tools/objtool/elf.c

   1 // SPDX-License-Identifier: GPL-2.0-or-later
   2 /*
   3  * elf.c - ELF access library
   4  *
   5  * Adapted from kpatch (https://github.com/dynup/kpatch):
   6  * Copyright (C) 2013-2015 Josh Poimboeuf <jpoimboe@redhat.com>
   7  * Copyright (C) 2014 Seth Jennings <sjenning@redhat.com>
   8  */
   9
  10 #include <sys/types.h>
  11 #include <sys/stat.h>
  12 #include <sys/mman.h>
  13 #include <fcntl.h>
  14 #include <stdio.h>
  15 #include <stdlib.h>
  16 #include <string.h>
  17 #include <unistd.h>
  18 #include <errno.h>
  19 #include <linux/interval_tree_generic.h>
  20 #include <objtool/builtin.h>
  21
  22 #include <objtool/elf.h>
  23 #include <objtool/warn.h>
  24
  25 #define MAX_NAME_LEN 128
  26
  27 static inline u32 str_hash(const char *str)
  28 {
  29         return jhash(str, strlen(str), 0);
  30 }
  31
  32 #define __elf_table(name)       (elf->name##_hash)
  33 #define __elf_bits(name)        (elf->name##_bits)
  34
  35 #define __elf_table_entry(name, key) \
  36         __elf_table(name)[hash_min(key, __elf_bits(name))]
  37
  38 #define elf_hash_add(name, node, key)                                   \
  39 ({                                                                      \
  40         struct elf_hash_node *__node = node;                            \
  41         __node->next = __elf_table_entry(name, key);                    \
  42         __elf_table_entry(name, key) = __node;                          \
  43 })
  44
  45 static inline void __elf_hash_del(struct elf_hash_node *node,
  46                                   struct elf_hash_node **head)
  47 {
  48         struct elf_hash_node *cur, *prev;
  49
  50         if (node == *head) {
  51                 *head = node->next;
  52                 return;
  53         }
  54
  55         for (prev = NULL, cur = *head; cur; prev = cur, cur = cur->next) {
  56                 if (cur == node) {
  57                         prev->next = cur->next;
  58                         break;
  59                 }
  60         }
  61 }
  62
  63 #define elf_hash_del(name, node, key) \
  64         __elf_hash_del(node, &__elf_table_entry(name, key))
  65
  66 #define elf_list_entry(ptr, type, member)                               \
  67 ({                                                                      \
  68         typeof(ptr) __ptr = (ptr);                                      \
  69         __ptr ? container_of(__ptr, type, member) : NULL;               \
  70 })
  71
  72 #define elf_hash_for_each_possible(name, obj, member, key)              \
  73         for (obj = elf_list_entry(__elf_table_entry(name, key), typeof(*obj), member); \
  74              obj;                                                       \
  75              obj = elf_list_entry(obj->member.next, typeof(*(obj)), member))
  76
  77 #define elf_alloc_hash(name, size) \
  78 ({ \
  79         __elf_bits(name) = max(10, ilog2(size)); \
  80         __elf_table(name) = mmap(NULL, sizeof(struct elf_hash_node *) << __elf_bits(name), \
  81                                  PROT_READ|PROT_WRITE, \
  82                                  MAP_PRIVATE|MAP_ANON, -1, 0); \
  83         if (__elf_table(name) == (void *)-1L) { \
  84                 WARN("mmap fail " #name); \
  85                 __elf_table(name) = NULL; \
  86         } \
  87         __elf_table(name); \
  88 })
  89
  90 static inline unsigned long __sym_start(struct symbol *s)
  91 {
  92         return s->offset;
  93 }
  94
  95 static inline unsigned long __sym_last(struct symbol *s)
  96 {
  97         return s->offset + s->len - 1;
  98 }
  99
 100 INTERVAL_TREE_DEFINE(struct symbol, node, unsigned long, __subtree_last,
 101                      __sym_start, __sym_last, static, __sym)
 102
 103 #define __sym_for_each(_iter, _tree, _start, _end)                      \
 104         for (_iter = __sym_iter_first((_tree), (_start), (_end));       \
 105              _iter; _iter = __sym_iter_next(_iter, (_start), (_end)))
 106
 107 struct symbol_hole {
 108         unsigned long key;
 109         const struct symbol *sym;
 110 };
 111
 112 /*
 113  * Find !section symbol where @offset is after it.
 114  */
 115 static int symbol_hole_by_offset(const void *key, const struct rb_node *node)
 116 {
 117         const struct symbol *s = rb_entry(node, struct symbol, node);
 118         struct symbol_hole *sh = (void *)key;
 119
 120         if (sh->key < s->offset)
 121                 return -1;
 122
 123         if (sh->key >= s->offset + s->len) {
 124                 if (s->type != STT_SECTION)
 125                         sh->sym = s;
 126                 return 1;
 127         }
 128
 129         return 0;
 130 }
 131
 132 struct section *find_section_by_name(const struct elf *elf, const char *name)
 133 {
 134         struct section *sec;
 135
 136         elf_hash_for_each_possible(section_name, sec, name_hash, str_hash(name)) {
 137                 if (!strcmp(sec->name, name))
 138                         return sec;
 139         }
 140
 141         return NULL;
 142 }
 143
 144 static struct section *find_section_by_index(struct elf *elf,
 145                                              unsigned int idx)
 146 {
 147         struct section *sec;
 148
 149         elf_hash_for_each_possible(section, sec, hash, idx) {
 150                 if (sec->idx == idx)
 151                         return sec;
 152         }
 153
 154         return NULL;
 155 }
 156
 157 static struct symbol *find_symbol_by_index(struct elf *elf, unsigned int idx)
 158 {
 159         struct symbol *sym;
 160
 161         elf_hash_for_each_possible(symbol, sym, hash, idx) {
 162                 if (sym->idx == idx)
 163                         return sym;
 164         }
 165
 166         return NULL;
 167 }
 168
 169 struct symbol *find_symbol_by_offset(struct section *sec, unsigned long offset)
 170 {
 171         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 172         struct symbol *iter;
 173
 174         __sym_for_each(iter, tree, offset, offset) {
 175                 if (iter->offset == offset && iter->type != STT_SECTION)
 176                         return iter;
 177         }
 178
 179         return NULL;
 180 }
 181
 182 struct symbol *find_func_by_offset(struct section *sec, unsigned long offset)
 183 {
 184         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 185         struct symbol *iter;
 186
 187         __sym_for_each(iter, tree, offset, offset) {
 188                 if (iter->offset == offset && iter->type == STT_FUNC)
 189                         return iter;
 190         }
 191
 192         return NULL;
 193 }
 194
 195 struct symbol *find_symbol_containing(const struct section *sec, unsigned long offset)
 196 {
 197         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 198         struct symbol *iter;
 199
 200         __sym_for_each(iter, tree, offset, offset) {
 201                 if (iter->type != STT_SECTION)
 202                         return iter;
 203         }
 204
 205         return NULL;
 206 }
 207
 208 /*
 209  * Returns size of hole starting at @offset.
