build/make/obj_int_extract.c

   1 /*
   2  *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
   3  *
   4  *  Use of this source code is governed by a BSD-style license
   5  *  that can be found in the LICENSE file in the root of the source
   6  *  tree. An additional intellectual property rights grant can be found
   7  *  in the file PATENTS.  All contributing project authors may
   8  *  be found in the AUTHORS file in the root of the source tree.
   9  */
  10
  11
  12 #include <stdarg.h>
  13 #include <stdio.h>
  14 #include <stdlib.h>
  15 #include <string.h>
  16
  17 #include "vpx_config.h"
  18 #include "vpx/vpx_integer.h"
  19
  20 typedef enum
  21 {
  22     OUTPUT_FMT_PLAIN,
  23     OUTPUT_FMT_RVDS,
  24     OUTPUT_FMT_GAS,
  25 } output_fmt_t;
  26
  27 int log_msg(const char *fmt, ...)
  28 {
  29     int res;
  30     va_list ap;
  31     va_start(ap, fmt);
  32     res = vfprintf(stderr, fmt, ap);
  33     va_end(ap);
  34     return res;
  35 }
  36
  37 #if defined(__GNUC__) && __GNUC__
  38 #if defined(__MACH__)
  39
  40 #include <mach-o/loader.h>
  41 #include <mach-o/nlist.h>
  42
  43 int parse_macho(uint8_t *base_buf, size_t sz)
  44 {
  45     int i, j;
  46     struct mach_header header;
  47     uint8_t *buf = base_buf;
  48     int base_data_section = 0;
  49     int bits = 0;
  50
  51     /* We can read in mach_header for 32 and 64 bit architectures
  52      * because it's identical to mach_header_64 except for the last
  53      * element (uint32_t reserved), which we don't use. Then, when
  54      * we know which architecture we're looking at, increment buf
  55      * appropriately.
  56      */
  57     memcpy(&header, buf, sizeof(struct mach_header));
  58
  59     if (header.magic == MH_MAGIC)
  60     {
  61         if (header.cputype == CPU_TYPE_ARM
  62             || header.cputype == CPU_TYPE_X86)
  63         {
  64             bits = 32;
  65             buf += sizeof(struct mach_header);
  66         }
  67         else
  68         {
  69             log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_[ARM|X86].\n");
  70             goto bail;
  71         }
  72     }
  73     else if (header.magic == MH_MAGIC_64)
  74     {
  75         if (header.cputype == CPU_TYPE_X86_64)
  76         {
  77             bits = 64;
  78             buf += sizeof(struct mach_header_64);
  79         }
  80         else
  81         {
  82             log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_X86_64.\n");
  83             goto bail;
  84         }
  85     }
  86     else
  87     {
  88         log_msg("Bad magic number for object file. 0x%x or 0x%x expected, 0x%x found.\n",
  89                 MH_MAGIC, MH_MAGIC_64, header.magic);
  90         goto bail;
  91     }
  92
  93     if (header.filetype != MH_OBJECT)
  94     {
  95         log_msg("Bad filetype for object file. Currently only tested for MH_OBJECT.\n");
  96         goto bail;
  97     }
  98
  99     for (i = 0; i < header.ncmds; i++)
 100     {
 101         struct load_command lc;
 102
 103         memcpy(&lc, buf, sizeof(struct load_command));
 104
 105         if (lc.cmd == LC_SEGMENT)
 106         {
 107             uint8_t *seg_buf = buf;
 108             struct section s;
 109             struct segment_command seg_c;
 110
 111             memcpy(&seg_c, seg_buf, sizeof(struct segment_command));
 112             seg_buf += sizeof(struct segment_command);
 113
 114             /* Although each section is given it's own offset, nlist.n_value
 115              * references the offset of the first section. This isn't
 116              * apparent without debug information because the offset of the
 117              * data section is the same as the first section. However, with
 118              * debug sections mixed in, the offset of the debug section
 119              * increases but n_value still references the first section.
