tools/sstrip/src/sstrip.c

   1 /* http://www.muppetlabs.com/~breadbox/software/elfkickers.html */
   2
   3 /* sstrip: Copyright (C) 1999-2001 by Brian Raiter, under the GNU
   4  * General Public License. No warranty. See COPYING for details.
   5  *
   6  * Aug 23, 2004 Hacked by Manuel Novoa III <mjn3@codepoet.org> to
   7  * handle targets of different endianness and/or elf class, making
   8  * it more useful in a cross-devel environment.
   9  */
  10
  11 /* ============== original README ===================
  12  *
  13  * sstrip is a small utility that removes the contents at the end of an
  14  * ELF file that are not part of the program's memory image.
  15  *
  16  * Most ELF executables are built with both a program header table and a
  17  * section header table. However, only the former is required in order
  18  * for the OS to load, link and execute a program. sstrip attempts to
  19  * extract the ELF header, the program header table, and its contents,
  20  * leaving everything else in the bit bucket. It can only remove parts of
  21  * the file that occur at the end, after the parts to be saved. However,
  22  * this almost always includes the section header table, and occasionally
  23  * a few random sections that are not used when running a program.
  24  *
  25  * It should be noted that the GNU bfd library is (understandably)
  26  * dependent on the section header table as an index to the file's
  27  * contents. Thus, an executable file that has no section header table
  28  * cannot be used with gdb, objdump, or any other program based upon the
  29  * bfd library, at all. In fact, the program will not even recognize the
  30  * file as a valid executable. (This limitation is noted in the source
  31  * code comments for bfd, and is marked "FIXME", so this may change at
  32  * some future date. However, I would imagine that it is a pretty
  33  * low-priority item, as executables without a section header table are
  34  * rare in the extreme.) This probably also explains why strip doesn't
  35  * offer the option to do this.
  36  *
  37  * Shared library files may also have their section header table removed.
  38  * Such a library will still function; however, it will no longer be
  39  * possible for a compiler to link a new program against it.
  40  *
  41  * As an added bonus, sstrip also tries to removes trailing zero bytes
  42  * from the end of the file. (This normally cannot be done with an
  43  * executable that has a section header table.)
  44  *
  45  * sstrip is a very simplistic program. It depends upon the common
  46  * practice of putting the parts of the file that contribute to the
  47  * memory image at the front, and the remaining material at the end. This
  48  * permits it to discard the latter material without affecting file
  49  * offsets and memory addresses in what remains. Of course, the ELF
  50  * standard permits files to be organized in almost any order, so if a
  51  * pathological linker decided to put its section headers at the top,
  52  * sstrip would be useless on such executables.
  53  */
  54
  55 #include        <stdio.h>
  56 #include        <stdlib.h>
  57 #include        <string.h>
  58 #include        <errno.h>
  59 #include        <unistd.h>
  60 #include        <fcntl.h>
  61 #include        <elf.h>
  62 #include        <byteswap.h>
  63
  64 #ifndef TRUE
  65 #define TRUE            1
  66 #define FALSE           0
  67 #endif
  68
  69 /* The name of the program.
  70  */
  71 static char const       *progname;
  72
  73 /* The name of the current file.
  74  */
  75 static char const       *filename;
  76
  77
  78 /* A simple error-handling function. FALSE is always returned for the
  79  * convenience of the caller.
  80  */
  81 static int err(char const *errmsg)
  82 {
  83         fprintf(stderr, "%s: %s: %s\n", progname, filename, errmsg);
  84         return FALSE;
  85 }
  86
  87 /* A flag to signal the need for endian reversal.
  88  */
  89 static int do_reverse_endian;
  90
  91 /* Get a value from the elf header, compensating for endianness.
  92  */
  93 #define EGET(X) \
  94         (__extension__ ({ \
  95                 uint64_t __res; \
  96                 if (!do_reverse_endian) { \
  97                         __res = (X); \
  98                 } else if (sizeof(X) == 1) { \
  99                         __res = (X); \
 100                 } else if (sizeof(X) == 2) { \
 101                         __res = bswap_16((X)); \
 102                 } else if (sizeof(X) == 4) { \
 103                         __res = bswap_32((X)); \
 104                 } else if (sizeof(X) == 8) { \
 105                         __res = bswap_64((X)); \
 106                 } else { \
 107                         fprintf(stderr, "%s: %s: EGET failed for size %d\n", \
 108                                         progname, filename, sizeof(X)); \
 109                         exit(EXIT_FAILURE); \
 110                 } \
 111                 __res; \
 112         }))
 113
 114 /* Set a value 'Y' in the elf header to 'X', compensating for endianness.
