xref: /openbmc/linux/kernel/irq/timings.c (revision 2326f3cd)
1 // SPDX-License-Identifier: GPL-2.0
2 // Copyright (C) 2016, Linaro Ltd - Daniel Lezcano <daniel.lezcano@linaro.org>
3 #define pr_fmt(fmt) "irq_timings: " fmt
4 
5 #include <linux/kernel.h>
6 #include <linux/percpu.h>
7 #include <linux/slab.h>
8 #include <linux/static_key.h>
9 #include <linux/init.h>
10 #include <linux/interrupt.h>
11 #include <linux/idr.h>
12 #include <linux/irq.h>
13 #include <linux/math64.h>
14 #include <linux/log2.h>
15 
16 #include <trace/events/irq.h>
17 
18 #include "internals.h"
19 
20 DEFINE_STATIC_KEY_FALSE(irq_timing_enabled);
21 
22 DEFINE_PER_CPU(struct irq_timings, irq_timings);
23 
24 static DEFINE_IDR(irqt_stats);
25 
26 void irq_timings_enable(void)
27 {
28 	static_branch_enable(&irq_timing_enabled);
29 }
30 
31 void irq_timings_disable(void)
32 {
33 	static_branch_disable(&irq_timing_enabled);
34 }
35 
36 /*
37  * The main goal of this algorithm is to predict the next interrupt
38  * occurrence on the current CPU.
39  *
40  * Currently, the interrupt timings are stored in a circular array
41  * buffer every time there is an interrupt, as a tuple: the interrupt
42  * number and the associated timestamp when the event occurred <irq,
43  * timestamp>.
44  *
45  * For every interrupt occurring in a short period of time, we can
46  * measure the elapsed time between the occurrences for the same
47  * interrupt and we end up with a suite of intervals. The experience
48  * showed the interrupts are often coming following a periodic
49  * pattern.
50  *
51  * The objective of the algorithm is to find out this periodic pattern
52  * in a fastest way and use its period to predict the next irq event.
53  *
54  * When the next interrupt event is requested, we are in the situation
55  * where the interrupts are disabled and the circular buffer
56  * containing the timings is filled with the events which happened
57  * after the previous next-interrupt-event request.
58  *
59  * At this point, we read the circular buffer and we fill the irq
60  * related statistics structure. After this step, the circular array
61  * containing the timings is empty because all the values are
62  * dispatched in their corresponding buffers.
63  *
64  * Now for each interrupt, we can predict the next event by using the
65  * suffix array, log interval and exponential moving average
66  *
67  * 1. Suffix array
68  *
69  * Suffix array is an array of all the suffixes of a string. It is
70  * widely used as a data structure for compression, text search, ...
71  * For instance for the word 'banana', the suffixes will be: 'banana'
72  * 'anana' 'nana' 'ana' 'na' 'a'
73  *
74  * Usually, the suffix array is sorted but for our purpose it is
75  * not necessary and won't provide any improvement in the context of
76  * the solved problem where we clearly define the boundaries of the
77  * search by a max period and min period.
78  *
79  * The suffix array will build a suite of intervals of different
80  * length and will look for the repetition of each suite. If the suite
81  * is repeating then we have the period because it is the length of
82  * the suite whatever its position in the buffer.
83  *
84  * 2. Log interval
85  *
86  * We saw the irq timings allow to compute the interval of the
87  * occurrences for a specific interrupt. We can reasonably assume the
88  * longer is the interval, the higher is the error for the next event
89  * and we can consider storing those interval values into an array
90  * where each slot in the array correspond to an interval at the power
91  * of 2 of the index. For example, index 12 will contain values
92  * between 2^11 and 2^12.
93  *
94  * At the end we have an array of values where at each index defines a
95  * [2^index - 1, 2 ^ index] interval values allowing to store a large
96  * number of values inside a small array.
97  *
98  * For example, if we have the value 1123, then we store it at
99  * ilog2(1123) = 10 index value.
100  *
101  * Storing those value at the specific index is done by computing an
102  * exponential moving average for this specific slot. For instance,
103  * for values 1800, 1123, 1453, ... fall under the same slot (10) and
104  * the exponential moving average is computed every time a new value
105  * is stored at this slot.
106  *
107  * 3. Exponential Moving Average
108  *
109  * The EMA is largely used to track a signal for stocks or as a low
110  * pass filter. The magic of the formula, is it is very simple and the
111  * reactivity of the average can be tuned with the factors called
112  * alpha.
