1 /* This is a cut-down version of loader.h from cctools-895, 2 shrunk to eliminate aspects unwanted in libdwarf and to avoid 3 #include entirely. All tab characters replaced with 4 spaces 4 so various things no line up as they used to. 5 cctools-895 in its original form 6 is available from https://opensource.apple.com/ 7 see Developer Tools version 8.2.1. cctools-895/include/loader.h */ 8 /* 9 * Copyright (c) 1999-2010 Apple Inc. All Rights Reserved. 10 * 11 * @APPLE_LICENSE_HEADER_START@ 12 * 13 * This file contains Original Code and/or Modifications of Original Code 14 * as defined in and that are subject to the Apple Public Source License 15 * Version 2.0 (the 'License'). You may not use this file except in 16 * compliance with the License. Please obtain a copy of the License at 17 * http://www.opensource.apple.com/apsl/ and read it before using this 18 * file. 19 * 20 * The Original Code and all software distributed under the License are 21 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER 22 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, 23 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, 24 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. 25 * Please see the License for the specific language governing rights and 26 * limitations under the License. 27 * 28 * @APPLE_LICENSE_HEADER_END@ 29 */ 30 #ifndef MACHO_LOADER_H 31 #define MACHO_LOADER_H 32 33 #ifdef __cplusplus 34 extern "C" { 35 #endif /* __cplusplus */ 36 37 38 #if 0 /* Not used here. DavidA. September 2018 */ 39 /* 40 * This file describes the format of mach object files. 41 */ 42 #include <stdint.h> 43 44 /* 45 * <mach/machine.h> is needed here for the cpu_type_t and cpu_subtype_t types 46 * and contains the constants for the possible values of these types. 47 */ 48 #include <mach/machine.h> 49 50 /* 51 * <mach/vm_prot.h> is needed here for the vm_prot_t type and contains the 52 * constants that are or'ed together for the possible values of this type. 53 */ 54 #include <mach/vm_prot.h> 55 56 /* 57 * <machine/thread_status.h> is expected to define the flavors of the thread 58 * states and the structures of those flavors for each machine. 59 */ 60 #include <mach/machine/thread_status.h> 61 #include <architecture/byte_order.h> 62 #endif /* 0 */ 63 64 #ifndef TYP 65 #define TYP(n,l) char n[l] 66 #endif /* TYP */ 67 68 /* 69 * The 32-bit mach header appears at the very beginning of the object file for 70 * 32-bit architectures. 71 */ 72 struct mach_header { 73 TYP(magic,4); /* mach magic number identifier */ 74 TYP(cputype,4); /* cpu specifier */ 75 TYP(cpusubtype,4); /* machine specifier */ 76 TYP(filetype,4); /* type of file */ 77 TYP(ncmds,4); /* number of load commands */ 78 TYP(sizeofcmds,4); /* the size of all the load commands */ 79 TYP(flags,4); /* flags */ 80 }; 81 82 /* Constant for the magic field of the 83 mach_header (32-bit architectures) 84 MH_MAGIC MH_MAGIC_64 appear in big-endian objects 85 MH_CIGAM MH_CIGAM_64 appear in little-endian objects */ 86 #define MH_MAGIC 0xfeedface /* the mach magic number */ 87 #define MH_CIGAM 0xcefaedfe /* NXSwapInt(MH_MAGIC) */ 88 89 /* 90 * The 64-bit mach header appears at the very beginning of object files for 91 * 64-bit architectures. 92 */ 93 struct mach_header_64 { 94 TYP(magic,4); /* mach magic number identifier */ 95 TYP(cputype,4); /* cpu specifier */ 96 TYP(cpusubtype,4); /* machine specifier */ 97 TYP(filetype,4); /* type of file */ 98 TYP(ncmds,4); /* number of load commands */ 99 TYP(sizeofcmds,4); /* the size of all the load commands */ 100 TYP(flags,4); /* flags */ 101 TYP(reserved,4); /* reserved */ 102 }; 103 104 /* Constant for the magic field of the mach_header_64 (64-bit architectures) */ 105 #define MH_MAGIC_64 0xfeedfacf /* the 64-bit mach magic number */ 106 #define MH_CIGAM_64 0xcffaedfe /* NXSwapInt(MH_MAGIC_64) */ 107 108 /* 109 * The layout of the file depends on the filetype. For all but the MH_OBJECT 110 * file type the segments are padded out and aligned on a segment alignment 111 * boundary for efficient demand pageing. The MH_EXECUTE, MH_FVMLIB, MH_DYLIB, 112 * MH_DYLINKER and MH_BUNDLE file types also have the headers included as part 113 * of their first segment. 114 * 115 * The file type MH_OBJECT is a compact format intended as output of the 116 * assembler and input (and possibly output) of the link editor (the .o 117 * format). All sections are in one unnamed segment with no segment padding. 118 * This format is used as an executable format when the file is so small the 119 * segment padding greatly increases its size. 120 * 121 * The file type MH_PRELOAD is an executable format intended for things that 122 * are not executed under the kernel (proms, stand alones, kernels, etc). The 123 * format can be executed under the kernel but may demand paged it and not 124 * preload it before execution. 125 * 126 * A core file is in MH_CORE format and can be any in an arbritray legal 127 * Mach-O file. 128 * 129 * Constants for the filetype field of the mach_header 130 */ 131 #define MH_OBJECT 0x1 /* relocatable object file */ 132 #define MH_EXECUTE 0x2 /* demand paged executable file */ 133 #define MH_FVMLIB 0x3 /* fixed VM shared library file */ 134 #define MH_CORE 0x4 /* core file */ 135 #define MH_PRELOAD 0x5 /* preloaded executable file */ 136 #define MH_DYLIB 0x6 /* dynamically bound shared library */ 137 #define MH_DYLINKER 0x7 /* dynamic link editor */ 138 #define MH_BUNDLE 0x8 /* dynamically bound bundle file */ 139 #define MH_DYLIB_STUB 0x9 /* shared library stub for static */ 140 /* linking only, no section contents */ 141 #define MH_DSYM 0xa /* companion file with only debug */ 142 /* sections */ 143 #define MH_KEXT_BUNDLE 0xb /* x86_64 kexts */ 144 145 /* Constants for the flags field of the mach_header */ 146 #define MH_NOUNDEFS 0x1 /* the object file has no undefined 147 references */ 148 #define MH_INCRLINK 0x2 /* the object file is the output of an 149 incremental link against a base file 150 and can't be link edited again */ 151 #define MH_DYLDLINK 0x4 /* the object file is input for the 152 dynamic linker and can't be staticly 153 link edited again */ 154 #define MH_BINDATLOAD 0x8 /* the object file's undefined 155 references are bound by the dynamic 156 linker when loaded. */ 157 #define MH_PREBOUND 0x10 /* the file has its dynamic undefined 158 references prebound. */ 159 #define MH_SPLIT_SEGS 0x20 /* the file has its read-only and 160 read-write segments split */ 161 #define MH_LAZY_INIT 0x40 /* the shared library init routine is 162 to be run lazily via catching memory 163 faults to its writeable segments 164 (obsolete) */ 165 #define MH_TWOLEVEL 0x80 /* the image is using two-level name 166 space bindings */ 167 #define MH_FORCE_FLAT 0x100 /* the executable is forcing all images 168 to use flat name space bindings */ 169 #define MH_NOMULTIDEFS 0x200 /* this umbrella guarantees no multiple 170 defintions of symbols in its 171 sub-images so the two-level namespace 172 hints can always be used. */ 173 #define MH_NOFIXPREBINDING 0x400 /* do not have dyld notify the 174 prebinding agent about this 175 executable */ 176 #define MH_PREBINDABLE 0x800 /* the binary is not prebound but can 177 have its prebinding redone. only used 178 when MH_PREBOUND is not set. */ 179 #define MH_ALLMODSBOUND 0x1000 /* indicates that this binary binds to 180 all two-level namespace modules of 181 its dependent libraries. only used 182 when MH_PREBINDABLE and MH_TWOLEVEL 183 are both set. */ 184 #define MH_SUBSECTIONS_VIA_SYMBOLS 0x2000/* safe to divide up the sections into 185 sub-sections via symbols for dead 186 code stripping */ 187 #define MH_CANONICAL 0x4000 /* the binary has been canonicalized 188 via the unprebind operation */ 189 #define MH_WEAK_DEFINES 0x8000 /* the final linked image contains 190 external weak symbols */ 191 #define MH_BINDS_TO_WEAK 0x10000 /* the final linked image uses 192 weak symbols */ 193 194 #define MH_ALLOW_STACK_EXECUTION 0x20000/* When this bit is set, all stacks 195 in the task will be given stack 196 execution privilege. Only used in 197 MH_EXECUTE filetypes. */ 198 #define MH_ROOT_SAFE 0x40000 /* When this bit is set, the binary 199 declares it is safe for use in 200 processes with uid zero */ 201 202 #define MH_SETUID_SAFE 0x80000 /* When this bit is set, the binary 203 declares it is safe for use in 204 processes when issetugid() is true */ 205 206 #define MH_NO_REEXPORTED_DYLIBS 0x100000 /* When this bit is set on a dylib, 207 the static linker does not need to 208 examine dependent dylibs to see 209 if any are re-exported */ 210 #define MH_PIE 0x200000 /* When this bit is set, the OS will 211 load the main executable at a 212 random address. Only used in 213 MH_EXECUTE filetypes. */ 214 #define MH_DEAD_STRIPPABLE_DYLIB 0x400000 /* Only