libkern/mach-o/loader.h

   1 /*
   2  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * The contents of this file constitute Original Code as defined in and
   7  * are subject to the Apple Public Source License Version 1.1 (the
   8  * "License").  You may not use this file except in compliance with the
   9  * License.  Please obtain a copy of the License at
  10  * http://www.apple.com/publicsource and read it before using this file.
  11  *
  12  * This Original Code and all software distributed under the License are
  13  * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  14  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  15  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  16  * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
  17  * License for the specific language governing rights and limitations
  18  * under the License.
  19  *
  20  * @APPLE_LICENSE_HEADER_END@
  21  */
  22 #ifndef _MACHO_LOADER_H_
  23 #define _MACHO_LOADER_H_
  24
  25 /*
  26  * This file describes the format of mach object files.
  27  */
  28
  29 /*
  30  * <mach/machine.h> is needed here for the cpu_type_t and cpu_subtype_t types
  31  * and contains the constants for the possible values of these types.
  32  */
  33 #include <mach/machine.h>
  34
  35 /*
  36  * <mach/vm_prot.h> is needed here for the vm_prot_t type and contains the
  37  * constants that are or'ed together for the possible values of this type.
  38  */
  39 #include <mach/vm_prot.h>
  40
  41 /*
  42  * <machine/thread_status.h> is expected to define the flavors of the thread
  43  * states and the structures of those flavors for each machine.
  44  */
  45 #include <mach/machine/thread_status.h>
  46
  47 /*
  48  * XXX historically, we have not included this header.  Continue to not do so.
  49  *
  50  * #include <architecture/byte_order.h>
  51  */
  52
  53 /*
  54  * The mach header appears at the very beginning of the object file; it
  55  * is the same for both 32-bit and 64-bit architectures.
  56  */
  57 struct mach_header {
  58         uint32_t        magic;          /* mach magic number identifier */
  59         cpu_type_t      cputype;        /* cpu specifier */
  60         cpu_subtype_t   cpusubtype;     /* machine specifier */
  61         uint32_t        filetype;       /* type of file */
  62         uint32_t        ncmds;          /* number of load commands */
  63         uint32_t        sizeofcmds;     /* the size of all the load commands */
  64         uint32_t        flags;          /* flags */
  65 };
  66
  67 /* Constant for the magic field of the mach_header (32-bit architectures) */
  68 #define MH_MAGIC        0xfeedface      /* the mach magic number */
  69 #define MH_CIGAM        NXSwapInt(MH_MAGIC)
  70
  71 /* Constant for the magic field of the mach_header_64 (64-bit architectures) */
  72 #define MH_MAGIC_64     0xfeedfacf      /* the 64-bit mach magic number */
  73 #define MH_CIGAM_64     NXSwapInt(MH_MAGIC_64)
  74
  75 /* Constants for the cmd field of new load commands, the type */
  76 #define LC_SEGMENT_64   0x19    /* 64-bit segment of this file to be mapped */
  77 #define LC_ROUTINES_64  0x1a    /* 64-bit image routines */
  78
  79
  80 /*
  81  * The layout of the file depends on the filetype.  For all but the MH_OBJECT
  82  * file type the segments are padded out and aligned on a segment alignment
  83  * boundary for efficient demand pageing.  The MH_EXECUTE, MH_FVMLIB, MH_DYLIB,
  84  * MH_DYLINKER and MH_BUNDLE file types also have the headers included as part
  85  * of their first segment.
  86  *
  87  * The file type MH_OBJECT is a compact format intended as output of the
  88  * assembler and input (and possibly output) of the link editor (the .o
  89  * format).  All sections are in one unnamed segment with no segment padding.
  90  * This format is used as an executable format when the file is so small the
  91  * segment padding greatly increases it's size.
  92  *
  93  * The file type MH_PRELOAD is an executable format intended for things that
  94  * not executed under the kernel (proms, stand alones, kernels, etc).  The
  95  * format can be executed under the kernel but may demand paged it and not
  96  * preload it before execution.
  97  *
  98  * A core file is in MH_CORE format and can be any in an arbritray legal
  99  * Mach-O file.
