tools/perf/builtin-lock.c

   1 #include "builtin.h"
   2 #include "perf.h"
   3
   4 #include "util/util.h"
   5 #include "util/cache.h"
   6 #include "util/symbol.h"
   7 #include "util/thread.h"
   8 #include "util/header.h"
   9
  10 #include "util/parse-options.h"
  11 #include "util/trace-event.h"
  12
  13 #include "util/debug.h"
  14 #include "util/session.h"
  15
  16 #include <sys/types.h>
  17 #include <sys/prctl.h>
  18 #include <semaphore.h>
  19 #include <pthread.h>
  20 #include <math.h>
  21 #include <limits.h>
  22
  23 #include <linux/list.h>
  24 #include <linux/hash.h>
  25
  26 static struct perf_session *session;
  27
  28 /* based on kernel/lockdep.c */
  29 #define LOCKHASH_BITS           12
  30 #define LOCKHASH_SIZE           (1UL << LOCKHASH_BITS)
  31
  32 static struct list_head lockhash_table[LOCKHASH_SIZE];
  33
  34 #define __lockhashfn(key)       hash_long((unsigned long)key, LOCKHASH_BITS)
  35 #define lockhashentry(key)      (lockhash_table + __lockhashfn((key)))
  36
  37 struct lock_stat {
  38         struct list_head        hash_entry;
  39         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          * Is there more better solution?
  45          */
  46         void                    *addr;          /* address of lockdep_map, used as ID */
  47         char                    *name;          /* for strcpy(), we cannot use const */
  48
  49         unsigned int            nr_acquire;
  50         unsigned int            nr_acquired;
  51         unsigned int            nr_contended;
  52         unsigned int            nr_release;
  53
  54         unsigned int            nr_readlock;
  55         unsigned int            nr_trylock;
  56         /* these times are in nano sec. */
  57         u64                     wait_time_total;
  58         u64                     wait_time_min;
  59         u64                     wait_time_max;
  60
  61         int                     discard; /* flag of blacklist */
  62 };
  63
  64 /*
  65  * States of lock_seq_stat
  66  *
  67  * UNINITIALIZED is required for detecting first event of acquire.
  68  * As the nature of lock events, there is no guarantee
  69  * that the first event for the locks are acquire,
  70  * it can be acquired, contended or release.
  71  */
  72 #define SEQ_STATE_UNINITIALIZED      0         /* initial state */
  73 #define SEQ_STATE_RELEASED      1
  74 #define SEQ_STATE_ACQUIRING     2
  75 #define SEQ_STATE_ACQUIRED      3
  76 #define SEQ_STATE_READ_ACQUIRED 4
  77 #define SEQ_STATE_CONTENDED     5
  78
  79 /*
  80  * MAX_LOCK_DEPTH
  81  * Imported from include/linux/sched.h.
  82  * Should this be synchronized?
  83  */
  84 #define MAX_LOCK_DEPTH 48
  85
  86 /*
  87  * struct lock_seq_stat:
  88  * Place to put on state of one lock sequence
  89  * 1) acquire -> acquired -> release
  90  * 2) acquire -> contended -> acquired -> release
  91  * 3) acquire (with read or try) -> release
  92  * 4) Are there other patterns?