 210  */
 211 int find_symbol_hole_containing(const struct section *sec, unsigned long offset)
 212 {
 213         struct symbol_hole hole = {
 214                 .key = offset,
 215                 .sym = NULL,
 216         };
 217         struct rb_node *n;
 218         struct symbol *s;
 219
 220         /*
 221          * Find the rightmost symbol for which @offset is after it.
 222          */
 223         n = rb_find(&hole, &sec->symbol_tree.rb_root, symbol_hole_by_offset);
 224
 225         /* found a symbol that contains @offset */
 226         if (n)
 227                 return 0; /* not a hole */
 228
 229         /* didn't find a symbol for which @offset is after it */
 230         if (!hole.sym)
 231                 return 0; /* not a hole */
 232
 233         /* @offset >= sym->offset + sym->len, find symbol after it */
 234         n = rb_next(&hole.sym->node);
 235         if (!n)
 236                 return -1; /* until end of address space */
 237
 238         /* hole until start of next symbol */
 239         s = rb_entry(n, struct symbol, node);
 240         return s->offset - offset;
 241 }
 242
 243 struct symbol *find_func_containing(struct section *sec, unsigned long offset)
 244 {
 245         struct rb_root_cached *tree = (struct rb_root_cached *)&sec->symbol_tree;
 246         struct symbol *iter;
 247
 248         __sym_for_each(iter, tree, offset, offset) {
 249                 if (iter->type == STT_FUNC)
 250                         return iter;
 251         }
 252
 253         return NULL;
 254 }
 255
 256 struct symbol *find_symbol_by_name(const struct elf *elf, const char *name)
 257 {
 258         struct symbol *sym;
 259
 260         elf_hash_for_each_possible(symbol_name, sym, name_hash, str_hash(name)) {
 261                 if (!strcmp(sym->name, name))
 262                         return sym;
 263         }
 264
 265         return NULL;
 266 }
 267
 268 struct reloc *find_reloc_by_dest_range(const struct elf *elf, struct section *sec,
 269                                      unsigned long offset, unsigned int len)
 270 {
 271         struct reloc *reloc, *r = NULL;
 272         struct section *rsec;
 273         unsigned long o;
 274
 275         rsec = sec->rsec;
 276         if (!rsec)
 277                 return NULL;
 278
 279         for_offset_range(o, offset, offset + len) {
 280                 elf_hash_for_each_possible(reloc, reloc, hash,
 281                                            sec_offset_hash(rsec, o)) {
 282                         if (reloc->sec != rsec)
 283                                 continue;
 284
 285                         if (reloc_offset(reloc) >= offset &&
 286                             reloc_offset(reloc) < offset + len) {
 287                                 if (!r || reloc_offset(reloc) < reloc_offset(r))
 288                                         r = reloc;
 289                         }
 290                 }
 291                 if (r)
 292                         return r;
 293         }
 294
 295         return NULL;
 296 }
 297
 298 struct reloc *find_reloc_by_dest(const struct elf *elf, struct section *sec, unsigned long offset)
 299 {
 300         return find_reloc_by_dest_range(elf, sec, offset, 1);
 301 }
 302
 303 static bool is_dwarf_section(struct section *sec)
 304 {
 305         return !strncmp(sec->name, ".debug_", 7);
 306 }
 307
 308 static int read_sections(struct elf *elf)
 309 {
 310         Elf_Scn *s = NULL;
 311         struct section *sec;
 312         size_t shstrndx, sections_nr;
 313         int i;
 314
 315         if (elf_getshdrnum(elf->elf, &sections_nr)) {
 316                 WARN_ELF("elf_getshdrnum");
 317                 return -1;
 318         }
 319
 320         if (elf_getshdrstrndx(elf->elf, &shstrndx)) {
 321                 WARN_ELF("elf_getshdrstrndx");
 322                 return -1;
 323         }
 324
 325         if (!elf_alloc_hash(section, sections_nr) ||
 326             !elf_alloc_hash(section_name, sections_nr))
 327                 return -1;
 328
 329         elf->section_data = calloc(sections_nr, sizeof(*sec));
 330         if (!elf->section_data) {
 331                 perror("calloc");
 332                 return -1;
 333         }
 334         for (i = 0; i < sections_nr; i++) {
 335                 sec = &elf->section_data[i];
 336
 337                 INIT_LIST_HEAD(&sec->symbol_list);
 338
 339                 s = elf_getscn(elf->elf, i);
 340                 if (!s) {
 341                         WARN_ELF("elf_getscn");
