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 "util/data.h"
37
#include "arch/common.h"
38

39
#include <dlfcn.h>
40 41
#include <linux/bitmap.h>

42
struct perf_report {
43
	struct perf_tool	tool;
44
	struct perf_session	*session;
45
	bool			force, use_tui, use_gtk, use_stdio;
46 47 48 49 50
	bool			hide_unresolved;
	bool			dont_use_callchains;
	bool			show_full_info;
	bool			show_threads;
	bool			inverted_callchain;
51
	bool			mem_mode;
52 53
	bool			header;
	bool			header_only;
54
	int			max_stack;
55 56 57
	struct perf_read_values	show_threads_values;
	const char		*pretty_printing_style;
	const char		*cpu_list;
58
	const char		*symbol_filter_str;
59
	float			min_percent;
60
	DECLARE_BITMAP(cpu_bitmap, MAX_NR_CPUS);
61
};
62

63 64 65 66 67 68
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;
	}
69 70 71 72 73
	if (!strcmp(var, "report.percent-limit")) {
		struct perf_report *rep = cb;
		rep->min_percent = strtof(value, NULL);
		return 0;
	}
74 75 76 77

	return perf_default_config(var, value, cb);
}

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
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,
96 97
						 sample, &parent, al,
						 rep->max_stack);
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119
		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.
	 */
120 121
	he = __hists__add_entry(&evsel->hists, al, parent, NULL, mi,
				cost, cost, 0);
122 123 124
	if (!he)
		return -ENOMEM;

125 126 127
	err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
	if (err)
		goto out;
128

129
	mx = he->mem_info;
130
	err = addr_map_symbol__inc_samples(&mx->daddr, evsel->idx);
131 132
	if (err)
		goto out;
133 134 135 136 137 138 139 140 141 142 143 144 145 146

	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;
}

147 148 149 150 151 152 153 154 155 156 157
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;
158
	struct branch_info *bi, *bx;
159 160 161

	if ((sort__has_parent || symbol_conf.use_callchain)
	    && sample->callchain) {
162
		err = machine__resolve_callchain(machine, evsel, al->thread,
163 164
						 sample, &parent, al,
						 rep->max_stack);
165 166 167 168 169 170 171 172 173 174 175 176
		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;
177 178 179

		err = -ENOMEM;

180 181 182 183
		/* overwrite the 'al' to branch-to info */
		al->map = bi[i].to.map;
		al->sym = bi[i].to.sym;
		al->addr = bi[i].to.addr;
184 185 186 187
		/*
		 * The report shows the percentage of total branches captured
		 * and not events sampled. Thus we use a pseudo period of 1.
		 */
188 189
		he = __hists__add_entry(&evsel->hists, al, parent, &bi[i], NULL,
					1, 1, 0);
190
		if (he) {
191
			bx = he->branch_info;
192
			err = addr_map_symbol__inc_samples(&bx->from, evsel->idx);
193 194 195
			if (err)
				goto out;

196
			err = addr_map_symbol__inc_samples(&bx->to, evsel->idx);
197 198 199
			if (err)
				goto out;

200 201 202
			evsel->hists.stats.total_period += 1;
			hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
		} else
203
			goto out;
204
	}
205
	err = 0;
206
out:
207
	free(bi);
208 209 210
	return err;
}

211 212
static int perf_evsel__add_hist_entry(struct perf_tool *tool,
				      struct perf_evsel *evsel,
213 214 215
				      struct addr_location *al,
				      struct perf_sample *sample,
				      struct machine *machine)
216
{
217
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
218
	struct symbol *parent = NULL;
219
	int err = 0;
220 221
	struct hist_entry *he;

222
	if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) {
223
		err = machine__resolve_callchain(machine, evsel, al->thread,
224 225
						 sample, &parent, al,
						 rep->max_stack);
226 227
		if (err)
			return err;
228
	}
229

230 231 232
	he = __hists__add_entry(&evsel->hists, al, parent, NULL, NULL,
				sample->period, sample->weight,
				sample->transaction);
233
	if (he == NULL)
234 235
		return -ENOMEM;

