timekeeping.c 45.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
/*
 *  linux/kernel/time/timekeeping.c
 *
 *  Kernel timekeeping code and accessor functions
 *
 *  This code was moved from linux/kernel/timer.c.
 *  Please see that file for copyright and history logs.
 *
 */

11
#include <linux/timekeeper_internal.h>
12 13 14 15 16
#include <linux/module.h>
#include <linux/interrupt.h>
#include <linux/percpu.h>
#include <linux/init.h>
#include <linux/mm.h>
17
#include <linux/sched.h>
18
#include <linux/syscore_ops.h>
19 20 21 22
#include <linux/clocksource.h>
#include <linux/jiffies.h>
#include <linux/time.h>
#include <linux/tick.h>
23
#include <linux/stop_machine.h>
24
#include <linux/pvclock_gtod.h>
25
#include <linux/compiler.h>
26

27
#include "tick-internal.h"
28
#include "ntp_internal.h"
29
#include "timekeeping_internal.h"
30

31 32
#define TK_CLEAR_NTP		(1 << 0)
#define TK_MIRROR		(1 << 1)
33
#define TK_CLOCK_WAS_SET	(1 << 2)
34

35
static struct timekeeper timekeeper;
36 37
static DEFINE_RAW_SPINLOCK(timekeeper_lock);
static seqcount_t timekeeper_seq;
38
static struct timekeeper shadow_timekeeper;
39

40 41 42
/* flag for if timekeeping is suspended */
int __read_mostly timekeeping_suspended;

43 44 45
/* Flag for if there is a persistent clock on this platform */
bool __read_mostly persistent_clock_exist = false;

46 47 48 49 50 51 52 53 54 55 56
static inline void tk_normalize_xtime(struct timekeeper *tk)
{
	while (tk->xtime_nsec >= ((u64)NSEC_PER_SEC << tk->shift)) {
		tk->xtime_nsec -= (u64)NSEC_PER_SEC << tk->shift;
		tk->xtime_sec++;
	}
}

static void tk_set_xtime(struct timekeeper *tk, const struct timespec *ts)
{
	tk->xtime_sec = ts->tv_sec;
57
	tk->xtime_nsec = (u64)ts->tv_nsec << tk->shift;
58 59 60 61 62
}

static void tk_xtime_add(struct timekeeper *tk, const struct timespec *ts)
{
	tk->xtime_sec += ts->tv_sec;
63
	tk->xtime_nsec += (u64)ts->tv_nsec << tk->shift;
64
	tk_normalize_xtime(tk);
65
}
66

67 68 69 70 71 72 73 74 75 76 77 78 79 80
static void tk_set_wall_to_mono(struct timekeeper *tk, struct timespec wtm)
{
	struct timespec tmp;

	/*
	 * Verify consistency of: offset_real = -wall_to_monotonic
	 * before modifying anything
	 */
	set_normalized_timespec(&tmp, -tk->wall_to_monotonic.tv_sec,
					-tk->wall_to_monotonic.tv_nsec);
	WARN_ON_ONCE(tk->offs_real.tv64 != timespec_to_ktime(tmp).tv64);
	tk->wall_to_monotonic = wtm;
	set_normalized_timespec(&tmp, -wtm.tv_sec, -wtm.tv_nsec);
	tk->offs_real = timespec_to_ktime(tmp);
81
	tk->offs_tai = ktime_add(tk->offs_real, ktime_set(tk->tai_offset, 0));
82 83 84 85 86 87 88 89 90 91 92
}

static void tk_set_sleep_time(struct timekeeper *tk, struct timespec t)
{
	/* Verify consistency before modifying */
	WARN_ON_ONCE(tk->offs_boot.tv64 != timespec_to_ktime(tk->total_sleep_time).tv64);

	tk->total_sleep_time	= t;
	tk->offs_boot		= timespec_to_ktime(t);
}

93
/**
94
 * tk_setup_internals - Set up internals to use clocksource clock.
95
 *
96
 * @tk:		The target timekeeper to setup.
97 98 99 100 101 102 103
 * @clock:		Pointer to clocksource.
 *
 * Calculates a fixed cycle/nsec interval for a given clocksource/adjustment
 * pair and interval request.
 *
 * Unless you're the timekeeping code, you should not be using this!
 */
104
static void tk_setup_internals(struct timekeeper *tk, struct clocksource *clock)
105 106
{
	cycle_t interval;
107
	u64 tmp, ntpinterval;
108
	struct clocksource *old_clock;
109

110 111
	old_clock = tk->clock;
	tk->clock = clock;
112
	tk->cycle_last = clock->cycle_last = clock->read(clock);
113 114 115 116

	/* Do the ns -> cycle conversion first, using original mult */
	tmp = NTP_INTERVAL_LENGTH;
	tmp <<= clock->shift;
117
	ntpinterval = tmp;
118 119
	tmp += clock->mult/2;
	do_div(tmp, clock->mult);
120 121 122 123
	if (tmp == 0)
		tmp = 1;

	interval = (cycle_t) tmp;
124
	tk->cycle_interval = interval;
125 126

	/* Go back from cycles -> shifted ns */
127 128 129
	tk->xtime_interval = (u64) interval * clock->mult;
	tk->xtime_remainder = ntpinterval - tk->xtime_interval;
	tk->raw_interval =
130
		((u64) interval * clock->mult) >> clock->shift;
131

132 133 134 135
	 /* if changing clocks, convert xtime_nsec shift units */
	if (old_clock) {
		int shift_change = clock->shift - old_clock->shift;
		if (shift_change < 0)
136
			tk->xtime_nsec >>= -shift_change;
137
		else
138
			tk->xtime_nsec <<= shift_change;
139
	}
140
	tk->shift = clock->shift;
141

142 143
	tk->ntp_error = 0;
	tk->ntp_error_shift = NTP_SCALE_SHIFT - clock->shift;
144 145 146 147 148 149

	/*
	 * The timekeeper keeps its own mult values for the currently
	 * active clocksource. These value will be adjusted via NTP
	 * to counteract clock drifting.
	 */
150
	tk->mult = clock->mult;
151
}
152

153
/* Timekeeper helper functions. */
154 155

#ifdef CONFIG_ARCH_USES_GETTIMEOFFSET
156 157
static u32 default_arch_gettimeoffset(void) { return 0; }
u32 (*arch_gettimeoffset)(void) = default_arch_gettimeoffset;
158
#else
159
static inline u32 arch_gettimeoffset(void) { return 0; }
160 161
#endif

162
static inline s64 timekeeping_get_ns(struct timekeeper *tk)
163 164 165
{
	cycle_t cycle_now, cycle_delta;
	struct clocksource *clock;
166
	s64 nsec;
167 168

	/* read clocksource: */
169
	clock = tk->clock;
170 171 172 173 174
	cycle_now = clock->read(clock);

	/* calculate the delta since the last update_wall_time: */
	cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;

175 176
	nsec = cycle_delta * tk->mult + tk->xtime_nsec;
	nsec >>= tk->shift;
177

178
	/* If arch requires, add in get_arch_timeoffset() */
179
	return nsec + arch_gettimeoffset();
180 181
}

182
static inline s64 timekeeping_get_ns_raw(struct timekeeper *tk)
183 184 185
{
	cycle_t cycle_now, cycle_delta;
	struct clocksource *clock;
186
	s64 nsec;
187 188

	/* read clocksource: */
189
	clock = tk->clock;
190 191 192 193 194
	cycle_now = clock->read(clock);

	/* calculate the delta since the last update_wall_time: */
	cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;

195 196 197
	/* convert delta to nanoseconds. */
	nsec = clocksource_cyc2ns(cycle_delta, clock->mult, clock->shift);

198
	/* If arch requires, add in get_arch_timeoffset() */
199
	return nsec + arch_gettimeoffset();
200 201
}

202 203
static RAW_NOTIFIER_HEAD(pvclock_gtod_chain);

204
static void update_pvclock_gtod(struct timekeeper *tk, bool was_set)
205
{
206
	raw_notifier_call_chain(&pvclock_gtod_chain, was_set, tk);
207 208 209 210 211 212 213 214 215 216 217
}