  93  */
  94 struct lock_seq_stat {
  95         struct list_head        list;
  96         int                     state;
  97         u64                     prev_event_time;
  98         void                    *addr;
  99
 100         int                     read_count;
 101 };
 102
 103 struct thread_stat {
 104         struct rb_node          rb;
 105
 106         u32                     tid;
 107         struct list_head        seq_list;
 108 };
 109
 110 static struct rb_root           thread_stats;
 111
 112 static struct thread_stat *thread_stat_find(u32 tid)
 113 {
 114         struct rb_node *node;
 115         struct thread_stat *st;
 116
 117         node = thread_stats.rb_node;
 118         while (node) {
 119                 st = container_of(node, struct thread_stat, rb);
 120                 if (st->tid == tid)
 121                         return st;
 122                 else if (tid < st->tid)
 123                         node = node->rb_left;
 124                 else
 125                         node = node->rb_right;
 126         }
 127
 128         return NULL;
 129 }
 130
 131 static void thread_stat_insert(struct thread_stat *new)
 132 {
 133         struct rb_node **rb = &thread_stats.rb_node;
 134         struct rb_node *parent = NULL;
 135         struct thread_stat *p;
 136
 137         while (*rb) {
 138                 p = container_of(*rb, struct thread_stat, rb);
 139                 parent = *rb;
 140
 141                 if (new->tid < p->tid)
 142                         rb = &(*rb)->rb_left;
 143                 else if (new->tid > p->tid)
 144                         rb = &(*rb)->rb_right;
 145                 else
 146                         BUG_ON("inserting invalid thread_stat\n");
 147         }
 148
 149         rb_link_node(&new->rb, parent, rb);
 150         rb_insert_color(&new->rb, &thread_stats);
 151 }
 152
 153 static struct thread_stat *thread_stat_findnew_after_first(u32 tid)
 154 {
 155         struct thread_stat *st;
 156
 157         st = thread_stat_find(tid);
 158         if (st)
 159                 return st;
 160
 161         st = zalloc(sizeof(struct thread_stat));
 162         if (!st)
 163                 die("memory allocation failed\n");
 164
 165         st->tid = tid;
 166         INIT_LIST_HEAD(&st->seq_list);
 167
 168         thread_stat_insert(st);
 169
 170         return st;
 171 }
 172
 173 static struct thread_stat *thread_stat_findnew_first(u32 tid);
 174 static struct thread_stat *(*thread_stat_findnew)(u32 tid) =
 175         thread_stat_findnew_first;
 176
 177 static struct thread_stat *thread_stat_findnew_first(u32 tid)
 178 {
 179         struct thread_stat *st;
 180
 181         st = zalloc(sizeof(struct thread_stat));
 182         if (!st)
 183                 die("memory allocation failed\n");
 184         st->tid = tid;
 185         INIT_LIST_HEAD(&st->seq_list);
 186
 187         rb_link_node(&st->rb, NULL, &thread_stats.rb_node);
 188         rb_insert_color(&st->rb, &thread_stats);
 189
 190         thread_stat_findnew = thread_stat_findnew_after_first;
 191         return st;
 192 }
 193
 194 /* build simple key function one is bigger than two */
 195 #define SINGLE_KEY(member)                                              \
 196         static int lock_stat_key_ ## member(struct lock_stat *one,      \
 197                                          struct lock_stat *two)         \
 198         {                                                               \
 199                 return one->member > two->member;                       \
 200         }
 201
 202 SINGLE_KEY(nr_acquired)
 203 SINGLE_KEY(nr_contended)
 204 SINGLE_KEY(wait_time_total)
 205 SINGLE_KEY(wait_time_max)
 206
 207 static int lock_stat_key_wait_time_min(struct lock_stat *one,
 208                                         struct lock_stat *two)
 209 {
 210         u64 s1 = one->wait_time_min;
 211         u64 s2 = two->wait_time_min;
 212         if (s1 == ULLONG_MAX)
 213                 s1 = 0;
 214         if (s2 == ULLONG_MAX)
 215                 s2 = 0;
 216         return s1 > s2;
 217 }
 218
 219 struct lock_key {
 220         /*
 221          * name: the value for specify by user
 222          * this should be simpler than raw name of member
 223          * e.g. nr_acquired -> acquired, wait_time_total -> wait_total
 224          */
 225         const char              *name;
 226         int                     (*key)(struct lock_stat*, struct lock_stat*);
 227 };
 228
