summaryrefslogtreecommitdiff
path: root/arch/i386/kernel/timers/timer_tsc.c
blob: d395e3b42485f68fafa1d7f73ca9a44273a0827c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
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
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
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
/*
 * This code largely moved from arch/i386/kernel/time.c.
 * See comments there for proper credits.
 *
 * 2004-06-25    Jesper Juhl
 *      moved mark_offset_tsc below cpufreq_delayed_get to avoid gcc 3.4
 *      failing to inline.
 */

#include <linux/spinlock.h>
#include <linux/init.h>
#include <linux/timex.h>
#include <linux/errno.h>
#include <linux/cpufreq.h>
#include <linux/string.h>
#include <linux/jiffies.h>

#include <asm/timer.h>
#include <asm/io.h>
/* processor.h for distable_tsc flag */
#include <asm/processor.h>

#include "io_ports.h"
#include "mach_timer.h"

#include <asm/hpet.h>
#include <asm/i8253.h>

#ifdef CONFIG_HPET_TIMER
static unsigned long hpet_usec_quotient;
static unsigned long hpet_last;
static struct timer_opts timer_tsc;
#endif

static inline void cpufreq_delayed_get(void);

int tsc_disable __devinitdata = 0;

static int use_tsc;
/* Number of usecs that the last interrupt was delayed */
static int delay_at_last_interrupt;

static unsigned long last_tsc_low; /* lsb 32 bits of Time Stamp Counter */
static unsigned long last_tsc_high; /* msb 32 bits of Time Stamp Counter */
static unsigned long long monotonic_base;
static seqlock_t monotonic_lock = SEQLOCK_UNLOCKED;

/* convert from cycles(64bits) => nanoseconds (64bits)
 *  basic equation:
 *		ns = cycles / (freq / ns_per_sec)
 *		ns = cycles * (ns_per_sec / freq)
 *		ns = cycles * (10^9 / (cpu_khz * 10^3))
 *		ns = cycles * (10^6 / cpu_khz)
 *
 *	Then we use scaling math (suggested by george@mvista.com) to get:
 *		ns = cycles * (10^6 * SC / cpu_khz) / SC
 *		ns = cycles * cyc2ns_scale / SC
 *
 *	And since SC is a constant power of two, we can convert the div
 *  into a shift.
 *
 *  We can use khz divisor instead of mhz to keep a better percision, since
 *  cyc2ns_scale is limited to 10^6 * 2^10, which fits in 32 bits.
 *  (mathieu.desnoyers@polymtl.ca)
 *
 *			-johnstul@us.ibm.com "math is hard, lets go shopping!"
 */
static unsigned long cyc2ns_scale; 
#define CYC2NS_SCALE_FACTOR 10 /* 2^10, carefully chosen */

static inline void set_cyc2ns_scale(unsigned long cpu_khz)
{
	cyc2ns_scale = (1000000 << CYC2NS_SCALE_FACTOR)/cpu_khz;
}

static inline unsigned long long cycles_2_ns(unsigned long long cyc)
{
	return (cyc * cyc2ns_scale) >> CYC2NS_SCALE_FACTOR;
}

static int count2; /* counter for mark_offset_tsc() */

/* Cached *multiplier* to convert TSC counts to microseconds.
 * (see the equation below).
 * Equal to 2^32 * (1 / (clocks per usec) ).
 * Initialized in time_init.
 */
static unsigned long fast_gettimeoffset_quotient;

static unsigned long get_offset_tsc(void)
{
	register unsigned long eax, edx;

	/* Read the Time Stamp Counter */

	rdtsc(eax,edx);

	/* .. relative to previous jiffy (32 bits is enough) */
	eax -= last_tsc_low;	/* tsc_low delta */

	/*
         * Time offset = (tsc_low delta) * fast_gettimeoffset_quotient
         *             = (tsc_low delta) * (usecs_per_clock)
         *             = (tsc_low delta) * (usecs_per_jiffy / clocks_per_jiffy)
	 *
	 * Using a mull instead of a divl saves up to 31 clock cycles
	 * in the critical path.
         */

