session.c 23.2 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;
104 105 106 107 108 109 110 111 112
	/*
	 * On 64bit we can mmap the data file in one go. No need for tiny mmap
	 * slices. On 32bit we use 32MB.
	 */
#if BITS_PER_LONG == 64
	self->mmap_window = ULLONG_MAX;
#else
	self->mmap_window = 32 * 1024 * 1024ULL;
#endif
113
	self->machines = RB_ROOT;
T
Tom Zanussi 已提交
114
	self->repipe = repipe;
115
	INIT_LIST_HEAD(&self->ordered_samples.samples);
116
	machine__init(&self->host_machine, "", HOST_KERNEL_ID);
117

118 119 120 121 122 123 124 125 126 127 128
	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;
	}
129

130
	perf_session__update_sample_type(self);
131 132
out:
	return self;
133
out_free:
134 135
	free(self);
	return NULL;
136 137 138
out_delete:
	perf_session__delete(self);
	return NULL;
139 140
}

141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
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);
	}
}

164 165 166
void perf_session__delete(struct perf_session *self)
{
	perf_header__exit(&self->header);
167
	perf_session__destroy_kernel_maps(self);
168 169 170
	perf_session__delete_dead_threads(self);
	perf_session__delete_threads(self);
	machine__exit(&self->host_machine);
171 172 173
	close(self->fd);
	free(self);
}
174

175 176
void perf_session__remove_thread(struct perf_session *self, struct thread *th)
{
177
	self->last_match = NULL;
178 179 180 181 182 183 184 185
	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);
}

186 187 188 189 190 191 192 193
static bool symbol__match_parent_regex(struct symbol *sym)
{
	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
		return 1;

	return 0;
}

194 195 196 197
struct map_symbol *perf_session__resolve_callchain(struct perf_session *self,
						   struct thread *thread,
						   struct ip_callchain *chain,
						   struct symbol **parent)
198 199 200
{
	u8 cpumode = PERF_RECORD_MISC_USER;
	unsigned int i;
201
	struct map_symbol *syms = calloc(chain->nr, sizeof(*syms));
202

203 204
	if (!syms)
		return NULL;
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223

	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;
		}

224
		al.filtered = false;
225
		thread__find_addr_location(thread, self, cpumode,
226
				MAP__FUNCTION, thread->pid, ip, &al, NULL);
227 228 229 230
		if (al.sym != NULL) {
			if (sort__has_parent && !*parent &&
			    symbol__match_parent_regex(al.sym))
				*parent = al.sym;
231
			if (!symbol_conf.use_callchain)
232
				break;
233 234
			syms[i].map = al.map;
			syms[i].sym = al.sym;
235 236 237 238 239
		}
	}

	return syms;
}
240 241 242 243 244 245 246 247

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

248 249 250 251 252 253 254 255 256 257 258 259
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);

260 261
static void perf_event_ops__fill_defaults(struct perf_event_ops *handler)
{
262 263 264 265 266 267 268 269 270 271 272
	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)
273
		handler->lost = event__process_lost;
274 275 276 277 278 279
	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;
280 281
	if (handler->attr == NULL)
		handler->attr = process_event_stub;
282 283
	if (handler->event_type == NULL)
		handler->event_type = process_event_stub;
284 285
	if (handler->tracing_data == NULL)
		handler->tracing_data = process_event_stub;
286 287
	if (handler->build_id == NULL)
		handler->build_id = process_event_stub;
288 289 290 291 292 293
	if (handler->finished_round == NULL) {
		if (handler->ordered_samples)
			handler->finished_round = process_finished_round;
		else
			handler->finished_round = process_finished_round_stub;
	}
294 295
}

296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
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);
}

347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
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);
}

367 368 369 370 371 372
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);
}

373 374 375 376 377
static void event__tracing_data_swap(event_t *self)
{
	self->tracing_data.size = bswap_32(self->tracing_data.size);
}

378 379 380 381 382 383 384 385 386 387
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,
388
	[PERF_RECORD_HEADER_ATTR]   = event__attr_swap,
389
	[PERF_RECORD_HEADER_EVENT_TYPE]   = event__event_type_swap,
390
	[PERF_RECORD_HEADER_TRACING_DATA]   = event__tracing_data_swap,
391
	[PERF_RECORD_HEADER_BUILD_ID]   = NULL,
392
	[PERF_RECORD_HEADER_MAX]    = NULL,
393 394
};

395 396
struct sample_queue {
	u64			timestamp;
397
	event_t			*event;
398 399 400 401 402 403
	struct list_head	list;
};

static void flush_sample_queue(struct perf_session *s,
			       struct perf_event_ops *ops)
{
404 405
	struct ordered_samples *os = &s->ordered_samples;
	struct list_head *head = &os->samples;
406
	struct sample_queue *tmp, *iter;
407 408
	u64 limit = os->next_flush;
	u64 last_ts = os->last_sample ? os->last_sample->timestamp : 0ULL;
409

