evsel.c 13.7 KB
Newer Older
1 2 3 4 5 6 7 8 9
/*
 * 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)
 */

10 11
#include <byteswap.h>
#include "asm/bug.h"
12
#include "evsel.h"
13
#include "evlist.h"
14
#include "util.h"
15
#include "cpumap.h"
16
#include "thread_map.h"
17

18
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
19
#define GROUP_FD(group_fd, cpu) (*(int *)xyarray__entry(group_fd, cpu, 0))
20

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
int __perf_evsel__sample_size(u64 sample_type)
{
	u64 mask = sample_type & PERF_SAMPLE_MASK;
	int size = 0;
	int i;

	for (i = 0; i < 64; i++) {
		if (mask & (1ULL << i))
			size++;
	}

	size *= sizeof(u64);

	return size;
}

37 38 39 40 41 42 43 44 45 46
static void hists__init(struct hists *hists)
{
	memset(hists, 0, sizeof(*hists));
	hists->entries_in_array[0] = hists->entries_in_array[1] = RB_ROOT;
	hists->entries_in = &hists->entries_in_array[0];
	hists->entries_collapsed = RB_ROOT;
	hists->entries = RB_ROOT;
	pthread_mutex_init(&hists->lock, NULL);
}

47 48 49 50 51 52
void perf_evsel__init(struct perf_evsel *evsel,
		      struct perf_event_attr *attr, int idx)
{
	evsel->idx	   = idx;
	evsel->attr	   = *attr;
	INIT_LIST_HEAD(&evsel->node);
53
	hists__init(&evsel->hists);
54 55
}

56
struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx)
57 58 59
{
	struct perf_evsel *evsel = zalloc(sizeof(*evsel));

60 61
	if (evsel != NULL)
		perf_evsel__init(evsel, attr, idx);
62 63 64 65

	return evsel;
}

66 67 68 69 70
void perf_evsel__config(struct perf_evsel *evsel, struct perf_record_opts *opts)
{
	struct perf_event_attr *attr = &evsel->attr;
	int track = !evsel->idx; /* only the first counter needs these */

71
	attr->sample_id_all = opts->sample_id_all_missing ? 0 : 1;
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
	attr->inherit	    = !opts->no_inherit;
	attr->read_format   = PERF_FORMAT_TOTAL_TIME_ENABLED |
			      PERF_FORMAT_TOTAL_TIME_RUNNING |
			      PERF_FORMAT_ID;

	attr->sample_type  |= PERF_SAMPLE_IP | PERF_SAMPLE_TID;

	/*
	 * We default some events to a 1 default interval. But keep
	 * it a weak assumption overridable by the user.
	 */
	if (!attr->sample_period || (opts->user_freq != UINT_MAX &&
				     opts->user_interval != ULLONG_MAX)) {
		if (opts->freq) {
			attr->sample_type	|= PERF_SAMPLE_PERIOD;
			attr->freq		= 1;
			attr->sample_freq	= opts->freq;
		} else {
			attr->sample_period = opts->default_interval;
		}
	}

	if (opts->no_samples)
		attr->sample_freq = 0;

	if (opts->inherit_stat)
		attr->inherit_stat = 1;

	if (opts->sample_address) {
		attr->sample_type	|= PERF_SAMPLE_ADDR;
		attr->mmap_data = track;
	}

	if (opts->call_graph)
		attr->sample_type	|= PERF_SAMPLE_CALLCHAIN;

	if (opts->system_wide)
		attr->sample_type	|= PERF_SAMPLE_CPU;

111 112 113
	if (opts->period)
		attr->sample_type	|= PERF_SAMPLE_PERIOD;

114
	if (!opts->sample_id_all_missing &&
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
	    (opts->sample_time || opts->system_wide ||
	     !opts->no_inherit || opts->cpu_list))
		attr->sample_type	|= PERF_SAMPLE_TIME;

	if (opts->raw_samples) {
		attr->sample_type	|= PERF_SAMPLE_TIME;
		attr->sample_type	|= PERF_SAMPLE_RAW;
		attr->sample_type	|= PERF_SAMPLE_CPU;
	}

	if (opts->no_delay) {
		attr->watermark = 0;
		attr->wakeup_events = 1;
	}

	attr->mmap = track;
	attr->comm = track;

