tools/objtool/elf.c

   1 // SPDX-License-Identifier: GPL-2.0-or-later
   2 /*
   3  * elf.c - ELF access library
   4  *
   5  * Adapted from kpatch (https://github.com/dynup/kpatch):
   6  * Copyright (C) 2013-2015 Josh Poimboeuf <[email protected]>
   7  * Copyright (C) 2014 Seth Jennings <[email protected]>
   8  */
   9
  10 #include <sys/types.h>
  11 #include <sys/stat.h>
  12 #include <sys/mman.h>
  13 #include <fcntl.h>
  14 #include <stdio.h>
  15 #include <stdlib.h>
  16 #include <string.h>
  17 #include <unistd.h>
  18 #include <errno.h>
  19 #include <linux/interval_tree_generic.h>
  20 #include <objtool/builtin.h>
  21
  22 #include <objtool/elf.h>
  23 #include <objtool/warn.h>
  24
  25 static inline u32 str_hash(const char *str)
  26 {
  27         return jhash(str, strlen(str), 0);
  28 }
  29
  30 #define __elf_table(name)       (elf->name##_hash)
  31 #define __elf_bits(name)        (elf->name##_bits)
  32
  33 #define __elf_table_entry(name, key) \
  34         __elf_table(name)[hash_min(key, __elf_bits(name))]
  35
  36 #define elf_hash_add(name, node, key)                                   \
  37 ({                                                                      \
  38         struct elf_hash_node *__node = node;                            \
  39         __node->next = __elf_table_entry(name, key);                    \
  40         __elf_table_entry(name, key) = __node;                          \
  41 })
  42
  43 static inline void __elf_hash_del(struct elf_hash_node *node,
  44                                   struct elf_hash_node **head)
  45 {
  46         struct elf_hash_node *cur, *prev;
  47
  48         if (node == *head) {
  49                 *head = node->next;
  50                 return;
  51         }
  52
  53         for (prev = NULL, cur = *head; cur; prev = cur, cur = cur->next) {
  54                 if (cur == node) {
  55                         prev->next = cur->next;
  56                         break;
  57                 }
  58         }
  59 }
  60
  61 #define elf_hash_del(name, node, key) \
  62         __elf_hash_del(node, &__elf_table_entry(name, key))
  63
  64 #define elf_list_entry(ptr, type, member)                               \
  65 ({                                                                      \
  66         typeof(ptr) __ptr = (ptr);                                      \
  67         __ptr ? container_of(__ptr, type, member) : NULL;               \
  68 })
  69
  70 #define elf_hash_for_each_possible(name, obj, member, key)              \
  71         for (obj = elf_list_entry(__elf_table_entry(name, key), typeof(*obj), member); \
  72              obj;                                                       \
  73              obj = elf_list_entry(obj->member.next, typeof(*(obj)), member))
  74
  75 #define elf_alloc_hash(name, size) \
  76 ({ \
  77         __elf_bits(name) = max(10, ilog2(size)); \
  78         __elf_table(name) = mmap(NULL, sizeof(struct elf_hash_node *) << __elf_bits(name), \
  79                                  PROT_READ|PROT_WRITE, \
  80                                  MAP_PRIVATE|MAP_ANON, -1, 0); \
  81         if (__elf_table(name) == (void *)-1L) { \
  82                 WARN("mmap fail " #name); \
  83                 __elf_table(name) = NULL; \
  84         } \
  85         __elf_table(name); \
  86 })
  87
  88 static inline unsigned long __sym_start(struct symbol *s)
  89 {
  90         return s->offset;
  91 }
  92
  93 static inline unsigned long __sym_last(struct symbol *s)
  94 {
  95         return s->offset + s->len - 1;
  96 }
  97
  98 INTERVAL_TREE_DEFINE(struct symbol, node, unsigned long, __subtree_last,
  99                      __sym_start, __sym_last, static, __sym)
 100
 101 #define __sym_for_each(_iter, _tree, _start, _end)                      \
 102         for (_iter = __sym_iter_first((_tree), (_start), (_end));       \
 103              _iter; _iter = __sym_iter_next(_iter, (_start), (_end)))
 104
 105 struct symbol_hole {
 106         unsigned long key;
 107         const struct symbol *sym;
 108 };
 109
 110 /*
 111  * Find !section symbol where @offset is after it.
 112  */
 113 static int symbol_hole_by_offset(const void *key, const struct rb_node *node)
 114 {
 115         const struct symbol *s = rb_entry(node, struct symbol, node);
 116         struct symbol_hole *sh = (void *)key;
 117
 118         if (sh->key < s->offset)
 119                 return -1;
 120
 121         if (sh->key >= s->offset + s->len) {
 122                 if (s->type != STT_SECTION)
 123                         sh->sym = s;
 124                 return 1;
 125         }
 126
 127         return 0;
 128 }
 129
 130 struct section *find_section_by_name(const struct elf *elf, const char *name)
 131 {
 132         struct section *sec;
 133
 134         elf_hash_for_each_possible(section_name, sec, name_hash, str_hash(name)) {
 135                 if (!strcmp(sec->name, name))
 136                         return sec;
 137         }
 138
 139         return NULL;
 140 }
 141
 142 static struct section *find_section_by_index(struct elf *elf,
 143                                              unsigned int idx)
 144 {
 145         struct section *sec;
 146
 147         elf_hash_for_each_possible(section, sec, hash, idx) {
 148                 if (sec->idx == idx)
 149                         return sec;
 150         }
 151
 152         return NULL;
 153 }
 154
 155 static struct symbol *find_symbol_by_index(struct elf *elf, unsigned int idx)
 156 {
 157         struct symbol *sym;
 158
 159         elf_hash_for_each_possible(symbol, sym, hash, idx) {
 160                 if (sym->idx == idx)
 161                         return sym;
 162         }
 163
 164         return NULL;
 165 }
 166
 167 struct symbol *find_symbol_by_offset(struct section *sec, unsigned long offset)
 168 {
 169         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 170         struct symbol *iter;
 171
 172         __sym_for_each(iter, tree, offset, offset) {
 173                 if (iter->offset == offset && iter->type != STT_SECTION)
 174                         return iter;
 175         }
 176
 177         return NULL;
 178 }
 179
 180 struct symbol *find_func_by_offset(struct section *sec, unsigned long offset)
 181 {
 182         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 183         struct symbol *iter;
 184
 185         __sym_for_each(iter, tree, offset, offset) {
 186                 if (iter->offset == offset && iter->type == STT_FUNC)
 187                         return iter;
 188         }
 189
 190         return NULL;
 191 }
 192
 193 struct symbol *find_symbol_containing(const struct section *sec, unsigned long offset)
 194 {
 195         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 196         struct symbol *iter;
 197
 198         __sym_for_each(iter, tree, offset, offset) {
 199                 if (iter->type != STT_SECTION)
 200                         return iter;
 201         }
 202
 203         return NULL;
 204 }
 205
 206 /*
 207  * Returns size of hole starting at @offset.
