| 1 |
/*
|
| 2 |
* Copyright 2003-2012 Gentoo Foundation
|
| 3 |
* Distributed under the terms of the GNU General Public License v2
|
| 4 |
* $Header: /var/cvsroot/gentoo-projects/pax-utils/paxmacho.c,v 1.19 2010/12/08 01:29:36 vapier Exp $
|
| 5 |
*
|
| 6 |
* Copyright 2005-2012 Ned Ludd - <solar@gentoo.org>
|
| 7 |
* Copyright 2005-2012 Mike Frysinger - <vapier@gentoo.org>
|
| 8 |
* 2008-2012 Fabian Groffen - <grobian@gentoo.org>
|
| 9 |
*/
|
| 10 |
|
| 11 |
#include "paxinc.h"
|
| 12 |
|
| 13 |
/* lil' static string pool */
|
| 14 |
static const char STR_BE[] = "BE";
|
| 15 |
static const char STR_LE[] = "LE";
|
| 16 |
static const char STR_PPC[] = "ppc";
|
| 17 |
static const char STR_PPC64[] = "ppc64";
|
| 18 |
static const char STR_I386[] = "i386";
|
| 19 |
static const char STR_X86_64[] = "x86_64";
|
| 20 |
static const char STR_ARM[] = "arm"; /* iPhone */
|
| 21 |
static const char STR_UNKNOWN[] = "unknown";
|
| 22 |
|
| 23 |
#define QUERY(n) { #n, n }
|
| 24 |
typedef const struct {
|
| 25 |
const char *str;
|
| 26 |
int value;
|
| 27 |
} pairtype;
|
| 28 |
|
| 29 |
static inline const char *find_pairtype(pairtype *pt, int type)
|
| 30 |
{
|
| 31 |
size_t i;
|
| 32 |
for (i = 0; pt[i].str; ++i)
|
| 33 |
if (type == pt[i].value)
|
| 34 |
return pt[i].str;
|
| 35 |
return "UNKNOWN TYPE";
|
| 36 |
}
|
| 37 |
|
| 38 |
/* translate misc mach-o MH_ defines */
|
| 39 |
static pairtype macho_mh_type[] = {
|
| 40 |
QUERY(MH_OBJECT),
|
| 41 |
QUERY(MH_EXECUTE),
|
| 42 |
QUERY(MH_BUNDLE),
|
| 43 |
QUERY(MH_DYLIB),
|
| 44 |
QUERY(MH_PRELOAD),
|
| 45 |
QUERY(MH_CORE),
|
| 46 |
QUERY(MH_DYLINKER),
|
| 47 |
QUERY(MH_DYLIB_STUB),
|
| 48 |
QUERY(MH_DSYM),
|
| 49 |
{ 0, 0 }
|
| 50 |
};
|
| 51 |
const char *get_machomhtype(fatobj *fobj)
|
| 52 |
{
|
| 53 |
/* can use 32-bits header, since 64 and 32 are aligned here */
|
| 54 |
return find_pairtype(macho_mh_type, MOBJGET(fobj, mhdr.hdr32->filetype));
|
| 55 |
}
|
| 56 |
|
| 57 |
/* translate misc mach-o MH_ flags */
|
| 58 |
static pairtype macho_mh_flag[] = {
|
| 59 |
QUERY(MH_NOUNDEFS),
|
| 60 |
QUERY(MH_INCRLINK),
|
| 61 |
QUERY(MH_DYLDLINK),
|
| 62 |
QUERY(MH_TWOLEVEL),
|
| 63 |
QUERY(MH_BINDATLOAD),
|
| 64 |
QUERY(MH_PREBOUND),
|
| 65 |
QUERY(MH_PREBINDABLE),
|
| 66 |
QUERY(MH_NOFIXPREBINDING),
|
| 67 |
QUERY(MH_ALLMODSBOUND),
|
| 68 |
QUERY(MH_CANONICAL),
|
| 69 |
QUERY(MH_SPLIT_SEGS),
|
| 70 |
QUERY(MH_FORCE_FLAT),
|
| 71 |
QUERY(MH_SUBSECTIONS_VIA_SYMBOLS),
|
| 72 |
QUERY(MH_NOMULTIDEFS),
|
| 73 |
{ 0, 0 }
|
| 74 |
};
|
| 75 |
void get_machomhflags(fatobj *fobj, char **ret, size_t *ret_len)
|
| 76 |
{
|
| 77 |
uint32_t flags;
|
| 78 |
int i;
|
| 79 |
char first = 1;
|
| 80 |
|
| 81 |
/* can use 32-bits header, since 64 and 32 are aligned here */
|
| 82 |
flags = MOBJGET(fobj, mhdr.hdr32->flags);
|
| 83 |
|
| 84 |
for (i = 0; macho_mh_flag[i].str; ++i)
