tools/perf/util/symbol.c

   1 #define _GNU_SOURCE
   2 #include <ctype.h>
   3 #include <dirent.h>
   4 #include <errno.h>
   5 #include <libgen.h>
   6 #include <stdlib.h>
   7 #include <stdio.h>
   8 #include <string.h>
   9 #include <sys/types.h>
  10 #include <sys/stat.h>
  11 #include <sys/param.h>
  12 #include <fcntl.h>
  13 #include <unistd.h>
  14 #include <inttypes.h>
  15 #include "build-id.h"
  16 #include "debug.h"
  17 #include "symbol.h"
  18 #include "strlist.h"
  19
  20 #include <libelf.h>
  21 #include <gelf.h>
  22 #include <elf.h>
  23 #include <limits.h>
  24 #include <sys/utsname.h>
  25
  26 #ifndef KSYM_NAME_LEN
  27 #define KSYM_NAME_LEN 128
  28 #endif
  29
  30 #ifndef NT_GNU_BUILD_ID
  31 #define NT_GNU_BUILD_ID 3
  32 #endif
  33
  34 static bool dso__build_id_equal(const struct dso *dso, u8 *build_id);
  35 static int elf_read_build_id(Elf *elf, void *bf, size_t size);
  36 static void dsos__add(struct list_head *head, struct dso *dso);
  37 static struct map *map__new2(u64 start, struct dso *dso, enum map_type type);
  38 static int dso__load_kernel_sym(struct dso *dso, struct map *map,
  39                                 symbol_filter_t filter);
  40 static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
  41                         symbol_filter_t filter);
  42 static int vmlinux_path__nr_entries;
  43 static char **vmlinux_path;
  44
  45 struct symbol_conf symbol_conf = {
  46         .exclude_other    = true,
  47         .use_modules      = true,
  48         .try_vmlinux_path = true,
  49         .symfs            = "",
  50 };
  51
  52 int dso__name_len(const struct dso *dso)
  53 {
  54         if (verbose)
  55                 return dso->long_name_len;
  56
  57         return dso->short_name_len;
  58 }
  59
  60 bool dso__loaded(const struct dso *dso, enum map_type type)
  61 {
  62         return dso->loaded & (1 << type);
  63 }
  64
  65 bool dso__sorted_by_name(const struct dso *dso, enum map_type type)
  66 {
  67         return dso->sorted_by_name & (1 << type);
  68 }
  69
  70 static void dso__set_sorted_by_name(struct dso *dso, enum map_type type)
  71 {
  72         dso->sorted_by_name |= (1 << type);
  73 }
  74
  75 bool symbol_type__is_a(char symbol_type, enum map_type map_type)
  76 {
  77         symbol_type = toupper(symbol_type);
  78
  79         switch (map_type) {
  80         case MAP__FUNCTION:
  81                 return symbol_type == 'T' || symbol_type == 'W';
  82         case MAP__VARIABLE:
  83                 return symbol_type == 'D';
  84         default:
  85                 return false;
  86         }
  87 }
  88
  89 static int prefix_underscores_count(const char *str)
  90 {
  91         const char *tail = str;
  92
  93         while (*tail == '_')
  94                 tail++;
  95
  96         return tail - str;
  97 }
  98
  99 #define SYMBOL_A 0
 100 #define SYMBOL_B 1
 101
 102 static int choose_best_symbol(struct symbol *syma, struct symbol *symb)
 103 {
 104         s64 a;
 105         s64 b;
 106
 107         /* Prefer a symbol with non zero length */
 108         a = syma->end - syma->start;
 109         b = symb->end - symb->start;
 110         if ((b == 0) && (a > 0))
 111                 return SYMBOL_A;
 112         else if ((a == 0) && (b > 0))
 113                 return SYMBOL_B;
 114
 115         /* Prefer a non weak symbol over a weak one */
 116         a = syma->binding == STB_WEAK;
 117         b = symb->binding == STB_WEAK;
 118         if (b && !a)
 119                 return SYMBOL_A;
 120         if (a && !b)
 121                 return SYMBOL_B;
 122
 123         /* Prefer a global symbol over a non global one */
 124         a = syma->binding == STB_GLOBAL;
 125         b = symb->binding == STB_GLOBAL;
 126         if (a && !b)
 127                 return SYMBOL_A;
 128         if (b && !a)
 129                 return SYMBOL_B;
 130
 131         /* Prefer a symbol with less underscores */
 132         a = prefix_underscores_count(syma->name);
 133         b = prefix_underscores_count(symb->name);
 134         if (b > a)
 135                 return SYMBOL_A;
 136         else if (a > b)
 137                 return SYMBOL_B;
 138
 139         /* If all else fails, choose the symbol with the longest name */
 140         if (strlen(syma->name) >= strlen(symb->name))
 141                 return SYMBOL_A;
 142         else
 143                 return SYMBOL_B;
 144 }
 145
 146 static void symbols__fixup_duplicate(struct rb_root *symbols)
 147 {
 148         struct rb_node *nd;
 149         struct symbol *curr, *next;
 150
 151         nd = rb_first(symbols);
 152
 153         while (nd) {
 154                 curr = rb_entry(nd, struct symbol, rb_node);
 155 again:
 156                 nd = rb_next(&curr->rb_node);
 157                 next = rb_entry(nd, struct symbol, rb_node);
 158
 159                 if (!nd)
 160                         break;
 161
 162                 if (curr->start != next->start)
 163                         continue;
 164
 165                 if (choose_best_symbol(curr, next) == SYMBOL_A) {
 166                         rb_erase(&next->rb_node, symbols);
 167                         goto again;
 168                 } else {
 169                         nd = rb_next(&curr->rb_node);
 170                         rb_erase(&curr->rb_node, symbols);
 171                 }
 172         }
 173 }
 174
 175 static void symbols__fixup_end(struct rb_root *symbols)
 176 {
 177         struct rb_node *nd, *prevnd = rb_first(symbols);
 178         struct symbol *curr, *prev;
 179
 180         if (prevnd == NULL)
 181                 return;
 182
 183         curr = rb_entry(prevnd, struct symbol, rb_node);
 184
 185         for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
 186                 prev = curr;
 187                 curr = rb_entry(nd, struct symbol, rb_node);
 188
 189                 if (prev->end == prev->start && prev->end != curr->start)
 190                         prev->end = curr->start - 1;
 191         }
 192
 193         /* Last entry */
 194         if (curr->end == curr->start)
 195                 curr->end = roundup(curr->start, 4096);
 196 }
 197
 198 static void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
 199 {
 200         struct map *prev, *curr;
 201         struct rb_node *nd, *prevnd = rb_first(&mg->maps[type]);
 202
 203         if (prevnd == NULL)
 204                 return;
 205
 206         curr = rb_entry(prevnd, struct map, rb_node);
 207
 208         for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
 209                 prev = curr;
 210                 curr = rb_entry(nd, struct map, rb_node);
 211                 prev->end = curr->start - 1;
 212         }
 213
 214         /*
 215          * We still haven't the actual symbols, so guess the
 216          * last map final address.
 217          */
 218         curr->end = ~0ULL;
 219 }
 220
 221 static void map_groups__fixup_end(struct map_groups *mg)
 222 {
 223         int i;
 224         for (i = 0; i < MAP__NR_TYPES; ++i)
 225                 __map_groups__fixup_end(mg, i);
 226 }
 227
 228 static struct symbol *symbol__new(u64 start, u64 len, u8 binding,
 229                                   const char *name)
 230 {
 231         size_t namelen = strlen(name) + 1;
 232         struct symbol *sym = calloc(1, (symbol_conf.priv_size +
 233                                         sizeof(*sym) + namelen));
 234         if (sym == NULL)
 235                 return NULL;
 236
 237         if (symbol_conf.priv_size)
 238                 sym = ((void *)sym) + symbol_conf.priv_size;
 239
 240         sym->start   = start;
 241         sym->end     = len ? start + len - 1 : start;
 242         sym->binding = binding;
 243         sym->namelen = namelen - 1;
 244
 245         pr_debug4("%s: %s %#" PRIx64 "-%#" PRIx64 "\n",
 246                   __func__, name, start, sym->end);
 247         memcpy(sym->name, name, namelen);
 248
 249         return sym;
 250 }
 251
 252 void symbol__delete(struct symbol *sym)
 253 {
 254         free(((void *)sym) - symbol_conf.priv_size);
 255 }
 256
 257 static size_t symbol__fprintf(struct symbol *sym, FILE *fp)
 258 {
 259         return fprintf(fp, " %" PRIx64 "-%" PRIx64 " %c %s\n",
 260                        sym->start, sym->end,
 261                        sym->binding == STB_GLOBAL ? 'g' :
 262                        sym->binding == STB_LOCAL  ? 'l' : 'w',
 263                        sym->name);
 264 }
 265
 266 void dso__set_long_name(struct dso *dso, char *name)
 267 {
 268         if (name == NULL)
 269                 return;
 270         dso->long_name = name;
 271         dso->long_name_len = strlen(name);
 272 }
 273
 274 static void dso__set_short_name(struct dso *dso, const char *name)
 275 {
 276         if (name == NULL)
 277                 return;
 278         dso->short_name = name;
 279         dso->short_name_len = strlen(name);
 280 }
 281
 282 static void dso__set_basename(struct dso *dso)
 283 {
 284         dso__set_short_name(dso, basename(dso->long_name));
 285 }
 286
 287 struct dso *dso__new(const char *name)
 288 {
 289         struct dso *dso = calloc(1, sizeof(*dso) + strlen(name) + 1);
 290
 291         if (dso != NULL) {
 292                 int i;
 293                 strcpy(dso->name, name);
 294                 dso__set_long_name(dso, dso->name);
 295                 dso__set_short_name(dso, dso->name);
 296                 for (i = 0; i < MAP__NR_TYPES; ++i)
 297                         dso->symbols[i] = dso->symbol_names[i] = RB_ROOT;
 298                 dso->symtab_type = SYMTAB__NOT_FOUND;
 299                 dso->loaded = 0;
 300                 dso->sorted_by_name = 0;
 301                 dso->has_build_id = 0;
 302                 dso->kernel = DSO_TYPE_USER;
 303                 INIT_LIST_HEAD(&dso->node);
 304         }
 305
 306         return dso;
 307 }
 308
 309 static void symbols__delete(struct rb_root *symbols)
 310 {
 311         struct symbol *pos;
 312         struct rb_node *next = rb_first(symbols);
 313
 314         while (next) {
 315                 pos = rb_entry(next, struct symbol, rb_node);
 316                 next = rb_next(&pos->rb_node);
 317                 rb_erase(&pos->rb_node, symbols);
 318                 symbol__delete(pos);
 319         }
 320 }
 321
 322 void dso__delete(struct dso *dso)
 323 {
 324         int i;
 325         for (i = 0; i < MAP__NR_TYPES; ++i)
 326                 symbols__delete(&dso->symbols[i]);
 327         if (dso->sname_alloc)
 328                 free((char *)dso->short_name);
 329         if (dso->lname_alloc)
 330                 free(dso->long_name);
 331         free(dso);
 332 }
 333
 334 void dso__set_build_id(struct dso *dso, void *build_id)
 335 {
 336         memcpy(dso->build_id, build_id, sizeof(dso->build_id));
 337         dso->has_build_id = 1;
 338 }
 339
 340 static void symbols__insert(struct rb_root *symbols, struct symbol *sym)
 341 {
 342         struct rb_node **p = &symbols->rb_node;
 343         struct rb_node *parent = NULL;
 344         const u64 ip = sym->start;
 345         struct symbol *s;
 346
 347         while (*p != NULL) {
 348                 parent = *p;
 349                 s = rb_entry(parent, struct symbol, rb_node);
 350                 if (ip < s->start)
 351                         p = &(*p)->rb_left;
 352                 else
 353                         p = &(*p)->rb_right;
 354         }
 355         rb_link_node(&sym->rb_node, parent, p);
 356         rb_insert_color(&sym->rb_node, symbols);
 357 }
 358
 359 static struct symbol *symbols__find(struct rb_root *symbols, u64 ip)
 360 {
 361         struct rb_node *n;
 362
 363         if (symbols == NULL)
 364                 return NULL;
 365
 366         n = symbols->rb_node;
 367
 368         while (n) {
 369                 struct symbol *s = rb_entry(n, struct symbol, rb_node);
 370
 371                 if (ip < s->start)