 208  */
 209 int find_symbol_hole_containing(const struct section *sec, unsigned long offset)
 210 {
 211         struct symbol_hole hole = {
 212                 .key = offset,
 213                 .sym = NULL,
 214         };
 215         struct rb_node *n;
 216         struct symbol *s;
 217
 218         /*
 219          * Find the rightmost symbol for which @offset is after it.
 220          */
 221         n = rb_find(&hole, &sec->symbol_tree.rb_root, symbol_hole_by_offset);
 222
 223         /* found a symbol that contains @offset */
 224         if (n)
 225                 return 0; /* not a hole */
 226
 227         /* didn't find a symbol for which @offset is after it */
 228         if (!hole.sym)
 229                 return 0; /* not a hole */
 230
 231         /* @offset >= sym->offset + sym->len, find symbol after it */
 232         n = rb_next(&hole.sym->node);
 233         if (!n)
 234                 return -1; /* until end of address space */
 235
 236         /* hole until start of next symbol */
 237         s = rb_entry(n, struct symbol, node);
 238         return s->offset - offset;
 239 }
 240
 241 struct symbol *find_func_containing(struct section *sec, unsigned long offset)
 242 {
 243         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 244         struct symbol *iter;
 245
 246         __sym_for_each(iter, tree, offset, offset) {
 247                 if (iter->type == STT_FUNC)
 248                         return iter;
 249         }
 250
 251         return NULL;
 252 }
 253
 254 struct symbol *find_symbol_by_name(const struct elf *elf, const char *name)
 255 {
 256         struct symbol *sym;
 257
 258         elf_hash_for_each_possible(symbol_name, sym, name_hash, str_hash(name)) {
 259                 if (!strcmp(sym->name, name))
 260                         return sym;
 261         }
 262
 263         return NULL;
 264 }
 265
 266 struct reloc *find_reloc_by_dest_range(const struct elf *elf, struct section *sec,
 267                                      unsigned long offset, unsigned int len)
 268 {
 269         struct reloc *reloc, *r = NULL;
 270         struct section *rsec;
 271         unsigned long o;
 272
 273         rsec = sec->rsec;
 274         if (!rsec)
 275                 return NULL;
 276
 277         for_offset_range(o, offset, offset + len) {
 278                 elf_hash_for_each_possible(reloc, reloc, hash,
 279                                            sec_offset_hash(rsec, o)) {
 280                         if (reloc->sec != rsec)
 281                                 continue;
 282
 283                         if (reloc_offset(reloc) >= offset &&
 284                             reloc_offset(reloc) < offset + len) {
 285                                 if (!r || reloc_offset(reloc) < reloc_offset(r))
 286                                         r = reloc;
 287                         }
 288                 }
 289                 if (r)
 290                         return r;
 291         }
 292
 293         return NULL;
 294 }
 295
 296 struct reloc *find_reloc_by_dest(const struct elf *elf, struct section *sec, unsigned long offset)
 297 {
 298         return find_reloc_by_dest_range(elf, sec, offset, 1);
 299 }
 300
 301 static bool is_dwarf_section(struct section *sec)
 302 {
 303         return !strncmp(sec->name, ".debug_", 7);
 304 }
 305
 306 static int read_sections(struct elf *elf)
 307 {
 308         Elf_Scn *s = NULL;
 309         struct section *sec;
 310         size_t shstrndx, sections_nr;
 311         int i;
 312
 313         if (elf_getshdrnum(elf->elf, &sections_nr)) {
 314                 WARN_ELF("elf_getshdrnum");
 315                 return -1;
 316         }
 317
 318         if (elf_getshdrstrndx(elf->elf, &shstrndx)) {
 319                 WARN_ELF("elf_getshdrstrndx");
 320                 return -1;
 321         }
 322
 323         if (!elf_alloc_hash(section, sections_nr) ||
 324             !elf_alloc_hash(section_name, sections_nr))
 325                 return -1;
 326
 327         elf->section_data = calloc(sections_nr, sizeof(*sec));
 328         if (!elf->section_data) {
 329                 perror("calloc");
 330                 return -1;
 331         }
 332         for (i = 0; i < sections_nr; i++) {
 333                 sec = &elf->section_data[i];
 334
 335                 INIT_LIST_HEAD(&sec->symbol_list);
 336
 337                 s = elf_getscn(elf->elf, i);
 338                 if (!s) {
 339                         WARN_ELF("elf_getscn");
