builtin-report.c 27.1 KB
Newer Older
1 2 3 4 5 6 7
/*
 * builtin-report.c
 *
 * Builtin report command: Analyze the perf.data input file,
 * look up and read DSOs and symbol information and display
 * a histogram of results, along various sorting keys.
 */
8
#include "builtin.h"
9

10
#include "util/util.h"
11
#include "util/cache.h"
12

13
#include "util/annotate.h"
14
#include "util/color.h"
15
#include <linux/list.h>
16
#include <linux/rbtree.h>
17
#include "util/symbol.h"
18
#include "util/callchain.h"
19
#include "util/strlist.h"
20
#include "util/values.h"
21

22
#include "perf.h"
23
#include "util/debug.h"
24 25
#include "util/evlist.h"
#include "util/evsel.h"
26
#include "util/header.h"
27
#include "util/session.h"
28
#include "util/tool.h"
29 30 31 32

#include "util/parse-options.h"
#include "util/parse-events.h"

33
#include "util/thread.h"
34
#include "util/sort.h"
35
#include "util/hist.h"
36
#include "arch/common.h"
37

38 39
#include <linux/bitmap.h>

40
struct perf_report {
41
	struct perf_tool	tool;
42
	struct perf_session	*session;
43
	bool			force, use_tui, use_gtk, use_stdio;
44 45 46 47 48
	bool			hide_unresolved;
	bool			dont_use_callchains;
	bool			show_full_info;
	bool			show_threads;
	bool			inverted_callchain;
49
	bool			mem_mode;
50 51 52 53
	struct perf_read_values	show_threads_values;
	const char		*pretty_printing_style;
	symbol_filter_t		annotate_init;
	const char		*cpu_list;
54
	const char		*symbol_filter_str;
55
	DECLARE_BITMAP(cpu_bitmap, MAX_NR_CPUS);
56
};
57

58 59 60 61 62 63 64 65 66 67
static int perf_report_config(const char *var, const char *value, void *cb)
{
	if (!strcmp(var, "report.group")) {
		symbol_conf.event_group = perf_config_bool(var, value);
		return 0;
	}

	return perf_default_config(var, value, cb);
}

68 69 70 71 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113
static int perf_report__add_mem_hist_entry(struct perf_tool *tool,
					   struct addr_location *al,
					   struct perf_sample *sample,
					   struct perf_evsel *evsel,
					   struct machine *machine,
					   union perf_event *event)
{
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
	struct symbol *parent = NULL;
	u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
	int err = 0;
	struct hist_entry *he;
	struct mem_info *mi, *mx;
	uint64_t cost;

	if ((sort__has_parent || symbol_conf.use_callchain) &&
	    sample->callchain) {
		err = machine__resolve_callchain(machine, evsel, al->thread,
						 sample, &parent);
		if (err)
			return err;
	}

	mi = machine__resolve_mem(machine, al->thread, sample, cpumode);
	if (!mi)
		return -ENOMEM;

	if (rep->hide_unresolved && !al->sym)
		return 0;

	cost = sample->weight;
	if (!cost)
		cost = 1;

	/*
	 * must pass period=weight in order to get the correct
	 * sorting from hists__collapse_resort() which is solely
	 * based on periods. We want sorting be done on nr_events * weight
	 * and this is indirectly achieved by passing period=weight here
	 * and the he_stat__add_period() function.
	 */
	he = __hists__add_mem_entry(&evsel->hists, al, parent, mi, cost, cost);
	if (!he)
		return -ENOMEM;

	/*
114
	 * In the TUI browser, we are doing integrated annotation,
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
	 * so we don't allocate the extra space needed because the stdio
	 * code will not use it.
	 */
	if (sort__has_sym && he->ms.sym && use_browser > 0) {
		struct annotation *notes = symbol__annotation(he->ms.sym);

		assert(evsel != NULL);

		if (notes->src == NULL && symbol__alloc_hist(he->ms.sym) < 0)
			goto out;

		err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
		if (err)
			goto out;
	}

	if (sort__has_sym && he->mem_info->daddr.sym && use_browser > 0) {
		struct annotation *notes;

		mx = he->mem_info;

		notes = symbol__annotation(mx->daddr.sym);
		if (notes->src == NULL && symbol__alloc_hist(mx->daddr.sym) < 0)
			goto out;

		err = symbol__inc_addr_samples(mx->daddr.sym,
					       mx->daddr.map,
					       evsel->idx,
					       mx->daddr.al_addr);
		if (err)
			goto out;
	}

	evsel->hists.stats.total_period += cost;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
	err = 0;

	if (symbol_conf.use_callchain) {
		err = callchain_append(he->callchain,
				       &callchain_cursor,
				       sample->period);
	}
out:
	return err;
}