133
	if (!opts->target_pid && !opts->target_tid && !opts->system_wide) {
134 135 136 137 138
		attr->disabled = 1;
		attr->enable_on_exec = 1;
	}
}

139 140
int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads)
{
141
	int cpu, thread;
142
	evsel->fd = xyarray__new(ncpus, nthreads, sizeof(int));
143 144 145 146 147 148 149 150 151

	if (evsel->fd) {
		for (cpu = 0; cpu < ncpus; cpu++) {
			for (thread = 0; thread < nthreads; thread++) {
				FD(evsel, cpu, thread) = -1;
			}
		}
	}

152 153 154
	return evsel->fd != NULL ? 0 : -ENOMEM;
}

155 156
int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads)
{
157 158 159 160 161 162 163 164 165 166 167 168
	evsel->sample_id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id));
	if (evsel->sample_id == NULL)
		return -ENOMEM;

	evsel->id = zalloc(ncpus * nthreads * sizeof(u64));
	if (evsel->id == NULL) {
		xyarray__delete(evsel->sample_id);
		evsel->sample_id = NULL;
		return -ENOMEM;
	}

	return 0;
169 170
}

171 172 173 174 175 176 177
int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus)
{
	evsel->counts = zalloc((sizeof(*evsel->counts) +
				(ncpus * sizeof(struct perf_counts_values))));
	return evsel->counts != NULL ? 0 : -ENOMEM;
}

178 179 180 181 182 183
void perf_evsel__free_fd(struct perf_evsel *evsel)
{
	xyarray__delete(evsel->fd);
	evsel->fd = NULL;
}

184 185
void perf_evsel__free_id(struct perf_evsel *evsel)
{
186 187 188
	xyarray__delete(evsel->sample_id);
	evsel->sample_id = NULL;
	free(evsel->id);
189 190 191
	evsel->id = NULL;
}

192 193 194 195 196 197 198 199 200 201 202
void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads)
{
	int cpu, thread;

	for (cpu = 0; cpu < ncpus; cpu++)
		for (thread = 0; thread < nthreads; ++thread) {
			close(FD(evsel, cpu, thread));
			FD(evsel, cpu, thread) = -1;
		}
}

203
void perf_evsel__exit(struct perf_evsel *evsel)
204 205 206
{
	assert(list_empty(&evsel->node));
	xyarray__delete(evsel->fd);
207 208
	xyarray__delete(evsel->sample_id);
	free(evsel->id);
209 210 211 212 213
}

void perf_evsel__delete(struct perf_evsel *evsel)
{
	perf_evsel__exit(evsel);
S
Stephane Eranian 已提交
214
	close_cgroup(evsel->cgrp);
215
	free(evsel->name);
216 217
	free(evsel);
}
218 219 220 221 222 223 224 225 226 227

int __perf_evsel__read_on_cpu(struct perf_evsel *evsel,
			      int cpu, int thread, bool scale)
{
	struct perf_counts_values count;
	size_t nv = scale ? 3 : 1;

	if (FD(evsel, cpu, thread) < 0)
		return -EINVAL;

228 229 230
	if (evsel->counts == NULL && perf_evsel__alloc_counts(evsel, cpu + 1) < 0)
		return -ENOMEM;

231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
	if (readn(FD(evsel, cpu, thread), &count, nv * sizeof(u64)) < 0)
		return -errno;

	if (scale) {
		if (count.run == 0)
			count.val = 0;
		else if (count.run < count.ena)
			count.val = (u64)((double)count.val * count.ena / count.run + 0.5);
	} else
		count.ena = count.run = 0;

	evsel->counts->cpu[cpu] = count;
	return 0;
}

int __perf_evsel__read(struct perf_evsel *evsel,
		       int ncpus, int nthreads, bool scale)
{
	size_t nv = scale ? 3 : 1;
	int cpu, thread;
	struct perf_counts_values *aggr = &evsel->counts->aggr, count;