236
	if (symbol_conf.use_callchain) {
237
		err = callchain_append(he->callchain,
238
				       &callchain_cursor,
239
				       sample->period);
240
		if (err)
241
			return err;
242
	}
243

244
	err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
245 246
	evsel->hists.stats.total_period += sample->period;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
247
	return err;
248 249
}

250

251
static int process_sample_event(struct perf_tool *tool,
252
				union perf_event *event,
253
				struct perf_sample *sample,
254
				struct perf_evsel *evsel,
255
				struct machine *machine)
256
{
257
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
258
	struct addr_location al;
259
	int ret;
260

261
	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
262
		fprintf(stderr, "problem processing %d event, skipping it.\n",
263 264 265
			event->header.type);
		return -1;
	}
266

267
	if (al.filtered || (rep->hide_unresolved && al.sym == NULL))
268
		return 0;
269

270
	if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap))
271 272
		return 0;

273
	if (sort__mode == SORT_MODE__BRANCH) {
274 275 276
		ret = perf_report__add_branch_hist_entry(tool, &al, sample,
							 evsel, machine);
		if (ret < 0)
277
			pr_debug("problem adding lbr entry, skipping event\n");
278
	} else if (rep->mem_mode == 1) {
279 280 281
		ret = perf_report__add_mem_hist_entry(tool, &al, sample,
						      evsel, machine, event);
		if (ret < 0)
282
			pr_debug("problem adding mem entry, skipping event\n");
283 284 285
	} else {
		if (al.map != NULL)
			al.map->dso->hit = 1;
286

287 288
		ret = perf_evsel__add_hist_entry(tool, evsel, &al, sample,
						 machine);
289
		if (ret < 0)
290
			pr_debug("problem incrementing symbol period, skipping event\n");
291
	}
292
	return ret;
293
}
I
Ingo Molnar 已提交
294

295
static int process_read_event(struct perf_tool *tool,
296
			      union perf_event *event,
297
			      struct perf_sample *sample __maybe_unused,
298
			      struct perf_evsel *evsel,
299
			      struct machine *machine __maybe_unused)
300
{
301
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
302

303
	if (rep->show_threads) {
304
		const char *name = evsel ? perf_evsel__name(evsel) : "unknown";
305
		perf_read_values_add_value(&rep->show_threads_values,
306 307 308 309 310 311
					   event->read.pid, event->read.tid,
					   event->read.id,
					   name,
					   event->read.value);
	}

312
	dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid,
313
		    evsel ? perf_evsel__name(evsel) : "FAIL",
314
		    event->read.value);
315 316 317 318

	return 0;
}

319
/* For pipe mode, sample_type is not currently set */
320
static int perf_report__setup_sample_type(struct perf_report *rep)
321
{
322 323 324
	struct perf_session *session = rep->session;
	u64 sample_type = perf_evlist__combined_sample_type(session->evlist);
	bool is_pipe = perf_data_file__is_pipe(session->file);
325

326
	if (!is_pipe && !(sample_type & PERF_SAMPLE_CALLCHAIN)) {
327
		if (sort__has_parent) {
328
			ui__error("Selected --sort parent, but no "
329 330
				    "callchain data. Did you call "
				    "'perf record' without -g?\n");
331
			return -EINVAL;
332
		}
333
		if (symbol_conf.use_callchain) {
334
			ui__error("Selected -g but no callchain data. Did "
335
				    "you call 'perf record' without -g?\n");
336
			return -1;
337
		}
338 339
	} else if (!rep->dont_use_callchains &&
		   callchain_param.mode != CHAIN_NONE &&
340
		   !symbol_conf.use_callchain) {
341
			symbol_conf.use_callchain = true;
342
			if (callchain_register_param(&callchain_param) < 0) {
343
				ui__error("Can't register callchain params.\n");
344
				return -EINVAL;
345
			}
346 347
	}