	__asm__("mull %2"
		:"=a" (eax), "=d" (edx)
		:"rm" (fast_gettimeoffset_quotient),
		 "0" (eax));

	/* our adjusted time offset in microseconds */
	return delay_at_last_interrupt + edx;
}

static unsigned long long monotonic_clock_tsc(void)
{
	unsigned long long last_offset, this_offset, base;
	unsigned seq;
	
	/* atomically read monotonic base & last_offset */
	do {
		seq = read_seqbegin(&monotonic_lock);
		last_offset = ((unsigned long long)last_tsc_high<<32)|last_tsc_low;
		base = monotonic_base;
	} while (read_seqretry(&monotonic_lock, seq));

	/* Read the Time Stamp Counter */
	rdtscll(this_offset);

	/* return the value in ns */
	return base + cycles_2_ns(this_offset - last_offset);
}

/*
 * Scheduler clock - returns current time in nanosec units.
 */
unsigned long long sched_clock(void)
{
	unsigned long long this_offset;

	/*
	 * In the NUMA case we dont use the TSC as they are not
	 * synchronized across all CPUs.
	 */
#ifndef CONFIG_NUMA
	if (!use_tsc)
#endif
		/* no locking but a rare wrong value is not a big deal */
		return jiffies_64 * (1000000000 / HZ);

	/* Read the Time Stamp Counter */
	rdtscll(this_offset);

	/* return the value in ns */
	return cycles_2_ns(this_offset);
}

static void delay_tsc(unsigned long loops)
{
	unsigned long bclock, now;
	
	rdtscl(bclock);
	do
	{
		rep_nop();
		rdtscl(now);
	} while ((now-bclock) < loops);
}

#ifdef CONFIG_HPET_TIMER
static void mark_offset_tsc_hpet(void)
{
	unsigned long long this_offset, last_offset;
 	unsigned long offset, temp, hpet_current;

	write_seqlock(&monotonic_lock);
	last_offset = ((unsigned long long)last_tsc_high<<32)|last_tsc_low;
	/*
	 * It is important that these two operations happen almost at
	 * the same time. We do the RDTSC stuff first, since it's
	 * faster. To avoid any inconsistencies, we need interrupts
	 * disabled locally.
	 */
	/*
	 * Interrupts are just disabled locally since the timer irq
	 * has the SA_INTERRUPT flag set. -arca
	 */
	/* read Pentium cycle counter */

	hpet_current = hpet_readl(HPET_COUNTER);
	rdtsc(last_tsc_low, last_tsc_high);

	/* lost tick compensation */
	offset = hpet_readl(HPET_T0_CMP) - hpet_tick;
	if (unlikely(((offset - hpet_last) > hpet_tick) && (hpet_last != 0))) {
		int lost_ticks = (offset - hpet_last) / hpet_tick;
		jiffies_64 += lost_ticks;
	}
	hpet_last = hpet_current;

	/* update the monotonic base value */
	this_offset = ((unsigned long long)last_tsc_high<<32)|last_tsc_low;
	monotonic_base += cycles_2_ns(this_offset - last_offset);
	write_sequnlock(&monotonic_lock);

	/* calculate delay_at_last_interrupt */
	/*
	 * Time offset = (hpet delta) * ( usecs per HPET clock )
	 *             = (hpet delta) * ( usecs per tick / HPET clocks per tick)
	 *             = (hpet delta) * ( hpet_usec_quotient ) / (2^32)
	 * Where,
	 * hpet_usec_quotient = (2^32 * usecs per tick)/HPET clocks per tick
	 */
	delay_at_last_interrupt = hpet_current - offset;
	ASM_MUL64_REG(temp, delay_at_last_interrupt,
			hpet_usec_quotient, delay_at_last_interrupt);
}
#endif


