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 <jpoimboe@redhat.com>
   7  * Copyright (C) 2014 Seth Jennings <sjenning@redhat.com>
   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         /*
 228          * @offset >= sym->offset + sym->len, find symbol after it.
 229          * When hole.sym is empty, use the first node to compute the hole.
 230          * If there is no symbol in the section, the first node will be NULL,
 231          * in which case, -1 is returned to skip the whole section.
 232          */
 233         if (hole.sym)
 234                 n = rb_next(&hole.sym->node);
 235         else
 236                 n = rb_first_cached(&sec->symbol_tree);
 237
 238         if (!n)
 239                 return -1; /* until end of address space */
 240
 241         /* hole until start of next symbol */
 242         s = rb_entry(n, struct symbol, node);
 243         return s->offset - offset;
 244 }
 245
 246 struct symbol *find_func_containing(struct section *sec, unsigned long offset)
 247 {
 248         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 249         struct symbol *iter;
 250
 251         __sym_for_each(iter, tree, offset, offset) {
 252                 if (iter->type == STT_FUNC)
 253                         return iter;
 254         }
 255
 256         return NULL;
 257 }
 258
 259 struct symbol *find_symbol_by_name(const struct elf *elf, const char *name)
 260 {
 261         struct symbol *sym;
 262
 263         elf_hash_for_each_possible(symbol_name, sym, name_hash, str_hash(name)) {
 264                 if (!strcmp(sym->name, name))
 265                         return sym;
 266         }
 267
 268         return NULL;
 269 }
 270
 271 struct reloc *find_reloc_by_dest_range(const struct elf *elf, struct section *sec,
 272                                      unsigned long offset, unsigned int len)
 273 {
 274         struct reloc *reloc, *r = NULL;
 275         struct section *rsec;
 276         unsigned long o;
 277
 278         rsec = sec->rsec;
 279         if (!rsec)
 280                 return NULL;
 281
 282         for_offset_range(o, offset, offset + len) {
 283                 elf_hash_for_each_possible(reloc, reloc, hash,
 284                                            sec_offset_hash(rsec, o)) {
 285                         if (reloc->sec != rsec)
 286                                 continue;
 287
 288                         if (reloc_offset(reloc) >= offset &&
 289                             reloc_offset(reloc) < offset + len) {
 290                                 if (!r || reloc_offset(reloc) < reloc_offset(r))
 291                                         r = reloc;
 292                         }
 293                 }
 294                 if (r)
 295                         return r;
 296         }
 297
 298         return NULL;
 299 }
 300
 301 struct reloc *find_reloc_by_dest(const struct elf *elf, struct section *sec, unsigned long offset)
 302 {
 303         return find_reloc_by_dest_range(elf, sec, offset, 1);
 304 }
 305
 306 static bool is_dwarf_section(struct section *sec)
 307 {
 308         return !strncmp(sec->name, ".debug_", 7);
 309 }
 310
 311 static int read_sections(struct elf *elf)
 312 {
 313         Elf_Scn *s = NULL;
 314         struct section *sec;
 315         size_t shstrndx, sections_nr;
 316         int i;
 317
 318         if (elf_getshdrnum(elf->elf, &sections_nr)) {
 319                 WARN_ELF("elf_getshdrnum");
 320                 return -1;
 321         }
 322
 323         if (elf_getshdrstrndx(elf->elf, &shstrndx)) {
 324                 WARN_ELF("elf_getshdrstrndx");
 325                 return -1;
 326         }
 327
 328         if (!elf_alloc_hash(section, sections_nr) ||
 329             !elf_alloc_hash(section_name, sections_nr))
 330                 return -1;
 331
 332         elf->section_data = calloc(sections_nr, sizeof(*sec));
 333         if (!elf->section_data) {
 334                 perror("calloc");
 335                 return -1;
 336         }
 337         for (i = 0; i < sections_nr; i++) {
 338                 sec = &elf->section_data[i];
 339
 340                 INIT_LIST_HEAD(&sec->symbol_list);
 341
 342                 s = elf_getscn(elf->elf, i);
 343                 if (!s) {
 344                         WARN_ELF("elf_getscn");
 345                         return -1;
 346                 }
 347
 348                 sec->idx = elf_ndxscn(s);