 372                         n = n->rb_left;
 373                 else if (ip > s->end)
 374                         n = n->rb_right;
 375                 else
 376                         return s;
 377         }
 378
 379         return NULL;
 380 }
 381
 382 struct symbol_name_rb_node {
 383         struct rb_node  rb_node;
 384         struct symbol   sym;
 385 };
 386
 387 static void symbols__insert_by_name(struct rb_root *symbols, struct symbol *sym)
 388 {
 389         struct rb_node **p = &symbols->rb_node;
 390         struct rb_node *parent = NULL;
 391         struct symbol_name_rb_node *symn, *s;
 392
 393         symn = container_of(sym, struct symbol_name_rb_node, sym);
 394
 395         while (*p != NULL) {
 396                 parent = *p;
 397                 s = rb_entry(parent, struct symbol_name_rb_node, rb_node);
 398                 if (strcmp(sym->name, s->sym.name) < 0)
 399                         p = &(*p)->rb_left;
 400                 else
 401                         p = &(*p)->rb_right;
 402         }
 403         rb_link_node(&symn->rb_node, parent, p);
 404         rb_insert_color(&symn->rb_node, symbols);
 405 }
 406
 407 static void symbols__sort_by_name(struct rb_root *symbols,
 408                                   struct rb_root *source)
 409 {
 410         struct rb_node *nd;
 411
 412         for (nd = rb_first(source); nd; nd = rb_next(nd)) {
 413                 struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
 414                 symbols__insert_by_name(symbols, pos);
 415         }
 416 }
 417
 418 static struct symbol *symbols__find_by_name(struct rb_root *symbols,
 419                                             const char *name)
 420 {
 421         struct rb_node *n;
 422
 423         if (symbols == NULL)
 424                 return NULL;
 425
 426         n = symbols->rb_node;
 427
 428         while (n) {
 429                 struct symbol_name_rb_node *s;
 430                 int cmp;
 431
 432                 s = rb_entry(n, struct symbol_name_rb_node, rb_node);
 433                 cmp = strcmp(name, s->sym.name);
 434
 435                 if (cmp < 0)
 436                         n = n->rb_left;
 437                 else if (cmp > 0)
 438                         n = n->rb_right;
 439                 else
 440                         return &s->sym;
 441         }
 442
 443         return NULL;
 444 }
 445
 446 struct symbol *dso__find_symbol(struct dso *dso,
 447                                 enum map_type type, u64 addr)
 448 {
 449         return symbols__find(&dso->symbols[type], addr);
 450 }
 451
 452 struct symbol *dso__find_symbol_by_name(struct dso *dso, enum map_type type,
 453                                         const char *name)
 454 {
 455         return symbols__find_by_name(&dso->symbol_names[type], name);
 456 }
 457
 458 void dso__sort_by_name(struct dso *dso, enum map_type type)
 459 {
 460         dso__set_sorted_by_name(dso, type);
 461         return symbols__sort_by_name(&dso->symbol_names[type],
 462                                      &dso->symbols[type]);
 463 }
 464
 465 int build_id__sprintf(const u8 *build_id, int len, char *bf)
 466 {
 467         char *bid = bf;
 468         const u8 *raw = build_id;
 469         int i;
 470
 471         for (i = 0; i < len; ++i) {
 472                 sprintf(bid, "%02x", *raw);
 473                 ++raw;
 474                 bid += 2;
 475         }
 476
 477         return raw - build_id;
 478 }
 479
 480 size_t dso__fprintf_buildid(struct dso *dso, FILE *fp)
 481 {
 482         char sbuild_id[BUILD_ID_SIZE * 2 + 1];
 483
 484         build_id__sprintf(dso->build_id, sizeof(dso->build_id), sbuild_id);
 485         return fprintf(fp, "%s", sbuild_id);
 486 }
 487
 488 size_t dso__fprintf_symbols_by_name(struct dso *dso,
 489                                     enum map_type type, FILE *fp)
 490 {
 491         size_t ret = 0;
 492         struct rb_node *nd;
 493         struct symbol_name_rb_node *pos;
 494
 495         for (nd = rb_first(&dso->symbol_names[type]); nd; nd = rb_next(nd)) {
 496                 pos = rb_entry(nd, struct symbol_name_rb_node, rb_node);
 497                 fprintf(fp, "%s\n", pos->sym.name);
 498         }
 499
 500         return ret;
 501 }
 502
 503 size_t dso__fprintf(struct dso *dso, enum map_type type, FILE *fp)
 504 {
 505         struct rb_node *nd;
 506         size_t ret = fprintf(fp, "dso: %s (", dso->short_name);
 507
 508         if (dso->short_name != dso->long_name)
 509                 ret += fprintf(fp, "%s, ", dso->long_name);
 510         ret += fprintf(fp, "%s, %sloaded, ", map_type__name[type],
 511                        dso->loaded ? "" : "NOT ");
 512         ret += dso__fprintf_buildid(dso, fp);
 513         ret += fprintf(fp, ")\n");
 514         for (nd = rb_first(&dso->symbols[type]); nd; nd = rb_next(nd)) {
 515                 struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
 516                 ret += symbol__fprintf(pos, fp);
 517         }
 518
 519         return ret;
 520 }
 521
 522 int kallsyms__parse(const char *filename, void *arg,
 523                     int (*process_symbol)(void *arg, const char *name,
 524                                           char type, u64 start, u64 end))
 525 {
 526         char *line = NULL;
 527         size_t n;
 528         int err = -1;
 529         FILE *file = fopen(filename, "r");
 530
 531         if (file == NULL)
 532                 goto out_failure;
 533
 534         err = 0;
 535
 536         while (!feof(file)) {
 537                 u64 start;
 538                 int line_len, len;
 539                 char symbol_type;
 540                 char *symbol_name;
 541
 542                 line_len = getline(&line, &n, file);
 543                 if (line_len < 0 || !line)
 544                         break;
 545
 546                 line[--line_len] = '\0'; /* \n */
 547
 548                 len = hex2u64(line, &start);
 549
 550                 len++;
 551                 if (len + 2 >= line_len)
 552                         continue;
 553
 554                 symbol_type = line[len];
 555                 len += 2;
 556                 symbol_name = line + len;
 557                 len = line_len - len;
 558
 559                 if (len >= KSYM_NAME_LEN) {
 560                         err = -1;
 561                         break;
 562                 }
 563
 564                 /*
 565                  * module symbols are not sorted so we add all
 566                  * symbols with zero length and rely on
 567                  * symbols__fixup_end() to fix it up.
 568                  */
 569                 err = process_symbol(arg, symbol_name,
 570                                      symbol_type, start, start);
 571                 if (err)
 572                         break;
 573         }
 574
 575         free(line);
 576         fclose(file);
 577         return err;
 578
 579 out_failure:
 580         return -1;
 581 }
 582
 583 struct process_kallsyms_args {
 584         struct map *map;
 585         struct dso *dso;
 586 };
 587
 588 static u8 kallsyms2elf_type(char type)
 589 {
 590         if (type == 'W')
 591                 return STB_WEAK;
 592
 593         return isupper(type) ? STB_GLOBAL : STB_LOCAL;
 594 }
 595
 596 static int map__process_kallsym_symbol(void *arg, const char *name,
 597                                        char type, u64 start, u64 end)
 598 {
 599         struct symbol *sym;
 600         struct process_kallsyms_args *a = arg;
 601         struct rb_root *root = &a->dso->symbols[a->map->type];
 602
 603         if (!symbol_type__is_a(type, a->map->type))
 604                 return 0;
 605
 606         sym = symbol__new(start, end - start + 1,
 607                           kallsyms2elf_type(type), name);
 608         if (sym == NULL)
 609                 return -ENOMEM;
 610         /*
 611          * We will pass the symbols to the filter later, in
 612          * map__split_kallsyms, when we have split the maps per module
 613          */
 614         symbols__insert(root, sym);
 615
 616         return 0;
 617 }
 618
 619 /*
 620  * Loads the function entries in /proc/kallsyms into kernel_map->dso,
 621  * so that we can in the next step set the symbol ->end address and then
 622  * call kernel_maps__split_kallsyms.
 623  */
 624 static int dso__load_all_kallsyms(struct dso *dso, const char *filename,
 625                                   struct map *map)
 626 {
 627         struct process_kallsyms_args args = { .map = map, .dso = dso, };
 628         return kallsyms__parse(filename, &args, map__process_kallsym_symbol);
 629 }
 630
 631 /*
 632  * Split the symbols into maps, making sure there are no overlaps, i.e. the
 633  * kernel range is broken in several maps, named [kernel].N, as we don't have
 634  * the original ELF section names vmlinux have.
 635  */
 636 static int dso__split_kallsyms(struct dso *dso, struct map *map,
 637                                symbol_filter_t filter)
 638 {
 639         struct map_groups *kmaps = map__kmap(map)->kmaps;
 640         struct machine *machine = kmaps->machine;
 641         struct map *curr_map = map;
 642         struct symbol *pos;
 643         int count = 0, moved = 0;
 644         struct rb_root *root = &dso->symbols[map->type];
 645         struct rb_node *next = rb_first(root);
 646         int kernel_range = 0;
 647
 648         while (next) {
 649                 char *module;
 650
 651                 pos = rb_entry(next, struct symbol, rb_node);
 652                 next = rb_next(&pos->rb_node);
 653
 654                 module = strchr(pos->name, '\t');
 655                 if (module) {
 656                         if (!symbol_conf.use_modules)
 657                                 goto discard_symbol;
 658
 659                         *module++ = '\0';
 660
 661                         if (strcmp(curr_map->dso->short_name, module)) {
 662                                 if (curr_map != map &&
 663                                     dso->kernel == DSO_TYPE_GUEST_KERNEL &&
 664                                     machine__is_default_guest(machine)) {
 665                                         /*
 666                                          * We assume all symbols of a module are
 667                                          * continuous in * kallsyms, so curr_map
 668                                          * points to a module and all its
 669                                          * symbols are in its kmap. Mark it as
 670                                          * loaded.
 671                                          */
 672                                         dso__set_loaded(curr_map->dso,
 673                                                         curr_map->type);
 674                                 }
 675
 676                                 curr_map = map_groups__find_by_name(kmaps,
 677                                                         map->type, module);
 678                                 if (curr_map == NULL) {
 679                                         pr_debug("%s/proc/{kallsyms,modules} "
 680                                                  "inconsistency while looking "
 681                                                  "for \"%s\" module!\n",
 682                                                  machine->root_dir, module);
 683                                         curr_map = map;
 684                                         goto discard_symbol;
 685                                 }
 686
 687                                 if (curr_map->dso->loaded &&
 688                                     !machine__is_default_guest(machine))
 689                                         goto discard_symbol;
 690                         }
 691                         /*
 692                          * So that we look just like we get from .ko files,
 693                          * i.e. not prelinked, relative to map->start.