 100  *
 101  * Constants for the filetype field of the mach_header
 102  */
 103 #define MH_OBJECT       0x1             /* relocatable object file */
 104 #define MH_EXECUTE      0x2             /* demand paged executable file */
 105 #define MH_FVMLIB       0x3             /* fixed VM shared library file */
 106 #define MH_CORE         0x4             /* core file */
 107 #define MH_PRELOAD      0x5             /* preloaded executable file */
 108 #define MH_DYLIB        0x6             /* dynamicly bound shared library file*/
 109 #define MH_DYLINKER     0x7             /* dynamic link editor */
 110 #define MH_BUNDLE       0x8             /* dynamicly bound bundle file */
 111
 112 /* Constants for the flags field of the mach_header */
 113 #define MH_NOUNDEFS     0x1             /* the object file has no undefined
 114                                            references, can be executed */
 115 #define MH_INCRLINK     0x2             /* the object file is the output of an
 116                                            incremental link against a base file
 117                                            and can't be link edited again */
 118 #define MH_DYLDLINK     0x4             /* the object file is input for the
 119                                            dynamic linker and can't be staticly
 120                                            link edited again */
 121 #define MH_BINDATLOAD   0x8             /* the object file's undefined
 122                                            references are bound by the dynamic
 123                                            linker when loaded. */
 124 #define MH_PREBOUND     0x10            /* the file has it's dynamic undefined
 125                                            references prebound. */
 126
 127 /*
 128  * The load commands directly follow the mach_header.  The total size of all
 129  * of the commands is given by the sizeofcmds field in the mach_header.  All
 130  * load commands must have as their first two fields cmd and cmdsize.  The cmd
 131  * field is filled in with a constant for that command type.  Each command type
 132  * has a structure specifically for it.  The cmdsize field is the size in bytes
 133  * of the particular load command structure plus anything that follows it that
 134  * is a part of the load command (i.e. section structures, strings, etc.).  To
 135  * advance to the next load command the cmdsize can be added to the offset or
 136  * pointer of the current load command.  The cmdsize for 32-bit architectures
 137  * MUST be a multiple of 4 bytes and for 64-bit architectures MUST be a multiple
 138  * of 8 bytes (these are forever the maximum alignment of any load commands).
 139  * sizeof(long) (this is forever the maximum alignment of any load commands).
 140  * The padded bytes must be zero.  All tables in the object file must also
 141  * follow these rules so the file can be memory mapped.  Otherwise the pointers
 142  * to these tables will not work well or at all on some machines.  With all
 143  * padding zeroed like objects will compare byte for byte.
 144  */
 145 struct load_command {
 146         unsigned long cmd;              /* type of load command */
 147         unsigned long cmdsize;          /* total size of command in bytes */
 148 };
 149
 150 /* Constants for the cmd field of all load commands, the type */
 151 #define LC_SEGMENT      0x1     /* segment of this file to be mapped */
 152 #define LC_SYMTAB       0x2     /* link-edit stab symbol table info */
 153 #define LC_SYMSEG       0x3     /* link-edit gdb symbol table info (obsolete) */
 154 #define LC_THREAD       0x4     /* thread */
 155 #define LC_UNIXTHREAD   0x5     /* unix thread (includes a stack) */
 156 #define LC_LOADFVMLIB   0x6     /* load a specified fixed VM shared library */
 157 #define LC_IDFVMLIB     0x7     /* fixed VM shared library identification */
 158 #define LC_IDENT        0x8     /* object identification info (obsolete) */
 159 #define LC_FVMFILE      0x9     /* fixed VM file inclusion (internal use) */
 160 #define LC_PREPAGE      0xa     /* prepage command (internal use) */
 161 #define LC_DYSYMTAB     0xb     /* dynamic link-edit symbol table info */
 162 #define LC_LOAD_DYLIB   0xc     /* load a dynamicly linked shared library */
 163 #define LC_ID_DYLIB     0xd     /* dynamicly linked shared lib identification */
 164 #define LC_LOAD_DYLINKER 0xe    /* load a dynamic linker */
 165 #define LC_ID_DYLINKER  0xf     /* dynamic linker identification */
 166 #define LC_PREBOUND_DYLIB 0x10  /* modules prebound for a dynamicly */
 167                                 /*  linked shared library */
 168
 169 /*
 170  * A variable length string in a load command is represented by an lc_str
 171  * union.  The strings are stored just after the load command structure and
 172  * the offset is from the start of the load command structure.  The size
 173  * of the string is reflected in the cmdsize field of the load command.