 340                         return -1;
 341                 }
 342
 343                 sec->idx = elf_ndxscn(s);
 344
 345                 if (!gelf_getshdr(s, &sec->sh)) {
 346                         WARN_ELF("gelf_getshdr");
 347                         return -1;
 348                 }
 349
 350                 sec->name = elf_strptr(elf->elf, shstrndx, sec->sh.sh_name);
 351                 if (!sec->name) {
 352                         WARN_ELF("elf_strptr");
 353                         return -1;
 354                 }
 355
 356                 if (sec->sh.sh_size != 0 && !is_dwarf_section(sec)) {
 357                         sec->data = elf_getdata(s, NULL);
 358                         if (!sec->data) {
 359                                 WARN_ELF("elf_getdata");
 360                                 return -1;
 361                         }
 362                         if (sec->data->d_off != 0 ||
 363                             sec->data->d_size != sec->sh.sh_size) {
 364                                 WARN("unexpected data attributes for %s",
 365                                      sec->name);
 366                                 return -1;
 367                         }
 368                 }
 369
 370                 list_add_tail(&sec->list, &elf->sections);
 371                 elf_hash_add(section, &sec->hash, sec->idx);
 372                 elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
 373
 374                 if (is_reloc_sec(sec))
 375                         elf->num_relocs += sec_num_entries(sec);
 376         }
 377
 378         if (opts.stats) {
 379                 printf("nr_sections: %lu\n", (unsigned long)sections_nr);
 380                 printf("section_bits: %d\n", elf->section_bits);
 381         }
 382
 383         /* sanity check, one more call to elf_nextscn() should return NULL */
 384         if (elf_nextscn(elf->elf, s)) {
 385                 WARN("section entry mismatch");
 386                 return -1;
 387         }
 388
 389         return 0;
 390 }
 391
 392 static void elf_add_symbol(struct elf *elf, struct symbol *sym)
 393 {
 394         struct list_head *entry;
 395         struct rb_node *pnode;
 396         struct symbol *iter;
 397
 398         INIT_LIST_HEAD(&sym->pv_target);
 399         sym->alias = sym;
 400
 401         sym->type = GELF_ST_TYPE(sym->sym.st_info);
 402         sym->bind = GELF_ST_BIND(sym->sym.st_info);
 403
 404         if (sym->type == STT_FILE)
 405                 elf->num_files++;
 406
 407         sym->offset = sym->sym.st_value;
 408         sym->len = sym->sym.st_size;
 409
 410         __sym_for_each(iter, &sym->sec->symbol_tree, sym->offset, sym->offset) {
 411                 if (iter->offset == sym->offset && iter->type == sym->type)
 412                         iter->alias = sym;
 413         }
 414
 415         __sym_insert(sym, &sym->sec->symbol_tree);
 416         pnode = rb_prev(&sym->node);
 417         if (pnode)
 418                 entry = &rb_entry(pnode, struct symbol, node)->list;
 419         else
 420                 entry = &sym->sec->symbol_list;
 421         list_add(&sym->list, entry);
 422         elf_hash_add(symbol, &sym->hash, sym->idx);
 423         elf_hash_add(symbol_name, &sym->name_hash, str_hash(sym->name));
 424
 425         /*
 426          * Don't store empty STT_NOTYPE symbols in the rbtree.  They
 427          * can exist within a function, confusing the sorting.
 428          */
 429         if (!sym->len)
 430                 __sym_remove(sym, &sym->sec->symbol_tree);
 431 }
 432
 433 static int read_symbols(struct elf *elf)
 434 {
 435         struct section *symtab, *symtab_shndx, *sec;
 436         struct symbol *sym, *pfunc;
 437         int symbols_nr, i;
 438         char *coldstr;
 439         Elf_Data *shndx_data = NULL;
 440         Elf32_Word shndx;
 441
 442         symtab = find_section_by_name(elf, ".symtab");
 443         if (symtab) {
 444                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 445                 if (symtab_shndx)
 446                         shndx_data = symtab_shndx->data;
 447
 448                 symbols_nr = sec_num_entries(symtab);
 449         } else {
 450                 /*
 451                  * A missing symbol table is actually possible if it's an empty
 452                  * .o file. This can happen for thunk_64.o. Make sure to at
 453                  * least allocate the symbol hash tables so we can do symbol
 454                  * lookups without crashing.
