EXTERNAL_HEADERS/mach-o/loader.h

   1 /*
   2  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * Copyright (c) 1999-2003 Apple Computer, Inc.  All Rights Reserved.
   7  *
   8  * This file contains Original Code and/or Modifications of Original Code
   9  * as defined in and that are subject to the Apple Public Source License
  10  * Version 2.0 (the 'License'). You may not use this file except in
  11  * compliance with the License. Please obtain a copy of the License at
  12  * http://www.opensource.apple.com/apsl/ and read it before using this
  13  * file.
  14  *
  15  * The Original Code and all software distributed under the License are
  16  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  17  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  18  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  19  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  20  * Please see the License for the specific language governing rights and
  21  * limitations under the License.
  22  *
  23  * @APPLE_LICENSE_HEADER_END@
  24  */
  25 #ifndef _MACHO_LOADER_H_
  26 #define _MACHO_LOADER_H_
  27
  28 /*
  29  * This file describes the format of mach object files.
  30  */
  31
  32 /*
  33  * <mach/machine.h> is needed here for the cpu_type_t and cpu_subtype_t types
  34  * and contains the constants for the possible values of these types.
  35  */
  36 #include <mach/machine.h>
  37
  38 /*
  39  * <mach/vm_prot.h> is needed here for the vm_prot_t type and contains the
  40  * constants that are or'ed together for the possible values of this type.
  41  */
  42 #include <mach/vm_prot.h>
  43
  44 /*
  45  * <machine/thread_status.h> is expected to define the flavors of the thread
  46  * states and the structures of those flavors for each machine.
  47  */
  48 #include <mach/machine/thread_status.h>
  49 #include <architecture/byte_order.h>
  50
  51 /*
  52  * The mach header appears at the very beginning of the object file.
  53  */
  54 struct mach_header {
  55         unsigned long   magic;          /* mach magic number identifier */
  56         cpu_type_t      cputype;        /* cpu specifier */
  57         cpu_subtype_t   cpusubtype;     /* machine specifier */
  58         unsigned long   filetype;       /* type of file */
  59         unsigned long   ncmds;          /* number of load commands */
  60         unsigned long   sizeofcmds;     /* the size of all the load commands */
  61         unsigned long   flags;          /* flags */
  62 };
  63
  64 /* Constant for the magic field of the mach_header */
  65 #define MH_MAGIC        0xfeedface      /* the mach magic number */
  66 #define MH_CIGAM        NXSwapInt(MH_MAGIC)
  67
  68 /*
  69  * The layout of the file depends on the filetype.  For all but the MH_OBJECT
  70  * file type the segments are padded out and aligned on a segment alignment
  71  * boundary for efficient demand pageing.  The MH_EXECUTE, MH_FVMLIB, MH_DYLIB,
  72  * MH_DYLINKER and MH_BUNDLE file types also have the headers included as part
  73  * of their first segment.
  74  *
  75  * The file type MH_OBJECT is a compact format intended as output of the
  76  * assembler and input (and possibly output) of the link editor (the .o
  77  * format).  All sections are in one unnamed segment with no segment padding.
  78  * This format is used as an executable format when the file is so small the
  79  * segment padding greatly increases it's size.
  80  *
  81  * The file type MH_PRELOAD is an executable format intended for things that
  82  * not executed under the kernel (proms, stand alones, kernels, etc).  The
  83  * format can be executed under the kernel but may demand paged it and not
  84  * preload it before execution.
  85  *
  86  * A core file is in MH_CORE format and can be any in an arbritray legal
  87  * Mach-O file.