348
	if (sort__mode == SORT_MODE__BRANCH) {
349
		if (!is_pipe &&
350
		    !(sample_type & PERF_SAMPLE_BRANCH_STACK)) {
351 352
			ui__error("Selected -b but no branch data. "
				  "Did you call perf record without -b?\n");
353 354 355 356
			return -1;
		}
	}

357 358
	return 0;
}
359

360
static void sig_handler(int sig __maybe_unused)
361 362 363 364
{
	session_done = 1;
}

365
static size_t hists__fprintf_nr_sample_events(struct perf_report *rep,
366
					      struct hists *hists,
367 368 369 370
					      const char *evname, FILE *fp)
{
	size_t ret;
	char unit;
371 372 373
	unsigned long nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE];
	u64 nr_events = hists->stats.total_period;
	struct perf_evsel *evsel = hists_to_evsel(hists);
374 375 376
	char buf[512];
	size_t size = sizeof(buf);

377
	if (perf_evsel__is_group_event(evsel)) {
378 379 380 381 382 383 384 385 386 387
		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;
		}
	}
388

389 390
	nr_samples = convert_unit(nr_samples, &unit);
	ret = fprintf(fp, "# Samples: %lu%c", nr_samples, unit);
391
	if (evname != NULL)
392 393
		ret += fprintf(fp, " of event '%s'", evname);

394 395 396 397 398
	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);
399 400 401
	return ret + fprintf(fp, "\n#\n");
}

402
static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist,
403
					 struct perf_report *rep,
404
					 const char *help)
405
{
406
	struct perf_evsel *pos;
407

408 409
	list_for_each_entry(pos, &evlist->entries, node) {
		struct hists *hists = &pos->hists;
410
		const char *evname = perf_evsel__name(pos);
411

412 413 414 415
		if (symbol_conf.event_group &&
		    !perf_evsel__is_group_leader(pos))
			continue;

416
		hists__fprintf_nr_sample_events(rep, hists, evname, stdout);
417
		hists__fprintf(hists, true, 0, 0, rep->min_percent, stdout);
418 419 420 421 422 423 424
		fprintf(stdout, "\n\n");
	}

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

425 426 427
		if (rep->show_threads) {
			bool style = !strcmp(rep->pretty_printing_style, "raw");
			perf_read_values_display(stdout, &rep->show_threads_values,
428
						 style);
429
			perf_read_values_destroy(&rep->show_threads_values);
430 431 432 433 434 435
		}
	}

	return 0;
}

436
static int __cmd_report(struct perf_report *rep)
437
{
438
	int ret = -EINVAL;
439
	u64 nr_samples;
440
	struct perf_session *session = rep->session;
441
	struct perf_evsel *pos;
442 443
	struct map *kernel_map;
	struct kmap *kernel_kmap;
444
	const char *help = "For a higher level overview, try: perf report --sort comm,dso";
445
	struct ui_progress prog;
446
	struct perf_data_file *file = session->file;
447

448 449
	signal(SIGINT, sig_handler);

450 451 452
	if (rep->cpu_list) {
		ret = perf_session__cpu_bitmap(session, rep->cpu_list,
					       rep->cpu_bitmap);
453
		if (ret)
454
			return ret;
455 456
	}

457 458
	if (rep->show_threads)
		perf_read_values_init(&rep->show_threads_values);
459

460
	ret = perf_report__setup_sample_type(rep);
461
	if (ret)
462
		return ret;
463

464
	ret = perf_session__process_events(session, &rep->tool);
465
	if (ret)
466
		return ret;
467

468
	kernel_map = session->machines.host.vmlinux_maps[MAP__FUNCTION];
469 470 471 472 473
	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))) {
474 475 476 477 478 479 480 481 482 483 484
		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.";
			}
		}
485

486 487 488 489
		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",
490
		desc);
491 492
	}

493
	if (verbose > 3)
494
		perf_session__fprintf(session, stdout);
495

496
	if (verbose > 2)
497
		perf_session__fprintf_dsos(session, stdout);
498

499 500
	if (dump_trace) {
		perf_session__fprintf_nr_events(session, stdout);
501
		return 0;
502 503
	}

