1 #include "builtin.h" 2 #include "perf.h" 3 4 #include "util/util.h" 5 #include "util/evlist.h" 6 #include "util/cache.h" 7 #include "util/evsel.h" 8 #include "util/symbol.h" 9 #include "util/thread.h" 10 #include "util/header.h" 11 #include "util/session.h" 12 #include "util/tool.h" 13 14 #include "util/parse-options.h" 15 #include "util/trace-event.h" 16 17 #include "util/debug.h" 18 19 #include <sys/prctl.h> 20 #include <sys/resource.h> 21 22 #include <semaphore.h> 23 #include <pthread.h> 24 #include <math.h> 25 26 #define PR_SET_NAME 15 /* Set process name */ 27 #define MAX_CPUS 4096 28 #define COMM_LEN 20 29 #define SYM_LEN 129 30 #define MAX_PID 65536 31 32 struct sched_atom; 33 34 struct task_desc { 35 unsigned long nr; 36 unsigned long pid; 37 char comm[COMM_LEN]; 38 39 unsigned long nr_events; 40 unsigned long curr_event; 41 struct sched_atom **atoms; 42 43 pthread_t thread; 44 sem_t sleep_sem; 45 46 sem_t ready_for_work; 47 sem_t work_done_sem; 48 49 u64 cpu_usage; 50 }; 51 52 enum sched_event_type { 53 SCHED_EVENT_RUN, 54 SCHED_EVENT_SLEEP, 55 SCHED_EVENT_WAKEUP, 56 SCHED_EVENT_MIGRATION, 57 }; 58 59 struct sched_atom { 60 enum sched_event_type type; 61 int specific_wait; 62 u64 timestamp; 63 u64 duration; 64 unsigned long nr; 65 sem_t *wait_sem; 66 struct task_desc *wakee; 67 }; 68 69 #define TASK_STATE_TO_CHAR_STR "RSDTtZX" 70 71 enum thread_state { 72 THREAD_SLEEPING = 0, 73 THREAD_WAIT_CPU, 74 THREAD_SCHED_IN, 75 THREAD_IGNORE 76 }; 77 78 struct work_atom { 79 struct list_head list; 80 enum thread_state state; 81 u64 sched_out_time; 82 u64 wake_up_time; 83 u64 sched_in_time; 84 u64 runtime; 85 }; 86 87 struct work_atoms { 88 struct list_head work_list; 89 struct thread *thread; 90 struct rb_node node; 91 u64 max_lat; 92 u64 max_lat_at; 93 u64 total_lat; 94 u64 nb_atoms; 95 u64 total_runtime; 96 }; 97 98 typedef int (*sort_fn_t)(struct work_atoms *, struct work_atoms *); 99 100 struct perf_sched; 101 102 struct trace_sched_handler { 103 int (*switch_event)(struct perf_sched *sched, struct perf_evsel *evsel, 104 struct perf_sample *sample, struct machine *machine); 105 106 int (*runtime_event)(struct perf_sched *sched, struct perf_evsel *evsel, 107 struct perf_sample *sample, struct machine *machine); 108 109 int (*wakeup_event)(struct perf_sched *sched, struct perf_evsel *evsel, 110 struct perf_sample *sample, struct machine *machine); 111 112 int (*fork_event)(struct perf_sched *sched, struct perf_evsel *evsel, 113 struct perf_sample *sample); 114 115 int (*migrate_task_event)(struct perf_sched *sched, 116 struct perf_evsel *evsel, 117 struct perf_sample *sample, 118 struct machine *machine); 119 }; 120 121 struct perf_sched { 122 struct perf_tool tool; 123 const char *input_name; 124 const char *sort_order; 125 unsigned long nr_tasks; 126 struct task_desc *pid_to_task[MAX_PID]; 127 struct task_desc **tasks; 128 const struct trace_sched_handler *tp_handler; 129 pthread_mutex_t start_work_mutex; 130 pthread_mutex_t work_done_wait_mutex; 131 int profile_cpu; 132 /* 133 * Track the current task - that way we can know whether there's any 134 * weird events, such as a task being switched away that is not current. 135 */ 136 int max_cpu; 137 u32 curr_pid[MAX_CPUS]; 138 struct thread *curr_thread[MAX_CPUS]; 139 char next_shortname1; 140 char next_shortname2; 141 unsigned int replay_repeat; 142 unsigned long nr_run_events; 143 unsigned long nr_sleep_events; 144 unsigned long nr_wakeup_events; 145 unsigned long nr_sleep_corrections; 146 unsigned long nr_run_events_optimized; 147 unsigned long targetless_wakeups; 148 unsigned long multitarget_wakeups; 149 unsigned long nr_runs; 150 unsigned long nr_timestamps; 151 unsigned long nr_unordered_timestamps; 152 unsigned long nr_state_machine_bugs; 153 unsigned long nr_context_switch_bugs; 154 unsigned long nr_events; 155 unsigned long nr_lost_chunks; 156 unsigned long nr_lost_events; 157 u64 run_measurement_overhead; 158 u64 sleep_measurement_overhead; 159 u64 start_time; 160 u64 cpu_usage; 161 u64 runavg_cpu_usage; 162 u64 parent_cpu_usage; 163 u64 runavg_parent_cpu_usage; 164 u64 sum_runtime; 165 u64 sum_fluct; 166 u64 run_avg; 167 u64 all_runtime; 168 u64 all_count; 169 u64 cpu_last_switched[MAX_CPUS]; 170 struct rb_root atom_root, sorted_atom_root; 171 struct list_head sort_list, cmp_pid; 172 }; 173 174 static u64 get_nsecs(void) 175 { 176 struct timespec ts; 177 178 clock_gettime(CLOCK_MONOTONIC, &ts); 179 180 return ts.tv_sec * 1000000000ULL + ts.tv_nsec; 181 } 182 183 static void burn_nsecs(struct perf_sched *sched, u64 nsecs) 184 { 185 u64 T0 = get_nsecs(), T1; 186 187 do { 188 T1 = get_nsecs(); 189 } while (T1 + sched->run_measurement_overhead < T0 + nsecs); 190 } 191 192 static void sleep_nsecs(u64 nsecs) 193 { 194 struct timespec ts; 195 196 ts.tv_nsec = nsecs % 999999999; 197 ts.tv_sec = nsecs / 999999999; 198 199 nanosleep(&ts, NULL); 200 } 201 202 static void calibrate_run_measurement_overhead(struct perf_sched *sched) 203 { 204 u64 T0, T1, delta, min_delta = 1000000000ULL; 205 int i; 206 207 for (i = 0; i < 10; i++) { 208 T0 = get_nsecs(); 209 burn_nsecs(sched, 0); 210 T1 = get_nsecs(); 211 delta = T1-T0; 212 min_delta = min(min_delta, delta); 213 } 214 sched->run_measurement_overhead = min_delta; 215 216 printf("run measurement overhead: %" PRIu64 " nsecs\n", min_delta); 217 } 218 219 static void calibrate_sleep_measurement_overhead(struct perf_sched *sched) 220 { 221 u64 T0, T1, delta, min_delta = 1000000000ULL; 222 int i; 223 224 for (i = 0; i < 10; i++) { 225 T0 = get_nsecs(); 226 sleep_nsecs(10000); 227 T1 = get_nsecs(); 228 delta = T1-T0; 229 min_delta = min(min_delta, delta); 230 } 231 min_delta -= 10000; 232 sched->sleep_measurement_overhead = min_delta; 233 234 printf("sleep measurement overhead: %" PRIu64 " nsecs\n", min_delta); 235 } 236 237 static struct sched_atom * 238 get_new_event(struct task_desc *task, u64 timestamp) 239 { 240 struct sched_atom *event = zalloc(sizeof(*event)); 241 unsigned long idx = task->nr_events; 242 size_t size; 243 244 event->timestamp = timestamp; 245 event->nr = idx; 246 247 task->nr_events++; 248 size = sizeof(struct sched_atom *) * task->nr_events; 249 task->atoms = realloc(task->atoms, size); 250 BUG_ON(!task->atoms); 251 252 task->atoms[idx] = event; 253 254 return event; 