builtin-report.c 28.0 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
	int			max_stack;
53 54 55
	struct perf_read_values	show_threads_values;
	const char		*pretty_printing_style;
	const char		*cpu_list;
56
	const char		*symbol_filter_str;
57
	float			min_percent;
58
	DECLARE_BITMAP(cpu_bitmap, MAX_NR_CPUS);
59
};
60

61 62 63 64 65 66
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;
	}
67 68 69 70 71
	if (!strcmp(var, "report.percent-limit")) {
		struct perf_report *rep = cb;
		rep->min_percent = strtof(value, NULL);
		return 0;
	}
72 73 74 75

	return perf_default_config(var, value, cb);
}

76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
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,
94 95
						 sample, &parent, al,
						 rep->max_stack);
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
		if (err)
			return err;
	}

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

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

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

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

	/*
123
	 * In the TUI browser, we are doing integrated annotation,
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169
	 * so we don't allocate the extra space needed because the stdio
	 * code will not use it.
	 */
	if (sort__has_sym && he->ms.sym && use_browser > 0) {
		struct annotation *notes = symbol__annotation(he->ms.sym);

		assert(evsel != NULL);

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

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

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

		mx = he->mem_info;

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

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

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

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

170 171 172 173 174 175 176 177 178 179 180
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;
181
	struct branch_info *bi, *bx;
182 183 184

	if ((sort__has_parent || symbol_conf.use_callchain)
	    && sample->callchain) {
185
		err = machine__resolve_callchain(machine, evsel, al->thread,
186 187
						 sample, &parent, al,
						 rep->max_stack);
188 189 190 191 192 193 194 195 196 197 198 199
		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;
200 201 202

		err = -ENOMEM;

203 204 205 206 207
		/*
		 * The report shows the percentage of total branches captured
		 * and not events sampled. Thus we use a pseudo period of 1.
		 */
		he = __hists__add_branch_entry(&evsel->hists, al, parent,
208
				&bi[i], 1, 1);
209
		if (he) {
210 211
			struct annotation *notes;
			bx = he->branch_info;
212
			if (bx->from.sym && use_browser == 1 && sort__has_sym) {
213 214 215 216 217 218 219 220 221 222 223 224 225
				notes = symbol__annotation(bx->from.sym);
				if (!notes->src
				    && symbol__alloc_hist(bx->from.sym) < 0)
					goto out;

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

226
			if (bx->to.sym && use_browser == 1 && sort__has_sym) {
227 228 229 230 231 232 233 234 235 236 237 238
				notes = symbol__annotation(bx->to.sym);
				if (!notes->src
				    && symbol__alloc_hist(bx->to.sym) < 0)
					goto out;

				err = symbol__inc_addr_samples(bx->to.sym,
							       bx->to.map,
							       evsel->idx,
							       bx->to.al_addr);
				if (err)
					goto out;
			}
239 240 241
			evsel->hists.stats.total_period += 1;
			hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
		} else
242
			goto out;
243
	}
244
	err = 0;
245
out:
246
	free(bi);
247 248 249
	return err;
}

250 251
static int perf_evsel__add_hist_entry(struct perf_tool *tool,
				      struct perf_evsel *evsel,
252 253 254
				      struct addr_location *al,
				      struct perf_sample *sample,
				      struct machine *machine)
255
{
256
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
257
	struct symbol *parent = NULL;
258
	int err = 0;
259 260
	struct hist_entry *he;

261
	if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) {
262
		err = machine__resolve_callchain(machine, evsel, al->thread,
263 264
						 sample, &parent, al,
						 rep->max_stack);
265 266
		if (err)
			return err;
267
	}
268

269
	he = __hists__add_entry(&evsel->hists, al, parent, sample->period,
270
				sample->weight, sample->transaction);
271
	if (he == NULL)
272 273
		return -ENOMEM;

