1// SPDX-License-Identifier: GPL-2.0-only 2/* 3 * builtin-ftrace.c 4 * 5 * Copyright (c) 2013 LG Electronics, Namhyung Kim <namhyung@kernel.org> 6 * Copyright (c) 2020 Changbin Du <changbin.du@gmail.com>, significant enhancement. 7 */ 8 9#include "builtin.h" 10 11#include <errno.h> 12#include <unistd.h> 13#include <signal.h> 14#include <stdlib.h> 15#include <fcntl.h> 16#include <math.h> 17#include <poll.h> 18#include <ctype.h> 19#include <linux/capability.h> 20#include <linux/string.h> 21 22#include "debug.h" 23#include <subcmd/pager.h> 24#include <subcmd/parse-options.h> 25#include <api/fs/tracing_path.h> 26#include "evlist.h" 27#include "target.h" 28#include "cpumap.h" 29#include "thread_map.h" 30#include "strfilter.h" 31#include "util/cap.h" 32#include "util/config.h" 33#include "util/ftrace.h" 34#include "util/units.h" 35#include "util/parse-sublevel-options.h" 36 37#define DEFAULT_TRACER "function_graph" 38 39static volatile sig_atomic_t workload_exec_errno; 40static volatile sig_atomic_t done; 41 42static void sig_handler(int sig __maybe_unused) 43{ 44 done = true; 45} 46 47/* 48 * evlist__prepare_workload will send a SIGUSR1 if the fork fails, since 49 * we asked by setting its exec_error to the function below, 50 * ftrace__workload_exec_failed_signal. 51 * 52 * XXX We need to handle this more appropriately, emitting an error, etc. 53 */ 54static void ftrace__workload_exec_failed_signal(int signo __maybe_unused, 55 siginfo_t *info __maybe_unused, 56 void *ucontext __maybe_unused) 57{ 58 workload_exec_errno = info->si_value.sival_int; 59 done = true; 60} 61 62static int __write_tracing_file(const char *name, const char *val, bool append) 63{ 64 char *file; 65 int fd, ret = -1; 66 ssize_t size = strlen(val); 67 int flags = O_WRONLY; 68 char errbuf[512]; 69 char *val_copy; 70 71 file = get_tracing_file(name); 72 if (!file) { 73 pr_debug("cannot get tracing file: %s\n", name); 74 return -1; 75 } 76 77 if (append) 78 flags |= O_APPEND; 79 else 80 flags |= O_TRUNC; 81 82 fd = open(file, flags); 83 if (fd < 0) { 84 pr_debug("cannot open tracing file: %s: %s\n", 85 name, str_error_r(errno, errbuf, sizeof(errbuf))); 86 goto out; 87 } 88 89 /* 90 * Copy the original value and append a '\n'. Without this, 91 * the kernel can hide possible errors. 92 */ 93 val_copy = strdup(val); 94 if (!val_copy) 95 goto out_close; 96 val_copy[size] = '\n'; 97 98 if (write(fd, val_copy, size + 1) == size + 1) 99 ret = 0; 100 else 101 pr_debug("write '%s' to tracing/%s failed: %s\n", 102 val, name, str_error_r(errno, errbuf, sizeof(errbuf))); 103 104 free(val_copy); 105out_close: 106 close(fd); 107out: 108 put_tracing_file(file); 109 return ret; 110} 111 112static int write_tracing_file(const char *name, const char *val) 113{ 114 return __write_tracing_file(name, val, false); 115} 116 117static int append_tracing_file(const char *name, const char *val) 118{ 119 return __write_tracing_file(name, val, true); 120} 121 122static int read_tracing_file_to_stdout(const char *name) 123{ 124 char buf[4096]; 125 char *file; 126 int fd; 127 int ret = -1; 128 129 file = get_tracing_file(name); 130 if (!file) { 131 pr_debug("cannot get tracing file: %s\n", name); 132 return -1; 133 } 134 135 fd = open(file, O_RDONLY); 136 if (fd < 0) { 137 pr_debug("cannot open tracing file: %s: %s\n", 138 name, str_error_r(errno, buf, sizeof(buf))); 139 goto out; 140 } 141 142 /* read contents to stdout */ 143 while (true) { 144 int n = read(fd, buf, sizeof(buf)); 145 if (n == 0) 146 break; 147 else if (n < 0) 148 goto out_close; 149 150 if (fwrite(buf, n, 1, stdout) != 1) 151 goto out_close; 152 } 153 ret = 0; 154 155out_close: 156 close(fd); 157out: 158 put_tracing_file(file); 159 return ret; 160} 161 162static int read_tracing_file_by_line(const char *name, 163 void (*cb)(char *str, void *arg), 164 void *cb_arg) 165{ 166 char *line = NULL; 167 size_t len = 0; 168 char *file; 169 FILE *fp; 170 171 file = get_tracing_file(name); 172 if (!file) { 173 pr_debug("cannot get tracing file: %s\n", name); 174 return -1; 175 } 176 177 fp = fopen(file, "r"); 178 if (fp == NULL) { 179 pr_debug("cannot open tracing file: %s\n", name); 180 put_tracing_file(file); 181 return -1; 182 } 183 184 while (getline(&line, &len, fp) != -1) { 185 cb(line, cb_arg); 186 } 187 188 if (line) 189 free(line); 190 191 fclose(fp); 192 put_tracing_file(file); 193 return 0; 194} 195 196static int write_tracing_file_int(const char *name, int value) 197{ 198 char buf[16]; 199 200 snprintf(buf, sizeof(buf), "%d", value); 201 if (write_tracing_file(name, buf) < 0) 202 return -1; 203 204 return 0; 205} 206 207static int write_tracing_option_file(const char *name, const char *val) 208{ 209 char *file; 210 int ret; 211 212 if (asprintf(&file, "options/%s", name) < 0) 213 return -1; 214 215 ret = __write_tracing_file(file, val, false); 216 free(file); 217 return ret; 218} 219 220static int reset_tracing_cpu(void); 221static void reset_tracing_filters(void); 222 223static void reset_tracing_options(struct perf_ftrace *ftrace __maybe_unused) 224{ 225 write_tracing_option_file("function-fork", "0"); 226 write_tracing_option_file("func_stack_trace", "0"); 227 write_tracing_option_file("sleep-time", "1"); 228 write_tracing_option_file("funcgraph-irqs", "1"); 229 write_tracing_option_file("funcgraph-proc", "0"); 230 write_tracing_option_file("funcgraph-abstime", "0"); 231 write_tracing_option_file("latency-format", "0"); 232 write_tracing_option_file("irq-info", "0"); 233} 234 235static int reset_tracing_files(struct perf_ftrace *ftrace __maybe_unused) 236{ 237 if (write_tracing_file("tracing_on", "0") < 0) 238 return -1; 239 240 if (write_tracing_file("current_tracer", "nop") < 0) 241 return -1; 242 243 if (write_tracing_file("set_ftrace_pid", " ") < 0) 244 return -1; 245 246 if (reset_tracing_cpu() < 0) 247 return -1; 248 249 if (write_tracing_file("max_graph_depth", "0") < 0) 250 return -1; 251 252 if (write_tracing_file("tracing_thresh", "0") < 0) 253 return -1; 254 255 reset_tracing_filters(); 256 reset_tracing_options(ftrace); 257 return 0; 258} 259 260static int set_tracing_pid(struct perf_ftrace *ftrace) 261{ 262 int i; 263 char buf[16]; 264 265 if (target__has_cpu(&ftrace->target)) 266 return 0; 267 268 for (i = 0; i < perf_thread_map__nr(ftrace->evlist->core.threads); i++) { 269 scnprintf(buf, sizeof(buf), "%d", 270 perf_thread_map__pid(ftrace->evlist->core.threads, i)); 271 if (append_tracing_file("set_ftrace_pid", buf) < 0) 272 return -1; 273 } 274 return 0; 275} 276 277static int set_tracing_cpumask(struct perf_cpu_map *cpumap) 278{ 279 char *cpumask; 280 size_t mask_size; 281 int ret; 282 int last_cpu; 283 284 last_cpu = perf_cpu_map__cpu(cpumap, perf_cpu_map__nr(cpumap) - 1).cpu; 285 mask_size = last_cpu / 4 + 2; /* one more byte for EOS */ 286 mask_size += last_cpu / 32; /* ',' is needed for every 32th cpus */ 287 288 cpumask = malloc(mask_size); 289 if (cpumask == NULL) { 290 pr_debug("failed to allocate cpu mask\n"); 291 return -1; 292 } 293 294 cpu_map__snprint_mask(cpumap, cpumask, mask_size); 295 296 ret = write_tracing_file("tracing_cpumask", cpumask); 297 298 free(cpumask); 299 return ret; 300} 301 302static int set_tracing_cpu(struct perf_ftrace *ftrace) 303{ 304 struct perf_cpu_map *cpumap = ftrace->evlist->core.user_requested_cpus; 305 306 if (!target__has_cpu(&ftrace->target)) 307 return 0; 308 309 return set_tracing_cpumask(cpumap); 310} 311 312static int set_tracing_func_stack_trace(struct perf_ftrace *ftrace) 313{ 314 if (!ftrace->func_stack_trace) 315 return 0; 316 317 if (write_tracing_option_file("func_stack_trace", "1") < 0) 318 return -1; 319 320 return 0; 321} 322 323static int set_tracing_func_irqinfo(struct perf_ftrace *ftrace) 324{ 325 if (!ftrace->func_irq_info) 326 return 0; 327 328 if (write_tracing_option_file("irq-info", "1") < 0) 329 return -1; 330 331 return 0; 332} 333 334static int reset_tracing_cpu(void) 335{ 336 struct perf_cpu_map *cpumap = perf_cpu_map__new_online_cpus(); 337 int ret; 338 339 ret = set_tracing_cpumask(cpumap); 340 perf_cpu_map__put(cpumap); 341 return ret; 342} 343 344static int __set_tracing_filter(const char *filter_file, struct list_head *funcs) 345{ 