builtin-report.c 26.6 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 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
static int report__config(const char *var, const char *value, void *cb)
64 65 66 67 68
{
	if (!strcmp(var, "report.group")) {
		symbol_conf.event_group = perf_config_bool(var, value);
		return 0;
	}
69
	if (!strcmp(var, "report.percent-limit")) {
70
		struct report *rep = cb;
71 72 73
		rep->min_percent = strtof(value, NULL);
		return 0;
	}
74 75 76 77

	return perf_default_config(var, value, cb);
}

78
static int report__add_mem_hist_entry(struct perf_tool *tool, struct addr_location *al,
79
				      struct perf_sample *sample, struct perf_evsel *evsel)
80
{
81
	struct report *rep = container_of(tool, struct report, tool);
82 83 84 85
	struct symbol *parent = NULL;
	struct hist_entry *he;
	struct mem_info *mi, *mx;
	uint64_t cost;
86
	int err = sample__resolve_callchain(sample, &parent, evsel, al, rep->max_stack);
87

88 89
	if (err)
		return err;
90

91
	mi = sample__resolve_mem(sample, al);
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108
	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.
	 */
109 110
	he = __hists__add_entry(&evsel->hists, al, parent, NULL, mi,
				cost, cost, 0);
111 112 113
	if (!he)
		return -ENOMEM;

114 115 116
	err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
	if (err)
		goto out;
117

118
	mx = he->mem_info;
119
	err = addr_map_symbol__inc_samples(&mx->daddr, evsel->idx);
120 121
	if (err)
		goto out;
122 123 124

	evsel->hists.stats.total_period += cost;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
125
	err = hist_entry__append_callchain(he, sample);
126 127 128 129
out:
	return err;
}

130
static int report__add_branch_hist_entry(struct perf_tool *tool, struct addr_location *al,
131
					 struct perf_sample *sample, struct perf_evsel *evsel)
132
{
133
	struct report *rep = container_of(tool, struct report, tool);
134 135 136
	struct symbol *parent = NULL;
	unsigned i;
	struct hist_entry *he;
137
	struct branch_info *bi, *bx;
138
	int err = sample__resolve_callchain(sample, &parent, evsel, al, rep->max_stack);
139

140 141
	if (err)
		return err;
142

143
	bi = sample__resolve_bstack(sample, al);
144 145 146 147 148 149
	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;
150 151 152

		err = -ENOMEM;

153 154 155 156
		/* 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;
157 158 159 160
		/*
		 * The report shows the percentage of total branches captured
		 * and not events sampled. Thus we use a pseudo period of 1.
		 */
161 162
		he = __hists__add_entry(&evsel->hists, al, parent, &bi[i], NULL,
					1, 1, 0);
163
		if (he) {
164
			bx = he->branch_info;
165
			err = addr_map_symbol__inc_samples(&bx->from, evsel->idx);
166 167 168
			if (err)
				goto out;

169
			err = addr_map_symbol__inc_samples(&bx->to, evsel->idx);
170 171 172
			if (err)
				goto out;

173 174 175
			evsel->hists.stats.total_period += 1;
			hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
		} else
176
			goto out;
177
	}
178
	err = 0;
179
out:
180
	free(bi);
181 182 183
	return err;
}

184
static int report__add_hist_entry(struct perf_tool *tool, struct perf_evsel *evsel,
185
				  struct addr_location *al, struct perf_sample *sample)
186
{
187
	struct report *rep = container_of(tool, struct report, tool);
188
	struct symbol *parent = NULL;
189
	struct hist_entry *he;
190
	int err = sample__resolve_callchain(sample, &parent, evsel, al, rep->max_stack);
191

192 193
	if (err)
		return err;
194

195 196 197
	he = __hists__add_entry(&evsel->hists, al, parent, NULL, NULL,
				sample->period, sample->weight,
				sample->transaction);
198
	if (he == NULL)
199 200
		return -ENOMEM;

201 202 203
	err = hist_entry__append_callchain(he, sample);
	if (err)
		goto out;
204

205
	err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
206 207
	evsel->hists.stats.total_period += sample->period;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
208
out:
209
	return err;
210 211
}