410
	if (!ops->ordered_samples || !limit)
411 412 413 414
		return;

	list_for_each_entry_safe(iter, tmp, head, list) {
		if (iter->timestamp > limit)
415
			break;
416

417
		ops->sample(iter->event, s);
418

419
		os->last_flush = iter->timestamp;
420 421 422
		list_del(&iter->list);
		free(iter);
	}
423 424 425 426 427 428 429

	if (list_empty(head)) {
		os->last_sample = NULL;
	} else if (last_ts <= limit) {
		os->last_sample =
			list_entry(head->prev, struct sample_queue, list);
	}
430 431
}

432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480
/*
 * 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;
}

481 482 483 484
/* The queue is ordered by time */
static void __queue_sample_event(struct sample_queue *new,
				 struct perf_session *s)
{
485 486 487 488
	struct ordered_samples *os = &s->ordered_samples;
	struct sample_queue *sample = os->last_sample;
	u64 timestamp = new->timestamp;
	struct list_head *p;
489

490
	os->last_sample = new;
491

492 493 494
	if (!sample) {
		list_add(&new->list, &os->samples);
		os->max_timestamp = timestamp;
495 496 497 498
		return;
	}

	/*
499 500 501
	 * 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.
502
	 */
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524
	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);
	}
525 526 527
}

static int queue_sample_event(event_t *event, struct sample_data *data,
528
			      struct perf_session *s)
529 530 531 532 533 534 535 536 537 538 539 540 541 542
{
	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;
543
	new->event = event;
544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560

	__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);

561
	queue_sample_event(event, &data, s);
562 563 564 565

	return 0;
}

566 567 568
static int perf_session__process_event(struct perf_session *self,
				       event_t *event,
				       struct perf_event_ops *ops,
569
				       u64 file_offset)
570 571 572
{
	trace_event(event);

573
	if (event->header.type < PERF_RECORD_HEADER_MAX) {
574
		dump_printf("%#Lx [%#x]: PERF_RECORD_%s",
575
			    file_offset, event->header.size,
576
			    event__name[event->header.type]);
577
		hists__inc_nr_events(&self->hists, event->header.type);
578 579
	}

580 581 582
	if (self->header.needs_swap && event__swap_ops[event->header.type])
		event__swap_ops[event->header.type](event);

583 584
	switch (event->header.type) {
	case PERF_RECORD_SAMPLE:
585
		return perf_session__process_sample(event, self, ops);
586
	case PERF_RECORD_MMAP:
587
		return ops->mmap(event, self);
588
	case PERF_RECORD_COMM:
589
		return ops->comm(event, self);
590
	case PERF_RECORD_FORK:
591
		return ops->fork(event, self);
592
	case PERF_RECORD_EXIT:
593
		return ops->exit(event, self);
594
	case PERF_RECORD_LOST:
595
		return ops->lost(event, self);
596
	case PERF_RECORD_READ:
597
		return ops->read(event, self);
598
	case PERF_RECORD_THROTTLE:
599
		return ops->throttle(event, self);
600
	case PERF_RECORD_UNTHROTTLE:
601
		return ops->unthrottle(event, self);
602 603
	case PERF_RECORD_HEADER_ATTR:
		return ops->attr(event, self);
604 605
	case PERF_RECORD_HEADER_EVENT_TYPE:
		return ops->event_type(event, self);
606 607
	case PERF_RECORD_HEADER_TRACING_DATA:
		/* setup for reading amidst mmap */
608
		lseek(self->fd, file_offset, SEEK_SET);
609
		return ops->tracing_data(event, self);
610 611
	case PERF_RECORD_HEADER_BUILD_ID:
		return ops->build_id(event, self);
612 613
	case PERF_RECORD_FINISHED_ROUND:
		return ops->finished_round(event, self, ops);
614
	default:
615
		++self->hists.stats.nr_unknown_events;
616 617 618 619
		return -1;
	}
}

620 621 622 623 624 625 626
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);
}

627 628 629 630 631 632 633 634 635 636 637 638
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;
}

639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691
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);

692 693 694 695 696 697 698 699
	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;
			}
700

701 702 703
			pr_err("failed to read event data\n");
			goto out_err;
		}
704 705 706
	}

	if (size == 0 ||
707
	    (skip = perf_session__process_event(self, &event, ops, head)) < 0) {
708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735
		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;
}

736
int __perf_session__process_events(struct perf_session *session,
737 738
				   u64 data_offset, u64 data_size,
				   u64 file_size, struct perf_event_ops *ops)
739
{
740
	u64 head, page_offset, file_offset, file_pos, progress_next;
741
	int err, mmap_prot, mmap_flags, map_idx = 0;
742
	struct ui_progress *progress;
743
	size_t	page_size, mmap_size;
744
	char *buf, *mmaps[8];
745 746
	event_t *event;
	uint32_t size;
747

748 749
	perf_event_ops__fill_defaults(ops);