 115  */
 116 #define ESET(Y,X) \
 117                 do if (!do_reverse_endian) { \
 118                         Y = (X); \
 119                 } else if (sizeof(Y) == 1) { \
 120                         Y = (X); \
 121                 } else if (sizeof(Y) == 2) { \
 122                         Y = bswap_16((uint16_t)(X)); \
 123                 } else if (sizeof(Y) == 4) { \
 124                         Y = bswap_32((uint32_t)(X)); \
 125                 } else if (sizeof(Y) == 8) { \
 126                         Y = bswap_64((uint64_t)(X)); \
 127                 } else { \
 128                         fprintf(stderr, "%s: %s: ESET failed for size %d\n", \
 129                                         progname, filename, sizeof(Y)); \
 130                         exit(EXIT_FAILURE); \
 131                 } while (0)
 132
 133
 134 /* A macro for I/O errors: The given error message is used only when
 135  * errno is not set.
 136  */
 137 #define ferr(msg)       (err(errno ? strerror(errno) : (msg)))
 138
 139
 140
 141 #define HEADER_FUNCTIONS(CLASS) \
 142  \
 143 /* readelfheader() reads the ELF header into our global variable, and \
 144  * checks to make sure that this is in fact a file that we should be \
 145  * munging. \
 146  */ \
 147 static int readelfheader ## CLASS (int fd, Elf ## CLASS ## _Ehdr *ehdr) \
 148 { \
 149          if (read(fd, ((char *)ehdr)+EI_NIDENT, sizeof(*ehdr) - EI_NIDENT) \
 150                 != sizeof(*ehdr) - EI_NIDENT) \
 151                 return ferr("missing or incomplete ELF header."); \
 152  \
 153         /* Verify the sizes of the ELF header and the program segment \
 154          * header table entries. \
 155          */ \
 156         if (EGET(ehdr->e_ehsize) != sizeof(Elf ## CLASS ## _Ehdr)) \
 157                 return err("unrecognized ELF header size."); \
 158         if (EGET(ehdr->e_phentsize) != sizeof(Elf ## CLASS ## _Phdr)) \
 159                 return err("unrecognized program segment header size."); \
 160  \
 161         /* Finally, check the file type. \
 162          */ \
 163         if (EGET(ehdr->e_type) != ET_EXEC && EGET(ehdr->e_type) != ET_DYN) \
 164                 return err("not an executable or shared-object library."); \
 165  \
 166         return TRUE; \
 167 } \
 168  \
 169 /* readphdrtable() loads the program segment header table into memory. \
 170  */ \
 171 static int readphdrtable ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 172                                                                    Elf ## CLASS ## _Phdr **phdrs) \
 173 { \
 174         size_t  size; \
 175  \
 176         if (!EGET(ehdr->e_phoff) || !EGET(ehdr->e_phnum) \
 177 )               return err("ELF file has no program header table."); \
 178  \
 179         size = EGET(ehdr->e_phnum) * sizeof **phdrs; \
 180         if (!(*phdrs = malloc(size))) \
 181                 return err("Out of memory!"); \
 182  \
 183         errno = 0; \
 184         if (read(fd, *phdrs, size) != (ssize_t)size) \
 185                 return ferr("missing or incomplete program segment header table."); \
 186  \
 187         return TRUE; \
 188 } \
 189  \
 190 /* getmemorysize() determines the offset of the last byte of the file \
 191  * that is referenced by an entry in the program segment header table. \
 192  * (Anything in the file after that point is not used when the program \
 193  * is executing, and thus can be safely discarded.) \
 194  */ \
 195 static int getmemorysize ## CLASS (Elf ## CLASS ## _Ehdr const *ehdr, \
 196                                                                    Elf ## CLASS ## _Phdr const *phdrs, \
 197                                                  unsigned long *newsize) \
 198 { \
 199         Elf ## CLASS ## _Phdr const   *phdr; \
 200         unsigned long   size, n; \
 201         int                     i; \
 202  \
 203         /* Start by setting the size to include the ELF header and the \
 204          * complete program segment header table. \
 205          */ \
 206         size = EGET(ehdr->e_phoff) + EGET(ehdr->e_phnum) * sizeof *phdrs; \
 207         if (size < sizeof *ehdr) \
 208                 size = sizeof *ehdr; \
 209  \
 210         /* Then keep extending the size to include whatever data the \
 211          * program segment header table references. \
 212          */ \
 213         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 214                 if (EGET(phdr->p_type) != PT_NULL) { \
 215                         n = EGET(phdr->p_offset) + EGET(phdr->p_filesz); \
 216                         if (n > size) \
 217                                 size = n; \
 218                 } \
 219         } \
 220  \
 221         *newsize = size; \
 222         return TRUE; \
 223 } \
 224  \
 225 /* modifyheaders() removes references to the section header table if \
 226  * it was stripped, and reduces program header table entries that \
 227  * included truncated bytes at the end of the file. \
 228  */ \
 229 static int modifyheaders ## CLASS (Elf ## CLASS ## _Ehdr *ehdr, \
