session.c 22.9 KB
Newer Older
1 2
#define _FILE_OFFSET_BITS 64

3 4
#include <linux/kernel.h>

5
#include <byteswap.h>
6 7
#include <unistd.h>
#include <sys/types.h>
8
#include <sys/mman.h>
9 10

#include "session.h"
11
#include "sort.h"
12 13 14 15 16 17
#include "util.h"

static int perf_session__open(struct perf_session *self, bool force)
{
	struct stat input_stat;

18 19 20 21 22 23 24 25 26 27
	if (!strcmp(self->filename, "-")) {
		self->fd_pipe = true;
		self->fd = STDIN_FILENO;

		if (perf_header__read(self, self->fd) < 0)
			pr_err("incompatible file format");

		return 0;
	}

28
	self->fd = open(self->filename, O_RDONLY);
29
	if (self->fd < 0) {
30 31 32 33
		int err = errno;

		pr_err("failed to open %s: %s", self->filename, strerror(err));
		if (err == ENOENT && !strcmp(self->filename, "perf.data"))
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
			pr_err("  (try 'perf record' first)");
		pr_err("\n");
		return -errno;
	}

	if (fstat(self->fd, &input_stat) < 0)
		goto out_close;

	if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) {
		pr_err("file %s not owned by current user or root\n",
		       self->filename);
		goto out_close;
	}

	if (!input_stat.st_size) {
		pr_info("zero-sized file (%s), nothing to do!\n",
			self->filename);
		goto out_close;
	}

54
	if (perf_header__read(self, self->fd) < 0) {
55 56 57 58 59 60 61 62 63 64 65 66 67
		pr_err("incompatible file format");
		goto out_close;
	}

	self->size = input_stat.st_size;
	return 0;

out_close:
	close(self->fd);
	self->fd = -1;
	return -1;
}

68 69 70 71 72
void perf_session__update_sample_type(struct perf_session *self)
{
	self->sample_type = perf_header__sample_type(&self->header);
}

73 74
int perf_session__create_kernel_maps(struct perf_session *self)
{
75
	int ret = machine__create_kernel_maps(&self->host_machine);
76 77

	if (ret >= 0)
78
		ret = machines__create_guest_kernel_maps(&self->machines);
79 80 81
	return ret;
}

82 83 84 85 86 87
static void perf_session__destroy_kernel_maps(struct perf_session *self)
{
	machine__destroy_kernel_maps(&self->host_machine);
	machines__destroy_guest_kernel_maps(&self->machines);
}

T
Tom Zanussi 已提交
88
struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe)
89
{
90
	size_t len = filename ? strlen(filename) + 1 : 0;
91 92 93 94 95 96
	struct perf_session *self = zalloc(sizeof(*self) + len);

	if (self == NULL)
		goto out;

	if (perf_header__init(&self->header) < 0)
97
		goto out_free;
98 99

	memcpy(self->filename, filename, len);
100
	self->threads = RB_ROOT;
101
	INIT_LIST_HEAD(&self->dead_threads);
102
	self->hists_tree = RB_ROOT;
103
	self->last_match = NULL;
104
	self->mmap_window = 32;
105
	self->machines = RB_ROOT;
T
Tom Zanussi 已提交
106
	self->repipe = repipe;
107
	INIT_LIST_HEAD(&self->ordered_samples.samples);
108
	machine__init(&self->host_machine, "", HOST_KERNEL_ID);
109

110 111 112 113 114 115 116 117 118 119 120
	if (mode == O_RDONLY) {
		if (perf_session__open(self, force) < 0)
			goto out_delete;
	} else if (mode == O_WRONLY) {
		/*
		 * In O_RDONLY mode this will be performed when reading the
		 * kernel MMAP event, in event__process_mmap().
		 */
		if (perf_session__create_kernel_maps(self) < 0)
			goto out_delete;
	}
121

122
	perf_session__update_sample_type(self);
123 124
out:
	return self;
125
out_free:
126 127
	free(self);
	return NULL;
128 129 130
out_delete:
	perf_session__delete(self);
	return NULL;
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
static void perf_session__delete_dead_threads(struct perf_session *self)
{
	struct thread *n, *t;

	list_for_each_entry_safe(t, n, &self->dead_threads, node) {
		list_del(&t->node);
		thread__delete(t);
	}
}

static void perf_session__delete_threads(struct perf_session *self)
{
	struct rb_node *nd = rb_first(&self->threads);

	while (nd) {
		struct thread *t = rb_entry(nd, struct thread, rb_node);

		rb_erase(&t->rb_node, &self->threads);
		nd = rb_next(nd);
		thread__delete(t);
	}
}

