tools/perf/util/annotate.c

   1 /*
   2  * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
   3  *
   4  * Parts came from builtin-annotate.c, see those files for further
   5  * copyright notes.
   6  *
   7  * Released under the GPL v2. (and only v2, not any later version)
   8  */
   9
  10 #include "util.h"
  11 #include "build-id.h"
  12 #include "color.h"
  13 #include "cache.h"
  14 #include "symbol.h"
  15 #include "debug.h"
  16 #include "annotate.h"
  17 #include "evsel.h"
  18 #include <pthread.h>
  19 #include <linux/bitops.h>
  20
  21 const char      *disassembler_style;
  22 const char      *objdump_path;
  23
  24 static struct ins *ins__find(const char *name);
  25 static int disasm_line__parse(char *line, char **namep, char **rawp);
  26
  27 static void ins__delete(struct ins_operands *ops)
  28 {
  29         free(ops->source.raw);
  30         free(ops->source.name);
  31         free(ops->target.raw);
  32         free(ops->target.name);
  33 }
  34
  35 static int ins__raw_scnprintf(struct ins *ins, char *bf, size_t size,
  36                               struct ins_operands *ops)
  37 {
  38         return scnprintf(bf, size, "%-6.6s %s", ins->name, ops->raw);
  39 }
  40
  41 int ins__scnprintf(struct ins *ins, char *bf, size_t size,
  42                   struct ins_operands *ops)
  43 {
  44         if (ins->ops->scnprintf)
  45                 return ins->ops->scnprintf(ins, bf, size, ops);
  46
  47         return ins__raw_scnprintf(ins, bf, size, ops);
  48 }
  49
  50 static int call__parse(struct ins_operands *ops)
  51 {
  52         char *endptr, *tok, *name;
  53
  54         ops->target.addr = strtoull(ops->raw, &endptr, 16);
  55
  56         name = strchr(endptr, '<');
  57         if (name == NULL)
  58                 goto indirect_call;
  59
  60         name++;
  61
  62         tok = strchr(name, '>');
  63         if (tok == NULL)
  64                 return -1;
  65
  66         *tok = '\0';
  67         ops->target.name = strdup(name);
  68         *tok = '>';
  69
  70         return ops->target.name == NULL ? -1 : 0;
  71
  72 indirect_call:
  73         tok = strchr(endptr, '(');
  74         if (tok != NULL) {
  75                 ops->target.addr = 0;
  76                 return 0;
  77         }
  78
  79         tok = strchr(endptr, '*');
  80         if (tok == NULL)
  81                 return -1;
  82
  83         ops->target.addr = strtoull(tok + 1, NULL, 16);
  84         return 0;
  85 }
  86
  87 static int call__scnprintf(struct ins *ins, char *bf, size_t size,
  88                            struct ins_operands *ops)
  89 {
  90         if (ops->target.name)
  91                 return scnprintf(bf, size, "%-6.6s %s", ins->name, ops->target.name);
  92
  93         if (ops->target.addr == 0)
  94                 return ins__raw_scnprintf(ins, bf, size, ops);
  95
  96         return scnprintf(bf, size, "%-6.6s *%" PRIx64, ins->name, ops->target.addr);
  97 }
  98
  99 static struct ins_ops call_ops = {
 100         .parse     = call__parse,
 101         .scnprintf = call__scnprintf,
 102 };
 103
 104 bool ins__is_call(const struct ins *ins)
 105 {
 106         return ins->ops == &call_ops;
 107 }
 108
 109 static int jump__parse(struct ins_operands *ops)
 110 {
 111         const char *s = strchr(ops->raw, '+');
 112
 113         ops->target.addr = strtoull(ops->raw, NULL, 16);
 114
 115         if (s++ != NULL)
 116                 ops->target.offset = strtoull(s, NULL, 16);
 117         else
 118                 ops->target.offset = UINT64_MAX;
 119
 120         return 0;
 121 }
 122
 123 static int jump__scnprintf(struct ins *ins, char *bf, size_t size,
 124                            struct ins_operands *ops)
 125 {
 126         return scnprintf(bf, size, "%-6.6s %" PRIx64, ins->name, ops->target.offset);
 127 }
 128
 129 static struct ins_ops jump_ops = {
 130         .parse     = jump__parse,
 131         .scnprintf = jump__scnprintf,
 132 };
 133
 134 bool ins__is_jump(const struct ins *ins)
 135 {
 136         return ins->ops == &jump_ops;
 137 }
 138
 139 static int comment__symbol(char *raw, char *comment, u64 *addrp, char **namep)
 140 {
 141         char *endptr, *name, *t;
 142
 143         if (strstr(raw, "(%rip)") == NULL)
 144                 return 0;
 145
 146         *addrp = strtoull(comment, &endptr, 16);
 147         name = strchr(endptr, '<');
 148         if (name == NULL)
 149                 return -1;
 150
 151         name++;
 152
 153         t = strchr(name, '>');
 154         if (t == NULL)
 155                 return 0;
 156
 157         *t = '\0';
 158         *namep = strdup(name);
 159         *t = '>';
 160
 161         return 0;
 162 }
 163
 164 static int lock__parse(struct ins_operands *ops)
 165 {
 166         char *name;
 167
 168         ops->locked.ops = zalloc(sizeof(*ops->locked.ops));
 169         if (ops->locked.ops == NULL)
 170                 return 0;
 171
 172         if (disasm_line__parse(ops->raw, &name, &ops->locked.ops->raw) < 0)
 173                 goto out_free_ops;
 174
 175         ops->locked.ins = ins__find(name);
 176         if (ops->locked.ins == NULL)
 177                 goto out_free_ops;
 178
 179         if (!ops->locked.ins->ops)
 180                 return 0;
 181
 182         if (ops->locked.ins->ops->parse)
 183                 ops->locked.ins->ops->parse(ops->locked.ops);
 184
 185         return 0;
 186
 187 out_free_ops:
 188         free(ops->locked.ops);
 189         ops->locked.ops = NULL;
 190         return 0;
 191 }
 192
 193 static int lock__scnprintf(struct ins *ins, char *bf, size_t size,
 194                            struct ins_operands *ops)
 195 {
 196         int printed;
 197
 198         if (ops->locked.ins == NULL)
 199                 return ins__raw_scnprintf(ins, bf, size, ops);
 200
 201         printed = scnprintf(bf, size, "%-6.6s ", ins->name);
 202         return printed + ins__scnprintf(ops->locked.ins, bf + printed,
 203                                         size - printed, ops->locked.ops);
 204 }
 205
 206 static void lock__delete(struct ins_operands *ops)
 207 {
 208         free(ops->locked.ops);
 209         free(ops->target.raw);
 210         free(ops->target.name);
 211 }
 212
 213 static struct ins_ops lock_ops = {
 214         .free      = lock__delete,
 215         .parse     = lock__parse,