346 struct filter_entry *pos; 347 348 list_for_each_entry(pos, funcs, list) { 349 if (append_tracing_file(filter_file, pos->name) < 0) 350 return -1; 351 } 352 353 return 0; 354} 355 356static int set_tracing_filters(struct perf_ftrace *ftrace) 357{ 358 int ret; 359 360 ret = __set_tracing_filter("set_ftrace_filter", &ftrace->filters); 361 if (ret < 0) 362 return ret; 363 364 ret = __set_tracing_filter("set_ftrace_notrace", &ftrace->notrace); 365 if (ret < 0) 366 return ret; 367 368 ret = __set_tracing_filter("set_graph_function", &ftrace->graph_funcs); 369 if (ret < 0) 370 return ret; 371 372 /* old kernels do not have this filter */ 373 __set_tracing_filter("set_graph_notrace", &ftrace->nograph_funcs); 374 375 return ret; 376} 377 378static void reset_tracing_filters(void) 379{ 380 write_tracing_file("set_ftrace_filter", " "); 381 write_tracing_file("set_ftrace_notrace", " "); 382 write_tracing_file("set_graph_function", " "); 383 write_tracing_file("set_graph_notrace", " "); 384} 385 386static int set_tracing_depth(struct perf_ftrace *ftrace) 387{ 388 if (ftrace->graph_depth == 0) 389 return 0; 390 391 if (ftrace->graph_depth < 0) { 392 pr_err("invalid graph depth: %d\n", ftrace->graph_depth); 393 return -1; 394 } 395 396 if (write_tracing_file_int("max_graph_depth", ftrace->graph_depth) < 0) 397 return -1; 398 399 return 0; 400} 401 402static int set_tracing_percpu_buffer_size(struct perf_ftrace *ftrace) 403{ 404 int ret; 405 406 if (ftrace->percpu_buffer_size == 0) 407 return 0; 408 409 ret = write_tracing_file_int("buffer_size_kb", 410 ftrace->percpu_buffer_size / 1024); 411 if (ret < 0) 412 return ret; 413 414 return 0; 415} 416 417static int set_tracing_trace_inherit(struct perf_ftrace *ftrace) 418{ 419 if (!ftrace->inherit) 420 return 0; 421 422 if (write_tracing_option_file("function-fork", "1") < 0) 423 return -1; 424 425 return 0; 426} 427 428static int set_tracing_sleep_time(struct perf_ftrace *ftrace) 429{ 430 if (!ftrace->graph_nosleep_time) 431 return 0; 432 433 if (write_tracing_option_file("sleep-time", "0") < 0) 434 return -1; 435 436 return 0; 437} 438 439static int set_tracing_funcgraph_irqs(struct perf_ftrace *ftrace) 440{ 441 if (!ftrace->graph_noirqs) 442 return 0; 443 444 if (write_tracing_option_file("funcgraph-irqs", "0") < 0) 445 return -1; 446 447 return 0; 448} 449 450static int set_tracing_funcgraph_verbose(struct perf_ftrace *ftrace) 451{ 452 if (!ftrace->graph_verbose) 453 return 0; 454 455 if (write_tracing_option_file("funcgraph-proc", "1") < 0) 456 return -1; 457 458 if (write_tracing_option_file("funcgraph-abstime", "1") < 0) 459 return -1; 460 461 if (write_tracing_option_file("latency-format", "1") < 0) 462 return -1; 463 464 return 0; 465} 466 467static int set_tracing_thresh(struct perf_ftrace *ftrace) 468{ 469 int ret; 470 471 if (ftrace->graph_thresh == 0) 472 return 0; 473 474 ret = write_tracing_file_int("tracing_thresh", ftrace->graph_thresh); 475 if (ret < 0) 476 return ret; 477 478 return 0; 479} 480 481static int set_tracing_options(struct perf_ftrace *ftrace) 482{ 483 if (set_tracing_pid(ftrace) < 0) { 484 pr_err("failed to set ftrace pid\n"); 485 return -1; 486 } 487 488 if (set_tracing_cpu(ftrace) < 0) { 489 pr_err("failed to set tracing cpumask\n"); 490 return -1; 491 } 492 493 if (set_tracing_func_stack_trace(ftrace) < 0) { 494 pr_err("failed to set tracing option func_stack_trace\n"); 495 return -1; 496 } 497 498 if (set_tracing_func_irqinfo(ftrace) < 0) { 499 pr_err("failed to set tracing option irq-info\n"); 500 return -1; 501 } 502 503 if (set_tracing_filters(ftrace) < 0) { 504 pr_err("failed to set tracing filters\n"); 505 return -1; 506 } 507 508 if (set_tracing_depth(ftrace) < 0) { 509 pr_err("failed to set graph depth\n"); 510 return -1; 511 } 512 513 if (set_tracing_percpu_buffer_size(ftrace) < 0) { 514 pr_err("failed to set tracing per-cpu buffer size\n"); 515 return -1; 516 } 517 518 if (set_tracing_trace_inherit(ftrace) < 0) { 519 pr_err("failed to set tracing option function-fork\n"); 520 return -1; 521 } 522 523 if (set_tracing_sleep_time(ftrace) < 