builtin-lock.c 22.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
#include "builtin.h"
#include "perf.h"

#include "util/util.h"
#include "util/cache.h"
#include "util/symbol.h"
#include "util/thread.h"
#include "util/header.h"

#include "util/parse-options.h"
#include "util/trace-event.h"

#include "util/debug.h"
#include "util/session.h"

#include <sys/types.h>
#include <sys/prctl.h>
#include <semaphore.h>
#include <pthread.h>
#include <math.h>
#include <limits.h>

#include <linux/list.h>
#include <linux/hash.h>

26 27
static struct perf_session *session;

28 29 30 31 32 33 34 35 36 37
/* based on kernel/lockdep.c */
#define LOCKHASH_BITS		12
#define LOCKHASH_SIZE		(1UL << LOCKHASH_BITS)

static struct list_head lockhash_table[LOCKHASH_SIZE];

#define __lockhashfn(key)	hash_long((unsigned long)key, LOCKHASH_BITS)
#define lockhashentry(key)	(lockhash_table + __lockhashfn((key)))

struct lock_stat {
38 39
	struct list_head	hash_entry;
	struct rb_node		rb;		/* used for sorting */
40

41 42
	/*
	 * FIXME: raw_field_value() returns unsigned long long,
43
	 * so address of lockdep_map should be dealed as 64bit.
44 45 46 47
	 * Is there more better solution?
	 */
	void			*addr;		/* address of lockdep_map, used as ID */
	char			*name;		/* for strcpy(), we cannot use const */
48

49
	unsigned int		nr_acquire;
50
	unsigned int		nr_acquired;
51 52
	unsigned int		nr_contended;
	unsigned int		nr_release;
53

54 55
	unsigned int		nr_readlock;
	unsigned int		nr_trylock;
56
	/* these times are in nano sec. */
57 58 59
	u64			wait_time_total;
	u64			wait_time_min;
	u64			wait_time_max;
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

	int			discard; /* flag of blacklist */
};

/*
 * States of lock_seq_stat
 *
 * UNINITIALIZED is required for detecting first event of acquire.
 * As the nature of lock events, there is no guarantee
 * that the first event for the locks are acquire,
 * it can be acquired, contended or release.
 */
#define SEQ_STATE_UNINITIALIZED      0	       /* initial state */
#define SEQ_STATE_RELEASED	1
#define SEQ_STATE_ACQUIRING	2
#define SEQ_STATE_ACQUIRED	3
#define SEQ_STATE_READ_ACQUIRED	4
#define SEQ_STATE_CONTENDED	5

/*
 * MAX_LOCK_DEPTH
 * Imported from include/linux/sched.h.
 * Should this be synchronized?
 */
#define MAX_LOCK_DEPTH 48

/*
 * struct lock_seq_stat:
 * Place to put on state of one lock sequence
 * 1) acquire -> acquired -> release
 * 2) acquire -> contended -> acquired -> release
 * 3) acquire (with read or try) -> release
 * 4) Are there other patterns?
 */
struct lock_seq_stat {
	struct list_head        list;
	int			state;
	u64			prev_event_time;
	void                    *addr;

	int                     read_count;
101 102
};

103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
struct thread_stat {
	struct rb_node		rb;

	u32                     tid;
	struct list_head        seq_list;
};

static struct rb_root		thread_stats;

static struct thread_stat *thread_stat_find(u32 tid)
{
	struct rb_node *node;
	struct thread_stat *st;

	node = thread_stats.rb_node;
	while (node) {
		st = container_of(node, struct thread_stat, rb);
		if (st->tid == tid)
			return st;
		else if (tid < st->tid)
			node = node->rb_left;
		else
			node = node->rb_right;
	}

	return NULL;
}

static void thread_stat_insert(struct thread_stat *new)
{
	struct rb_node **rb = &thread_stats.rb_node;
	struct rb_node *parent = NULL;
	struct thread_stat *p;

	while (*rb) {
		p = container_of(*rb, struct thread_stat, rb);
		parent = *rb;

		if (new->tid < p->tid)
			rb = &(*rb)->rb_left;
		else if (new->tid > p->tid)
			rb = &(*rb)->rb_right;
		else
			BUG_ON("inserting invalid thread_stat\n");
	}