212

213
static int process_sample_event(struct perf_tool *tool,
214
				union perf_event *event,
215
				struct perf_sample *sample,
216
				struct perf_evsel *evsel,
217
				struct machine *machine)
218
{
219
	struct report *rep = container_of(tool, struct report, tool);
220
	struct addr_location al;
221
	int ret;
222

223
	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
224 225
		pr_debug("problem processing %d event, skipping it.\n",
			 event->header.type);
226 227
		return -1;
	}
228

229
	if (al.filtered || (rep->hide_unresolved && al.sym == NULL))
230
		return 0;
231

232
	if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap))
233 234
		return 0;

235
	if (sort__mode == SORT_MODE__BRANCH) {
236
		ret = report__add_branch_hist_entry(tool, &al, sample, evsel);
237
		if (ret < 0)
238
			pr_debug("problem adding lbr entry, skipping event\n");
239
	} else if (rep->mem_mode == 1) {
240
		ret = report__add_mem_hist_entry(tool, &al, sample, evsel);
241
		if (ret < 0)
242
			pr_debug("problem adding mem entry, skipping event\n");
243 244 245
	} else {
		if (al.map != NULL)
			al.map->dso->hit = 1;
246

247
		ret = report__add_hist_entry(tool, evsel, &al, sample);
248
		if (ret < 0)
249
			pr_debug("problem incrementing symbol period, skipping event\n");
250
	}
251
	return ret;
252
}
I
Ingo Molnar 已提交
253

254
static int process_read_event(struct perf_tool *tool,
255
			      union perf_event *event,
256
			      struct perf_sample *sample __maybe_unused,
257
			      struct perf_evsel *evsel,
258
			      struct machine *machine __maybe_unused)
259
{
260
	struct report *rep = container_of(tool, struct report, tool);
261

262
	if (rep->show_threads) {
263
		const char *name = evsel ? perf_evsel__name(evsel) : "unknown";
264
		perf_read_values_add_value(&rep->show_threads_values,
265 266 267 268 269 270
					   event->read.pid, event->read.tid,
					   event->read.id,
					   name,
					   event->read.value);
	}

271
	dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid,
272
		    evsel ? perf_evsel__name(evsel) : "FAIL",
273
		    event->read.value);
274 275 276 277

	return 0;
}

278
/* For pipe mode, sample_type is not currently set */
279
static int report__setup_sample_type(struct report *rep)
280
{
281 282 283
	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);
284

285
	if (!is_pipe && !(sample_type & PERF_SAMPLE_CALLCHAIN)) {
286
		if (sort__has_parent) {
287
			ui__error("Selected --sort parent, but no "
288 289
				    "callchain data. Did you call "
				    "'perf record' without -g?\n");
290
			return -EINVAL;
291
		}
292
		if (symbol_conf.use_callchain) {
293
			ui__error("Selected -g but no callchain data. Did "
294
				    "you call 'perf record' without -g?\n");
295
			return -1;
296
		}
297 298
	} else if (!rep->dont_use_callchains &&
		   callchain_param.mode != CHAIN_NONE &&
299
		   !symbol_conf.use_callchain) {
300
			symbol_conf.use_callchain = true;
301
			if (callchain_register_param(&callchain_param) < 0) {
302
				ui__error("Can't register callchain params.\n");
303
				return -EINVAL;
304
			}
305 306
	}

307
	if (sort__mode == SORT_MODE__BRANCH) {
308
		if (!is_pipe &&
309
		    !(sample_type & PERF_SAMPLE_BRANCH_STACK)) {
310 311
			ui__error("Selected -b but no branch data. "
				  "Did you call perf record without -b?\n");
312 313 314 315
			return -1;
		}
	}

316 317
	return 0;
}
318

319
static void sig_handler(int sig __maybe_unused)
320 321 322 323
{
	session_done = 1;
}

324
static size_t hists__fprintf_nr_sample_events(struct hists *hists, struct report *rep,
325 326 327 328
					      const char *evname, FILE *fp)
{
	size_t ret;
	char unit;
329 330 331
	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);
332 333 334
	char buf[512];
	size_t size = sizeof(buf);

335
	if (perf_evsel__is_group_event(evsel)) {
336 337 338 339 340 341 342 343 344 345
		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;
		}
	}
346