 120              */
 121             if (seg_c.nsects < 1)
 122             {
 123                 log_msg("Not enough sections\n");
 124                 goto bail;
 125             }
 126
 127             memcpy(&s, seg_buf, sizeof(struct section));
 128             base_data_section = s.offset;
 129         }
 130         else if (lc.cmd == LC_SEGMENT_64)
 131         {
 132             uint8_t *seg_buf = buf;
 133             struct section_64 s;
 134             struct segment_command_64 seg_c;
 135
 136             memcpy(&seg_c, seg_buf, sizeof(struct segment_command_64));
 137             seg_buf += sizeof(struct segment_command_64);
 138
 139             /* Explanation in LG_SEGMENT */
 140             if (seg_c.nsects < 1)
 141             {
 142                 log_msg("Not enough sections\n");
 143                 goto bail;
 144             }
 145
 146             memcpy(&s, seg_buf, sizeof(struct section_64));
 147             base_data_section = s.offset;
 148         }
 149         else if (lc.cmd == LC_SYMTAB)
 150         {
 151             if (base_data_section != 0)
 152             {
 153                 struct symtab_command sc;
 154                 uint8_t *sym_buf = base_buf;
 155                 uint8_t *str_buf = base_buf;
 156
 157                 memcpy(&sc, buf, sizeof(struct symtab_command));
 158
 159                 if (sc.cmdsize != sizeof(struct symtab_command))
 160                 {
 161                     log_msg("Can't find symbol table!\n");
 162                     goto bail;
 163                 }
 164
 165                 sym_buf += sc.symoff;
 166                 str_buf += sc.stroff;
 167
 168                 for (j = 0; j < sc.nsyms; j++)
 169                 {
 170                     /* Location of string is cacluated each time from the
 171                      * start of the string buffer.  On darwin the symbols
 172                      * are prefixed by "_", so we bump the pointer by 1.
 173                      * The target value is defined as an int in asm_*_offsets.c,
 174                      * which is 4 bytes on all targets we currently use.
 175                      */
 176                     if (bits == 32)
 177                     {
 178                         struct nlist nl;
 179                         int val;
 180
 181                         memcpy(&nl, sym_buf, sizeof(struct nlist));
 182                         sym_buf += sizeof(struct nlist);
 183
 184                         memcpy(&val, base_buf + base_data_section + nl.n_value,
 185                                sizeof(val));
 186                         printf("%-40s EQU %5d\n",
 187                                str_buf + nl.n_un.n_strx + 1, val);
 188                     }
 189                     else /* if (bits == 64) */
 190                     {
 191                         struct nlist_64 nl;
 192                         int val;
 193
 194                         memcpy(&nl, sym_buf, sizeof(struct nlist_64));
 195                         sym_buf += sizeof(struct nlist_64);
 196
 197                         memcpy(&val, base_buf + base_data_section + nl.n_value,
 198                                sizeof(val));
 199                         printf("%-40s EQU %5d\n",
 200                                str_buf + nl.n_un.n_strx + 1, val);
 201                     }
 202                 }
 203             }
 204         }
 205
 206         buf += lc.cmdsize;
 207     }
 208
 209     return 0;
 210 bail:
 211     return 1;
 212
 213 }
 214
 215 #elif defined(__ELF__)
 216 #include "elf.h"
 217
 218 #define COPY_STRUCT(dst, buf, ofst, sz) do {\
 219         if(ofst + sizeof((*(dst))) > sz) goto bail;\
 220         memcpy(dst, buf+ofst, sizeof((*(dst))));\
 221     } while(0)
 222
 223 #define ENDIAN_ASSIGN(val, memb) do {\
 224         if(!elf->le_data) {log_msg("Big Endian data not supported yet!\n");goto bail;}\
 225         (val) = (memb);\
 226     } while(0)
 227
 228 #define ENDIAN_ASSIGN_IN_PLACE(memb) do {\
 229         ENDIAN_ASSIGN(memb, memb);\
 230     } while(0)
 231
 232 typedef struct
 233 {
 234     uint8_t      *buf; /* Buffer containing ELF data */
 235     size_t        sz;  /* Buffer size */
 236     int           le_data; /* Data is little-endian */
 237     unsigned char e_ident[EI_NIDENT]; /* Magic number and other info */
 238     int           bits; /* 32 or 64 */
 239     Elf32_Ehdr    hdr32;
 240     Elf64_Ehdr    hdr64;
 241 } elf_obj_t;
 242
 243 int parse_elf_header(elf_obj_t *elf)
 244 {
 245     int res;
 246     /* Verify ELF Magic numbers */
 247     COPY_STRUCT(&elf->e_ident, elf->buf, 0, elf->sz);
 248     res = elf->e_ident[EI_MAG0] == ELFMAG0;