 349
 350                 if (!gelf_getshdr(s, &sec->sh)) {
 351                         WARN_ELF("gelf_getshdr");
 352                         return -1;
 353                 }
 354
 355                 sec->name = elf_strptr(elf->elf, shstrndx, sec->sh.sh_name);
 356                 if (!sec->name) {
 357                         WARN_ELF("elf_strptr");
 358                         return -1;
 359                 }
 360
 361                 if (sec->sh.sh_size != 0 && !is_dwarf_section(sec)) {
 362                         sec->data = elf_getdata(s, NULL);
 363                         if (!sec->data) {
 364                                 WARN_ELF("elf_getdata");
 365                                 return -1;
 366                         }
 367                         if (sec->data->d_off != 0 ||
 368                             sec->data->d_size != sec->sh.sh_size) {
 369                                 WARN("unexpected data attributes for %s",
 370                                      sec->name);
 371                                 return -1;
 372                         }
 373                 }
 374
 375                 list_add_tail(&sec->list, &elf->sections);
 376                 elf_hash_add(section, &sec->hash, sec->idx);
 377                 elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
 378
 379                 if (is_reloc_sec(sec))
 380                         elf->num_relocs += sec_num_entries(sec);
 381         }
 382
 383         if (opts.stats) {
 384                 printf("nr_sections: %lu\n", (unsigned long)sections_nr);
 385                 printf("section_bits: %d\n", elf->section_bits);
 386         }
 387
 388         /* sanity check, one more call to elf_nextscn() should return NULL */
 389         if (elf_nextscn(elf->elf, s)) {
 390                 WARN("section entry mismatch");
 391                 return -1;
 392         }
 393
 394         return 0;
 395 }
 396
 397 static void elf_add_symbol(struct elf *elf, struct symbol *sym)
 398 {
 399         struct list_head *entry;
 400         struct rb_node *pnode;
 401         struct symbol *iter;
 402
 403         INIT_LIST_HEAD(&sym->pv_target);
 404         sym->alias = sym;
 405
 406         sym->type = GELF_ST_TYPE(sym->sym.st_info);
 407         sym->bind = GELF_ST_BIND(sym->sym.st_info);
 408
 409         if (sym->type == STT_FILE)
 410                 elf->num_files++;
 411
 412         sym->offset = sym->sym.st_value;
 413         sym->len = sym->sym.st_size;
 414
 415         __sym_for_each(iter, &sym->sec->symbol_tree, sym->offset, sym->offset) {
 416                 if (iter->offset == sym->offset && iter->type == sym->type)
 417                         iter->alias = sym;
 418         }
 419
 420         __sym_insert(sym, &sym->sec->symbol_tree);
 421         pnode = rb_prev(&sym->node);
 422         if (pnode)
 423                 entry = &rb_entry(pnode, struct symbol, node)->list;
 424         else
 425                 entry = &sym->sec->symbol_list;
 426         list_add(&sym->list, entry);
 427         elf_hash_add(symbol, &sym->hash, sym->idx);
 428         elf_hash_add(symbol_name, &sym->name_hash, str_hash(sym->name));
 429
 430         /*
 431          * Don't store empty STT_NOTYPE symbols in the rbtree.  They
 432          * can exist within a function, confusing the sorting.
 433          */
 434         if (!sym->len)
 435                 __sym_remove(sym, &sym->sec->symbol_tree);
 436 }
 437
 438 static int read_symbols(struct elf *elf)
 439 {
 440         struct section *symtab, *symtab_shndx, *sec;
 441         struct symbol *sym, *pfunc;
 442         int symbols_nr, i;
 443         char *coldstr;
 444         Elf_Data *shndx_data = NULL;
 445         Elf32_Word shndx;
 446
 447         symtab = find_section_by_name(elf, ".symtab");
 448         if (symtab) {
 449                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 450                 if (symtab_shndx)
 451                         shndx_data = symtab_shndx->data;
 452
 453                 symbols_nr = sec_num_entries(symtab);
 454         } else {
 455                 /*
 456                  * A missing symbol table is actually possible if it's an empty
 457                  * .o file. This can happen for thunk_64.o. Make sure to at
 458                  * least allocate the symbol hash tables so we can do symbol
 459                  * lookups without crashing.