	rb_link_node(&new->rb, parent, rb);
	rb_insert_color(&new->rb, &thread_stats);
}

static struct thread_stat *thread_stat_findnew_after_first(u32 tid)
{
	struct thread_stat *st;

	st = thread_stat_find(tid);
	if (st)
		return st;

	st = zalloc(sizeof(struct thread_stat));
	if (!st)
		die("memory allocation failed\n");

	st->tid = tid;
	INIT_LIST_HEAD(&st->seq_list);

	thread_stat_insert(st);

	return st;
}

static struct thread_stat *thread_stat_findnew_first(u32 tid);
static struct thread_stat *(*thread_stat_findnew)(u32 tid) =
	thread_stat_findnew_first;

static struct thread_stat *thread_stat_findnew_first(u32 tid)
{
	struct thread_stat *st;

	st = zalloc(sizeof(struct thread_stat));
	if (!st)
		die("memory allocation failed\n");
	st->tid = tid;
	INIT_LIST_HEAD(&st->seq_list);

	rb_link_node(&st->rb, NULL, &thread_stats.rb_node);
	rb_insert_color(&st->rb, &thread_stats);

	thread_stat_findnew = thread_stat_findnew_after_first;
	return st;
}

194
/* build simple key function one is bigger than two */
195
#define SINGLE_KEY(member)						\
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
	static int lock_stat_key_ ## member(struct lock_stat *one,	\
					 struct lock_stat *two)		\
	{								\
		return one->member > two->member;			\
	}

SINGLE_KEY(nr_acquired)
SINGLE_KEY(nr_contended)
SINGLE_KEY(wait_time_total)
SINGLE_KEY(wait_time_min)
SINGLE_KEY(wait_time_max)

struct lock_key {
	/*
	 * name: the value for specify by user
	 * this should be simpler than raw name of member
	 * e.g. nr_acquired -> acquired, wait_time_total -> wait_total
	 */
214 215
	const char		*name;
	int			(*key)(struct lock_stat*, struct lock_stat*);
216 217
};

218 219 220 221 222
static const char		*sort_key = "acquired";

static int			(*compare)(struct lock_stat *, struct lock_stat *);

static struct rb_root		result;	/* place to store sorted data */
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252

#define DEF_KEY_LOCK(name, fn_suffix)	\
	{ #name, lock_stat_key_ ## fn_suffix }
struct lock_key keys[] = {
	DEF_KEY_LOCK(acquired, nr_acquired),
	DEF_KEY_LOCK(contended, nr_contended),
	DEF_KEY_LOCK(wait_total, wait_time_total),
	DEF_KEY_LOCK(wait_min, wait_time_min),
	DEF_KEY_LOCK(wait_max, wait_time_max),

	/* extra comparisons much complicated should be here */

	{ NULL, NULL }
};

static void select_key(void)
{
	int i;

	for (i = 0; keys[i].name; i++) {
		if (!strcmp(keys[i].name, sort_key)) {
			compare = keys[i].key;
			return;
		}
	}

	die("Unknown compare key:%s\n", sort_key);
}

static void insert_to_result(struct lock_stat *st,
253
			     int (*bigger)(struct lock_stat *, struct lock_stat *))
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
{
	struct rb_node **rb = &result.rb_node;
	struct rb_node *parent = NULL;
	struct lock_stat *p;

	while (*rb) {
		p = container_of(*rb, struct lock_stat, rb);
		parent = *rb;

		if (bigger(st, p))
			rb = &(*rb)->rb_left;
		else
			rb = &(*rb)->rb_right;
	}

	rb_link_node(&st->rb, parent, rb);
	rb_insert_color(&st->rb, &result);
}

/* returns left most element of result, and erase it */
static struct lock_stat *pop_from_result(void)
{
	struct rb_node *node = result.rb_node;

	if (!node)
		return NULL;

	while (node->rb_left)
		node = node->rb_left;

	rb_erase(node, &result);
	return container_of(node, struct lock_stat, rb);
}

288
static struct lock_stat *lock_stat_findnew(void *addr, const char *name)
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319
{
	struct list_head *entry = lockhashentry(addr);
	struct lock_stat *ret, *new;