 174  * Once again any padded bytes to bring the cmdsize field to a multiple
 175  * of sizeof(long) must be zero.
 176  */
 177 union lc_str {
 178         unsigned long   offset; /* offset to the string */
 179         char            *ptr;   /* pointer to the string */
 180 };
 181
 182 /*
 183  * The segment load command indicates that a part of this file is to be
 184  * mapped into the task's address space.  The size of this segment in memory,
 185  * vmsize, maybe equal to or larger than the amount to map from this file,
 186  * filesize.  The file is mapped starting at fileoff to the beginning of
 187  * the segment in memory, vmaddr.  The rest of the memory of the segment,
 188  * if any, is allocated zero fill on demand.  The segment's maximum virtual
 189  * memory protection and initial virtual memory protection are specified
 190  * by the maxprot and initprot fields.  If the segment has sections then the
 191  * section structures directly follow the segment command and their size is
 192  * reflected in cmdsize.
 193  */
 194 struct segment_command {        /* for 32-bit architectures */
 195         unsigned long   cmd;            /* LC_SEGMENT */
 196         unsigned long   cmdsize;        /* includes sizeof section structs */
 197         char            segname[16];    /* segment name */
 198         unsigned long   vmaddr;         /* memory address of this segment */
 199         unsigned long   vmsize;         /* memory size of this segment */
 200         unsigned long   fileoff;        /* file offset of this segment */
 201         unsigned long   filesize;       /* amount to map from the file */
 202         vm_prot_t       maxprot;        /* maximum VM protection */
 203         vm_prot_t       initprot;       /* initial VM protection */
 204         unsigned long   nsects;         /* number of sections in segment */
 205         unsigned long   flags;          /* flags */
 206 };
 207
 208 /*
 209  * The 64-bit segment load command indicates that a part of this file is to be
 210  * mapped into a 64-bit task's address space.  If the 64-bit segment has
 211  * sections then section_64 structures directly follow the 64-bit segment
 212  * command and their size is reflected in cmdsize.
 213  */
 214 struct segment_command_64 {     /* for 64-bit architectures */
 215         uint32_t        cmd;            /* LC_SEGMENT_64 */
 216         uint32_t        cmdsize;        /* includes sizeof section_64 structs */
 217         char            segname[16];    /* segment name */
 218         uint64_t        vmaddr;         /* memory address of this segment */
 219         uint64_t        vmsize;         /* memory size of this segment */
 220         uint32_t        fileoff;        /* file offset of this segment */
 221         uint32_t        filesize;       /* amount to map from the file */
 222         vm_prot_t       maxprot;        /* maximum VM protection */
 223         vm_prot_t       initprot;       /* initial VM protection */
 224         uint32_t        nsects;         /* number of sections in segment */
 225         uint32_t        flags;          /* flags */
 226 };
 227
 228
 229 /* Constants for the flags field of the segment_command */
 230 #define SG_HIGHVM       0x1     /* the file contents for this segment is for
 231                                    the high part of the VM space, the low part
 232                                    is zero filled (for stacks in core files) */
 233 #define SG_FVMLIB       0x2     /* this segment is the VM that is allocated by
 234                                    a fixed VM library, for overlap checking in
 235                                    the link editor */
 236 #define SG_NORELOC      0x4     /* this segment has nothing that was relocated
 237                                    in it and nothing relocated to it, that is
 238                                    it maybe safely replaced without relocation*/
 239
 240 /*
 241  * A segment is made up of zero or more sections.  Non-MH_OBJECT files have
 242  * all of their segments with the proper sections in each, and padded to the
 243  * specified segment alignment when produced by the link editor.  The first
 244  * segment of a MH_EXECUTE and MH_FVMLIB format file contains the mach_header
 245  * and load commands of the object file before it's first section.  The zero
 246  * fill sections are always last in their segment (in all formats).  This
 247  * allows the zeroed segment padding to be mapped into memory where zero fill
 248  * sections might be. The gigabyte zero fill sections, those with the section
 249  * type S_GB_ZEROFILL, can only be in a segment with sections of this type.
 250  * These segments are then placed after all other segments.