161 162 163 164 165 166 167 168 169 170 171
static int perf_report__add_branch_hist_entry(struct perf_tool *tool,
					struct addr_location *al,
					struct perf_sample *sample,
					struct perf_evsel *evsel,
				      struct machine *machine)
{
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
	struct symbol *parent = NULL;
	int err = 0;
	unsigned i;
	struct hist_entry *he;
172
	struct branch_info *bi, *bx;
173 174 175

	if ((sort__has_parent || symbol_conf.use_callchain)
	    && sample->callchain) {
176 177
		err = machine__resolve_callchain(machine, evsel, al->thread,
						 sample, &parent);
178 179 180 181 182 183 184 185 186 187 188 189
		if (err)
			return err;
	}

	bi = machine__resolve_bstack(machine, al->thread,
				     sample->branch_stack);
	if (!bi)
		return -ENOMEM;

	for (i = 0; i < sample->branch_stack->nr; i++) {
		if (rep->hide_unresolved && !(bi[i].from.sym && bi[i].to.sym))
			continue;
190 191 192

		err = -ENOMEM;

193 194 195 196 197
		/*
		 * The report shows the percentage of total branches captured
		 * and not events sampled. Thus we use a pseudo period of 1.
		 */
		he = __hists__add_branch_entry(&evsel->hists, al, parent,
198
				&bi[i], 1, 1);
199
		if (he) {
200 201
			struct annotation *notes;
			bx = he->branch_info;
202
			if (bx->from.sym && use_browser == 1 && sort__has_sym) {
203 204 205 206 207 208 209 210 211 212 213 214 215
				notes = symbol__annotation(bx->from.sym);
				if (!notes->src
				    && symbol__alloc_hist(bx->from.sym) < 0)
					goto out;

				err = symbol__inc_addr_samples(bx->from.sym,
							       bx->from.map,
							       evsel->idx,
							       bx->from.al_addr);
				if (err)
					goto out;
			}

216
			if (bx->to.sym && use_browser == 1 && sort__has_sym) {
217 218 219 220 221 222 223 224 225 226 227 228
				notes = symbol__annotation(bx->to.sym);
				if (!notes->src
				    && symbol__alloc_hist(bx->to.sym) < 0)
					goto out;

				err = symbol__inc_addr_samples(bx->to.sym,
							       bx->to.map,
							       evsel->idx,
							       bx->to.al_addr);
				if (err)
					goto out;
			}
229 230 231
			evsel->hists.stats.total_period += 1;
			hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
		} else
232
			goto out;
233
	}
234
	err = 0;
235
out:
236
	free(bi);
237 238 239
	return err;
}

240 241 242 243
static int perf_evsel__add_hist_entry(struct perf_evsel *evsel,
				      struct addr_location *al,
				      struct perf_sample *sample,
				      struct machine *machine)
244
{
245
	struct symbol *parent = NULL;
246
	int err = 0;
247 248
	struct hist_entry *he;

249
	if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) {
250 251
		err = machine__resolve_callchain(machine, evsel, al->thread,
						 sample, &parent);
252 253
		if (err)
			return err;
254
	}
255

256 257
	he = __hists__add_entry(&evsel->hists, al, parent, sample->period,
					sample->weight);
258
	if (he == NULL)
259 260
		return -ENOMEM;

261
	if (symbol_conf.use_callchain) {
262
		err = callchain_append(he->callchain,
263
				       &callchain_cursor,
264
				       sample->period);
265
		if (err)
266
			return err;
267 268
	}
	/*
269
	 * Only in the TUI browser we are doing integrated annotation,
270 271 272
	 * so we don't allocated the extra space needed because the stdio
	 * code will not use it.
	 */
273
	if (he->ms.sym != NULL && use_browser == 1 && sort__has_sym) {
274
		struct annotation *notes = symbol__annotation(he->ms.sym);
275 276 277 278

		assert(evsel != NULL);

		err = -ENOMEM;
279
		if (notes->src == NULL && symbol__alloc_hist(he->ms.sym) < 0)
280 281 282
			goto out;

		err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
283
	}
284

285 286 287
	evsel->hists.stats.total_period += sample->period;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
out:
288
	return err;
289 290
}