/**
 * pvclock_gtod_register_notifier - register a pvclock timedata update listener
 */
int pvclock_gtod_register_notifier(struct notifier_block *nb)
{
	struct timekeeper *tk = &timekeeper;
	unsigned long flags;
	int ret;

218
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
219
	ret = raw_notifier_chain_register(&pvclock_gtod_chain, nb);
220
	update_pvclock_gtod(tk, true);
221
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
222 223 224 225 226 227 228 229 230 231 232 233 234 235

	return ret;
}
EXPORT_SYMBOL_GPL(pvclock_gtod_register_notifier);

/**
 * pvclock_gtod_unregister_notifier - unregister a pvclock
 * timedata update listener
 */
int pvclock_gtod_unregister_notifier(struct notifier_block *nb)
{
	unsigned long flags;
	int ret;

236
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
237
	ret = raw_notifier_chain_unregister(&pvclock_gtod_chain, nb);
238
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
239 240 241 242 243

	return ret;
}
EXPORT_SYMBOL_GPL(pvclock_gtod_unregister_notifier);

244
/* must hold timekeeper_lock */
245
static void timekeeping_update(struct timekeeper *tk, unsigned int action)
246
{
247
	if (action & TK_CLEAR_NTP) {
248
		tk->ntp_error = 0;
249 250
		ntp_clear();
	}
251
	update_vsyscall(tk);
252
	update_pvclock_gtod(tk, action & TK_CLOCK_WAS_SET);
253

254
	if (action & TK_MIRROR)
255
		memcpy(&shadow_timekeeper, &timekeeper, sizeof(timekeeper));
256 257
}

258
/**
259
 * timekeeping_forward_now - update clock to the current time
260
 *
261 262 263
 * Forward the current clock to update its state since the last call to
 * update_wall_time(). This is useful before significant clock changes,
 * as it avoids having to deal with this time offset explicitly.
264
 */
265
static void timekeeping_forward_now(struct timekeeper *tk)
266 267
{
	cycle_t cycle_now, cycle_delta;
268
	struct clocksource *clock;
269
	s64 nsec;
270

271
	clock = tk->clock;
272
	cycle_now = clock->read(clock);
273
	cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;
274
	tk->cycle_last = clock->cycle_last = cycle_now;
275

276
	tk->xtime_nsec += cycle_delta * tk->mult;
277

278
	/* If arch requires, add in get_arch_timeoffset() */
279
	tk->xtime_nsec += (u64)arch_gettimeoffset() << tk->shift;
280

281
	tk_normalize_xtime(tk);
282

283
	nsec = clocksource_cyc2ns(cycle_delta, clock->mult, clock->shift);
284
	timespec_add_ns(&tk->raw_time, nsec);
285 286 287
}

/**
288
 * __getnstimeofday - Returns the time of day in a timespec.
289 290
 * @ts:		pointer to the timespec to be set
 *
291 292
 * Updates the time of day in the timespec.
 * Returns 0 on success, or -ve when suspended (timespec will be undefined).
293
 */
294
int __getnstimeofday(struct timespec *ts)
295
{
296
	struct timekeeper *tk = &timekeeper;
297
	unsigned long seq;
298
	s64 nsecs = 0;
299 300

	do {
301
		seq = read_seqcount_begin(&timekeeper_seq);
302

303
		ts->tv_sec = tk->xtime_sec;
304
		nsecs = timekeeping_get_ns(tk);
305

306
	} while (read_seqcount_retry(&timekeeper_seq, seq));
307

308
	ts->tv_nsec = 0;
309
	timespec_add_ns(ts, nsecs);
310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329

	/*
	 * Do not bail out early, in case there were callers still using
	 * the value, even in the face of the WARN_ON.
	 */
	if (unlikely(timekeeping_suspended))
		return -EAGAIN;
	return 0;
}
EXPORT_SYMBOL(__getnstimeofday);

/**
 * getnstimeofday - Returns the time of day in a timespec.
 * @ts:		pointer to the timespec to be set
 *
 * Returns the time of day in a timespec (WARN if suspended).
 */
void getnstimeofday(struct timespec *ts)
{
	WARN_ON(__getnstimeofday(ts));
330 331 332
}
EXPORT_SYMBOL(getnstimeofday);

333 334
ktime_t ktime_get(void)
{
335
	struct timekeeper *tk = &timekeeper;
336 337 338 339 340 341
	unsigned int seq;
	s64 secs, nsecs;

	WARN_ON(timekeeping_suspended);

	do {
342
		seq = read_seqcount_begin(&timekeeper_seq);
343 344
		secs = tk->xtime_sec + tk->wall_to_monotonic.tv_sec;
		nsecs = timekeeping_get_ns(tk) + tk->wall_to_monotonic.tv_nsec;
345

346
	} while (read_seqcount_retry(&timekeeper_seq, seq));
347 348

	return ktime_set(secs, nsecs);
349 350 351 352 353 354 355 356 357 358 359 360 361
}
EXPORT_SYMBOL_GPL(ktime_get);

/**
 * ktime_get_ts - get the monotonic clock in timespec format
 * @ts:		pointer to timespec variable
 *
 * The function calculates the monotonic clock from the realtime
 * clock and the wall_to_monotonic offset and stores the result
 * in normalized timespec format in the variable pointed to by @ts.
 */
void ktime_get_ts(struct timespec *ts)
{
362
	struct timekeeper *tk = &timekeeper;
363
	struct timespec tomono;
364
	s64 nsec;
365 366 367 368 369
	unsigned int seq;

	WARN_ON(timekeeping_suspended);

	do {
370
		seq = read_seqcount_begin(&timekeeper_seq);
371
		ts->tv_sec = tk->xtime_sec;
372
		nsec = timekeeping_get_ns(tk);
373
		tomono = tk->wall_to_monotonic;
374

375
	} while (read_seqcount_retry(&timekeeper_seq, seq));
376

377 378 379
	ts->tv_sec += tomono.tv_sec;
	ts->tv_nsec = 0;
	timespec_add_ns(ts, nsec + tomono.tv_nsec);
380 381 382
}
EXPORT_SYMBOL_GPL(ktime_get_ts);

J
John Stultz 已提交
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398

/**
 * timekeeping_clocktai - Returns the TAI time of day in a timespec
 * @ts:		pointer to the timespec to be set
 *
 * Returns the time of day in a timespec.
 */
void timekeeping_clocktai(struct timespec *ts)
{
	struct timekeeper *tk = &timekeeper;
	unsigned long seq;
	u64 nsecs;

	WARN_ON(timekeeping_suspended);

	do {
399
		seq = read_seqcount_begin(&timekeeper_seq);
J
John Stultz 已提交
400 401 402 403

		ts->tv_sec = tk->xtime_sec + tk->tai_offset;
		nsecs = timekeeping_get_ns(tk);

404
	} while (read_seqcount_retry(&timekeeper_seq, seq));
J
John Stultz 已提交
405 406 407 408 409 410 411 412

	ts->tv_nsec = 0;
	timespec_add_ns(ts, nsecs);

}
EXPORT_SYMBOL(timekeeping_clocktai);


413 414 415 416 417 418 419 420 421 422 423 424 425 426
/**
 * ktime_get_clocktai - Returns the TAI time of day in a ktime
 *
 * Returns the time of day in a ktime.
 */
ktime_t ktime_get_clocktai(void)
{
	struct timespec ts;

	timekeeping_clocktai(&ts);
	return timespec_to_ktime(ts);
}
EXPORT_SYMBOL(ktime_get_clocktai);

427 428 429 430 431 432 433 434 435 436 437 438 439
#ifdef CONFIG_NTP_PPS

/**
 * getnstime_raw_and_real - get day and raw monotonic time in timespec format
 * @ts_raw:	pointer to the timespec to be set to raw monotonic time
 * @ts_real:	pointer to the timespec to be set to the time of day
 *
 * This function reads both the time of day and raw monotonic time at the
 * same time atomically and stores the resulting timestamps in timespec
 * format.
 */
void getnstime_raw_and_real(struct timespec *ts_raw, struct timespec *ts_real)
{
440
	struct timekeeper *tk = &timekeeper;
441 442 443 444 445 446
	unsigned long seq;
	s64 nsecs_raw, nsecs_real;