 342                         return -1;
 343                 }
 344
 345                 sec->idx = elf_ndxscn(s);
 346
 347                 if (!gelf_getshdr(s, &sec->sh)) {
 348                         WARN_ELF("gelf_getshdr");
 349                         return -1;
 350                 }
 351
 352                 sec->name = elf_strptr(elf->elf, shstrndx, sec->sh.sh_name);
 353                 if (!sec->name) {
 354                         WARN_ELF("elf_strptr");
 355                         return -1;
 356                 }
 357
 358                 if (sec->sh.sh_size != 0 && !is_dwarf_section(sec)) {
 359                         sec->data = elf_getdata(s, NULL);
 360                         if (!sec->data) {
 361                                 WARN_ELF("elf_getdata");
 362                                 return -1;
 363                         }
 364                         if (sec->data->d_off != 0 ||
 365                             sec->data->d_size != sec->sh.sh_size) {
 366                                 WARN("unexpected data attributes for %s",
 367                                      sec->name);
 368                                 return -1;
 369                         }
 370                 }
 371
 372                 list_add_tail(&sec->list, &elf->sections);
 373                 elf_hash_add(section, &sec->hash, sec->idx);
 374                 elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
 375
 376                 if (is_reloc_sec(sec))
 377                         elf->num_relocs += sec_num_entries(sec);
 378         }
 379
 380         if (opts.stats) {
 381                 printf("nr_sections: %lu\n", (unsigned long)sections_nr);
 382                 printf("section_bits: %d\n", elf->section_bits);
 383         }
 384
 385         /* sanity check, one more call to elf_nextscn() should return NULL */
 386         if (elf_nextscn(elf->elf, s)) {
 387                 WARN("section entry mismatch");
 388                 return -1;
 389         }
 390
 391         return 0;
 392 }
 393
 394 static void elf_add_symbol(struct elf *elf, struct symbol *sym)
 395 {
 396         struct list_head *entry;
 397         struct rb_node *pnode;
 398         struct symbol *iter;
 399
 400         INIT_LIST_HEAD(&sym->pv_target);
 401         sym->alias = sym;
 402
 403         sym->type = GELF_ST_TYPE(sym->sym.st_info);
 404         sym->bind = GELF_ST_BIND(sym->sym.st_info);
 405
 406         if (sym->type == STT_FILE)
 407                 elf->num_files++;
 408
 409         sym->offset = sym->sym.st_value;
 410         sym->len = sym->sym.st_size;
 411
 412         __sym_for_each(iter, &sym->sec->symbol_tree, sym->offset, sym->offset) {
 413                 if (iter->offset == sym->offset && iter->type == sym->type)
 414                         iter->alias = sym;
 415         }
 416
 417         __sym_insert(sym, &sym->sec->symbol_tree);
 418         pnode = rb_prev(&sym->node);
 419         if (pnode)
 420                 entry = &rb_entry(pnode, struct symbol, node)->list;
 421         else
 422                 entry = &sym->sec->symbol_list;
 423         list_add(&sym->list, entry);
 424         elf_hash_add(symbol, &sym->hash, sym->idx);
 425         elf_hash_add(symbol_name, &sym->name_hash, str_hash(sym->name));
 426
 427         /*
 428          * Don't store empty STT_NOTYPE symbols in the rbtree.  They
 429          * can exist within a function, confusing the sorting.
 430          */
 431         if (!sym->len)
 432                 __sym_remove(sym, &sym->sec->symbol_tree);
 433 }
 434
 435 static int read_symbols(struct elf *elf)
 436 {
 437         struct section *symtab, *symtab_shndx, *sec;
 438         struct symbol *sym, *pfunc;
 439         int symbols_nr, i;
 440         char *coldstr;
 441         Elf_Data *shndx_data = NULL;
 442         Elf32_Word shndx;
 443
 444         symtab = find_section_by_name(elf, ".symtab");
 445         if (symtab) {
 446                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 447                 if (symtab_shndx)
 448                         shndx_data = symtab_shndx->data;
 449
 450                 symbols_nr = sec_num_entries(symtab);
 451         } else {
 452                 /*
 453                  * A missing symbol table is actually possible if it's an empty
 454                  * .o file. This can happen for thunk_64.o. Make sure to at
 455                  * least allocate the symbol hash tables so we can do symbol
 456                  * lookups without crashing.