347 348
	nr_samples = convert_unit(nr_samples, &unit);
	ret = fprintf(fp, "# Samples: %lu%c", nr_samples, unit);
349
	if (evname != NULL)
350 351
		ret += fprintf(fp, " of event '%s'", evname);

352 353 354 355 356
	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);
357 358 359
	return ret + fprintf(fp, "\n#\n");
}

360
static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist,
361
					 struct report *rep,
362
					 const char *help)
363
{
364
	struct perf_evsel *pos;
365

366
	evlist__for_each(evlist, pos) {
367
		struct hists *hists = &pos->hists;
368
		const char *evname = perf_evsel__name(pos);
369

370 371 372 373
		if (symbol_conf.event_group &&
		    !perf_evsel__is_group_leader(pos))
			continue;

374
		hists__fprintf_nr_sample_events(hists, rep, evname, stdout);
375
		hists__fprintf(hists, true, 0, 0, rep->min_percent, stdout);
376 377 378 379 380 381 382
		fprintf(stdout, "\n\n");
	}

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

383 384 385
		if (rep->show_threads) {
			bool style = !strcmp(rep->pretty_printing_style, "raw");
			perf_read_values_display(stdout, &rep->show_threads_values,
386
						 style);
387
			perf_read_values_destroy(&rep->show_threads_values);
388 389 390 391 392 393
		}
	}

	return 0;
}

394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422
static void report__warn_kptr_restrict(const struct report *rep)
{
	struct map *kernel_map = rep->session->machines.host.vmlinux_maps[MAP__FUNCTION];
	struct kmap *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))) {
		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.";
			}
		}

		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",
		desc);
	}
}

423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
static int report__gtk_browse_hists(struct report *rep, const char *help)
{
	int (*hist_browser)(struct perf_evlist *evlist, const char *help,
			    struct hist_browser_timer *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 -1;
	}

	return hist_browser(rep->session->evlist, help, NULL, rep->min_percent);
}

static int report__browse_hists(struct report *rep)
{
	int ret;
	struct perf_session *session = rep->session;
	struct perf_evlist *evlist = session->evlist;
	const char *help = "For a higher level overview, try: perf report --sort comm,dso";

	switch (use_browser) {
	case 1:
		ret = perf_evlist__tui_browse_hists(evlist, help, NULL,
						    rep->min_percent,
						    &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;
		break;
	case 2:
		ret = report__gtk_browse_hists(rep, help);
		break;
	default:
		ret = perf_evlist__tty_browse_hists(evlist, rep, help);
		break;
	}

	return ret;
}

468 469 470 471 472 473 474 475 476
static u64 report__collapse_hists(struct report *rep)
{
	struct ui_progress prog;
	struct perf_evsel *pos;
	u64 nr_samples = 0;
	/*
 	 * Count number of histogram entries to use when showing progress,
 	 * reusing nr_samples variable.
 	 */
477
	evlist__for_each(rep->session->evlist, pos)
478 479 480 481 482 483 484 485 486
		nr_samples += pos->hists.nr_entries;

	ui_progress__init(&prog, nr_samples, "Merging related events...");
	/*
	 * Count total number of samples, will be used to check if this
 	 * session had any.
 	 */
	nr_samples = 0;

487
	evlist__for_each(rep->session->evlist, pos) {
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
		struct hists *hists = &pos->hists;

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

		hists__collapse_resort(hists, &prog);
		nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE];

		/* 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);
		}
	}

	ui_progress__finish();

	return nr_samples;
}

511
static int __cmd_report(struct report *rep)
512
{
513
	int ret;
514
	u64 nr_samples;
515
	struct perf_session *session = rep->session;
516
	struct perf_evsel *pos;
517
	struct perf_data_file *file = session->file;
518

519 520
	signal(SIGINT, sig_handler);

521 522 523
	if (rep->cpu_list) {
		ret = perf_session__cpu_bitmap(session, rep->cpu_list,
					       rep->cpu_bitmap);
524
		if (ret)
525
			return ret;
526 527
	}

528 529
	if (rep->show_threads)
		perf_read_values_init(&rep->show_threads_values);
530