 229 static const char               *sort_key = "acquired";
 230
 231 static int                      (*compare)(struct lock_stat *, struct lock_stat *);
 232
 233 static struct rb_root           result; /* place to store sorted data */
 234
 235 #define DEF_KEY_LOCK(name, fn_suffix)   \
 236         { #name, lock_stat_key_ ## fn_suffix }
 237 struct lock_key keys[] = {
 238         DEF_KEY_LOCK(acquired, nr_acquired),
 239         DEF_KEY_LOCK(contended, nr_contended),
 240         DEF_KEY_LOCK(wait_total, wait_time_total),
 241         DEF_KEY_LOCK(wait_min, wait_time_min),
 242         DEF_KEY_LOCK(wait_max, wait_time_max),
 243
 244         /* extra comparisons much complicated should be here */
 245
 246         { NULL, NULL }
 247 };
 248
 249 static void select_key(void)
 250 {
 251         int i;
 252
 253         for (i = 0; keys[i].name; i++) {
 254                 if (!strcmp(keys[i].name, sort_key)) {
 255                         compare = keys[i].key;
 256                         return;
 257                 }
 258         }
 259
 260         die("Unknown compare key:%s\n", sort_key);
 261 }
 262
 263 static void insert_to_result(struct lock_stat *st,
 264                              int (*bigger)(struct lock_stat *, struct lock_stat *))
 265 {
 266         struct rb_node **rb = &result.rb_node;
 267         struct rb_node *parent = NULL;
 268         struct lock_stat *p;
 269
 270         while (*rb) {
 271                 p = container_of(*rb, struct lock_stat, rb);
 272                 parent = *rb;
 273
 274                 if (bigger(st, p))
 275                         rb = &(*rb)->rb_left;
 276                 else
 277                         rb = &(*rb)->rb_right;
 278         }
 279
 280         rb_link_node(&st->rb, parent, rb);
 281         rb_insert_color(&st->rb, &result);
 282 }
 283
 284 /* returns left most element of result, and erase it */
 285 static struct lock_stat *pop_from_result(void)
 286 {
 287         struct rb_node *node = result.rb_node;
 288
 289         if (!node)
 290                 return NULL;
 291
 292         while (node->rb_left)
 293                 node = node->rb_left;
 294
 295         rb_erase(node, &result);
 296         return container_of(node, struct lock_stat, rb);
 297 }
 298
 299 static struct lock_stat *lock_stat_findnew(void *addr, const char *name)
 300 {
 301         struct list_head *entry = lockhashentry(addr);
 302         struct lock_stat *ret, *new;
 303
 304         list_for_each_entry(ret, entry, hash_entry) {
 305                 if (ret->addr == addr)
 306                         return ret;
 307         }
 308
 309         new = zalloc(sizeof(struct lock_stat));
 310         if (!new)
 311                 goto alloc_failed;
 312
 313         new->addr = addr;
 314         new->name = zalloc(sizeof(char) * strlen(name) + 1);
 315         if (!new->name)
 316                 goto alloc_failed;
 317         strcpy(new->name, name);
 318
 319         new->wait_time_min = ULLONG_MAX;
 320
 321         list_add(&new->hash_entry, entry);
 322         return new;
 323
 324 alloc_failed:
 325         die("memory allocation failed\n");
 326 }
 327
 328 static char                     const *input_name = "perf.data";
 329
 330 struct raw_event_sample {
 331         u32                     size;
 332         char                    data[0];
 333 };
 334
 335 struct trace_acquire_event {
 336         void                    *addr;
 337         const char              *name;
 338         int                     flag;
 339 };
 340
 341 struct trace_acquired_event {
 342         void                    *addr;
 343         const char              *name;
 344 };
 345
 346 struct trace_contended_event {
 347         void                    *addr;
 348         const char              *name;
 349 };
 350
 351 struct trace_release_event {
 352         void                    *addr;
 353         const char              *name;
 354 };
 355
 356 struct trace_lock_handler {
 357         void (*acquire_event)(struct trace_acquire_event *,
 358                               struct event *,
 359                               int cpu,
 360                               u64 timestamp,
 361                               struct thread *thread);
 362
 363         void (*acquired_event)(struct trace_acquired_event *,
 364                                struct event *,
 365                                int cpu,
 366                                u64 timestamp,
 367                                struct thread *thread);
 368
