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 section *find_section_by_name(const struct elf *elf, const char *name)
  87 {
  88         struct section *sec;
  89
  90         elf_hash_for_each_possible(section_name, sec, name_hash, str_hash(name)) {
  91                 if (!strcmp(sec->name, name))
  92                         return sec;
  93         }
  94
  95         return NULL;
  96 }
  97
  98 static struct section *find_section_by_index(struct elf *elf,
  99                                              unsigned int idx)
 100 {
 101         struct section *sec;
 102
 103         elf_hash_for_each_possible(section, sec, hash, idx) {
 104                 if (sec->idx == idx)
 105                         return sec;
 106         }
 107
 108         return NULL;
 109 }
 110
 111 static struct symbol *find_symbol_by_index(struct elf *elf, unsigned int idx)
 112 {
 113         struct symbol *sym;
 114
 115         elf_hash_for_each_possible(symbol, sym, hash, idx) {
 116                 if (sym->idx == idx)
 117                         return sym;
 118         }
 119
 120         return NULL;
 121 }
 122
 123 struct symbol *find_symbol_by_offset(struct section *sec, unsigned long offset)
 124 {
 125         struct rb_node *node;
 126
 127         rb_for_each(node, &offset, &sec->symbol_tree, symbol_by_offset) {
 128                 struct symbol *s = rb_entry(node, struct symbol, node);
 129
 130                 if (s->offset == offset && s->type != STT_SECTION)
 131                         return s;
 132         }
 133
 134         return NULL;
 135 }
 136
 137 struct symbol *find_func_by_offset(struct section *sec, unsigned long offset)
 138 {
 139         struct rb_node *node;
 140
 141         rb_for_each(node, &offset, &sec->symbol_tree, symbol_by_offset) {
 142                 struct symbol *s = rb_entry(node, struct symbol, node);
 143
 144                 if (s->offset == offset && s->type == STT_FUNC)
 145                         return s;
 146         }
 147
 148         return NULL;
 149 }
 150
 151 struct symbol *find_symbol_containing(const struct section *sec, unsigned long offset)
 152 {
 153         struct rb_node *node;
 154
 155         rb_for_each(node, &offset, &sec->symbol_tree, symbol_by_offset) {
 156                 struct symbol *s = rb_entry(node, struct symbol, node);
 157
 158                 if (s->type != STT_SECTION)
 159                         return s;
 160         }
 161
 162         return NULL;
 163 }
 164
 165 struct symbol *find_func_containing(struct section *sec, unsigned long offset)
 166 {
 167         struct rb_node *node;
 168
 169         rb_for_each(node, &offset, &sec->symbol_tree, symbol_by_offset) {
 170                 struct symbol *s = rb_entry(node, struct symbol, node);
 171
 172                 if (s->type == STT_FUNC)
 173                         return s;
 174         }
 175
 176         return NULL;
 177 }
 178
 179 struct symbol *find_symbol_by_name(const struct elf *elf, const char *name)
 180 {
 181         struct symbol *sym;
 182
 183         elf_hash_for_each_possible(symbol_name, sym, name_hash, str_hash(name)) {
 184                 if (!strcmp(sym->name, name))
 185                         return sym;
 186         }
 187
 188         return NULL;
 189 }
 190
 191 struct reloc *find_reloc_by_dest_range(const struct elf *elf, struct section *sec,
 192                                      unsigned long offset, unsigned int len)
 193 {
 194         struct reloc *reloc, *r = NULL;
 195         unsigned long o;
 196
 197         if (!sec->reloc)
 198                 return NULL;
 199
 200         sec = sec->reloc;
 201
 202         for_offset_range(o, offset, offset + len) {
 203                 elf_hash_for_each_possible(reloc, reloc, hash,
 204                                            sec_offset_hash(sec, o)) {
 205                         if (reloc->sec != sec)
 206                                 continue;
 207
 208                         if (reloc->offset >= offset && reloc->offset < offset + len) {
 209                                 if (!r || reloc->offset < r->offset)
 210                                         r = reloc;
 211                         }
 212                 }
 213                 if (r)
 214                         return r;
 215         }
 216
 217         return NULL;
 218 }
 219
 220 struct reloc *find_reloc_by_dest(const struct elf *elf, struct section *sec, unsigned long offset)
 221 {
 222         return find_reloc_by_dest_range(elf, sec, offset, 1);
 223 }
 224
 225 static int read_sections(struct elf *elf)
 226 {
 227         Elf_Scn *s = NULL;
 228         struct section *sec;
 229         size_t shstrndx, sections_nr;
 230         int i;
 231
 232         if (elf_getshdrnum(elf->elf, &sections_nr)) {
 233                 WARN_ELF("elf_getshdrnum");
 234                 return -1;
 235         }
 236
 237         if (elf_getshdrstrndx(elf->elf, &shstrndx)) {
