tools/perf/util/symbol.c

   1 #include "util.h"
   2 #include "../perf.h"
   3 #include "string.h"
   4 #include "symbol.h"
   5
   6 #include <libelf.h>
   7 #include <gelf.h>
   8 #include <elf.h>
   9
  10 const char *sym_hist_filter;
  11
  12 static struct symbol *symbol__new(u64 start, u64 len,
  13                                   const char *name, unsigned int priv_size,
  14                                   u64 obj_start, int verbose)
  15 {
  16         size_t namelen = strlen(name) + 1;
  17         struct symbol *self = calloc(1, priv_size + sizeof(*self) + namelen);
  18
  19         if (!self)
  20                 return NULL;
  21
  22         if (verbose >= 2)
  23                 printf("new symbol: %016Lx [%08lx]: %s, hist: %p, obj_start: %p\n",
  24                         (u64)start, (unsigned long)len, name, self->hist, (void *)(unsigned long)obj_start);
  25
  26         self->obj_start= obj_start;
  27         self->hist = NULL;
  28         self->hist_sum = 0;
  29
  30         if (sym_hist_filter && !strcmp(name, sym_hist_filter))
  31                 self->hist = calloc(sizeof(u64), len);
  32
  33         if (priv_size) {
  34                 memset(self, 0, priv_size);
  35                 self = ((void *)self) + priv_size;
  36         }
  37         self->start = start;
  38         self->end   = start + len - 1;
  39         memcpy(self->name, name, namelen);
  40
  41         return self;
  42 }
  43
  44 static void symbol__delete(struct symbol *self, unsigned int priv_size)
  45 {
  46         free(((void *)self) - priv_size);
  47 }
  48
  49 static size_t symbol__fprintf(struct symbol *self, FILE *fp)
  50 {
  51         return fprintf(fp, " %llx-%llx %s\n",
  52                        self->start, self->end, self->name);
  53 }
  54
  55 struct dso *dso__new(const char *name, unsigned int sym_priv_size)
  56 {
  57         struct dso *self = malloc(sizeof(*self) + strlen(name) + 1);
  58
  59         if (self != NULL) {
  60                 strcpy(self->name, name);
  61                 self->syms = RB_ROOT;
  62                 self->sym_priv_size = sym_priv_size;
  63                 self->find_symbol = dso__find_symbol;
  64         }
  65
  66         return self;
  67 }
  68
  69 static void dso__delete_symbols(struct dso *self)
  70 {
  71         struct symbol *pos;
  72         struct rb_node *next = rb_first(&self->syms);
  73
  74         while (next) {
  75                 pos = rb_entry(next, struct symbol, rb_node);
  76                 next = rb_next(&pos->rb_node);
  77                 rb_erase(&pos->rb_node, &self->syms);
  78                 symbol__delete(pos, self->sym_priv_size);
  79         }
  80 }
  81
  82 void dso__delete(struct dso *self)
  83 {
  84         dso__delete_symbols(self);
  85         free(self);
  86 }
  87
  88 static void dso__insert_symbol(struct dso *self, struct symbol *sym)
  89 {
  90         struct rb_node **p = &self->syms.rb_node;
  91         struct rb_node *parent = NULL;
  92         const u64 ip = sym->start;
  93         struct symbol *s;
  94
  95         while (*p != NULL) {
  96                 parent = *p;
  97                 s = rb_entry(parent, struct symbol, rb_node);
  98                 if (ip < s->start)
  99                         p = &(*p)->rb_left;
 100                 else
 101                         p = &(*p)->rb_right;
 102         }
 103         rb_link_node(&sym->rb_node, parent, p);
 104         rb_insert_color(&sym->rb_node, &self->syms);
 105 }
 106
 107 struct symbol *dso__find_symbol(struct dso *self, u64 ip)
 108 {
 109         struct rb_node *n;
 110
 111         if (self == NULL)
 112                 return NULL;
 113
 114         n = self->syms.rb_node;
 115
 116         while (n) {
 117                 struct symbol *s = rb_entry(n, struct symbol, rb_node);
 118
 119                 if (ip < s->start)
 120                         n = n->rb_left;
 121                 else if (ip > s->end)
 122                         n = n->rb_right;
 123                 else
 124                         return s;
 125         }
 126
 127         return NULL;
 128 }
 129
 130 size_t dso__fprintf(struct dso *self, FILE *fp)
 131 {
 132         size_t ret = fprintf(fp, "dso: %s\n", self->name);
 133