274
	if (symbol_conf.use_callchain) {
275
		err = callchain_append(he->callchain,
276
				       &callchain_cursor,
277
				       sample->period);
278
		if (err)
279
			return err;
280 281
	}
	/*
282
	 * Only in the TUI browser we are doing integrated annotation,
283 284 285
	 * so we don't allocated the extra space needed because the stdio
	 * code will not use it.
	 */
286
	if (he->ms.sym != NULL && use_browser == 1 && sort__has_sym) {
287
		struct annotation *notes = symbol__annotation(he->ms.sym);
288 289 290 291

		assert(evsel != NULL);

		err = -ENOMEM;
292
		if (notes->src == NULL && symbol__alloc_hist(he->ms.sym) < 0)
293 294 295
			goto out;

		err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr);
296
	}
297

298 299 300
	evsel->hists.stats.total_period += sample->period;
	hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE);
out:
301
	return err;
302 303
}

304

305
static int process_sample_event(struct perf_tool *tool,
306
				union perf_event *event,
307
				struct perf_sample *sample,
308
				struct perf_evsel *evsel,
309
				struct machine *machine)
310
{
311
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
312
	struct addr_location al;
313
	int ret;
314

315
	if (perf_event__preprocess_sample(event, machine, &al, sample) < 0) {
316
		fprintf(stderr, "problem processing %d event, skipping it.\n",
317 318 319
			event->header.type);
		return -1;
	}
320

321
	if (al.filtered || (rep->hide_unresolved && al.sym == NULL))
322
		return 0;
323

324
	if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap))
325 326
		return 0;

327
	if (sort__mode == SORT_MODE__BRANCH) {
328 329 330
		ret = perf_report__add_branch_hist_entry(tool, &al, sample,
							 evsel, machine);
		if (ret < 0)
331
			pr_debug("problem adding lbr entry, skipping event\n");
332
	} else if (rep->mem_mode == 1) {
333 334 335
		ret = perf_report__add_mem_hist_entry(tool, &al, sample,
						      evsel, machine, event);
		if (ret < 0)
336
			pr_debug("problem adding mem entry, skipping event\n");
337 338 339
	} else {
		if (al.map != NULL)
			al.map->dso->hit = 1;
340

341 342
		ret = perf_evsel__add_hist_entry(tool, evsel, &al, sample,
						 machine);
343
		if (ret < 0)
344
			pr_debug("problem incrementing symbol period, skipping event\n");
345
	}
346
	return ret;
347
}
I
Ingo Molnar 已提交
348

349
static int process_read_event(struct perf_tool *tool,
350
			      union perf_event *event,
351
			      struct perf_sample *sample __maybe_unused,
352
			      struct perf_evsel *evsel,
353
			      struct machine *machine __maybe_unused)
354
{
355
	struct perf_report *rep = container_of(tool, struct perf_report, tool);
356

357
	if (rep->show_threads) {
358
		const char *name = evsel ? perf_evsel__name(evsel) : "unknown";
359
		perf_read_values_add_value(&rep->show_threads_values,
360 361 362 363 364 365
					   event->read.pid, event->read.tid,
					   event->read.id,
					   name,
					   event->read.value);
	}

366
	dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid,
367
		    evsel ? perf_evsel__name(evsel) : "FAIL",
368
		    event->read.value);
369 370 371 372

	return 0;
}

373
/* For pipe mode, sample_type is not currently set */
374
static int perf_report__setup_sample_type(struct perf_report *rep)
375
{
376 377 378
	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);
379

380
	if (!is_pipe && !(sample_type & PERF_SAMPLE_CALLCHAIN)) {
381
		if (sort__has_parent) {
382
			ui__error("Selected --sort parent, but no "
383 384
				    "callchain data. Did you call "
				    "'perf record' without -g?\n");
385
			return -EINVAL;
386
		}
387
		if (symbol_conf.use_callchain) {
388
			ui__error("Selected -g but no callchain data. Did "
389
				    "you call 'perf record' without -g?\n");
390
			return -1;
391
		}
392 393
	} else if (!rep->dont_use_callchains &&
		   callchain_param.mode != CHAIN_NONE &&
394
		   !symbol_conf.use_callchain) {
395
			symbol_conf.use_callchain = true;
396
			if (callchain_register_param(&callchain_param) < 0) {
397
				ui__error("Can't register callchain params.\n");
398
				return -EINVAL;
399
			}
400 401
	}

402
	if (sort__mode == SORT_MODE__BRANCH) {
403
		if (!is_pipe &&
404
		    !(sample_type & PERF_SAMPLE_BRANCH_STACK)) {
405 406
			ui__error("Selected -b but no branch data. "
				  "Did you call perf record without -b?\n");
407 408 409 410
			return -1;
		}
	}