253
	aggr->val = aggr->ena = aggr->run = 0;
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

	for (cpu = 0; cpu < ncpus; cpu++) {
		for (thread = 0; thread < nthreads; thread++) {
			if (FD(evsel, cpu, thread) < 0)
				continue;

			if (readn(FD(evsel, cpu, thread),
				  &count, nv * sizeof(u64)) < 0)
				return -errno;

			aggr->val += count.val;
			if (scale) {
				aggr->ena += count.ena;
				aggr->run += count.run;
			}
		}
	}

	evsel->counts->scaled = 0;
	if (scale) {
		if (aggr->run == 0) {
			evsel->counts->scaled = -1;
			aggr->val = 0;
			return 0;
		}

		if (aggr->run < aggr->ena) {
			evsel->counts->scaled = 1;
			aggr->val = (u64)((double)aggr->val * aggr->ena / aggr->run + 0.5);
		}
	} else
		aggr->ena = aggr->run = 0;

	return 0;
}
289

290
static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
291 292
			      struct thread_map *threads, bool group,
			      struct xyarray *group_fds)
293
{
294
	int cpu, thread;
S
Stephane Eranian 已提交
295
	unsigned long flags = 0;
296
	int pid = -1, err;
297

298 299
	if (evsel->fd == NULL &&
	    perf_evsel__alloc_fd(evsel, cpus->nr, threads->nr) < 0)
300
		return -ENOMEM;
301

S
Stephane Eranian 已提交
302 303 304 305 306
	if (evsel->cgrp) {
		flags = PERF_FLAG_PID_CGROUP;
		pid = evsel->cgrp->fd;
	}

307
	for (cpu = 0; cpu < cpus->nr; cpu++) {
308
		int group_fd = group_fds ? GROUP_FD(group_fds, cpu) : -1;
309

310
		for (thread = 0; thread < threads->nr; thread++) {
S
Stephane Eranian 已提交
311 312 313 314

			if (!evsel->cgrp)
				pid = threads->map[thread];

315
			FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr,
S
Stephane Eranian 已提交
316
								     pid,
317
								     cpus->map[cpu],
S
Stephane Eranian 已提交
318
								     group_fd, flags);
319 320
			if (FD(evsel, cpu, thread) < 0) {
				err = -errno;
321
				goto out_close;
322
			}
323 324 325

			if (group && group_fd == -1)
				group_fd = FD(evsel, cpu, thread);
326
		}
327 328 329 330 331
	}

	return 0;

out_close:
332 333 334 335 336 337 338
	do {
		while (--thread >= 0) {
			close(FD(evsel, cpu, thread));
			FD(evsel, cpu, thread) = -1;
		}
		thread = threads->nr;
	} while (--cpu >= 0);
339 340 341 342 343 344 345 346 347 348 349
	return err;
}

void perf_evsel__close(struct perf_evsel *evsel, int ncpus, int nthreads)
{
	if (evsel->fd == NULL)
		return;

	perf_evsel__close_fd(evsel, ncpus, nthreads);
	perf_evsel__free_fd(evsel);
	evsel->fd = NULL;
350 351
}

352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
static struct {
	struct cpu_map map;
	int cpus[1];
} empty_cpu_map = {
	.map.nr	= 1,
	.cpus	= { -1, },
};

static struct {
	struct thread_map map;
	int threads[1];
} empty_thread_map = {
	.map.nr	 = 1,
	.threads = { -1, },
};

368
int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
369 370
		     struct thread_map *threads, bool group,
		     struct xyarray *group_fd)
371
{
372 373 374
	if (cpus == NULL) {
		/* Work around old compiler warnings about strict aliasing */
		cpus = &empty_cpu_map.map;
375 376
	}

