tools/perf/builtin-sched.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 #include "util/session.h"
  10
  11 #include "util/parse-options.h"
  12 #include "util/trace-event.h"
  13
  14 #include "util/debug.h"
  15
  16 #include <sys/prctl.h>
  17
  18 #include <semaphore.h>
  19 #include <pthread.h>
  20 #include <math.h>
  21
  22 static char                     const *input_name = "perf.data";
  23
  24 static char                     default_sort_order[] = "avg, max, switch, runtime";
  25 static const char               *sort_order = default_sort_order;
  26
  27 static int                      profile_cpu = -1;
  28
  29 #define PR_SET_NAME             15               /* Set process name */
  30 #define MAX_CPUS                4096
  31
  32 static u64                      run_measurement_overhead;
  33 static u64                      sleep_measurement_overhead;
  34
  35 #define COMM_LEN                20
  36 #define SYM_LEN                 129
  37
  38 #define MAX_PID                 65536
  39
  40 static unsigned long            nr_tasks;
  41
  42 struct sched_atom;
  43
  44 struct task_desc {
  45         unsigned long           nr;
  46         unsigned long           pid;
  47         char                    comm[COMM_LEN];
  48
  49         unsigned long           nr_events;
  50         unsigned long           curr_event;
  51         struct sched_atom       **atoms;
  52
  53         pthread_t               thread;
  54         sem_t                   sleep_sem;
  55
  56         sem_t                   ready_for_work;
  57         sem_t                   work_done_sem;
  58
  59         u64                     cpu_usage;
  60 };
  61
  62 enum sched_event_type {
  63         SCHED_EVENT_RUN,
  64         SCHED_EVENT_SLEEP,
  65         SCHED_EVENT_WAKEUP,
  66         SCHED_EVENT_MIGRATION,
  67 };
  68
  69 struct sched_atom {
  70         enum sched_event_type   type;
  71         int                     specific_wait;
  72         u64                     timestamp;
  73         u64                     duration;
  74         unsigned long           nr;
  75         sem_t                   *wait_sem;
  76         struct task_desc        *wakee;
  77 };
  78
  79 static struct task_desc         *pid_to_task[MAX_PID];
  80
  81 static struct task_desc         **tasks;
  82
  83 static pthread_mutex_t          start_work_mutex = PTHREAD_MUTEX_INITIALIZER;
  84 static u64                      start_time;
  85
  86 static pthread_mutex_t          work_done_wait_mutex = PTHREAD_MUTEX_INITIALIZER;
  87
  88 static unsigned long            nr_run_events;
  89 static unsigned long            nr_sleep_events;
  90 static unsigned long            nr_wakeup_events;
  91
  92 static unsigned long            nr_sleep_corrections;
  93 static unsigned long            nr_run_events_optimized;
  94
  95 static unsigned long            targetless_wakeups;
  96 static unsigned long            multitarget_wakeups;
  97
  98 static u64                      cpu_usage;
  99 static u64                      runavg_cpu_usage;
 100 static u64                      parent_cpu_usage;
 101 static u64                      runavg_parent_cpu_usage;
 102
 103 static unsigned long            nr_runs;
 104 static u64                      sum_runtime;
 105 static u64                      sum_fluct;
 106 static u64                      run_avg;
 107
 108 static unsigned int             replay_repeat = 10;
 109 static unsigned long            nr_timestamps;
 110 static unsigned long            nr_unordered_timestamps;
 111 static unsigned long            nr_state_machine_bugs;
 112 static unsigned long            nr_context_switch_bugs;
 113 static unsigned long            nr_events;
 114 static unsigned long            nr_lost_chunks;
 115 static unsigned long            nr_lost_events;
 116
 117 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
 118
 119 enum thread_state {
 120         THREAD_SLEEPING = 0,
 121         THREAD_WAIT_CPU,
 122         THREAD_SCHED_IN,
 123         THREAD_IGNORE
 124 };
 125
 126 struct work_atom {
 127         struct list_head        list;
 128         enum thread_state       state;
 129         u64                     sched_out_time;
 130         u64                     wake_up_time;
 131         u64                     sched_in_time;
 132         u64                     runtime;
 133 };
 134
 135 struct work_atoms {
 136         struct list_head        work_list;
 137         struct thread           *thread;
 138         struct rb_node          node;
 139         u64                     max_lat;
 140         u64                     max_lat_at;
 141         u64                     total_lat;
 142         u64                     nb_atoms;
 143         u64                     total_runtime;
 144 };
 145
 146 typedef int (*sort_fn_t)(struct work_atoms *, struct work_atoms *);
 147
 148 static struct rb_root           atom_root, sorted_atom_root;
 149
 150 static u64                      all_runtime;
 151 static u64                      all_count;
 152
 153
 154 static u64 get_nsecs(void)
 155 {
 156         struct timespec ts;
 157
 158         clock_gettime(CLOCK_MONOTONIC, &ts);
 159
 160         return ts.tv_sec * 1000000000ULL + ts.tv_nsec;
 161 }
 162
 163 static void burn_nsecs(u64 nsecs)
 164 {
 165         u64 T0 = get_nsecs(), T1;
 166
 167         do {
 168                 T1 = get_nsecs();
 169         } while (T1 + run_measurement_overhead < T0 + nsecs);
 170 }
 171
 172 static void sleep_nsecs(u64 nsecs)
 173 {
 174         struct timespec ts;
 175
 176         ts.tv_nsec = nsecs % 999999999;
 177         ts.tv_sec = nsecs / 999999999;
 178
 179         nanosleep(&ts, NULL);
 180 }
 181
 182 static void calibrate_run_measurement_overhead(void)
 183 {
 184         u64 T0, T1, delta, min_delta = 1000000000ULL;
 185         int i;
 186
 187         for (i = 0; i < 10; i++) {
 188                 T0 = get_nsecs();
 189                 burn_nsecs(0);
 190                 T1 = get_nsecs();
 191                 delta = T1-T0;
 192                 min_delta = min(min_delta, delta);
 193         }
 194         run_measurement_overhead = min_delta;
 195
 196         printf("run measurement overhead: %" PRIu64 " nsecs\n", min_delta);
 197 }
 198
 199 static void calibrate_sleep_measurement_overhead(void)
 200 {
 201         u64 T0, T1, delta, min_delta = 1000000000ULL;
 202         int i;
 203
 204         for (i = 0; i < 10; i++) {
 205                 T0 = get_nsecs();
 206                 sleep_nsecs(10000);
 207                 T1 = get_nsecs();
 208                 delta = T1-T0;
 209                 min_delta = min(min_delta, delta);
 210         }
 211         min_delta -= 10000;
 212         sleep_measurement_overhead = min_delta;
 213
 214         printf("sleep measurement overhead: %" PRIu64 " nsecs\n", min_delta);
 215 }
 216
 217 static struct sched_atom *
 218 get_new_event(struct task_desc *task, u64 timestamp)
 219 {
 220         struct sched_atom *event = zalloc(sizeof(*event));
 221         unsigned long idx = task->nr_events;
 222         size_t size;
 223
 224         event->timestamp = timestamp;
 225         event->nr = idx;
 226
 227         task->nr_events++;
 228         size = sizeof(struct sched_atom *) * task->nr_events;
 229         task->atoms = realloc(task->atoms, size);
 230         BUG_ON(!task->atoms);
 231
 232         task->atoms[idx] = event;
 233
 234         return event;
 235 }
 236
 237 static struct sched_atom *last_event(struct task_desc *task)
 238 {
 239         if (!task->nr_events)
 240                 return NULL;
 241
 242         return task->atoms[task->nr_events - 1];
 243 }
 244
 245 static void
 246 add_sched_event_run(struct task_desc *task, u64 timestamp, u64 duration)
 247 {
 248         struct sched_atom *event, *curr_event = last_event(task);
 249
 250         /*
 251          * optimize an existing RUN event by merging this one
 252          * to it:
 253          */
 254         if (curr_event && curr_event->type == SCHED_EVENT_RUN) {
 255                 nr_run_events_optimized++;
 256                 curr_event->duration += duration;
 257                 return;
 258         }
 259
 260         event = get_new_event(task, timestamp);
 261
 262         event->type = SCHED_EVENT_RUN;
 263         event->duration = duration;
 264
 265         nr_run_events++;