531
	ret = report__setup_sample_type(rep);
532
	if (ret)
533
		return ret;
534

535
	ret = perf_session__process_events(session, &rep->tool);
536
	if (ret)
537
		return ret;
538

539
	report__warn_kptr_restrict(rep);
540

541 542 543
	if (use_browser == 0) {
		if (verbose > 3)
			perf_session__fprintf(session, stdout);
544

545 546
		if (verbose > 2)
			perf_session__fprintf_dsos(session, stdout);
547

548 549 550 551
		if (dump_trace) {
			perf_session__fprintf_nr_events(session, stdout);
			return 0;
		}
552 553
	}

554
	nr_samples = report__collapse_hists(rep);
555

556 557 558
	if (session_done())
		return 0;

559
	if (nr_samples == 0) {
560
		ui__error("The %s file has no samples!\n", file->path);
561
		return 0;
562 563
	}

564
	evlist__for_each(session->evlist, pos)
565 566
		hists__output_resort(&pos->hists);

567
	return report__browse_hists(rep);
568 569
}

570
static int
571
parse_callchain_opt(const struct option *opt, const char *arg, int unset)
572
{
573
	struct report *rep = (struct report *)opt->value;
574
	char *tok, *tok2;
575 576
	char *endptr;

577 578 579 580
	/*
	 * --no-call-graph
	 */
	if (unset) {
581
		rep->dont_use_callchains = true;
582 583 584
		return 0;
	}

585
	symbol_conf.use_callchain = true;
586 587 588 589

	if (!arg)
		return 0;

590 591 592 593 594 595
	tok = strtok((char *)arg, ",");
	if (!tok)
		return -1;

	/* get the output mode */
	if (!strncmp(tok, "graph", strlen(arg)))
596
		callchain_param.mode = CHAIN_GRAPH_ABS;
597

598
	else if (!strncmp(tok, "flat", strlen(arg)))
599 600 601 602 603
		callchain_param.mode = CHAIN_FLAT;

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

604 605
	else if (!strncmp(tok, "none", strlen(arg))) {
		callchain_param.mode = CHAIN_NONE;
606
		symbol_conf.use_callchain = false;
607 608 609 610

		return 0;
	}

611 612 613
	else
		return -1;

614 615 616
	/* get the min percentage */
	tok = strtok(NULL, ",");
	if (!tok)
617
		goto setup;
618

619
	callchain_param.min_percent = strtod(tok, &endptr);
620 621 622
	if (tok == endptr)
		return -1;

623 624 625 626 627 628
	/* get the print limit */
	tok2 = strtok(NULL, ",");
	if (!tok2)
		goto setup;

	if (tok2[0] != 'c') {
629
		callchain_param.print_limit = strtoul(tok2, &endptr, 0);
630 631 632 633 634 635
		tok2 = strtok(NULL, ",");
		if (!tok2)
			goto setup;
	}

	/* get the call chain order */
636
	if (!strncmp(tok2, "caller", strlen("caller")))
637
		callchain_param.order = ORDER_CALLER;
638
	else if (!strncmp(tok2, "callee", strlen("callee")))
639 640 641
		callchain_param.order = ORDER_CALLEE;
	else
		return -1;
642 643 644 645 646 647 648 649 650 651 652

	/* 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;
653
setup:
654
	if (callchain_register_param(&callchain_param) < 0) {
655
		pr_err("Can't register callchain params\n");
656 657
		return -1;
	}
658 659 660
	return 0;
}

661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678
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;
}

679
static int
680 681
parse_branch_mode(const struct option *opt __maybe_unused,
		  const char *str __maybe_unused, int unset)
682
{
683 684 685
	int *branch_mode = opt->value;

	*branch_mode = !unset;
686 687 688
	return 0;
}

689 690 691 692
static int
parse_percent_limit(const struct option *opt, const char *str,
		    int unset __maybe_unused)
{
693
	struct report *rep = opt->value;
694 695 696 697 698