	WARN_ON_ONCE(timekeeping_suspended);

	do {
447
		seq = read_seqcount_begin(&timekeeper_seq);
448

449 450
		*ts_raw = tk->raw_time;
		ts_real->tv_sec = tk->xtime_sec;
451
		ts_real->tv_nsec = 0;
452

453 454
		nsecs_raw = timekeeping_get_ns_raw(tk);
		nsecs_real = timekeeping_get_ns(tk);
455

456
	} while (read_seqcount_retry(&timekeeper_seq, seq));
457 458 459 460 461 462 463 464

	timespec_add_ns(ts_raw, nsecs_raw);
	timespec_add_ns(ts_real, nsecs_real);
}
EXPORT_SYMBOL(getnstime_raw_and_real);

#endif /* CONFIG_NTP_PPS */

465 466 467 468
/**
 * do_gettimeofday - Returns the time of day in a timeval
 * @tv:		pointer to the timeval to be set
 *
469
 * NOTE: Users should be converted to using getnstimeofday()
470 471 472 473 474
 */
void do_gettimeofday(struct timeval *tv)
{
	struct timespec now;

475
	getnstimeofday(&now);
476 477 478 479
	tv->tv_sec = now.tv_sec;
	tv->tv_usec = now.tv_nsec/1000;
}
EXPORT_SYMBOL(do_gettimeofday);
480

481 482 483 484 485 486
/**
 * do_settimeofday - Sets the time of day
 * @tv:		pointer to the timespec variable containing the new time
 *
 * Sets the time of day to the new time and update NTP and notify hrtimers
 */
487
int do_settimeofday(const struct timespec *tv)
488
{
489
	struct timekeeper *tk = &timekeeper;
490
	struct timespec ts_delta, xt;
491
	unsigned long flags;
492

493
	if (!timespec_valid_strict(tv))
494 495
		return -EINVAL;

496 497
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
498

499
	timekeeping_forward_now(tk);
500

501
	xt = tk_xtime(tk);
502 503 504
	ts_delta.tv_sec = tv->tv_sec - xt.tv_sec;
	ts_delta.tv_nsec = tv->tv_nsec - xt.tv_nsec;

505
	tk_set_wall_to_mono(tk, timespec_sub(tk->wall_to_monotonic, ts_delta));
506

507
	tk_set_xtime(tk, tv);
508

509
	timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET);
510

511 512
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
513 514 515 516 517 518 519 520

	/* signal hrtimers about time change */
	clock_was_set();

	return 0;
}
EXPORT_SYMBOL(do_settimeofday);

521 522 523 524 525 526 527 528
/**
 * timekeeping_inject_offset - Adds or subtracts from the current time.
 * @tv:		pointer to the timespec variable containing the offset
 *
 * Adds or subtracts an offset value from the current time.
 */
int timekeeping_inject_offset(struct timespec *ts)
{
529
	struct timekeeper *tk = &timekeeper;
530
	unsigned long flags;
531 532
	struct timespec tmp;
	int ret = 0;
533 534 535 536

	if ((unsigned long)ts->tv_nsec >= NSEC_PER_SEC)
		return -EINVAL;

537 538
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
539

540
	timekeeping_forward_now(tk);
541

542 543
	/* Make sure the proposed value is valid */
	tmp = timespec_add(tk_xtime(tk),  *ts);
544
	if (!timespec_valid_strict(&tmp)) {
545 546 547
		ret = -EINVAL;
		goto error;
	}
548

549 550
	tk_xtime_add(tk, ts);
	tk_set_wall_to_mono(tk, timespec_sub(tk->wall_to_monotonic, *ts));
551

552
error: /* even if we error out, we forwarded the time, so call update */
553
	timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET);
554

555 556
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
557 558 559 560

	/* signal hrtimers about time change */
	clock_was_set();

561
	return ret;
562 563 564
}
EXPORT_SYMBOL(timekeeping_inject_offset);

565 566 567 568 569 570 571 572 573 574 575 576

/**
 * timekeeping_get_tai_offset - Returns current TAI offset from UTC
 *
 */
s32 timekeeping_get_tai_offset(void)
{
	struct timekeeper *tk = &timekeeper;
	unsigned int seq;
	s32 ret;

	do {
577
		seq = read_seqcount_begin(&timekeeper_seq);
578
		ret = tk->tai_offset;
579
	} while (read_seqcount_retry(&timekeeper_seq, seq));
580 581 582 583 584 585 586 587

	return ret;
}

/**
 * __timekeeping_set_tai_offset - Lock free worker function
 *
 */
588
static void __timekeeping_set_tai_offset(struct timekeeper *tk, s32 tai_offset)
589 590
{
	tk->tai_offset = tai_offset;
591
	tk->offs_tai = ktime_add(tk->offs_real, ktime_set(tai_offset, 0));
592 593 594 595 596 597 598 599 600 601 602
}

/**
 * timekeeping_set_tai_offset - Sets the current TAI offset from UTC
 *
 */
void timekeeping_set_tai_offset(s32 tai_offset)
{
	struct timekeeper *tk = &timekeeper;
	unsigned long flags;

603 604
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
605
	__timekeeping_set_tai_offset(tk, tai_offset);
606
	timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET);
607 608
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
609
	clock_was_set();
610 611
}

612 613 614 615 616
/**
 * change_clocksource - Swaps clocksources if a new one is available
 *
 * Accumulates current time interval and initializes new clocksource
 */
617
static int change_clocksource(void *data)
618
{
619
	struct timekeeper *tk = &timekeeper;
620
	struct clocksource *new, *old;
621
	unsigned long flags;
622

623
	new = (struct clocksource *) data;
624

625 626
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
627

628
	timekeeping_forward_now(tk);
629 630 631 632 633 634 635 636 637 638 639 640 641 642
	/*
	 * If the cs is in module, get a module reference. Succeeds
	 * for built-in code (owner == NULL) as well.
	 */
	if (try_module_get(new->owner)) {
		if (!new->enable || new->enable(new) == 0) {
			old = tk->clock;
			tk_setup_internals(tk, new);
			if (old->disable)
				old->disable(old);
			module_put(old->owner);
		} else {
			module_put(new->owner);
		}
643
	}
644
	timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET);
645

646 647
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
648

649 650
	return 0;
}
651

652 653 654 655 656 657 658
/**
 * timekeeping_notify - Install a new clock source
 * @clock:		pointer to the clock source
 *
 * This function is called from clocksource.c after a new, better clock
 * source has been registered. The caller holds the clocksource_mutex.
 */
659
int timekeeping_notify(struct clocksource *clock)
660
{
661 662 663
	struct timekeeper *tk = &timekeeper;

	if (tk->clock == clock)
664
		return 0;
665
	stop_machine(change_clocksource, clock, NULL);
666
	tick_clock_notify();
667
	return tk->clock == clock ? 0 : -1;
668
}
669

670 671 672 673 674 675 676 677 678 679 680 681 682 683
/**
 * ktime_get_real - get the real (wall-) time in ktime_t format
 *
 * returns the time in ktime_t format
 */
ktime_t ktime_get_real(void)
{
	struct timespec now;

	getnstimeofday(&now);

	return timespec_to_ktime(now);
}
EXPORT_SYMBOL_GPL(ktime_get_real);
684

685 686 687 688 689 690 691 692
/**
 * getrawmonotonic - Returns the raw monotonic time in a timespec
 * @ts:		pointer to the timespec to be set
 *
 * Returns the raw monotonic time (completely un-modified by ntp)
 */
void getrawmonotonic(struct timespec *ts)
{
693
	struct timekeeper *tk = &timekeeper;
694 695 696 697
	unsigned long seq;
	s64 nsecs;

	do {
698
		seq = read_seqcount_begin(&timekeeper_seq);
699 700
		nsecs = timekeeping_get_ns_raw(tk);
		*ts = tk->raw_time;
701