 266 }
 267
 268 static void
 269 add_sched_event_wakeup(struct task_desc *task, u64 timestamp,
 270                        struct task_desc *wakee)
 271 {
 272         struct sched_atom *event, *wakee_event;
 273
 274         event = get_new_event(task, timestamp);
 275         event->type = SCHED_EVENT_WAKEUP;
 276         event->wakee = wakee;
 277
 278         wakee_event = last_event(wakee);
 279         if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) {
 280                 targetless_wakeups++;
 281                 return;
 282         }
 283         if (wakee_event->wait_sem) {
 284                 multitarget_wakeups++;
 285                 return;
 286         }
 287
 288         wakee_event->wait_sem = zalloc(sizeof(*wakee_event->wait_sem));
 289         sem_init(wakee_event->wait_sem, 0, 0);
 290         wakee_event->specific_wait = 1;
 291         event->wait_sem = wakee_event->wait_sem;
 292
 293         nr_wakeup_events++;
 294 }
 295
 296 static void
 297 add_sched_event_sleep(struct task_desc *task, u64 timestamp,
 298                       u64 task_state __used)
 299 {
 300         struct sched_atom *event = get_new_event(task, timestamp);
 301
 302         event->type = SCHED_EVENT_SLEEP;
 303
 304         nr_sleep_events++;
 305 }
 306
 307 static struct task_desc *register_pid(unsigned long pid, const char *comm)
 308 {
 309         struct task_desc *task;
 310
 311         BUG_ON(pid >= MAX_PID);
 312
 313         task = pid_to_task[pid];
 314
 315         if (task)
 316                 return task;
 317
 318         task = zalloc(sizeof(*task));
 319         task->pid = pid;
 320         task->nr = nr_tasks;
 321         strcpy(task->comm, comm);
 322         /*
 323          * every task starts in sleeping state - this gets ignored
 324          * if there's no wakeup pointing to this sleep state:
 325          */
 326         add_sched_event_sleep(task, 0, 0);
 327
 328         pid_to_task[pid] = task;
 329         nr_tasks++;
 330         tasks = realloc(tasks, nr_tasks*sizeof(struct task_task *));
 331         BUG_ON(!tasks);
 332         tasks[task->nr] = task;
 333
 334         if (verbose)
 335                 printf("registered task #%ld, PID %ld (%s)\n", nr_tasks, pid, comm);
 336
 337         return task;
 338 }
 339
 340
 341 static void print_task_traces(void)
 342 {
 343         struct task_desc *task;
 344         unsigned long i;
 345
 346         for (i = 0; i < nr_tasks; i++) {
 347                 task = tasks[i];
 348                 printf("task %6ld (%20s:%10ld), nr_events: %ld\n",
 349                         task->nr, task->comm, task->pid, task->nr_events);
 350         }
 351 }
 352
 353 static void add_cross_task_wakeups(void)
 354 {
 355         struct task_desc *task1, *task2;
 356         unsigned long i, j;
 357
 358         for (i = 0; i < nr_tasks; i++) {
 359                 task1 = tasks[i];
 360                 j = i + 1;
 361                 if (j == nr_tasks)
 362                         j = 0;
 363                 task2 = tasks[j];
 364                 add_sched_event_wakeup(task1, 0, task2);
 365         }
 366 }
 367
 368 static void
 369 process_sched_event(struct task_desc *this_task __used, struct sched_atom *atom)
 370 {
 371         int ret = 0;
 372
 373         switch (atom->type) {
 374                 case SCHED_EVENT_RUN:
 375                         burn_nsecs(atom->duration);
 376                         break;
 377                 case SCHED_EVENT_SLEEP:
 378                         if (atom->wait_sem)
 379                                 ret = sem_wait(atom->wait_sem);
 380                         BUG_ON(ret);
 381                         break;
 382                 case SCHED_EVENT_WAKEUP:
 383                         if (atom->wait_sem)
 384                                 ret = sem_post(atom->wait_sem);
 385                         BUG_ON(ret);
 386                         break;
 387                 case SCHED_EVENT_MIGRATION:
 388                         break;
 389                 default:
 390                         BUG_ON(1);
 391         }
 392 }
 393
 394 static u64 get_cpu_usage_nsec_parent(void)
 395 {
 396         struct rusage ru;
 397         u64 sum;
 398         int err;
 399
 400         err = getrusage(RUSAGE_SELF, &ru);
 401         BUG_ON(err);
 402
 403         sum =  ru.ru_utime.tv_sec*1e9 + ru.ru_utime.tv_usec*1e3;
 404         sum += ru.ru_stime.tv_sec*1e9 + ru.ru_stime.tv_usec*1e3;
 405
 406         return sum;
 407 }
 408
 409 static int self_open_counters(void)
 410 {
 411         struct perf_event_attr attr;
 412         int fd;
 413
 414         memset(&attr, 0, sizeof(attr));
 415
 416         attr.type = PERF_TYPE_SOFTWARE;
 417         attr.config = PERF_COUNT_SW_TASK_CLOCK;
 418
 419         fd = sys_perf_event_open(&attr, 0, -1, -1, 0);
 420
 421         if (fd < 0)
 422                 die("Error: sys_perf_event_open() syscall returned"
 423                     "with %d (%s)\n", fd, strerror(errno));
 424         return fd;
 425 }
 426
 427 static u64 get_cpu_usage_nsec_self(int fd)
 428 {
 429         u64 runtime;
 430         int ret;
 431
 432         ret = read(fd, &runtime, sizeof(runtime));
 433         BUG_ON(ret != sizeof(runtime));
 434
 435         return runtime;
 436 }
 437
 438 static void *thread_func(void *ctx)
 439 {
 440         struct task_desc *this_task = ctx;
 441         u64 cpu_usage_0, cpu_usage_1;
 442         unsigned long i, ret;
 443         char comm2[22];
 444         int fd;
 445
 446         sprintf(comm2, ":%s", this_task->comm);
 447         prctl(PR_SET_NAME, comm2);
 448         fd = self_open_counters();
 449
 450 again:
 451         ret = sem_post(&this_task->ready_for_work);
 452         BUG_ON(ret);
 453         ret = pthread_mutex_lock(&start_work_mutex);
 454         BUG_ON(ret);
 455         ret = pthread_mutex_unlock(&start_work_mutex);
 456         BUG_ON(ret);
 457
 458         cpu_usage_0 = get_cpu_usage_nsec_self(fd);
 459
 460         for (i = 0; i < this_task->nr_events; i++) {
 461                 this_task->curr_event = i;
 462                 process_sched_event(this_task, this_task->atoms[i]);
 463         }
 464
 465         cpu_usage_1 = get_cpu_usage_nsec_self(fd);
 466         this_task->cpu_usage = cpu_usage_1 - cpu_usage_0;
 467         ret = sem_post(&this_task->work_done_sem);
 468         BUG_ON(ret);
 469
 470         ret = pthread_mutex_lock(&work_done_wait_mutex);
 471         BUG_ON(ret);
 472         ret = pthread_mutex_unlock(&work_done_wait_mutex);
 473         BUG_ON(ret);
 474
 475         goto again;
 476 }
 477
 478 static void create_tasks(void)
 479 {
 480         struct task_desc *task;
 481         pthread_attr_t attr;
 482         unsigned long i;
 483         int err;
 484
 485         err = pthread_attr_init(&attr);
 486         BUG_ON(err);
 487         err = pthread_attr_setstacksize(&attr,
 488                         (size_t) max(16 * 1024, PTHREAD_STACK_MIN));
 489         BUG_ON(err);
 490         err = pthread_mutex_lock(&start_work_mutex);
 491         BUG_ON(err);
 492         err = pthread_mutex_lock(&work_done_wait_mutex);
 493         BUG_ON(err);
 494         for (i = 0; i < nr_tasks; i++) {
 495                 task = tasks[i];
 496                 sem_init(&task->sleep_sem, 0, 0);
 497                 sem_init(&task->ready_for_work, 0, 0);
 498                 sem_init(&task->work_done_sem, 0, 0);
 499                 task->curr_event = 0;
 500                 err = pthread_create(&task->thread, &attr, thread_func, task);
 501                 BUG_ON(err);
 502         }
 503 }
 504
 505 static void wait_for_tasks(void)
 506 {
 507         u64 cpu_usage_0, cpu_usage_1;
 508         struct task_desc *task;
 509         unsigned long i, ret;
 510
 511         start_time = get_nsecs();
 512         cpu_usage = 0;
 513         pthread_mutex_unlock(&work_done_wait_mutex);
 514
 515         for (i = 0; i < nr_tasks; i++) {
 516                 task = tasks[i];
 517                 ret = sem_wait(&task->ready_for_work);
 518                 BUG_ON(ret);
 519                 sem_init(&task->ready_for_work, 0, 0);
 520         }
 521         ret = pthread_mutex_lock(&work_done_wait_mutex);
 522         BUG_ON(ret);
 523
 524         cpu_usage_0 = get_cpu_usage_nsec_parent();
 525
 526         pthread_mutex_unlock(&start_work_mutex);