 238                 WARN_ELF("elf_getshdrstrndx");
 239                 return -1;
 240         }
 241
 242         if (!elf_alloc_hash(section, sections_nr) ||
 243             !elf_alloc_hash(section_name, sections_nr))
 244                 return -1;
 245
 246         for (i = 0; i < sections_nr; i++) {
 247                 sec = malloc(sizeof(*sec));
 248                 if (!sec) {
 249                         perror("malloc");
 250                         return -1;
 251                 }
 252                 memset(sec, 0, sizeof(*sec));
 253
 254                 INIT_LIST_HEAD(&sec->symbol_list);
 255                 INIT_LIST_HEAD(&sec->reloc_list);
 256
 257                 s = elf_getscn(elf->elf, i);
 258                 if (!s) {
 259                         WARN_ELF("elf_getscn");
 260                         return -1;
 261                 }
 262
 263                 sec->idx = elf_ndxscn(s);
 264
 265                 if (!gelf_getshdr(s, &sec->sh)) {
 266                         WARN_ELF("gelf_getshdr");
 267                         return -1;
 268                 }
 269
 270                 sec->name = elf_strptr(elf->elf, shstrndx, sec->sh.sh_name);
 271                 if (!sec->name) {
 272                         WARN_ELF("elf_strptr");
 273                         return -1;
 274                 }
 275
 276                 if (sec->sh.sh_size != 0) {
 277                         sec->data = elf_getdata(s, NULL);
 278                         if (!sec->data) {
 279                                 WARN_ELF("elf_getdata");
 280                                 return -1;
 281                         }
 282                         if (sec->data->d_off != 0 ||
 283                             sec->data->d_size != sec->sh.sh_size) {
 284                                 WARN("unexpected data attributes for %s",
 285                                      sec->name);
 286                                 return -1;
 287                         }
 288                 }
 289
 290                 if (sec->sh.sh_flags & SHF_EXECINSTR)
 291                         elf->text_size += sec->sh.sh_size;
 292
 293                 list_add_tail(&sec->list, &elf->sections);
 294                 elf_hash_add(section, &sec->hash, sec->idx);
 295                 elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
 296         }
 297
 298         if (stats) {
 299                 printf("nr_sections: %lu\n", (unsigned long)sections_nr);
 300                 printf("section_bits: %d\n", elf->section_bits);
 301         }
 302
 303         /* sanity check, one more call to elf_nextscn() should return NULL */
 304         if (elf_nextscn(elf->elf, s)) {
 305                 WARN("section entry mismatch");
 306                 return -1;
 307         }
 308
 309         return 0;
 310 }
 311
 312 static void elf_add_symbol(struct elf *elf, struct symbol *sym)
 313 {
 314         struct list_head *entry;
 315         struct rb_node *pnode;
 316
 317         sym->alias = sym;
 318
 319         sym->type = GELF_ST_TYPE(sym->sym.st_info);
 320         sym->bind = GELF_ST_BIND(sym->sym.st_info);
 321
 322         sym->offset = sym->sym.st_value;
 323         sym->len = sym->sym.st_size;
 324
 325         rb_add(&sym->node, &sym->sec->symbol_tree, symbol_to_offset);
 326         pnode = rb_prev(&sym->node);
 327         if (pnode)
 328                 entry = &rb_entry(pnode, struct symbol, node)->list;
 329         else
 330                 entry = &sym->sec->symbol_list;
 331         list_add(&sym->list, entry);
 332         elf_hash_add(symbol, &sym->hash, sym->idx);
 333         elf_hash_add(symbol_name, &sym->name_hash, str_hash(sym->name));
 334
 335         /*
 336          * Don't store empty STT_NOTYPE symbols in the rbtree.  They
 337          * can exist within a function, confusing the sorting.
 338          */
 339         if (!sym->len)
 340                 rb_erase(&sym->node, &sym->sec->symbol_tree);
 341 }
 342
 343 static int read_symbols(struct elf *elf)
 344 {
 345         struct section *symtab, *symtab_shndx, *sec;
 346         struct symbol *sym, *pfunc;
 347         int symbols_nr, i;
 348         char *coldstr;
 349         Elf_Data *shndx_data = NULL;
 350         Elf32_Word shndx;
 351
 352         symtab = find_section_by_name(elf, ".symtab");
 353         if (symtab) {
 354                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 355                 if (symtab_shndx)
 356                         shndx_data = symtab_shndx->data;
 357
 358                 symbols_nr = symtab->sh.sh_size / symtab->sh.sh_entsize;
 359         } else {
 360                 /*
 361                  * A missing symbol table is actually possible if it's an empty
 362                  * .o file. This can happen for thunk_64.o. Make sure to at
 363                  * least allocate the symbol hash tables so we can do symbol
 364                  * lookups without crashing.
