evlist.c 27.6 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 "util.h"
10
#include <lk/debugfs.h>
11
#include <poll.h>
12 13
#include "cpumap.h"
#include "thread_map.h"
14
#include "target.h"
15 16
#include "evlist.h"
#include "evsel.h"
A
Adrian Hunter 已提交
17
#include "debug.h"
18
#include <unistd.h>
19

20
#include "parse-events.h"
21
#include "parse-options.h"
22

23 24
#include <sys/mman.h>

25 26 27
#include <linux/bitops.h>
#include <linux/hash.h>

28
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
29
#define SID(e, x, y) xyarray__entry(e->sample_id, x, y)
30

31 32
void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus,
		       struct thread_map *threads)
33 34 35 36 37 38
{
	int i;

	for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i)
		INIT_HLIST_HEAD(&evlist->heads[i]);
	INIT_LIST_HEAD(&evlist->entries);
39
	perf_evlist__set_maps(evlist, cpus, threads);
40
	evlist->workload.pid = -1;
41 42
}

43
struct perf_evlist *perf_evlist__new(void)
44 45 46
{
	struct perf_evlist *evlist = zalloc(sizeof(*evlist));

47
	if (evlist != NULL)
48
		perf_evlist__init(evlist, NULL, NULL);
49 50 51 52

	return evlist;
}

53 54 55 56 57 58 59 60 61 62 63 64
struct perf_evlist *perf_evlist__new_default(void)
{
	struct perf_evlist *evlist = perf_evlist__new();

	if (evlist && perf_evlist__add_default(evlist)) {
		perf_evlist__delete(evlist);
		evlist = NULL;
	}

	return evlist;
}

65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
/**
 * perf_evlist__set_id_pos - set the positions of event ids.
 * @evlist: selected event list
 *
 * Events with compatible sample types all have the same id_pos
 * and is_pos.  For convenience, put a copy on evlist.
 */
void perf_evlist__set_id_pos(struct perf_evlist *evlist)
{
	struct perf_evsel *first = perf_evlist__first(evlist);

	evlist->id_pos = first->id_pos;
	evlist->is_pos = first->is_pos;
}

80 81 82 83 84 85 86 87 88 89
static void perf_evlist__update_id_pos(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel;

	list_for_each_entry(evsel, &evlist->entries, node)
		perf_evsel__calc_id_pos(evsel);

	perf_evlist__set_id_pos(evlist);
}

90 91 92 93 94 95 96 97 98 99 100 101
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;
}

102
void perf_evlist__exit(struct perf_evlist *evlist)
103
{
104
	free(evlist->mmap);
105
	free(evlist->pollfd);
106 107 108 109 110 111 112 113
	evlist->mmap = NULL;
	evlist->pollfd = NULL;
}

void perf_evlist__delete(struct perf_evlist *evlist)
{
	perf_evlist__purge(evlist);
	perf_evlist__exit(evlist);
114 115 116 117 118 119
	free(evlist);
}

void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry)
{
	list_add_tail(&entry->node, &evlist->entries);
120 121
	entry->idx = evlist->nr_entries;

122 123
	if (!evlist->nr_entries++)
		perf_evlist__set_id_pos(evlist);
124 125
}

126 127 128
void perf_evlist__splice_list_tail(struct perf_evlist *evlist,
				   struct list_head *list,
				   int nr_entries)
129
{
130 131
	bool set_id_pos = !evlist->nr_entries;

132 133
	list_splice_tail(list, &evlist->entries);
	evlist->nr_entries += nr_entries;
134 135
	if (set_id_pos)
		perf_evlist__set_id_pos(evlist);
136 137
}

138 139 140 141 142
void __perf_evlist__set_leader(struct list_head *list)
{
	struct perf_evsel *evsel, *leader;

	leader = list_entry(list->next, struct perf_evsel, node);
143 144 145
	evsel = list_entry(list->prev, struct perf_evsel, node);

	leader->nr_members = evsel->idx - leader->idx + 1;
146 147

	list_for_each_entry(evsel, list, node) {
148
		evsel->leader = leader;
149 150 151 152
	}
}

void perf_evlist__set_leader(struct perf_evlist *evlist)
153
{
154 155
	if (evlist->nr_entries) {
		evlist->nr_groups = evlist->nr_entries > 1 ? 1 : 0;
156
		__perf_evlist__set_leader(&evlist->entries);
157
	}
158 159
}