 249     res &= elf->e_ident[EI_MAG1] == ELFMAG1;
 250     res &= elf->e_ident[EI_MAG2] == ELFMAG2;
 251     res &= elf->e_ident[EI_MAG3] == ELFMAG3;
 252     res &= elf->e_ident[EI_CLASS] == ELFCLASS32
 253         || elf->e_ident[EI_CLASS] == ELFCLASS64;
 254     res &= elf->e_ident[EI_DATA] == ELFDATA2LSB;
 255
 256     if (!res) goto bail;
 257
 258     elf->le_data = elf->e_ident[EI_DATA] == ELFDATA2LSB;
 259
 260     /* Read in relevant values */
 261     if (elf->e_ident[EI_CLASS] == ELFCLASS32)
 262     {
 263         elf->bits = 32;
 264         COPY_STRUCT(&elf->hdr32, elf->buf, 0, elf->sz);
 265
 266         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_type);
 267         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_machine);
 268         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_version);
 269         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_entry);
 270         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phoff);
 271         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shoff);
 272         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_flags);
 273         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_ehsize);
 274         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phentsize);
 275         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phnum);
 276         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shentsize);
 277         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shnum);
 278         ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shstrndx);
 279     }
 280     else /* if (elf->e_ident[EI_CLASS] == ELFCLASS64) */
 281     {
 282         elf->bits = 64;
 283         COPY_STRUCT(&elf->hdr64, elf->buf, 0, elf->sz);
 284
 285         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_type);
 286         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_machine);
 287         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_version);
 288         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_entry);
 289         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phoff);
 290         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shoff);
 291         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_flags);
 292         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_ehsize);
 293         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phentsize);
 294         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phnum);
 295         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shentsize);
 296         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shnum);
 297         ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shstrndx);
 298     }
 299
 300     return 0;
 301 bail:
 302     log_msg("Failed to parse ELF file header");
 303     return 1;
 304 }
 305
 306 int parse_elf_section(elf_obj_t *elf, int idx, Elf32_Shdr *hdr32, Elf64_Shdr *hdr64)
 307 {
 308     if (hdr32)
 309     {
 310         if (idx >= elf->hdr32.e_shnum)
 311             goto bail;
 312
 313         COPY_STRUCT(hdr32, elf->buf, elf->hdr32.e_shoff + idx * elf->hdr32.e_shentsize,
 314                     elf->sz);
 315         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_name);
 316         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_type);
 317         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_flags);
 318         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addr);
 319         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_offset);
 320         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_size);
 321         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_link);
 322         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_info);
 323         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addralign);
 324         ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_entsize);
 325     }
 326     else /* if (hdr64) */
 327     {
 328         if (idx >= elf->hdr64.e_shnum)
 329             goto bail;
 330
 331         COPY_STRUCT(hdr64, elf->buf, elf->hdr64.e_shoff + idx * elf->hdr64.e_shentsize,
 332                     elf->sz);
 333         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_name);
 334         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_type);
 335         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_flags);
 336         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addr);
 337         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_offset);
 338         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_size);
