time.c 34.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10
/*
 *  linux/arch/x86-64/kernel/time.c
 *
 *  "High Precision Event Timer" based timekeeping.
 *
 *  Copyright (c) 1991,1992,1995  Linus Torvalds
 *  Copyright (c) 1994  Alan Modra
 *  Copyright (c) 1995  Markus Kuhn
 *  Copyright (c) 1996  Ingo Molnar
 *  Copyright (c) 1998  Andrea Arcangeli
11
 *  Copyright (c) 2002,2006  Vojtech Pavlik
L
Linus Torvalds 已提交
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
 *  Copyright (c) 2003  Andi Kleen
 *  RTC support code taken from arch/i386/kernel/timers/time_hpet.c
 */

#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/mc146818rtc.h>
#include <linux/time.h>
#include <linux/ioport.h>
#include <linux/module.h>
#include <linux/device.h>
#include <linux/sysdev.h>
#include <linux/bcd.h>
27 28
#include <linux/notifier.h>
#include <linux/cpu.h>
L
Linus Torvalds 已提交
29
#include <linux/kallsyms.h>
A
Andi Kleen 已提交
30
#include <linux/acpi.h>
31
#ifdef CONFIG_ACPI
A
Andi Kleen 已提交
32
#include <acpi/achware.h>	/* for PM timer frequency */
33
#include <acpi/acpi_bus.h>
34
#endif
L
Linus Torvalds 已提交
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
#include <asm/8253pit.h>
#include <asm/pgtable.h>
#include <asm/vsyscall.h>
#include <asm/timex.h>
#include <asm/proto.h>
#include <asm/hpet.h>
#include <asm/sections.h>
#include <linux/cpufreq.h>
#include <linux/hpet.h>
#include <asm/apic.h>

#ifdef CONFIG_CPU_FREQ
static void cpufreq_delayed_get(void);
#endif
extern void i8254_timer_resume(void);
extern int using_apic_timer;

52
static char *timename = NULL;
53

L
Linus Torvalds 已提交
54
DEFINE_SPINLOCK(rtc_lock);
55
EXPORT_SYMBOL(rtc_lock);
L
Linus Torvalds 已提交
56 57
DEFINE_SPINLOCK(i8253_lock);

58
int nohpet __initdata = 0;
L
Linus Torvalds 已提交
59 60
static int notsc __initdata = 0;

61 62 63 64 65 66 67
#define USEC_PER_TICK (USEC_PER_SEC / HZ)
#define NSEC_PER_TICK (NSEC_PER_SEC / HZ)
#define FSEC_PER_TICK (FSEC_PER_SEC / HZ)

#define NS_SCALE	10 /* 2^10, carefully chosen */
#define US_SCALE	32 /* 2^32, arbitralrily chosen */

L
Linus Torvalds 已提交
68
unsigned int cpu_khz;					/* TSC clocks / usec, not used here */
69
EXPORT_SYMBOL(cpu_khz);
L
Linus Torvalds 已提交
70 71
static unsigned long hpet_period;			/* fsecs / HPET clock */
unsigned long hpet_tick;				/* HPET clocks / interrupt */
72
int hpet_use_timer;				/* Use counter of hpet for time keeping, otherwise PIT */
L
Linus Torvalds 已提交
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
unsigned long vxtime_hz = PIT_TICK_RATE;
int report_lost_ticks;				/* command line option */
unsigned long long monotonic_base;

struct vxtime_data __vxtime __section_vxtime;	/* for vsyscalls */

volatile unsigned long __jiffies __section_jiffies = INITIAL_JIFFIES;
struct timespec __xtime __section_xtime;
struct timezone __sys_tz __section_sys_tz;

/*
 * do_gettimeoffset() returns microseconds since last timer interrupt was
 * triggered by hardware. A memory read of HPET is slower than a register read
 * of TSC, but much more reliable. It's also synchronized to the timer
 * interrupt. Note that do_gettimeoffset() may return more than hpet_tick, if a
 * timer interrupt has happened already, but vxtime.trigger wasn't updated yet.
 * This is not a problem, because jiffies hasn't updated either. They are bound
 * together by xtime_lock.
 */

static inline unsigned int do_gettimeoffset_tsc(void)
{
	unsigned long t;
	unsigned long x;
97
	t = get_cycles_sync();
98 99
	if (t < vxtime.last_tsc) 
		t = vxtime.last_tsc; /* hack */
100
	x = ((t - vxtime.last_tsc) * vxtime.tsc_quot) >> US_SCALE;
L
Linus Torvalds 已提交
101 102 103 104 105
	return x;
}

static inline unsigned int do_gettimeoffset_hpet(void)
{
106 107
	/* cap counter read to one tick to avoid inconsistencies */
	unsigned long counter = hpet_readl(HPET_COUNTER) - vxtime.last;
108
	return (min(counter,hpet_tick) * vxtime.quot) >> US_SCALE;
L
Linus Torvalds 已提交
109 110 111 112 113 114 115 116 117 118 119 120
}

unsigned int (*do_gettimeoffset)(void) = do_gettimeoffset_tsc;

/*
 * This version of gettimeofday() has microsecond resolution and better than
 * microsecond precision, as we're using at least a 10 MHz (usually 14.31818
 * MHz) HPET timer.
 */

void do_gettimeofday(struct timeval *tv)
{
A
Atsushi Nemoto 已提交
121
	unsigned long seq;
L
Linus Torvalds 已提交
122 123 124 125 126 127
 	unsigned int sec, usec;

	do {
		seq = read_seqbegin(&xtime_lock);

		sec = xtime.tv_sec;
128
		usec = xtime.tv_nsec / NSEC_PER_USEC;
L
Linus Torvalds 已提交
129 130 131 132 133 134 135 136 137

		/* i386 does some correction here to keep the clock 
		   monotonous even when ntpd is fixing drift.
		   But they didn't work for me, there is a non monotonic
		   clock anyways with ntp.
		   I dropped all corrections now until a real solution can
		   be found. Note when you fix it here you need to do the same
		   in arch/x86_64/kernel/vsyscall.c and export all needed
		   variables in vmlinux.lds. -AK */ 
A
Atsushi Nemoto 已提交
138
		usec += do_gettimeoffset();
L
Linus Torvalds 已提交
139 140 141

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

142 143
	tv->tv_sec = sec + usec / USEC_PER_SEC;
	tv->tv_usec = usec % USEC_PER_SEC;
L
Linus Torvalds 已提交
144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
}

EXPORT_SYMBOL(do_gettimeofday);

/*
 * settimeofday() first undoes the correction that gettimeofday would do
 * on the time, and then saves it. This is ugly, but has been like this for
 * ages already.
 */

int do_settimeofday(struct timespec *tv)
{
	time_t wtm_sec, sec = tv->tv_sec;
	long wtm_nsec, nsec = tv->tv_nsec;

