session.c 16.5 KB
Newer Older
1 2
#define _FILE_OFFSET_BITS 64

3 4
#include <linux/kernel.h>

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

#include "session.h"
10
#include "sort.h"
11 12 13 14 15 16
#include "util.h"

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

17 18 19 20 21 22 23 24 25 26
	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;
	}

27
	self->fd = open(self->filename, O_RDONLY);
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
	if (self->fd < 0) {
		pr_err("failed to open file: %s", self->filename);
		if (!strcmp(self->filename, "perf.data"))
			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;
	}

51
	if (perf_header__read(self, self->fd) < 0) {
52 53 54 55 56 57 58 59 60 61 62 63 64
		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;
}

65 66 67 68 69
void perf_session__update_sample_type(struct perf_session *self)
{
	self->sample_type = perf_header__sample_type(&self->header);
}

70
struct perf_session *perf_session__new(const char *filename, int mode, bool force)
71
{
72
	size_t len = filename ? strlen(filename) + 1 : 0;
73 74 75 76 77 78
	struct perf_session *self = zalloc(sizeof(*self) + len);

	if (self == NULL)
		goto out;

	if (perf_header__init(&self->header) < 0)
79
		goto out_free;
80 81

	memcpy(self->filename, filename, len);
82
	self->threads = RB_ROOT;
83
	self->stats_by_id = RB_ROOT;
84
	self->last_match = NULL;
85 86 87
	self->mmap_window = 32;
	self->cwd = NULL;
	self->cwdlen = 0;
88
	self->unknown_events = 0;
89
	map_groups__init(&self->kmaps);
90

91 92 93 94 95 96 97 98 99 100 101
	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;
	}
102

103
	perf_session__update_sample_type(self);
104 105
out:
	return self;
106
out_free:
107 108
	free(self);
	return NULL;
109 110 111
out_delete:
	perf_session__delete(self);
	return NULL;
112 113 114 115 116 117
}

void perf_session__delete(struct perf_session *self)
{
	perf_header__exit(&self->header);
	close(self->fd);
118
	free(self->cwd);
119 120
	free(self);
}
121 122 123 124 125 126 127 128 129

static bool symbol__match_parent_regex(struct symbol *sym)
{
	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
		return 1;

	return 0;
}

130 131 132 133
struct map_symbol *perf_session__resolve_callchain(struct perf_session *self,
						   struct thread *thread,
						   struct ip_callchain *chain,
						   struct symbol **parent)
134 135 136
{
	u8 cpumode = PERF_RECORD_MISC_USER;
	unsigned int i;
137
	struct map_symbol *syms = calloc(chain->nr, sizeof(*syms));
138

139 140
	if (!syms)
		return NULL;
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165

	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;
		}

		thread__find_addr_location(thread, self, cpumode,
					   MAP__FUNCTION, ip, &al, NULL);
		if (al.sym != NULL) {
			if (sort__has_parent && !*parent &&
			    symbol__match_parent_regex(al.sym))
				*parent = al.sym;
166
			if (!symbol_conf.use_callchain)
167
				break;
168 169
			syms[i].map = al.map;
			syms[i].sym = al.sym;
170 171 172 173 174
		}
	}

	return syms;
}
175 176 177 178 179 180 181 182 183 184

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

static void perf_event_ops__fill_defaults(struct perf_event_ops *handler)
{
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202
	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)
		handler->lost = process_event_stub;
	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;
203 204
	if (handler->attr == NULL)
		handler->attr = process_event_stub;
205 206
	if (handler->event_type == NULL)
		handler->event_type = process_event_stub;
207 208
	if (handler->tracing_data == NULL)
		handler->tracing_data = process_event_stub;
209 210 211 212 213 214 215 216 217 218 219 220 221
}

static const char *event__name[] = {
	[0]			 = "TOTAL",
	[PERF_RECORD_MMAP]	 = "MMAP",
	[PERF_RECORD_LOST]	 = "LOST",
	[PERF_RECORD_COMM]	 = "COMM",
	[PERF_RECORD_EXIT]	 = "EXIT",
	[PERF_RECORD_THROTTLE]	 = "THROTTLE",
	[PERF_RECORD_UNTHROTTLE] = "UNTHROTTLE",
	[PERF_RECORD_FORK]	 = "FORK",
	[PERF_RECORD_READ]	 = "READ",
	[PERF_RECORD_SAMPLE]	 = "SAMPLE",
222
	[PERF_RECORD_HEADER_ATTR]	 = "ATTR",
223
	[PERF_RECORD_HEADER_EVENT_TYPE]	 = "EVENT_TYPE",
224
	[PERF_RECORD_HEADER_TRACING_DATA]	 = "TRACING_DATA",
225 226
};