156 157 158
void perf_session__delete(struct perf_session *self)
{
	perf_header__exit(&self->header);
159
	perf_session__destroy_kernel_maps(self);
160 161 162
	perf_session__delete_dead_threads(self);
	perf_session__delete_threads(self);
	machine__exit(&self->host_machine);
163 164 165
	close(self->fd);
	free(self);
}
166

167 168
void perf_session__remove_thread(struct perf_session *self, struct thread *th)
{
169
	self->last_match = NULL;
170 171 172 173 174 175 176 177
	rb_erase(&th->rb_node, &self->threads);
	/*
	 * We may have references to this thread, for instance in some hist_entry
	 * instances, so just move them to a separate list.
	 */
	list_add_tail(&th->node, &self->dead_threads);
}

178 179 180 181 182 183 184 185
static bool symbol__match_parent_regex(struct symbol *sym)
{
	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
		return 1;

	return 0;
}

186 187 188 189
struct map_symbol *perf_session__resolve_callchain(struct perf_session *self,
						   struct thread *thread,
						   struct ip_callchain *chain,
						   struct symbol **parent)
190 191 192
{
	u8 cpumode = PERF_RECORD_MISC_USER;
	unsigned int i;
193
	struct map_symbol *syms = calloc(chain->nr, sizeof(*syms));
194

195 196
	if (!syms)
		return NULL;
197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215

	for (i = 0; i < chain->nr; i++) {
		u64 ip = chain->ips[i];
		struct addr_location al;

		if (ip >= PERF_CONTEXT_MAX) {
			switch (ip) {
			case PERF_CONTEXT_HV:
				cpumode = PERF_RECORD_MISC_HYPERVISOR;	break;
			case PERF_CONTEXT_KERNEL:
				cpumode = PERF_RECORD_MISC_KERNEL;	break;
			case PERF_CONTEXT_USER:
				cpumode = PERF_RECORD_MISC_USER;	break;
			default:
				break;
			}
			continue;
		}

216
		al.filtered = false;
217
		thread__find_addr_location(thread, self, cpumode,
218
				MAP__FUNCTION, thread->pid, ip, &al, NULL);
219 220 221 222
		if (al.sym != NULL) {
			if (sort__has_parent && !*parent &&
			    symbol__match_parent_regex(al.sym))
				*parent = al.sym;
223
			if (!symbol_conf.use_callchain)
224
				break;
225 226
			syms[i].map = al.map;
			syms[i].sym = al.sym;
227 228 229 230 231
		}
	}

	return syms;
}
232 233 234 235 236 237 238 239

static int process_event_stub(event_t *event __used,
			      struct perf_session *session __used)
{
	dump_printf(": unhandled!\n");
	return 0;
}

240 241 242 243 244 245 246 247 248 249 250 251
static int process_finished_round_stub(event_t *event __used,
				       struct perf_session *session __used,
				       struct perf_event_ops *ops __used)
{
	dump_printf(": unhandled!\n");
	return 0;
}

static int process_finished_round(event_t *event,
				  struct perf_session *session,
				  struct perf_event_ops *ops);

252 253
static void perf_event_ops__fill_defaults(struct perf_event_ops *handler)
{
254 255 256 257 258 259 260 261 262 263 264
	if (handler->sample == NULL)
		handler->sample = process_event_stub;
	if (handler->mmap == NULL)
		handler->mmap = process_event_stub;
	if (handler->comm == NULL)
		handler->comm = process_event_stub;
	if (handler->fork == NULL)
		handler->fork = process_event_stub;
	if (handler->exit == NULL)
		handler->exit = process_event_stub;
	if (handler->lost == NULL)
265
		handler->lost = event__process_lost;
266 267 268 269 270 271
	if (handler->read == NULL)
		handler->read = process_event_stub;
	if (handler->throttle == NULL)
		handler->throttle = process_event_stub;
	if (handler->unthrottle == NULL)
		handler->unthrottle = process_event_stub;
272 273
	if (handler->attr == NULL)
		handler->attr = process_event_stub;
274 275
	if (handler->event_type == NULL)
		handler->event_type = process_event_stub;
276 277
	if (handler->tracing_data == NULL)
		handler->tracing_data = process_event_stub;
278 279
	if (handler->build_id == NULL)
		handler->build_id = process_event_stub;
280 281 282 283 284 285
	if (handler->finished_round == NULL) {
		if (handler->ordered_samples)
			handler->finished_round = process_finished_round;
		else
			handler->finished_round = process_finished_round_stub;
	}
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
void mem_bswap_64(void *src, int byte_size)
{
	u64 *m = src;

