package/openwrt/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 #ifdef __FreeBSD__
  63 /**
  64  * This seems to work on FreeBSD 5.3, should
  65  * work on all newer versions as well. I have
  66  * no idea if it will work on versions < 5.3
  67  *
  68  * Joe Estock (guru) <jestock at nutextonline.com>
  69  */
  70 #include <sys/endian.h>
  71 #define bswap_64 __bswap64
  72 #define bswap_32 __bswap32
  73 #define bswap_16 __bswap16
  74 #else
  75 #include        <endian.h>
  76 #include        <byteswap.h>
  77 #endif /* defined(__FreeBSD__) */
  78
  79
  80 #ifndef TRUE
  81 #define TRUE            1
  82 #define FALSE           0
  83 #endif
  84
  85 /* The name of the program.
  86  */
  87 static char const       *progname;
  88
  89 /* The name of the current file.
  90  */
  91 static char const       *filename;
  92
  93
  94 /* A simple error-handling function. FALSE is always returned for the
  95  * convenience of the caller.
  96  */
  97 static int err(char const *errmsg)
  98 {
  99         fprintf(stderr, "%s: %s: %s\n", progname, filename, errmsg);
 100         return FALSE;
 101 }
 102
 103 /* A flag to signal the need for endian reversal.
 104  */
 105 static int do_reverse_endian;
 106
 107 /* Get a value from the elf header, compensating for endianness.
 108  */
 109 #define EGET(X) \
 110         (__extension__ ({ \
 111                 uint64_t __res; \
 112                 if (!do_reverse_endian) { \
 113                         __res = (X); \
 114                 } else if (sizeof(X) == 1) { \
 115                         __res = (X); \
 116                 } else if (sizeof(X) == 2) { \
 117                         __res = bswap_16((X)); \
 118                 } else if (sizeof(X) == 4) { \
 119                         __res = bswap_32((X)); \
 120                 } else if (sizeof(X) == 8) { \
 121                         __res = bswap_64((X)); \
 122                 } else { \
 123                         fprintf(stderr, "%s: %s: EGET failed for size %d\n", \
 124                                         progname, filename, sizeof(X)); \
 125                         exit(EXIT_FAILURE); \
 126                 } \
 127                 __res; \
 128         }))
 129
 130 /* Set a value 'Y' in the elf header to 'X', compensating for endianness.
 131  */
 132 #define ESET(Y,X) \
 133                 do if (!do_reverse_endian) { \
 134                         Y = (X); \
 135                 } else if (sizeof(Y) == 1) { \
 136                         Y = (X); \
 137                 } else if (sizeof(Y) == 2) { \
 138                         Y = bswap_16((uint16_t)(X)); \
 139                 } else if (sizeof(Y) == 4) { \
 140                         Y = bswap_32((uint32_t)(X)); \
 141                 } else if (sizeof(Y) == 8) { \
 142                         Y = bswap_64((uint64_t)(X)); \
 143                 } else { \
 144                         fprintf(stderr, "%s: %s: ESET failed for size %d\n", \
 145                                         progname, filename, sizeof(Y)); \
 146                         exit(EXIT_FAILURE); \
 147                 } while (0)
 148
 149
 150 /* A macro for I/O errors: The given error message is used only when
 151  * errno is not set.