702
	} while (read_seqcount_retry(&timekeeper_seq, seq));
703 704 705 706 707

	timespec_add_ns(ts, nsecs);
}
EXPORT_SYMBOL(getrawmonotonic);

708
/**
709
 * timekeeping_valid_for_hres - Check if timekeeping is suitable for hres
710
 */
711
int timekeeping_valid_for_hres(void)
712
{
713
	struct timekeeper *tk = &timekeeper;
714 715 716 717
	unsigned long seq;
	int ret;

	do {
718
		seq = read_seqcount_begin(&timekeeper_seq);
719

720
		ret = tk->clock->flags & CLOCK_SOURCE_VALID_FOR_HRES;
721

722
	} while (read_seqcount_retry(&timekeeper_seq, seq));
723 724 725 726

	return ret;
}

727 728 729 730 731
/**
 * timekeeping_max_deferment - Returns max time the clocksource can be deferred
 */
u64 timekeeping_max_deferment(void)
{
732
	struct timekeeper *tk = &timekeeper;
J
John Stultz 已提交
733 734
	unsigned long seq;
	u64 ret;
735

J
John Stultz 已提交
736
	do {
737
		seq = read_seqcount_begin(&timekeeper_seq);
J
John Stultz 已提交
738

739
		ret = tk->clock->max_idle_ns;
J
John Stultz 已提交
740

741
	} while (read_seqcount_retry(&timekeeper_seq, seq));
J
John Stultz 已提交
742 743

	return ret;
744 745
}

746
/**
747
 * read_persistent_clock -  Return time from the persistent clock.
748 749
 *
 * Weak dummy function for arches that do not yet support it.
750 751
 * Reads the time from the battery backed persistent clock.
 * Returns a timespec with tv_sec=0 and tv_nsec=0 if unsupported.
752 753 754
 *
 *  XXX - Do be sure to remove it once all arches implement it.
 */
755
void __weak read_persistent_clock(struct timespec *ts)
756
{
757 758
	ts->tv_sec = 0;
	ts->tv_nsec = 0;
759 760
}

761 762 763 764 765 766 767 768 769
/**
 * read_boot_clock -  Return time of the system start.
 *
 * Weak dummy function for arches that do not yet support it.
 * Function to read the exact time the system has been started.
 * Returns a timespec with tv_sec=0 and tv_nsec=0 if unsupported.
 *
 *  XXX - Do be sure to remove it once all arches implement it.
 */
770
void __weak read_boot_clock(struct timespec *ts)
771 772 773 774 775
{
	ts->tv_sec = 0;
	ts->tv_nsec = 0;
}

776 777 778 779 780
/*
 * timekeeping_init - Initializes the clocksource and common timekeeping values
 */
void __init timekeeping_init(void)
{
781
	struct timekeeper *tk = &timekeeper;
782
	struct clocksource *clock;
783
	unsigned long flags;
784
	struct timespec now, boot, tmp;
785 786

	read_persistent_clock(&now);
787

788
	if (!timespec_valid_strict(&now)) {
789 790 791 792
		pr_warn("WARNING: Persistent clock returned invalid value!\n"
			"         Check your CMOS/BIOS settings.\n");
		now.tv_sec = 0;
		now.tv_nsec = 0;
793 794
	} else if (now.tv_sec || now.tv_nsec)
		persistent_clock_exist = true;
795

796
	read_boot_clock(&boot);
797
	if (!timespec_valid_strict(&boot)) {
798 799 800 801 802
		pr_warn("WARNING: Boot clock returned invalid value!\n"
			"         Check your CMOS/BIOS settings.\n");
		boot.tv_sec = 0;
		boot.tv_nsec = 0;
	}
803

804 805
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
806 807
	ntp_init();

808
	clock = clocksource_default_clock();
809 810
	if (clock->enable)
		clock->enable(clock);
811
	tk_setup_internals(tk, clock);
812

813 814 815
	tk_set_xtime(tk, &now);
	tk->raw_time.tv_sec = 0;
	tk->raw_time.tv_nsec = 0;
816
	if (boot.tv_sec == 0 && boot.tv_nsec == 0)
817
		boot = tk_xtime(tk);
818

819
	set_normalized_timespec(&tmp, -boot.tv_sec, -boot.tv_nsec);
820
	tk_set_wall_to_mono(tk, tmp);
821 822 823

	tmp.tv_sec = 0;
	tmp.tv_nsec = 0;
824
	tk_set_sleep_time(tk, tmp);
825

826 827
	memcpy(&shadow_timekeeper, &timekeeper, sizeof(timekeeper));

828 829
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
830 831 832
}

/* time in seconds when suspend began */
833
static struct timespec timekeeping_suspend_time;
834

835 836 837 838 839 840 841
/**
 * __timekeeping_inject_sleeptime - Internal function to add sleep interval
 * @delta: pointer to a timespec delta value
 *
 * Takes a timespec offset measuring a suspend interval and properly
 * adds the sleep offset to the timekeeping variables.
 */
842 843
static void __timekeeping_inject_sleeptime(struct timekeeper *tk,
							struct timespec *delta)
844
{
845
	if (!timespec_valid_strict(delta)) {
846 847 848
		printk_deferred(KERN_WARNING
				"__timekeeping_inject_sleeptime: Invalid "
				"sleep delta value!\n");
849 850
		return;
	}
851
	tk_xtime_add(tk, delta);
852 853
	tk_set_wall_to_mono(tk, timespec_sub(tk->wall_to_monotonic, *delta));
	tk_set_sleep_time(tk, timespec_add(tk->total_sleep_time, *delta));
854
	tk_debug_account_sleep_time(delta);
855 856 857 858 859 860 861 862 863 864 865 866 867 868
}

/**
 * timekeeping_inject_sleeptime - Adds suspend interval to timeekeeping values
 * @delta: pointer to a timespec delta value
 *
 * This hook is for architectures that cannot support read_persistent_clock
 * because their RTC/persistent clock is only accessible when irqs are enabled.
 *
 * This function should only be called by rtc_resume(), and allows
 * a suspend offset to be injected into the timekeeping values.
 */
void timekeeping_inject_sleeptime(struct timespec *delta)
{
869
	struct timekeeper *tk = &timekeeper;
870
	unsigned long flags;
871

872 873 874 875 876
	/*
	 * Make sure we don't set the clock twice, as timekeeping_resume()
	 * already did it
	 */
	if (has_persistent_clock())
877 878
		return;

879 880
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
J
John Stultz 已提交
881

882
	timekeeping_forward_now(tk);
883

884
	__timekeeping_inject_sleeptime(tk, delta);
885

886
	timekeeping_update(tk, TK_CLEAR_NTP | TK_MIRROR | TK_CLOCK_WAS_SET);
887

888 889
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
890 891 892 893 894

	/* signal hrtimers about time change */
	clock_was_set();
}

895 896 897 898 899 900 901
/**
 * timekeeping_resume - Resumes the generic timekeeping subsystem.
 *
 * This is for the generic clocksource timekeeping.
 * xtime/wall_to_monotonic/jiffies/etc are
 * still managed by arch specific suspend/resume code.
 */
902
static void timekeeping_resume(void)
903
{
904
	struct timekeeper *tk = &timekeeper;
905
	struct clocksource *clock = tk->clock;
906
	unsigned long flags;
907 908 909
	struct timespec ts_new, ts_delta;
	cycle_t cycle_now, cycle_delta;
	bool suspendtime_found = false;
910

911
	read_persistent_clock(&ts_new);
912

913
	clockevents_resume();
914 915
	clocksource_resume();

