evlist.c 13.7 KB
Newer Older
1
2
3
4
5
6
7
8
/*
 * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
 *
 * Parts came from builtin-{top,stat,record}.c, see those files for further
 * copyright notes.
 *
 * Released under the GPL v2. (and only v2, not any later version)
 */
9
#include <poll.h>
10
11
#include "cpumap.h"
#include "thread_map.h"
12
13
14
15
#include "evlist.h"
#include "evsel.h"
#include "util.h"

16
17
#include <sys/mman.h>

18
19
20
#include <linux/bitops.h>
#include <linux/hash.h>

21
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
22
#define SID(e, x, y) xyarray__entry(e->sample_id, x, y)
23

24
25
void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus,
		       struct thread_map *threads)
26
27
28
29
30
31
{
	int i;

	for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i)
		INIT_HLIST_HEAD(&evlist->heads[i]);
	INIT_LIST_HEAD(&evlist->entries);
32
	perf_evlist__set_maps(evlist, cpus, threads);
33
34
}

35
36
struct perf_evlist *perf_evlist__new(struct cpu_map *cpus,
				     struct thread_map *threads)
37
38
39
{
	struct perf_evlist *evlist = zalloc(sizeof(*evlist));

40
	if (evlist != NULL)
41
		perf_evlist__init(evlist, cpus, threads);
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57

	return evlist;
}

static void perf_evlist__purge(struct perf_evlist *evlist)
{
	struct perf_evsel *pos, *n;

	list_for_each_entry_safe(pos, n, &evlist->entries, node) {
		list_del_init(&pos->node);
		perf_evsel__delete(pos);
	}

	evlist->nr_entries = 0;
}

58
void perf_evlist__exit(struct perf_evlist *evlist)
59
{
60
	free(evlist->mmap);
61
	free(evlist->pollfd);
62
63
64
65
66
67
68
69
	evlist->mmap = NULL;
	evlist->pollfd = NULL;
}

void perf_evlist__delete(struct perf_evlist *evlist)
{
	perf_evlist__purge(evlist);
	perf_evlist__exit(evlist);
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
	free(evlist);
}

void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry)
{
	list_add_tail(&entry->node, &evlist->entries);
	++evlist->nr_entries;
}

int perf_evlist__add_default(struct perf_evlist *evlist)
{
	struct perf_event_attr attr = {
		.type = PERF_TYPE_HARDWARE,
		.config = PERF_COUNT_HW_CPU_CYCLES,
	};
	struct perf_evsel *evsel = perf_evsel__new(&attr, 0);

	if (evsel == NULL)
88
89
90
91
92
93
		goto error;

	/* use strdup() because free(evsel) assumes name is allocated */
	evsel->name = strdup("cycles");
	if (!evsel->name)
		goto error_free;
94
95
96

	perf_evlist__add(evlist, evsel);
	return 0;
97
98
99
100
error_free:
	perf_evsel__delete(evsel);
error:
	return -ENOMEM;
101
}
102

103
104
105
106
107
108
109
110
111
112
113
114
115
void perf_evlist__disable(struct perf_evlist *evlist)
{
	int cpu, thread;
	struct perf_evsel *pos;

	for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
		list_for_each_entry(pos, &evlist->entries, node) {
			for (thread = 0; thread < evlist->threads->nr; thread++)
				ioctl(FD(pos, cpu, thread), PERF_EVENT_IOC_DISABLE);
		}
	}
}

116
117
118
119
120
121
122
123
124
125
126
127
128
void perf_evlist__enable(struct perf_evlist *evlist)
{
	int cpu, thread;
	struct perf_evsel *pos;

	for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
		list_for_each_entry(pos, &evlist->entries, node) {
			for (thread = 0; thread < evlist->threads->nr; thread++)
				ioctl(FD(pos, cpu, thread), PERF_EVENT_IOC_ENABLE);
		}
	}
}

129
int perf_evlist__alloc_pollfd(struct perf_evlist *evlist)
130
{
131
	int nfds = evlist->cpus->nr * evlist->threads->nr * evlist->nr_entries;
132
133
134
	evlist->pollfd = malloc(sizeof(struct pollfd) * nfds);
	return evlist->pollfd != NULL ? 0 : -ENOMEM;
}
135
136
137
138
139
140
141
142

void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd)
{
	fcntl(fd, F_SETFL, O_NONBLOCK);
	evlist->pollfd[evlist->nr_fds].fd = fd;
	evlist->pollfd[evlist->nr_fds].events = POLLIN;
	evlist->nr_fds++;
}
143