 455                  */
 456                 symbols_nr = 0;
 457         }
 458
 459         if (!elf_alloc_hash(symbol, symbols_nr) ||
 460             !elf_alloc_hash(symbol_name, symbols_nr))
 461                 return -1;
 462
 463         elf->symbol_data = calloc(symbols_nr, sizeof(*sym));
 464         if (!elf->symbol_data) {
 465                 perror("calloc");
 466                 return -1;
 467         }
 468         for (i = 0; i < symbols_nr; i++) {
 469                 sym = &elf->symbol_data[i];
 470
 471                 sym->idx = i;
 472
 473                 if (!gelf_getsymshndx(symtab->data, shndx_data, i, &sym->sym,
 474                                       &shndx)) {
 475                         WARN_ELF("gelf_getsymshndx");
 476                         goto err;
 477                 }
 478
 479                 sym->name = elf_strptr(elf->elf, symtab->sh.sh_link,
 480                                        sym->sym.st_name);
 481                 if (!sym->name) {
 482                         WARN_ELF("elf_strptr");
 483                         goto err;
 484                 }
 485
 486                 if ((sym->sym.st_shndx > SHN_UNDEF &&
 487                      sym->sym.st_shndx < SHN_LORESERVE) ||
 488                     (shndx_data && sym->sym.st_shndx == SHN_XINDEX)) {
 489                         if (sym->sym.st_shndx != SHN_XINDEX)
 490                                 shndx = sym->sym.st_shndx;
 491
 492                         sym->sec = find_section_by_index(elf, shndx);
 493                         if (!sym->sec) {
 494                                 WARN("couldn't find section for symbol %s",
 495                                      sym->name);
 496                                 goto err;
 497                         }
 498                         if (GELF_ST_TYPE(sym->sym.st_info) == STT_SECTION) {
 499                                 sym->name = sym->sec->name;
 500                                 sym->sec->sym = sym;
 501                         }
 502                 } else
 503                         sym->sec = find_section_by_index(elf, 0);
 504
 505                 elf_add_symbol(elf, sym);
 506         }
 507
 508         if (opts.stats) {
 509                 printf("nr_symbols: %lu\n", (unsigned long)symbols_nr);
 510                 printf("symbol_bits: %d\n", elf->symbol_bits);
 511         }
 512
 513         /* Create parent/child links for any cold subfunctions */
 514         list_for_each_entry(sec, &elf->sections, list) {
 515                 sec_for_each_sym(sec, sym) {
 516                         char *pname;
 517                         size_t pnamelen;
 518                         if (sym->type != STT_FUNC)
 519                                 continue;
 520
 521                         if (sym->pfunc == NULL)
 522                                 sym->pfunc = sym;
 523
 524                         if (sym->cfunc == NULL)
 525                                 sym->cfunc = sym;
 526
 527                         coldstr = strstr(sym->name, ".cold");
 528                         if (!coldstr)
 529                                 continue;
 530
 531                         pnamelen = coldstr - sym->name;
 532                         pname = strndup(sym->name, pnamelen);
 533                         if (!pname) {
 534                                 WARN("%s(): failed to allocate memory",
 535                                      sym->name);
 536                                 return -1;
 537                         }
 538
 539                         pfunc = find_symbol_by_name(elf, pname);
 540                         free(pname);
 541
 542                         if (!pfunc) {
 543                                 WARN("%s(): can't find parent function",
 544                                      sym->name);
 545                                 return -1;
 546                         }
 547
 548                         sym->pfunc = pfunc;
 549                         pfunc->cfunc = sym;
 550
 551                         /*
 552                          * Unfortunately, -fnoreorder-functions puts the child
 553                          * inside the parent.  Remove the overlap so we can
 554                          * have sane assumptions.
 555                          *
 556                          * Note that pfunc->len now no longer matches
 557                          * pfunc->sym.st_size.
 558                          */
 559                         if (sym->sec == pfunc->sec &&
 560                             sym->offset >= pfunc->offset &&
 561                             sym->offset + sym->len == pfunc->offset + pfunc->len) {
 562                                 pfunc->len -= sym->len;
 563                         }
 564                 }
 565         }
 566
 567         return 0;
 568
 569 err:
 570         free(sym);
 571         return -1;
 572 }
 573
 574 /*
 575  * @sym's idx has changed.  Update the relocs which reference it.
 576  */
 577 static int elf_update_sym_relocs(struct elf *elf, struct symbol *sym)
 578 {
 579         struct reloc *reloc;
 580
 581         for (reloc = sym->relocs; reloc; reloc = reloc->sym_next_reloc)
 582                 set_reloc_sym(elf, reloc, reloc->sym->idx);
 583
 584         return 0;
 585 }
 586
 587 /*
 588  * The libelf API is terrible; gelf_update_sym*() takes a data block relative
 589  * index value, *NOT* the symbol index. As such, iterate the data blocks and
 590  * adjust index until it fits.
 591  *
 592  * If no data block is found, allow adding a new data block provided the index
 593  * is only one past the end.
 594  */
 595 static int elf_update_symbol(struct elf *elf, struct section *symtab,
 596                              struct section *symtab_shndx, struct symbol *sym)
 597 {
 598         Elf32_Word shndx = sym->sec ? sym->sec->idx : SHN_UNDEF;
 599         Elf_Data *symtab_data = NULL, *shndx_data = NULL;
 600         Elf64_Xword entsize = symtab->sh.sh_entsize;
 601         int max_idx, idx = sym->idx;
 602         Elf_Scn *s, *t = NULL;
 603         bool is_special_shndx = sym->sym.st_shndx >= SHN_LORESERVE &&
 604                                 sym->sym.st_shndx != SHN_XINDEX;
 605
 606         if (is_special_shndx)
 607                 shndx = sym->sym.st_shndx;
 608
 609         s = elf_getscn(elf->elf, symtab->idx);
 610         if (!s) {
 611                 WARN_ELF("elf_getscn");
 612                 return -1;
 613         }
 614
 615         if (symtab_shndx) {
 616                 t = elf_getscn(elf->elf, symtab_shndx->idx);
 617                 if (!t) {
 618                         WARN_ELF("elf_getscn");
 619                         return -1;
 620                 }
 621         }
 622
 623         for (;;) {
 624                 /* get next data descriptor for the relevant sections */
 625                 symtab_data = elf_getdata(s, symtab_data);
 626                 if (t)
 627                         shndx_data = elf_getdata(t, shndx_data);
 628
 629                 /* end-of-list */
 630                 if (!symtab_data) {
 631                         /*
 632                          * Over-allocate to avoid O(n^2) symbol creation
 633                          * behaviour.  The down side is that libelf doesn't
 634                          * like this; see elf_truncate_section() for the fixup.