255 } 256 257 static struct sched_atom *last_event(struct task_desc *task) 258 { 259 if (!task->nr_events) 260 return NULL; 261 262 return task->atoms[task->nr_events - 1]; 263 } 264 265 static void add_sched_event_run(struct perf_sched *sched, struct task_desc *task, 266 u64 timestamp, u64 duration) 267 { 268 struct sched_atom *event, *curr_event = last_event(task); 269 270 /* 271 * optimize an existing RUN event by merging this one 272 * to it: 273 */ 274 if (curr_event && curr_event->type == SCHED_EVENT_RUN) { 275 sched->nr_run_events_optimized++; 276 curr_event->duration += duration; 277 return; 278 } 279 280 event = get_new_event(task, timestamp); 281 282 event->type = SCHED_EVENT_RUN; 283 event->duration = duration; 284 285 sched->nr_run_events++; 286 } 287 288 static void add_sched_event_wakeup(struct perf_sched *sched, struct task_desc *task, 289 u64 timestamp, struct task_desc *wakee) 290 { 291 struct sched_atom *event, *wakee_event; 292 293 event = get_new_event(task, timestamp); 294 event->type = SCHED_EVENT_WAKEUP; 295 event->wakee = wakee; 296 297 wakee_event = last_event(wakee); 298 if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) { 299 sched->targetless_wakeups++; 300 return; 301 } 302 if (wakee_event->wait_sem) { 303 sched->multitarget_wakeups++; 304 return; 305 } 306 307 wakee_event->wait_sem = zalloc(sizeof(*wakee_event->wait_sem)); 308 sem_init(wakee_event->wait_sem, 0, 0); 309 wakee_event->specific_wait = 1; 310 event->wait_sem = wakee_event->wait_sem; 311 312 sched->nr_wakeup_events++; 313 } 314 315 static void add_sched_event_sleep(struct perf_sched *sched, struct task_desc *task, 316 u64 timestamp, u64 task_state __maybe_unused) 317 { 318 struct sched_atom *event = get_new_event(task, timestamp); 319 320 event->type = SCHED_EVENT_SLEEP; 321 322 sched->nr_sleep_events++; 323 } 324 325 static struct task_desc *register_pid(struct perf_sched *sched, 326 unsigned long pid, const char *comm) 327 { 328 struct task_desc *task; 329 330 BUG_ON(pid >= MAX_PID); 331 332 task = sched->pid_to_task[pid]; 333 334 if (task) 335 return task; 336 337 task = zalloc(sizeof(*task)); 338 task->pid = pid; 339 task->nr = sched->nr_tasks; 340 strcpy(task->comm, comm); 341 /* 342 * every task starts in sleeping state - this gets ignored 343 * if there's no wakeup pointing to this sleep state: 344 */ 345 add_sched_event_sleep(sched, task, 0, 0); 346 347 sched->pid_to_task[pid] = task; 348 sched->nr_tasks++; 349 sched->tasks = realloc(sched->tasks, sched->nr_tasks * sizeof(struct task_task *)); 350 BUG_ON(!sched->tasks); 351 sched->tasks[task->nr] = task; 352 353 if (verbose) 354 printf("registered task #%ld, PID %ld (%s)\n", sched->nr_tasks, pid, comm); 355 356 return task; 357 } 358 359 360 static void print_task_traces(struct perf_sched *sched) 361 { 362 struct task_desc *task; 363 unsigned long i; 364 365 for (i = 0; i < sched->nr_tasks; i++) { 366 task = sched->tasks[i]; 367 printf("task %6ld (%20s:%10ld), nr_events: %ld\n", 368 task->nr, task->comm, task->pid, task->nr_events); 369 } 370 } 371 372 static void add_cross_task_wakeups(struct perf_sched *sched) 373 { 374 struct task_desc *task1, *task2; 375 unsigned long i, j; 376 377 for (i = 0; i < sched->nr_tasks; i++) { 378 task1 = sched->tasks[i]; 379 j = i + 1; 380 if (j == sched->nr_tasks) 381 j = 0; 382 task2 = sched->tasks[j]; 383 add_sched_event_wakeup(sched, task1, 0, task2); 384 } 385 } 386 387 static void perf_sched__process_event(struct perf_sched *sched, 388 struct sched_atom *atom) 389 { 390 int ret = 0; 391 392 switch (atom->type) { 393 case SCHED_EVENT_RUN: 394 burn_nsecs(sched, atom->duration); 395 break; 396 case SCHED_EVENT_SLEEP: 397 if (atom->wait_sem) 398 ret = sem_wait(atom->wait_sem); 399 BUG_ON(ret); 400 break; 401 case SCHED_EVENT_WAKEUP: 402 if (atom->wait_sem) 403 ret = sem_post(atom->wait_sem); 404 BUG_ON(ret); 405 break; 406 case SCHED_EVENT_MIGRATION: 407 break; 408 default: 409 BUG_ON(1); 410 } 411 } 412 413 static u64 get_cpu_usage_nsec_parent(void) 414 { 415 struct rusage ru; 416 u64 sum; 417 int err; 418 419 err = getrusage(RUSAGE_SELF, &ru); 420 BUG_ON(err); 421 422 sum = ru.ru_utime.tv_sec*1e9 + ru.ru_utime.tv_usec*1e3; 423 sum += ru.ru_stime.tv_sec*1e9 + ru.ru_stime.tv_usec*1e3; 424 425 return sum; 426 } 427 428 static int self_open_counters(void) 429 { 430 struct perf_event_attr attr; 431 int fd; 432 433 memset(&attr, 0, sizeof(attr)); 434 435 attr.type = PERF_TYPE_SOFTWARE; 436 attr.config = PERF_COUNT_SW_TASK_CLOCK; 437 438 fd = sys_perf_event_open(&attr, 0, -1, -1, 0); 439 440 if (fd < 0) 441 pr_err("Error: sys_perf_event_open() syscall returned " 442 "with %d (%s)\n", fd, strerror(errno)); 443 return fd; 444 } 445 446 static u64 get_cpu_usage_nsec_self(int fd) 447 { 448 u64 runtime; 449 int ret; 450 451 ret = read(fd, &runtime, sizeof(runtime)); 452 BUG_ON(ret != sizeof(runtime)); 453 454 return runtime; 455 } 456 457 struct sched_thread_parms { 458 struct task_desc *task; 459 struct perf_sched *sched; 460 }; 461 462 static void *thread_func(void *ctx) 463 { 464 struct sched_thread_parms *parms = ctx; 465 struct task_desc *this_task = parms->task; 466 struct perf_sched *sched = parms->sched; 467 u64 cpu_usage_0, cpu_usage_1; 468 unsigned long i, ret; 469 char comm2[22]; 470 int fd; 471 472 free(parms); 473 474 sprintf(comm2, ":%s", this_task->comm); 475 prctl(PR_SET_NAME, comm2); 476 fd = self_open_counters(); 477 if (fd < 0) 478 return NULL; 479 again: 480 ret = sem_post(&this_task->ready_for_work); 481 BUG_ON(ret); 482 ret = pthread_mutex_lock(&sched->start_work_mutex); 483 BUG_ON(ret); 484 ret = pthread_mutex_unlock(&sched->start_work_mutex); 485 BUG_ON(ret); 486 487 cpu_usage_0 = get_cpu_usage_nsec_self(fd); 488 489 for (i = 0; i < this_task->nr_events; i++) { 490 this_task->curr_event = i; 491 perf_sched__process_event(sched, this_task->atoms[i]); 492 } 493 494 cpu_usage_1 = get_cpu_usage_nsec_self(fd); 495 this_task->cpu_usage = cpu_usage_1 - cpu_usage_0; 496 ret = sem_post(&this_task->work_done_sem); 497 BUG_ON(ret); 498 499 ret = pthread_mutex_lock(&sched->work_done_wait_mutex); 500 BUG_ON(ret); 501 ret = pthread_mutex_unlock(&sched->work_done_wait_mutex); 502 BUG_ON(ret); 503 504 goto again; 505 } 506 507 static void create_tasks(struct perf_sched *sched) 508 { 509 struct