 369         void (*contended_event)(struct trace_contended_event *,
 370                                 struct event *,
 371                                 int cpu,
 372                                 u64 timestamp,
 373                                 struct thread *thread);
 374
 375         void (*release_event)(struct trace_release_event *,
 376                               struct event *,
 377                               int cpu,
 378                               u64 timestamp,
 379                               struct thread *thread);
 380 };
 381
 382 static struct lock_seq_stat *get_seq(struct thread_stat *ts, void *addr)
 383 {
 384         struct lock_seq_stat *seq;
 385
 386         list_for_each_entry(seq, &ts->seq_list, list) {
 387                 if (seq->addr == addr)
 388                         return seq;
 389         }
 390
 391         seq = zalloc(sizeof(struct lock_seq_stat));
 392         if (!seq)
 393                 die("Not enough memory\n");
 394         seq->state = SEQ_STATE_UNINITIALIZED;
 395         seq->addr = addr;
 396
 397         list_add(&seq->list, &ts->seq_list);
 398         return seq;
 399 }
 400
 401 enum broken_state {
 402         BROKEN_ACQUIRE,
 403         BROKEN_ACQUIRED,
 404         BROKEN_CONTENDED,
 405         BROKEN_RELEASE,
 406         BROKEN_MAX,
 407 };
 408
 409 static int bad_hist[BROKEN_MAX];
 410
 411 enum acquire_flags {
 412         TRY_LOCK = 1,
 413         READ_LOCK = 2,
 414 };
 415
 416 static void
 417 report_lock_acquire_event(struct trace_acquire_event *acquire_event,
 418                         struct event *__event __used,
 419                         int cpu __used,
 420                         u64 timestamp __used,
 421                         struct thread *thread __used)
 422 {
 423         struct lock_stat *ls;
 424         struct thread_stat *ts;
 425         struct lock_seq_stat *seq;
 426
 427         ls = lock_stat_findnew(acquire_event->addr, acquire_event->name);
 428         if (ls->discard)
 429                 return;
 430
 431         ts = thread_stat_findnew(thread->pid);
 432         seq = get_seq(ts, acquire_event->addr);
 433
 434         switch (seq->state) {
 435         case SEQ_STATE_UNINITIALIZED:
 436         case SEQ_STATE_RELEASED:
 437                 if (!acquire_event->flag) {
 438                         seq->state = SEQ_STATE_ACQUIRING;
 439                 } else {
 440                         if (acquire_event->flag & TRY_LOCK)
 441                                 ls->nr_trylock++;
 442                         if (acquire_event->flag & READ_LOCK)
 443                                 ls->nr_readlock++;
 444                         seq->state = SEQ_STATE_READ_ACQUIRED;
 445                         seq->read_count = 1;
 446                         ls->nr_acquired++;
 447                 }
 448                 break;
 449         case SEQ_STATE_READ_ACQUIRED:
 450                 if (acquire_event->flag & READ_LOCK) {
 451                         seq->read_count++;
 452                         ls->nr_acquired++;
 453                         goto end;
 454                 } else {
 455                         goto broken;
 456                 }
 457                 break;
 458         case SEQ_STATE_ACQUIRED:
 459         case SEQ_STATE_ACQUIRING:
 460         case SEQ_STATE_CONTENDED:
 461 broken:
 462                 /* broken lock sequence, discard it */
 463                 ls->discard = 1;
 464                 bad_hist[BROKEN_ACQUIRE]++;
 465                 list_del(&seq->list);
 466                 free(seq);
 467                 goto end;
 468                 break;
 469         default:
 470                 BUG_ON("Unknown state of lock sequence found!\n");
 471                 break;
 472         }
 473
 474         ls->nr_acquire++;
 475         seq->prev_event_time = timestamp;
 476 end:
 477         return;
 478 }
 479
 480 static void
 481 report_lock_acquired_event(struct trace_acquired_event *acquired_event,
 482                          struct event *__event __used,
 483                          int cpu __used,
 484                          u64 timestamp __used,
 485                          struct thread *thread __used)
 486 {
 487         struct lock_stat *ls;
 488         struct thread_stat *ts;
 489         struct lock_seq_stat *seq;
 490         u64 contended_term;
 491
 492         ls = lock_stat_findnew(acquired_event->addr, acquired_event->name);
 493         if (ls->discard)
 494                 return;
 495