	while (byte_size > 0) {
		*m = bswap_64(*m);
		byte_size -= sizeof(u64);
		++m;
	}
}

static void event__all64_swap(event_t *self)
{
	struct perf_event_header *hdr = &self->header;
	mem_bswap_64(hdr + 1, self->header.size - sizeof(*hdr));
}

static void event__comm_swap(event_t *self)
{
	self->comm.pid = bswap_32(self->comm.pid);
	self->comm.tid = bswap_32(self->comm.tid);
}

static void event__mmap_swap(event_t *self)
{
	self->mmap.pid	 = bswap_32(self->mmap.pid);
	self->mmap.tid	 = bswap_32(self->mmap.tid);
	self->mmap.start = bswap_64(self->mmap.start);
	self->mmap.len	 = bswap_64(self->mmap.len);
	self->mmap.pgoff = bswap_64(self->mmap.pgoff);
}

static void event__task_swap(event_t *self)
{
	self->fork.pid	= bswap_32(self->fork.pid);
	self->fork.tid	= bswap_32(self->fork.tid);
	self->fork.ppid	= bswap_32(self->fork.ppid);
	self->fork.ptid	= bswap_32(self->fork.ptid);
	self->fork.time	= bswap_64(self->fork.time);
}

static void event__read_swap(event_t *self)
{
	self->read.pid		= bswap_32(self->read.pid);
	self->read.tid		= bswap_32(self->read.tid);
	self->read.value	= bswap_64(self->read.value);
	self->read.time_enabled	= bswap_64(self->read.time_enabled);
	self->read.time_running	= bswap_64(self->read.time_running);
	self->read.id		= bswap_64(self->read.id);
}

339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
static void event__attr_swap(event_t *self)
{
	size_t size;

	self->attr.attr.type		= bswap_32(self->attr.attr.type);
	self->attr.attr.size		= bswap_32(self->attr.attr.size);
	self->attr.attr.config		= bswap_64(self->attr.attr.config);
	self->attr.attr.sample_period	= bswap_64(self->attr.attr.sample_period);
	self->attr.attr.sample_type	= bswap_64(self->attr.attr.sample_type);
	self->attr.attr.read_format	= bswap_64(self->attr.attr.read_format);
	self->attr.attr.wakeup_events	= bswap_32(self->attr.attr.wakeup_events);
	self->attr.attr.bp_type		= bswap_32(self->attr.attr.bp_type);
	self->attr.attr.bp_addr		= bswap_64(self->attr.attr.bp_addr);
	self->attr.attr.bp_len		= bswap_64(self->attr.attr.bp_len);

	size = self->header.size;
	size -= (void *)&self->attr.id - (void *)self;
	mem_bswap_64(self->attr.id, size);
}

359 360 361 362 363 364
static void event__event_type_swap(event_t *self)
{
	self->event_type.event_type.event_id =
		bswap_64(self->event_type.event_type.event_id);
}

365 366 367 368 369
static void event__tracing_data_swap(event_t *self)
{
	self->tracing_data.size = bswap_32(self->tracing_data.size);
}

370 371 372 373 374 375 376 377 378 379
typedef void (*event__swap_op)(event_t *self);

static event__swap_op event__swap_ops[] = {
	[PERF_RECORD_MMAP]   = event__mmap_swap,
	[PERF_RECORD_COMM]   = event__comm_swap,
	[PERF_RECORD_FORK]   = event__task_swap,
	[PERF_RECORD_EXIT]   = event__task_swap,
	[PERF_RECORD_LOST]   = event__all64_swap,
	[PERF_RECORD_READ]   = event__read_swap,
	[PERF_RECORD_SAMPLE] = event__all64_swap,
380
	[PERF_RECORD_HEADER_ATTR]   = event__attr_swap,
381
	[PERF_RECORD_HEADER_EVENT_TYPE]   = event__event_type_swap,
382
	[PERF_RECORD_HEADER_TRACING_DATA]   = event__tracing_data_swap,
383
	[PERF_RECORD_HEADER_BUILD_ID]   = NULL,
384
	[PERF_RECORD_HEADER_MAX]    = NULL,
385 386
};