for use on dylibs. When 215 linking against a dylib that 216 has this bit set, the static linker 217 will automatically not create a 218 LC_LOAD_DYLIB load command to the 219 dylib if no symbols are being 220 referenced from the dylib. */ 221 #define MH_HAS_TLV_DESCRIPTORS 0x800000 /* Contains a section of type 222 S_THREAD_LOCAL_VARIABLES */ 223 224 #define MH_NO_HEAP_EXECUTION 0x1000000 /* When this bit is set, the OS will 225 run the main executable with 226 a non-executable heap even on 227 platforms (e.g. i386) that don't 228 require it. Only used in MH_EXECUTE 229 filetypes. */ 230 231 #define MH_APP_EXTENSION_SAFE 0x02000000 /* The code was linked for use in an 232 application extension. */ 233 234 /* 235 * The load commands directly follow the mach_header. The total size of all 236 * of the commands is given by the sizeofcmds field in the mach_header. All 237 * load commands must have as their first two fields cmd and cmdsize. The cmd 238 * field is filled in with a constant for that command type. Each command type 239 * has a structure specifically for it. The cmdsize field is the size in bytes 240 * of the particular load command structure plus anything that follows it that 241 * is a part of the load command (i.e. section structures, strings, etc.). To 242 * advance to the next load command the cmdsize can be added to the offset or 243 * pointer of the current load command. The cmdsize for 32-bit architectures 244 * MUST be a multiple of 4 bytes and for 64-bit architectures MUST be a multiple 245 * of 8 bytes (these are forever the maximum alignment of any load commands). 246 * The padded bytes must be zero. All tables in the object file must also 247 * follow these rules so the file can be memory mapped. Otherwise the pointers 248 * to these tables will not work well or at all on some machines. With all 249 * padding zeroed like objects will compare byte for byte. 250 */ 251 struct load_command { 252 TYP(cmd,4); /* type of load command */ 253 TYP(cmdsize,4); /* total size of command in bytes */ 254 }; 255 256 /* 257 * After MacOS X 10.1 when a new load command is added that is required to be 258 * understood by the dynamic linker for the image to execute properly the 259 * LC_REQ_DYLD bit will be or'ed into the load command constant. If the dynamic 260 * linker sees such a load command it it does not understand will issue a 261 * "unknown load command required for execution" error and refuse to use the 262 * image. Other load commands without this bit that are not understood will 263 * simply be ignored. 264 */ 265 #define LC_REQ_DYLD 0x80000000 266 267 /* Constants for the cmd field of all load commands, the type */ 268 #define LC_SEGMENT 0x1 /* segment of this file to be mapped */ 269 #define LC_SYMTAB 0x2 /* link-edit stab symbol table info */ 270 #define LC_SYMSEG 0x3 /* link-edit gdb symbol table info (obsolete) */ 271 #define LC_THREAD 0x4 /* thread */ 272 #define LC_UNIXTHREAD 0x5 /* unix thread (includes a stack) */ 273 #define LC_LOADFVMLIB 0x6 /* load a specified fixed VM shared library */ 274 #define LC_IDFVMLIB 0x7 /* fixed VM shared library identification */ 275 #define LC_IDENT 0x8 /* object identification info (obsolete) */ 276 #define LC_FVMFILE 0x9 /* fixed VM file inclusion (internal use) */ 277 #define LC_PREPAGE 0xa /* prepage command (internal use) */ 278 #define LC_DYSYMTAB 0xb /* dynamic link-edit symbol table info */ 279 #define LC_LOAD_DYLIB 0xc /* load a dynamically linked shared library */ 280 #define LC_ID_DYLIB 0xd /* dynamically linked shared lib ident */ 281 #define LC_LOAD_DYLINKER 0xe /* load a dynamic linker */ 282 #define LC_ID_DYLINKER 0xf /* dynamic linker identification */ 283 #define LC_PREBOUND_DYLIB 0x10 /* modules prebound for a dynamically */ 284 /* linked shared library */ 285 #define LC_ROUTINES 0x11 /* image routines */ 286 #define LC_SUB_FRAMEWORK 0x12 /* sub framework */ 287 #define LC_SUB_UMBRELLA 0x13 /* sub umbrella */ 288 #define LC_SUB_CLIENT 0x14 /* sub client */ 289 #define LC_SUB_LIBRARY 0x15 /* sub library */ 290 #define LC_TWOLEVEL_HINTS 0x16 /* two-level namespace lookup hints */ 291 #define LC_PREBIND_CKSUM 0x17 /* prebind checksum */ 292 293 /* 294 * load a dynamically linked shared library that is allowed to be missing 295 * (all symbols are weak imported). 296 */ 297 #define LC_LOAD_WEAK_DYLIB (0x18 | LC_REQ_DYLD) 298 299 #define LC_SEGMENT_64 0x19 /* 64-bit segment of this file to be 300 mapped */ 301 #define LC_ROUTINES_64 0x1a /* 64-bit image routines */ 302 #define LC_UUID 0x1b /* the uuid */ 303 #define LC_RPATH (0x1c | LC_REQ_DYLD) /* runpath additions */ 304 #define LC_CODE_SIGNATURE 0x1d /* local of code signature */ 305 #define LC_SEGMENT_SPLIT_INFO 0x1e /* local of info to split segments */ 306 #define LC_REEXPORT_DYLIB (0x1f | LC_REQ_DYLD) /* load and re-export dylib */ 307 #define LC_LAZY_LOAD_DYLIB 0x20 /* delay load of dylib until first use */ 308 #define LC_ENCRYPTION_INFO 0x21 /* encrypted segment information */ 309 #define LC_DYLD_INFO 0x22 /* compressed dyld information */ 310 #define LC_DYLD_INFO_ONLY (0x22|LC_REQ_DYLD) /* compressed dyld information only */ 311 #define LC_LOAD_UPWARD_DYLIB (0x23 | LC_REQ_DYLD) /* load upward dylib */ 312 #define LC_VERSION_MIN_MACOSX 0x24 /* build for MacOSX min OS version */ 313 #define LC_VERSION_MIN_IPHONEOS 0x25 /* build for iPhoneOS min OS version */ 314 #define LC_FUNCTION_STARTS 0x26 /* compressed table of function start addresses */ 315 #define LC_DYLD_ENVIRONMENT 0x27 /* string for dyld to treat 316 like environment variable */ 317 #define LC_MAIN (0x28|LC_REQ_DYLD) /* replacement for LC_UNIXTHREAD */ 318 #define LC_DATA_IN_CODE 0x29 /* table of non-instructions in __text */ 319 #define LC_SOURCE_VERSION 0x2A /* source version used to build binary */ 320 #define LC_DYLIB_CODE_SIGN_DRS 0x2B /* Code signing DRs copied from linked dylibs */ 321 #define LC_ENCRYPTION_INFO_64 0x2C /* 64-bit encrypted segment information */ 322 #define LC_LINKER_OPTION 0x2D /* linker options in MH_OBJECT files */ 323 #define LC_LINKER_OPTIMIZATION_HINT 0x2E /* optimization hints in MH_OBJECT files */ 324 #define LC_VERSION_MIN_TVOS 0x2F /* build for AppleTV min OS version */ 325 #define LC_VERSION_MIN_WATCHOS 0x30 /* build for Watch min OS version */ 326 327 /* 328 * A variable length string in a load command is represented by an lc_str 329 * union. The strings are stored just after the load command structure and 330 * the offset is from the start of the load command structure. The size 331 * of the string is reflected in the cmdsize field of the load command. 332 * Once again any padded bytes to bring the cmdsize field to a multiple 333 * of 4 bytes must be zero. 334 */ 335 union lc_str { 336 TYP(offset,4); /* offset to the string */ 337 #ifndef __LP64__ 338 char *ptr; /* pointer to the string */ 339 #endif 340 }; 341 342 /* 343 * The segment load command indicates that a part of this file is to be 344 * mapped into the task's address space. The size of this segment in memory, 345 * vmsize, maybe equal to or larger than the amount to map from this file, 346 * filesize. The file is mapped starting at fileoff to the beginning of 347 * the segment in memory, vmaddr. The rest of the memory of the segment, 348 * if any, is allocated zero fill on demand. The segment's maximum virtual 349 * memory protection and initial virtual memory protection are specified 350 * by the maxprot and initprot fields. If the segment has sections then the 351 * section structures directly follow the segment command and their size is 352 * reflected in cmdsize. 353 */ 354 struct segment_command { /* for 32-bit architectures */ 355 TYP(cmd,4); /* LC_SEGMENT */ 356 TYP(cmdsize,4); /* includes sizeof section structs */ 357 char segname[16]; /* segment name */ 358 TYP(vmaddr,4); /* memory address of this segment */ 359 TYP(vmsize,4); /* memory size of this segment */ 360 TYP(fileoff,4); /* file offset of this segment */ 361 TYP(filesize,4); /* amount to map from the file */ 362 TYP(maxprot,4); /* maximum VM protection */ 363 TYP(initprot,4); /* initial VM protection */ 364 TYP(nsects,4); /* number of sections in segment */ 365 TYP(flags,4); /* flags */ 366 }; 367 368 /* 369 * The 64-bit segment load command indicates that a part of this file is to be 370 * mapped into a 64-bit task's address space. If the 64-bit segment has 371 * sections then section_64 structures directly follow the 64-bit segment 372 * command and their size is reflected in cmdsize. 