 365                  */
 366                 symbols_nr = 0;
 367         }
 368
 369         if (!elf_alloc_hash(symbol, symbols_nr) ||
 370             !elf_alloc_hash(symbol_name, symbols_nr))
 371                 return -1;
 372
 373         for (i = 0; i < symbols_nr; i++) {
 374                 sym = malloc(sizeof(*sym));
 375                 if (!sym) {
 376                         perror("malloc");
 377                         return -1;
 378                 }
 379                 memset(sym, 0, sizeof(*sym));
 380
 381                 sym->idx = i;
 382
 383                 if (!gelf_getsymshndx(symtab->data, shndx_data, i, &sym->sym,
 384                                       &shndx)) {
 385                         WARN_ELF("gelf_getsymshndx");
 386                         goto err;
 387                 }
 388
 389                 sym->name = elf_strptr(elf->elf, symtab->sh.sh_link,
 390                                        sym->sym.st_name);
 391                 if (!sym->name) {
 392                         WARN_ELF("elf_strptr");
 393                         goto err;
 394                 }
 395
 396                 if ((sym->sym.st_shndx > SHN_UNDEF &&
 397                      sym->sym.st_shndx < SHN_LORESERVE) ||
 398                     (shndx_data && sym->sym.st_shndx == SHN_XINDEX)) {
 399                         if (sym->sym.st_shndx != SHN_XINDEX)
 400                                 shndx = sym->sym.st_shndx;
 401
 402                         sym->sec = find_section_by_index(elf, shndx);
 403                         if (!sym->sec) {
 404                                 WARN("couldn't find section for symbol %s",
 405                                      sym->name);
 406                                 goto err;
 407                         }
 408                         if (GELF_ST_TYPE(sym->sym.st_info) == STT_SECTION) {
 409                                 sym->name = sym->sec->name;
 410                                 sym->sec->sym = sym;
 411                         }
 412                 } else
 413                         sym->sec = find_section_by_index(elf, 0);
 414
 415                 elf_add_symbol(elf, sym);
 416         }
 417
 418         if (stats) {
 419                 printf("nr_symbols: %lu\n", (unsigned long)symbols_nr);
 420                 printf("symbol_bits: %d\n", elf->symbol_bits);
 421         }
 422
 423         /* Create parent/child links for any cold subfunctions */
 424         list_for_each_entry(sec, &elf->sections, list) {
 425                 list_for_each_entry(sym, &sec->symbol_list, list) {
 426                         char pname[MAX_NAME_LEN + 1];
 427                         size_t pnamelen;
 428                         if (sym->type != STT_FUNC)
 429                                 continue;
 430
 431                         if (sym->pfunc == NULL)
 432                                 sym->pfunc = sym;
 433
 434                         if (sym->cfunc == NULL)
 435                                 sym->cfunc = sym;
 436
 437                         coldstr = strstr(sym->name, ".cold");
 438                         if (!coldstr)
 439                                 continue;
 440
 441                         pnamelen = coldstr - sym->name;
 442                         if (pnamelen > MAX_NAME_LEN) {
 443                                 WARN("%s(): parent function name exceeds maximum length of %d characters",
 444                                      sym->name, MAX_NAME_LEN);
 445                                 return -1;
 446                         }
 447
 448                         strncpy(pname, sym->name, pnamelen);
 449                         pname[pnamelen] = '\0';
 450                         pfunc = find_symbol_by_name(elf, pname);
 451
 452                         if (!pfunc) {
 453                                 WARN("%s(): can't find parent function",
 454                                      sym->name);
 455                                 return -1;
 456                         }
 457
 458                         sym->pfunc = pfunc;
 459                         pfunc->cfunc = sym;
 460
 461                         /*
 462                          * Unfortunately, -fnoreorder-functions puts the child
 463                          * inside the parent.  Remove the overlap so we can
 464                          * have sane assumptions.
 465                          *
 466                          * Note that pfunc->len now no longer matches
 467                          * pfunc->sym.st_size.
 468                          */
 469                         if (sym->sec == pfunc->sec &&
 470                             sym->offset >= pfunc->offset &&
 471                             sym->offset + sym->len == pfunc->offset + pfunc->len) {
 472                                 pfunc->len -= sym->len;
 473                         }
 474                 }
 475         }
 476
 477         return 0;
 478
 479 err:
 480         free(sym);
 481         return -1;
 482 }
 483
 484 static struct section *elf_create_reloc_section(struct elf *elf,
 485                                                 struct section *base,
 486                                                 int reltype);
 487
 488 int elf_add_reloc(struct elf *elf, struct section *sec, unsigned long offset,
 489                   unsigned int type, struct symbol *sym, s64 addend)
 490 {
 491         struct reloc *reloc;
 492
 493         if (!sec->reloc && !elf_create_reloc_section(elf, sec, SHT_RELA))
 494                 return -1;
 495
 496         reloc = malloc(sizeof(*reloc));
 497         if (!reloc) {
 498                 perror("malloc");
 499                 return -1;
 500         }
 501         memset(reloc, 0, sizeof(*reloc));
 502
 503         reloc->sec = sec->reloc;
 504         reloc->offset = offset;
 505         reloc->type = type;
 506         reloc->sym = sym;
 507         reloc->addend = addend;
 508
 509         list_add_tail(&reloc->list, &sec->reloc->reloc_list);
 510         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 511
 512         sec->reloc->sh.sh_size += sec->reloc->sh.sh_entsize;
 513         sec->reloc->changed = true;
 514
 515         return 0;
 516 }
 517
 518 /*
 519  * Ensure that any reloc section containing references to @sym is marked
 520  * changed such that it will get re-generated in elf_rebuild_reloc_sections()
 521  * with the new symbol index.
 522  */
 523 static void elf_dirty_reloc_sym(struct elf *elf, struct symbol *sym)
 524 {
 525         struct section *sec;
 526
 527         list_for_each_entry(sec, &elf->sections, list) {
 528                 struct reloc *reloc;
 529
 530                 if (sec->changed)
 531                         continue;
 532
 533                 list_for_each_entry(reloc, &sec->reloc_list, list) {
 534                         if (reloc->sym == sym) {
 535                                 sec->changed = true;
 536                                 break;
 537                         }
 538                 }
 539         }
 540 }
 541
 542 /*
 543  * The libelf API is terrible; gelf_update_sym*() takes a data block relative
 544  * index value, *NOT* the symbol index. As such, iterate the data blocks and
 545  * adjust index until it fits.
 546  *
 547  * If no data block is found, allow adding a new data block provided the index
 548  * is only one past the end.