 230                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 231                                                                    unsigned long newsize) \
 232 { \
 233         Elf ## CLASS ## _Phdr *phdr; \
 234         int             i; \
 235  \
 236         /* If the section header table is gone, then remove all references \
 237          * to it in the ELF header. \
 238          */ \
 239         if (EGET(ehdr->e_shoff) >= newsize) { \
 240                 ESET(ehdr->e_shoff,0); \
 241                 ESET(ehdr->e_shnum,0); \
 242                 ESET(ehdr->e_shentsize,0); \
 243                 ESET(ehdr->e_shstrndx,0); \
 244         } \
 245  \
 246         /* The program adjusts the file size of any segment that was \
 247          * truncated. The case of a segment being completely stripped out \
 248          * is handled separately. \
 249          */ \
 250         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 251                 if (EGET(phdr->p_offset) >= newsize) { \
 252                         ESET(phdr->p_offset,newsize); \
 253                         ESET(phdr->p_filesz,0); \
 254                 } else if (EGET(phdr->p_offset) + EGET(phdr->p_filesz) > newsize) { \
 255                         ESET(phdr->p_filesz, newsize - EGET(phdr->p_offset)); \
 256                 } \
 257         } \
 258  \
 259         return TRUE; \
 260 } \
 261  \
 262 /* commitchanges() writes the new headers back to the original file \
 263  * and sets the file to its new size. \
 264  */ \
 265 static int commitchanges ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 266                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 267                                                                    unsigned long newsize) \
 268 { \
 269         size_t  n; \
 270  \
 271         /* Save the changes to the ELF header, if any. \
 272          */ \
 273         if (lseek(fd, 0, SEEK_SET)) \
 274                 return ferr("could not rewind file"); \
 275         errno = 0; \
 276         if (write(fd, ehdr, sizeof *ehdr) != sizeof *ehdr) \
 277                 return err("could not modify file"); \
 278  \
 279         /* Save the changes to the program segment header table, if any. \
 280          */ \
 281         if (lseek(fd, EGET(ehdr->e_phoff), SEEK_SET) == (off_t)-1) { \
 282                 err("could not seek in file."); \
 283                 goto warning; \
 284         } \
 285         n = EGET(ehdr->e_phnum) * sizeof *phdrs; \
 286         if (write(fd, phdrs, n) != (ssize_t)n) { \
 287                 err("could not write to file"); \
 288                 goto warning; \
 289         } \
 290  \
 291         /* Eleventh-hour sanity check: don't truncate before the end of \
 292          * the program segment header table. \
 293          */ \
 294         if (newsize < EGET(ehdr->e_phoff) + n) \
 295                 newsize = EGET(ehdr->e_phoff) + n; \
 296  \
 297         /* Chop off the end of the file. \
 298          */ \
 299         if (ftruncate(fd, newsize)) { \
 300                 err("could not resize file"); \
 301                 goto warning; \
 302         } \
 303  \
 304         return TRUE; \
 305  \
 306  warning: \
 307         return err("ELF file may have been corrupted!"); \
 308 }
 309
 310
 311 /* First elements of Elf32_Ehdr and Elf64_Ehdr are common.
 312  */
 313 static int readelfheaderident(int fd, Elf32_Ehdr *ehdr)
 314 {
 315         errno = 0;
 316         if (read(fd, ehdr, EI_NIDENT) != EI_NIDENT)
 317                 return ferr("missing or incomplete ELF header.");
 318
 319         /* Check the ELF signature.
 320          */
 321         if (!(ehdr->e_ident[EI_MAG0] == ELFMAG0 &&
 322                   ehdr->e_ident[EI_MAG1] == ELFMAG1 &&
 323                   ehdr->e_ident[EI_MAG2] == ELFMAG2 &&
 324                   ehdr->e_ident[EI_MAG3] == ELFMAG3))
 325         {
 326                 err("missing ELF signature.");
 327                 return -1;
 328         }
 329
 330         /* Compare the file's class and endianness with the program's.
 331          */
 332 #if __BYTE_ORDER == __LITTLE_ENDIAN
 333         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 334                 do_reverse_endian = 0;
 335         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 336 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 337                 do_reverse_endian = 1;
 338         }
 339 #elif __BYTE_ORDER == __BIG_ENDIAN
 340         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 341 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 342                 do_reverse_endian = 1;
 343         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 344                 do_reverse_endian = 0;
 345         }
 346 #else
 347 #error unkown endianness
 348 #endif
 349         else {
 350                 err("Unsupported endianness");
 351                 return -1;
 352         }
 353
 354         /* Check the target architecture.