144
145
146
static void perf_evlist__id_hash(struct perf_evlist *evlist,
				 struct perf_evsel *evsel,
				 int cpu, int thread, u64 id)
147
148
149
150
151
152
153
154
155
156
{
	int hash;
	struct perf_sample_id *sid = SID(evsel, cpu, thread);

	sid->id = id;
	sid->evsel = evsel;
	hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS);
	hlist_add_head(&sid->node, &evlist->heads[hash]);
}

157
158
159
160
161
162
163
164
165
166
void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel,
			 int cpu, int thread, u64 id)
{
	perf_evlist__id_hash(evlist, evsel, cpu, thread, id);
	evsel->id[evsel->ids++] = id;
}

static int perf_evlist__id_add_fd(struct perf_evlist *evlist,
				  struct perf_evsel *evsel,
				  int cpu, int thread, int fd)
167
168
{
	u64 read_data[4] = { 0, };
169
	int id_idx = 1; /* The first entry is the counter value */
170
171
172
173
174
175
176
177
178
179

	if (!(evsel->attr.read_format & PERF_FORMAT_ID) ||
	    read(fd, &read_data, sizeof(read_data)) == -1)
		return -1;

	if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
		++id_idx;
	if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
		++id_idx;

180
	perf_evlist__id_add(evlist, evsel, cpu, thread, read_data[id_idx]);
181
182
183
	return 0;
}

184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id)
{
	struct hlist_head *head;
	struct hlist_node *pos;
	struct perf_sample_id *sid;
	int hash;

	if (evlist->nr_entries == 1)
		return list_entry(evlist->entries.next, struct perf_evsel, node);

	hash = hash_64(id, PERF_EVLIST__HLIST_BITS);
	head = &evlist->heads[hash];

	hlist_for_each_entry(sid, pos, head, node)
		if (sid->id == id)
			return sid->evsel;
	return NULL;
}
202

203
union perf_event *perf_evlist__mmap_read(struct perf_evlist *evlist, int idx)
204
205
206
{
	/* XXX Move this to perf.c, making it generally available */
	unsigned int page_size = sysconf(_SC_PAGE_SIZE);
207
	struct perf_mmap *md = &evlist->mmap[idx];
208
209
210
	unsigned int head = perf_mmap__read_head(md);
	unsigned int old = md->prev;
	unsigned char *data = md->base + page_size;
211
	union perf_event *event = NULL;
212

213
	if (evlist->overwrite) {
214
		/*
215
216
217
218
219
220
		 * If we're further behind than half the buffer, there's a chance
		 * the writer will bite our tail and mess up the samples under us.
		 *
		 * If we somehow ended up ahead of the head, we got messed up.
		 *
		 * In either case, truncate and restart at head.
221
		 */
222
223
224
225
226
227
228
229
230
		int diff = head - old;
		if (diff > md->mask / 2 || diff < 0) {
			fprintf(stderr, "WARNING: failed to keep up with mmap data.\n");

			/*
			 * head points to a known good entry, start there.
			 */
			old = head;
		}
231
232
233
234
235
	}

	if (old != head) {
		size_t size;

236
		event = (union perf_event *)&data[old & md->mask];
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
		size = event->header.size;

		/*
		 * Event straddles the mmap boundary -- header should always
		 * be inside due to u64 alignment of output.
		 */
		if ((old & md->mask) + size != ((old + size) & md->mask)) {
			unsigned int offset = old;
			unsigned int len = min(sizeof(*event), size), cpy;
			void *dst = &evlist->event_copy;

			do {
				cpy = min(md->mask + 1 - (offset & md->mask), len);
				memcpy(dst, &data[offset & md->mask], cpy);
				offset += cpy;
				dst += cpy;
				len -= cpy;
			} while (len);

			event = &evlist->event_copy;
		}

		old += size;
	}

	md->prev = old;
263
264
265
266

	if (!evlist->overwrite)
		perf_mmap__write_tail(md, old);

267
268
	return event;
}
269

270
void perf_evlist__munmap(struct perf_evlist *evlist)
271
{
272
	int i;
273

274
275
276
277
	for (i = 0; i < evlist->nr_mmaps; i++) {
		if (evlist->mmap[i].base != NULL) {
			munmap(evlist->mmap[i].base, evlist->mmap_len);
			evlist->mmap[i].base = NULL;
278
279
		}
	}
280
281
282

	free(evlist->mmap);
	evlist->mmap = NULL;
283
284
}

285
int perf_evlist__alloc_mmap(struct perf_evlist *evlist)
286
{
287
288
289
290
	evlist->nr_mmaps = evlist->cpus->nr;
	if (evlist->cpus->map[0] == -1)
		evlist->nr_mmaps = evlist->threads->nr;
	evlist->mmap = zalloc(evlist->nr_mmaps * sizeof(struct perf_mmap));
291
292
293
	return evlist->mmap != NULL ? 0 : -ENOMEM;
}