411 412
	return 0;
}
413

414
static void sig_handler(int sig __maybe_unused)
415 416 417 418
{
	session_done = 1;
}

419
static size_t hists__fprintf_nr_sample_events(struct perf_report *rep,
420
					      struct hists *hists,
421 422 423 424
					      const char *evname, FILE *fp)
{
	size_t ret;
	char unit;
425 426 427
	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);
428 429 430
	char buf[512];
	size_t size = sizeof(buf);

431
	if (perf_evsel__is_group_event(evsel)) {
432 433 434 435 436 437 438 439 440 441
		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;
		}
	}
442

443 444
	nr_samples = convert_unit(nr_samples, &unit);
	ret = fprintf(fp, "# Samples: %lu%c", nr_samples, unit);
445
	if (evname != NULL)
446 447
		ret += fprintf(fp, " of event '%s'", evname);

448 449 450 451 452
	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);
453 454 455
	return ret + fprintf(fp, "\n#\n");
}

456
static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist,
457
					 struct perf_report *rep,
458
					 const char *help)
459
{
460
	struct perf_evsel *pos;
461

462 463
	list_for_each_entry(pos, &evlist->entries, node) {
		struct hists *hists = &pos->hists;
464
		const char *evname = perf_evsel__name(pos);
465

466 467 468 469
		if (symbol_conf.event_group &&
		    !perf_evsel__is_group_leader(pos))
			continue;

470
		hists__fprintf_nr_sample_events(rep, hists, evname, stdout);
471
		hists__fprintf(hists, true, 0, 0, rep->min_percent, stdout);
472 473 474 475 476 477 478
		fprintf(stdout, "\n\n");
	}

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

479 480 481
		if (rep->show_threads) {
			bool style = !strcmp(rep->pretty_printing_style, "raw");
			perf_read_values_display(stdout, &rep->show_threads_values,
482
						 style);
483
			perf_read_values_destroy(&rep->show_threads_values);
484 485 486 487 488 489
		}
	}

	return 0;
}

490
static int __cmd_report(struct perf_report *rep)
491
{
492
	int ret = -EINVAL;
493
	u64 nr_samples;
494
	struct perf_session *session = rep->session;
495
	struct perf_evsel *pos;
496 497
	struct map *kernel_map;
	struct kmap *kernel_kmap;
498
	const char *help = "For a higher level overview, try: perf report --sort comm,dso";
499
	struct ui_progress prog;
500
	struct perf_data_file *file = session->file;
501

502 503
	signal(SIGINT, sig_handler);

504 505 506
	if (rep->cpu_list) {
		ret = perf_session__cpu_bitmap(session, rep->cpu_list,
					       rep->cpu_bitmap);
507
		if (ret)
508
			return ret;
509 510
	}

511
	if (use_browser <= 0)
512
		perf_session__fprintf_info(session, stdout, rep->show_full_info);
513

514 515
	if (rep->show_threads)
		perf_read_values_init(&rep->show_threads_values);
516

517
	ret = perf_report__setup_sample_type(rep);
518
	if (ret)
519
		return ret;
520

521
	ret = perf_session__process_events(session, &rep->tool);
522
	if (ret)
523
		return ret;
524

525
	kernel_map = session->machines.host.vmlinux_maps[MAP__FUNCTION];
526 527 528 529 530
	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))) {
531 532 533 534 535 536 537 538 539 540 541
		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.";
			}
		}
542

543 544 545 546
		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",
547
		desc);
548 549
	}

550
	if (verbose > 3)
551
		perf_session__fprintf(session, stdout);
552

553
	if (verbose > 2)
554
		perf_session__fprintf_dsos(session, stdout);
555

556 557
	if (dump_trace) {
		perf_session__fprintf_nr_events(session, stdout);
558
		return 0;
559 560
	}

561 562 563 564 565 566
	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...");

567 568 569
	nr_samples = 0;
	list_for_each_entry(pos, &session->evlist->entries, node) {
		struct hists *hists = &pos->hists;
570

571 572 573
		if (pos->idx == 0)
			hists->symbol_filter_str = rep->symbol_filter_str;