 355          */
 356 /*       if (EGET(ehdr->e_machine) != ELF_ARCH) { */
 357 /*              /\* return err("ELF file created for different architecture."); *\/ */
 358 /*              fprintf(stderr, "ELF file created for different architecture.\n"); */
 359 /*      } */
 360         return ehdr->e_ident[EI_CLASS];
 361 }
 362
 363
 364 HEADER_FUNCTIONS(32)
 365
 366 HEADER_FUNCTIONS(64)
 367
 368 /* truncatezeros() examines the bytes at the end of the file's
 369  * size-to-be, and reduces the size to exclude any trailing zero
 370  * bytes.
 371  */
 372 static int truncatezeros(int fd, unsigned long *newsize)
 373 {
 374         unsigned char   contents[1024];
 375         unsigned long   size, n;
 376
 377         size = *newsize;
 378         do {
 379                 n = sizeof contents;
 380                 if (n > size)
 381                         n = size;
 382                 if (lseek(fd, size - n, SEEK_SET) == (off_t)-1)
 383                         return ferr("cannot seek in file.");
 384                 if (read(fd, contents, n) != (ssize_t)n)
 385                         return ferr("cannot read file contents");
 386                 while (n && !contents[--n])
 387                         --size;
 388         } while (size && !n);
 389
 390         /* Sanity check.
 391          */
 392         if (!size)
 393                 return err("ELF file is completely blank!");
 394
 395         *newsize = size;
 396         return TRUE;
 397 }
 398
 399 /* main() loops over the cmdline arguments, leaving all the real work
 400  * to the other functions.
 401  */
 402 int main(int argc, char *argv[])
 403 {
 404         int                             fd;
 405         union {
 406                 Elf32_Ehdr      ehdr32;
 407                 Elf64_Ehdr      ehdr64;
 408         } e;
 409         union {
 410                 Elf32_Phdr      *phdrs32;
 411                 Elf64_Phdr      *phdrs64;
 412         } p;
 413         unsigned long   newsize;
 414         char                    **arg;
 415         int                             failures = 0;
 416
 417         if (argc < 2 || argv[1][0] == '-') {
 418                 printf("Usage: sstrip FILE...\n"
 419                            "sstrip discards all nonessential bytes from an executable.\n\n"
 420                            "Version 2.0-X Copyright (C) 2000,2001 Brian Raiter.\n"
 421                            "Cross-devel hacks Copyright (C) 2004 Manuel Novoa III.\n"
 422                            "This program is free software, licensed under the GNU\n"
 423                            "General Public License. There is absolutely no warranty.\n");
 424                 return EXIT_SUCCESS;
 425         }
 426
 427         progname = argv[0];
 428
 429         for (arg = argv + 1 ; *arg != NULL ; ++arg) {
 430                 filename = *arg;
 431
 432                 fd = open(*arg, O_RDWR);
 433                 if (fd < 0) {
 434                         ferr("can't open");
 435                         ++failures;
 436                         continue;
 437                 }
 438
 439                 switch (readelfheaderident(fd, &e.ehdr32)) {
 440                         case ELFCLASS32:
 441                                 if (!(readelfheader32(fd, &e.ehdr32)                                    &&
 442                                           readphdrtable32(fd, &e.ehdr32, &p.phdrs32)            &&
 443                                           getmemorysize32(&e.ehdr32, p.phdrs32, &newsize)       &&
 444                                           truncatezeros(fd, &newsize)                                           &&
 445                                           modifyheaders32(&e.ehdr32, p.phdrs32, newsize)        &&
 446                                           commitchanges32(fd, &e.ehdr32, p.phdrs32, newsize)))
 447                                         ++failures;
 448                                 break;
 449                         case ELFCLASS64:
 450                                 if (!(readelfheader64(fd, &e.ehdr64)                                    &&
 451                                           readphdrtable64(fd, &e.ehdr64, &p.phdrs64)            &&
 452                                           getmemorysize64(&e.ehdr64, p.phdrs64, &newsize)       &&
 453                                           truncatezeros(fd, &newsize)                                           &&
 454                                           modifyheaders64(&e.ehdr64, p.phdrs64, newsize)        &&
 455                                           commitchanges64(fd, &e.ehdr64, p.phdrs64, newsize)))
 456                                         ++failures;
 457                                 break;
 458                         default:
 459                                 ++failures;
 460                                 break;
 461                 }
 462                 close(fd);
 463         }
 464
 465         return failures ? EXIT_FAILURE : EXIT_SUCCESS;
 466 }