160 161 162 163 164 165
int perf_evlist__add_default(struct perf_evlist *evlist)
{
	struct perf_event_attr attr = {
		.type = PERF_TYPE_HARDWARE,
		.config = PERF_COUNT_HW_CPU_CYCLES,
	};
166 167 168
	struct perf_evsel *evsel;

	event_attr_init(&attr);
169

170
	evsel = perf_evsel__new(&attr);
171
	if (evsel == NULL)
172 173 174 175 176 177
		goto error;

	/* use strdup() because free(evsel) assumes name is allocated */
	evsel->name = strdup("cycles");
	if (!evsel->name)
		goto error_free;
178 179 180

	perf_evlist__add(evlist, evsel);
	return 0;
181 182 183 184
error_free:
	perf_evsel__delete(evsel);
error:
	return -ENOMEM;
185
}
186

187 188
static int perf_evlist__add_attrs(struct perf_evlist *evlist,
				  struct perf_event_attr *attrs, size_t nr_attrs)
189 190 191 192 193 194
{
	struct perf_evsel *evsel, *n;
	LIST_HEAD(head);
	size_t i;

	for (i = 0; i < nr_attrs; i++) {
195
		evsel = perf_evsel__new_idx(attrs + i, evlist->nr_entries + i);
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210
		if (evsel == NULL)
			goto out_delete_partial_list;
		list_add_tail(&evsel->node, &head);
	}

	perf_evlist__splice_list_tail(evlist, &head, nr_attrs);

	return 0;

out_delete_partial_list:
	list_for_each_entry_safe(evsel, n, &head, node)
		perf_evsel__delete(evsel);
	return -1;
}

211 212 213 214 215 216 217 218 219 220 221
int __perf_evlist__add_default_attrs(struct perf_evlist *evlist,
				     struct perf_event_attr *attrs, size_t nr_attrs)
{
	size_t i;

	for (i = 0; i < nr_attrs; i++)
		event_attr_init(attrs + i);

	return perf_evlist__add_attrs(evlist, attrs, nr_attrs);
}

222 223
struct perf_evsel *
perf_evlist__find_tracepoint_by_id(struct perf_evlist *evlist, int id)
224 225 226 227 228 229 230 231 232 233 234 235
{
	struct perf_evsel *evsel;

	list_for_each_entry(evsel, &evlist->entries, node) {
		if (evsel->attr.type   == PERF_TYPE_TRACEPOINT &&
		    (int)evsel->attr.config == id)
			return evsel;
	}

	return NULL;
}

236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
struct perf_evsel *
perf_evlist__find_tracepoint_by_name(struct perf_evlist *evlist,
				     const char *name)
{
	struct perf_evsel *evsel;

	list_for_each_entry(evsel, &evlist->entries, node) {
		if ((evsel->attr.type == PERF_TYPE_TRACEPOINT) &&
		    (strcmp(evsel->name, name) == 0))
			return evsel;
	}

	return NULL;
}

251 252 253
int perf_evlist__add_newtp(struct perf_evlist *evlist,
			   const char *sys, const char *name, void *handler)
{
254
	struct perf_evsel *evsel = perf_evsel__newtp(sys, name);
255 256 257 258

	if (evsel == NULL)
		return -1;

259
	evsel->handler = handler;
260 261 262 263
	perf_evlist__add(evlist, evsel);
	return 0;
}

264 265 266 267
void perf_evlist__disable(struct perf_evlist *evlist)
{
	int cpu, thread;
	struct perf_evsel *pos;
268 269
	int nr_cpus = cpu_map__nr(evlist->cpus);
	int nr_threads = thread_map__nr(evlist->threads);
270

271
	for (cpu = 0; cpu < nr_cpus; cpu++) {
272
		list_for_each_entry(pos, &evlist->entries, node) {
273
			if (!perf_evsel__is_group_leader(pos) || !pos->fd)
274
				continue;
275
			for (thread = 0; thread < nr_threads; thread++)
276 277
				ioctl(FD(pos, cpu, thread),
				      PERF_EVENT_IOC_DISABLE, 0);
278 279 280 281
		}
	}
}

282 283 284 285
void perf_evlist__enable(struct perf_evlist *evlist)
{
	int cpu, thread;
	struct perf_evsel *pos;
286 287
	int nr_cpus = cpu_map__nr(evlist->cpus);
	int nr_threads = thread_map__nr(evlist->threads);
288