 339         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_link);
 340         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_info);
 341         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addralign);
 342         ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_entsize);
 343     }
 344
 345     return 0;
 346 bail:
 347     return 1;
 348 }
 349
 350 char *parse_elf_string_table(elf_obj_t *elf, int s_idx, int idx)
 351 {
 352     if (elf->bits == 32)
 353     {
 354         Elf32_Shdr shdr;
 355
 356         if (parse_elf_section(elf, s_idx, &shdr, NULL))
 357         {
 358             log_msg("Failed to parse ELF string table: section %d, index %d\n",
 359                     s_idx, idx);
 360             return "";
 361         }
 362
 363         return (char *)(elf->buf + shdr.sh_offset + idx);
 364     }
 365     else /* if (elf->bits == 64) */
 366     {
 367         Elf64_Shdr shdr;
 368
 369         if (parse_elf_section(elf, s_idx, NULL, &shdr))
 370         {
 371             log_msg("Failed to parse ELF string table: section %d, index %d\n",
 372                     s_idx, idx);
 373             return "";
 374         }
 375
 376         return (char *)(elf->buf + shdr.sh_offset + idx);
 377     }
 378 }
 379
 380 int parse_elf_symbol(elf_obj_t *elf, unsigned int ofst, Elf32_Sym *sym32, Elf64_Sym *sym64)
 381 {
 382     if (sym32)
 383     {
 384         COPY_STRUCT(sym32, elf->buf, ofst, elf->sz);
 385         ENDIAN_ASSIGN_IN_PLACE(sym32->st_name);
 386         ENDIAN_ASSIGN_IN_PLACE(sym32->st_value);
 387         ENDIAN_ASSIGN_IN_PLACE(sym32->st_size);
 388         ENDIAN_ASSIGN_IN_PLACE(sym32->st_info);
 389         ENDIAN_ASSIGN_IN_PLACE(sym32->st_other);
 390         ENDIAN_ASSIGN_IN_PLACE(sym32->st_shndx);
 391     }
 392     else /* if (sym64) */
 393     {
 394         COPY_STRUCT(sym64, elf->buf, ofst, elf->sz);
 395         ENDIAN_ASSIGN_IN_PLACE(sym64->st_name);
 396         ENDIAN_ASSIGN_IN_PLACE(sym64->st_value);
 397         ENDIAN_ASSIGN_IN_PLACE(sym64->st_size);
 398         ENDIAN_ASSIGN_IN_PLACE(sym64->st_info);
 399         ENDIAN_ASSIGN_IN_PLACE(sym64->st_other);
 400         ENDIAN_ASSIGN_IN_PLACE(sym64->st_shndx);
 401     }
 402     return 0;
 403 bail:
 404     return 1;
 405 }
 406
 407 int parse_elf(uint8_t *buf, size_t sz, output_fmt_t mode)
 408 {
 409     elf_obj_t    elf;
 410     unsigned int ofst;
 411     int          i;
 412     Elf32_Off    strtab_off32;
 413     Elf64_Off    strtab_off64; /* save String Table offset for later use */
 414
 415     memset(&elf, 0, sizeof(elf));
 416     elf.buf = buf;
 417     elf.sz = sz;
 418
 419     /* Parse Header */
 420     if (parse_elf_header(&elf))
 421       goto bail;
 422
 423     if (elf.bits == 32)
 424     {
 425         Elf32_Shdr shdr;
 426         for (i = 0; i < elf.hdr32.e_shnum; i++)
 427         {
 428             parse_elf_section(&elf, i, &shdr, NULL);
 429
 430             if (shdr.sh_type == SHT_STRTAB)
 431             {
 432                 char strtsb_name[128];
 433
 434                 strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name));
 435
 436                 if (!(strcmp(strtsb_name, ".shstrtab")))
 437                 {
 438                     /* log_msg("found section: %s\n", strtsb_name); */
 439                     strtab_off32 = shdr.sh_offset;
 440                     break;
 441                 }
 442             }
 443         }
 444     }
 445     else /* if (elf.bits == 64) */
 446     {
 447         Elf64_Shdr shdr;
 448         for (i = 0; i < elf.hdr64.e_shnum; i++)
 449         {
 450             parse_elf_section(&elf, i, NULL, &shdr);
 451
 452             if (shdr.sh_type == SHT_STRTAB)
 453             {
 454                 char strtsb_name[128];
 455
 456                 strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name));
 457
 458                 if (!(strcmp(strtsb_name, ".shstrtab")))
 459                 {
 460                     /* log_msg("found section: %s\n", strtsb_name); */
 461                     strtab_off64 = shdr.sh_offset;
 462                     break;
 463                 }
 464             }
 465         }
 466     }
 467
 468     /* Parse all Symbol Tables */
 469     if (elf.bits == 32)
 470     {
 471         Elf32_Shdr shdr;
 472         for (i = 0; i < elf.hdr32.e_shnum; i++)
 473         {
 474             parse_elf_section(&elf, i, &shdr, NULL);
 475
 476             if (shdr.sh_type == SHT_SYMTAB)
 477             {
 478                 for (ofst = shdr.sh_offset;
 479                      ofst < shdr.sh_offset + shdr.sh_size;
 480                      ofst += shdr.sh_entsize)
 481                 {
