timekeeping.c 22.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 *  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.
 *
 */

#include <linux/module.h>
#include <linux/interrupt.h>
#include <linux/percpu.h>
#include <linux/init.h>
#include <linux/mm.h>
16
#include <linux/sched.h>
17 18 19 20 21
#include <linux/sysdev.h>
#include <linux/clocksource.h>
#include <linux/jiffies.h>
#include <linux/time.h>
#include <linux/tick.h>
22
#include <linux/stop_machine.h>
23

24 25 26 27
/* Structure holding internal timekeeping values. */
struct timekeeper {
	/* Current clocksource used for timekeeping. */
	struct clocksource *clock;
28 29
	/* The shift value of the current clocksource. */
	int	shift;
30 31 32 33 34 35 36 37 38 39 40 41 42

	/* Number of clock cycles in one NTP interval. */
	cycle_t cycle_interval;
	/* Number of clock shifted nano seconds in one NTP interval. */
	u64	xtime_interval;
	/* Raw nano seconds accumulated per NTP interval. */
	u32	raw_interval;

	/* Clock shifted nano seconds remainder not stored in xtime.tv_nsec. */
	u64	xtime_nsec;
	/* Difference between accumulated time and NTP time in ntp
	 * shifted nano seconds. */
	s64	ntp_error;
43 44 45
	/* Shift conversion between clock shifted nano seconds and
	 * ntp shifted nano seconds. */
	int	ntp_error_shift;
46 47
	/* NTP adjusted clock multiplier */
	u32	mult;
48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72
};

struct timekeeper timekeeper;

/**
 * timekeeper_setup_internals - Set up internals to use clocksource clock.
 *
 * @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!
 */
static void timekeeper_setup_internals(struct clocksource *clock)
{
	cycle_t interval;
	u64 tmp;

	timekeeper.clock = clock;
	clock->cycle_last = clock->read(clock);

	/* Do the ns -> cycle conversion first, using original mult */
	tmp = NTP_INTERVAL_LENGTH;
	tmp <<= clock->shift;
73 74
	tmp += clock->mult/2;
	do_div(tmp, clock->mult);
75 76 77 78 79 80 81 82 83
	if (tmp == 0)
		tmp = 1;

	interval = (cycle_t) tmp;
	timekeeper.cycle_interval = interval;

	/* Go back from cycles -> shifted ns */
	timekeeper.xtime_interval = (u64) interval * clock->mult;
	timekeeper.raw_interval =
84
		((u64) interval * clock->mult) >> clock->shift;
85 86

	timekeeper.xtime_nsec = 0;
87
	timekeeper.shift = clock->shift;
88 89

	timekeeper.ntp_error = 0;
90
	timekeeper.ntp_error_shift = NTP_SCALE_SHIFT - clock->shift;
91 92 93 94 95 96 97

	/*
	 * The timekeeper keeps its own mult values for the currently
	 * active clocksource. These value will be adjusted via NTP
	 * to counteract clock drifting.
	 */
	timekeeper.mult = clock->mult;
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
/* Timekeeper helper functions. */
static inline s64 timekeeping_get_ns(void)
{
	cycle_t cycle_now, cycle_delta;
	struct clocksource *clock;

	/* read clocksource: */
	clock = timekeeper.clock;
	cycle_now = clock->read(clock);

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

	/* return delta convert to nanoseconds using ntp adjusted mult. */
	return clocksource_cyc2ns(cycle_delta, timekeeper.mult,
				  timekeeper.shift);
}

static inline s64 timekeeping_get_ns_raw(void)
{
	cycle_t cycle_now, cycle_delta;
	struct clocksource *clock;

	/* read clocksource: */
	clock = timekeeper.clock;
	cycle_now = clock->read(clock);

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

	/* return delta convert to nanoseconds using ntp adjusted mult. */
	return clocksource_cyc2ns(cycle_delta, clock->mult, clock->shift);
}