 152  */
 153 #define ferr(msg)       (err(errno ? strerror(errno) : (msg)))
 154
 155
 156
 157 #define HEADER_FUNCTIONS(CLASS) \
 158  \
 159 /* readelfheader() reads the ELF header into our global variable, and \
 160  * checks to make sure that this is in fact a file that we should be \
 161  * munging. \
 162  */ \
 163 static int readelfheader ## CLASS (int fd, Elf ## CLASS ## _Ehdr *ehdr) \
 164 { \
 165          if (read(fd, ((char *)ehdr)+EI_NIDENT, sizeof(*ehdr) - EI_NIDENT) \
 166                 != sizeof(*ehdr) - EI_NIDENT) \
 167                 return ferr("missing or incomplete ELF header."); \
 168  \
 169         /* Verify the sizes of the ELF header and the program segment \
 170          * header table entries. \
 171          */ \
 172         if (EGET(ehdr->e_ehsize) != sizeof(Elf ## CLASS ## _Ehdr)) \
 173                 return err("unrecognized ELF header size."); \
 174         if (EGET(ehdr->e_phentsize) != sizeof(Elf ## CLASS ## _Phdr)) \
 175                 return err("unrecognized program segment header size."); \
 176  \
 177         /* Finally, check the file type. \
 178          */ \
 179         if (EGET(ehdr->e_type) != ET_EXEC && EGET(ehdr->e_type) != ET_DYN) \
 180                 return err("not an executable or shared-object library."); \
 181  \
 182         return TRUE; \
 183 } \
 184  \
 185 /* readphdrtable() loads the program segment header table into memory. \
 186  */ \
 187 static int readphdrtable ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 188                                                                    Elf ## CLASS ## _Phdr **phdrs) \
 189 { \
 190         size_t  size; \
 191  \
 192         if (!EGET(ehdr->e_phoff) || !EGET(ehdr->e_phnum) \
 193 )               return err("ELF file has no program header table."); \
 194  \
 195         size = EGET(ehdr->e_phnum) * sizeof **phdrs; \
 196         if (!(*phdrs = malloc(size))) \
 197                 return err("Out of memory!"); \
 198  \
 199         errno = 0; \
 200         if (read(fd, *phdrs, size) != (ssize_t)size) \
 201                 return ferr("missing or incomplete program segment header table."); \
 202  \
 203         return TRUE; \
 204 } \
 205  \
 206 /* getmemorysize() determines the offset of the last byte of the file \
 207  * that is referenced by an entry in the program segment header table. \
 208  * (Anything in the file after that point is not used when the program \
 209  * is executing, and thus can be safely discarded.) \
 210  */ \
 211 static int getmemorysize ## CLASS (Elf ## CLASS ## _Ehdr const *ehdr, \
 212                                                                    Elf ## CLASS ## _Phdr const *phdrs, \
 213                                                  unsigned long *newsize) \
 214 { \
 215         Elf ## CLASS ## _Phdr const   *phdr; \
 216         unsigned long   size, n; \
 217         int                     i; \
 218  \
 219         /* Start by setting the size to include the ELF header and the \
 220          * complete program segment header table. \
 221          */ \
 222         size = EGET(ehdr->e_phoff) + EGET(ehdr->e_phnum) * sizeof *phdrs; \
 223         if (size < sizeof *ehdr) \
 224                 size = sizeof *ehdr; \
 225  \
 226         /* Then keep extending the size to include whatever data the \
 227          * program segment header table references. \
 228          */ \
 229         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 230                 if (EGET(phdr->p_type) != PT_NULL) { \
 231                         n = EGET(phdr->p_offset) + EGET(phdr->p_filesz); \
 232                         if (n > size) \
 233                                 size = n; \
 234                 } \
 235         } \
 236  \
 237         *newsize = size; \
 238         return TRUE; \
 239 } \
 240  \
 241 /* modifyheaders() removes references to the section header table if \
 242  * it was stripped, and reduces program header table entries that \
 243  * included truncated bytes at the end of the file. \
 244  */ \
 245 static int modifyheaders ## CLASS (Elf ## CLASS ## _Ehdr *ehdr, \
 246                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 247                                                                    unsigned long newsize) \
 248 { \
 249         Elf ## CLASS ## _Phdr *phdr; \
 250         int             i; \
 251  \
 252         /* If the section header table is gone, then remove all references \
 253          * to it in the ELF header. \
 254          */ \
 255         if (EGET(ehdr->e_shoff) >= newsize) { \
 256                 ESET(ehdr->e_shoff,0); \