504 505 506 507 508 509
	nr_samples = 0;
	list_for_each_entry(pos, &session->evlist->entries, node)
		nr_samples += pos->hists.nr_entries;

	ui_progress__init(&prog, nr_samples, "Merging related events...");

510 511 512
	nr_samples = 0;
	list_for_each_entry(pos, &session->evlist->entries, node) {
		struct hists *hists = &pos->hists;
513

514 515 516
		if (pos->idx == 0)
			hists->symbol_filter_str = rep->symbol_filter_str;

517
		hists__collapse_resort(hists, &prog);
518
		nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE];
519 520 521 522 523 524 525 526 527

		/* 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);
		}
528
	}
529
	ui_progress__finish();
530

531 532 533
	if (session_done())
		return 0;

534
	if (nr_samples == 0) {
535
		ui__error("The %s file has no samples!\n", file->path);
536
		return 0;
537 538
	}

539 540 541
	list_for_each_entry(pos, &session->evlist->entries, node)
		hists__output_resort(&pos->hists);

542
	if (use_browser > 0) {
543
		if (use_browser == 1) {
544
			ret = perf_evlist__tui_browse_hists(session->evlist,
545 546
							help, NULL,
							rep->min_percent,
547 548 549 550 551 552 553 554
							&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;

555
		} else if (use_browser == 2) {
556 557 558 559 560 561 562 563 564 565 566 567 568
			int (*hist_browser)(struct perf_evlist *,
					    const char *,
					    struct hist_browser_timer *,
					    float min_pcnt);

			hist_browser = dlsym(perf_gtk_handle,
					     "perf_evlist__gtk_browse_hists");
			if (hist_browser == NULL) {
				ui__error("GTK browser not found!\n");
				return ret;
			}
			hist_browser(session->evlist, help, NULL,
				     rep->min_percent);
569
		}
570
	} else
571
		perf_evlist__tty_browse_hists(session->evlist, rep, help);
572

573
	return ret;
574 575
}

576
static int
577
parse_callchain_opt(const struct option *opt, const char *arg, int unset)
578
{
579
	struct perf_report *rep = (struct perf_report *)opt->value;
580
	char *tok, *tok2;
581 582
	char *endptr;

583 584 585 586
	/*
	 * --no-call-graph
	 */
	if (unset) {
587
		rep->dont_use_callchains = true;
588 589 590
		return 0;
	}

591
	symbol_conf.use_callchain = true;
592 593 594 595

	if (!arg)
		return 0;

596 597 598 599 600 601
	tok = strtok((char *)arg, ",");
	if (!tok)
		return -1;

	/* get the output mode */
	if (!strncmp(tok, "graph", strlen(arg)))
602
		callchain_param.mode = CHAIN_GRAPH_ABS;
603

604
	else if (!strncmp(tok, "flat", strlen(arg)))
605 606 607 608 609
		callchain_param.mode = CHAIN_FLAT;

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

610 611
	else if (!strncmp(tok, "none", strlen(arg))) {
		callchain_param.mode = CHAIN_NONE;
612
		symbol_conf.use_callchain = false;
613 614 615 616

		return 0;
	}

617 618 619
	else
		return -1;

620 621 622
	/* get the min percentage */
	tok = strtok(NULL, ",");
	if (!tok)
623
		goto setup;
624

625
	callchain_param.min_percent = strtod(tok, &endptr);
626 627 628
	if (tok == endptr)
		return -1;

629 630 631 632 633 634
	/* get the print limit */
	tok2 = strtok(NULL, ",");
	if (!tok2)
		goto setup;

	if (tok2[0] != 'c') {
635
		callchain_param.print_limit = strtoul(tok2, &endptr, 0);
636 637 638 639 640 641
		tok2 = strtok(NULL, ",");
		if (!tok2)
			goto setup;
	}

	/* get the call chain order */
642
	if (!strncmp(tok2, "caller", strlen("caller")))
643
		callchain_param.order = ORDER_CALLER;
644
	else if (!strncmp(tok2, "callee", strlen("callee")))
645 646 647
		callchain_param.order = ORDER_CALLEE;
	else
		return -1;
648 649 650 651 652 653 654 655 656 657 658

