builtin-report.c 27.2 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

129 130 131 132 133
	mx = he->mem_info;
	err = symbol__inc_addr_samples(mx->daddr.sym, mx->daddr.map,
				       evsel->idx, mx->daddr.al_addr);
	if (err)
		goto out;
134 135 136 137 138 139 140 141 142 143 144 145 146 147

	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;
}

148 149 150 151 152 153 154 155 156 157 158
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;
159
	struct branch_info *bi, *bx;
160 161 162

	if ((sort__has_parent || symbol_conf.use_callchain)
	    && sample->callchain) {
163
		err = machine__resolve_callchain(machine, evsel, al->thread,
164 165
						 sample, &parent, al,
						 rep->max_stack);
166 167 168 169 170 171 172 173 174 175 176 177
		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;
178 179 180

		err = -ENOMEM;

181 182 183 184
		/* 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;
185 186 187 188
		/*
		 * The report shows the percentage of total branches captured
		 * and not events sampled. Thus we use a pseudo period of 1.
		 */
189 190
		he = __hists__add_entry(&evsel->hists, al, parent, &bi[i], NULL,
					1, 1, 0);
191
		if (he) {
192
			bx = he->branch_info;
193 194 195 196 197 198 199 200 201 202 203 204
			err = symbol__inc_addr_samples(bx->from.sym,
						       bx->from.map, evsel->idx,
						       bx->from.al_addr);
			if (err)
				goto out;

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

205 206 207
			evsel->hists.stats.total_period += 1;
			hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
		} else
208
			goto out;
209
	}
210
	err = 0;
211
out:
212
	free(bi);
213 214 215
	return err;
}

216 217
static int perf_evsel__add_hist_entry(struct perf_tool *tool,
				      struct perf_evsel *evsel,
218 219 220
				      struct addr_location *al,
				      struct perf_sample *sample,
				      struct machine *machine)
221
{
222
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
223
	struct symbol *parent = NULL;
224
	int err = 0;
225 226
	struct hist_entry *he;

227
	if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) {
228
		err = machine__resolve_callchain(machine, evsel, al->thread,
229 230
						 sample, &parent, al,
						 rep->max_stack);
231 232
		if (err)
			return err;
233
	}
234

235 236 237
	he = __hists__add_entry(&evsel->hists, al, parent, NULL, NULL,
				sample->period, sample->weight,
				sample->transaction);
238
	if (he == NULL)
239 240
		return -ENOMEM;

241
	if (symbol_conf.use_callchain) {
242
		err = callchain_append(he->callchain,
243
				       &callchain_cursor,
244
				       sample->period);
245
		if (err)
246
			return err;
247
	}
248

249
	err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
250 251
	evsel->hists.stats.total_period += sample->period;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
252
	return err;
253 254
}

255

256
static int process_sample_event(struct perf_tool *tool,
257
				union perf_event *event,
258
				struct perf_sample *sample,
259
				struct perf_evsel *evsel,
260
				struct machine *machine)
261
{
262
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
263
	struct addr_location al;
264
	int ret;
265

266
	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
267
		fprintf(stderr, "problem processing %d event, skipping it.\n",
268 269 270
			event->header.type);
		return -1;
	}
271

272
	if (al.filtered || (rep->hide_unresolved && al.sym == NULL))
273
		return 0;
274

275
	if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap))
276 277
		return 0;

278
	if (sort__mode == SORT_MODE__BRANCH) {
279 280 281
		ret = perf_report__add_branch_hist_entry(tool, &al, sample,
							 evsel, machine);
		if (ret < 0)
282
			pr_debug("problem adding lbr entry, skipping event\n");
283
	} else if (rep->mem_mode == 1) {
284 285 286
		ret = perf_report__add_mem_hist_entry(tool, &al, sample,
						      evsel, machine, event);
		if (ret < 0)
287
			pr_debug("problem adding mem entry, skipping event\n");
288 289 290
	} else {
		if (al.map != NULL)
			al.map->dso->hit = 1;
291