	list_for_each_entry(ret, entry, hash_entry) {
		if (ret->addr == addr)
			return ret;
	}

	new = zalloc(sizeof(struct lock_stat));
	if (!new)
		goto alloc_failed;

	new->addr = addr;
	new->name = zalloc(sizeof(char) * strlen(name) + 1);
	if (!new->name)
		goto alloc_failed;
	strcpy(new->name, name);

	new->wait_time_min = ULLONG_MAX;

	list_add(&new->hash_entry, entry);
	return new;

alloc_failed:
	die("memory allocation failed\n");
}

static char			const *input_name = "perf.data";

struct raw_event_sample {
320 321
	u32			size;
	char			data[0];
322 323 324
};

struct trace_acquire_event {
325 326
	void			*addr;
	const char		*name;
327
	int			flag;
328 329 330
};

struct trace_acquired_event {
331 332
	void			*addr;
	const char		*name;
333 334 335
};

struct trace_contended_event {
336 337
	void			*addr;
	const char		*name;
338 339 340
};

struct trace_release_event {
341 342
	void			*addr;
	const char		*name;
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370
};

struct trace_lock_handler {
	void (*acquire_event)(struct trace_acquire_event *,
			      struct event *,
			      int cpu,
			      u64 timestamp,
			      struct thread *thread);

	void (*acquired_event)(struct trace_acquired_event *,
			       struct event *,
			       int cpu,
			       u64 timestamp,
			       struct thread *thread);

	void (*contended_event)(struct trace_contended_event *,
				struct event *,
				int cpu,
				u64 timestamp,
				struct thread *thread);

	void (*release_event)(struct trace_release_event *,
			      struct event *,
			      int cpu,
			      u64 timestamp,
			      struct thread *thread);
};

371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
static struct lock_seq_stat *get_seq(struct thread_stat *ts, void *addr)
{
	struct lock_seq_stat *seq;

	list_for_each_entry(seq, &ts->seq_list, list) {
		if (seq->addr == addr)
			return seq;
	}

	seq = zalloc(sizeof(struct lock_seq_stat));
	if (!seq)
		die("Not enough memory\n");
	seq->state = SEQ_STATE_UNINITIALIZED;
	seq->addr = addr;

	list_add(&seq->list, &ts->seq_list);
	return seq;
}

static int bad_hist[4];

392 393
static void
report_lock_acquire_event(struct trace_acquire_event *acquire_event,
394 395
			struct event *__event __used,
			int cpu __used,
396
			u64 timestamp __used,
397 398
			struct thread *thread __used)
{
399 400 401 402 403 404 405
	struct lock_stat *ls;
	struct thread_stat *ts;
	struct lock_seq_stat *seq;

	ls = lock_stat_findnew(acquire_event->addr, acquire_event->name);
	if (ls->discard)
		return;
406

407 408
	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, acquire_event->addr);
409

410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
	case SEQ_STATE_RELEASED:
		if (!acquire_event->flag) {
			seq->state = SEQ_STATE_ACQUIRING;
		} else {
			if (acquire_event->flag & 1)
				ls->nr_trylock++;
			if (acquire_event->flag & 2)
				ls->nr_readlock++;
			seq->state = SEQ_STATE_READ_ACQUIRED;
			seq->read_count = 1;
			ls->nr_acquired++;
		}
		break;
	case SEQ_STATE_READ_ACQUIRED:
		if (acquire_event->flag & 2) {
			seq->read_count++;
			ls->nr_acquired++;
			goto end;
		} else {
			goto broken;
		}
433
		break;
434 435 436 437 438 439 440 441 442 443
	case SEQ_STATE_ACQUIRED:
	case SEQ_STATE_ACQUIRING:
	case SEQ_STATE_CONTENDED:
broken:
		/* broken lock sequence, discard it */
		ls->discard = 1;
		bad_hist[0]++;
		list_del(&seq->list);
		free(seq);
		goto end;
444 445
		break;
	default:
446
		BUG_ON("Unknown state of lock sequence found!\n");
447 448 449
		break;
	}

450 451 452 453
	ls->nr_acquire++;
	seq->prev_event_time = timestamp;
end:
	return;
454 455
}