 694                          */
 695                         pos->start = curr_map->map_ip(curr_map, pos->start);
 696                         pos->end   = curr_map->map_ip(curr_map, pos->end);
 697                 } else if (curr_map != map) {
 698                         char dso_name[PATH_MAX];
 699                         struct dso *ndso;
 700
 701                         if (count == 0) {
 702                                 curr_map = map;
 703                                 goto filter_symbol;
 704                         }
 705
 706                         if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
 707                                 snprintf(dso_name, sizeof(dso_name),
 708                                         "[guest.kernel].%d",
 709                                         kernel_range++);
 710                         else
 711                                 snprintf(dso_name, sizeof(dso_name),
 712                                         "[kernel].%d",
 713                                         kernel_range++);
 714
 715                         ndso = dso__new(dso_name);
 716                         if (ndso == NULL)
 717                                 return -1;
 718
 719                         ndso->kernel = dso->kernel;
 720
 721                         curr_map = map__new2(pos->start, ndso, map->type);
 722                         if (curr_map == NULL) {
 723                                 dso__delete(ndso);
 724                                 return -1;
 725                         }
 726
 727                         curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
 728                         map_groups__insert(kmaps, curr_map);
 729                         ++kernel_range;
 730                 }
 731 filter_symbol:
 732                 if (filter && filter(curr_map, pos)) {
 733 discard_symbol:         rb_erase(&pos->rb_node, root);
 734                         symbol__delete(pos);
 735                 } else {
 736                         if (curr_map != map) {
 737                                 rb_erase(&pos->rb_node, root);
 738                                 symbols__insert(&curr_map->dso->symbols[curr_map->type], pos);
 739                                 ++moved;
 740                         } else
 741                                 ++count;
 742                 }
 743         }
 744
 745         if (curr_map != map &&
 746             dso->kernel == DSO_TYPE_GUEST_KERNEL &&
 747             machine__is_default_guest(kmaps->machine)) {
 748                 dso__set_loaded(curr_map->dso, curr_map->type);
 749         }
 750
 751         return count + moved;
 752 }
 753
 754 static bool symbol__restricted_filename(const char *filename,
 755                                         const char *restricted_filename)
 756 {
 757         bool restricted = false;
 758
 759         if (symbol_conf.kptr_restrict) {
 760                 char *r = realpath(filename, NULL);
 761
 762                 if (r != NULL) {
 763                         restricted = strcmp(r, restricted_filename) == 0;
 764                         free(r);
 765                         return restricted;
 766                 }
 767         }
 768
 769         return restricted;
 770 }
 771
 772 int dso__load_kallsyms(struct dso *dso, const char *filename,
 773                        struct map *map, symbol_filter_t filter)
 774 {
 775         if (symbol__restricted_filename(filename, "/proc/kallsyms"))
 776                 return -1;
 777
 778         if (dso__load_all_kallsyms(dso, filename, map) < 0)
 779                 return -1;
 780
 781         symbols__fixup_duplicate(&dso->symbols[map->type]);
 782         symbols__fixup_end(&dso->symbols[map->type]);
 783
 784         if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
 785                 dso->symtab_type = SYMTAB__GUEST_KALLSYMS;
 786         else
 787                 dso->symtab_type = SYMTAB__KALLSYMS;
 788
 789         return dso__split_kallsyms(dso, map, filter);
 790 }
 791
 792 static int dso__load_perf_map(struct dso *dso, struct map *map,
 793                               symbol_filter_t filter)
 794 {
 795         char *line = NULL;
 796         size_t n;
 797         FILE *file;
 798         int nr_syms = 0;
 799
 800         file = fopen(dso->long_name, "r");
 801         if (file == NULL)
 802                 goto out_failure;
 803
 804         while (!feof(file)) {
 805                 u64 start, size;
 806                 struct symbol *sym;
 807                 int line_len, len;
 808
 809                 line_len = getline(&line, &n, file);
 810                 if (line_len < 0)
 811                         break;
 812
 813                 if (!line)
 814                         goto out_failure;
 815
 816                 line[--line_len] = '\0'; /* \n */
 817
 818                 len = hex2u64(line, &start);
 819
 820                 len++;
 821                 if (len + 2 >= line_len)
 822                         continue;
 823
 824                 len += hex2u64(line + len, &size);
 825
 826                 len++;
 827                 if (len + 2 >= line_len)
 828                         continue;
 829
 830                 sym = symbol__new(start, size, STB_GLOBAL, line + len);
 831
 832                 if (sym == NULL)
 833                         goto out_delete_line;
 834
 835                 if (filter && filter(map, sym))
 836                         symbol__delete(sym);
 837                 else {
 838                         symbols__insert(&dso->symbols[map->type], sym);
 839                         nr_syms++;
 840                 }
 841         }
 842
 843         free(line);
 844         fclose(file);
 845
 846         return nr_syms;
 847
 848 out_delete_line:
 849         free(line);
 850 out_failure:
 851         return -1;
 852 }
 853
 854 /**
 855  * elf_symtab__for_each_symbol - iterate thru all the symbols
 856  *
 857  * @syms: struct elf_symtab instance to iterate
 858  * @idx: uint32_t idx
 859  * @sym: GElf_Sym iterator
 860  */
 861 #define elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) \
 862         for (idx = 0, gelf_getsym(syms, idx, &sym);\
 863              idx < nr_syms; \
 864              idx++, gelf_getsym(syms, idx, &sym))
 865
 866 static inline uint8_t elf_sym__type(const GElf_Sym *sym)
 867 {
 868         return GELF_ST_TYPE(sym->st_info);
 869 }
 870
 871 static inline int elf_sym__is_function(const GElf_Sym *sym)
 872 {
 873         return elf_sym__type(sym) == STT_FUNC &&
 874                sym->st_name != 0 &&
 875                sym->st_shndx != SHN_UNDEF;
 876 }
 877
 878 static inline bool elf_sym__is_object(const GElf_Sym *sym)
 879 {
 880         return elf_sym__type(sym) == STT_OBJECT &&
 881                 sym->st_name != 0 &&
 882                 sym->st_shndx != SHN_UNDEF;
 883 }
 884
 885 static inline int elf_sym__is_label(const GElf_Sym *sym)
 886 {
 887         return elf_sym__type(sym) == STT_NOTYPE &&
 888                 sym->st_name != 0 &&
 889                 sym->st_shndx != SHN_UNDEF &&
 890                 sym->st_shndx != SHN_ABS;
 891 }
 892
 893 static inline const char *elf_sec__name(const GElf_Shdr *shdr,
 894                                         const Elf_Data *secstrs)
 895 {
 896         return secstrs->d_buf + shdr->sh_name;
 897 }
 898
 899 static inline int elf_sec__is_text(const GElf_Shdr *shdr,
 900                                         const Elf_Data *secstrs)
 901 {
 902         return strstr(elf_sec__name(shdr, secstrs), "text") != NULL;
 903 }
 904
 905 static inline bool elf_sec__is_data(const GElf_Shdr *shdr,
 906                                     const Elf_Data *secstrs)
 907 {
 908         return strstr(elf_sec__name(shdr, secstrs), "data") != NULL;
 909 }
 910
 911 static inline const char *elf_sym__name(const GElf_Sym *sym,
 912                                         const Elf_Data *symstrs)
 913 {
 914         return symstrs->d_buf + sym->st_name;
 915 }
 916
 917 static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep,
 918                                     GElf_Shdr *shp, const char *name,
 919                                     size_t *idx)
 920 {
 921         Elf_Scn *sec = NULL;
 922         size_t cnt = 1;
 923
 924         while ((sec = elf_nextscn(elf, sec)) != NULL) {
 925                 char *str;
 926
 927                 gelf_getshdr(sec, shp);
 928                 str = elf_strptr(elf, ep->e_shstrndx, shp->sh_name);
 929                 if (!strcmp(name, str)) {
 930                         if (idx)
 931                                 *idx = cnt;
 932                         break;
 933                 }
 934                 ++cnt;
 935         }
 936
 937         return sec;
 938 }
 939
 940 #define elf_section__for_each_rel(reldata, pos, pos_mem, idx, nr_entries) \
 941         for (idx = 0, pos = gelf_getrel(reldata, 0, &pos_mem); \
 942              idx < nr_entries; \
 943              ++idx, pos = gelf_getrel(reldata, idx, &pos_mem))
 944
 945 #define elf_section__for_each_rela(reldata, pos, pos_mem, idx, nr_entries) \
 946         for (idx = 0, pos = gelf_getrela(reldata, 0, &pos_mem); \
 947              idx < nr_entries; \
 948              ++idx, pos = gelf_getrela(reldata, idx, &pos_mem))
 949
 950 /*
 951  * We need to check if we have a .dynsym, so that we can handle the
 952  * .plt, synthesizing its symbols, that aren't on the symtabs (be it
 953  * .dynsym or .symtab).
 954  * And always look at the original dso, not at debuginfo packages, that
 955  * have the PLT data stripped out (shdr_rel_plt.sh_type == SHT_NOBITS).
 956  */
 957 static int dso__synthesize_plt_symbols(struct  dso *dso, struct map *map,
 958                                        symbol_filter_t filter)
 959 {
 960         uint32_t nr_rel_entries, idx;
 961         GElf_Sym sym;
 962         u64 plt_offset;
 963         GElf_Shdr shdr_plt;
 964         struct symbol *f;
 965         GElf_Shdr shdr_rel_plt, shdr_dynsym;
 966         Elf_Data *reldata, *syms, *symstrs;
 967         Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym;
 968         size_t dynsym_idx;
 969         GElf_Ehdr ehdr;
 970         char sympltname[1024];
 971         Elf *elf;
 972         int nr = 0, symidx, fd, err = 0;
 973         char name[PATH_MAX];
 974
 975         snprintf(name, sizeof(name), "%s%s",
 976                  symbol_conf.symfs, dso->long_name);
 977         fd = open(name, O_RDONLY);
 978         if (fd < 0)
 979                 goto out;
 980
 981         elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
 982         if (elf == NULL)
 983                 goto out_close;
 984
 985         if (gelf_getehdr(elf, &ehdr) == NULL)
 986                 goto out_elf_end;
 987
 988         scn_dynsym = elf_section_by_name(elf, &ehdr, &shdr_dynsym,
 989                                          ".dynsym", &dynsym_idx);
 990         if (scn_dynsym == NULL)
 991                 goto out_elf_end;
 992
 993         scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
 994                                           ".rela.plt", NULL);
 995         if (scn_plt_rel == NULL) {
 996                 scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
 997                                                   ".rel.plt", NULL);
 998                 if (scn_plt_rel == NULL)
 999                         goto out_elf_end;
1000         }
1001
1002         err = -1;
1003
1004         if (shdr_rel_plt.sh_link != dynsym_idx)
1005                 goto out_elf_end;
1006
1007         if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL)
1008                 goto out_elf_end;
1009
1010         /*
1011          * Fetch the relocation section to find the idxes to the GOT
1012          * and the symbols in the .dynsym they refer to.