 635                          */
 636                         int num = max(1U, sym->idx/3);
 637                         void *buf;
 638
 639                         if (idx) {
 640                                 /* we don't do holes in symbol tables */
 641                                 WARN("index out of range");
 642                                 return -1;
 643                         }
 644
 645                         /* if @idx == 0, it's the next contiguous entry, create it */
 646                         symtab_data = elf_newdata(s);
 647                         if (t)
 648                                 shndx_data = elf_newdata(t);
 649
 650                         buf = calloc(num, entsize);
 651                         if (!buf) {
 652                                 WARN("malloc");
 653                                 return -1;
 654                         }
 655
 656                         symtab_data->d_buf = buf;
 657                         symtab_data->d_size = num * entsize;
 658                         symtab_data->d_align = 1;
 659                         symtab_data->d_type = ELF_T_SYM;
 660
 661                         mark_sec_changed(elf, symtab, true);
 662                         symtab->truncate = true;
 663
 664                         if (t) {
 665                                 buf = calloc(num, sizeof(Elf32_Word));
 666                                 if (!buf) {
 667                                         WARN("malloc");
 668                                         return -1;
 669                                 }
 670
 671                                 shndx_data->d_buf = buf;
 672                                 shndx_data->d_size = num * sizeof(Elf32_Word);
 673                                 shndx_data->d_align = sizeof(Elf32_Word);
 674                                 shndx_data->d_type = ELF_T_WORD;
 675
 676                                 mark_sec_changed(elf, symtab_shndx, true);
 677                                 symtab_shndx->truncate = true;
 678                         }
 679
 680                         break;
 681                 }
 682
 683                 /* empty blocks should not happen */
 684                 if (!symtab_data->d_size) {
 685                         WARN("zero size data");
 686                         return -1;
 687                 }
 688
 689                 /* is this the right block? */
 690                 max_idx = symtab_data->d_size / entsize;
 691                 if (idx < max_idx)
 692                         break;
 693
 694                 /* adjust index and try again */
 695                 idx -= max_idx;
 696         }
 697
 698         /* something went side-ways */
 699         if (idx < 0) {
 700                 WARN("negative index");
 701                 return -1;
 702         }
 703
 704         /* setup extended section index magic and write the symbol */
 705         if ((shndx >= SHN_UNDEF && shndx < SHN_LORESERVE) || is_special_shndx) {
 706                 sym->sym.st_shndx = shndx;
 707                 if (!shndx_data)
 708                         shndx = 0;
 709         } else {
 710                 sym->sym.st_shndx = SHN_XINDEX;
 711                 if (!shndx_data) {
 712                         WARN("no .symtab_shndx");
 713                         return -1;
 714                 }
 715         }
 716
 717         if (!gelf_update_symshndx(symtab_data, shndx_data, idx, &sym->sym, shndx)) {
 718                 WARN_ELF("gelf_update_symshndx");
 719                 return -1;
 720         }
 721
 722         return 0;
 723 }
 724
 725 static struct symbol *
 726 __elf_create_symbol(struct elf *elf, struct symbol *sym)
 727 {
 728         struct section *symtab, *symtab_shndx;
 729         Elf32_Word first_non_local, new_idx;
 730         struct symbol *old;
 731
 732         symtab = find_section_by_name(elf, ".symtab");
 733         if (symtab) {
 734                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 735         } else {
 736                 WARN("no .symtab");
 737                 return NULL;
 738         }
 739
 740         new_idx = sec_num_entries(symtab);
 741
 742         if (GELF_ST_BIND(sym->sym.st_info) != STB_LOCAL)
 743                 goto non_local;
 744
 745         /*
 746          * Move the first global symbol, as per sh_info, into a new, higher
 747          * symbol index. This fees up a spot for a new local symbol.
 748          */
 749         first_non_local = symtab->sh.sh_info;
 750         old = find_symbol_by_index(elf, first_non_local);
 751         if (old) {
 752
 753                 elf_hash_del(symbol, &old->hash, old->idx);
 754                 elf_hash_add(symbol, &old->hash, new_idx);
 755                 old->idx = new_idx;
 756
 757                 if (elf_update_symbol(elf, symtab, symtab_shndx, old)) {
 758                         WARN("elf_update_symbol move");
 759                         return NULL;
 760                 }
 761
 762                 if (elf_update_sym_relocs(elf, old))
 763                         return NULL;
 764
 765                 new_idx = first_non_local;
 766         }
 767
 768         /*
 769          * Either way, we will add a LOCAL symbol.
 770          */
 771         symtab->sh.sh_info += 1;
 772
 773 non_local:
 774         sym->idx = new_idx;
 775         if (elf_update_symbol(elf, symtab, symtab_shndx, sym)) {
 776                 WARN("elf_update_symbol");
 777                 return NULL;
 778         }
 779
 780         symtab->sh.sh_size += symtab->sh.sh_entsize;
 781         mark_sec_changed(elf, symtab, true);
 782
 783         if (symtab_shndx) {
 784                 symtab_shndx->sh.sh_size += sizeof(Elf32_Word);
 785                 mark_sec_changed(elf, symtab_shndx, true);
 786         }
 787
 788         return sym;
 789 }
 790
 791 static struct symbol *
 792 elf_create_section_symbol(struct elf *elf, struct section *sec)
 793 {
 794         struct symbol *sym = calloc(1, sizeof(*sym));
 795
 796         if (!sym) {
 797                 perror("malloc");
 798                 return NULL;
 799         }
 800
 801         sym->name = sec->name;
 802         sym->sec = sec;
 803
 804         // st_name 0