291

292
static int process_sample_event(struct perf_tool *tool,
293
				union perf_event *event,
294
				struct perf_sample *sample,
295
				struct perf_evsel *evsel,
296
				struct machine *machine)
297
{
298
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
299
	struct addr_location al;
300

301
	if (perf_event__preprocess_sample(event, machine, &al, sample,
302
					  rep->annotate_init) < 0) {
303
		fprintf(stderr, "problem processing %d event, skipping it.\n",
304 305 306
			event->header.type);
		return -1;
	}
307

308
	if (al.filtered || (rep->hide_unresolved && al.sym == NULL))
309
		return 0;
310

311
	if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap))
312 313
		return 0;

314
	if (sort__mode == SORT_MODE__BRANCH) {
315 316 317 318 319
		if (perf_report__add_branch_hist_entry(tool, &al, sample,
						       evsel, machine)) {
			pr_debug("problem adding lbr entry, skipping event\n");
			return -1;
		}
320 321 322 323 324 325
	} else if (rep->mem_mode == 1) {
		if (perf_report__add_mem_hist_entry(tool, &al, sample,
						    evsel, machine, event)) {
			pr_debug("problem adding mem entry, skipping event\n");
			return -1;
		}
326 327 328
	} else {
		if (al.map != NULL)
			al.map->dso->hit = 1;
329

330 331 332 333
		if (perf_evsel__add_hist_entry(evsel, &al, sample, machine)) {
			pr_debug("problem incrementing symbol period, skipping event\n");
			return -1;
		}
334
	}
335 336
	return 0;
}
I
Ingo Molnar 已提交
337

338
static int process_read_event(struct perf_tool *tool,
339
			      union perf_event *event,
340
			      struct perf_sample *sample __maybe_unused,
341
			      struct perf_evsel *evsel,
342
			      struct machine *machine __maybe_unused)
343
{
344
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
345

346
	if (rep->show_threads) {
347
		const char *name = evsel ? perf_evsel__name(evsel) : "unknown";
348
		perf_read_values_add_value(&rep->show_threads_values,
349 350 351 352 353 354
					   event->read.pid, event->read.tid,
					   event->read.id,
					   name,
					   event->read.value);
	}

355
	dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid,
356
		    evsel ? perf_evsel__name(evsel) : "FAIL",
357
		    event->read.value);
358 359 360 361

	return 0;
}

362
/* For pipe mode, sample_type is not currently set */
363
static int perf_report__setup_sample_type(struct perf_report *rep)
364
{
365
	struct perf_session *self = rep->session;
366
	u64 sample_type = perf_evlist__sample_type(self->evlist);
367

368
	if (!self->fd_pipe && !(sample_type & PERF_SAMPLE_CALLCHAIN)) {
369
		if (sort__has_parent) {
370
			ui__error("Selected --sort parent, but no "
371 372
				    "callchain data. Did you call "
				    "'perf record' without -g?\n");
373
			return -EINVAL;
374
		}
375
		if (symbol_conf.use_callchain) {
376
			ui__error("Selected -g but no callchain data. Did "
377
				    "you call 'perf record' without -g?\n");
378
			return -1;
379
		}
380 381
	} else if (!rep->dont_use_callchains &&
		   callchain_param.mode != CHAIN_NONE &&
382
		   !symbol_conf.use_callchain) {
383
			symbol_conf.use_callchain = true;
384
			if (callchain_register_param(&callchain_param) < 0) {
385
				ui__error("Can't register callchain params.\n");
386
				return -EINVAL;
387
			}
388 389
	}

390
	if (sort__mode == SORT_MODE__BRANCH) {
391
		if (!self->fd_pipe &&
392
		    !(sample_type & PERF_SAMPLE_BRANCH_STACK)) {
393 394
			ui__error("Selected -b but no branch data. "
				  "Did you call perf record without -b?\n");
395 396 397 398
			return -1;
		}
	}

399 400
	return 0;
}
401

402 403
extern volatile int session_done;

404
static void sig_handler(int sig __maybe_unused)
405 406 407 408
{
	session_done = 1;
}