113  *
114  * The higher the alphas are, the faster the average respond to the
115  * signal change. In our case, if a slot in the array is a big
116  * interval, we can have numbers with a big difference between
117  * them. The impact of those differences in the average computation
118  * can be tuned by changing the alpha value.
119  *
120  *
121  *  -- The algorithm --
122  *
123  * We saw the different processing above, now let's see how they are
124  * used together.
125  *
126  * For each interrupt:
127  *	For each interval:
128  *		Compute the index = ilog2(interval)
129  *		Compute a new_ema(buffer[index], interval)
130  *		Store the index in a circular buffer
131  *
132  *	Compute the suffix array of the indexes
133  *
134  *	For each suffix:
135  *		If the suffix is reverse-found 3 times
136  *			Return suffix
137  *
138  *	Return Not found
139  *
140  * However we can not have endless suffix array to be build, it won't
141  * make sense and it will add an extra overhead, so we can restrict
142  * this to a maximum suffix length of 5 and a minimum suffix length of
143  * 2. The experience showed 5 is the majority of the maximum pattern
144  * period found for different devices.
145  *
146  * The result is a pattern finding less than 1us for an interrupt.
147  *
148  * Example based on real values:
149  *
150  * Example 1 : MMC write/read interrupt interval:
151  *
152  *	223947, 1240, 1384, 1386, 1386,
153  *	217416, 1236, 1384, 1386, 1387,
154  *	214719, 1241, 1386, 1387, 1384,
155  *	213696, 1234, 1384, 1386, 1388,
156  *	219904, 1240, 1385, 1389, 1385,
157  *	212240, 1240, 1386, 1386, 1386,
158  *	214415, 1236, 1384, 1386, 1387,
159  *	214276, 1234, 1384, 1388, ?
160  *
161  * For each element, apply ilog2(value)
162  *
163  *	15, 8, 8, 8, 8,
164  *	15, 8, 8, 8, 8,
165  *	15, 8, 8, 8, 8,
166  *	15, 8, 8, 8, 8,
167  *	15, 8, 8, 8, 8,
168  *	15, 8, 8, 8, 8,
169  *	15, 8, 8, 8, 8,
170  *	15, 8, 8, 8, ?
171  *
172  * Max period of 5, we take the last (max_period * 3) 15 elements as
173  * we can be confident if the pattern repeats itself three times it is
174  * a repeating pattern.
175  *
176  *	             8,
177  *	15, 8, 8, 8, 8,
178  *	15, 8, 8, 8, 8,
179  *	15, 8, 8, 8, ?
180  *
181  * Suffixes are:
182  *
183  *  1) 8, 15, 8, 8, 8  <- max period
184  *  2) 8, 15, 8, 8
185  *  3) 8, 15, 8
186  *  4) 8, 15           <- min period
187  *
188  * From there we search the repeating pattern for each suffix.
189  *
190  * buffer: 8, 15, 8, 8, 8, 8, 15, 8, 8, 8, 8, 15, 8, 8, 8
191  *         |   |  |  |  |  |   |  |  |  |  |   |  |  |  |
192  *         8, 15, 8, 8, 8  |   |  |  |  |  |   |  |  |  |
193  *                         8, 15, 8, 8, 8  |   |  |  |  |
194  *                                         8, 15, 8, 8, 8
195  *
196  * When moving the suffix, we found exactly 3 matches.
197  *
198  * The first suffix with period 5 is repeating.