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

	write_seqlock_irq(&xtime_lock);

A
Atsushi Nemoto 已提交
164
	nsec -= do_gettimeoffset() * NSEC_PER_USEC;
L
Linus Torvalds 已提交
165 166 167 168 169 170 171

	wtm_sec  = wall_to_monotonic.tv_sec + (xtime.tv_sec - sec);
	wtm_nsec = wall_to_monotonic.tv_nsec + (xtime.tv_nsec - nsec);

	set_normalized_timespec(&xtime, sec, nsec);
	set_normalized_timespec(&wall_to_monotonic, wtm_sec, wtm_nsec);

J
john stultz 已提交
172
	ntp_clear();
L
Linus Torvalds 已提交
173 174 175 176 177 178 179 180 181 182 183 184

	write_sequnlock_irq(&xtime_lock);
	clock_was_set();
	return 0;
}

EXPORT_SYMBOL(do_settimeofday);

unsigned long profile_pc(struct pt_regs *regs)
{
	unsigned long pc = instruction_pointer(regs);

185 186 187
	/* Assume the lock function has either no stack frame or a copy
	   of eflags from PUSHF
	   Eflags always has bits 22 and up cleared unlike kernel addresses. */
188
	if (!user_mode(regs) && in_lock_functions(pc)) {
189 190 191 192 193
		unsigned long *sp = (unsigned long *)regs->rsp;
		if (sp[0] >> 22)
			return sp[0];
		if (sp[1] >> 22)
			return sp[1];
L
Linus Torvalds 已提交
194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
	}
	return pc;
}
EXPORT_SYMBOL(profile_pc);

/*
 * In order to set the CMOS clock precisely, set_rtc_mmss has to be called 500
 * ms after the second nowtime has started, because when nowtime is written
 * into the registers of the CMOS clock, it will jump to the next second
 * precisely 500 ms later. Check the Motorola MC146818A or Dallas DS12887 data
 * sheet for details.
 */

static void set_rtc_mmss(unsigned long nowtime)
{
	int real_seconds, real_minutes, cmos_minutes;
	unsigned char control, freq_select;

/*
 * IRQs are disabled when we're called from the timer interrupt,
 * no need for spin_lock_irqsave()
 */

	spin_lock(&rtc_lock);

/*
 * Tell the clock it's being set and stop it.
 */

	control = CMOS_READ(RTC_CONTROL);
	CMOS_WRITE(control | RTC_SET, RTC_CONTROL);

	freq_select = CMOS_READ(RTC_FREQ_SELECT);
	CMOS_WRITE(freq_select | RTC_DIV_RESET2, RTC_FREQ_SELECT);

	cmos_minutes = CMOS_READ(RTC_MINUTES);
		BCD_TO_BIN(cmos_minutes);

/*
 * since we're only adjusting minutes and seconds, don't interfere with hour
 * overflow. This avoids messing with unknown time zones but requires your RTC
 * not to be off by more than 15 minutes. Since we're calling it only when
 * our clock is externally synchronized using NTP, this shouldn't be a problem.
 */

	real_seconds = nowtime % 60;
	real_minutes = nowtime / 60;
	if (((abs(real_minutes - cmos_minutes) + 15) / 30) & 1)
		real_minutes += 30;		/* correct for half hour time zone */
	real_minutes %= 60;

	if (abs(real_minutes - cmos_minutes) >= 30) {
		printk(KERN_WARNING "time.c: can't update CMOS clock "
		       "from %d to %d\n", cmos_minutes, real_minutes);
248
	} else {
249 250
		BIN_TO_BCD(real_seconds);
		BIN_TO_BCD(real_minutes);
L
Linus Torvalds 已提交
251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
		CMOS_WRITE(real_seconds, RTC_SECONDS);
		CMOS_WRITE(real_minutes, RTC_MINUTES);
	}

/*
 * The following flags have to be released exactly in this order, otherwise the
 * DS12887 (popular MC146818A clone with integrated battery and quartz) will
 * not reset the oscillator and will not update precisely 500 ms later. You
 * won't find this mentioned in the Dallas Semiconductor data sheets, but who
 * believes data sheets anyway ... -- Markus Kuhn
 */

	CMOS_WRITE(control, RTC_CONTROL);
	CMOS_WRITE(freq_select, RTC_FREQ_SELECT);

	spin_unlock(&rtc_lock);
}


/* monotonic_clock(): returns # of nanoseconds passed since time_init()
 *		Note: This function is required to return accurate
 *		time even in the absence of multiple timer ticks.
 */
274
static inline unsigned long long cycles_2_ns(unsigned long long cyc);
L
Linus Torvalds 已提交
275 276 277 278 279 280 281 282 283 284 285 286
unsigned long long monotonic_clock(void)
{
	unsigned long seq;
 	u32 last_offset, this_offset, offset;
	unsigned long long base;

	if (vxtime.mode == VXTIME_HPET) {
		do {
			seq = read_seqbegin(&xtime_lock);

			last_offset = vxtime.last;
			base = monotonic_base;
287
			this_offset = hpet_readl(HPET_COUNTER);
L
Linus Torvalds 已提交
288 289
		} while (read_seqretry(&xtime_lock, seq));
		offset = (this_offset - last_offset);
290
		offset *= NSEC_PER_TICK / hpet_tick;
291
	} else {
L
Linus Torvalds 已提交
292 293 294 295 296 297
		do {
			seq = read_seqbegin(&xtime_lock);

			last_offset = vxtime.last_tsc;
			base = monotonic_base;
		} while (read_seqretry(&xtime_lock, seq));
298
		this_offset = get_cycles_sync();
299
		offset = cycles_2_ns(this_offset - last_offset);
L
Linus Torvalds 已提交
300
	}
301
	return base + offset;
L
Linus Torvalds 已提交
302 303 304 305 306
}
EXPORT_SYMBOL(monotonic_clock);

static noinline void handle_lost_ticks(int lost, struct pt_regs *regs)
{
307 308 309 310 311 312 313 314 315 316
	static long lost_count;
	static int warned;
	if (report_lost_ticks) {
		printk(KERN_WARNING "time.c: Lost %d timer tick(s)! ", lost);
		print_symbol("rip %s)\n", regs->rip);
	}

	if (lost_count == 1000 && !warned) {
		printk(KERN_WARNING "warning: many lost ticks.\n"
		       KERN_WARNING "Your time source seems to be instable or "
L
Linus Torvalds 已提交
317
		   		"some driver is hogging interupts\n");
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332
		print_symbol("rip %s\n", regs->rip);
		if (vxtime.mode == VXTIME_TSC && vxtime.hpet_address) {
			printk(KERN_WARNING "Falling back to HPET\n");
			if (hpet_use_timer)
				vxtime.last = hpet_readl(HPET_T0_CMP) - 
							hpet_tick;
			else
				vxtime.last = hpet_readl(HPET_COUNTER);
			vxtime.mode = VXTIME_HPET;
			do_gettimeoffset = do_gettimeoffset_hpet;
		}
		/* else should fall back to PIT, but code missing. */
		warned = 1;
	} else
		lost_count++;
L
Linus Torvalds 已提交
333 334