409 410
static size_t hists__fprintf_nr_sample_events(struct perf_report *rep,
					      struct hists *self,
411 412 413 414
					      const char *evname, FILE *fp)
{
	size_t ret;
	char unit;
415 416
	unsigned long nr_samples = self->stats.nr_events[PERF_RECORD_SAMPLE];
	u64 nr_events = self->stats.total_period;
417 418 419 420
	struct perf_evsel *evsel = hists_to_evsel(self);
	char buf[512];
	size_t size = sizeof(buf);

421
	if (perf_evsel__is_group_event(evsel)) {
422 423 424 425 426 427 428 429 430 431
		struct perf_evsel *pos;

		perf_evsel__group_desc(evsel, buf, size);
		evname = buf;

		for_each_group_member(pos, evsel) {
			nr_samples += pos->hists.stats.nr_events[PERF_RECORD_SAMPLE];
			nr_events += pos->hists.stats.total_period;
		}
	}
432

433 434
	nr_samples = convert_unit(nr_samples, &unit);
	ret = fprintf(fp, "# Samples: %lu%c", nr_samples, unit);
435
	if (evname != NULL)
436 437
		ret += fprintf(fp, " of event '%s'", evname);

438 439 440 441 442
	if (rep->mem_mode) {
		ret += fprintf(fp, "\n# Total weight : %" PRIu64, nr_events);
		ret += fprintf(fp, "\n# Sort order   : %s", sort_order);
	} else
		ret += fprintf(fp, "\n# Event count (approx.): %" PRIu64, nr_events);
443 444 445
	return ret + fprintf(fp, "\n#\n");
}

446
static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist,
447
					 struct perf_report *rep,
448
					 const char *help)
449
{
450
	struct perf_evsel *pos;
451

452 453
	list_for_each_entry(pos, &evlist->entries, node) {
		struct hists *hists = &pos->hists;
454
		const char *evname = perf_evsel__name(pos);
455

456 457 458 459
		if (symbol_conf.event_group &&
		    !perf_evsel__is_group_leader(pos))
			continue;

460
		hists__fprintf_nr_sample_events(rep, hists, evname, stdout);
461
		hists__fprintf(hists, true, 0, 0, stdout);
462 463 464 465 466 467 468
		fprintf(stdout, "\n\n");
	}

	if (sort_order == default_sort_order &&
	    parent_pattern == default_parent_pattern) {
		fprintf(stdout, "#\n# (%s)\n#\n", help);

469 470 471
		if (rep->show_threads) {
			bool style = !strcmp(rep->pretty_printing_style, "raw");
			perf_read_values_display(stdout, &rep->show_threads_values,
472
						 style);
473
			perf_read_values_destroy(&rep->show_threads_values);
474 475 476 477 478 479
		}
	}

	return 0;
}

480
static int __cmd_report(struct perf_report *rep)
481
{
482
	int ret = -EINVAL;
483
	u64 nr_samples;
484
	struct perf_session *session = rep->session;
485
	struct perf_evsel *pos;
486 487
	struct map *kernel_map;
	struct kmap *kernel_kmap;
488
	const char *help = "For a higher level overview, try: perf report --sort comm,dso";
489

490 491
	signal(SIGINT, sig_handler);

492 493 494
	if (rep->cpu_list) {
		ret = perf_session__cpu_bitmap(session, rep->cpu_list,
					       rep->cpu_bitmap);
495 496 497 498
		if (ret)
			goto out_delete;
	}

499
	if (use_browser <= 0)
500
		perf_session__fprintf_info(session, stdout, rep->show_full_info);
501

502 503
	if (rep->show_threads)
		perf_read_values_init(&rep->show_threads_values);
504

505
	ret = perf_report__setup_sample_type(rep);
506 507 508
	if (ret)
		goto out_delete;

509
	ret = perf_session__process_events(session, &rep->tool);
510
	if (ret)
511
		goto out_delete;
512