292 293
		ret = perf_evsel__add_hist_entry(tool, evsel, &al, sample,
						 machine);
294
		if (ret < 0)
295
			pr_debug("problem incrementing symbol period, skipping event\n");
296
	}
297
	return ret;
298
}
I
Ingo Molnar 已提交
299

300
static int process_read_event(struct perf_tool *tool,
301
			      union perf_event *event,
302
			      struct perf_sample *sample __maybe_unused,
303
			      struct perf_evsel *evsel,
304
			      struct machine *machine __maybe_unused)
305
{
306
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
307

308
	if (rep->show_threads) {
309
		const char *name = evsel ? perf_evsel__name(evsel) : "unknown";
310
		perf_read_values_add_value(&rep->show_threads_values,
311 312 313 314 315 316
					   event->read.pid, event->read.tid,
					   event->read.id,
					   name,
					   event->read.value);
	}

317
	dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid,
318
		    evsel ? perf_evsel__name(evsel) : "FAIL",
319
		    event->read.value);
320 321 322 323

	return 0;
}

324
/* For pipe mode, sample_type is not currently set */
325
static int perf_report__setup_sample_type(struct perf_report *rep)
326
{
327 328 329
	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);
330

331
	if (!is_pipe && !(sample_type & PERF_SAMPLE_CALLCHAIN)) {
332
		if (sort__has_parent) {
333
			ui__error("Selected --sort parent, but no "
334 335
				    "callchain data. Did you call "
				    "'perf record' without -g?\n");
336
			return -EINVAL;
337
		}
338
		if (symbol_conf.use_callchain) {
339
			ui__error("Selected -g but no callchain data. Did "
340
				    "you call 'perf record' without -g?\n");
341
			return -1;
342
		}
343 344
	} else if (!rep->dont_use_callchains &&
		   callchain_param.mode != CHAIN_NONE &&
345
		   !symbol_conf.use_callchain) {
346
			symbol_conf.use_callchain = true;
347
			if (callchain_register_param(&callchain_param) < 0) {
348
				ui__error("Can't register callchain params.\n");
349
				return -EINVAL;
350
			}
351 352
	}

353
	if (sort__mode == SORT_MODE__BRANCH) {
354
		if (!is_pipe &&
355
		    !(sample_type & PERF_SAMPLE_BRANCH_STACK)) {
356 357
			ui__error("Selected -b but no branch data. "
				  "Did you call perf record without -b?\n");
358 359 360 361
			return -1;
		}
	}

362 363
	return 0;
}
364

365
static void sig_handler(int sig __maybe_unused)
366 367 368 369
{
	session_done = 1;
}

370
static size_t hists__fprintf_nr_sample_events(struct perf_report *rep,
371
					      struct hists *hists,
372 373 374 375
					      const char *evname, FILE *fp)
{
	size_t ret;
	char unit;
376 377 378
	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);
379 380 381
	char buf[512];
	size_t size = sizeof(buf);

382
	if (perf_evsel__is_group_event(evsel)) {
383 384 385 386 387 388 389 390 391 392
		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;
		}
	}
393

394 395
	nr_samples = convert_unit(nr_samples, &unit);
	ret = fprintf(fp, "# Samples: %lu%c", nr_samples, unit);
396
	if (evname != NULL)
397 398
		ret += fprintf(fp, " of event '%s'", evname);

399 400 401 402 403
	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);
404 405 406
	return ret + fprintf(fp, "\n#\n");
}

407
static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist,
408
					 struct perf_report *rep,
409
					 const char *help)
410
{
411
	struct perf_evsel *pos;
412

413 414
	list_for_each_entry(pos, &evlist->entries, node) {
		struct hists *hists = &pos->hists;
415
		const char *evname = perf_evsel__name(pos);
416

417 418 419 420
		if (symbol_conf.event_group &&
		    !perf_evsel__is_group_leader(pos))
			continue;