 457                  */
 458                 symbols_nr = 0;
 459         }
 460
 461         if (!elf_alloc_hash(symbol, symbols_nr) ||
 462             !elf_alloc_hash(symbol_name, symbols_nr))
 463                 return -1;
 464
 465         elf->symbol_data = calloc(symbols_nr, sizeof(*sym));
 466         if (!elf->symbol_data) {
 467                 perror("calloc");
 468                 return -1;
 469         }
 470         for (i = 0; i < symbols_nr; i++) {
 471                 sym = &elf->symbol_data[i];
 472
 473                 sym->idx = i;
 474
 475                 if (!gelf_getsymshndx(symtab->data, shndx_data, i, &sym->sym,
 476                                       &shndx)) {
 477                         WARN_ELF("gelf_getsymshndx");
 478                         goto err;
 479                 }
 480
 481                 sym->name = elf_strptr(elf->elf, symtab->sh.sh_link,
 482                                        sym->sym.st_name);
 483                 if (!sym->name) {
 484                         WARN_ELF("elf_strptr");
 485                         goto err;
 486                 }
 487
 488                 if ((sym->sym.st_shndx > SHN_UNDEF &&
 489                      sym->sym.st_shndx < SHN_LORESERVE) ||
 490                     (shndx_data && sym->sym.st_shndx == SHN_XINDEX)) {
 491                         if (sym->sym.st_shndx != SHN_XINDEX)
 492                                 shndx = sym->sym.st_shndx;
 493
 494                         sym->sec = find_section_by_index(elf, shndx);
 495                         if (!sym->sec) {
 496                                 WARN("couldn't find section for symbol %s",
 497                                      sym->name);
 498                                 goto err;
 499                         }
 500                         if (GELF_ST_TYPE(sym->sym.st_info) == STT_SECTION) {
 501                                 sym->name = sym->sec->name;
 502                                 sym->sec->sym = sym;
 503                         }
 504                 } else
 505                         sym->sec = find_section_by_index(elf, 0);
 506
 507                 elf_add_symbol(elf, sym);
 508         }
 509
 510         if (opts.stats) {
 511                 printf("nr_symbols: %lu\n", (unsigned long)symbols_nr);
 512                 printf("symbol_bits: %d\n", elf->symbol_bits);
 513         }
 514
 515         /* Create parent/child links for any cold subfunctions */
 516         list_for_each_entry(sec, &elf->sections, list) {
 517                 sec_for_each_sym(sec, sym) {
 518                         char pname[MAX_NAME_LEN + 1];
 519                         size_t pnamelen;
 520                         if (sym->type != STT_FUNC)
 521                                 continue;
 522
 523                         if (sym->pfunc == NULL)
 524                                 sym->pfunc = sym;
 525
 526                         if (sym->cfunc == NULL)
 527                                 sym->cfunc = sym;
 528
 529                         coldstr = strstr(sym->name, ".cold");
 530                         if (!coldstr)
 531                                 continue;
 532
 533                         pnamelen = coldstr - sym->name;
 534                         if (pnamelen > MAX_NAME_LEN) {
 535                                 WARN("%s(): parent function name exceeds maximum length of %d characters",
 536                                      sym->name, MAX_NAME_LEN);
 537                                 return -1;
 538                         }
 539
 540                         strncpy(pname, sym->name, pnamelen);
 541                         pname[pnamelen] = '\0';
 542                         pfunc = find_symbol_by_name(elf, pname);
 543
 544                         if (!pfunc) {
 545                                 WARN("%s(): can't find parent function",
 546                                      sym->name);
 547                                 return -1;
 548                         }
 549
 550                         sym->pfunc = pfunc;
 551                         pfunc->cfunc = sym;
 552
 553                         /*
 554                          * Unfortunately, -fnoreorder-functions puts the child
 555                          * inside the parent.  Remove the overlap so we can
 556                          * have sane assumptions.
 557                          *
 558                          * Note that pfunc->len now no longer matches
 559                          * pfunc->sym.st_size.
 560                          */
 561                         if (sym->sec == pfunc->sec &&
 562                             sym->offset >= pfunc->offset &&
 563                             sym->offset + sym->len == pfunc->offset + pfunc->len) {
 564                                 pfunc->len -= sym->len;
 565                         }
 566                 }
 567         }
 568
 569         return 0;
 570
 571 err:
 572         free(sym);
 573         return -1;
 574 }
 575
 576 /*
 577  * @sym's idx has changed.  Update the relocs which reference it.
 578  */
 579 static int elf_update_sym_relocs(struct elf *elf, struct symbol *sym)
 580 {
 581         struct reloc *reloc;
 582
 583         for (reloc = sym->relocs; reloc; reloc = reloc->sym_next_reloc)
 584                 set_reloc_sym(elf, reloc, reloc->sym->idx);
 585
 586         return 0;
 587 }
 588
 589 /*
 590  * The libelf API is terrible; gelf_update_sym*() takes a data block relative
 591  * index value, *NOT* the symbol index. As such, iterate the data blocks and
 592  * adjust index until it fits.
 593  *
 594  * If no data block is found, allow adding a new data block provided the index
 595  * is only one past the end.
 596  */
 597 static int elf_update_symbol(struct elf *elf, struct section *symtab,
 598                              struct section *symtab_shndx, struct symbol *sym)
 599 {
 600         Elf32_Word shndx = sym->sec ? sym->sec->idx : SHN_UNDEF;
 601         Elf_Data *symtab_data = NULL, *shndx_data = NULL;
 602         Elf64_Xword entsize = symtab->sh.sh_entsize;
 603         int max_idx, idx = sym->idx;
 604         Elf_Scn *s, *t = NULL;
 605         bool is_special_shndx = sym->sym.st_shndx >= SHN_LORESERVE &&
 606                                 sym->sym.st_shndx != SHN_XINDEX;
 607
 608         if (is_special_shndx)
 609                 shndx = sym->sym.st_shndx;
 610
 611         s = elf_getscn(elf->elf, symtab->idx);
 612         if (!s) {
 613                 WARN_ELF("elf_getscn");
 614                 return -1;
 615         }
 616
 617         if (symtab_shndx) {
 618                 t = elf_getscn(elf->elf, symtab_shndx->idx);
 619                 if (!t) {
 620                         WARN_ELF("elf_getscn");
 621                         return -1;
 622                 }
 623         }
 624
 625         for (;;) {
 626                 /* get next data descriptor for the relevant sections */
 627                 symtab_data = elf_getdata(s, symtab_data);
 628                 if (t)
 629                         shndx_data = elf_getdata(t, shndx_data);
 630
 631                 /* end-of-list */
 632                 if (!symtab_data) {
 633                         /*
 634                          * Over-allocate to avoid O(n^2) symbol creation
 635                          * behaviour.  The down side is that libelf doesn't
 636                          * like this; see elf_truncate_section() for the fixup.