 549  */
 550 static int elf_update_symbol(struct elf *elf, struct section *symtab,
 551                              struct section *symtab_shndx, struct symbol *sym)
 552 {
 553         Elf32_Word shndx = sym->sec ? sym->sec->idx : SHN_UNDEF;
 554         Elf_Data *symtab_data = NULL, *shndx_data = NULL;
 555         Elf64_Xword entsize = symtab->sh.sh_entsize;
 556         int max_idx, idx = sym->idx;
 557         Elf_Scn *s, *t = NULL;
 558
 559         s = elf_getscn(elf->elf, symtab->idx);
 560         if (!s) {
 561                 WARN_ELF("elf_getscn");
 562                 return -1;
 563         }
 564
 565         if (symtab_shndx) {
 566                 t = elf_getscn(elf->elf, symtab_shndx->idx);
 567                 if (!t) {
 568                         WARN_ELF("elf_getscn");
 569                         return -1;
 570                 }
 571         }
 572
 573         for (;;) {
 574                 /* get next data descriptor for the relevant sections */
 575                 symtab_data = elf_getdata(s, symtab_data);
 576                 if (t)
 577                         shndx_data = elf_getdata(t, shndx_data);
 578
 579                 /* end-of-list */
 580                 if (!symtab_data) {
 581                         void *buf;
 582
 583                         if (idx) {
 584                                 /* we don't do holes in symbol tables */
 585                                 WARN("index out of range");
 586                                 return -1;
 587                         }
 588
 589                         /* if @idx == 0, it's the next contiguous entry, create it */
 590                         symtab_data = elf_newdata(s);
 591                         if (t)
 592                                 shndx_data = elf_newdata(t);
 593
 594                         buf = calloc(1, entsize);
 595                         if (!buf) {
 596                                 WARN("malloc");
 597                                 return -1;
 598                         }
 599
 600                         symtab_data->d_buf = buf;
 601                         symtab_data->d_size = entsize;
 602                         symtab_data->d_align = 1;
 603                         symtab_data->d_type = ELF_T_SYM;
 604
 605                         symtab->sh.sh_size += entsize;
 606                         symtab->changed = true;
 607
 608                         if (t) {
 609                                 shndx_data->d_buf = &sym->sec->idx;
 610                                 shndx_data->d_size = sizeof(Elf32_Word);
 611                                 shndx_data->d_align = sizeof(Elf32_Word);
 612                                 shndx_data->d_type = ELF_T_WORD;
 613
 614                                 symtab_shndx->sh.sh_size += sizeof(Elf32_Word);
 615                                 symtab_shndx->changed = true;
 616                         }
 617
 618                         break;
 619                 }
 620
 621                 /* empty blocks should not happen */
 622                 if (!symtab_data->d_size) {
 623                         WARN("zero size data");
 624                         return -1;
 625                 }
 626
 627                 /* is this the right block? */
 628                 max_idx = symtab_data->d_size / entsize;
 629                 if (idx < max_idx)
 630                         break;
 631
 632                 /* adjust index and try again */
 633                 idx -= max_idx;
 634         }
 635
 636         /* something went side-ways */
 637         if (idx < 0) {
 638                 WARN("negative index");
 639                 return -1;
 640         }
 641
 642         /* setup extended section index magic and write the symbol */
 643         if (shndx >= SHN_UNDEF && shndx < SHN_LORESERVE) {
 644                 sym->sym.st_shndx = shndx;
 645                 if (!shndx_data)
 646                         shndx = 0;
 647         } else {
 648                 sym->sym.st_shndx = SHN_XINDEX;
 649                 if (!shndx_data) {
 650                         WARN("no .symtab_shndx");
 651                         return -1;
 652                 }
 653         }
 654
 655         if (!gelf_update_symshndx(symtab_data, shndx_data, idx, &sym->sym, shndx)) {
 656                 WARN_ELF("gelf_update_symshndx");
 657                 return -1;
 658         }
 659
 660         return 0;
 661 }
 662
 663 static struct symbol *
 664 elf_create_section_symbol(struct elf *elf, struct section *sec)
 665 {
 666         struct section *symtab, *symtab_shndx;
 667         Elf32_Word first_non_local, new_idx;
 668         struct symbol *sym, *old;
 669
 670         symtab = find_section_by_name(elf, ".symtab");
 671         if (symtab) {
 672                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 673         } else {
 674                 WARN("no .symtab");
 675                 return NULL;
 676         }
 677
 678         sym = calloc(1, sizeof(*sym));
 679         if (!sym) {
 680                 perror("malloc");
 681                 return NULL;
 682         }
 683
 684         sym->name = sec->name;
 685         sym->sec = sec;
 686
 687         // st_name 0
 688         sym->sym.st_info = GELF_ST_INFO(STB_LOCAL, STT_SECTION);
 689         // st_other 0
 690         // st_value 0
 691         // st_size 0
 692
 693         /*
 694          * Move the first global symbol, as per sh_info, into a new, higher
 695          * symbol index. This fees up a spot for a new local symbol.
 696          */
 697         first_non_local = symtab->sh.sh_info;
 698         new_idx = symtab->sh.sh_size / symtab->sh.sh_entsize;
 699         old = find_symbol_by_index(elf, first_non_local);
 700         if (old) {
 701                 old->idx = new_idx;
 702
 703                 hlist_del(&old->hash);
 704                 elf_hash_add(symbol, &old->hash, old->idx);
 705
 706                 elf_dirty_reloc_sym(elf, old);
 707
 708                 if (elf_update_symbol(elf, symtab, symtab_shndx, old)) {
 709                         WARN("elf_update_symbol move");
 710                         return NULL;
 711                 }
 712
 713                 new_idx = first_non_local;
 714         }
 715
 716         sym->idx = new_idx;
 717         if (elf_update_symbol(elf, symtab, symtab_shndx, sym)) {
 718                 WARN("elf_update_symbol");
 719                 return NULL;
 720         }
 721
 722         /*
 723          * Either way, we added a LOCAL symbol.