#ifdef CONFIG_CPU_FREQ
#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 inline void cpufreq_delayed_get(void) 
{
	if (cpufreq_init && !cpufreq_delayed_issched) {
		cpufreq_delayed_issched = 1;
		printk(KERN_DEBUG "Losing some ticks... checking if CPU frequency changed.\n");
		schedule_work(&cpufreq_delayed_get_work);
	}
}

/* If the CPU frequency is scaled, TSC-based delays will need a different
 * loops_per_jiffy value to function properly.
 */

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

#ifndef CONFIG_SMP
static unsigned long fast_gettimeoffset_ref = 0;
static unsigned int cpu_khz_ref = 0;
#endif

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

	if (val != CPUFREQ_RESUMECHANGE)
		write_seqlock_irq(&xtime_lock);
	if (!ref_freq) {
		ref_freq = freq->old;
		loops_per_jiffy_ref = cpu_data[freq->cpu].loops_per_jiffy;
#ifndef CONFIG_SMP
		fast_gettimeoffset_ref = fast_gettimeoffset_quotient;
		cpu_khz_ref = cpu_khz;
#endif
	}

	if ((val == CPUFREQ_PRECHANGE  && freq->old < freq->new) ||
	    (val == CPUFREQ_POSTCHANGE && freq->old > freq->new) ||
	    (val == CPUFREQ_RESUMECHANGE)) {
		if (!(freq->flags & CPUFREQ_CONST_LOOPS))
			cpu_data[freq->cpu].loops_per_jiffy = cpufreq_scale(loops_per_jiffy_ref, ref_freq, freq->new);
#ifndef CONFIG_SMP
		if (cpu_khz)
			cpu_khz = cpufreq_scale(cpu_khz_ref, ref_freq, freq->new);
		if (use_tsc) {
			if (!(freq->flags & CPUFREQ_CONST_LOOPS)) {
				fast_gettimeoffset_quotient = cpufreq_scale(fast_gettimeoffset_ref, freq->new, ref_freq);
				set_cyc2ns_scale(cpu_khz);
			}
		}
#endif
	}

	if (val != CPUFREQ_RESUMECHANGE)
		write_sequnlock_irq(&xtime_lock);

	return 0;
}

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


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

#else /* CONFIG_CPU_FREQ */
static inline void cpufreq_delayed_get(void) { return; }
#endif 

int recalibrate_cpu_khz(void)
{
#ifndef CONFIG_SMP
	unsigned int cpu_khz_old = cpu_khz;

	if (cpu_has_tsc) {
		init_cpu_khz();
		cpu_data[0].loops_per_jiffy =
		    cpufreq_scale(cpu_data[0].loops_per_jiffy,
			          cpu_khz_old,
				  cpu_khz);
		return 0;
	} else
		return -ENODEV;
#else
	return -ENODEV;
#endif
}
EXPORT_SYMBOL(recalibrate_cpu_khz);

static void mark_offset_tsc(void)
{
	unsigned long lost,delay;
	unsigned long delta = last_tsc_low;
	int count;
	int countmp;
	static int count1 = 0;
	unsigned long long this_offset, last_offset;
	static int lost_count = 0;

	write_seqlock(&monotonic_lock);
	last_offset = ((unsigned long long)last_tsc_high<<32)|last_tsc_low;
	/*
	 * It is important that these two operations happen almost at
	 * the same time. We do the RDTSC stuff first, since it's
	 * faster. To avoid any inconsistencies, we need interrupts
	 * disabled locally.
	 */

	/*
	 * Interrupts are just disabled locally since the timer irq
	 * has the SA_INTERRUPT flag set. -arca
	 */

	/* read Pentium cycle counter */

	rdtsc(last_tsc_low, last_tsc_high);

	spin_lock(&i8253_lock);
	outb_p(0x00, PIT_MODE);     /* latch the count ASAP */

	count = inb_p(PIT_CH0);    /* read the latched count */
	count |= inb(PIT_CH0) << 8;