456 457
static void
report_lock_acquired_event(struct trace_acquired_event *acquired_event,
458 459
			 struct event *__event __used,
			 int cpu __used,
460
			 u64 timestamp __used,
461 462
			 struct thread *thread __used)
{
463 464 465 466
	struct lock_stat *ls;
	struct thread_stat *ts;
	struct lock_seq_stat *seq;
	u64 contended_term;
467

468 469 470 471 472 473
	ls = lock_stat_findnew(acquired_event->addr, acquired_event->name);
	if (ls->discard)
		return;

	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, acquired_event->addr);
474

475 476 477 478 479
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
		/* orphan event, do nothing */
		return;
	case SEQ_STATE_ACQUIRING:
480
		break;
481 482 483 484 485 486 487 488
	case SEQ_STATE_CONTENDED:
		contended_term = timestamp - seq->prev_event_time;
		ls->wait_time_total += contended_term;

		if (contended_term < ls->wait_time_min)
			ls->wait_time_min = contended_term;
		else if (ls->wait_time_max < contended_term)
			ls->wait_time_max = contended_term;
489
		break;
490 491 492 493 494 495 496 497 498 499 500
	case SEQ_STATE_RELEASED:
	case SEQ_STATE_ACQUIRED:
	case SEQ_STATE_READ_ACQUIRED:
		/* broken lock sequence, discard it */
		ls->discard = 1;
		bad_hist[1]++;
		list_del(&seq->list);
		free(seq);
		goto end;
		break;

501
	default:
502
		BUG_ON("Unknown state of lock sequence found!\n");
503 504 505
		break;
	}

506 507 508 509 510
	seq->state = SEQ_STATE_ACQUIRED;
	ls->nr_acquired++;
	seq->prev_event_time = timestamp;
end:
	return;
511 512
}

513 514
static void
report_lock_contended_event(struct trace_contended_event *contended_event,
515 516
			  struct event *__event __used,
			  int cpu __used,
517
			  u64 timestamp __used,
518 519
			  struct thread *thread __used)
{
520 521 522 523 524 525 526
	struct lock_stat *ls;
	struct thread_stat *ts;
	struct lock_seq_stat *seq;

	ls = lock_stat_findnew(contended_event->addr, contended_event->name);
	if (ls->discard)
		return;
527

528 529
	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, contended_event->addr);
530

531 532 533 534 535
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
		/* orphan event, do nothing */
		return;
	case SEQ_STATE_ACQUIRING:
536
		break;
537 538 539 540 541 542 543 544 545 546
	case SEQ_STATE_RELEASED:
	case SEQ_STATE_ACQUIRED:
	case SEQ_STATE_READ_ACQUIRED:
	case SEQ_STATE_CONTENDED:
		/* broken lock sequence, discard it */
		ls->discard = 1;
		bad_hist[2]++;
		list_del(&seq->list);
		free(seq);
		goto end;
547 548
		break;
	default:
549
		BUG_ON("Unknown state of lock sequence found!\n");
550 551 552
		break;
	}

553 554 555 556 557
	seq->state = SEQ_STATE_CONTENDED;
	ls->nr_contended++;
	seq->prev_event_time = timestamp;
end:
	return;
558 559
}

560 561
static void
report_lock_release_event(struct trace_release_event *release_event,
562 563
			struct event *__event __used,
			int cpu __used,
564
			u64 timestamp __used,
565 566
			struct thread *thread __used)
{
567 568 569
	struct lock_stat *ls;
	struct thread_stat *ts;
	struct lock_seq_stat *seq;
570

571 572 573
	ls = lock_stat_findnew(release_event->addr, release_event->name);
	if (ls->discard)
		return;
574

575 576
	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, release_event->addr);
577

578 579 580 581 582 583 584 585 586 587 588
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
		goto end;
		break;
	case SEQ_STATE_ACQUIRED:
		break;
	case SEQ_STATE_READ_ACQUIRED:
		seq->read_count--;
		BUG_ON(seq->read_count < 0);
		if (!seq->read_count) {
			ls->nr_release++;
589 590
			goto end;
		}
591 592 593 594 595 596 597 598
		break;
	case SEQ_STATE_ACQUIRING:
	case SEQ_STATE_CONTENDED:
	case SEQ_STATE_RELEASED:
		/* broken lock sequence, discard it */
		ls->discard = 1;
		bad_hist[3]++;
		goto free_seq;
599 600
		break;
	default:
601
		BUG_ON("Unknown state of lock sequence found!\n");
602 603 604
		break;
	}