  88  *
  89  * Constants for the filetype field of the mach_header
  90  */
  91 #define MH_OBJECT       0x1             /* relocatable object file */
  92 #define MH_EXECUTE      0x2             /* demand paged executable file */
  93 #define MH_FVMLIB       0x3             /* fixed VM shared library file */
  94 #define MH_CORE         0x4             /* core file */
  95 #define MH_PRELOAD      0x5             /* preloaded executable file */
  96 #define MH_DYLIB        0x6             /* dynamicly bound shared library file*/
  97 #define MH_DYLINKER     0x7             /* dynamic link editor */
  98 #define MH_BUNDLE       0x8             /* dynamicly bound bundle file */
  99
 100 /* Constants for the flags field of the mach_header */
 101 #define MH_NOUNDEFS     0x1             /* the object file has no undefined
 102                                            references, can be executed */
 103 #define MH_INCRLINK     0x2             /* the object file is the output of an
 104                                            incremental link against a base file
 105                                            and can't be link edited again */
 106 #define MH_DYLDLINK     0x4             /* the object file is input for the
 107                                            dynamic linker and can't be staticly
 108                                            link edited again */
 109 #define MH_BINDATLOAD   0x8             /* the object file's undefined
 110                                            references are bound by the dynamic
 111                                            linker when loaded. */
 112 #define MH_PREBOUND     0x10            /* the file has it's dynamic undefined
 113                                            references prebound. */
 114
 115 /*
 116  * The load commands directly follow the mach_header.  The total size of all
 117  * of the commands is given by the sizeofcmds field in the mach_header.  All
 118  * load commands must have as their first two fields cmd and cmdsize.  The cmd
 119  * field is filled in with a constant for that command type.  Each command type
 120  * has a structure specifically for it.  The cmdsize field is the size in bytes
 121  * of the particular load command structure plus anything that follows it that
 122  * is a part of the load command (i.e. section structures, strings, etc.).  To
 123  * advance to the next load command the cmdsize can be added to the offset or
 124  * pointer of the current load command.  The cmdsize MUST be a multiple of
 125  * sizeof(long) (this is forever the maximum alignment of any load commands).
 126  * The padded bytes must be zero.  All tables in the object file must also
 127  * follow these rules so the file can be memory mapped.  Otherwise the pointers
 128  * to these tables will not work well or at all on some machines.  With all
 129  * padding zeroed like objects will compare byte for byte.
 130  */
 131 struct load_command {
 132         unsigned long cmd;              /* type of load command */
 133         unsigned long cmdsize;          /* total size of command in bytes */
 134 };
 135
 136 /* Constants for the cmd field of all load commands, the type */
 137 #define LC_SEGMENT      0x1     /* segment of this file to be mapped */
 138 #define LC_SYMTAB       0x2     /* link-edit stab symbol table info */
 139 #define LC_SYMSEG       0x3     /* link-edit gdb symbol table info (obsolete) */
 140 #define LC_THREAD       0x4     /* thread */
 141 #define LC_UNIXTHREAD   0x5     /* unix thread (includes a stack) */
 142 #define LC_LOADFVMLIB   0x6     /* load a specified fixed VM shared library */
 143 #define LC_IDFVMLIB     0x7     /* fixed VM shared library identification */
 144 #define LC_IDENT        0x8     /* object identification info (obsolete) */
 145 #define LC_FVMFILE      0x9     /* fixed VM file inclusion (internal use) */
 146 #define LC_PREPAGE      0xa     /* prepage command (internal use) */
 147 #define LC_DYSYMTAB     0xb     /* dynamic link-edit symbol table info */
 148 #define LC_LOAD_DYLIB   0xc     /* load a dynamicly linked shared library */
 149 #define LC_ID_DYLIB     0xd     /* dynamicly linked shared lib identification */
 150 #define LC_LOAD_DYLINKER 0xe    /* load a dynamic linker */
 151 #define LC_ID_DYLINKER  0xf     /* dynamic linker identification */
 152 #define LC_PREBOUND_DYLIB 0x10  /* modules prebound for a dynamicly */
 153                                 /*  linked shared library */
 154
 155 /*
 156  * A variable length string in a load command is represented by an lc_str
 157  * union.  The strings are stored just after the load command structure and
 158  * the offset is from the start of the load command structure.  The size
 159  * of the string is reflected in the cmdsize field of the load command.
 160  * Once again any padded bytes to bring the cmdsize field to a multiple
 161  * of sizeof(long) must be zero.
 162  */
 163 union lc_str {
 164         unsigned long   offset; /* offset to the string */
 165         char            *ptr;   /* pointer to the string */
 166 };
 167
 168 /*
 169  * The segment load command indicates that a part of this file is to be
 170  * mapped into the task's address space.  The size of this segment in memory,
 171  * vmsize, maybe equal to or larger than the amount to map from this file,
 172  * filesize.  The file is mapped starting at fileoff to the beginning of
 173  * the segment in memory, vmaddr.  The rest of the memory of the segment,
 174  * if any, is allocated zero fill on demand.  The segment's maximum virtual
 175  * memory protection and initial virtual memory protection are specified
 176  * by the maxprot and initprot fields.  If the segment has sections then the
 177  * section structures directly follow the segment command and their size is
 178  * reflected in cmdsize.