 637                          */
 638                         int num = max(1U, sym->idx/3);
 639                         void *buf;
 640
 641                         if (idx) {
 642                                 /* we don't do holes in symbol tables */
 643                                 WARN("index out of range");
 644                                 return -1;
 645                         }
 646
 647                         /* if @idx == 0, it's the next contiguous entry, create it */
 648                         symtab_data = elf_newdata(s);
 649                         if (t)
 650                                 shndx_data = elf_newdata(t);
 651
 652                         buf = calloc(num, entsize);
 653                         if (!buf) {
 654                                 WARN("malloc");
 655                                 return -1;
 656                         }
 657
 658                         symtab_data->d_buf = buf;
 659                         symtab_data->d_size = num * entsize;
 660                         symtab_data->d_align = 1;
 661                         symtab_data->d_type = ELF_T_SYM;
 662
 663                         mark_sec_changed(elf, symtab, true);
 664                         symtab->truncate = true;
 665
 666                         if (t) {
 667                                 buf = calloc(num, sizeof(Elf32_Word));
 668                                 if (!buf) {
 669                                         WARN("malloc");
 670                                         return -1;
 671                                 }
 672
 673                                 shndx_data->d_buf = buf;
 674                                 shndx_data->d_size = num * sizeof(Elf32_Word);
 675                                 shndx_data->d_align = sizeof(Elf32_Word);
 676                                 shndx_data->d_type = ELF_T_WORD;
 677
 678                                 mark_sec_changed(elf, symtab_shndx, true);
 679                                 symtab_shndx->truncate = true;
 680                         }
 681
 682                         break;
 683                 }
 684
 685                 /* empty blocks should not happen */
 686                 if (!symtab_data->d_size) {
 687                         WARN("zero size data");
 688                         return -1;
 689                 }
 690
 691                 /* is this the right block? */
 692                 max_idx = symtab_data->d_size / entsize;
 693                 if (idx < max_idx)
 694                         break;
 695
 696                 /* adjust index and try again */
 697                 idx -= max_idx;
 698         }
 699
 700         /* something went side-ways */
 701         if (idx < 0) {
 702                 WARN("negative index");
 703                 return -1;
 704         }
 705
 706         /* setup extended section index magic and write the symbol */
 707         if ((shndx >= SHN_UNDEF && shndx < SHN_LORESERVE) || is_special_shndx) {
 708                 sym->sym.st_shndx = shndx;
 709                 if (!shndx_data)
 710                         shndx = 0;
 711         } else {
 712                 sym->sym.st_shndx = SHN_XINDEX;
 713                 if (!shndx_data) {
 714                         WARN("no .symtab_shndx");
 715                         return -1;
 716                 }
 717         }
 718
 719         if (!gelf_update_symshndx(symtab_data, shndx_data, idx, &sym->sym, shndx)) {
 720                 WARN_ELF("gelf_update_symshndx");
 721                 return -1;
 722         }
 723
 724         return 0;
 725 }
 726
 727 static struct symbol *
 728 __elf_create_symbol(struct elf *elf, struct symbol *sym)
 729 {
 730         struct section *symtab, *symtab_shndx;
 731         Elf32_Word first_non_local, new_idx;
 732         struct symbol *old;
 733
 734         symtab = find_section_by_name(elf, ".symtab");
 735         if (symtab) {
 736                 symtab_shndx = find_section_by_name(elf, ".symtab_shndx");
 737         } else {
 738                 WARN("no .symtab");
 739                 return NULL;
 740         }
 741
 742         new_idx = sec_num_entries(symtab);
 743
 744         if (GELF_ST_BIND(sym->sym.st_info) != STB_LOCAL)
 745                 goto non_local;
 746
 747         /*
 748          * Move the first global symbol, as per sh_info, into a new, higher
 749          * symbol index. This fees up a spot for a new local symbol.
 750          */
 751         first_non_local = symtab->sh.sh_info;
 752         old = find_symbol_by_index(elf, first_non_local);
 753         if (old) {
 754
 755                 elf_hash_del(symbol, &old->hash, old->idx);
 756                 elf_hash_add(symbol, &old->hash, new_idx);
 757                 old->idx = new_idx;
 758
 759                 if (elf_update_symbol(elf, symtab, symtab_shndx, old)) {
 760                         WARN("elf_update_symbol move");
 761                         return NULL;
 762                 }
 763
 764                 if (elf_update_sym_relocs(elf, old))
 765                         return NULL;
 766
 767                 new_idx = first_non_local;
 768         }
 769
 770         /*
 771          * Either way, we will add a LOCAL symbol.
 772          */
 773         symtab->sh.sh_info += 1;
 774
 775 non_local:
 776         sym->idx = new_idx;
 777         if (elf_update_symbol(elf, symtab, symtab_shndx, sym)) {
 778                 WARN("elf_update_symbol");
 779                 return NULL;
 780         }
 781
 782         symtab->sh.sh_size += symtab->sh.sh_entsize;
 783         mark_sec_changed(elf, symtab, true);
 784
 785         if (symtab_shndx) {
 786                 symtab_shndx->sh.sh_size += sizeof(Elf32_Word);
 787                 mark_sec_changed(elf, symtab_shndx, true);
 788         }
 789
 790         return sym;
 791 }
 792
 793 static struct symbol *
 794 elf_create_section_symbol(struct elf *elf, struct section *sec)
 795 {
 796         struct symbol *sym = calloc(1, sizeof(*sym));
 797
 798         if (!sym) {
 799                 perror("malloc");
 800                 return NULL;
 801         }
 802
 803         sym->name = sec->name;
 804         sym->sec = sec;
 805
 806         // st_name 0