513
	kernel_map = session->machines.host.vmlinux_maps[MAP__FUNCTION];
514 515 516 517 518
	kernel_kmap = map__kmap(kernel_map);
	if (kernel_map == NULL ||
	    (kernel_map->dso->hit &&
	     (kernel_kmap->ref_reloc_sym == NULL ||
	      kernel_kmap->ref_reloc_sym->addr == 0))) {
519 520 521 522 523 524 525 526 527 528 529
		const char *desc =
		    "As no suitable kallsyms nor vmlinux was found, kernel samples\n"
		    "can't be resolved.";

		if (kernel_map) {
			const struct dso *kdso = kernel_map->dso;
			if (!RB_EMPTY_ROOT(&kdso->symbols[MAP__FUNCTION])) {
				desc = "If some relocation was applied (e.g. "
				       "kexec) symbols may be misresolved.";
			}
		}
530

531 532 533 534
		ui__warning(
"Kernel address maps (/proc/{kallsyms,modules}) were restricted.\n\n"
"Check /proc/sys/kernel/kptr_restrict before running 'perf record'.\n\n%s\n\n"
"Samples in kernel modules can't be resolved as well.\n\n",
535
		desc);
536 537
	}

538
	if (verbose > 3)
539
		perf_session__fprintf(session, stdout);
540

541
	if (verbose > 2)
542
		perf_session__fprintf_dsos(session, stdout);
543

544 545 546 547 548
	if (dump_trace) {
		perf_session__fprintf_nr_events(session, stdout);
		goto out_delete;
	}

549 550 551
	nr_samples = 0;
	list_for_each_entry(pos, &session->evlist->entries, node) {
		struct hists *hists = &pos->hists;
552

553 554 555
		if (pos->idx == 0)
			hists->symbol_filter_str = rep->symbol_filter_str;

556
		hists__collapse_resort(hists);
557
		nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE];
558 559 560 561 562 563 564 565 566

		/* Non-group events are considered as leader */
		if (symbol_conf.event_group &&
		    !perf_evsel__is_group_leader(pos)) {
			struct hists *leader_hists = &pos->leader->hists;

			hists__match(leader_hists, hists);
			hists__link(leader_hists, hists);
		}
567 568 569
	}

	if (nr_samples == 0) {
570
		ui__error("The %s file has no samples!\n", session->filename);
571
		goto out_delete;
572 573
	}

574 575 576
	list_for_each_entry(pos, &session->evlist->entries, node)
		hists__output_resort(&pos->hists);

577
	if (use_browser > 0) {
578
		if (use_browser == 1) {
579 580 581 582 583 584 585 586 587 588 589
			ret = perf_evlist__tui_browse_hists(session->evlist,
							help,
							NULL,
							&session->header.env);
			/*
			 * Usually "ret" is the last pressed key, and we only
			 * care if the key notifies us to switch data file.
			 */
			if (ret != K_SWITCH_INPUT_DATA)
				ret = 0;

590 591
		} else if (use_browser == 2) {
			perf_evlist__gtk_browse_hists(session->evlist, help,
592
						      NULL);
593
		}
594
	} else
595
		perf_evlist__tty_browse_hists(session->evlist, rep, help);
596

597
out_delete:
598 599 600 601 602 603 604 605 606 607 608 609
	/*
	 * Speed up the exit process, for large files this can
	 * take quite a while.
	 *
	 * XXX Enable this when using valgrind or if we ever
	 * librarize this command.
	 *
	 * Also experiment with obstacks to see how much speed
	 * up we'll get here.
	 *
 	 * perf_session__delete(session);
 	 */
610
	return ret;
611 612
}

613
static int
614
parse_callchain_opt(const struct option *opt, const char *arg, int unset)
615
{
616
	struct perf_report *rep = (struct perf_report *)opt->value;
617
	char *tok, *tok2;
618 619
	char *endptr;

620 621 622 623
	/*
	 * --no-call-graph
	 */
	if (unset) {
624
		rep->dont_use_callchains = true;
625 626 627
		return 0;
	}

628
	symbol_conf.use_callchain = true;
629 630 631 632

	if (!arg)
		return 0;

633 634 635 636 637 638
	tok = strtok((char *)arg, ",");
	if (!tok)
		return -1;

	/* get the output mode */
	if (!strncmp(tok, "graph", strlen(arg)))
639
		callchain_param.mode = CHAIN_GRAPH_ABS;
640

641
	else if (!strncmp(tok, "flat", strlen(arg)))
642 643 644 645 646
		callchain_param.mode = CHAIN_FLAT;

	else if (!strncmp(tok, "fractal", strlen(arg)))
		callchain_param.mode = CHAIN_GRAPH_REL;