 216         .scnprintf = lock__scnprintf,
 217 };
 218
 219 static int mov__parse(struct ins_operands *ops)
 220 {
 221         char *s = strchr(ops->raw, ','), *target, *comment, prev;
 222
 223         if (s == NULL)
 224                 return -1;
 225
 226         *s = '\0';
 227         ops->source.raw = strdup(ops->raw);
 228         *s = ',';
 229
 230         if (ops->source.raw == NULL)
 231                 return -1;
 232
 233         target = ++s;
 234
 235         while (s[0] != '\0' && !isspace(s[0]))
 236                 ++s;
 237         prev = *s;
 238         *s = '\0';
 239
 240         ops->target.raw = strdup(target);
 241         *s = prev;
 242
 243         if (ops->target.raw == NULL)
 244                 goto out_free_source;
 245
 246         comment = strchr(s, '#');
 247         if (comment == NULL)
 248                 return 0;
 249
 250         while (comment[0] != '\0' && isspace(comment[0]))
 251                 ++comment;
 252
 253         comment__symbol(ops->source.raw, comment, &ops->source.addr, &ops->source.name);
 254         comment__symbol(ops->target.raw, comment, &ops->target.addr, &ops->target.name);
 255
 256         return 0;
 257
 258 out_free_source:
 259         free(ops->source.raw);
 260         ops->source.raw = NULL;
 261         return -1;
 262 }
 263
 264 static int mov__scnprintf(struct ins *ins, char *bf, size_t size,
 265                            struct ins_operands *ops)
 266 {
 267         return scnprintf(bf, size, "%-6.6s %s,%s", ins->name,
 268                          ops->source.name ?: ops->source.raw,
 269                          ops->target.name ?: ops->target.raw);
 270 }
 271
 272 static struct ins_ops mov_ops = {
 273         .parse     = mov__parse,
 274         .scnprintf = mov__scnprintf,
 275 };
 276
 277 static int dec__parse(struct ins_operands *ops)
 278 {
 279         char *target, *comment, *s, prev;
 280
 281         target = s = ops->raw;
 282
 283         while (s[0] != '\0' && !isspace(s[0]))
 284                 ++s;
 285         prev = *s;
 286         *s = '\0';
 287
 288         ops->target.raw = strdup(target);
 289         *s = prev;
 290
 291         if (ops->target.raw == NULL)
 292                 return -1;
 293
 294         comment = strchr(s, '#');
 295         if (comment == NULL)
 296                 return 0;
 297
 298         while (comment[0] != '\0' && isspace(comment[0]))
 299                 ++comment;
 300
 301         comment__symbol(ops->target.raw, comment, &ops->target.addr, &ops->target.name);
 302
 303         return 0;
 304 }
 305
 306 static int dec__scnprintf(struct ins *ins, char *bf, size_t size,
 307                            struct ins_operands *ops)
 308 {
 309         return scnprintf(bf, size, "%-6.6s %s", ins->name,
 310                          ops->target.name ?: ops->target.raw);
 311 }
 312
 313 static struct ins_ops dec_ops = {
 314         .parse     = dec__parse,
 315         .scnprintf = dec__scnprintf,
 316 };
 317
 318 static int nop__scnprintf(struct ins *ins __maybe_unused, char *bf, size_t size,
 319                           struct ins_operands *ops __maybe_unused)
 320 {
 321         return scnprintf(bf, size, "%-6.6s", "nop");
 322 }
 323
 324 static struct ins_ops nop_ops = {
 325         .scnprintf = nop__scnprintf,
 326 };
 327
 328 /*
 329  * Must be sorted by name!
 330  */
 331 static struct ins instructions[] = {
 332         { .name = "add",   .ops  = &mov_ops, },
 333         { .name = "addl",  .ops  = &mov_ops, },
 334         { .name = "addq",  .ops  = &mov_ops, },
 335         { .name = "addw",  .ops  = &mov_ops, },
 336         { .name = "and",   .ops  = &mov_ops, },
 337         { .name = "bts",   .ops  = &mov_ops, },
 338         { .name = "call",  .ops  = &call_ops, },
 339         { .name = "callq", .ops  = &call_ops, },
 340         { .name = "cmp",   .ops  = &mov_ops, },
 341         { .name = "cmpb",  .ops  = &mov_ops, },
 342         { .name = "cmpl",  .ops  = &mov_ops, },
 343         { .name = "cmpq",  .ops  = &mov_ops, },
 344         { .name = "cmpw",  .ops  = &mov_ops, },
 345         { .name = "cmpxch", .ops  = &mov_ops, },
 346         { .name = "dec",   .ops  = &dec_ops, },
 347         { .name = "decl",  .ops  = &dec_ops, },
 348         { .name = "imul",  .ops  = &mov_ops, },
 349         { .name = "inc",   .ops  = &dec_ops, },
 350         { .name = "incl",  .ops  = &dec_ops, },
 351         { .name = "ja",    .ops  = &jump_ops, },
 352         { .name = "jae",   .ops  = &jump_ops, },
 353         { .name = "jb",    .ops  = &jump_ops, },
 354         { .name = "jbe",   .ops  = &jump_ops, },
 355         { .name = "jc",    .ops  = &jump_ops, },
 356         { .name = "jcxz",  .ops  = &jump_ops, },
 357         { .name = "je",    .ops  = &jump_ops, },
 358         { .name = "jecxz", .ops  = &jump_ops, },
 359         { .name = "jg",    .ops  = &jump_ops, },
 360         { .name = "jge",   .ops  = &jump_ops, },
 361         { .name = "jl",    .ops  = &jump_ops, },
 362         { .name = "jle",   .ops  = &jump_ops, },
 363         { .name = "jmp",   .ops  = &jump_ops, },
 364         { .name = "jmpq",  .ops  = &jump_ops, },
 365         { .name = "jna",   .ops  = &jump_ops, },
 366         { .name = "jnae",  .ops  = &jump_ops, },
 367         { .name = "jnb",   .ops  = &jump_ops, },
 368         { .name = "jnbe",  .ops  = &jump_ops, },
 369         { .name = "jnc",   .ops  = &jump_ops, },
 370         { .name = "jne",   .ops  = &jump_ops, },
 371         { .name = "jng",   .ops  = &jump_ops, },
 372         { .name = "jnge",  .ops  = &jump_ops, },
 373         { .name = "jnl",   .ops  = &jump_ops, },
 374         { .name = "jnle",  .ops  = &jump_ops, },
 375         { .name = "jno",   .ops  = &jump_ops, },
 376         { .name = "jnp",   .ops  = &jump_ops, },
 377         { .name = "jns",   .ops  = &jump_ops, },
 378         { .name = "jnz",   .ops  = &jump_ops, },
 379         { .name = "jo",    .ops  = &jump_ops, },
 380         { .name = "jp",    .ops  = &jump_ops, },
 381         { .name = "jpe",   .ops  = &jump_ops, },
 382         { .name = "jpo",   .ops  = &jump_ops, },
 383         { .name = "jrcxz", .ops  = &jump_ops, },
 384         { .name = "js",    .ops  = &jump_ops, },
 385         { .name = "jz",    .ops  = &jump_ops, },
 386         { .name = "lea",   .ops  = &mov_ops, },
 387         { .name = "lock",  .ops  = &lock_ops, },