134 135
/*
 * This read-write spinlock protects us from races in SMP while
136
 * playing with xtime.
137
 */
A
Adrian Bunk 已提交
138
__cacheline_aligned_in_smp DEFINE_SEQLOCK(xtime_lock);
139 140 141 142 143 144 145 146 147


/*
 * The current time
 * wall_to_monotonic is what we need to add to xtime (or xtime corrected
 * for sub jiffie times) to get to monotonic time.  Monotonic is pegged
 * at zero at system boot time, so wall_to_monotonic will be negative,
 * however, we will ALWAYS keep the tv_nsec part positive so we can use
 * the usual normalization.
T
Tomas Janousek 已提交
148 149 150 151 152 153 154
 *
 * wall_to_monotonic is moved after resume from suspend for the monotonic
 * time not to jump. We need to add total_sleep_time to wall_to_monotonic
 * to get the real boot based time offset.
 *
 * - wall_to_monotonic is no longer the boot time, getboottime must be
 * used instead.
155 156 157
 */
struct timespec xtime __attribute__ ((aligned (16)));
struct timespec wall_to_monotonic __attribute__ ((aligned (16)));
158
static struct timespec total_sleep_time;
159

160 161 162 163 164
/*
 * The raw monotonic time for the CLOCK_MONOTONIC_RAW posix clock.
 */
struct timespec raw_time;

165 166 167
/* flag for if timekeeping is suspended */
int __read_mostly timekeeping_suspended;

168 169 170 171 172
/* must hold xtime_lock */
void timekeeping_leap_insert(int leapsecond)
{
	xtime.tv_sec += leapsecond;
	wall_to_monotonic.tv_sec -= leapsecond;
173
	update_vsyscall(&xtime, timekeeper.clock, timekeeper.mult);
174
}
175 176

/**
177
 * timekeeping_forward_now - update clock to the current time
178
 *
179 180 181
 * 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.
182
 */
183
static void timekeeping_forward_now(void)
184 185
{
	cycle_t cycle_now, cycle_delta;
186
	struct clocksource *clock;
187
	s64 nsec;
188

189
	clock = timekeeper.clock;
190
	cycle_now = clock->read(clock);
191
	cycle_delta = (cycle_now - clock->cycle_last) & clock->mask;
192
	clock->cycle_last = cycle_now;
193

194 195
	nsec = clocksource_cyc2ns(cycle_delta, timekeeper.mult,
				  timekeeper.shift);
196 197 198 199

	/* If arch requires, add in gettimeoffset() */
	nsec += arch_gettimeoffset();

200
	timespec_add_ns(&xtime, nsec);
201

202
	nsec = clocksource_cyc2ns(cycle_delta, clock->mult, clock->shift);
203
	timespec_add_ns(&raw_time, nsec);
204 205 206
}

/**
207
 * getnstimeofday - Returns the time of day in a timespec
208 209
 * @ts:		pointer to the timespec to be set
 *
210
 * Returns the time of day in a timespec.
211
 */
212
void getnstimeofday(struct timespec *ts)
213 214 215 216
{
	unsigned long seq;
	s64 nsecs;

217 218
	WARN_ON(timekeeping_suspended);

219 220 221 222
	do {
		seq = read_seqbegin(&xtime_lock);

		*ts = xtime;
223
		nsecs = timekeeping_get_ns();
224

225 226 227
		/* If arch requires, add in gettimeoffset() */
		nsecs += arch_gettimeoffset();

228 229 230 231 232 233 234
	} while (read_seqretry(&xtime_lock, seq));

	timespec_add_ns(ts, nsecs);
}

EXPORT_SYMBOL(getnstimeofday);

235 236 237 238 239 240 241 242 243 244 245
ktime_t ktime_get(void)
{
	unsigned int seq;
	s64 secs, nsecs;

