Free Electrons

Embedded Linux Experts

  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
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
/*
 *  drivers/cpufreq/cpufreq_conservative.c
 *
 *  Copyright (C)  2001 Russell King
 *            (C)  2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
 *                      Jun Nakajima <jun.nakajima@intel.com>
 *            (C)  2009 Alexander Clouter <alex@digriz.org.uk>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/cpufreq.h>
#include <linux/cpu.h>
#include <linux/jiffies.h>
#include <linux/kernel_stat.h>
#include <linux/mutex.h>
#include <linux/hrtimer.h>
#include <linux/tick.h>
#include <linux/ktime.h>
#include <linux/sched.h>

/*
 * dbs is used in this file as a shortform for demandbased switching
 * It helps to keep variable names smaller, simpler
 */

#define DEF_FREQUENCY_UP_THRESHOLD		(80)
#define DEF_FREQUENCY_DOWN_THRESHOLD		(20)

/*
 * The polling frequency of this governor depends on the capability of
 * the processor. Default polling frequency is 1000 times the transition
 * latency of the processor. The governor will work on any processor with
 * transition latency <= 10mS, using appropriate sampling
 * rate.
 * For CPUs with transition latency > 10mS (mostly drivers with CPUFREQ_ETERNAL)
 * this governor will not work.
 * All times here are in uS.
 */
#define MIN_SAMPLING_RATE_RATIO			(2)

static unsigned int min_sampling_rate;

#define LATENCY_MULTIPLIER			(1000)
#define MIN_LATENCY_MULTIPLIER			(100)
#define DEF_SAMPLING_DOWN_FACTOR		(1)
#define MAX_SAMPLING_DOWN_FACTOR		(10)
#define TRANSITION_LATENCY_LIMIT		(10 * 1000 * 1000)

static void do_dbs_timer(struct work_struct *work);

struct cpu_dbs_info_s {
	cputime64_t prev_cpu_idle;
	cputime64_t prev_cpu_wall;
	cputime64_t prev_cpu_nice;
	struct cpufreq_policy *cur_policy;
	struct delayed_work work;
	unsigned int down_skip;
	unsigned int requested_freq;
	int cpu;
	unsigned int enable:1;
	/*
	 * percpu mutex that serializes governor limit change with
	 * do_dbs_timer invocation. We do not want do_dbs_timer to run
	 * when user is changing the governor or limits.
	 */
	struct mutex timer_mutex;
};
static DEFINE_PER_CPU(struct cpu_dbs_info_s, cs_cpu_dbs_info);

static unsigned int dbs_enable;	/* number of CPUs using this policy */

/*
 * dbs_mutex protects dbs_enable in governor start/stop.
 */
static DEFINE_MUTEX(dbs_mutex);

static struct dbs_tuners {
	unsigned int sampling_rate;
	unsigned int sampling_down_factor;
	unsigned int up_threshold;
	unsigned int down_threshold;
	unsigned int ignore_nice;
	unsigned int freq_step;
} dbs_tuners_ins = {
	.up_threshold = DEF_FREQUENCY_UP_THRESHOLD,
	.down_threshold = DEF_FREQUENCY_DOWN_THRESHOLD,
	.sampling_down_factor = DEF_SAMPLING_DOWN_FACTOR,
	.ignore_nice = 0,
	.freq_step = 5,
};

static inline u64 get_cpu_idle_time_jiffy(unsigned int cpu, u64 *wall)
{
	u64 idle_time;
	u64 cur_wall_time;
	u64 busy_time;

	cur_wall_time = jiffies64_to_cputime64(get_jiffies_64());

	busy_time  = kcpustat_cpu(cpu).cpustat[CPUTIME_USER];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_SYSTEM];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_IRQ];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_SOFTIRQ];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_STEAL];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_NICE];

	idle_time = cur_wall_time - busy_time;
	if (wall)
		*wall = jiffies_to_usecs(cur_wall_time);

	return jiffies_to_usecs(idle_time);
}

static inline cputime64_t get_cpu_idle_time(unsigned int cpu, cputime64_t *wall)
{
	u64 idle_time = get_cpu_idle_time_us(cpu, NULL);