 388         { .name = "mov",   .ops  = &mov_ops, },
 389         { .name = "movb",  .ops  = &mov_ops, },
 390         { .name = "movdqa",.ops  = &mov_ops, },
 391         { .name = "movl",  .ops  = &mov_ops, },
 392         { .name = "movq",  .ops  = &mov_ops, },
 393         { .name = "movslq", .ops  = &mov_ops, },
 394         { .name = "movzbl", .ops  = &mov_ops, },
 395         { .name = "movzwl", .ops  = &mov_ops, },
 396         { .name = "nop",   .ops  = &nop_ops, },
 397         { .name = "nopl",  .ops  = &nop_ops, },
 398         { .name = "nopw",  .ops  = &nop_ops, },
 399         { .name = "or",    .ops  = &mov_ops, },
 400         { .name = "orl",   .ops  = &mov_ops, },
 401         { .name = "test",  .ops  = &mov_ops, },
 402         { .name = "testb", .ops  = &mov_ops, },
 403         { .name = "testl", .ops  = &mov_ops, },
 404         { .name = "xadd",  .ops  = &mov_ops, },
 405         { .name = "xbeginl", .ops  = &jump_ops, },
 406         { .name = "xbeginq", .ops  = &jump_ops, },
 407 };
 408
 409 static int ins__cmp(const void *name, const void *insp)
 410 {
 411         const struct ins *ins = insp;
 412
 413         return strcmp(name, ins->name);
 414 }
 415
 416 static struct ins *ins__find(const char *name)
 417 {
 418         const int nmemb = ARRAY_SIZE(instructions);
 419
 420         return bsearch(name, instructions, nmemb, sizeof(struct ins), ins__cmp);
 421 }
 422
 423 int symbol__annotate_init(struct map *map __maybe_unused, struct symbol *sym)
 424 {
 425         struct annotation *notes = symbol__annotation(sym);
 426         pthread_mutex_init(&notes->lock, NULL);
 427         return 0;
 428 }
 429
 430 int symbol__alloc_hist(struct symbol *sym)
 431 {
 432         struct annotation *notes = symbol__annotation(sym);
 433         const size_t size = symbol__size(sym);
 434         size_t sizeof_sym_hist;
 435
 436         /* Check for overflow when calculating sizeof_sym_hist */
 437         if (size > (SIZE_MAX - sizeof(struct sym_hist)) / sizeof(u64))
 438                 return -1;
 439
 440         sizeof_sym_hist = (sizeof(struct sym_hist) + size * sizeof(u64));
 441
 442         /* Check for overflow in zalloc argument */
 443         if (sizeof_sym_hist > (SIZE_MAX - sizeof(*notes->src))
 444                                 / symbol_conf.nr_events)
 445                 return -1;
 446
 447         notes->src = zalloc(sizeof(*notes->src) + symbol_conf.nr_events * sizeof_sym_hist);
 448         if (notes->src == NULL)
 449                 return -1;
 450         notes->src->sizeof_sym_hist = sizeof_sym_hist;
 451         notes->src->nr_histograms   = symbol_conf.nr_events;
 452         INIT_LIST_HEAD(&notes->src->source);
 453         return 0;
 454 }
 455
 456 void symbol__annotate_zero_histograms(struct symbol *sym)
 457 {
 458         struct annotation *notes = symbol__annotation(sym);
 459
 460         pthread_mutex_lock(&notes->lock);
 461         if (notes->src != NULL)
 462                 memset(notes->src->histograms, 0,
 463                        notes->src->nr_histograms * notes->src->sizeof_sym_hist);
 464         pthread_mutex_unlock(&notes->lock);
 465 }
 466
 467 int symbol__inc_addr_samples(struct symbol *sym, struct map *map,
 468                              int evidx, u64 addr)
 469 {
 470         unsigned offset;
 471         struct annotation *notes;
 472         struct sym_hist *h;
 473
 474         notes = symbol__annotation(sym);
 475         if (notes->src == NULL)
 476                 return -ENOMEM;
 477
 478         pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map->unmap_ip(map, addr));
 479
 480         if (addr < sym->start || addr > sym->end)
 481                 return -ERANGE;
 482
 483         offset = addr - sym->start;
 484         h = annotation__histogram(notes, evidx);
 485         h->sum++;
 486         h->addr[offset]++;
 487
 488         pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64
 489                   ", evidx=%d] => %" PRIu64 "\n", sym->start, sym->name,
 490                   addr, addr - sym->start, evidx, h->addr[offset]);
 491         return 0;
 492 }
 493
 494 static void disasm_line__init_ins(struct disasm_line *dl)
 495 {
 496         dl->ins = ins__find(dl->name);
 497
 498         if (dl->ins == NULL)
 499                 return;
 500
 501         if (!dl->ins->ops)
 502                 return;
 503
 504         if (dl->ins->ops->parse)
 505                 dl->ins->ops->parse(&dl->ops);
 506 }
 507
 508 static int disasm_line__parse(char *line, char **namep, char **rawp)
 509 {
 510         char *name = line, tmp;
 511
 512         while (isspace(name[0]))
 513                 ++name;
 514
 515         if (name[0] == '\0')
 516                 return -1;
 517
 518         *rawp = name + 1;
 519
 520         while ((*rawp)[0] != '\0' && !isspace((*rawp)[0]))
 521                 ++*rawp;
 522
 523         tmp = (*rawp)[0];
 524         (*rawp)[0] = '\0';
 525         *namep = strdup(name);
 526
 527         if (*namep == NULL)
 528                 goto out_free_name;
 529
 530         (*rawp)[0] = tmp;
 531
 532         if ((*rawp)[0] != '\0') {
 533                 (*rawp)++;
 534                 while (isspace((*rawp)[0]))
 535                         ++(*rawp);
 536         }
 537
 538         return 0;
 539
 540 out_free_name:
 541         free(*namep);
 542         *namep = NULL;
 543         return -1;
 544 }
 545
 546 static struct disasm_line *disasm_line__new(s64 offset, char *line, size_t privsize)
 547 {
 548         struct disasm_line *dl = zalloc(sizeof(*dl) + privsize);
 549
 550         if (dl != NULL) {
 551                 dl->offset = offset;
 552                 dl->line = strdup(line);
 553                 if (dl->line == NULL)
 554                         goto out_delete;
 555
 556                 if (offset != -1) {
 557                         if (disasm_line__parse(dl->line, &dl->name, &dl->ops.raw) < 0)
 558                                 goto out_free_line;
 559
 560                         disasm_line__init_ins(dl);
 561                 }
 562         }
 563
 564         return dl;
 565
 566 out_free_line:
 567         free(dl->line);
 568 out_delete:
 569         free(dl);
 570         return NULL;
 571 }
 572
 573 void disasm_line__free(struct disasm_line *dl)
 574 {
 575         free(dl->line);
 576         free(dl->name);