289
	for (cpu = 0; cpu < nr_cpus; cpu++) {
290
		list_for_each_entry(pos, &evlist->entries, node) {
291
			if (!perf_evsel__is_group_leader(pos) || !pos->fd)
292
				continue;
293
			for (thread = 0; thread < nr_threads; thread++)
294 295
				ioctl(FD(pos, cpu, thread),
				      PERF_EVENT_IOC_ENABLE, 0);
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
int perf_evlist__disable_event(struct perf_evlist *evlist,
			       struct perf_evsel *evsel)
{
	int cpu, thread, err;

	if (!evsel->fd)
		return 0;

	for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
		for (thread = 0; thread < evlist->threads->nr; thread++) {
			err = ioctl(FD(evsel, cpu, thread),
				    PERF_EVENT_IOC_DISABLE, 0);
			if (err)
				return err;
		}
	}
	return 0;
}

int perf_evlist__enable_event(struct perf_evlist *evlist,
			      struct perf_evsel *evsel)
{
	int cpu, thread, err;

	if (!evsel->fd)
		return -EINVAL;

	for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
		for (thread = 0; thread < evlist->threads->nr; thread++) {
			err = ioctl(FD(evsel, cpu, thread),
				    PERF_EVENT_IOC_ENABLE, 0);
			if (err)
				return err;
		}
	}
	return 0;
}

338
static int perf_evlist__alloc_pollfd(struct perf_evlist *evlist)
339
{
340 341 342
	int nr_cpus = cpu_map__nr(evlist->cpus);
	int nr_threads = thread_map__nr(evlist->threads);
	int nfds = nr_cpus * nr_threads * evlist->nr_entries;
343 344 345
	evlist->pollfd = malloc(sizeof(struct pollfd) * nfds);
	return evlist->pollfd != NULL ? 0 : -ENOMEM;
}
346 347 348 349 350 351 352 353

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++;
}
354

355 356 357
static void perf_evlist__id_hash(struct perf_evlist *evlist,
				 struct perf_evsel *evsel,
				 int cpu, int thread, u64 id)
358 359 360 361 362 363 364 365 366 367
{
	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]);
}

368 369 370 371 372 373 374 375 376 377
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)
378 379
{
	u64 read_data[4] = { 0, };
380
	int id_idx = 1; /* The first entry is the counter value */
381 382 383 384 385 386 387 388 389 390 391
	u64 id;
	int ret;

	ret = ioctl(fd, PERF_EVENT_IOC_ID, &id);
	if (!ret)
		goto add;

	if (errno != ENOTTY)
		return -1;

	/* Legacy way to get event id.. All hail to old kernels! */
392

393 394 395 396 397 398 399
	/*
	 * This way does not work with group format read, so bail
	 * out in that case.
	 */
	if (perf_evlist__read_format(evlist) & PERF_FORMAT_GROUP)
		return -1;

400 401 402 403 404 405 406 407 408
	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;

409 410 411 412
	id = read_data[id_idx];

 add:
	perf_evlist__id_add(evlist, evsel, cpu, thread, id);
413 414 415
	return 0;
}

416
struct perf_sample_id *perf_evlist__id2sid(struct perf_evlist *evlist, u64 id)
417 418 419 420 421 422 423 424
{
	struct hlist_head *head;
	struct perf_sample_id *sid;
	int hash;

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

425
	hlist_for_each_entry(sid, head, node)
426
		if (sid->id == id)
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
			return sid;

	return NULL;
}

struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id)
{
	struct perf_sample_id *sid;

	if (evlist->nr_entries == 1)
		return perf_evlist__first(evlist);

	sid = perf_evlist__id2sid(evlist, id);
	if (sid)
		return sid->evsel;
442 443

	if (!perf_evlist__sample_id_all(evlist))
444
		return perf_evlist__first(evlist);
445

446 447
	return NULL;
}
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
static int perf_evlist__event2id(struct perf_evlist *evlist,
				 union perf_event *event, u64 *id)
{
	const u64 *array = event->sample.array;
	ssize_t n;

	n = (event->header.size - sizeof(event->header)) >> 3;

	if (event->header.type == PERF_RECORD_SAMPLE) {
		if (evlist->id_pos >= n)
			return -1;
		*id = array[evlist->id_pos];
	} else {
		if (evlist->is_pos > n)
			return -1;
		n -= evlist->is_pos;
		*id = array[n];
	}
	return 0;
}

static struct perf_evsel *perf_evlist__event2evsel(struct perf_evlist *evlist,
						   union perf_event *event)
{
473
	struct perf_evsel *first = perf_evlist__first(evlist);
474 475 476 477 478 479
	struct hlist_head *head;
	struct perf_sample_id *sid;
	int hash;
	u64 id;

	if (evlist->nr_entries == 1)
480 481 482 483 484
		return first;

	if (!first->attr.sample_id_all &&
	    event->header.type != PERF_RECORD_SAMPLE)
		return first;
485 486 487 488 489 490

	if (perf_evlist__event2id(evlist, event, &id))
		return NULL;