 460                  */
 461                 symbols_nr = 0;
 462         }
 463
 464         if (!elf_alloc_hash(symbol, symbols_nr) ||
 465             !elf_alloc_hash(symbol_name, symbols_nr))
 466                 return -1;
 467
 468         elf->symbol_data = calloc(symbols_nr, sizeof(*sym));
 469         if (!elf->symbol_data) {
 470                 perror("calloc");
 471                 return -1;
 472         }
 473         for (i = 0; i < symbols_nr; i++) {
 474                 sym = &elf->symbol_data[i];
 475
 476                 sym->idx = i;
 477
 478                 if (!gelf_getsymshndx(symtab->data, shndx_data, i, &sym->sym,
 479                                       &shndx)) {
 480                         WARN_ELF("gelf_getsymshndx");
 481                         goto err;
 482                 }
 483
 484                 sym->name = elf_strptr(elf->elf, symtab->sh.sh_link,
 485                                        sym->sym.st_name);
 486                 if (!sym->name) {
 487                         WARN_ELF("elf_strptr");
 488                         goto err;
 489                 }
 490
 491                 if ((sym->sym.st_shndx > SHN_UNDEF &&
 492                      sym->sym.st_shndx < SHN_LORESERVE) ||
 493                     (shndx_data && sym->sym.st_shndx == SHN_XINDEX)) {
 494                         if (sym->sym.st_shndx != SHN_XINDEX)
 495                                 shndx = sym->sym.st_shndx;
 496
 497                         sym->sec = find_section_by_index(elf, shndx);
 498                         if (!sym->sec) {
 499                                 WARN("couldn't find section for symbol %s",
 500                                      sym->name);
 501                                 goto err;
 502                         }
 503                         if (GELF_ST_TYPE(sym->sym.st_info) == STT_SECTION) {
 504                                 sym->name = sym->sec->name;
 505                                 sym->sec->sym = sym;
 506                         }
 507                 } else
 508                         sym->sec = find_section_by_index(elf, 0);
 509
 510                 elf_add_symbol(elf, sym);
 511         }
 512
 513         if (opts.stats) {
 514                 printf("nr_symbols: %lu\n", (unsigned long)symbols_nr);
 515                 printf("symbol_bits: %d\n", elf->symbol_bits);
 516         }
 517
 518         /* Create parent/child links for any cold subfunctions */
 519         list_for_each_entry(sec, &elf->sections, list) {
 520                 sec_for_each_sym(sec, sym) {
 521                         char *pname;
 522                         size_t pnamelen;
 523                         if (sym->type != STT_FUNC)
 524                                 continue;
 525
 526                         if (sym->pfunc == NULL)
 527                                 sym->pfunc = sym;
 528
 529                         if (sym->cfunc == NULL)
 530                                 sym->cfunc = sym;
 531
 532                         coldstr = strstr(sym->name, ".cold");
 533                         if (!coldstr)
 534                                 continue;
 535
 536                         pnamelen = coldstr - sym->name;
 537                         pname = strndup(sym->name, pnamelen);
 538                         if (!pname) {
 539                                 WARN("%s(): failed to allocate memory",
 540                                      sym->name);
 541                                 return -1;
 542                         }
 543
 544                         pfunc = find_symbol_by_name(elf, pname);
 545                         free(pname);
 546
 547                         if (!pfunc) {
 548                                 WARN("%s(): can't find parent function",
 549                                      sym->name);
 550                                 return -1;
 551                         }
 552
 553                         sym->pfunc = pfunc;
 554                         pfunc->cfunc = sym;
 555
 556                         /*
 557                          * Unfortunately, -fnoreorder-functions puts the child
 558                          * inside the parent.  Remove the overlap so we can
 559                          * have sane assumptions.
 560                          *
 561                          * Note that pfunc->len now no longer matches
 562                          * pfunc->sym.st_size.
 563                          */
 564                         if (sym->sec == pfunc->sec &&
 565                             sym->offset >= pfunc->offset &&
 566                             sym->offset + sym->len == pfunc->offset + pfunc->len) {
 567                                 pfunc->len -= sym->len;
 568                         }
 569                 }
 570         }
 571
 572         return 0;
 573
 574 err:
 575         free(sym);
 576         return -1;
 577 }
 578
 579 /*
 580  * @sym's idx has changed.  Update the relocs which reference it.
 581  */
 582 static int elf_update_sym_relocs(struct elf *elf, struct symbol *sym)
 583 {
 584         struct reloc *reloc;
 585
 586         for (reloc = sym->relocs; reloc; reloc = reloc->sym_next_reloc)
 587                 set_reloc_sym(elf, reloc, reloc->sym->idx);
 588
 589         return 0;
 590 }
 591
 592 /*
 593  * The libelf API is terrible; gelf_update_sym*() takes a data block relative
 594  * index value, *NOT* the symbol index. As such, iterate the data blocks and
 595  * adjust index until it fits.
 596  *
 597  * If no data block is found, allow adding a new data block provided the index
 598  * is only one past the end.
 599  */
 600 static int elf_update_symbol(struct elf *elf, struct section *symtab,
 601                              struct section *symtab_shndx, struct symbol *sym)
 602 {
 603         Elf32_Word shndx = sym->sec ? sym->sec->idx : SHN_UNDEF;
 604         Elf_Data *symtab_data = NULL, *shndx_data = NULL;
 605         Elf64_Xword entsize = symtab->sh.sh_entsize;
 606         int max_idx, idx = sym->idx;
 607         Elf_Scn *s, *t = NULL;
 608         bool is_special_shndx = sym->sym.st_shndx >= SHN_LORESERVE &&
 609                                 sym->sym.st_shndx != SHN_XINDEX;
 610
 611         if (is_special_shndx)
 612                 shndx = sym->sym.st_shndx;
 613
 614         s = elf_getscn(elf->elf, symtab->idx);
 615         if (!s) {
 616                 WARN_ELF("elf_getscn");
 617                 return -1;
 618         }
 619
 620         if (symtab_shndx) {
 621                 t = elf_getscn(elf->elf, symtab_shndx->idx);
 622                 if (!t) {
 623                         WARN_ELF("elf_getscn");
 624                         return -1;
 625                 }
 626         }
 627
 628         for (;;) {
 629                 /* get next data descriptor for the relevant sections */
 630                 symtab_data = elf_getdata(s, symtab_data);
 631                 if (t)
 632                         shndx_data = elf_getdata(t, shndx_data);
 633
 634                 /* end-of-list */
 635                 if (!symtab_data) {
 636                         /*
 637                          * Over-allocate to avoid O(n^2) symbol creation
 638                          * behaviour.  The down side is that libelf doesn't
 639                          * like this; see elf_truncate_section() for the fixup.