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

699
int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
700
{
701
	struct perf_session *session;
702
	struct stat st;
703
	bool has_br_stack = false;
704
	int branch_mode = -1;
705
	int ret = -1;
706 707
	char callchain_default_opt[] = "fractal,0.5,callee";
	const char * const report_usage[] = {
N
Namhyung Kim 已提交
708
		"perf report [<options>]",
709 710
		NULL
	};
711
	struct report report = {
712
		.tool = {
713 714
			.sample		 = process_sample_event,
			.mmap		 = perf_event__process_mmap,
715
			.mmap2		 = perf_event__process_mmap2,
716
			.comm		 = perf_event__process_comm,
717 718
			.exit		 = perf_event__process_exit,
			.fork		 = perf_event__process_fork,
719 720 721 722 723 724 725 726
			.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,
		},
727
		.max_stack		 = PERF_MAX_STACK_DEPTH,
728 729 730
		.pretty_printing_style	 = "normal",
	};
	const struct option options[] = {
731
	OPT_STRING('i', "input", &input_name, "file",
732
		    "input file name"),
733
	OPT_INCR('v', "verbose", &verbose,
734
		    "be more verbose (show symbol address, etc)"),
735 736
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
		    "dump raw trace in ASCII"),
737 738
	OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
		   "file", "vmlinux pathname"),
739 740
	OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name,
		   "file", "kallsyms pathname"),
741
	OPT_BOOLEAN('f', "force", &report.force, "don't complain, do it"),
742
	OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules,
743
		    "load module symbols - WARNING: use only with -k and LIVE kernel"),
744
	OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples,
745
		    "Show a column with the number of samples"),
746
	OPT_BOOLEAN('T', "threads", &report.show_threads,
747
		    "Show per-thread event counters"),
748
	OPT_STRING(0, "pretty", &report.pretty_printing_style, "key",
749
		   "pretty printing style key: normal raw"),
750
	OPT_BOOLEAN(0, "tui", &report.use_tui, "Use the TUI interface"),
751
	OPT_BOOLEAN(0, "gtk", &report.use_gtk, "Use the GTK2 interface"),
752 753
	OPT_BOOLEAN(0, "stdio", &report.use_stdio,
		    "Use the stdio interface"),
754 755 756
	OPT_BOOLEAN(0, "header", &report.header, "Show data header."),
	OPT_BOOLEAN(0, "header-only", &report.header_only,
		    "Show only data header."),
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
		   " weight, local_weight, mem, symbol_daddr, dso_daddr, tlb, "
761
		   "snoop, locked, abort, in_tx, transaction"),
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
	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order",
769 770
		     "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),
771 772 773 774
	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)),
775 776
	OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
		    "alias for inverted call graph"),
777 778 779
	OPT_CALLBACK(0, "ignore-callees", NULL, "regex",
		   "ignore callees of these functions in call graphs",
		   report_parse_ignore_callees_opt),
780
	OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
781
		   "only consider symbols in these dsos"),
782
	OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
783
		   "only consider symbols in these comms"),
784
	OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]",
785
		   "only consider these symbols"),
786 787
	OPT_STRING(0, "symbol-filter", &report.symbol_filter_str, "filter",
		   "only show symbols that (partially) match with this filter"),
788
	OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str,
789 790
		   "width[,width...]",
		   "don't try to adjust column width, use these fixed values"),
791
	OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator",
792 793
		   "separator for columns, no spaces will be added between "
		   "columns '.' is reserved."),
794
	OPT_BOOLEAN('U', "hide-unresolved", &report.hide_unresolved,
795
		    "Only display entries resolved to a symbol"),
796 797
	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
		    "Look for files with symbols relative to this directory"),
798
	OPT_STRING('C', "cpu", &report.cpu_list, "cpu",
799 800
		   "list of cpus to profile"),
	OPT_BOOLEAN('I', "show-info", &report.show_full_info,
801
		    "Display extended information about perf.data file"),
802 803 804 805
	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)"),
806 807
	OPT_STRING('M', "disassembler-style", &disassembler_style, "disassembler style",
		   "Specify disassembler style (e.g. -M intel for intel syntax)"),
808 809
	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
810 811
	OPT_BOOLEAN(0, "group", &symbol_conf.event_group,
		    "Show event group information together"),
812
	OPT_CALLBACK_NOOPT('b', "branch-stack", &branch_mode, "",
813
		    "use branch records for histogram filling", parse_branch_mode),
814 815
	OPT_STRING(0, "objdump", &objdump_path, "path",
		   "objdump binary to use for disassembly and annotations"),
816 817
	OPT_BOOLEAN(0, "demangle", &symbol_conf.demangle,
		    "Disable symbol demangling"),
818
	OPT_BOOLEAN(0, "mem-mode", &report.mem_mode, "mem access profile"),
819 820
	OPT_CALLBACK(0, "percent-limit", &report, "percent",
		     "Don't show entries under that percent", parse_percent_limit),
821
	OPT_END()
822
	};
823 824 825
	struct perf_data_file file = {
		.mode  = PERF_DATA_MODE_READ,
	};
826