387 388
struct sample_queue {
	u64			timestamp;
389
	event_t			*event;
390 391 392 393 394 395
	struct list_head	list;
};

static void flush_sample_queue(struct perf_session *s,
			       struct perf_event_ops *ops)
{
396 397
	struct ordered_samples *os = &s->ordered_samples;
	struct list_head *head = &os->samples;
398
	struct sample_queue *tmp, *iter;
399 400
	u64 limit = os->next_flush;
	u64 last_ts = os->last_sample ? os->last_sample->timestamp : 0ULL;
401

402
	if (!ops->ordered_samples || !limit)
403 404 405 406
		return;

	list_for_each_entry_safe(iter, tmp, head, list) {
		if (iter->timestamp > limit)
407
			break;
408

409
		ops->sample(iter->event, s);
410

411
		os->last_flush = iter->timestamp;
412 413 414 415
		list_del(&iter->list);
		free(iter->event);
		free(iter);
	}
416 417 418 419 420 421 422

	if (list_empty(head)) {
		os->last_sample = NULL;
	} else if (last_ts <= limit) {
		os->last_sample =
			list_entry(head->prev, struct sample_queue, list);
	}
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
/*
 * When perf record finishes a pass on every buffers, it records this pseudo
 * event.
 * We record the max timestamp t found in the pass n.
 * Assuming these timestamps are monotonic across cpus, we know that if
 * a buffer still has events with timestamps below t, they will be all
 * available and then read in the pass n + 1.
 * Hence when we start to read the pass n + 2, we can safely flush every
 * events with timestamps below t.
 *
 *    ============ PASS n =================
 *       CPU 0         |   CPU 1
 *                     |
 *    cnt1 timestamps  |   cnt2 timestamps
 *          1          |         2
 *          2          |         3
 *          -          |         4  <--- max recorded
 *
 *    ============ PASS n + 1 ==============
 *       CPU 0         |   CPU 1
 *                     |
 *    cnt1 timestamps  |   cnt2 timestamps
 *          3          |         5
 *          4          |         6
 *          5          |         7 <---- max recorded
 *
 *      Flush every events below timestamp 4
 *
 *    ============ PASS n + 2 ==============
 *       CPU 0         |   CPU 1
 *                     |
 *    cnt1 timestamps  |   cnt2 timestamps
 *          6          |         8
 *          7          |         9
 *          -          |         10
 *
 *      Flush every events below timestamp 7
 *      etc...
 */
static int process_finished_round(event_t *event __used,
				  struct perf_session *session,
				  struct perf_event_ops *ops)
{
	flush_sample_queue(session, ops);
	session->ordered_samples.next_flush = session->ordered_samples.max_timestamp;

	return 0;
}

474 475 476 477
/* The queue is ordered by time */
static void __queue_sample_event(struct sample_queue *new,
				 struct perf_session *s)
{
478 479 480 481
	struct ordered_samples *os = &s->ordered_samples;
	struct sample_queue *sample = os->last_sample;
	u64 timestamp = new->timestamp;
	struct list_head *p;
482

483
	os->last_sample = new;
484

485 486 487
	if (!sample) {
		list_add(&new->list, &os->samples);
		os->max_timestamp = timestamp;
488 489 490 491
		return;
	}