	/* Get the sort key */
	tok2 = strtok(NULL, ",");
	if (!tok2)
		goto setup;
	if (!strncmp(tok2, "function", strlen("function")))
		callchain_param.key = CCKEY_FUNCTION;
	else if (!strncmp(tok2, "address", strlen("address")))
		callchain_param.key = CCKEY_ADDRESS;
	else
		return -1;
659
setup:
660
	if (callchain_register_param(&callchain_param) < 0) {
661 662 663
		fprintf(stderr, "Can't register callchain params\n");
		return -1;
	}
664 665 666
	return 0;
}

667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684
int
report_parse_ignore_callees_opt(const struct option *opt __maybe_unused,
				const char *arg, int unset __maybe_unused)
{
	if (arg) {
		int err = regcomp(&ignore_callees_regex, arg, REG_EXTENDED);
		if (err) {
			char buf[BUFSIZ];
			regerror(err, &ignore_callees_regex, buf, sizeof(buf));
			pr_err("Invalid --ignore-callees regex: %s\n%s", arg, buf);
			return -1;
		}
		have_ignore_callees = 1;
	}

	return 0;
}

685
static int
686 687
parse_branch_mode(const struct option *opt __maybe_unused,
		  const char *str __maybe_unused, int unset)
688
{
689 690 691
	int *branch_mode = opt->value;

	*branch_mode = !unset;
692 693 694
	return 0;
}

695 696 697 698 699 700 701 702 703 704
static int
parse_percent_limit(const struct option *opt, const char *str,
		    int unset __maybe_unused)
{
	struct perf_report *rep = opt->value;