 640                          */
 641                         int num = max(1U, sym->idx/3);
 642                         void *buf;
 643
 644                         if (idx) {
 645                                 /* we don't do holes in symbol tables */
 646                                 WARN("index out of range");
 647                                 return -1;
 648                         }
 649
 650                         /* if @idx == 0, it's the next contiguous entry, create it */
 651                         symtab_data = elf_newdata(s);
 652                         if (t)
 653                                 shndx_data = elf_newdata(t);
 654
 655                         buf = calloc(num, entsize);
 656                         if (!buf) {
 657                                 WARN("malloc");
 658                                 return -1;
 659                         }
 660
 661                         symtab_data->d_buf = buf;
 662                         symtab_data->d_size = num * entsize;
 663                         symtab_data->d_align = 1;
 664                         symtab_data->d_type = ELF_T_SYM;
 665
 666                         mark_sec_changed(elf, symtab, true);
 667                         symtab->truncate = true;
 668
 669                         if (t) {
 670                                 buf = calloc(num, sizeof(Elf32_Word));
 671                                 if (!buf) {
 672                                         WARN("malloc");
 673                                         return -1;
 674                                 }
 675
 676                                 shndx_data->d_buf = buf;
 677                                 shndx_data->d_size = num * sizeof(Elf32_Word);
 678                                 shndx_data->d_align = sizeof(Elf32_Word);
 679                                 shndx_data->d_type = ELF_T_WORD;
 680
 681                                 mark_sec_changed(elf, symtab_shndx, true);
 682                                 symtab_shndx->truncate = true;
 683                         }
 684
 685                         break;
 686                 }
 687
 688                 /* empty blocks should not happen */
 689                 if (!symtab_data->d_size) {
 690                         WARN("zero size data");
 691                         return -1;
 692                 }
 693
 694                 /* is this the right block? */
 695                 max_idx = symtab_data->d_size / entsize;
 696                 if (idx < max_idx)
 697                         break;
 698
 699                 /* adjust index and try again */
 700                 idx -= max_idx;
 701         }
 702
 703         /* something went side-ways */
 704         if (idx < 0) {
 705                 WARN("negative index");
 706                 return -1;
 707         }
 708
 709         /* setup extended section index magic and write the symbol */
 710         if ((shndx >= SHN_UNDEF && shndx < SHN_LORESERVE) || is_special_shndx) {
 711                 sym->sym.st_shndx = shndx;
 712                 if (!shndx_data)
 713                         shndx = 0;
 714         } else {
 715                 sym->sym.st_shndx = SHN_XINDEX;
 716                 if (!shndx_data) {
 717                         WARN("no .symtab_shndx");
 718                         return -1;
 719                 }
 720         }
 721
 722         if (!gelf_update_symshndx(symtab_data, shndx_data, idx, &sym->sym, shndx)) {
 723                 WARN_ELF("gelf_update_symshndx");
 724                 return -1;
 725         }
 726
 727         return 0;
 728 }
 729
 730 static struct symbol *
 731 __elf_create_symbol(struct elf *elf, struct symbol *sym)
 732 {
 733         struct section *symtab, *symtab_shndx;
 734         Elf32_Word first_non_local, new_idx;
 735         struct symbol *old;
 736
 737         symtab = find_section_by_name(elf, ".symtab");
 738         if (symtab) {
 739                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 740         } else {
 741                 WARN("no .symtab");
 742                 return NULL;
 743         }
 744
 745         new_idx = sec_num_entries(symtab);
 746
 747         if (GELF_ST_BIND(sym->sym.st_info) != STB_LOCAL)
 748                 goto non_local;
 749
 750         /*
 751          * Move the first global symbol, as per sh_info, into a new, higher
 752          * symbol index. This fees up a spot for a new local symbol.
 753          */
 754         first_non_local = symtab->sh.sh_info;
 755         old = find_symbol_by_index(elf, first_non_local);
 756         if (old) {
 757
 758                 elf_hash_del(symbol, &old->hash, old->idx);
 759                 elf_hash_add(symbol, &old->hash, new_idx);
 760                 old->idx = new_idx;
 761
 762                 if (elf_update_symbol(elf, symtab, symtab_shndx, old)) {
 763                         WARN("elf_update_symbol move");
 764                         return NULL;
 765                 }
 766
 767                 if (elf_update_sym_relocs(elf, old))
 768                         return NULL;
 769
 770                 new_idx = first_non_local;
 771         }
 772
 773         /*
 774          * Either way, we will add a LOCAL symbol.