#ifdef CONFIG_CPU_FREQ
335 336 337 338
	/* In some cases the CPU can change frequency without us noticing
	   Give cpufreq a change to catch up. */
	if ((lost_count+1) % 25 == 0)
		cpufreq_delayed_get();
L
Linus Torvalds 已提交
339 340 341
#endif
}

342
void main_timer_handler(struct pt_regs *regs)
L
Linus Torvalds 已提交
343 344 345
{
	static unsigned long rtc_update = 0;
	unsigned long tsc;
346
	int delay = 0, offset = 0, lost = 0;
L
Linus Torvalds 已提交
347 348 349 350 351 352 353 354 355 356

/*
 * Here we are in the timer irq handler. We have irqs locally disabled (so we
 * don't need spin_lock_irqsave()) but we don't know if the timer_bh is running
 * on the other CPU, so we need a lock. We also need to lock the vsyscall
 * variables, because both do_timer() and us change them -arca+vojtech
 */

	write_seqlock(&xtime_lock);

357 358 359 360 361 362 363 364
	if (vxtime.hpet_address)
		offset = hpet_readl(HPET_COUNTER);

	if (hpet_use_timer) {
		/* if we're using the hpet timer functionality,
		 * we can more accurately know the counter value
		 * when the timer interrupt occured.
		 */
L
Linus Torvalds 已提交
365 366
		offset = hpet_readl(HPET_T0_CMP) - hpet_tick;
		delay = hpet_readl(HPET_COUNTER) - offset;
367
	} else if (!pmtmr_ioport) {
L
Linus Torvalds 已提交
368 369 370 371 372 373 374 375
		spin_lock(&i8253_lock);
		outb_p(0x00, 0x43);
		delay = inb_p(0x40);
		delay |= inb(0x40) << 8;
		spin_unlock(&i8253_lock);
		delay = LATCH - 1 - delay;
	}

376
	tsc = get_cycles_sync();
L
Linus Torvalds 已提交
377 378 379 380 381 382 383

	if (vxtime.mode == VXTIME_HPET) {
		if (offset - vxtime.last > hpet_tick) {
			lost = (offset - vxtime.last) / hpet_tick - 1;
		}

		monotonic_base += 
384
			(offset - vxtime.last) * NSEC_PER_TICK / hpet_tick;
L
Linus Torvalds 已提交
385 386

		vxtime.last = offset;
A
Andi Kleen 已提交
387 388 389 390
#ifdef CONFIG_X86_PM_TIMER
	} else if (vxtime.mode == VXTIME_PMTMR) {
		lost = pmtimer_mark_offset();
#endif
L
Linus Torvalds 已提交
391 392
	} else {
		offset = (((tsc - vxtime.last_tsc) *
393
			   vxtime.tsc_quot) >> US_SCALE) - USEC_PER_TICK;
L
Linus Torvalds 已提交
394 395 396 397

		if (offset < 0)
			offset = 0;

398 399 400
		if (offset > USEC_PER_TICK) {
			lost = offset / USEC_PER_TICK;
			offset %= USEC_PER_TICK;
L
Linus Torvalds 已提交
401 402
		}

403
		monotonic_base += cycles_2_ns(tsc - vxtime.last_tsc);
L
Linus Torvalds 已提交
404 405 406 407

		vxtime.last_tsc = tsc - vxtime.quot * delay / vxtime.tsc_quot;

		if ((((tsc - vxtime.last_tsc) *
408
		      vxtime.tsc_quot) >> US_SCALE) < offset)
L
Linus Torvalds 已提交
409
			vxtime.last_tsc = tsc -
410
				(((long) offset << US_SCALE) / vxtime.tsc_quot) - 1;
L
Linus Torvalds 已提交
411 412
	}

413
	if (lost > 0)
L
Linus Torvalds 已提交
414
		handle_lost_ticks(lost, regs);
415 416
	else
		lost = 0;
L
Linus Torvalds 已提交
417 418 419 420 421

/*
 * Do the timer stuff.
 */

422
	do_timer(lost + 1);
L
Linus Torvalds 已提交
423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
#ifndef CONFIG_SMP
	update_process_times(user_mode(regs));
#endif

/*
 * In the SMP case we use the local APIC timer interrupt to do the profiling,
 * except when we simulate SMP mode on a uniprocessor system, in that case we
 * have to call the local interrupt handler.
 */

	if (!using_apic_timer)
		smp_local_timer_interrupt(regs);

/*
 * If we have an externally synchronized Linux clock, then update CMOS clock
 * accordingly every ~11 minutes. set_rtc_mmss() will be called in the jiffy
 * closest to exactly 500 ms before the next second. If the update fails, we
 * don't care, as it'll be updated on the next turn, and the problem (time way
 * off) isn't likely to go away much sooner anyway.
 */

J
john stultz 已提交
444
	if (ntp_synced() && xtime.tv_sec > rtc_update &&
L
Linus Torvalds 已提交
445 446 447 448 449 450
		abs(xtime.tv_nsec - 500000000) <= tick_nsec / 2) {
		set_rtc_mmss(xtime.tv_sec);
		rtc_update = xtime.tv_sec + 660;
	}
 
	write_sequnlock(&xtime_lock);
451
}
L
Linus Torvalds 已提交
452

453 454 455 456 457
static irqreturn_t timer_interrupt(int irq, void *dev_id, struct pt_regs *regs)
{
	if (apic_runs_main_timer > 1)
		return IRQ_HANDLED;
	main_timer_handler(regs);
458 459
	if (using_apic_timer)
		smp_send_timer_broadcast_ipi();
L
Linus Torvalds 已提交
460 461 462
	return IRQ_HANDLED;
}

463
static unsigned int cyc2ns_scale __read_mostly;
L
Linus Torvalds 已提交
464

465
static inline void set_cyc2ns_scale(unsigned long cpu_khz)
L
Linus Torvalds 已提交
466
{
467
	cyc2ns_scale = (NSEC_PER_MSEC << NS_SCALE) / cpu_khz;
L
Linus Torvalds 已提交
468 469 470 471
}

static inline unsigned long long cycles_2_ns(unsigned long long cyc)
{
472
	return (cyc * cyc2ns_scale) >> NS_SCALE;
L
Linus Torvalds 已提交
473 474 475 476 477 478 479 480 481 482 483 484
}

unsigned long long sched_clock(void)
{
	unsigned long a = 0;

