builtin-lock.c 23.0 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
	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_max)

207 208 209 210 211 212 213 214 215 216 217 218
static int lock_stat_key_wait_time_min(struct lock_stat *one,
					struct lock_stat *two)
{
	u64 s1 = one->wait_time_min;
	u64 s2 = two->wait_time_min;
	if (s1 == ULLONG_MAX)
		s1 = 0;
	if (s2 == ULLONG_MAX)
		s2 = 0;
	return s1 > s2;
}

219 220 221 222 223 224
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
	 */
225 226
	const char		*name;
	int			(*key)(struct lock_stat*, struct lock_stat*);
227 228
};

229 230 231 232 233
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 */
234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263

#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,
264
			     int (*bigger)(struct lock_stat *, struct lock_stat *))
265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298
{
	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);
}

299
static struct lock_stat *lock_stat_findnew(void *addr, const char *name)
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
{
	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 {
331 332
	u32			size;
	char			data[0];
333 334 335
};

struct trace_acquire_event {
336 337
	void			*addr;
	const char		*name;
338
	int			flag;
339 340 341
};

struct trace_acquired_event {
342 343
	void			*addr;
	const char		*name;
344 345 346
};

struct trace_contended_event {
347 348
	void			*addr;
	const char		*name;
349 350 351
};

struct trace_release_event {
352 353
	void			*addr;
	const char		*name;
354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381
};

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);
};

382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
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;
}

401 402 403 404 405 406 407 408 409
enum broken_state {
	BROKEN_ACQUIRE,
	BROKEN_ACQUIRED,
	BROKEN_CONTENDED,
	BROKEN_RELEASE,
	BROKEN_MAX,
};

static int bad_hist[BROKEN_MAX];
410

411 412 413 414 415
enum acquire_flags {
	TRY_LOCK = 1,
	READ_LOCK = 2,
};

416 417
static void
report_lock_acquire_event(struct trace_acquire_event *acquire_event,
418 419
			struct event *__event __used,
			int cpu __used,
420
			u64 timestamp __used,
421 422
			struct thread *thread __used)
{
423 424 425 426 427 428 429
	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;
430

431 432
	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, acquire_event->addr);
433

434 435 436 437 438 439
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
	case SEQ_STATE_RELEASED:
		if (!acquire_event->flag) {
			seq->state = SEQ_STATE_ACQUIRING;
		} else {
440
			if (acquire_event->flag & TRY_LOCK)
441
				ls->nr_trylock++;
442
			if (acquire_event->flag & READ_LOCK)
443 444 445 446 447 448 449
				ls->nr_readlock++;
			seq->state = SEQ_STATE_READ_ACQUIRED;
			seq->read_count = 1;
			ls->nr_acquired++;
		}
		break;
	case SEQ_STATE_READ_ACQUIRED:
450
		if (acquire_event->flag & READ_LOCK) {
451 452 453 454 455 456
			seq->read_count++;
			ls->nr_acquired++;
			goto end;
		} else {
			goto broken;
		}
457
		break;
458 459 460 461 462 463
	case SEQ_STATE_ACQUIRED:
	case SEQ_STATE_ACQUIRING:
	case SEQ_STATE_CONTENDED:
broken:
		/* broken lock sequence, discard it */
		ls->discard = 1;
464
		bad_hist[BROKEN_ACQUIRE]++;
465 466 467
		list_del(&seq->list);
		free(seq);
		goto end;
468 469
		break;
	default:
470
		BUG_ON("Unknown state of lock sequence found!\n");
471 472 473
		break;
	}

474 475 476 477
	ls->nr_acquire++;
	seq->prev_event_time = timestamp;
end:
	return;
478 479
}

480 481
static void
report_lock_acquired_event(struct trace_acquired_event *acquired_event,
482 483
			 struct event *__event __used,
			 int cpu __used,
484
			 u64 timestamp __used,
485 486
			 struct thread *thread __used)
{
487 488 489 490
	struct lock_stat *ls;
	struct thread_stat *ts;
	struct lock_seq_stat *seq;
	u64 contended_term;
491

492 493 494 495 496 497
	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);
498

499 500 501 502 503
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
		/* orphan event, do nothing */
		return;
	case SEQ_STATE_ACQUIRING:
504
		break;
505 506 507 508 509
	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;
510
		if (ls->wait_time_max < contended_term)
511
			ls->wait_time_max = contended_term;
512
		break;
513 514 515 516 517
	case SEQ_STATE_RELEASED:
	case SEQ_STATE_ACQUIRED:
	case SEQ_STATE_READ_ACQUIRED:
		/* broken lock sequence, discard it */
		ls->discard = 1;
518
		bad_hist[BROKEN_ACQUIRED]++;
519 520 521 522 523
		list_del(&seq->list);
		free(seq);
		goto end;
		break;

524
	default:
525
		BUG_ON("Unknown state of lock sequence found!\n");
526 527 528
		break;
	}