 577         if (dl->ins && dl->ins->ops->free)
 578                 dl->ins->ops->free(&dl->ops);
 579         else
 580                 ins__delete(&dl->ops);
 581         free(dl);
 582 }
 583
 584 int disasm_line__scnprintf(struct disasm_line *dl, char *bf, size_t size, bool raw)
 585 {
 586         if (raw || !dl->ins)
 587                 return scnprintf(bf, size, "%-6.6s %s", dl->name, dl->ops.raw);
 588
 589         return ins__scnprintf(dl->ins, bf, size, &dl->ops);
 590 }
 591
 592 static void disasm__add(struct list_head *head, struct disasm_line *line)
 593 {
 594         list_add_tail(&line->node, head);
 595 }
 596
 597 struct disasm_line *disasm__get_next_ip_line(struct list_head *head, struct disasm_line *pos)
 598 {
 599         list_for_each_entry_continue(pos, head, node)
 600                 if (pos->offset >= 0)
 601                         return pos;
 602
 603         return NULL;
 604 }
 605
 606 double disasm__calc_percent(struct annotation *notes, int evidx, s64 offset,
 607                             s64 end, const char **path)
 608 {
 609         struct source_line *src_line = notes->src->lines;
 610         double percent = 0.0;
 611
 612         if (src_line) {
 613                 size_t sizeof_src_line = sizeof(*src_line) +
 614                                 sizeof(src_line->p) * (src_line->nr_pcnt - 1);
 615
 616                 while (offset < end) {
 617                         src_line = (void *)notes->src->lines +
 618                                         (sizeof_src_line * offset);
 619
 620                         if (*path == NULL)
 621                                 *path = src_line->path;
 622
 623                         percent += src_line->p[evidx].percent;
 624                         offset++;
 625                 }
 626         } else {
 627                 struct sym_hist *h = annotation__histogram(notes, evidx);
 628                 unsigned int hits = 0;
 629
 630                 while (offset < end)
 631                         hits += h->addr[offset++];
 632
 633                 if (h->sum)
 634                         percent = 100.0 * hits / h->sum;
 635         }
 636
 637         return percent;
 638 }
 639
 640 static int disasm_line__print(struct disasm_line *dl, struct symbol *sym, u64 start,
 641                       struct perf_evsel *evsel, u64 len, int min_pcnt, int printed,
 642                       int max_lines, struct disasm_line *queue)
 643 {
 644         static const char *prev_line;
 645         static const char *prev_color;
 646
 647         if (dl->offset != -1) {
 648                 const char *path = NULL;
 649                 double percent, max_percent = 0.0;
 650                 double *ppercents = &percent;
 651                 int i, nr_percent = 1;
 652                 const char *color;
 653                 struct annotation *notes = symbol__annotation(sym);
 654                 s64 offset = dl->offset;
 655                 const u64 addr = start + offset;
 656                 struct disasm_line *next;
 657
 658                 next = disasm__get_next_ip_line(&notes->src->source, dl);
 659
 660                 if (perf_evsel__is_group_event(evsel)) {
 661                         nr_percent = evsel->nr_members;
 662                         ppercents = calloc(nr_percent, sizeof(double));
 663                         if (ppercents == NULL)
 664                                 return -1;
 665                 }
 666
 667                 for (i = 0; i < nr_percent; i++) {
 668                         percent = disasm__calc_percent(notes,
 669                                         notes->src->lines ? i : evsel->idx + i,
 670                                         offset,
 671                                         next ? next->offset : (s64) len,
 672                                         &path);
 673
 674                         ppercents[i] = percent;
 675                         if (percent > max_percent)
 676                                 max_percent = percent;
 677                 }
 678
 679                 if (max_percent < min_pcnt)
 680                         return -1;
 681
 682                 if (max_lines && printed >= max_lines)
 683                         return 1;
 684
 685                 if (queue != NULL) {
 686                         list_for_each_entry_from(queue, &notes->src->source, node) {
 687                                 if (queue == dl)
 688                                         break;
 689                                 disasm_line__print(queue, sym, start, evsel, len,
 690                                                     0, 0, 1, NULL);
 691                         }
 692                 }
 693
 694                 color = get_percent_color(max_percent);
 695
 696                 /*
 697                  * Also color the filename and line if needed, with
 698                  * the same color than the percentage. Don't print it
 699                  * twice for close colored addr with the same filename:line
 700                  */
 701                 if (path) {
 702                         if (!prev_line || strcmp(prev_line, path)
 703                                        || color != prev_color) {
 704                                 color_fprintf(stdout, color, " %s", path);
 705                                 prev_line = path;
 706                                 prev_color = color;
 707                         }
 708                 }
 709
 710                 for (i = 0; i < nr_percent; i++) {
 711                         percent = ppercents[i];
 712                         color = get_percent_color(percent);
 713                         color_fprintf(stdout, color, " %7.2f", percent);
 714                 }
 715
 716                 printf(" :      ");
 717                 color_fprintf(stdout, PERF_COLOR_MAGENTA, "  %" PRIx64 ":", addr);
 718                 color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", dl->line);
 719
 720                 if (ppercents != &percent)
 721                         free(ppercents);
 722
 723         } else if (max_lines && printed >= max_lines)
 724                 return 1;
 725         else {
 726                 int width = 8;
 727
 728                 if (queue)
 729                         return -1;
 730
 731                 if (perf_evsel__is_group_event(evsel))