 724          */
 725         symtab->sh.sh_info += 1;
 726
 727         elf_add_symbol(elf, sym);
 728
 729         return sym;
 730 }
 731
 732 int elf_add_reloc_to_insn(struct elf *elf, struct section *sec,
 733                           unsigned long offset, unsigned int type,
 734                           struct section *insn_sec, unsigned long insn_off)
 735 {
 736         struct symbol *sym = insn_sec->sym;
 737         int addend = insn_off;
 738
 739         if (!sym) {
 740                 /*
 741                  * Due to how weak functions work, we must use section based
 742                  * relocations. Symbol based relocations would result in the
 743                  * weak and non-weak function annotations being overlaid on the
 744                  * non-weak function after linking.
 745                  */
 746                 sym = elf_create_section_symbol(elf, insn_sec);
 747                 if (!sym)
 748                         return -1;
 749
 750                 insn_sec->sym = sym;
 751         }
 752
 753         return elf_add_reloc(elf, sec, offset, type, sym, addend);
 754 }
 755
 756 static int read_rel_reloc(struct section *sec, int i, struct reloc *reloc, unsigned int *symndx)
 757 {
 758         if (!gelf_getrel(sec->data, i, &reloc->rel)) {
 759                 WARN_ELF("gelf_getrel");
 760                 return -1;
 761         }
 762         reloc->type = GELF_R_TYPE(reloc->rel.r_info);
 763         reloc->addend = 0;
 764         reloc->offset = reloc->rel.r_offset;
 765         *symndx = GELF_R_SYM(reloc->rel.r_info);
 766         return 0;
 767 }
 768
 769 static int read_rela_reloc(struct section *sec, int i, struct reloc *reloc, unsigned int *symndx)
 770 {
 771         if (!gelf_getrela(sec->data, i, &reloc->rela)) {
 772                 WARN_ELF("gelf_getrela");
 773                 return -1;
 774         }
 775         reloc->type = GELF_R_TYPE(reloc->rela.r_info);
 776         reloc->addend = reloc->rela.r_addend;
 777         reloc->offset = reloc->rela.r_offset;
 778         *symndx = GELF_R_SYM(reloc->rela.r_info);
 779         return 0;
 780 }
 781
 782 static int read_relocs(struct elf *elf)
 783 {
 784         struct section *sec;
 785         struct reloc *reloc;
 786         int i;
 787         unsigned int symndx;
 788         unsigned long nr_reloc, max_reloc = 0, tot_reloc = 0;
 789
 790         if (!elf_alloc_hash(reloc, elf->text_size / 16))
 791                 return -1;
 792
 793         list_for_each_entry(sec, &elf->sections, list) {
 794                 if ((sec->sh.sh_type != SHT_RELA) &&
 795                     (sec->sh.sh_type != SHT_REL))
 796                         continue;
 797
 798                 sec->base = find_section_by_index(elf, sec->sh.sh_info);
 799                 if (!sec->base) {
 800                         WARN("can't find base section for reloc section %s",
 801                              sec->name);
 802                         return -1;
 803                 }
 804
 805                 sec->base->reloc = sec;
 806
 807                 nr_reloc = 0;
 808                 for (i = 0; i < sec->sh.sh_size / sec->sh.sh_entsize; i++) {
 809                         reloc = malloc(sizeof(*reloc));
 810                         if (!reloc) {
 811                                 perror("malloc");
 812                                 return -1;
 813                         }
 814                         memset(reloc, 0, sizeof(*reloc));
 815                         switch (sec->sh.sh_type) {
 816                         case SHT_REL:
 817                                 if (read_rel_reloc(sec, i, reloc, &symndx))
 818                                         return -1;
 819                                 break;
 820                         case SHT_RELA:
 821                                 if (read_rela_reloc(sec, i, reloc, &symndx))
 822                                         return -1;
 823                                 break;
 824                         default: return -1;
 825                         }
 826
 827                         reloc->sec = sec;
 828                         reloc->idx = i;
 829                         reloc->sym = find_symbol_by_index(elf, symndx);
 830                         if (!reloc->sym) {
 831                                 WARN("can't find reloc entry symbol %d for %s",
 832                                      symndx, sec->name);
 833                                 return -1;
 834                         }
 835
 836                         list_add_tail(&reloc->list, &sec->reloc_list);
 837                         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 838
 839                         nr_reloc++;
 840                 }
 841                 max_reloc = max(max_reloc, nr_reloc);
 842                 tot_reloc += nr_reloc;
 843         }
 844
 845         if (stats) {
 846                 printf("max_reloc: %lu\n", max_reloc);
 847                 printf("tot_reloc: %lu\n", tot_reloc);
 848                 printf("reloc_bits: %d\n", elf->reloc_bits);
 849         }
 850
 851         return 0;
 852 }
 853
 854 struct elf *elf_open_read(const char *name, int flags)
 855 {
 856         struct elf *elf;
 857         Elf_Cmd cmd;
 858
 859         elf_version(EV_CURRENT);
 860
 861         elf = malloc(sizeof(*elf));
 862         if (!elf) {
 863                 perror("malloc");
 864                 return NULL;
 865         }
 866         memset(elf, 0, offsetof(struct elf, sections));
 867
 868         INIT_LIST_HEAD(&elf->sections);