199  *
200  * The next event is (3 * max_period) % suffix_period
201  *
202  * In this example, the result 0, so the next event is suffix[0] => 8
203  *
204  * However, 8 is the index in the array of exponential moving average
205  * which was calculated on the fly when storing the values, so the
206  * interval is ema[8] = 1366
207  *
208  *
209  * Example 2:
210  *
211  *	4, 3, 5, 100,
212  *	3, 3, 5, 117,
213  *	4, 4, 5, 112,
214  *	4, 3, 4, 110,
215  *	3, 5, 3, 117,
216  *	4, 4, 5, 112,
217  *	4, 3, 4, 110,
218  *	3, 4, 5, 112,
219  *	4, 3, 4, 110
220  *
221  * ilog2
222  *
223  *	0, 0, 0, 4,
224  *	0, 0, 0, 4,
225  *	0, 0, 0, 4,
226  *	0, 0, 0, 4,
227  *	0, 0, 0, 4,
228  *	0, 0, 0, 4,
229  *	0, 0, 0, 4,
230  *	0, 0, 0, 4,
231  *	0, 0, 0, 4
232  *
233  * Max period 5:
234  *	   0, 0, 4,
235  *	0, 0, 0, 4,
236  *	0, 0, 0, 4,
237  *	0, 0, 0, 4
238  *
239  * Suffixes:
240  *
241  *  1) 0, 0, 4, 0, 0
242  *  2) 0, 0, 4, 0
243  *  3) 0, 0, 4
244  *  4) 0, 0
245  *
246  * buffer: 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4
247  *         |  |  |  |  |  |  X
248  *         0, 0, 4, 0, 0, |  X
249  *                        0, 0
250  *
251  * buffer: 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4
252  *         |  |  |  |  |  |  |  |  |  |  |  |  |  |  |
253  *         0, 0, 4, 0, |  |  |  |  |  |  |  |  |  |  |
254  *                     0, 0, 4, 0, |  |  |  |  |  |  |
255  *                                 0, 0, 4, 0, |  |  |
256  *                                             0  0  4
257  *
258  * Pattern is found 3 times, the remaining is 1 which results from
259  * (max_period * 3) % suffix_period. This value is the index in the
260  * suffix arrays. The suffix array for a period 4 has the value 4
261  * at index 1.
262  */
263 #define EMA_ALPHA_VAL		64
264 #define EMA_ALPHA_SHIFT		7
265 
266 #define PREDICTION_PERIOD_MIN	3
267 #define PREDICTION_PERIOD_MAX	5
268 #define PREDICTION_FACTOR	4
269 #define PREDICTION_MAX		10 /* 2 ^ PREDICTION_MAX useconds */
270 #define PREDICTION_BUFFER_SIZE	16 /* slots for EMAs, hardly more than 16 */
271 
272 /*
273  * Number of elements in the circular buffer: If it happens it was
274  * flushed before, then the number of elements could be smaller than
275  * IRQ_TIMINGS_SIZE, so the count is used, otherwise the array size is
276  * used as we wrapped. The index begins from zero when we did not
277  * wrap. That could be done in a nicer way with the proper circular
278  * array structure type but with the cost of extra computation in the
279  * interrupt handler hot path. We choose efficiency.
280  */
281 #define for_each_irqts(i, irqts)					\
282 	for (i = irqts->count < IRQ_TIMINGS_SIZE ?			\
283 		     0 : irqts->count & IRQ_TIMINGS_MASK,		\
284 		     irqts->count = min(IRQ_TIMINGS_SIZE,		\
285 					irqts->count);			\
286 	     irqts->count > 0; irqts->count--,				\
287 		     i = (i + 1) & IRQ_TIMINGS_MASK)
288 
289 struct irqt_stat {
290 	u64	last_ts;
291 	u64	ema_time[PREDICTION_BUFFER_SIZE];
292 	int	timings[IRQ_TIMINGS_SIZE];
293 	int	circ_timings[IRQ_TIMINGS_SIZE];
294 	int	count;
295 };
296 
297 /*
298  * Exponential moving average computation
299  */
300 static u64 irq_timings_ema_new(u64 value, u64 ema_old)
301 {
302 	s64 diff;
303 
304 	if (unlikely(!ema_old))
305 		return value;
306 
307 	diff = (value - ema_old) * EMA_ALPHA_VAL;
308 	/*
309 	 * We can use a s64 type variable to be added with the u64
310 	 * ema_old variable as this one will never have its topmost
311 	 * bit set, it will be always smaller than 2^63 nanosec
312 	 * interrupt interval (292 years).
313 	 */
314 	return ema_old + (diff >> EMA_ALPHA_SHIFT);
315 }
316 
317 static int irq_timings_next_event_index(int *buffer, size_t len, int period_max)
318 {
319 	int period;
320 
321 	/*
322 	 * Move the beginning pointer to the end minus the max period x 3.
323 	 * We are at the point we can begin searching the pattern
324 	 */
325 	buffer = &buffer[len - (period_max * 3)];
326 
327 	/* Adjust the length to the maximum allowed period x 3 */
328 	len = period_max * 3;
329 
330 	/*
331 	 * The buffer contains the suite of intervals, in a ilog2
332 	 * basis, we are looking for a repetition. We point the
333 	 * beginning of the search three times the length of the
334 	 * period beginning at the end of the buffer. We do that for
335 	 * each suffix.