	/* Synthesized events have an id of zero */
	if (!id)
491
		return first;
492 493 494 495 496 497 498 499 500 501 502

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

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

503
union perf_event *perf_evlist__mmap_read(struct perf_evlist *evlist, int idx)
504
{
505
	struct perf_mmap *md = &evlist->mmap[idx];
506 507 508
	unsigned int head = perf_mmap__read_head(md);
	unsigned int old = md->prev;
	unsigned char *data = md->base + page_size;
509
	union perf_event *event = NULL;
510

511
	if (evlist->overwrite) {
512
		/*
513 514 515 516 517 518
		 * 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.
519
		 */
520 521 522 523 524 525 526 527 528
		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;
		}
529 530 531 532 533
	}

	if (old != head) {
		size_t size;

534
		event = (union perf_event *)&data[old & md->mask];
535 536 537 538 539 540 541 542 543
		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;
544
			void *dst = md->event_copy;
545 546 547 548 549 550 551 552 553

			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);

554
			event = (union perf_event *) md->event_copy;
555 556 557 558 559 560
		}

		old += size;
	}

	md->prev = old;
561

562 563
	return event;
}
564

565 566 567 568 569 570 571 572 573 574
void perf_evlist__mmap_consume(struct perf_evlist *evlist, int idx)
{
	if (!evlist->overwrite) {
		struct perf_mmap *md = &evlist->mmap[idx];
		unsigned int old = md->prev;

		perf_mmap__write_tail(md, old);
	}
}

575 576 577 578 579 580 581 582
static void __perf_evlist__munmap(struct perf_evlist *evlist, int idx)
{
	if (evlist->mmap[idx].base != NULL) {
		munmap(evlist->mmap[idx].base, evlist->mmap_len);
		evlist->mmap[idx].base = NULL;
	}
}

583
void perf_evlist__munmap(struct perf_evlist *evlist)
584
{
585
	int i;
586

587 588
	for (i = 0; i < evlist->nr_mmaps; i++)
		__perf_evlist__munmap(evlist, i);
589 590 591

	free(evlist->mmap);
	evlist->mmap = NULL;
592 593
}

594
static int perf_evlist__alloc_mmap(struct perf_evlist *evlist)
595
{
596
	evlist->nr_mmaps = cpu_map__nr(evlist->cpus);
597
	if (cpu_map__empty(evlist->cpus))
598
		evlist->nr_mmaps = thread_map__nr(evlist->threads);
599
	evlist->mmap = zalloc(evlist->nr_mmaps * sizeof(struct perf_mmap));
600 601 602
	return evlist->mmap != NULL ? 0 : -ENOMEM;
}

603
static int __perf_evlist__mmap(struct perf_evlist *evlist,
604
			       int idx, int prot, int mask, int fd)
605
{
606 607 608
	evlist->mmap[idx].prev = 0;
	evlist->mmap[idx].mask = mask;
	evlist->mmap[idx].base = mmap(NULL, evlist->mmap_len, prot,
609
				      MAP_SHARED, fd, 0);
610
	if (evlist->mmap[idx].base == MAP_FAILED) {
611 612
		pr_debug2("failed to mmap perf event ring buffer, error %d\n",
			  errno);
613
		evlist->mmap[idx].base = NULL;
614
		return -1;
615
	}
616 617 618 619 620

	perf_evlist__add_pollfd(evlist, fd);
	return 0;
}

621 622 623
static int perf_evlist__mmap_per_evsel(struct perf_evlist *evlist, int idx,
				       int prot, int mask, int cpu, int thread,
				       int *output)
624 625
{
	struct perf_evsel *evsel;
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

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

		if (*output == -1) {
			*output = fd;
			if (__perf_evlist__mmap(evlist, idx, prot, mask,
						*output) < 0)
				return -1;
		} else {
			if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, *output) != 0)
				return -1;
		}

		if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
		    perf_evlist__id_add_fd(evlist, evsel, cpu, thread, fd) < 0)
			return -1;
	}

	return 0;
}

static int perf_evlist__mmap_per_cpu(struct perf_evlist *evlist, int prot,
				     int mask)
{
651
	int cpu, thread;
652 653
	int nr_cpus = cpu_map__nr(evlist->cpus);
	int nr_threads = thread_map__nr(evlist->threads);
654