0) { 524 pr_err("failed to set tracing option sleep-time\n"); 525 return -1; 526 } 527 528 if (set_tracing_funcgraph_irqs(ftrace) < 0) { 529 pr_err("failed to set tracing option funcgraph-irqs\n"); 530 return -1; 531 } 532 533 if (set_tracing_funcgraph_verbose(ftrace) < 0) { 534 pr_err("failed to set tracing option funcgraph-proc/funcgraph-abstime\n"); 535 return -1; 536 } 537 538 if (set_tracing_thresh(ftrace) < 0) { 539 pr_err("failed to set tracing thresh\n"); 540 return -1; 541 } 542 543 return 0; 544} 545 546static void select_tracer(struct perf_ftrace *ftrace) 547{ 548 bool graph = !list_empty(&ftrace->graph_funcs) || 549 !list_empty(&ftrace->nograph_funcs); 550 bool func = !list_empty(&ftrace->filters) || 551 !list_empty(&ftrace->notrace); 552 553 /* The function_graph has priority over function tracer. */ 554 if (graph) 555 ftrace->tracer = "function_graph"; 556 else if (func) 557 ftrace->tracer = "function"; 558 /* Otherwise, the default tracer is used. */ 559 560 pr_debug("%s tracer is used\n", ftrace->tracer); 561} 562 563static int __cmd_ftrace(struct perf_ftrace *ftrace) 564{ 565 char *trace_file; 566 int trace_fd; 567 char buf[4096]; 568 struct pollfd pollfd = { 569 .events = POLLIN, 570 }; 571 572 if (!(perf_cap__capable(CAP_PERFMON) || 573 perf_cap__capable(CAP_SYS_ADMIN))) { 574 pr_err("ftrace only works for %s!\n", 575#ifdef HAVE_LIBCAP_SUPPORT 576 "users with the CAP_PERFMON or CAP_SYS_ADMIN capability" 577#else 578 "root" 579#endif 580 ); 581 return -1; 582 } 583 584 select_tracer(ftrace); 585 586 if (reset_tracing_files(ftrace) < 0) { 587 pr_err("failed to reset ftrace\n"); 588 goto out; 589 } 590 591 /* reset ftrace buffer */ 592 if (write_tracing_file("trace", "0") < 0) 593 goto out; 594 595 if (set_tracing_options(ftrace) < 0) 596 goto out_reset; 597 598 if (write_tracing_file("current_tracer", ftrace->tracer) < 0) { 599 pr_err("failed to set current_tracer to %s\n", ftrace->tracer); 600 goto out_reset; 601 } 602 603 setup_pager(); 604 605 trace_file = get_tracing_file("trace_pipe"); 606 if (!trace_file) { 607 pr_err("failed to open trace_pipe\n"); 608 goto out_reset; 609 } 610 611 trace_fd = open(trace_file, O_RDONLY); 612 613 put_tracing_file(trace_file); 614 615 if (trace_fd < 0) { 616 pr_err("failed to open trace_pipe\n"); 617 goto out_reset; 618 } 619 620 fcntl(trace_fd, F_SETFL, O_NONBLOCK); 621 pollfd.fd = trace_fd; 622 623 /* display column headers */ 624 read_tracing_file_to_stdout("trace"); 625 626 if (!ftrace->target.initial_delay) { 627 if (write_tracing_file("tracing_on", "1") < 0) { 628 pr_err("can't enable tracing\n"); 629 goto out_close_fd; 630 } 631 } 632 633 evlist__start_workload(ftrace->evlist); 634 635 if (ftrace->target.initial_delay > 0) { 636 usleep(ftrace->target.initial_delay * 1000); 637 if (write_tracing_file("tracing_on", "1") < 0) { 638 pr_err("can't enable tracing\n"); 639 goto out_close_fd; 640 } 641 } 642 643 while (!done) { 644 if (poll(&pollfd, 1, -1) < 0) 645 break; 646 647 if (pollfd.revents & POLLIN) { 648 int n = read(trace_fd, buf, sizeof(buf)); 649 if (n < 0) 650 break; 651 if (fwrite(buf, n, 1, stdout) != 1) 652 break; 653 /* flush output since stdout is in full buffering mode due to pager */ 654 fflush(stdout); 655 } 656 } 657 658 write_tracing_file("tracing_on", "0"); 659 660 if (workload_exec_errno) { 661 const char *emsg = str_error_r(workload_exec_errno, buf, sizeof(buf)); 662 /* flush stdout first so below error msg appears at the end. */ 663 fflush(stdout); 664 pr_err("workload failed: %s\n", emsg); 665 goto out_close_fd; 666 } 667 668 /* read remaining buffer contents */ 669 while (true) { 670 int n = read(trace_fd, buf, sizeof(buf)); 671 if (n <= 0) 672 break; 673 if (fwrite(buf, n, 1, stdout) != 1) 674 break; 675 } 676 677out_close_fd: 678 close(trace_fd); 679out_reset: 680 reset_tracing_files(ftrace); 681out: 682 return (done && !workload_exec_errno) ? 