 775          */
 776         symtab->sh.sh_info += 1;
 777
 778 non_local:
 779         sym->idx = new_idx;
 780         if (elf_update_symbol(elf, symtab, symtab_shndx, sym)) {
 781                 WARN("elf_update_symbol");
 782                 return NULL;
 783         }
 784
 785         symtab->sh.sh_size += symtab->sh.sh_entsize;
 786         mark_sec_changed(elf, symtab, true);
 787
 788         if (symtab_shndx) {
 789                 symtab_shndx->sh.sh_size += sizeof(Elf32_Word);
 790                 mark_sec_changed(elf, symtab_shndx, true);
 791         }
 792
 793         return sym;
 794 }
 795
 796 static struct symbol *
 797 elf_create_section_symbol(struct elf *elf, struct section *sec)
 798 {
 799         struct symbol *sym = calloc(1, sizeof(*sym));
 800
 801         if (!sym) {
 802                 perror("malloc");
 803                 return NULL;
 804         }
 805
 806         sym->name = sec->name;
 807         sym->sec = sec;
 808
 809         // st_name 0
 810         sym->sym.st_info = GELF_ST_INFO(STB_LOCAL, STT_SECTION);
 811         // st_other 0
 812         // st_value 0
 813         // st_size 0
 814
 815         sym = __elf_create_symbol(elf, sym);
 816         if (sym)
 817                 elf_add_symbol(elf, sym);
 818
 819         return sym;
 820 }
 821
 822 static int elf_add_string(struct elf *elf, struct section *strtab, char *str);
 823
 824 struct symbol *
 825 elf_create_prefix_symbol(struct elf *elf, struct symbol *orig, long size)
 826 {
 827         struct symbol *sym = calloc(1, sizeof(*sym));
 828         size_t namelen = strlen(orig->name) + sizeof("__pfx_");
 829         char *name = malloc(namelen);
 830
 831         if (!sym || !name) {
 832                 perror("malloc");
 833                 return NULL;
 834         }
 835
 836         snprintf(name, namelen, "__pfx_%s", orig->name);
 837
 838         sym->name = name;
 839         sym->sec = orig->sec;
 840
 841         sym->sym.st_name = elf_add_string(elf, NULL, name);
 842         sym->sym.st_info = orig->sym.st_info;
 843         sym->sym.st_value = orig->sym.st_value - size;
 844         sym->sym.st_size = size;
 845
 846         sym = __elf_create_symbol(elf, sym);
 847         if (sym)
 848                 elf_add_symbol(elf, sym);
 849
 850         return sym;
 851 }
 852
 853 static struct reloc *elf_init_reloc(struct elf *elf, struct section *rsec,
 854                                     unsigned int reloc_idx,
 855                                     unsigned long offset, struct symbol *sym,
 856                                     s64 addend, unsigned int type)
 857 {
 858         struct reloc *reloc, empty = { 0 };
 859
 860         if (reloc_idx >= sec_num_entries(rsec)) {
 861                 WARN("%s: bad reloc_idx %u for %s with %d relocs",
 862                      __func__, reloc_idx, rsec->name, sec_num_entries(rsec));
 863                 return NULL;
 864         }
 865
 866         reloc = &rsec->relocs[reloc_idx];
 867
 868         if (memcmp(reloc, &empty, sizeof(empty))) {
 869                 WARN("%s: %s: reloc %d already initialized!",
 870                      __func__, rsec->name, reloc_idx);
 871                 return NULL;
 872         }
 873
 874         reloc->sec = rsec;
 875         reloc->sym = sym;
 876
 877         set_reloc_offset(elf, reloc, offset);
 878         set_reloc_sym(elf, reloc, sym->idx);
 879         set_reloc_type(elf, reloc, type);
 880         set_reloc_addend(elf, reloc, addend);
 881
 882         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 883         reloc->sym_next_reloc = sym->relocs;
 884         sym->relocs = reloc;
 885
 886         return reloc;
 887 }
 888
 889 struct reloc *elf_init_reloc_text_sym(struct elf *elf, struct section *sec,
 890                                       unsigned long offset,
 891                                       unsigned int reloc_idx,
 892                                       struct section *insn_sec,
 893                                       unsigned long insn_off)
 894 {
 895         struct symbol *sym = insn_sec->sym;
 896         int addend = insn_off;
 897
 898         if (!(insn_sec->sh.sh_flags & SHF_EXECINSTR)) {
 899                 WARN("bad call to %s() for data symbol %s",
 900                      __func__, sym->name);
 901                 return NULL;
 902         }
 903
 904         if (!sym) {
 905                 /*
 906                  * Due to how weak functions work, we must use section based
 907                  * relocations. Symbol based relocations would result in the
 908                  * weak and non-weak function annotations being overlaid on the
 909                  * non-weak function after linking.