750
	page_size = sysconf(_SC_PAGESIZE);
751

752 753 754
	page_offset = page_size * (data_offset / page_size);
	file_offset = page_offset;
	head = data_offset - page_offset;
755

756 757 758
	if (data_offset + data_size < file_size)
		file_size = data_offset + data_size;

759 760 761 762 763 764 765 766 767
	progress_next = file_size / 16;
	progress = ui_progress__new("Processing events...", file_size);
	if (progress == NULL)
		return -1;

	mmap_size = session->mmap_window;
	if (mmap_size > file_size)
		mmap_size = file_size;

768 769
	memset(mmaps, 0, sizeof(mmaps));

770 771 772
	mmap_prot  = PROT_READ;
	mmap_flags = MAP_SHARED;

773
	if (session->header.needs_swap) {
774 775 776
		mmap_prot  |= PROT_WRITE;
		mmap_flags = MAP_PRIVATE;
	}
777
remap:
778 779
	buf = mmap(NULL, mmap_size, mmap_prot, mmap_flags, session->fd,
		   file_offset);
780 781 782 783 784
	if (buf == MAP_FAILED) {
		pr_err("failed to mmap file\n");
		err = -errno;
		goto out_err;
	}
785 786
	mmaps[map_idx] = buf;
	map_idx = (map_idx + 1) & (ARRAY_SIZE(mmaps) - 1);
787
	file_pos = file_offset + head;
788 789 790 791

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

792
	if (session->header.needs_swap)
793
		perf_event_header__bswap(&event->header);
794 795 796 797
	size = event->header.size;
	if (size == 0)
		size = 8;

798
	if (head + event->header.size >= mmap_size) {
799 800 801 802
		if (mmaps[map_idx]) {
			munmap(mmaps[map_idx], mmap_size);
			mmaps[map_idx] = NULL;
		}
803

804 805 806
		page_offset = page_size * (head / page_size);
		file_offset += page_offset;
		head -= page_offset;
807 808 809 810 811
		goto remap;
	}

	size = event->header.size;

812
	dump_printf("\n%#Lx [%#x]: event: %d\n",
813
		    file_pos, event->header.size, event->header.type);
814

815 816
	if (size == 0 ||
	    perf_session__process_event(session, event, ops, file_pos) < 0) {
817
		dump_printf("%#Lx [%#x]: skipping unknown header type: %d\n",
818
			    file_offset + head, event->header.size,
819 820 821 822 823 824 825 826 827 828 829 830
			    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;
831
	file_pos += size;
832

833 834 835 836 837
	if (file_pos >= progress_next) {
		progress_next += file_size / 16;
		ui_progress__update(progress, file_pos);
	}

838
	if (file_pos < file_size)
839
		goto more;
840

841
	err = 0;
842
	/* do the final flush for ordered samples */
843 844
	session->ordered_samples.next_flush = ULLONG_MAX;
	flush_sample_queue(session, ops);
845
out_err:
846
	ui_progress__delete(progress);
847 848

	if (ops->lost == event__process_lost &&
849
	    session->hists.stats.total_lost != 0) {
850 851
		ui__warning("Processed %Lu events and LOST %Lu!\n\n"
			    "Check IO/CPU overload!\n\n",
852 853
			    session->hists.stats.total_period,
			    session->hists.stats.total_lost);
854
	}
855 856

	if (session->hists.stats.nr_unknown_events != 0) {
857 858 859 860 861
		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",
862
			    session->hists.stats.nr_unknown_events);
863
	}
864

865 866
	return err;
}
867

868 869 870 871 872 873 874 875
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;

876 877 878 879 880 881 882
	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);
883

884 885 886
	return err;
}

887
bool perf_session__has_traces(struct perf_session *self, const char *msg)
888 889
{
	if (!(self->sample_type & PERF_SAMPLE_RAW)) {
890 891
		pr_err("No trace sample to read. Did you call 'perf %s'?\n", msg);
		return false;
892 893
	}

894
	return true;
895
}
896

897
int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps,
898 899 900 901
					     const char *symbol_name,
					     u64 addr)
{
	char *bracket;
902
	enum map_type i;
903 904 905 906 907
	struct ref_reloc_sym *ref;

	ref = zalloc(sizeof(struct ref_reloc_sym));
	if (ref == NULL)
		return -ENOMEM;
908

909 910 911
	ref->name = strdup(symbol_name);
	if (ref->name == NULL) {
		free(ref);
912
		return -ENOMEM;
913
	}
914

915
	bracket = strchr(ref->name, ']');
916 917 918
	if (bracket)
		*bracket = '\0';

919
	ref->addr = addr;
920 921

	for (i = 0; i < MAP__NR_TYPES; ++i) {
922 923
		struct kmap *kmap = map__kmap(maps[i]);
		kmap->ref_reloc_sym = ref;
924 925
	}

926 927
	return 0;
}
928 929 930 931 932 933 934

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);
}
935 936 937 938 939 940 941

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);
}