0 : -1; 683} 684 685static void make_histogram(int buckets[], char *buf, size_t len, char *linebuf, 686 bool use_nsec) 687{ 688 char *p, *q; 689 char *unit; 690 double num; 691 int i; 692 693 /* ensure NUL termination */ 694 buf[len] = '\0'; 695 696 /* handle data line by line */ 697 for (p = buf; (q = strchr(p, '\n')) != NULL; p = q + 1) { 698 *q = '\0'; 699 /* move it to the line buffer */ 700 strcat(linebuf, p); 701 702 /* 703 * parse trace output to get function duration like in 704 * 705 * # tracer: function_graph 706 * # 707 * # CPU DURATION FUNCTION CALLS 708 * # | | | | | | | 709 * 1) + 10.291 us | do_filp_open(); 710 * 1) 4.889 us | do_filp_open(); 711 * 1) 6.086 us | do_filp_open(); 712 * 713 */ 714 if (linebuf[0] == '#') 715 goto next; 716 717 /* ignore CPU */ 718 p = strchr(linebuf, ')'); 719 if (p == NULL) 720 p = linebuf; 721 722 while (*p && !isdigit(*p) && (*p != '|')) 723 p++; 724 725 /* no duration */ 726 if (*p == '\0' || *p == '|') 727 goto next; 728 729 num = strtod(p, &unit); 730 if (!unit || strncmp(unit, " us", 3)) 731 goto next; 732 733 if (use_nsec) 734 num *= 1000; 735 736 i = log2(num); 737 if (i < 0) 738 i = 0; 739 if (i >= NUM_BUCKET) 740 i = NUM_BUCKET - 1; 741 742 buckets[i]++; 743 744next: 745 /* empty the line buffer for the next output */ 746 linebuf[0] = '\0'; 747 } 748 749 /* preserve any remaining output (before newline) */ 750 strcat(linebuf, p); 751} 752 753static void display_histogram(int buckets[], bool use_nsec) 754{ 755 int i; 756 int total = 0; 757 int bar_total = 46; /* to fit in 80 column */ 758 char bar[] = "###############################################"; 759 int bar_len; 760 761 for (i = 0; i < NUM_BUCKET; i++) 762 total += buckets[i]; 763 764 if (total == 0) { 765 printf("No data found\n"); 766 return; 767 } 768 769 printf("# %14s | %10s | %-*s |\n", 770 " DURATION ", "COUNT", bar_total, "GRAPH"); 771 772 bar_len = buckets[0] * bar_total / total; 773 printf(" %4d - %-4d %s | %10d | %.*s%*s |\n", 774 0, 1, "us", buckets[0], bar_len, bar, bar_total - bar_len, ""); 775 776 for (i = 1; i < NUM_BUCKET - 1; i++) { 777 int start = (1 << (i - 1)); 778 int stop = 1 << i; 779 const char *unit = use_nsec ? "ns" : "us"; 780 781 if (start >= 1024) { 782 start >>= 10; 783 stop >>= 10; 784 unit = use_nsec ? "us" : "ms"; 785 } 786 bar_len = buckets[i] * bar_total / total; 787 printf(" %4d - %-4d %s | %10d | %.*s%*s |\n", 788 start, stop, unit, buckets[i], bar_len, bar, 789 bar_total - bar_len, ""); 790 } 791 792 bar_len = buckets[NUM_BUCKET - 1] * bar_total / total; 793 printf(" %4d - %-4s %s | %10d | %.*s%*s |\n", 794 1, "...", use_nsec ? "ms" : " s", buckets[NUM_BUCKET - 1], 795 bar_len, bar, bar_total - bar_len, ""); 796 797} 798 799static int prepare_func_latency(struct perf_ftrace *ftrace) 800{ 801 char *trace_file; 802 int fd; 803 804 if (ftrace->target.use_bpf) 805 return perf_ftrace__latency_prepare_bpf(ftrace); 806 807 if (reset_tracing_files(ftrace) < 0) { 808 pr_err("failed to reset ftrace\n"); 809 return -1; 810 } 811 812 /* reset ftrace buffer */ 813 if (write_tracing_file("trace", "0") < 0) 814 return -1; 815 816 if (set_tracing_options(ftrace) < 0) 817 return -1; 818 819 /* force to use the function_graph tracer to track duration */ 820 if (write_tracing_file("current_tracer", "function_graph") < 0) { 821 pr_err("failed to set current_tracer to function_graph\n"); 822 return -1; 823 } 824 825 trace_file = get_tracing_file("trace_pipe"); 826 if (!trace_file) { 827 pr_err("failed to open trace_pipe\n"); 828 return -1; 829 } 830 831 fd = open(trace_file, O_RDONLY); 832 if (fd < 0) 833 pr_err("failed to open trace_pipe\n"); 834 835 put_tracing_file(trace_file); 836 return fd; 837} 838 839static int start_func_latency(struct perf_ftrace *ftrace) 840{ 841 if (ftrace->target.use_bpf) 842 return perf_ftrace__latency_start_bpf(ftrace); 843 844 if (write_tracing_file("tracing_on", "1") < 0) { 845 pr_err("can't enable tracing\n"); 846 return -1; 847 } 848 849 return 0; 850} 851 852static int stop_func_latency(struct perf_ftrace *ftrace) 853{ 854 if (ftrace->target.use_bpf) 855 return perf_ftrace__latency_stop_bpf(ftrace); 856 857 write_tracing_file("tracing_on", "0"); 858 return 0; 859} 860 861static int read_func_latency(struct perf_ftrace *ftrace, int buckets[]) 862{ 863 if (ftrace->target.use_bpf) 864 return perf_ftrace__latency_read_bpf(ftrace, buckets); 865 866 return 0; 867} 868 869static int cleanup_func_latency(struct perf_ftrace *ftrace) 870{ 871 if (ftrace->target.use_bpf) 872 return perf_ftrace__latency_cleanup_bpf(ftrace); 873 874 reset_tracing_files(ftrace); 875 return 0; 876} 877 878static int __cmd_latency(struct perf_ftrace *ftrace) 879{ 880 int trace_fd; 881 char buf[4096]; 882 char line[256]; 883 struct pollfd pollfd = { 884 .events = POLLIN, 885 }; 886 int buckets[NUM_BUCKET] = { }; 887 888 if (!