summaryrefslogtreecommitdiffstats
path: root/drivers/perf/fsl_imx8_ddr_perf.c
blob: a11bfd8a08230c2760b2969ab74f25a20f7fff96 (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
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
626
627
628
629
630
631
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
675
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
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
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
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright 2017 NXP
 * Copyright 2016 Freescale Semiconductor, Inc.
 */

#include <linux/bitfield.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/io.h>
#include <linux/module.h>
#include <linux/of.h>
#include <linux/of_address.h>
#include <linux/of_device.h>
#include <linux/of_irq.h>
#include <linux/perf_event.h>
#include <linux/slab.h>

#define COUNTER_CNTL		0x0
#define COUNTER_READ		0x20

#define COUNTER_DPCR1		0x30

#define CNTL_OVER		0x1
#define CNTL_CLEAR		0x2
#define CNTL_EN			0x4
#define CNTL_EN_MASK		0xFFFFFFFB
#define CNTL_CLEAR_MASK		0xFFFFFFFD
#define CNTL_OVER_MASK		0xFFFFFFFE

#define CNTL_CSV_SHIFT		24
#define CNTL_CSV_MASK		(0xFF << CNTL_CSV_SHIFT)

#define EVENT_CYCLES_ID		0
#define EVENT_CYCLES_COUNTER	0
#define NUM_COUNTERS		4

#define AXI_MASKING_REVERT	0xffff0000	/* AXI_MASKING(MSB 16bits) + AXI_ID(LSB 16bits) */

#define to_ddr_pmu(p)		container_of(p, struct ddr_pmu, pmu)

#define DDR_PERF_DEV_NAME	"imx8_ddr"
#define DDR_CPUHP_CB_NAME	DDR_PERF_DEV_NAME "_perf_pmu"

static DEFINE_IDA(ddr_ida);

/* DDR Perf hardware feature */
#define DDR_CAP_AXI_ID_FILTER			0x1     /* support AXI ID filter */
#define DDR_CAP_AXI_ID_FILTER_ENHANCED		0x3     /* support enhanced AXI ID filter */

struct fsl_ddr_devtype_data {
	unsigned int quirks;    /* quirks needed for different DDR Perf core */
	const char *identifier;	/* system PMU identifier for userspace */
};

static const struct fsl_ddr_devtype_data imx8_devtype_data;

static const struct fsl_ddr_devtype_data imx8m_devtype_data = {
	.quirks = DDR_CAP_AXI_ID_FILTER,
};

static const struct fsl_ddr_devtype_data imx8mq_devtype_data = {
	.quirks = DDR_CAP_AXI_ID_FILTER,
	.identifier = "i.MX8MQ",
};

static const struct fsl_ddr_devtype_data imx8mm_devtype_data = {
	.quirks = DDR_CAP_AXI_ID_FILTER,
	.identifier = "i.MX8MM",
};

static const struct fsl_ddr_devtype_data imx8mn_devtype_data = {
	.quirks = DDR_CAP_AXI_ID_FILTER,
	.identifier = "i.MX8MN",
};

static const struct fsl_ddr_devtype_data imx8mp_devtype_data = {
	.quirks = DDR_CAP_AXI_ID_FILTER_ENHANCED,
	.identifier = "i.MX8MP",
};

static const struct of_device_id imx_ddr_pmu_dt_ids[] = {
	{ .compatible = "fsl,imx8-ddr-pmu", .data = &imx8_devtype_data},
	{ .compatible = "fsl,imx8m-ddr-pmu", .data = &imx8m_devtype_data},
	{ .compatible = "fsl,imx8mq-ddr-pmu", .data = &imx8mq_devtype_data},
	{ .compatible = "fsl,imx8mm-ddr-pmu", .data = &imx8mm_devtype_data},
	{ .compatible = "fsl,imx8mn-ddr-pmu", .data = &imx8mn_devtype_data},
	{ .compatible = "fsl,imx8mp-ddr-pmu", .data = &imx8mp_devtype_data},
	{ /* sentinel */ }
};
MODULE_DEVICE_TABLE(of, imx_ddr_pmu_dt_ids);