	/*
492 493 494
	 * last_sample might point to some random place in the list as it's
	 * the last queued event. We expect that the new event is close to
	 * this.
495
	 */
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
	if (sample->timestamp <= timestamp) {
		while (sample->timestamp <= timestamp) {
			p = sample->list.next;
			if (p == &os->samples) {
				list_add_tail(&new->list, &os->samples);
				os->max_timestamp = timestamp;
				return;
			}
			sample = list_entry(p, struct sample_queue, list);
		}
		list_add_tail(&new->list, &sample->list);
	} else {
		while (sample->timestamp > timestamp) {
			p = sample->list.prev;
			if (p == &os->samples) {
				list_add(&new->list, &os->samples);
				return;
			}
			sample = list_entry(p, struct sample_queue, list);
		}
		list_add(&new->list, &sample->list);
	}
518 519 520
}

static int queue_sample_event(event_t *event, struct sample_data *data,
521
			      struct perf_session *s)
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
{
	u64 timestamp = data->time;
	struct sample_queue *new;


	if (timestamp < s->ordered_samples.last_flush) {
		printf("Warning: Timestamp below last timeslice flush\n");
		return -EINVAL;
	}

	new = malloc(sizeof(*new));
	if (!new)
		return -ENOMEM;

	new->timestamp = timestamp;

	new->event = malloc(event->header.size);
	if (!new->event) {
		free(new);
		return -ENOMEM;
	}

	memcpy(new->event, event, event->header.size);

	__queue_sample_event(new, s);

	return 0;
}

static int perf_session__process_sample(event_t *event, struct perf_session *s,
					struct perf_event_ops *ops)
{
	struct sample_data data;

	if (!ops->ordered_samples)
		return ops->sample(event, s);

	bzero(&data, sizeof(struct sample_data));
	event__parse_sample(event, s->sample_type, &data);

562
	queue_sample_event(event, &data, s);
563 564 565 566

	return 0;
}

567 568 569
static int perf_session__process_event(struct perf_session *self,
				       event_t *event,
				       struct perf_event_ops *ops,
570
				       u64 file_offset)
571 572 573
{
	trace_event(event);

574
	if (event->header.type < PERF_RECORD_HEADER_MAX) {
575
		dump_printf("%#Lx [%#x]: PERF_RECORD_%s",
576
			    file_offset, event->header.size,
577
			    event__name[event->header.type]);
578
		hists__inc_nr_events(&self->hists, event->header.type);
579 580
	}

581 582 583
	if (self->header.needs_swap && event__swap_ops[event->header.type])
		event__swap_ops[event->header.type](event);

584 585
	switch (event->header.type) {
	case PERF_RECORD_SAMPLE:
586
		return perf_session__process_sample(event, self, ops);
587
	case PERF_RECORD_MMAP:
588
		return ops->mmap(event, self);
589
	case PERF_RECORD_COMM:
590
		return ops->comm(event, self);
591
	case PERF_RECORD_FORK:
592
		return ops->fork(event, self);
593
	case PERF_RECORD_EXIT:
594
		return ops->exit(event, self);
595
	case PERF_RECORD_LOST:
596
		return ops->lost(event, self);
597
	case PERF_RECORD_READ:
598
		return ops->read(event, self);
599
	case PERF_RECORD_THROTTLE:
600
		return ops->throttle(event, self);
601
	case PERF_RECORD_UNTHROTTLE:
602
		return ops->unthrottle(event, self);
603 604
	case PERF_RECORD_HEADER_ATTR:
		return ops->attr(event, self);
605 606
	case PERF_RECORD_HEADER_EVENT_TYPE:
		return ops->event_type(event, self);
607 608
	case PERF_RECORD_HEADER_TRACING_DATA:
		/* setup for reading amidst mmap */
609
		lseek(self->fd, file_offset, SEEK_SET);
610
		return ops->tracing_data(event, self);
611 612
	case PERF_RECORD_HEADER_BUILD_ID:
		return ops->build_id(event, self);
613 614
	case PERF_RECORD_FINISHED_ROUND:
		return ops->finished_round(event, self, ops);
615
	default:
616
		++self->hists.stats.nr_unknown_events;
617 618 619 620
		return -1;
	}
}

621 622 623 624 625 626 627
void perf_event_header__bswap(struct perf_event_header *self)
{
	self->type = bswap_32(self->type);
	self->misc = bswap_16(self->misc);
	self->size = bswap_16(self->size);
}