|
| 85 |
if ((flags & macho_mh_flag[i].value) == macho_mh_flag[i].value) {
|
| 86 |
if (!first)
|
| 87 |
xchrcat(ret, ',', ret_len);
|
| 88 |
xstrcat(ret, macho_mh_flag[i].str, ret_len);
|
| 89 |
first = 0;
|
| 90 |
}
|
| 91 |
}
|
| 92 |
|
| 93 |
static pairtype macho_cputype[] = {
|
| 94 |
QUERY(CPU_TYPE_POWERPC),
|
| 95 |
QUERY(CPU_TYPE_I386),
|
| 96 |
QUERY(CPU_TYPE_ARM),
|
| 97 |
QUERY(CPU_TYPE_POWERPC64),
|
| 98 |
QUERY(CPU_TYPE_X86_64),
|
| 99 |
{ 0, 0 }
|
| 100 |
};
|
| 101 |
const char *get_machocputype(fatobj *fobj)
|
| 102 |
{
|
| 103 |
/* can use 32-bits header, since 64 and 32 are aligned here */
|
| 104 |
const char *ret = find_pairtype(macho_cputype, MOBJGET(fobj, mhdr.hdr32->cputype));
|
| 105 |
return ret + sizeof("CPU_TYPE_") - 1;
|
| 106 |
}
|
| 107 |
|
| 108 |
/* translate cpusubtypes */
|
| 109 |
static pairtype macho_cpusubtypeppc[] = {
|
| 110 |
QUERY(CPU_SUBTYPE_POWERPC_ALL),
|
| 111 |
QUERY(CPU_SUBTYPE_POWERPC_601),
|
| 112 |
QUERY(CPU_SUBTYPE_POWERPC_602),
|
| 113 |
QUERY(CPU_SUBTYPE_POWERPC_603),
|
| 114 |
QUERY(CPU_SUBTYPE_POWERPC_603e),
|
| 115 |
QUERY(CPU_SUBTYPE_POWERPC_603ev),
|
| 116 |
QUERY(CPU_SUBTYPE_POWERPC_604),
|
| 117 |
QUERY(CPU_SUBTYPE_POWERPC_604e),
|
| 118 |
QUERY(CPU_SUBTYPE_POWERPC_620),
|
| 119 |
QUERY(CPU_SUBTYPE_POWERPC_750),
|
| 120 |
QUERY(CPU_SUBTYPE_POWERPC_7400),
|
| 121 |
QUERY(CPU_SUBTYPE_POWERPC_7450),
|
| 122 |
QUERY(CPU_SUBTYPE_POWERPC_970),
|
| 123 |
{ 0, 0 }
|
| 124 |
};
|
| 125 |
static pairtype macho_cpusubtypex86[] = {
|
| 126 |
QUERY(CPU_SUBTYPE_I386_ALL),
|
| 127 |
QUERY(CPU_SUBTYPE_486),
|
| 128 |
QUERY(CPU_SUBTYPE_586),
|
| 129 |
QUERY(CPU_SUBTYPE_PENTIUM_3),
|
| 130 |
QUERY(CPU_SUBTYPE_PENTIUM_M),
|
| 131 |
QUERY(CPU_SUBTYPE_PENTIUM_4),
|
| 132 |
QUERY(CPU_SUBTYPE_ITANIUM),
|
| 133 |
QUERY(CPU_SUBTYPE_XEON),
|
| 134 |
{ 0, 0 }
|
| 135 |
};
|
| 136 |
const char *get_machosubcputype(fatobj *fobj)
|
| 137 |
{
|
| 138 |
const char *ret;
|
| 139 |
/* can use 32-bits header, since 64 and 32 are aligned here */
|
| 140 |
uint32_t type = MOBJGET(fobj, mhdr.hdr32->cputype);
|
| 141 |
pairtype *pt = NULL;
|
| 142 |
|
| 143 |
if (type == CPU_TYPE_I386 || type == CPU_TYPE_X86_64)
|
| 144 |
pt = macho_cpusubtypex86;
|
| 145 |
else if (type == CPU_TYPE_POWERPC || type == CPU_TYPE_POWERPC64)
|
| 146 |
pt = macho_cpusubtypeppc;
|
| 147 |
|
| 148 |
if (pt) {
|
| 149 |
type = MOBJGET(fobj, mhdr.hdr32->cpusubtype);
|
| 150 |
ret = find_pairtype(pt, type);
|
| 151 |
return ret + sizeof("CPU_SUBTYPE_") - 1;
|
| 152 |
} else
|
| 153 |
return STR_UNKNOWN;
|
| 154 |
}
|
| 155 |
|
| 156 |
/* Determines the type of this object, and sets the right 32-bit or
|
| 157 |
* 64-bits pointer. The ismach64 flag is filled in appropriately. The
|
| 158 |
* return of this function is the read magic value, or 0 when the file
|
| 159 |
* is not recognised.