struct ddr_pmu {
	struct pmu pmu;
	void __iomem *base;
	unsigned int cpu;
	struct	hlist_node node;
	struct	device *dev;
	struct perf_event *events[NUM_COUNTERS];
	int active_events;
	enum cpuhp_state cpuhp_state;
	const struct fsl_ddr_devtype_data *devtype_data;
	int irq;
	int id;
};

static ssize_t ddr_perf_identifier_show(struct device *dev,
					struct device_attribute *attr,
					char *page)
{
	struct ddr_pmu *pmu = dev_get_drvdata(dev);

	return sprintf(page, "%s\n", pmu->devtype_data->identifier);
}

static umode_t ddr_perf_identifier_attr_visible(struct kobject *kobj,
						struct attribute *attr,
						int n)
{
	struct device *dev = kobj_to_dev(kobj);
	struct ddr_pmu *pmu = dev_get_drvdata(dev);

	if (!pmu->devtype_data->identifier)
		return 0;
	return attr->mode;
};

static struct device_attribute ddr_perf_identifier_attr =
	__ATTR(identifier, 0444, ddr_perf_identifier_show, NULL);

static struct attribute *ddr_perf_identifier_attrs[] = {
	&ddr_perf_identifier_attr.attr,
	NULL,
};

static struct attribute_group ddr_perf_identifier_attr_group = {
	.attrs = ddr_perf_identifier_attrs,
	.is_visible = ddr_perf_identifier_attr_visible,
};

enum ddr_perf_filter_capabilities {
	PERF_CAP_AXI_ID_FILTER = 0,
	PERF_CAP_AXI_ID_FILTER_ENHANCED,
	PERF_CAP_AXI_ID_FEAT_MAX,
};

static u32 ddr_perf_filter_cap_get(struct ddr_pmu *pmu, int cap)
{
	u32 quirks = pmu->devtype_data->quirks;

	switch (cap) {
	case PERF_CAP_AXI_ID_FILTER:
		return !!(quirks & DDR_CAP_AXI_ID_FILTER);
	case PERF_CAP_AXI_ID_FILTER_ENHANCED:
		quirks &= DDR_CAP_AXI_ID_FILTER_ENHANCED;
		return quirks == DDR_CAP_AXI_ID_FILTER_ENHANCED;
	default:
		WARN(1, "unknown filter cap %d\n", cap);
	}

	return 0;
}

static ssize_t ddr_perf_filter_cap_show(struct device *dev,
					struct device_attribute *attr,
					char *buf)
{
	struct ddr_pmu *pmu = dev_get_drvdata(dev);
	struct dev_ext_attribute *ea =
		container_of(attr, struct dev_ext_attribute, attr);
	int cap = (long)ea->var;

	return snprintf(buf, PAGE_SIZE, "%u\n",
			ddr_perf_filter_cap_get(pmu, cap));
}

#define PERF_EXT_ATTR_ENTRY(_name, _func, _var)				\
	(&((struct dev_ext_attribute) {					\
		__ATTR(_name, 0444, _func, NULL), (void *)_var		\
	}).attr.attr)

#define PERF_FILTER_EXT_ATTR_ENTRY(_name, _var)				\
	PERF_EXT_ATTR_ENTRY(_name, ddr_perf_filter_cap_show, _var)

static struct attribute *ddr_perf_filter_cap_attr[] = {
	PERF_FILTER_EXT_ATTR_ENTRY(filter, PERF_CAP_AXI_ID_FILTER),
	PERF_FILTER_EXT_ATTR_ENTRY(enhanced_filter, PERF_CAP_AXI_ID_FILTER_ENHANCED),
	NULL,
};

static struct attribute_group ddr_perf_filter_cap_attr_group = {
	.name = "caps",
	.attrs = ddr_perf_filter_cap_attr,
};

static ssize_t ddr_perf_cpumask_show(struct device *dev,
				struct device_attribute *attr, char *buf)
{
	struct ddr_pmu *pmu = dev_get_drvdata(dev);