 496         ts = thread_stat_findnew(thread->pid);
 497         seq = get_seq(ts, acquired_event->addr);
 498
 499         switch (seq->state) {
 500         case SEQ_STATE_UNINITIALIZED:
 501                 /* orphan event, do nothing */
 502                 return;
 503         case SEQ_STATE_ACQUIRING:
 504                 break;
 505         case SEQ_STATE_CONTENDED:
 506                 contended_term = timestamp - seq->prev_event_time;
 507                 ls->wait_time_total += contended_term;
 508                 if (contended_term < ls->wait_time_min)
 509                         ls->wait_time_min = contended_term;
 510                 if (ls->wait_time_max < contended_term)
 511                         ls->wait_time_max = contended_term;
 512                 break;
 513         case SEQ_STATE_RELEASED:
 514         case SEQ_STATE_ACQUIRED:
 515         case SEQ_STATE_READ_ACQUIRED:
 516                 /* broken lock sequence, discard it */
 517                 ls->discard = 1;
 518                 bad_hist[BROKEN_ACQUIRED]++;
 519                 list_del(&seq->list);
 520                 free(seq);
 521                 goto end;
 522                 break;
 523
 524         default:
 525                 BUG_ON("Unknown state of lock sequence found!\n");
 526                 break;
 527         }
 528
 529         seq->state = SEQ_STATE_ACQUIRED;
 530         ls->nr_acquired++;
 531         seq->prev_event_time = timestamp;
 532 end:
 533         return;
 534 }
 535
 536 static void
 537 report_lock_contended_event(struct trace_contended_event *contended_event,
 538                           struct event *__event __used,
 539                           int cpu __used,
 540                           u64 timestamp __used,
 541                           struct thread *thread __used)
 542 {
 543         struct lock_stat *ls;
 544         struct thread_stat *ts;
 545         struct lock_seq_stat *seq;
 546
 547         ls = lock_stat_findnew(contended_event->addr, contended_event->name);
 548         if (ls->discard)
 549                 return;
 550
 551         ts = thread_stat_findnew(thread->pid);
 552         seq = get_seq(ts, contended_event->addr);
 553
 554         switch (seq->state) {
 555         case SEQ_STATE_UNINITIALIZED:
 556                 /* orphan event, do nothing */
 557                 return;
 558         case SEQ_STATE_ACQUIRING:
 559                 break;
 560         case SEQ_STATE_RELEASED:
 561         case SEQ_STATE_ACQUIRED:
 562         case SEQ_STATE_READ_ACQUIRED:
 563         case SEQ_STATE_CONTENDED:
 564                 /* broken lock sequence, discard it */
 565                 ls->discard = 1;
 566                 bad_hist[BROKEN_CONTENDED]++;
 567                 list_del(&seq->list);
 568                 free(seq);
 569                 goto end;
 570                 break;
 571         default:
 572                 BUG_ON("Unknown state of lock sequence found!\n");
 573                 break;
 574         }
 575
 576         seq->state = SEQ_STATE_CONTENDED;
 577         ls->nr_contended++;
 578         seq->prev_event_time = timestamp;
 579 end:
 580         return;
 581 }
 582
 583 static void
 584 report_lock_release_event(struct trace_release_event *release_event,
 585                         struct event *__event __used,
 586                         int cpu __used,
 587                         u64 timestamp __used,
 588                         struct thread *thread __used)
 589 {
 590         struct lock_stat *ls;
 591         struct thread_stat *ts;
 592         struct lock_seq_stat *seq;
 593
 594         ls = lock_stat_findnew(release_event->addr, release_event->name);
 595         if (ls->discard)
 596                 return;
 597
 598         ts = thread_stat_findnew(thread->pid);
 599         seq = get_seq(ts, release_event->addr);
 600
 601         switch (seq->state) {
 602         case SEQ_STATE_UNINITIALIZED:
 603                 goto end;
 604                 break;
 605         case SEQ_STATE_ACQUIRED:
 606                 break;
 607         case SEQ_STATE_READ_ACQUIRED:
 608                 seq->read_count--;
 609                 BUG_ON(seq->read_count < 0);
 610                 if (!seq->read_count) {
 611                         ls->nr_release++;
 612                         goto end;
 613                 }
 614                 break;
 615         case SEQ_STATE_ACQUIRING:
 616         case SEQ_STATE_CONTENDED:
 617         case SEQ_STATE_RELEASED:
 618                 /* broken lock sequence, discard it */