 179  */
 180 struct segment_command {
 181         unsigned long   cmd;            /* LC_SEGMENT */
 182         unsigned long   cmdsize;        /* includes sizeof section structs */
 183         char            segname[16];    /* segment name */
 184         unsigned long   vmaddr;         /* memory address of this segment */
 185         unsigned long   vmsize;         /* memory size of this segment */
 186         unsigned long   fileoff;        /* file offset of this segment */
 187         unsigned long   filesize;       /* amount to map from the file */
 188         vm_prot_t       maxprot;        /* maximum VM protection */
 189         vm_prot_t       initprot;       /* initial VM protection */
 190         unsigned long   nsects;         /* number of sections in segment */
 191         unsigned long   flags;          /* flags */
 192 };
 193
 194 /* Constants for the flags field of the segment_command */
 195 #define SG_HIGHVM       0x1     /* the file contents for this segment is for
 196                                    the high part of the VM space, the low part
 197                                    is zero filled (for stacks in core files) */
 198 #define SG_FVMLIB       0x2     /* this segment is the VM that is allocated by
 199                                    a fixed VM library, for overlap checking in
 200                                    the link editor */
 201 #define SG_NORELOC      0x4     /* this segment has nothing that was relocated
 202                                    in it and nothing relocated to it, that is
 203                                    it maybe safely replaced without relocation*/
 204
 205 /*
 206  * A segment is made up of zero or more sections.  Non-MH_OBJECT files have
 207  * all of their segments with the proper sections in each, and padded to the
 208  * specified segment alignment when produced by the link editor.  The first
 209  * segment of a MH_EXECUTE and MH_FVMLIB format file contains the mach_header
 210  * and load commands of the object file before it's first section.  The zero
 211  * fill sections are always last in their segment (in all formats).  This
 212  * allows the zeroed segment padding to be mapped into memory where zero fill
 213  * sections might be.
 214  *
 215  * The MH_OBJECT format has all of it's sections in one segment for
 216  * compactness.  There is no padding to a specified segment boundary and the
 217  * mach_header and load commands are not part of the segment.
 218  *
 219  * Sections with the same section name, sectname, going into the same segment,
 220  * segname, are combined by the link editor.  The resulting section is aligned
 221  * to the maximum alignment of the combined sections and is the new section's
 222  * alignment.  The combined sections are aligned to their original alignment in
 223  * the combined section.  Any padded bytes to get the specified alignment are
 224  * zeroed.
 225  *
 226  * The format of the relocation entries referenced by the reloff and nreloc
 227  * fields of the section structure for mach object files is described in the
 228  * header file <reloc.h>.
 229  */
 230 struct section {
 231         char            sectname[16];   /* name of this section */
 232         char            segname[16];    /* segment this section goes in */
 233         unsigned long   addr;           /* memory address of this section */
 234         unsigned long   size;           /* size in bytes of this section */
 235         unsigned long   offset;         /* file offset of this section */
 236         unsigned long   align;          /* section alignment (power of 2) */
 237         unsigned long   reloff;         /* file offset of relocation entries */
 238         unsigned long   nreloc;         /* number of relocation entries */
 239         unsigned long   flags;          /* flags (section type and attributes)*/
 240         unsigned long   reserved1;      /* reserved */
 241         unsigned long   reserved2;      /* reserved */
 242 };
 243
 244 /*
 245  * The flags field of a section structure is separated into two parts a section
 246  * type and section attributes.  The section types are mutually exclusive (it
 247  * can only have one type) but the section attributes are not (it may have more
 248  * than one attribute).
 249  */
 250 #define SECTION_TYPE             0x000000ff     /* 256 section types */
 251 #define SECTION_ATTRIBUTES       0xffffff00     /*  24 section attributes */
 252
 253 /* Constants for the type of a section */
 254 #define S_REGULAR               0x0     /* regular section */
 255 #define S_ZEROFILL              0x1     /* zero fill on demand section */
 256 #define S_CSTRING_LITERALS      0x2     /* section with only literal C strings*/
 257 #define S_4BYTE_LITERALS        0x3     /* section with only 4 byte literals */
 258 #define S_8BYTE_LITERALS        0x4     /* section with only 8 byte literals */
 259 #define S_LITERAL_POINTERS      0x5     /* section with only pointers to */
 260                                         /*  literals */
 261 /*
 262  * For the two types of symbol pointers sections and the symbol stubs section
 263  * they have indirect symbol table entries.  For each of the entries in the
 264  * section the indirect symbol table entries, in corresponding order in the
 265  * indirect symbol table, start at the index stored in the reserved1 field
 266  * of the section structure.  Since the indirect symbol table entries
 267  * correspond to the entries in the section the number of indirect symbol table
 268  * entries is inferred from the size of the section divided by the size of the
 269  * entries in the section.  For symbol pointers sections the size of the entries
 270  * in the section is 4 bytes and for symbol stubs sections the byte size of the
 271  * stubs is stored in the reserved2 field of the section structure.