647 648
	else if (!strncmp(tok, "none", strlen(arg))) {
		callchain_param.mode = CHAIN_NONE;
649
		symbol_conf.use_callchain = false;
650 651 652 653

		return 0;
	}

654 655 656
	else
		return -1;

657 658 659
	/* get the min percentage */
	tok = strtok(NULL, ",");
	if (!tok)
660
		goto setup;
661

662
	callchain_param.min_percent = strtod(tok, &endptr);
663 664 665
	if (tok == endptr)
		return -1;

666 667 668 669 670 671
	/* get the print limit */
	tok2 = strtok(NULL, ",");
	if (!tok2)
		goto setup;

	if (tok2[0] != 'c') {
672
		callchain_param.print_limit = strtoul(tok2, &endptr, 0);
673 674 675 676 677 678 679 680 681 682 683 684
		tok2 = strtok(NULL, ",");
		if (!tok2)
			goto setup;
	}

	/* get the call chain order */
	if (!strcmp(tok2, "caller"))
		callchain_param.order = ORDER_CALLER;
	else if (!strcmp(tok2, "callee"))
		callchain_param.order = ORDER_CALLEE;
	else
		return -1;
685
setup:
686
	if (callchain_register_param(&callchain_param) < 0) {
687 688 689
		fprintf(stderr, "Can't register callchain params\n");
		return -1;
	}
690 691 692
	return 0;
}

693
static int
694 695
parse_branch_mode(const struct option *opt __maybe_unused,
		  const char *str __maybe_unused, int unset)
696
{
697 698 699
	int *branch_mode = opt->value;