421
		hists__fprintf_nr_sample_events(rep, hists, evname, stdout);
422
		hists__fprintf(hists, true, 0, 0, rep->min_percent, stdout);
423 424 425 426 427 428 429
		fprintf(stdout, "\n\n");
	}

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

430 431 432
		if (rep->show_threads) {
			bool style = !strcmp(rep->pretty_printing_style, "raw");
			perf_read_values_display(stdout, &rep->show_threads_values,
433
						 style);
434
			perf_read_values_destroy(&rep->show_threads_values);
435 436 437 438 439 440
		}
	}

	return 0;
}

441
static int __cmd_report(struct perf_report *rep)
442
{
443
	int ret = -EINVAL;
444
	u64 nr_samples;
445
	struct perf_session *session = rep->session;
446
	struct perf_evsel *pos;
447 448
	struct map *kernel_map;
	struct kmap *kernel_kmap;
449
	const char *help = "For a higher level overview, try: perf report --sort comm,dso";
450
	struct ui_progress prog;
451
	struct perf_data_file *file = session->file;
452

453 454
	signal(SIGINT, sig_handler);

455 456 457
	if (rep->cpu_list) {
		ret = perf_session__cpu_bitmap(session, rep->cpu_list,
					       rep->cpu_bitmap);
458
		if (ret)
459
			return ret;
460 461
	}

462 463
	if (rep->show_threads)
		perf_read_values_init(&rep->show_threads_values);
464

465
	ret = perf_report__setup_sample_type(rep);
466
	if (ret)
467
		return ret;
468

469
	ret = perf_session__process_events(session, &rep->tool);
470
	if (ret)
471
		return ret;
472

473
	kernel_map = session->machines.host.vmlinux_maps[MAP__FUNCTION];
474 475 476 477 478
	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))) {
479 480 481 482 483 484 485 486 487 488 489
		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.";
			}
		}
490

491 492 493 494
		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",
495
		desc);
496 497
	}

498
	if (verbose > 3)
499
		perf_session__fprintf(session, stdout);
500

501
	if (verbose > 2)
502
		perf_session__fprintf_dsos(session, stdout);
503

504 505
	if (dump_trace) {
		perf_session__fprintf_nr_events(session, stdout);
506
		return 0;
507 508
	}

509 510 511 512 513 514
	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...");

515 516 517
	nr_samples = 0;
	list_for_each_entry(pos, &session->evlist->entries, node) {
		struct hists *hists = &pos->hists;
518

519 520 521
		if (pos->idx == 0)
			hists->symbol_filter_str = rep->symbol_filter_str;

522
		hists__collapse_resort(hists, &prog);
523
		nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE];
524 525 526 527 528 529 530 531 532

		/* 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);
		}
533
	}
534
	ui_progress__finish();
535

536 537 538
	if (session_done())
		return 0;

539
	if (nr_samples == 0) {
540
		ui__error("The %s file has no samples!\n", file->path);
541
		return 0;
542 543
	}

544 545 546
	list_for_each_entry(pos, &session->evlist->entries, node)
		hists__output_resort(&pos->hists);

547
	if (use_browser > 0) {
548
		if (use_browser == 1) {
549
			ret = perf_evlist__tui_browse_hists(session->evlist,
550 551
							help, NULL,
							rep->min_percent,
552 553 554 555 556 557 558 559
							&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;

560
		} else if (use_browser == 2) {
561 562 563 564 565 566 567 568 569 570 571 572 573
			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);
574
		}
575
	} else
576
		perf_evlist__tty_browse_hists(session->evlist, rep, help);
577

578
	return ret;
579 580
}

581
static int
582
parse_callchain_opt(const struct option *opt, const char *arg, int unset)
583
{
584
	struct perf_report *rep = (struct perf_report *)opt->value;
585
	char *tok, *tok2;
586 587
	char *endptr;