 272  */
 273 #define S_NON_LAZY_SYMBOL_POINTERS      0x6     /* section with only non-lazy
 274                                                    symbol pointers */
 275 #define S_LAZY_SYMBOL_POINTERS          0x7     /* section with only lazy symbol
 276                                                    pointers */
 277 #define S_SYMBOL_STUBS                  0x8     /* section with only symbol
 278                                                    stubs, byte size of stub in
 279                                                    the reserved2 field */
 280 #define S_MOD_INIT_FUNC_POINTERS        0x9     /* section with only function
 281                                                    pointers for initialization*/
 282 /*
 283  * Constants for the section attributes part of the flags field of a section
 284  * structure.
 285  */
 286 #define SECTION_ATTRIBUTES_USR   0xff000000     /* User setable attributes */
 287 #define S_ATTR_PURE_INSTRUCTIONS 0x80000000     /* section contains only true
 288                                                    machine instructions */
 289 #define SECTION_ATTRIBUTES_SYS   0x00ffff00     /* system setable attributes */
 290 #define S_ATTR_SOME_INSTRUCTIONS 0x00000400     /* section contains some
 291                                                    machine instructions */
 292 #define S_ATTR_EXT_RELOC         0x00000200     /* section has external
 293                                                    relocation entries */
 294 #define S_ATTR_LOC_RELOC         0x00000100     /* section has local
 295                                                    relocation entries */
 296
 297
 298 /*
 299  * The names of segments and sections in them are mostly meaningless to the
 300  * link-editor.  But there are few things to support traditional UNIX
 301  * executables that require the link-editor and assembler to use some names
 302  * agreed upon by convention.
 303  *
 304  * The initial protection of the "__TEXT" segment has write protection turned
 305  * off (not writeable).
 306  *
 307  * The link-editor will allocate common symbols at the end of the "__common"
 308  * section in the "__DATA" segment.  It will create the section and segment
 309  * if needed.
 310  */
 311
 312 /* The currently known segment names and the section names in those segments */
 313
 314 #define SEG_PAGEZERO    "__PAGEZERO"    /* the pagezero segment which has no */
 315                                         /* protections and catches NULL */
 316                                         /* references for MH_EXECUTE files */
 317
 318
 319 #define SEG_TEXT        "__TEXT"        /* the tradition UNIX text segment */
 320 #define SECT_TEXT       "__text"        /* the real text part of the text */
 321                                         /* section no headers, and no padding */
 322 #define SECT_FVMLIB_INIT0 "__fvmlib_init0"      /* the fvmlib initialization */
 323                                                 /*  section */
 324 #define SECT_FVMLIB_INIT1 "__fvmlib_init1"      /* the section following the */
 325                                                 /*  fvmlib initialization */
 326                                                 /*  section */
 327
 328 #define SEG_DATA        "__DATA"        /* the tradition UNIX data segment */
 329 #define SECT_DATA       "__data"        /* the real initialized data section */
 330                                         /* no padding, no bss overlap */
 331 #define SECT_BSS        "__bss"         /* the real uninitialized data section*/
 332                                         /* no padding */
 333 #define SECT_COMMON     "__common"      /* the section common symbols are */
 334                                         /* allocated in by the link editor */
 335
 336 #define SEG_OBJC        "__OBJC"        /* objective-C runtime segment */
 337 #define SECT_OBJC_SYMBOLS "__symbol_table"      /* symbol table */
 338 #define SECT_OBJC_MODULES "__module_info"       /* module information */
 339 #define SECT_OBJC_STRINGS "__selector_strs"     /* string table */
 340 #define SECT_OBJC_REFS "__selector_refs"        /* string table */
 341
 342 #define SEG_ICON         "__ICON"       /* the NeXT icon segment */
 343 #define SECT_ICON_HEADER "__header"     /* the icon headers */
 344 #define SECT_ICON_TIFF   "__tiff"       /* the icons in tiff format */
 345
 346 #define SEG_LINKEDIT    "__LINKEDIT"    /* the segment containing all structs */
 347                                         /* created and maintained by the link */
 348                                         /* editor.  Created with -seglinkedit */
 349                                         /* option to ld(1) for MH_EXECUTE and */
 350                                         /* FVMLIB file types only */
 351
 352 #define SEG_UNIXSTACK   "__UNIXSTACK"   /* the unix stack segment */
 353
 354 /*
 355  * Fixed virtual memory shared libraries are identified by two things.  The
 356  * target pathname (the name of the library as found for execution), and the
 357  * minor version number.  The address of where the headers are loaded is in
 358  * header_addr.