1013          */
1014         reldata = elf_getdata(scn_plt_rel, NULL);
1015         if (reldata == NULL)
1016                 goto out_elf_end;
1017
1018         syms = elf_getdata(scn_dynsym, NULL);
1019         if (syms == NULL)
1020                 goto out_elf_end;
1021
1022         scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link);
1023         if (scn_symstrs == NULL)
1024                 goto out_elf_end;
1025
1026         symstrs = elf_getdata(scn_symstrs, NULL);
1027         if (symstrs == NULL)
1028                 goto out_elf_end;
1029
1030         nr_rel_entries = shdr_rel_plt.sh_size / shdr_rel_plt.sh_entsize;
1031         plt_offset = shdr_plt.sh_offset;
1032
1033         if (shdr_rel_plt.sh_type == SHT_RELA) {
1034                 GElf_Rela pos_mem, *pos;
1035
1036                 elf_section__for_each_rela(reldata, pos, pos_mem, idx,
1037                                            nr_rel_entries) {
1038                         symidx = GELF_R_SYM(pos->r_info);
1039                         plt_offset += shdr_plt.sh_entsize;
1040                         gelf_getsym(syms, symidx, &sym);
1041                         snprintf(sympltname, sizeof(sympltname),
1042                                  "%s@plt", elf_sym__name(&sym, symstrs));
1043
1044                         f = symbol__new(plt_offset, shdr_plt.sh_entsize,
1045                                         STB_GLOBAL, sympltname);
1046                         if (!f)
1047                                 goto out_elf_end;
1048
1049                         if (filter && filter(map, f))
1050                                 symbol__delete(f);
1051                         else {
1052                                 symbols__insert(&dso->symbols[map->type], f);
1053                                 ++nr;
1054                         }
1055                 }
1056         } else if (shdr_rel_plt.sh_type == SHT_REL) {
1057                 GElf_Rel pos_mem, *pos;
1058                 elf_section__for_each_rel(reldata, pos, pos_mem, idx,
1059                                           nr_rel_entries) {
1060                         symidx = GELF_R_SYM(pos->r_info);
1061                         plt_offset += shdr_plt.sh_entsize;
1062                         gelf_getsym(syms, symidx, &sym);
1063                         snprintf(sympltname, sizeof(sympltname),
1064                                  "%s@plt", elf_sym__name(&sym, symstrs));
1065
1066                         f = symbol__new(plt_offset, shdr_plt.sh_entsize,
1067                                         STB_GLOBAL, sympltname);
1068                         if (!f)
1069                                 goto out_elf_end;
1070
1071                         if (filter && filter(map, f))
1072                                 symbol__delete(f);
1073                         else {
1074                                 symbols__insert(&dso->symbols[map->type], f);
1075                                 ++nr;
1076                         }
1077                 }
1078         }
1079
1080         err = 0;
1081 out_elf_end:
1082         elf_end(elf);
1083 out_close:
1084         close(fd);
1085
1086         if (err == 0)
1087                 return nr;
1088 out:
1089         pr_debug("%s: problems reading %s PLT info.\n",
1090                  __func__, dso->long_name);
1091         return 0;
1092 }
1093
1094 static bool elf_sym__is_a(GElf_Sym *sym, enum map_type type)
1095 {
1096         switch (type) {
1097         case MAP__FUNCTION:
1098                 return elf_sym__is_function(sym);
1099         case MAP__VARIABLE:
1100                 return elf_sym__is_object(sym);
1101         default:
1102                 return false;
1103         }
1104 }
1105
1106 static bool elf_sec__is_a(GElf_Shdr *shdr, Elf_Data *secstrs,
1107                           enum map_type type)
1108 {
1109         switch (type) {
1110         case MAP__FUNCTION:
1111                 return elf_sec__is_text(shdr, secstrs);
1112         case MAP__VARIABLE:
1113                 return elf_sec__is_data(shdr, secstrs);
1114         default:
1115                 return false;
1116         }
1117 }
1118
1119 static size_t elf_addr_to_index(Elf *elf, GElf_Addr addr)
1120 {
1121         Elf_Scn *sec = NULL;
1122         GElf_Shdr shdr;
1123         size_t cnt = 1;
1124
1125         while ((sec = elf_nextscn(elf, sec)) != NULL) {
1126                 gelf_getshdr(sec, &shdr);
1127
1128                 if ((addr >= shdr.sh_addr) &&
1129                     (addr < (shdr.sh_addr + shdr.sh_size)))
1130                         return cnt;
1131
1132                 ++cnt;
1133         }
1134
1135         return -1;
1136 }
1137
1138 static int dso__load_sym(struct dso *dso, struct map *map, const char *name,
1139                          int fd, symbol_filter_t filter, int kmodule,
1140                          int want_symtab)
1141 {
1142         struct kmap *kmap = dso->kernel ? map__kmap(map) : NULL;
1143         struct map *curr_map = map;
1144         struct dso *curr_dso = dso;
1145         Elf_Data *symstrs, *secstrs;
1146         uint32_t nr_syms;
1147         int err = -1;
1148         uint32_t idx;
1149         GElf_Ehdr ehdr;
1150         GElf_Shdr shdr, opdshdr;
1151         Elf_Data *syms, *opddata = NULL;
1152         GElf_Sym sym;
1153         Elf_Scn *sec, *sec_strndx, *opdsec;
1154         Elf *elf;
1155         int nr = 0;
1156         size_t opdidx = 0;
1157
1158         elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
1159         if (elf == NULL) {
1160                 pr_debug("%s: cannot read %s ELF file.\n", __func__, name);
1161                 goto out_close;
1162         }
1163
1164         if (gelf_getehdr(elf, &ehdr) == NULL) {
1165                 pr_debug("%s: cannot get elf header.\n", __func__);
1166                 goto out_elf_end;
1167         }
1168
1169         /* Always reject images with a mismatched build-id: */
1170         if (dso->has_build_id) {
1171                 u8 build_id[BUILD_ID_SIZE];
1172
1173                 if (elf_read_build_id(elf, build_id, BUILD_ID_SIZE) < 0)
1174                         goto out_elf_end;
1175
1176                 if (!dso__build_id_equal(dso, build_id))
1177                         goto out_elf_end;
1178         }
1179
1180         sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL);
1181         if (sec == NULL) {
1182                 if (want_symtab)
1183                         goto out_elf_end;
1184
1185                 sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL);
1186                 if (sec == NULL)
1187                         goto out_elf_end;
1188         }
1189
1190         opdsec = elf_section_by_name(elf, &ehdr, &opdshdr, ".opd", &opdidx);
1191         if (opdshdr.sh_type != SHT_PROGBITS)
1192                 opdsec = NULL;
1193         if (opdsec)
1194                 opddata = elf_rawdata(opdsec, NULL);
1195
1196         syms = elf_getdata(sec, NULL);
1197         if (syms == NULL)
1198                 goto out_elf_end;
1199
1200         sec = elf_getscn(elf, shdr.sh_link);
1201         if (sec == NULL)
1202                 goto out_elf_end;
1203
1204         symstrs = elf_getdata(sec, NULL);
1205         if (symstrs == NULL)
1206                 goto out_elf_end;
1207
1208         sec_strndx = elf_getscn(elf, ehdr.e_shstrndx);
1209         if (sec_strndx == NULL)
1210                 goto out_elf_end;
1211
1212         secstrs = elf_getdata(sec_strndx, NULL);
1213         if (secstrs == NULL)
1214                 goto out_elf_end;
1215
1216         nr_syms = shdr.sh_size / shdr.sh_entsize;
1217
1218         memset(&sym, 0, sizeof(sym));
1219         if (dso->kernel == DSO_TYPE_USER) {
1220                 dso->adjust_symbols = (ehdr.e_type == ET_EXEC ||
1221                                 elf_section_by_name(elf, &ehdr, &shdr,
1222                                                      ".gnu.prelink_undo",
1223                                                      NULL) != NULL);
1224         } else {
1225                 dso->adjust_symbols = 0;
1226         }
1227         elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) {
1228                 struct symbol *f;
1229                 const char *elf_name = elf_sym__name(&sym, symstrs);
1230                 char *demangled = NULL;
1231                 int is_label = elf_sym__is_label(&sym);
1232                 const char *section_name;
1233
1234                 if (kmap && kmap->ref_reloc_sym && kmap->ref_reloc_sym->name &&
1235                     strcmp(elf_name, kmap->ref_reloc_sym->name) == 0)
1236                         kmap->ref_reloc_sym->unrelocated_addr = sym.st_value;
1237
1238                 if (!is_label && !elf_sym__is_a(&sym, map->type))
1239                         continue;
1240
1241                 /* Reject ARM ELF "mapping symbols": these aren't unique and
1242                  * don't identify functions, so will confuse the profile
1243                  * output: */
1244                 if (ehdr.e_machine == EM_ARM) {
1245                         if (!strcmp(elf_name, "$a") ||
1246                             !strcmp(elf_name, "$d") ||
1247                             !strcmp(elf_name, "$t"))
1248                                 continue;
1249                 }
1250
1251                 if (opdsec && sym.st_shndx == opdidx) {
1252                         u32 offset = sym.st_value - opdshdr.sh_addr;
1253                         u64 *opd = opddata->d_buf + offset;
1254                         sym.st_value = *opd;
1255                         sym.st_shndx = elf_addr_to_index(elf, sym.st_value);
1256                 }
1257
1258                 sec = elf_getscn(elf, sym.st_shndx);
1259                 if (!sec)
1260                         goto out_elf_end;
1261
1262                 gelf_getshdr(sec, &shdr);
1263
1264                 if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type))
1265                         continue;
1266
1267                 section_name = elf_sec__name(&shdr, secstrs);
1268
1269                 /* On ARM, symbols for thumb functions have 1 added to
1270                  * the symbol address as a flag - remove it */
1271                 if ((ehdr.e_machine == EM_ARM) &&
1272                     (map->type == MAP__FUNCTION) &&
1273                     (sym.st_value & 1))
1274                         --sym.st_value;
1275
1276                 if (dso->kernel != DSO_TYPE_USER || kmodule) {
1277                         char dso_name[PATH_MAX];
1278
1279                         if (strcmp(section_name,
1280                                    (curr_dso->short_name +
1281                                     dso->short_name_len)) == 0)
1282                                 goto new_symbol;
1283
1284                         if (strcmp(section_name, ".text") == 0) {
1285                                 curr_map = map;
1286                                 curr_dso = dso;
1287                                 goto new_symbol;
1288                         }
1289
1290                         snprintf(dso_name, sizeof(dso_name),
1291                                  "%s%s", dso->short_name, section_name);
1292
1293                         curr_map = map_groups__find_by_name(kmap->kmaps, map->type, dso_name);
1294                         if (curr_map == NULL) {
1295                                 u64 start = sym.st_value;
1296
1297                                 if (kmodule)
1298                                         start += map->start + shdr.sh_offset;
1299
1300                                 curr_dso = dso__new(dso_name);
1301                                 if (curr_dso == NULL)
1302                                         goto out_elf_end;
1303                                 curr_dso->kernel = dso->kernel;
1304                                 curr_dso->long_name = dso->long_name;
1305                                 curr_dso->long_name_len = dso->long_name_len;
1306                                 curr_map = map__new2(start, curr_dso,
1307                                                      map->type);
1308                                 if (curr_map == NULL) {
1309                                         dso__delete(curr_dso);
1310                                         goto out_elf_end;
1311                                 }
1312                                 curr_map->map_ip = identity__map_ip;
1313                                 curr_map->unmap_ip = identity__map_ip;
1314                                 curr_dso->symtab_type = dso->symtab_type;
1315                                 map_groups__insert(kmap->kmaps, curr_map);
1316                                 dsos__add(&dso->node, curr_dso);
1317                                 dso__set_loaded(curr_dso, map->type);
1318                         } else
1319                                 curr_dso = curr_map->dso;
1320
1321                         goto new_symbol;
1322                 }
1323
1324                 if (curr_dso->adjust_symbols) {
1325                         pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " "
1326                                   "sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__,
1327                                   (u64)sym.st_value, (u64)shdr.sh_addr,
1328                                   (u64)shdr.sh_offset);
1329                         sym.st_value -= shdr.sh_addr - shdr.sh_offset;
1330                 }
1331                 /*
1332                  * We need to figure out if the object was created from C++ sources
1333                  * DWARF DW_compile_unit has this, but we don't always have access
1334                  * to it...
1335                  */
1336                 demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI);
1337                 if (demangled != NULL)
1338                         elf_name = demangled;
1339 new_symbol:
1340                 f = symbol__new(sym.st_value, sym.st_size,
1341                                 GELF_ST_BIND(sym.st_info), elf_name);
1342                 free(demangled);
1343                 if (!f)
1344                         goto out_elf_end;
1345
1346                 if (filter && filter(curr_map, f))
1347                         symbol__delete(f);
1348                 else {
1349                         symbols__insert(&curr_dso->symbols[curr_map->type], f);
1350                         nr++;
1351                 }
1352         }
1353
1354         /*
1355          * For misannotated, zeroed, ASM function sizes.
1356          */
1357         if (nr > 0) {
1358                 symbols__fixup_duplicate(&dso->symbols[map->type]);
1359                 symbols__fixup_end(&dso->symbols[map->type]);
1360                 if (kmap) {
1361                         /*
1362                          * We need to fixup this here too because we create new
1363                          * maps here, for things like vsyscall sections.
