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
  63 #ifndef TRUE
  64 #define TRUE            1
  65 #define FALSE           0
  66 #endif
  67
  68 /* The name of the program.
  69  */
  70 static char const       *progname;
  71
  72 /* The name of the current file.
  73  */
  74 static char const       *filename;
  75
  76
  77 /* A simple error-handling function. FALSE is always returned for the
  78  * convenience of the caller.
  79  */
  80 static int err(char const *errmsg)
  81 {
  82         fprintf(stderr, "%s: %s: %s\n", progname, filename, errmsg);
  83         return FALSE;
  84 }
  85
  86 /* A flag to signal the need for endian reversal.
  87  */
  88 static int do_reverse_endian;
  89
  90 /* Get a value from the elf header, compensating for endianness.
  91  */
  92 #define EGET(X) \
  93         (__extension__ ({ \
  94                 uint64_t __res; \
  95                 if (!do_reverse_endian) { \
  96                         __res = (X); \
  97                 } else if (sizeof(X) == 1) { \
  98                         __res = (X); \
  99                 } else if (sizeof(X) == 2) { \
 100                         __res = bswap_16((X)); \
 101                 } else if (sizeof(X) == 4) { \
 102                         __res = bswap_32((X)); \
 103                 } else if (sizeof(X) == 8) { \
 104                         __res = bswap_64((X)); \
 105                 } else { \
 106                         fprintf(stderr, "%s: %s: EGET failed for size %d\n", \
 107                                         progname, filename, sizeof(X)); \
 108                         exit(EXIT_FAILURE); \
 109                 } \
 110                 __res; \
 111         }))
 112
 113 /* Set a value 'Y' in the elf header to 'X', compensating for endianness.
 114  */
 115 #define ESET(Y,X) \
 116                 do if (!do_reverse_endian) { \
 117                         Y = (X); \
 118                 } else if (sizeof(Y) == 1) { \
 119                         Y = (X); \
 120                 } else if (sizeof(Y) == 2) { \
 121                         Y = bswap_16((uint16_t)(X)); \
 122                 } else if (sizeof(Y) == 4) { \
 123                         Y = bswap_32((uint32_t)(X)); \
 124                 } else if (sizeof(Y) == 8) { \
 125                         Y = bswap_64((uint64_t)(X)); \
 126                 } else { \
 127                         fprintf(stderr, "%s: %s: ESET failed for size %d\n", \
 128                                         progname, filename, sizeof(Y)); \
 129                         exit(EXIT_FAILURE); \
 130                 } while (0)
 131
 132
 133 /* A macro for I/O errors: The given error message is used only when
 134  * errno is not set.
 135  */
 136 #define ferr(msg)       (err(errno ? strerror(errno) : (msg)))
 137
 138
 139
 140 #define HEADER_FUNCTIONS(CLASS) \
 141  \
 142 /* readelfheader() reads the ELF header into our global variable, and \
 143  * checks to make sure that this is in fact a file that we should be \
 144  * munging. \
 145  */ \
 146 static int readelfheader ## CLASS (int fd, Elf ## CLASS ## _Ehdr *ehdr) \
 147 { \
 148          if (read(fd, ((char *)ehdr)+EI_NIDENT, sizeof(*ehdr) - EI_NIDENT) \
 149                 != sizeof(*ehdr) - EI_NIDENT) \
 150                 return ferr("missing or incomplete ELF header."); \
 151  \
 152         /* Verify the sizes of the ELF header and the program segment \
 153          * header table entries. \
 154          */ \
 155         if (EGET(ehdr->e_ehsize) != sizeof(Elf ## CLASS ## _Ehdr)) \
 156                 return err("unrecognized ELF header size."); \
 157         if (EGET(ehdr->e_phentsize) != sizeof(Elf ## CLASS ## _Phdr)) \
 158                 return err("unrecognized program segment header size."); \
 159  \
 160         /* Finally, check the file type. \
 161          */ \
 162         if (EGET(ehdr->e_type) != ET_EXEC && EGET(ehdr->e_type) != ET_DYN) \
 163                 return err("not an executable or shared-object library."); \
 164  \
 165         return TRUE; \
 166 } \
 167  \
 168 /* readphdrtable() loads the program segment header table into memory. \