 805         sym->sym.st_info = GELF_ST_INFO(STB_LOCAL, STT_SECTION);
 806         // st_other 0
 807         // st_value 0
 808         // st_size 0
 809
 810         sym = __elf_create_symbol(elf, sym);
 811         if (sym)
 812                 elf_add_symbol(elf, sym);
 813
 814         return sym;
 815 }
 816
 817 static int elf_add_string(struct elf *elf, struct section *strtab, char *str);
 818
 819 struct symbol *
 820 elf_create_prefix_symbol(struct elf *elf, struct symbol *orig, long size)
 821 {
 822         struct symbol *sym = calloc(1, sizeof(*sym));
 823         size_t namelen = strlen(orig->name) + sizeof("__pfx_");
 824         char *name = malloc(namelen);
 825
 826         if (!sym || !name) {
 827                 perror("malloc");
 828                 return NULL;
 829         }
 830
 831         snprintf(name, namelen, "__pfx_%s", orig->name);
 832
 833         sym->name = name;
 834         sym->sec = orig->sec;
 835
 836         sym->sym.st_name = elf_add_string(elf, NULL, name);
 837         sym->sym.st_info = orig->sym.st_info;
 838         sym->sym.st_value = orig->sym.st_value - size;
 839         sym->sym.st_size = size;
 840
 841         sym = __elf_create_symbol(elf, sym);
 842         if (sym)
 843                 elf_add_symbol(elf, sym);
 844
 845         return sym;
 846 }
 847
 848 static struct reloc *elf_init_reloc(struct elf *elf, struct section *rsec,
 849                                     unsigned int reloc_idx,
 850                                     unsigned long offset, struct symbol *sym,
 851                                     s64 addend, unsigned int type)
 852 {
 853         struct reloc *reloc, empty = { 0 };
 854
 855         if (reloc_idx >= sec_num_entries(rsec)) {
 856                 WARN("%s: bad reloc_idx %u for %s with %d relocs",
 857                      __func__, reloc_idx, rsec->name, sec_num_entries(rsec));
 858                 return NULL;
 859         }
 860
 861         reloc = &rsec->relocs[reloc_idx];
 862
 863         if (memcmp(reloc, &empty, sizeof(empty))) {
 864                 WARN("%s: %s: reloc %d already initialized!",
 865                      __func__, rsec->name, reloc_idx);
 866                 return NULL;
 867         }
 868
 869         reloc->sec = rsec;
 870         reloc->sym = sym;
 871
 872         set_reloc_offset(elf, reloc, offset);
 873         set_reloc_sym(elf, reloc, sym->idx);
 874         set_reloc_type(elf, reloc, type);
 875         set_reloc_addend(elf, reloc, addend);
 876
 877         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 878         reloc->sym_next_reloc = sym->relocs;
 879         sym->relocs = reloc;
 880
 881         return reloc;
 882 }
 883
 884 struct reloc *elf_init_reloc_text_sym(struct elf *elf, struct section *sec,
 885                                       unsigned long offset,
 886                                       unsigned int reloc_idx,
 887                                       struct section *insn_sec,
 888                                       unsigned long insn_off)
 889 {
 890         struct symbol *sym = insn_sec->sym;
 891         int addend = insn_off;
 892
 893         if (!(insn_sec->sh.sh_flags & SHF_EXECINSTR)) {
 894                 WARN("bad call to %s() for data symbol %s",
 895                      __func__, sym->name);
 896                 return NULL;
 897         }
 898
 899         if (!sym) {
 900                 /*
 901                  * Due to how weak functions work, we must use section based
 902                  * relocations. Symbol based relocations would result in the
 903                  * weak and non-weak function annotations being overlaid on the
 904                  * non-weak function after linking.
 905                  */
 906                 sym = elf_create_section_symbol(elf, insn_sec);
 907                 if (!sym)
 908                         return NULL;
 909
 910                 insn_sec->sym = sym;
 911         }
 912
 913         return elf_init_reloc(elf, sec->rsec, reloc_idx, offset, sym, addend,
 914                               elf_text_rela_type(elf));
 915 }
 916
 917 struct reloc *elf_init_reloc_data_sym(struct elf *elf, struct section *sec,
 918                                       unsigned long offset,
 919                                       unsigned int reloc_idx,
 920                                       struct symbol *sym,
 921                                       s64 addend)
 922 {
 923         if (sym->sec && (sec->sh.sh_flags & SHF_EXECINSTR)) {
 924                 WARN("bad call to %s() for text symbol %s",
 925                      __func__, sym->name);
 926                 return NULL;
 927         }
 928
 929         return elf_init_reloc(elf, sec->rsec, reloc_idx, offset, sym, addend,
 930                               elf_data_rela_type(elf));
 931 }
 932
 933 static int read_relocs(struct elf *elf)
 934 {
 935         unsigned long nr_reloc, max_reloc = 0;
 936         struct section *rsec;
 937         struct reloc *reloc;
 938         unsigned int symndx;
 939         struct symbol *sym;
 940         int i;
 941
 942         if (!elf_alloc_hash(reloc, elf->num_relocs))
 943                 return -1;
 944
 945         list_for_each_entry(rsec, &elf->sections, list) {
 946                 if (!is_reloc_sec(rsec))
 947                         continue;
 948
 949                 rsec->base = find_section_by_index(elf, rsec->sh.sh_info);
 950                 if (!rsec->base) {
 951                         WARN("can't find base section for reloc section %s",
 952                              rsec->name);
 953                         return -1;
 954                 }
 955
 956                 rsec->base->rsec = rsec;
 957
 958                 nr_reloc = 0;
 959                 rsec->relocs = calloc(sec_num_entries(rsec), sizeof(*reloc));
 960                 if (!rsec->relocs) {
 961                         perror("calloc");
 962                         return -1;
 963                 }
 964                 for (i = 0; i < sec_num_entries(rsec); i++) {
 965                         reloc = &rsec->relocs[i];
 966
 967                         reloc->sec = rsec;
 968                         symndx = reloc_sym(reloc);