 807         sym->sym.st_info = GELF_ST_INFO(STB_LOCAL, STT_SECTION);
 808         // st_other 0
 809         // st_value 0
 810         // st_size 0
 811
 812         sym = __elf_create_symbol(elf, sym);
 813         if (sym)
 814                 elf_add_symbol(elf, sym);
 815
 816         return sym;
 817 }
 818
 819 static int elf_add_string(struct elf *elf, struct section *strtab, char *str);
 820
 821 struct symbol *
 822 elf_create_prefix_symbol(struct elf *elf, struct symbol *orig, long size)
 823 {
 824         struct symbol *sym = calloc(1, sizeof(*sym));
 825         size_t namelen = strlen(orig->name) + sizeof("__pfx_");
 826         char *name = malloc(namelen);
 827
 828         if (!sym || !name) {
 829                 perror("malloc");
 830                 return NULL;
 831         }
 832
 833         snprintf(name, namelen, "__pfx_%s", orig->name);
 834
 835         sym->name = name;
 836         sym->sec = orig->sec;
 837
 838         sym->sym.st_name = elf_add_string(elf, NULL, name);
 839         sym->sym.st_info = orig->sym.st_info;
 840         sym->sym.st_value = orig->sym.st_value - size;
 841         sym->sym.st_size = size;
 842
 843         sym = __elf_create_symbol(elf, sym);
 844         if (sym)
 845                 elf_add_symbol(elf, sym);
 846
 847         return sym;
 848 }
 849
 850 static struct reloc *elf_init_reloc(struct elf *elf, struct section *rsec,
 851                                     unsigned int reloc_idx,
 852                                     unsigned long offset, struct symbol *sym,
 853                                     s64 addend, unsigned int type)
 854 {
 855         struct reloc *reloc, empty = { 0 };
 856
 857         if (reloc_idx >= sec_num_entries(rsec)) {
 858                 WARN("%s: bad reloc_idx %u for %s with %d relocs",
 859                      __func__, reloc_idx, rsec->name, sec_num_entries(rsec));
 860                 return NULL;
 861         }
 862
 863         reloc = &rsec->relocs[reloc_idx];
 864
 865         if (memcmp(reloc, &empty, sizeof(empty))) {
 866                 WARN("%s: %s: reloc %d already initialized!",
 867                      __func__, rsec->name, reloc_idx);
 868                 return NULL;
 869         }
 870
 871         reloc->sec = rsec;
 872         reloc->sym = sym;
 873
 874         set_reloc_offset(elf, reloc, offset);
 875         set_reloc_sym(elf, reloc, sym->idx);
 876         set_reloc_type(elf, reloc, type);
 877         set_reloc_addend(elf, reloc, addend);
 878
 879         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 880         reloc->sym_next_reloc = sym->relocs;
 881         sym->relocs = reloc;
 882
 883         return reloc;
 884 }
 885
 886 struct reloc *elf_init_reloc_text_sym(struct elf *elf, struct section *sec,
 887                                       unsigned long offset,
 888                                       unsigned int reloc_idx,
 889                                       struct section *insn_sec,
 890                                       unsigned long insn_off)
 891 {
 892         struct symbol *sym = insn_sec->sym;
 893         int addend = insn_off;
 894
 895         if (!(insn_sec->sh.sh_flags & SHF_EXECINSTR)) {
 896                 WARN("bad call to %s() for data symbol %s",
 897                      __func__, sym->name);
 898                 return NULL;
 899         }
 900
 901         if (!sym) {
 902                 /*
 903                  * Due to how weak functions work, we must use section based
 904                  * relocations. Symbol based relocations would result in the
 905                  * weak and non-weak function annotations being overlaid on the
 906                  * non-weak function after linking.
 907                  */
 908                 sym = elf_create_section_symbol(elf, insn_sec);
 909                 if (!sym)
 910                         return NULL;
 911
 912                 insn_sec->sym = sym;
 913         }
 914
 915         return elf_init_reloc(elf, sec->rsec, reloc_idx, offset, sym, addend,
 916                               elf_text_rela_type(elf));
 917 }
 918
 919 struct reloc *elf_init_reloc_data_sym(struct elf *elf, struct section *sec,
 920                                       unsigned long offset,
 921                                       unsigned int reloc_idx,
 922                                       struct symbol *sym,
 923                                       s64 addend)
 924 {
 925         if (sym->sec && (sec->sh.sh_flags & SHF_EXECINSTR)) {
 926                 WARN("bad call to %s() for text symbol %s",
 927                      __func__, sym->name);
 928                 return NULL;
 929         }
 930
 931         return elf_init_reloc(elf, sec->rsec, reloc_idx, offset, sym, addend,
 932                               elf_data_rela_type(elf));
 933 }
 934
 935 static int read_relocs(struct elf *elf)
 936 {
 937         unsigned long nr_reloc, max_reloc = 0;
 938         struct section *rsec;
 939         struct reloc *reloc;
 940         unsigned int symndx;
 941         struct symbol *sym;
 942         int i;
 943
 944         if (!elf_alloc_hash(reloc, elf->num_relocs))
 945                 return -1;
 946
 947         list_for_each_entry(rsec, &elf->sections, list) {
 948                 if (!is_reloc_sec(rsec))
 949                         continue;
 950
 951                 rsec->base = find_section_by_index(elf, rsec->sh.sh_info);
 952                 if (!rsec->base) {
 953                         WARN("can't find base section for reloc section %s",
 954                              rsec->name);
 955                         return -1;
 956                 }
 957
 958                 rsec->base->rsec = rsec;
 959
 960                 nr_reloc = 0;
 961                 rsec->relocs = calloc(sec_num_entries(rsec), sizeof(*reloc));
 962                 if (!rsec->relocs) {
 963                         perror("calloc");
 964                         return -1;
 965                 }
 966                 for (i = 0; i < sec_num_entries(rsec); i++) {
 967                         reloc = &rsec->relocs[i];
 968
 969                         reloc->sec = rsec;
 970                         symndx = reloc_sym(reloc);