 527
 528         for (i = 0; i < nr_tasks; i++) {
 529                 task = tasks[i];
 530                 ret = sem_wait(&task->work_done_sem);
 531                 BUG_ON(ret);
 532                 sem_init(&task->work_done_sem, 0, 0);
 533                 cpu_usage += task->cpu_usage;
 534                 task->cpu_usage = 0;
 535         }
 536
 537         cpu_usage_1 = get_cpu_usage_nsec_parent();
 538         if (!runavg_cpu_usage)
 539                 runavg_cpu_usage = cpu_usage;
 540         runavg_cpu_usage = (runavg_cpu_usage*9 + cpu_usage)/10;
 541
 542         parent_cpu_usage = cpu_usage_1 - cpu_usage_0;
 543         if (!runavg_parent_cpu_usage)
 544                 runavg_parent_cpu_usage = parent_cpu_usage;
 545         runavg_parent_cpu_usage = (runavg_parent_cpu_usage*9 +
 546                                    parent_cpu_usage)/10;
 547
 548         ret = pthread_mutex_lock(&start_work_mutex);
 549         BUG_ON(ret);
 550
 551         for (i = 0; i < nr_tasks; i++) {
 552                 task = tasks[i];
 553                 sem_init(&task->sleep_sem, 0, 0);
 554                 task->curr_event = 0;
 555         }
 556 }
 557
 558 static void run_one_test(void)
 559 {
 560         u64 T0, T1, delta, avg_delta, fluct;
 561
 562         T0 = get_nsecs();
 563         wait_for_tasks();
 564         T1 = get_nsecs();
 565
 566         delta = T1 - T0;
 567         sum_runtime += delta;
 568         nr_runs++;
 569
 570         avg_delta = sum_runtime / nr_runs;
 571         if (delta < avg_delta)
 572                 fluct = avg_delta - delta;
 573         else
 574                 fluct = delta - avg_delta;
 575         sum_fluct += fluct;
 576         if (!run_avg)
 577                 run_avg = delta;
 578         run_avg = (run_avg*9 + delta)/10;
 579
 580         printf("#%-3ld: %0.3f, ",
 581                 nr_runs, (double)delta/1000000.0);
 582
 583         printf("ravg: %0.2f, ",
 584                 (double)run_avg/1e6);
 585
 586         printf("cpu: %0.2f / %0.2f",
 587                 (double)cpu_usage/1e6, (double)runavg_cpu_usage/1e6);
 588
 589 #if 0
 590         /*
 591          * rusage statistics done by the parent, these are less
 592          * accurate than the sum_exec_runtime based statistics:
 593          */
 594         printf(" [%0.2f / %0.2f]",
 595                 (double)parent_cpu_usage/1e6,
 596                 (double)runavg_parent_cpu_usage/1e6);
 597 #endif
 598
 599         printf("\n");
 600
 601         if (nr_sleep_corrections)
 602                 printf(" (%ld sleep corrections)\n", nr_sleep_corrections);
 603         nr_sleep_corrections = 0;
 604 }
 605
 606 static void test_calibrations(void)
 607 {
 608         u64 T0, T1;
 609
 610         T0 = get_nsecs();
 611         burn_nsecs(1e6);
 612         T1 = get_nsecs();
 613
 614         printf("the run test took %" PRIu64 " nsecs\n", T1 - T0);
 615
 616         T0 = get_nsecs();
 617         sleep_nsecs(1e6);
 618         T1 = get_nsecs();
 619
 620         printf("the sleep test took %" PRIu64 " nsecs\n", T1 - T0);
 621 }
 622
 623 #define FILL_FIELD(ptr, field, event, data)     \
 624         ptr.field = (typeof(ptr.field)) raw_field_value(event, #field, data)
 625
 626 #define FILL_ARRAY(ptr, array, event, data)                     \
 627 do {                                                            \
 628         void *__array = raw_field_ptr(event, #array, data);     \
 629         memcpy(ptr.array, __array, sizeof(ptr.array));  \
 630 } while(0)
 631
 632 #define FILL_COMMON_FIELDS(ptr, event, data)                    \
 633 do {                                                            \
 634         FILL_FIELD(ptr, common_type, event, data);              \
 635         FILL_FIELD(ptr, common_flags, event, data);             \
 636         FILL_FIELD(ptr, common_preempt_count, event, data);     \
 637         FILL_FIELD(ptr, common_pid, event, data);               \
 638         FILL_FIELD(ptr, common_tgid, event, data);              \
 639 } while (0)
 640
 641
 642
 643 struct trace_switch_event {
 644         u32 size;
 645
 646         u16 common_type;
 647         u8 common_flags;
 648         u8 common_preempt_count;
 649         u32 common_pid;
 650         u32 common_tgid;
 651
 652         char prev_comm[16];
 653         u32 prev_pid;
 654         u32 prev_prio;
 655         u64 prev_state;
 656         char next_comm[16];
 657         u32 next_pid;
 658         u32 next_prio;
 659 };
 660
 661 struct trace_runtime_event {
 662         u32 size;
 663
 664         u16 common_type;
 665         u8 common_flags;
 666         u8 common_preempt_count;
 667         u32 common_pid;
 668         u32 common_tgid;
 669
 670         char comm[16];
 671         u32 pid;
 672         u64 runtime;
 673         u64 vruntime;
 674 };
 675
 676 struct trace_wakeup_event {
 677         u32 size;
 678
 679         u16 common_type;
 680         u8 common_flags;
 681         u8 common_preempt_count;
 682         u32 common_pid;
 683         u32 common_tgid;
 684
 685         char comm[16];
 686         u32 pid;
 687
 688         u32 prio;
 689         u32 success;
 690         u32 cpu;
 691 };
 692
 693 struct trace_fork_event {
 694         u32 size;
 695
 696         u16 common_type;
 697         u8 common_flags;
 698         u8 common_preempt_count;
 699         u32 common_pid;
 700         u32 common_tgid;
 701
 702         char parent_comm[16];
 703         u32 parent_pid;
 704         char child_comm[16];
 705         u32 child_pid;
 706 };
 707
 708 struct trace_migrate_task_event {
 709         u32 size;
 710
 711         u16 common_type;
 712         u8 common_flags;
 713         u8 common_preempt_count;
 714         u32 common_pid;
 715         u32 common_tgid;
 716
 717         char comm[16];
 718         u32 pid;
 719
 720         u32 prio;
 721         u32 cpu;
 722 };
 723
 724 struct trace_sched_handler {
 725         void (*switch_event)(struct trace_switch_event *,
 726                              struct perf_session *,
 727                              struct event *,
 728                              int cpu,
 729                              u64 timestamp,
 730                              struct thread *thread);
 731
 732         void (*runtime_event)(struct trace_runtime_event *,
 733                               struct perf_session *,
 734                               struct event *,
 735                               int cpu,
 736                               u64 timestamp,
 737                               struct thread *thread);
 738
 739         void (*wakeup_event)(struct trace_wakeup_event *,
 740                              struct perf_session *,
 741                              struct event *,
 742                              int cpu,
 743                              u64 timestamp,
 744                              struct thread *thread);
 745
 746         void (*fork_event)(struct trace_fork_event *,
 747                            struct event *,
 748                            int cpu,
 749                            u64 timestamp,
 750                            struct thread *thread);
 751
 752         void (*migrate_task_event)(struct trace_migrate_task_event *,
 753                            struct perf_session *session,
 754                            struct event *,
 755                            int cpu,
 756                            u64 timestamp,
 757                            struct thread *thread);
 758 };
 759
 760
 761 static void
 762 replay_wakeup_event(struct trace_wakeup_event *wakeup_event,
 763                     struct perf_session *session __used,
 764                     struct event *event,
 765                     int cpu __used,
 766                     u64 timestamp __used,
 767                     struct thread *thread __used)
 768 {
 769         struct task_desc *waker, *wakee;
 770
 771         if (verbose) {
 772                 printf("sched_wakeup event %p\n", event);
 773
 774                 printf(" ... pid %d woke up %s/%d\n",
 775                         wakeup_event->common_pid,
 776                         wakeup_event->comm,
 777                         wakeup_event->pid);
 778         }
 779
 780         waker = register_pid(wakeup_event->common_pid, "<unknown>");
 781         wakee = register_pid(wakeup_event->pid, wakeup_event->comm);