916 917
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
918

919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958
	/*
	 * After system resumes, we need to calculate the suspended time and
	 * compensate it for the OS time. There are 3 sources that could be
	 * used: Nonstop clocksource during suspend, persistent clock and rtc
	 * device.
	 *
	 * One specific platform may have 1 or 2 or all of them, and the
	 * preference will be:
	 *	suspend-nonstop clocksource -> persistent clock -> rtc
	 * The less preferred source will only be tried if there is no better
	 * usable source. The rtc part is handled separately in rtc core code.
	 */
	cycle_now = clock->read(clock);
	if ((clock->flags & CLOCK_SOURCE_SUSPEND_NONSTOP) &&
		cycle_now > clock->cycle_last) {
		u64 num, max = ULLONG_MAX;
		u32 mult = clock->mult;
		u32 shift = clock->shift;
		s64 nsec = 0;

		cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;

		/*
		 * "cycle_delta * mutl" may cause 64 bits overflow, if the
		 * suspended time is too long. In that case we need do the
		 * 64 bits math carefully
		 */
		do_div(max, mult);
		if (cycle_delta > max) {
			num = div64_u64(cycle_delta, max);
			nsec = (((u64) max * mult) >> shift) * num;
			cycle_delta -= num * max;
		}
		nsec += ((u64) cycle_delta * mult) >> shift;

		ts_delta = ns_to_timespec(nsec);
		suspendtime_found = true;
	} else if (timespec_compare(&ts_new, &timekeeping_suspend_time) > 0) {
		ts_delta = timespec_sub(ts_new, timekeeping_suspend_time);
		suspendtime_found = true;
959
	}
960 961 962 963 964

	if (suspendtime_found)
		__timekeeping_inject_sleeptime(tk, &ts_delta);

	/* Re-base the last cycle value */
965
	tk->cycle_last = clock->cycle_last = cycle_now;
966
	tk->ntp_error = 0;
967
	timekeeping_suspended = 0;
968
	timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET);
969 970
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
971 972 973 974 975 976

	touch_softlockup_watchdog();

	clockevents_notify(CLOCK_EVT_NOTIFY_RESUME, NULL);

	/* Resume hrtimers */
977
	hrtimers_resume();
978 979
}

980
static int timekeeping_suspend(void)
981
{
982
	struct timekeeper *tk = &timekeeper;
983
	unsigned long flags;
984 985
	struct timespec		delta, delta_delta;
	static struct timespec	old_delta;
986

987
	read_persistent_clock(&timekeeping_suspend_time);
988

989 990 991 992 993 994 995 996
	/*
	 * On some systems the persistent_clock can not be detected at
	 * timekeeping_init by its return value, so if we see a valid
	 * value returned, update the persistent_clock_exists flag.
	 */
	if (timekeeping_suspend_time.tv_sec || timekeeping_suspend_time.tv_nsec)
		persistent_clock_exist = true;

997 998
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);
999
	timekeeping_forward_now(tk);
1000
	timekeeping_suspended = 1;
1001 1002 1003 1004 1005 1006 1007

	/*
	 * To avoid drift caused by repeated suspend/resumes,
	 * which each can add ~1 second drift error,
	 * try to compensate so the difference in system time
	 * and persistent_clock time stays close to constant.
	 */
1008
	delta = timespec_sub(tk_xtime(tk), timekeeping_suspend_time);
1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
	delta_delta = timespec_sub(delta, old_delta);
	if (abs(delta_delta.tv_sec)  >= 2) {
		/*
		 * if delta_delta is too large, assume time correction
		 * has occured and set old_delta to the current delta.
		 */
		old_delta = delta;
	} else {
		/* Otherwise try to adjust old_system to compensate */
		timekeeping_suspend_time =
			timespec_add(timekeeping_suspend_time, delta_delta);
	}
1021 1022

	timekeeping_update(tk, TK_MIRROR);
1023 1024
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
1025 1026

	clockevents_notify(CLOCK_EVT_NOTIFY_SUSPEND, NULL);
M
Magnus Damm 已提交
1027
	clocksource_suspend();
1028
	clockevents_suspend();
1029 1030 1031 1032 1033

	return 0;
}

/* sysfs resume/suspend bits for timekeeping */
1034
static struct syscore_ops timekeeping_syscore_ops = {
1035 1036 1037 1038
	.resume		= timekeeping_resume,
	.suspend	= timekeeping_suspend,
};

1039
static int __init timekeeping_init_ops(void)
1040
{
1041 1042
	register_syscore_ops(&timekeeping_syscore_ops);
	return 0;
1043 1044
}

1045
device_initcall(timekeeping_init_ops);
1046 1047 1048 1049 1050

/*
 * If the error is already larger, we look ahead even further
 * to compensate for late or lost adjustments.
 */
1051 1052
static __always_inline int timekeeping_bigadjust(struct timekeeper *tk,
						 s64 error, s64 *interval,
1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
						 s64 *offset)
{
	s64 tick_error, i;
	u32 look_ahead, adj;
	s32 error2, mult;

	/*
	 * Use the current error value to determine how much to look ahead.
	 * The larger the error the slower we adjust for it to avoid problems
	 * with losing too many ticks, otherwise we would overadjust and
	 * produce an even larger error.  The smaller the adjustment the
	 * faster we try to adjust for it, as lost ticks can do less harm
L
Li Zefan 已提交
1065
	 * here.  This is tuned so that an error of about 1 msec is adjusted
1066 1067
	 * within about 1 sec (or 2^20 nsec in 2^SHIFT_HZ ticks).
	 */
1068
	error2 = tk->ntp_error >> (NTP_SCALE_SHIFT + 22 - 2 * SHIFT_HZ);
1069 1070 1071 1072 1073 1074 1075 1076
	error2 = abs(error2);
	for (look_ahead = 0; error2 > 0; look_ahead++)
		error2 >>= 2;

	/*
	 * Now calculate the error in (1 << look_ahead) ticks, but first
	 * remove the single look ahead already included in the error.
	 */
1077 1078
	tick_error = ntp_tick_length() >> (tk->ntp_error_shift + 1);
	tick_error -= tk->xtime_interval >> 1;
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102
	error = ((error - tick_error) >> look_ahead) + tick_error;

	/* Finally calculate the adjustment shift value.  */
	i = *interval;
	mult = 1;
	if (error < 0) {
		error = -error;
		*interval = -*interval;
		*offset = -*offset;
		mult = -1;
	}
	for (adj = 0; error > i; adj++)
		error >>= 1;

	*interval <<= adj;
	*offset <<= adj;
	return mult << adj;
}

/*
 * Adjust the multiplier to reduce the error value,
 * this is optimized for the most common adjustments of -1,0,1,
 * for other values we can do a bit more work.
 */
1103
static void timekeeping_adjust(struct timekeeper *tk, s64 offset)
1104
{
1105
	s64 error, interval = tk->cycle_interval;
1106 1107
	int adj;

1108
	/*
1109
	 * The point of this is to check if the error is greater than half
1110 1111 1112 1113 1114
	 * an interval.
	 *
	 * First we shift it down from NTP_SHIFT to clocksource->shifted nsecs.
	 *
	 * Note we subtract one in the shift, so that error is really error*2.
1115 1116
	 * This "saves" dividing(shifting) interval twice, but keeps the
	 * (error > interval) comparison as still measuring if error is
1117
	 * larger than half an interval.
1118
	 *
1119
	 * Note: It does not "save" on aggravation when reading the code.
1120
	 */
1121
	error = tk->ntp_error >> (tk->ntp_error_shift - 1);
1122
	if (error > interval) {
1123 1124
		/*
		 * We now divide error by 4(via shift), which checks if
1125
		 * the error is greater than twice the interval.
1126 1127 1128
		 * If it is greater, we need a bigadjust, if its smaller,
		 * we can adjust by 1.
		 */
1129 1130 1131 1132
		error >>= 2;
		if (likely(error <= interval))
			adj = 1;
		else
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
			adj = timekeeping_bigadjust(tk, error, &interval, &offset);
	} else {
		if (error < -interval) {
			/* See comment above, this is just switched for the negative */
			error >>= 2;
			if (likely(error >= -interval)) {
				adj = -1;
				interval = -interval;
				offset = -offset;
			} else {
				adj = timekeeping_bigadjust(tk, error, &interval, &offset);
			}
		} else {
			goto out_adjust;
		}
	}
1149