 169  */ \
 170 static int readphdrtable ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 171                                                                    Elf ## CLASS ## _Phdr **phdrs) \
 172 { \
 173         size_t  size; \
 174  \
 175         if (!EGET(ehdr->e_phoff) || !EGET(ehdr->e_phnum) \
 176 )               return err("ELF file has no program header table."); \
 177  \
 178         size = EGET(ehdr->e_phnum) * sizeof **phdrs; \
 179         if (!(*phdrs = malloc(size))) \
 180                 return err("Out of memory!"); \
 181  \
 182         errno = 0; \
 183         if (read(fd, *phdrs, size) != (ssize_t)size) \
 184                 return ferr("missing or incomplete program segment header table."); \
 185  \
 186         return TRUE; \
 187 } \
 188  \
 189 /* getmemorysize() determines the offset of the last byte of the file \
 190  * that is referenced by an entry in the program segment header table. \
 191  * (Anything in the file after that point is not used when the program \
 192  * is executing, and thus can be safely discarded.) \
 193  */ \
 194 static int getmemorysize ## CLASS (Elf ## CLASS ## _Ehdr const *ehdr, \
 195                                                                    Elf ## CLASS ## _Phdr const *phdrs, \
 196                                                  unsigned long *newsize) \
 197 { \
 198         Elf ## CLASS ## _Phdr const   *phdr; \
 199         unsigned long   size, n; \
 200         int                     i; \
 201  \
 202         /* Start by setting the size to include the ELF header and the \
 203          * complete program segment header table. \
 204          */ \
 205         size = EGET(ehdr->e_phoff) + EGET(ehdr->e_phnum) * sizeof *phdrs; \
 206         if (size < sizeof *ehdr) \
 207                 size = sizeof *ehdr; \
 208  \
 209         /* Then keep extending the size to include whatever data the \
 210          * program segment header table references. \
 211          */ \
 212         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 213                 if (EGET(phdr->p_type) != PT_NULL) { \
 214                         n = EGET(phdr->p_offset) + EGET(phdr->p_filesz); \
 215                         if (n > size) \
 216                                 size = n; \
 217                 } \
 218         } \
 219  \
 220         *newsize = size; \
 221         return TRUE; \
 222 } \
 223  \
 224 /* modifyheaders() removes references to the section header table if \
 225  * it was stripped, and reduces program header table entries that \
 226  * included truncated bytes at the end of the file. \
 227  */ \
 228 static int modifyheaders ## CLASS (Elf ## CLASS ## _Ehdr *ehdr, \
 229                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 230                                                                    unsigned long newsize) \
 231 { \
 232         Elf ## CLASS ## _Phdr *phdr; \
 233         int             i; \
 234  \
 235         /* If the section header table is gone, then remove all references \
 236          * to it in the ELF header. \
 237          */ \
 238         if (EGET(ehdr->e_shoff) >= newsize) { \
 239                 ESET(ehdr->e_shoff,0); \
 240                 ESET(ehdr->e_shnum,0); \
 241                 ESET(ehdr->e_shentsize,0); \
 242                 ESET(ehdr->e_shstrndx,0); \
 243         } \
 244  \
 245         /* The program adjusts the file size of any segment that was \
 246          * truncated. The case of a segment being completely stripped out \
 247          * is handled separately. \
 248          */ \
 249         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 250                 if (EGET(phdr->p_offset) >= newsize) { \
 251                         ESET(phdr->p_offset,newsize); \
 252                         ESET(phdr->p_filesz,0); \
 253                 } else if (EGET(phdr->p_offset) + EGET(phdr->p_filesz) > newsize) { \
 254                         ESET(phdr->p_filesz, newsize - EGET(phdr->p_offset)); \
 255                 } \
 256         } \
 257  \
 258         return TRUE; \
 259 } \
 260  \