 482                     Elf32_Sym sym;
 483
 484                     parse_elf_symbol(&elf, ofst, &sym, NULL);
 485
 486                     /* For all OBJECTS (data objects), extract the value from the
 487                      * proper data segment.
 488                      */
 489                     /* if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name)
 490                         log_msg("found data object %s\n",
 491                                 parse_elf_string_table(&elf,
 492                                                        shdr.sh_link,
 493                                                        sym.st_name));
 494                      */
 495
 496                     if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT
 497                         && sym.st_size == 4)
 498                     {
 499                         Elf32_Shdr dhdr;
 500                         int val = 0;
 501                         char section_name[128];
 502
 503                         parse_elf_section(&elf, sym.st_shndx, &dhdr, NULL);
 504
 505                         /* For explanition - refer to _MSC_VER version of code */
 506                         strcpy(section_name, (char *)(elf.buf + strtab_off32 + dhdr.sh_name));
 507                         /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */
 508
 509                         if (strcmp(section_name, ".bss"))
 510                         {
 511                             if (sizeof(val) != sym.st_size)
 512                             {
 513                                 /* The target value is declared as an int in
 514                                  * asm_*_offsets.c, which is 4 bytes on all
 515                                  * targets we currently use. Complain loudly if
 516                                  * this is not true.
 517                                  */
 518                                 log_msg("Symbol size is wrong\n");
 519                                 goto bail;
 520                             }
 521
 522                             memcpy(&val,
 523                                    elf.buf + dhdr.sh_offset + sym.st_value,
 524                                    sym.st_size);
 525                         }
 526
 527                         if (!elf.le_data)
 528                         {
 529                             log_msg("Big Endian data not supported yet!\n");
 530                             goto bail;
 531                         }
 532
 533                         switch (mode)
 534                         {
 535                             case OUTPUT_FMT_RVDS:
 536                                 printf("%-40s EQU %5d\n",
 537                                        parse_elf_string_table(&elf,
 538                                                               shdr.sh_link,
 539                                                               sym.st_name),
 540                                        val);
 541                                 break;
 542                             case OUTPUT_FMT_GAS:
 543                                 printf(".equ %-40s, %5d\n",
 544                                        parse_elf_string_table(&elf,
 545                                                               shdr.sh_link,
 546                                                               sym.st_name),
 547                                        val);
 548                                 break;
 549                             default:
 550                                 printf("%s = %d\n",
 551                                        parse_elf_string_table(&elf,
 552                                                               shdr.sh_link,
 553                                                               sym.st_name),
 554                                        val);
 555                         }
 556                     }
 557                 }
 558             }
 559         }
 560     }
 561     else /* if (elf.bits == 64) */
 562     {
 563         Elf64_Shdr shdr;
 564         for (i = 0; i < elf.hdr64.e_shnum; i++)
 565         {
 566             parse_elf_section(&elf, i, NULL, &shdr);
 567
 568             if (shdr.sh_type == SHT_SYMTAB)
 569             {
 570                 for (ofst = shdr.sh_offset;
 571                      ofst < shdr.sh_offset + shdr.sh_size;
 572                      ofst += shdr.sh_entsize)
 573                 {
 574                     Elf64_Sym sym;
 575
 576                     parse_elf_symbol(&elf, ofst, NULL, &sym);
 577
 578                     /* For all OBJECTS (data objects), extract the value from the
 579                      * proper data segment.