	WARN_ON(timekeeping_suspended);

	do {
		seq = read_seqbegin(&xtime_lock);
		secs = xtime.tv_sec + wall_to_monotonic.tv_sec;
		nsecs = xtime.tv_nsec + wall_to_monotonic.tv_nsec;
246
		nsecs += timekeeping_get_ns();
247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276

	} while (read_seqretry(&xtime_lock, seq));
	/*
	 * Use ktime_set/ktime_add_ns to create a proper ktime on
	 * 32-bit architectures without CONFIG_KTIME_SCALAR.
	 */
	return ktime_add_ns(ktime_set(secs, 0), nsecs);
}
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)
{
	struct timespec tomono;
	unsigned int seq;
	s64 nsecs;

	WARN_ON(timekeeping_suspended);

	do {
		seq = read_seqbegin(&xtime_lock);
		*ts = xtime;
		tomono = wall_to_monotonic;
277
		nsecs = timekeeping_get_ns();
278 279 280 281 282 283 284 285

	} while (read_seqretry(&xtime_lock, seq));

	set_normalized_timespec(ts, ts->tv_sec + tomono.tv_sec,
				ts->tv_nsec + tomono.tv_nsec + nsecs);
}
EXPORT_SYMBOL_GPL(ktime_get_ts);

286 287 288 289
/**
 * do_gettimeofday - Returns the time of day in a timeval
 * @tv:		pointer to the timeval to be set
 *
290
 * NOTE: Users should be converted to using getnstimeofday()
291 292 293 294 295
 */
void do_gettimeofday(struct timeval *tv)
{
	struct timespec now;

296
	getnstimeofday(&now);
297 298 299 300 301 302 303 304 305 306 307 308 309
	tv->tv_sec = now.tv_sec;
	tv->tv_usec = now.tv_nsec/1000;
}

EXPORT_SYMBOL(do_gettimeofday);
/**
 * 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
 */
int do_settimeofday(struct timespec *tv)
{
310
	struct timespec ts_delta;
311 312 313 314 315 316 317
	unsigned long flags;

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

	write_seqlock_irqsave(&xtime_lock, flags);

318
	timekeeping_forward_now();
319 320 321 322

	ts_delta.tv_sec = tv->tv_sec - xtime.tv_sec;
	ts_delta.tv_nsec = tv->tv_nsec - xtime.tv_nsec;
	wall_to_monotonic = timespec_sub(wall_to_monotonic, ts_delta);
323

324
	xtime = *tv;
325

326
	timekeeper.ntp_error = 0;
327 328
	ntp_clear();

329
	update_vsyscall(&xtime, timekeeper.clock, timekeeper.mult);
330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345

	write_sequnlock_irqrestore(&xtime_lock, flags);

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

	return 0;
}

EXPORT_SYMBOL(do_settimeofday);

/**
 * change_clocksource - Swaps clocksources if a new one is available
 *
 * Accumulates current time interval and initializes new clocksource
 */
346
static int change_clocksource(void *data)
347
{
348
	struct clocksource *new, *old;
349

350
	new = (struct clocksource *) data;
351

352
	timekeeping_forward_now();
353 354 355 356 357 358 359 360
	if (!new->enable || new->enable(new) == 0) {
		old = timekeeper.clock;
		timekeeper_setup_internals(new);
		if (old->disable)
			old->disable(old);
	}
	return 0;
}
361

362 363 364 365 366 367 368 369 370 371
/**
 * 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.
 */
void timekeeping_notify(struct clocksource *clock)
{
	if (timekeeper.clock == clock)
372
		return;
373
	stop_machine(change_clocksource, clock, NULL);
374 375
	tick_clock_notify();
}
376

377 378 379 380 381 382 383 384 385 386 387 388 389 390
/**
 * 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);
391

392 393 394 395 396 397 398 399 400 401 402 403 404
/**
 * 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)
{
	unsigned long seq;
	s64 nsecs;

	do {
		seq = read_seqbegin(&xtime_lock);
405
		nsecs = timekeeping_get_ns_raw();
406
		*ts = raw_time;
407 408 409 410 411 412 413 414