373 */ 374 struct segment_command_64 { /* for 64-bit architectures */ 375 TYP(cmd,4); /* LC_SEGMENT_64 */ 376 TYP(cmdsize,4); /* includes sizeof section_64 structs */ 377 char segname[16]; /* segment name */ 378 TYP(vmaddr,8); /* memory address of this segment */ 379 TYP(vmsize,8); /* memory size of this segment */ 380 TYP(fileoff,8); /* file offset of this segment */ 381 TYP(filesize,8); /* amount to map from the file */ 382 TYP(maxprot,4); /* maximum VM protection */ 383 TYP(initprot,4); /* initial VM protection */ 384 TYP(nsects,4); /* number of sections in segment */ 385 TYP(flags,4); /* flags */ 386 }; 387 388 /* Constants for the flags field of the segment_command */ 389 #define SG_HIGHVM 0x1 /* the file contents for this segment is for 390 the high part of the VM space, the low part 391 is zero filled (for stacks in core files) */ 392 #define SG_FVMLIB 0x2 /* this segment is the VM that is allocated by 393 a fixed VM library, for overlap checking in 394 the link editor */ 395 #define SG_NORELOC 0x4 /* this segment has nothing that was relocated 396 in it and nothing relocated to it, that is 397 it maybe safely replaced without relocation*/ 398 #define SG_PROTECTED_VERSION_1 0x8 /* This segment is protected. If the 399 segment starts at file offset 0, the 400 first page of the segment is not 401 protected. All other pages of the 402 segment are protected. */ 403 404 /* 405 * A segment is made up of zero or more sections. Non-MH_OBJECT files have 406 * all of their segments with the proper sections in each, and padded to the 407 * specified segment alignment when produced by the link editor. The first 408 * segment of a MH_EXECUTE and MH_FVMLIB format file contains the mach_header 409 * and load commands of the object file before its first section. The zero 410 * fill sections are always last in their segment (in all formats). This 411 * allows the zeroed segment padding to be mapped into memory where zero fill 412 * sections might be. The gigabyte zero fill sections, those with the section 413 * type S_GB_ZEROFILL, can only be in a segment with sections of this type. 414 * These segments are then placed after all other segments. 415 * 416 * The MH_OBJECT format has all of its sections in one segment for 417 * compactness. There is no padding to a specified segment boundary and the 418 * mach_header and load commands are not part of the segment. 419 * 420 * Sections with the same section name, sectname, going into the same segment, 421 * segname, are combined by the link editor. The resulting section is aligned 422 * to the maximum alignment of the combined sections and is the new section's 423 * alignment. The combined sections are aligned to their original alignment in 424 * the combined section. Any padded bytes to get the specified alignment are 425 * zeroed. 426 * 427 * The format of the relocation entries referenced by the reloff and nreloc 428 * fields of the section structure for mach object files is described in the 429 * header file <reloc.h>. 430 */ 431 struct section { /* for 32-bit architectures */ 432 char sectname[16]; /* name of this section */ 433 char segname[16]; /* segment this section goes in */ 434 TYP(addr,4); /* memory address of this section */ 435 TYP(size,4); /* size in bytes of this section */ 436 TYP(offset,4); /* file offset of this section */ 437 TYP(align,4); /* section alignment (power of 2) */ 438 TYP(reloff,4); /* file offset of relocation entries */ 439 TYP(nreloc,4); /* number of relocation entries */ 440 TYP(flags,4); /* flags (section type and attributes)*/ 441 TYP(reserved1,4); /* reserved (for offset or index) */ 442 TYP(reserved2,4); /* reserved (for count or sizeof) */ 443 }; 444 445 struct section_64 { /* for 64-bit architectures */ 446 char sectname[16]; /* name of this section */ 447 char segname[16]; /* segment this section goes in */ 448 TYP(addr,8); /* memory address of this section */ 449 TYP(size,8); /* size in bytes of this section */ 450 TYP(offset,4); /* file offset of this section */ 451 TYP(align,4); /* section alignment (power of 2) */ 452 TYP(reloff,4); /* file offset of relocation entries */ 453 TYP(nreloc,4); /* number of relocation entries */ 454 TYP(flags,4); /* flags (section type and attributes)*/ 455 TYP(reserved1,4); /* reserved (for offset or index) */ 456 TYP(reserved2,4); /* reserved (for count or sizeof) */ 457 TYP(reserved3,4); /* reserved */ 458 }; 459 460 /* 461 * The flags field of a section structure is separated into two parts a section 462 * type and section attributes. The section types are mutually exclusive (it 463 * can only have one type) but the section attributes are not (it may have more 464 * than one attribute). 465 */ 466 #define SECTION_TYPE 0x000000ff /* 256 section types */ 467 #define SECTION_ATTRIBUTES 0xffffff00 /* 24 section attributes */ 468 469 /* Constants for the type of a section */ 470 #define S_REGULAR 0x0 /* regular section */ 471 #define S_ZEROFILL 0x1 /* zero fill on demand section */ 472 #define S_CSTRING_LITERALS 0x2 /* section with only literal C strings*/ 473 #define S_4BYTE_LITERALS 0x3 /* section with only 4 byte literals */ 474 #define S_8BYTE_LITERALS 0x4 /* section with only 8 byte literals */ 475 #define S_LITERAL_POINTERS 0x5 /* section with only pointers to */ 476 /* literals */ 477 /* 478 * For the two types of symbol pointers sections and the symbol stubs section 479 * they have indirect symbol table entries. For each of the entries in the 480 * section the indirect symbol table entries, in corresponding order in the 481 * indirect symbol table, start at the index stored in the reserved1 field 482 * of the section structure. Since the indirect symbol table entries 483 * correspond to the entries in the section the number of indirect symbol table 484 * entries is inferred from the size of the section divided by the size of the 485 * entries in the section. For symbol pointers sections the size of the entries 486 * in the section is 4 bytes and for symbol stubs sections the byte size of the 487 * stubs is stored in the reserved2 field of the section structure. 488 */ 489 #define S_NON_LAZY_SYMBOL_POINTERS 0x6 /* section with only non-lazy 490 symbol pointers */ 491 #define S_LAZY_SYMBOL_POINTERS 0x7 /* section with only lazy symbol 492 pointers */ 493 #define S_SYMBOL_STUBS 0x8 /* section with only symbol 494 stubs, byte size of stub in 495 the reserved2 field */ 496 #define S_MOD_INIT_FUNC_POINTERS 0x9 /* section with only function 497 pointers for initialization*/ 498 #define S_MOD_TERM_FUNC_POINTERS 0xa /* section with only function 499 pointers for termination */ 500 #define S_COALESCED 0xb /* section contains symbols that 501 are to be coalesced */ 502 #define S_GB_ZEROFILL 0xc /* zero fill on demand section 503 (that can be larger than 4 504 gigabytes) */ 505 #define S_INTERPOSING 0xd /* section with only pairs of 506 function pointers for 507 interposing */ 508 #define S_16BYTE_LITERALS 0xe /* section with only 16 byte 509 literals */ 510 #define S_DTRACE_DOF 0xf /* section contains 511 DTrace Object Format */ 512 #define S_LAZY_DYLIB_SYMBOL_POINTERS 0x10 /* section with only lazy 513 symbol pointers to lazy 514 loaded dylibs */ 515 /* 516 * Section types to support thread local variables 517 */ 518 #define S_THREAD_LOCAL_REGULAR 0x11 /* template of initial 519 values for TLVs */ 520 #define S_THREAD_LOCAL_ZEROFILL 0x12 /* template of initial 521 values for TLVs */ 522 #define S_THREAD_LOCAL_VARIABLES 0x13 /* TLV descriptors */ 523 #define S_THREAD_LOCAL_VARIABLE_POINTERS 0x14 /* pointers to TLV 524 descriptors */ 525 #define S_THREAD_LOCAL_INIT_FUNCTION_POINTERS 0x15 /* functions to call 526 to initialize TLV 527 values */ 528 529 /* 530 * Constants for the section attributes part of the flags field of a section 531 * structure. 532 */ 533 #define SECTION_ATTRIBUTES_USR 0xff000000 /* User setable attributes */ 534 #define S_ATTR_PURE_INSTRUCTIONS 0x80000000 /* section contains only true 535 machine instructions */ 536 #define S_ATTR_NO_TOC 0x40000000 /* section contains coalesced 537 symbols that are not to be 538 in a ranlib table of 539 contents */ 540 #define S_ATTR_STRIP_STATIC_SYMS 0x20000000 /* ok to strip static symbols 541 in this section in files 542 with the MH_DYLDLINK flag */ 543 #define S_ATTR_NO_DEAD_STRIP 0x10000000 /* no dead stripping */ 544 #define S_ATTR_LIVE_SUPPORT 0x08000000 /* blocks are live if they 545 reference live blocks */ 546 #define S_ATTR_SELF_MODIFYING_CODE 0x04000000 /* Used with i386 code stubs 547 written on by dyld */ 548 /* 549 * If a segment contains any sections marked with S_ATTR_DEBUG then all 550 * sections in that segment must have this attribute. No section other than 551 * a section marked with this attribute may reference the contents of this 552 * section. A section with this attribute may contain no symbols and must have 553 * a section type S_REGULAR. The static linker will not copy section contents 554 * from sections with this attribute into its output file. These sections 555 * generally contain DWARF debugging info. 556 */ 557 #define S_ATTR_DEBUG 0x02000000 /* a debug section */ 558 #define SECTION_ATTRIBUTES_SYS 0x00ffff00 /* system setable attributes */ 559 #define S_ATTR_SOME_INSTRUCTIONS 0x00000400 /* section contains some 560 machine instructions */ 561 #define S_ATTR_EXT_RELOC 0x00000200 /* section has external 562 relocation entries */ 563 #define S_ATTR_LOC_RELOC 0x00000100 /* section has local 564 relocation entries */ 565 566 567 /* 568 * The names of segments and sections in them are mostly meaningless to the 569 * link-editor. But there are few things to support traditional UNIX 570 * executables that require the link-editor and assembler to use some names 571 * agreed upon by convention. 