task_desc *task; 510 pthread_attr_t attr; 511 unsigned long i; 512 int err; 513 514 err = pthread_attr_init(&attr); 515 BUG_ON(err); 516 err = pthread_attr_setstacksize(&attr, 517 (size_t) max(16 * 1024, PTHREAD_STACK_MIN)); 518 BUG_ON(err); 519 err = pthread_mutex_lock(&sched->start_work_mutex); 520 BUG_ON(err); 521 err = pthread_mutex_lock(&sched->work_done_wait_mutex); 522 BUG_ON(err); 523 for (i = 0; i < sched->nr_tasks; i++) { 524 struct sched_thread_parms *parms = malloc(sizeof(*parms)); 525 BUG_ON(parms == NULL); 526 parms->task = task = sched->tasks[i]; 527 parms->sched = sched; 528 sem_init(&task->sleep_sem, 0, 0); 529 sem_init(&task->ready_for_work, 0, 0); 530 sem_init(&task->work_done_sem, 0, 0); 531 task->curr_event = 0; 532 err = pthread_create(&task->thread, &attr, thread_func, parms); 533 BUG_ON(err); 534 } 535 } 536 537 static void wait_for_tasks(struct perf_sched *sched) 538 { 539 u64 cpu_usage_0, cpu_usage_1; 540 struct task_desc *task; 541 unsigned long i, ret; 542 543 sched->start_time = get_nsecs(); 544 sched->cpu_usage = 0; 545 pthread_mutex_unlock(&sched->work_done_wait_mutex); 546 547 for (i = 0; i < sched->nr_tasks; i++) { 548 task = sched->tasks[i]; 549 ret = sem_wait(&task->ready_for_work); 550 BUG_ON(ret); 551 sem_init(&task->ready_for_work, 0, 0); 552 } 553 ret = pthread_mutex_lock(&sched->work_done_wait_mutex); 554 BUG_ON(ret); 555 556 cpu_usage_0 = get_cpu_usage_nsec_parent(); 557 558 pthread_mutex_unlock(&sched->start_work_mutex); 559 560 for (i = 0; i < sched->nr_tasks; i++) { 561 task = sched->tasks[i]; 562 ret = sem_wait(&task->work_done_sem); 563 BUG_ON(ret); 564 sem_init(&task->work_done_sem, 0, 0); 565 sched->cpu_usage += task->cpu_usage; 566 task->cpu_usage = 0; 567 } 568 569 cpu_usage_1 = get_cpu_usage_nsec_parent(); 570 if (!sched->runavg_cpu_usage) 571 sched->runavg_cpu_usage = sched->cpu_usage; 572 sched->runavg_cpu_usage = (sched->runavg_cpu_usage * 9 + sched->cpu_usage) / 10; 573 574 sched->parent_cpu_usage = cpu_usage_1 - cpu_usage_0; 575 if (!sched->runavg_parent_cpu_usage) 576 sched->runavg_parent_cpu_usage = sched->parent_cpu_usage; 577 sched->runavg_parent_cpu_usage = (sched->runavg_parent_cpu_usage * 9 + 578 sched->parent_cpu_usage)/10; 579 580 ret = pthread_mutex_lock(&sched->start_work_mutex); 581 BUG_ON(ret); 582 583 for (i = 0; i < sched->nr_tasks; i++) { 584 task = sched->tasks[i]; 585 sem_init(&task->sleep_sem, 0, 0); 586 task->curr_event = 0; 587 } 588 } 589 590 static void run_one_test(struct perf_sched *sched) 591 { 592 u64 T0, T1, delta, avg_delta, fluct; 593 594 T0 = get_nsecs(); 595 wait_for_tasks(sched); 596 T1 = get_nsecs(); 597 598 delta = T1 - T0; 599 sched->sum_runtime += delta; 600 sched->nr_runs++; 601 602 avg_delta = sched->sum_runtime / sched->nr_runs; 603 if (delta < avg_delta) 604 fluct = avg_delta - delta; 605 else 606 fluct = delta - avg_delta; 607 sched->sum_fluct += fluct; 608 if (!sched->run_avg) 609 sched->run_avg = delta; 610 sched->run_avg = (sched->run_avg * 9 + delta) / 10; 611 612 printf("#%-3ld: %0.3f, ", sched->nr_runs, (double)delta / 1000000.0); 613 614 printf("ravg: %0.2f, ", (double)sched->run_avg / 1e6); 615 616 printf("cpu: %0.2f / %0.2f", 617 (double)sched->cpu_usage / 1e6, (double)sched->runavg_cpu_usage / 1e6); 618 619 #if 0 620 /* 621 * rusage statistics done by the parent, these are less 622 * accurate than the sched->sum_exec_runtime based statistics: 623 */ 624 printf(" [%0.2f / %0.2f]", 625 (double)sched->parent_cpu_usage/1e6, 626 (double)sched->runavg_parent_cpu_usage/1e6); 627 #endif 628 629 printf("\n"); 630 631 if (sched->nr_sleep_corrections) 632 printf(" (%ld sleep corrections)\n", sched->nr_sleep_corrections); 633 sched->nr_sleep_corrections = 0; 634 } 635 636 static void test_calibrations(struct perf_sched *sched) 637 { 638 u64 T0, T1; 639 640 T0 = get_nsecs(); 641 burn_nsecs(sched, 1e6); 642 T1 = get_nsecs(); 643 644 printf("the run test took %" PRIu64 " nsecs\n", T1 - T0); 645 646 T0 = get_nsecs(); 647 sleep_nsecs(1e6); 648 T1 = get_nsecs(); 649 650 printf("the sleep test took %" PRIu64 " nsecs\n", T1 - T0); 651 } 652 653 static int 654 replay_wakeup_event(struct perf_sched *sched, 655 struct perf_evsel *evsel, struct perf_sample *sample, 656 struct machine *machine __maybe_unused) 657 { 658 const char *comm = perf_evsel__strval(evsel, sample, "comm"); 659 const u32 pid = perf_evsel__intval(evsel, sample, "pid"); 660 struct task_desc *waker, *wakee; 661 662 if (verbose) { 663 printf("sched_wakeup event %p\n", evsel); 664 665 printf(" ... pid %d woke up %s/%d\n", sample->tid, comm, pid); 666 } 667 668 waker = register_pid(sched, sample->tid, "<unknown>"); 669 wakee = register_pid(sched, pid, comm); 670 671 add_sched_event_wakeup(sched, waker, sample->time, wakee); 672 return 0; 673 } 674 675 static int replay_switch_event(struct perf_sched *sched, 676 struct perf_evsel *evsel, 677 struct perf_sample *sample, 678 struct machine *machine __maybe_unused) 679 { 680 const char *prev_comm = perf_evsel__strval(evsel, sample, "prev_comm"), 681 *next_comm = perf_evsel__strval(evsel, sample, "next_comm"); 682 const u32 prev_pid = perf_evsel__intval(evsel, sample, "prev_pid"), 683 next_pid = perf_evsel__intval(evsel, sample, "next_pid"); 684 const u64 prev_state = perf_evsel__intval(evsel, sample, "prev_state"); 685 struct task_desc *prev, __maybe_unused *next; 686 u64 timestamp0, timestamp = sample->time; 687 int cpu = sample->cpu; 688 s64 delta; 689 690 if (verbose) 691 printf("sched_switch event %p\n", evsel); 692 693 if (cpu >= MAX_CPUS || cpu < 0) 694 return 0; 695 696 timestamp0 = sched->cpu_last_switched[cpu]; 697 if (timestamp0) 698 delta = timestamp - timestamp0; 699 else 700 delta = 0; 701 702 if (delta < 0) { 703 pr_err("hm, delta: %" PRIu64 " < 0 ?