(perf_cap__capable(CAP_PERFMON) || 889 perf_cap__capable(CAP_SYS_ADMIN))) { 890 pr_err("ftrace only works for %s!\n", 891#ifdef HAVE_LIBCAP_SUPPORT 892 "users with the CAP_PERFMON or CAP_SYS_ADMIN capability" 893#else 894 "root" 895#endif 896 ); 897 return -1; 898 } 899 900 trace_fd = prepare_func_latency(ftrace); 901 if (trace_fd < 0) 902 goto out; 903 904 fcntl(trace_fd, F_SETFL, O_NONBLOCK); 905 pollfd.fd = trace_fd; 906 907 if (start_func_latency(ftrace) < 0) 908 goto out; 909 910 evlist__start_workload(ftrace->evlist); 911 912 line[0] = '\0'; 913 while (!done) { 914 if (poll(&pollfd, 1, -1) < 0) 915 break; 916 917 if (pollfd.revents & POLLIN) { 918 int n = read(trace_fd, buf, sizeof(buf) - 1); 919 if (n < 0) 920 break; 921 922 make_histogram(buckets, buf, n, line, ftrace->use_nsec); 923 } 924 } 925 926 stop_func_latency(ftrace); 927 928 if (workload_exec_errno) { 929 const char *emsg = str_error_r(workload_exec_errno, buf, sizeof(buf)); 930 pr_err("workload failed: %s\n", emsg); 931 goto out; 932 } 933 934 /* read remaining buffer contents */ 935 while (!ftrace->target.use_bpf) { 936 int n = read(trace_fd, buf, sizeof(buf) - 1); 937 if (n <= 0) 938 break; 939 make_histogram(buckets, buf, n, line, ftrace->use_nsec); 940 } 941 942 read_func_latency(ftrace, buckets); 943 944 display_histogram(buckets, ftrace->use_nsec); 945 946out: 947 close(trace_fd); 948 cleanup_func_latency(ftrace); 949 950 return (done && !workload_exec_errno) ? 0 : -1; 951} 952 953static int perf_ftrace_config(const char *var, const char *value, void *cb) 954{ 955 struct perf_ftrace *ftrace = cb; 956 957 if (!strstarts(var, "ftrace.")) 958 return 0; 959 960 if (strcmp(var, "ftrace.tracer")) 961 return -1; 962 963 if (!strcmp(value, "function_graph") || 964 !strcmp(value, "function")) { 965 ftrace->tracer = value; 966 return 0; 967 } 968 969 pr_err("Please select \"function_graph\" (default) or \"function\"\n"); 970 return -1; 971} 972 973static void list_function_cb(char *str, void *arg) 974{ 975 struct strfilter *filter = (struct strfilter *)arg; 976 977 if (strfilter__compare(filter, str)) 978 printf("%s", str); 979} 980 981static int opt_list_avail_functions(const struct option *opt __maybe_unused, 982 const char *str, int unset) 983{ 984 struct strfilter *filter; 985 const char *err = NULL; 986 int ret; 987 988 if (unset || !str) 989 return -1; 990 991 filter = strfilter__new(str, &err); 992 if (!filter) 993 return err ? -EINVAL : -ENOMEM; 994 995 ret = strfilter__or(filter, str, &err); 996 if (ret == -EINVAL) { 997 pr_err("Filter parse error at %td.\n", err - str + 1); 998 pr_err("Source: \"%s\"\n", str); 999 pr_err(" %*c\n", (int)(err - str + 1), '^'); 1000 strfilter__delete(filter); 1001 return ret; 1002 } 1003 1004 ret = read_tracing_file_by_line("available_filter_functions", 1005 list_function_cb, filter); 1006 strfilter__delete(filter); 1007 if (ret < 0) 1008 return ret; 1009 1010 exit(0); 1011} 1012 1013static int parse_filter_func(const struct option *opt, const char *str, 1014 int unset __maybe_unused) 1015{ 1016 struct list_head *head = opt->value; 1017 struct filter_entry *entry; 1018 1019 entry = malloc(sizeof(*entry) + strlen(str) + 1); 1020 if (entry == NULL) 1021 return -ENOMEM; 1022 1023 strcpy(entry->name, str); 1024 list_add_tail(&entry->list, head); 1025 1026 return 0; 1027} 1028 1029static void delete_filter_func(struct list_head *head) 1030{ 1031 struct filter_entry *pos, *tmp; 1032 1033 list_for_each_entry_safe(pos, tmp, head, list) { 1034 list_del_init(&pos->list); 