 782
 783         add_sched_event_wakeup(waker, timestamp, wakee);
 784 }
 785
 786 static u64 cpu_last_switched[MAX_CPUS];
 787
 788 static void
 789 replay_switch_event(struct trace_switch_event *switch_event,
 790                     struct perf_session *session __used,
 791                     struct event *event,
 792                     int cpu,
 793                     u64 timestamp,
 794                     struct thread *thread __used)
 795 {
 796         struct task_desc *prev, __used *next;
 797         u64 timestamp0;
 798         s64 delta;
 799
 800         if (verbose)
 801                 printf("sched_switch event %p\n", event);
 802
 803         if (cpu >= MAX_CPUS || cpu < 0)
 804                 return;
 805
 806         timestamp0 = cpu_last_switched[cpu];
 807         if (timestamp0)
 808                 delta = timestamp - timestamp0;
 809         else
 810                 delta = 0;
 811
 812         if (delta < 0)
 813                 die("hm, delta: %" PRIu64 " < 0 ?\n", delta);
 814
 815         if (verbose) {
 816                 printf(" ... switch from %s/%d to %s/%d [ran %" PRIu64 " nsecs]\n",
 817                         switch_event->prev_comm, switch_event->prev_pid,
 818                         switch_event->next_comm, switch_event->next_pid,
 819                         delta);
 820         }
 821
 822         prev = register_pid(switch_event->prev_pid, switch_event->prev_comm);
 823         next = register_pid(switch_event->next_pid, switch_event->next_comm);
 824
 825         cpu_last_switched[cpu] = timestamp;
 826
 827         add_sched_event_run(prev, timestamp, delta);
 828         add_sched_event_sleep(prev, timestamp, switch_event->prev_state);
 829 }
 830
 831
 832 static void
 833 replay_fork_event(struct trace_fork_event *fork_event,
 834                   struct event *event,
 835                   int cpu __used,
 836                   u64 timestamp __used,
 837                   struct thread *thread __used)
 838 {
 839         if (verbose) {
 840                 printf("sched_fork event %p\n", event);
 841                 printf("... parent: %s/%d\n", fork_event->parent_comm, fork_event->parent_pid);
 842                 printf("...  child: %s/%d\n", fork_event->child_comm, fork_event->child_pid);
 843         }
 844         register_pid(fork_event->parent_pid, fork_event->parent_comm);
 845         register_pid(fork_event->child_pid, fork_event->child_comm);
 846 }
 847
 848 static struct trace_sched_handler replay_ops  = {
 849         .wakeup_event           = replay_wakeup_event,
 850         .switch_event           = replay_switch_event,
 851         .fork_event             = replay_fork_event,
 852 };
 853
 854 struct sort_dimension {
 855         const char              *name;
 856         sort_fn_t               cmp;
 857         struct list_head        list;
 858 };
 859
 860 static LIST_HEAD(cmp_pid);
 861
 862 static int
 863 thread_lat_cmp(struct list_head *list, struct work_atoms *l, struct work_atoms *r)
 864 {
 865         struct sort_dimension *sort;
 866         int ret = 0;
 867
 868         BUG_ON(list_empty(list));
 869
 870         list_for_each_entry(sort, list, list) {
 871                 ret = sort->cmp(l, r);
 872                 if (ret)
 873                         return ret;
 874         }
 875
 876         return ret;
 877 }
 878
 879 static struct work_atoms *
 880 thread_atoms_search(struct rb_root *root, struct thread *thread,
 881                          struct list_head *sort_list)
 882 {
 883         struct rb_node *node = root->rb_node;
 884         struct work_atoms key = { .thread = thread };
 885
 886         while (node) {
 887                 struct work_atoms *atoms;
 888                 int cmp;
 889
 890                 atoms = container_of(node, struct work_atoms, node);
 891
 892                 cmp = thread_lat_cmp(sort_list, &key, atoms);
 893                 if (cmp > 0)
 894                         node = node->rb_left;
 895                 else if (cmp < 0)
 896                         node = node->rb_right;
 897                 else {
 898                         BUG_ON(thread != atoms->thread);
 899                         return atoms;
 900                 }
 901         }
 902         return NULL;
 903 }
 904
 905 static void
 906 __thread_latency_insert(struct rb_root *root, struct work_atoms *data,
 907                          struct list_head *sort_list)
 908 {
 909         struct rb_node **new = &(root->rb_node), *parent = NULL;
 910
 911         while (*new) {
 912                 struct work_atoms *this;
 913                 int cmp;
 914
 915                 this = container_of(*new, struct work_atoms, node);
 916                 parent = *new;
 917
 918                 cmp = thread_lat_cmp(sort_list, data, this);
 919
 920                 if (cmp > 0)
 921                         new = &((*new)->rb_left);
 922                 else
 923                         new = &((*new)->rb_right);
 924         }
 925
 926         rb_link_node(&data->node, parent, new);
 927         rb_insert_color(&data->node, root);
 928 }
 929
 930 static void thread_atoms_insert(struct thread *thread)
 931 {
 932         struct work_atoms *atoms = zalloc(sizeof(*atoms));
 933         if (!atoms)
 934                 die("No memory");
 935
 936         atoms->thread = thread;
 937         INIT_LIST_HEAD(&atoms->work_list);
 938         __thread_latency_insert(&atom_root, atoms, &cmp_pid);
 939 }
 940
 941 static void
 942 latency_fork_event(struct trace_fork_event *fork_event __used,
 943                    struct event *event __used,
 944                    int cpu __used,
 945                    u64 timestamp __used,
 946                    struct thread *thread __used)
 947 {
 948         /* should insert the newcomer */
 949 }
 950
 951 __used
 952 static char sched_out_state(struct trace_switch_event *switch_event)
 953 {
 954         const char *str = TASK_STATE_TO_CHAR_STR;
 955
 956         return str[switch_event->prev_state];
 957 }
 958
 959 static void
 960 add_sched_out_event(struct work_atoms *atoms,
 961                     char run_state,
 962                     u64 timestamp)
 963 {
 964         struct work_atom *atom = zalloc(sizeof(*atom));
 965         if (!atom)
 966                 die("Non memory");
 967
 968         atom->sched_out_time = timestamp;
 969
 970         if (run_state == 'R') {
 971                 atom->state = THREAD_WAIT_CPU;
 972                 atom->wake_up_time = atom->sched_out_time;
 973         }
 974
 975         list_add_tail(&atom->list, &atoms->work_list);
 976 }
 977
 978 static void
 979 add_runtime_event(struct work_atoms *atoms, u64 delta, u64 timestamp __used)
 980 {
 981         struct work_atom *atom;
 982
 983         BUG_ON(list_empty(&atoms->work_list));
 984
 985         atom = list_entry(atoms->work_list.prev, struct work_atom, list);
 986
 987         atom->runtime += delta;
 988         atoms->total_runtime += delta;
 989 }
 990
 991 static void
 992 add_sched_in_event(struct work_atoms *atoms, u64 timestamp)
 993 {
 994         struct work_atom *atom;
 995         u64 delta;
 996
 997         if (list_empty(&atoms->work_list))
 998                 return;
 999
1000         atom = list_entry(atoms->work_list.prev, struct work_atom, list);
1001
1002         if (atom->state != THREAD_WAIT_CPU)
1003                 return;
1004
1005         if (timestamp < atom->wake_up_time) {
1006                 atom->state = THREAD_IGNORE;
1007                 return;
1008         }
1009
1010         atom->state = THREAD_SCHED_IN;
1011         atom->sched_in_time = timestamp;
1012
1013         delta = atom->sched_in_time - atom->wake_up_time;
1014         atoms->total_lat += delta;
1015         if (delta > atoms->max_lat) {
1016                 atoms->max_lat = delta;
1017                 atoms->max_lat_at = timestamp;
1018         }
1019         atoms->nb_atoms++;
1020 }
1021
1022 static void
1023 latency_switch_event(struct trace_switch_event *switch_event,
1024                      struct perf_session *session,
1025                      struct event *event __used,
1026                      int cpu,
1027                      u64 timestamp,
1028                      struct thread *thread __used)
1029 {
1030         struct work_atoms *out_events, *in_events;
