lib/bin-info.c

   1 /*
   2  * bin-info.c
   3  *
   4  * Babeltrace - Executable and Shared Object Debug Info Reader
   5  *
   6  * Copyright 2015 Antoine Busque <abusque@efficios.com>
   7  *
   8  * Author: Antoine Busque <abusque@efficios.com>
   9  *
  10  * Permission is hereby granted, free of charge, to any person obtaining a copy
  11  * of this software and associated documentation files (the "Software"), to deal
  12  * in the Software without restriction, including without limitation the rights
  13  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  14  * copies of the Software, and to permit persons to whom the Software is
  15  * furnished to do so, subject to the following conditions:
  16  *
  17  * The above copyright notice and this permission notice shall be included in
  18  * all copies or substantial portions of the Software.
  19  *
  20  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  21  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  23  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  24  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  25  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  26  * SOFTWARE.
  27  */
  28
  29 #include <fcntl.h>
  30 #include <math.h>
  31 #include <libgen.h>
  32 #include <stdio.h>
  33 #include <inttypes.h>
  34 #include <stdlib.h>
  35 #include <string.h>
  36 #include <unistd.h>
  37 #include <dwarf.h>
  38 #include <glib.h>
  39 #include <babeltrace/dwarf.h>
  40 #include <babeltrace/bin-info.h>
  41 #include <babeltrace/crc32.h>
  42 #include <babeltrace/babeltrace-internal.h>
  43 #include <babeltrace/utils.h>
  44 #include <errno.h>
  45
  46 /*
  47  * An address printed in hex is at most 20 bytes (16 for 64-bits +
  48  * leading 0x + optional leading '+' if addr is an offset + null
  49  * character).
  50  */
  51 #define ADDR_STR_LEN 20
  52
  53 BT_HIDDEN
  54 int bin_info_init(void)
  55 {
  56         int ret = 0;
  57
  58         if (elf_version(EV_CURRENT) == EV_NONE) {
  59                 printf_debug("ELF library initialization failed: %s\n",
  60                                 elf_errmsg(-1));
  61                 ret = -1;
  62         }
  63
  64         return ret;
  65 }
  66
  67 BT_HIDDEN
  68 struct bin_info *bin_info_create(const char *path, uint64_t low_addr,
  69                 uint64_t memsz, bool is_pic)
  70 {
  71         struct bin_info *bin = NULL;
  72
  73         if (!path) {
  74                 goto error;
  75         }
  76
  77         bin = g_new0(struct bin_info, 1);
  78         if (!bin) {
  79                 goto error;
  80         }
  81
  82         if (opt_debug_info_target_prefix) {
  83                 bin->elf_path = g_build_path("/", opt_debug_info_target_prefix,
  84                                                 path, NULL);
  85         } else {
  86                 bin->elf_path = strdup(path);
  87         }
  88
  89         if (!bin->elf_path) {
  90                 goto error;
  91         }
  92
  93         bin->is_pic = is_pic;
  94         bin->memsz = memsz;
  95         bin->low_addr = low_addr;
  96         bin->high_addr = bin->low_addr + bin->memsz;
  97
  98         return bin;
  99
 100 error:
 101         bin_info_destroy(bin);
 102         return NULL;
 103 }
 104
 105 BT_HIDDEN
 106 void bin_info_destroy(struct bin_info *bin)
 107 {
 108         if (!bin) {
 109                 return;
 110         }
 111
 112         dwarf_end(bin->dwarf_info);
 113
 114         free(bin->elf_path);
 115         free(bin->dwarf_path);
 116         free(bin->build_id);
 117         free(bin->dbg_link_filename);
 118
 119         elf_end(bin->elf_file);
 120
 121         close(bin->elf_fd);
 122         close(bin->dwarf_fd);
 123
 124         g_free(bin);
 125 }
 126
 127
 128 BT_HIDDEN
 129 int bin_info_set_build_id(struct bin_info *bin, uint8_t *build_id,
 130                 size_t build_id_len)
 131 {
 132         if (!bin || !build_id) {
 133                 goto error;
 134         }
 135
 136         bin->build_id = malloc(build_id_len);
 137         if (!bin->build_id) {
 138                 goto error;
 139         }
 140
 141         memcpy(bin->build_id, build_id, build_id_len);
 142         bin->build_id_len = build_id_len;
 143
 144         /*
 145          * Reset the is_elf_only flag in case it had been set
 146          * previously, because we might find separate debug info using
 147          * the new build id information.
 148          */
 149         bin->is_elf_only = false;
 150
 151         return 0;
 152
 153 error:
 154
 155         return -1;
 156 }
 157
 158 BT_HIDDEN
 159 int bin_info_set_debug_link(struct bin_info *bin, char *filename, uint32_t crc)
 160 {
 161         if (!bin || !filename) {
 162                 goto error;
 163         }
 164
 165         bin->dbg_link_filename = strdup(filename);
 166         if (!bin->dbg_link_filename) {
 167                 goto error;
 168         }
 169
 170         bin->dbg_link_crc = crc;
 171
 172         /*
 173          * Reset the is_elf_only flag in case it had been set
 174          * previously, because we might find separate debug info using
 175          * the new build id information.