336 	 */
337 	for (period = period_max; period >= PREDICTION_PERIOD_MIN; period--) {
338 
339 		/*
340 		 * The first comparison always succeed because the
341 		 * suffix is deduced from the first n-period bytes of
342 		 * the buffer and we compare the initial suffix with
343 		 * itself, so we can skip the first iteration.
344 		 */
345 		int idx = period;
346 		size_t size = period;
347 
348 		/*
349 		 * We look if the suite with period 'i' repeat
350 		 * itself. If it is truncated at the end, as it
351 		 * repeats we can use the period to find out the next
352 		 * element with the modulo.
353 		 */
354 		while (!memcmp(buffer, &buffer[idx], size * sizeof(int))) {
355 
356 			/*
357 			 * Move the index in a period basis
358 			 */
359 			idx += size;
360 
361 			/*
362 			 * If this condition is reached, all previous
363 			 * memcmp were successful, so the period is
364 			 * found.
365 			 */
366 			if (idx == len)
367 				return buffer[len % period];
368 
369 			/*
370 			 * If the remaining elements to compare are
371 			 * smaller than the period, readjust the size
372 			 * of the comparison for the last iteration.
373 			 */
374 			if (len - idx < period)
375 				size = len - idx;
376 		}
377 	}
378 
379 	return -1;
380 }
381 
382 static u64 __irq_timings_next_event(struct irqt_stat *irqs, int irq, u64 now)
383 {
384 	int index, i, period_max, count, start, min = INT_MAX;
385 
386 	if ((now - irqs->last_ts) >= NSEC_PER_SEC) {
387 		irqs->count = irqs->last_ts = 0;
388 		return U64_MAX;
389 	}
390 
391 	/*
392 	 * As we want to find three times the repetition, we need a
393 	 * number of intervals greater or equal to three times the
394 	 * maximum period, otherwise we truncate the max period.
395 	 */
396 	period_max = irqs->count > (3 * PREDICTION_PERIOD_MAX) ?
397 		PREDICTION_PERIOD_MAX : irqs->count / 3;
398 
399 	/*
400 	 * If we don't have enough irq timings for this prediction,
401 	 * just bail out.
402 	 */
403 	if (period_max <= PREDICTION_PERIOD_MIN)
404 		return U64_MAX;
405 
406 	/*
407 	 * 'count' will depends if the circular buffer wrapped or not
408 	 */
409 	count = irqs->count < IRQ_TIMINGS_SIZE ?
410 		irqs->count : IRQ_TIMINGS_SIZE;
411 
412 	start = irqs->count < IRQ_TIMINGS_SIZE ?
413 		0 : (irqs->count & IRQ_TIMINGS_MASK);
414 
415 	/*
416 	 * Copy the content of the circular buffer into another buffer
417 	 * in order to linearize the buffer instead of dealing with
418 	 * wrapping indexes and shifted array which will be prone to
419 	 * error and extremely difficult to debug.
420 	 */
421 	for (i = 0; i < count; i++) {
422 		int index = (start + i) & IRQ_TIMINGS_MASK;
423 
424 		irqs->timings[i] = irqs->circ_timings[index];
425 		min = min_t(int, irqs->timings[i], min);
426 	}
427 
428 	index = irq_timings_next_event_index(irqs->timings, count, period_max);
429 	if (index < 0)
430 		return irqs->last_ts + irqs->ema_time[min];
431 
432 	return irqs->last_ts + irqs->ema_time[index];
433 }
434 
435 static __always_inline int irq_timings_interval_index(u64 interval)
436 {
437 	/*
438 	 * The PREDICTION_FACTOR increase the interval size for the
439 	 * array of exponential average.
440 	 */
441 	u64 interval_us = (interval >> 10) / PREDICTION_FACTOR;
442 
443 	return likely(interval_us) ? ilog2(interval_us) : 0;
444 }
445 
446 static __always_inline void __irq_timings_store(int irq, struct irqt_stat *irqs,
447 						u64 interval)
448 {
449 	int index;
450 
451 	/*
452 	 * Get the index in the ema table for this interrupt.