	return cpumap_print_to_pagebuf(true, buf, cpumask_of(pmu->cpu));
}

static struct device_attribute ddr_perf_cpumask_attr =
	__ATTR(cpumask, 0444, ddr_perf_cpumask_show, NULL);

static struct attribute *ddr_perf_cpumask_attrs[] = {
	&ddr_perf_cpumask_attr.attr,
	NULL,
};

static struct attribute_group ddr_perf_cpumask_attr_group = {
	.attrs = ddr_perf_cpumask_attrs,
};

static ssize_t
ddr_pmu_event_show(struct device *dev, struct device_attribute *attr,
		   char *page)
{
	struct perf_pmu_events_attr *pmu_attr;

	pmu_attr = container_of(attr, struct perf_pmu_events_attr, attr);
	return sprintf(page, "event=0x%02llx\n", pmu_attr->id);
}

#define IMX8_DDR_PMU_EVENT_ATTR(_name, _id)				\
	(&((struct perf_pmu_events_attr[]) {				\
		{ .attr = __ATTR(_name, 0444, ddr_pmu_event_show, NULL),\
		  .id = _id, }						\
	})[0].attr.attr)

static struct attribute *ddr_perf_events_attrs[] = {
	IMX8_DDR_PMU_EVENT_ATTR(cycles, EVENT_CYCLES_ID),
	IMX8_DDR_PMU_EVENT_ATTR(selfresh, 0x01),
	IMX8_DDR_PMU_EVENT_ATTR(read-accesses, 0x04),
	IMX8_DDR_PMU_EVENT_ATTR(write-accesses, 0x05),
	IMX8_DDR_PMU_EVENT_ATTR(read-queue-depth, 0x08),
	IMX8_DDR_PMU_EVENT_ATTR(write-queue-depth, 0x09),
	IMX8_DDR_PMU_EVENT_ATTR(lp-read-credit-cnt, 0x10),
	IMX8_DDR_PMU_EVENT_ATTR(hp-read-credit-cnt, 0x11),
	IMX8_DDR_PMU_EVENT_ATTR(write-credit-cnt, 0x12),
	IMX8_DDR_PMU_EVENT_ATTR(read-command, 0x20),
	IMX8_DDR_PMU_EVENT_ATTR(write-command, 0x21),
	IMX8_DDR_PMU_EVENT_ATTR(read-modify-write-command, 0x22),
	IMX8_DDR_PMU_EVENT_ATTR(hp-read, 0x23),
	IMX8_DDR_PMU_EVENT_ATTR(hp-req-nocredit, 0x24),
	IMX8_DDR_PMU_EVENT_ATTR(hp-xact-credit, 0x25),
	IMX8_DDR_PMU_EVENT_ATTR(lp-req-nocredit, 0x26),
	IMX8_DDR_PMU_EVENT_ATTR(lp-xact-credit, 0x27),
	IMX8_DDR_PMU_EVENT_ATTR(wr-xact-credit, 0x29),
	IMX8_DDR_PMU_EVENT_ATTR(read-cycles, 0x2a),
	IMX8_DDR_PMU_EVENT_ATTR(write-cycles, 0x2b),
	IMX8_DDR_PMU_EVENT_ATTR(read-write-transition, 0x30),
	IMX8_DDR_PMU_EVENT_ATTR(precharge, 0x31),
	IMX8_DDR_PMU_EVENT_ATTR(activate, 0x32),
	IMX8_DDR_PMU_EVENT_ATTR(load-mode, 0x33),
	IMX8_DDR_PMU_EVENT_ATTR(perf-mwr, 0x34),
	IMX8_DDR_PMU_EVENT_ATTR(read, 0x35),
	IMX8_DDR_PMU_EVENT_ATTR(read-activate, 0x36),
	IMX8_DDR_PMU_EVENT_ATTR(refresh, 0x37),
	IMX8_DDR_PMU_EVENT_ATTR(write, 0x38),
	IMX8_DDR_PMU_EVENT_ATTR(raw-hazard, 0x39),
	IMX8_DDR_PMU_EVENT_ATTR(axid-read, 0x41),
	IMX8_DDR_PMU_EVENT_ATTR(axid-write, 0x42),
	NULL,
};

static struct attribute_group ddr_perf_events_attr_group = {
	.name = "events",
	.attrs = ddr_perf_events_attrs,
};