1031         struct thread *sched_out, *sched_in;
1032         u64 timestamp0;
1033         s64 delta;
1034
1035         BUG_ON(cpu >= MAX_CPUS || cpu < 0);
1036
1037         timestamp0 = cpu_last_switched[cpu];
1038         cpu_last_switched[cpu] = timestamp;
1039         if (timestamp0)
1040                 delta = timestamp - timestamp0;
1041         else
1042                 delta = 0;
1043
1044         if (delta < 0)
1045                 die("hm, delta: %" PRIu64 " < 0 ?\n", delta);
1046
1047
1048         sched_out = perf_session__findnew(session, switch_event->prev_pid);
1049         sched_in = perf_session__findnew(session, switch_event->next_pid);
1050
1051         out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid);
1052         if (!out_events) {
1053                 thread_atoms_insert(sched_out);
1054                 out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid);
1055                 if (!out_events)
1056                         die("out-event: Internal tree error");
1057         }
1058         add_sched_out_event(out_events, sched_out_state(switch_event), timestamp);
1059
1060         in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid);
1061         if (!in_events) {
1062                 thread_atoms_insert(sched_in);
1063                 in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid);
1064                 if (!in_events)
1065                         die("in-event: Internal tree error");
1066                 /*
1067                  * Take came in we have not heard about yet,
1068                  * add in an initial atom in runnable state:
1069                  */
1070                 add_sched_out_event(in_events, 'R', timestamp);
1071         }
1072         add_sched_in_event(in_events, timestamp);
1073 }
1074
1075 static void
1076 latency_runtime_event(struct trace_runtime_event *runtime_event,
1077                      struct perf_session *session,
1078                      struct event *event __used,
1079                      int cpu,
1080                      u64 timestamp,
1081                      struct thread *this_thread __used)
1082 {
1083         struct thread *thread = perf_session__findnew(session, runtime_event->pid);
1084         struct work_atoms *atoms = thread_atoms_search(&atom_root, thread, &cmp_pid);
1085
1086         BUG_ON(cpu >= MAX_CPUS || cpu < 0);
1087         if (!atoms) {
1088                 thread_atoms_insert(thread);
1089                 atoms = thread_atoms_search(&atom_root, thread, &cmp_pid);
1090                 if (!atoms)
1091                         die("in-event: Internal tree error");
1092                 add_sched_out_event(atoms, 'R', timestamp);
1093         }
1094
1095         add_runtime_event(atoms, runtime_event->runtime, timestamp);
1096 }
1097
1098 static void
1099 latency_wakeup_event(struct trace_wakeup_event *wakeup_event,
1100                      struct perf_session *session,
1101                      struct event *__event __used,
1102                      int cpu __used,
1103                      u64 timestamp,
1104                      struct thread *thread __used)
1105 {
1106         struct work_atoms *atoms;
1107         struct work_atom *atom;
1108         struct thread *wakee;
1109
1110         /* Note for later, it may be interesting to observe the failing cases */
1111         if (!wakeup_event->success)
1112                 return;
1113
1114         wakee = perf_session__findnew(session, wakeup_event->pid);
1115         atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid);
1116         if (!atoms) {
1117                 thread_atoms_insert(wakee);
1118                 atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid);
1119                 if (!atoms)
1120                         die("wakeup-event: Internal tree error");
1121                 add_sched_out_event(atoms, 'S', timestamp);
1122         }
1123
1124         BUG_ON(list_empty(&atoms->work_list));
1125
1126         atom = list_entry(atoms->work_list.prev, struct work_atom, list);
1127
1128         /*
1129          * You WILL be missing events if you've recorded only
1130          * one CPU, or are only looking at only one, so don't
1131          * make useless noise.
1132          */
1133         if (profile_cpu == -1 && atom->state != THREAD_SLEEPING)
1134                 nr_state_machine_bugs++;
1135
1136         nr_timestamps++;
1137         if (atom->sched_out_time > timestamp) {
1138                 nr_unordered_timestamps++;
1139                 return;
1140         }
1141
1142         atom->state = THREAD_WAIT_CPU;
1143         atom->wake_up_time = timestamp;
1144 }
1145
1146 static void
1147 latency_migrate_task_event(struct trace_migrate_task_event *migrate_task_event,
1148                      struct perf_session *session,
1149                      struct event *__event __used,
1150                      int cpu __used,
1151                      u64 timestamp,
1152                      struct thread *thread __used)
1153 {
1154         struct work_atoms *atoms;
1155         struct work_atom *atom;
1156         struct thread *migrant;
1157
1158         /*
1159          * Only need to worry about migration when profiling one CPU.
1160          */
1161         if (profile_cpu == -1)
1162                 return;
1163
1164         migrant = perf_session__findnew(session, migrate_task_event->pid);
1165         atoms = thread_atoms_search(&atom_root, migrant, &cmp_pid);
1166         if (!atoms) {
1167                 thread_atoms_insert(migrant);
1168                 register_pid(migrant->pid, migrant->comm);
1169                 atoms = thread_atoms_search(&atom_root, migrant, &cmp_pid);
1170                 if (!atoms)
1171                         die("migration-event: Internal tree error");
1172                 add_sched_out_event(atoms, 'R', timestamp);
1173         }
1174
1175         BUG_ON(list_empty(&atoms->work_list));
1176
1177         atom = list_entry(atoms->work_list.prev, struct work_atom, list);
1178         atom->sched_in_time = atom->sched_out_time = atom->wake_up_time = timestamp;
1179
1180         nr_timestamps++;
1181
1182         if (atom->sched_out_time > timestamp)
1183                 nr_unordered_timestamps++;
1184 }
1185
1186 static struct trace_sched_handler lat_ops  = {
1187         .wakeup_event           = latency_wakeup_event,
1188         .switch_event           = latency_switch_event,
1189         .runtime_event          = latency_runtime_event,
1190         .fork_event             = latency_fork_event,
1191         .migrate_task_event     = latency_migrate_task_event,
1192 };
1193
1194 static void output_lat_thread(struct work_atoms *work_list)
1195 {
1196         int i;
1197         int ret;
1198         u64 avg;
1199
1200         if (!work_list->nb_atoms)
1201                 return;
1202         /*
1203          * Ignore idle threads:
1204          */
1205         if (!strcmp(work_list->thread->comm, "swapper"))
1206                 return;
1207
1208         all_runtime += work_list->total_runtime;
1209         all_count += work_list->nb_atoms;
1210
1211         ret = printf("  %s:%d ", work_list->thread->comm, work_list->thread->pid);
1212
1213         for (i = 0; i < 24 - ret; i++)
1214                 printf(" ");
1215
1216         avg = work_list->total_lat / work_list->nb_atoms;
1217
1218         printf("|%11.3f ms |%9" PRIu64 " | avg:%9.3f ms | max:%9.3f ms | max at: %9.6f s\n",
1219               (double)work_list->total_runtime / 1e6,
1220                  work_list->nb_atoms, (double)avg / 1e6,
1221                  (double)work_list->max_lat / 1e6,
1222                  (double)work_list->max_lat_at / 1e9);
1223 }
1224
1225 static int pid_cmp(struct work_atoms *l, struct work_atoms *r)
1226 {
1227         if (l->thread->pid < r->thread->pid)
1228                 return -1;
1229         if (l->thread->pid > r->thread->pid)
1230                 return 1;
1231
1232         return 0;
1233 }
1234
1235 static struct sort_dimension pid_sort_dimension = {
1236         .name                   = "pid",
1237         .cmp                    = pid_cmp,
1238 };
1239
1240 static int avg_cmp(struct work_atoms *l, struct work_atoms *r)
1241 {
1242         u64 avgl, avgr;
1243
1244         if (!l->nb_atoms)
1245                 return -1;
1246
1247         if (!r->nb_atoms)
1248                 return 1;
1249
1250         avgl = l->total_lat / l->nb_atoms;