	} while (read_seqretry(&xtime_lock, seq));

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


415
/**
416
 * timekeeping_valid_for_hres - Check if timekeeping is suitable for hres
417
 */
418
int timekeeping_valid_for_hres(void)
419 420 421 422 423 424 425
{
	unsigned long seq;
	int ret;

	do {
		seq = read_seqbegin(&xtime_lock);

426
		ret = timekeeper.clock->flags & CLOCK_SOURCE_VALID_FOR_HRES;
427 428 429 430 431 432

	} while (read_seqretry(&xtime_lock, seq));

	return ret;
}

433 434 435 436 437 438 439 440 441 442 443
/**
 * timekeeping_max_deferment - Returns max time the clocksource can be deferred
 *
 * Caller must observe xtime_lock via read_seqbegin/read_seqretry to
 * ensure that the clocksource does not change!
 */
u64 timekeeping_max_deferment(void)
{
	return timekeeper.clock->max_idle_ns;
}

444
/**
445
 * read_persistent_clock -  Return time from the persistent clock.
446 447
 *
 * Weak dummy function for arches that do not yet support it.
448 449
 * Reads the time from the battery backed persistent clock.
 * Returns a timespec with tv_sec=0 and tv_nsec=0 if unsupported.
450 451 452
 *
 *  XXX - Do be sure to remove it once all arches implement it.
 */
453
void __attribute__((weak)) read_persistent_clock(struct timespec *ts)
454
{
455 456
	ts->tv_sec = 0;
	ts->tv_nsec = 0;
457 458
}

459 460 461 462 463 464 465 466 467 468 469 470 471 472 473
/**
 * 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.
 */
void __attribute__((weak)) read_boot_clock(struct timespec *ts)
{
	ts->tv_sec = 0;
	ts->tv_nsec = 0;
}

474 475 476 477 478
/*
 * timekeeping_init - Initializes the clocksource and common timekeeping values
 */
void __init timekeeping_init(void)
{
479
	struct clocksource *clock;
480
	unsigned long flags;
481
	struct timespec now, boot;
482 483

	read_persistent_clock(&now);
484
	read_boot_clock(&boot);
485 486 487

	write_seqlock_irqsave(&xtime_lock, flags);

R
Roman Zippel 已提交
488
	ntp_init();
489

490
	clock = clocksource_default_clock();
491 492
	if (clock->enable)
		clock->enable(clock);
493
	timekeeper_setup_internals(clock);
494

495 496
	xtime.tv_sec = now.tv_sec;
	xtime.tv_nsec = now.tv_nsec;
497 498
	raw_time.tv_sec = 0;
	raw_time.tv_nsec = 0;
499 500 501 502
	if (boot.tv_sec == 0 && boot.tv_nsec == 0) {
		boot.tv_sec = xtime.tv_sec;
		boot.tv_nsec = xtime.tv_nsec;
	}
503
	set_normalized_timespec(&wall_to_monotonic,
504
				-boot.tv_sec, -boot.tv_nsec);
505 506
	total_sleep_time.tv_sec = 0;
	total_sleep_time.tv_nsec = 0;
507 508 509 510
	write_sequnlock_irqrestore(&xtime_lock, flags);
}

/* time in seconds when suspend began */
511
static struct timespec timekeeping_suspend_time;
512 513 514 515 516 517 518 519 520 521 522 523

/**
 * timekeeping_resume - Resumes the generic timekeeping subsystem.
 * @dev:	unused
 *
 * This is for the generic clocksource timekeeping.
 * xtime/wall_to_monotonic/jiffies/etc are
 * still managed by arch specific suspend/resume code.
 */
static int timekeeping_resume(struct sys_device *dev)
{
	unsigned long flags;
524 525 526
	struct timespec ts;

	read_persistent_clock(&ts);
527

528 529
	clocksource_resume();