453 	 */
454 	index = irq_timings_interval_index(interval);
455 
456 	if (index > PREDICTION_BUFFER_SIZE - 1) {
457 		irqs->count = 0;
458 		return;
459 	}
460 
461 	/*
462 	 * Store the index as an element of the pattern in another
463 	 * circular array.
464 	 */
465 	irqs->circ_timings[irqs->count & IRQ_TIMINGS_MASK] = index;
466 
467 	irqs->ema_time[index] = irq_timings_ema_new(interval,
468 						    irqs->ema_time[index]);
469 
470 	irqs->count++;
471 }
472 
473 static inline void irq_timings_store(int irq, struct irqt_stat *irqs, u64 ts)
474 {
475 	u64 old_ts = irqs->last_ts;
476 	u64 interval;
477 
478 	/*
479 	 * The timestamps are absolute time values, we need to compute
480 	 * the timing interval between two interrupts.
481 	 */
482 	irqs->last_ts = ts;
483 
484 	/*
485 	 * The interval type is u64 in order to deal with the same
486 	 * type in our computation, that prevent mindfuck issues with
487 	 * overflow, sign and division.
488 	 */
489 	interval = ts - old_ts;
490 
491 	/*
492 	 * The interrupt triggered more than one second apart, that
493 	 * ends the sequence as predictable for our purpose. In this
494 	 * case, assume we have the beginning of a sequence and the
495 	 * timestamp is the first value. As it is impossible to
496 	 * predict anything at this point, return.
497 	 *
498 	 * Note the first timestamp of the sequence will always fall
499 	 * in this test because the old_ts is zero. That is what we
500 	 * want as we need another timestamp to compute an interval.
501 	 */
502 	if (interval >= NSEC_PER_SEC) {
503 		irqs->count = 0;
504 		return;
505 	}
506 
507 	__irq_timings_store(irq, irqs, interval);
508 }
509 
510 /**
511  * irq_timings_next_event - Return when the next event is supposed to arrive
512  *
513  * During the last busy cycle, the number of interrupts is incremented
514  * and stored in the irq_timings structure. This information is
515  * necessary to:
516  *
517  * - know if the index in the table wrapped up:
518  *
519  *      If more than the array size interrupts happened during the
520  *      last busy/idle cycle, the index wrapped up and we have to
521  *      begin with the next element in the array which is the last one
522  *      in the sequence, otherwise it is at the index 0.
523  *
524  * - have an indication of the interrupts activity on this CPU
525  *   (eg. irq/sec)
526  *
527  * The values are 'consumed' after inserting in the statistical model,
528  * thus the count is reinitialized.
529  *
530  * The array of values **must** be browsed in the time direction, the
531  * timestamp must increase between an element and the next one.
532  *
533  * Returns a nanosec time based estimation of the earliest interrupt,
534  * U64_MAX otherwise.
535  */
536 u64 irq_timings_next_event(u64 now)
537 {
538 	struct irq_timings *irqts = this_cpu_ptr(&irq_timings);
539 	struct irqt_stat *irqs;
540 	struct irqt_stat __percpu *s;
541 	u64 ts, next_evt = U64_MAX;
542 	int i, irq = 0;
543 
544 	/*
545 	 * This function must be called with the local irq disabled in
546 	 * order to prevent the timings circular buffer to be updated
547 	 * while we are reading it.
548 	 */
549 	lockdep_assert_irqs_disabled();
550 
551 	if (!irqts->count)
552 		return next_evt;
553 
554 	/*
555 	 * Number of elements in the circular buffer: If it happens it
556 	 * was flushed before, then the number of elements could be
557 	 * smaller than IRQ_TIMINGS_SIZE, so the count is used,
558 	 * otherwise the array size is used as we wrapped. The index
559 	 * begins from zero when we did not wrap. That could be done
560 	 * in a nicer way with the proper circular array structure
561 	 * type but with the cost of extra computation in the
562 	 * interrupt handler hot path. We choose efficiency.
563 	 *
564 	 * Inject measured irq/timestamp to the pattern prediction
565 	 * model while decrementing the counter because we consume the
566 	 * data from our circular buffer.
567 	 */
568 	for_each_irqts(i, irqts) {
569 		irq = irq_timing_decode(irqts->values[i], &ts);
570 		s = idr_find(&irqt_stats, irq);
571 		if (s)
572 			irq_timings_store(irq, this_cpu_ptr(s), ts);
573 	}
574 
575 	/*
576 	 * Look in the list of interrupts' statistics, the earliest
577 	 * next event.
