toolchain/sstrip/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        <endian.h>
  63 #include        <byteswap.h>
  64
  65 #ifndef TRUE
  66 #define TRUE            1
  67 #define FALSE           0
  68 #endif
  69
  70 /* The name of the program.
  71  */
  72 static char const       *progname;
  73
  74 /* The name of the current file.
  75  */
  76 static char const       *filename;
  77
  78
  79 /* A simple error-handling function. FALSE is always returned for the
  80  * convenience of the caller.
  81  */
  82 static int err(char const *errmsg)
  83 {
  84         fprintf(stderr, "%s: %s: %s\n", progname, filename, errmsg);
  85         return FALSE;
  86 }
  87
  88 /* A flag to signal the need for endian reversal.
  89  */
  90 static int do_reverse_endian;
  91
  92 /* Get a value from the elf header, compensating for endianness.
  93  */
  94 #define EGET(X) \
  95         (__extension__ ({ \
  96                 uint64_t __res; \
  97                 if (!do_reverse_endian) { \
  98                         __res = (X); \
  99                 } else if (sizeof(X) == 1) { \
 100                         __res = (X); \
 101                 } else if (sizeof(X) == 2) { \
 102                         __res = bswap_16((X)); \
 103                 } else if (sizeof(X) == 4) { \
 104                         __res = bswap_32((X)); \
 105                 } else if (sizeof(X) == 8) { \
 106                         __res = bswap_64((X)); \
 107                 } else { \
 108                         fprintf(stderr, "%s: %s: EGET failed for size %d\n", \
 109                                         progname, filename, sizeof(X)); \
 110                         exit(EXIT_FAILURE); \
 111                 } \
 112                 __res; \
 113         }))
 114
 115 /* Set a value 'Y' in the elf header to 'X', compensating for endianness.
 116  */
 117 #define ESET(Y,X) \
 118                 do if (!do_reverse_endian) { \
 119                         Y = (X); \
 120                 } else if (sizeof(Y) == 1) { \
 121                         Y = (X); \
 122                 } else if (sizeof(Y) == 2) { \
 123                         Y = bswap_16((uint16_t)(X)); \
 124                 } else if (sizeof(Y) == 4) { \
 125                         Y = bswap_32((uint32_t)(X)); \
 126                 } else if (sizeof(Y) == 8) { \
 127                         Y = bswap_64((uint64_t)(X)); \
 128                 } else { \
 129                         fprintf(stderr, "%s: %s: ESET failed for size %d\n", \
 130                                         progname, filename, sizeof(Y)); \
 131                         exit(EXIT_FAILURE); \
 132                 } while (0)
 133
 134
 135 /* A macro for I/O errors: The given error message is used only when
 136  * errno is not set.
 137  */
 138 #define ferr(msg)       (err(errno ? strerror(errno) : (msg)))
 139
 140
 141
 142 #define HEADER_FUNCTIONS(CLASS) \
 143  \
 144 /* readelfheader() reads the ELF header into our global variable, and \
 145  * checks to make sure that this is in fact a file that we should be \
 146  * munging. \
 147  */ \
 148 static int readelfheader ## CLASS (int fd, Elf ## CLASS ## _Ehdr *ehdr) \
 149 { \
 150          if (read(fd, ((char *)ehdr)+EI_NIDENT, sizeof(*ehdr) - EI_NIDENT) \
 151                 != sizeof(*ehdr) - EI_NIDENT) \
 152                 return ferr("missing or incomplete ELF header."); \
 153  \
 154         /* Verify the sizes of the ELF header and the program segment \
 155          * header table entries. \
 156          */ \
 157         if (EGET(ehdr->e_ehsize) != sizeof(Elf ## CLASS ## _Ehdr)) \
 158                 return err("unrecognized ELF header size."); \
 159         if (EGET(ehdr->e_phentsize) != sizeof(Elf ## CLASS ## _Phdr)) \
 160                 return err("unrecognized program segment header size."); \
 161  \
 162         /* Finally, check the file type. \
 163          */ \
 164         if (EGET(ehdr->e_type) != ET_EXEC && EGET(ehdr->e_type) != ET_DYN) \
 165                 return err("not an executable or shared-object library."); \
 166  \
 167         return TRUE; \
 168 } \
 169  \