	*branch_mode = !unset;
700 701 702
	return 0;
}

703
int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
704
{
705
	struct perf_session *session;
706
	struct stat st;
707
	bool has_br_stack = false;
708
	int branch_mode = -1;
709
	int ret = -1;
710 711
	char callchain_default_opt[] = "fractal,0.5,callee";
	const char * const report_usage[] = {
N
Namhyung Kim 已提交
712
		"perf report [<options>]",
713 714 715
		NULL
	};
	struct perf_report report = {
716
		.tool = {
717 718 719
			.sample		 = process_sample_event,
			.mmap		 = perf_event__process_mmap,
			.comm		 = perf_event__process_comm,
720 721
			.exit		 = perf_event__process_exit,
			.fork		 = perf_event__process_fork,
722 723 724 725 726 727 728 729 730 731 732 733
			.lost		 = perf_event__process_lost,
			.read		 = process_read_event,
			.attr		 = perf_event__process_attr,
			.event_type	 = perf_event__process_event_type,
			.tracing_data	 = perf_event__process_tracing_data,
			.build_id	 = perf_event__process_build_id,
			.ordered_samples = true,
			.ordering_requires_timestamps = true,
		},
		.pretty_printing_style	 = "normal",
	};
	const struct option options[] = {
734
	OPT_STRING('i', "input", &input_name, "file",
735
		    "input file name"),
736
	OPT_INCR('v', "verbose", &verbose,
737
		    "be more verbose (show symbol address, etc)"),
738 739
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
		    "dump raw trace in ASCII"),
740 741
	OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
		   "file", "vmlinux pathname"),
742 743
	OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name,
		   "file", "kallsyms pathname"),
744
	OPT_BOOLEAN('f', "force", &report.force, "don't complain, do it"),
745
	OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules,
746
		    "load module symbols - WARNING: use only with -k and LIVE kernel"),
747
	OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples,
748
		    "Show a column with the number of samples"),
749
	OPT_BOOLEAN('T', "threads", &report.show_threads,
750
		    "Show per-thread event counters"),
751
	OPT_STRING(0, "pretty", &report.pretty_printing_style, "key",
752
		   "pretty printing style key: normal raw"),
753
	OPT_BOOLEAN(0, "tui", &report.use_tui, "Use the TUI interface"),
754
	OPT_BOOLEAN(0, "gtk", &report.use_gtk, "Use the GTK2 interface"),
755 756
	OPT_BOOLEAN(0, "stdio", &report.use_stdio,
		    "Use the stdio interface"),
757
	OPT_STRING('s', "sort", &sort_order, "key[,key2...]",
758
		   "sort by key(s): pid, comm, dso, symbol, parent, cpu, srcline,"
759
		   " dso_to, dso_from, symbol_to, symbol_from, mispredict,"
760 761
		   " weight, local_weight, mem, symbol_daddr, dso_daddr, tlb, "
		   "snoop, locked"),
762 763
	OPT_BOOLEAN(0, "showcpuutilization", &symbol_conf.show_cpu_utilization,
		    "Show sample percentage for different cpu modes"),
764 765
	OPT_STRING('p', "parent", &parent_pattern, "regex",
		   "regex filter to identify parent, see: '--sort parent'"),
766
	OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other,
767
		    "Only display entries with parent-match"),
768 769
	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order",
		     "Display callchains using output_type (graph, flat, fractal, or none) , min percent threshold, optional print limit and callchain order. "
770
		     "Default: fractal,0.5,callee", &parse_callchain_opt, callchain_default_opt),
771 772
	OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
		    "alias for inverted call graph"),
773
	OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
774
		   "only consider symbols in these dsos"),
775
	OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
776
		   "only consider symbols in these comms"),
777
	OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]",
778
		   "only consider these symbols"),
779 780
	OPT_STRING(0, "symbol-filter", &report.symbol_filter_str, "filter",
		   "only show symbols that (partially) match with this filter"),
781
	OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str,
782 783
		   "width[,width...]",
		   "don't try to adjust column width, use these fixed values"),
784
	OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator",
785 786
		   "separator for columns, no spaces will be added between "
		   "columns '.' is reserved."),
787
	OPT_BOOLEAN('U', "hide-unresolved", &report.hide_unresolved,
788
		    "Only display entries resolved to a symbol"),
789 790
	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
		    "Look for files with symbols relative to this directory"),
791
	OPT_STRING('C', "cpu", &report.cpu_list, "cpu",
792 793
		   "list of cpus to profile"),
	OPT_BOOLEAN('I', "show-info", &report.show_full_info,
794
		    "Display extended information about perf.data file"),
795 796 797 798
	OPT_BOOLEAN(0, "source", &symbol_conf.annotate_src,
		    "Interleave source code with assembly code (default)"),
	OPT_BOOLEAN(0, "asm-raw", &symbol_conf.annotate_asm_raw,
		    "Display raw encoding of assembly instructions (default)"),
799 800
	OPT_STRING('M', "disassembler-style", &disassembler_style, "disassembler style",
		   "Specify disassembler style (e.g. -M intel for intel syntax)"),
801 802
	OPT_BOOLEAN(0, "show-total-period", &symbol_conf.show_total_period,
		    "Show a column with the sum of periods"),
N
Namhyung Kim 已提交
803 804
	OPT_BOOLEAN(0, "group", &symbol_conf.event_group,
		    "Show event group information together"),
805
	OPT_CALLBACK_NOOPT('b', "branch-stack", &branch_mode, "",
806
		    "use branch records for histogram filling", parse_branch_mode),
807 808
	OPT_STRING(0, "objdump", &objdump_path, "path",
		   "objdump binary to use for disassembly and annotations"),
809 810
	OPT_BOOLEAN(0, "demangle", &symbol_conf.demangle,
		    "Disable symbol demangling"),
811
	OPT_BOOLEAN(0, "mem-mode", &report.mem_mode, "mem access profile"),
812
	OPT_END()
813
	};
814

815 816
	perf_config(perf_report_config, NULL);

817 818
	argc = parse_options(argc, argv, options, report_usage, 0);

819
	if (report.use_stdio)
820
		use_browser = 0;
821
	else if (report.use_tui)
822
		use_browser = 1;
823 824
	else if (report.use_gtk)
		use_browser = 2;
825

826
	if (report.inverted_callchain)
827 828
		callchain_param.order = ORDER_CALLER;

829
	if (!input_name || !strlen(input_name)) {
830
		if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode))
831
			input_name = "-";
832
		else
833
			input_name = "perf.data";
834
	}
835 836 837 838 839 840 841 842 843 844

	if (strcmp(input_name, "-") != 0)
		setup_browser(true);
	else {
		use_browser = 0;
		perf_hpp__column_enable(PERF_HPP__OVERHEAD);
		perf_hpp__init();
	}

repeat:
845
	session = perf_session__new(input_name, O_RDONLY,
846 847 848 849 850 851 852 853
				    report.force, false, &report.tool);
	if (session == NULL)
		return -ENOMEM;

	report.session = session;

	has_br_stack = perf_header__has_feat(&session->header,
					     HEADER_BRANCH_STACK);
854