 580                      */
 581                     /* if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name)
 582                         log_msg("found data object %s\n",
 583                                 parse_elf_string_table(&elf,
 584                                                        shdr.sh_link,
 585                                                        sym.st_name));
 586                      */
 587
 588                     if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT
 589                         && sym.st_size == 4)
 590                     {
 591                         Elf64_Shdr dhdr;
 592                         int val = 0;
 593                         char section_name[128];
 594
 595                         parse_elf_section(&elf, sym.st_shndx, NULL, &dhdr);
 596
 597                         /* For explanition - refer to _MSC_VER version of code */
 598                         strcpy(section_name, (char *)(elf.buf + strtab_off64 + dhdr.sh_name));
 599                         /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */
 600
 601                         if ((strcmp(section_name, ".bss")))
 602                         {
 603                             if (sizeof(val) != sym.st_size)
 604                             {
 605                                 /* The target value is declared as an int in
 606                                  * asm_*_offsets.c, which is 4 bytes on all
 607                                  * targets we currently use. Complain loudly if
 608                                  * this is not true.
 609                                  */
 610                                 log_msg("Symbol size is wrong\n");
 611                                 goto bail;
 612                             }
 613
 614                             memcpy(&val,
 615                                    elf.buf + dhdr.sh_offset + sym.st_value,
 616                                    sym.st_size);
 617                         }
 618
 619                         if (!elf.le_data)
 620                         {
 621                             log_msg("Big Endian data not supported yet!\n");
 622                             goto bail;
 623                         }
 624
 625                         switch (mode)
 626                         {
 627                             case OUTPUT_FMT_RVDS:
 628                                 printf("%-40s EQU %5d\n",
 629                                        parse_elf_string_table(&elf,
 630                                                               shdr.sh_link,
 631                                                               sym.st_name),
 632                                        val);
 633                                 break;
 634                             case OUTPUT_FMT_GAS:
 635                                 printf(".equ %-40s, %5d\n",
 636                                        parse_elf_string_table(&elf,
 637                                                               shdr.sh_link,
 638                                                               sym.st_name),
 639                                        val);
 640                                 break;
 641                             default:
 642                                 printf("%s = %d\n",
 643                                        parse_elf_string_table(&elf,
 644                                                               shdr.sh_link,
 645                                                               sym.st_name),
 646                                        val);
 647                         }
 648                     }
 649                 }
 650             }
 651         }
 652     }
 653
 654     if (mode == OUTPUT_FMT_RVDS)
 655         printf("    END\n");
 656
 657     return 0;
 658 bail:
 659     log_msg("Parse error: File does not appear to be valid ELF32 or ELF64\n");
 660     return 1;
 661 }
 662
 663 #endif
 664 #endif /* defined(__GNUC__) && __GNUC__ */
 665
 666
 667 #if defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__)
 668 /*  See "Microsoft Portable Executable and Common Object File Format Specification"
 669     for reference.
 670 */
 671 #define get_le32(x) ((*(x)) | (*(x+1)) << 8 |(*(x+2)) << 16 | (*(x+3)) << 24 )
 672 #define get_le16(x) ((*(x)) | (*(x+1)) << 8)
 673
 674 int parse_coff(uint8_t *buf, size_t sz)
 675 {
 676     unsigned int nsections, symtab_ptr, symtab_sz, strtab_ptr;
 677     unsigned int sectionrawdata_ptr;
 678     unsigned int i;
 679     uint8_t *ptr;
 680     uint32_t symoffset;
 681
 682     char **sectionlist;  //this array holds all section names in their correct order.