377 378
	if (threads == NULL)
		threads = &empty_thread_map.map;
379

380
	return __perf_evsel__open(evsel, cpus, threads, group, group_fd);
381 382
}

383
int perf_evsel__open_per_cpu(struct perf_evsel *evsel,
384 385
			     struct cpu_map *cpus, bool group,
			     struct xyarray *group_fd)
386
{
387 388
	return __perf_evsel__open(evsel, cpus, &empty_thread_map.map, group,
				  group_fd);
389
}
390

391
int perf_evsel__open_per_thread(struct perf_evsel *evsel,
392 393
				struct thread_map *threads, bool group,
				struct xyarray *group_fd)
394
{
395 396
	return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group,
				  group_fd);
397
}
398

399 400
static int perf_event__parse_id_sample(const union perf_event *event, u64 type,
				       struct perf_sample *sample)
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
{
	const u64 *array = event->sample.array;

	array += ((event->header.size -
		   sizeof(event->header)) / sizeof(u64)) - 1;

	if (type & PERF_SAMPLE_CPU) {
		u32 *p = (u32 *)array;
		sample->cpu = *p;
		array--;
	}

	if (type & PERF_SAMPLE_STREAM_ID) {
		sample->stream_id = *array;
		array--;
	}

	if (type & PERF_SAMPLE_ID) {
		sample->id = *array;
		array--;
	}

	if (type & PERF_SAMPLE_TIME) {
		sample->time = *array;
		array--;
	}

	if (type & PERF_SAMPLE_TID) {
		u32 *p = (u32 *)array;
		sample->pid = p[0];
		sample->tid = p[1];
	}

	return 0;
}

437 438 439 440 441 442 443 444 445 446 447
static bool sample_overlap(const union perf_event *event,
			   const void *offset, u64 size)
{
	const void *base = event;

	if (offset + size > base + event->header.size)
		return true;

	return false;
}

448
int perf_event__parse_sample(const union perf_event *event, u64 type,
449
			     int sample_size, bool sample_id_all,
450
			     struct perf_sample *data, bool swapped)
451 452 453
{
	const u64 *array;

454 455 456 457 458 459 460 461 462
	/*
	 * used for cross-endian analysis. See git commit 65014ab3
	 * for why this goofiness is needed.
	 */
	union {
		u64 val64;
		u32 val32[2];
	} u;

463
	memset(data, 0, sizeof(*data));
464 465
	data->cpu = data->pid = data->tid = -1;
	data->stream_id = data->id = data->time = -1ULL;
466
	data->period = 1;
467 468 469 470

	if (event->header.type != PERF_RECORD_SAMPLE) {
		if (!sample_id_all)
			return 0;
471
		return perf_event__parse_id_sample(event, type, data);
472 473 474 475
	}

	array = event->sample.array;

476 477 478
	if (sample_size + sizeof(event->header) > event->header.size)
		return -EFAULT;

479 480 481 482 483 484
	if (type & PERF_SAMPLE_IP) {
		data->ip = event->ip.ip;
		array++;
	}

	if (type & PERF_SAMPLE_TID) {
485 486 487 488 489 490 491 492 493 494
		u.val64 = *array;
		if (swapped) {
			/* undo swap of u64, then swap on individual u32s */
			u.val64 = bswap_64(u.val64);
			u.val32[0] = bswap_32(u.val32[0]);
			u.val32[1] = bswap_32(u.val32[1]);
		}

		data->pid = u.val32[0];
		data->tid = u.val32[1];
495 496 497 498 499 500 501 502
		array++;
	}

	if (type & PERF_SAMPLE_TIME) {
		data->time = *array;
		array++;
	}

503
	data->addr = 0;
504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
	if (type & PERF_SAMPLE_ADDR) {
		data->addr = *array;
		array++;
	}

	data->id = -1ULL;
	if (type & PERF_SAMPLE_ID) {
		data->id = *array;
		array++;
	}