1150 1151
	if (unlikely(tk->clock->maxadj &&
		(tk->mult + adj > tk->clock->mult + tk->clock->maxadj))) {
1152
		printk_deferred_once(KERN_WARNING
1153
			"Adjusting %s more than 11%% (%ld vs %ld)\n",
1154 1155
			tk->clock->name, (long)tk->mult + adj,
			(long)tk->clock->mult + tk->clock->maxadj);
1156
	}
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
	/*
	 * So the following can be confusing.
	 *
	 * To keep things simple, lets assume adj == 1 for now.
	 *
	 * When adj != 1, remember that the interval and offset values
	 * have been appropriately scaled so the math is the same.
	 *
	 * The basic idea here is that we're increasing the multiplier
	 * by one, this causes the xtime_interval to be incremented by
	 * one cycle_interval. This is because:
	 *	xtime_interval = cycle_interval * mult
	 * So if mult is being incremented by one:
	 *	xtime_interval = cycle_interval * (mult + 1)
	 * Its the same as:
	 *	xtime_interval = (cycle_interval * mult) + cycle_interval
	 * Which can be shortened to:
	 *	xtime_interval += cycle_interval
	 *
	 * So offset stores the non-accumulated cycles. Thus the current
	 * time (in shifted nanoseconds) is:
	 *	now = (offset * adj) + xtime_nsec
	 * Now, even though we're adjusting the clock frequency, we have
	 * to keep time consistent. In other words, we can't jump back
	 * in time, and we also want to avoid jumping forward in time.
	 *
	 * So given the same offset value, we need the time to be the same
	 * both before and after the freq adjustment.
	 *	now = (offset * adj_1) + xtime_nsec_1
	 *	now = (offset * adj_2) + xtime_nsec_2
	 * So:
	 *	(offset * adj_1) + xtime_nsec_1 =
	 *		(offset * adj_2) + xtime_nsec_2
	 * And we know:
	 *	adj_2 = adj_1 + 1
	 * So:
	 *	(offset * adj_1) + xtime_nsec_1 =
	 *		(offset * (adj_1+1)) + xtime_nsec_2
	 *	(offset * adj_1) + xtime_nsec_1 =
	 *		(offset * adj_1) + offset + xtime_nsec_2
	 * Canceling the sides:
	 *	xtime_nsec_1 = offset + xtime_nsec_2
	 * Which gives us:
	 *	xtime_nsec_2 = xtime_nsec_1 - offset
	 * Which simplfies to:
	 *	xtime_nsec -= offset
	 *
	 * XXX - TODO: Doc ntp_error calculation.
	 */
1206 1207 1208 1209
	tk->mult += adj;
	tk->xtime_interval += interval;
	tk->xtime_nsec -= offset;
	tk->ntp_error -= (interval - offset) << tk->ntp_error_shift;
1210

1211
out_adjust:
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
	/*
	 * It may be possible that when we entered this function, xtime_nsec
	 * was very small.  Further, if we're slightly speeding the clocksource
	 * in the code above, its possible the required corrective factor to
	 * xtime_nsec could cause it to underflow.
	 *
	 * Now, since we already accumulated the second, cannot simply roll
	 * the accumulated second back, since the NTP subsystem has been
	 * notified via second_overflow. So instead we push xtime_nsec forward
	 * by the amount we underflowed, and add that amount into the error.
	 *
	 * We'll correct this error next time through this function, when
	 * xtime_nsec is not as small.
	 */
1226 1227 1228 1229
	if (unlikely((s64)tk->xtime_nsec < 0)) {
		s64 neg = -(s64)tk->xtime_nsec;
		tk->xtime_nsec = 0;
		tk->ntp_error += neg << tk->ntp_error_shift;
1230 1231
	}

1232 1233
}

1234 1235 1236 1237 1238 1239 1240 1241
/**
 * accumulate_nsecs_to_secs - Accumulates nsecs into secs
 *
 * Helper function that accumulates a the nsecs greater then a second
 * from the xtime_nsec field to the xtime_secs field.
 * It also calls into the NTP code to handle leapsecond processing.
 *
 */
1242
static inline unsigned int accumulate_nsecs_to_secs(struct timekeeper *tk)
1243 1244
{
	u64 nsecps = (u64)NSEC_PER_SEC << tk->shift;
1245
	unsigned int clock_set = 0;
1246 1247 1248 1249 1250 1251 1252 1253 1254

	while (tk->xtime_nsec >= nsecps) {
		int leap;

		tk->xtime_nsec -= nsecps;
		tk->xtime_sec++;

		/* Figure out if its a leap sec and apply if needed */
		leap = second_overflow(tk->xtime_sec);
1255 1256 1257 1258
		if (unlikely(leap)) {
			struct timespec ts;

			tk->xtime_sec += leap;
1259

1260 1261 1262 1263 1264
			ts.tv_sec = leap;
			ts.tv_nsec = 0;
			tk_set_wall_to_mono(tk,
				timespec_sub(tk->wall_to_monotonic, ts));

1265 1266
			__timekeeping_set_tai_offset(tk, tk->tai_offset - leap);

1267
			clock_set = TK_CLOCK_WAS_SET;
1268
		}
1269
	}
1270
	return clock_set;
1271 1272
}

1273 1274 1275 1276 1277 1278 1279 1280 1281
/**
 * logarithmic_accumulation - shifted accumulation of cycles
 *
 * This functions accumulates a shifted interval of cycles into
 * into a shifted interval nanoseconds. Allows for O(log) accumulation
 * loop.
 *
 * Returns the unconsumed cycles.
 */
1282
static cycle_t logarithmic_accumulation(struct timekeeper *tk, cycle_t offset,
1283 1284
						u32 shift,
						unsigned int *clock_set)
1285
{
T
Thomas Gleixner 已提交
1286
	cycle_t interval = tk->cycle_interval << shift;
1287
	u64 raw_nsecs;
1288

1289
	/* If the offset is smaller then a shifted interval, do nothing */
T
Thomas Gleixner 已提交
1290
	if (offset < interval)
1291 1292 1293
		return offset;

	/* Accumulate one shifted interval */
T
Thomas Gleixner 已提交
1294
	offset -= interval;
1295
	tk->cycle_last += interval;
1296

1297
	tk->xtime_nsec += tk->xtime_interval << shift;
1298
	*clock_set |= accumulate_nsecs_to_secs(tk);
1299

1300
	/* Accumulate raw time */
1301
	raw_nsecs = (u64)tk->raw_interval << shift;
1302
	raw_nsecs += tk->raw_time.tv_nsec;
1303 1304 1305
	if (raw_nsecs >= NSEC_PER_SEC) {
		u64 raw_secs = raw_nsecs;
		raw_nsecs = do_div(raw_secs, NSEC_PER_SEC);
1306
		tk->raw_time.tv_sec += raw_secs;
1307
	}
1308
	tk->raw_time.tv_nsec = raw_nsecs;
1309 1310

	/* Accumulate error between NTP and clock interval */
1311 1312 1313
	tk->ntp_error += ntp_tick_length() << shift;
	tk->ntp_error -= (tk->xtime_interval + tk->xtime_remainder) <<
						(tk->ntp_error_shift + shift);
1314 1315 1316 1317

	return offset;
}

1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336
#ifdef CONFIG_GENERIC_TIME_VSYSCALL_OLD
static inline void old_vsyscall_fixup(struct timekeeper *tk)
{
	s64 remainder;

	/*
	* Store only full nanoseconds into xtime_nsec after rounding
	* it up and add the remainder to the error difference.
	* XXX - This is necessary to avoid small 1ns inconsistnecies caused
	* by truncating the remainder in vsyscalls. However, it causes
	* additional work to be done in timekeeping_adjust(). Once
	* the vsyscall implementations are converted to use xtime_nsec
	* (shifted nanoseconds), and CONFIG_GENERIC_TIME_VSYSCALL_OLD
	* users are removed, this can be killed.
	*/
	remainder = tk->xtime_nsec & ((1ULL << tk->shift) - 1);
	tk->xtime_nsec -= remainder;
	tk->xtime_nsec += 1ULL << tk->shift;
	tk->ntp_error += remainder << tk->ntp_error_shift;
1337
	tk->ntp_error -= (1ULL << tk->shift) << tk->ntp_error_shift;
1338 1339 1340 1341 1342 1343 1344
}
#else
#define old_vsyscall_fixup(tk)
#endif