572 * 573 * The initial protection of the "__TEXT" segment has write protection turned 574 * off (not writeable). 575 * 576 * The link-editor will allocate common symbols at the end of the "__common" 577 * section in the "__DATA" segment. It will create the section and segment 578 * if needed. 579 */ 580 581 /* The currently known segment names and the section names in those segments */ 582 583 #define SEG_PAGEZERO "__PAGEZERO" /* the pagezero segment which has no */ 584 /* protections and catches NULL */ 585 /* references for MH_EXECUTE files */ 586 587 588 #define SEG_TEXT "__TEXT" /* the tradition UNIX text segment */ 589 #define SECT_TEXT "__text" /* the real text part of the text */ 590 /* section no headers, and no padding */ 591 #define SECT_FVMLIB_INIT0 "__fvmlib_init0" /* the fvmlib initialization */ 592 /* section */ 593 #define SECT_FVMLIB_INIT1 "__fvmlib_init1" /* the section following the */ 594 /* fvmlib initialization */ 595 /* section */ 596 597 #define SEG_DATA "__DATA" /* the tradition UNIX data segment */ 598 #define SECT_DATA "__data" /* the real initialized data section */ 599 /* no padding, no bss overlap */ 600 #define SECT_BSS "__bss" /* the real uninitialized data section*/ 601 /* no padding */ 602 #define SECT_COMMON "__common" /* the section common symbols are */ 603 /* allocated in by the link editor */ 604 605 #define SEG_OBJC "__OBJC" /* objective-C runtime segment */ 606 #define SECT_OBJC_SYMBOLS "__symbol_table" /* symbol table */ 607 #define SECT_OBJC_MODULES "__module_info" /* module information */ 608 #define SECT_OBJC_STRINGS "__selector_strs" /* string table */ 609 #define SECT_OBJC_REFS "__selector_refs" /* string table */ 610 611 #define SEG_ICON "__ICON" /* the icon segment */ 612 #define SECT_ICON_HEADER "__header" /* the icon headers */ 613 #define SECT_ICON_TIFF "__tiff" /* the icons in tiff format */ 614 615 #define SEG_LINKEDIT "__LINKEDIT" /* the segment containing all structs */ 616 /* created and maintained by the link */ 617 /* editor. Created with -seglinkedit */ 618 /* option to ld(1) for MH_EXECUTE and */ 619 /* FVMLIB file types only */ 620 621 #define SEG_UNIXSTACK "__UNIXSTACK" /* the unix stack segment */ 622 623 #define SEG_IMPORT "__IMPORT" /* the segment for the self (dyld) */ 624 /* modifing code stubs that has read, */ 625 /* write and execute permissions */ 626 627 /* 628 * Fixed virtual memory shared libraries are identified by two things. The 629 * target pathname (the name of the library as found for execution), and the 630 * minor version number. The address of where the headers are loaded is in 631 * header_addr. (THIS IS OBSOLETE and no longer supported). 632 */ 633 struct fvmlib { 634 union lc_str name; /* library's target pathname */ 635 TYP(minor_version,4); /* library's minor version number */ 636 TYP(header_addr,4); /* library's header address */ 637 }; 638 639 /* 640 * A fixed virtual shared library (filetype == MH_FVMLIB in the mach header) 641 * contains a fvmlib_command (cmd == LC_IDFVMLIB) to identify the library. 642 * An object that uses a fixed virtual shared library also contains a 643 * fvmlib_command (cmd == LC_LOADFVMLIB) for each library it uses. 644 * (THIS IS OBSOLETE and no longer supported). 645 */ 646 struct fvmlib_command { 647 TYP(cmd,4); /* LC_IDFVMLIB or LC_LOADFVMLIB */ 648 TYP(cmdsize,4); /* includes pathname string */ 649 struct fvmlib fvmlib; /* the library identification */ 650 }; 651 652 /* 653 * Dynamicly linked shared libraries are identified by two things. The 654 * pathname (the name of the library as found for execution), and the 655 * compatibility version number. The pathname must match and the compatibility 656 * number in the user of the library must be greater than or equal to the 657 * library being used. The time stamp is used to record the time a library was 658 * built and copied into user so it can be use to determined if the library used 659 * at runtime is exactly the same as used to built the program. 660 */ 661 struct dylib { 662 union lc_str name; /* library's path name */ 663 TYP(timestamp,4); /* library's build time stamp */ 664 TYP(current_version,4); /* library's current version number */ 665 TYP(compatibility_version,4); /* library's compatibility vers number*/ 666 }; 667 668 /* 669 * A dynamically linked shared library (filetype == MH_DYLIB in the mach header) 670 * contains a dylib_command (cmd == LC_ID_DYLIB) to identify the library. 671 * An object that uses a dynamically linked shared library also contains a 672 * dylib_command (cmd == LC_LOAD_DYLIB, LC_LOAD_WEAK_DYLIB, or 673 * LC_REEXPORT_DYLIB) for each library it uses. 674 */ 675 struct dylib_command { 676 TYP(cmd,4); /* LC_ID_DYLIB, LC_LOAD_{,WEAK_}DYLIB, 677 LC_REEXPORT_DYLIB */ 678 TYP(cmdsize,4); /* includes pathname string */ 679 struct dylib dylib; /* the library identification */ 680 }; 681 682 /* 683 * A dynamically linked shared library may be a subframework of an umbrella 684 * framework. If so it will be linked with "-umbrella umbrella_name" where 685 * Where "umbrella_name" is the name of the umbrella framework. A subframework 686 * can only be linked against by its umbrella framework or other subframeworks 687 * that are part of the same umbrella framework. Otherwise the static link 688 * editor produces an error and states to link against the umbrella framework. 689 * The name of the umbrella framework for subframeworks is recorded in the 690 * following structure. 691 */ 692 struct sub_framework_command { 693 TYP(cmd,4); /* LC_SUB_FRAMEWORK */ 694 TYP(cmdsize,4); /* includes umbrella string */ 695 union lc_str umbrella; /* the umbrella framework name */ 696 }; 697 698 /* 699 * For dynamically linked shared libraries that are subframework of an umbrella 700 * framework they can allow clients other than the umbrella framework or other 701 * subframeworks in the same umbrella framework. To do this the subframework 702 * is built with "-allowable_client client_name" and an LC_SUB_CLIENT load 703 * command is created for each -allowable_client flag. The client_name is 704 * usually a framework name. It can also be a name used for bundles clients 705 * where the bundle is built with "-client_name client_name". 706 */ 707 struct sub_client_command { 708 TYP(cmd,4); /* LC_SUB_CLIENT */ 709 TYP(cmdsize,4); /* includes client string */ 710 union lc_str client; /* the client name */ 711 }; 712 713 /* 714 * A dynamically linked shared library may be a sub_umbrella of an umbrella 715 * framework. If so it will be linked with "-sub_umbrella umbrella_name" where 716 * Where "umbrella_name" is the name of the sub_umbrella framework. When 717 * staticly linking when -twolevel_namespace is in effect a twolevel namespace 718 * umbrella framework will only cause its subframeworks and those frameworks 719 * listed as sub_umbrella frameworks to be implicited linked in. Any other 720 * dependent dynamic libraries will not be linked it when -twolevel_namespace 721 * is in effect. The primary library recorded by the static linker when 722 * resolving a symbol in these libraries will be the umbrella framework. 723 * Zero or more sub_umbrella frameworks may be use by an umbrella framework. 724 * The name of a sub_umbrella framework is recorded in the following structure. 725 */ 726 struct sub_umbrella_command { 727 TYP(cmd,4); /* LC_SUB_UMBRELLA */ 728 TYP(cmdsize,4); /* includes sub_umbrella string */ 729 union lc_str sub_umbrella; /* the sub_umbrella framework name */ 730 }; 731 732 /* 733 * A dynamically linked shared library may be a sub_library of another shared 734 * library. If so it will be linked with "-sub_library library_name" where 735 * Where "library_name" is the name of the sub_library shared library. When 736 * staticly linking when -twolevel_namespace is in effect a twolevel namespace 737 * shared library will only cause its subframeworks and those frameworks 738 * listed as sub_umbrella frameworks and libraries listed as sub_libraries to 739 * be implicited linked in. Any other dependent dynamic libraries will not be 740 * linked it when -twolevel_namespace is in effect. The primary library 741 * recorded by the static linker when resolving a symbol in these libraries 742 * will be the umbrella framework (or dynamic library). Zero or more sub_library 743 * shared libraries may be use by an umbrella framework or (or dynamic library). 