227
unsigned long event__total[PERF_RECORD_HEADER_MAX];
228 229 230 231

void event__print_totals(void)
{
	int i;
232 233 234
	for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) {
		if (!event__name[i])
			continue;
235 236
		pr_info("%10s events: %10ld\n",
			event__name[i], event__total[i]);
237
	}
238 239
}

240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
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);
}

291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
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);
}

311 312 313 314 315 316
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);
}

317 318 319 320 321
static void event__tracing_data_swap(event_t *self)
{
	self->tracing_data.size = bswap_32(self->tracing_data.size);
}

322 323 324 325 326 327 328 329 330 331
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,
332
	[PERF_RECORD_HEADER_ATTR]   = event__attr_swap,
333
	[PERF_RECORD_HEADER_EVENT_TYPE]   = event__event_type_swap,
334
	[PERF_RECORD_HEADER_TRACING_DATA]   = event__tracing_data_swap,
335
	[PERF_RECORD_HEADER_MAX]    = NULL,
336 337
};

338 339 340
static int perf_session__process_event(struct perf_session *self,
				       event_t *event,
				       struct perf_event_ops *ops,
341
				       u64 offset, u64 head)
342 343 344
{
	trace_event(event);

345
	if (event->header.type < PERF_RECORD_HEADER_MAX) {
346
		dump_printf("%#Lx [%#x]: PERF_RECORD_%s",
347
			    offset + head, event->header.size,
348 349 350 351 352
			    event__name[event->header.type]);
		++event__total[0];
		++event__total[event->header.type];
	}

353 354 355
	if (self->header.needs_swap && event__swap_ops[event->header.type])
		event__swap_ops[event->header.type](event);

356 357
	switch (event->header.type) {
	case PERF_RECORD_SAMPLE:
358
		return ops->sample(event, self);
359
	case PERF_RECORD_MMAP:
360
		return ops->mmap(event, self);
361
	case PERF_RECORD_COMM:
362
		return ops->comm(event, self);
363
	case PERF_RECORD_FORK:
364
		return ops->fork(event, self);
365
	case PERF_RECORD_EXIT:
366
		return ops->exit(event, self);
367
	case PERF_RECORD_LOST:
368
		return ops->lost(event, self);
369
	case PERF_RECORD_READ:
370
		return ops->read(event, self);
371
	case PERF_RECORD_THROTTLE:
372
		return ops->throttle(event, self);
373
	case PERF_RECORD_UNTHROTTLE:
374
		return ops->unthrottle(event, self);
375 376
	case PERF_RECORD_HEADER_ATTR:
		return ops->attr(event, self);
377 378
	case PERF_RECORD_HEADER_EVENT_TYPE:
		return ops->event_type(event, self);
379 380 381 382
	case PERF_RECORD_HEADER_TRACING_DATA:
		/* setup for reading amidst mmap */
		lseek(self->fd, offset + head, SEEK_SET);
		return ops->tracing_data(event, self);
383
	default:
384
		self->unknown_events++;
385 386 387 388
		return -1;
	}
}

389 390 391 392 393 394 395 396 397
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);
}

int perf_header__read_build_ids(struct perf_header *self,
				int input, u64 offset, u64 size)
398 399 400 401 402 403 404 405 406
{
	struct build_id_event bev;
	char filename[PATH_MAX];
	u64 limit = offset + size;
	int err = -1;

	while (offset < limit) {
		struct dso *dso;
		ssize_t len;
407
		struct list_head *head = &dsos__user;
408 409 410 411

		if (read(input, &bev, sizeof(bev)) != sizeof(bev))
			goto out;

412 413 414
		if (self->needs_swap)
			perf_event_header__bswap(&bev.header);

415 416 417 418
		len = bev.header.size - sizeof(bev);
		if (read(input, filename, len) != len)
			goto out;

419 420 421 422
		if (bev.header.misc & PERF_RECORD_MISC_KERNEL)
			head = &dsos__kernel;

		dso = __dsos__findnew(head, filename);
423
		if (dso != NULL) {
424
			dso__set_build_id(dso, &bev.build_id);
425 426 427
			if (head == &dsos__kernel && filename[0] == '[')
				dso->kernel = 1;
		}
428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447

		offset += bev.header.size;
	}
	err = 0;
out:
	return err;
}

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;
}

448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
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);