827
	perf_config(report__config, &report);
828

829 830
	argc = parse_options(argc, argv, options, report_usage, 0);

831
	if (report.use_stdio)
832
		use_browser = 0;
833
	else if (report.use_tui)
834
		use_browser = 1;
835 836
	else if (report.use_gtk)
		use_browser = 2;
837

838
	if (report.inverted_callchain)
839 840
		callchain_param.order = ORDER_CALLER;

841
	if (!input_name || !strlen(input_name)) {
842
		if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode))
843
			input_name = "-";
844
		else
845
			input_name = "perf.data";
846
	}
847

848 849 850
	file.path  = input_name;
	file.force = report.force;

851
repeat:
852
	session = perf_session__new(&file, false, &report.tool);
853 854 855 856 857 858 859
	if (session == NULL)
		return -ENOMEM;

	report.session = session;

	has_br_stack = perf_header__has_feat(&session->header,
					     HEADER_BRANCH_STACK);
860

861 862
	if (branch_mode == -1 && has_br_stack)
		sort__mode = SORT_MODE__BRANCH;
863

864 865
	/* sort__mode could be NORMAL if --no-branch-stack */
	if (sort__mode == SORT_MODE__BRANCH) {
866
		/*
867 868
		 * if no sort_order is provided, then specify
		 * branch-mode specific order
869 870 871 872 873
		 */
		if (sort_order == default_sort_order)
			sort_order = "comm,dso_from,symbol_from,"
				     "dso_to,symbol_to";

874
	}
875
	if (report.mem_mode) {
876
		if (sort__mode == SORT_MODE__BRANCH) {
877
			pr_err("branch and mem mode incompatible\n");
878 879
			goto error;
		}
880 881
		sort__mode = SORT_MODE__MEMORY;

882 883 884 885 886 887 888
		/*
		 * 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";
	}
889

890 891 892 893
	if (setup_sorting() < 0) {
		parse_options_usage(report_usage, options, "s", 1);
		goto error;
	}
894

895 896 897 898 899
	if (parent_pattern != default_parent_pattern) {
		if (sort_dimension__add("parent") < 0)
			goto error;
	}

900 901 902 903
	/* Force tty output for header output. */
	if (report.header || report.header_only)
		use_browser = 0;

904 905 906 907 908 909 910
	if (strcmp(input_name, "-") != 0)
		setup_browser(true);
	else {
		use_browser = 0;
		perf_hpp__init();
	}

911 912 913 914 915 916 917 918 919 920
	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);
	}

921
	/*
922
	 * Only in the TUI browser we are doing integrated annotation,
923 924 925
	 * so don't allocate extra space that won't be used in the stdio
	 * implementation.
	 */
926
	if (use_browser == 1 && sort__has_sym) {
927
		symbol_conf.priv_size = sizeof(struct annotation);
928 929
		machines__set_symbol_filter(&session->machines,
					    symbol__annotate_init);
930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945
		/*
 		 * 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;
		}
	}
946

947
	if (symbol__init() < 0)
948
		goto error;
949

950 951 952 953 954 955 956 957 958 959
	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];
	}
960

961
	sort__setup_elide(stdout);
962

963
	ret = __cmd_report(&report);
964 965 966 967 968 969
	if (ret == K_SWITCH_INPUT_DATA) {
		perf_session__delete(session);
		goto repeat;
	} else
		ret = 0;

970 971 972
error:
	perf_session__delete(session);
	return ret;
973
}