 170 /* readphdrtable() loads the program segment header table into memory. \
 171  */ \
 172 static int readphdrtable ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 173                                                                    Elf ## CLASS ## _Phdr **phdrs) \
 174 { \
 175         size_t  size; \
 176  \
 177         if (!EGET(ehdr->e_phoff) || !EGET(ehdr->e_phnum) \
 178 )               return err("ELF file has no program header table."); \
 179  \
 180         size = EGET(ehdr->e_phnum) * sizeof **phdrs; \
 181         if (!(*phdrs = malloc(size))) \
 182                 return err("Out of memory!"); \
 183  \
 184         errno = 0; \
 185         if (read(fd, *phdrs, size) != (ssize_t)size) \
 186                 return ferr("missing or incomplete program segment header table."); \
 187  \
 188         return TRUE; \
 189 } \
 190  \
 191 /* getmemorysize() determines the offset of the last byte of the file \
 192  * that is referenced by an entry in the program segment header table. \
 193  * (Anything in the file after that point is not used when the program \
 194  * is executing, and thus can be safely discarded.) \
 195  */ \
 196 static int getmemorysize ## CLASS (Elf ## CLASS ## _Ehdr const *ehdr, \
 197                                                                    Elf ## CLASS ## _Phdr const *phdrs, \
 198                                                  unsigned long *newsize) \
 199 { \
 200         Elf ## CLASS ## _Phdr const   *phdr; \
 201         unsigned long   size, n; \
 202         int                     i; \
 203  \
 204         /* Start by setting the size to include the ELF header and the \
 205          * complete program segment header table. \
 206          */ \
 207         size = EGET(ehdr->e_phoff) + EGET(ehdr->e_phnum) * sizeof *phdrs; \
 208         if (size < sizeof *ehdr) \
 209                 size = sizeof *ehdr; \
 210  \
 211         /* Then keep extending the size to include whatever data the \
 212          * program segment header table references. \
 213          */ \
 214         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 215                 if (EGET(phdr->p_type) != PT_NULL) { \
 216                         n = EGET(phdr->p_offset) + EGET(phdr->p_filesz); \
 217                         if (n > size) \
 218                                 size = n; \
 219                 } \
 220         } \
 221  \
 222         *newsize = size; \
 223         return TRUE; \
 224 } \
 225  \
 226 /* modifyheaders() removes references to the section header table if \
 227  * it was stripped, and reduces program header table entries that \
 228  * included truncated bytes at the end of the file. \
 229  */ \
 230 static int modifyheaders ## CLASS (Elf ## CLASS ## _Ehdr *ehdr, \
 231                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 232                                                                    unsigned long newsize) \
 233 { \
 234         Elf ## CLASS ## _Phdr *phdr; \
 235         int             i; \
 236  \
 237         /* If the section header table is gone, then remove all references \
 238          * to it in the ELF header. \
 239          */ \
 240         if (EGET(ehdr->e_shoff) >= newsize) { \
 241                 ESET(ehdr->e_shoff,0); \
 242                 ESET(ehdr->e_shnum,0); \
 243                 ESET(ehdr->e_shentsize,0); \
 244                 ESET(ehdr->e_shstrndx,0); \
 245         } \
 246  \
 247         /* The program adjusts the file size of any segment that was \
 248          * truncated. The case of a segment being completely stripped out \
 249          * is handled separately. \
 250          */ \
 251         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 252                 if (EGET(phdr->p_offset) >= newsize) { \
 253                         ESET(phdr->p_offset,newsize); \
 254                         ESET(phdr->p_filesz,0); \
 255                 } else if (EGET(phdr->p_offset) + EGET(phdr->p_filesz) > newsize) { \
 256                         newsize -= EGET(phdr->p_offset); \
 257                         ESET(phdr->p_filesz, newsize); \
 258                 } \
 259         } \
 260  \
 261         return TRUE; \