\n", delta); 704 return -1; 705 } 706 707 pr_debug(" ... switch from %s/%d to %s/%d [ran %" PRIu64 " nsecs]\n", 708 prev_comm, prev_pid, next_comm, next_pid, delta); 709 710 prev = register_pid(sched, prev_pid, prev_comm); 711 next = register_pid(sched, next_pid, next_comm); 712 713 sched->cpu_last_switched[cpu] = timestamp; 714 715 add_sched_event_run(sched, prev, timestamp, delta); 716 add_sched_event_sleep(sched, prev, timestamp, prev_state); 717 718 return 0; 719 } 720 721 static int replay_fork_event(struct perf_sched *sched, struct perf_evsel *evsel, 722 struct perf_sample *sample) 723 { 724 const char *parent_comm = perf_evsel__strval(evsel, sample, "parent_comm"), 725 *child_comm = perf_evsel__strval(evsel, sample, "child_comm"); 726 const u32 parent_pid = perf_evsel__intval(evsel, sample, "parent_pid"), 727 child_pid = perf_evsel__intval(evsel, sample, "child_pid"); 728 729 if (verbose) { 730 printf("sched_fork event %p\n", evsel); 731 printf("... parent: %s/%d\n", parent_comm, parent_pid); 732 printf("... child: %s/%d\n", child_comm, child_pid); 733 } 734 735 register_pid(sched, parent_pid, parent_comm); 736 register_pid(sched, child_pid, child_comm); 737 return 0; 738 } 739 740 struct sort_dimension { 741 const char *name; 742 sort_fn_t cmp; 743 struct list_head list; 744 }; 745 746 static int 747 thread_lat_cmp(struct list_head *list, struct work_atoms *l, struct work_atoms *r) 748 { 749 struct sort_dimension *sort; 750 int ret = 0; 751 752 BUG_ON(list_empty(list)); 753 754 list_for_each_entry(sort, list, list) { 755 ret = sort->cmp(l, r); 756 if (ret) 757 return ret; 758 } 759 760 return ret; 761 } 762 763 static struct work_atoms * 764 thread_atoms_search(struct rb_root *root, struct thread *thread, 765 struct list_head *sort_list) 766 { 767 struct rb_node *node = root->rb_node; 768 struct work_atoms key = { .thread = thread }; 769 770 while (node) { 771 struct work_atoms *atoms; 772 int cmp; 773 774 atoms = container_of(node, struct work_atoms, node); 775 776 cmp = thread_lat_cmp(sort_list, &key, atoms); 777 if (cmp > 0) 778 node = node->rb_left; 779 else if (cmp < 0) 780 node = node->rb_right; 781 else { 782 BUG_ON(thread != atoms->thread); 783 return atoms; 784 } 785 } 786 return NULL; 787 } 788 789 static void 790 __thread_latency_insert(struct rb_root *root, struct work_atoms *data, 791 struct list_head *sort_list) 792 { 793 struct rb_node **new = &(root->rb_node), *parent = NULL; 794 795 while (*new) { 796 struct work_atoms *this; 797 int cmp; 798 799 this = container_of(*new, struct work_atoms, node); 800 parent = *new; 801 802 cmp = thread_lat_cmp(sort_list, data, this); 803 804 if (cmp > 0) 805 new = &((*new)->rb_left); 806 else 807 new = &((*new)->rb_right); 808 } 809 810 rb_link_node(&data->node, parent, new); 811 rb_insert_color(&data->node, root); 812 } 813 814 static int thread_atoms_insert(struct perf_sched *sched, struct thread *thread) 815 { 816 struct work_atoms *atoms = zalloc(sizeof(*atoms)); 817 if (!atoms) { 818 pr_err("No memory at %s\n", __func__); 819 return -1; 820 } 821 822 atoms->thread = thread; 823 INIT_LIST_HEAD(&atoms->work_list); 824 __thread_latency_insert(&sched->atom_root, atoms, &sched->cmp_pid); 825 return 0; 826 } 827 828 static int latency_fork_event(struct perf_sched *sched __maybe_unused, 829 struct perf_evsel *evsel __maybe_unused, 830 struct perf_sample *sample __maybe_unused) 831 { 832 /* should insert the newcomer */ 833 return 0; 834 } 835 836 static char sched_out_state(u64 prev_state) 837 { 838 const char *str = TASK_STATE_TO_CHAR_STR; 839 840 return str[prev_state]; 841 } 842 843 static int 844 add_sched_out_event(struct work_atoms *atoms, 845 char run_state, 846 u64 timestamp) 847 { 848 struct work_atom *atom = zalloc(sizeof(*atom)); 849 if (!atom) { 850 pr_err("Non memory at %s", __func__); 851 return -1; 852 } 853 854 atom->sched_out_time = timestamp; 855 856 if (run_state == 'R') { 857 atom->state = THREAD_WAIT_CPU; 858 atom->wake_up_time = atom->sched_out_time; 859 } 860 861 list_add_tail(&atom->list, &atoms->work_list); 862 return 0; 863 } 864 865 static void 866 add_runtime_event(struct work_atoms *atoms, u64 delta, 867 u64 timestamp __maybe_unused) 868 { 869 struct work_atom *atom; 870 871 BUG_ON(list_empty(&atoms->work_list)); 872 873 atom = list_entry(atoms->work_list.prev, struct work_atom, list); 874 875 atom->runtime += delta; 876 atoms->total_runtime += delta; 877 } 878 879 static void 880 add_sched_in_event(struct work_atoms *atoms, u64 timestamp) 881 { 882 struct work_atom *atom; 883 u64 delta; 884 885 if (list_empty(&atoms->work_list)) 886 return; 887 888 atom = list_entry(atoms->work_list.prev, struct work_atom, list); 889 890 if (atom->state != THREAD_WAIT_CPU) 891 return; 892 893 if (timestamp < atom->wake_up_time) { 894 atom->state = THREAD_IGNORE; 895 return; 896 } 897 898 atom->state = THREAD_SCHED_IN; 899 atom->sched_in_time = timestamp; 900 901 delta = atom->sched_in_time - atom->wake_up_time; 902 atoms->total_lat += delta; 903 if (delta > atoms->max_lat) { 904 atoms->max_lat = delta; 905 atoms->max_lat_at = timestamp; 906 } 907 atoms->nb_atoms++; 908 } 909 910 static int latency_switch_event(struct perf_sched *sched, 911 struct perf_evsel *evsel, 912 struct perf_sample *sample, 913 struct machine *machine) 914 { 915 const u32 prev_pid = perf_evsel__intval(evsel, sample, "prev_pid"), 916 next_pid = perf_evsel__intval(evsel, sample, "next_pid"); 917 const u64 prev_state = perf_evsel__intval(evsel, sample, "prev_state"); 918 struct work_atoms *out_events, *in_events; 919 struct thread *sched_out, *sched_in; 920 u64 timestamp0, timestamp = sample->time; 921 int cpu = sample->cpu; 922 s64 delta; 923 924 BUG_ON(cpu >= MAX_CPUS || cpu < 0); 925 926 timestamp0 = sched->cpu_last_switched[cpu]; 927 sched->cpu_last_switched[cpu] = timestamp; 928 if (timestamp0) 929 delta = timestamp - timestamp0; 930 else 931 delta = 0; 932 933 if (delta < 0) { 934 pr_err("hm, delta: %" PRIu64 " < 0 ?