628 629 630 631 632 633 634 635 636 637 638 639
static struct thread *perf_session__register_idle_thread(struct perf_session *self)
{
	struct thread *thread = perf_session__findnew(self, 0);

	if (thread == NULL || thread__set_comm(thread, "swapper")) {
		pr_err("problem inserting idle task.\n");
		thread = NULL;
	}

	return thread;
}

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
int do_read(int fd, void *buf, size_t size)
{
	void *buf_start = buf;

	while (size) {
		int ret = read(fd, buf, size);

		if (ret <= 0)
			return ret;

		size -= ret;
		buf += ret;
	}

	return buf - buf_start;
}

#define session_done()	(*(volatile int *)(&session_done))
volatile int session_done;

static int __perf_session__process_pipe_events(struct perf_session *self,
					       struct perf_event_ops *ops)
{
	event_t event;
	uint32_t size;
	int skip = 0;
	u64 head;
	int err;
	void *p;

	perf_event_ops__fill_defaults(ops);

	head = 0;
more:
	err = do_read(self->fd, &event, sizeof(struct perf_event_header));
	if (err <= 0) {
		if (err == 0)
			goto done;

		pr_err("failed to read event header\n");
		goto out_err;
	}

	if (self->header.needs_swap)
		perf_event_header__bswap(&event.header);

	size = event.header.size;
	if (size == 0)
		size = 8;

	p = &event;
	p += sizeof(struct perf_event_header);

693 694 695 696 697 698 699 700
	if (size - sizeof(struct perf_event_header)) {
		err = do_read(self->fd, p,
			      size - sizeof(struct perf_event_header));
		if (err <= 0) {
			if (err == 0) {
				pr_err("unexpected end of event stream\n");
				goto done;
			}
701

702 703 704
			pr_err("failed to read event data\n");
			goto out_err;
		}
705 706 707
	}

	if (size == 0 ||
708
	    (skip = perf_session__process_event(self, &event, ops, head)) < 0) {
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
		dump_printf("%#Lx [%#x]: skipping unknown header type: %d\n",
			    head, event.header.size, event.header.type);
		/*
		 * assume we lost track of the stream, check alignment, and
		 * increment a single u64 in the hope to catch on again 'soon'.
		 */
		if (unlikely(head & 7))
			head &= ~7ULL;

		size = 8;
	}

	head += size;

	dump_printf("\n%#Lx [%#x]: event: %d\n",
		    head, event.header.size, event.header.type);

	if (skip > 0)
		head += skip;

	if (!session_done())
		goto more;
done:
	err = 0;
out_err:
	return err;
}

737
int __perf_session__process_events(struct perf_session *session,
738 739
				   u64 data_offset, u64 data_size,
				   u64 file_size, struct perf_event_ops *ops)
740
{
741
	u64 head, page_offset, file_offset, file_pos;
742
	int err, mmap_prot, mmap_flags;
743
	struct ui_progress *progress;
744 745 746 747
	size_t	page_size;
	event_t *event;
	uint32_t size;
	char *buf;
748 749

	progress = ui_progress__new("Processing events...", session->size);
750 751
	if (progress == NULL)
		return -1;
752 753 754

	perf_event_ops__fill_defaults(ops);

755
	page_size = sysconf(_SC_PAGESIZE);
756

757 758 759
	page_offset = page_size * (data_offset / page_size);
	file_offset = page_offset;
	head = data_offset - page_offset;
760

761 762 763
	if (data_offset + data_size < file_size)
		file_size = data_offset + data_size;

764 765 766
	mmap_prot  = PROT_READ;
	mmap_flags = MAP_SHARED;

767
	if (session->header.needs_swap) {
768 769 770
		mmap_prot  |= PROT_WRITE;
		mmap_flags = MAP_PRIVATE;
	}
771
remap:
772 773
	buf = mmap(NULL, page_size * session->mmap_window, mmap_prot,
		   mmap_flags, session->fd, file_offset);
774 775 776 777 778
	if (buf == MAP_FAILED) {
		pr_err("failed to mmap file\n");
		err = -errno;
		goto out_err;
	}
779
	file_pos = file_offset + head;
780
	ui_progress__update(progress, file_offset);
781 782 783 784

more:
	event = (event_t *)(buf + head);

785
	if (session->header.needs_swap)
786
		perf_event_header__bswap(&event->header);
787 788 789 790
	size = event->header.size;
	if (size == 0)
		size = 8;

791
	if (head + event->header.size >= page_size * session->mmap_window) {
792 793
		int munmap_ret;

794
		munmap_ret = munmap(buf, page_size * session->mmap_window);
795 796
		assert(munmap_ret == 0);