PMU_FORMAT_ATTR(event, "config:0-7");
PMU_FORMAT_ATTR(axi_id, "config1:0-15");
PMU_FORMAT_ATTR(axi_mask, "config1:16-31");

static struct attribute *ddr_perf_format_attrs[] = {
	&format_attr_event.attr,
	&format_attr_axi_id.attr,
	&format_attr_axi_mask.attr,
	NULL,
};

static struct attribute_group ddr_perf_format_attr_group = {
	.name = "format",
	.attrs = ddr_perf_format_attrs,
};

static const struct attribute_group *attr_groups[] = {
	&ddr_perf_events_attr_group,
	&ddr_perf_format_attr_group,
	&ddr_perf_cpumask_attr_group,
	&ddr_perf_filter_cap_attr_group,
	&ddr_perf_identifier_attr_group,
	NULL,
};

static bool ddr_perf_is_filtered(struct perf_event *event)
{
	return event->attr.config == 0x41 || event->attr.config == 0x42;
}

static u32 ddr_perf_filter_val(struct perf_event *event)
{
	return event->attr.config1;
}

static bool ddr_perf_filters_compatible(struct perf_event *a,
					struct perf_event *b)
{
	if (!ddr_perf_is_filtered(a))
		return true;
	if (!ddr_perf_is_filtered(b))
		return true;
	return ddr_perf_filter_val(a) == ddr_perf_filter_val(b);
}

static bool ddr_perf_is_enhanced_filtered(struct perf_event *event)
{
	unsigned int filt;
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);

	filt = pmu->devtype_data->quirks & DDR_CAP_AXI_ID_FILTER_ENHANCED;
	return (filt == DDR_CAP_AXI_ID_FILTER_ENHANCED) &&
		ddr_perf_is_filtered(event);
}

static u32 ddr_perf_alloc_counter(struct ddr_pmu *pmu, int event)
{
	int i;

	/*
	 * Always map cycle event to counter 0
	 * Cycles counter is dedicated for cycle event
	 * can't used for the other events
	 */
	if (event == EVENT_CYCLES_ID) {
		if (pmu->events[EVENT_CYCLES_COUNTER] == NULL)
			return EVENT_CYCLES_COUNTER;
		else
			return -ENOENT;
	}

	for (i = 1; i < NUM_COUNTERS; i++) {
		if (pmu->events[i] == NULL)
			return i;
	}

	return -ENOENT;
}

static void ddr_perf_free_counter(struct ddr_pmu *pmu, int counter)
{
	pmu->events[counter] = NULL;
}

static u32 ddr_perf_read_counter(struct ddr_pmu *pmu, int counter)
{
	struct perf_event *event = pmu->events[counter];
	void __iomem *base = pmu->base;

	/*
	 * return bytes instead of bursts from ddr transaction for
	 * axid-read and axid-write event if PMU core supports enhanced
	 * filter.
	 */
	base += ddr_perf_is_enhanced_filtered(event) ? COUNTER_DPCR1 :
						       COUNTER_READ;
	return readl_relaxed(base + counter * 4);
}

static int ddr_perf_event_init(struct perf_event *event)
{
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
	struct hw_perf_event *hwc = &event->hw;
	struct perf_event *sibling;

	if (event->attr.type != event->pmu->type)
		return -ENOENT;

	if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
		return -EOPNOTSUPP;

	if (event->cpu < 0) {
		dev_warn(pmu->dev, "Can't provide per-task data!\n");
		return -EOPNOTSUPP;
	}