605 606 607 608
	ls->nr_release++;
free_seq:
	list_del(&seq->list);
	free(seq);
609
end:
610
	return;
611 612 613 614
}

/* lock oriented handlers */
/* TODO: handlers for CPU oriented, thread oriented */
615 616 617 618 619
static struct trace_lock_handler report_lock_ops  = {
	.acquire_event		= report_lock_acquire_event,
	.acquired_event		= report_lock_acquired_event,
	.contended_event	= report_lock_contended_event,
	.release_event		= report_lock_release_event,
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
};

static struct trace_lock_handler *trace_handler;

static void
process_lock_acquire_event(void *data,
			   struct event *event __used,
			   int cpu __used,
			   u64 timestamp __used,
			   struct thread *thread __used)
{
	struct trace_acquire_event acquire_event;
	u64 tmp;		/* this is required for casting... */

	tmp = raw_field_value(event, "lockdep_addr", data);
	memcpy(&acquire_event.addr, &tmp, sizeof(void *));
	acquire_event.name = (char *)raw_field_ptr(event, "name", data);
637
	acquire_event.flag = (int)raw_field_value(event, "flag", data);
638

639 640
	if (trace_handler->acquire_event)
		trace_handler->acquire_event(&acquire_event, event, cpu, timestamp, thread);
641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656
}

static void
process_lock_acquired_event(void *data,
			    struct event *event __used,
			    int cpu __used,
			    u64 timestamp __used,
			    struct thread *thread __used)
{
	struct trace_acquired_event acquired_event;
	u64 tmp;		/* this is required for casting... */

	tmp = raw_field_value(event, "lockdep_addr", data);
	memcpy(&acquired_event.addr, &tmp, sizeof(void *));
	acquired_event.name = (char *)raw_field_ptr(event, "name", data);

657 658
	if (trace_handler->acquire_event)
		trace_handler->acquired_event(&acquired_event, event, cpu, timestamp, thread);
659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674
}

static void
process_lock_contended_event(void *data,
			     struct event *event __used,
			     int cpu __used,
			     u64 timestamp __used,
			     struct thread *thread __used)
{
	struct trace_contended_event contended_event;
	u64 tmp;		/* this is required for casting... */

	tmp = raw_field_value(event, "lockdep_addr", data);
	memcpy(&contended_event.addr, &tmp, sizeof(void *));
	contended_event.name = (char *)raw_field_ptr(event, "name", data);

675 676
	if (trace_handler->acquire_event)
		trace_handler->contended_event(&contended_event, event, cpu, timestamp, thread);
677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692
}

static void
process_lock_release_event(void *data,
			   struct event *event __used,
			   int cpu __used,
			   u64 timestamp __used,
			   struct thread *thread __used)
{
	struct trace_release_event release_event;
	u64 tmp;		/* this is required for casting... */

	tmp = raw_field_value(event, "lockdep_addr", data);
	memcpy(&release_event.addr, &tmp, sizeof(void *));
	release_event.name = (char *)raw_field_ptr(event, "name", data);

693 694
	if (trace_handler->acquire_event)
		trace_handler->release_event(&release_event, event, cpu, timestamp, thread);
695 696 697
}

static void
698
process_raw_event(void *data, int cpu, u64 timestamp, struct thread *thread)
699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720
{
	struct event *event;
	int type;

	type = trace_parse_common_type(data);
	event = trace_find_event(type);

	if (!strcmp(event->name, "lock_acquire"))
		process_lock_acquire_event(data, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "lock_acquired"))
		process_lock_acquired_event(data, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "lock_contended"))
		process_lock_contended_event(data, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "lock_release"))
		process_lock_release_event(data, event, cpu, timestamp, thread);
}