 969                         reloc->sym = sym = find_symbol_by_index(elf, symndx);
 970                         if (!reloc->sym) {
 971                                 WARN("can't find reloc entry symbol %d for %s",
 972                                      symndx, rsec->name);
 973                                 return -1;
 974                         }
 975
 976                         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 977                         reloc->sym_next_reloc = sym->relocs;
 978                         sym->relocs = reloc;
 979
 980                         nr_reloc++;
 981                 }
 982                 max_reloc = max(max_reloc, nr_reloc);
 983         }
 984
 985         if (opts.stats) {
 986                 printf("max_reloc: %lu\n", max_reloc);
 987                 printf("num_relocs: %lu\n", elf->num_relocs);
 988                 printf("reloc_bits: %d\n", elf->reloc_bits);
 989         }
 990
 991         return 0;
 992 }
 993
 994 struct elf *elf_open_read(const char *name, int flags)
 995 {
 996         struct elf *elf;
 997         Elf_Cmd cmd;
 998
 999         elf_version(EV_CURRENT);
1000
1001         elf = malloc(sizeof(*elf));
1002         if (!elf) {
1003                 perror("malloc");
1004                 return NULL;
1005         }
1006         memset(elf, 0, sizeof(*elf));
1007
1008         INIT_LIST_HEAD(&elf->sections);
1009
1010         elf->fd = open(name, flags);
1011         if (elf->fd == -1) {
1012                 fprintf(stderr, "objtool: Can't open '%s': %s\n",
1013                         name, strerror(errno));
1014                 goto err;
1015         }
1016
1017         if ((flags & O_ACCMODE) == O_RDONLY)
1018                 cmd = ELF_C_READ_MMAP;
1019         else if ((flags & O_ACCMODE) == O_RDWR)
1020                 cmd = ELF_C_RDWR;
1021         else /* O_WRONLY */
1022                 cmd = ELF_C_WRITE;
1023
1024         elf->elf = elf_begin(elf->fd, cmd, NULL);
1025         if (!elf->elf) {
1026                 WARN_ELF("elf_begin");
1027                 goto err;
1028         }
1029
1030         if (!gelf_getehdr(elf->elf, &elf->ehdr)) {
1031                 WARN_ELF("gelf_getehdr");
1032                 goto err;
1033         }
1034
1035         if (read_sections(elf))
1036                 goto err;
1037
1038         if (read_symbols(elf))
1039                 goto err;
1040
1041         if (read_relocs(elf))
1042                 goto err;
1043
1044         return elf;
1045
1046 err:
1047         elf_close(elf);
1048         return NULL;
1049 }
1050
1051 static int elf_add_string(struct elf *elf, struct section *strtab, char *str)
1052 {
1053         Elf_Data *data;
1054         Elf_Scn *s;
1055         int len;
1056
1057         if (!strtab)
1058                 strtab = find_section_by_name(elf, ".strtab");
1059         if (!strtab) {
1060                 WARN("can't find .strtab section");
1061                 return -1;
1062         }
1063
1064         s = elf_getscn(elf->elf, strtab->idx);
1065         if (!s) {
1066                 WARN_ELF("elf_getscn");
1067                 return -1;
1068         }
1069
1070         data = elf_newdata(s);
1071         if (!data) {
1072                 WARN_ELF("elf_newdata");
1073                 return -1;
1074         }
1075
1076         data->d_buf = str;
1077         data->d_size = strlen(str) + 1;
1078         data->d_align = 1;
1079
1080         len = strtab->sh.sh_size;
1081         strtab->sh.sh_size += data->d_size;
1082
1083         mark_sec_changed(elf, strtab, true);
1084
1085         return len;
1086 }
1087
1088 struct section *elf_create_section(struct elf *elf, const char *name,
1089                                    size_t entsize, unsigned int nr)
1090 {
1091         struct section *sec, *shstrtab;
1092         size_t size = entsize * nr;
1093         Elf_Scn *s;
1094
1095         sec = malloc(sizeof(*sec));
1096         if (!sec) {
1097                 perror("malloc");
1098                 return NULL;
1099         }
1100         memset(sec, 0, sizeof(*sec));
1101
1102         INIT_LIST_HEAD(&sec->symbol_list);
1103
1104         s = elf_newscn(elf->elf);
1105         if (!s) {
1106                 WARN_ELF("elf_newscn");
1107                 return NULL;
1108         }
1109
1110         sec->name = strdup(name);
1111         if (!sec->name) {
1112                 perror("strdup");
1113                 return NULL;
1114         }
1115
1116         sec->idx = elf_ndxscn(s);
1117
1118         sec->data = elf_newdata(s);
1119         if (!sec->data) {
1120                 WARN_ELF("elf_newdata");
1121                 return NULL;
1122         }
1123
1124         sec->data->d_size = size;
1125         sec->data->d_align = 1;
1126
1127         if (size) {
1128                 sec->data->d_buf = malloc(size);
1129                 if (!sec->data->d_buf) {
1130                         perror("malloc");
1131                         return NULL;
1132                 }
1133                 memset(sec->data->d_buf, 0, size);
1134         }
1135
1136         if (!gelf_getshdr(s, &sec->sh)) {
1137                 WARN_ELF("gelf_getshdr");
1138                 return NULL;
1139         }
1140
1141         sec->sh.sh_size = size;
1142         sec->sh.sh_entsize = entsize;
1143         sec->sh.sh_type = SHT_PROGBITS;
1144         sec->sh.sh_addralign = 1;
1145         sec->sh.sh_flags = SHF_ALLOC;
1146
1147         /* Add section name to .shstrtab (or .strtab for Clang) */
1148         shstrtab = find_section_by_name(elf, ".shstrtab");
1149         if (!shstrtab)
1150                 shstrtab = find_section_by_name(elf, ".strtab");
1151         if (!shstrtab) {
1152                 WARN("can't find .shstrtab or .strtab section");
1153                 return NULL;
1154         }
1155         sec->sh.sh_name = elf_add_string(elf, shstrtab, sec->name);
1156         if (sec->sh.sh_name == -1)
1157                 return NULL;
1158
1159         list_add_tail(&sec->list, &elf->sections);
1160         elf_hash_add(section, &sec->hash, sec->idx);
1161         elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
1162
1163         mark_sec_changed(elf, sec, true);
1164
1165         return sec;