294
static int __perf_evlist__mmap(struct perf_evlist *evlist,
295
			       int idx, int prot, int mask, int fd)
296
{
297
298
299
	evlist->mmap[idx].prev = 0;
	evlist->mmap[idx].mask = mask;
	evlist->mmap[idx].base = mmap(NULL, evlist->mmap_len, prot,
300
				      MAP_SHARED, fd, 0);
301
	if (evlist->mmap[idx].base == MAP_FAILED)
302
303
304
305
306
307
		return -1;

	perf_evlist__add_pollfd(evlist, fd);
	return 0;
}

308
309
310
311
312
313
314
315
316
317
318
319
320
321
static int perf_evlist__mmap_per_cpu(struct perf_evlist *evlist, int prot, int mask)
{
	struct perf_evsel *evsel;
	int cpu, thread;

	for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
		int output = -1;

		for (thread = 0; thread < evlist->threads->nr; thread++) {
			list_for_each_entry(evsel, &evlist->entries, node) {
				int fd = FD(evsel, cpu, thread);

				if (output == -1) {
					output = fd;
322
					if (__perf_evlist__mmap(evlist, cpu,
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
								prot, mask, output) < 0)
						goto out_unmap;
				} else {
					if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, output) != 0)
						goto out_unmap;
				}

				if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
				    perf_evlist__id_add_fd(evlist, evsel, cpu, thread, fd) < 0)
					goto out_unmap;
			}
		}
	}

	return 0;

out_unmap:
	for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
		if (evlist->mmap[cpu].base != NULL) {
			munmap(evlist->mmap[cpu].base, evlist->mmap_len);
			evlist->mmap[cpu].base = NULL;
		}
	}
	return -1;
}

static int perf_evlist__mmap_per_thread(struct perf_evlist *evlist, int prot, int mask)
{
	struct perf_evsel *evsel;
	int thread;

	for (thread = 0; thread < evlist->threads->nr; thread++) {
		int output = -1;

		list_for_each_entry(evsel, &evlist->entries, node) {
			int fd = FD(evsel, 0, thread);

			if (output == -1) {
				output = fd;
362
				if (__perf_evlist__mmap(evlist, thread,
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
							prot, mask, output) < 0)
					goto out_unmap;
			} else {
				if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, output) != 0)
					goto out_unmap;
			}

			if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
			    perf_evlist__id_add_fd(evlist, evsel, 0, thread, fd) < 0)
				goto out_unmap;
		}
	}

	return 0;

out_unmap:
	for (thread = 0; thread < evlist->threads->nr; thread++) {
		if (evlist->mmap[thread].base != NULL) {
			munmap(evlist->mmap[thread].base, evlist->mmap_len);
			evlist->mmap[thread].base = NULL;
		}
	}
	return -1;
}

388
389
390
391
392
393
394
395
396
397
398
399
/** perf_evlist__mmap - Create per cpu maps to receive events
 *
 * @evlist - list of events
 * @pages - map length in pages
 * @overwrite - overwrite older events?
 *
 * If overwrite is false the user needs to signal event consuption using:
 *
 *	struct perf_mmap *m = &evlist->mmap[cpu];
 *	unsigned int head = perf_mmap__read_head(m);
 *
 *	perf_mmap__write_tail(m, head)
400
401
 *
 * Using perf_evlist__read_on_cpu does this automatically.
402
 */
403
int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite)
404
405
{
	unsigned int page_size = sysconf(_SC_PAGE_SIZE);
406
407
	int mask = pages * page_size - 1;
	struct perf_evsel *evsel;
408
409
	const struct cpu_map *cpus = evlist->cpus;
	const struct thread_map *threads = evlist->threads;
410
	int prot = PROT_READ | (overwrite ? 0 : PROT_WRITE);
411

412
	if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0)
413
414
		return -ENOMEM;

415
	if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0)
416
417
418
419
420
421
422
		return -ENOMEM;

	evlist->overwrite = overwrite;
	evlist->mmap_len = (pages + 1) * page_size;

	list_for_each_entry(evsel, &evlist->entries, node) {
		if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
423
		    evsel->sample_id == NULL &&
424
425
426
427
		    perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0)
			return -ENOMEM;
	}