 619                 ls->discard = 1;
 620                 bad_hist[BROKEN_RELEASE]++;
 621                 goto free_seq;
 622                 break;
 623         default:
 624                 BUG_ON("Unknown state of lock sequence found!\n");
 625                 break;
 626         }
 627
 628         ls->nr_release++;
 629 free_seq:
 630         list_del(&seq->list);
 631         free(seq);
 632 end:
 633         return;
 634 }
 635
 636 /* lock oriented handlers */
 637 /* TODO: handlers for CPU oriented, thread oriented */
 638 static struct trace_lock_handler report_lock_ops  = {
 639         .acquire_event          = report_lock_acquire_event,
 640         .acquired_event         = report_lock_acquired_event,
 641         .contended_event        = report_lock_contended_event,
 642         .release_event          = report_lock_release_event,
 643 };
 644
 645 static struct trace_lock_handler *trace_handler;
 646
 647 static void
 648 process_lock_acquire_event(void *data,
 649                            struct event *event __used,
 650                            int cpu __used,
 651                            u64 timestamp __used,
 652                            struct thread *thread __used)
 653 {
 654         struct trace_acquire_event acquire_event;
 655         u64 tmp;                /* this is required for casting... */
 656
 657         tmp = raw_field_value(event, "lockdep_addr", data);
 658         memcpy(&acquire_event.addr, &tmp, sizeof(void *));
 659         acquire_event.name = (char *)raw_field_ptr(event, "name", data);
 660         acquire_event.flag = (int)raw_field_value(event, "flag", data);
 661
 662         if (trace_handler->acquire_event)
 663                 trace_handler->acquire_event(&acquire_event, event, cpu, timestamp, thread);
 664 }
 665
 666 static void
 667 process_lock_acquired_event(void *data,
 668                             struct event *event __used,
 669                             int cpu __used,
 670                             u64 timestamp __used,
 671                             struct thread *thread __used)
 672 {
 673         struct trace_acquired_event acquired_event;
 674         u64 tmp;                /* this is required for casting... */
 675
 676         tmp = raw_field_value(event, "lockdep_addr", data);
 677         memcpy(&acquired_event.addr, &tmp, sizeof(void *));
 678         acquired_event.name = (char *)raw_field_ptr(event, "name", data);
 679
 680         if (trace_handler->acquire_event)
 681                 trace_handler->acquired_event(&acquired_event, event, cpu, timestamp, thread);
 682 }
 683
 684 static void
 685 process_lock_contended_event(void *data,
 686                              struct event *event __used,
 687                              int cpu __used,
 688                              u64 timestamp __used,
 689                              struct thread *thread __used)
 690 {
 691         struct trace_contended_event contended_event;
 692         u64 tmp;                /* this is required for casting... */
 693
 694         tmp = raw_field_value(event, "lockdep_addr", data);
 695         memcpy(&contended_event.addr, &tmp, sizeof(void *));
 696         contended_event.name = (char *)raw_field_ptr(event, "name", data);
 697
 698         if (trace_handler->acquire_event)
 699                 trace_handler->contended_event(&contended_event, event, cpu, timestamp, thread);
 700 }
 701
 702 static void
 703 process_lock_release_event(void *data,
 704                            struct event *event __used,
 705                            int cpu __used,
 706                            u64 timestamp __used,
 707                            struct thread *thread __used)
 708 {
 709         struct trace_release_event release_event;
 710         u64 tmp;                /* this is required for casting... */
 711
 712         tmp = raw_field_value(event, "lockdep_addr", data);
 713         memcpy(&release_event.addr, &tmp, sizeof(void *));
 714         release_event.name = (char *)raw_field_ptr(event, "name", data);
 715
 716         if (trace_handler->acquire_event)
 717                 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         struct event *event;
 724         int type;
 725
 726         type = trace_parse_common_type(data);
 727         event = trace_find_event(type);
 728
 729         if (!strcmp(event->name, "lock_acquire"))
 730                 process_lock_acquire_event(data, event, cpu, timestamp, thread);
 731         if (!strcmp(event->name, "lock_acquired"))