	rep->min_percent = strtof(str, NULL);
	return 0;
}

705
int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
706
{
707
	struct perf_session *session;
708
	struct stat st;
709
	bool has_br_stack = false;
710
	int branch_mode = -1;
711
	int ret = -1;
712 713
	char callchain_default_opt[] = "fractal,0.5,callee";
	const char * const report_usage[] = {
N
Namhyung Kim 已提交
714
		"perf report [<options>]",
715 716 717
		NULL
	};
	struct perf_report report = {
718
		.tool = {
719 720
			.sample		 = process_sample_event,
			.mmap		 = perf_event__process_mmap,
721
			.mmap2		 = perf_event__process_mmap2,
722
			.comm		 = perf_event__process_comm,
723 724
			.exit		 = perf_event__process_exit,
			.fork		 = perf_event__process_fork,
725 726 727 728 729 730 731 732
			.lost		 = perf_event__process_lost,
			.read		 = process_read_event,
			.attr		 = perf_event__process_attr,
			.tracing_data	 = perf_event__process_tracing_data,
			.build_id	 = perf_event__process_build_id,
			.ordered_samples = true,
			.ordering_requires_timestamps = true,
		},
733
		.max_stack		 = PERF_MAX_STACK_DEPTH,
734 735 736
		.pretty_printing_style	 = "normal",
	};
	const struct option options[] = {
737
	OPT_STRING('i', "input", &input_name, "file",
738
		    "input file name"),
739
	OPT_INCR('v', "verbose", &verbose,
740
		    "be more verbose (show symbol address, etc)"),
741 742
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
		    "dump raw trace in ASCII"),
743 744
	OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
		   "file", "vmlinux pathname"),
745 746
	OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name,
		   "file", "kallsyms pathname"),
747
	OPT_BOOLEAN('f', "force", &report.force, "don't complain, do it"),
748
	OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules,
749
		    "load module symbols - WARNING: use only with -k and LIVE kernel"),
750
	OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples,
751
		    "Show a column with the number of samples"),
752
	OPT_BOOLEAN('T', "threads", &report.show_threads,
753
		    "Show per-thread event counters"),
754
	OPT_STRING(0, "pretty", &report.pretty_printing_style, "key",
755
		   "pretty printing style key: normal raw"),
756
	OPT_BOOLEAN(0, "tui", &report.use_tui, "Use the TUI interface"),
757
	OPT_BOOLEAN(0, "gtk", &report.use_gtk, "Use the GTK2 interface"),
758 759
	OPT_BOOLEAN(0, "stdio", &report.use_stdio,
		    "Use the stdio interface"),
760 761 762
	OPT_BOOLEAN(0, "header", &report.header, "Show data header."),
	OPT_BOOLEAN(0, "header-only", &report.header_only,
		    "Show only data header."),
763
	OPT_STRING('s', "sort", &sort_order, "key[,key2...]",
764
		   "sort by key(s): pid, comm, dso, symbol, parent, cpu, srcline,"
765
		   " dso_to, dso_from, symbol_to, symbol_from, mispredict,"
766
		   " weight, local_weight, mem, symbol_daddr, dso_daddr, tlb, "
767
		   "snoop, locked, abort, in_tx, transaction"),
768 769
	OPT_BOOLEAN(0, "showcpuutilization", &symbol_conf.show_cpu_utilization,
		    "Show sample percentage for different cpu modes"),
770 771
	OPT_STRING('p', "parent", &parent_pattern, "regex",
		   "regex filter to identify parent, see: '--sort parent'"),
772
	OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other,
773
		    "Only display entries with parent-match"),
774
	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order",
775 776
		     "Display callchains using output_type (graph, flat, fractal, or none) , min percent threshold, optional print limit, callchain order, key (function or address). "
		     "Default: fractal,0.5,callee,function", &parse_callchain_opt, callchain_default_opt),
777 778 779 780
	OPT_INTEGER(0, "max-stack", &report.max_stack,
		    "Set the maximum stack depth when parsing the callchain, "
		    "anything beyond the specified depth will be ignored. "
		    "Default: " __stringify(PERF_MAX_STACK_DEPTH)),
781 782
	OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
		    "alias for inverted call graph"),
783 784 785
	OPT_CALLBACK(0, "ignore-callees", NULL, "regex",
		   "ignore callees of these functions in call graphs",
		   report_parse_ignore_callees_opt),
786
	OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
787
		   "only consider symbols in these dsos"),
788
	OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
789
		   "only consider symbols in these comms"),
790
	OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]",
791
		   "only consider these symbols"),
792 793
	OPT_STRING(0, "symbol-filter", &report.symbol_filter_str, "filter",
		   "only show symbols that (partially) match with this filter"),
794
	OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str,
795 796
		   "width[,width...]",
		   "don't try to adjust column width, use these fixed values"),
797
	OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator",
798 799
		   "separator for columns, no spaces will be added between "
		   "columns '.' is reserved."),
800
	OPT_BOOLEAN('U', "hide-unresolved", &report.hide_unresolved,
801
		    "Only display entries resolved to a symbol"),
802 803
	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
		    "Look for files with symbols relative to this directory"),
804
	OPT_STRING('C', "cpu", &report.cpu_list, "cpu",
805 806
		   "list of cpus to profile"),
	OPT_BOOLEAN('I', "show-info", &report.show_full_info,
807
		    "Display extended information about perf.data file"),
808 809 810 811
	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)"),
812 813
	OPT_STRING('M', "disassembler-style", &disassembler_style, "disassembler style",
		   "Specify disassembler style (e.g. -M intel for intel syntax)"),
814 815
	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
816 817
	OPT_BOOLEAN(0, "group", &symbol_conf.event_group,
		    "Show event group information together"),
818
	OPT_CALLBACK_NOOPT('b', "branch-stack", &branch_mode, "",
819
		    "use branch records for histogram filling", parse_branch_mode),
820 821
	OPT_STRING(0, "objdump", &objdump_path, "path",
		   "objdump binary to use for disassembly and annotations"),
822 823
	OPT_BOOLEAN(0, "demangle", &symbol_conf.demangle,
		    "Disable symbol demangling"),
824
	OPT_BOOLEAN(0, "mem-mode", &report.mem_mode, "mem access profile"),
825 826
	OPT_CALLBACK(0, "percent-limit", &report, "percent",
		     "Don't show entries under that percent", parse_percent_limit),
827
	OPT_END()
828
	};
829 830 831
	struct perf_data_file file = {
		.mode  = PERF_DATA_MODE_READ,
	};
832