 261 /* commitchanges() writes the new headers back to the original file \
 262  * and sets the file to its new size. \
 263  */ \
 264 static int commitchanges ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 265                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 266                                                                    unsigned long newsize) \
 267 { \
 268         size_t  n; \
 269  \
 270         /* Save the changes to the ELF header, if any. \
 271          */ \
 272         if (lseek(fd, 0, SEEK_SET)) \
 273                 return ferr("could not rewind file"); \
 274         errno = 0; \
 275         if (write(fd, ehdr, sizeof *ehdr) != sizeof *ehdr) \
 276                 return err("could not modify file"); \
 277  \
 278         /* Save the changes to the program segment header table, if any. \
 279          */ \
 280         if (lseek(fd, EGET(ehdr->e_phoff), SEEK_SET) == (off_t)-1) { \
 281                 err("could not seek in file."); \
 282                 goto warning; \
 283         } \
 284         n = EGET(ehdr->e_phnum) * sizeof *phdrs; \
 285         if (write(fd, phdrs, n) != (ssize_t)n) { \
 286                 err("could not write to file"); \
 287                 goto warning; \
 288         } \
 289  \
 290         /* Eleventh-hour sanity check: don't truncate before the end of \
 291          * the program segment header table. \
 292          */ \
 293         if (newsize < EGET(ehdr->e_phoff) + n) \
 294                 newsize = EGET(ehdr->e_phoff) + n; \
 295  \
 296         /* Chop off the end of the file. \
 297          */ \
 298         if (ftruncate(fd, newsize)) { \
 299                 err("could not resize file"); \
 300                 goto warning; \
 301         } \
 302  \
 303         return TRUE; \
 304  \
 305  warning: \
 306         return err("ELF file may have been corrupted!"); \
 307 }
 308
 309
 310 /* First elements of Elf32_Ehdr and Elf64_Ehdr are common.
 311  */
 312 static int readelfheaderident(int fd, Elf32_Ehdr *ehdr)
 313 {
 314         errno = 0;
 315         if (read(fd, ehdr, EI_NIDENT) != EI_NIDENT)
 316                 return ferr("missing or incomplete ELF header.");
 317
 318         /* Check the ELF signature.
 319          */
 320         if (!(ehdr->e_ident[EI_MAG0] == ELFMAG0 &&
 321                   ehdr->e_ident[EI_MAG1] == ELFMAG1 &&
 322                   ehdr->e_ident[EI_MAG2] == ELFMAG2 &&
 323                   ehdr->e_ident[EI_MAG3] == ELFMAG3))
 324         {
 325                 err("missing ELF signature.");
 326                 return -1;
 327         }
 328
 329         /* Compare the file's class and endianness with the program's.
 330          */
 331 #if __BYTE_ORDER == __LITTLE_ENDIAN
 332         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 333                 do_reverse_endian = 0;
 334         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 335 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 336                 do_reverse_endian = 1;
 337         }
 338 #elif __BYTE_ORDER == __BIG_ENDIAN
 339         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 340 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 341                 do_reverse_endian = 1;
 342         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 343                 do_reverse_endian = 0;
 344         }
 345 #else
 346 #error unkown endianness
 347 #endif
 348         else {
 349                 err("Unsupported endianness");
 350                 return -1;
 351         }
 352
 353         /* Check the target architecture.
 354          */
 355 /*       if (EGET(ehdr->e_machine) != ELF_ARCH) { */
 356 /*              /\* return err("ELF file created for different architecture."); *\/ */
 357 /*              fprintf(stderr, "ELF file created for different architecture.\n"); */
 358 /*      } */
 359         return ehdr->e_ident[EI_CLASS];
 360 }
 361
 362
 363 HEADER_FUNCTIONS(32)
 364
 365 HEADER_FUNCTIONS(64)
 366
 367 /* truncatezeros() examines the bytes at the end of the file's
 368  * size-to-be, and reduces the size to exclude any trailing zero
 369  * bytes.