744 * The name of a sub_library framework is recorded in the following structure. 745 * For example /usr/lib/libobjc_profile.A.dylib would be recorded as "libobjc". 746 */ 747 struct sub_library_command { 748 TYP(cmd,4); /* LC_SUB_LIBRARY */ 749 TYP(cmdsize,4); /* includes sub_library string */ 750 union lc_str sub_library; /* the sub_library name */ 751 }; 752 753 /* 754 * A program (filetype == MH_EXECUTE) that is 755 * prebound to its dynamic libraries has one of these for each library that 756 * the static linker used in prebinding. It contains a bit vector for the 757 * modules in the library. The bits indicate which modules are bound (1) and 758 * which are not (0) from the library. The bit for module 0 is the low bit 759 * of the first byte. So the bit for the Nth module is: 760 * (linked_modules[N/8] >> N%8) & 1 761 */ 762 struct prebound_dylib_command { 763 TYP(cmd,4); /* LC_PREBOUND_DYLIB */ 764 TYP(cmdsize,4); /* includes strings */ 765 union lc_str name; /* library's path name */ 766 TYP(nmodules,4); /* number of modules in library */ 767 union lc_str linked_modules; /* bit vector of linked modules */ 768 }; 769 770 /* 771 * A program that uses a dynamic linker contains a dylinker_command to identify 772 * the name of the dynamic linker (LC_LOAD_DYLINKER). And a dynamic linker 773 * contains a dylinker_command to identify the dynamic linker (LC_ID_DYLINKER). 774 * A file can have at most one of these. 775 * This struct is also used for the LC_DYLD_ENVIRONMENT load command and 776 * contains string for dyld to treat like environment variable. 777 */ 778 struct dylinker_command { 779 TYP(cmd,4); /* LC_ID_DYLINKER, LC_LOAD_DYLINKER or 780 LC_DYLD_ENVIRONMENT */ 781 TYP(cmdsize,4); /* includes pathname string */ 782 union lc_str name; /* dynamic linker's path name */ 783 }; 784 785 /* 786 * Thread commands contain machine-specific data structures suitable for 787 * use in the thread state primitives. The machine specific data structures 788 * follow the struct thread_command as follows. 789 * Each flavor of machine specific data structure is preceded by an unsigned 790 * long constant for the flavor of that data structure, an uint32_t 791 * that is the count of longs of the size of the state data structure and then 792 * the state data structure follows. This triple may be repeated for many 793 * flavors. The constants for the flavors, counts and state data structure 794 * definitions are expected to be in the header file <machine/thread_status.h>. 795 * These machine specific data structures sizes must be multiples of 796 * 4 bytes The cmdsize reflects the total size of the thread_command 797 * and all of the sizes of the constants for the flavors, counts and state 798 * data structures. 799 * 800 * For executable objects that are unix processes there will be one 801 * thread_command (cmd == LC_UNIXTHREAD) created for it by the link-editor. 802 * This is the same as a LC_THREAD, except that a stack is automatically 803 * created (based on the shell's limit for the stack size). Command arguments 804 * and environment variables are copied onto that stack. 805 */ 806 struct thread_command { 807 TYP(cmd,4); /* LC_THREAD or LC_UNIXTHREAD */ 808 TYP(cmdsize,4); /* total size of this command */ 809 /* uint32_t flavor flavor of thread state */ 810 /* uint32_t count count of longs in thread state */ 811 /* struct XXX_thread_state state thread state for this flavor */ 812 /* ... */ 813 }; 814 815 /* 816 * The routines command contains the address of the dynamic shared library 817 * initialization routine and an index into the module table for the module 818 * that defines the routine. Before any modules are used from the library the 819 * dynamic linker fully binds the module that defines the initialization routine 820 * and then calls it. This gets called before any module initialization 821 * routines (used for C++ static constructors) in the library. 822 */ 823 struct routines_command { /* for 32-bit architectures */ 824 TYP(cmd,4); /* LC_ROUTINES */ 825 TYP(cmdsize,4); /* total size of this command */ 826 TYP(init_address,4); /* address of initialization routine */ 827 TYP(init_module,4); /* index into the module table that */ 828 /* the init routine is defined in */ 829 TYP(reserved1,4); 830 TYP(reserved2,4); 831 TYP(reserved3,4); 832 TYP(reserved4,4); 833 TYP(reserved5,4); 834 TYP(reserved6,4); 835 }; 836 837 /* 838 * The 64-bit routines command. Same use as above. 839 */ 840 struct routines_command_64 { /* for 64-bit architectures */ 841 TYP(cmd,4); /* LC_ROUTINES_64 */ 842 TYP(cmdsize,4); /* total size of this command */ 843 TYP(init_address,8); /* address of initialization routine */ 844 TYP(init_module,8); /* index into the module table that */ 845 /* the init routine is defined in */ 846 TYP(reserved1,8); 847 TYP(reserved2,8); 848 TYP(reserved3,8); 849 TYP(reserved4,8); 850 TYP(reserved5,8); 851 TYP(reserved6,8); 852 }; 853 854 /* 855 * The symtab_command contains the offsets and sizes of the link-edit 4.3BSD 856 * "stab" style symbol table information as described in the header files 857 * <nlist.h> and <stab.h>. 858 */ 859 struct symtab_command { 860 TYP(cmd,4); /* LC_SYMTAB */ 861 TYP(cmdsize,4); /* sizeof(struct symtab_command) */ 862 TYP(symoff,4); /* symbol table offset */ 863 TYP(nsyms,4); /* number of symbol table entries */ 864 TYP(stroff,4); /* string table offset */ 865 TYP(strsize,4); /* string table size in bytes */ 866 }; 867 868 /* 869 * This is the second set of the symbolic information which is used to support 870 * the data structures for the dynamically link editor. 871 * 872 * The original set of symbolic information in the symtab_command which contains 873 * the symbol and string tables must also be present when this load command is 874 * present. When this load command is present the symbol table is organized 875 * into three groups of symbols: 876 * local symbols (static and debugging symbols) - grouped by module 877 * defined external symbols - grouped by module (sorted by name if not lib) 878 * undefined external symbols (sorted by name if MH_BINDATLOAD is not set, 879 * and in order the were seen by the static 880 * linker if MH_BINDATLOAD is set) 881 * In this load command there are offsets and counts to each of the three groups 882 * of symbols. 883 * 884 * This load command contains a the offsets and sizes of the following new 885 * symbolic information tables: 886 * table of contents 887 * module table 888 * reference symbol table 889 * indirect symbol table 890 * The first three tables above (the table of contents, module table and 891 * reference symbol table) are only present if the file is a dynamically linked 892 * shared library. For executable and object modules, which are files 893 * containing only one module, the information that would be in these three 894 * tables is determined as follows: 895 * table of contents - the defined external symbols are sorted by name 896 * module table - the file contains only one module so everything in the 897 * file is part of the module. 898 * reference symbol table - is the defined and undefined external symbols 899 * 900 * For dynamically linked shared library files this load command also contains 901 * offsets and sizes to the pool of relocation entries for all sections 902 * separated into two groups: 903 * external relocation entries 904 * local relocation entries 905 * For executable and object modules the relocation entries continue to hang 906 * off the section structures. 907 */ 908 struct dysymtab_command { 909 TYP(cmd,4); /* LC_DYSYMTAB */ 910 TYP(cmdsize,4); /* sizeof(struct dysymtab_command) */ 911 912 /* 913 * The symbols indicated by symoff and nsyms of the LC_SYMTAB load command 914 * are grouped into the following three groups: 915 * local symbols (further grouped by the module they are from) 916 * defined external symbols (further grouped by the module they are from) 917 * undefined symbols 918 * 919 * The local symbols are used only for debugging. The dynamic binding 920 * process may have to use them to indicate to the debugger the local 921 * symbols for a module that is being bound. 922 * 923 * The last two groups are used by the dynamic binding process to do the 924 * binding (indirectly through the module table and the reference symbol 925 * table when this is a dynamically linked shared library file). 