 359  */
 360 struct fvmlib {
 361         union lc_str    name;           /* library's target pathname */
 362         unsigned long   minor_version;  /* library's minor version number */
 363         unsigned long   header_addr;    /* library's header address */
 364 };
 365
 366 /*
 367  * A fixed virtual shared library (filetype == MH_FVMLIB in the mach header)
 368  * contains a fvmlib_command (cmd == LC_IDFVMLIB) to identify the library.
 369  * An object that uses a fixed virtual shared library also contains a
 370  * fvmlib_command (cmd == LC_LOADFVMLIB) for each library it uses.
 371  */
 372 struct fvmlib_command {
 373         unsigned long   cmd;            /* LC_IDFVMLIB or LC_LOADFVMLIB */
 374         unsigned long   cmdsize;        /* includes pathname string */
 375         struct fvmlib   fvmlib;         /* the library identification */
 376 };
 377
 378 /*
 379  * Dynamicly linked shared libraries are identified by two things.  The
 380  * pathname (the name of the library as found for execution), and the
 381  * compatibility version number.  The pathname must match and the compatibility
 382  * number in the user of the library must be greater than or equal to the
 383  * library being used.  The time stamp is used to record the time a library was
 384  * built and copied into user so it can be use to determined if the library used
 385  * at runtime is exactly the same as used to built the program.
 386  */
 387 struct dylib {
 388     union lc_str  name;                 /* library's path name */
 389     unsigned long timestamp;            /* library's build time stamp */
 390     unsigned long current_version;      /* library's current version number */
 391     unsigned long compatibility_version;/* library's compatibility vers number*/
 392 };
 393
 394 /*
 395  * A dynamicly linked shared library (filetype == MH_DYLIB in the mach header)
 396  * contains a dylib_command (cmd == LC_ID_DYLIB) to identify the library.
 397  * An object that uses a dynamicly linked shared library also contains a
 398  * dylib_command (cmd == LC_LOAD_DYLIB) for each library it uses.
 399  */
 400 struct dylib_command {
 401         unsigned long   cmd;            /* LC_ID_DYLIB or LC_LOAD_DYLIB */
 402         unsigned long   cmdsize;        /* includes pathname string */
 403         struct dylib    dylib;          /* the library identification */
 404 };
 405
 406 /*
 407  * A program (filetype == MH_EXECUTE) or bundle (filetype == MH_BUNDLE) that is
 408  * prebound to it's dynamic libraries has one of these for each library that
 409  * the static linker used in prebinding.  It contains a bit vector for the
 410  * modules in the library.  The bits indicate which modules are bound (1) and
 411  * which are not (0) from the library.  The bit for module 0 is the low bit
 412  * of the first byte.  So the bit for the Nth module is:
 413  * (linked_modules[N/8] >> N%8) & 1
 414  */
 415 struct prebound_dylib_command {
 416         unsigned long   cmd;            /* LC_PREBOUND_DYLIB */
 417         unsigned long   cmdsize;        /* includes strings */
 418         union lc_str    name;           /* library's path name */
 419         unsigned long   nmodules;       /* number of modules in library */
 420         union lc_str    linked_modules; /* bit vector of linked modules */
 421 };
 422
 423 /*
 424  * A program that uses a dynamic linker contains a dylinker_command to identify
 425  * the name of the dynamic linker (LC_LOAD_DYLINKER).  And a dynamic linker
 426  * contains a dylinker_command to identify the dynamic linker (LC_ID_DYLINKER).
 427  * A file can have at most one of these.
 428  */
 429 struct dylinker_command {
 430         unsigned long   cmd;            /* LC_ID_DYLINKER or LC_LOAD_DYLINKER */
 431         unsigned long   cmdsize;        /* includes pathname string */
 432         union lc_str    name;           /* dynamic linker's path name */
 433 };
 434
 435 /*
 436  * Thread commands contain machine-specific data structures suitable for
 437  * use in the thread state primitives.  The machine specific data structures
 438  * follow the struct thread_command as follows.