	/*
	 * We must NOT create groups containing mixed PMUs, although software
	 * events are acceptable (for example to create a CCN group
	 * periodically read when a hrtimer aka cpu-clock leader triggers).
	 */
	if (event->group_leader->pmu != event->pmu &&
			!is_software_event(event->group_leader))
		return -EINVAL;

	if (pmu->devtype_data->quirks & DDR_CAP_AXI_ID_FILTER) {
		if (!ddr_perf_filters_compatible(event, event->group_leader))
			return -EINVAL;
		for_each_sibling_event(sibling, event->group_leader) {
			if (!ddr_perf_filters_compatible(event, sibling))
				return -EINVAL;
		}
	}

	for_each_sibling_event(sibling, event->group_leader) {
		if (sibling->pmu != event->pmu &&
				!is_software_event(sibling))
			return -EINVAL;
	}

	event->cpu = pmu->cpu;
	hwc->idx = -1;

	return 0;
}

static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config,
				  int counter, bool enable)
{
	u8 reg = counter * 4 + COUNTER_CNTL;
	int val;

	if (enable) {
		/*
		 * cycle counter is special which should firstly write 0 then
		 * write 1 into CLEAR bit to clear it. Other counters only
		 * need write 0 into CLEAR bit and it turns out to be 1 by
		 * hardware. Below enable flow is harmless for all counters.
		 */
		writel(0, pmu->base + reg);
		val = CNTL_EN | CNTL_CLEAR;
		val |= FIELD_PREP(CNTL_CSV_MASK, config);
		writel(val, pmu->base + reg);
	} else {
		/* Disable counter */
		val = readl_relaxed(pmu->base + reg) & CNTL_EN_MASK;
		writel(val, pmu->base + reg);
	}
}

static bool ddr_perf_counter_overflow(struct ddr_pmu *pmu, int counter)
{
	int val;

	val = readl_relaxed(pmu->base + counter * 4 + COUNTER_CNTL);

	return val & CNTL_OVER;
}

static void ddr_perf_counter_clear(struct ddr_pmu *pmu, int counter)
{
	u8 reg = counter * 4 + COUNTER_CNTL;
	int val;

	val = readl_relaxed(pmu->base + reg);
	val &= ~CNTL_CLEAR;
	writel(val, pmu->base + reg);

	val |= CNTL_CLEAR;
	writel(val, pmu->base + reg);
}

static void ddr_perf_event_update(struct perf_event *event)
{
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
	struct hw_perf_event *hwc = &event->hw;
	u64 new_raw_count;
	int counter = hwc->idx;
	int ret;

	new_raw_count = ddr_perf_read_counter(pmu, counter);
	local64_add(new_raw_count, &event->count);

	/*
	 * For legacy SoCs: event counter continue counting when overflow,
	 *                  no need to clear the counter.
	 * For new SoCs: event counter stop counting when overflow, need
	 *               clear counter to let it count again.
	 */
	if (counter != EVENT_CYCLES_COUNTER) {
		ret = ddr_perf_counter_overflow(pmu, counter);
		if (ret)
			dev_warn_ratelimited(pmu->dev,  "events lost due to counter overflow (config 0x%llx)\n",
					     event->attr.config);
	}

	/* clear counter every time for both cycle counter and event counter */
	ddr_perf_counter_clear(pmu, counter);
}

static void ddr_perf_event_start(struct perf_event *event, int flags)
{
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
	struct hw_perf_event *hwc = &event->hw;
	int counter = hwc->idx;

	local64_set(&hwc->prev_count, 0);

	ddr_perf_counter_enable(pmu, event->attr.config, counter, true);

	hwc->state = 0;
}

static int ddr_perf_event_add(struct perf_event *event, int flags)
{
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
	struct hw_perf_event *hwc = &event->hw;
	int counter;
	int cfg = event->attr.config;
	int cfg1 = event->attr.config1;

	if (pmu->devtype_data->quirks & DDR_CAP_AXI_ID_FILTER) {
		int i;

		for (i = 1; i < NUM_COUNTERS; i++) {
			if (pmu->events[i] &&
			    !ddr_perf_filters_compatible(event, pmu->events[i]))
				return -EINVAL;
		}

		if (ddr_perf_is_filtered(event)) {
			/* revert axi id masking(axi_mask) value */
			cfg1 ^= AXI_MASKING_REVERT;
			writel(cfg1, pmu->base + COUNTER_DPCR1);
		}
	}

	counter = ddr_perf_alloc_counter(pmu, cfg);
	if (counter < 0) {
		dev_dbg(pmu->dev, "There are not enough counters\n");
		return -EOPNOTSUPP;
	}

