
		}
		/* Update jiffies and reread time */
		tick_do_update_jiffies64(now);
		now = ktime_get();
	}
}

450
/**
451
 * tick_nohz_restart_sched_tick - restart the idle tick from the idle task
452 453 454 455 456 457 458
 *
 * Restart the idle tick when the CPU is woken up from idle
 */
void tick_nohz_restart_sched_tick(void)
{
	int cpu = smp_processor_id();
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
459
#ifndef CONFIG_VIRT_CPU_ACCOUNTING
460
	unsigned long ticks;
461
#endif
462
	ktime_t now;
463

464
	local_irq_disable();
465 466 467 468 469
	if (ts->idle_active || (ts->inidle && ts->tick_stopped))
		now = ktime_get();

	if (ts->idle_active)
		tick_nohz_stop_idle(cpu, now);
470

471 472
	if (!ts->inidle || !ts->tick_stopped) {
		ts->inidle = 0;
473
		local_irq_enable();
474
		return;
475
	}
476

477 478
	ts->inidle = 0;

479 480
	rcu_exit_nohz();

481
	/* Update jiffies first */
482
	select_nohz_load_balancer(0);
483
	tick_do_update_jiffies64(now);
484
	cpumask_clear_cpu(cpu, nohz_cpu_mask);
485

486
#ifndef CONFIG_VIRT_CPU_ACCOUNTING
487 488 489 490 491 492 493 494 495
	/*
	 * We stopped the tick in idle. Update process times would miss the
	 * time we slept as update_process_times does only a 1 tick
	 * accounting. Enforce that this is accounted to idle !
	 */
	ticks = jiffies - ts->idle_jiffies;
	/*
	 * We might be one off. Do not randomly account a huge number of ticks!
	 */
496 497 498
	if (ticks && ticks < LONG_MAX)
		account_idle_ticks(ticks);
#endif
499

I
Ingo Molnar 已提交
500
	touch_softlockup_watchdog();
501 502 503 504
	/*
	 * Cancel the scheduled timer and restore the tick
	 */
	ts->tick_stopped  = 0;
505
	ts->idle_exittime = now;
506

507
	tick_nohz_restart(ts, now);
508 509 510 511 512 513 514

	local_irq_enable();
}

static int tick_nohz_reprogram(struct tick_sched *ts, ktime_t now)
{
	hrtimer_forward(&ts->sched_timer, now, tick_period);
515
	return tick_program_event(hrtimer_get_expires(&ts->sched_timer), 0);
516 517 518 519 520 521 522 523 524
}

/*
 * The nohz low res interrupt handler
 */
static void tick_nohz_handler(struct clock_event_device *dev)
{
	struct tick_sched *ts = &__get_cpu_var(tick_cpu_sched);
	struct pt_regs *regs = get_irq_regs();
525
	int cpu = smp_processor_id();
526 527 528 529
	ktime_t now = ktime_get();

	dev->next_event.tv64 = KTIME_MAX;

530 531 532 533 534 535 536
	/*
	 * Check if the do_timer duty was dropped. We don't care about
	 * concurrency: This happens only when the cpu in charge went
	 * into a long sleep. If two cpus happen to assign themself to
	 * this duty, then the jiffies update is still serialized by
	 * xtime_lock.
	 */
537
	if (unlikely(tick_do_timer_cpu == TICK_DO_TIMER_NONE))
538 539
		tick_do_timer_cpu = cpu;

540
	/* Check, if the jiffies need an update */
541 542
	if (tick_do_timer_cpu == cpu)
		tick_do_update_jiffies64(now);
543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593

	/*
	 * When we are idle and the tick is stopped, we have to touch
	 * the watchdog as we might not schedule for a really long
	 * time. This happens on complete idle SMP systems while
	 * waiting on the login prompt. We also increment the "start
	 * of idle" jiffy stamp so the idle accounting adjustment we
	 * do when we go busy again does not account too much ticks.
	 */
	if (ts->tick_stopped) {
		touch_softlockup_watchdog();
		ts->idle_jiffies++;
	}

	update_process_times(user_mode(regs));
	profile_tick(CPU_PROFILING);

	while (tick_nohz_reprogram(ts, now)) {
		now = ktime_get();
		tick_do_update_jiffies64(now);
	}
}