 439  * Each flavor of machine specific data structure is preceded by an unsigned
 440  * long constant for the flavor of that data structure, an unsigned long
 441  * that is the count of longs of the size of the state data structure and then
 442  * the state data structure follows.  This triple may be repeated for many
 443  * flavors.  The constants for the flavors, counts and state data structure
 444  * definitions are expected to be in the header file <machine/thread_status.h>.
 445  * These machine specific data structures sizes must be multiples of
 446  * sizeof(long).  The cmdsize reflects the total size of the thread_command
 447  * and all of the sizes of the constants for the flavors, counts and state
 448  * data structures.
 449  *
 450  * For executable objects that are unix processes there will be one
 451  * thread_command (cmd == LC_UNIXTHREAD) created for it by the link-editor.
 452  * This is the same as a LC_THREAD, except that a stack is automatically
 453  * created (based on the shell's limit for the stack size).  Command arguments
 454  * and environment variables are copied onto that stack.
 455  */
 456 struct thread_command {
 457         unsigned long   cmd;            /* LC_THREAD or  LC_UNIXTHREAD */
 458         unsigned long   cmdsize;        /* total size of this command */
 459         /* unsigned long flavor            flavor of thread state */
 460         /* unsigned long count             count of longs in thread state */
 461         /* struct XXX_thread_state state   thread state for this flavor */
 462         /* ... */
 463 };
 464
 465 /*
 466  * The symtab_command contains the offsets and sizes of the link-edit 4.3BSD
 467  * "stab" style symbol table information as described in the header files
 468  * <nlist.h> and <stab.h>.
 469  */
 470 struct symtab_command {
 471         unsigned long   cmd;            /* LC_SYMTAB */
 472         unsigned long   cmdsize;        /* sizeof(struct symtab_command) */
 473         unsigned long   symoff;         /* symbol table offset */
 474         unsigned long   nsyms;          /* number of symbol table entries */
 475         unsigned long   stroff;         /* string table offset */
 476         unsigned long   strsize;        /* string table size in bytes */
 477 };
 478
 479 /*
 480  * This is the second set of the symbolic information which is used to support
 481  * the data structures for the dynamicly link editor.
 482  *
 483  * The original set of symbolic information in the symtab_command which contains
 484  * the symbol and string tables must also be present when this load command is
 485  * present.  When this load command is present the symbol table is organized
 486  * into three groups of symbols:
 487  *      local symbols (static and debugging symbols) - grouped by module
 488  *      defined external symbols - grouped by module (sorted by name if not lib)
 489  *      undefined external symbols (sorted by name)
 490  * In this load command there are offsets and counts to each of the three groups
 491  * of symbols.
 492  *
 493  * This load command contains a the offsets and sizes of the following new
 494  * symbolic information tables:
 495  *      table of contents
 496  *      module table
 497  *      reference symbol table
 498  *      indirect symbol table
 499  * The first three tables above (the table of contents, module table and
 500  * reference symbol table) are only present if the file is a dynamicly linked
 501  * shared library.  For executable and object modules, which are files
 502  * containing only one module, the information that would be in these three
 503  * tables is determined as follows:
 504  *      table of contents - the defined external symbols are sorted by name
 505  *      module table - the file contains only one module so everything in the
 506  *                     file is part of the module.
 507  *      reference symbol table - is the defined and undefined external symbols
 508  *
 509  * For dynamicly linked shared library files this load command also contains
 510  * offsets and sizes to the pool of relocation entries for all sections
 511  * separated into two groups:
 512  *      external relocation entries
 513  *      local relocation entries
 514  * For executable and object modules the relocation entries continue to hang
 515  * off the section structures.
 516  */
 517 struct dysymtab_command {
 518     unsigned long cmd;          /* LC_DYSYMTAB */
 519     unsigned long cmdsize;      /* sizeof(struct dysymtab_command) */
 520
 521     /*
 522      * The symbols indicated by symoff and nsyms of the LC_SYMTAB load command
 523      * are grouped into the following three groups:
 524      *    local symbols (further grouped by the module they are from)
 525      *    defined external symbols (further grouped by the module they are from)
 526      *    undefined symbols
 527      *
 528      * The local symbols are used only for debugging.  The dynamic binding
 529      * process may have to use them to indicate to the debugger the local
 530      * symbols for a module that is being bound.
 531      *
 532      * The last two groups are used by the dynamic binding process to do the
 533      * binding (indirectly through the module table and the reference symbol
 534      * table when this is a dynamicly linked shared library file).