530 531
	write_seqlock_irqsave(&xtime_lock, flags);

532 533
	if (timespec_compare(&ts, &timekeeping_suspend_time) > 0) {
		ts = timespec_sub(ts, timekeeping_suspend_time);
J
John Stultz 已提交
534
		xtime = timespec_add(xtime, ts);
535
		wall_to_monotonic = timespec_sub(wall_to_monotonic, ts);
J
John Stultz 已提交
536
		total_sleep_time = timespec_add(total_sleep_time, ts);
537 538
	}
	/* re-base the last cycle value */
539 540
	timekeeper.clock->cycle_last = timekeeper.clock->read(timekeeper.clock);
	timekeeper.ntp_error = 0;
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
	timekeeping_suspended = 0;
	write_sequnlock_irqrestore(&xtime_lock, flags);

	touch_softlockup_watchdog();

	clockevents_notify(CLOCK_EVT_NOTIFY_RESUME, NULL);

	/* Resume hrtimers */
	hres_timers_resume();

	return 0;
}

static int timekeeping_suspend(struct sys_device *dev, pm_message_t state)
{
	unsigned long flags;

558
	read_persistent_clock(&timekeeping_suspend_time);
559

560
	write_seqlock_irqsave(&xtime_lock, flags);
561
	timekeeping_forward_now();
562 563 564 565
	timekeeping_suspended = 1;
	write_sequnlock_irqrestore(&xtime_lock, flags);

	clockevents_notify(CLOCK_EVT_NOTIFY_SUSPEND, NULL);
M
Magnus Damm 已提交
566
	clocksource_suspend();
567 568 569 570 571 572

	return 0;
}

/* sysfs resume/suspend bits for timekeeping */
static struct sysdev_class timekeeping_sysclass = {
573
	.name		= "timekeeping",
574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596
	.resume		= timekeeping_resume,
	.suspend	= timekeeping_suspend,
};

static struct sys_device device_timer = {
	.id		= 0,
	.cls		= &timekeeping_sysclass,
};

static int __init timekeeping_init_device(void)
{
	int error = sysdev_class_register(&timekeeping_sysclass);
	if (!error)
		error = sysdev_register(&device_timer);
	return error;
}

device_initcall(timekeeping_init_device);

/*
 * If the error is already larger, we look ahead even further
 * to compensate for late or lost adjustments.
 */
597
static __always_inline int timekeeping_bigadjust(s64 error, s64 *interval,
598 599 600 601 602 603 604 605 606 607 608 609
						 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
610
	 * here.  This is tuned so that an error of about 1 msec is adjusted
611 612
	 * within about 1 sec (or 2^20 nsec in 2^SHIFT_HZ ticks).
	 */
613
	error2 = timekeeper.ntp_error >> (NTP_SCALE_SHIFT + 22 - 2 * SHIFT_HZ);
614 615 616 617 618 619 620 621
	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.
	 */
622
	tick_error = tick_length >> (timekeeper.ntp_error_shift + 1);
623
	tick_error -= timekeeper.xtime_interval >> 1;
624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
	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.
 */
648
static void timekeeping_adjust(s64 offset)
649
{
650
	s64 error, interval = timekeeper.cycle_interval;
651 652
	int adj;

653
	error = timekeeper.ntp_error >> (timekeeper.ntp_error_shift - 1);
654 655 656 657 658
	if (error > interval) {
		error >>= 2;
		if (likely(error <= interval))
			adj = 1;
		else
659
			adj = timekeeping_bigadjust(error, &interval, &offset);
660 661 662 663 664 665 666
	} else if (error < -interval) {
		error >>= 2;
		if (likely(error >= -interval)) {
			adj = -1;
			interval = -interval;
			offset = -offset;
		} else
667
			adj = timekeeping_bigadjust(error, &interval, &offset);
668 669 670
	} else
		return;

