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