574
		hists__collapse_resort(hists, &prog);
575
		nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE];
576 577 578 579 580 581 582 583 584

		/* 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);
		}
585
	}
586
	ui_progress__finish();
587

588 589 590
	if (session_done())
		return 0;

591
	if (nr_samples == 0) {
592
		ui__error("The %s file has no samples!\n", file->path);
593
		return 0;
594 595
	}

596 597 598
	list_for_each_entry(pos, &session->evlist->entries, node)
		hists__output_resort(&pos->hists);

599
	if (use_browser > 0) {
600
		if (use_browser == 1) {
601
			ret = perf_evlist__tui_browse_hists(session->evlist,
602 603
							help, NULL,
							rep->min_percent,
604 605 606 607 608 609 610 611
							&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;

612
		} else if (use_browser == 2) {
613 614 615 616 617 618 619 620 621 622 623 624 625
			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);
626
		}
627
	} else
628
		perf_evlist__tty_browse_hists(session->evlist, rep, help);
629

630
	return ret;
631 632
}

633
static int
634
parse_callchain_opt(const struct option *opt, const char *arg, int unset)
635
{
636
	struct perf_report *rep = (struct perf_report *)opt->value;
637
	char *tok, *tok2;
638 639
	char *endptr;

640 641 642 643
	/*
	 * --no-call-graph
	 */
	if (unset) {
644
		rep->dont_use_callchains = true;
645 646 647
		return 0;
	}

648
	symbol_conf.use_callchain = true;
649 650 651 652

	if (!arg)
		return 0;

653 654 655 656 657 658
	tok = strtok((char *)arg, ",");
	if (!tok)
		return -1;

	/* get the output mode */
	if (!strncmp(tok, "graph", strlen(arg)))
659
		callchain_param.mode = CHAIN_GRAPH_ABS;
660

661
	else if (!strncmp(tok, "flat", strlen(arg)))
662 663 664 665 666
		callchain_param.mode = CHAIN_FLAT;

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

667 668
	else if (!strncmp(tok, "none", strlen(arg))) {
		callchain_param.mode = CHAIN_NONE;
669
		symbol_conf.use_callchain = false;
670 671 672 673

		return 0;
	}

674 675 676
	else
		return -1;

677 678 679
	/* get the min percentage */
	tok = strtok(NULL, ",");
	if (!tok)
680
		goto setup;
681

682
	callchain_param.min_percent = strtod(tok, &endptr);
683 684 685
	if (tok == endptr)
		return -1;

686 687 688 689 690 691
	/* get the print limit */
	tok2 = strtok(NULL, ",");
	if (!tok2)
		goto setup;

	if (tok2[0] != 'c') {
692
		callchain_param.print_limit = strtoul(tok2, &endptr, 0);
693 694 695 696 697 698
		tok2 = strtok(NULL, ",");
		if (!tok2)
			goto setup;
	}

	/* get the call chain order */
699
	if (!strncmp(tok2, "caller", strlen("caller")))
700
		callchain_param.order = ORDER_CALLER;
701
	else if (!strncmp(tok2, "callee", strlen("callee")))
702 703 704
		callchain_param.order = ORDER_CALLEE;
	else
		return -1;
705 706 707 708 709 710 711 712 713 714 715

	/* 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;
716
setup:
717
	if (callchain_register_param(&callchain_param) < 0) {
718 719 720
		fprintf(stderr, "Can't register callchain params\n");
		return -1;
	}
721 722 723
	return 0;
}

724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741
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;
}

742
static int
743 744
parse_branch_mode(const struct option *opt __maybe_unused,
		  const char *str __maybe_unused, int unset)
745
{
746 747 748
	int *branch_mode = opt->value;

	*branch_mode = !unset;
749 750 751
	return 0;
}

752 753 754 755 756 757 758 759 760 761
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;
}