	pmu->events[counter] = event;
	pmu->active_events++;
	hwc->idx = counter;

	hwc->state |= PERF_HES_STOPPED;

	if (flags & PERF_EF_START)
		ddr_perf_event_start(event, flags);

	return 0;
}

static void ddr_perf_event_stop(struct perf_event *event, int flags)
{
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
	struct hw_perf_event *hwc = &event->hw;
	int counter = hwc->idx;

	ddr_perf_counter_enable(pmu, event->attr.config, counter, false);
	ddr_perf_event_update(event);

	hwc->state |= PERF_HES_STOPPED;
}

static void ddr_perf_event_del(struct perf_event *event, int flags)
{
	struct ddr_pmu *pmu = to_ddr_pmu(event->pmu);
	struct hw_perf_event *hwc = &event->hw;
	int counter = hwc->idx;

	ddr_perf_event_stop(event, PERF_EF_UPDATE);

	ddr_perf_free_counter(pmu, counter);
	pmu->active_events--;
	hwc->idx = -1;
}

static void ddr_perf_pmu_enable(struct pmu *pmu)
{
	struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);

	/* enable cycle counter if cycle is not active event list */
	if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
		ddr_perf_counter_enable(ddr_pmu,
				      EVENT_CYCLES_ID,
				      EVENT_CYCLES_COUNTER,
				      true);
}

static void ddr_perf_pmu_disable(struct pmu *pmu)
{
	struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu);

	if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL)
		ddr_perf_counter_enable(ddr_pmu,
				      EVENT_CYCLES_ID,
				      EVENT_CYCLES_COUNTER,
				      false);
}

static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base,
			 struct device *dev)
{
	*pmu = (struct ddr_pmu) {
		.pmu = (struct pmu) {
			.module	      = THIS_MODULE,
			.capabilities = PERF_PMU_CAP_NO_EXCLUDE,
			.task_ctx_nr = perf_invalid_context,
			.attr_groups = attr_groups,
			.event_init  = ddr_perf_event_init,
			.add	     = ddr_perf_event_add,
			.del	     = ddr_perf_event_del,
			.start	     = ddr_perf_event_start,
			.stop	     = ddr_perf_event_stop,
			.read	     = ddr_perf_event_update,
			.pmu_enable  = ddr_perf_pmu_enable,
			.pmu_disable = ddr_perf_pmu_disable,
		},
		.base = base,
		.dev = dev,
	};

	pmu->id = ida_simple_get(&ddr_ida, 0, 0, GFP_KERNEL);
	return pmu->id;
}

static irqreturn_t ddr_perf_irq_handler(int irq, void *p)
{
	int i;
	struct ddr_pmu *pmu = (struct ddr_pmu *) p;
	struct perf_event *event;

	/* all counter will stop if cycle counter disabled */
	ddr_perf_counter_enable(pmu,
			      EVENT_CYCLES_ID,
			      EVENT_CYCLES_COUNTER,
			      false);
	/*
	 * When the cycle counter overflows, all counters are stopped,
	 * and an IRQ is raised. If any other counter overflows, it
	 * continues counting, and no IRQ is raised. But for new SoCs,
	 * such as i.MX8MP, event counter would stop when overflow, so
	 * we need use cycle counter to stop overflow of event counter.
	 *
	 * Cycles occur at least 4 times as often as other events, so we
	 * can update all events on a cycle counter overflow and not
	 * lose events.
	 *
	 */
	for (i = 0; i < NUM_COUNTERS; i++) {

		if (!pmu->events[i])
			continue;

		event = pmu->events[i];

		ddr_perf_event_update(event);
	}

	ddr_perf_counter_enable(pmu,
			      EVENT_CYCLES_ID,
			      EVENT_CYCLES_COUNTER,
			      true);

	return IRQ_HANDLED;
}

static int ddr_perf_offline_cpu(unsigned int cpu, struct hlist_node *node)
{
	struct ddr_pmu *pmu = hlist_entry_safe(node, struct ddr_pmu, node);
	int target;

	if (cpu != pmu->cpu)
		return 0;

	target = cpumask_any_but(cpu_online_mask, cpu);
	if (target >= nr_cpu_ids)
		return 0;

	perf_pmu_migrate_context(&pmu->pmu, cpu, target);
	pmu->cpu = target;