797 798 799
		page_offset = page_size * (head / page_size);
		file_offset += page_offset;
		head -= page_offset;
800 801 802 803 804
		goto remap;
	}

	size = event->header.size;

805
	dump_printf("\n%#Lx [%#x]: event: %d\n",
806
		    file_pos, event->header.size, event->header.type);
807

808 809
	if (size == 0 ||
	    perf_session__process_event(session, event, ops, file_pos) < 0) {
810
		dump_printf("%#Lx [%#x]: skipping unknown header type: %d\n",
811
			    file_offset + head, event->header.size,
812 813 814 815 816 817 818 819 820 821 822 823
			    event->header.type);
		/*
		 * assume we lost track of the stream, check alignment, and
		 * increment a single u64 in the hope to catch on again 'soon'.
		 */
		if (unlikely(head & 7))
			head &= ~7ULL;

		size = 8;
	}

	head += size;
824
	file_pos += size;
825

826
	if (file_pos < file_size)
827
		goto more;
828

829
	err = 0;
830
	/* do the final flush for ordered samples */
831 832
	session->ordered_samples.next_flush = ULLONG_MAX;
	flush_sample_queue(session, ops);
833
out_err:
834
	ui_progress__delete(progress);
835 836

	if (ops->lost == event__process_lost &&
837
	    session->hists.stats.total_lost != 0) {
838 839
		ui__warning("Processed %Lu events and LOST %Lu!\n\n"
			    "Check IO/CPU overload!\n\n",
840 841
			    session->hists.stats.total_period,
			    session->hists.stats.total_lost);
842
	}
843 844

	if (session->hists.stats.nr_unknown_events != 0) {
845 846 847 848 849
		ui__warning("Found %u unknown events!\n\n"
			    "Is this an older tool processing a perf.data "
			    "file generated by a more recent tool?\n\n"
			    "If that is not the case, consider "
			    "reporting to linux-kernel@vger.kernel.org.\n\n",
850
			    session->hists.stats.nr_unknown_events);
851
	}
852

853 854
	return err;
}
855

856 857 858 859 860 861 862 863
int perf_session__process_events(struct perf_session *self,
				 struct perf_event_ops *ops)
{
	int err;

	if (perf_session__register_idle_thread(self) == NULL)
		return -ENOMEM;

864 865 866 867 868 869 870
	if (!self->fd_pipe)
		err = __perf_session__process_events(self,
						     self->header.data_offset,
						     self->header.data_size,
						     self->size, ops);
	else
		err = __perf_session__process_pipe_events(self, ops);
871

872 873 874
	return err;
}

875
bool perf_session__has_traces(struct perf_session *self, const char *msg)
876 877
{
	if (!(self->sample_type & PERF_SAMPLE_RAW)) {
878 879
		pr_err("No trace sample to read. Did you call 'perf %s'?\n", msg);
		return false;
880 881
	}

882
	return true;
883
}
884

885
int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps,
886 887 888 889
					     const char *symbol_name,
					     u64 addr)
{
	char *bracket;
890
	enum map_type i;
891 892 893 894 895
	struct ref_reloc_sym *ref;

	ref = zalloc(sizeof(struct ref_reloc_sym));
	if (ref == NULL)
		return -ENOMEM;
896

897 898 899
	ref->name = strdup(symbol_name);
	if (ref->name == NULL) {
		free(ref);
900
		return -ENOMEM;
901
	}
902

903
	bracket = strchr(ref->name, ']');
904 905 906
	if (bracket)
		*bracket = '\0';

907
	ref->addr = addr;
908 909

	for (i = 0; i < MAP__NR_TYPES; ++i) {
910 911
		struct kmap *kmap = map__kmap(maps[i]);
		kmap->ref_reloc_sym = ref;
912 913
	}

914 915
	return 0;
}
916 917 918 919 920 921 922

size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp)
{
	return __dsos__fprintf(&self->host_machine.kernel_dsos, fp) +
	       __dsos__fprintf(&self->host_machine.user_dsos, fp) +
	       machines__fprintf_dsos(&self->machines, fp);
}
923 924 925 926 927 928 929

size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, FILE *fp,
					  bool with_hits)
{
	size_t ret = machine__fprintf_dsos_buildid(&self->host_machine, fp, with_hits);
	return ret + machines__fprintf_dsos_buildid(&self->machines, fp, with_hits);
}