1035 free(pos); 1036 } 1037} 1038 1039static int parse_buffer_size(const struct option *opt, 1040 const char *str, int unset) 1041{ 1042 unsigned long *s = (unsigned long *)opt->value; 1043 static struct parse_tag tags_size[] = { 1044 { .tag = 'B', .mult = 1 }, 1045 { .tag = 'K', .mult = 1 << 10 }, 1046 { .tag = 'M', .mult = 1 << 20 }, 1047 { .tag = 'G', .mult = 1 << 30 }, 1048 { .tag = 0 }, 1049 }; 1050 unsigned long val; 1051 1052 if (unset) { 1053 *s = 0; 1054 return 0; 1055 } 1056 1057 val = parse_tag_value(str, tags_size); 1058 if (val != (unsigned long) -1) { 1059 if (val < 1024) { 1060 pr_err("buffer size too small, must larger than 1KB."); 1061 return -1; 1062 } 1063 *s = val; 1064 return 0; 1065 } 1066 1067 return -1; 1068} 1069 1070static int parse_func_tracer_opts(const struct option *opt, 1071 const char *str, int unset) 1072{ 1073 int ret; 1074 struct perf_ftrace *ftrace = (struct perf_ftrace *) opt->value; 1075 struct sublevel_option func_tracer_opts[] = { 1076 { .name = "call-graph", .value_ptr = &ftrace->func_stack_trace }, 1077 { .name = "irq-info", .value_ptr = &ftrace->func_irq_info }, 1078 { .name = NULL, } 1079 }; 1080 1081 if (unset) 1082 return 0; 1083 1084 ret = perf_parse_sublevel_options(str, func_tracer_opts); 1085 if (ret) 1086 return ret; 1087 1088 return 0; 1089} 1090 1091static int parse_graph_tracer_opts(const struct option *opt, 1092 const char *str, int unset) 1093{ 1094 int ret; 1095 struct perf_ftrace *ftrace = (struct perf_ftrace *) opt->value; 1096 struct sublevel_option graph_tracer_opts[] = { 1097 { .name = "nosleep-time", .value_ptr = &ftrace->graph_nosleep_time }, 1098 { .name = "noirqs", .value_ptr = &ftrace->graph_noirqs }, 1099 { .name = "verbose", .value_ptr = &ftrace->graph_verbose }, 1100 { .name = "thresh", .value_ptr = &ftrace->graph_thresh }, 1101 { .name = "depth", .value_ptr = &ftrace->graph_depth }, 1102 { .name = NULL, } 1103 }; 1104 1105 if (unset) 1106 return 0; 1107 1108 ret = perf_parse_sublevel_options(str, graph_tracer_opts); 1109 if (ret) 1110 return ret; 1111 1112 return 0; 1113} 1114 1115enum perf_ftrace_subcommand { 1116 PERF_FTRACE_NONE, 1117 PERF_FTRACE_TRACE, 1118 PERF_FTRACE_LATENCY, 1119}; 1120 1121int cmd_ftrace(int argc, const char **argv) 1122{ 1123 int ret; 1124 int (*cmd_func)(struct perf_ftrace *) = NULL; 1125 struct perf_ftrace ftrace = { 1126 .tracer = DEFAULT_TRACER, 1127 .target = { .uid = UINT_MAX, }, 1128 }; 1129 const struct option common_options[] = { 1130 OPT_STRING('p', "pid", &ftrace.target.pid, "pid", 1131 "Trace on existing process id"), 1132 /* TODO: Add short option -t after -t/--tracer can be removed. */ 1133 OPT_STRING(0, "tid", &ftrace.target.tid, "tid", 1134 "Trace on existing thread id (exclusive to --pid)"), 1135 OPT_INCR('v', "verbose", &verbose, 1136 "Be more verbose"), 1137 OPT_BOOLEAN('a', "all-cpus", &ftrace.target.system_wide, 1138 "System-wide collection from all CPUs"), 1139 OPT_STRING('C', "cpu", &ftrace.target.cpu_list, "cpu", 1140 "List of cpus to monitor"), 1141 OPT_END() 1142 }; 1143 const struct option ftrace_options[] = { 1144 OPT_STRING('t', "tracer", &ftrace.tracer, "tracer", 1145 "Tracer to use: function_graph(default) or function"), 1146 OPT_CALLBACK_DEFAULT('F', "funcs", NULL, "[FILTER]", 1147 "Show available functions to filter", 1148 opt_list_avail_functions, "*"), 1149 OPT_CALLBACK('T', "trace-funcs", &ftrace.filters, "func", 1150 "Trace given functions using function tracer", 1151 parse_filter_func), 1152 OPT_CALLBACK('N', "notrace-funcs", &ftrace.notrace, "func", 1153 "Do not trace given functions", parse_filter_func), 1154 OPT_CALLBACK(0, "func-opts", &ftrace, "options", 1155 "Function tracer options, available options: call-graph,irq-info", 1156 parse_func_tracer_opts), 1157 OPT_CALLBACK('G', "graph-funcs", &ftrace.graph_funcs, "func", 1158 "Trace given functions using function_graph tracer", 1159 parse_filter_func), 1160 OPT_CALLBACK('g', "nograph-funcs", &ftrace.nograph_funcs, "func", 1161 "Set nograph filter on given functions", parse_filter_func), 1162 OPT_CALLBACK(0, "graph-opts", &ftrace, "options", 