 683     //it is used to check if the symbol is in .bss or .rdata section.
 684
 685     nsections = get_le16(buf + 2);
 686     symtab_ptr = get_le32(buf + 8);
 687     symtab_sz = get_le32(buf + 12);
 688     strtab_ptr = symtab_ptr + symtab_sz * 18;
 689
 690     if (nsections > 96)
 691     {
 692         log_msg("Too many sections\n");
 693         return 1;
 694     }
 695
 696     sectionlist = malloc(nsections * sizeof(sectionlist));
 697
 698     if (sectionlist == NULL)
 699     {
 700         log_msg("Allocating first level of section list failed\n");
 701         return 1;
 702     }
 703
 704     //log_msg("COFF: Found %u symbols in %u sections.\n", symtab_sz, nsections);
 705
 706     /*
 707     The size of optional header is always zero for an obj file. So, the section header
 708     follows the file header immediately.
 709     */
 710
 711     ptr = buf + 20;     //section header
 712
 713     for (i = 0; i < nsections; i++)
 714     {
 715         char sectionname[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0};
 716         strncpy(sectionname, ptr, 8);
 717         //log_msg("COFF: Parsing section %s\n",sectionname);
 718
 719         sectionlist[i] = malloc(strlen(sectionname) + 1);
 720
 721         if (sectionlist[i] == NULL)
 722         {
 723             log_msg("Allocating storage for %s failed\n", sectionname);
 724             goto bail;
 725         }
 726         strcpy(sectionlist[i], sectionname);
 727
 728         if (!strcmp(sectionname, ".rdata")) sectionrawdata_ptr = get_le32(ptr + 20);
 729
 730         ptr += 40;
 731     }
 732
 733     //log_msg("COFF: Symbol table at offset %u\n", symtab_ptr);
 734     //log_msg("COFF: raw data pointer ofset for section .rdata is %u\n", sectionrawdata_ptr);
 735
 736     /*  The compiler puts the data with non-zero offset in .rdata section, but puts the data with
 737         zero offset in .bss section. So, if the data in in .bss section, set offset=0.
 738         Note from Wiki: In an object module compiled from C, the bss section contains
 739         the local variables (but not functions) that were declared with the static keyword,
 740         except for those with non-zero initial values. (In C, static variables are initialized
 741         to zero by default.) It also contains the non-local (both extern and static) variables
 742         that are also initialized to zero (either explicitly or by default).
 743         */
 744     //move to symbol table
 745     /* COFF symbol table:
 746         offset      field
 747         0           Name(*)
 748         8           Value
 749         12          SectionNumber
 750         14          Type
 751         16          StorageClass
 752         17          NumberOfAuxSymbols
 753         */
 754     ptr = buf + symtab_ptr;
 755
 756     for (i = 0; i < symtab_sz; i++)
 757     {
 758         int16_t section = get_le16(ptr + 12); //section number
 759
 760         if (section > 0 && ptr[16] == 2)
 761         {
 762             //if(section > 0 && ptr[16] == 3 && get_le32(ptr+8)) {
 763
 764             if (get_le32(ptr))
 765             {
 766                 char name[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0};
 767                 strncpy(name, ptr, 8);
 768                 //log_msg("COFF: Parsing symbol %s\n",name);
 769                 /* The 64bit Windows compiler doesn't prefix with an _.