762
int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
763
{
764
	struct perf_session *session;
765
	struct stat st;
766
	bool has_br_stack = false;
767
	int branch_mode = -1;
768
	int ret = -1;
769 770
	char callchain_default_opt[] = "fractal,0.5,callee";
	const char * const report_usage[] = {
N
Namhyung Kim 已提交
771
		"perf report [<options>]",
772 773 774
		NULL
	};
	struct perf_report report = {
775
		.tool = {
776 777
			.sample		 = process_sample_event,
			.mmap		 = perf_event__process_mmap,
778
			.mmap2		 = perf_event__process_mmap2,
779
			.comm		 = perf_event__process_comm,
780 781
			.exit		 = perf_event__process_exit,
			.fork		 = perf_event__process_fork,
782 783 784 785 786 787 788 789
			.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,
		},
790
		.max_stack		 = PERF_MAX_STACK_DEPTH,
791 792 793
		.pretty_printing_style	 = "normal",
	};
	const struct option options[] = {
794
	OPT_STRING('i', "input", &input_name, "file",
795
		    "input file name"),
796
	OPT_INCR('v', "verbose", &verbose,
797
		    "be more verbose (show symbol address, etc)"),
798 799
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
		    "dump raw trace in ASCII"),
800 801
	OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
		   "file", "vmlinux pathname"),
802 803
	OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name,
		   "file", "kallsyms pathname"),
804
	OPT_BOOLEAN('f', "force", &report.force, "don't complain, do it"),
805
	OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules,
806
		    "load module symbols - WARNING: use only with -k and LIVE kernel"),
807
	OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples,
808
		    "Show a column with the number of samples"),
809
	OPT_BOOLEAN('T', "threads", &report.show_threads,
810
		    "Show per-thread event counters"),
811
	OPT_STRING(0, "pretty", &report.pretty_printing_style, "key",
812
		   "pretty printing style key: normal raw"),
813
	OPT_BOOLEAN(0, "tui", &report.use_tui, "Use the TUI interface"),
814
	OPT_BOOLEAN(0, "gtk", &report.use_gtk, "Use the GTK2 interface"),
815 816
	OPT_BOOLEAN(0, "stdio", &report.use_stdio,
		    "Use the stdio interface"),
817
	OPT_STRING('s', "sort", &sort_order, "key[,key2...]",
818
		   "sort by key(s): pid, comm, dso, symbol, parent, cpu, srcline,"
819
		   " dso_to, dso_from, symbol_to, symbol_from, mispredict,"
820
		   " weight, local_weight, mem, symbol_daddr, dso_daddr, tlb, "
821
		   "snoop, locked, abort, in_tx, transaction"),
822 823
	OPT_BOOLEAN(0, "showcpuutilization", &symbol_conf.show_cpu_utilization,
		    "Show sample percentage for different cpu modes"),
824 825
	OPT_STRING('p', "parent", &parent_pattern, "regex",
		   "regex filter to identify parent, see: '--sort parent'"),
826
	OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other,
827
		    "Only display entries with parent-match"),
828
	OPT_CALLBACK_DEFAULT('g', "call-graph", &report, "output_type,min_percent[,print_limit],call_order",
829 830
		     "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),
831 832 833 834
	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)),
835 836
	OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
		    "alias for inverted call graph"),
837 838 839
	OPT_CALLBACK(0, "ignore-callees", NULL, "regex",
		   "ignore callees of these functions in call graphs",
		   report_parse_ignore_callees_opt),
840
	OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
841
		   "only consider symbols in these dsos"),
842
	OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
843
		   "only consider symbols in these comms"),
844
	OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]",
845
		   "only consider these symbols"),
846 847
	OPT_STRING(0, "symbol-filter", &report.symbol_filter_str, "filter",
		   "only show symbols that (partially) match with this filter"),
848
	OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str,
849 850
		   "width[,width...]",
		   "don't try to adjust column width, use these fixed values"),
851
	OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator",
852 853
		   "separator for columns, no spaces will be added between "
		   "columns '.' is reserved."),
854
	OPT_BOOLEAN('U', "hide-unresolved", &report.hide_unresolved,
855
		    "Only display entries resolved to a symbol"),
856 857
	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
		    "Look for files with symbols relative to this directory"),
858
	OPT_STRING('C', "cpu", &report.cpu_list, "cpu",
859 860
		   "list of cpus to profile"),
	OPT_BOOLEAN('I', "show-info", &report.show_full_info,
861
		    "Display extended information about perf.data file"),
862 863 864 865
	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)"),
866 867
	OPT_STRING('M', "disassembler-style", &disassembler_style, "disassembler style",
		   "Specify disassembler style (e.g. -M intel for intel syntax)"),
868 869
	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
870 871
	OPT_BOOLEAN(0, "group", &symbol_conf.event_group,
		    "Show event group information together"),
872
	OPT_CALLBACK_NOOPT('b', "branch-stack", &branch_mode, "",
873
		    "use branch records for histogram filling", parse_branch_mode),
874 875
	OPT_STRING(0, "objdump", &objdump_path, "path",
		   "objdump binary to use for disassembly and annotations"),
876 877
	OPT_BOOLEAN(0, "demangle", &symbol_conf.demangle,
		    "Disable symbol demangling"),
878
	OPT_BOOLEAN(0, "mem-mode", &report.mem_mode, "mem access profile"),
879 880
	OPT_CALLBACK(0, "percent-limit", &report, "percent",
		     "Don't show entries under that percent", parse_percent_limit),
881
	OPT_END()
882
	};
883 884 885
	struct perf_data_file file = {
		.mode  = PERF_DATA_MODE_READ,
	};
886