926 */ 927 TYP(ilocalsym,4); /* index to local symbols */ 928 TYP(nlocalsym,4); /* number of local symbols */ 929 930 TYP(iextdefsym,4); /* index to externally defined symbols */ 931 TYP(nextdefsym,4); /* number of externally defined symbols */ 932 933 TYP(iundefsym,4); /* index to undefined symbols */ 934 TYP(nundefsym,4); /* number of undefined symbols */ 935 936 /* 937 * For the for the dynamic binding process to find which module a symbol 938 * is defined in the table of contents is used (analogous to the ranlib 939 * structure in an archive) which maps defined external symbols to modules 940 * they are defined in. This exists only in a dynamically linked shared 941 * library file. For executable and object modules the defined external 942 * symbols are sorted by name and is use as the table of contents. 943 */ 944 TYP(tocoff,4); /* file offset to table of contents */ 945 TYP(ntoc,4); /* number of entries in table of contents */ 946 947 /* 948 * To support dynamic binding of "modules" (whole object files) the symbol 949 * table must reflect the modules that the file was created from. This is 950 * done by having a module table that has indexes and counts into the merged 951 * tables for each module. The module structure that these two entries 952 * refer to is described below. This exists only in a dynamically linked 953 * shared library file. For executable and object modules the file only 954 * contains one module so everything in the file belongs to the module. 955 */ 956 TYP(modtaboff,4); /* file offset to module table */ 957 TYP(nmodtab,4); /* number of module table entries */ 958 959 /* 960 * To support dynamic module binding the module structure for each module 961 * indicates the external references (defined and undefined) each module 962 * makes. For each module there is an offset and a count into the 963 * reference symbol table for the symbols that the module references. 964 * This exists only in a dynamically linked shared library file. For 965 * executable and object modules the defined external symbols and the 966 * undefined external symbols indicates the external references. 967 */ 968 TYP(extrefsymoff,4); /* offset to referenced symbol table */ 969 TYP(nextrefsyms,4); /* number of referenced symbol table entries */ 970 971 /* 972 * The sections that contain "symbol pointers" and "routine stubs" have 973 * indexes and (implied counts based on the size of the section and fixed 974 * size of the entry) into the "indirect symbol" table for each pointer 975 * and stub. For every section of these two types the index into the 976 * indirect symbol table is stored in the section header in the field 977 * reserved1. An indirect symbol table entry is simply a 32bit index into 978 * the symbol table to the symbol that the pointer or stub is referring to. 979 * The indirect symbol table is ordered to match the entries in the section. 980 */ 981 TYP(indirectsymoff,4); /* file offset to the indirect symbol table */ 982 TYP(nindirectsyms,4); /* number of indirect symbol table entries */ 983 984 /* 985 * To support relocating an individual module in a library file quickly the 986 * external relocation entries for each module in the library need to be 987 * accessed efficiently. Since the relocation entries can't be accessed 988 * through the section headers for a library file they are separated into 989 * groups of local and external entries further grouped by module. In this 990 * case the presents of this load command who's extreloff, nextrel, 991 * locreloff and nlocrel fields are non-zero indicates that the relocation 992 * entries of non-merged sections are not referenced through the section 993 * structures (and the reloff and nreloc fields in the section headers are 994 * set to zero). 995 * 996 * Since the relocation entries are not accessed through the section headers 997 * this requires the r_address field to be something other than a section 998 * offset to identify the item to be relocated. In this case r_address is 999 * set to the offset from the vmaddr of the first LC_SEGMENT command. 1000 * For MH_SPLIT_SEGS images r_address is set to the the offset from the 1001 * vmaddr of the first read-write LC_SEGMENT command. 1002 * 1003 * The relocation entries are grouped by module and the module table 1004 * entries have indexes and counts into them for the group of external 1005 * relocation entries for that the module. 1006 * 1007 * For sections that are merged across modules there must not be any 1008 * remaining external relocation entries for them (for merged sections 1009 * remaining relocation entries must be local). 1010 */ 1011 TYP(extreloff,4); /* offset to external relocation entries */ 1012 TYP(nextrel,4); /* number of external relocation entries */ 1013 1014 /* 1015 * All the local relocation entries are grouped together (they are not 1016 * grouped by their module since they are only used if the object is moved 1017 * from it staticly link edited address). 1018 */ 1019 TYP(locreloff,4); /* offset to local relocation entries */ 1020 TYP(nlocrel,4); /* number of local relocation entries */ 1021 }; 1022 1023 /* 1024 * An indirect symbol table entry is simply a 32bit index into the symbol table 1025 * to the symbol that the pointer or stub is refering to. Unless it is for a 1026 * non-lazy symbol pointer section for a defined symbol which strip(1) as 1027 * removed. In which case it has the value INDIRECT_SYMBOL_LOCAL. If the 1028 * symbol was also absolute INDIRECT_SYMBOL_ABS is or'ed with that. 1029 */ 1030 #define INDIRECT_SYMBOL_LOCAL 0x80000000 1031 #define INDIRECT_SYMBOL_ABS 0x40000000 1032 1033 1034 /* a table of contents entry */ 1035 struct dylib_table_of_contents { 1036 TYP(symbol_index,4); /* the defined external symbol 1037 (index into the symbol table) */ 1038 TYP(module_index,4); /* index into the module table this symbol 1039 is defined in */ 1040 }; 1041 1042 /* a module table entry */ 1043 struct dylib_module { 1044 TYP(module_name,4); /* the module name (index into string table) */ 1045 1046 TYP(iextdefsym,4); /* index into externally defined symbols */ 1047 TYP(nextdefsym,4); /* number of externally defined symbols */ 1048 TYP(irefsym,4); /* index into reference symbol table */ 1049 TYP(nrefsym,4); /* number of reference symbol table entries */ 1050 TYP(ilocalsym,4); /* index into symbols for local symbols */ 1051 TYP(nlocalsym,4); /* number of local symbols */ 1052 1053 TYP(iextrel,4); /* index into external relocation entries */ 1054 TYP(nextrel,4); /* number of external relocation entries */ 1055 1056 TYP(iinit_iterm,4); /* low 16 bits are the index into the init 1057 section, high 16 bits are the index into 1058 the term section */ 1059 TYP(ninit_nterm,4); /* low 16 bits are the number of init section 1060 entries, high 16 bits are the number of 1061 term section entries */ 1062 1063 /* for this module address of the start of */ 1064 /* the (__OBJC,__module_info) section */ 1065 TYP(objc_module_info_addr,4); 1066 1067 /* for this module size of */ 1068 /* the (__OBJC,__module_info) section */ 1069 TYP(objc_module_info_size,4); 1070 }; 1071 1072 /* a 64-bit module table entry */ 1073 struct dylib_module_64 { 1074 TYP(module_name,4); /* the module name (index into string table) */ 1075 1076 TYP(iextdefsym,4); /* index into externally defined symbols */ 1077 TYP(nextdefsym,4); /* number of externally defined symbols */ 1078 TYP(irefsym,4); /* index into reference symbol table */ 1079 TYP(nrefsym,4); /* number of reference symbol table entries */ 1080 TYP(ilocalsym,4); /* index into symbols for local symbols */ 1081 TYP(nlocalsym,4); /* number of local symbols */ 1082 1083 TYP(iextrel,4); /* index into external relocation entries */ 1084 TYP(nextrel,4); /* number of external relocation entries */ 1085 1086 TYP(iinit_iterm,4); /* low 16 bits are the index into the init 1087 section, high 16 bits are the index into 1088 the term section */ 1089 TYP(ninit_nterm,4); /* low 16 bits are the number of init section 1090 entries, high 16 bits are the number of 1091 term section entries */ 1092 1093 TYP(objc_module_info_size,4); /* for this module size of */ 1094 /* the (__OBJC,__module_info) section */ 1095 TYP(objc_module_info_addr,8); /* for this module address of the start of */ 1096 /* the (__OBJC,__module_info) section */ 1097 }; 1098 1099 /* 1100 * The entries in the reference symbol table are used when loading the module 1101 * (both by the static and dynamic link editors) and if the module is unloaded 1102 * or replaced. Therefore all external symbols (defined and undefined) are 1103 * listed in the module's reference table. The flags describe the type of 1104 * reference that is being made. The constants for the flags are defined in 1105 * <mach-o/nlist.h> as they are also used for symbol table entries. 1106 */ 1107 #if 0 /* dwarf readers not using this */ 1108 struct dylib_reference { 1109 UNUSED uint32_t isym:24, /* index into the symbol table */ 1110 UNUSED flags:8; /* flags to indicate the type of reference */ 1111 }; 1112 #endif /* 0 */ 1113 1114 /* 1115 * The twolevel_hints_command contains the offset and number of hints in the 1116 * two-level namespace lookup hints table. 