428
429
	if (evlist->cpus->map[0] == -1)
		return perf_evlist__mmap_per_thread(evlist, prot, mask);
430

431
	return perf_evlist__mmap_per_cpu(evlist, prot, mask);
432
}
433
434
435
436
437
438
439
440
441

int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid,
			     pid_t target_tid, const char *cpu_list)
{
	evlist->threads = thread_map__new(target_pid, target_tid);

	if (evlist->threads == NULL)
		return -1;

442
	if (cpu_list == NULL && target_tid != -1)
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
		evlist->cpus = cpu_map__dummy_new();
	else
		evlist->cpus = cpu_map__new(cpu_list);

	if (evlist->cpus == NULL)
		goto out_delete_threads;

	return 0;

out_delete_threads:
	thread_map__delete(evlist->threads);
	return -1;
}

void perf_evlist__delete_maps(struct perf_evlist *evlist)
{
	cpu_map__delete(evlist->cpus);
	thread_map__delete(evlist->threads);
	evlist->cpus	= NULL;
	evlist->threads = NULL;
}
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

int perf_evlist__set_filters(struct perf_evlist *evlist)
{
	const struct thread_map *threads = evlist->threads;
	const struct cpu_map *cpus = evlist->cpus;
	struct perf_evsel *evsel;
	char *filter;
	int thread;
	int cpu;
	int err;
	int fd;

	list_for_each_entry(evsel, &evlist->entries, node) {
		filter = evsel->filter;
		if (!filter)
			continue;
		for (cpu = 0; cpu < cpus->nr; cpu++) {
			for (thread = 0; thread < threads->nr; thread++) {
				fd = FD(evsel, cpu, thread);
				err = ioctl(fd, PERF_EVENT_IOC_SET_FILTER, filter);
				if (err)
					return err;
			}
		}
	}

	return 0;
}
492

493
bool perf_evlist__valid_sample_type(const struct perf_evlist *evlist)
494
{
495
496
497
498
499
500
501
	struct perf_evsel *pos, *first;

	pos = first = list_entry(evlist->entries.next, struct perf_evsel, node);

	list_for_each_entry_continue(pos, &evlist->entries, node) {
		if (first->attr.sample_type != pos->attr.sample_type)
			return false;
502
503
	}

504
	return true;
505
506
}

507
508
509
510
511
512
513
514
515
u64 perf_evlist__sample_type(const struct perf_evlist *evlist)
{
	struct perf_evsel *first;

	first = list_entry(evlist->entries.next, struct perf_evsel, node);
	return first->attr.sample_type;
}

bool perf_evlist__valid_sample_id_all(const struct perf_evlist *evlist)
516
{
517
518
519
520
521
522
523
	struct perf_evsel *pos, *first;

	pos = first = list_entry(evlist->entries.next, struct perf_evsel, node);

	list_for_each_entry_continue(pos, &evlist->entries, node) {
		if (first->attr.sample_id_all != pos->attr.sample_id_all)
			return false;
524
525
	}

526
527
528
529
530
531
532
533
534
	return true;
}

bool perf_evlist__sample_id_all(const struct perf_evlist *evlist)
{
	struct perf_evsel *first;

	first = list_entry(evlist->entries.next, struct perf_evsel, node);
	return first->attr.sample_id_all;
535
}
536
537
538
539
540
541

void perf_evlist__set_selected(struct perf_evlist *evlist,
			       struct perf_evsel *evsel)
{
	evlist->selected = evsel;
}
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

int perf_evlist__open(struct perf_evlist *evlist, bool group)
{
	struct perf_evsel *evsel, *first;
	int err, ncpus, nthreads;

	first = list_entry(evlist->entries.next, struct perf_evsel, node);

	list_for_each_entry(evsel, &evlist->entries, node) {
		struct xyarray *group_fd = NULL;

		if (group && evsel != first)
			group_fd = first->fd;

		err = perf_evsel__open(evsel, evlist->cpus, evlist->threads,
				       group, group_fd);
		if (err < 0)
			goto out_err;
	}

	return 0;
out_err:
	ncpus = evlist->cpus ? evlist->cpus->nr : 1;
	nthreads = evlist->threads ? evlist->threads->nr : 1;

	list_for_each_entry_reverse(evsel, &evlist->entries, node)
		perf_evsel__close(evsel, ncpus, nthreads);

	return err;
}