 971                         reloc->sym = sym = find_symbol_by_index(elf, symndx);
 972                         if (!reloc->sym) {
 973                                 WARN("can't find reloc entry symbol %d for %s",
 974                                      symndx, rsec->name);
 975                                 return -1;
 976                         }
 977
 978                         elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
 979                         reloc->sym_next_reloc = sym->relocs;
 980                         sym->relocs = reloc;
 981
 982                         nr_reloc++;
 983                 }
 984                 max_reloc = max(max_reloc, nr_reloc);
 985         }
 986
 987         if (opts.stats) {
 988                 printf("max_reloc: %lu\n", max_reloc);
 989                 printf("num_relocs: %lu\n", elf->num_relocs);
 990                 printf("reloc_bits: %d\n", elf->reloc_bits);
 991         }
 992
 993         return 0;
 994 }
 995
 996 struct elf *elf_open_read(const char *name, int flags)
 997 {
 998         struct elf *elf;
 999         Elf_Cmd cmd;
1000
1001         elf_version(EV_CURRENT);
1002
1003         elf = malloc(sizeof(*elf));
1004         if (!elf) {
1005                 perror("malloc");
1006                 return NULL;
1007         }
1008         memset(elf, 0, sizeof(*elf));
1009
1010         INIT_LIST_HEAD(&elf->sections);
1011
1012         elf->fd = open(name, flags);
1013         if (elf->fd == -1) {
1014                 fprintf(stderr, "objtool: Can't open '%s': %s\n",
1015                         name, strerror(errno));
1016                 goto err;
1017         }
1018
1019         if ((flags & O_ACCMODE) == O_RDONLY)
1020                 cmd = ELF_C_READ_MMAP;
1021         else if ((flags & O_ACCMODE) == O_RDWR)
1022                 cmd = ELF_C_RDWR;
1023         else /* O_WRONLY */
1024                 cmd = ELF_C_WRITE;
1025
1026         elf->elf = elf_begin(elf->fd, cmd, NULL);
1027         if (!elf->elf) {
1028                 WARN_ELF("elf_begin");
1029                 goto err;
1030         }
1031
1032         if (!gelf_getehdr(elf->elf, &elf->ehdr)) {
1033                 WARN_ELF("gelf_getehdr");
1034                 goto err;
1035         }
1036
1037         if (read_sections(elf))
1038                 goto err;
1039
1040         if (read_symbols(elf))
1041                 goto err;
1042
1043         if (read_relocs(elf))
1044                 goto err;
1045
1046         return elf;
1047
1048 err:
1049         elf_close(elf);
1050         return NULL;
1051 }
1052
1053 static int elf_add_string(struct elf *elf, struct section *strtab, char *str)
1054 {
1055         Elf_Data *data;
1056         Elf_Scn *s;
1057         int len;
1058
1059         if (!strtab)
1060                 strtab = find_section_by_name(elf, ".strtab");
1061         if (!strtab) {
1062                 WARN("can't find .strtab section");
1063                 return -1;
1064         }
1065
1066         s = elf_getscn(elf->elf, strtab->idx);
1067         if (!s) {
1068                 WARN_ELF("elf_getscn");
1069                 return -1;
1070         }
1071
1072         data = elf_newdata(s);
1073         if (!data) {
1074                 WARN_ELF("elf_newdata");
1075                 return -1;
1076         }
1077
1078         data->d_buf = str;
1079         data->d_size = strlen(str) + 1;
1080         data->d_align = 1;
1081
1082         len = strtab->sh.sh_size;
1083         strtab->sh.sh_size += data->d_size;
1084
1085         mark_sec_changed(elf, strtab, true);
1086
1087         return len;
1088 }
1089
1090 struct section *elf_create_section(struct elf *elf, const char *name,
1091                                    size_t entsize, unsigned int nr)
1092 {
1093         struct section *sec, *shstrtab;
1094         size_t size = entsize * nr;
1095         Elf_Scn *s;
1096
1097         sec = malloc(sizeof(*sec));
1098         if (!sec) {
1099                 perror("malloc");
1100                 return NULL;
1101         }
1102         memset(sec, 0, sizeof(*sec));
1103
1104         INIT_LIST_HEAD(&sec->symbol_list);
1105
1106         s = elf_newscn(elf->elf);
1107         if (!s) {
1108                 WARN_ELF("elf_newscn");
1109                 return NULL;
1110         }
1111
1112         sec->name = strdup(name);
1113         if (!sec->name) {
1114                 perror("strdup");
1115                 return NULL;
1116         }
1117
1118         sec->idx = elf_ndxscn(s);
1119
1120         sec->data = elf_newdata(s);
1121         if (!sec->data) {
1122                 WARN_ELF("elf_newdata");
1123                 return NULL;
1124         }
1125
1126         sec->data->d_size = size;
1127         sec->data->d_align = 1;
1128
1129         if (size) {
1130                 sec->data->d_buf = malloc(size);
1131                 if (!sec->data->d_buf) {
1132                         perror("malloc");
1133                         return NULL;
1134                 }
1135                 memset(sec->data->d_buf, 0, size);
1136         }
1137
1138         if (!gelf_getshdr(s, &sec->sh)) {
1139                 WARN_ELF("gelf_getshdr");
1140                 return NULL;
1141         }
1142
1143         sec->sh.sh_size = size;
1144         sec->sh.sh_entsize = entsize;
1145         sec->sh.sh_type = SHT_PROGBITS;
1146         sec->sh.sh_addralign = 1;
1147         sec->sh.sh_flags = SHF_ALLOC;
1148
1149         /* Add section name to .shstrtab (or .strtab for Clang) */
1150         shstrtab = find_section_by_name(elf, ".shstrtab");
1151         if (!shstrtab)
1152                 shstrtab = find_section_by_name(elf, ".strtab");
1153         if (!shstrtab) {
1154                 WARN("can't find .shstrtab or .strtab section");
1155                 return NULL;
1156         }
1157         sec->sh.sh_name = elf_add_string(elf, shstrtab, sec->name);
1158         if (sec->sh.sh_name == -1)
1159                 return NULL;
1160
1161         list_add_tail(&sec->list, &elf->sections);
1162         elf_hash_add(section, &sec->hash, sec->idx);
1163         elf_hash_add(section_name, &sec->name_hash, str_hash(sec->name));
1164
1165         mark_sec_changed(elf, sec, true);
1166
1167         return sec;