	if (idle_time == -1ULL)
		return get_cpu_idle_time_jiffy(cpu, wall);
	else
		idle_time += get_cpu_iowait_time_us(cpu, wall);

	return idle_time;
}

/* keep track of frequency transitions */
static int
dbs_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
		     void *data)
{
	struct cpufreq_freqs *freq = data;
	struct cpu_dbs_info_s *this_dbs_info = &per_cpu(cs_cpu_dbs_info,
							freq->cpu);

	struct cpufreq_policy *policy;

	if (!this_dbs_info->enable)
		return 0;

	policy = this_dbs_info->cur_policy;

	/*
	 * we only care if our internally tracked freq moves outside
	 * the 'valid' ranges of freqency available to us otherwise
	 * we do not change it
	*/
	if (this_dbs_info->requested_freq > policy->max
			|| this_dbs_info->requested_freq < policy->min)
		this_dbs_info->requested_freq = freq->new;

	return 0;
}

static struct notifier_block dbs_cpufreq_notifier_block = {
	.notifier_call = dbs_cpufreq_notifier
};

/************************** sysfs interface ************************/
static ssize_t show_sampling_rate_min(struct kobject *kobj,
				      struct attribute *attr, char *buf)
{
	return sprintf(buf, "%u\n", min_sampling_rate);
}

define_one_global_ro(sampling_rate_min);

/* cpufreq_conservative Governor Tunables */
#define show_one(file_name, object)					\
static ssize_t show_##file_name						\
(struct kobject *kobj, struct attribute *attr, char *buf)		\
{									\
	return sprintf(buf, "%u\n", dbs_tuners_ins.object);		\
}
show_one(sampling_rate, sampling_rate);
show_one(sampling_down_factor, sampling_down_factor);
show_one(up_threshold, up_threshold);
show_one(down_threshold, down_threshold);
show_one(ignore_nice_load, ignore_nice);
show_one(freq_step, freq_step);

static ssize_t store_sampling_down_factor(struct kobject *a,
					  struct attribute *b,
					  const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf(buf, "%u", &input);

	if (ret != 1 || input > MAX_SAMPLING_DOWN_FACTOR || input < 1)
		return -EINVAL;

	dbs_tuners_ins.sampling_down_factor = input;
	return count;
}

static ssize_t store_sampling_rate(struct kobject *a, struct attribute *b,
				   const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf(buf, "%u", &input);

	if (ret != 1)
		return -EINVAL;

	dbs_tuners_ins.sampling_rate = max(input, min_sampling_rate);
	return count;
}

static ssize_t store_up_threshold(struct kobject *a, struct attribute *b,
				  const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf(buf, "%u", &input);

	if (ret != 1 || input > 100 ||
			input <= dbs_tuners_ins.down_threshold)
		return -EINVAL;

	dbs_tuners_ins.up_threshold = input;
	return count;
}

static ssize_t store_down_threshold(struct kobject *a, struct attribute *b,
				    const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf(buf, "%u", &input);

	/* cannot be lower than 11 otherwise freq will not fall */
	if (ret != 1 || input < 11 || input > 100 ||
			input >= dbs_tuners_ins.up_threshold)
		return -EINVAL;

	dbs_tuners_ins.down_threshold = input;
	return count;
}

static ssize_t store_ignore_nice_load(struct kobject *a, struct attribute *b,
				      const char *buf, size_t count)
{
	unsigned int input;
	int ret;

	unsigned int j;

	ret = sscanf(buf, "%u", &input);
	if (ret != 1)
		return -EINVAL;

	if (input > 1)
		input = 1;

	if (input == dbs_tuners_ins.ignore_nice) /* nothing to do */
		return count;

	dbs_tuners_ins.ignore_nice = input;