1364                          */
1365                         __map_groups__fixup_end(kmap->kmaps, map->type);
1366                 }
1367         }
1368         err = nr;
1369 out_elf_end:
1370         elf_end(elf);
1371 out_close:
1372         return err;
1373 }
1374
1375 static bool dso__build_id_equal(const struct dso *dso, u8 *build_id)
1376 {
1377         return memcmp(dso->build_id, build_id, sizeof(dso->build_id)) == 0;
1378 }
1379
1380 bool __dsos__read_build_ids(struct list_head *head, bool with_hits)
1381 {
1382         bool have_build_id = false;
1383         struct dso *pos;
1384
1385         list_for_each_entry(pos, head, node) {
1386                 if (with_hits && !pos->hit)
1387                         continue;
1388                 if (pos->has_build_id) {
1389                         have_build_id = true;
1390                         continue;
1391                 }
1392                 if (filename__read_build_id(pos->long_name, pos->build_id,
1393                                             sizeof(pos->build_id)) > 0) {
1394                         have_build_id     = true;
1395                         pos->has_build_id = true;
1396                 }
1397         }
1398
1399         return have_build_id;
1400 }
1401
1402 /*
1403  * Align offset to 4 bytes as needed for note name and descriptor data.
1404  */
1405 #define NOTE_ALIGN(n) (((n) + 3) & -4U)
1406
1407 static int elf_read_build_id(Elf *elf, void *bf, size_t size)
1408 {
1409         int err = -1;
1410         GElf_Ehdr ehdr;
1411         GElf_Shdr shdr;
1412         Elf_Data *data;
1413         Elf_Scn *sec;
1414         Elf_Kind ek;
1415         void *ptr;
1416
1417         if (size < BUILD_ID_SIZE)
1418                 goto out;
1419
1420         ek = elf_kind(elf);
1421         if (ek != ELF_K_ELF)
1422                 goto out;
1423
1424         if (gelf_getehdr(elf, &ehdr) == NULL) {
1425                 pr_err("%s: cannot get elf header.\n", __func__);
1426                 goto out;
1427         }
1428
1429         sec = elf_section_by_name(elf, &ehdr, &shdr,
1430                                   ".note.gnu.build-id", NULL);
1431         if (sec == NULL) {
1432                 sec = elf_section_by_name(elf, &ehdr, &shdr,
1433                                           ".notes", NULL);
1434                 if (sec == NULL)
1435                         goto out;
1436         }
1437
1438         data = elf_getdata(sec, NULL);
1439         if (data == NULL)
1440                 goto out;
1441
1442         ptr = data->d_buf;
1443         while (ptr < (data->d_buf + data->d_size)) {
1444                 GElf_Nhdr *nhdr = ptr;
1445                 size_t namesz = NOTE_ALIGN(nhdr->n_namesz),
1446                        descsz = NOTE_ALIGN(nhdr->n_descsz);
1447                 const char *name;
1448
1449                 ptr += sizeof(*nhdr);
1450                 name = ptr;
1451                 ptr += namesz;
1452                 if (nhdr->n_type == NT_GNU_BUILD_ID &&
1453                     nhdr->n_namesz == sizeof("GNU")) {
1454                         if (memcmp(name, "GNU", sizeof("GNU")) == 0) {
1455                                 size_t sz = min(size, descsz);
1456                                 memcpy(bf, ptr, sz);
1457                                 memset(bf + sz, 0, size - sz);
1458                                 err = descsz;
1459                                 break;
1460                         }
1461                 }
1462                 ptr += descsz;
1463         }
1464
1465 out:
1466         return err;
1467 }
1468
1469 int filename__read_build_id(const char *filename, void *bf, size_t size)
1470 {
1471         int fd, err = -1;
1472         Elf *elf;
1473
1474         if (size < BUILD_ID_SIZE)
1475                 goto out;
1476
1477         fd = open(filename, O_RDONLY);
1478         if (fd < 0)
1479                 goto out;
1480
1481         elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
1482         if (elf == NULL) {
1483                 pr_debug2("%s: cannot read %s ELF file.\n", __func__, filename);
1484                 goto out_close;
1485         }
1486
1487         err = elf_read_build_id(elf, bf, size);
1488
1489         elf_end(elf);
1490 out_close:
1491         close(fd);
1492 out:
1493         return err;
1494 }
1495
1496 int sysfs__read_build_id(const char *filename, void *build_id, size_t size)
1497 {
1498         int fd, err = -1;
1499
1500         if (size < BUILD_ID_SIZE)
1501                 goto out;
1502
1503         fd = open(filename, O_RDONLY);
1504         if (fd < 0)
1505                 goto out;
1506
1507         while (1) {
1508                 char bf[BUFSIZ];
1509                 GElf_Nhdr nhdr;
1510                 size_t namesz, descsz;
1511
1512                 if (read(fd, &nhdr, sizeof(nhdr)) != sizeof(nhdr))
1513                         break;
1514
1515                 namesz = NOTE_ALIGN(nhdr.n_namesz);
1516                 descsz = NOTE_ALIGN(nhdr.n_descsz);
1517                 if (nhdr.n_type == NT_GNU_BUILD_ID &&
1518                     nhdr.n_namesz == sizeof("GNU")) {
1519                         if (read(fd, bf, namesz) != (ssize_t)namesz)
1520                                 break;
1521                         if (memcmp(bf, "GNU", sizeof("GNU")) == 0) {
1522                                 size_t sz = min(descsz, size);
1523                                 if (read(fd, build_id, sz) == (ssize_t)sz) {
1524                                         memset(build_id + sz, 0, size - sz);
1525                                         err = 0;
1526                                         break;
1527                                 }
1528                         } else if (read(fd, bf, descsz) != (ssize_t)descsz)
1529                                 break;
1530                 } else {
1531                         int n = namesz + descsz;
1532                         if (read(fd, bf, n) != n)
1533                                 break;
1534                 }
1535         }
1536         close(fd);
1537 out:
1538         return err;
1539 }
1540
1541 char dso__symtab_origin(const struct dso *dso)
1542 {
1543         static const char origin[] = {
1544                 [SYMTAB__KALLSYMS]            = 'k',
1545                 [SYMTAB__JAVA_JIT]            = 'j',
1546                 [SYMTAB__BUILD_ID_CACHE]      = 'B',
1547                 [SYMTAB__FEDORA_DEBUGINFO]    = 'f',
1548                 [SYMTAB__UBUNTU_DEBUGINFO]    = 'u',
1549                 [SYMTAB__BUILDID_DEBUGINFO]   = 'b',
1550                 [SYMTAB__SYSTEM_PATH_DSO]     = 'd',
1551                 [SYMTAB__SYSTEM_PATH_KMODULE] = 'K',
1552                 [SYMTAB__GUEST_KALLSYMS]      =  'g',
1553                 [SYMTAB__GUEST_KMODULE]       =  'G',
1554         };
1555
1556         if (dso == NULL || dso->symtab_type == SYMTAB__NOT_FOUND)
1557                 return '!';
1558         return origin[dso->symtab_type];
1559 }
1560
1561 int dso__load(struct dso *dso, struct map *map, symbol_filter_t filter)
1562 {
1563         int size = PATH_MAX;
1564         char *name;
1565         int ret = -1;
1566         int fd;
1567         struct machine *machine;
1568         const char *root_dir;
1569         int want_symtab;
1570
1571         dso__set_loaded(dso, map->type);
1572
1573         if (dso->kernel == DSO_TYPE_KERNEL)
1574                 return dso__load_kernel_sym(dso, map, filter);
1575         else if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
1576                 return dso__load_guest_kernel_sym(dso, map, filter);
1577
1578         if (map->groups && map->groups->machine)
1579                 machine = map->groups->machine;
1580         else
1581                 machine = NULL;
1582
1583         name = malloc(size);
1584         if (!name)
1585                 return -1;
1586
1587         dso->adjust_symbols = 0;
1588
1589         if (strncmp(dso->name, "/tmp/perf-", 10) == 0) {
1590                 struct stat st;
1591
1592                 if (lstat(dso->name, &st) < 0)
1593                         return -1;
1594
1595                 if (st.st_uid && (st.st_uid != geteuid())) {
1596                         pr_warning("File %s not owned by current user or root, "
1597                                 "ignoring it.\n", dso->name);
1598                         return -1;
1599                 }
1600
1601                 ret = dso__load_perf_map(dso, map, filter);
1602                 dso->symtab_type = ret > 0 ? SYMTAB__JAVA_JIT :
1603                                               SYMTAB__NOT_FOUND;
1604                 return ret;
1605         }
1606
1607         /* Iterate over candidate debug images.
1608          * On the first pass, only load images if they have a full symtab.
1609          * Failing that, do a second pass where we accept .dynsym also
1610          */
1611         want_symtab = 1;
1612 restart:
1613         for (dso->symtab_type = SYMTAB__BUILD_ID_CACHE;
1614              dso->symtab_type != SYMTAB__NOT_FOUND;
1615              dso->symtab_type++) {
1616                 switch (dso->symtab_type) {
1617                 case SYMTAB__BUILD_ID_CACHE:
1618                         /* skip the locally configured cache if a symfs is given */
1619                         if (symbol_conf.symfs[0] ||
1620                             (dso__build_id_filename(dso, name, size) == NULL)) {
1621                                 continue;
1622                         }
1623                         break;
1624                 case SYMTAB__FEDORA_DEBUGINFO:
1625                         snprintf(name, size, "%s/usr/lib/debug%s.debug",
1626                                  symbol_conf.symfs, dso->long_name);
1627                         break;
1628                 case SYMTAB__UBUNTU_DEBUGINFO:
1629                         snprintf(name, size, "%s/usr/lib/debug%s",
1630                                  symbol_conf.symfs, dso->long_name);
1631                         break;
1632                 case SYMTAB__BUILDID_DEBUGINFO: {
1633                         char build_id_hex[BUILD_ID_SIZE * 2 + 1];
1634
1635                         if (!dso->has_build_id)
1636                                 continue;
1637
1638                         build_id__sprintf(dso->build_id,
1639                                           sizeof(dso->build_id),
1640                                           build_id_hex);
1641                         snprintf(name, size,
1642                                  "%s/usr/lib/debug/.build-id/%.2s/%s.debug",
1643                                  symbol_conf.symfs, build_id_hex, build_id_hex + 2);
1644                         }
1645                         break;
1646                 case SYMTAB__SYSTEM_PATH_DSO:
1647                         snprintf(name, size, "%s%s",
1648                              symbol_conf.symfs, dso->long_name);
1649                         break;
1650                 case SYMTAB__GUEST_KMODULE:
1651                         if (map->groups && machine)
1652                                 root_dir = machine->root_dir;
1653                         else
1654                                 root_dir = "";
1655                         snprintf(name, size, "%s%s%s", symbol_conf.symfs,
1656                                  root_dir, dso->long_name);
1657                         break;
1658
1659                 case SYMTAB__SYSTEM_PATH_KMODULE:
1660                         snprintf(name, size, "%s%s", symbol_conf.symfs,
1661                                  dso->long_name);
1662                         break;
1663                 default:;
1664                 }
1665
1666                 /* Name is now the name of the next image to try */
1667                 fd = open(name, O_RDONLY);
1668                 if (fd < 0)
1669                         continue;
1670
1671                 ret = dso__load_sym(dso, map, name, fd, filter, 0,
1672                                     want_symtab);
1673                 close(fd);
1674
1675                 /*
1676                  * Some people seem to have debuginfo files _WITHOUT_ debug
1677                  * info!?!?
1678                  */
1679                 if (!ret)
1680                         continue;
1681
1682                 if (ret > 0) {
1683                         int nr_plt = dso__synthesize_plt_symbols(dso, map,
1684                                                                  filter);
1685                         if (nr_plt > 0)
1686                                 ret += nr_plt;
1687                         break;
1688                 }
1689         }
1690
1691         /*
1692          * If we wanted a full symtab but no image had one,
1693          * relax our requirements and repeat the search.