 732                         width *= evsel->nr_members;
 733
 734                 if (!*dl->line)
 735                         printf(" %*s:\n", width, " ");
 736                 else
 737                         printf(" %*s:   %s\n", width, " ", dl->line);
 738         }
 739
 740         return 0;
 741 }
 742
 743 /*
 744  * symbol__parse_objdump_line() parses objdump output (with -d --no-show-raw)
 745  * which looks like following
 746  *
 747  *  0000000000415500 <_init>:
 748  *    415500:       sub    $0x8,%rsp
 749  *    415504:       mov    0x2f5ad5(%rip),%rax        # 70afe0 <_DYNAMIC+0x2f8>
 750  *    41550b:       test   %rax,%rax
 751  *    41550e:       je     415515 <_init+0x15>
 752  *    415510:       callq  416e70 <__gmon_start__@plt>
 753  *    415515:       add    $0x8,%rsp
 754  *    415519:       retq
 755  *
 756  * it will be parsed and saved into struct disasm_line as
 757  *  <offset>       <name>  <ops.raw>
 758  *
 759  * The offset will be a relative offset from the start of the symbol and -1
 760  * means that it's not a disassembly line so should be treated differently.
 761  * The ops.raw part will be parsed further according to type of the instruction.
 762  */
 763 static int symbol__parse_objdump_line(struct symbol *sym, struct map *map,
 764                                       FILE *file, size_t privsize)
 765 {
 766         struct annotation *notes = symbol__annotation(sym);
 767         struct disasm_line *dl;
 768         char *line = NULL, *parsed_line, *tmp, *tmp2, *c;
 769         size_t line_len;
 770         s64 line_ip, offset = -1;
 771
 772         if (getline(&line, &line_len, file) < 0)
 773                 return -1;
 774
 775         if (!line)
 776                 return -1;
 777
 778         while (line_len != 0 && isspace(line[line_len - 1]))
 779                 line[--line_len] = '\0';
 780
 781         c = strchr(line, '\n');
 782         if (c)
 783                 *c = 0;
 784
 785         line_ip = -1;
 786         parsed_line = line;
 787
 788         /*
 789          * Strip leading spaces:
 790          */
 791         tmp = line;
 792         while (*tmp) {
 793                 if (*tmp != ' ')
 794                         break;
 795                 tmp++;
 796         }
 797
 798         if (*tmp) {
 799                 /*
 800                  * Parse hexa addresses followed by ':'
 801                  */
 802                 line_ip = strtoull(tmp, &tmp2, 16);
 803                 if (*tmp2 != ':' || tmp == tmp2 || tmp2[1] == '\0')
 804                         line_ip = -1;
 805         }
 806
 807         if (line_ip != -1) {
 808                 u64 start = map__rip_2objdump(map, sym->start),
 809                     end = map__rip_2objdump(map, sym->end);
 810
 811                 offset = line_ip - start;
 812                 if ((u64)line_ip < start || (u64)line_ip > end)
 813                         offset = -1;
 814                 else
 815                         parsed_line = tmp2 + 1;
 816         }
 817
 818         dl = disasm_line__new(offset, parsed_line, privsize);
 819         free(line);
 820
 821         if (dl == NULL)
 822                 return -1;
 823
 824         if (dl->ops.target.offset == UINT64_MAX)
 825                 dl->ops.target.offset = dl->ops.target.addr -
 826                                         map__rip_2objdump(map, sym->start);
 827
 828         /* kcore has no symbols, so add the call target name */
 829         if (dl->ins && ins__is_call(dl->ins) && !dl->ops.target.name) {
 830                 struct addr_map_symbol target = {
 831                         .map = map,
 832                         .addr = dl->ops.target.addr,
 833                 };
 834
 835                 if (!map_groups__find_ams(&target, NULL) &&
 836                     target.sym->start == target.al_addr)
 837                         dl->ops.target.name = strdup(target.sym->name);
 838         }
 839
 840         disasm__add(&notes->src->source, dl);
 841
 842         return 0;
 843 }
 844
 845 static void delete_last_nop(struct symbol *sym)
 846 {
 847         struct annotation *notes = symbol__annotation(sym);
 848         struct list_head *list = &notes->src->source;
 849         struct disasm_line *dl;
 850
 851         while (!list_empty(list)) {
 852                 dl = list_entry(list->prev, struct disasm_line, node);
 853
 854                 if (dl->ins && dl->ins->ops) {
 855                         if (dl->ins->ops != &nop_ops)
 856                                 return;
 857                 } else {
 858                         if (!strstr(dl->line, " nop ") &&
 859                             !strstr(dl->line, " nopl ") &&
 860                             !strstr(dl->line, " nopw "))
 861                                 return;
 862                 }
 863
 864                 list_del(&dl->node);
 865                 disasm_line__free(dl);
 866         }
 867 }
 868
 869 int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize)
 870 {
 871         struct dso *dso = map->dso;
 872         char *filename = dso__build_id_filename(dso, NULL, 0);
 873         bool free_filename = true;
 874         char command[PATH_MAX * 2];
 875         FILE *file;
 876         int err = 0;
 877         char symfs_filename[PATH_MAX];
 878         struct kcore_extract kce;
 879         bool delete_extract = false;
 880
 881         if (filename) {
 882                 snprintf(symfs_filename, sizeof(symfs_filename), "%s%s",
 883                          symbol_conf.symfs, filename);
 884         }
 885
 886         if (filename == NULL) {
 887                 if (dso->has_build_id) {
 888                         pr_err("Can't annotate %s: not enough memory\n",
 889                                sym->name);
 890                         return -ENOMEM;
 891                 }
 892                 goto fallback;
 893         } else if (readlink(symfs_filename, command, sizeof(command)) < 0 ||
 894                    strstr(command, "[kernel.kallsyms]") ||
 895                    access(symfs_filename, R_OK)) {
 896                 free(filename);
 897 fallback:
 898                 /*
 899                  * If we don't have build-ids or the build-id file isn't in the
 900                  * cache, or is just a kallsyms file, well, lets hope that this
 901                  * DSO is the same as when 'perf record' ran.