 176          */
 177         bin->is_elf_only = false;
 178
 179         return 0;
 180
 181 error:
 182
 183         return -1;
 184 }
 185
 186 /**
 187  * Tries to read DWARF info from the location given by path, and
 188  * attach it to the given bin_info instance if it exists.
 189  *
 190  * @param bin   bin_info instance for which to set DWARF info
 191  * @param path  Presumed location of the DWARF info
 192  * @returns     0 on success, negative value on failure
 193  */
 194 static
 195 int bin_info_set_dwarf_info_from_path(struct bin_info *bin, char *path)
 196 {
 197         int fd = -1, ret = 0;
 198         struct bt_dwarf_cu *cu = NULL;
 199         Dwarf *dwarf_info = NULL;
 200
 201         if (!bin || !path) {
 202                 goto error;
 203         }
 204
 205         fd = open(path, O_RDONLY);
 206         if (fd < 0) {
 207                 fd = -errno;
 208                 goto error;
 209         }
 210
 211         dwarf_info = dwarf_begin(fd, DWARF_C_READ);
 212         if (!dwarf_info) {
 213                 goto error;
 214         }
 215
 216         /*
 217          * Check if the dwarf info has any CU. If not, the
 218          * executable's object file contains no DWARF info.
 219          */
 220         cu = bt_dwarf_cu_create(dwarf_info);
 221         if (!cu) {
 222                 goto error;
 223         }
 224
 225         ret = bt_dwarf_cu_next(cu);
 226         if (ret) {
 227                 goto error;
 228         }
 229
 230         bin->dwarf_fd = fd;
 231         bin->dwarf_path = strdup(path);
 232         if (!bin->dwarf_path) {
 233                 goto error;
 234         }
 235         bin->dwarf_info = dwarf_info;
 236         free(cu);
 237
 238         return 0;
 239
 240 error:
 241         if (fd >= 0) {
 242                 close(fd);
 243                 fd = -1;
 244         }
 245         dwarf_end(dwarf_info);
 246         g_free(dwarf_info);
 247         free(cu);
 248
 249         return fd;
 250 }
 251
 252 /**
 253  * Try to set the dwarf_info for a given bin_info instance via the
 254  * build ID method.
 255  *
 256  * @param bin           bin_info instance for which to retrieve the
 257  *                      DWARF info via build ID
 258  * @returns             0 on success (i.e. dwarf_info set), -1 on failure
 259  */
 260 static
 261 int bin_info_set_dwarf_info_build_id(struct bin_info *bin)
 262 {
 263         int i = 0, ret = 0;
 264         char *path = NULL, *build_id_file = NULL;
 265         const char *dbg_dir = NULL;
 266         size_t build_id_file_len;
 267
 268         if (!bin || !bin->build_id) {
 269                 goto error;
 270         }
 271
 272         dbg_dir = opt_debug_info_dir ? : DEFAULT_DEBUG_DIR;
 273
 274         /* 2 characters per byte printed in hex, +1 for '/' and +1 for '\0' */
 275         build_id_file_len = (2 * bin->build_id_len) + 1 +
 276                         strlen(BUILD_ID_SUFFIX) + 1;
 277         build_id_file = malloc(build_id_file_len);
 278         if (!build_id_file) {
 279                 goto error;
 280         }
 281
 282         snprintf(build_id_file, 4, "%02x/", bin->build_id[0]);
 283         for (i = 1; i < bin->build_id_len; ++i) {
 284                 int path_idx = 3 + 2 * (i - 1);
 285
 286                 snprintf(&build_id_file[path_idx], 3, "%02x", bin->build_id[i]);
 287         }
 288         strcat(build_id_file, BUILD_ID_SUFFIX);
 289
 290         path = g_build_path("/", dbg_dir, BUILD_ID_SUBDIR, build_id_file, NULL);
 291         if (!path) {
 292                 goto error;
 293         }
 294
 295         ret = bin_info_set_dwarf_info_from_path(bin, path);
 296         if (ret) {
 297                 goto error;
 298         }
 299
 300         goto end;
 301
 302 error:
 303         ret = -1;
 304 end:
 305         free(build_id_file);
 306         free(path);
 307
 308         return ret;
 309 }
 310
 311 /**
 312  * Tests whether the file located at path exists and has the expected
 313  * checksum.
 314  *
 315  * This predicate is used when looking up separate debug info via the
 316  * GNU debuglink method. The expected crc can be found .gnu_debuglink
 317  * section in the original ELF file, along with the filename for the
 318  * file containing the debug info.
 319  *
 320  * @param path  Full path at which to look for the debug file
 321  * @param crc   Expected checksum for the debug file
 322  * @returns     1 if the file exists and has the correct checksum,
 323  *              0 otherwise
 324  */
 325 static
 326 int is_valid_debug_file(char *path, uint32_t crc)
 327 {
 328         int ret = 0, fd = -1;
 329         uint32_t _crc = 0;
 330
 331         if (!path) {
 332                 goto end_noclose;
 333         }
 334
 335         fd = open(path, O_RDONLY);
 336         if (fd < 0) {
 337                 goto end_noclose;
 338         }
 339
 340         ret = crc32(fd, &_crc);
 341         if (ret) {
 342                 ret = 0;
 343                 goto end;
 344         }
 345
 346         ret = (crc == _crc);
 347
 348 end:
 349         close(fd);
 350 end_noclose:
 351         return ret;
 352 }
 353
 354 /**
 355  * Try to set the dwarf_info for a given bin_info instance via the
 356  * build ID method.