/* TODO: various way to print, coloring, nano or milli sec */
static void print_result(void)
{
	struct lock_stat *st;
	char cut_name[20];
721
	int bad, total;
722

723 724 725
	pr_info("%20s ", "Name");
	pr_info("%10s ", "acquired");
	pr_info("%10s ", "contended");
726

727 728 729
	pr_info("%15s ", "total wait (ns)");
	pr_info("%15s ", "max wait (ns)");
	pr_info("%15s ", "min wait (ns)");
730

731
	pr_info("\n\n");
732

733
	bad = total = 0;
734
	while ((st = pop_from_result())) {
735 736 737 738 739
		total++;
		if (st->discard) {
			bad++;
			continue;
		}
740 741 742 743
		bzero(cut_name, 20);

		if (strlen(st->name) < 16) {
			/* output raw name */
744
			pr_info("%20s ", st->name);
745 746 747 748 749 750 751
		} else {
			strncpy(cut_name, st->name, 16);
			cut_name[16] = '.';
			cut_name[17] = '.';
			cut_name[18] = '.';
			cut_name[19] = '\0';
			/* cut off name for saving output style */
752
			pr_info("%20s ", cut_name);
753 754
		}

755 756
		pr_info("%10u ", st->nr_acquired);
		pr_info("%10u ", st->nr_contended);
757

758 759 760
		pr_info("%15llu ", st->wait_time_total);
		pr_info("%15llu ", st->wait_time_max);
		pr_info("%15llu ", st->wait_time_min == ULLONG_MAX ?
761
		       0 : st->wait_time_min);
762
		pr_info("\n");
763
	}
764 765 766 767 768 769 770

	{
		/* Output for debug, this have to be removed */
		int i;
		const char *name[4] =
			{ "acquire", "acquired", "contended", "release" };

771 772 773
		pr_debug("\n=== output for debug===\n\n");
		pr_debug("bad:%d, total:%d\n", bad, total);
		pr_debug("bad rate:%f\n", (double)(bad / total));
774

775
		pr_debug("histogram of events caused bad sequence\n");
776
		for (i = 0; i < 4; i++)
777
			pr_debug(" %10s: %d\n", name[i], bad_hist[i]);
778
	}
779 780
}

781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800
static int			info_threads;
static int			info_map;

static void dump_threads(void)
{
	struct thread_stat *st;
	struct rb_node *node;
	struct thread *t;

	pr_info("%10s: comm\n", "Thread ID");

	node = rb_first(&thread_stats);
	while (node) {
		st = container_of(node, struct thread_stat, rb);
		t = perf_session__findnew(session, st->tid);
		pr_info("%10d: %s\n", st->tid, t->comm);
		node = rb_next(node);
	};
}

801 802 803 804 805
static void dump_map(void)
{
	unsigned int i;
	struct lock_stat *st;

806
	pr_info("Address of instance: name of class\n");
807 808
	for (i = 0; i < LOCKHASH_SIZE; i++) {
		list_for_each_entry(st, &lockhash_table[i], hash_entry) {
809
			pr_info(" %p: %s\n", st->addr, st->name);
810 811 812 813
		}
	}
}

814 815 816 817 818 819 820 821 822 823
static void dump_info(void)
{
	if (info_threads)
		dump_threads();
	else if (info_map)
		dump_map();
	else
		die("Unknown type of information\n");
}

824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843
static int process_sample_event(event_t *self, struct perf_session *s)
{
	struct sample_data data;
	struct thread *thread;

	bzero(&data, sizeof(data));
	event__parse_sample(self, s->sample_type, &data);

	thread = perf_session__findnew(s, data.tid);
	if (thread == NULL) {
		pr_debug("problem processing %d event, skipping it.\n",
			self->header.type);
		return -1;
	}

	process_raw_event(data.raw_data, data.cpu, data.time, thread);

	return 0;
}

844
static struct perf_event_ops eops = {
845 846
	.sample			= process_sample_event,
	.comm			= event__process_comm,
847
	.ordered_samples	= true,
848 849 850 851
};

static int read_events(void)
{
T
Tom Zanussi 已提交
852
	session = perf_session__new(input_name, O_RDONLY, 0, false);
853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
	if (!session)
		die("Initializing perf session failed\n");

	return perf_session__process_events(session, &eops);
}

static void sort_result(void)
{
	unsigned int i;
	struct lock_stat *st;

	for (i = 0; i < LOCKHASH_SIZE; i++) {
		list_for_each_entry(st, &lockhash_table[i], hash_entry) {
			insert_to_result(st, compare);
		}
	}
}