1694          */
1695         if (ret <= 0 && want_symtab) {
1696                 want_symtab = 0;
1697                 goto restart;
1698         }
1699
1700         free(name);
1701         if (ret < 0 && strstr(dso->name, " (deleted)") != NULL)
1702                 return 0;
1703         return ret;
1704 }
1705
1706 struct map *map_groups__find_by_name(struct map_groups *mg,
1707                                      enum map_type type, const char *name)
1708 {
1709         struct rb_node *nd;
1710
1711         for (nd = rb_first(&mg->maps[type]); nd; nd = rb_next(nd)) {
1712                 struct map *map = rb_entry(nd, struct map, rb_node);
1713
1714                 if (map->dso && strcmp(map->dso->short_name, name) == 0)
1715                         return map;
1716         }
1717
1718         return NULL;
1719 }
1720
1721 static int dso__kernel_module_get_build_id(struct dso *dso,
1722                                            const char *root_dir)
1723 {
1724         char filename[PATH_MAX];
1725         /*
1726          * kernel module short names are of the form "[module]" and
1727          * we need just "module" here.
1728          */
1729         const char *name = dso->short_name + 1;
1730
1731         snprintf(filename, sizeof(filename),
1732                  "%s/sys/module/%.*s/notes/.note.gnu.build-id",
1733                  root_dir, (int)strlen(name) - 1, name);
1734
1735         if (sysfs__read_build_id(filename, dso->build_id,
1736                                  sizeof(dso->build_id)) == 0)
1737                 dso->has_build_id = true;
1738
1739         return 0;
1740 }
1741
1742 static int map_groups__set_modules_path_dir(struct map_groups *mg,
1743                                 const char *dir_name)
1744 {
1745         struct dirent *dent;
1746         DIR *dir = opendir(dir_name);
1747         int ret = 0;
1748
1749         if (!dir) {
1750                 pr_debug("%s: cannot open %s dir\n", __func__, dir_name);
1751                 return -1;
1752         }
1753
1754         while ((dent = readdir(dir)) != NULL) {
1755                 char path[PATH_MAX];
1756                 struct stat st;
1757
1758                 /*sshfs might return bad dent->d_type, so we have to stat*/
1759                 sprintf(path, "%s/%s", dir_name, dent->d_name);
1760                 if (stat(path, &st))
1761                         continue;
1762
1763                 if (S_ISDIR(st.st_mode)) {
1764                         if (!strcmp(dent->d_name, ".") ||
1765                             !strcmp(dent->d_name, ".."))
1766                                 continue;
1767
1768                         snprintf(path, sizeof(path), "%s/%s",
1769                                  dir_name, dent->d_name);
1770                         ret = map_groups__set_modules_path_dir(mg, path);
1771                         if (ret < 0)
1772                                 goto out;
1773                 } else {
1774                         char *dot = strrchr(dent->d_name, '.'),
1775                              dso_name[PATH_MAX];
1776                         struct map *map;
1777                         char *long_name;
1778
1779                         if (dot == NULL || strcmp(dot, ".ko"))
1780                                 continue;
1781                         snprintf(dso_name, sizeof(dso_name), "[%.*s]",
1782                                  (int)(dot - dent->d_name), dent->d_name);
1783
1784                         strxfrchar(dso_name, '-', '_');
1785                         map = map_groups__find_by_name(mg, MAP__FUNCTION,
1786                                                        dso_name);
1787                         if (map == NULL)
1788                                 continue;
1789
1790                         snprintf(path, sizeof(path), "%s/%s",
1791                                  dir_name, dent->d_name);
1792
1793                         long_name = strdup(path);
1794                         if (long_name == NULL) {
1795                                 ret = -1;
1796                                 goto out;
1797                         }
1798                         dso__set_long_name(map->dso, long_name);
1799                         map->dso->lname_alloc = 1;
1800                         dso__kernel_module_get_build_id(map->dso, "");
1801                 }
1802         }
1803
1804 out:
1805         closedir(dir);
1806         return ret;
1807 }
1808
1809 static char *get_kernel_version(const char *root_dir)
1810 {
1811         char version[PATH_MAX];
1812         FILE *file;
1813         char *name, *tmp;
1814         const char *prefix = "Linux version ";
1815
1816         sprintf(version, "%s/proc/version", root_dir);
1817         file = fopen(version, "r");
1818         if (!file)
1819                 return NULL;
1820
1821         version[0] = '\0';
1822         tmp = fgets(version, sizeof(version), file);
1823         fclose(file);
1824
1825         name = strstr(version, prefix);
1826         if (!name)
1827                 return NULL;
1828         name += strlen(prefix);
1829         tmp = strchr(name, ' ');
1830         if (tmp)
1831                 *tmp = '\0';
1832
1833         return strdup(name);
1834 }
1835
1836 static int machine__set_modules_path(struct machine *machine)
1837 {
1838         char *version;
1839         char modules_path[PATH_MAX];
1840
1841         version = get_kernel_version(machine->root_dir);
1842         if (!version)
1843                 return -1;
1844
1845         snprintf(modules_path, sizeof(modules_path), "%s/lib/modules/%s/kernel",
1846                  machine->root_dir, version);
1847         free(version);
1848
1849         return map_groups__set_modules_path_dir(&machine->kmaps, modules_path);
1850 }
1851
1852 /*
1853  * Constructor variant for modules (where we know from /proc/modules where
1854  * they are loaded) and for vmlinux, where only after we load all the
1855  * symbols we'll know where it starts and ends.
1856  */
1857 static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
1858 {
1859         struct map *map = calloc(1, (sizeof(*map) +
1860                                      (dso->kernel ? sizeof(struct kmap) : 0)));
1861         if (map != NULL) {
1862                 /*
1863                  * ->end will be filled after we load all the symbols
1864                  */
1865                 map__init(map, type, start, 0, 0, dso);
1866         }
1867
1868         return map;
1869 }
1870
1871 struct map *machine__new_module(struct machine *machine, u64 start,
1872                                 const char *filename)
1873 {
1874         struct map *map;
1875         struct dso *dso = __dsos__findnew(&machine->kernel_dsos, filename);
1876
1877         if (dso == NULL)
1878                 return NULL;
1879
1880         map = map__new2(start, dso, MAP__FUNCTION);
1881         if (map == NULL)
1882                 return NULL;
1883
1884         if (machine__is_host(machine))
1885                 dso->symtab_type = SYMTAB__SYSTEM_PATH_KMODULE;
1886         else
1887                 dso->symtab_type = SYMTAB__GUEST_KMODULE;
1888         map_groups__insert(&machine->kmaps, map);
1889         return map;
1890 }
1891
1892 static int machine__create_modules(struct machine *machine)
1893 {
1894         char *line = NULL;
1895         size_t n;
1896         FILE *file;
1897         struct map *map;
1898         const char *modules;
1899         char path[PATH_MAX];
1900
1901         if (machine__is_default_guest(machine))
1902                 modules = symbol_conf.default_guest_modules;
1903         else {
1904                 sprintf(path, "%s/proc/modules", machine->root_dir);
1905                 modules = path;
1906         }
1907
1908         if (symbol__restricted_filename(path, "/proc/modules"))
1909                 return -1;
1910
1911         file = fopen(modules, "r");
1912         if (file == NULL)
1913                 return -1;
1914
1915         while (!feof(file)) {
1916                 char name[PATH_MAX];
1917                 u64 start;
1918                 char *sep;
1919                 int line_len;
1920
1921                 line_len = getline(&line, &n, file);
1922                 if (line_len < 0)
1923                         break;
1924
1925                 if (!line)
1926                         goto out_failure;
1927
1928                 line[--line_len] = '\0'; /* \n */
1929
1930                 sep = strrchr(line, 'x');
1931                 if (sep == NULL)
1932                         continue;
1933
1934                 hex2u64(sep + 1, &start);
1935
1936                 sep = strchr(line, ' ');
1937                 if (sep == NULL)
1938                         continue;
1939
1940                 *sep = '\0';
1941
1942                 snprintf(name, sizeof(name), "[%s]", line);
1943                 map = machine__new_module(machine, start, name);
1944                 if (map == NULL)
1945                         goto out_delete_line;
1946                 dso__kernel_module_get_build_id(map->dso, machine->root_dir);
1947         }
1948
1949         free(line);
1950         fclose(file);
1951
1952         return machine__set_modules_path(machine);
1953
1954 out_delete_line:
1955         free(line);
1956 out_failure:
1957         return -1;
1958 }
1959
1960 int dso__load_vmlinux(struct dso *dso, struct map *map,
1961                       const char *vmlinux, symbol_filter_t filter)
1962 {
1963         int err = -1, fd;
1964         char symfs_vmlinux[PATH_MAX];
1965
1966         snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s%s",
1967                  symbol_conf.symfs, vmlinux);
1968         fd = open(symfs_vmlinux, O_RDONLY);
1969         if (fd < 0)
1970                 return -1;
1971
1972         dso__set_long_name(dso, (char *)vmlinux);
1973         dso__set_loaded(dso, map->type);
1974         err = dso__load_sym(dso, map, symfs_vmlinux, fd, filter, 0, 0);
1975         close(fd);
1976
1977         if (err > 0)
1978                 pr_debug("Using %s for symbols\n", symfs_vmlinux);
1979
1980         return err;
1981 }
1982
1983 int dso__load_vmlinux_path(struct dso *dso, struct map *map,
1984                            symbol_filter_t filter)
1985 {
1986         int i, err = 0;
1987         char *filename;
1988
1989         pr_debug("Looking at the vmlinux_path (%d entries long)\n",
1990                  vmlinux_path__nr_entries + 1);
1991
1992         filename = dso__build_id_filename(dso, NULL, 0);
1993         if (filename != NULL) {
1994                 err = dso__load_vmlinux(dso, map, filename, filter);
1995                 if (err > 0) {
1996                         dso__set_long_name(dso, filename);
1997                         goto out;
1998                 }
1999                 free(filename);
2000         }
2001
2002         for (i = 0; i < vmlinux_path__nr_entries; ++i) {
2003                 err = dso__load_vmlinux(dso, map, vmlinux_path[i], filter);
2004                 if (err > 0) {
2005                         dso__set_long_name(dso, strdup(vmlinux_path[i]));
2006                         break;
2007                 }
2008         }
2009 out:
2010         return err;
2011 }
2012
2013 static int dso__load_kernel_sym(struct dso *dso, struct map *map,
2014                                 symbol_filter_t filter)
2015 {
2016         int err;
2017         const char *kallsyms_filename = NULL;
2018         char *kallsyms_allocated_filename = NULL;
2019         /*
2020          * Step 1: if the user specified a kallsyms or vmlinux filename, use
2021          * it and only it, reporting errors to the user if it cannot be used.
2022          *
2023          * For instance, try to analyse an ARM perf.data file _without_ a
2024          * build-id, or if the user specifies the wrong path to the right
2025          * vmlinux file, obviously we can't fallback to another vmlinux (a
2026          * x86_86 one, on the machine where analysis is being performed, say),
2027          * or worse, /proc/kallsyms.
2028          *
2029          * If the specified file _has_ a build-id and there is a build-id
2030          * section in the perf.data file, we will still do the expected
2031          * validation in dso__load_vmlinux and will bail out if they don't
2032          * match.
2033          */
2034         if (symbol_conf.kallsyms_name != NULL) {
2035                 kallsyms_filename = symbol_conf.kallsyms_name;
2036                 goto do_kallsyms;
2037         }
2038
2039         if (symbol_conf.vmlinux_name != NULL) {
2040                 err = dso__load_vmlinux(dso, map,
2041                                         symbol_conf.vmlinux_name, filter);
2042                 if (err > 0) {
2043                         dso__set_long_name(dso,
2044                                            strdup(symbol_conf.vmlinux_name));
2045                         goto out_fixup;
2046                 }
2047                 return err;
2048         }
2049
2050         if (vmlinux_path != NULL) {
2051                 err = dso__load_vmlinux_path(dso, map, filter);
2052                 if (err > 0)
2053                         goto out_fixup;
2054         }
2055
2056         /* do not try local files if a symfs was given */
2057         if (symbol_conf.symfs[0] != 0)
2058                 return -1;
2059
2060         /*
2061          * Say the kernel DSO was created when processing the build-id header table,
2062          * we have a build-id, so check if it is the same as the running kernel,
2063          * using it if it is.