	WARN_ON(irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu)));

	return 0;
}

static int ddr_perf_probe(struct platform_device *pdev)
{
	struct ddr_pmu *pmu;
	struct device_node *np;
	void __iomem *base;
	char *name;
	int num;
	int ret;
	int irq;

	base = devm_platform_ioremap_resource(pdev, 0);
	if (IS_ERR(base))
		return PTR_ERR(base);

	np = pdev->dev.of_node;

	pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL);
	if (!pmu)
		return -ENOMEM;

	num = ddr_perf_init(pmu, base, &pdev->dev);

	platform_set_drvdata(pdev, pmu);

	name = devm_kasprintf(&pdev->dev, GFP_KERNEL, DDR_PERF_DEV_NAME "%d",
			      num);
	if (!name)
		return -ENOMEM;

	pmu->devtype_data = of_device_get_match_data(&pdev->dev);

	pmu->cpu = raw_smp_processor_id();
	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
				      DDR_CPUHP_CB_NAME,
				      NULL,
				      ddr_perf_offline_cpu);

	if (ret < 0) {
		dev_err(&pdev->dev, "cpuhp_setup_state_multi failed\n");
		goto cpuhp_state_err;
	}

	pmu->cpuhp_state = ret;

	/* Register the pmu instance for cpu hotplug */
	ret = cpuhp_state_add_instance_nocalls(pmu->cpuhp_state, &pmu->node);
	if (ret) {
		dev_err(&pdev->dev, "Error %d registering hotplug\n", ret);
		goto cpuhp_instance_err;
	}

	/* Request irq */
	irq = of_irq_get(np, 0);
	if (irq < 0) {
		dev_err(&pdev->dev, "Failed to get irq: %d", irq);
		ret = irq;
		goto ddr_perf_err;
	}

	ret = devm_request_irq(&pdev->dev, irq,
					ddr_perf_irq_handler,
					IRQF_NOBALANCING | IRQF_NO_THREAD,
					DDR_CPUHP_CB_NAME,
					pmu);
	if (ret < 0) {
		dev_err(&pdev->dev, "Request irq failed: %d", ret);
		goto ddr_perf_err;
	}

	pmu->irq = irq;
	ret = irq_set_affinity_hint(pmu->irq, cpumask_of(pmu->cpu));
	if (ret) {
		dev_err(pmu->dev, "Failed to set interrupt affinity!\n");
		goto ddr_perf_err;
	}

	ret = perf_pmu_register(&pmu->pmu, name, -1);
	if (ret)
		goto ddr_perf_err;

	return 0;

ddr_perf_err:
	cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
cpuhp_instance_err:
	cpuhp_remove_multi_state(pmu->cpuhp_state);
cpuhp_state_err:
	ida_simple_remove(&ddr_ida, pmu->id);
	dev_warn(&pdev->dev, "i.MX8 DDR Perf PMU failed (%d), disabled\n", ret);
	return ret;
}

static int ddr_perf_remove(struct platform_device *pdev)
{
	struct ddr_pmu *pmu = platform_get_drvdata(pdev);

	cpuhp_state_remove_instance_nocalls(pmu->cpuhp_state, &pmu->node);
	cpuhp_remove_multi_state(pmu->cpuhp_state);
	irq_set_affinity_hint(pmu->irq, NULL);

	perf_pmu_unregister(&pmu->pmu);

	ida_simple_remove(&ddr_ida, pmu->id);
	return 0;
}

static struct platform_driver imx_ddr_pmu_driver = {
	.driver         = {
		.name   = "imx-ddr-pmu",
		.of_match_table = imx_ddr_pmu_dt_ids,
		.suppress_bind_attrs = true,
	},
	.probe          = ddr_perf_probe,
	.remove         = ddr_perf_remove,
};

module_platform_driver(imx_ddr_pmu_driver);
MODULE_LICENSE("GPL v2");