1251         avgr = r->total_lat / r->nb_atoms;
1252
1253         if (avgl < avgr)
1254                 return -1;
1255         if (avgl > avgr)
1256                 return 1;
1257
1258         return 0;
1259 }
1260
1261 static struct sort_dimension avg_sort_dimension = {
1262         .name                   = "avg",
1263         .cmp                    = avg_cmp,
1264 };
1265
1266 static int max_cmp(struct work_atoms *l, struct work_atoms *r)
1267 {
1268         if (l->max_lat < r->max_lat)
1269                 return -1;
1270         if (l->max_lat > r->max_lat)
1271                 return 1;
1272
1273         return 0;
1274 }
1275
1276 static struct sort_dimension max_sort_dimension = {
1277         .name                   = "max",
1278         .cmp                    = max_cmp,
1279 };
1280
1281 static int switch_cmp(struct work_atoms *l, struct work_atoms *r)
1282 {
1283         if (l->nb_atoms < r->nb_atoms)
1284                 return -1;
1285         if (l->nb_atoms > r->nb_atoms)
1286                 return 1;
1287
1288         return 0;
1289 }
1290
1291 static struct sort_dimension switch_sort_dimension = {
1292         .name                   = "switch",
1293         .cmp                    = switch_cmp,
1294 };
1295
1296 static int runtime_cmp(struct work_atoms *l, struct work_atoms *r)
1297 {
1298         if (l->total_runtime < r->total_runtime)
1299                 return -1;
1300         if (l->total_runtime > r->total_runtime)
1301                 return 1;
1302
1303         return 0;
1304 }
1305
1306 static struct sort_dimension runtime_sort_dimension = {
1307         .name                   = "runtime",
1308         .cmp                    = runtime_cmp,
1309 };
1310
1311 static struct sort_dimension *available_sorts[] = {
1312         &pid_sort_dimension,
1313         &avg_sort_dimension,
1314         &max_sort_dimension,
1315         &switch_sort_dimension,
1316         &runtime_sort_dimension,
1317 };
1318
1319 #define NB_AVAILABLE_SORTS      (int)(sizeof(available_sorts) / sizeof(struct sort_dimension *))
1320
1321 static LIST_HEAD(sort_list);
1322
1323 static int sort_dimension__add(const char *tok, struct list_head *list)
1324 {
1325         int i;
1326
1327         for (i = 0; i < NB_AVAILABLE_SORTS; i++) {
1328                 if (!strcmp(available_sorts[i]->name, tok)) {
1329                         list_add_tail(&available_sorts[i]->list, list);
1330
1331                         return 0;
1332                 }
1333         }
1334
1335         return -1;
1336 }
1337
1338 static void setup_sorting(void);
1339
1340 static void sort_lat(void)
1341 {
1342         struct rb_node *node;
1343
1344         for (;;) {
1345                 struct work_atoms *data;
1346                 node = rb_first(&atom_root);
1347                 if (!node)
1348                         break;
1349
1350                 rb_erase(node, &atom_root);
1351                 data = rb_entry(node, struct work_atoms, node);
1352                 __thread_latency_insert(&sorted_atom_root, data, &sort_list);
1353         }
1354 }
1355
1356 static struct trace_sched_handler *trace_handler;
1357
1358 static void
1359 process_sched_wakeup_event(void *data, struct perf_session *session,
1360                            struct event *event,
1361                            int cpu __used,
1362                            u64 timestamp __used,
1363                            struct thread *thread __used)
1364 {
1365         struct trace_wakeup_event wakeup_event;
1366
1367         FILL_COMMON_FIELDS(wakeup_event, event, data);
1368
1369         FILL_ARRAY(wakeup_event, comm, event, data);
1370         FILL_FIELD(wakeup_event, pid, event, data);
1371         FILL_FIELD(wakeup_event, prio, event, data);
1372         FILL_FIELD(wakeup_event, success, event, data);
1373         FILL_FIELD(wakeup_event, cpu, event, data);
1374
1375         if (trace_handler->wakeup_event)
1376                 trace_handler->wakeup_event(&wakeup_event, session, event,
1377                                             cpu, timestamp, thread);
1378 }
1379
1380 /*
1381  * Track the current task - that way we can know whether there's any
1382  * weird events, such as a task being switched away that is not current.
1383  */
1384 static int max_cpu;
1385
1386 static u32 curr_pid[MAX_CPUS] = { [0 ... MAX_CPUS-1] = -1 };
1387
1388 static struct thread *curr_thread[MAX_CPUS];
1389
1390 static char next_shortname1 = 'A';
1391 static char next_shortname2 = '0';
1392
1393 static void
1394 map_switch_event(struct trace_switch_event *switch_event,
1395                  struct perf_session *session,
1396                  struct event *event __used,
1397                  int this_cpu,
1398                  u64 timestamp,
1399                  struct thread *thread __used)
1400 {
1401         struct thread *sched_out __used, *sched_in;
1402         int new_shortname;
1403         u64 timestamp0;
1404         s64 delta;
1405         int cpu;
1406
1407         BUG_ON(this_cpu >= MAX_CPUS || this_cpu < 0);
1408
1409         if (this_cpu > max_cpu)
1410                 max_cpu = this_cpu;
1411
1412         timestamp0 = cpu_last_switched[this_cpu];
1413         cpu_last_switched[this_cpu] = timestamp;
1414         if (timestamp0)
1415                 delta = timestamp - timestamp0;
1416         else
1417                 delta = 0;
1418
1419         if (delta < 0)
1420                 die("hm, delta: %" PRIu64 " < 0 ?\n", delta);
1421
1422
1423         sched_out = perf_session__findnew(session, switch_event->prev_pid);
1424         sched_in = perf_session__findnew(session, switch_event->next_pid);
1425
1426         curr_thread[this_cpu] = sched_in;
1427
1428         printf("  ");
1429
1430         new_shortname = 0;
1431         if (!sched_in->shortname[0]) {
1432                 sched_in->shortname[0] = next_shortname1;
1433                 sched_in->shortname[1] = next_shortname2;
1434
1435                 if (next_shortname1 < 'Z') {
1436                         next_shortname1++;
1437                 } else {
1438                         next_shortname1='A';
1439                         if (next_shortname2 < '9') {
1440                                 next_shortname2++;
1441                         } else {
1442                                 next_shortname2='0';
1443                         }
1444                 }
1445                 new_shortname = 1;
1446         }
1447
1448         for (cpu = 0; cpu <= max_cpu; cpu++) {
1449                 if (cpu != this_cpu)
1450                         printf(" ");
1451                 else
1452                         printf("*");
1453
1454                 if (curr_thread[cpu]) {
1455                         if (curr_thread[cpu]->pid)
1456                                 printf("%2s ", curr_thread[cpu]->shortname);
1457                         else
1458                                 printf(".  ");
1459                 } else
1460                         printf("   ");
1461         }
1462
1463         printf("  %12.6f secs ", (double)timestamp/1e9);
1464         if (new_shortname) {
1465                 printf("%s => %s:%d\n",
1466                         sched_in->shortname, sched_in->comm, sched_in->pid);
1467         } else {
1468                 printf("\n");
1469         }
1470 }
1471
1472
1473 static void
1474 process_sched_switch_event(void *data, struct perf_session *session,
1475                            struct event *event,
1476                            int this_cpu,
1477                            u64 timestamp __used,
1478                            struct thread *thread __used)
1479 {
1480         struct trace_switch_event switch_event;
1481
1482         FILL_COMMON_FIELDS(switch_event, event, data);
1483
1484         FILL_ARRAY(switch_event, prev_comm, event, data);
1485         FILL_FIELD(switch_event, prev_pid, event, data);
1486         FILL_FIELD(switch_event, prev_prio, event, data);
1487         FILL_FIELD(switch_event, prev_state, event, data);
1488         FILL_ARRAY(switch_event, next_comm, event, data);
1489         FILL_FIELD(switch_event, next_pid, event, data);
1490         FILL_FIELD(switch_event, next_prio, event, data);
1491
1492         if (curr_pid[this_cpu] != (u32)-1) {
1493                 /*
1494                  * Are we trying to switch away a PID that is
1495                  * not current?