 357  *
 358  * @param bin           bin_info instance for which to retrieve the
 359  *                      DWARF info via debug link
 360  * @returns             0 on success (i.e. dwarf_info set), -1 on failure
 361  */
 362 static
 363 int bin_info_set_dwarf_info_debug_link(struct bin_info *bin)
 364 {
 365         int ret = 0;
 366         const char *dbg_dir = NULL;
 367         char *dir_name = NULL, *bin_dir = NULL, *path = NULL;
 368         size_t max_path_len = 0;
 369
 370         if (!bin || !bin->dbg_link_filename) {
 371                 goto error;
 372         }
 373
 374         dbg_dir = opt_debug_info_dir ? : DEFAULT_DEBUG_DIR;
 375
 376         dir_name = dirname(bin->elf_path);
 377         if (!dir_name) {
 378                 goto error;
 379         }
 380
 381         /* bin_dir is just dir_name with a trailing slash */
 382         bin_dir = malloc(strlen(dir_name) + 2);
 383         if (!bin_dir) {
 384                 goto error;
 385         }
 386
 387         strcpy(bin_dir, dir_name);
 388         strcat(bin_dir, "/");
 389
 390         max_path_len = strlen(dbg_dir) + strlen(bin_dir) +
 391                         strlen(DEBUG_SUBDIR) + strlen(bin->dbg_link_filename)
 392                         + 1;
 393         path = malloc(max_path_len);
 394         if (!path) {
 395                 goto error;
 396         }
 397
 398         /* First look in the executable's dir */
 399         strcpy(path, bin_dir);
 400         strcat(path, bin->dbg_link_filename);
 401
 402         if (is_valid_debug_file(path, bin->dbg_link_crc)) {
 403                 goto found;
 404         }
 405
 406         /* If not found, look in .debug subdir */
 407         strcpy(path, bin_dir);
 408         strcat(path, DEBUG_SUBDIR);
 409         strcat(path, bin->dbg_link_filename);
 410
 411         if (is_valid_debug_file(path, bin->dbg_link_crc)) {
 412                 goto found;
 413         }
 414
 415         /* Lastly, look under the global debug directory */
 416         strcpy(path, dbg_dir);
 417         strcat(path, bin_dir);
 418         strcat(path, bin->dbg_link_filename);
 419
 420         if (is_valid_debug_file(path, bin->dbg_link_crc)) {
 421                 goto found;
 422         }
 423
 424 error:
 425         ret = -1;
 426 end:
 427         free(path);
 428         free(bin_dir);
 429
 430         return ret;
 431
 432 found:
 433         ret = bin_info_set_dwarf_info_from_path(bin, path);
 434         if (ret) {
 435                 goto error;
 436         }
 437
 438         goto end;
 439 }
 440
 441 /**
 442  * Initialize the DWARF info for a given executable.
 443  *
 444  * @param bin   bin_info instance
 445  * @returns     0 on success, negative value on failure
 446  */
 447 static
 448 int bin_info_set_dwarf_info(struct bin_info *bin)
 449 {
 450         int ret = 0;
 451
 452         if (!bin) {
 453                 ret = -1;
 454                 goto end;
 455         }
 456
 457         /* First try to set the DWARF info from the ELF file */
 458         ret = bin_info_set_dwarf_info_from_path(bin, bin->elf_path);
 459         if (!ret) {
 460                 goto end;
 461         }
 462
 463         /*
 464          * If that fails, try to find separate debug info via build ID
 465          * and debug link.
 466          */
 467         ret = bin_info_set_dwarf_info_build_id(bin);
 468         if (!ret) {
 469                 goto end;
 470         }
 471
 472         ret = bin_info_set_dwarf_info_debug_link(bin);
 473         if (!ret) {
 474                 goto end;
 475         }
 476
 477 end:
 478         return ret;
 479 }
 480
 481 /**
 482  * Initialize the ELF file for a given executable.
 483  *
 484  * @param bin   bin_info instance
 485  * @returns     0 on success, negative value on error.
 486  */
 487 static
 488 int bin_info_set_elf_file(struct bin_info *bin)
 489 {
 490         int elf_fd = -1;
 491         Elf *elf_file = NULL;
 492
 493         if (!bin) {
 494                 goto error;
 495         }
 496
 497         elf_fd = open(bin->elf_path, O_RDONLY);
 498         if (elf_fd < 0) {
 499                 elf_fd = -errno;
 500                 printf_verbose("Failed to open %s\n", bin->elf_path);
 501                 goto error;
 502         }
 503
 504         elf_file = elf_begin(elf_fd, ELF_C_READ, NULL);
 505         if (!elf_file) {
 506                 printf_debug("elf_begin failed: %s\n", elf_errmsg(-1));
 507                 goto error;
 508         }
 509
 510         if (elf_kind(elf_file) != ELF_K_ELF) {
 511                 printf_verbose("Error: %s is not an ELF object\n",
 512                                 bin->elf_path);
 513                 goto error;
 514         }
 515
 516         bin->elf_fd = elf_fd;
 517         bin->elf_file = elf_file;
 518         return 0;
 519
 520 error:
 521         if (elf_fd >= 0) {
 522                 close(elf_fd);
 523                 elf_fd = -1;
 524         }
 525         elf_end(elf_file);
 526         return elf_fd;
 527 }
 528
 529 BT_HIDDEN
 530 void source_location_destroy(struct source_location *src_loc)
 531 {
 532         if (!src_loc) {
 533                 return;
 534         }
 535
 536         free(src_loc->filename);
 537         g_free(src_loc);
 538 }
 539
 540 /**
 541  * Append a string representation of an address offset to an existing
 542  * string.