 134         struct rb_node *nd;
 135         for (nd = rb_first(&self->syms); nd; nd = rb_next(nd)) {
 136                 struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
 137                 ret += symbol__fprintf(pos, fp);
 138         }
 139
 140         return ret;
 141 }
 142
 143 static int dso__load_kallsyms(struct dso *self, symbol_filter_t filter, int verbose)
 144 {
 145         struct rb_node *nd, *prevnd;
 146         char *line = NULL;
 147         size_t n;
 148         FILE *file = fopen("/proc/kallsyms", "r");
 149
 150         if (file == NULL)
 151                 goto out_failure;
 152
 153         while (!feof(file)) {
 154                 u64 start;
 155                 struct symbol *sym;
 156                 int line_len, len;
 157                 char symbol_type;
 158
 159                 line_len = getline(&line, &n, file);
 160                 if (line_len < 0)
 161                         break;
 162
 163                 if (!line)
 164                         goto out_failure;
 165
 166                 line[--line_len] = '\0'; /* \n */
 167
 168                 len = hex2u64(line, &start);
 169
 170                 len++;
 171                 if (len + 2 >= line_len)
 172                         continue;
 173
 174                 symbol_type = toupper(line[len]);
 175                 /*
 176                  * We're interested only in code ('T'ext)
 177                  */
 178                 if (symbol_type != 'T' && symbol_type != 'W')
 179                         continue;
 180                 /*
 181                  * Well fix up the end later, when we have all sorted.
 182                  */
 183                 sym = symbol__new(start, 0xdead, line + len + 2,
 184                                   self->sym_priv_size, 0, verbose);
 185
 186                 if (sym == NULL)
 187                         goto out_delete_line;
 188
 189                 if (filter && filter(self, sym))
 190                         symbol__delete(sym, self->sym_priv_size);
 191                 else
 192                         dso__insert_symbol(self, sym);
 193         }
 194
 195         /*
 196          * Now that we have all sorted out, just set the ->end of all
 197          * symbols
 198          */
 199         prevnd = rb_first(&self->syms);
 200
 201         if (prevnd == NULL)
 202                 goto out_delete_line;
 203
 204         for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
 205                 struct symbol *prev = rb_entry(prevnd, struct symbol, rb_node),
 206                               *curr = rb_entry(nd, struct symbol, rb_node);
 207
 208                 prev->end = curr->start - 1;
 209                 prevnd = nd;
 210         }
 211
 212         free(line);
 213         fclose(file);
 214
 215         return 0;
 216
 217 out_delete_line:
 218         free(line);
 219 out_failure:
 220         return -1;
 221 }
 222
 223 static int dso__load_perf_map(struct dso *self, symbol_filter_t filter, int verbose)
 224 {
 225         char *line = NULL;
 226         size_t n;
 227         FILE *file;
 228         int nr_syms = 0;
 229
 230         file = fopen(self->name, "r");
 231         if (file == NULL)
 232                 goto out_failure;
 233
 234         while (!feof(file)) {
 235                 u64 start, size;
 236                 struct symbol *sym;
 237                 int line_len, len;
 238
 239                 line_len = getline(&line, &n, file);
 240                 if (line_len < 0)
 241                         break;
 242
 243                 if (!line)
 244                         goto out_failure;
 245
 246                 line[--line_len] = '\0'; /* \n */
 247
 248                 len = hex2u64(line, &start);
 249
 250                 len++;
 251                 if (len + 2 >= line_len)
 252                         continue;
 253
 254                 len += hex2u64(line + len, &size);
 255
 256                 len++;
 257                 if (len + 2 >= line_len)
 258                         continue;
 259
 260                 sym = symbol__new(start, size, line + len,
 261                                   self->sym_priv_size, start, verbose);
 262
 263                 if (sym == NULL)
 264                         goto out_delete_line;
 265