1117 */ 1118 struct twolevel_hints_command { 1119 TYP(cmd,4); /* LC_TWOLEVEL_HINTS */ 1120 TYP(cmdsize,4); /* sizeof(struct twolevel_hints_command) */ 1121 TYP(offset,4); /* offset to the hint table */ 1122 TYP(nhints,4); /* number of hints in the hint table */ 1123 }; 1124 1125 /* 1126 * The entries in the two-level namespace lookup hints table are twolevel_hint 1127 * structs. These provide hints to the dynamic link editor where to start 1128 * looking for an undefined symbol in a two-level namespace image. The 1129 * isub_image field is an index into the sub-images (sub-frameworks and 1130 * sub-umbrellas list) that made up the two-level image that the undefined 1131 * symbol was found in when it was built by the static link editor. If 1132 * isub-image is 0 the the symbol is expected to be defined in library and not 1133 * in the sub-images. If isub-image is non-zero it is an index into the array 1134 * of sub-images for the umbrella with the first index in the sub-images being 1135 * 1. The array of sub-images is the ordered list of sub-images of the umbrella 1136 * that would be searched for a symbol that has the umbrella recorded as its 1137 * primary library. The table of contents index is an index into the 1138 * library's table of contents. This is used as the starting point of the 1139 * binary search or a directed linear search. 1140 */ 1141 #if 0 1142 /* Not used by dwarf readers. */ 1143 struct twolevel_hint { 1144 UNUSED uint32_t 1145 isub_image:8, /* index into the sub images */ 1146 itoc:24; /* index into the table of contents */ 1147 }; 1148 #endif 1149 1150 /* 1151 * The prebind_cksum_command contains the value of the original check sum for 1152 * prebound files or zero. When a prebound file is first created or modified 1153 * for other than updating its prebinding information the value of the check sum 1154 * is set to zero. When the file has it prebinding re-done and if the value of 1155 * the check sum is zero the original check sum is calculated and stored in 1156 * cksum field of this load command in the output file. If when the prebinding 1157 * is re-done and the cksum field is non-zero it is left unchanged from the 1158 * input file. 1159 */ 1160 struct prebind_cksum_command { 1161 TYP(cmd,4); /* LC_PREBIND_CKSUM */ 1162 TYP(cmdsize,4); /* sizeof(struct prebind_cksum_command) */ 1163 TYP(cksum,4); /* the check sum or zero */ 1164 }; 1165 1166 /* 1167 * The uuid load command contains a single 128-bit unique random number that 1168 * identifies an object produced by the static link editor. 1169 */ 1170 struct uuid_command { 1171 TYP(cmd,4); /* LC_UUID */ 1172 TYP(cmdsize,4); /* sizeof(struct uuid_command) */ 1173 unsigned char uuid[16]; /* the 128-bit uuid */ 1174 }; 1175 1176 /* 1177 * The rpath_command contains a path which at runtime should be added to 1178 * the current run path used to find @rpath prefixed dylibs. 1179 */ 1180 struct rpath_command { 1181 TYP(cmd,4); /* LC_RPATH */ 1182 TYP(cmdsize,4); /* includes string */ 1183 union lc_str path; /* path to add to run path */ 1184 }; 1185 1186 /* 1187 * The linkedit_data_command contains the offsets and sizes of a blob 1188 * of data in the __LINKEDIT segment. 1189 */ 1190 struct linkedit_data_command { 1191 TYP(cmd,4); /* LC_CODE_SIGNATURE, LC_SEGMENT_SPLIT_INFO, 1192 LC_FUNCTION_STARTS, LC_DATA_IN_CODE, 1193 LC_DYLIB_CODE_SIGN_DRS or 1194 LC_LINKER_OPTIMIZATION_HINT. */ 1195 TYP(cmdsize,4); /* sizeof(struct linkedit_data_command) */ 1196 TYP(dataoff,4); /* file offset of data in __LINKEDIT segment */ 1197 TYP(datasize,4); /* file size of data in __LINKEDIT segment */ 1198 }; 1199 1200 /* 1201 * The encryption_info_command contains the file offset and size of an 1202 * of an encrypted segment. 1203 */ 1204 struct encryption_info_command { 1205 TYP(cmd,4); /* LC_ENCRYPTION_INFO */ 1206 TYP(cmdsize,4); /* sizeof(struct encryption_info_command) */ 1207 TYP(cryptoff,4); /* file offset of encrypted range */ 1208 TYP(cryptsize,4); /* file size of encrypted range */ 1209 TYP(cryptid,4); /* which enryption system, 1210 0 means not-encrypted yet */ 1211 }; 1212 1213 /* 1214 * The encryption_info_command_64 contains the file offset and size of an 1215 * of an encrypted segment (for use in x86_64 targets). 1216 */ 1217 struct encryption_info_command_64 { 1218 TYP(cmd,4); /* LC_ENCRYPTION_INFO_64 */ 1219 TYP(cmdsize,4); /* sizeof(struct encryption_info_command_64) */ 1220 TYP(cryptoff,4); /* file offset of encrypted range */ 1221 TYP(cryptsize,4); /* file size of encrypted range */ 1222 TYP(cryptid,4); /* which enryption system, 1223 0 means not-encrypted yet */ 1224 TYP(pad,4); /* padding to make this struct's size a multiple 1225 of 8 bytes */ 1226 }; 1227 1228 /* 1229 * The version_min_command contains the min OS version on which this 1230 * binary was built to run. 1231 */ 1232 struct version_min_command { 1233 TYP(cmd,4); /* LC_VERSION_MIN_MACOSX or 1234 LC_VERSION_MIN_IPHONEOS or 1235 LC_VERSION_MIN_WATCHOS or 1236 LC_VERSION_MIN_TVOS */ 1237 TYP(cmdsize,4); /* sizeof(struct min_version_command) */ 1238 TYP(version,4); /* X.Y.Z is encoded in nibbles xxxx.yy.zz */ 1239 TYP(sdk,4); /* X.Y.Z is encoded in nibbles xxxx.yy.zz */ 1240 }; 1241 1242 /* 1243 * The dyld_info_command contains the file offsets and sizes of 1244 * the new compressed form of the information dyld needs to 1245 * load the image. This information is used by dyld on Mac OS X 1246 * 10.6 and later. All information pointed to by this command 1247 * is encoded using byte streams, so no endian swapping is needed 1248 * to interpret it. 1249 */ 1250 struct dyld_info_command { 1251 TYP(cmd,4); /* LC_DYLD_INFO or LC_DYLD_INFO_ONLY */ 1252 TYP(cmdsize,4); /* sizeof(struct dyld_info_command) */ 1253 1254 /* 1255 * Dyld rebases an image whenever dyld loads it at an address different 1256 * from its preferred address. The rebase information is a stream 1257 * of byte sized opcodes whose symbolic names start with REBASE_OPCODE_. 1258 * Conceptually the rebase information is a table of tuples: 1259 * <seg-index, seg-offset, type> 1260 * The opcodes are a compressed way to encode the table by only 1261 * encoding when a column changes. In addition simple patterns 1262 * like "every n'th offset for m times" can be encoded in a few 1263 * bytes. 1264 */ 1265 TYP(rebase_off,4); /* file offset to rebase info */ 1266 TYP(rebase_size,4); /* size of rebase info */ 1267 1268 /* 1269 * Dyld binds an image during the loading process, if the image 1270 * requires any pointers to be initialized to symbols in other images. 1271 * The bind information is a stream of byte sized 1272 * opcodes whose symbolic names start with BIND_OPCODE_. 1273 * Conceptually the bind information is a table of tuples: 1274 * <seg-index, seg-offset, type, symbol-library-ordinal, symbol-name, addend> 1275 * The opcodes are a compressed way to encode the table by only 1276 * encoding when a column changes. In addition simple patterns 1277 * like for runs of pointers initialzed to the same value can be 1278 * encoded in a few bytes. 1279 */ 1280 TYP(bind_off,4); /* file offset to binding info */ 1281 TYP(bind_size,4); /* size of binding info */ 1282 1283 /* 1284 * Some C++ programs require dyld to unique symbols so that all 1285 * images in the process use the same copy of some code/data. 1286 * This step is done after binding. The content of the weak_bind 1287 * info is an opcode stream like the bind_info. But it is sorted 1288 * alphabetically by symbol name. This enable dyld to walk 1289 * all images with weak binding information in order and look 1290 * for collisions. If there are no collisions, dyld does 1291 * no updating. That means that some fixups are also encoded 1292 * in the bind_info. For instance, all calls to "operator new" 1293 * are first bound to libstdc++.dylib using the information 1294 * in bind_info. Then if some image overrides operator new 1295 * that is detected when the weak_bind information is processed 1296 * and the call to operator new is then rebound. 1297 */ 1298 TYP(weak_bind_off,4); /* file offset to weak binding info */ 1299 TYP(weak_bind_size,4); /* size of weak binding info */ 1300 1301 /* 1302 * Some uses of external symbols do not need to be bound immediately. 1303 * Instead they can be lazily bound on first use. The lazy_bind 1304 * are contains a stream of BIND opcodes to bind all lazy symbols. 