/**
 * tick_nohz_switch_to_nohz - switch to nohz mode
 */
static void tick_nohz_switch_to_nohz(void)
{
	struct tick_sched *ts = &__get_cpu_var(tick_cpu_sched);
	ktime_t next;

	if (!tick_nohz_enabled)
		return;

	local_irq_disable();
	if (tick_switch_to_oneshot(tick_nohz_handler)) {
		local_irq_enable();
		return;
	}

	ts->nohz_mode = NOHZ_MODE_LOWRES;

	/*
	 * Recycle the hrtimer in ts, so we can share the
	 * hrtimer_forward with the highres code.
	 */
	hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
	/* Get the next period */
	next = tick_init_jiffy_update();

	for (;;) {
594
		hrtimer_set_expires(&ts->sched_timer, next);
595 596 597 598 599 600 601 602 603 604
		if (!tick_program_event(next, 0))
			break;
		next = ktime_add(next, tick_period);
	}
	local_irq_enable();

	printk(KERN_INFO "Switched to NOHz mode on CPU #%d\n",
	       smp_processor_id());
}

605 606 607 608 609 610 611 612 613 614 615
/*
 * When NOHZ is enabled and the tick is stopped, we need to kick the
 * tick timer from irq_enter() so that the jiffies update is kept
 * alive during long running softirqs. That's ugly as hell, but
 * correctness is key even if we need to fix the offending softirq in
 * the first place.
 *
 * Note, this is different to tick_nohz_restart. We just kick the
 * timer and do not touch the other magic bits which need to be done
 * when idle is left.
 */
616
static void tick_nohz_kick_tick(int cpu, ktime_t now)
617
{
618 619 620
#if 0
	/* Switch back to 2.6.27 behaviour */

621
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
622
	ktime_t delta;
623

624 625 626 627
	/*
	 * Do not touch the tick device, when the next expiry is either
	 * already reached or less/equal than the tick period.
	 */
628
	delta =	ktime_sub(hrtimer_get_expires(&ts->sched_timer), now);
629 630 631 632
	if (delta.tv64 <= tick_period.tv64)
		return;

	tick_nohz_restart(ts, now);
633
#endif
634 635
}

636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
static inline void tick_check_nohz(int cpu)
{
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);
	ktime_t now;

	if (!ts->idle_active && !ts->tick_stopped)
		return;
	now = ktime_get();
	if (ts->idle_active)
		tick_nohz_stop_idle(cpu, now);
	if (ts->tick_stopped) {
		tick_nohz_update_jiffies(now);
		tick_nohz_kick_tick(cpu, now);
	}
}

652 653 654
#else

static inline void tick_nohz_switch_to_nohz(void) { }
655
static inline void tick_check_nohz(int cpu) { }
656 657 658

#endif /* NO_HZ */

659 660 661 662 663
/*
 * Called from irq_enter to notify about the possible interruption of idle()
 */
void tick_check_idle(int cpu)
{
664
	tick_check_oneshot_broadcast(cpu);
665
	tick_check_nohz(cpu);
666 667
}

668 669 670 671 672
/*
 * High resolution timer specific code
 */
#ifdef CONFIG_HIGH_RES_TIMERS
/*
P
Pavel Machek 已提交
673
 * We rearm the timer until we get disabled by the idle code.
674 675 676 677 678 679 680 681
 * Called with interrupts disabled and timer->base->cpu_base->lock held.
 */
static enum hrtimer_restart tick_sched_timer(struct hrtimer *timer)
{
	struct tick_sched *ts =
		container_of(timer, struct tick_sched, sched_timer);
	struct pt_regs *regs = get_irq_regs();
	ktime_t now = ktime_get();
682 683 684 685 686 687 688 689 690 691
	int cpu = smp_processor_id();

#ifdef CONFIG_NO_HZ
	/*
	 * Check if the do_timer duty was dropped. We don't care about
	 * concurrency: This happens only when the cpu in charge went
	 * into a long sleep. If two cpus happen to assign themself to
	 * this duty, then the jiffies update is still serialized by
	 * xtime_lock.
	 */
692
	if (unlikely(tick_do_timer_cpu == TICK_DO_TIMER_NONE))
693 694
		tick_do_timer_cpu = cpu;
#endif
695 696