A
Adrian Hunter 已提交
655
	pr_debug2("perf event ring buffer mmapped per cpu\n");
656
	for (cpu = 0; cpu < nr_cpus; cpu++) {
657 658
		int output = -1;

659
		for (thread = 0; thread < nr_threads; thread++) {
660 661 662
			if (perf_evlist__mmap_per_evsel(evlist, cpu, prot, mask,
							cpu, thread, &output))
				goto out_unmap;
663 664 665 666 667 668
		}
	}

	return 0;

out_unmap:
669 670
	for (cpu = 0; cpu < nr_cpus; cpu++)
		__perf_evlist__munmap(evlist, cpu);
671 672 673
	return -1;
}

674 675
static int perf_evlist__mmap_per_thread(struct perf_evlist *evlist, int prot,
					int mask)
676 677
{
	int thread;
678
	int nr_threads = thread_map__nr(evlist->threads);
679

A
Adrian Hunter 已提交
680
	pr_debug2("perf event ring buffer mmapped per thread\n");
681
	for (thread = 0; thread < nr_threads; thread++) {
682 683
		int output = -1;

684 685 686
		if (perf_evlist__mmap_per_evsel(evlist, thread, prot, mask, 0,
						thread, &output))
			goto out_unmap;
687 688 689 690 691
	}

	return 0;

out_unmap:
692 693
	for (thread = 0; thread < nr_threads; thread++)
		__perf_evlist__munmap(evlist, thread);
694 695 696
	return -1;
}

697 698 699 700 701 702 703 704 705 706 707 708 709 710
static size_t perf_evlist__mmap_size(unsigned long pages)
{
	/* 512 kiB: default amount of unprivileged mlocked memory */
	if (pages == UINT_MAX)
		pages = (512 * 1024) / page_size;
	else if (!is_power_of_2(pages))
		return 0;

	return (pages + 1) * page_size;
}

int perf_evlist__parse_mmap_pages(const struct option *opt, const char *str,
				  int unset __maybe_unused)
{
711 712
	unsigned int *mmap_pages = opt->value;
	unsigned long pages, val;
713
	size_t size;
714 715 716 717 718 719 720
	static struct parse_tag tags[] = {
		{ .tag  = 'B', .mult = 1       },
		{ .tag  = 'K', .mult = 1 << 10 },
		{ .tag  = 'M', .mult = 1 << 20 },
		{ .tag  = 'G', .mult = 1 << 30 },
		{ .tag  = 0 },
	};
721

722
	val = parse_tag_value(str, tags);
723
	if (val != (unsigned long) -1) {
724 725
		/* we got file size value */
		pages = PERF_ALIGN(val, page_size) / page_size;
726
		if (pages < (1UL << 31) && !is_power_of_2(pages)) {
727
			pages = next_pow2(pages);
728
			pr_info("rounding mmap pages size to %lu (%lu pages)\n",
729 730 731 732 733 734 735 736 737 738
				pages * page_size, pages);
		}
	} else {
		/* we got pages count value */
		char *eptr;
		pages = strtoul(str, &eptr, 10);
		if (*eptr != '\0') {
			pr_err("failed to parse --mmap_pages/-m value\n");
			return -1;
		}
739 740
	}

741 742 743 744 745
	if (pages > UINT_MAX || pages > SIZE_MAX / page_size) {
		pr_err("--mmap_pages/-m value too big\n");
		return -1;
	}

746 747 748 749 750 751 752 753 754 755
	size = perf_evlist__mmap_size(pages);
	if (!size) {
		pr_err("--mmap_pages/-m value must be a power of two.");
		return -1;
	}

	*mmap_pages = pages;
	return 0;
}

756 757 758 759 760
/**
 * perf_evlist__mmap - Create mmaps to receive events.
 * @evlist: list of events
 * @pages: map length in pages
 * @overwrite: overwrite older events?
761
 *
762 763 764
 * If @overwrite is %false the user needs to signal event consumption using
 * perf_mmap__write_tail().  Using perf_evlist__mmap_read() does this
 * automatically.
765
 *
766
 * Return: %0 on success, negative error code otherwise.
767
 */
768 769
int perf_evlist__mmap(struct perf_evlist *evlist, unsigned int pages,
		      bool overwrite)
770
{
771
	struct perf_evsel *evsel;
772 773
	const struct cpu_map *cpus = evlist->cpus;
	const struct thread_map *threads = evlist->threads;
774 775
	int prot = PROT_READ | (overwrite ? 0 : PROT_WRITE), mask;