1496                  */
1497                 if (curr_pid[this_cpu] != switch_event.prev_pid)
1498                         nr_context_switch_bugs++;
1499         }
1500         if (trace_handler->switch_event)
1501                 trace_handler->switch_event(&switch_event, session, event,
1502                                             this_cpu, timestamp, thread);
1503
1504         curr_pid[this_cpu] = switch_event.next_pid;
1505 }
1506
1507 static void
1508 process_sched_runtime_event(void *data, struct perf_session *session,
1509                            struct event *event,
1510                            int cpu __used,
1511                            u64 timestamp __used,
1512                            struct thread *thread __used)
1513 {
1514         struct trace_runtime_event runtime_event;
1515
1516         FILL_ARRAY(runtime_event, comm, event, data);
1517         FILL_FIELD(runtime_event, pid, event, data);
1518         FILL_FIELD(runtime_event, runtime, event, data);
1519         FILL_FIELD(runtime_event, vruntime, event, data);
1520
1521         if (trace_handler->runtime_event)
1522                 trace_handler->runtime_event(&runtime_event, session, event, cpu, timestamp, thread);
1523 }
1524
1525 static void
1526 process_sched_fork_event(void *data,
1527                          struct event *event,
1528                          int cpu __used,
1529                          u64 timestamp __used,
1530                          struct thread *thread __used)
1531 {
1532         struct trace_fork_event fork_event;
1533
1534         FILL_COMMON_FIELDS(fork_event, event, data);
1535
1536         FILL_ARRAY(fork_event, parent_comm, event, data);
1537         FILL_FIELD(fork_event, parent_pid, event, data);
1538         FILL_ARRAY(fork_event, child_comm, event, data);
1539         FILL_FIELD(fork_event, child_pid, event, data);
1540
1541         if (trace_handler->fork_event)
1542                 trace_handler->fork_event(&fork_event, event,
1543                                           cpu, timestamp, thread);
1544 }
1545
1546 static void
1547 process_sched_exit_event(struct event *event,
1548                          int cpu __used,
1549                          u64 timestamp __used,
1550                          struct thread *thread __used)
1551 {
1552         if (verbose)
1553                 printf("sched_exit event %p\n", event);
1554 }
1555
1556 static void
1557 process_sched_migrate_task_event(void *data, struct perf_session *session,
1558                            struct event *event,
1559                            int cpu __used,
1560                            u64 timestamp __used,
1561                            struct thread *thread __used)
1562 {
1563         struct trace_migrate_task_event migrate_task_event;
1564
1565         FILL_COMMON_FIELDS(migrate_task_event, event, data);
1566
1567         FILL_ARRAY(migrate_task_event, comm, event, data);
1568         FILL_FIELD(migrate_task_event, pid, event, data);
1569         FILL_FIELD(migrate_task_event, prio, event, data);
1570         FILL_FIELD(migrate_task_event, cpu, event, data);
1571
1572         if (trace_handler->migrate_task_event)
1573                 trace_handler->migrate_task_event(&migrate_task_event, session,
1574                                                  event, cpu, timestamp, thread);
1575 }
1576
1577 static void process_raw_event(union perf_event *raw_event __used,
1578                               struct perf_session *session, void *data, int cpu,
1579                               u64 timestamp, struct thread *thread)
1580 {
1581         struct event *event;
1582         int type;
1583
1584
1585         type = trace_parse_common_type(data);
1586         event = trace_find_event(type);
1587
1588         if (!strcmp(event->name, "sched_switch"))
1589                 process_sched_switch_event(data, session, event, cpu, timestamp, thread);
1590         if (!strcmp(event->name, "sched_stat_runtime"))
1591                 process_sched_runtime_event(data, session, event, cpu, timestamp, thread);
1592         if (!strcmp(event->name, "sched_wakeup"))
1593                 process_sched_wakeup_event(data, session, event, cpu, timestamp, thread);
1594         if (!strcmp(event->name, "sched_wakeup_new"))
1595                 process_sched_wakeup_event(data, session, event, cpu, timestamp, thread);
1596         if (!strcmp(event->name, "sched_process_fork"))
1597                 process_sched_fork_event(data, event, cpu, timestamp, thread);
1598         if (!strcmp(event->name, "sched_process_exit"))
1599                 process_sched_exit_event(event, cpu, timestamp, thread);
1600         if (!strcmp(event->name, "sched_migrate_task"))
1601                 process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread);
1602 }
1603
1604 static int process_sample_event(union perf_event *event,
1605                                 struct perf_sample *sample,
1606                                 struct perf_evsel *evsel __used,
1607                                 struct perf_session *session)
1608 {
1609         struct thread *thread;
1610
1611         if (!(session->sample_type & PERF_SAMPLE_RAW))
1612                 return 0;
1613
1614         thread = perf_session__findnew(session, sample->pid);
1615         if (thread == NULL) {
1616                 pr_debug("problem processing %d event, skipping it.\n",
1617                          event->header.type);
1618                 return -1;
1619         }
1620
1621         dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
1622
1623         if (profile_cpu != -1 && profile_cpu != (int)sample->cpu)
1624                 return 0;
1625
1626         process_raw_event(event, session, sample->raw_data, sample->cpu,
1627                           sample->time, thread);
1628
1629         return 0;
1630 }
1631
1632 static struct perf_event_ops event_ops = {
1633         .sample                 = process_sample_event,
1634         .comm                   = perf_event__process_comm,
1635         .lost                   = perf_event__process_lost,
1636         .fork                   = perf_event__process_task,
1637         .ordered_samples        = true,
1638 };
1639
1640 static void read_events(bool destroy, struct perf_session **psession)
1641 {
1642         int err = -EINVAL;
1643         struct perf_session *session = perf_session__new(input_name, O_RDONLY,
1644                                                          0, false, &event_ops);
1645         if (session == NULL)
1646                 die("No Memory");
1647
1648         if (perf_session__has_traces(session, "record -R")) {
1649                 err = perf_session__process_events(session, &event_ops);
1650                 if (err)
1651                         die("Failed to process events, error %d", err);
1652
1653                 nr_events      = session->hists.stats.nr_events[0];
1654                 nr_lost_events = session->hists.stats.total_lost;
1655                 nr_lost_chunks = session->hists.stats.nr_events[PERF_RECORD_LOST];
1656         }
1657
1658         if (destroy)
1659                 perf_session__delete(session);
1660
1661         if (psession)
1662                 *psession = session;
1663 }
1664
1665 static void print_bad_events(void)
1666 {
1667         if (nr_unordered_timestamps && nr_timestamps) {
1668                 printf("  INFO: %.3f%% unordered timestamps (%ld out of %ld)\n",
1669                         (double)nr_unordered_timestamps/(double)nr_timestamps*100.0,
1670                         nr_unordered_timestamps, nr_timestamps);
1671         }
1672         if (nr_lost_events && nr_events) {
1673                 printf("  INFO: %.3f%% lost events (%ld out of %ld, in %ld chunks)\n",
1674                         (double)nr_lost_events/(double)nr_events*100.0,
1675                         nr_lost_events, nr_events, nr_lost_chunks);
1676         }
1677         if (nr_state_machine_bugs && nr_timestamps) {
1678                 printf("  INFO: %.3f%% state machine bugs (%ld out of %ld)",
1679                         (double)nr_state_machine_bugs/(double)nr_timestamps*100.0,
1680                         nr_state_machine_bugs, nr_timestamps);
1681                 if (nr_lost_events)
1682                         printf(" (due to lost events?)");
1683                 printf("\n");
1684         }
1685         if (nr_context_switch_bugs && nr_timestamps) {
1686                 printf("  INFO: %.3f%% context switch bugs (%ld out of %ld)",
1687                         (double)nr_context_switch_bugs/(double)nr_timestamps*100.0,
1688                         nr_context_switch_bugs, nr_timestamps);
1689                 if (nr_lost_events)