 262 } \
 263  \
 264 /* commitchanges() writes the new headers back to the original file \
 265  * and sets the file to its new size. \
 266  */ \
 267 static int commitchanges ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 268                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 269                                                                    unsigned long newsize) \
 270 { \
 271         size_t  n; \
 272  \
 273         /* Save the changes to the ELF header, if any. \
 274          */ \
 275         if (lseek(fd, 0, SEEK_SET)) \
 276                 return ferr("could not rewind file"); \
 277         errno = 0; \
 278         if (write(fd, ehdr, sizeof *ehdr) != sizeof *ehdr) \
 279                 return err("could not modify file"); \
 280  \
 281         /* Save the changes to the program segment header table, if any. \
 282          */ \
 283         if (lseek(fd, EGET(ehdr->e_phoff), SEEK_SET) == (off_t)-1) { \
 284                 err("could not seek in file."); \
 285                 goto warning; \
 286         } \
 287         n = EGET(ehdr->e_phnum) * sizeof *phdrs; \
 288         if (write(fd, phdrs, n) != (ssize_t)n) { \
 289                 err("could not write to file"); \
 290                 goto warning; \
 291         } \
 292  \
 293         /* Eleventh-hour sanity check: don't truncate before the end of \
 294          * the program segment header table. \
 295          */ \
 296         if (newsize < EGET(ehdr->e_phoff) + n) \
 297                 newsize = EGET(ehdr->e_phoff) + n; \
 298  \
 299         /* Chop off the end of the file. \
 300          */ \
 301         if (ftruncate(fd, newsize)) { \
 302                 err("could not resize file"); \
 303                 goto warning; \
 304         } \
 305  \
 306         return TRUE; \
 307  \
 308  warning: \
 309         return err("ELF file may have been corrupted!"); \
 310 }
 311
 312
 313 /* First elements of Elf32_Ehdr and Elf64_Ehdr are common.
 314  */
 315 static int readelfheaderident(int fd, Elf32_Ehdr *ehdr)
 316 {
 317         errno = 0;
 318         if (read(fd, ehdr, EI_NIDENT) != EI_NIDENT)
 319                 return ferr("missing or incomplete ELF header.");
 320
 321         /* Check the ELF signature.
 322          */
 323         if (!(ehdr->e_ident[EI_MAG0] == ELFMAG0 &&
 324                   ehdr->e_ident[EI_MAG1] == ELFMAG1 &&
 325                   ehdr->e_ident[EI_MAG2] == ELFMAG2 &&
 326                   ehdr->e_ident[EI_MAG3] == ELFMAG3))
 327         {
 328                 err("missing ELF signature.");
 329                 return -1;
 330         }
 331
 332         /* Compare the file's class and endianness with the program's.
 333          */
 334 #if __BYTE_ORDER == __LITTLE_ENDIAN
 335         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 336                 do_reverse_endian = 0;
 337         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 338 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 339                 do_reverse_endian = 1;
 340         }
 341 #elif __BYTE_ORDER == __BIG_ENDIAN
 342         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 343 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 344                 do_reverse_endian = 1;
 345         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 346                 do_reverse_endian = 0;
 347         }
 348 #else
 349 #error unkown endianness
 350 #endif
 351         else {
 352                 err("Unsupported endianness");
 353                 return -1;
 354         }
 355
 356         /* Check the target architecture.
 357          */
 358 /*       if (EGET(ehdr->e_machine) != ELF_ARCH) { */
 359 /*              /\* return err("ELF file created for different architecture."); *\/ */
 360 /*              fprintf(stderr, "ELF file created for different architecture.\n"); */
 361 /*      } */
 362         return ehdr->e_ident[EI_CLASS];
 363 }
 364
 365
 366 HEADER_FUNCTIONS(32)
 367
 368 HEADER_FUNCTIONS(64)
 369
 370 /* truncatezeros() examines the bytes at the end of the file's
 371  * size-to-be, and reduces the size to exclude any trailing zero
 372  * bytes.