 370  */
 371 static int truncatezeros(int fd, unsigned long *newsize)
 372 {
 373         unsigned char   contents[1024];
 374         unsigned long   size, n;
 375
 376         size = *newsize;
 377         do {
 378                 n = sizeof contents;
 379                 if (n > size)
 380                         n = size;
 381                 if (lseek(fd, size - n, SEEK_SET) == (off_t)-1)
 382                         return ferr("cannot seek in file.");
 383                 if (read(fd, contents, n) != (ssize_t)n)
 384                         return ferr("cannot read file contents");
 385                 while (n && !contents[--n])
 386                         --size;
 387         } while (size && !n);
 388
 389         /* Sanity check.
 390          */
 391         if (!size)
 392                 return err("ELF file is completely blank!");
 393
 394         *newsize = size;
 395         return TRUE;
 396 }
 397
 398 /* main() loops over the cmdline arguments, leaving all the real work
 399  * to the other functions.
 400  */
 401 int main(int argc, char *argv[])
 402 {
 403         int                             fd;
 404         union {
 405                 Elf32_Ehdr      ehdr32;
 406                 Elf64_Ehdr      ehdr64;
 407         } e;
 408         union {
 409                 Elf32_Phdr      *phdrs32;
 410                 Elf64_Phdr      *phdrs64;
 411         } p;
 412         unsigned long   newsize;
 413         char                    **arg;
 414         int                             failures = 0;
 415
 416         if (argc < 2 || argv[1][0] == '-') {
 417                 printf("Usage: sstrip FILE...\n"
 418                            "sstrip discards all nonessential bytes from an executable.\n\n"
 419                            "Version 2.0-X Copyright (C) 2000,2001 Brian Raiter.\n"
 420                            "Cross-devel hacks Copyright (C) 2004 Manuel Novoa III.\n"
 421                            "This program is free software, licensed under the GNU\n"
 422                            "General Public License. There is absolutely no warranty.\n");
 423                 return EXIT_SUCCESS;
 424         }
 425
 426         progname = argv[0];
 427
 428         for (arg = argv + 1 ; *arg != NULL ; ++arg) {
 429                 filename = *arg;
 430
 431                 fd = open(*arg, O_RDWR);
 432                 if (fd < 0) {
 433                         ferr("can't open");
 434                         ++failures;
 435                         continue;
 436                 }
 437
 438                 switch (readelfheaderident(fd, &e.ehdr32)) {
 439                         case ELFCLASS32:
 440                                 if (!(readelfheader32(fd, &e.ehdr32)                                    &&
 441                                           readphdrtable32(fd, &e.ehdr32, &p.phdrs32)            &&
 442                                           getmemorysize32(&e.ehdr32, p.phdrs32, &newsize)       &&
 443                                           truncatezeros(fd, &newsize)                                           &&
 444                                           modifyheaders32(&e.ehdr32, p.phdrs32, newsize)        &&
 445                                           commitchanges32(fd, &e.ehdr32, p.phdrs32, newsize)))
 446                                         ++failures;
 447                                 break;
 448                         case ELFCLASS64:
 449                                 if (!(readelfheader64(fd, &e.ehdr64)                                    &&
 450                                           readphdrtable64(fd, &e.ehdr64, &p.phdrs64)            &&
 451                                           getmemorysize64(&e.ehdr64, p.phdrs64, &newsize)       &&
 452                                           truncatezeros(fd, &newsize)                                           &&
 453                                           modifyheaders64(&e.ehdr64, p.phdrs64, newsize)        &&
 454                                           commitchanges64(fd, &e.ehdr64, p.phdrs64, newsize)))
 455                                         ++failures;
 456                                 break;
 457                         default:
 458                                 ++failures;
 459                                 break;
 460                 }
 461                 close(fd);
 462         }
 463
 464         return failures ? EXIT_FAILURE : EXIT_SUCCESS;
 465 }