588 589 590 591
	/*
	 * --no-call-graph
	 */
	if (unset) {
592
		rep->dont_use_callchains = true;
593 594 595
		return 0;
	}

596
	symbol_conf.use_callchain = true;
597 598 599 600

	if (!arg)
		return 0;

601 602 603 604 605 606
	tok = strtok((char *)arg, ",");
	if (!tok)
		return -1;

	/* get the output mode */
	if (!strncmp(tok, "graph", strlen(arg)))
607
		callchain_param.mode = CHAIN_GRAPH_ABS;
608

609
	else if (!strncmp(tok, "flat", strlen(arg)))
610 611 612 613 614
		callchain_param.mode = CHAIN_FLAT;

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

615 616
	else if (!strncmp(tok, "none", strlen(arg))) {
		callchain_param.mode = CHAIN_NONE;
617
		symbol_conf.use_callchain = false;
618 619 620 621

		return 0;
	}

622 623 624
	else
		return -1;

625 626 627
	/* get the min percentage */
	tok = strtok(NULL, ",");
	if (!tok)
628
		goto setup;
629

630
	callchain_param.min_percent = strtod(tok, &endptr);
631 632 633
	if (tok == endptr)
		return -1;

634 635 636 637 638 639
	/* get the print limit */
	tok2 = strtok(NULL, ",");
	if (!tok2)
		goto setup;

	if (tok2[0] != 'c') {
640
		callchain_param.print_limit = strtoul(tok2, &endptr, 0);
641 642 643 644 645 646
		tok2 = strtok(NULL, ",");
		if (!tok2)
			goto setup;
	}

	/* get the call chain order */
647
	if (!strncmp(tok2, "caller", strlen("caller")))
648
		callchain_param.order = ORDER_CALLER;
649
	else if (!strncmp(tok2, "callee", strlen("callee")))
650 651 652
		callchain_param.order = ORDER_CALLEE;
	else
		return -1;
653 654 655 656 657 658 659 660 661 662 663

	/* 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;
664
setup:
665
	if (callchain_register_param(&callchain_param) < 0) {
666 667 668
		fprintf(stderr, "Can't register callchain params\n");
		return -1;
	}
669 670 671
	return 0;
}

672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
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;
}

690
static int
691 692
parse_branch_mode(const struct option *opt __maybe_unused,
		  const char *str __maybe_unused, int unset)
693
{
694 695 696
	int *branch_mode = opt->value;

	*branch_mode = !unset;
697 698 699
	return 0;
}

700 701 702 703 704 705 706 707 708 709
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;
}