529 530 531 532 533
	seq->state = SEQ_STATE_ACQUIRED;
	ls->nr_acquired++;
	seq->prev_event_time = timestamp;
end:
	return;
534 535
}

536 537
static void
report_lock_contended_event(struct trace_contended_event *contended_event,
538 539
			  struct event *__event __used,
			  int cpu __used,
540
			  u64 timestamp __used,
541 542
			  struct thread *thread __used)
{
543 544 545 546 547 548 549
	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;
550

551 552
	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, contended_event->addr);
553

554 555 556 557 558
	switch (seq->state) {
	case SEQ_STATE_UNINITIALIZED:
		/* orphan event, do nothing */
		return;
	case SEQ_STATE_ACQUIRING:
559
		break;
560 561 562 563 564 565
	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;
566
		bad_hist[BROKEN_CONTENDED]++;
567 568 569
		list_del(&seq->list);
		free(seq);
		goto end;
570 571
		break;
	default:
572
		BUG_ON("Unknown state of lock sequence found!\n");
573 574 575
		break;
	}

576 577 578 579 580
	seq->state = SEQ_STATE_CONTENDED;
	ls->nr_contended++;
	seq->prev_event_time = timestamp;
end:
	return;
581 582
}

583 584
static void
report_lock_release_event(struct trace_release_event *release_event,
585 586
			struct event *__event __used,
			int cpu __used,
587
			u64 timestamp __used,
588 589
			struct thread *thread __used)
{
590 591 592
	struct lock_stat *ls;
	struct thread_stat *ts;
	struct lock_seq_stat *seq;
593

594 595 596
	ls = lock_stat_findnew(release_event->addr, release_event->name);
	if (ls->discard)
		return;
597

598 599
	ts = thread_stat_findnew(thread->pid);
	seq = get_seq(ts, release_event->addr);
600

601 602 603 604 605 606 607 608 609 610 611
	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++;
612 613
			goto end;
		}
614 615 616 617 618 619
		break;
	case SEQ_STATE_ACQUIRING:
	case SEQ_STATE_CONTENDED:
	case SEQ_STATE_RELEASED:
		/* broken lock sequence, discard it */
		ls->discard = 1;
620
		bad_hist[BROKEN_RELEASE]++;
621
		goto free_seq;
622 623
		break;
	default:
624
		BUG_ON("Unknown state of lock sequence found!\n");
625 626 627
		break;
	}

628 629 630 631
	ls->nr_release++;
free_seq:
	list_del(&seq->list);
	free(seq);
632
end:
633
	return;
634 635 636 637
}

/* lock oriented handlers */
/* TODO: handlers for CPU oriented, thread oriented */
638 639 640 641 642
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,
643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659
};

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);
660
	acquire_event.flag = (int)raw_field_value(event, "flag", data);
661

662 663
	if (trace_handler->acquire_event)
		trace_handler->acquire_event(&acquire_event, event, cpu, timestamp, thread);
664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
}

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);

680 681
	if (trace_handler->acquire_event)
		trace_handler->acquired_event(&acquired_event, event, cpu, timestamp, thread);
682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697
}

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);

698 699
	if (trace_handler->acquire_event)
		trace_handler->contended_event(&contended_event, event, cpu, timestamp, thread);
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
}

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);

716 717
	if (trace_handler->acquire_event)
		trace_handler->release_event(&release_event, event, cpu, timestamp, thread);
718 719 720
}

static void
721
process_raw_event(void *data, int cpu, u64 timestamp, struct thread *thread)
722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738
{
	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);
}

739 740 741 742 743 744 745 746
static void print_bad_events(int bad, int total)
{
	/* Output for debug, this have to be removed */
	int i;
	const char *name[4] =
		{ "acquire", "acquired", "contended", "release" };

	pr_info("\n=== output for debug===\n\n");
747 748
	pr_info("bad: %d, total: %d\n", bad, total);
	pr_info("bad rate: %f %%\n", (double)bad / (double)total * 100);
749 750 751 752 753
	pr_info("histogram of events caused bad sequence\n");
	for (i = 0; i < BROKEN_MAX; i++)
		pr_info(" %10s: %d\n", name[i], bad_hist[i]);
}

754 755 756 757 758
/* TODO: various way to print, coloring, nano or milli sec */
static void print_result(void)
{
	struct lock_stat *st;
	char cut_name[20];
759
	int bad, total;
760

761 762 763
	pr_info("%20s ", "Name");
	pr_info("%10s ", "acquired");
	pr_info("%10s ", "contended");
764

765 766 767
	pr_info("%15s ", "total wait (ns)");
	pr_info("%15s ", "max wait (ns)");
	pr_info("%15s ", "min wait (ns)");
768