 266                 if (filter && filter(self, sym))
 267                         symbol__delete(sym, self->sym_priv_size);
 268                 else {
 269                         dso__insert_symbol(self, sym);
 270                         nr_syms++;
 271                 }
 272         }
 273
 274         free(line);
 275         fclose(file);
 276
 277         return nr_syms;
 278
 279 out_delete_line:
 280         free(line);
 281 out_failure:
 282         return -1;
 283 }
 284
 285 /**
 286  * elf_symtab__for_each_symbol - iterate thru all the symbols
 287  *
 288  * @self: struct elf_symtab instance to iterate
 289  * @index: uint32_t index
 290  * @sym: GElf_Sym iterator
 291  */
 292 #define elf_symtab__for_each_symbol(syms, nr_syms, index, sym) \
 293         for (index = 0, gelf_getsym(syms, index, &sym);\
 294              index < nr_syms; \
 295              index++, gelf_getsym(syms, index, &sym))
 296
 297 static inline uint8_t elf_sym__type(const GElf_Sym *sym)
 298 {
 299         return GELF_ST_TYPE(sym->st_info);
 300 }
 301
 302 static inline int elf_sym__is_function(const GElf_Sym *sym)
 303 {
 304         return elf_sym__type(sym) == STT_FUNC &&
 305                sym->st_name != 0 &&
 306                sym->st_shndx != SHN_UNDEF &&
 307                sym->st_size != 0;
 308 }
 309
 310 static inline const char *elf_sym__name(const GElf_Sym *sym,
 311                                         const Elf_Data *symstrs)
 312 {
 313         return symstrs->d_buf + sym->st_name;
 314 }
 315
 316 static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep,
 317                                     GElf_Shdr *shp, const char *name,
 318                                     size_t *index)
 319 {
 320         Elf_Scn *sec = NULL;
 321         size_t cnt = 1;
 322
 323         while ((sec = elf_nextscn(elf, sec)) != NULL) {
 324                 char *str;
 325
 326                 gelf_getshdr(sec, shp);
 327                 str = elf_strptr(elf, ep->e_shstrndx, shp->sh_name);
 328                 if (!strcmp(name, str)) {
 329                         if (index)
 330                                 *index = cnt;
 331                         break;
 332                 }
 333                 ++cnt;
 334         }
 335
 336         return sec;
 337 }
 338
 339 #define elf_section__for_each_rel(reldata, pos, pos_mem, idx, nr_entries) \
 340         for (idx = 0, pos = gelf_getrel(reldata, 0, &pos_mem); \
 341              idx < nr_entries; \
 342              ++idx, pos = gelf_getrel(reldata, idx, &pos_mem))
 343
 344 #define elf_section__for_each_rela(reldata, pos, pos_mem, idx, nr_entries) \
 345         for (idx = 0, pos = gelf_getrela(reldata, 0, &pos_mem); \
 346              idx < nr_entries; \
 347              ++idx, pos = gelf_getrela(reldata, idx, &pos_mem))
 348
 349 static int dso__synthesize_plt_symbols(struct  dso *self, Elf *elf,
 350                                        GElf_Ehdr *ehdr, Elf_Scn *scn_dynsym,
 351                                        GElf_Shdr *shdr_dynsym,
 352                                        size_t dynsym_idx, int verbose)
 353 {
 354         uint32_t nr_rel_entries, idx;
 355         GElf_Sym sym;
 356         u64 plt_offset;
 357         GElf_Shdr shdr_plt;
 358         struct symbol *f;
 359         GElf_Shdr shdr_rel_plt;
 360         Elf_Data *reldata, *syms, *symstrs;
 361         Elf_Scn *scn_plt_rel, *scn_symstrs;
 362         char sympltname[1024];
 363         int nr = 0, symidx;
 364
 365         scn_plt_rel = elf_section_by_name(elf, ehdr, &shdr_rel_plt,
 366                                           ".rela.plt", NULL);
 367         if (scn_plt_rel == NULL) {
 368                 scn_plt_rel = elf_section_by_name(elf, ehdr, &shdr_rel_plt,
 369                                                   ".rel.plt", NULL);
 370                 if (scn_plt_rel == NULL)
 371                         return 0;
 372         }
 373
 374         if (shdr_rel_plt.sh_link != dynsym_idx)
 375                 return 0;
 376
 377         if (elf_section_by_name(elf, ehdr, &shdr_plt, ".plt", NULL) == NULL)
 378                 return 0;
 379
 380         /*
 381          * Fetch the relocation section to find the indexes to the GOT
 382          * and the symbols in the .dynsym they refer to.