1345 1346 1347 1348
/**
 * update_wall_time - Uses the current clocksource to increment the wall time
 *
 */
1349
void update_wall_time(void)
1350
{
1351
	struct clocksource *clock;
1352 1353
	struct timekeeper *real_tk = &timekeeper;
	struct timekeeper *tk = &shadow_timekeeper;
1354
	cycle_t offset;
1355
	int shift = 0, maxshift;
1356
	unsigned int clock_set = 0;
J
John Stultz 已提交
1357 1358
	unsigned long flags;

1359
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
1360 1361 1362

	/* Make sure we're fully resumed: */
	if (unlikely(timekeeping_suspended))
J
John Stultz 已提交
1363
		goto out;
1364

1365
	clock = real_tk->clock;
J
John Stultz 已提交
1366 1367

#ifdef CONFIG_ARCH_USES_GETTIMEOFFSET
1368
	offset = real_tk->cycle_interval;
J
John Stultz 已提交
1369 1370
#else
	offset = (clock->read(clock) - clock->cycle_last) & clock->mask;
1371 1372
#endif

1373
	/* Check if there's really nothing to do */
1374
	if (offset < real_tk->cycle_interval)
1375 1376
		goto out;

1377 1378 1379 1380
	/*
	 * With NO_HZ we may have to accumulate many cycle_intervals
	 * (think "ticks") worth of time at once. To do this efficiently,
	 * we calculate the largest doubling multiple of cycle_intervals
1381
	 * that is smaller than the offset.  We then accumulate that
1382 1383
	 * chunk in one go, and then try to consume the next smaller
	 * doubled multiple.
1384
	 */
1385
	shift = ilog2(offset) - ilog2(tk->cycle_interval);
1386
	shift = max(0, shift);
1387
	/* Bound shift to one less than what overflows tick_length */
1388
	maxshift = (64 - (ilog2(ntp_tick_length())+1)) - 1;
1389
	shift = min(shift, maxshift);
1390
	while (offset >= tk->cycle_interval) {
1391 1392
		offset = logarithmic_accumulation(tk, offset, shift,
							&clock_set);
1393
		if (offset < tk->cycle_interval<<shift)
1394
			shift--;
1395 1396 1397
	}

	/* correct the clock when NTP error is too big */
1398
	timekeeping_adjust(tk, offset);
1399

J
John Stultz 已提交
1400
	/*
1401 1402 1403 1404
	 * XXX This can be killed once everyone converts
	 * to the new update_vsyscall.
	 */
	old_vsyscall_fixup(tk);
1405

J
John Stultz 已提交
1406 1407
	/*
	 * Finally, make sure that after the rounding
1408
	 * xtime_nsec isn't larger than NSEC_PER_SEC
J
John Stultz 已提交
1409
	 */
1410
	clock_set |= accumulate_nsecs_to_secs(tk);
L
Linus Torvalds 已提交
1411

1412
	write_seqcount_begin(&timekeeper_seq);
1413 1414
	/* Update clock->cycle_last with the new value */
	clock->cycle_last = tk->cycle_last;
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
	/*
	 * Update the real timekeeper.
	 *
	 * We could avoid this memcpy by switching pointers, but that
	 * requires changes to all other timekeeper usage sites as
	 * well, i.e. move the timekeeper pointer getter into the
	 * spinlocked/seqcount protected sections. And we trade this
	 * memcpy under the timekeeper_seq against one before we start
	 * updating.
	 */
	memcpy(real_tk, tk, sizeof(*tk));
1426
	timekeeping_update(real_tk, clock_set);
1427
	write_seqcount_end(&timekeeper_seq);
1428
out:
1429
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
1430
	if (clock_set)
1431 1432
		/* Have to call _delayed version, since in irq context*/
		clock_was_set_delayed();
1433
}
T
Tomas Janousek 已提交
1434 1435 1436 1437 1438

/**
 * getboottime - Return the real time of system boot.
 * @ts:		pointer to the timespec to be set
 *
1439
 * Returns the wall-time of boot in a timespec.
T
Tomas Janousek 已提交
1440 1441 1442 1443 1444 1445 1446 1447
 *
 * This is based on the wall_to_monotonic offset and the total suspend
 * time. Calls to settimeofday will affect the value returned (which
 * basically means that however wrong your real time clock is at boot time,
 * you get the right time here).
 */
void getboottime(struct timespec *ts)
{
1448
	struct timekeeper *tk = &timekeeper;
1449
	struct timespec boottime = {
1450 1451 1452 1453
		.tv_sec = tk->wall_to_monotonic.tv_sec +
				tk->total_sleep_time.tv_sec,
		.tv_nsec = tk->wall_to_monotonic.tv_nsec +
				tk->total_sleep_time.tv_nsec
1454
	};
1455 1456

	set_normalized_timespec(ts, -boottime.tv_sec, -boottime.tv_nsec);
T
Tomas Janousek 已提交
1457
}
1458
EXPORT_SYMBOL_GPL(getboottime);
T
Tomas Janousek 已提交
1459

1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470
/**
 * get_monotonic_boottime - Returns monotonic time since boot
 * @ts:		pointer to the timespec to be set
 *
 * Returns the monotonic time since boot in a timespec.
 *
 * This is similar to CLOCK_MONTONIC/ktime_get_ts, but also
 * includes the time spent in suspend.
 */
void get_monotonic_boottime(struct timespec *ts)
{
1471
	struct timekeeper *tk = &timekeeper;
1472
	struct timespec tomono, sleep;
1473
	s64 nsec;
1474 1475 1476 1477 1478
	unsigned int seq;

	WARN_ON(timekeeping_suspended);

	do {
1479
		seq = read_seqcount_begin(&timekeeper_seq);
1480
		ts->tv_sec = tk->xtime_sec;
1481
		nsec = timekeeping_get_ns(tk);
1482 1483
		tomono = tk->wall_to_monotonic;
		sleep = tk->total_sleep_time;
1484

1485
	} while (read_seqcount_retry(&timekeeper_seq, seq));
1486

1487 1488 1489
	ts->tv_sec += tomono.tv_sec + sleep.tv_sec;
	ts->tv_nsec = 0;
	timespec_add_ns(ts, nsec + tomono.tv_nsec + sleep.tv_nsec);
1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
}
EXPORT_SYMBOL_GPL(get_monotonic_boottime);

/**
 * ktime_get_boottime - Returns monotonic time since boot in a ktime
 *
 * Returns the monotonic time since boot in a ktime
 *
 * This is similar to CLOCK_MONTONIC/ktime_get, but also
 * includes the time spent in suspend.
 */
ktime_t ktime_get_boottime(void)
{
	struct timespec ts;

	get_monotonic_boottime(&ts);
	return timespec_to_ktime(ts);
}
EXPORT_SYMBOL_GPL(ktime_get_boottime);

T
Tomas Janousek 已提交
1510 1511 1512 1513 1514 1515
/**
 * monotonic_to_bootbased - Convert the monotonic time to boot based.
 * @ts:		pointer to the timespec to be converted
 */
void monotonic_to_bootbased(struct timespec *ts)
{
1516 1517 1518
	struct timekeeper *tk = &timekeeper;

	*ts = timespec_add(*ts, tk->total_sleep_time);
T
Tomas Janousek 已提交
1519
}
1520
EXPORT_SYMBOL_GPL(monotonic_to_bootbased);
1521

1522 1523
unsigned long get_seconds(void)
{
1524 1525 1526
	struct timekeeper *tk = &timekeeper;

	return tk->xtime_sec;
1527 1528 1529
}
EXPORT_SYMBOL(get_seconds);