\n", delta); 935 return -1; 936 } 937 938 sched_out = machine__findnew_thread(machine, prev_pid); 939 sched_in = machine__findnew_thread(machine, next_pid); 940 941 out_events = thread_atoms_search(&sched->atom_root, sched_out, &sched->cmp_pid); 942 if (!out_events) { 943 if (thread_atoms_insert(sched, sched_out)) 944 return -1; 945 out_events = thread_atoms_search(&sched->atom_root, sched_out, &sched->cmp_pid); 946 if (!out_events) { 947 pr_err("out-event: Internal tree error"); 948 return -1; 949 } 950 } 951 if (add_sched_out_event(out_events, sched_out_state(prev_state), timestamp)) 952 return -1; 953 954 in_events = thread_atoms_search(&sched->atom_root, sched_in, &sched->cmp_pid); 955 if (!in_events) { 956 if (thread_atoms_insert(sched, sched_in)) 957 return -1; 958 in_events = thread_atoms_search(&sched->atom_root, sched_in, &sched->cmp_pid); 959 if (!in_events) { 960 pr_err("in-event: Internal tree error"); 961 return -1; 962 } 963 /* 964 * Take came in we have not heard about yet, 965 * add in an initial atom in runnable state: 966 */ 967 if (add_sched_out_event(in_events, 'R', timestamp)) 968 return -1; 969 } 970 add_sched_in_event(in_events, timestamp); 971 972 return 0; 973 } 974 975 static int latency_runtime_event(struct perf_sched *sched, 976 struct perf_evsel *evsel, 977 struct perf_sample *sample, 978 struct machine *machine) 979 { 980 const u32 pid = perf_evsel__intval(evsel, sample, "pid"); 981 const u64 runtime = perf_evsel__intval(evsel, sample, "runtime"); 982 struct thread *thread = machine__findnew_thread(machine, pid); 983 struct work_atoms *atoms = thread_atoms_search(&sched->atom_root, thread, &sched->cmp_pid); 984 u64 timestamp = sample->time; 985 int cpu = sample->cpu; 986 987 BUG_ON(cpu >= MAX_CPUS || cpu < 0); 988 if (!atoms) { 989 if (thread_atoms_insert(sched, thread)) 990 return -1; 991 atoms = thread_atoms_search(&sched->atom_root, thread, &sched->cmp_pid); 992 if (!atoms) { 993 pr_err("in-event: Internal tree error"); 994 return -1; 995 } 996 if (add_sched_out_event(atoms, 'R', timestamp)) 997 return -1; 998 } 999 1000 add_runtime_event(atoms, runtime, timestamp); 1001 return 0; 1002 } 1003 1004 static int latency_wakeup_event(struct perf_sched *sched, 1005 struct perf_evsel *evsel, 1006 struct perf_sample *sample, 1007 struct machine *machine) 1008 { 1009 const u32 pid = perf_evsel__intval(evsel, sample, "pid"), 1010 success = perf_evsel__intval(evsel, sample, "success"); 1011 struct work_atoms *atoms; 1012 struct work_atom *atom; 1013 struct thread *wakee; 1014 u64 timestamp = sample->time; 1015 1016 /* Note for later, it may be interesting to observe the failing cases */ 1017 if (!success) 1018 return 0; 1019 1020 wakee = machine__findnew_thread(machine, pid); 1021 atoms = thread_atoms_search(&sched->atom_root, wakee, &sched->cmp_pid); 1022 if (!atoms) { 1023 if (thread_atoms_insert(sched, wakee)) 1024 return -1; 1025 atoms = thread_atoms_search(&sched->atom_root, wakee, &sched->cmp_pid); 1026 if (!atoms) { 1027 pr_err("wakeup-event: Internal tree error"); 1028 return -1; 1029 } 1030 if (add_sched_out_event(atoms, 'S', timestamp)) 1031 return -1; 1032 } 1033 1034 BUG_ON(list_empty(&atoms->work_list)); 1035 1036 atom = list_entry(atoms->work_list.prev, struct work_atom, list); 1037 1038 /* 1039 * You WILL be missing events if you've recorded only 1040 * one CPU, or are only looking at only one, so don't 1041 * make useless noise. 1042 */ 1043 if (sched->profile_cpu == -1 && atom->state != THREAD_SLEEPING) 1044 sched->nr_state_machine_bugs++; 1045 1046 sched->nr_timestamps++; 1047 if (atom->sched_out_time > timestamp) { 1048 sched->nr_unordered_timestamps++; 1049 return 0; 1050 } 1051 1052 atom->state = THREAD_WAIT_CPU; 1053 atom->wake_up_time = timestamp; 1054 return 0; 1055 } 1056 1057 static int latency_migrate_task_event(struct perf_sched *sched, 1058 struct perf_evsel *evsel, 1059 struct perf_sample *sample, 1060 struct machine *machine) 1061 { 1062 const u32 pid = perf_evsel__intval(evsel, sample, "pid"); 1063 u64 timestamp = sample->time; 1064 struct work_atoms *atoms; 1065 struct work_atom *atom; 1066 struct thread *migrant; 1067 1068 /* 1069 * Only need to worry about migration when profiling one CPU. 1070 */ 1071 if (sched->profile_cpu == -1) 1072 return 0; 1073 1074 migrant = machine__findnew_thread(machine, pid); 1075 atoms = thread_atoms_search(&sched->atom_root, migrant, &sched->cmp_pid); 1076 if (!atoms) { 1077 if (thread_atoms_insert(sched, migrant)) 1078 return -1; 1079 register_pid(sched, migrant->pid, migrant->comm); 1080 atoms = thread_atoms_search(&sched->atom_root, migrant, &sched->cmp_pid); 1081 if (!atoms) { 1082 pr_err("migration-event: Internal tree error"); 1083 return -1; 1084 } 1085 if (add_sched_out_event(atoms, 'R', timestamp)) 1086 return -1; 1087 } 1088 1089 BUG_ON(list_empty(&atoms->work_list)); 1090 1091 atom = list_entry(atoms->work_list.prev, struct work_atom, list); 1092 atom->sched_in_time = atom->sched_out_time = atom->wake_up_time = timestamp; 1093 1094 sched->nr_timestamps++; 1095 1096 if (atom->sched_out_time > timestamp) 1097 sched->nr_unordered_timestamps++; 1098 1099 return 0; 1100 } 1101 1102 static void output_lat_thread(struct perf_sched *sched, struct work_atoms *work_list) 1103 { 1104 int i; 1105 int ret; 1106 u64 avg; 1107 1108 if (!work_list->nb_atoms) 1109 return; 1110 /* 1111 * Ignore idle threads: 1112 */ 1113 if (!strcmp(work_list->thread->comm, "swapper")) 1114 return; 1115 1116 sched->all_runtime += work_list->total_runtime; 1117 sched->all_count += work_list->nb_atoms; 1118 1119 ret = printf(" %s:%d ", work_list->thread->comm, work_list->thread->pid); 1120 1121 for (i = 0; i < 24 - ret; i++) 1122 printf(" "); 1123 1124 avg = work_list->total_lat / work_list->nb_atoms; 1125 1126 printf("|%11.3f ms |%9" PRIu64 " | avg:%9.3f ms | max:%9.3f ms | max at: %9.6f s\n", 1127 (double)work_list->total_runtime / 1e6, 1128 work_list->nb_atoms, (double)avg / 1e6, 1129 (double)work_list->max_lat / 1e6, 1130 (double)work_list->max_lat_at / 1e9); 1131 } 1132 1133 static int pid_cmp(struct work_atoms *l, struct work_atoms *r) 1134 { 1135 if (l->thread->pid < r->thread->pid) 1136 return -1; 1137 if (l->thread->pid > r->thread->pid) 1138 return 1; 1139 1140 return 0; 1141 } 1142 1143 static int avg_cmp(struct work_atoms *l, struct work_atoms *r) 1144 { 1145 u64 avgl, avgr; 1146 1147 if (!l->nb_atoms) 1148 return -1; 1149 1150 if (!r->nb_atoms) 1151 return 1; 1152 1153 avgl = l->total_lat / l->nb_atoms; 1154 avgr = r->total_lat / r->nb_atoms; 1155 1156 if (avgl < avgr) 1157 return -1; 1158 if (avgl > avgr) 1159 return 1; 1160 1161 return 0; 1162 } 1163 1164 static int max_cmp(struct work_atoms *l, struct work_atoms *r) 1165 { 1166 if (l->max_lat < r->max_lat) 1167 return -1; 1168 if (l->max_lat > r->max_lat) 1169 return 1; 1170 1171 return 0; 1172 } 1173 1174 static int switch_cmp(struct work_atoms *l, struct work_atoms *r) 1175 { 1176 if (l->nb_atoms < r->nb_atoms) 1177 return -1; 1178 if (l->nb_atoms > r->nb_atoms) 1179 return 1; 1180 1181 return 0; 1182 } 1183 1184 static int runtime_cmp(struct work_atoms *l, struct work_atoms *r) 1185 { 1186 if (l->total_runtime < r->total_runtime) 1187 return -1; 1188 if (l->total_runtime > r->total_runtime) 1189 return 1; 1190 1191 return 0; 1192 } 1193 1194 static int sort_dimension__add(const char *tok, struct list_head *list) 1195 { 1196 size_t i; 1197 static struct sort_dimension avg_sort_dimension = { 1198 .name = "avg", 1199 .cmp = avg_cmp, 1200 }; 1201 static struct sort_dimension max_sort_dimension = { 1202 .name = "max", 1203 .cmp = max_cmp, 1204 }; 1205 static struct sort_dimension pid_sort_dimension = { 1206 .name = "pid", 1207 .cmp = pid_cmp, 1208 }; 1209 static struct sort_dimension runtime_sort_dimension = { 1210 .name = "runtime", 1211 .cmp = runtime_cmp, 1212 }; 1213 static struct sort_dimension switch_sort_dimension = { 1214 .name = "switch", 1215 .cmp = switch_cmp, 1216 }; 1217 struct sort_dimension *available_sorts[] = { 1218 &pid_sort_dimension, 1219 &avg_sort_dimension, 1220 &max_sort_dimension, 1221 &switch_sort_dimension, 1222 &runtime_sort_dimension, 1223 }; 1224 1225 for (i = 0; i < ARRAY_SIZE(available_sorts); i++) { 1226 if (!strcmp(available_sorts[i]->name, tok)) { 1227 list_add_tail(&available_sorts[i]->list, list); 1228 1229 return 0; 1230 } 1231 } 1232 1233 return -1; 1234 } 1235 1236 static void perf_sched__sort_lat(struct perf_sched *sched) 1237 { 1238 struct rb_node *node; 1239 1240 for (;;) { 1241 struct work_atoms *data; 1242 node = rb_first(&sched->atom_root); 1243 if (!node) 1244 break; 1245 1246 rb_erase(node, &sched->atom_root); 1247 data = rb_entry(node, struct work_atoms, node); 1248 __thread_latency_insert(&sched->sorted_atom_root, data, &sched->sort_list); 1249 } 1250 } 1251 1252 static int process_sched_wakeup_event(struct perf_tool *tool, 1253 struct perf_evsel *evsel, 1254 struct perf_sample *sample, 1255 struct machine *machine) 1256 { 1257 struct perf_sched *sched = container_of(tool, struct perf_sched, tool); 1258 1259 if (sched->tp_handler->wakeup_event) 1260 return sched->tp_handler->wakeup_event(sched, evsel, sample, machine); 1261 1262 return 0; 1263 } 1264 1265 static int map_switch_event(struct perf_sched *sched, struct perf_evsel *evsel, 1266 struct perf_sample *sample, struct machine *machine) 1267 { 1268 const u32 prev_pid = perf_evsel__intval(evsel, sample, "prev_pid"), 1269 next_pid = perf_evsel__intval(evsel, sample, "next_pid"); 1270 struct thread *sched_out __maybe_unused, *sched_in; 1271 int new_shortname; 1272 u64 timestamp0, timestamp = sample->time; 1273 s64 delta; 1274 int cpu, this_cpu = sample->cpu; 1275 1276 BUG_ON(this_cpu >= MAX_CPUS || this_cpu < 0); 1277 1278 if (this_cpu > sched->max_cpu) 1279 sched->max_cpu = this_cpu; 1280 1281 timestamp0 = sched->cpu_last_switched[this_cpu]; 1282 sched->cpu_last_switched[this_cpu] = timestamp; 1283 if (timestamp0) 1284 delta = timestamp - timestamp0; 1285 else 1286 delta = 0; 1287 1288 if (delta < 0) { 1289 pr_err("hm, delta: %" PRIu64 " < 0 ?\n", delta); 1290 return -1; 1291 } 1292 1293 sched_out = machine__findnew_thread(machine, prev_pid); 1294 sched_in = machine__findnew_thread(machine, next_pid); 1295 1296 sched->curr_thread[this_cpu] = sched_in; 1297 1298 printf(" "); 1299 1300 new_shortname = 0; 1301 if (!sched_in->shortname[0]) { 1302 sched_in->shortname[0] = sched->next_shortname1; 1303 sched_in->shortname[1] = sched->next_shortname2; 1304 1305 if (sched->next_shortname1 < 'Z') { 1306 sched->next_shortname1++; 1307 } else { 1308 sched->next_shortname1='A'; 1309 if (sched->next_shortname2 < '9') { 1310 sched->next_shortname2++; 1311 } else { 1312 sched->next_shortname2='0'; 1313 } 1314 } 1315 new_shortname = 1; 1316 } 1317 1318 for (cpu = 0; cpu <= sched->max_cpu; cpu++) { 1319 if (cpu != this_cpu) 1320 printf(" "); 1321 else 1322 printf("*"); 1323 1324 if (sched->curr_thread[cpu]) { 1325 if (sched->curr_thread[cpu]->pid) 1326 printf("%2s ", sched->curr_thread[cpu]->shortname); 1327 else 1328 printf(". "); 1329 } else 1330 printf(" "); 1331 } 1332 1333 printf(" %12.6f secs ", (double)timestamp/1e9); 1334 if (new_shortname) { 1335 printf("%s => %s:%d\n", 1336 sched_in->shortname, sched_in->comm, sched_in->pid); 1337 } else { 1338 printf("\n"); 1339 } 1340 1341 return 0; 1342 } 1343 1344 static int process_sched_switch_event(struct perf_tool *tool, 1345 struct perf_evsel *evsel, 1346 struct perf_sample *sample, 1347 struct machine *machine) 1348 { 1349 struct perf_sched *sched = container_of(tool, struct perf_sched, tool); 1350 int this_cpu = sample->cpu, err = 0; 1351 u32 prev_pid = perf_evsel__intval(evsel, sample, "prev_pid"), 1352 next_pid = perf_evsel__intval(evsel, sample, "next_pid"); 1353 1354 if (sched->curr_pid[this_cpu] != (u32)-1) { 1355 /* 1356 * Are we trying to switch away a PID that is 1357 * not current? 1358 */ 1359 if (sched->curr_pid[this_cpu] != prev_pid) 1360 sched->nr_context_switch_bugs++; 1361 } 1362 1363 if (sched->tp_handler->switch_event) 1364 err = sched->tp_handler->switch_event(sched, evsel, sample, machine); 1365 1366 sched->curr_pid[this_cpu] = next_pid; 1367 return err; 1368 } 1369 1370 static int process_sched_runtime_event(struct perf_tool *tool, 1371 struct perf_evsel *evsel, 1372 struct perf_sample *sample, 1373 struct machine *machine) 1374 { 1375 struct perf_sched *sched = container_of(tool, struct perf_sched, tool); 1376 1377 if (sched->tp_handler->runtime_event) 1378 return sched->tp_handler->runtime_event(sched, evsel, sample, machine); 1379 1380 return 0; 1381 } 1382 1383 static int process_sched_fork_event(struct perf_tool *tool, 1384 struct perf_evsel *evsel, 1385 struct perf_sample *sample, 1386 struct machine *machine __maybe_unused) 1387 { 1388 struct perf_sched *sched = container_of(tool, struct perf_sched, tool); 1389 1390 if (sched->tp_handler->fork_event) 1391 return sched->tp_handler->fork_event(sched, evsel, sample); 1392 1393 return 0; 1394 } 1395 1396 static int process_sched_exit_event(struct perf_tool *tool __maybe_unused, 1397 struct perf_evsel *evsel, 1398 struct perf_sample *sample __maybe_unused, 1399 struct machine *machine __maybe_unused) 1400 { 1401 pr_debug("sched_exit event %p\n", evsel); 1402 return 0; 1403 } 1404 1405 static int process_sched_migrate_task_event(struct perf_tool *tool, 1406 struct perf_evsel *evsel, 1407 struct perf_sample *sample, 1408 struct machine *machine) 1409 { 1410 struct perf_sched *sched = container_of(tool, struct perf_sched, tool); 1411 1412 if (sched->tp_handler->migrate_task_event) 1413 return sched->tp_handler->migrate_task_event(sched, evsel, sample, machine); 1414 1415 return 0; 1416 } 1417 1418 typedef int (*tracepoint_handler)(struct perf_tool *tool, 1419 struct perf_evsel *evsel, 1420 struct perf_sample *sample, 1421 struct machine *machine); 1422 1423 static int perf_sched__process_tracepoint_sample(struct perf_tool *tool __maybe_unused, 1424 union perf_event *event __maybe_unused, 1425 struct perf_sample *sample, 1426 struct perf_evsel *evsel, 1427 struct machine *machine) 1428 { 1429 struct thread *thread = machine__findnew_thread(machine, sample->tid); 1430 int err = 0; 1431 1432 if (thread == NULL) { 1433 pr_debug("problem processing %s event, skipping it.