 869
 870         elf->fd = open(name, flags);
 871         if (elf->fd == -1) {
 872                 fprintf(stderr, "objtool: Can't open '%s': %s\n",
 873                         name, strerror(errno));
 874                 goto err;
 875         }
 876
 877         if ((flags & O_ACCMODE) == O_RDONLY)
 878                 cmd = ELF_C_READ_MMAP;
 879         else if ((flags & O_ACCMODE) == O_RDWR)
 880                 cmd = ELF_C_RDWR;
 881         else /* O_WRONLY */
 882                 cmd = ELF_C_WRITE;
 883
 884         elf->elf = elf_begin(elf->fd, cmd, NULL);
 885         if (!elf->elf) {
 886                 WARN_ELF("elf_begin");
 887                 goto err;
 888         }
 889
 890         if (!gelf_getehdr(elf->elf, &elf->ehdr)) {
 891                 WARN_ELF("gelf_getehdr");
 892                 goto err;
 893         }
 894
 895         if (read_sections(elf))
 896                 goto err;
 897
 898         if (read_symbols(elf))
 899                 goto err;
 900
 901         if (read_relocs(elf))
 902                 goto err;
 903
 904         return elf;
 905
 906 err:
 907         elf_close(elf);
 908         return NULL;
 909 }
 910
 911 static int elf_add_string(struct elf *elf, struct section *strtab, char *str)
 912 {
 913         Elf_Data *data;
 914         Elf_Scn *s;
 915         int len;
 916
 917         if (!strtab)
 918                 strtab = find_section_by_name(elf, ".strtab");
 919         if (!strtab) {
 920                 WARN("can't find .strtab section");
 921                 return -1;
 922         }
 923
 924         s = elf_getscn(elf->elf, strtab->idx);
 925         if (!s) {
 926                 WARN_ELF("elf_getscn");
 927                 return -1;
 928         }
 929
 930         data = elf_newdata(s);
 931         if (!data) {
 932                 WARN_ELF("elf_newdata");
 933                 return -1;
 934         }
 935
 936         data->d_buf = str;
 937         data->d_size = strlen(str) + 1;
 938         data->d_align = 1;
 939
 940         len = strtab->sh.sh_size;
 941         strtab->sh.sh_size += data->d_size;
 942         strtab->changed = true;
 943
 944         return len;
 945 }
 946
 947 struct symbol *elf_create_undef_symbol(struct elf *elf, const char *name)
 948 {
 949         struct section *symtab, *symtab_shndx;
 950         struct symbol *sym;
 951         Elf_Data *data;
 952         Elf_Scn *s;
 953
 954         sym = malloc(sizeof(*sym));
 955         if (!sym) {
 956                 perror("malloc");
 957                 return NULL;
 958         }
 959         memset(sym, 0, sizeof(*sym));
 960
 961         sym->name = strdup(name);
 962
 963         sym->sym.st_name = elf_add_string(elf, NULL, sym->name);
 964         if (sym->sym.st_name == -1)
 965                 return NULL;
 966
 967         sym->sym.st_info = GELF_ST_INFO(STB_GLOBAL, STT_NOTYPE);
 968         // st_other 0
 969         // st_shndx 0
 970         // st_value 0
 971         // st_size 0
 972
 973         symtab = find_section_by_name(elf, ".symtab");
 974         if (!symtab) {
 975                 WARN("can't find .symtab");
 976                 return NULL;
 977         }
 978
 979         s = elf_getscn(elf->elf, symtab->idx);
 980         if (!s) {
 981                 WARN_ELF("elf_getscn");
 982                 return NULL;
 983         }
 984
 985         data = elf_newdata(s);
 986         if (!data) {
 987                 WARN_ELF("elf_newdata");
 988                 return NULL;
 989         }
 990
 991         data->d_buf = &sym->sym;
 992         data->d_size = sizeof(sym->sym);
 993         data->d_align = 1;
 994         data->d_type = ELF_T_SYM;
 995
 996         sym->idx = symtab->sh.sh_size / sizeof(sym->sym);
 997
 998         symtab->sh.sh_size += data->d_size;
 999         symtab->changed = true;
1000
1001         symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
1002         if (symtab_shndx) {
1003                 s = elf_getscn(elf->elf, symtab_shndx->idx);
1004                 if (!s) {
1005                         WARN_ELF("elf_getscn");
1006                         return NULL;
1007                 }
1008
1009                 data = elf_newdata(s);
1010                 if (!data) {
1011                         WARN_ELF("elf_newdata");
1012                         return NULL;
1013                 }
1014
1015                 data->d_buf = &sym->sym.st_size; /* conveniently 0 */
1016                 data->d_size = sizeof(Elf32_Word);
1017                 data->d_align = 4;
1018                 data->d_type = ELF_T_WORD;
1019
1020                 symtab_shndx->sh.sh_size += 4;
1021                 symtab_shndx->changed = true;
1022         }
1023
1024         sym->sec = find_section_by_index(elf, 0);
1025
1026         elf_add_symbol(elf, sym);
1027
1028         return sym;
1029 }
1030
1031 struct section *elf_create_section(struct elf *elf, const char *name,
1032                                    unsigned int sh_flags, size_t entsize, int nr)
1033 {
1034         struct section *sec, *shstrtab;
1035         size_t size = entsize * nr;
1036         Elf_Scn *s;
1037
1038         sec = malloc(sizeof(*sec));
1039         if (!sec) {
1040                 perror("malloc");
1041                 return NULL;
1042         }
1043         memset(sec, 0, sizeof(*sec));
1044
1045         INIT_LIST_HEAD(&sec->symbol_list);
1046         INIT_LIST_HEAD(&sec->reloc_list);
1047
1048         s = elf_newscn(elf->elf);
1049         if (!s) {
1050                 WARN_ELF("elf_newscn");
1051                 return NULL;