1530 1531
struct timespec __current_kernel_time(void)
{
1532 1533 1534
	struct timekeeper *tk = &timekeeper;

	return tk_xtime(tk);
1535
}
1536

1537 1538
struct timespec current_kernel_time(void)
{
1539
	struct timekeeper *tk = &timekeeper;
1540 1541 1542 1543
	struct timespec now;
	unsigned long seq;

	do {
1544
		seq = read_seqcount_begin(&timekeeper_seq);
L
Linus Torvalds 已提交
1545

1546
		now = tk_xtime(tk);
1547
	} while (read_seqcount_retry(&timekeeper_seq, seq));
1548 1549 1550 1551

	return now;
}
EXPORT_SYMBOL(current_kernel_time);
1552 1553 1554

struct timespec get_monotonic_coarse(void)
{
1555
	struct timekeeper *tk = &timekeeper;
1556 1557 1558 1559
	struct timespec now, mono;
	unsigned long seq;

	do {
1560
		seq = read_seqcount_begin(&timekeeper_seq);
L
Linus Torvalds 已提交
1561

1562 1563
		now = tk_xtime(tk);
		mono = tk->wall_to_monotonic;
1564
	} while (read_seqcount_retry(&timekeeper_seq, seq));
1565 1566 1567 1568 1569

	set_normalized_timespec(&now, now.tv_sec + mono.tv_sec,
				now.tv_nsec + mono.tv_nsec);
	return now;
}
1570 1571

/*
1572
 * Must hold jiffies_lock
1573 1574 1575 1576 1577 1578
 */
void do_timer(unsigned long ticks)
{
	jiffies_64 += ticks;
	calc_global_load(ticks);
}
1579 1580

/**
1581 1582 1583 1584 1585 1586
 * ktime_get_update_offsets_tick - hrtimer helper
 * @offs_real:	pointer to storage for monotonic -> realtime offset
 * @offs_boot:	pointer to storage for monotonic -> boottime offset
 * @offs_tai:	pointer to storage for monotonic -> clock tai offset
 *
 * Returns monotonic time at last tick and various offsets
1587
 */
1588 1589
ktime_t ktime_get_update_offsets_tick(ktime_t *offs_real, ktime_t *offs_boot,
							ktime_t *offs_tai)
1590
{
1591
	struct timekeeper *tk = &timekeeper;
1592 1593 1594
	struct timespec ts;
	ktime_t now;
	unsigned int seq;
1595 1596

	do {
1597
		seq = read_seqcount_begin(&timekeeper_seq);
1598 1599 1600 1601 1602 1603

		ts = tk_xtime(tk);

		*offs_real = tk->offs_real;
		*offs_boot = tk->offs_boot;
		*offs_tai = tk->offs_tai;
1604
	} while (read_seqcount_retry(&timekeeper_seq, seq));
1605 1606 1607 1608

	now = ktime_set(ts.tv_sec, ts.tv_nsec);
	now = ktime_sub(now, *offs_real);
	return now;
1609
}
T
Torben Hohn 已提交
1610

1611 1612
#ifdef CONFIG_HIGH_RES_TIMERS
/**
1613
 * ktime_get_update_offsets_now - hrtimer helper
1614 1615
 * @offs_real:	pointer to storage for monotonic -> realtime offset
 * @offs_boot:	pointer to storage for monotonic -> boottime offset
1616
 * @offs_tai:	pointer to storage for monotonic -> clock tai offset
1617 1618
 *
 * Returns current monotonic time and updates the offsets
1619
 * Called from hrtimer_interrupt() or retrigger_next_event()
1620
 */
1621
ktime_t ktime_get_update_offsets_now(ktime_t *offs_real, ktime_t *offs_boot,
1622
							ktime_t *offs_tai)
1623
{
1624
	struct timekeeper *tk = &timekeeper;
1625 1626 1627 1628 1629
	ktime_t now;
	unsigned int seq;
	u64 secs, nsecs;

	do {
1630
		seq = read_seqcount_begin(&timekeeper_seq);
1631

1632 1633
		secs = tk->xtime_sec;
		nsecs = timekeeping_get_ns(tk);
1634

1635 1636
		*offs_real = tk->offs_real;
		*offs_boot = tk->offs_boot;
1637
		*offs_tai = tk->offs_tai;
1638
	} while (read_seqcount_retry(&timekeeper_seq, seq));
1639 1640 1641 1642 1643 1644 1645

	now = ktime_add_ns(ktime_set(secs, 0), nsecs);
	now = ktime_sub(now, *offs_real);
	return now;
}
#endif

1646 1647 1648 1649 1650
/**
 * ktime_get_monotonic_offset() - get wall_to_monotonic in ktime_t format
 */
ktime_t ktime_get_monotonic_offset(void)
{
1651
	struct timekeeper *tk = &timekeeper;
1652 1653 1654 1655
	unsigned long seq;
	struct timespec wtom;

	do {
1656
		seq = read_seqcount_begin(&timekeeper_seq);
1657
		wtom = tk->wall_to_monotonic;
1658
	} while (read_seqcount_retry(&timekeeper_seq, seq));
J
John Stultz 已提交
1659

1660 1661
	return timespec_to_ktime(wtom);
}
1662 1663
EXPORT_SYMBOL_GPL(ktime_get_monotonic_offset);

1664 1665 1666 1667 1668
/**
 * do_adjtimex() - Accessor function to NTP __do_adjtimex function
 */
int do_adjtimex(struct timex *txc)
{
1669
	struct timekeeper *tk = &timekeeper;
1670
	unsigned long flags;
1671
	struct timespec ts;
1672
	s32 orig_tai, tai;
1673 1674 1675 1676 1677 1678 1679
	int ret;

	/* Validate the data before disabling interrupts */
	ret = ntp_validate_timex(txc);
	if (ret)
		return ret;

1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
	if (txc->modes & ADJ_SETOFFSET) {
		struct timespec delta;
		delta.tv_sec  = txc->time.tv_sec;
		delta.tv_nsec = txc->time.tv_usec;
		if (!(txc->modes & ADJ_NANO))
			delta.tv_nsec *= 1000;
		ret = timekeeping_inject_offset(&delta);
		if (ret)
			return ret;
	}

1691 1692
	getnstimeofday(&ts);

1693 1694 1695
	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);

1696
	orig_tai = tai = tk->tai_offset;
1697
	ret = __do_adjtimex(txc, &ts, &tai);
1698

1699 1700
	if (tai != orig_tai) {
		__timekeeping_set_tai_offset(tk, tai);
1701
		timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET);
1702
	}
1703 1704 1705
	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);

1706 1707 1708
	if (tai != orig_tai)
		clock_was_set();

1709 1710
	ntp_notify_cmos_timer();

1711 1712
	return ret;
}
1713 1714 1715 1716 1717 1718 1719

#ifdef CONFIG_NTP_PPS
/**
 * hardpps() - Accessor function to NTP __hardpps function
 */
void hardpps(const struct timespec *phase_ts, const struct timespec *raw_ts)
{
1720 1721 1722 1723 1724
	unsigned long flags;

	raw_spin_lock_irqsave(&timekeeper_lock, flags);
	write_seqcount_begin(&timekeeper_seq);

1725
	__hardpps(phase_ts, raw_ts);
1726 1727 1728

	write_seqcount_end(&timekeeper_seq);
	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
1729 1730 1731 1732
}
EXPORT_SYMBOL(hardpps);
#endif

T
Torben Hohn 已提交
1733 1734 1735 1736 1737 1738 1739 1740
/**
 * xtime_update() - advances the timekeeping infrastructure
 * @ticks:	number of ticks, that have elapsed since the last call.
 *
 * Must be called with interrupts disabled.
 */
void xtime_update(unsigned long ticks)
{
1741
	write_seqlock(&jiffies_lock);
T
Torben Hohn 已提交
1742
	do_timer(ticks);
1743
	write_sequnlock(&jiffies_lock);
1744
	update_wall_time();
T
Torben Hohn 已提交
1745
}