 543  *
 544  * On success, the out parameter `result` will contain the base string
 545  * followed by the offset string of the form "+0x1234". On failure,
 546  * `result` remains unchanged.
 547  *
 548  * @param base_str      The string to which to append an offset string
 549  * @param low_addr      The lower virtual memory address, the base from
 550  *                      which the offset is computed
 551  * @param high_addr     The higher virtual memory address
 552  * @param result        Out parameter, the base string followed by the
 553  *                      offset string
 554  * @returns             0 on success, -1 on failure
 555  */
 556 static
 557 int bin_info_append_offset_str(const char *base_str, uint64_t low_addr,
 558                                 uint64_t high_addr, char **result)
 559 {
 560         int ret;
 561         uint64_t offset;
 562         char *_result = NULL;
 563         char offset_str[ADDR_STR_LEN];
 564
 565         if (!base_str || !result) {
 566                 goto error;
 567         }
 568
 569         offset = high_addr - low_addr;
 570
 571         _result = malloc(strlen(base_str) + ADDR_STR_LEN);
 572         if (!_result) {
 573                 goto error;
 574         }
 575
 576         ret = snprintf(offset_str, ADDR_STR_LEN, "+%#0" PRIx64, offset);
 577         if (ret < 0) {
 578                 goto error;
 579         }
 580         strcpy(_result, base_str);
 581         strcat(_result, offset_str);
 582         *result = _result;
 583
 584         return 0;
 585
 586 error:
 587         free(_result);
 588         return -1;
 589 }
 590
 591 /**
 592  * Try to find the symbol closest to an address within a given ELF
 593  * section.
 594  *
 595  * Only function symbols are taken into account. The symbol's address
 596  * must precede `addr`. A symbol with a closer address might exist
 597  * after `addr` but is irrelevant because it cannot encompass `addr`.
 598  *
 599  * On success, if found, the out parameters `sym` and `shdr` are
 600  * set. On failure or if none are found, they remain unchanged.
 601  *
 602  * @param scn           ELF section in which to look for the address
 603  * @param addr          Virtual memory address for which to find the
 604  *                      nearest function symbol
 605  * @param sym           Out parameter, the nearest function symbol
 606  * @param shdr          Out parameter, the section header for scn
 607  * @returns             0 on success, -1 on failure
 608  */
 609 static
 610 int bin_info_get_nearest_symbol_from_section(Elf_Scn *scn, uint64_t addr,
 611                 GElf_Sym **sym, GElf_Shdr **shdr)
 612 {
 613         int i;
 614         size_t symbol_count;
 615         Elf_Data *data = NULL;
 616         GElf_Shdr *_shdr = NULL;
 617         GElf_Sym *nearest_sym = NULL;
 618
 619         if (!scn || !sym || !shdr) {
 620                 goto error;
 621         }
 622
 623         _shdr = g_new0(GElf_Shdr, 1);
 624         if (!_shdr) {
 625                 goto error;
 626         }
 627
 628         _shdr = gelf_getshdr(scn, _shdr);
 629         if (!_shdr) {
 630                 goto error;
 631         }
 632
 633         if (_shdr->sh_type != SHT_SYMTAB) {
 634                 /*
 635                  * We are only interested in symbol table (symtab)
 636                  * sections, skip this one.
 637                  */
 638                 goto end;
 639         }
 640
 641         data = elf_getdata(scn, NULL);
 642         if (!data) {
 643                 goto error;
 644         }
 645
 646         symbol_count = _shdr->sh_size / _shdr->sh_entsize;
 647
 648         for (i = 0; i < symbol_count; ++i) {
 649                 GElf_Sym *cur_sym = NULL;
 650
 651                 cur_sym = g_new0(GElf_Sym, 1);
 652                 if (!cur_sym) {
 653                         goto error;
 654                 }
 655                 cur_sym = gelf_getsym(data, i, cur_sym);
 656                 if (!cur_sym) {
 657                         goto error;
 658                 }
 659                 if (GELF_ST_TYPE(cur_sym->st_info) != STT_FUNC) {
 660                         /* We're only interested in the functions. */
 661                         g_free(cur_sym);
 662                         continue;
 663                 }
 664
 665                 if (cur_sym->st_value <= addr &&
 666                                 (!nearest_sym ||
 667                                 cur_sym->st_value > nearest_sym->st_value)) {
 668                         g_free(nearest_sym);
 669                         nearest_sym = cur_sym;
 670                 } else {
 671                         g_free(cur_sym);
 672                 }
 673         }
 674
 675 end:
 676         if (nearest_sym) {
 677                 *sym = nearest_sym;
 678                 *shdr = _shdr;
 679         } else {
 680                 g_free(_shdr);
 681         }
 682
 683         return 0;
 684
 685 error:
 686         g_free(nearest_sym);
 687         g_free(_shdr);
 688         return -1;
 689 }
 690
 691 /**
 692  * Get the name of the function containing a given address within an
 693  * executable using ELF symbols.