#if 0
	/* Don't do a HPET read here. Using TSC always is much faster
	   and HPET may not be mapped yet when the scheduler first runs.
           Disadvantage is a small drift between CPUs in some configurations,
	   but that should be tolerable. */
	if (__vxtime.mode == VXTIME_HPET)
485
		return (hpet_readl(HPET_COUNTER) * vxtime.quot) >> US_SCALE;
L
Linus Torvalds 已提交
486 487 488 489 490 491 492 493 494 495 496
#endif

	/* Could do CPU core sync here. Opteron can execute rdtsc speculatively,
	   which means it is not completely exact and may not be monotonous between
	   CPUs. But the errors should be too small to matter for scheduling
	   purposes. */

	rdtscll(a);
	return cycles_2_ns(a);
}

497
static unsigned long get_cmos_time(void)
L
Linus Torvalds 已提交
498
{
499
	unsigned int year, mon, day, hour, min, sec;
L
Linus Torvalds 已提交
500
	unsigned long flags;
501
	unsigned extyear = 0;
L
Linus Torvalds 已提交
502 503 504

	spin_lock_irqsave(&rtc_lock, flags);

505 506 507 508 509 510 511
	do {
		sec = CMOS_READ(RTC_SECONDS);
		min = CMOS_READ(RTC_MINUTES);
		hour = CMOS_READ(RTC_HOURS);
		day = CMOS_READ(RTC_DAY_OF_MONTH);
		mon = CMOS_READ(RTC_MONTH);
		year = CMOS_READ(RTC_YEAR);
512
#ifdef CONFIG_ACPI
513 514 515
		if (acpi_fadt.revision >= FADT2_REVISION_ID &&
					acpi_fadt.century)
			extyear = CMOS_READ(acpi_fadt.century);
516
#endif
517
	} while (sec != CMOS_READ(RTC_SECONDS));
518

L
Linus Torvalds 已提交
519 520
	spin_unlock_irqrestore(&rtc_lock, flags);

521 522 523
	/*
	 * We know that x86-64 always uses BCD format, no need to check the
	 * config register.
524
 	 */
L
Linus Torvalds 已提交
525

526 527 528 529 530 531
	BCD_TO_BIN(sec);
	BCD_TO_BIN(min);
	BCD_TO_BIN(hour);
	BCD_TO_BIN(day);
	BCD_TO_BIN(mon);
	BCD_TO_BIN(year);
L
Linus Torvalds 已提交
532

533 534 535 536 537 538 539 540 541 542 543
	if (extyear) {
		BCD_TO_BIN(extyear);
		year += extyear;
		printk(KERN_INFO "Extended CMOS year: %d\n", extyear);
	} else { 
		/*
		 * x86-64 systems only exists since 2002.
		 * This will work up to Dec 31, 2100
	 	 */
		year += 2000;
	}
L
Linus Torvalds 已提交
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

	return mktime(year, mon, day, hour, min, sec);
}

#ifdef CONFIG_CPU_FREQ

/* Frequency scaling support. Adjust the TSC based timer when the cpu frequency
   changes.
   
   RED-PEN: On SMP we assume all CPUs run with the same frequency.  It's
   not that important because current Opteron setups do not support
   scaling on SMP anyroads.

   Should fix up last_tsc too. Currently gettimeofday in the
   first tick after the change will be slightly wrong. */

#include <linux/workqueue.h>

static unsigned int cpufreq_delayed_issched = 0;
static unsigned int cpufreq_init = 0;
static struct work_struct cpufreq_delayed_get_work;

static void handle_cpufreq_delayed_get(void *v)
{
	unsigned int cpu;
	for_each_online_cpu(cpu) {
		cpufreq_get(cpu);
	}
	cpufreq_delayed_issched = 0;
}

/* if we notice lost ticks, schedule a call to cpufreq_get() as it tries
 * to verify the CPU frequency the timing core thinks the CPU is running
 * at is still correct.
 */
static void cpufreq_delayed_get(void)
{
	static int warned;
	if (cpufreq_init && !cpufreq_delayed_issched) {
		cpufreq_delayed_issched = 1;
		if (!warned) {
			warned = 1;
586 587
			printk(KERN_DEBUG 
	"Losing some ticks... checking if CPU frequency changed.\n");
L
Linus Torvalds 已提交
588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603
		}
		schedule_work(&cpufreq_delayed_get_work);
	}
}

static unsigned int  ref_freq = 0;
static unsigned long loops_per_jiffy_ref = 0;

static unsigned long cpu_khz_ref = 0;

static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
				 void *data)
{
        struct cpufreq_freqs *freq = data;
	unsigned long *lpj, dummy;

604 605 606
	if (cpu_has(&cpu_data[freq->cpu], X86_FEATURE_CONSTANT_TSC))
		return 0;

L
Linus Torvalds 已提交
607 608 609
	lpj = &dummy;
	if (!(freq->flags & CPUFREQ_CONST_LOOPS))
#ifdef CONFIG_SMP
610
		lpj = &cpu_data[freq->cpu].loops_per_jiffy;
L
Linus Torvalds 已提交
611
#else
612
		lpj = &boot_cpu_data.loops_per_jiffy;
L
Linus Torvalds 已提交
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
#endif

	if (!ref_freq) {
		ref_freq = freq->old;
		loops_per_jiffy_ref = *lpj;
		cpu_khz_ref = cpu_khz;
	}
        if ((val == CPUFREQ_PRECHANGE  && freq->old < freq->new) ||
            (val == CPUFREQ_POSTCHANGE && freq->old > freq->new) ||
	    (val == CPUFREQ_RESUMECHANGE)) {
                *lpj =
		cpufreq_scale(loops_per_jiffy_ref, ref_freq, freq->new);

		cpu_khz = cpufreq_scale(cpu_khz_ref, ref_freq, freq->new);
		if (!(freq->flags & CPUFREQ_CONST_LOOPS))
628
			vxtime.tsc_quot = (USEC_PER_MSEC << US_SCALE) / cpu_khz;
L
Linus Torvalds 已提交
629 630
	}
	
631
	set_cyc2ns_scale(cpu_khz_ref);
L
Linus Torvalds 已提交
632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674

	return 0;
}
 
static struct notifier_block time_cpufreq_notifier_block = {
         .notifier_call  = time_cpufreq_notifier
};

static int __init cpufreq_tsc(void)
{
	INIT_WORK(&cpufreq_delayed_get_work, handle_cpufreq_delayed_get, NULL);
	if (!cpufreq_register_notifier(&time_cpufreq_notifier_block,
				       CPUFREQ_TRANSITION_NOTIFIER))
		cpufreq_init = 1;
	return 0;
}

core_initcall(cpufreq_tsc);

#endif

/*
 * calibrate_tsc() calibrates the processor TSC in a very simple way, comparing
 * it to the HPET timer of known frequency.
 */