776
	if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0)
777 778
		return -ENOMEM;

779
	if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0)
780 781 782
		return -ENOMEM;

	evlist->overwrite = overwrite;
783
	evlist->mmap_len = perf_evlist__mmap_size(pages);
784
	pr_debug("mmap size %zuB\n", evlist->mmap_len);
785
	mask = evlist->mmap_len - page_size - 1;
786 787 788

	list_for_each_entry(evsel, &evlist->entries, node) {
		if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
789
		    evsel->sample_id == NULL &&
790
		    perf_evsel__alloc_id(evsel, cpu_map__nr(cpus), threads->nr) < 0)
791 792 793
			return -ENOMEM;
	}

794
	if (cpu_map__empty(cpus))
795
		return perf_evlist__mmap_per_thread(evlist, prot, mask);
796

797
	return perf_evlist__mmap_per_cpu(evlist, prot, mask);
798
}
799

800 801
int perf_evlist__create_maps(struct perf_evlist *evlist,
			     struct perf_target *target)
802
{
803 804
	evlist->threads = thread_map__new_str(target->pid, target->tid,
					      target->uid);
805 806 807 808

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

809
	if (perf_target__has_task(target))
810
		evlist->cpus = cpu_map__dummy_new();
N
Namhyung Kim 已提交
811 812
	else if (!perf_target__has_cpu(target) && !target->uses_mmap)
		evlist->cpus = cpu_map__dummy_new();
813 814
	else
		evlist->cpus = cpu_map__new(target->cpu_list);
815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832

	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;
}
833

834
int perf_evlist__apply_filters(struct perf_evlist *evlist)
835 836
{
	struct perf_evsel *evsel;
837 838
	int err = 0;
	const int ncpus = cpu_map__nr(evlist->cpus),
839
		  nthreads = thread_map__nr(evlist->threads);
840 841

	list_for_each_entry(evsel, &evlist->entries, node) {
842
		if (evsel->filter == NULL)
843
			continue;
844 845 846 847

		err = perf_evsel__set_filter(evsel, ncpus, nthreads, evsel->filter);
		if (err)
			break;
848 849
	}

850 851 852 853 854 855 856 857
	return err;
}

int perf_evlist__set_filter(struct perf_evlist *evlist, const char *filter)
{
	struct perf_evsel *evsel;
	int err = 0;
	const int ncpus = cpu_map__nr(evlist->cpus),
858
		  nthreads = thread_map__nr(evlist->threads);
859 860 861 862 863 864 865 866

	list_for_each_entry(evsel, &evlist->entries, node) {
		err = perf_evsel__set_filter(evsel, ncpus, nthreads, filter);
		if (err)
			break;
	}

	return err;
867
}
868

869
bool perf_evlist__valid_sample_type(struct perf_evlist *evlist)
870
{
871
	struct perf_evsel *pos;
872

873 874 875 876 877 878 879 880 881
	if (evlist->nr_entries == 1)
		return true;

	if (evlist->id_pos < 0 || evlist->is_pos < 0)
		return false;

	list_for_each_entry(pos, &evlist->entries, node) {
		if (pos->id_pos != evlist->id_pos ||
		    pos->is_pos != evlist->is_pos)
882
			return false;
883 884
	}

885
	return true;
886 887
}

888
u64 __perf_evlist__combined_sample_type(struct perf_evlist *evlist)
889
{
890 891 892 893 894 895 896 897 898 899 900 901 902 903 904
	struct perf_evsel *evsel;

	if (evlist->combined_sample_type)
		return evlist->combined_sample_type;

	list_for_each_entry(evsel, &evlist->entries, node)
		evlist->combined_sample_type |= evsel->attr.sample_type;

	return evlist->combined_sample_type;
}

u64 perf_evlist__combined_sample_type(struct perf_evlist *evlist)
{
	evlist->combined_sample_type = 0;
	return __perf_evlist__combined_sample_type(evlist);
905 906
}

907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932
bool perf_evlist__valid_read_format(struct perf_evlist *evlist)
{
	struct perf_evsel *first = perf_evlist__first(evlist), *pos = first;
	u64 read_format = first->attr.read_format;
	u64 sample_type = first->attr.sample_type;