	/* we need to re-evaluate prev_cpu_idle */
	for_each_online_cpu(j) {
		struct cpu_dbs_info_s *dbs_info;
		dbs_info = &per_cpu(cs_cpu_dbs_info, j);
		dbs_info->prev_cpu_idle = get_cpu_idle_time(j,
						&dbs_info->prev_cpu_wall);
		if (dbs_tuners_ins.ignore_nice)
			dbs_info->prev_cpu_nice = kcpustat_cpu(j).cpustat[CPUTIME_NICE];
	}
	return count;
}

static ssize_t store_freq_step(struct kobject *a, struct attribute *b,
			       const char *buf, size_t count)
{
	unsigned int input;
	int ret;
	ret = sscanf(buf, "%u", &input);

	if (ret != 1)
		return -EINVAL;

	if (input > 100)
		input = 100;

	/* no need to test here if freq_step is zero as the user might actually
	 * want this, they would be crazy though :) */
	dbs_tuners_ins.freq_step = input;
	return count;
}

define_one_global_rw(sampling_rate);
define_one_global_rw(sampling_down_factor);
define_one_global_rw(up_threshold);
define_one_global_rw(down_threshold);
define_one_global_rw(ignore_nice_load);
define_one_global_rw(freq_step);

static struct attribute *dbs_attributes[] = {
	&sampling_rate_min.attr,
	&sampling_rate.attr,
	&sampling_down_factor.attr,
	&up_threshold.attr,
	&down_threshold.attr,
	&ignore_nice_load.attr,
	&freq_step.attr,
	NULL
};

static struct attribute_group dbs_attr_group = {
	.attrs = dbs_attributes,
	.name = "conservative",
};

/************************** sysfs end ************************/

static void dbs_check_cpu(struct cpu_dbs_info_s *this_dbs_info)
{
	unsigned int load = 0;
	unsigned int max_load = 0;
	unsigned int freq_target;

	struct cpufreq_policy *policy;
	unsigned int j;

	policy = this_dbs_info->cur_policy;

	/*
	 * Every sampling_rate, we check, if current idle time is less
	 * than 20% (default), then we try to increase frequency
	 * Every sampling_rate*sampling_down_factor, we check, if current
	 * idle time is more than 80%, then we try to decrease frequency
	 *
	 * Any frequency increase takes it to the maximum frequency.
	 * Frequency reduction happens at minimum steps of
	 * 5% (default) of maximum frequency
	 */

	/* Get Absolute Load */
	for_each_cpu(j, policy->cpus) {
		struct cpu_dbs_info_s *j_dbs_info;
		cputime64_t cur_wall_time, cur_idle_time;
		unsigned int idle_time, wall_time;

		j_dbs_info = &per_cpu(cs_cpu_dbs_info, j);

		cur_idle_time = get_cpu_idle_time(j, &cur_wall_time);

		wall_time = (unsigned int)
			(cur_wall_time - j_dbs_info->prev_cpu_wall);
		j_dbs_info->prev_cpu_wall = cur_wall_time;

		idle_time = (unsigned int)
			(cur_idle_time - j_dbs_info->prev_cpu_idle);
		j_dbs_info->prev_cpu_idle = cur_idle_time;

		if (dbs_tuners_ins.ignore_nice) {
			u64 cur_nice;
			unsigned long cur_nice_jiffies;

			cur_nice = kcpustat_cpu(j).cpustat[CPUTIME_NICE] -
					 j_dbs_info->prev_cpu_nice;
			/*
			 * Assumption: nice time between sampling periods will
			 * be less than 2^32 jiffies for 32 bit sys
			 */
			cur_nice_jiffies = (unsigned long)
					cputime64_to_jiffies64(cur_nice);

			j_dbs_info->prev_cpu_nice = kcpustat_cpu(j).cpustat[CPUTIME_NICE];
			idle_time += jiffies_to_usecs(cur_nice_jiffies);
		}

		if (unlikely(!wall_time || wall_time < idle_time))
			continue;

		load = 100 * (wall_time - idle_time) / wall_time;

		if (load > max_load)
			max_load = load;
	}

	/*
	 * break out if we 'cannot' reduce the speed as the user might
	 * want freq_step to be zero
	 */
	if (dbs_tuners_ins.freq_step == 0)
		return;