	/*
	 * VIA686a test code... reset the latch if count > max + 1
	 * from timer_pit.c - cjb
	 */
	if (count > LATCH) {
		outb_p(0x34, PIT_MODE);
		outb_p(LATCH & 0xff, PIT_CH0);
		outb(LATCH >> 8, PIT_CH0);
		count = LATCH - 1;
	}

	spin_unlock(&i8253_lock);

	if (pit_latch_buggy) {
		/* get center value of last 3 time lutch */
		if ((count2 >= count && count >= count1)
		    || (count1 >= count && count >= count2)) {
			count2 = count1; count1 = count;
		} else if ((count1 >= count2 && count2 >= count)
			   || (count >= count2 && count2 >= count1)) {
			countmp = count;count = count2;
			count2 = count1;count1 = countmp;
		} else {
			count2 = count1; count1 = count; count = count1;
		}
	}

	/* lost tick compensation */
	delta = last_tsc_low - delta;
	{
		register unsigned long eax, edx;
		eax = delta;
		__asm__("mull %2"
		:"=a" (eax), "=d" (edx)
		:"rm" (fast_gettimeoffset_quotient),
		 "0" (eax));
		delta = edx;
	}
	delta += delay_at_last_interrupt;
	lost = delta/(1000000/HZ);
	delay = delta%(1000000/HZ);
	if (lost >= 2) {
		jiffies_64 += lost-1;

		/* sanity check to ensure we're not always losing ticks */
		if (lost_count++ > 100) {
			printk(KERN_WARNING "Losing too many ticks!\n");
			printk(KERN_WARNING "TSC cannot be used as a timesource.  \n");
			printk(KERN_WARNING "Possible reasons for this are:\n");
			printk(KERN_WARNING "  You're running with Speedstep,\n");
			printk(KERN_WARNING "  You don't have DMA enabled for your hard disk (see hdparm),\n");
			printk(KERN_WARNING "  Incorrect TSC synchronization on an SMP system (see dmesg).\n");
			printk(KERN_WARNING "Falling back to a sane timesource now.\n");

			clock_fallback();
		}
		/* ... but give the TSC a fair chance */
		if (lost_count > 25)
			cpufreq_delayed_get();
	} else
		lost_count = 0;
	/* update the monotonic base value */
	this_offset = ((unsigned long long)last_tsc_high<<32)|last_tsc_low;
	monotonic_base += cycles_2_ns(this_offset - last_offset);
	write_sequnlock(&monotonic_lock);

	/* calculate delay_at_last_interrupt */
	count = ((LATCH-1) - count) * TICK_SIZE;
	delay_at_last_interrupt = (count + LATCH/2) / LATCH;

	/* catch corner case where tick rollover occured
	 * between tsc and pit reads (as noted when
	 * usec delta is > 90% # of usecs/tick)
	 */
	if (lost && abs(delay - delay_at_last_interrupt) > (900000/HZ))
		jiffies_64++;
}

static int __init init_tsc(char* override)
{

	/* check clock override */
	if (override[0] && strncmp(override,"tsc",3)) {
#ifdef CONFIG_HPET_TIMER
		if (is_hpet_enabled()) {
			printk(KERN_ERR "Warning: clock= override failed. Defaulting to tsc\n");
		} else
#endif
		{
			return -ENODEV;
		}
	}