 694  *
 695  * The function name is in fact the name of the nearest ELF symbol,
 696  * followed by the offset in bytes between the address and the symbol
 697  * (in hex), separated by a '+' character.
 698  *
 699  * If found, the out parameter `func_name` is set on success. On failure,
 700  * it remains unchanged.
 701  *
 702  * @param bin           bin_info instance for the executable containing
 703  *                      the address
 704  * @param addr          Virtual memory address for which to find the
 705  *                      function name
 706  * @param func_name     Out parameter, the function name
 707  * @returns             0 on success, -1 on failure
 708  */
 709 static
 710 int bin_info_lookup_elf_function_name(struct bin_info *bin, uint64_t addr,
 711                 char **func_name)
 712 {
 713         /*
 714          * TODO (possible optimisation): if an ELF has no symtab
 715          * section, it has been stripped. Therefore, it would be wise
 716          * to store a flag indicating the stripped status after the
 717          * first iteration to prevent subsequent ones.
 718          */
 719         int ret = 0;
 720         Elf_Scn *scn = NULL;
 721         GElf_Sym *sym = NULL;
 722         GElf_Shdr *shdr = NULL;
 723         char *sym_name = NULL;
 724
 725         /* Set ELF file if it hasn't been accessed yet. */
 726         if (!bin->elf_file) {
 727                 ret = bin_info_set_elf_file(bin);
 728                 if (ret) {
 729                         /* Failed to set ELF file. */
 730                         goto error;
 731                 }
 732         }
 733
 734         scn = elf_nextscn(bin->elf_file, scn);
 735         if (!scn) {
 736                 goto error;
 737         }
 738
 739         while (scn && !sym) {
 740                 ret = bin_info_get_nearest_symbol_from_section(
 741                                 scn, addr, &sym, &shdr);
 742                 if (ret) {
 743                         goto error;
 744                 }
 745
 746                 scn = elf_nextscn(bin->elf_file, scn);
 747         }
 748
 749         if (sym) {
 750                 sym_name = elf_strptr(bin->elf_file, shdr->sh_link,
 751                                 sym->st_name);
 752                 if (!sym_name) {
 753                         goto error;
 754                 }
 755
 756                 ret = bin_info_append_offset_str(sym_name, sym->st_value, addr,
 757                                                 func_name);
 758                 if (ret) {
 759                         goto error;
 760                 }
 761         }
 762
 763         g_free(shdr);
 764         g_free(sym);
 765         return 0;
 766
 767 error:
 768         g_free(shdr);
 769         g_free(sym);
 770         return ret;
 771 }
 772
 773 /**
 774  * Get the name of the function containing a given address within a
 775  * given compile unit (CU).
 776  *
 777  * If found, the out parameter `func_name` is set on success. On
 778  * failure, it remains unchanged.
 779  *
 780  * @param cu            bt_dwarf_cu instance which may contain the address
 781  * @param addr          Virtual memory address for which to find the
 782  *                      function name
 783  * @param func_name     Out parameter, the function name
 784  * @returns             0 on success, -1 on failure
 785  */
 786 static
 787 int bin_info_lookup_cu_function_name(struct bt_dwarf_cu *cu, uint64_t addr,
 788                 char **func_name)
 789 {
 790         int ret = 0;
 791         bool found = false;
 792         struct bt_dwarf_die *die = NULL;
 793
 794         if (!cu || !func_name) {
 795                 goto error;
 796         }
 797
 798         die = bt_dwarf_die_create(cu);
 799         if (!die) {
 800                 goto error;
 801         }
 802
 803         while (bt_dwarf_die_next(die) == 0) {
 804                 int tag;
 805
 806                 ret = bt_dwarf_die_get_tag(die, &tag);
 807                 if (ret) {
 808                         goto error;
 809                 }
 810
 811                 if (tag == DW_TAG_subprogram) {
 812                         ret = bt_dwarf_die_contains_addr(die, addr, &found);
 813                         if (ret) {
 814                                 goto error;
 815                         }
 816
 817                         if (found) {
 818                                 break;
 819                         }
 820                 }
 821         }
 822
 823         if (found) {
 824                 uint64_t low_addr = 0;
 825                 char *die_name = NULL;
 826
 827                 ret = bt_dwarf_die_get_name(die, &die_name);
 828                 if (ret) {
 829                         goto error;
 830                 }
 831
 832                 ret = dwarf_lowpc(die->dwarf_die, &low_addr);
 833                 if (ret) {
 834                         free(die_name);
 835                         goto error;
 836                 }
 837
 838                 ret = bin_info_append_offset_str(die_name, low_addr, addr,
 839                                                 func_name);
 840                 free(die_name);
 841                 if (ret) {
 842                         goto error;
 843                 }
 844         }
 845
 846         bt_dwarf_die_destroy(die);
 847         return 0;
 848
 849 error:
 850         bt_dwarf_die_destroy(die);
 851         return -1;
 852 }
 853
 854 /**
 855  * Get the name of the function containing a given address within an
 856  * executable using DWARF debug info.