 383          */
 384         reldata = elf_getdata(scn_plt_rel, NULL);
 385         if (reldata == NULL)
 386                 return -1;
 387
 388         syms = elf_getdata(scn_dynsym, NULL);
 389         if (syms == NULL)
 390                 return -1;
 391
 392         scn_symstrs = elf_getscn(elf, shdr_dynsym->sh_link);
 393         if (scn_symstrs == NULL)
 394                 return -1;
 395
 396         symstrs = elf_getdata(scn_symstrs, NULL);
 397         if (symstrs == NULL)
 398                 return -1;
 399
 400         nr_rel_entries = shdr_rel_plt.sh_size / shdr_rel_plt.sh_entsize;
 401         plt_offset = shdr_plt.sh_offset;
 402
 403         if (shdr_rel_plt.sh_type == SHT_RELA) {
 404                 GElf_Rela pos_mem, *pos;
 405
 406                 elf_section__for_each_rela(reldata, pos, pos_mem, idx,
 407                                            nr_rel_entries) {
 408                         symidx = GELF_R_SYM(pos->r_info);
 409                         plt_offset += shdr_plt.sh_entsize;
 410                         gelf_getsym(syms, symidx, &sym);
 411                         snprintf(sympltname, sizeof(sympltname),
 412                                  "%s@plt", elf_sym__name(&sym, symstrs));
 413
 414                         f = symbol__new(plt_offset, shdr_plt.sh_entsize,
 415                                         sympltname, self->sym_priv_size, 0, verbose);
 416                         if (!f)
 417                                 return -1;
 418
 419                         dso__insert_symbol(self, f);
 420                         ++nr;
 421                 }
 422         } else if (shdr_rel_plt.sh_type == SHT_REL) {
 423                 GElf_Rel pos_mem, *pos;
 424                 elf_section__for_each_rel(reldata, pos, pos_mem, idx,
 425                                           nr_rel_entries) {
 426                         symidx = GELF_R_SYM(pos->r_info);
 427                         plt_offset += shdr_plt.sh_entsize;
 428                         gelf_getsym(syms, symidx, &sym);
 429                         snprintf(sympltname, sizeof(sympltname),
 430                                  "%s@plt", elf_sym__name(&sym, symstrs));
 431
 432                         f = symbol__new(plt_offset, shdr_plt.sh_entsize,
 433                                         sympltname, self->sym_priv_size, 0, verbose);
 434                         if (!f)
 435                                 return -1;
 436
 437                         dso__insert_symbol(self, f);
 438                         ++nr;
 439                 }
 440         } else {
 441                 /*
 442                  * TODO: There are still one more shdr_rel_plt.sh_type
 443                  * I have to investigate, but probably should be ignored.
 444                  */
 445         }
 446
 447         return nr;
 448 }
 449
 450 static int dso__load_sym(struct dso *self, int fd, const char *name,
 451                          symbol_filter_t filter, int verbose)
 452 {
 453         Elf_Data *symstrs;
 454         uint32_t nr_syms;
 455         int err = -1;
 456         uint32_t index;
 457         GElf_Ehdr ehdr;
 458         GElf_Shdr shdr;
 459         Elf_Data *syms;
 460         GElf_Sym sym;
 461         Elf_Scn *sec, *sec_dynsym;
 462         Elf *elf;
 463         size_t dynsym_idx;
 464         int nr = 0;
 465
 466         elf = elf_begin(fd, ELF_C_READ_MMAP, NULL);
 467         if (elf == NULL) {
 468                 if (verbose)
 469                         fprintf(stderr, "%s: cannot read %s ELF file.\n",
 470                                 __func__, name);
 471                 goto out_close;
 472         }
 473
 474         if (gelf_getehdr(elf, &ehdr) == NULL) {
 475                 if (verbose)
 476                         fprintf(stderr, "%s: cannot get elf header.\n", __func__);
 477                 goto out_elf_end;
 478         }
 479
 480         /*
 481          * We need to check if we have a .dynsym, so that we can handle the
 482          * .plt, synthesizing its symbols, that aren't on the symtabs (be it
 483          * .dynsym or .symtab)
 484          */
 485         sec_dynsym = elf_section_by_name(elf, &ehdr, &shdr,
 486                                          ".dynsym", &dynsym_idx);
 487         if (sec_dynsym != NULL) {
 488                 nr = dso__synthesize_plt_symbols(self, elf, &ehdr,
 489                                                  sec_dynsym, &shdr,
 490                                                  dynsym_idx, verbose);
 491                 if (nr < 0)
 492                         goto out_elf_end;
 493         }
 494
 495         /*
