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