|
| 160 |
* Note: the input addr must be enough to map on struct mach_header! */
|
| 161 |
inline static uint32_t read_mach_header(fatobj *fobj, void *addr)
|
| 162 |
{
|
| 163 |
struct mach_header *mhdr = addr;
|
| 164 |
fobj->mhdata = addr;
|
| 165 |
switch (mhdr->magic) {
|
| 166 |
case MH_CIGAM:
|
| 167 |
fobj->swapped = 1;
|
| 168 |
case MH_MAGIC:
|
| 169 |
/* 32-bits */
|
| 170 |
fobj->ismach64 = 0;
|
| 171 |
fobj->mhdr.hdr32 = mhdr;
|
| 172 |
fobj->isbigendian = (*fobj->mhdata == (char)(MH_MAGIC >> 24) ? 1 : 0);
|
| 173 |
return mhdr->magic;
|
| 174 |
case MH_CIGAM_64:
|
| 175 |
fobj->swapped = 1;
|
| 176 |
case MH_MAGIC_64:
|
| 177 |
/* 64-bits */
|
| 178 |
fobj->ismach64 = 1;
|
| 179 |
fobj->mhdr.hdr64 = addr;
|
| 180 |
fobj->isbigendian = (*fobj->mhdata == (char)(MH_MAGIC_64 >> 24) ? 1 : 0);
|
| 181 |
return mhdr->magic;
|
| 182 |
default:
|
| 183 |
return 0; /* unrecognised file */
|
| 184 |
}
|
| 185 |
}
|
| 186 |
|
| 187 |
/* Read a macho into memory, returning a fatobj struct with at least one
|
| 188 |
* arch. */
|
| 189 |
fatobj *readmacho(const char *filename)
|
| 190 |
{
|
| 191 |
struct stat st;
|
| 192 |
int fd;
|
| 193 |
|
| 194 |
if (stat(filename, &st) == -1)
|
| 195 |
return NULL;
|
| 196 |
|
| 197 |
if ((fd = open(filename, O_RDONLY)) == -1)
|
| 198 |
return NULL;
|
| 199 |
|
| 200 |
return readmacho_fd(filename, fd, st.st_size);
|
| 201 |
}
|
| 202 |
|
| 203 |
fatobj *readmacho_fd(const char *filename, int fd, size_t len)
|
| 204 |
{
|
| 205 |
char *data;
|
| 206 |
fatobj *ret;
|
| 207 |
|
| 208 |
if (len == 0) {
|
| 209 |
struct stat st;
|
| 210 |
if (fstat(fd, &st) == -1)
|
| 211 |
return NULL;
|
| 212 |
len = st.st_size;
|
| 213 |
if (len == 0)
|
| 214 |
return NULL;
|
| 215 |
}
|
| 216 |
|
| 217 |
data = mmap(0, len, PROT_READ, MAP_PRIVATE, fd, 0);
|
| 218 |
if (data == MAP_FAILED) {
|
| 219 |
warn("mmap on '%s' of %zu bytes failed :(", filename, len);
|
| 220 |
return NULL;
|
| 221 |
}
|
| 222 |
|
| 223 |
ret = readmacho_buffer(filename, data, len);
|
| 224 |
if (ret != NULL) {
|
| 225 |
ret->fd = fd;
|
| 226 |
return ret;
|
| 227 |
}
|
| 228 |
|
| 229 |
munmap(data, len);
|
| 230 |
return NULL;
|
| 231 |
}
|
| 232 |
|
| 233 |
fatobj *readmacho_buffer(const char *filename, char *buffer, size_t buffer_len)
|
| 234 |
{
|
| 235 |
struct fat_header *fhdr;
|
| 236 |
fatobj *ret = xmalloc(sizeof(*ret));
|
| 237 |
|
| 238 |
ret->fd = -1;
|
| 239 |
ret->filename = filename;
|
| 240 |
ret->base_filename = strrchr(ret->filename, '/');
|
| 241 |
ret->base_filename =
|
| 242 |
(ret->base_filename == NULL ? ret->filename : ret->base_filename + 1);
|
| 243 |
ret->len = buffer_len;
|
| 244 |
ret->data = buffer;