1163 "Graph tracer options, available options: nosleep-time,noirqs,verbose,thresh=<n>,depth=<n>", 1164 parse_graph_tracer_opts), 1165 OPT_CALLBACK('m', "buffer-size", &ftrace.percpu_buffer_size, "size", 1166 "Size of per cpu buffer, needs to use a B, K, M or G suffix.", parse_buffer_size), 1167 OPT_BOOLEAN(0, "inherit", &ftrace.inherit, 1168 "Trace children processes"), 1169 OPT_INTEGER('D', "delay", &ftrace.target.initial_delay, 1170 "Number of milliseconds to wait before starting tracing after program start"), 1171 OPT_PARENT(common_options), 1172 }; 1173 const struct option latency_options[] = { 1174 OPT_CALLBACK('T', "trace-funcs", &ftrace.filters, "func", 1175 "Show latency of given function", parse_filter_func), 1176#ifdef HAVE_BPF_SKEL 1177 OPT_BOOLEAN('b', "use-bpf", &ftrace.target.use_bpf, 1178 "Use BPF to measure function latency"), 1179#endif 1180 OPT_BOOLEAN('n', "use-nsec", &ftrace.use_nsec, 1181 "Use nano-second histogram"), 1182 OPT_PARENT(common_options), 1183 }; 1184 const struct option *options = ftrace_options; 1185 1186 const char * const ftrace_usage[] = { 1187 "perf ftrace [<options>] [<command>]", 1188 "perf ftrace [<options>] -- [<command>] [<options>]", 1189 "perf ftrace {trace|latency} [<options>] [<command>]", 1190 "perf ftrace {trace|latency} [<options>] -- [<command>] [<options>]", 1191 NULL 1192 }; 1193 enum perf_ftrace_subcommand subcmd = PERF_FTRACE_NONE; 1194 1195 INIT_LIST_HEAD(&ftrace.filters); 1196 INIT_LIST_HEAD(&ftrace.notrace); 1197 INIT_LIST_HEAD(&ftrace.graph_funcs); 1198 INIT_LIST_HEAD(&ftrace.nograph_funcs); 1199 1200 signal(SIGINT, sig_handler); 1201 signal(SIGUSR1, sig_handler); 1202 signal(SIGCHLD, sig_handler); 1203 signal(SIGPIPE, sig_handler); 1204 1205 ret = perf_config(perf_ftrace_config, &ftrace); 1206 if (ret < 0) 1207 return -1; 1208 1209 if (argc > 1) { 1210 if (!strcmp(argv[1], "trace")) { 1211 subcmd = PERF_FTRACE_TRACE; 1212 } else if (!strcmp(argv[1], "latency")) { 1213 subcmd = PERF_FTRACE_LATENCY; 1214 options = latency_options; 1215 } 1216 1217 if (subcmd != PERF_FTRACE_NONE) { 1218 argc--; 1219 argv++; 1220 } 1221 } 1222 /* for backward compatibility */ 1223 if (subcmd == PERF_FTRACE_NONE) 1224 subcmd = PERF_FTRACE_TRACE; 1225 1226 argc = parse_options(argc, argv, options, ftrace_usage, 1227 PARSE_OPT_STOP_AT_NON_OPTION); 1228 if (argc < 0) { 1229 ret = -EINVAL; 1230 goto out_delete_filters; 1231 } 1232 1233 /* Make system wide (-a) the default target. */ 1234 if (!argc && target__none(&ftrace.target)) 1235 ftrace.target.system_wide = true; 1236 1237 switch (subcmd) { 1238 case PERF_FTRACE_TRACE: 1239 cmd_func = __cmd_ftrace; 1240 break; 1241 case PERF_FTRACE_LATENCY: 1242 if (list_empty(&ftrace.filters)) { 1243 pr_err("Should provide a function to measure\n"); 1244 parse_options_usage(ftrace_usage, options, "T", 1); 1245 ret = -EINVAL; 1246 goto out_delete_filters; 1247 } 1248 cmd_func = __cmd_latency; 1249 break; 1250 case PERF_FTRACE_NONE: 1251 default: 1252 pr_err("Invalid subcommand\n"); 1253 ret = -EINVAL; 1254 goto out_delete_filters; 1255 } 1256 1257 ret = target__validate(&ftrace.target); 1258 if (ret) { 1259 char errbuf[512]; 1260 1261 target__strerror(&ftrace.target, ret, errbuf, 512); 1262 pr_err("%s\n", errbuf); 1263 goto out_delete_filters; 1264 } 1265 1266 ftrace.evlist = evlist__new(); 1267 if (ftrace.evlist == NULL) { 1268 ret = -ENOMEM; 1269 goto out_delete_filters; 1270 } 1271 1272 ret = evlist__create_maps(ftrace.evlist, &ftrace.target); 1273 if (ret < 0) 1274 goto out_delete_evlist; 1275 1276 if (argc) { 1277 ret = evlist__prepare_workload(ftrace.evlist, &ftrace.target, 1278 argv, false, 1279 ftrace__workload_exec_failed_signal); 1280 if (ret < 0) 1281 goto out_delete_evlist; 1282 } 1283 1284 ret = cmd_func(&ftrace); 1285 1286out_delete_evlist: 1287 evlist__delete(ftrace.evlist); 1288 1289out_delete_filters: 1290 delete_filter_func(&ftrace.filters); 1291 delete_filter_func(&ftrace.notrace); 1292 delete_filter_func(&ftrace.graph_funcs); 1293 delete_filter_func(&ftrace.nograph_funcs); 1294 1295 return ret; 1296} 1297