 902                  */
 903                 filename = dso->long_name;
 904                 snprintf(symfs_filename, sizeof(symfs_filename), "%s%s",
 905                          symbol_conf.symfs, filename);
 906                 free_filename = false;
 907         }
 908
 909         if (dso->symtab_type == DSO_BINARY_TYPE__KALLSYMS &&
 910             !dso__is_kcore(dso)) {
 911                 char bf[BUILD_ID_SIZE * 2 + 16] = " with build id ";
 912                 char *build_id_msg = NULL;
 913
 914                 if (dso->annotate_warned)
 915                         goto out_free_filename;
 916
 917                 if (dso->has_build_id) {
 918                         build_id__sprintf(dso->build_id,
 919                                           sizeof(dso->build_id), bf + 15);
 920                         build_id_msg = bf;
 921                 }
 922                 err = -ENOENT;
 923                 dso->annotate_warned = 1;
 924                 pr_err("Can't annotate %s:\n\n"
 925                        "No vmlinux file%s\nwas found in the path.\n\n"
 926                        "Please use:\n\n"
 927                        "  perf buildid-cache -vu vmlinux\n\n"
 928                        "or:\n\n"
 929                        "  --vmlinux vmlinux\n",
 930                        sym->name, build_id_msg ?: "");
 931                 goto out_free_filename;
 932         }
 933
 934         pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__,
 935                  filename, sym->name, map->unmap_ip(map, sym->start),
 936                  map->unmap_ip(map, sym->end));
 937
 938         pr_debug("annotating [%p] %30s : [%p] %30s\n",
 939                  dso, dso->long_name, sym, sym->name);
 940
 941         if (dso__is_kcore(dso)) {
 942                 kce.kcore_filename = symfs_filename;
 943                 kce.addr = map__rip_2objdump(map, sym->start);
 944                 kce.offs = sym->start;
 945                 kce.len = sym->end + 1 - sym->start;
 946                 if (!kcore_extract__create(&kce)) {
 947                         delete_extract = true;
 948                         strlcpy(symfs_filename, kce.extract_filename,
 949                                 sizeof(symfs_filename));
 950                         if (free_filename) {
 951                                 free(filename);
 952                                 free_filename = false;
 953                         }
 954                         filename = symfs_filename;
 955                 }
 956         }
 957
 958         snprintf(command, sizeof(command),
 959                  "%s %s%s --start-address=0x%016" PRIx64
 960                  " --stop-address=0x%016" PRIx64
 961                  " -d %s %s -C %s 2>/dev/null|grep -v %s|expand",
 962                  objdump_path ? objdump_path : "objdump",
 963                  disassembler_style ? "-M " : "",
 964                  disassembler_style ? disassembler_style : "",
 965                  map__rip_2objdump(map, sym->start),
 966                  map__rip_2objdump(map, sym->end+1),
 967                  symbol_conf.annotate_asm_raw ? "" : "--no-show-raw",
 968                  symbol_conf.annotate_src ? "-S" : "",
 969                  symfs_filename, filename);
 970
 971         pr_debug("Executing: %s\n", command);
 972
 973         file = popen(command, "r");
 974         if (!file)
 975                 goto out_free_filename;
 976
 977         while (!feof(file))
 978                 if (symbol__parse_objdump_line(sym, map, file, privsize) < 0)
 979                         break;
 980
 981         /*
 982          * kallsyms does not have symbol sizes so there may a nop at the end.
 983          * Remove it.