 257                 ESET(ehdr->e_shnum,0); \
 258                 ESET(ehdr->e_shentsize,0); \
 259                 ESET(ehdr->e_shstrndx,0); \
 260         } \
 261  \
 262         /* The program adjusts the file size of any segment that was \
 263          * truncated. The case of a segment being completely stripped out \
 264          * is handled separately. \
 265          */ \
 266         for (i = 0, phdr = phdrs ; i < EGET(ehdr->e_phnum) ; ++i, ++phdr) { \
 267                 if (EGET(phdr->p_offset) >= newsize) { \
 268                         ESET(phdr->p_offset,newsize); \
 269                         ESET(phdr->p_filesz,0); \
 270                 } else if (EGET(phdr->p_offset) + EGET(phdr->p_filesz) > newsize) { \
 271                         newsize -= EGET(phdr->p_offset); \
 272                         ESET(phdr->p_filesz, newsize); \
 273                 } \
 274         } \
 275  \
 276         return TRUE; \
 277 } \
 278  \
 279 /* commitchanges() writes the new headers back to the original file \
 280  * and sets the file to its new size. \
 281  */ \
 282 static int commitchanges ## CLASS (int fd, Elf ## CLASS ## _Ehdr const *ehdr, \
 283                                                                    Elf ## CLASS ## _Phdr *phdrs, \
 284                                                                    unsigned long newsize) \
 285 { \
 286         size_t  n; \
 287  \
 288         /* Save the changes to the ELF header, if any. \
 289          */ \
 290         if (lseek(fd, 0, SEEK_SET)) \
 291                 return ferr("could not rewind file"); \
 292         errno = 0; \
 293         if (write(fd, ehdr, sizeof *ehdr) != sizeof *ehdr) \
 294                 return err("could not modify file"); \
 295  \
 296         /* Save the changes to the program segment header table, if any. \
 297          */ \
 298         if (lseek(fd, EGET(ehdr->e_phoff), SEEK_SET) == (off_t)-1) { \
 299                 err("could not seek in file."); \
 300                 goto warning; \
 301         } \
 302         n = EGET(ehdr->e_phnum) * sizeof *phdrs; \
 303         if (write(fd, phdrs, n) != (ssize_t)n) { \
 304                 err("could not write to file"); \
 305                 goto warning; \
 306         } \
 307  \
 308         /* Eleventh-hour sanity check: don't truncate before the end of \
 309          * the program segment header table. \
 310          */ \
 311         if (newsize < EGET(ehdr->e_phoff) + n) \
 312                 newsize = EGET(ehdr->e_phoff) + n; \
 313  \
 314         /* Chop off the end of the file. \
 315          */ \
 316         if (ftruncate(fd, newsize)) { \
 317                 err("could not resize file"); \
 318                 goto warning; \
 319         } \
 320  \
 321         return TRUE; \
 322  \
 323  warning: \
 324         return err("ELF file may have been corrupted!"); \
 325 }
 326
 327
 328 /* First elements of Elf32_Ehdr and Elf64_Ehdr are common.
 329  */
 330 static int readelfheaderident(int fd, Elf32_Ehdr *ehdr)
 331 {
 332         errno = 0;
 333         if (read(fd, ehdr, EI_NIDENT) != EI_NIDENT)
 334                 return ferr("missing or incomplete ELF header.");
 335
 336         /* Check the ELF signature.
 337          */
 338         if (!(ehdr->e_ident[EI_MAG0] == ELFMAG0 &&
 339                   ehdr->e_ident[EI_MAG1] == ELFMAG1 &&
 340                   ehdr->e_ident[EI_MAG2] == ELFMAG2 &&
 341                   ehdr->e_ident[EI_MAG3] == ELFMAG3))
 342         {
 343                 err("missing ELF signature.");
 344                 return -1;
 345         }
 346
 347         /* Compare the file's class and endianness with the program's.
 348          */
 349 #if __BYTE_ORDER == __LITTLE_ENDIAN
 350         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 351                 do_reverse_endian = 0;
 352         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 353 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 354                 do_reverse_endian = 1;
 355         }
 356 #elif __BYTE_ORDER == __BIG_ENDIAN
 357         if (ehdr->e_ident[EI_DATA] == ELFDATA2LSB) {
 358 /*              fprintf(stderr, "ELF file has different endianness.\n"); */
 359                 do_reverse_endian = 1;
 360         } else if (ehdr->e_ident[EI_DATA] == ELFDATA2MSB) {
 361                 do_reverse_endian = 0;
 362         }
 363 #else
 364 #error unkown endianness
 365 #endif
 366         else {
 367                 err("Unsupported endianness");
 368                 return -1;
 369         }
 370
 371         /* Check the target architecture.