1052         }
1053
1054         sec->name = strdup(name);
1055         if (!sec->name) {
1056                 perror("strdup");
1057                 return NULL;
1058         }
1059
1060         sec->idx = elf_ndxscn(s);
1061         sec->changed = true;
1062
1063         sec->data = elf_newdata(s);
1064         if (!sec->data) {
1065                 WARN_ELF("elf_newdata");
1066                 return NULL;
1067         }
1068
1069         sec->data->d_size = size;
1070         sec->data->d_align = 1;
1071
1072         if (size) {
1073                 sec->data->d_buf = malloc(size);
1074                 if (!sec->data->d_buf) {
1075                         perror("malloc");
1076                         return NULL;
1077                 }
1078                 memset(sec->data->d_buf, 0, size);
1079         }
1080
1081         if (!gelf_getshdr(s, &sec->sh)) {
1082                 WARN_ELF("gelf_getshdr");
1083                 return NULL;
1084         }
1085
1086         sec->sh.sh_size = size;
1087         sec->sh.sh_entsize = entsize;
1088         sec->sh.sh_type = SHT_PROGBITS;
1089         sec->sh.sh_addralign = 1;
1090         sec->sh.sh_flags = SHF_ALLOC | sh_flags;
1091
1092         /* Add section name to .shstrtab (or .strtab for Clang) */
1093         shstrtab = find_section_by_name(elf, ".shstrtab");
1094         if (!shstrtab)
1095                 shstrtab = find_section_by_name(elf, ".strtab");
1096         if (!shstrtab) {
1097                 WARN("can't find .shstrtab or .strtab section");
1098                 return NULL;
1099         }
1100         sec->sh.sh_name = elf_add_string(elf, shstrtab, sec->name);
1101         if (sec->sh.sh_name == -1)
1102                 return NULL;
1103
1104         list_add_tail(&sec->list, &elf->sections);
1105         elf_hash_add(section, &sec->hash, sec->idx);
1106         elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
1107
1108         elf->changed = true;
1109
1110         return sec;
1111 }
1112
1113 static struct section *elf_create_rel_reloc_section(struct elf *elf, struct section *base)
1114 {
1115         char *relocname;
1116         struct section *sec;
1117
1118         relocname = malloc(strlen(base->name) + strlen(".rel") + 1);
1119         if (!relocname) {
1120                 perror("malloc");
1121                 return NULL;
1122         }
1123         strcpy(relocname, ".rel");
1124         strcat(relocname, base->name);
1125
1126         sec = elf_create_section(elf, relocname, 0, sizeof(GElf_Rel), 0);
1127         free(relocname);
1128         if (!sec)
1129                 return NULL;
1130
1131         base->reloc = sec;
1132         sec->base = base;
1133
1134         sec->sh.sh_type = SHT_REL;
1135         sec->sh.sh_addralign = 8;
1136         sec->sh.sh_link = find_section_by_name(elf, ".symtab")->idx;
1137         sec->sh.sh_info = base->idx;
1138         sec->sh.sh_flags = SHF_INFO_LINK;
1139
1140         return sec;
1141 }
1142
1143 static struct section *elf_create_rela_reloc_section(struct elf *elf, struct section *base)
1144 {
1145         char *relocname;
1146         struct section *sec;
1147
1148         relocname = malloc(strlen(base->name) + strlen(".rela") + 1);
1149         if (!relocname) {
1150                 perror("malloc");
1151                 return NULL;
1152         }
1153         strcpy(relocname, ".rela");
1154         strcat(relocname, base->name);
1155
1156         sec = elf_create_section(elf, relocname, 0, sizeof(GElf_Rela), 0);
1157         free(relocname);
1158         if (!sec)
1159                 return NULL;
1160
1161         base->reloc = sec;
1162         sec->base = base;
1163
1164         sec->sh.sh_type = SHT_RELA;
1165         sec->sh.sh_addralign = 8;
1166         sec->sh.sh_link = find_section_by_name(elf, ".symtab")->idx;
1167         sec->sh.sh_info = base->idx;
1168         sec->sh.sh_flags = SHF_INFO_LINK;
1169
1170         return sec;
1171 }
1172
1173 static struct section *elf_create_reloc_section(struct elf *elf,
1174                                          struct section *base,
1175                                          int reltype)
1176 {
1177         switch (reltype) {
1178         case SHT_REL:  return elf_create_rel_reloc_section(elf, base);
1179         case SHT_RELA: return elf_create_rela_reloc_section(elf, base);
1180         default:       return NULL;
1181         }
1182 }
1183
1184 static int elf_rebuild_rel_reloc_section(struct section *sec)
1185 {
1186         struct reloc *reloc;
1187         int idx = 0;
1188         void *buf;
1189
1190         /* Allocate a buffer for relocations */
1191         buf = malloc(sec->sh.sh_size);
1192         if (!buf) {
1193                 perror("malloc");
1194                 return -1;
1195         }
1196
1197         sec->data->d_buf = buf;
1198         sec->data->d_size = sec->sh.sh_size;
1199         sec->data->d_type = ELF_T_REL;
1200
1201         idx = 0;
1202         list_for_each_entry(reloc, &sec->reloc_list, list) {
1203                 reloc->rel.r_offset = reloc->offset;
1204                 reloc->rel.r_info = GELF_R_INFO(reloc->sym->idx, reloc->type);
1205                 if (!gelf_update_rel(sec->data, idx, &reloc->rel)) {
1206                         WARN_ELF("gelf_update_rel");
1207                         return -1;
1208                 }
1209                 idx++;
1210         }
1211
1212         return 0;
1213 }
1214
1215 static int elf_rebuild_rela_reloc_section(struct section *sec)
1216 {
1217         struct reloc *reloc;
1218         int idx = 0;