1690                         printf(" (due to lost events?)");
1691                 printf("\n");
1692         }
1693 }
1694
1695 static void __cmd_lat(void)
1696 {
1697         struct rb_node *next;
1698         struct perf_session *session;
1699
1700         setup_pager();
1701         read_events(false, &session);
1702         sort_lat();
1703
1704         printf("\n ---------------------------------------------------------------------------------------------------------------\n");
1705         printf("  Task                  |   Runtime ms  | Switches | Average delay ms | Maximum delay ms | Maximum delay at     |\n");
1706         printf(" ---------------------------------------------------------------------------------------------------------------\n");
1707
1708         next = rb_first(&sorted_atom_root);
1709
1710         while (next) {
1711                 struct work_atoms *work_list;
1712
1713                 work_list = rb_entry(next, struct work_atoms, node);
1714                 output_lat_thread(work_list);
1715                 next = rb_next(next);
1716         }
1717
1718         printf(" -----------------------------------------------------------------------------------------\n");
1719         printf("  TOTAL:                |%11.3f ms |%9" PRIu64 " |\n",
1720                 (double)all_runtime/1e6, all_count);
1721
1722         printf(" ---------------------------------------------------\n");
1723
1724         print_bad_events();
1725         printf("\n");
1726
1727         perf_session__delete(session);
1728 }
1729
1730 static struct trace_sched_handler map_ops  = {
1731         .wakeup_event           = NULL,
1732         .switch_event           = map_switch_event,
1733         .runtime_event          = NULL,
1734         .fork_event             = NULL,
1735 };
1736
1737 static void __cmd_map(void)
1738 {
1739         max_cpu = sysconf(_SC_NPROCESSORS_CONF);
1740
1741         setup_pager();
1742         read_events(true, NULL);
1743         print_bad_events();
1744 }
1745
1746 static void __cmd_replay(void)
1747 {
1748         unsigned long i;
1749
1750         calibrate_run_measurement_overhead();
1751         calibrate_sleep_measurement_overhead();
1752
1753         test_calibrations();
1754
1755         read_events(true, NULL);
1756
1757         printf("nr_run_events:        %ld\n", nr_run_events);
1758         printf("nr_sleep_events:      %ld\n", nr_sleep_events);
1759         printf("nr_wakeup_events:     %ld\n", nr_wakeup_events);
1760
1761         if (targetless_wakeups)
1762                 printf("target-less wakeups:  %ld\n", targetless_wakeups);
1763         if (multitarget_wakeups)
1764                 printf("multi-target wakeups: %ld\n", multitarget_wakeups);
1765         if (nr_run_events_optimized)
1766                 printf("run atoms optimized: %ld\n",
1767                         nr_run_events_optimized);
1768
1769         print_task_traces();
1770         add_cross_task_wakeups();
1771
1772         create_tasks();
1773         printf("------------------------------------------------------------\n");
1774         for (i = 0; i < replay_repeat; i++)
1775                 run_one_test();
1776 }
1777
1778
1779 static const char * const sched_usage[] = {
1780         "perf sched [<options>] {record|latency|map|replay|script}",
1781         NULL
1782 };
1783
1784 static const struct option sched_options[] = {
1785         OPT_STRING('i', "input", &input_name, "file",
1786                     "input file name"),
1787         OPT_INCR('v', "verbose", &verbose,
1788                     "be more verbose (show symbol address, etc)"),
1789         OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
1790                     "dump raw trace in ASCII"),
1791         OPT_END()
1792 };
1793
1794 static const char * const latency_usage[] = {
1795         "perf sched latency [<options>]",
1796         NULL
1797 };
1798
1799 static const struct option latency_options[] = {
1800         OPT_STRING('s', "sort", &sort_order, "key[,key2...]",
1801                    "sort by key(s): runtime, switch, avg, max"),
1802         OPT_INCR('v', "verbose", &verbose,
1803                     "be more verbose (show symbol address, etc)"),
1804         OPT_INTEGER('C', "CPU", &profile_cpu,
1805                     "CPU to profile on"),
1806         OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
1807                     "dump raw trace in ASCII"),
1808         OPT_END()
1809 };
1810
1811 static const char * const replay_usage[] = {
1812         "perf sched replay [<options>]",
1813         NULL
1814 };
1815
1816 static const struct option replay_options[] = {
1817         OPT_UINTEGER('r', "repeat", &replay_repeat,
1818                      "repeat the workload replay N times (-1: infinite)"),
1819         OPT_INCR('v', "verbose", &verbose,
1820                     "be more verbose (show symbol address, etc)"),
1821         OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
1822                     "dump raw trace in ASCII"),
1823         OPT_END()
1824 };
1825
1826 static void setup_sorting(void)
1827 {
1828         char *tmp, *tok, *str = strdup(sort_order);
1829
1830         for (tok = strtok_r(str, ", ", &tmp);
1831                         tok; tok = strtok_r(NULL, ", ", &tmp)) {
1832                 if (sort_dimension__add(tok, &sort_list) < 0) {
1833                         error("Unknown --sort key: `%s'", tok);
1834                         usage_with_options(latency_usage, latency_options);
1835                 }
1836         }
1837
1838         free(str);
1839
1840         sort_dimension__add("pid", &cmp_pid);
1841 }
1842
1843 static const char *record_args[] = {
1844         "record",
1845         "-a",
1846         "-R",
1847         "-f",
1848         "-m", "1024",
1849         "-c", "1",
1850         "-e", "sched:sched_switch",
1851         "-e", "sched:sched_stat_wait",
1852         "-e", "sched:sched_stat_sleep",
1853         "-e", "sched:sched_stat_iowait",
1854         "-e", "sched:sched_stat_runtime",
1855         "-e", "sched:sched_process_exit",
1856         "-e", "sched:sched_process_fork",
1857         "-e", "sched:sched_wakeup",
1858         "-e", "sched:sched_migrate_task",
1859 };
1860
1861 static int __cmd_record(int argc, const char **argv)
1862 {
1863         unsigned int rec_argc, i, j;
1864         const char **rec_argv;
1865
1866         rec_argc = ARRAY_SIZE(record_args) + argc - 1;
1867         rec_argv = calloc(rec_argc + 1, sizeof(char *));
1868
1869         if (rec_argv == NULL)
1870                 return -ENOMEM;
1871
1872         for (i = 0; i < ARRAY_SIZE(record_args); i++)
1873                 rec_argv[i] = strdup(record_args[i]);
1874
1875         for (j = 1; j < (unsigned int)argc; j++, i++)
1876                 rec_argv[i] = argv[j];
1877
1878         BUG_ON(i != rec_argc);
1879
1880         return cmd_record(i, rec_argv, NULL);
1881 }
1882
1883 int cmd_sched(int argc, const char **argv, const char *prefix __used)
1884 {
1885         argc = parse_options(argc, argv, sched_options, sched_usage,
1886                              PARSE_OPT_STOP_AT_NON_OPTION);
1887         if (!argc)
1888                 usage_with_options(sched_usage, sched_options);
1889
1890         /*
1891          * Aliased to 'perf script' for now:
1892          */
1893         if (!strcmp(argv[0], "script"))
1894                 return cmd_script(argc, argv, prefix);
1895
1896         symbol__init();
1897         if (!strncmp(argv[0], "rec", 3)) {
1898                 return __cmd_record(argc, argv);
1899         } else if (!strncmp(argv[0], "lat", 3)) {
1900                 trace_handler = &lat_ops;
1901                 if (argc > 1) {
1902                         argc = parse_options(argc, argv, latency_options, latency_usage, 0);
1903                         if (argc)
1904                                 usage_with_options(latency_usage, latency_options);
1905                 }
1906                 setup_sorting();
1907                 __cmd_lat();
1908         } else if (!strcmp(argv[0], "map")) {
1909                 trace_handler = &map_ops;
1910                 setup_sorting();
1911                 __cmd_map();
1912         } else if (!strncmp(argv[0], "rep", 3)) {
1913                 trace_handler = &replay_ops;
1914                 if (argc) {
1915                         argc = parse_options(argc, argv, replay_options, replay_usage, 0);
1916                         if (argc)
1917                                 usage_with_options(replay_usage, replay_options);
1918                 }
1919                 __cmd_replay();
1920         } else {
1921                 usage_with_options(sched_usage, sched_options);
1922         }
1923
1924         return 0;
1925 }