2064          */
2065         if (dso->has_build_id) {
2066                 u8 kallsyms_build_id[BUILD_ID_SIZE];
2067                 char sbuild_id[BUILD_ID_SIZE * 2 + 1];
2068
2069                 if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id,
2070                                          sizeof(kallsyms_build_id)) == 0) {
2071                         if (dso__build_id_equal(dso, kallsyms_build_id)) {
2072                                 kallsyms_filename = "/proc/kallsyms";
2073                                 goto do_kallsyms;
2074                         }
2075                 }
2076                 /*
2077                  * Now look if we have it on the build-id cache in
2078                  * $HOME/.debug/[kernel.kallsyms].
2079                  */
2080                 build_id__sprintf(dso->build_id, sizeof(dso->build_id),
2081                                   sbuild_id);
2082
2083                 if (asprintf(&kallsyms_allocated_filename,
2084                              "%s/.debug/[kernel.kallsyms]/%s",
2085                              getenv("HOME"), sbuild_id) == -1) {
2086                         pr_err("Not enough memory for kallsyms file lookup\n");
2087                         return -1;
2088                 }
2089
2090                 kallsyms_filename = kallsyms_allocated_filename;
2091
2092                 if (access(kallsyms_filename, F_OK)) {
2093                         pr_err("No kallsyms or vmlinux with build-id %s "
2094                                "was found\n", sbuild_id);
2095                         free(kallsyms_allocated_filename);
2096                         return -1;
2097                 }
2098         } else {
2099                 /*
2100                  * Last resort, if we don't have a build-id and couldn't find
2101                  * any vmlinux file, try the running kernel kallsyms table.
2102                  */
2103                 kallsyms_filename = "/proc/kallsyms";
2104         }
2105
2106 do_kallsyms:
2107         err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
2108         if (err > 0)
2109                 pr_debug("Using %s for symbols\n", kallsyms_filename);
2110         free(kallsyms_allocated_filename);
2111
2112         if (err > 0) {
2113 out_fixup:
2114                 if (kallsyms_filename != NULL)
2115                         dso__set_long_name(dso, strdup("[kernel.kallsyms]"));
2116                 map__fixup_start(map);
2117                 map__fixup_end(map);
2118         }
2119
2120         return err;
2121 }
2122
2123 static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
2124                                       symbol_filter_t filter)
2125 {
2126         int err;
2127         const char *kallsyms_filename = NULL;
2128         struct machine *machine;
2129         char path[PATH_MAX];
2130
2131         if (!map->groups) {
2132                 pr_debug("Guest kernel map hasn't the point to groups\n");
2133                 return -1;
2134         }
2135         machine = map->groups->machine;
2136
2137         if (machine__is_default_guest(machine)) {
2138                 /*
2139                  * if the user specified a vmlinux filename, use it and only
2140                  * it, reporting errors to the user if it cannot be used.
2141                  * Or use file guest_kallsyms inputted by user on commandline
2142                  */
2143                 if (symbol_conf.default_guest_vmlinux_name != NULL) {
2144                         err = dso__load_vmlinux(dso, map,
2145                                 symbol_conf.default_guest_vmlinux_name, filter);
2146                         goto out_try_fixup;
2147                 }
2148
2149                 kallsyms_filename = symbol_conf.default_guest_kallsyms;
2150                 if (!kallsyms_filename)
2151                         return -1;
2152         } else {
2153                 sprintf(path, "%s/proc/kallsyms", machine->root_dir);
2154                 kallsyms_filename = path;
2155         }
2156
2157         err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
2158         if (err > 0)
2159                 pr_debug("Using %s for symbols\n", kallsyms_filename);
2160
2161 out_try_fixup:
2162         if (err > 0) {
2163                 if (kallsyms_filename != NULL) {
2164                         machine__mmap_name(machine, path, sizeof(path));
2165                         dso__set_long_name(dso, strdup(path));
2166                 }
2167                 map__fixup_start(map);
2168                 map__fixup_end(map);
2169         }
2170
2171         return err;
2172 }
2173
2174 static void dsos__add(struct list_head *head, struct dso *dso)
2175 {
2176         list_add_tail(&dso->node, head);
2177 }
2178
2179 static struct dso *dsos__find(struct list_head *head, const char *name)
2180 {
2181         struct dso *pos;
2182
2183         list_for_each_entry(pos, head, node)
2184                 if (strcmp(pos->long_name, name) == 0)
2185                         return pos;
2186         return NULL;
2187 }
2188
2189 struct dso *__dsos__findnew(struct list_head *head, const char *name)
2190 {
2191         struct dso *dso = dsos__find(head, name);
2192
2193         if (!dso) {
2194                 dso = dso__new(name);
2195                 if (dso != NULL) {
2196                         dsos__add(head, dso);
2197                         dso__set_basename(dso);
2198                 }
2199         }
2200
2201         return dso;
2202 }
2203
2204 size_t __dsos__fprintf(struct list_head *head, FILE *fp)
2205 {
2206         struct dso *pos;
2207         size_t ret = 0;
2208
2209         list_for_each_entry(pos, head, node) {
2210                 int i;
2211                 for (i = 0; i < MAP__NR_TYPES; ++i)
2212                         ret += dso__fprintf(pos, i, fp);
2213         }
2214
2215         return ret;
2216 }
2217
2218 size_t machines__fprintf_dsos(struct rb_root *machines, FILE *fp)
2219 {
2220         struct rb_node *nd;
2221         size_t ret = 0;
2222
2223         for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
2224                 struct machine *pos = rb_entry(nd, struct machine, rb_node);
2225                 ret += __dsos__fprintf(&pos->kernel_dsos, fp);
2226                 ret += __dsos__fprintf(&pos->user_dsos, fp);
2227         }
2228
2229         return ret;
2230 }
2231
2232 static size_t __dsos__fprintf_buildid(struct list_head *head, FILE *fp,
2233                                       bool with_hits)
2234 {
2235         struct dso *pos;
2236         size_t ret = 0;
2237
2238         list_for_each_entry(pos, head, node) {
2239                 if (with_hits && !pos->hit)
2240                         continue;
2241                 ret += dso__fprintf_buildid(pos, fp);
2242                 ret += fprintf(fp, " %s\n", pos->long_name);
2243         }
2244         return ret;
2245 }
2246
2247 size_t machine__fprintf_dsos_buildid(struct machine *machine, FILE *fp,
2248                                      bool with_hits)
2249 {
2250         return __dsos__fprintf_buildid(&machine->kernel_dsos, fp, with_hits) +
2251                __dsos__fprintf_buildid(&machine->user_dsos, fp, with_hits);
2252 }
2253
2254 size_t machines__fprintf_dsos_buildid(struct rb_root *machines,
2255                                       FILE *fp, bool with_hits)
2256 {
2257         struct rb_node *nd;
2258         size_t ret = 0;
2259
2260         for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
2261                 struct machine *pos = rb_entry(nd, struct machine, rb_node);
2262                 ret += machine__fprintf_dsos_buildid(pos, fp, with_hits);
2263         }
2264         return ret;
2265 }
2266
2267 static struct dso*
2268 dso__kernel_findnew(struct machine *machine, const char *name,
2269                     const char *short_name, int dso_type)
2270 {
2271         /*
2272          * The kernel dso could be created by build_id processing.
2273          */
2274         struct dso *dso = __dsos__findnew(&machine->kernel_dsos, name);
2275
2276         /*
2277          * We need to run this in all cases, since during the build_id
2278          * processing we had no idea this was the kernel dso.
2279          */
2280         if (dso != NULL) {
2281                 dso__set_short_name(dso, short_name);
2282                 dso->kernel = dso_type;
2283         }
2284
2285         return dso;
2286 }
2287
2288 void dso__read_running_kernel_build_id(struct dso *dso, struct machine *machine)
2289 {
2290         char path[PATH_MAX];
2291
2292         if (machine__is_default_guest(machine))
2293                 return;
2294         sprintf(path, "%s/sys/kernel/notes", machine->root_dir);
2295         if (sysfs__read_build_id(path, dso->build_id,
2296                                  sizeof(dso->build_id)) == 0)
2297                 dso->has_build_id = true;
2298 }
2299
2300 static struct dso *machine__get_kernel(struct machine *machine)
2301 {
2302         const char *vmlinux_name = NULL;
2303         struct dso *kernel;
2304
2305         if (machine__is_host(machine)) {
2306                 vmlinux_name = symbol_conf.vmlinux_name;
2307                 if (!vmlinux_name)
2308                         vmlinux_name = "[kernel.kallsyms]";
2309
2310                 kernel = dso__kernel_findnew(machine, vmlinux_name,
2311                                              "[kernel]",
2312                                              DSO_TYPE_KERNEL);
2313         } else {
2314                 char bf[PATH_MAX];
2315
2316                 if (machine__is_default_guest(machine))
2317                         vmlinux_name = symbol_conf.default_guest_vmlinux_name;
2318                 if (!vmlinux_name)
2319                         vmlinux_name = machine__mmap_name(machine, bf,
2320                                                           sizeof(bf));
2321
2322                 kernel = dso__kernel_findnew(machine, vmlinux_name,
2323                                              "[guest.kernel]",
2324                                              DSO_TYPE_GUEST_KERNEL);
2325         }
2326
2327         if (kernel != NULL && (!kernel->has_build_id))
2328                 dso__read_running_kernel_build_id(kernel, machine);
2329
2330         return kernel;
2331 }
2332
2333 struct process_args {
2334         u64 start;
2335 };
2336
2337 static int symbol__in_kernel(void *arg, const char *name,
2338                              char type __used, u64 start, u64 end __used)
2339 {
2340         struct process_args *args = arg;
2341
2342         if (strchr(name, '['))
2343                 return 0;
2344
2345         args->start = start;
2346         return 1;
2347 }
2348
2349 /* Figure out the start address of kernel map from /proc/kallsyms */
2350 static u64 machine__get_kernel_start_addr(struct machine *machine)
2351 {
2352         const char *filename;
2353         char path[PATH_MAX];
2354         struct process_args args;
2355
2356         if (machine__is_host(machine)) {
2357                 filename = "/proc/kallsyms";
2358         } else {
2359                 if (machine__is_default_guest(machine))
2360                         filename = (char *)symbol_conf.default_guest_kallsyms;
2361                 else {
2362                         sprintf(path, "%s/proc/kallsyms", machine->root_dir);
2363                         filename = path;
2364                 }
2365         }
2366
2367         if (symbol__restricted_filename(filename, "/proc/kallsyms"))
2368                 return 0;
2369
2370         if (kallsyms__parse(filename, &args, symbol__in_kernel) <= 0)
2371                 return 0;
2372
2373         return args.start;
2374 }
2375
2376 int __machine__create_kernel_maps(struct machine *machine, struct dso *kernel)
2377 {
2378         enum map_type type;
2379         u64 start = machine__get_kernel_start_addr(machine);
2380
2381         for (type = 0; type < MAP__NR_TYPES; ++type) {
2382                 struct kmap *kmap;
2383
2384                 machine->vmlinux_maps[type] = map__new2(start, kernel, type);
2385                 if (machine->vmlinux_maps[type] == NULL)
2386                         return -1;
2387
2388                 machine->vmlinux_maps[type]->map_ip =
2389                         machine->vmlinux_maps[type]->unmap_ip =
2390                                 identity__map_ip;
2391                 kmap = map__kmap(machine->vmlinux_maps[type]);
2392                 kmap->kmaps = &machine->kmaps;
2393                 map_groups__insert(&machine->kmaps,
2394                                    machine->vmlinux_maps[type]);
2395         }
2396
2397         return 0;
2398 }
2399
2400 void machine__destroy_kernel_maps(struct machine *machine)
2401 {
2402         enum map_type type;
2403
2404         for (type = 0; type < MAP__NR_TYPES; ++type) {
2405                 struct kmap *kmap;
2406
2407                 if (machine->vmlinux_maps[type] == NULL)
2408                         continue;
2409
2410                 kmap = map__kmap(machine->vmlinux_maps[type]);
2411                 map_groups__remove(&machine->kmaps,
2412                                    machine->vmlinux_maps[type]);
2413                 if (kmap->ref_reloc_sym) {
2414                         /*
2415                          * ref_reloc_sym is shared among all maps, so free just
2416                          * on one of them.