 732                 process_lock_acquired_event(data, event, cpu, timestamp, thread);
 733         if (!strcmp(event->name, "lock_contended"))
 734                 process_lock_contended_event(data, event, cpu, timestamp, thread);
 735         if (!strcmp(event->name, "lock_release"))
 736                 process_lock_release_event(data, event, cpu, timestamp, thread);
 737 }
 738
 739 static void print_bad_events(int bad, int total)
 740 {
 741         /* Output for debug, this have to be removed */
 742         int i;
 743         const char *name[4] =
 744                 { "acquire", "acquired", "contended", "release" };
 745
 746         pr_info("\n=== output for debug===\n\n");
 747         pr_info("bad: %d, total: %d\n", bad, total);
 748         pr_info("bad rate: %f %%\n", (double)bad / (double)total * 100);
 749         pr_info("histogram of events caused bad sequence\n");
 750         for (i = 0; i < BROKEN_MAX; i++)
 751                 pr_info(" %10s: %d\n", name[i], bad_hist[i]);
 752 }
 753
 754 /* TODO: various way to print, coloring, nano or milli sec */
 755 static void print_result(void)
 756 {
 757         struct lock_stat *st;
 758         char cut_name[20];
 759         int bad, total;
 760
 761         pr_info("%20s ", "Name");
 762         pr_info("%10s ", "acquired");
 763         pr_info("%10s ", "contended");
 764
 765         pr_info("%15s ", "total wait (ns)");
 766         pr_info("%15s ", "max wait (ns)");
 767         pr_info("%15s ", "min wait (ns)");
 768
 769         pr_info("\n\n");
 770
 771         bad = total = 0;
 772         while ((st = pop_from_result())) {
 773                 total++;
 774                 if (st->discard) {
 775                         bad++;
 776                         continue;
 777                 }
 778                 bzero(cut_name, 20);
 779
 780                 if (strlen(st->name) < 16) {
 781                         /* output raw name */
 782                         pr_info("%20s ", st->name);
 783                 } else {
 784                         strncpy(cut_name, st->name, 16);
 785                         cut_name[16] = '.';
 786                         cut_name[17] = '.';
 787                         cut_name[18] = '.';
 788                         cut_name[19] = '\0';
 789                         /* cut off name for saving output style */
 790                         pr_info("%20s ", cut_name);
 791                 }
 792
 793                 pr_info("%10u ", st->nr_acquired);
 794                 pr_info("%10u ", st->nr_contended);
 795
 796                 pr_info("%15" PRIu64 " ", st->wait_time_total);
 797                 pr_info("%15" PRIu64 " ", st->wait_time_max);
 798                 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 static void dump_threads(void)
 809 {
 810         struct thread_stat *st;
 811         struct rb_node *node;
 812         struct thread *t;
 813
 814         pr_info("%10s: comm\n", "Thread ID");
 815
 816         node = rb_first(&thread_stats);
 817         while (node) {
 818                 st = container_of(node, struct thread_stat, rb);
 819                 t = perf_session__findnew(session, st->tid);
 820                 pr_info("%10d: %s\n", st->tid, t->comm);
 821                 node = rb_next(node);
 822         };
 823 }
 824
 825 static void dump_map(void)
 826 {
 827         unsigned int i;
 828         struct lock_stat *st;
 829
 830         pr_info("Address of instance: name of class\n");
 831         for (i = 0; i < LOCKHASH_SIZE; i++) {
 832                 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 static void dump_info(void)
 839 {
 840         if (info_threads)
 841                 dump_threads();
 842         else if (info_map)
 843                 dump_map();
 844         else
 845                 die("Unknown type of information\n");
 846 }
 847
 848 static int process_sample_event(union perf_event *event,
 849                                 struct perf_sample *sample,
 850                                 struct perf_evsel *evsel __used,
 851                                 struct perf_session *s)
 852 {
 853         struct thread *thread = perf_session__findnew(s, sample->tid);
 854
 855         if (thread == NULL) {
 856                 pr_debug("problem processing %d event, skipping it.\n",
 857                         event->header.type);
 858                 return -1;
 859         }
 860
 861         process_raw_event(sample->raw_data, sample->cpu, sample->time, thread);
 862
 863         return 0;