 535      */
 536     unsigned long ilocalsym;    /* index to local symbols */
 537     unsigned long nlocalsym;    /* number of local symbols */
 538
 539     unsigned long iextdefsym;   /* index to externally defined symbols */
 540     unsigned long nextdefsym;   /* number of externally defined symbols */
 541
 542     unsigned long iundefsym;    /* index to undefined symbols */
 543     unsigned long nundefsym;    /* number of undefined symbols */
 544
 545     /*
 546      * For the for the dynamic binding process to find which module a symbol
 547      * is defined in the table of contents is used (analogous to the ranlib
 548      * structure in an archive) which maps defined external symbols to modules
 549      * they are defined in.  This exists only in a dynamicly linked shared
 550      * library file.  For executable and object modules the defined external
 551      * symbols are sorted by name and is use as the table of contents.
 552      */
 553     unsigned long tocoff;       /* file offset to table of contents */
 554     unsigned long ntoc;         /* number of entries in table of contents */
 555
 556     /*
 557      * To support dynamic binding of "modules" (whole object files) the symbol
 558      * table must reflect the modules that the file was created from.  This is
 559      * done by having a module table that has indexes and counts into the merged
 560      * tables for each module.  The module structure that these two entries
 561      * refer to is described below.  This exists only in a dynamicly linked
 562      * shared library file.  For executable and object modules the file only
 563      * contains one module so everything in the file belongs to the module.
 564      */
 565     unsigned long modtaboff;    /* file offset to module table */
 566     unsigned long nmodtab;      /* number of module table entries */
 567
 568     /*
 569      * To support dynamic module binding the module structure for each module
 570      * indicates the external references (defined and undefined) each module
 571      * makes.  For each module there is an offset and a count into the
 572      * reference symbol table for the symbols that the module references.
 573      * This exists only in a dynamicly linked shared library file.  For
 574      * executable and object modules the defined external symbols and the
 575      * undefined external symbols indicates the external references.
 576      */
 577     unsigned long extrefsymoff;  /* offset to referenced symbol table */
 578     unsigned long nextrefsyms;   /* number of referenced symbol table entries */
 579
 580     /*
 581      * The sections that contain "symbol pointers" and "routine stubs" have
 582      * indexes and (implied counts based on the size of the section and fixed
 583      * size of the entry) into the "indirect symbol" table for each pointer
 584      * and stub.  For every section of these two types the index into the
 585      * indirect symbol table is stored in the section header in the field
 586      * reserved1.  An indirect symbol table entry is simply a 32bit index into
 587      * the symbol table to the symbol that the pointer or stub is referring to.
 588      * The indirect symbol table is ordered to match the entries in the section.
 589      */
 590     unsigned long indirectsymoff; /* file offset to the indirect symbol table */
 591     unsigned long nindirectsyms;  /* number of indirect symbol table entries */
 592
 593     /*
 594      * To support relocating an individual module in a library file quickly the
 595      * external relocation entries for each module in the library need to be
 596      * accessed efficiently.  Since the relocation entries can't be accessed
 597      * through the section headers for a library file they are separated into
 598      * groups of local and external entries further grouped by module.  In this
 599      * case the presents of this load command who's extreloff, nextrel,
 600      * locreloff and nlocrel fields are non-zero indicates that the relocation
 601      * entries of non-merged sections are not referenced through the section
 602      * structures (and the reloff and nreloc fields in the section headers are
 603      * set to zero).
 604      *
 605      * Since the relocation entries are not accessed through the section headers
 606      * this requires the r_address field to be something other than a section
 607      * offset to identify the item to be relocated.  In this case r_address is
 608      * set to the offset from the vmaddr of the first LC_SEGMENT command.
 609      *
 610      * The relocation entries are grouped by module and the module table
 611      * entries have indexes and counts into them for the group of external
 612      * relocation entries for that the module.
 613      *
 614      * For sections that are merged across modules there must not be any
 615      * remaining external relocation entries for them (for merged sections
 616      * remaining relocation entries must be local).
 617      */
 618     unsigned long extreloff;    /* offset to external relocation entries */
 619     unsigned long nextrel;      /* number of external relocation entries */
 620
 621     /*
 622      * All the local relocation entries are grouped together (they are not
 623      * grouped by their module since they are only used if the object is moved
 624      * from it staticly link edited address).