 910                  */
 911                 sym = elf_create_section_symbol(elf, insn_sec);
 912                 if (!sym)
 913                         return NULL;
 914
 915                 insn_sec->sym = sym;
 916         }
 917
 918         return elf_init_reloc(elf, sec->rsec, reloc_idx, offset, sym, addend,
 919                               elf_text_rela_type(elf));
 920 }
 921
 922 struct reloc *elf_init_reloc_data_sym(struct elf *elf, struct section *sec,
 923                                       unsigned long offset,
 924                                       unsigned int reloc_idx,
 925                                       struct symbol *sym,
 926                                       s64 addend)
 927 {
 928         if (sym->sec && (sec->sh.sh_flags & SHF_EXECINSTR)) {
 929                 WARN("bad call to %s() for text symbol %s",
 930                      __func__, sym->name);
 931                 return NULL;
 932         }
 933
 934         return elf_init_reloc(elf, sec->rsec, reloc_idx, offset, sym, addend,
 935                               elf_data_rela_type(elf));
 936 }
 937
 938 static int read_relocs(struct elf *elf)
 939 {
 940         unsigned long nr_reloc, max_reloc = 0;
 941         struct section *rsec;
 942         struct reloc *reloc;
 943         unsigned int symndx;
 944         struct symbol *sym;
 945         int i;
 946
 947         if (!elf_alloc_hash(reloc, elf->num_relocs))
 948                 return -1;
 949
 950         list_for_each_entry(rsec, &elf->sections, list) {
 951                 if (!is_reloc_sec(rsec))
 952                         continue;
 953
 954                 rsec->base = find_section_by_index(elf, rsec->sh.sh_info);
 955                 if (!rsec->base) {
 956                         WARN("can't find base section for reloc section %s",
 957                              rsec->name);
 958                         return -1;
 959                 }
 960
 961                 rsec->base->rsec = rsec;
 962
 963                 nr_reloc = 0;
 964                 rsec->relocs = calloc(sec_num_entries(rsec), sizeof(*reloc));
 965                 if (!rsec->relocs) {
 966                         perror("calloc");
 967                         return -1;
 968                 }
 969                 for (i = 0; i < sec_num_entries(rsec); i++) {
 970                         reloc = &rsec->relocs[i];
 971
 972                         reloc->sec = rsec;
 973                         symndx = reloc_sym(reloc);
 974                         reloc->sym = sym = find_symbol_by_index(elf, symndx);
 975                         if (!reloc->sym) {
 976                                 WARN("can't find reloc entry symbol %d for %s",
 977                                      symndx, rsec->name);
 978                                 return -1;
 979                         }
 980
 981                         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 982                         reloc->sym_next_reloc = sym->relocs;
 983                         sym->relocs = reloc;
 984
 985                         nr_reloc++;
 986                 }
 987                 max_reloc = max(max_reloc, nr_reloc);
 988         }
 989
 990         if (opts.stats) {
 991                 printf("max_reloc: %lu\n", max_reloc);
 992                 printf("num_relocs: %lu\n", elf->num_relocs);
 993                 printf("reloc_bits: %d\n", elf->reloc_bits);
 994         }
 995
 996         return 0;
 997 }
 998
 999 struct elf *elf_open_read(const char *name, int flags)
1000 {
1001         struct elf *elf;
1002         Elf_Cmd cmd;
1003
1004         elf_version(EV_CURRENT);
1005
1006         elf = malloc(sizeof(*elf));
1007         if (!elf) {
1008                 perror("malloc");
1009                 return NULL;
1010         }
1011         memset(elf, 0, sizeof(*elf));
1012
1013         INIT_LIST_HEAD(&elf->sections);
1014
1015         elf->fd = open(name, flags);
1016         if (elf->fd == -1) {
1017                 fprintf(stderr, "objtool: Can't open '%s': %s\n",
1018                         name, strerror(errno));
1019                 goto err;
1020         }
1021
1022         if ((flags & O_ACCMODE) == O_RDONLY)
1023                 cmd = ELF_C_READ_MMAP;
1024         else if ((flags & O_ACCMODE) == O_RDWR)
1025                 cmd = ELF_C_RDWR;
1026         else /* O_WRONLY */
1027                 cmd = ELF_C_WRITE;
1028
1029         elf->elf = elf_begin(elf->fd, cmd, NULL);
1030         if (!elf->elf) {
1031                 WARN_ELF("elf_begin");
1032                 goto err;
1033         }
1034
1035         if (!gelf_getehdr(elf->elf, &elf->ehdr)) {
1036                 WARN_ELF("gelf_getehdr");
1037                 goto err;
1038         }
1039
1040         if (read_sections(elf))
1041                 goto err;
1042
1043         if (read_symbols(elf))
1044                 goto err;
1045
1046         if (read_relocs(elf))
1047                 goto err;
1048
1049         return elf;