710
int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
711
{
712
	struct perf_session *session;
713
	struct stat st;
714
	bool has_br_stack = false;
715
	int branch_mode = -1;
716
	int ret = -1;
717 718
	char callchain_default_opt[] = "fractal,0.5,callee";
	const char * const report_usage[] = {
N
Namhyung Kim 已提交
719
		"perf report [<options>]",
720 721 722
		NULL
	};
	struct perf_report report = {
723
		.tool = {
724 725
			.sample		 = process_sample_event,
			.mmap		 = perf_event__process_mmap,
726
			.mmap2		 = perf_event__process_mmap2,
727
			.comm		 = perf_event__process_comm,
728 729
			.exit		 = perf_event__process_exit,
			.fork		 = perf_event__process_fork,
730 731 732 733 734 735 736 737
			.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,
		},
738
		.max_stack		 = PERF_MAX_STACK_DEPTH,
739 740 741
		.pretty_printing_style	 = "normal",
	};
	const struct option options[] = {
742
	OPT_STRING('i', "input", &input_name, "file",
743
		    "input file name"),
744
	OPT_INCR('v', "verbose", &verbose,
745
		    "be more verbose (show symbol address, etc)"),
746 747
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
		    "dump raw trace in ASCII"),
748 749
	OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
		   "file", "vmlinux pathname"),
750 751
	OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name,
		   "file", "kallsyms pathname"),
752
	OPT_BOOLEAN('f', "force", &report.force, "don't complain, do it"),
753
	OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules,
754
		    "load module symbols - WARNING: use only with -k and LIVE kernel"),
755
	OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples,
756
		    "Show a column with the number of samples"),
757
	OPT_BOOLEAN('T', "threads", &report.show_threads,
758
		    "Show per-thread event counters"),
759
	OPT_STRING(0, "pretty", &report.pretty_printing_style, "key",
760
		   "pretty printing style key: normal raw"),
761
	OPT_BOOLEAN(0, "tui", &report.use_tui, "Use the TUI interface"),
762
	OPT_BOOLEAN(0, "gtk", &report.use_gtk, "Use the GTK2 interface"),
763 764
	OPT_BOOLEAN(0, "stdio", &report.use_stdio,
		    "Use the stdio interface"),
765 766 767
	OPT_BOOLEAN(0, "header", &report.header, "Show data header."),
	OPT_BOOLEAN(0, "header-only", &report.header_only,
		    "Show only data header."),
768
	OPT_STRING('s', "sort", &sort_order, "key[,key2...]",
769
		   "sort by key(s): pid, comm, dso, symbol, parent, cpu, srcline,"
770
		   " dso_to, dso_from, symbol_to, symbol_from, mispredict,"
771
		   " weight, local_weight, mem, symbol_daddr, dso_daddr, tlb, "
772
		   "snoop, locked, abort, in_tx, transaction"),
773 774
	OPT_BOOLEAN(0, "showcpuutilization", &symbol_conf.show_cpu_utilization,
		    "Show sample percentage for different cpu modes"),
775 776
	OPT_STRING('p', "parent", &parent_pattern, "regex",
		   "regex filter to identify parent, see: '--sort parent'"),
777
	OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other,
778
		    "Only display entries with parent-match"),
779
	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order",
780 781
		     "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),
782 783 784 785
	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)),
786 787
	OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
		    "alias for inverted call graph"),
788 789 790
	OPT_CALLBACK(0, "ignore-callees", NULL, "regex",
		   "ignore callees of these functions in call graphs",
		   report_parse_ignore_callees_opt),
791
	OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
792
		   "only consider symbols in these dsos"),
793
	OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
794
		   "only consider symbols in these comms"),
795
	OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]",
796
		   "only consider these symbols"),
797 798
	OPT_STRING(0, "symbol-filter", &report.symbol_filter_str, "filter",
		   "only show symbols that (partially) match with this filter"),
799
	OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str,
800 801
		   "width[,width...]",
		   "don't try to adjust column width, use these fixed values"),
802
	OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator",
803 804
		   "separator for columns, no spaces will be added between "
		   "columns '.' is reserved."),
805
	OPT_BOOLEAN('U', "hide-unresolved", &report.hide_unresolved,
806
		    "Only display entries resolved to a symbol"),
807 808
	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
		    "Look for files with symbols relative to this directory"),
809
	OPT_STRING('C', "cpu", &report.cpu_list, "cpu",
810 811
		   "list of cpus to profile"),
	OPT_BOOLEAN('I', "show-info", &report.show_full_info,
812
		    "Display extended information about perf.data file"),
813 814 815 816
	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)"),
817 818
	OPT_STRING('M', "disassembler-style", &disassembler_style, "disassembler style",
		   "Specify disassembler style (e.g. -M intel for intel syntax)"),
819 820
	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
821 822
	OPT_BOOLEAN(0, "group", &symbol_conf.event_group,
		    "Show event group information together"),
823
	OPT_CALLBACK_NOOPT('b', "branch-stack", &branch_mode, "",
824
		    "use branch records for histogram filling", parse_branch_mode),
825 826
	OPT_STRING(0, "objdump", &objdump_path, "path",
		   "objdump binary to use for disassembly and annotations"),
827 828
	OPT_BOOLEAN(0, "demangle", &symbol_conf.demangle,
		    "Disable symbol demangling"),
829
	OPT_BOOLEAN(0, "mem-mode", &report.mem_mode, "mem access profile"),
830 831
	OPT_CALLBACK(0, "percent-limit", &report, "percent",
		     "Don't show entries under that percent", parse_percent_limit),
832
	OPT_END()
833
	};
834 835 836
	struct perf_data_file file = {
		.mode  = PERF_DATA_MODE_READ,
	};
837