1219         void *buf;
1220
1221         /* Allocate a buffer for relocations with addends */
1222         buf = malloc(sec->sh.sh_size);
1223         if (!buf) {
1224                 perror("malloc");
1225                 return -1;
1226         }
1227
1228         sec->data->d_buf = buf;
1229         sec->data->d_size = sec->sh.sh_size;
1230         sec->data->d_type = ELF_T_RELA;
1231
1232         idx = 0;
1233         list_for_each_entry(reloc, &sec->reloc_list, list) {
1234                 reloc->rela.r_offset = reloc->offset;
1235                 reloc->rela.r_addend = reloc->addend;
1236                 reloc->rela.r_info = GELF_R_INFO(reloc->sym->idx, reloc->type);
1237                 if (!gelf_update_rela(sec->data, idx, &reloc->rela)) {
1238                         WARN_ELF("gelf_update_rela");
1239                         return -1;
1240                 }
1241                 idx++;
1242         }
1243
1244         return 0;
1245 }
1246
1247 static int elf_rebuild_reloc_section(struct elf *elf, struct section *sec)
1248 {
1249         switch (sec->sh.sh_type) {
1250         case SHT_REL:  return elf_rebuild_rel_reloc_section(sec);
1251         case SHT_RELA: return elf_rebuild_rela_reloc_section(sec);
1252         default:       return -1;
1253         }
1254 }
1255
1256 int elf_write_insn(struct elf *elf, struct section *sec,
1257                    unsigned long offset, unsigned int len,
1258                    const char *insn)
1259 {
1260         Elf_Data *data = sec->data;
1261
1262         if (data->d_type != ELF_T_BYTE || data->d_off) {
1263                 WARN("write to unexpected data for section: %s", sec->name);
1264                 return -1;
1265         }
1266
1267         memcpy(data->d_buf + offset, insn, len);
1268         elf_flagdata(data, ELF_C_SET, ELF_F_DIRTY);
1269
1270         elf->changed = true;
1271
1272         return 0;
1273 }
1274
1275 int elf_write_reloc(struct elf *elf, struct reloc *reloc)
1276 {
1277         struct section *sec = reloc->sec;
1278
1279         if (sec->sh.sh_type == SHT_REL) {
1280                 reloc->rel.r_info = GELF_R_INFO(reloc->sym->idx, reloc->type);
1281                 reloc->rel.r_offset = reloc->offset;
1282
1283                 if (!gelf_update_rel(sec->data, reloc->idx, &reloc->rel)) {
1284                         WARN_ELF("gelf_update_rel");
1285                         return -1;
1286                 }
1287         } else {
1288                 reloc->rela.r_info = GELF_R_INFO(reloc->sym->idx, reloc->type);
1289                 reloc->rela.r_addend = reloc->addend;
1290                 reloc->rela.r_offset = reloc->offset;
1291
1292                 if (!gelf_update_rela(sec->data, reloc->idx, &reloc->rela)) {
1293                         WARN_ELF("gelf_update_rela");
1294                         return -1;
1295                 }
1296         }
1297
1298         elf->changed = true;
1299
1300         return 0;
1301 }
1302
1303 int elf_write(struct elf *elf)
1304 {
1305         struct section *sec;
1306         Elf_Scn *s;
1307
1308         /* Update changed relocation sections and section headers: */
1309         list_for_each_entry(sec, &elf->sections, list) {
1310                 if (sec->changed) {
1311                         s = elf_getscn(elf->elf, sec->idx);
1312                         if (!s) {
1313                                 WARN_ELF("elf_getscn");
1314                                 return -1;
1315                         }
1316                         if (!gelf_update_shdr(s, &sec->sh)) {
1317                                 WARN_ELF("gelf_update_shdr");
1318                                 return -1;
1319                         }
1320
1321                         if (sec->base &&
1322                             elf_rebuild_reloc_section(elf, sec)) {
1323                                 WARN("elf_rebuild_reloc_section");
1324                                 return -1;
1325                         }
1326
1327                         sec->changed = false;
1328                         elf->changed = true;
1329                 }
1330         }
1331
1332         /* Make sure the new section header entries get updated properly. */
1333         elf_flagelf(elf->elf, ELF_C_SET, ELF_F_DIRTY);
1334
1335         /* Write all changes to the file. */
1336         if (elf_update(elf->elf, ELF_C_WRITE) < 0) {
1337                 WARN_ELF("elf_update");
1338                 return -1;
1339         }
1340
1341         elf->changed = false;
1342
1343         return 0;
1344 }
1345
1346 void elf_close(struct elf *elf)
1347 {
1348         struct section *sec, *tmpsec;
1349         struct symbol *sym, *tmpsym;
1350         struct reloc *reloc, *tmpreloc;
1351
1352         if (elf->elf)
1353                 elf_end(elf->elf);
1354
1355         if (elf->fd > 0)
1356                 close(elf->fd);
1357
1358         list_for_each_entry_safe(sec, tmpsec, &elf->sections, list) {
1359                 list_for_each_entry_safe(sym, tmpsym, &sec->symbol_list, list) {
1360                         list_del(&sym->list);
1361                         hash_del(&sym->hash);
1362                         free(sym);
1363                 }
1364                 list_for_each_entry_safe(reloc, tmpreloc, &sec->reloc_list, list) {
1365                         list_del(&reloc->list);
1366                         hash_del(&reloc->hash);
1367                         free(reloc);
1368                 }
1369                 list_del(&sec->list);
1370                 free(sec);
1371         }
1372
1373         free(elf);
1374 }