 625      */
 626     unsigned long locreloff;    /* offset to local relocation entries */
 627     unsigned long nlocrel;      /* number of local relocation entries */
 628
 629 };
 630
 631 /*
 632  * An indirect symbol table entry is simply a 32bit index into the symbol table
 633  * to the symbol that the pointer or stub is refering to.  Unless it is for a
 634  * non-lazy symbol pointer section for a defined symbol which strip(1) as
 635  * removed.  In which case it has the value INDIRECT_SYMBOL_LOCAL.  If the
 636  * symbol was also absolute INDIRECT_SYMBOL_ABS is or'ed with that.
 637  */
 638 #define INDIRECT_SYMBOL_LOCAL   0x80000000
 639 #define INDIRECT_SYMBOL_ABS     0x40000000
 640
 641
 642 /* a table of contents entry */
 643 struct dylib_table_of_contents {
 644     unsigned long symbol_index; /* the defined external symbol
 645                                    (index into the symbol table) */
 646     unsigned long module_index; /* index into the module table this symbol
 647                                    is defined in */
 648 };
 649
 650 /* a module table entry */
 651 struct dylib_module {
 652     unsigned long module_name;  /* the module name (index into string table) */
 653
 654     unsigned long iextdefsym;   /* index into externally defined symbols */
 655     unsigned long nextdefsym;   /* number of externally defined symbols */
 656     unsigned long irefsym;              /* index into reference symbol table */
 657     unsigned long nrefsym;      /* number of reference symbol table entries */
 658     unsigned long ilocalsym;    /* index into symbols for local symbols */
 659     unsigned long nlocalsym;    /* number of local symbols */
 660
 661     unsigned long iextrel;      /* index into external relocation entries */
 662     unsigned long nextrel;      /* number of external relocation entries */
 663
 664     unsigned long iinit;        /* index into the init section */
 665     unsigned long ninit;        /* number of init section entries */
 666
 667     unsigned long               /* for this module address of the start of */
 668         objc_module_info_addr;  /*  the (__OBJC,__module_info) section */
 669     unsigned long               /* for this module size of */
 670         objc_module_info_size;  /*  the (__OBJC,__module_info) section */
 671 };
 672
 673 /*
 674  * The entries in the reference symbol table are used when loading the module
 675  * (both by the static and dynamic link editors) and if the module is unloaded
 676  * or replaced.  Therefore all external symbols (defined and undefined) are
 677  * listed in the module's reference table.  The flags describe the type of
 678  * reference that is being made.  The constants for the flags are defined in
 679  * <mach-o/nlist.h> as they are also used for symbol table entries.
 680  */
 681 struct dylib_reference {
 682     unsigned long isym:24,      /* index into the symbol table */
 683                   flags:8;      /* flags to indicate the type of reference */
 684 };
 685
 686 /*
 687  * The symseg_command contains the offset and size of the GNU style
 688  * symbol table information as described in the header file <symseg.h>.
 689  * The symbol roots of the symbol segments must also be aligned properly
 690  * in the file.  So the requirement of keeping the offsets aligned to a
 691  * multiple of a sizeof(long) translates to the length field of the symbol
 692  * roots also being a multiple of a long.  Also the padding must again be
 693  * zeroed. (THIS IS OBSOLETE and no longer supported).
 694  */
 695 struct symseg_command {
 696         unsigned long   cmd;            /* LC_SYMSEG */
 697         unsigned long   cmdsize;        /* sizeof(struct symseg_command) */
 698         unsigned long   offset;         /* symbol segment offset */
 699         unsigned long   size;           /* symbol segment size in bytes */
 700 };
 701
 702 /*
 703  * The ident_command contains a free format string table following the
 704  * ident_command structure.  The strings are null terminated and the size of
 705  * the command is padded out with zero bytes to a multiple of sizeof(long).
 706  * (THIS IS OBSOLETE and no longer supported).
 707  */
 708 struct ident_command {
 709         unsigned long cmd;              /* LC_IDENT */
 710         unsigned long cmdsize;          /* strings that follow this command */
 711 };
 712
 713 /*
 714  * The fvmfile_command contains a reference to a file to be loaded at the
 715  * specified virtual address.  (Presently, this command is reserved for NeXT
 716  * internal use.  The kernel ignores this command when loading a program into
 717  * memory).
 718  */
 719 struct fvmfile_command {
 720         unsigned long cmd;              /* LC_FVMFILE */
 721         unsigned long cmdsize;          /* includes pathname string */
 722         union lc_str    name;           /* files pathname */
 723         unsigned long   header_addr;    /* files virtual address */
 724 };
 725
 726 #endif _MACHO_LOADER_H_