#define TICK_COUNT 100000000

static unsigned int __init hpet_calibrate_tsc(void)
{
	int tsc_start, hpet_start;
	int tsc_now, hpet_now;
	unsigned long flags;

	local_irq_save(flags);
	local_irq_disable();

	hpet_start = hpet_readl(HPET_COUNTER);
	rdtscl(tsc_start);

	do {
		local_irq_disable();
		hpet_now = hpet_readl(HPET_COUNTER);
675
		tsc_now = get_cycles_sync();
L
Linus Torvalds 已提交
676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
		local_irq_restore(flags);
	} while ((tsc_now - tsc_start) < TICK_COUNT &&
		 (hpet_now - hpet_start) < TICK_COUNT);

	return (tsc_now - tsc_start) * 1000000000L
		/ ((hpet_now - hpet_start) * hpet_period / 1000);
}


/*
 * pit_calibrate_tsc() uses the speaker output (channel 2) of
 * the PIT. This is better than using the timer interrupt output,
 * because we can read the value of the speaker with just one inb(),
 * where we need three i/o operations for the interrupt channel.
 * We count how many ticks the TSC does in 50 ms.
 */

static unsigned int __init pit_calibrate_tsc(void)
{
	unsigned long start, end;
	unsigned long flags;

	spin_lock_irqsave(&i8253_lock, flags);

	outb((inb(0x61) & ~0x02) | 0x01, 0x61);

	outb(0xb0, 0x43);
	outb((PIT_TICK_RATE / (1000 / 50)) & 0xff, 0x42);
	outb((PIT_TICK_RATE / (1000 / 50)) >> 8, 0x42);
705
	start = get_cycles_sync();
L
Linus Torvalds 已提交
706
	while ((inb(0x61) & 0x20) == 0);
707
	end = get_cycles_sync();
L
Linus Torvalds 已提交
708 709 710 711 712 713 714 715 716 717 718 719 720

	spin_unlock_irqrestore(&i8253_lock, flags);
	
	return (end - start) / 50;
}

#ifdef	CONFIG_HPET
static __init int late_hpet_init(void)
{
	struct hpet_data	hd;
	unsigned int 		ntimer;

	if (!vxtime.hpet_address)
721
        	return 0;
L
Linus Torvalds 已提交
722 723 724 725 726 727 728 729 730 731 732 733

	memset(&hd, 0, sizeof (hd));

	ntimer = hpet_readl(HPET_ID);
	ntimer = (ntimer & HPET_ID_NUMBER) >> HPET_ID_NUMBER_SHIFT;
	ntimer++;

	/*
	 * Register with driver.
	 * Timer0 and Timer1 is used by platform.
	 */
	hd.hd_phys_address = vxtime.hpet_address;
734
	hd.hd_address = (void __iomem *)fix_to_virt(FIX_HPET_BASE);
L
Linus Torvalds 已提交
735 736 737 738 739 740 741 742 743 744 745 746 747 748
	hd.hd_nirqs = ntimer;
	hd.hd_flags = HPET_DATA_PLATFORM;
	hpet_reserve_timer(&hd, 0);
#ifdef	CONFIG_HPET_EMULATE_RTC
	hpet_reserve_timer(&hd, 1);
#endif
	hd.hd_irq[0] = HPET_LEGACY_8254;
	hd.hd_irq[1] = HPET_LEGACY_RTC;
	if (ntimer > 2) {
		struct hpet		*hpet;
		struct hpet_timer	*timer;
		int			i;

		hpet = (struct hpet *) fix_to_virt(FIX_HPET_BASE);
749 750
		timer = &hpet->hpet_timers[2];
		for (i = 2; i < ntimer; timer++, i++)
L
Linus Torvalds 已提交
751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780
			hd.hd_irq[i] = (timer->hpet_config &
					Tn_INT_ROUTE_CNF_MASK) >>
				Tn_INT_ROUTE_CNF_SHIFT;

	}

	hpet_alloc(&hd);
	return 0;
}
fs_initcall(late_hpet_init);
#endif

static int hpet_timer_stop_set_go(unsigned long tick)
{
	unsigned int cfg;

/*
 * Stop the timers and reset the main counter.
 */

	cfg = hpet_readl(HPET_CFG);
	cfg &= ~(HPET_CFG_ENABLE | HPET_CFG_LEGACY);
	hpet_writel(cfg, HPET_CFG);
	hpet_writel(0, HPET_COUNTER);
	hpet_writel(0, HPET_COUNTER + 4);

/*
 * Set up timer 0, as periodic with first interrupt to happen at hpet_tick,
 * and period also hpet_tick.
 */
781 782
	if (hpet_use_timer) {
		hpet_writel(HPET_TN_ENABLE | HPET_TN_PERIODIC | HPET_TN_SETVAL |
L
Linus Torvalds 已提交
783
		    HPET_TN_32BIT, HPET_T0_CFG);
784 785
		hpet_writel(hpet_tick, HPET_T0_CMP); /* next interrupt */
		hpet_writel(hpet_tick, HPET_T0_CMP); /* period */
786 787
		cfg |= HPET_CFG_LEGACY;
	}
L
Linus Torvalds 已提交
788 789 790 791
/*
 * Go!
 */

792
	cfg |= HPET_CFG_ENABLE;
L
Linus Torvalds 已提交
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
	hpet_writel(cfg, HPET_CFG);

	return 0;
}

static int hpet_init(void)
{
	unsigned int id;

	if (!vxtime.hpet_address)
		return -1;
	set_fixmap_nocache(FIX_HPET_BASE, vxtime.hpet_address);
	__set_fixmap(VSYSCALL_HPET, vxtime.hpet_address, PAGE_KERNEL_VSYSCALL_NOCACHE);

/*
 * Read the period, compute tick and quotient.
 */

	id = hpet_readl(HPET_ID);

813
	if (!(id & HPET_ID_VENDOR) || !(id & HPET_ID_NUMBER))
L
Linus Torvalds 已提交
814 815 816 817 818 819
		return -1;

	hpet_period = hpet_readl(HPET_PERIOD);
	if (hpet_period < 100000 || hpet_period > 100000000)
		return -1;

820
	hpet_tick = (FSEC_PER_TICK + hpet_period / 2) / hpet_period;
L
Linus Torvalds 已提交
821

822 823
	hpet_use_timer = (id & HPET_ID_LEGSUP);