	if (type & PERF_SAMPLE_STREAM_ID) {
		data->stream_id = *array;
		array++;
	}

	if (type & PERF_SAMPLE_CPU) {
521 522 523 524 525 526 527 528 529

		u.val64 = *array;
		if (swapped) {
			/* undo swap of u64, then swap on individual u32s */
			u.val64 = bswap_64(u.val64);
			u.val32[0] = bswap_32(u.val32[0]);
		}

		data->cpu = u.val32[0];
530 531 532 533 534 535 536 537 538
		array++;
	}

	if (type & PERF_SAMPLE_PERIOD) {
		data->period = *array;
		array++;
	}

	if (type & PERF_SAMPLE_READ) {
M
Masanari Iida 已提交
539
		fprintf(stderr, "PERF_SAMPLE_READ is unsupported for now\n");
540 541 542 543
		return -1;
	}

	if (type & PERF_SAMPLE_CALLCHAIN) {
544 545 546
		if (sample_overlap(event, array, sizeof(data->callchain->nr)))
			return -EFAULT;

547
		data->callchain = (struct ip_callchain *)array;
548 549 550 551

		if (sample_overlap(event, array, data->callchain->nr))
			return -EFAULT;

552 553 554 555
		array += 1 + data->callchain->nr;
	}

	if (type & PERF_SAMPLE_RAW) {
J
Jiri Olsa 已提交
556 557
		const u64 *pdata;

558 559 560 561 562 563 564 565
		u.val64 = *array;
		if (WARN_ONCE(swapped,
			      "Endianness of raw data not corrected!\n")) {
			/* undo swap of u64, then swap on individual u32s */
			u.val64 = bswap_64(u.val64);
			u.val32[0] = bswap_32(u.val32[0]);
			u.val32[1] = bswap_32(u.val32[1]);
		}
566 567 568 569

		if (sample_overlap(event, array, sizeof(u32)))
			return -EFAULT;

570
		data->raw_size = u.val32[0];
J
Jiri Olsa 已提交
571
		pdata = (void *) array + sizeof(u32);
572

J
Jiri Olsa 已提交
573
		if (sample_overlap(event, pdata, data->raw_size))
574 575
			return -EFAULT;

J
Jiri Olsa 已提交
576
		data->raw_data = (void *) pdata;
577 578 579 580
	}

	return 0;
}
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

int perf_event__synthesize_sample(union perf_event *event, u64 type,
				  const struct perf_sample *sample,
				  bool swapped)
{
	u64 *array;

	/*
	 * used for cross-endian analysis. See git commit 65014ab3
	 * for why this goofiness is needed.
	 */
	union {
		u64 val64;
		u32 val32[2];
	} u;

	array = event->sample.array;

	if (type & PERF_SAMPLE_IP) {
		event->ip.ip = sample->ip;
		array++;
	}

	if (type & PERF_SAMPLE_TID) {
		u.val32[0] = sample->pid;
		u.val32[1] = sample->tid;
		if (swapped) {
			/*
			 * Inverse of what is done in perf_event__parse_sample
			 */
			u.val32[0] = bswap_32(u.val32[0]);
			u.val32[1] = bswap_32(u.val32[1]);
			u.val64 = bswap_64(u.val64);
		}

		*array = u.val64;
		array++;
	}

	if (type & PERF_SAMPLE_TIME) {
		*array = sample->time;
		array++;
	}

	if (type & PERF_SAMPLE_ADDR) {
		*array = sample->addr;
		array++;
	}

	if (type & PERF_SAMPLE_ID) {
		*array = sample->id;
		array++;
	}

	if (type & PERF_SAMPLE_STREAM_ID) {
		*array = sample->stream_id;
		array++;
	}

	if (type & PERF_SAMPLE_CPU) {
		u.val32[0] = sample->cpu;
		if (swapped) {
			/*
			 * Inverse of what is done in perf_event__parse_sample
			 */
			u.val32[0] = bswap_32(u.val32[0]);
			u.val64 = bswap_64(u.val64);
		}
		*array = u.val64;
		array++;
	}

	if (type & PERF_SAMPLE_PERIOD) {
		*array = sample->period;
		array++;
	}

	return 0;
}