838
	perf_config(perf_report_config, &report);
839

840 841
	argc = parse_options(argc, argv, options, report_usage, 0);

842
	if (report.use_stdio)
843
		use_browser = 0;
844
	else if (report.use_tui)
845
		use_browser = 1;
846 847
	else if (report.use_gtk)
		use_browser = 2;
848

849
	if (report.inverted_callchain)
850 851
		callchain_param.order = ORDER_CALLER;

852
	if (!input_name || !strlen(input_name)) {
853
		if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode))
854
			input_name = "-";
855
		else
856
			input_name = "perf.data";
857
	}
858

859 860 861
	file.path  = input_name;
	file.force = report.force;

862
repeat:
863
	session = perf_session__new(&file, false, &report.tool);
864 865 866 867 868 869 870
	if (session == NULL)
		return -ENOMEM;

	report.session = session;

	has_br_stack = perf_header__has_feat(&session->header,
					     HEADER_BRANCH_STACK);
871

872 873
	if (branch_mode == -1 && has_br_stack)
		sort__mode = SORT_MODE__BRANCH;
874

875 876
	/* sort__mode could be NORMAL if --no-branch-stack */
	if (sort__mode == SORT_MODE__BRANCH) {
877
		/*
878 879
		 * if no sort_order is provided, then specify
		 * branch-mode specific order
880 881 882 883 884
		 */
		if (sort_order == default_sort_order)
			sort_order = "comm,dso_from,symbol_from,"
				     "dso_to,symbol_to";

885
	}
886
	if (report.mem_mode) {
887
		if (sort__mode == SORT_MODE__BRANCH) {
888 889 890
			fprintf(stderr, "branch and mem mode incompatible\n");
			goto error;
		}
891 892
		sort__mode = SORT_MODE__MEMORY;

893 894 895 896 897 898 899
		/*
		 * 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";
	}
900

901 902 903 904
	if (setup_sorting() < 0) {
		parse_options_usage(report_usage, options, "s", 1);
		goto error;
	}
905

906 907 908 909 910
	if (parent_pattern != default_parent_pattern) {
		if (sort_dimension__add("parent") < 0)
			goto error;
	}

911 912 913 914
	/* Force tty output for header output. */
	if (report.header || report.header_only)
		use_browser = 0;

915 916 917 918 919 920 921
	if (strcmp(input_name, "-") != 0)
		setup_browser(true);
	else {
		use_browser = 0;
		perf_hpp__init();
	}

922 923 924 925 926 927 928 929 930 931
	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);
	}

932
	/*
933
	 * Only in the TUI browser we are doing integrated annotation,
934 935 936
	 * so don't allocate extra space that won't be used in the stdio
	 * implementation.
	 */
937
	if (use_browser == 1 && sort__has_sym) {
938
		symbol_conf.priv_size = sizeof(struct annotation);
939 940
		machines__set_symbol_filter(&session->machines,
					    symbol__annotate_init);
941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
		/*
 		 * 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;
		}
	}
957

958
	if (symbol__init() < 0)
959
		goto error;
960

961 962 963 964 965 966 967 968 969 970
	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];
	}
971

972
	sort__setup_elide(stdout);
973

974
	ret = __cmd_report(&report);
975 976 977 978 979 980
	if (ret == K_SWITCH_INPUT_DATA) {
		perf_session__delete(session);
		goto repeat;
	} else
		ret = 0;

981 982 983
error:
	perf_session__delete(session);
	return ret;
984
}