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

	/* PERF_SAMPLE_READ imples PERF_FORMAT_ID. */
	if ((sample_type & PERF_SAMPLE_READ) &&
	    !(read_format & PERF_FORMAT_ID)) {
		return false;
	}

	return true;
}

u64 perf_evlist__read_format(struct perf_evlist *evlist)
{
	struct perf_evsel *first = perf_evlist__first(evlist);
	return first->attr.read_format;
}

933
u16 perf_evlist__id_hdr_size(struct perf_evlist *evlist)
934
{
935
	struct perf_evsel *first = perf_evlist__first(evlist);
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958
	struct perf_sample *data;
	u64 sample_type;
	u16 size = 0;

	if (!first->attr.sample_id_all)
		goto out;

	sample_type = first->attr.sample_type;

	if (sample_type & PERF_SAMPLE_TID)
		size += sizeof(data->tid) * 2;

       if (sample_type & PERF_SAMPLE_TIME)
		size += sizeof(data->time);

	if (sample_type & PERF_SAMPLE_ID)
		size += sizeof(data->id);

	if (sample_type & PERF_SAMPLE_STREAM_ID)
		size += sizeof(data->stream_id);

	if (sample_type & PERF_SAMPLE_CPU)
		size += sizeof(data->cpu) * 2;
959 960 961

	if (sample_type & PERF_SAMPLE_IDENTIFIER)
		size += sizeof(data->id);
962 963 964 965
out:
	return size;
}

966
bool perf_evlist__valid_sample_id_all(struct perf_evlist *evlist)
967
{
968
	struct perf_evsel *first = perf_evlist__first(evlist), *pos = first;
969 970 971 972

	list_for_each_entry_continue(pos, &evlist->entries, node) {
		if (first->attr.sample_id_all != pos->attr.sample_id_all)
			return false;
973 974
	}

975 976 977
	return true;
}

978
bool perf_evlist__sample_id_all(struct perf_evlist *evlist)
979
{
980
	struct perf_evsel *first = perf_evlist__first(evlist);
981
	return first->attr.sample_id_all;
982
}
983 984 985 986 987 988

void perf_evlist__set_selected(struct perf_evlist *evlist,
			       struct perf_evsel *evsel)
{
	evlist->selected = evsel;
}
989

990 991 992 993 994 995 996 997 998 999
void perf_evlist__close(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel;
	int ncpus = cpu_map__nr(evlist->cpus);
	int nthreads = thread_map__nr(evlist->threads);

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

1000
int perf_evlist__open(struct perf_evlist *evlist)
1001
{
1002
	struct perf_evsel *evsel;
1003
	int err;
1004

1005 1006
	perf_evlist__update_id_pos(evlist);

1007
	list_for_each_entry(evsel, &evlist->entries, node) {
1008
		err = perf_evsel__open(evsel, evlist->cpus, evlist->threads);
1009 1010 1011 1012 1013 1014
		if (err < 0)
			goto out_err;
	}

	return 0;
out_err:
1015
	perf_evlist__close(evlist);
1016
	errno = -err;
1017 1018
	return err;
}
1019 1020

int perf_evlist__prepare_workload(struct perf_evlist *evlist,
1021
				  struct perf_target *target,
1022 1023
				  const char *argv[], bool pipe_output,
				  bool want_signal)
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
{
	int child_ready_pipe[2], go_pipe[2];
	char bf;

	if (pipe(child_ready_pipe) < 0) {
		perror("failed to create 'ready' pipe");
		return -1;
	}

	if (pipe(go_pipe) < 0) {
		perror("failed to create 'go' pipe");
		goto out_close_ready_pipe;
	}

	evlist->workload.pid = fork();
	if (evlist->workload.pid < 0) {
		perror("failed to fork");
		goto out_close_pipes;
	}

	if (!evlist->workload.pid) {
1045
		if (pipe_output)
1046 1047
			dup2(2, 1);

1048 1049
		signal(SIGTERM, SIG_DFL);

1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
		close(child_ready_pipe[0]);
		close(go_pipe[1]);
		fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC);

		/*
		 * Tell the parent we're ready to go
		 */
		close(child_ready_pipe[1]);

		/*
		 * Wait until the parent tells us to go.
		 */
		if (read(go_pipe[0], &bf, 1) == -1)
			perror("unable to read pipe");

		execvp(argv[0], (char **)argv);

		perror(argv[0]);
1068 1069
		if (want_signal)
			kill(getppid(), SIGUSR1);
1070 1071 1072
		exit(-1);
	}