578 	 */
579 	idr_for_each_entry(&irqt_stats, s, i) {
580 
581 		irqs = this_cpu_ptr(s);
582 
583 		ts = __irq_timings_next_event(irqs, i, now);
584 		if (ts <= now)
585 			return now;
586 
587 		if (ts < next_evt)
588 			next_evt = ts;
589 	}
590 
591 	return next_evt;
592 }
593 
594 void irq_timings_free(int irq)
595 {
596 	struct irqt_stat __percpu *s;
597 
598 	s = idr_find(&irqt_stats, irq);
599 	if (s) {
600 		free_percpu(s);
601 		idr_remove(&irqt_stats, irq);
602 	}
603 }
604 
605 int irq_timings_alloc(int irq)
606 {
607 	struct irqt_stat __percpu *s;
608 	int id;
609 
610 	/*
611 	 * Some platforms can have the same private interrupt per cpu,
612 	 * so this function may be called several times with the
613 	 * same interrupt number. Just bail out in case the per cpu
614 	 * stat structure is already allocated.
615 	 */
616 	s = idr_find(&irqt_stats, irq);
617 	if (s)
618 		return 0;
619 
620 	s = alloc_percpu(*s);
621 	if (!s)
622 		return -ENOMEM;
623 
624 	idr_preload(GFP_KERNEL);
625 	id = idr_alloc(&irqt_stats, s, irq, irq + 1, GFP_NOWAIT);
626 	idr_preload_end();
627 
628 	if (id < 0) {
629 		free_percpu(s);
630 		return id;
631 	}
632 
633 	return 0;
634 }
635 
636 #ifdef CONFIG_TEST_IRQ_TIMINGS
637 struct timings_intervals {
638 	u64 *intervals;
639 	size_t count;
640 };
641 
642 /*
643  * Intervals are given in nanosecond base
644  */
645 static u64 intervals0[] __initdata = {
646 	10000, 50000, 200000, 500000,
647 	10000, 50000, 200000, 500000,
648 	10000, 50000, 200000, 500000,
649 	10000, 50000, 200000, 500000,
650 	10000, 50000, 200000, 500000,
651 	10000, 50000, 200000, 500000,
652 	10000, 50000, 200000, 500000,
653 	10000, 50000, 200000, 500000,
654 	10000, 50000, 200000,
655 };
656 
657 static u64 intervals1[] __initdata = {
658 	223947000, 1240000, 1384000, 1386000, 1386000,
659 	217416000, 1236000, 1384000, 1386000, 1387000,
660 	214719000, 1241000, 1386000, 1387000, 1384000,
661 	213696000, 1234000, 1384000, 1386000, 1388000,
662 	219904000, 1240000, 1385000, 1389000, 1385000,
663 	212240000, 1240000, 1386000, 1386000, 1386000,
664 	214415000, 1236000, 1384000, 1386000, 1387000,
665 	214276000, 1234000,
666 };
667 
668 static u64 intervals2[] __initdata = {
669 	4000, 3000, 5000, 100000,
670 	3000, 3000, 5000, 117000,
671 	4000, 4000, 5000, 112000,
672 	4000, 3000, 4000, 110000,
673 	3000, 5000, 3000, 117000,
674 	4000, 4000, 5000, 112000,
675 	4000, 3000, 4000, 110000,
676 	3000, 4000, 5000, 112000,
677 	4000,
678 };
679 
680 static u64 intervals3[] __initdata = {
681 	1385000, 212240000, 1240000,
682 	1386000, 214415000, 1236000,
683 	1384000, 214276000, 1234000,
684 	1386000, 214415000, 1236000,
685 	1385000, 212240000, 1240000,
686 	1386000, 214415000, 1236000,
687 	1384000, 214276000, 1234000,
688 	1386000, 214415000, 1236000,
689 	1385000, 212240000, 1240000,
690 };
691 
692 static u64 intervals4[] __initdata = {
693 	10000, 50000, 10000, 50000,
694 	10000, 50000, 10000, 50000,
695 	10000, 50000, 10000, 50000,
696 	10000, 50000, 10000, 50000,
697 	10000, 50000, 10000, 50000,
698 	10000, 50000, 10000, 50000,
699 	10000, 50000, 10000, 50000,
700 	10000, 50000, 10000, 50000,
701 	10000,
702 };
703 
704 static struct timings_intervals tis[] __initdata = {
705 	{ intervals0, ARRAY_SIZE(intervals0) },
706 	{ intervals1, ARRAY_SIZE(intervals1) },
707 	{ intervals2, ARRAY_SIZE(intervals2) },
708 	{ intervals3, ARRAY_SIZE(intervals3) },
709 	{ intervals4, ARRAY_SIZE(intervals4) },
710 };
711 
712 static int __init irq_timings_test_next_index(struct timings_intervals *ti)
713 {
714 	int _buffer[IRQ_TIMINGS_SIZE];
715 	int buffer[IRQ_TIMINGS_SIZE];
716 	int index, start, i, count, period_max;
717 
718 	count = ti->count - 1;
719 
720 	period_max = count > (3 * PREDICTION_PERIOD_MAX) ?