 984          */
 985         if (dso__is_kcore(dso))
 986                 delete_last_nop(sym);
 987
 988         pclose(file);
 989 out_free_filename:
 990         if (delete_extract)
 991                 kcore_extract__delete(&kce);
 992         if (free_filename)
 993                 free(filename);
 994         return err;
 995 }
 996
 997 static void insert_source_line(struct rb_root *root, struct source_line *src_line)
 998 {
 999         struct source_line *iter;
1000         struct rb_node **p = &root->rb_node;
1001         struct rb_node *parent = NULL;
1002         int i, ret;
1003
1004         while (*p != NULL) {
1005                 parent = *p;
1006                 iter = rb_entry(parent, struct source_line, node);
1007
1008                 ret = strcmp(iter->path, src_line->path);
1009                 if (ret == 0) {
1010                         for (i = 0; i < src_line->nr_pcnt; i++)
1011                                 iter->p[i].percent_sum += src_line->p[i].percent;
1012                         return;
1013                 }
1014
1015                 if (ret < 0)
1016                         p = &(*p)->rb_left;
1017                 else
1018                         p = &(*p)->rb_right;
1019         }
1020
1021         for (i = 0; i < src_line->nr_pcnt; i++)
1022                 src_line->p[i].percent_sum = src_line->p[i].percent;
1023
1024         rb_link_node(&src_line->node, parent, p);
1025         rb_insert_color(&src_line->node, root);
1026 }
1027
1028 static int cmp_source_line(struct source_line *a, struct source_line *b)
1029 {
1030         int i;
1031
1032         for (i = 0; i < a->nr_pcnt; i++) {
1033                 if (a->p[i].percent_sum == b->p[i].percent_sum)
1034                         continue;
1035                 return a->p[i].percent_sum > b->p[i].percent_sum;
1036         }
1037
1038         return 0;
1039 }
1040
1041 static void __resort_source_line(struct rb_root *root, struct source_line *src_line)
1042 {
1043         struct source_line *iter;
1044         struct rb_node **p = &root->rb_node;
1045         struct rb_node *parent = NULL;
1046
1047         while (*p != NULL) {
1048                 parent = *p;
1049                 iter = rb_entry(parent, struct source_line, node);
1050
1051                 if (cmp_source_line(src_line, iter))
1052                         p = &(*p)->rb_left;
1053                 else
1054                         p = &(*p)->rb_right;
1055         }
1056
1057         rb_link_node(&src_line->node, parent, p);
1058         rb_insert_color(&src_line->node, root);
1059 }
1060
1061 static void resort_source_line(struct rb_root *dest_root, struct rb_root *src_root)
1062 {
1063         struct source_line *src_line;
1064         struct rb_node *node;
1065
1066         node = rb_first(src_root);
1067         while (node) {
1068                 struct rb_node *next;
1069
1070                 src_line = rb_entry(node, struct source_line, node);
1071                 next = rb_next(node);
1072                 rb_erase(node, src_root);
1073
1074                 __resort_source_line(dest_root, src_line);
1075                 node = next;
1076         }
1077 }
1078
1079 static void symbol__free_source_line(struct symbol *sym, int len)
1080 {
1081         struct annotation *notes = symbol__annotation(sym);
1082         struct source_line *src_line = notes->src->lines;
1083         size_t sizeof_src_line;
1084         int i;
1085
1086         sizeof_src_line = sizeof(*src_line) +
1087                           (sizeof(src_line->p) * (src_line->nr_pcnt - 1));
1088
1089         for (i = 0; i < len; i++) {
1090                 free_srcline(src_line->path);
1091                 src_line = (void *)src_line + sizeof_src_line;
1092         }
1093
1094         free(notes->src->lines);
1095         notes->src->lines = NULL;
1096 }
1097
1098 /* Get the filename:line for the colored entries */
1099 static int symbol__get_source_line(struct symbol *sym, struct map *map,
1100                                    struct perf_evsel *evsel,
1101                                    struct rb_root *root, int len)
1102 {
1103         u64 start;
1104         int i, k;
1105         int evidx = evsel->idx;
1106         struct source_line *src_line;
1107         struct annotation *notes = symbol__annotation(sym);
1108         struct sym_hist *h = annotation__histogram(notes, evidx);
1109         struct rb_root tmp_root = RB_ROOT;
1110         int nr_pcnt = 1;
1111         u64 h_sum = h->sum;
1112         size_t sizeof_src_line = sizeof(struct source_line);
1113
1114         if (perf_evsel__is_group_event(evsel)) {
1115                 for (i = 1; i < evsel->nr_members; i++) {
1116                         h = annotation__histogram(notes, evidx + i);
1117                         h_sum += h->sum;
1118                 }
1119                 nr_pcnt = evsel->nr_members;
1120                 sizeof_src_line += (nr_pcnt - 1) * sizeof(src_line->p);
1121         }
1122
1123         if (!h_sum)
1124                 return 0;
1125
1126         src_line = notes->src->lines = calloc(len, sizeof_src_line);
1127         if (!notes->src->lines)
1128                 return -1;
1129
1130         start = map__rip_2objdump(map, sym->start);
1131
1132         for (i = 0; i < len; i++) {
1133                 u64 offset;
1134                 double percent_max = 0.0;
1135
1136                 src_line->nr_pcnt = nr_pcnt;
1137
1138                 for (k = 0; k < nr_pcnt; k++) {
1139                         h = annotation__histogram(notes, evidx + k);
1140                         src_line->p[k].percent = 100.0 * h->addr[i] / h->sum;
1141
1142                         if (src_line->p[k].percent > percent_max)
1143                                 percent_max = src_line->p[k].percent;
1144                 }
1145
1146                 if (percent_max <= 0.5)
1147                         goto next;
1148
1149                 offset = start + i;
1150                 src_line->path = get_srcline(map->dso, offset);
1151                 insert_source_line(&tmp_root, src_line);
1152
1153         next:
1154                 src_line = (void *)src_line + sizeof_src_line;
1155         }
1156
1157         resort_source_line(root, &tmp_root);
1158         return 0;
1159 }
1160
1161 static void print_summary(struct rb_root *root, const char *filename)
1162 {
1163         struct source_line *src_line;
1164         struct rb_node *node;
1165
1166         printf("\nSorted summary for file %s\n", filename);
1167         printf("----------------------------------------------\n\n");
1168
1169         if (RB_EMPTY_ROOT(root)) {
1170                 printf(" Nothing higher than %1.1f%%\n", MIN_GREEN);
1171                 return;
1172         }
1173
1174         node = rb_first(root);
1175         while (node) {
1176                 double percent, percent_max = 0.0;
1177                 const char *color;
1178                 char *path;
1179                 int i;
1180
1181                 src_line = rb_entry(node, struct source_line, node);
1182                 for (i = 0; i < src_line->nr_pcnt; i++) {
1183                         percent = src_line->p[i].percent_sum;