 770                  * Check what's there, and bump if necessary
 771                  */
 772                 if (name[0] == '_')
 773                     printf("%-40s EQU ", name + 1);
 774                 else
 775                     printf("%-40s EQU ", name);
 776             }
 777             else
 778             {
 779                 //log_msg("COFF: Parsing symbol %s\n",
 780                 //        buf + strtab_ptr + get_le32(ptr+4));
 781                 if ((buf + strtab_ptr + get_le32(ptr + 4))[0] == '_')
 782                     printf("%-40s EQU ",
 783                            buf + strtab_ptr + get_le32(ptr + 4) + 1);
 784                 else
 785                     printf("%-40s EQU ", buf + strtab_ptr + get_le32(ptr + 4));
 786             }
 787
 788             if (!(strcmp(sectionlist[section-1], ".bss")))
 789             {
 790                 symoffset = 0;
 791             }
 792             else
 793             {
 794                 symoffset = get_le32(buf + sectionrawdata_ptr + get_le32(ptr + 8));
 795             }
 796
 797             //log_msg("      Section: %d\n",section);
 798             //log_msg("      Class:   %d\n",ptr[16]);
 799             //log_msg("      Address: %u\n",get_le32(ptr+8));
 800             //log_msg("      Offset: %u\n", symoffset);
 801
 802             printf("%5d\n", symoffset);
 803         }
 804
 805         ptr += 18;
 806     }
 807
 808     printf("    END\n");
 809
 810     for (i = 0; i < nsections; i++)
 811     {
 812         free(sectionlist[i]);
 813     }
 814
 815     free(sectionlist);
 816
 817     return 0;
 818 bail:
 819
 820     for (i = 0; i < nsections; i++)
 821     {
 822         free(sectionlist[i]);
 823     }
 824
 825     free(sectionlist);
 826
 827     return 1;
 828 }
 829 #endif /* defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__) */
 830
 831 int main(int argc, char **argv)
 832 {
 833     output_fmt_t mode = OUTPUT_FMT_PLAIN;
 834     const char *f;
 835     uint8_t *file_buf;
 836     int res;
 837     FILE *fp;
 838     long int file_size;
 839
 840     if (argc < 2 || argc > 3)
 841     {
 842         fprintf(stderr, "Usage: %s [output format] <obj file>\n\n", argv[0]);
 843         fprintf(stderr, "  <obj file>\tobject file to parse\n");
 844         fprintf(stderr, "Output Formats:\n");
 845         fprintf(stderr, "  gas  - compatible with GNU assembler\n");
 846         fprintf(stderr, "  rvds - compatible with armasm\n");
 847         goto bail;
 848     }
 849
 850     f = argv[2];
 851
 852     if (!strcmp(argv[1], "rvds"))
 853         mode = OUTPUT_FMT_RVDS;
 854     else if (!strcmp(argv[1], "gas"))
 855         mode = OUTPUT_FMT_GAS;
 856     else
 857         f = argv[1];
 858
 859     fp = fopen(f, "rb");
 860
 861     if (!fp)
 862     {
 863         perror("Unable to open file");
 864         goto bail;
 865     }
 866
 867     if (fseek(fp, 0, SEEK_END))
 868     {
 869         perror("stat");
 870         goto bail;
 871     }
 872
 873     file_size = ftell(fp);
 874     file_buf = malloc(file_size);
 875
 876     if (!file_buf)
 877     {
 878         perror("malloc");
 879         goto bail;
 880     }
 881
 882     rewind(fp);
 883
 884     if (fread(file_buf, sizeof(char), file_size, fp) != file_size)
 885     {
 886         perror("read");
 887         goto bail;
 888     }
 889
 890     if (fclose(fp))
 891     {
 892         perror("close");
 893         goto bail;
 894     }
 895
 896 #if defined(__GNUC__) && __GNUC__
 897 #if defined(__MACH__)
 898     res = parse_macho(file_buf, file_size);
 899 #elif defined(__ELF__)
 900     res = parse_elf(file_buf, file_size, mode);
 901 #endif
 902 #endif
 903 #if defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__)
 904     res = parse_coff(file_buf, file_size);
 905 #endif
 906
 907     free(file_buf);
 908
 909     if (!res)
 910         return EXIT_SUCCESS;
 911
 912 bail:
 913     return EXIT_FAILURE;
 914 }