2417                          */
2418                         if (type == MAP__FUNCTION) {
2419                                 free((char *)kmap->ref_reloc_sym->name);
2420                                 kmap->ref_reloc_sym->name = NULL;
2421                                 free(kmap->ref_reloc_sym);
2422                         }
2423                         kmap->ref_reloc_sym = NULL;
2424                 }
2425
2426                 map__delete(machine->vmlinux_maps[type]);
2427                 machine->vmlinux_maps[type] = NULL;
2428         }
2429 }
2430
2431 int machine__create_kernel_maps(struct machine *machine)
2432 {
2433         struct dso *kernel = machine__get_kernel(machine);
2434
2435         if (kernel == NULL ||
2436             __machine__create_kernel_maps(machine, kernel) < 0)
2437                 return -1;
2438
2439         if (symbol_conf.use_modules && machine__create_modules(machine) < 0)
2440                 pr_debug("Problems creating module maps, continuing anyway...\n");
2441         /*
2442          * Now that we have all the maps created, just set the ->end of them:
2443          */
2444         map_groups__fixup_end(&machine->kmaps);
2445         return 0;
2446 }
2447
2448 static void vmlinux_path__exit(void)
2449 {
2450         while (--vmlinux_path__nr_entries >= 0) {
2451                 free(vmlinux_path[vmlinux_path__nr_entries]);
2452                 vmlinux_path[vmlinux_path__nr_entries] = NULL;
2453         }
2454
2455         free(vmlinux_path);
2456         vmlinux_path = NULL;
2457 }
2458
2459 static int vmlinux_path__init(void)
2460 {
2461         struct utsname uts;
2462         char bf[PATH_MAX];
2463
2464         vmlinux_path = malloc(sizeof(char *) * 5);
2465         if (vmlinux_path == NULL)
2466                 return -1;
2467
2468         vmlinux_path[vmlinux_path__nr_entries] = strdup("vmlinux");
2469         if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
2470                 goto out_fail;
2471         ++vmlinux_path__nr_entries;
2472         vmlinux_path[vmlinux_path__nr_entries] = strdup("/boot/vmlinux");
2473         if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
2474                 goto out_fail;
2475         ++vmlinux_path__nr_entries;
2476
2477         /* only try running kernel version if no symfs was given */
2478         if (symbol_conf.symfs[0] != 0)
2479                 return 0;
2480
2481         if (uname(&uts) < 0)
2482                 return -1;
2483
2484         snprintf(bf, sizeof(bf), "/boot/vmlinux-%s", uts.release);
2485         vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
2486         if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
2487                 goto out_fail;
2488         ++vmlinux_path__nr_entries;
2489         snprintf(bf, sizeof(bf), "/lib/modules/%s/build/vmlinux", uts.release);
2490         vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
2491         if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
2492                 goto out_fail;
2493         ++vmlinux_path__nr_entries;
2494         snprintf(bf, sizeof(bf), "/usr/lib/debug/lib/modules/%s/vmlinux",
2495                  uts.release);
2496         vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
2497         if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
2498                 goto out_fail;
2499         ++vmlinux_path__nr_entries;
2500
2501         return 0;
2502
2503 out_fail:
2504         vmlinux_path__exit();
2505         return -1;
2506 }
2507
2508 size_t machine__fprintf_vmlinux_path(struct machine *machine, FILE *fp)
2509 {
2510         int i;
2511         size_t printed = 0;
2512         struct dso *kdso = machine->vmlinux_maps[MAP__FUNCTION]->dso;
2513
2514         if (kdso->has_build_id) {
2515                 char filename[PATH_MAX];
2516                 if (dso__build_id_filename(kdso, filename, sizeof(filename)))
2517                         printed += fprintf(fp, "[0] %s\n", filename);
2518         }
2519
2520         for (i = 0; i < vmlinux_path__nr_entries; ++i)
2521                 printed += fprintf(fp, "[%d] %s\n",
2522                                    i + kdso->has_build_id, vmlinux_path[i]);
2523
2524         return printed;
2525 }
2526
2527 static int setup_list(struct strlist **list, const char *list_str,
2528                       const char *list_name)
2529 {
2530         if (list_str == NULL)
2531                 return 0;
2532
2533         *list = strlist__new(true, list_str);
2534         if (!*list) {
2535                 pr_err("problems parsing %s list\n", list_name);
2536                 return -1;
2537         }
2538         return 0;
2539 }
2540
2541 static bool symbol__read_kptr_restrict(void)
2542 {
2543         bool value = false;
2544
2545         if (geteuid() != 0) {
2546                 FILE *fp = fopen("/proc/sys/kernel/kptr_restrict", "r");
2547                 if (fp != NULL) {
2548                         char line[8];
2549
2550                         if (fgets(line, sizeof(line), fp) != NULL)
2551                                 value = atoi(line) != 0;
2552
2553                         fclose(fp);
2554                 }
2555         }
2556
2557         return value;
2558 }
2559
2560 int symbol__init(void)
2561 {
2562         const char *symfs;
2563
2564         if (symbol_conf.initialized)
2565                 return 0;
2566
2567         symbol_conf.priv_size = ALIGN(symbol_conf.priv_size, sizeof(u64));
2568
2569         elf_version(EV_CURRENT);
2570         if (symbol_conf.sort_by_name)
2571                 symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) -
2572                                           sizeof(struct symbol));
2573
2574         if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0)
2575                 return -1;
2576
2577         if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') {
2578                 pr_err("'.' is the only non valid --field-separator argument\n");
2579                 return -1;
2580         }
2581
2582         if (setup_list(&symbol_conf.dso_list,
2583                        symbol_conf.dso_list_str, "dso") < 0)
2584                 return -1;
2585
2586         if (setup_list(&symbol_conf.comm_list,
2587                        symbol_conf.comm_list_str, "comm") < 0)
2588                 goto out_free_dso_list;
2589
2590         if (setup_list(&symbol_conf.sym_list,
2591                        symbol_conf.sym_list_str, "symbol") < 0)
2592                 goto out_free_comm_list;
2593
2594         /*
2595          * A path to symbols of "/" is identical to ""
2596          * reset here for simplicity.
2597          */
2598         symfs = realpath(symbol_conf.symfs, NULL);
2599         if (symfs == NULL)
2600                 symfs = symbol_conf.symfs;
2601         if (strcmp(symfs, "/") == 0)
2602                 symbol_conf.symfs = "";
2603         if (symfs != symbol_conf.symfs)
2604                 free((void *)symfs);
2605
2606         symbol_conf.kptr_restrict = symbol__read_kptr_restrict();
2607
2608         symbol_conf.initialized = true;
2609         return 0;
2610
2611 out_free_dso_list:
2612         strlist__delete(symbol_conf.dso_list);
2613 out_free_comm_list:
2614         strlist__delete(symbol_conf.comm_list);
2615         return -1;
2616 }
2617
2618 void symbol__exit(void)
2619 {
2620         if (!symbol_conf.initialized)
2621                 return;
2622         strlist__delete(symbol_conf.sym_list);
2623         strlist__delete(symbol_conf.dso_list);
2624         strlist__delete(symbol_conf.comm_list);
2625         vmlinux_path__exit();
2626         symbol_conf.sym_list = symbol_conf.dso_list = symbol_conf.comm_list = NULL;
2627         symbol_conf.initialized = false;
2628 }
2629
2630 int machines__create_kernel_maps(struct rb_root *machines, pid_t pid)
2631 {
2632         struct machine *machine = machines__findnew(machines, pid);
2633
2634         if (machine == NULL)
2635                 return -1;
2636
2637         return machine__create_kernel_maps(machine);
2638 }
2639
2640 static int hex(char ch)
2641 {
2642         if ((ch >= '0') && (ch <= '9'))
2643                 return ch - '0';
2644         if ((ch >= 'a') && (ch <= 'f'))
2645                 return ch - 'a' + 10;
2646         if ((ch >= 'A') && (ch <= 'F'))
2647                 return ch - 'A' + 10;
2648         return -1;
2649 }
2650
2651 /*
2652  * While we find nice hex chars, build a long_val.
2653  * Return number of chars processed.
2654  */
2655 int hex2u64(const char *ptr, u64 *long_val)
2656 {
2657         const char *p = ptr;
2658         *long_val = 0;
2659
2660         while (*p) {
2661                 const int hex_val = hex(*p);
2662
2663                 if (hex_val < 0)
2664                         break;
2665
2666                 *long_val = (*long_val << 4) | hex_val;
2667                 p++;
2668         }
2669
2670         return p - ptr;
2671 }
2672
2673 char *strxfrchar(char *s, char from, char to)
2674 {
2675         char *p = s;
2676
2677         while ((p = strchr(p, from)) != NULL)
2678                 *p++ = to;
2679
2680         return s;
2681 }
2682
2683 int machines__create_guest_kernel_maps(struct rb_root *machines)
2684 {
2685         int ret = 0;
2686         struct dirent **namelist = NULL;
2687         int i, items = 0;
2688         char path[PATH_MAX];
2689         pid_t pid;
2690
2691         if (symbol_conf.default_guest_vmlinux_name ||
2692             symbol_conf.default_guest_modules ||
2693             symbol_conf.default_guest_kallsyms) {
2694                 machines__create_kernel_maps(machines, DEFAULT_GUEST_KERNEL_ID);
2695         }
2696
2697         if (symbol_conf.guestmount) {
2698                 items = scandir(symbol_conf.guestmount, &namelist, NULL, NULL);
2699                 if (items <= 0)
2700                         return -ENOENT;
2701                 for (i = 0; i < items; i++) {
2702                         if (!isdigit(namelist[i]->d_name[0])) {
2703                                 /* Filter out . and .. */
2704                                 continue;
2705                         }
2706                         pid = atoi(namelist[i]->d_name);
2707                         sprintf(path, "%s/%s/proc/kallsyms",
2708                                 symbol_conf.guestmount,
2709                                 namelist[i]->d_name);
2710                         ret = access(path, R_OK);
2711                         if (ret) {
2712                                 pr_debug("Can't access file %s\n", path);
2713                                 goto failure;
2714                         }
2715                         machines__create_kernel_maps(machines, pid);
2716                 }
2717 failure:
2718                 free(namelist);
2719         }
2720
2721         return ret;
2722 }
2723
2724 void machines__destroy_guest_kernel_maps(struct rb_root *machines)
2725 {
2726         struct rb_node *next = rb_first(machines);
2727
2728         while (next) {
2729                 struct machine *pos = rb_entry(next, struct machine, rb_node);
2730
2731                 next = rb_next(&pos->rb_node);
2732                 rb_erase(&pos->rb_node, machines);
2733                 machine__delete(pos);
2734         }
2735 }
2736
2737 int machine__load_kallsyms(struct machine *machine, const char *filename,
2738                            enum map_type type, symbol_filter_t filter)
2739 {
2740         struct map *map = machine->vmlinux_maps[type];
2741         int ret = dso__load_kallsyms(map->dso, filename, map, filter);
2742
2743         if (ret > 0) {
2744                 dso__set_loaded(map->dso, type);
2745                 /*
2746                  * Since /proc/kallsyms will have multiple sessions for the
2747                  * kernel, with modules between them, fixup the end of all
2748                  * sections.
2749                  */
2750                 __map_groups__fixup_end(&machine->kmaps, type);
2751         }
2752
2753         return ret;
2754 }
2755
2756 int machine__load_vmlinux_path(struct machine *machine, enum map_type type,
2757                                symbol_filter_t filter)
2758 {
2759         struct map *map = machine->vmlinux_maps[type];
2760         int ret = dso__load_vmlinux_path(map->dso, map, filter);
2761
2762         if (ret > 0) {
2763                 dso__set_loaded(map->dso, type);
2764                 map__reloc_vmlinux(map);
2765         }
2766
2767         return ret;
2768 }