L
Linus Torvalds 已提交
824 825 826 827 828 829 830 831
	return hpet_timer_stop_set_go(hpet_tick);
}

static int hpet_reenable(void)
{
	return hpet_timer_stop_set_go(hpet_tick);
}

832 833 834 835
#define PIT_MODE 0x43
#define PIT_CH0  0x40

static void __init __pit_init(int val, u8 mode)
L
Linus Torvalds 已提交
836 837 838 839
{
	unsigned long flags;

	spin_lock_irqsave(&i8253_lock, flags);
840 841 842
	outb_p(mode, PIT_MODE);
	outb_p(val & 0xff, PIT_CH0);	/* LSB */
	outb_p(val >> 8, PIT_CH0);	/* MSB */
L
Linus Torvalds 已提交
843 844 845
	spin_unlock_irqrestore(&i8253_lock, flags);
}

846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
void __init pit_init(void)
{
	__pit_init(LATCH, 0x34); /* binary, mode 2, LSB/MSB, ch 0 */
}

void __init pit_stop_interrupt(void)
{
	__pit_init(0, 0x30); /* mode 0 */
}

void __init stop_timer_interrupt(void)
{
	char *name;
	if (vxtime.hpet_address) {
		name = "HPET";
		hpet_timer_stop_set_go(0);
	} else {
		name = "PIT";
		pit_stop_interrupt();
	}
	printk(KERN_INFO "timer: %s interrupt stopped.\n", name);
}

L
Linus Torvalds 已提交
869 870 871 872 873 874 875
int __init time_setup(char *str)
{
	report_lost_ticks = 1;
	return 1;
}

static struct irqaction irq0 = {
876
	timer_interrupt, IRQF_DISABLED, CPU_MASK_NONE, "timer", NULL, NULL
L
Linus Torvalds 已提交
877 878
};

879 880
static int __cpuinit
time_cpu_notifier(struct notifier_block *nb, unsigned long action, void *hcpu)
L
Linus Torvalds 已提交
881
{
882
	unsigned cpu = (unsigned long) hcpu;
883 884
	if (action == CPU_ONLINE)
		vsyscall_set_cpu(cpu);
885 886
	return NOTIFY_DONE;
}
L
Linus Torvalds 已提交
887

888 889
void __init time_init(void)
{
L
Linus Torvalds 已提交
890 891 892 893 894 895 896 897 898
	if (nohpet)
		vxtime.hpet_address = 0;

	xtime.tv_sec = get_cmos_time();
	xtime.tv_nsec = 0;

	set_normalized_timespec(&wall_to_monotonic,
	                        -xtime.tv_sec, -xtime.tv_nsec);

899
	if (!hpet_init())
900
                vxtime_hz = (FSEC_PER_SEC + hpet_period / 2) / hpet_period;
901 902
	else
		vxtime.hpet_address = 0;
903 904

	if (hpet_use_timer) {
905 906
		/* set tick_nsec to use the proper rate for HPET */
	  	tick_nsec = TICK_NSEC_HPET;
L
Linus Torvalds 已提交
907 908
		cpu_khz = hpet_calibrate_tsc();
		timename = "HPET";
A
Andi Kleen 已提交
909
#ifdef CONFIG_X86_PM_TIMER
910
	} else if (pmtmr_ioport && !vxtime.hpet_address) {
A
Andi Kleen 已提交
911 912 913 914 915
		vxtime_hz = PM_TIMER_FREQUENCY;
		timename = "PM";
		pit_init();
		cpu_khz = pit_calibrate_tsc();
#endif
L
Linus Torvalds 已提交
916 917 918 919 920 921
	} else {
		pit_init();
		cpu_khz = pit_calibrate_tsc();
		timename = "PIT";
	}

922
	vxtime.mode = VXTIME_TSC;
923 924
	vxtime.quot = (USEC_PER_SEC << US_SCALE) / vxtime_hz;
	vxtime.tsc_quot = (USEC_PER_MSEC << US_SCALE) / cpu_khz;
925
	vxtime.last_tsc = get_cycles_sync();
926
	set_cyc2ns_scale(cpu_khz);
L
Linus Torvalds 已提交
927
	setup_irq(0, &irq0);
928 929
	hotcpu_notifier(time_cpu_notifier, 0);
	time_cpu_notifier(NULL, CPU_ONLINE, (void *)(long)smp_processor_id());
L
Linus Torvalds 已提交
930

931 932 933
#ifndef CONFIG_SMP
	time_init_gtod();
#endif
L
Linus Torvalds 已提交
934 935
}

A
Andi Kleen 已提交
936 937 938 939
/*
 * Make an educated guess if the TSC is trustworthy and synchronized
 * over all CPUs.
 */
940
__cpuinit int unsynchronized_tsc(void)
A
Andi Kleen 已提交
941 942
{
#ifdef CONFIG_SMP
943
	if (apic_is_clustered_box())
A
Andi Kleen 已提交
944 945
		return 1;
#endif
946 947 948 949 950 951 952 953 954 955 956
	/* Most intel systems have synchronized TSCs except for
	   multi node systems */
 	if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL) {
#ifdef CONFIG_ACPI
		/* But TSC doesn't tick in C3 so don't use it there */
		if (acpi_fadt.length > 0 && acpi_fadt.plvl3_lat < 100)
			return 1;
#endif
 		return 0;
	}

A
Andi Kleen 已提交
957
 	/* Assume multi socket systems are not synchronized */
958
 	return num_present_cpus() > 1;
A
Andi Kleen 已提交
959 960
}

961
/*
962
 * Decide what mode gettimeofday should use.
963
 */
964
void time_init_gtod(void)
L
Linus Torvalds 已提交
965 966 967
{
	char *timetype;

A
Andi Kleen 已提交
968
	if (unsynchronized_tsc())
L
Linus Torvalds 已提交
969
		notsc = 1;
970

971 972 973 974 975
 	if (cpu_has(&boot_cpu_data, X86_FEATURE_RDTSCP))
		vgetcpu_mode = VGETCPU_RDTSCP;
	else
		vgetcpu_mode = VGETCPU_LSL;

L
Linus Torvalds 已提交
976
	if (vxtime.hpet_address && notsc) {
977
		timetype = hpet_use_timer ? "HPET" : "PIT/HPET";
978 979 980 981
		if (hpet_use_timer)
			vxtime.last = hpet_readl(HPET_T0_CMP) - hpet_tick;
		else
			vxtime.last = hpet_readl(HPET_COUNTER);
L
Linus Torvalds 已提交
982 983
		vxtime.mode = VXTIME_HPET;
		do_gettimeoffset = do_gettimeoffset_hpet;
A
Andi Kleen 已提交
984 985 986 987 988 989 990 991 992 993
#ifdef CONFIG_X86_PM_TIMER
	/* Using PM for gettimeofday is quite slow, but we have no other
	   choice because the TSC is too unreliable on some systems. */
	} else if (pmtmr_ioport && !vxtime.hpet_address && notsc) {
		timetype = "PM";
		do_gettimeoffset = do_gettimeoffset_pm;
		vxtime.mode = VXTIME_PMTMR;
		sysctl_vsyscall = 0;
		printk(KERN_INFO "Disabling vsyscall due to use of PM timer\n");
#endif
L
Linus Torvalds 已提交
994
	} else {
995
		timetype = hpet_use_timer ? "HPET/TSC" : "PIT/TSC";
L
Linus Torvalds 已提交
996 997
		vxtime.mode = VXTIME_TSC;
	}
998 999 1000 1001 1002 1003 1004 1005 1006 1007