	/*
	 * If we have APM enabled or the CPU clock speed is variable
	 * (CPU stops clock on HLT or slows clock to save power)
	 * then the TSC timestamps may diverge by up to 1 jiffy from
	 * 'real time' but nothing will break.
	 * The most frequent case is that the CPU is "woken" from a halt
	 * state by the timer interrupt itself, so we get 0 error. In the
	 * rare cases where a driver would "wake" the CPU and request a
	 * timestamp, the maximum error is < 1 jiffy. But timestamps are
	 * still perfectly ordered.
	 * Note that the TSC counter will be reset if APM suspends
	 * to disk; this won't break the kernel, though, 'cuz we're
	 * smart.  See arch/i386/kernel/apm.c.
	 */
 	/*
 	 *	Firstly we have to do a CPU check for chips with
 	 * 	a potentially buggy TSC. At this point we haven't run
 	 *	the ident/bugs checks so we must run this hook as it
 	 *	may turn off the TSC flag.
 	 *
 	 *	NOTE: this doesn't yet handle SMP 486 machines where only
 	 *	some CPU's have a TSC. Thats never worked and nobody has
 	 *	moaned if you have the only one in the world - you fix it!
 	 */

	count2 = LATCH; /* initialize counter for mark_offset_tsc() */

	if (cpu_has_tsc) {
		unsigned long tsc_quotient;
#ifdef CONFIG_HPET_TIMER
		if (is_hpet_enabled() && hpet_use_timer) {
			unsigned long result, remain;
			printk("Using TSC for gettimeofday\n");
			tsc_quotient = calibrate_tsc_hpet(NULL);
			timer_tsc.mark_offset = &mark_offset_tsc_hpet;
			/*
			 * Math to calculate hpet to usec multiplier
			 * Look for the comments at get_offset_tsc_hpet()
			 */
			ASM_DIV64_REG(result, remain, hpet_tick,
					0, KERNEL_TICK_USEC);
			if (remain > (hpet_tick >> 1))
				result++; /* rounding the result */

			hpet_usec_quotient = result;
		} else
#endif
		{
			tsc_quotient = calibrate_tsc();
		}

		if (tsc_quotient) {
			fast_gettimeoffset_quotient = tsc_quotient;
			use_tsc = 1;
			/*
			 *	We could be more selective here I suspect
			 *	and just enable this for the next intel chips ?
			 */
			/* report CPU clock rate in Hz.
			 * The formula is (10^6 * 2^32) / (2^32 * 1 / (clocks/us)) =
			 * clock/second. Our precision is about 100 ppm.
			 */
			{	unsigned long eax=0, edx=1000;
				__asm__("divl %2"
		       		:"=a" (cpu_khz), "=d" (edx)
        	       		:"r" (tsc_quotient),
	                	"0" (eax), "1" (edx));
				printk("Detected %u.%03u MHz processor.\n",
					cpu_khz / 1000, cpu_khz % 1000);
			}
			set_cyc2ns_scale(cpu_khz);
			return 0;
		}
	}
	return -ENODEV;
}

static int tsc_resume(void)
{
	write_seqlock(&monotonic_lock);
	/* Assume this is the last mark offset time */
	rdtsc(last_tsc_low, last_tsc_high);
#ifdef CONFIG_HPET_TIMER
	if (is_hpet_enabled() && hpet_use_timer)
		hpet_last = hpet_readl(HPET_COUNTER);
#endif
	write_sequnlock(&monotonic_lock);
	return 0;
}

#ifndef CONFIG_X86_TSC
/* disable flag for tsc.  Takes effect by clearing the TSC cpu flag
 * in cpu/common.c */
static int __init tsc_setup(char *str)
{
	tsc_disable = 1;
	return 1;
}
#else
static int __init tsc_setup(char *str)
{
	printk(KERN_WARNING "notsc: Kernel compiled with CONFIG_X86_TSC, "
				"cannot disable TSC.\n");
	return 1;
}
#endif
__setup("notsc", tsc_setup);



/************************************************************/

/* tsc timer_opts struct */
static struct timer_opts timer_tsc = {
	.name = "tsc",
	.mark_offset = mark_offset_tsc, 
	.get_offset = get_offset_tsc,
	.monotonic_clock = monotonic_clock_tsc,
	.delay = delay_tsc,
	.read_timer = read_timer_tsc,
	.resume	= tsc_resume,
};

struct init_timer_opts __initdata timer_tsc_init = {
	.init = init_tsc,
	.opts = &timer_tsc,
};