769
	pr_info("\n\n");
770

771
	bad = total = 0;
772
	while ((st = pop_from_result())) {
773 774 775 776 777
		total++;
		if (st->discard) {
			bad++;
			continue;
		}
778 779 780 781
		bzero(cut_name, 20);

		if (strlen(st->name) < 16) {
			/* output raw name */
782
			pr_info("%20s ", st->name);
783 784 785 786 787 788 789
		} 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 */
790
			pr_info("%20s ", cut_name);
791 792
		}

793 794
		pr_info("%10u ", st->nr_acquired);
		pr_info("%10u ", st->nr_contended);
795

796 797 798
		pr_info("%15" PRIu64 " ", st->wait_time_total);
		pr_info("%15" PRIu64 " ", st->wait_time_max);
		pr_info("%15" PRIu64 " ", st->wait_time_min == ULLONG_MAX ?
799
		       0 : st->wait_time_min);
800
		pr_info("\n");
801
	}
802

803
	print_bad_events(bad, total);
804 805
}

806
static bool info_threads, info_map;
807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824

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);
	};
}

825 826 827 828 829
static void dump_map(void)
{
	unsigned int i;
	struct lock_stat *st;

830
	pr_info("Address of instance: name of class\n");
831 832
	for (i = 0; i < LOCKHASH_SIZE; i++) {
		list_for_each_entry(st, &lockhash_table[i], hash_entry) {
833
			pr_info(" %p: %s\n", st->addr, st->name);
834 835 836 837
		}
	}
}

838 839 840 841 842 843 844 845 846 847
static void dump_info(void)
{
	if (info_threads)
		dump_threads();
	else if (info_map)
		dump_map();
	else
		die("Unknown type of information\n");
}

848 849 850
static int process_sample_event(union perf_event *event,
				struct perf_sample *sample,
				struct perf_evsel *evsel __used,
851
				struct perf_session *s)
852
{
853
	struct thread *thread = perf_session__findnew(s, sample->tid);
854 855 856

	if (thread == NULL) {
		pr_debug("problem processing %d event, skipping it.\n",
857
			event->header.type);
858 859 860
		return -1;
	}

861
	process_raw_event(sample->raw_data, sample->cpu, sample->time, thread);
862 863 864 865

	return 0;
}

866
static struct perf_event_ops eops = {
867
	.sample			= process_sample_event,
868
	.comm			= perf_event__process_comm,
869
	.ordered_samples	= true,
870 871 872 873
};

static int read_events(void)
{
874
	session = perf_session__new(input_name, O_RDONLY, 0, false, &eops);
875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892
	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);
		}
	}
}

893
static void __cmd_report(void)
894 895 896 897 898 899 900 901
{
	setup_pager();
	select_key();
	read_events();
	sort_result();
	print_result();
}

902 903
static const char * const report_usage[] = {
	"perf lock report [<options>]",
904 905 906
	NULL
};

907
static const struct option report_options[] = {
908
	OPT_STRING('k', "key", &sort_key, "acquired",
909
		    "key for sorting (acquired / contended / wait_total / wait_max / wait_min)"),
910 911 912 913
	/* TODO: type */
	OPT_END()
};

914 915 916 917 918 919 920 921 922 923 924 925 926
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()
};

927
static const char * const lock_usage[] = {
928
	"perf lock [<options>] {record|trace|report}",
929 930 931 932
	NULL
};

static const struct option lock_options[] = {
933
	OPT_STRING('i', "input", &input_name, "file", "input file name"),
934
	OPT_INCR('v', "verbose", &verbose, "be more verbose (show symbol address, etc)"),
935
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"),
936 937 938 939 940 941 942 943 944
	OPT_END()
};

static const char *record_args[] = {
	"record",
	"-R",
	"-f",
	"-m", "1024",
	"-c", "1",
945 946 947 948
	"-e", "lock:lock_acquire",
	"-e", "lock:lock_acquired",
	"-e", "lock:lock_contended",
	"-e", "lock:lock_release",
949 950 951 952 953 954 955 956 957 958
};

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 *));

959 960 961
	if (rec_argv == NULL)
		return -ENOMEM;

962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987
	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);
988 989
	} else if (!strncmp(argv[0], "report", 6)) {
		trace_handler = &report_lock_ops;
990 991
		if (argc) {
			argc = parse_options(argc, argv,
992
					     report_options, report_usage, 0);
993
			if (argc)
994
				usage_with_options(report_usage, report_options);
995
		}
996
		__cmd_report();
997 998 999
	} else if (!strcmp(argv[0], "script")) {
		/* Aliased to 'perf script' */
		return cmd_script(argc, argv, prefix);
1000 1001 1002 1003 1004 1005 1006
	} 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);
		}
1007 1008
		/* recycling report_lock_ops */
		trace_handler = &report_lock_ops;
1009 1010
		setup_pager();
		read_events();
1011
		dump_info();
1012 1013 1014 1015 1016 1017
	} else {
		usage_with_options(lock_usage, lock_options);
	}

	return 0;
}