1166 }
1167
1168 static struct section *elf_create_rela_section(struct elf *elf,
1169                                                struct section *sec,
1170                                                unsigned int reloc_nr)
1171 {
1172         struct section *rsec;
1173         char *rsec_name;
1174
1175         rsec_name = malloc(strlen(sec->name) + strlen(".rela") + 1);
1176         if (!rsec_name) {
1177                 perror("malloc");
1178                 return NULL;
1179         }
1180         strcpy(rsec_name, ".rela");
1181         strcat(rsec_name, sec->name);
1182
1183         rsec = elf_create_section(elf, rsec_name, elf_rela_size(elf), reloc_nr);
1184         free(rsec_name);
1185         if (!rsec)
1186                 return NULL;
1187
1188         rsec->data->d_type = ELF_T_RELA;
1189         rsec->sh.sh_type = SHT_RELA;
1190         rsec->sh.sh_addralign = elf_addr_size(elf);
1191         rsec->sh.sh_link = find_section_by_name(elf, ".symtab")->idx;
1192         rsec->sh.sh_info = sec->idx;
1193         rsec->sh.sh_flags = SHF_INFO_LINK;
1194
1195         rsec->relocs = calloc(sec_num_entries(rsec), sizeof(struct reloc));
1196         if (!rsec->relocs) {
1197                 perror("calloc");
1198                 return NULL;
1199         }
1200
1201         sec->rsec = rsec;
1202         rsec->base = sec;
1203
1204         return rsec;
1205 }
1206
1207 struct section *elf_create_section_pair(struct elf *elf, const char *name,
1208                                         size_t entsize, unsigned int nr,
1209                                         unsigned int reloc_nr)
1210 {
1211         struct section *sec;
1212
1213         sec = elf_create_section(elf, name, entsize, nr);
1214         if (!sec)
1215                 return NULL;
1216
1217         if (!elf_create_rela_section(elf, sec, reloc_nr))
1218                 return NULL;
1219
1220         return sec;
1221 }
1222
1223 int elf_write_insn(struct elf *elf, struct section *sec,
1224                    unsigned long offset, unsigned int len,
1225                    const char *insn)
1226 {
1227         Elf_Data *data = sec->data;
1228
1229         if (data->d_type != ELF_T_BYTE || data->d_off) {
1230                 WARN("write to unexpected data for section: %s", sec->name);
1231                 return -1;
1232         }
1233
1234         memcpy(data->d_buf + offset, insn, len);
1235
1236         mark_sec_changed(elf, sec, true);
1237
1238         return 0;
1239 }
1240
1241 /*
1242  * When Elf_Scn::sh_size is smaller than the combined Elf_Data::d_size
1243  * do you:
1244  *
1245  *   A) adhere to the section header and truncate the data, or
1246  *   B) ignore the section header and write out all the data you've got?
1247  *
1248  * Yes, libelf sucks and we need to manually truncate if we over-allocate data.
1249  */
1250 static int elf_truncate_section(struct elf *elf, struct section *sec)
1251 {
1252         u64 size = sec->sh.sh_size;
1253         bool truncated = false;
1254         Elf_Data *data = NULL;
1255         Elf_Scn *s;
1256
1257         s = elf_getscn(elf->elf, sec->idx);
1258         if (!s) {
1259                 WARN_ELF("elf_getscn");
1260                 return -1;
1261         }
1262
1263         for (;;) {
1264                 /* get next data descriptor for the relevant section */
1265                 data = elf_getdata(s, data);
1266
1267                 if (!data) {
1268                         if (size) {
1269                                 WARN("end of section data but non-zero size left\n");
1270                                 return -1;
1271                         }
1272                         return 0;
1273                 }
1274
1275                 if (truncated) {
1276                         /* when we remove symbols */
1277                         WARN("truncated; but more data\n");
1278                         return -1;
1279                 }
1280
1281                 if (!data->d_size) {
1282                         WARN("zero size data");
1283                         return -1;
1284                 }
1285
1286                 if (data->d_size > size) {
1287                         truncated = true;
1288                         data->d_size = size;
1289                 }
1290
1291                 size -= data->d_size;
1292         }
1293 }
1294
1295 int elf_write(struct elf *elf)
1296 {
1297         struct section *sec;
1298         Elf_Scn *s;
1299
1300         if (opts.dryrun)
1301                 return 0;
1302
1303         /* Update changed relocation sections and section headers: */
1304         list_for_each_entry(sec, &elf->sections, list) {
1305                 if (sec->truncate)
1306                         elf_truncate_section(elf, sec);
1307
1308                 if (sec_changed(sec)) {
1309                         s = elf_getscn(elf->elf, sec->idx);
1310                         if (!s) {
1311                                 WARN_ELF("elf_getscn");
1312                                 return -1;
1313                         }
1314
1315                         /* Note this also flags the section dirty */
1316                         if (!gelf_update_shdr(s, &sec->sh)) {
1317                                 WARN_ELF("gelf_update_shdr");
1318                                 return -1;
1319                         }
1320
1321                         mark_sec_changed(elf, sec, false);
1322                 }
1323         }
1324
1325         /* Make sure the new section header entries get updated properly. */
1326         elf_flagelf(elf->elf, ELF_C_SET, ELF_F_DIRTY);
1327
1328         /* Write all changes to the file. */
1329         if (elf_update(elf->elf, ELF_C_WRITE) < 0) {
1330                 WARN_ELF("elf_update");
1331                 return -1;
1332         }
1333
1334         elf->changed = false;
1335
1336         return 0;
1337 }
1338
1339 void elf_close(struct elf *elf)
1340 {
1341         if (elf->elf)
1342                 elf_end(elf->elf);
1343
1344         if (elf->fd > 0)
1345                 close(elf->fd);
1346
1347         /*
1348          * NOTE: All remaining allocations are leaked on purpose.  Objtool is
1349          * about to exit anyway.
1350          */
1351 }