855 856
	if (branch_mode == -1 && has_br_stack)
		sort__mode = SORT_MODE__BRANCH;
857

858 859
	/* sort__mode could be NORMAL if --no-branch-stack */
	if (sort__mode == SORT_MODE__BRANCH) {
860
		/*
861 862
		 * if no sort_order is provided, then specify
		 * branch-mode specific order
863 864 865 866 867
		 */
		if (sort_order == default_sort_order)
			sort_order = "comm,dso_from,symbol_from,"
				     "dso_to,symbol_to";

868
	}
869
	if (report.mem_mode) {
870
		if (sort__mode == SORT_MODE__BRANCH) {
871 872 873
			fprintf(stderr, "branch and mem mode incompatible\n");
			goto error;
		}
874 875
		sort__mode = SORT_MODE__MEMORY;

876 877 878 879 880 881 882
		/*
		 * if no sort_order is provided, then specify
		 * branch-mode specific order
		 */
		if (sort_order == default_sort_order)
			sort_order = "local_weight,mem,sym,dso,symbol_daddr,dso_daddr,snoop,tlb,locked";
	}
883

884 885
	if (setup_sorting() < 0)
		usage_with_options(report_usage, options);
886

887
	/*
888
	 * Only in the TUI browser we are doing integrated annotation,
889 890 891
	 * so don't allocate extra space that won't be used in the stdio
	 * implementation.
	 */
892
	if (use_browser == 1 && sort__has_sym) {
893
		symbol_conf.priv_size = sizeof(struct annotation);
894
		report.annotate_init  = symbol__annotate_init;
895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910
		/*
 		 * For searching by name on the "Browse map details".
 		 * providing it only in verbose mode not to bloat too
 		 * much struct symbol.
 		 */
		if (verbose) {
			/*
			 * XXX: Need to provide a less kludgy way to ask for
			 * more space per symbol, the u32 is for the index on
			 * the ui browser.
			 * See symbol__browser_index.
			 */
			symbol_conf.priv_size += sizeof(u32);
			symbol_conf.sort_by_name = true;
		}
	}
911

912
	if (symbol__init() < 0)
913
		goto error;
914

915
	if (parent_pattern != default_parent_pattern) {
916
		if (sort_dimension__add("parent") < 0)
917
			goto error;
918 919 920 921 922 923 924 925

		/*
		 * Only show the parent fields if we explicitly
		 * sort that way. If we only use parent machinery
		 * for filtering, we don't want it.
		 */
		if (!strstr(sort_order, "parent"))
			sort_parent.elide = 1;
926
	} else
927
		symbol_conf.exclude_other = false;
928

929 930 931 932 933 934 935 936 937 938
	if (argc) {
		/*
		 * Special case: if there's an argument left then assume that
		 * it's a symbol filter:
		 */
		if (argc > 1)
			usage_with_options(report_usage, options);

		report.symbol_filter_str = argv[0];
	}
939

940
	sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", stdout);
941

942
	if (sort__mode == SORT_MODE__BRANCH) {
943 944 945 946 947
		sort_entry__setup_elide(&sort_dso_from, symbol_conf.dso_from_list, "dso_from", stdout);
		sort_entry__setup_elide(&sort_dso_to, symbol_conf.dso_to_list, "dso_to", stdout);
		sort_entry__setup_elide(&sort_sym_from, symbol_conf.sym_from_list, "sym_from", stdout);
		sort_entry__setup_elide(&sort_sym_to, symbol_conf.sym_to_list, "sym_to", stdout);
	} else {
948 949 950 951 952 953 954 955
		if (report.mem_mode) {
			sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "symbol_daddr", stdout);
			sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso_daddr", stdout);
			sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "mem", stdout);
			sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "local_weight", stdout);
			sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "tlb", stdout);
			sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "snoop", stdout);
		}
956 957 958
		sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso", stdout);
		sort_entry__setup_elide(&sort_sym, symbol_conf.sym_list, "symbol", stdout);
	}
959

960
	ret = __cmd_report(&report);
961 962 963 964 965 966
	if (ret == K_SWITCH_INPUT_DATA) {
		perf_session__delete(session);
		goto repeat;
	} else
		ret = 0;

967 968 969
error:
	perf_session__delete(session);
	return ret;
970
}