 857  *
 858  * If found, the out parameter `func_name` is set on success. On
 859  * failure, it remains unchanged.
 860  *
 861  * @param bin           bin_info instance for the executable containing
 862  *                      the address
 863  * @param addr          Virtual memory address for which to find the
 864  *                      function name
 865  * @param func_name     Out parameter, the function name
 866  * @returns             0 on success, -1 on failure
 867  */
 868 static
 869 int bin_info_lookup_dwarf_function_name(struct bin_info *bin, uint64_t addr,
 870                 char **func_name)
 871 {
 872         int ret = 0;
 873         char *_func_name = NULL;
 874         struct bt_dwarf_cu *cu = NULL;
 875
 876         if (!bin || !func_name) {
 877                 goto error;
 878         }
 879
 880         cu = bt_dwarf_cu_create(bin->dwarf_info);
 881         if (!cu) {
 882                 goto error;
 883         }
 884
 885         while (bt_dwarf_cu_next(cu) == 0) {
 886                 ret = bin_info_lookup_cu_function_name(cu, addr, &_func_name);
 887                 if (ret) {
 888                         goto error;
 889                 }
 890
 891                 if (_func_name) {
 892                         break;
 893                 }
 894         }
 895
 896         if (_func_name) {
 897                 *func_name = _func_name;
 898         } else {
 899                 goto error;
 900         }
 901
 902         bt_dwarf_cu_destroy(cu);
 903         return 0;
 904
 905 error:
 906         bt_dwarf_cu_destroy(cu);
 907         return -1;
 908 }
 909
 910 BT_HIDDEN
 911 int bin_info_lookup_function_name(struct bin_info *bin, uint64_t addr,
 912                 char **func_name)
 913 {
 914         int ret = 0;
 915         char *_func_name = NULL;
 916
 917         if (!bin || !func_name) {
 918                 goto error;
 919         }
 920
 921         /* Set DWARF info if it hasn't been accessed yet. */
 922         if (!bin->dwarf_info && !bin->is_elf_only) {
 923                 ret = bin_info_set_dwarf_info(bin);
 924                 if (ret) {
 925                         printf_verbose("Failed to set bin dwarf info, falling back to ELF lookup.\n");
 926                         /* Failed to set DWARF info, fallback to ELF. */
 927                         bin->is_elf_only = true;
 928                 }
 929         }
 930
 931         if (!bin_info_has_address(bin, addr)) {
 932                 goto error;
 933         }
 934
 935         /*
 936          * Addresses in ELF and DWARF are relative to base address for
 937          * PIC, so make the address argument relative too if needed.
 938          */
 939         if (bin->is_pic) {
 940                 addr -= bin->low_addr;
 941         }
 942
 943         if (bin->is_elf_only) {
 944                 ret = bin_info_lookup_elf_function_name(bin, addr, &_func_name);
 945                 printf_verbose("Failed to lookup function name (elf), error %i\n", ret);
 946         } else {
 947                 ret = bin_info_lookup_dwarf_function_name(bin, addr, &_func_name);
 948                 printf_verbose("Failed to lookup function name (dwarf), error %i\n", ret);
 949         }
 950
 951         *func_name = _func_name;
 952         return 0;
 953
 954 error:
 955         return -1;
 956 }
 957
 958 BT_HIDDEN
 959 int bin_info_get_bin_loc(struct bin_info *bin, uint64_t addr, char **bin_loc)
 960 {
 961         int ret = 0;
 962         char *_bin_loc = NULL;
 963
 964         if (!bin || !bin_loc) {
 965                 goto error;
 966         }
 967
 968         if (bin->is_pic) {
 969                 addr -= bin->low_addr;
 970                 ret = asprintf(&_bin_loc, "+%#0" PRIx64, addr);
 971         } else {
 972                 ret = asprintf(&_bin_loc, "@%#0" PRIx64, addr);
 973         }
 974
 975         if (ret == -1 || !_bin_loc) {
 976                 goto error;
 977         }
 978
 979         *bin_loc = _bin_loc;
 980         return 0;
 981
 982 error:
 983         return -1;
 984 }
 985
 986 /**
 987  * Predicate used to determine whether the children of a given DIE
 988  * contain a specific address.
 989  *
 990  * More specifically, the parameter `die` is expected to be a
 991  * subprogram (function) DIE, and this predicate tells whether any
 992  * subroutines are inlined within this function and would contain
 993  * `addr`.
 994  *
 995  * On success, the out parameter `contains` is set with the boolean
 996  * value indicating whether the DIE's range covers `addr`. On failure,
 997  * it remains unchanged.
 998  *
 999  * Do note that this function advances the position of `die`. If the
1000  * address is found within one of its children, `die` will be pointing
1001  * to that child upon returning from the function, allowing to extract
1002  * the information deemed necessary.