|
| 245 |
ret->swapped = 0;
|
| 246 |
|
| 247 |
/* make sure we have enough bytes to scan */
|
| 248 |
if (ret->len <= sizeof(struct fat_header))
|
| 249 |
return NULL;
|
| 250 |
|
| 251 |
fhdr = ret->data;
|
| 252 |
/* Check what kind of file this is. Unfortunately we don't have
|
| 253 |
* configure, so we don't know if we're on big or little endian, so
|
| 254 |
* we cannot check if the fat_header is in bigendian like it should.
|
| 255 |
*/
|
| 256 |
if (fhdr->magic == FAT_MAGIC || fhdr->magic == FAT_CIGAM) {
|
| 257 |
/* we're indeed in a FAT file */
|
| 258 |
int i;
|
| 259 |
fatobj *fobj = ret;
|
| 260 |
struct fat_arch *farch;
|
| 261 |
void *dptr = ret->data + sizeof(struct fat_header);
|
| 262 |
uint32_t bufleft = ret->len - sizeof(struct fat_header);
|
| 263 |
char swapped = 0;
|
| 264 |
uint32_t narchs = fhdr->nfat_arch;
|
| 265 |
uint32_t offset;
|
| 266 |
|
| 267 |
/* FAT headers are always big-endian, so swap if on little
|
| 268 |
* machines... */
|
| 269 |
if (fhdr->magic == FAT_CIGAM) {
|
| 270 |
swapped = 1;
|
| 271 |
narchs = bswap_32(narchs);
|
| 272 |
}
|
| 273 |
|
| 274 |
/* can we read the headers at all?
|
| 275 |
* beware of corrupt files and Java bytecode which shares
|
| 276 |
* the same magic with us :( */
|
| 277 |
if (sizeof(struct fat_arch) * narchs > bufleft)
|
| 278 |
return NULL;
|
| 279 |
|
| 280 |
for (i = 1; i <= narchs; i++) {
|
| 281 |
farch = (struct fat_arch *)dptr;
|
| 282 |
offset = MGET(swapped, farch->offset);
|
| 283 |
if (offset + sizeof(struct mach_header) >= bufleft ||
|
| 284 |
read_mach_header(fobj, ret->data + offset) == 0)
|
| 285 |
return NULL;
|
| 286 |
if (i < narchs) {
|
| 287 |
fobj = fobj->next = xzalloc(sizeof(*fobj));
|
| 288 |
/* filename and size are necessary for printing */
|
| 289 |
fobj->filename = ret->filename;
|
| 290 |
fobj->base_filename = ret->base_filename;
|
| 291 |
fobj->len = ret->len;
|
| 292 |
} else {
|
| 293 |
fobj->next = NULL;
|
| 294 |
}
|
| 295 |
dptr += sizeof(struct fat_arch);
|
| 296 |
bufleft -= sizeof(struct fat_arch);
|
| 297 |
}
|
| 298 |
} else {
|
| 299 |
/* simple Mach-O file, treat as single arch FAT file */
|
| 300 |
if (ret->len < sizeof(struct mach_header) ||
|
| 301 |
read_mach_header(ret, ret->data) == 0)
|
| 302 |
return NULL;
|
| 303 |
ret->next = NULL;
|
| 304 |
}
|
| 305 |
|
| 306 |
return ret;
|
| 307 |
}
|
| 308 |
|
| 309 |
/* undo the readmacho() stuff */
|
| 310 |
void unreadmacho(fatobj *macho)
|
| 311 |
{
|
| 312 |
if (macho->data != NULL) {
|
| 313 |
munmap(macho->data, macho->len);
|
| 314 |
close(macho->fd);
|
| 315 |
}
|
| 316 |
/* free all arches recursively */
|
| 317 |
if (macho->next != NULL)
|
| 318 |