1184                         color = get_percent_color(percent);
1185                         color_fprintf(stdout, color, " %7.2f", percent);
1186
1187                         if (percent > percent_max)
1188                                 percent_max = percent;
1189                 }
1190
1191                 path = src_line->path;
1192                 color = get_percent_color(percent_max);
1193                 color_fprintf(stdout, color, " %s\n", path);
1194
1195                 node = rb_next(node);
1196         }
1197 }
1198
1199 static void symbol__annotate_hits(struct symbol *sym, struct perf_evsel *evsel)
1200 {
1201         struct annotation *notes = symbol__annotation(sym);
1202         struct sym_hist *h = annotation__histogram(notes, evsel->idx);
1203         u64 len = symbol__size(sym), offset;
1204
1205         for (offset = 0; offset < len; ++offset)
1206                 if (h->addr[offset] != 0)
1207                         printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2,
1208                                sym->start + offset, h->addr[offset]);
1209         printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum);
1210 }
1211
1212 int symbol__annotate_printf(struct symbol *sym, struct map *map,
1213                             struct perf_evsel *evsel, bool full_paths,
1214                             int min_pcnt, int max_lines, int context)
1215 {
1216         struct dso *dso = map->dso;
1217         char *filename;
1218         const char *d_filename;
1219         struct annotation *notes = symbol__annotation(sym);
1220         struct disasm_line *pos, *queue = NULL;
1221         u64 start = map__rip_2objdump(map, sym->start);
1222         int printed = 2, queue_len = 0;
1223         int more = 0;
1224         u64 len;
1225         int width = 8;
1226         int namelen;
1227
1228         filename = strdup(dso->long_name);
1229         if (!filename)
1230                 return -ENOMEM;
1231
1232         if (full_paths)
1233                 d_filename = filename;
1234         else
1235                 d_filename = basename(filename);
1236
1237         len = symbol__size(sym);
1238         namelen = strlen(d_filename);
1239
1240         if (perf_evsel__is_group_event(evsel))
1241                 width *= evsel->nr_members;
1242
1243         printf(" %-*.*s|        Source code & Disassembly of %s\n",
1244                width, width, "Percent", d_filename);
1245         printf("-%-*.*s-------------------------------------\n",
1246                width+namelen, width+namelen, graph_dotted_line);
1247
1248         if (verbose)
1249                 symbol__annotate_hits(sym, evsel);
1250
1251         list_for_each_entry(pos, &notes->src->source, node) {
1252                 if (context && queue == NULL) {
1253                         queue = pos;
1254                         queue_len = 0;
1255                 }
1256
1257                 switch (disasm_line__print(pos, sym, start, evsel, len,
1258                                             min_pcnt, printed, max_lines,
1259                                             queue)) {
1260                 case 0:
1261                         ++printed;
1262                         if (context) {
1263                                 printed += queue_len;
1264                                 queue = NULL;
1265                                 queue_len = 0;
1266                         }
1267                         break;
1268                 case 1:
1269                         /* filtered by max_lines */
1270                         ++more;
1271                         break;
1272                 case -1:
1273                 default:
1274                         /*
1275                          * Filtered by min_pcnt or non IP lines when
1276                          * context != 0
1277                          */
1278                         if (!context)
1279                                 break;
1280                         if (queue_len == context)
1281                                 queue = list_entry(queue->node.next, typeof(*queue), node);
1282                         else
1283                                 ++queue_len;
1284                         break;
1285                 }
1286         }
1287
1288         free(filename);
1289
1290         return more;
1291 }
1292
1293 void symbol__annotate_zero_histogram(struct symbol *sym, int evidx)
1294 {
1295         struct annotation *notes = symbol__annotation(sym);
1296         struct sym_hist *h = annotation__histogram(notes, evidx);
1297
1298         memset(h, 0, notes->src->sizeof_sym_hist);
1299 }
1300
1301 void symbol__annotate_decay_histogram(struct symbol *sym, int evidx)
1302 {
1303         struct annotation *notes = symbol__annotation(sym);
1304         struct sym_hist *h = annotation__histogram(notes, evidx);
1305         int len = symbol__size(sym), offset;
1306
1307         h->sum = 0;
1308         for (offset = 0; offset < len; ++offset) {
1309                 h->addr[offset] = h->addr[offset] * 7 / 8;
1310                 h->sum += h->addr[offset];
1311         }
1312 }
1313
1314 void disasm__purge(struct list_head *head)
1315 {
1316         struct disasm_line *pos, *n;
1317
1318         list_for_each_entry_safe(pos, n, head, node) {
1319                 list_del(&pos->node);
1320                 disasm_line__free(pos);
1321         }
1322 }
1323
1324 static size_t disasm_line__fprintf(struct disasm_line *dl, FILE *fp)
1325 {
1326         size_t printed;
1327
1328         if (dl->offset == -1)
1329                 return fprintf(fp, "%s\n", dl->line);
1330
1331         printed = fprintf(fp, "%#" PRIx64 " %s", dl->offset, dl->name);
1332
1333         if (dl->ops.raw[0] != '\0') {
1334                 printed += fprintf(fp, "%.*s %s\n", 6 - (int)printed, " ",
1335                                    dl->ops.raw);
1336         }
1337
1338         return printed + fprintf(fp, "\n");
1339 }
1340
1341 size_t disasm__fprintf(struct list_head *head, FILE *fp)
1342 {
1343         struct disasm_line *pos;
1344         size_t printed = 0;
1345
1346         list_for_each_entry(pos, head, node)
1347                 printed += disasm_line__fprintf(pos, fp);
1348
1349         return printed;
1350 }
1351
1352 int symbol__tty_annotate(struct symbol *sym, struct map *map,
1353                          struct perf_evsel *evsel, bool print_lines,
1354                          bool full_paths, int min_pcnt, int max_lines)
1355 {
1356         struct dso *dso = map->dso;
1357         struct rb_root source_line = RB_ROOT;
1358         u64 len;
1359
1360         if (symbol__annotate(sym, map, 0) < 0)
1361                 return -1;
1362
1363         len = symbol__size(sym);
1364
1365         if (print_lines) {
1366                 symbol__get_source_line(sym, map, evsel, &source_line, len);
1367                 print_summary(&source_line, dso->long_name);
1368         }
1369
1370         symbol__annotate_printf(sym, map, evsel, full_paths,
1371                                 min_pcnt, max_lines, 0);
1372         if (print_lines)
1373                 symbol__free_source_line(sym, len);
1374
1375         disasm__purge(&symbol__annotation(sym)->src->source);
1376
1377         return 0;
1378 }