	printk(KERN_INFO "time.c: Using %ld.%06ld MHz WALL %s GTOD %s timer.\n",
	       vxtime_hz / 1000000, vxtime_hz % 1000000, timename, timetype);
	printk(KERN_INFO "time.c: Detected %d.%03d MHz processor.\n",
		cpu_khz / 1000, cpu_khz % 1000);
	vxtime.quot = (USEC_PER_SEC << US_SCALE) / vxtime_hz;
	vxtime.tsc_quot = (USEC_PER_MSEC << US_SCALE) / cpu_khz;
	vxtime.last_tsc = get_cycles_sync();

	set_cyc2ns_scale(cpu_khz);
L
Linus Torvalds 已提交
1008 1009 1010 1011 1012 1013 1014
}

__setup("report_lost_ticks", time_setup);

static long clock_cmos_diff;
static unsigned long sleep_start;

1015 1016 1017 1018
/*
 * sysfs support for the timer.
 */

1019
static int timer_suspend(struct sys_device *dev, pm_message_t state)
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
{
	/*
	 * Estimate time zone so that set_time can update the clock
	 */
	long cmos_time =  get_cmos_time();

	clock_cmos_diff = -cmos_time;
	clock_cmos_diff += get_seconds();
	sleep_start = cmos_time;
	return 0;
}

static int timer_resume(struct sys_device *dev)
{
	unsigned long flags;
	unsigned long sec;
	unsigned long ctime = get_cmos_time();
1037
	long sleep_length = (ctime - sleep_start) * HZ;
L
Linus Torvalds 已提交
1038

1039 1040 1041 1042 1043 1044 1045 1046
	if (sleep_length < 0) {
		printk(KERN_WARNING "Time skew detected in timer resume!\n");
		/* The time after the resume must not be earlier than the time
		 * before the suspend or some nasty things will happen
		 */
		sleep_length = 0;
		ctime = sleep_start;
	}
L
Linus Torvalds 已提交
1047 1048 1049 1050 1051 1052 1053 1054 1055
	if (vxtime.hpet_address)
		hpet_reenable();
	else
		i8254_timer_resume();

	sec = ctime + clock_cmos_diff;
	write_seqlock_irqsave(&xtime_lock,flags);
	xtime.tv_sec = sec;
	xtime.tv_nsec = 0;
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
	if (vxtime.mode == VXTIME_HPET) {
		if (hpet_use_timer)
			vxtime.last = hpet_readl(HPET_T0_CMP) - hpet_tick;
		else
			vxtime.last = hpet_readl(HPET_COUNTER);
#ifdef CONFIG_X86_PM_TIMER
	} else if (vxtime.mode == VXTIME_PMTMR) {
		pmtimer_resume();
#endif
	} else
		vxtime.last_tsc = get_cycles_sync();
L
Linus Torvalds 已提交
1067 1068
	write_sequnlock_irqrestore(&xtime_lock,flags);
	jiffies += sleep_length;
1069
	monotonic_base += sleep_length * (NSEC_PER_SEC/HZ);
I
Ingo Molnar 已提交
1070
	touch_softlockup_watchdog();
L
Linus Torvalds 已提交
1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
	return 0;
}

static struct sysdev_class timer_sysclass = {
	.resume = timer_resume,
	.suspend = timer_suspend,
	set_kset_name("timer"),
};

/* XXX this driverfs stuff should probably go elsewhere later -john */
static struct sys_device device_timer = {
	.id	= 0,
	.cls	= &timer_sysclass,
};

static int time_init_device(void)
{
	int error = sysdev_class_register(&timer_sysclass);
	if (!error)
		error = sysdev_register(&device_timer);
	return error;
}

device_initcall(time_init_device);

#ifdef CONFIG_HPET_EMULATE_RTC
/* HPET in LegacyReplacement Mode eats up RTC interrupt line. When, HPET
 * is enabled, we support RTC interrupt functionality in software.
 * RTC has 3 kinds of interrupts:
 * 1) Update Interrupt - generate an interrupt, every sec, when RTC clock
 *    is updated
 * 2) Alarm Interrupt - generate an interrupt at a specific time of day
 * 3) Periodic Interrupt - generate periodic interrupt, with frequencies
 *    2Hz-8192Hz (2Hz-64Hz for non-root user) (all freqs in powers of 2)
 * (1) and (2) above are implemented using polling at a frequency of
 * 64 Hz. The exact frequency is a tradeoff between accuracy and interrupt
 * overhead. (DEFAULT_RTC_INT_FREQ)
 * For (3), we use interrupts at 64Hz or user specified periodic
 * frequency, whichever is higher.
 */
#include <linux/rtc.h>

#define DEFAULT_RTC_INT_FREQ 	64
#define RTC_NUM_INTS 		1

static unsigned long UIE_on;
static unsigned long prev_update_sec;

static unsigned long AIE_on;
static struct rtc_time alarm_time;

static unsigned long PIE_on;
static unsigned long PIE_freq = DEFAULT_RTC_INT_FREQ;
static unsigned long PIE_count;

static unsigned long hpet_rtc_int_freq; /* RTC interrupt frequency */
1127
static unsigned int hpet_t1_cmp; /* cached comparator register */
L
Linus Torvalds 已提交
1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160

int is_hpet_enabled(void)
{
	return vxtime.hpet_address != 0;
}

/*
 * Timer 1 for RTC, we do not use periodic interrupt feature,
 * even if HPET supports periodic interrupts on Timer 1.
 * The reason being, to set up a periodic interrupt in HPET, we need to
 * stop the main counter. And if we do that everytime someone diables/enables
 * RTC, we will have adverse effect on main kernel timer running on Timer 0.
 * So, for the time being, simulate the periodic interrupt in software.
 *
 * hpet_rtc_timer_init() is called for the first time and during subsequent
 * interuppts reinit happens through hpet_rtc_timer_reinit().
 */
int hpet_rtc_timer_init(void)
{
	unsigned int cfg, cnt;
	unsigned long flags;

	if (!is_hpet_enabled())
		return 0;
	/*
	 * Set the counter 1 and enable the interrupts.
	 */
	if (PIE_on && (PIE_freq > DEFAULT_RTC_INT_FREQ))
		hpet_rtc_int_freq = PIE_freq;
	else
		hpet_rtc_int_freq = DEFAULT_RTC_INT_FREQ;

	local_irq_save(flags);
1161

L
Linus Torvalds 已提交
1162 1163 1164
	cnt = hpet_readl(HPET_COUNTER);
	cnt += ((hpet_tick*HZ)/hpet_rtc_int_freq);
	hpet_writel(cnt, HPET_T1_CMP);
1165
	hpet_t1_cmp = cnt;
L
Linus Torvalds 已提交
1166 1167

	cfg = hpet_readl(HPET_T1_CFG);
1168 1169
	cfg &= ~HPET_TN_PERIODIC;
	cfg |= HPET_TN_ENABLE | HPET_TN_32BIT;
L
Linus Torvalds 已提交
1170 1171
	hpet_writel(cfg, HPET_T1_CFG);