1003  *
1004  * @param die           The parent DIE in whose children the address will be
1005  *                      looked for
1006  * @param addr          The address for which to look for in the DIEs
1007  * @param contains      Out parameter, true if addr is contained,
1008  *                      false if not
1009  * @returns             Returns 0 on success, -1 on failure
1010  */
1011 static
1012 int bin_info_child_die_has_address(struct bt_dwarf_die *die, uint64_t addr, bool *contains)
1013 {
1014         int ret = 0;
1015         bool _contains = false;
1016
1017         if (!die) {
1018                 goto error;
1019         }
1020
1021         ret = bt_dwarf_die_child(die);
1022         if (ret) {
1023                 goto error;
1024         }
1025
1026         do {
1027                 int tag;
1028
1029                 ret = bt_dwarf_die_get_tag(die, &tag);
1030                 if (ret) {
1031                         goto error;
1032                 }
1033
1034                 if (tag == DW_TAG_inlined_subroutine) {
1035                         ret = bt_dwarf_die_contains_addr(die, addr, &_contains);
1036                         if (ret) {
1037                                 goto error;
1038                         }
1039
1040                         if (_contains) {
1041                                 goto end;
1042                         }
1043                 }
1044         } while (bt_dwarf_die_next(die) == 0);
1045
1046 end:
1047         *contains = _contains;
1048         return 0;
1049
1050 error:
1051         return -1;
1052 }
1053
1054 /**
1055  * Lookup the source location for a given address within a CU, making
1056  * the assumption that it is contained within an inline routine in a
1057  * function.
1058  *
1059  * @param cu            bt_dwarf_cu instance in which to look for the address
1060  * @param addr          The address for which to look for
1061  * @param src_loc       Out parameter, the source location (filename and
1062  *                      line number) for the address
1063  * @returns             0 on success, -1 on failure
1064  */
1065 static
1066 int bin_info_lookup_cu_src_loc_inl(struct bt_dwarf_cu *cu, uint64_t addr,
1067                 struct source_location **src_loc)
1068 {
1069         int ret = 0;
1070         bool found = false;
1071         struct bt_dwarf_die *die = NULL;
1072         struct source_location *_src_loc = NULL;
1073
1074         if (!cu || !src_loc) {
1075                 goto error;
1076         }
1077
1078         die = bt_dwarf_die_create(cu);
1079         if (!die) {
1080                 goto error;
1081         }
1082
1083         while (bt_dwarf_die_next(die) == 0) {
1084                 int tag;
1085
1086                 ret = bt_dwarf_die_get_tag(die, &tag);
1087                 if (ret) {
1088                         goto error;
1089                 }
1090
1091                 if (tag == DW_TAG_subprogram) {
1092                         bool contains = false;
1093
1094                         ret = bt_dwarf_die_contains_addr(die, addr, &contains);
1095                         if (ret) {
1096                                 goto error;
1097                         }
1098
1099                         if (contains) {
1100                                 /*
1101                                  * Try to find an inlined subroutine
1102                                  * child of this DIE containing addr.
1103                                  */
1104                                 ret = bin_info_child_die_has_address(die, addr,
1105                                                 &found);
1106                                 if(ret) {
1107                                         goto error;
1108                                 }
1109
1110                                 goto end;
1111                         }
1112                 }
1113         }
1114
1115 end:
1116         if (found) {
1117                 char *filename = NULL;
1118                 uint64_t line_no;
1119
1120                 _src_loc = g_new0(struct source_location, 1);
1121                 if (!_src_loc) {
1122                         goto error;
1123                 }
1124
1125                 ret = bt_dwarf_die_get_call_file(die, &filename);
1126                 if (ret) {
1127                         goto error;
1128                 }
1129                 ret = bt_dwarf_die_get_call_line(die, &line_no);
1130                 if (ret) {
1131                         free(filename);
1132                         goto error;
1133                 }
1134
1135                 _src_loc->filename = filename;
1136                 _src_loc->line_no = line_no;
1137                 *src_loc = _src_loc;
1138         }
1139
1140         bt_dwarf_die_destroy(die);
1141         return 0;
1142
1143 error:
1144         source_location_destroy(_src_loc);
1145         bt_dwarf_die_destroy(die);
1146         return -1;
1147 }
1148
1149 /**
1150  * Lookup the source location for a given address within a CU,
1151  * assuming that it is contained within an inlined function.
1152  *
1153  * A source location can be found regardless of inlining status for
1154  * this method, but in the case of an inlined function, the returned
1155  * source location will point not to the callsite but rather to the
1156  * definition site of the inline function.