1073
	if (perf_target__none(target))
1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
		evlist->threads->map[0] = evlist->workload.pid;

	close(child_ready_pipe[1]);
	close(go_pipe[0]);
	/*
	 * wait for child to settle
	 */
	if (read(child_ready_pipe[0], &bf, 1) == -1) {
		perror("unable to read pipe");
		goto out_close_pipes;
	}

1086
	fcntl(go_pipe[1], F_SETFD, FD_CLOEXEC);
1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102
	evlist->workload.cork_fd = go_pipe[1];
	close(child_ready_pipe[0]);
	return 0;

out_close_pipes:
	close(go_pipe[0]);
	close(go_pipe[1]);
out_close_ready_pipe:
	close(child_ready_pipe[0]);
	close(child_ready_pipe[1]);
	return -1;
}

int perf_evlist__start_workload(struct perf_evlist *evlist)
{
	if (evlist->workload.cork_fd > 0) {
1103
		char bf = 0;
1104
		int ret;
1105 1106 1107
		/*
		 * Remove the cork, let it rip!
		 */
1108 1109 1110 1111 1112 1113
		ret = write(evlist->workload.cork_fd, &bf, 1);
		if (ret < 0)
			perror("enable to write to pipe");

		close(evlist->workload.cork_fd);
		return ret;
1114 1115 1116 1117
	}

	return 0;
}
1118

1119
int perf_evlist__parse_sample(struct perf_evlist *evlist, union perf_event *event,
1120
			      struct perf_sample *sample)
1121
{
1122 1123 1124 1125
	struct perf_evsel *evsel = perf_evlist__event2evsel(evlist, event);

	if (!evsel)
		return -EFAULT;
1126
	return perf_evsel__parse_sample(evsel, event, sample);
1127
}
1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140

size_t perf_evlist__fprintf(struct perf_evlist *evlist, FILE *fp)
{
	struct perf_evsel *evsel;
	size_t printed = 0;

	list_for_each_entry(evsel, &evlist->entries, node) {
		printed += fprintf(fp, "%s%s", evsel->idx ? ", " : "",
				   perf_evsel__name(evsel));
	}

	return printed + fprintf(fp, "\n");;
}
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167

int perf_evlist__strerror_tp(struct perf_evlist *evlist __maybe_unused,
			     int err, char *buf, size_t size)
{
	char sbuf[128];

	switch (err) {
	case ENOENT:
		scnprintf(buf, size, "%s",
			  "Error:\tUnable to find debugfs\n"
			  "Hint:\tWas your kernel was compiled with debugfs support?\n"
			  "Hint:\tIs the debugfs filesystem mounted?\n"
			  "Hint:\tTry 'sudo mount -t debugfs nodev /sys/kernel/debug'");
		break;
	case EACCES:
		scnprintf(buf, size,
			  "Error:\tNo permissions to read %s/tracing/events/raw_syscalls\n"
			  "Hint:\tTry 'sudo mount -o remount,mode=755 %s'\n",
			  debugfs_mountpoint, debugfs_mountpoint);
		break;
	default:
		scnprintf(buf, size, "%s", strerror_r(err, sbuf, sizeof(sbuf)));
		break;
	}

	return 0;
}
1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203

int perf_evlist__strerror_open(struct perf_evlist *evlist __maybe_unused,
			       int err, char *buf, size_t size)
{
	int printed, value;
	char sbuf[128], *emsg = strerror_r(err, sbuf, sizeof(sbuf));

	switch (err) {
	case EACCES:
	case EPERM:
		printed = scnprintf(buf, size,
				    "Error:\t%s.\n"
				    "Hint:\tCheck /proc/sys/kernel/perf_event_paranoid setting.", emsg);

		if (filename__read_int("/proc/sys/kernel/perf_event_paranoid", &value))
			break;

		printed += scnprintf(buf + printed, size - printed, "\nHint:\t");

		if (value >= 2) {
			printed += scnprintf(buf + printed, size - printed,
					     "For your workloads it needs to be <= 1\nHint:\t");
		}
		printed += scnprintf(buf + printed, size - printed,
				     "For system wide tracing it needs to be set to -1");

		printed += scnprintf(buf + printed, size - printed,
				    ".\nHint:\tThe current value is %d.", value);
		break;
	default:
		scnprintf(buf, size, "%s", emsg);
		break;
	}

	return 0;
}