671
	timekeeper.mult += adj;
672 673 674
	timekeeper.xtime_interval += interval;
	timekeeper.xtime_nsec -= offset;
	timekeeper.ntp_error -= (interval - offset) <<
675
				timekeeper.ntp_error_shift;
676 677
}

L
Linus Torvalds 已提交
678

679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
/**
 * 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.
 */
static cycle_t logarithmic_accumulation(cycle_t offset, int shift)
{
	u64 nsecps = (u64)NSEC_PER_SEC << timekeeper.shift;

	/* If the offset is smaller then a shifted interval, do nothing */
	if (offset < timekeeper.cycle_interval<<shift)
		return offset;

	/* Accumulate one shifted interval */
	offset -= timekeeper.cycle_interval << shift;
	timekeeper.clock->cycle_last += timekeeper.cycle_interval << shift;

	timekeeper.xtime_nsec += timekeeper.xtime_interval << shift;
	while (timekeeper.xtime_nsec >= nsecps) {
		timekeeper.xtime_nsec -= nsecps;
		xtime.tv_sec++;
		second_overflow();
	}

	/* Accumulate into raw time */
	raw_time.tv_nsec += timekeeper.raw_interval << shift;;
	while (raw_time.tv_nsec >= NSEC_PER_SEC) {
		raw_time.tv_nsec -= NSEC_PER_SEC;
		raw_time.tv_sec++;
	}

	/* Accumulate error between NTP and clock interval */
	timekeeper.ntp_error += tick_length << shift;
	timekeeper.ntp_error -= timekeeper.xtime_interval <<
				(timekeeper.ntp_error_shift + shift);

	return offset;
}

L
Linus Torvalds 已提交
722

723 724 725 726 727 728 729
/**
 * update_wall_time - Uses the current clocksource to increment the wall time
 *
 * Called from the timer interrupt, must hold a write on xtime_lock.
 */
void update_wall_time(void)
{
730
	struct clocksource *clock;
731
	cycle_t offset;
732
	int shift = 0, maxshift;
733 734 735 736 737

	/* Make sure we're fully resumed: */
	if (unlikely(timekeeping_suspended))
		return;

738
	clock = timekeeper.clock;
J
John Stultz 已提交
739 740

#ifdef CONFIG_ARCH_USES_GETTIMEOFFSET
741
	offset = timekeeper.cycle_interval;
J
John Stultz 已提交
742 743
#else
	offset = (clock->read(clock) - clock->cycle_last) & clock->mask;
744
#endif
745
	timekeeper.xtime_nsec = (s64)xtime.tv_nsec << timekeeper.shift;
746

747 748 749 750 751 752 753
	/*
	 * 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
	 * that is smaller then the offset. We then accumulate that
	 * chunk in one go, and then try to consume the next smaller
	 * doubled multiple.
754
	 */
755 756 757 758 759
	shift = ilog2(offset) - ilog2(timekeeper.cycle_interval);
	shift = max(0, shift);
	/* Bound shift to one less then what overflows tick_length */
	maxshift = (8*sizeof(tick_length) - (ilog2(tick_length)+1)) - 1;
	shift = min(shift, maxshift);
760
	while (offset >= timekeeper.cycle_interval) {
761
		offset = logarithmic_accumulation(offset, shift);
762 763
		if(offset < timekeeper.cycle_interval<<shift)
			shift--;
764 765 766
	}

	/* correct the clock when NTP error is too big */
767
	timekeeping_adjust(offset);
768