1168 }
1169
1170 static struct section *elf_create_rela_section(struct elf *elf,
1171                                                struct section *sec,
1172                                                unsigned int reloc_nr)
1173 {
1174         struct section *rsec;
1175         char *rsec_name;
1176
1177         rsec_name = malloc(strlen(sec->name) + strlen(".rela") + 1);
1178         if (!rsec_name) {
1179                 perror("malloc");
1180                 return NULL;
1181         }
1182         strcpy(rsec_name, ".rela");
1183         strcat(rsec_name, sec->name);
1184
1185         rsec = elf_create_section(elf, rsec_name, elf_rela_size(elf), reloc_nr);
1186         free(rsec_name);
1187         if (!rsec)
1188                 return NULL;
1189
1190         rsec->data->d_type = ELF_T_RELA;
1191         rsec->sh.sh_type = SHT_RELA;
1192         rsec->sh.sh_addralign = elf_addr_size(elf);
1193         rsec->sh.sh_link = find_section_by_name(elf, ".symtab")->idx;
1194         rsec->sh.sh_info = sec->idx;
1195         rsec->sh.sh_flags = SHF_INFO_LINK;
1196
1197         rsec->relocs = calloc(sec_num_entries(rsec), sizeof(struct reloc));
1198         if (!rsec->relocs) {
1199                 perror("calloc");
1200                 return NULL;
1201         }
1202
1203         sec->rsec = rsec;
1204         rsec->base = sec;
1205
1206         return rsec;
1207 }
1208
1209 struct section *elf_create_section_pair(struct elf *elf, const char *name,
1210                                         size_t entsize, unsigned int nr,
1211                                         unsigned int reloc_nr)
1212 {
1213         struct section *sec;
1214
1215         sec = elf_create_section(elf, name, entsize, nr);
1216         if (!sec)
1217                 return NULL;
1218
1219         if (!elf_create_rela_section(elf, sec, reloc_nr))
1220                 return NULL;
1221
1222         return sec;
1223 }
1224
1225 int elf_write_insn(struct elf *elf, struct section *sec,
1226                    unsigned long offset, unsigned int len,
1227                    const char *insn)
1228 {
1229         Elf_Data *data = sec->data;
1230
1231         if (data->d_type != ELF_T_BYTE || data->d_off) {
1232                 WARN("write to unexpected data for section: %s", sec->name);
1233                 return -1;
1234         }
1235
1236         memcpy(data->d_buf + offset, insn, len);
1237
1238         mark_sec_changed(elf, sec, true);
1239
1240         return 0;
1241 }
1242
1243 /*
1244  * When Elf_Scn::sh_size is smaller than the combined Elf_Data::d_size
1245  * do you:
1246  *
1247  *   A) adhere to the section header and truncate the data, or
1248  *   B) ignore the section header and write out all the data you've got?
1249  *
1250  * Yes, libelf sucks and we need to manually truncate if we over-allocate data.
1251  */
1252 static int elf_truncate_section(struct elf *elf, struct section *sec)
1253 {
1254         u64 size = sec->sh.sh_size;
1255         bool truncated = false;
1256         Elf_Data *data = NULL;
1257         Elf_Scn *s;
1258
1259         s = elf_getscn(elf->elf, sec->idx);
1260         if (!s) {
1261                 WARN_ELF("elf_getscn");
1262                 return -1;
1263         }
1264
1265         for (;;) {
1266                 /* get next data descriptor for the relevant section */
1267                 data = elf_getdata(s, data);
1268
1269                 if (!data) {
1270                         if (size) {
1271                                 WARN("end of section data but non-zero size left\n");
1272                                 return -1;
1273                         }
1274                         return 0;
1275                 }
1276
1277                 if (truncated) {
1278                         /* when we remove symbols */
1279                         WARN("truncated; but more data\n");
1280                         return -1;
1281                 }
1282
1283                 if (!data->d_size) {
1284                         WARN("zero size data");
1285                         return -1;
1286                 }
1287
1288                 if (data->d_size > size) {
1289                         truncated = true;
1290                         data->d_size = size;
1291                 }
1292
1293                 size -= data->d_size;
1294         }
1295 }
1296
1297 int elf_write(struct elf *elf)
1298 {
1299         struct section *sec;
1300         Elf_Scn *s;
1301
1302         if (opts.dryrun)
1303                 return 0;
1304
1305         /* Update changed relocation sections and section headers: */
1306         list_for_each_entry(sec, &elf->sections, list) {
1307                 if (sec->truncate)
1308                         elf_truncate_section(elf, sec);
1309
1310                 if (sec_changed(sec)) {
1311                         s = elf_getscn(elf->elf, sec->idx);
1312                         if (!s) {
1313                                 WARN_ELF("elf_getscn");
1314                                 return -1;
1315                         }
1316
1317                         /* Note this also flags the section dirty */
1318                         if (!gelf_update_shdr(s, &sec->sh)) {
1319                                 WARN_ELF("gelf_update_shdr");
1320                                 return -1;
1321                         }
1322
1323                         mark_sec_changed(elf, sec, false);
1324                 }
1325         }
1326
1327         /* Make sure the new section header entries get updated properly. */
1328         elf_flagelf(elf->elf, ELF_C_SET, ELF_F_DIRTY);
1329
1330         /* Write all changes to the file. */
1331         if (elf_update(elf->elf, ELF_C_WRITE) < 0) {
1332                 WARN_ELF("elf_update");
1333                 return -1;
1334         }
1335
1336         elf->changed = false;
1337
1338         return 0;
1339 }
1340
1341 void elf_close(struct elf *elf)
1342 {
1343         if (elf->elf)
1344                 elf_end(elf->elf);
1345
1346         if (elf->fd > 0)
1347                 close(elf->fd);
1348
1349         /*
1350          * NOTE: All remaining allocations are leaked on purpose.  Objtool is
1351          * about to exit anyway.
1352          */
1353 }