721 		PREDICTION_PERIOD_MAX : count / 3;
722 
723 	/*
724 	 * Inject all values except the last one which will be used
725 	 * to compare with the next index result.
726 	 */
727 	pr_debug("index suite: ");
728 
729 	for (i = 0; i < count; i++) {
730 		index = irq_timings_interval_index(ti->intervals[i]);
731 		_buffer[i & IRQ_TIMINGS_MASK] = index;
732 		pr_cont("%d ", index);
733 	}
734 
735 	start = count < IRQ_TIMINGS_SIZE ? 0 :
736 		count & IRQ_TIMINGS_MASK;
737 
738 	count = min_t(int, count, IRQ_TIMINGS_SIZE);
739 
740 	for (i = 0; i < count; i++) {
741 		int index = (start + i) & IRQ_TIMINGS_MASK;
742 		buffer[i] = _buffer[index];
743 	}
744 
745 	index = irq_timings_next_event_index(buffer, count, period_max);
746 	i = irq_timings_interval_index(ti->intervals[ti->count - 1]);
747 
748 	if (index != i) {
749 		pr_err("Expected (%d) and computed (%d) next indexes differ\n",
750 		       i, index);
751 		return -EINVAL;
752 	}
753 
754 	return 0;
755 }
756 
757 static int __init irq_timings_next_index_selftest(void)
758 {
759 	int i, ret;
760 
761 	for (i = 0; i < ARRAY_SIZE(tis); i++) {
762 
763 		pr_info("---> Injecting intervals number #%d (count=%zd)\n",
764 			i, tis[i].count);
765 
766 		ret = irq_timings_test_next_index(&tis[i]);
767 		if (ret)
768 			break;
769 	}
770 
771 	return ret;
772 }
773 
774 static int __init irq_timings_test_irqs(struct timings_intervals *ti)
775 {
776 	struct irqt_stat __percpu *s;
777 	struct irqt_stat *irqs;
778 	int i, index, ret, irq = 0xACE5;
779 
780 	ret = irq_timings_alloc(irq);
781 	if (ret) {
782 		pr_err("Failed to allocate irq timings\n");
783 		return ret;
784 	}
785 
786 	s = idr_find(&irqt_stats, irq);
787 	if (!s) {
788 		ret = -EIDRM;
789 		goto out;
790 	}
791 
792 	irqs = this_cpu_ptr(s);
793 
794 	for (i = 0; i < ti->count; i++) {
795 
796 		index = irq_timings_interval_index(ti->intervals[i]);
797 		pr_debug("%d: interval=%llu ema_index=%d\n",
798 			 i, ti->intervals[i], index);
799 
800 		__irq_timings_store(irq, irqs, ti->intervals[i]);
801 		if (irqs->circ_timings[i & IRQ_TIMINGS_MASK] != index) {
802 			ret = -EBADSLT;
803 			pr_err("Failed to store in the circular buffer\n");
804 			goto out;
805 		}
806 	}
807 
808 	if (irqs->count != ti->count) {
809 		ret = -ERANGE;
810 		pr_err("Count differs\n");
811 		goto out;
812 	}
813 
814 	ret = 0;
815 out:
816 	irq_timings_free(irq);
817 
818 	return ret;
819 }
820 
821 static int __init irq_timings_irqs_selftest(void)
822 {
823 	int i, ret;
824 
825 	for (i = 0; i < ARRAY_SIZE(tis); i++) {
826 		pr_info("---> Injecting intervals number #%d (count=%zd)\n",
827 			i, tis[i].count);
828 		ret = irq_timings_test_irqs(&tis[i]);
829 		if (ret)
830 			break;
831 	}
832 
833 	return ret;
834 }
835 
836 static int __init irq_timings_test_irqts(struct irq_timings *irqts,
837 					 unsigned count)
838 {
839 	int start = count >= IRQ_TIMINGS_SIZE ? count - IRQ_TIMINGS_SIZE : 0;
840 	int i, irq, oirq = 0xBEEF;
841 	u64 ots = 0xDEAD, ts;
842 
843 	/*
844 	 * Fill the circular buffer by using the dedicated function.