 496          * But if we have a full .symtab (that is a superset of .dynsym) we
 497          * should add the symbols not in the .dynsyn
 498          */
 499         sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL);
 500         if (sec == NULL) {
 501                 if (sec_dynsym == NULL)
 502                         goto out_elf_end;
 503
 504                 sec = sec_dynsym;
 505                 gelf_getshdr(sec, &shdr);
 506         }
 507
 508         syms = elf_getdata(sec, NULL);
 509         if (syms == NULL)
 510                 goto out_elf_end;
 511
 512         sec = elf_getscn(elf, shdr.sh_link);
 513         if (sec == NULL)
 514                 goto out_elf_end;
 515
 516         symstrs = elf_getdata(sec, NULL);
 517         if (symstrs == NULL)
 518                 goto out_elf_end;
 519
 520         nr_syms = shdr.sh_size / shdr.sh_entsize;
 521
 522         memset(&sym, 0, sizeof(sym));
 523
 524         elf_symtab__for_each_symbol(syms, nr_syms, index, sym) {
 525                 struct symbol *f;
 526                 u64 obj_start;
 527
 528                 if (!elf_sym__is_function(&sym))
 529                         continue;
 530
 531                 sec = elf_getscn(elf, sym.st_shndx);
 532                 if (!sec)
 533                         goto out_elf_end;
 534
 535                 gelf_getshdr(sec, &shdr);
 536                 obj_start = sym.st_value;
 537
 538                 sym.st_value -= shdr.sh_addr - shdr.sh_offset;
 539
 540                 f = symbol__new(sym.st_value, sym.st_size,
 541                                 elf_sym__name(&sym, symstrs),
 542                                 self->sym_priv_size, obj_start, verbose);
 543                 if (!f)
 544                         goto out_elf_end;
 545
 546                 if (filter && filter(self, f))
 547                         symbol__delete(f, self->sym_priv_size);
 548                 else {
 549                         dso__insert_symbol(self, f);
 550                         nr++;
 551                 }
 552         }
 553
 554         err = nr;
 555 out_elf_end:
 556         elf_end(elf);
 557 out_close:
 558         return err;
 559 }
 560
 561 int dso__load(struct dso *self, symbol_filter_t filter, int verbose)
 562 {
 563         int size = strlen(self->name) + sizeof("/usr/lib/debug%s.debug");
 564         char *name = malloc(size);
 565         int variant = 0;
 566         int ret = -1;
 567         int fd;
 568
 569         if (!name)
 570                 return -1;
 571
 572         if (strncmp(self->name, "/tmp/perf-", 10) == 0)
 573                 return dso__load_perf_map(self, filter, verbose);
 574
 575 more:
 576         do {
 577                 switch (variant) {
 578                 case 0: /* Fedora */
 579                         snprintf(name, size, "/usr/lib/debug%s.debug", self->name);
 580                         break;
 581                 case 1: /* Ubuntu */
 582                         snprintf(name, size, "/usr/lib/debug%s", self->name);
 583                         break;
 584                 case 2: /* Sane people */
 585                         snprintf(name, size, "%s", self->name);
 586                         break;
 587
 588                 default:
 589                         goto out;
 590                 }
 591                 variant++;
 592
 593                 fd = open(name, O_RDONLY);
 594         } while (fd < 0);
 595
 596         ret = dso__load_sym(self, fd, name, filter, verbose);
 597         close(fd);
 598
 599         /*
 600          * Some people seem to have debuginfo files _WITHOUT_ debug info!?!?
 601          */
 602         if (!ret)
 603                 goto more;
 604
 605 out:
 606         free(name);
 607         return ret;
 608 }
 609
 610 static int dso__load_vmlinux(struct dso *self, const char *vmlinux,
 611                              symbol_filter_t filter, int verbose)
 612 {
 613         int err, fd = open(vmlinux, O_RDONLY);
 614
 615         if (fd < 0)
 616                 return -1;
 617
 618         err = dso__load_sym(self, fd, vmlinux, filter, verbose);
 619         close(fd);
 620
 621         return err;
 622 }
 623
 624 int dso__load_kernel(struct dso *self, const char *vmlinux,
 625                      symbol_filter_t filter, int verbose)
 626 {
 627         int err = -1;
 628
 629         if (vmlinux)
 630                 err = dso__load_vmlinux(self, vmlinux, filter, verbose);
 631
 632         if (err < 0)
 633                 err = dso__load_kallsyms(self, filter, verbose);
 634
 635         return err;
 636 }
 637
 638 void symbol__init(void)
 639 {
 640         elf_version(EV_CURRENT);
 641 }