\n", 1434 perf_evsel__name(evsel)); 1435 return -1; 1436 } 1437 1438 evsel->hists.stats.total_period += sample->period; 1439 hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE); 1440 1441 if (evsel->handler.func != NULL) { 1442 tracepoint_handler f = evsel->handler.func; 1443 err = f(tool, evsel, sample, machine); 1444 } 1445 1446 return err; 1447 } 1448 1449 static int perf_sched__read_events(struct perf_sched *sched, bool destroy, 1450 struct perf_session **psession) 1451 { 1452 const struct perf_evsel_str_handler handlers[] = { 1453 { "sched:sched_switch", process_sched_switch_event, }, 1454 { "sched:sched_stat_runtime", process_sched_runtime_event, }, 1455 { "sched:sched_wakeup", process_sched_wakeup_event, }, 1456 { "sched:sched_wakeup_new", process_sched_wakeup_event, }, 1457 { "sched:sched_process_fork", process_sched_fork_event, }, 1458 { "sched:sched_process_exit", process_sched_exit_event, }, 1459 { "sched:sched_migrate_task", process_sched_migrate_task_event, }, 1460 }; 1461 struct perf_session *session; 1462 1463 session = perf_session__new(sched->input_name, O_RDONLY, 0, false, &sched->tool); 1464 if (session == NULL) { 1465 pr_debug("No Memory for session\n"); 1466 return -1; 1467 } 1468 1469 if (perf_session__set_tracepoints_handlers(session, handlers)) 1470 goto out_delete; 1471 1472 if (perf_session__has_traces(session, "record -R")) { 1473 int err = perf_session__process_events(session, &sched->tool); 1474 if (err) { 1475 pr_err("Failed to process events, error %d", err); 1476 goto out_delete; 1477 } 1478 1479 sched->nr_events = session->hists.stats.nr_events[0]; 1480 sched->nr_lost_events = session->hists.stats.total_lost; 1481 sched->nr_lost_chunks = session->hists.stats.nr_events[PERF_RECORD_LOST]; 1482 } 1483 1484 if (destroy) 1485 perf_session__delete(session); 1486 1487 if (psession) 1488 *psession = session; 1489 1490 return 0; 1491 1492 out_delete: 1493 perf_session__delete(session); 1494 return -1; 1495 } 1496 1497 static void print_bad_events(struct perf_sched *sched) 1498 { 1499 if (sched->nr_unordered_timestamps && sched->nr_timestamps) { 1500 printf(" INFO: %.3f%% unordered timestamps (%ld out of %ld)\n", 1501 (double)sched->nr_unordered_timestamps/(double)sched->nr_timestamps*100.0, 1502 sched->nr_unordered_timestamps, sched->nr_timestamps); 1503 } 1504 if (sched->nr_lost_events && sched->nr_events) { 1505 printf(" INFO: %.3f%% lost events (%ld out of %ld, in %ld chunks)\n", 1506 (double)sched->nr_lost_events/(double)sched->nr_events * 100.0, 1507 sched->nr_lost_events, sched->nr_events, sched->nr_lost_chunks); 1508 } 1509 if (sched->nr_state_machine_bugs && sched->nr_timestamps) { 1510 printf(" INFO: %.3f%% state machine bugs (%ld out of %ld)", 1511 (double)sched->nr_state_machine_bugs/(double)sched->nr_timestamps*100.0, 1512 sched->nr_state_machine_bugs, sched->nr_timestamps); 1513 if (sched->nr_lost_events) 1514 printf(" (due to lost events?)"); 1515 printf("\n"); 1516 } 1517 if (sched->nr_context_switch_bugs && sched->nr_timestamps) { 1518 printf(" INFO: %.3f%% context switch bugs (%ld out of %ld)", 1519 (double)sched->nr_context_switch_bugs/(double)sched->nr_timestamps*100.0, 1520 sched->nr_context_switch_bugs, sched->nr_timestamps); 1521 if (sched->nr_lost_events) 1522 printf(" (due to lost events?)"); 1523 printf("\n"); 1524 } 1525 } 1526 1527 static int perf_sched__lat(struct perf_sched *sched) 1528 { 1529 struct rb_node *next; 1530 struct perf_session *session; 1531 1532 setup_pager(); 1533 if (perf_sched__read_events(sched, false, &session)) 1534 return -1; 1535 perf_sched__sort_lat(sched); 1536 1537 printf("\n ---------------------------------------------------------------------------------------------------------------\n"); 1538 printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms | Maximum delay at |\n"); 1539 printf(" ---------------------------------------------------------------------------------------------------------------\n"); 1540 1541 next = rb_first(&sched->sorted_atom_root); 1542 1543 while (next) { 1544 struct work_atoms *work_list; 1545 1546 work_list = rb_entry(next, struct work_atoms, node); 1547 output_lat_thread(sched, work_list); 1548 next = rb_next(next); 1549 } 1550 1551 printf(" -----------------------------------------------------------------------------------------\n"); 1552 printf(" TOTAL: |%11.3f ms |%9" PRIu64 " |\n", 1553 (double)sched->all_runtime / 1e6, sched->all_count); 1554 1555 printf(" ---------------------------------------------------\n"); 1556 1557 print_bad_events(sched); 1558 printf("\n"); 1559 1560 perf_session__delete(session); 1561 return 0; 1562 } 1563 1564 static int perf_sched__map(struct perf_sched *sched) 1565 { 1566 sched->max_cpu = sysconf(_SC_NPROCESSORS_CONF); 1567 1568 setup_pager(); 1569 if (perf_sched__read_events(sched, true, NULL)) 1570 return -1; 1571 print_bad_events(sched); 1572 return 0; 1573 } 1574 1575 static int perf_sched__replay(struct perf_sched *sched) 1576 { 1577 unsigned long i; 1578 1579 calibrate_run_measurement_overhead(sched); 1580 calibrate_sleep_measurement_overhead(sched); 1581 1582 test_calibrations(sched); 1583 1584 if (perf_sched__read_events(sched, true, NULL)) 1585 return -1; 1586 1587 printf("nr_run_events: %ld\n", sched->nr_run_events); 1588 printf("nr_sleep_events: %ld\n", sched->nr_sleep_events); 1589 printf("nr_wakeup_events: %ld\n", sched->nr_wakeup_events); 1590 1591 if (sched->targetless_wakeups) 1592 printf("target-less wakeups: %ld\n", sched->targetless_wakeups); 1593 if (sched->multitarget_wakeups) 1594 printf("multi-target wakeups: %ld\n", sched->multitarget_wakeups); 1595 if (sched->nr_run_events_optimized) 1596 printf("run atoms optimized: %ld\n", 1597 sched->nr_run_events_optimized); 1598 1599 print_task_traces(sched); 1600 add_cross_task_wakeups(sched); 1601 1602 create_tasks(sched); 1603 printf("------------------------------------------------------------\n"); 1604 for (i = 0; i < sched->replay_repeat; i++) 1605 run_one_test(sched); 1606 1607 return 0; 1608 } 1609 1610 static void setup_sorting(struct perf_sched *sched, const struct option *options, 1611 const char * const usage_msg[]) 1612 { 1613 char *tmp, *tok, *str = strdup(sched->sort_order); 1614 1615 for (tok = strtok_r(str, ", ", &tmp); 1616 tok; tok = strtok_r(NULL, ", ", &tmp)) { 1617 if (sort_dimension__add(tok, &sched->sort_list) < 0) { 1618 error("Unknown --sort key: `%s'", tok); 1619 usage_with_options(usage_msg, options); 1620 } 1621 } 1622 1623 free(str); 1624 1625 sort_dimension__add("pid", &sched->cmp_pid); 1626 } 1627 1628 static int __cmd_record(int argc, const char **argv) 1629 { 1630 unsigned int rec_argc, i, j; 1631 const char **rec_argv; 1632 const char * const record_args[] = { 1633 "record", 1634 "-a", 1635 "-R", 1636 "-f", 1637 "-m", "1024", 1638 "-c", "1", 1639 "-e", "sched:sched_switch", 1640 "-e", "sched:sched_stat_wait", 1641 "-e", "sched:sched_stat_sleep", 1642 "-e", "sched:sched_stat_iowait", 1643 "-e", "sched:sched_stat_runtime", 1644 "-e", "sched:sched_process_exit", 1645 "-e", "sched:sched_process_fork", 1646 "-e", "sched:sched_wakeup", 1647 "-e", "sched:sched_migrate_task", 1648 }; 1649 1650 rec_argc = ARRAY_SIZE(record_args) + argc - 1; 1651 rec_argv = calloc(rec_argc + 1, sizeof(char *)); 1652 1653 if (rec_argv == NULL) 1654 return -ENOMEM; 1655 1656 for (i = 0; i < ARRAY_SIZE(record_args); i++) 1657 rec_argv[i] = strdup(record_args[i]); 1658 1659 for (j = 1; j < (unsigned int)argc; j++, i++) 1660 rec_argv[i] = argv[j]; 1661 1662 BUG_ON(i != rec_argc); 1663 1664 return cmd_record(i, rec_argv, NULL); 1665 } 1666 1667 int cmd_sched(int argc, const char **argv, const char *prefix __maybe_unused) 1668 { 1669 const char default_sort_order[] = "avg, max, switch, runtime"; 1670 struct perf_sched sched = { 1671 .tool = { 1672 .sample = perf_sched__process_tracepoint_sample, 1673 .comm = perf_event__process_comm, 1674 .lost = perf_event__process_lost, 1675 .fork = perf_event__process_task, 1676 .ordered_samples = true, 1677 }, 1678 .cmp_pid = LIST_HEAD_INIT(sched.cmp_pid), 1679 .sort_list = LIST_HEAD_INIT(sched.sort_list), 1680 .start_work_mutex = PTHREAD_MUTEX_INITIALIZER, 1681 .work_done_wait_mutex = PTHREAD_MUTEX_INITIALIZER, 1682 .curr_pid = { [0 ... MAX_CPUS - 1] = -1 }, 1683 .sort_order = default_sort_order, 1684 .replay_repeat = 10, 1685 .profile_cpu = -1, 1686 .next_shortname1 = 'A', 1687 .next_shortname2 = '0', 1688 }; 1689 const struct option latency_options[] = { 1690 OPT_STRING('s', "sort", &sched.sort_order, "key[,key2...]", 1691 "sort by key(s): runtime, switch, avg, max"), 1692 OPT_INCR('v', "verbose", &verbose, 1693 "be more verbose (show symbol address, etc)"), 1694 OPT_INTEGER('C', "CPU", &sched.profile_cpu, 1695 "CPU to profile on"), 1696 OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, 1697 "dump raw trace in ASCII"), 1698 OPT_END() 1699 }; 1700 const struct option replay_options[] = { 1701 OPT_UINTEGER('r', "repeat", &sched.replay_repeat, 1702 "repeat the workload replay N times (-1: infinite)"), 1703 OPT_INCR('v', "verbose", &verbose, 1704 "be more verbose (show symbol address, etc)"), 1705 OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, 1706 "dump raw trace in ASCII"), 1707 OPT_END() 1708 }; 1709 const struct option sched_options[] = { 1710 OPT_STRING('i', "input", &sched.input_name, "file", 1711 "input file name"), 1712 OPT_INCR('v', "verbose", &verbose, 1713 "be more verbose (show symbol address, etc)"), 1714 OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, 1715 "dump raw trace in ASCII"), 1716 OPT_END() 1717 }; 1718 const char * const latency_usage[] = { 1719 "perf sched latency [<options>]", 1720 NULL 1721 }; 1722 const char * const replay_usage[] = { 1723 "perf sched replay [<options>]", 1724 NULL 1725 }; 1726 const char * const sched_usage[] = { 1727 "perf sched [<options>] {record|latency|map|replay|script}", 1728 NULL 1729 }; 1730 struct trace_sched_handler lat_ops = { 1731 .wakeup_event = latency_wakeup_event, 1732 .switch_event = latency_switch_event, 1733 .runtime_event = latency_runtime_event, 1734 .fork_event = latency_fork_event, 1735 .migrate_task_event = latency_migrate_task_event, 1736 }; 1737 struct trace_sched_handler map_ops = { 1738 .switch_event = map_switch_event, 1739 }; 1740 struct trace_sched_handler replay_ops = { 1741 .wakeup_event = replay_wakeup_event, 1742 .switch_event = replay_switch_event, 1743 .fork_event = replay_fork_event, 1744 }; 1745 1746 argc = parse_options(argc, argv, sched_options, sched_usage, 1747 PARSE_OPT_STOP_AT_NON_OPTION); 1748 if (!argc) 1749 usage_with_options(sched_usage, sched_options); 1750 1751 /* 1752 * Aliased to 'perf script' for now: 1753 */ 1754 if (!strcmp(argv[0], "script")) 1755 return cmd_script(argc, argv, prefix); 1756 1757 symbol__init(); 1758 if (!strncmp(argv[0], "rec", 3)) { 1759 return __cmd_record(argc, argv); 1760 } else if (!strncmp(argv[0], "lat", 3)) { 1761 sched.tp_handler = &lat_ops; 1762 if (argc > 1) { 1763 argc = parse_options(argc, argv, latency_options, latency_usage, 0); 1764 if (argc) 1765 usage_with_options(latency_usage, latency_options); 1766 } 1767 setup_sorting(&sched, latency_options, latency_usage); 1768 return perf_sched__lat(&sched); 1769 } else if (!strcmp(argv[0], "map")) { 1770 sched.tp_handler = &map_ops; 1771 setup_sorting(&sched, latency_options, latency_usage); 1772 return perf_sched__map(&sched); 1773 } else if (!strncmp(argv[0], "rep", 3)) { 1774 sched.tp_handler = &replay_ops; 1775 if (argc) { 1776 argc = parse_options(argc, argv, replay_options, replay_usage, 0); 1777 if (argc) 1778 usage_with_options(replay_usage, replay_options); 1779 } 1780 return perf_sched__replay(&sched); 1781 } else { 1782 usage_with_options(sched_usage, sched_options); 1783 } 1784 1785 return 0; 1786 } 1787