1050
1051 err:
1052         elf_close(elf);
1053         return NULL;
1054 }
1055
1056 static int elf_add_string(struct elf *elf, struct section *strtab, char *str)
1057 {
1058         Elf_Data *data;
1059         Elf_Scn *s;
1060         int len;
1061
1062         if (!strtab)
1063                 strtab = find_section_by_name(elf, ".strtab");
1064         if (!strtab) {
1065                 WARN("can't find .strtab section");
1066                 return -1;
1067         }
1068
1069         s = elf_getscn(elf->elf, strtab->idx);
1070         if (!s) {
1071                 WARN_ELF("elf_getscn");
1072                 return -1;
1073         }
1074
1075         data = elf_newdata(s);
1076         if (!data) {
1077                 WARN_ELF("elf_newdata");
1078                 return -1;
1079         }
1080
1081         data->d_buf = str;
1082         data->d_size = strlen(str) + 1;
1083         data->d_align = 1;
1084
1085         len = strtab->sh.sh_size;
1086         strtab->sh.sh_size += data->d_size;
1087
1088         mark_sec_changed(elf, strtab, true);
1089
1090         return len;
1091 }
1092
1093 struct section *elf_create_section(struct elf *elf, const char *name,
1094                                    size_t entsize, unsigned int nr)
1095 {
1096         struct section *sec, *shstrtab;
1097         size_t size = entsize * nr;
1098         Elf_Scn *s;
1099
1100         sec = malloc(sizeof(*sec));
1101         if (!sec) {
1102                 perror("malloc");
1103                 return NULL;
1104         }
1105         memset(sec, 0, sizeof(*sec));
1106
1107         INIT_LIST_HEAD(&sec->symbol_list);
1108
1109         s = elf_newscn(elf->elf);
1110         if (!s) {
1111                 WARN_ELF("elf_newscn");
1112                 return NULL;
1113         }
1114
1115         sec->name = strdup(name);
1116         if (!sec->name) {
1117                 perror("strdup");
1118                 return NULL;
1119         }
1120
1121         sec->idx = elf_ndxscn(s);
1122
1123         sec->data = elf_newdata(s);
1124         if (!sec->data) {
1125                 WARN_ELF("elf_newdata");
1126                 return NULL;
1127         }
1128
1129         sec->data->d_size = size;
1130         sec->data->d_align = 1;
1131
1132         if (size) {
1133                 sec->data->d_buf = malloc(size);
1134                 if (!sec->data->d_buf) {
1135                         perror("malloc");
1136                         return NULL;
1137                 }
1138                 memset(sec->data->d_buf, 0, size);
1139         }
1140
1141         if (!gelf_getshdr(s, &sec->sh)) {
1142                 WARN_ELF("gelf_getshdr");
1143                 return NULL;
1144         }
1145
1146         sec->sh.sh_size = size;
1147         sec->sh.sh_entsize = entsize;
1148         sec->sh.sh_type = SHT_PROGBITS;
1149         sec->sh.sh_addralign = 1;
1150         sec->sh.sh_flags = SHF_ALLOC;
1151
1152         /* Add section name to .shstrtab (or .strtab for Clang) */
1153         shstrtab = find_section_by_name(elf, ".shstrtab");
1154         if (!shstrtab)
1155                 shstrtab = find_section_by_name(elf, ".strtab");
1156         if (!shstrtab) {
1157                 WARN("can't find .shstrtab or .strtab section");
1158                 return NULL;
1159         }
1160         sec->sh.sh_name = elf_add_string(elf, shstrtab, sec->name);
1161         if (sec->sh.sh_name == -1)
1162                 return NULL;
1163
1164         list_add_tail(&sec->list, &elf->sections);
1165         elf_hash_add(section, &sec->hash, sec->idx);
1166         elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
1167
1168         mark_sec_changed(elf, sec, true);
1169
1170         return sec;
1171 }
1172
1173 static struct section *elf_create_rela_section(struct elf *elf,
1174                                                struct section *sec,
1175                                                unsigned int reloc_nr)
1176 {
1177         struct section *rsec;
1178         char *rsec_name;
1179
1180         rsec_name = malloc(strlen(sec->name) + strlen(".rela") + 1);
1181         if (!rsec_name) {
1182                 perror("malloc");
1183                 return NULL;
1184         }
1185         strcpy(rsec_name, ".rela");
1186         strcat(rsec_name, sec->name);
1187
1188         rsec = elf_create_section(elf, rsec_name, elf_rela_size(elf), reloc_nr);
1189         free(rsec_name);
1190         if (!rsec)
1191                 return NULL;
1192
1193         rsec->data->d_type = ELF_T_RELA;
1194         rsec->sh.sh_type = SHT_RELA;
1195         rsec->sh.sh_addralign = elf_addr_size(elf);
1196         rsec->sh.sh_link = find_section_by_name(elf, ".symtab")->idx;
1197         rsec->sh.sh_info = sec->idx;
1198         rsec->sh.sh_flags = SHF_INFO_LINK;
1199
1200         rsec->relocs = calloc(sec_num_entries(rsec), sizeof(struct reloc));
1201         if (!rsec->relocs) {
1202                 perror("calloc");