unreadmacho(macho->next);
|
| 319 |
free(macho);
|
| 320 |
}
|
| 321 |
|
| 322 |
/* Returns the first load_command in the file (after the mach_header)
|
| 323 |
* and allocates a loadcmd struct to store it together with some
|
| 324 |
* convenience data. The struct can be manually freed, if not traversed
|
| 325 |
* until the end of the load section. */
|
| 326 |
loadcmd *firstloadcmd(fatobj *fobj)
|
| 327 |
{
|
| 328 |
loadcmd *ret = xmalloc(sizeof(*ret));
|
| 329 |
ret->data = fobj->mhdata +
|
| 330 |
(fobj->ismach64 ? sizeof(struct mach_header_64) : sizeof(struct mach_header));
|
| 331 |
ret->lcmd = ret->data;
|
| 332 |
ret->cleft = MOBJGET(fobj, mhdr.hdr32->ncmds); /* 32 and 64 bits are aligned here */
|
| 333 |
ret->align = (fobj->ismach64 ? 8 : 4);
|
| 334 |
ret->swapped = fobj->swapped;
|
| 335 |
/* a bit useless, but a nice consistency check for ourselves now */
|
| 336 |
if (ret->lcmd->cmdsize % ret->align != 0)
|
| 337 |
warn("cmdsize isn't properly aligned on %d bytes boundary (%d)",
|
| 338 |
ret->align, ret->lcmd->cmdsize);
|
| 339 |
return ret;
|
| 340 |
}
|
| 341 |
|
| 342 |
/* Sets up the given loadcmd struct with the next load command, or frees
|
| 343 |
* it if there are no more load commands. If a new load command was
|
| 344 |
* loaded, 1 is returned, 0 otherwise. This behaviour is useful when
|
| 345 |
* looping over all load commands, since firstloadcmd will allocate the
|
| 346 |
* loadcmd struct, and nextloadcmd will free it once all load commands
|
| 347 |
* have been seen. */
|
| 348 |
int nextloadcmd(loadcmd *lcmd)
|
| 349 |
{
|
| 350 |
uint32_t size = MOBJGET(lcmd, lcmd->cmdsize);
|
| 351 |
|
| 352 |
if (--(lcmd->cleft) == 0) {
|
| 353 |
free(lcmd);
|
| 354 |
return 0;
|
| 355 |
}
|
| 356 |
|
| 357 |
if (size % lcmd->align != 0) {
|
| 358 |
/* fix alignment, this should actually never happen, but the doc
|
| 359 |
* says we have to pad if the alignment sucks */
|
| 360 |
size += lcmd->align - (size % lcmd->align);
|
| 361 |
}
|
| 362 |
lcmd->data += size;
|
| 363 |
lcmd->lcmd = lcmd->data;
|
| 364 |
|
| 365 |
return 1;
|
| 366 |
}
|
| 367 |
|
| 368 |
const char *get_machoendian(fatobj *fobj)
|
| 369 |
{
|
| 370 |
return fobj->isbigendian ? STR_BE : STR_LE;
|
| 371 |
}
|
| 372 |
|
| 373 |
const char *get_machomtype(fatobj *fobj)
|
| 374 |
{
|
| 375 |
switch (MOBJGET(fobj, mhdr.hdr32->cputype)) {
|
| 376 |
case CPU_TYPE_POWERPC: return STR_PPC;
|
| 377 |
case CPU_TYPE_I386: return STR_I386;
|
| 378 |
case CPU_TYPE_ARM: return STR_ARM;
|
| 379 |
case CPU_TYPE_POWERPC64: return STR_PPC64;
|
| 380 |
case CPU_TYPE_X86_64: return STR_X86_64;
|
| 381 |
default: return STR_UNKNOWN;
|
| 382 |
}
|
| 383 |
}
|