	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;
		}

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

	if (size == 0 ||
	    (skip = perf_session__process_event(self, &event, ops,
						0, head)) < 0) {
		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;
}

543 544 545
int __perf_session__process_events(struct perf_session *self,
				   u64 data_offset, u64 data_size,
				   u64 file_size, struct perf_event_ops *ops)
546
{
547 548 549
	int err, mmap_prot, mmap_flags;
	u64 head, shift;
	u64 offset = 0;
550 551 552 553
	size_t	page_size;
	event_t *event;
	uint32_t size;
	char *buf;
554 555 556 557
	struct ui_progress *progress = ui_progress__new("Processing events...",
							self->size);
	if (progress == NULL)
		return -1;
558 559 560

	perf_event_ops__fill_defaults(ops);

561
	page_size = sysconf(_SC_PAGESIZE);
562

563
	head = data_offset;
564 565 566 567
	shift = page_size * (head / page_size);
	offset += shift;
	head -= shift;

568 569 570 571 572 573 574
	mmap_prot  = PROT_READ;
	mmap_flags = MAP_SHARED;

	if (self->header.needs_swap) {
		mmap_prot  |= PROT_WRITE;
		mmap_flags = MAP_PRIVATE;
	}
575
remap:
576 577
	buf = mmap(NULL, page_size * self->mmap_window, mmap_prot,
		   mmap_flags, self->fd, offset);
578 579 580 581 582 583 584 585
	if (buf == MAP_FAILED) {
		pr_err("failed to mmap file\n");
		err = -errno;
		goto out_err;
	}

more:
	event = (event_t *)(buf + head);
586
	ui_progress__update(progress, offset);
587

588 589
	if (self->header.needs_swap)
		perf_event_header__bswap(&event->header);
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608
	size = event->header.size;
	if (size == 0)
		size = 8;

	if (head + event->header.size >= page_size * self->mmap_window) {
		int munmap_ret;

		shift = page_size * (head / page_size);

		munmap_ret = munmap(buf, page_size * self->mmap_window);
		assert(munmap_ret == 0);

		offset += shift;
		head -= shift;
		goto remap;
	}

	size = event->header.size;

609
	dump_printf("\n%#Lx [%#x]: event: %d\n",
610
		    offset + head, event->header.size, event->header.type);
611 612 613

	if (size == 0 ||
	    perf_session__process_event(self, event, ops, offset, head) < 0) {
614
		dump_printf("%#Lx [%#x]: skipping unknown header type: %d\n",
615
			    offset + head, event->header.size,
616 617 618 619 620 621 622 623 624 625 626 627 628
			    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;

629
	if (offset + head >= data_offset + data_size)
630 631
		goto done;

632
	if (offset + head < file_size)
633 634 635 636
		goto more;
done:
	err = 0;
out_err:
637
	ui_progress__delete(progress);
638 639
	return err;
}
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
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;

	if (!symbol_conf.full_paths) {
		char bf[PATH_MAX];

		if (getcwd(bf, sizeof(bf)) == NULL) {
			err = -errno;
out_getcwd_err:
			pr_err("failed to get the current directory\n");
			goto out_err;
		}
		self->cwd = strdup(bf);
		if (self->cwd == NULL) {
			err = -ENOMEM;
			goto out_getcwd_err;
		}
		self->cwdlen = strlen(self->cwd);
	}

666 667 668 669 670 671 672
	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);
673 674 675 676
out_err:
	return err;
}

677
bool perf_session__has_traces(struct perf_session *self, const char *msg)
678 679
{
	if (!(self->sample_type & PERF_SAMPLE_RAW)) {
680 681
		pr_err("No trace sample to read. Did you call 'perf %s'?\n", msg);
		return false;
682 683
	}

684
	return true;
685
}
686 687 688 689 690 691

int perf_session__set_kallsyms_ref_reloc_sym(struct perf_session *self,
					     const char *symbol_name,
					     u64 addr)
{
	char *bracket;
692
	enum map_type i;
693 694 695 696 697 698 699 700 701 702

	self->ref_reloc_sym.name = strdup(symbol_name);
	if (self->ref_reloc_sym.name == NULL)
		return -ENOMEM;

	bracket = strchr(self->ref_reloc_sym.name, ']');
	if (bracket)
		*bracket = '\0';

	self->ref_reloc_sym.addr = addr;
703 704 705 706 707 708

	for (i = 0; i < MAP__NR_TYPES; ++i) {
		struct kmap *kmap = map__kmap(self->vmlinux_maps[i]);
		kmap->ref_reloc_sym = &self->ref_reloc_sym;
	}

709 710
	return 0;
}