1305 * Normal use is that dyld ignores the lazy_bind section when 1306 * loading an image. Instead the static linker arranged for the 1307 * lazy pointer to initially point to a helper function which 1308 * pushes the offset into the lazy_bind area for the symbol 1309 * needing to be bound, then jumps to dyld which simply adds 1310 * the offset to lazy_bind_off to get the information on what 1311 * to bind. 1312 */ 1313 TYP(lazy_bind_off,4); /* file offset to lazy binding info */ 1314 TYP(lazy_bind_size,4); /* size of lazy binding infs */ 1315 1316 /* 1317 * The symbols exported by a dylib are encoded in a trie. This 1318 * is a compact representation that factors out common prefixes. 1319 * It also reduces LINKEDIT pages in RAM because it encodes all 1320 * information (name, address, flags) in one small, contiguous range. 1321 * The export area is a stream of nodes. The first node sequentially 1322 * is the start node for the trie. 1323 * 1324 * Nodes for a symbol start with a uleb128 that is the length of 1325 * the exported symbol information for the string so far. 1326 * If there is no exported symbol, the node starts with a zero byte. 1327 * If there is exported info, it follows the length. 1328 * 1329 * First is a uleb128 containing flags. Normally, it is followed by 1330 * a uleb128 encoded offset which is location of the content named 1331 * by the symbol from the mach_header for the image. If the flags 1332 * is EXPORT_SYMBOL_FLAGS_REEXPORT, then following the flags is 1333 * a uleb128 encoded library ordinal, then a zero terminated 1334 * UTF8 string. If the string is zero length, then the symbol 1335 * is re-export from the specified dylib with the same name. 1336 * If the flags is EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER, then following 1337 * the flags is two uleb128s: the stub offset and the resolver offset. 1338 * The stub is used by non-lazy pointers. The resolver is used 1339 * by lazy pointers and must be called to get the actual address to use. 1340 * 1341 * After the optional exported symbol information is a byte of 1342 * how many edges (0-255) that this node has leaving it, 1343 * followed by each edge. 1344 * Each edge is a zero terminated UTF8 of the addition chars 1345 * in the symbol, followed by a uleb128 offset for the node that 1346 * edge points to. 1347 * 1348 */ 1349 TYP(export_off,4); /* file offset to lazy binding info */ 1350 TYP(export_size,4); /* size of lazy binding infs */ 1351 }; 1352 1353 /* 1354 * The following are used to encode rebasing information 1355 */ 1356 #define REBASE_TYPE_POINTER 1 1357 #define REBASE_TYPE_TEXT_ABSOLUTE32 2 1358 #define REBASE_TYPE_TEXT_PCREL32 3 1359 1360 #define REBASE_OPCODE_MASK 0xF0 1361 #define REBASE_IMMEDIATE_MASK 0x0F 1362 #define REBASE_OPCODE_DONE 0x00 1363 #define REBASE_OPCODE_SET_TYPE_IMM 0x10 1364 #define REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB 0x20 1365 #define REBASE_OPCODE_ADD_ADDR_ULEB 0x30 1366 #define REBASE_OPCODE_ADD_ADDR_IMM_SCALED 0x40 1367 #define REBASE_OPCODE_DO_REBASE_IMM_TIMES 0x50 1368 #define REBASE_OPCODE_DO_REBASE_ULEB_TIMES 0x60 1369 #define REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB 0x70 1370 #define REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_ULEB 0x80 1371 1372 1373 /* 1374 * The following are used to encode binding information 1375 */ 1376 #define BIND_TYPE_POINTER 1 1377 #define BIND_TYPE_TEXT_ABSOLUTE32 2 1378 #define BIND_TYPE_TEXT_PCREL32 3 1379 1380 #define BIND_SPECIAL_DYLIB_SELF 0 1381 #define BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE -1 1382 #define BIND_SPECIAL_DYLIB_FLAT_LOOKUP -2 1383 1384 #define BIND_SYMBOL_FLAGS_WEAK_IMPORT 0x1 1385 #define BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION 0x8 1386 1387 #define BIND_OPCODE_MASK 0xF0 1388 #define BIND_IMMEDIATE_MASK 0x0F 1389 #define BIND_OPCODE_DONE 0x00 1390 #define BIND_OPCODE_SET_DYLIB_ORDINAL_IMM 0x10 1391 #define BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB 0x20 1392 #define BIND_OPCODE_SET_DYLIB_SPECIAL_IMM 0x30 1393 #define BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM 0x40 1394 #define BIND_OPCODE_SET_TYPE_IMM 0x50 1395 #define BIND_OPCODE_SET_ADDEND_SLEB 0x60 1396 #define BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB 0x70 1397 #define BIND_OPCODE_ADD_ADDR_ULEB 0x80 1398 #define BIND_OPCODE_DO_BIND 0x90 1399 #define BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB 0xA0 1400 #define BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED 0xB0 1401 #define BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB 0xC0 1402 1403 1404 /* 1405 * The following are used on the flags byte of a terminal node 1406 * in the export information. 1407 */ 1408 #define EXPORT_SYMBOL_FLAGS_KIND_MASK 0x03 1409 #define EXPORT_SYMBOL_FLAGS_KIND_REGULAR 0x00 1410 #define EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL 0x01 1411 #define EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION 0x04 1412 #define EXPORT_SYMBOL_FLAGS_REEXPORT 0x08 1413 #define EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER 0x10 1414 1415 /* 1416 * The linker_option_command contains linker options embedded in object files. 1417 */ 1418 struct linker_option_command { 1419 TYP(cmd,4); /* LC_LINKER_OPTION only used in MH_OBJECT filetypes */ 1420 TYP(cmdsize,4); 1421 TYP(count,4); /* number of strings */ 1422 /* concatenation of zero terminated UTF8 strings. 1423 Zero filled at end to align */ 1424 }; 1425 1426 /* 1427 * The symseg_command contains the offset and size of the GNU style 1428 * symbol table information as described in the header file <symseg.h>. 1429 * The symbol roots of the symbol segments must also be aligned properly 1430 * in the file. So the requirement of keeping the offsets aligned to a 1431 * multiple of a 4 bytes translates to the length field of the symbol 1432 * roots also being a multiple of a long. Also the padding must again be 1433 * zeroed. (THIS IS OBSOLETE and no longer supported). 1434 */ 1435 struct symseg_command { 1436 TYP(cmd,4); /* LC_SYMSEG */ 1437 TYP(cmdsize,4); /* sizeof(struct symseg_command) */ 1438 TYP(offset,4); /* symbol segment offset */ 1439 TYP(size,4); /* symbol segment size in bytes */ 1440 }; 1441 1442 /* 1443 * The ident_command contains a free format string table following the 1444 * ident_command structure. The strings are null terminated and the size of 1445 * the command is padded out with zero bytes to a multiple of 4 bytes/ 1446 * (THIS IS OBSOLETE and no longer supported). 1447 */ 1448 struct ident_command { 1449 TYP(cmd,4); /* LC_IDENT */ 1450 TYP(cmdsize,4); /* strings that follow this command */ 1451 }; 1452 1453 /* 1454 * The fvmfile_command contains a reference to a file to be loaded at the 1455 * specified virtual address. (Presently, this command is reserved for 1456 * internal use. The kernel ignores this command when loading a program into 1457 * memory). 1458 */ 1459 struct fvmfile_command { 1460 TYP(cmd,4); /* LC_FVMFILE */ 1461 TYP(cmdsize,4); /* includes pathname string */ 1462 union lc_str name; /* files pathname */ 1463 TYP(header_addr,4); /* files virtual address */ 1464 }; 1465 1466 1467 /* 1468 * The entry_point_command is a replacement for thread_command. 1469 * It is used for main executables to specify the location (file offset) 1470 * of main(). If -stack_size was used at link time, the stacksize 1471 * field will contain the stack size need for the main thread. 1472 */ 1473 struct entry_point_command { 1474 TYP(cmd,4); /* LC_MAIN only used in MH_EXECUTE filetypes */ 1475 TYP(cmdsize,4); /* 24 */ 1476 TYP(entryoff,8); /* file (__TEXT) offset of main() */ 1477 TYP(stacksize,8); /* if not zero, initial stack size */ 1478 }; 1479 1480 1481 /* 1482 * The source_version_command is an optional load command containing 1483 * the version of the sources used to build the binary. 1484 */ 1485 struct source_version_command { 1486 TYP(cmd,4); /* LC_SOURCE_VERSION */ 1487 TYP(cmdsize,4); /* 16 */ 1488 TYP(version,8); /* A.B.C.D.E packed as a24.b10.c10.d10.e10 */ 1489 }; 1490 1491 1492 /* 1493 * The LC_DATA_IN_CODE load commands uses a linkedit_data_command 1494 * to point to an array of data_in_code_entry entries. Each entry 1495 * describes a range of data in a code section. 1496 */ 1497 struct data_in_code_entry { 1498 TYP(offset,4); /* from mach_header to start of data range*/ 1499 TYP(length,2); /* number of bytes in data range */ 1500 TYP(kind,2); /* a DICE_KIND_* value */ 1501 }; 1502 #define DICE_KIND_DATA 0x0001 1503 #define DICE_KIND_JUMP_TABLE8 0x0002 1504 #define DICE_KIND_JUMP_TABLE16 0x0003 1505 #define DICE_KIND_JUMP_TABLE32 0x0004 1506 #define DICE_KIND_ABS_JUMP_TABLE32 0x0005 1507 1508 1509 1510 /* 1511 * Sections of type S_THREAD_LOCAL_VARIABLES contain an array 1512 * of tlv_descriptor structures. 1513 */ 1514 struct tlv_descriptor 1515 { 1516 void* (*thunk)(struct tlv_descriptor*); 1517 unsigned long key; 1518 unsigned long offset; 1519 }; 1520 #ifdef __cplusplus 1521 } 1522 #endif /* __cplusplus */ 1523 #endif /* MACHO_LOADER_H */ 1524