	/* Check for frequency increase */
	if (max_load > dbs_tuners_ins.up_threshold) {
		this_dbs_info->down_skip = 0;

		/* if we are already at full speed then break out early */
		if (this_dbs_info->requested_freq == policy->max)
			return;

		freq_target = (dbs_tuners_ins.freq_step * policy->max) / 100;

		/* max freq cannot be less than 100. But who knows.... */
		if (unlikely(freq_target == 0))
			freq_target = 5;

		this_dbs_info->requested_freq += freq_target;
		if (this_dbs_info->requested_freq > policy->max)
			this_dbs_info->requested_freq = policy->max;

		__cpufreq_driver_target(policy, this_dbs_info->requested_freq,
			CPUFREQ_RELATION_H);
		return;
	}

	/*
	 * The optimal frequency is the frequency that is the lowest that
	 * can support the current CPU usage without triggering the up
	 * policy. To be safe, we focus 10 points under the threshold.
	 */
	if (max_load < (dbs_tuners_ins.down_threshold - 10)) {
		freq_target = (dbs_tuners_ins.freq_step * policy->max) / 100;

		this_dbs_info->requested_freq -= freq_target;
		if (this_dbs_info->requested_freq < policy->min)
			this_dbs_info->requested_freq = policy->min;

		/*
		 * if we cannot reduce the frequency anymore, break out early
		 */
		if (policy->cur == policy->min)
			return;

		__cpufreq_driver_target(policy, this_dbs_info->requested_freq,
				CPUFREQ_RELATION_H);
		return;
	}
}

static void do_dbs_timer(struct work_struct *work)
{
	struct cpu_dbs_info_s *dbs_info =
		container_of(work, struct cpu_dbs_info_s, work.work);
	unsigned int cpu = dbs_info->cpu;

	/* We want all CPUs to do sampling nearly on same jiffy */
	int delay = usecs_to_jiffies(dbs_tuners_ins.sampling_rate);

	delay -= jiffies % delay;

	mutex_lock(&dbs_info->timer_mutex);

	dbs_check_cpu(dbs_info);

	schedule_delayed_work_on(cpu, &dbs_info->work, delay);
	mutex_unlock(&dbs_info->timer_mutex);
}

static inline void dbs_timer_init(struct cpu_dbs_info_s *dbs_info)
{
	/* We want all CPUs to do sampling nearly on same jiffy */
	int delay = usecs_to_jiffies(dbs_tuners_ins.sampling_rate);
	delay -= jiffies % delay;

	dbs_info->enable = 1;
	INIT_DELAYED_WORK_DEFERRABLE(&dbs_info->work, do_dbs_timer);
	schedule_delayed_work_on(dbs_info->cpu, &dbs_info->work, delay);
}

static inline void dbs_timer_exit(struct cpu_dbs_info_s *dbs_info)
{
	dbs_info->enable = 0;
	cancel_delayed_work_sync(&dbs_info->work);
}

static int cpufreq_governor_dbs(struct cpufreq_policy *policy,
				   unsigned int event)
{
	unsigned int cpu = policy->cpu;
	struct cpu_dbs_info_s *this_dbs_info;
	unsigned int j;
	int rc;

	this_dbs_info = &per_cpu(cs_cpu_dbs_info, cpu);

	switch (event) {
	case CPUFREQ_GOV_START:
		if ((!cpu_online(cpu)) || (!policy->cur))
			return -EINVAL;

		mutex_lock(&dbs_mutex);

		for_each_cpu(j, policy->cpus) {
			struct cpu_dbs_info_s *j_dbs_info;
			j_dbs_info = &per_cpu(cs_cpu_dbs_info, j);
			j_dbs_info->cur_policy = policy;

			j_dbs_info->prev_cpu_idle = get_cpu_idle_time(j,
						&j_dbs_info->prev_cpu_wall);
			if (dbs_tuners_ins.ignore_nice)
				j_dbs_info->prev_cpu_nice =
						kcpustat_cpu(j).cpustat[CPUTIME_NICE];
		}
		this_dbs_info->down_skip = 0;
		this_dbs_info->requested_freq = policy->cur;