	/* Check, if the jiffies need an update */
697 698
	if (tick_do_timer_cpu == cpu)
		tick_do_update_jiffies64(now);
699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732

	/*
	 * Do not call, when we are not in irq context and have
	 * no valid regs pointer
	 */
	if (regs) {
		/*
		 * When we are idle and the tick is stopped, we have to touch
		 * the watchdog as we might not schedule for a really long
		 * time. This happens on complete idle SMP systems while
		 * waiting on the login prompt. We also increment the "start of
		 * idle" jiffy stamp so the idle accounting adjustment we do
		 * when we go busy again does not account too much ticks.
		 */
		if (ts->tick_stopped) {
			touch_softlockup_watchdog();
			ts->idle_jiffies++;
		}
		update_process_times(user_mode(regs));
		profile_tick(CPU_PROFILING);
	}

	hrtimer_forward(timer, now, tick_period);

	return HRTIMER_RESTART;
}

/**
 * tick_setup_sched_timer - setup the tick emulation timer
 */
void tick_setup_sched_timer(void)
{
	struct tick_sched *ts = &__get_cpu_var(tick_cpu_sched);
	ktime_t now = ktime_get();
733
	u64 offset;
734 735 736 737 738 739 740

	/*
	 * Emulate tick processing via per-CPU hrtimers:
	 */
	hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
	ts->sched_timer.function = tick_sched_timer;

741
	/* Get the next period (per cpu) */
742
	hrtimer_set_expires(&ts->sched_timer, tick_init_jiffy_update());
743
	offset = ktime_to_ns(tick_period) >> 1;
744
	do_div(offset, num_possible_cpus());
745
	offset *= smp_processor_id();
746
	hrtimer_add_expires_ns(&ts->sched_timer, offset);
747 748 749

	for (;;) {
		hrtimer_forward(&ts->sched_timer, now, tick_period);
750 751
		hrtimer_start_expires(&ts->sched_timer,
				      HRTIMER_MODE_ABS_PINNED);
752 753 754 755 756 757 758 759 760 761 762
		/* Check, if the timer was already in the past */
		if (hrtimer_active(&ts->sched_timer))
			break;
		now = ktime_get();
	}

#ifdef CONFIG_NO_HZ
	if (tick_nohz_enabled)
		ts->nohz_mode = NOHZ_MODE_HIGHRES;
#endif
}
763
#endif /* HIGH_RES_TIMERS */
764

765
#if defined CONFIG_NO_HZ || defined CONFIG_HIGH_RES_TIMERS
766 767 768 769
void tick_cancel_sched_timer(int cpu)
{
	struct tick_sched *ts = &per_cpu(tick_cpu_sched, cpu);

770
# ifdef CONFIG_HIGH_RES_TIMERS
771 772
	if (ts->sched_timer.base)
		hrtimer_cancel(&ts->sched_timer);
773
# endif
774

775 776
	ts->nohz_mode = NOHZ_MODE_INACTIVE;
}
777
#endif
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817

/**
 * Async notification about clocksource changes
 */
void tick_clock_notify(void)
{
	int cpu;

	for_each_possible_cpu(cpu)
		set_bit(0, &per_cpu(tick_cpu_sched, cpu).check_clocks);
}

/*
 * Async notification about clock event changes
 */
void tick_oneshot_notify(void)
{
	struct tick_sched *ts = &__get_cpu_var(tick_cpu_sched);

	set_bit(0, &ts->check_clocks);
}

/**
 * Check, if a change happened, which makes oneshot possible.
 *
 * Called cyclic from the hrtimer softirq (driven by the timer
 * softirq) allow_nohz signals, that we can switch into low-res nohz
 * mode, because high resolution timers are disabled (either compile
 * or runtime).
 */
int tick_check_oneshot_change(int allow_nohz)
{
	struct tick_sched *ts = &__get_cpu_var(tick_cpu_sched);

	if (!test_and_clear_bit(0, &ts->check_clocks))
		return 0;

	if (ts->nohz_mode != NOHZ_MODE_INACTIVE)
		return 0;

818
	if (!timekeeping_valid_for_hres() || !tick_is_oneshot_available())
819 820 821 822 823 824 825 826
		return 0;

	if (!allow_nohz)
		return 1;

	tick_nohz_switch_to_nohz();
	return 0;
}