833
	perf_config(perf_report_config, &report);
834

835 836
	argc = parse_options(argc, argv, options, report_usage, 0);

837
	if (report.use_stdio)
838
		use_browser = 0;
839
	else if (report.use_tui)
840
		use_browser = 1;
841 842
	else if (report.use_gtk)
		use_browser = 2;
843

844
	if (report.inverted_callchain)
845 846
		callchain_param.order = ORDER_CALLER;

847
	if (!input_name || !strlen(input_name)) {
848
		if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode))
849
			input_name = "-";
850
		else
851
			input_name = "perf.data";
852
	}
853

854 855 856
	file.path  = input_name;
	file.force = report.force;

857
repeat:
858
	session = perf_session__new(&file, false, &report.tool);
859 860 861 862 863 864 865
	if (session == NULL)
		return -ENOMEM;

	report.session = session;

	has_br_stack = perf_header__has_feat(&session->header,
					     HEADER_BRANCH_STACK);
866

867 868
	if (branch_mode == -1 && has_br_stack)
		sort__mode = SORT_MODE__BRANCH;
869

870 871
	/* sort__mode could be NORMAL if --no-branch-stack */
	if (sort__mode == SORT_MODE__BRANCH) {
872
		/*
873 874
		 * if no sort_order is provided, then specify
		 * branch-mode specific order
875 876 877 878 879
		 */
		if (sort_order == default_sort_order)
			sort_order = "comm,dso_from,symbol_from,"
				     "dso_to,symbol_to";

880
	}
881
	if (report.mem_mode) {
882
		if (sort__mode == SORT_MODE__BRANCH) {
883 884 885
			fprintf(stderr, "branch and mem mode incompatible\n");
			goto error;
		}
886 887
		sort__mode = SORT_MODE__MEMORY;

888 889 890 891 892 893 894
		/*
		 * 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";
	}
895

896 897 898 899
	if (setup_sorting() < 0) {
		parse_options_usage(report_usage, options, "s", 1);
		goto error;
	}
900

901 902 903 904 905
	if (parent_pattern != default_parent_pattern) {
		if (sort_dimension__add("parent") < 0)
			goto error;
	}

906 907 908 909
	/* Force tty output for header output. */
	if (report.header || report.header_only)
		use_browser = 0;

910 911 912 913 914 915 916
	if (strcmp(input_name, "-") != 0)
		setup_browser(true);
	else {
		use_browser = 0;
		perf_hpp__init();
	}

917 918 919 920 921 922 923 924 925 926
	if (report.header || report.header_only) {
		perf_session__fprintf_info(session, stdout,
					   report.show_full_info);
		if (report.header_only)
			return 0;
	} else if (use_browser == 0) {
		fputs("# To display the perf.data header info, please use --header/--header-only options.\n#\n",
		      stdout);
	}

927
	/*
928
	 * Only in the TUI browser we are doing integrated annotation,
929 930 931
	 * so don't allocate extra space that won't be used in the stdio
	 * implementation.
	 */
932
	if (use_browser == 1 && sort__has_sym) {
933
		symbol_conf.priv_size = sizeof(struct annotation);
934 935
		machines__set_symbol_filter(&session->machines,
					    symbol__annotate_init);
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951
		/*
 		 * 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;
		}
	}
952

953
	if (symbol__init() < 0)
954
		goto error;
955

956 957 958 959 960 961 962 963 964 965
	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];
	}
966

967
	sort__setup_elide(stdout);
968

969
	ret = __cmd_report(&report);
970 971 972 973 974 975
	if (ret == K_SWITCH_INPUT_DATA) {
		perf_session__delete(session);
		goto repeat;
	} else
		ret = 0;

976 977 978
error:
	perf_session__delete(session);
	return ret;
979
}