769 770 771 772
	/*
	 * Since in the loop above, we accumulate any amount of time
	 * in xtime_nsec over a second into xtime.tv_sec, its possible for
	 * xtime_nsec to be fairly small after the loop. Further, if we're
773
	 * slightly speeding the clocksource up in timekeeping_adjust(),
774 775 776 777 778 779 780 781 782 783 784
	 * its possible the required corrective factor to xtime_nsec could
	 * cause it to underflow.
	 *
	 * Now, we 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.
	 */
785 786 787
	if (unlikely((s64)timekeeper.xtime_nsec < 0)) {
		s64 neg = -(s64)timekeeper.xtime_nsec;
		timekeeper.xtime_nsec = 0;
788
		timekeeper.ntp_error += neg << timekeeper.ntp_error_shift;
789 790
	}

J
John Stultz 已提交
791 792 793

	/*
	 * Store full nanoseconds into xtime after rounding it up and
794 795
	 * add the remainder to the error difference.
	 */
796 797 798 799
	xtime.tv_nsec =	((s64) timekeeper.xtime_nsec >> timekeeper.shift) + 1;
	timekeeper.xtime_nsec -= (s64) xtime.tv_nsec << timekeeper.shift;
	timekeeper.ntp_error +=	timekeeper.xtime_nsec <<
				timekeeper.ntp_error_shift;
800

J
John Stultz 已提交
801 802 803 804 805 806 807 808 809
	/*
	 * Finally, make sure that after the rounding
	 * xtime.tv_nsec isn't larger then NSEC_PER_SEC
	 */
	if (unlikely(xtime.tv_nsec >= NSEC_PER_SEC)) {
		xtime.tv_nsec -= NSEC_PER_SEC;
		xtime.tv_sec++;
		second_overflow();
	}
L
Linus Torvalds 已提交
810

811
	/* check to see if there is a new clocksource to use */
812
	update_vsyscall(&xtime, timekeeper.clock, timekeeper.mult);
813
}
T
Tomas Janousek 已提交
814 815 816 817 818 819 820 821 822 823 824 825 826 827

/**
 * getboottime - Return the real time of system boot.
 * @ts:		pointer to the timespec to be set
 *
 * Returns the time of day in a timespec.
 *
 * 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)
{
828 829 830 831
	struct timespec boottime = {
		.tv_sec = wall_to_monotonic.tv_sec + total_sleep_time.tv_sec,
		.tv_nsec = wall_to_monotonic.tv_nsec + total_sleep_time.tv_nsec
	};
832 833

	set_normalized_timespec(ts, -boottime.tv_sec, -boottime.tv_nsec);
T
Tomas Janousek 已提交
834
}
835
EXPORT_SYMBOL_GPL(getboottime);
T
Tomas Janousek 已提交
836 837 838 839 840 841 842

/**
 * 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)
{
J
John Stultz 已提交
843
	*ts = timespec_add(*ts, total_sleep_time);
T
Tomas Janousek 已提交
844
}
845
EXPORT_SYMBOL_GPL(monotonic_to_bootbased);
846

847 848
unsigned long get_seconds(void)
{
J
John Stultz 已提交
849
	return xtime.tv_sec;
850 851 852
}
EXPORT_SYMBOL(get_seconds);

853 854
struct timespec __current_kernel_time(void)
{
J
John Stultz 已提交
855
	return xtime;
856
}
857

858 859 860 861 862 863 864
struct timespec current_kernel_time(void)
{
	struct timespec now;
	unsigned long seq;

	do {
		seq = read_seqbegin(&xtime_lock);
L
Linus Torvalds 已提交
865

J
John Stultz 已提交
866
		now = xtime;
867 868 869 870 871
	} while (read_seqretry(&xtime_lock, seq));

	return now;
}
EXPORT_SYMBOL(current_kernel_time);
872 873 874 875 876 877 878 879

struct timespec get_monotonic_coarse(void)
{
	struct timespec now, mono;
	unsigned long seq;

	do {
		seq = read_seqbegin(&xtime_lock);
L
Linus Torvalds 已提交
880

J
John Stultz 已提交
881
		now = xtime;
882 883 884 885 886 887 888
		mono = wall_to_monotonic;
	} while (read_seqretry(&xtime_lock, seq));

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