871
static void __cmd_report(void)
872 873 874 875 876 877 878 879
{
	setup_pager();
	select_key();
	read_events();
	sort_result();
	print_result();
}

880 881
static const char * const report_usage[] = {
	"perf lock report [<options>]",
882 883 884
	NULL
};

885
static const struct option report_options[] = {
886 887 888 889 890 891
	OPT_STRING('k', "key", &sort_key, "acquired",
		    "key for sorting"),
	/* TODO: type */
	OPT_END()
};

892 893 894 895 896 897 898 899 900 901 902 903 904
static const char * const info_usage[] = {
	"perf lock info [<options>]",
	NULL
};

static const struct option info_options[] = {
	OPT_BOOLEAN('t', "threads", &info_threads,
		    "dump thread list in perf.data"),
	OPT_BOOLEAN('m', "map", &info_map,
		    "map of lock instances (name:address table)"),
	OPT_END()
};

905
static const char * const lock_usage[] = {
906
	"perf lock [<options>] {record|trace|report}",
907 908 909 910
	NULL
};

static const struct option lock_options[] = {
911
	OPT_STRING('i', "input", &input_name, "file", "input file name"),
912
	OPT_INCR('v', "verbose", &verbose, "be more verbose (show symbol address, etc)"),
913
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"),
914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963
	OPT_END()
};

static const char *record_args[] = {
	"record",
	"-a",
	"-R",
	"-f",
	"-m", "1024",
	"-c", "1",
	"-e", "lock:lock_acquire:r",
	"-e", "lock:lock_acquired:r",
	"-e", "lock:lock_contended:r",
	"-e", "lock:lock_release:r",
};

static int __cmd_record(int argc, const char **argv)
{
	unsigned int rec_argc, i, j;
	const char **rec_argv;

	rec_argc = ARRAY_SIZE(record_args) + argc - 1;
	rec_argv = calloc(rec_argc + 1, sizeof(char *));

	for (i = 0; i < ARRAY_SIZE(record_args); i++)
		rec_argv[i] = strdup(record_args[i]);

	for (j = 1; j < (unsigned int)argc; j++, i++)
		rec_argv[i] = argv[j];

	BUG_ON(i != rec_argc);

	return cmd_record(i, rec_argv, NULL);
}

int cmd_lock(int argc, const char **argv, const char *prefix __used)
{
	unsigned int i;

	symbol__init();
	for (i = 0; i < LOCKHASH_SIZE; i++)
		INIT_LIST_HEAD(lockhash_table + i);

	argc = parse_options(argc, argv, lock_options, lock_usage,
			     PARSE_OPT_STOP_AT_NON_OPTION);
	if (!argc)
		usage_with_options(lock_usage, lock_options);

	if (!strncmp(argv[0], "rec", 3)) {
		return __cmd_record(argc, argv);
964 965
	} else if (!strncmp(argv[0], "report", 6)) {
		trace_handler = &report_lock_ops;
966 967
		if (argc) {
			argc = parse_options(argc, argv,
968
					     report_options, report_usage, 0);
969
			if (argc)
970
				usage_with_options(report_usage, report_options);
971
		}
972
		__cmd_report();
973 974 975
	} else if (!strcmp(argv[0], "trace")) {
		/* Aliased to 'perf trace' */
		return cmd_trace(argc, argv, prefix);
976 977 978 979 980 981 982
	} else if (!strcmp(argv[0], "info")) {
		if (argc) {
			argc = parse_options(argc, argv,
					     info_options, info_usage, 0);
			if (argc)
				usage_with_options(info_usage, info_options);
		}
983 984
		/* recycling report_lock_ops */
		trace_handler = &report_lock_ops;
985 986
		setup_pager();
		read_events();
987
		dump_info();
988 989 990 991 992 993
	} else {
		usage_with_options(lock_usage, lock_options);
	}

	return 0;
}