 864 }
 865
 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 static int read_events(void)
 873 {
 874         session = perf_session__new(input_name, O_RDONLY, 0, false, &eops);
 875         if (!session)
 876                 die("Initializing perf session failed\n");
 877
 878         return perf_session__process_events(session, &eops);
 879 }
 880
 881 static void sort_result(void)
 882 {
 883         unsigned int i;
 884         struct lock_stat *st;
 885
 886         for (i = 0; i < LOCKHASH_SIZE; i++) {
 887                 list_for_each_entry(st, &lockhash_table[i], hash_entry) {
 888                         insert_to_result(st, compare);
 889                 }
 890         }
 891 }
 892
 893 static void __cmd_report(void)
 894 {
 895         setup_pager();
 896         select_key();
 897         read_events();
 898         sort_result();
 899         print_result();
 900 }
 901
 902 static const char * const report_usage[] = {
 903         "perf lock report [<options>]",
 904         NULL
 905 };
 906
 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         /* TODO: type */
 911         OPT_END()
 912 };
 913
 914 static const char * const info_usage[] = {
 915         "perf lock info [<options>]",
 916         NULL
 917 };
 918
 919 static const struct option info_options[] = {
 920         OPT_BOOLEAN('t', "threads", &info_threads,
 921                     "dump thread list in perf.data"),
 922         OPT_BOOLEAN('m', "map", &info_map,
 923                     "map of lock instances (name:address table)"),
 924         OPT_END()
 925 };
 926
 927 static const char * const lock_usage[] = {
 928         "perf lock [<options>] {record|trace|report}",
 929         NULL
 930 };
 931
 932 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         OPT_END()
 937 };
 938
 939 static const char *record_args[] = {
 940         "record",
 941         "-R",
 942         "-f",
 943         "-m", "1024",
 944         "-c", "1",
 945         "-e", "lock:lock_acquire",
 946         "-e", "lock:lock_acquired",
 947         "-e", "lock:lock_contended",
 948         "-e", "lock:lock_release",
 949 };
 950
 951 static int __cmd_record(int argc, const char **argv)
 952 {
 953         unsigned int rec_argc, i, j;
 954         const char **rec_argv;
 955
 956         rec_argc = ARRAY_SIZE(record_args) + argc - 1;
 957         rec_argv = calloc(rec_argc + 1, sizeof(char *));
 958
 959         if (rec_argv == NULL)
 960                 return -ENOMEM;
 961
 962         for (i = 0; i < ARRAY_SIZE(record_args); i++)
 963                 rec_argv[i] = strdup(record_args[i]);
 964
 965         for (j = 1; j < (unsigned int)argc; j++, i++)
 966                 rec_argv[i] = argv[j];
 967
 968         BUG_ON(i != rec_argc);
 969
 970         return cmd_record(i, rec_argv, NULL);
 971 }
 972
 973 int cmd_lock(int argc, const char **argv, const char *prefix __used)
 974 {
 975         unsigned int i;
 976
 977         symbol__init();
 978         for (i = 0; i < LOCKHASH_SIZE; i++)
 979                 INIT_LIST_HEAD(lockhash_table + i);
 980
 981         argc = parse_options(argc, argv, lock_options, lock_usage,
 982                              PARSE_OPT_STOP_AT_NON_OPTION);
 983         if (!argc)
 984                 usage_with_options(lock_usage, lock_options);
 985
 986         if (!strncmp(argv[0], "rec", 3)) {
 987                 return __cmd_record(argc, argv);
 988         } else if (!strncmp(argv[0], "report", 6)) {
 989                 trace_handler = &report_lock_ops;
 990                 if (argc) {
 991                         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         } else if (!strcmp(argv[0], "script")) {
 998                 /* Aliased to 'perf script' */
 999                 return cmd_script(argc, argv, prefix);
1000         } else if (!strcmp(argv[0], "info")) {
1001                 if (argc) {
1002                         argc = parse_options(argc, argv,
1003                                              info_options, info_usage, 0);
1004                         if (argc)
1005                                 usage_with_options(info_usage, info_options);
1006                 }
1007                 /* recycling report_lock_ops */
1008                 trace_handler = &report_lock_ops;
1009                 setup_pager();
1010                 read_events();
1011                 dump_info();
1012         } else {
1013                 usage_with_options(lock_usage, lock_options);
1014         }
1015
1016         return 0;
1017 }