845 	 */
846 	for (i = 0; i < count; i++) {
847 		pr_debug("%d: index=%d, ts=%llX irq=%X\n",
848 			 i, i & IRQ_TIMINGS_MASK, ots + i, oirq + i);
849 
850 		irq_timings_push(ots + i, oirq + i);
851 	}
852 
853 	/*
854 	 * Compute the first elements values after the index wrapped
855 	 * up or not.
856 	 */
857 	ots += start;
858 	oirq += start;
859 
860 	/*
861 	 * Test the circular buffer count is correct.
862 	 */
863 	pr_debug("---> Checking timings array count (%d) is right\n", count);
864 	if (WARN_ON(irqts->count != count))
865 		return -EINVAL;
866 
867 	/*
868 	 * Test the macro allowing to browse all the irqts.
869 	 */
870 	pr_debug("---> Checking the for_each_irqts() macro\n");
871 	for_each_irqts(i, irqts) {
872 
873 		irq = irq_timing_decode(irqts->values[i], &ts);
874 
875 		pr_debug("index=%d, ts=%llX / %llX, irq=%X / %X\n",
876 			 i, ts, ots, irq, oirq);
877 
878 		if (WARN_ON(ts != ots || irq != oirq))
879 			return -EINVAL;
880 
881 		ots++; oirq++;
882 	}
883 
884 	/*
885 	 * The circular buffer should have be flushed when browsed
886 	 * with for_each_irqts
887 	 */
888 	pr_debug("---> Checking timings array is empty after browsing it\n");
889 	if (WARN_ON(irqts->count))
890 		return -EINVAL;
891 
892 	return 0;
893 }
894 
895 static int __init irq_timings_irqts_selftest(void)
896 {
897 	struct irq_timings *irqts = this_cpu_ptr(&irq_timings);
898 	int i, ret;
899 
900 	/*
901 	 * Test the circular buffer with different number of
902 	 * elements. The purpose is to test at the limits (empty, half
903 	 * full, full, wrapped with the cursor at the boundaries,
904 	 * wrapped several times, etc ...
905 	 */
906 	int count[] = { 0,
907 			IRQ_TIMINGS_SIZE >> 1,
908 			IRQ_TIMINGS_SIZE,
909 			IRQ_TIMINGS_SIZE + (IRQ_TIMINGS_SIZE >> 1),
910 			2 * IRQ_TIMINGS_SIZE,
911 			(2 * IRQ_TIMINGS_SIZE) + 3,
912 	};
913 
914 	for (i = 0; i < ARRAY_SIZE(count); i++) {
915 
916 		pr_info("---> Checking the timings with %d/%d values\n",
917 			count[i], IRQ_TIMINGS_SIZE);
918 
919 		ret = irq_timings_test_irqts(irqts, count[i]);
920 		if (ret)
921 			break;
922 	}
923 
924 	return ret;
925 }
926 
927 static int __init irq_timings_selftest(void)
928 {
929 	int ret;
930 
931 	pr_info("------------------- selftest start -----------------\n");
932 
933 	/*
934 	 * At this point, we don't except any subsystem to use the irq
935 	 * timings but us, so it should not be enabled.
936 	 */
937 	if (static_branch_unlikely(&irq_timing_enabled)) {
938 		pr_warn("irq timings already initialized, skipping selftest\n");
939 		return 0;
940 	}
941 
942 	ret = irq_timings_irqts_selftest();
943 	if (ret)
944 		goto out;
945 
946 	ret = irq_timings_irqs_selftest();
947 	if (ret)
948 		goto out;
949 
950 	ret = irq_timings_next_index_selftest();
951 out:
952 	pr_info("---------- selftest end with %s -----------\n",
953 		ret ? "failure" : "success");
954 
955 	return ret;
956 }
957 early_initcall(irq_timings_selftest);
958 #endif
959