1172 1173
	local_irq_restore(flags);

L
Linus Torvalds 已提交
1174 1175 1176 1177 1178
	return 1;
}

static void hpet_rtc_timer_reinit(void)
{
1179
	unsigned int cfg, cnt, ticks_per_int, lost_ints;
L
Linus Torvalds 已提交
1180

1181 1182 1183 1184
	if (unlikely(!(PIE_on | AIE_on | UIE_on))) {
		cfg = hpet_readl(HPET_T1_CFG);
		cfg &= ~HPET_TN_ENABLE;
		hpet_writel(cfg, HPET_T1_CFG);
L
Linus Torvalds 已提交
1185
		return;
1186
	}
L
Linus Torvalds 已提交
1187 1188 1189 1190 1191 1192 1193

	if (PIE_on && (PIE_freq > DEFAULT_RTC_INT_FREQ))
		hpet_rtc_int_freq = PIE_freq;
	else
		hpet_rtc_int_freq = DEFAULT_RTC_INT_FREQ;

	/* It is more accurate to use the comparator value than current count.*/
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220
	ticks_per_int = hpet_tick * HZ / hpet_rtc_int_freq;
	hpet_t1_cmp += ticks_per_int;
	hpet_writel(hpet_t1_cmp, HPET_T1_CMP);

	/*
	 * If the interrupt handler was delayed too long, the write above tries
	 * to schedule the next interrupt in the past and the hardware would
	 * not interrupt until the counter had wrapped around.
	 * So we have to check that the comparator wasn't set to a past time.
	 */
	cnt = hpet_readl(HPET_COUNTER);
	if (unlikely((int)(cnt - hpet_t1_cmp) > 0)) {
		lost_ints = (cnt - hpet_t1_cmp) / ticks_per_int + 1;
		/* Make sure that, even with the time needed to execute
		 * this code, the next scheduled interrupt has been moved
		 * back to the future: */
		lost_ints++;

		hpet_t1_cmp += lost_ints * ticks_per_int;
		hpet_writel(hpet_t1_cmp, HPET_T1_CMP);

		if (PIE_on)
			PIE_count += lost_ints;

		printk(KERN_WARNING "rtc: lost some interrupts at %ldHz.\n",
		       hpet_rtc_int_freq);
	}
L
Linus Torvalds 已提交
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348
}

/*
 * The functions below are called from rtc driver.
 * Return 0 if HPET is not being used.
 * Otherwise do the necessary changes and return 1.
 */
int hpet_mask_rtc_irq_bit(unsigned long bit_mask)
{
	if (!is_hpet_enabled())
		return 0;

	if (bit_mask & RTC_UIE)
		UIE_on = 0;
	if (bit_mask & RTC_PIE)
		PIE_on = 0;
	if (bit_mask & RTC_AIE)
		AIE_on = 0;

	return 1;
}

int hpet_set_rtc_irq_bit(unsigned long bit_mask)
{
	int timer_init_reqd = 0;

	if (!is_hpet_enabled())
		return 0;

	if (!(PIE_on | AIE_on | UIE_on))
		timer_init_reqd = 1;

	if (bit_mask & RTC_UIE) {
		UIE_on = 1;
	}
	if (bit_mask & RTC_PIE) {
		PIE_on = 1;
		PIE_count = 0;
	}
	if (bit_mask & RTC_AIE) {
		AIE_on = 1;
	}

	if (timer_init_reqd)
		hpet_rtc_timer_init();

	return 1;
}

int hpet_set_alarm_time(unsigned char hrs, unsigned char min, unsigned char sec)
{
	if (!is_hpet_enabled())
		return 0;

	alarm_time.tm_hour = hrs;
	alarm_time.tm_min = min;
	alarm_time.tm_sec = sec;

	return 1;
}

int hpet_set_periodic_freq(unsigned long freq)
{
	if (!is_hpet_enabled())
		return 0;

	PIE_freq = freq;
	PIE_count = 0;

	return 1;
}

int hpet_rtc_dropped_irq(void)
{
	if (!is_hpet_enabled())
		return 0;

	return 1;
}

irqreturn_t hpet_rtc_interrupt(int irq, void *dev_id, struct pt_regs *regs)
{
	struct rtc_time curr_time;
	unsigned long rtc_int_flag = 0;
	int call_rtc_interrupt = 0;

	hpet_rtc_timer_reinit();

	if (UIE_on | AIE_on) {
		rtc_get_rtc_time(&curr_time);
	}
	if (UIE_on) {
		if (curr_time.tm_sec != prev_update_sec) {
			/* Set update int info, call real rtc int routine */
			call_rtc_interrupt = 1;
			rtc_int_flag = RTC_UF;
			prev_update_sec = curr_time.tm_sec;
		}
	}
	if (PIE_on) {
		PIE_count++;
		if (PIE_count >= hpet_rtc_int_freq/PIE_freq) {
			/* Set periodic int info, call real rtc int routine */
			call_rtc_interrupt = 1;
			rtc_int_flag |= RTC_PF;
			PIE_count = 0;
		}
	}
	if (AIE_on) {
		if ((curr_time.tm_sec == alarm_time.tm_sec) &&
		    (curr_time.tm_min == alarm_time.tm_min) &&
		    (curr_time.tm_hour == alarm_time.tm_hour)) {
			/* Set alarm int info, call real rtc int routine */
			call_rtc_interrupt = 1;
			rtc_int_flag |= RTC_AF;
		}
	}
	if (call_rtc_interrupt) {
		rtc_int_flag |= (RTC_IRQF | (RTC_NUM_INTS << 8));
		rtc_interrupt(rtc_int_flag, dev_id, regs);
	}
	return IRQ_HANDLED;
}
#endif

static int __init nohpet_setup(char *s) 
{ 
	nohpet = 1;
1349
	return 1;
L
Linus Torvalds 已提交
1350 1351 1352 1353
} 

__setup("nohpet", nohpet_setup);

1354
int __init notsc_setup(char *s)
L
Linus Torvalds 已提交
1355 1356
{
	notsc = 1;
1357
	return 1;
L
Linus Torvalds 已提交
1358 1359 1360
}

__setup("notsc", notsc_setup);