 373  */
 374 static int truncatezeros(int fd, unsigned long *newsize)
 375 {
 376         unsigned char   contents[1024];
 377         unsigned long   size, n;
 378
 379         size = *newsize;
 380         do {
 381                 n = sizeof contents;
 382                 if (n > size)
 383                         n = size;
 384                 if (lseek(fd, size - n, SEEK_SET) == (off_t)-1)
 385                         return ferr("cannot seek in file.");
 386                 if (read(fd, contents, n) != (ssize_t)n)
 387                         return ferr("cannot read file contents");
 388                 while (n && !contents[--n])
 389                         --size;
 390         } while (size && !n);
 391
 392         /* Sanity check.
 393          */
 394         if (!size)
 395                 return err("ELF file is completely blank!");
 396
 397         *newsize = size;
 398         return TRUE;
 399 }
 400
 401 /* main() loops over the cmdline arguments, leaving all the real work
 402  * to the other functions.
 403  */
 404 int main(int argc, char *argv[])
 405 {
 406         int                             fd;
 407         union {
 408                 Elf32_Ehdr      ehdr32;
 409                 Elf64_Ehdr      ehdr64;
 410         } e;
 411         union {
 412                 Elf32_Phdr      *phdrs32;
 413                 Elf64_Phdr      *phdrs64;
 414         } p;
 415         unsigned long   newsize;
 416         char                    **arg;
 417         int                             failures = 0;
 418
 419         if (argc < 2 || argv[1][0] == '-') {
 420                 printf("Usage: sstrip FILE...\n"
 421                            "sstrip discards all nonessential bytes from an executable.\n\n"
 422                            "Version 2.0-X Copyright (C) 2000,2001 Brian Raiter.\n"
 423                            "Cross-devel hacks Copyright (C) 2004 Manuel Novoa III.\n"
 424                            "This program is free software, licensed under the GNU\n"
 425                            "General Public License. There is absolutely no warranty.\n");
 426                 return EXIT_SUCCESS;
 427         }
 428
 429         progname = argv[0];
 430
 431         for (arg = argv + 1 ; *arg != NULL ; ++arg) {
 432                 filename = *arg;
 433
 434                 fd = open(*arg, O_RDWR);
 435                 if (fd < 0) {
 436                         ferr("can't open");
 437                         ++failures;
 438                         continue;
 439                 }
 440
 441                 switch (readelfheaderident(fd, &e.ehdr32)) {
 442                         case ELFCLASS32:
 443                                 if (!(readelfheader32(fd, &e.ehdr32)                                    &&
 444                                           readphdrtable32(fd, &e.ehdr32, &p.phdrs32)            &&
 445                                           getmemorysize32(&e.ehdr32, p.phdrs32, &newsize)       &&
 446                                           truncatezeros(fd, &newsize)                                           &&
 447                                           modifyheaders32(&e.ehdr32, p.phdrs32, newsize)        &&
 448                                           commitchanges32(fd, &e.ehdr32, p.phdrs32, newsize)))
 449                                         ++failures;
 450                                 break;
 451                         case ELFCLASS64:
 452                                 if (!(readelfheader64(fd, &e.ehdr64)                                    &&
 453                                           readphdrtable64(fd, &e.ehdr64, &p.phdrs64)            &&
 454                                           getmemorysize64(&e.ehdr64, p.phdrs64, &newsize)       &&
 455                                           truncatezeros(fd, &newsize)                                           &&
 456                                           modifyheaders64(&e.ehdr64, p.phdrs64, newsize)        &&
 457                                           commitchanges64(fd, &e.ehdr64, p.phdrs64, newsize)))
 458                                         ++failures;
 459                                 break;
 460                         default:
 461                                 ++failures;
 462                                 break;
 463                 }
 464                 close(fd);
 465         }
 466
 467         return failures ? EXIT_FAILURE : EXIT_SUCCESS;
 468 }