1157  *
1158  * @param cu            bt_dwarf_cu instance in which to look for the address
1159  * @param addr          The address for which to look for
1160  * @param src_loc       Out parameter, the source location (filename and
1161  *                      line number) for the address
1162  * @returns             0 on success, -1 on failure
1163  */
1164 static
1165 int bin_info_lookup_cu_src_loc_no_inl(struct bt_dwarf_cu *cu, uint64_t addr,
1166                 struct source_location **src_loc)
1167 {
1168         struct source_location *_src_loc = NULL;
1169         struct bt_dwarf_die *die = NULL;
1170         const char *filename = NULL;
1171         Dwarf_Line *line = NULL;
1172         Dwarf_Addr line_addr;
1173         int ret, line_no;
1174
1175         if (!cu || !src_loc) {
1176                 goto error;
1177         }
1178
1179         die = bt_dwarf_die_create(cu);
1180         if (!die) {
1181                 goto error;
1182         }
1183
1184         line = dwarf_getsrc_die(die->dwarf_die, addr);
1185         if (!line) {
1186                 goto error;
1187         }
1188
1189         ret = dwarf_lineaddr(line, &line_addr);
1190         if (ret) {
1191                 goto error;
1192         }
1193
1194         filename = dwarf_linesrc(line, NULL, NULL);
1195         if (!filename) {
1196                 goto error;
1197         }
1198
1199         if (addr == line_addr) {
1200                 _src_loc = g_new0(struct source_location, 1);
1201                 if (!_src_loc) {
1202                         goto error;
1203                 }
1204
1205                 ret = dwarf_lineno(line, &line_no);
1206                 if (ret) {
1207                         goto error;
1208                 }
1209
1210                 _src_loc->line_no = line_no;
1211                 _src_loc->filename = strdup(filename);
1212         }
1213
1214         bt_dwarf_die_destroy(die);
1215
1216         if (_src_loc) {
1217                 *src_loc = _src_loc;
1218         }
1219
1220         return 0;
1221
1222 error:
1223         source_location_destroy(_src_loc);
1224         bt_dwarf_die_destroy(die);
1225         return -1;
1226 }
1227
1228 /**
1229  * Get the source location (file name and line number) for a given
1230  * address within a compile unit (CU).
1231  *
1232  * On success, the out parameter `src_loc` is set if found. On
1233  * failure, it remains unchanged.
1234  *
1235  * @param cu            bt_dwarf_cu instance for the compile unit which
1236  *                      may contain the address
1237  * @param addr          Virtual memory address for which to find the
1238  *                      source location
1239  * @param src_loc       Out parameter, the source location
1240  * @returns             0 on success, -1 on failure
1241  */
1242 static
1243 int bin_info_lookup_cu_src_loc(struct bt_dwarf_cu *cu, uint64_t addr,
1244                 struct source_location **src_loc)
1245 {
1246         int ret = 0;
1247         struct source_location *_src_loc = NULL;
1248
1249         if (!cu || !src_loc) {
1250                 goto error;
1251         }
1252
1253         ret = bin_info_lookup_cu_src_loc_inl(cu, addr, &_src_loc);
1254         if (ret) {
1255                 goto error;
1256         }
1257
1258         if (_src_loc) {
1259                 goto end;
1260         }
1261
1262         ret = bin_info_lookup_cu_src_loc_no_inl(cu, addr, &_src_loc);
1263         if (ret) {
1264                 goto error;
1265         }
1266
1267         if (_src_loc) {
1268                 goto end;
1269         }
1270
1271 end:
1272         if (_src_loc) {
1273                 *src_loc = _src_loc;
1274         }
1275
1276         return 0;
1277
1278 error:
1279         source_location_destroy(_src_loc);
1280         return -1;
1281 }
1282
1283 BT_HIDDEN
1284 int bin_info_lookup_source_location(struct bin_info *bin, uint64_t addr,
1285                 struct source_location **src_loc)
1286 {
1287         struct bt_dwarf_cu *cu = NULL;
1288         struct source_location *_src_loc = NULL;
1289
1290         if (!bin || !src_loc) {
1291                 goto error;
1292         }
1293
1294         /* Set DWARF info if it hasn't been accessed yet. */
1295         if (!bin->dwarf_info && !bin->is_elf_only) {
1296                 if (bin_info_set_dwarf_info(bin)) {
1297                         /* Failed to set DWARF info. */
1298                         bin->is_elf_only = true;
1299                 }
1300         }
1301
1302         if (bin->is_elf_only) {
1303                 /* We cannot lookup source location without DWARF info. */
1304                 goto error;
1305         }
1306
1307         if (!bin_info_has_address(bin, addr)) {
1308                 goto error;
1309         }
1310
1311         /*
1312          * Addresses in ELF and DWARF are relative to base address for
1313          * PIC, so make the address argument relative too if needed.
1314          */
1315         if (bin->is_pic) {
1316                 addr -= bin->low_addr;
1317         }
1318
1319         cu = bt_dwarf_cu_create(bin->dwarf_info);
1320         if (!cu) {
1321                 goto error;
1322         }
1323
1324         while (bt_dwarf_cu_next(cu) == 0) {
1325                 int ret;
1326
1327                 ret = bin_info_lookup_cu_src_loc(cu, addr, &_src_loc);
1328                 if (ret) {
1329                         goto error;
1330                 }
1331
1332                 if (_src_loc) {
1333                         break;
1334                 }
1335         }
1336
1337         bt_dwarf_cu_destroy(cu);
1338         if (_src_loc) {
1339                 *src_loc = _src_loc;
1340         }
1341
1342         return 0;
1343
1344 error:
1345         source_location_destroy(_src_loc);
1346         bt_dwarf_cu_destroy(cu);
1347         return -1;
1348 }