 372          */
 373 /*       if (EGET(ehdr->e_machine) != ELF_ARCH) { */
 374 /*              /\* return err("ELF file created for different architecture."); *\/ */
 375 /*              fprintf(stderr, "ELF file created for different architecture.\n"); */
 376 /*      } */
 377         return ehdr->e_ident[EI_CLASS];
 378 }
 379
 380
 381 HEADER_FUNCTIONS(32)
 382
 383 HEADER_FUNCTIONS(64)
 384
 385 /* truncatezeros() examines the bytes at the end of the file's
 386  * size-to-be, and reduces the size to exclude any trailing zero
 387  * bytes.
 388  */
 389 static int truncatezeros(int fd, unsigned long *newsize)
 390 {
 391         unsigned char   contents[1024];
 392         unsigned long   size, n;
 393
 394         size = *newsize;
 395         do {
 396                 n = sizeof contents;
 397                 if (n > size)
 398                         n = size;
 399                 if (lseek(fd, size - n, SEEK_SET) == (off_t)-1)
 400                         return ferr("cannot seek in file.");
 401                 if (read(fd, contents, n) != (ssize_t)n)
 402                         return ferr("cannot read file contents");
 403                 while (n && !contents[--n])
 404                         --size;
 405         } while (size && !n);
 406
 407         /* Sanity check.
 408          */
 409         if (!size)
 410                 return err("ELF file is completely blank!");
 411
 412         *newsize = size;
 413         return TRUE;
 414 }
 415
 416 /* main() loops over the cmdline arguments, leaving all the real work
 417  * to the other functions.
 418  */
 419 int main(int argc, char *argv[])
 420 {
 421         int                             fd;
 422         union {
 423                 Elf32_Ehdr      ehdr32;
 424                 Elf64_Ehdr      ehdr64;
 425         } e;
 426         union {
 427                 Elf32_Phdr      *phdrs32;
 428                 Elf64_Phdr      *phdrs64;
 429         } p;
 430         unsigned long   newsize;
 431         char                    **arg;
 432         int                             failures = 0;
 433
 434         if (argc < 2 || argv[1][0] == '-') {
 435                 printf("Usage: sstrip FILE...\n"
 436                            "sstrip discards all nonessential bytes from an executable.\n\n"
 437                            "Version 2.0-X Copyright (C) 2000,2001 Brian Raiter.\n"
 438                            "Cross-devel hacks Copyright (C) 2004 Manuel Novoa III.\n"
 439                            "This program is free software, licensed under the GNU\n"
 440                            "General Public License. There is absolutely no warranty.\n");
 441                 return EXIT_SUCCESS;
 442         }
 443
 444         progname = argv[0];
 445
 446         for (arg = argv + 1 ; *arg != NULL ; ++arg) {
 447                 filename = *arg;
 448
 449                 fd = open(*arg, O_RDWR);
 450                 if (fd < 0) {
 451                         ferr("can't open");
 452                         ++failures;
 453                         continue;
 454                 }
 455
 456                 switch (readelfheaderident(fd, &e.ehdr32)) {
 457                         case ELFCLASS32:
 458                                 if (!(readelfheader32(fd, &e.ehdr32)                                    &&
 459                                           readphdrtable32(fd, &e.ehdr32, &p.phdrs32)            &&
 460                                           getmemorysize32(&e.ehdr32, p.phdrs32, &newsize)       &&
 461                                           truncatezeros(fd, &newsize)                                           &&
 462                                           modifyheaders32(&e.ehdr32, p.phdrs32, newsize)        &&
 463                                           commitchanges32(fd, &e.ehdr32, p.phdrs32, newsize)))
 464                                         ++failures;
 465                                 break;
 466                         case ELFCLASS64:
 467                                 if (!(readelfheader64(fd, &e.ehdr64)                                    &&
 468                                           readphdrtable64(fd, &e.ehdr64, &p.phdrs64)            &&
 469                                           getmemorysize64(&e.ehdr64, p.phdrs64, &newsize)       &&
 470                                           truncatezeros(fd, &newsize)                                           &&
 471                                           modifyheaders64(&e.ehdr64, p.phdrs64, newsize)        &&
 472                                           commitchanges64(fd, &e.ehdr64, p.phdrs64, newsize)))
 473                                         ++failures;
 474                                 break;
 475                         default:
 476                                 ++failures;
 477                                 break;
 478                 }
 479                 close(fd);
 480         }
 481
 482         return failures ? EXIT_FAILURE : EXIT_SUCCESS;
 483 }