1203                 return NULL;
1204         }
1205
1206         sec->rsec = rsec;
1207         rsec->base = sec;
1208
1209         return rsec;
1210 }
1211
1212 struct section *elf_create_section_pair(struct elf *elf, const char *name,
1213                                         size_t entsize, unsigned int nr,
1214                                         unsigned int reloc_nr)
1215 {
1216         struct section *sec;
1217
1218         sec = elf_create_section(elf, name, entsize, nr);
1219         if (!sec)
1220                 return NULL;
1221
1222         if (!elf_create_rela_section(elf, sec, reloc_nr))
1223                 return NULL;
1224
1225         return sec;
1226 }
1227
1228 int elf_write_insn(struct elf *elf, struct section *sec,
1229                    unsigned long offset, unsigned int len,
1230                    const char *insn)
1231 {
1232         Elf_Data *data = sec->data;
1233
1234         if (data->d_type != ELF_T_BYTE || data->d_off) {
1235                 WARN("write to unexpected data for section: %s", sec->name);
1236                 return -1;
1237         }
1238
1239         memcpy(data->d_buf + offset, insn, len);
1240
1241         mark_sec_changed(elf, sec, true);
1242
1243         return 0;
1244 }
1245
1246 /*
1247  * When Elf_Scn::sh_size is smaller than the combined Elf_Data::d_size
1248  * do you:
1249  *
1250  *   A) adhere to the section header and truncate the data, or
1251  *   B) ignore the section header and write out all the data you've got?
1252  *
1253  * Yes, libelf sucks and we need to manually truncate if we over-allocate data.
1254  */
1255 static int elf_truncate_section(struct elf *elf, struct section *sec)
1256 {
1257         u64 size = sec->sh.sh_size;
1258         bool truncated = false;
1259         Elf_Data *data = NULL;
1260         Elf_Scn *s;
1261
1262         s = elf_getscn(elf->elf, sec->idx);
1263         if (!s) {
1264                 WARN_ELF("elf_getscn");
1265                 return -1;
1266         }
1267
1268         for (;;) {
1269                 /* get next data descriptor for the relevant section */
1270                 data = elf_getdata(s, data);
1271
1272                 if (!data) {
1273                         if (size) {
1274                                 WARN("end of section data but non-zero size left\n");
1275                                 return -1;
1276                         }
1277                         return 0;
1278                 }
1279
1280                 if (truncated) {
1281                         /* when we remove symbols */
1282                         WARN("truncated; but more data\n");
1283                         return -1;
1284                 }
1285
1286                 if (!data->d_size) {
1287                         WARN("zero size data");
1288                         return -1;
1289                 }
1290
1291                 if (data->d_size > size) {
1292                         truncated = true;
1293                         data->d_size = size;
1294                 }
1295
1296                 size -= data->d_size;
1297         }
1298 }
1299
1300 int elf_write(struct elf *elf)
1301 {
1302         struct section *sec;
1303         Elf_Scn *s;
1304
1305         if (opts.dryrun)
1306                 return 0;
1307
1308         /* Update changed relocation sections and section headers: */
1309         list_for_each_entry(sec, &elf->sections, list) {
1310                 if (sec->truncate)
1311                         elf_truncate_section(elf, sec);
1312
1313                 if (sec_changed(sec)) {
1314                         s = elf_getscn(elf->elf, sec->idx);
1315                         if (!s) {
1316                                 WARN_ELF("elf_getscn");
1317                                 return -1;
1318                         }
1319
1320                         /* Note this also flags the section dirty */
1321                         if (!gelf_update_shdr(s, &sec->sh)) {
1322                                 WARN_ELF("gelf_update_shdr");
1323                                 return -1;
1324                         }
1325
1326                         mark_sec_changed(elf, sec, false);
1327                 }
1328         }
1329
1330         /* Make sure the new section header entries get updated properly. */
1331         elf_flagelf(elf->elf, ELF_C_SET, ELF_F_DIRTY);
1332
1333         /* Write all changes to the file. */
1334         if (elf_update(elf->elf, ELF_C_WRITE) < 0) {
1335                 WARN_ELF("elf_update");
1336                 return -1;
1337         }
1338
1339         elf->changed = false;
1340
1341         return 0;
1342 }
1343
1344 void elf_close(struct elf *elf)
1345 {
1346         if (elf->elf)
1347                 elf_end(elf->elf);
1348
1349         if (elf->fd > 0)
1350                 close(elf->fd);
1351
1352         /*
1353          * NOTE: All remaining allocations are leaked on purpose.  Objtool is
1354          * about to exit anyway.
1355          */
1356 }