887
	perf_config(perf_report_config, &report);
888

889 890
	argc = parse_options(argc, argv, options, report_usage, 0);

891
	if (report.use_stdio)
892
		use_browser = 0;
893
	else if (report.use_tui)
894
		use_browser = 1;
895 896
	else if (report.use_gtk)
		use_browser = 2;
897

898
	if (report.inverted_callchain)
899 900
		callchain_param.order = ORDER_CALLER;

901
	if (!input_name || !strlen(input_name)) {
902
		if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode))
903
			input_name = "-";
904
		else
905
			input_name = "perf.data";
906
	}
907

908 909 910
	file.path  = input_name;
	file.force = report.force;

911
repeat:
912
	session = perf_session__new(&file, false, &report.tool);
913 914 915 916 917 918 919
	if (session == NULL)
		return -ENOMEM;

	report.session = session;

	has_br_stack = perf_header__has_feat(&session->header,
					     HEADER_BRANCH_STACK);
920

921 922
	if (branch_mode == -1 && has_br_stack)
		sort__mode = SORT_MODE__BRANCH;
923

924 925
	/* sort__mode could be NORMAL if --no-branch-stack */
	if (sort__mode == SORT_MODE__BRANCH) {
926
		/*
927 928
		 * if no sort_order is provided, then specify
		 * branch-mode specific order
929 930 931 932 933
		 */
		if (sort_order == default_sort_order)
			sort_order = "comm,dso_from,symbol_from,"
				     "dso_to,symbol_to";

934
	}
935
	if (report.mem_mode) {
936
		if (sort__mode == SORT_MODE__BRANCH) {
937 938 939
			fprintf(stderr, "branch and mem mode incompatible\n");
			goto error;
		}
940 941
		sort__mode = SORT_MODE__MEMORY;

942 943 944 945 946 947 948
		/*
		 * 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";
	}
949

950 951
	if (setup_sorting() < 0)
		usage_with_options(report_usage, options);
952

953 954 955 956 957 958 959 960 961 962 963 964
	if (parent_pattern != default_parent_pattern) {
		if (sort_dimension__add("parent") < 0)
			goto error;
	}

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

965
	/*
966
	 * Only in the TUI browser we are doing integrated annotation,
967 968 969
	 * so don't allocate extra space that won't be used in the stdio
	 * implementation.
	 */
970
	if (use_browser == 1 && sort__has_sym) {
971
		symbol_conf.priv_size = sizeof(struct annotation);
972 973
		machines__set_symbol_filter(&session->machines,
					    symbol__annotate_init);
974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989
		/*
 		 * 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;
		}
	}
990

991
	if (symbol__init() < 0)
992
		goto error;
993

994 995 996 997 998 999 1000 1001 1002 1003
	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];
	}
1004

1005
	sort__setup_elide(stdout);
1006

1007
	ret = __cmd_report(&report);
1008 1009 1010 1011 1012 1013
	if (ret == K_SWITCH_INPUT_DATA) {
		perf_session__delete(session);
		goto repeat;
	} else
		ret = 0;

1014 1015 1016
error:
	perf_session__delete(session);
	return ret;
1017
}