		mutex_init(&this_dbs_info->timer_mutex);
		dbs_enable++;
		/*
		 * Start the timerschedule work, when this governor
		 * is used for first time
		 */
		if (dbs_enable == 1) {
			unsigned int latency;
			/* policy latency is in nS. Convert it to uS first */
			latency = policy->cpuinfo.transition_latency / 1000;
			if (latency == 0)
				latency = 1;

			rc = sysfs_create_group(cpufreq_global_kobject,
						&dbs_attr_group);
			if (rc) {
				mutex_unlock(&dbs_mutex);
				return rc;
			}

			/*
			 * conservative does not implement micro like ondemand
			 * governor, thus we are bound to jiffes/HZ
			 */
			min_sampling_rate =
				MIN_SAMPLING_RATE_RATIO * jiffies_to_usecs(10);
			/* Bring kernel and HW constraints together */
			min_sampling_rate = max(min_sampling_rate,
					MIN_LATENCY_MULTIPLIER * latency);
			dbs_tuners_ins.sampling_rate =
				max(min_sampling_rate,
				    latency * LATENCY_MULTIPLIER);

			cpufreq_register_notifier(
					&dbs_cpufreq_notifier_block,
					CPUFREQ_TRANSITION_NOTIFIER);
		}
		mutex_unlock(&dbs_mutex);

		dbs_timer_init(this_dbs_info);

		break;

	case CPUFREQ_GOV_STOP:
		dbs_timer_exit(this_dbs_info);

		mutex_lock(&dbs_mutex);
		dbs_enable--;
		mutex_destroy(&this_dbs_info->timer_mutex);

		/*
		 * Stop the timerschedule work, when this governor
		 * is used for first time
		 */
		if (dbs_enable == 0)
			cpufreq_unregister_notifier(
					&dbs_cpufreq_notifier_block,
					CPUFREQ_TRANSITION_NOTIFIER);

		mutex_unlock(&dbs_mutex);
		if (!dbs_enable)
			sysfs_remove_group(cpufreq_global_kobject,
					   &dbs_attr_group);

		break;

	case CPUFREQ_GOV_LIMITS:
		mutex_lock(&this_dbs_info->timer_mutex);
		if (policy->max < this_dbs_info->cur_policy->cur)
			__cpufreq_driver_target(
					this_dbs_info->cur_policy,
					policy->max, CPUFREQ_RELATION_H);
		else if (policy->min > this_dbs_info->cur_policy->cur)
			__cpufreq_driver_target(
					this_dbs_info->cur_policy,
					policy->min, CPUFREQ_RELATION_L);
		mutex_unlock(&this_dbs_info->timer_mutex);

		break;
	}
	return 0;
}

#ifndef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE
static
#endif
struct cpufreq_governor cpufreq_gov_conservative = {
	.name			= "conservative",
	.governor		= cpufreq_governor_dbs,
	.max_transition_latency	= TRANSITION_LATENCY_LIMIT,
	.owner			= THIS_MODULE,
};

static int __init cpufreq_gov_dbs_init(void)
{
	return cpufreq_register_governor(&cpufreq_gov_conservative);
}

static void __exit cpufreq_gov_dbs_exit(void)
{
	cpufreq_unregister_governor(&cpufreq_gov_conservative);
}


MODULE_AUTHOR("Alexander Clouter <alex@digriz.org.uk>");
MODULE_DESCRIPTION("'cpufreq_conservative' - A dynamic cpufreq governor for "
		"Low Latency Frequency Transition capable processors "
		"optimised for use in a battery environment");
MODULE_LICENSE("GPL");

#ifdef CONFIG_CPU_FREQ_DEFAULT_GOV_CONSERVATIVE
fs_initcall(cpufreq_gov_dbs_init);
#else
module_init(cpufreq_gov_dbs_init);
#endif
module_exit(cpufreq_gov_dbs_exit);