 251  *
 252  * The MH_OBJECT format has all of it's sections in one segment for
 253  * compactness.  There is no padding to a specified segment boundary and the
 254  * mach_header and load commands are not part of the segment.
 255  *
 256  * Sections with the same section name, sectname, going into the same segment,
 257  * segname, are combined by the link editor.  The resulting section is aligned
 258  * to the maximum alignment of the combined sections and is the new section's
 259  * alignment.  The combined sections are aligned to their original alignment in
 260  * the combined section.  Any padded bytes to get the specified alignment are
 261  * zeroed.
 262  *
 263  * The format of the relocation entries referenced by the reloff and nreloc
 264  * fields of the section structure for mach object files is described in the
 265  * header file <reloc.h>.
 266  */
 267 struct section {                /* for 32-bit architectures */
 268         char            sectname[16];   /* name of this section */
 269         char            segname[16];    /* segment this section goes in */
 270         unsigned long   addr;           /* memory address of this section */
 271         unsigned long   size;           /* size in bytes of this section */
 272         unsigned long   offset;         /* file offset of this section */
 273         unsigned long   align;          /* section alignment (power of 2) */
 274         unsigned long   reloff;         /* file offset of relocation entries */
 275         unsigned long   nreloc;         /* number of relocation entries */
 276         unsigned long   flags;          /* flags (section type and attributes)*/
 277         unsigned long   reserved1;      /* reserved */
 278         unsigned long   reserved2;      /* reserved */
 279 };
 280
 281 struct section_64 { /* for 64-bit architectures */
 282         char            sectname[16];   /* name of this section */
 283         char            segname[16];    /* segment this section goes in */
 284         uint64_t        addr;           /* memory address of this section */
 285         uint64_t        size;           /* size in bytes of this section */
 286         uint32_t        offset;         /* file offset of this section */
 287         uint32_t        align;          /* section alignment (power of 2) */
 288         uint32_t        reloff;         /* file offset of relocation entries */
 289         uint32_t        nreloc;         /* number of relocation entries */
 290         uint32_t        flags;          /* flags (section type and attributes)*/
 291         uint32_t        reserved1;      /* reserved (for offset or index) */
 292         uint32_t        reserved2;      /* reserved (for count or sizeof) */
 293         uint32_t        reserved3;      /* reserved */
 294 };
 295
 296
 297 /*
 298  * The flags field of a section structure is separated into two parts a section
 299  * type and section attributes.  The section types are mutually exclusive (it
 300  * can only have one type) but the section attributes are not (it may have more
 301  * than one attribute).
 302  */
 303 #define SECTION_TYPE             0x000000ff     /* 256 section types */
 304 #define SECTION_ATTRIBUTES       0xffffff00     /*  24 section attributes */
 305
 306 /* Constants for the type of a section */
 307 #define S_REGULAR               0x0     /* regular section */
 308 #define S_ZEROFILL              0x1     /* zero fill on demand section */
 309 #define S_CSTRING_LITERALS      0x2     /* section with only literal C strings*/
 310 #define S_4BYTE_LITERALS        0x3     /* section with only 4 byte literals */
 311 #define S_8BYTE_LITERALS        0x4     /* section with only 8 byte literals */
 312 #define S_LITERAL_POINTERS      0x5     /* section with only pointers to */
 313                                         /*  literals */
 314 /*
 315  * For the two types of symbol pointers sections and the symbol stubs section
 316  * they have indirect symbol table entries.  For each of the entries in the
 317  * section the indirect symbol table entries, in corresponding order in the
 318  * indirect symbol table, start at the index stored in the reserved1 field
 319  * of the section structure.  Since the indirect symbol table entries
 320  * correspond to the entries in the section the number of indirect symbol table
 321  * entries is inferred from the size of the section divided by the size of the
 322  * entries in the section.  For symbol pointers sections the size of the entries
 323  * in the section is 4 bytes and for symbol stubs sections the byte size of the
 324  * stubs is stored in the reserved2 field of the section structure.
 325  */
 326 #define S_NON_LAZY_SYMBOL_POINTERS      0x6     /* section with only non-lazy
 327                                                    symbol pointers */
 328 #define S_LAZY_SYMBOL_POINTERS          0x7     /* section with only lazy symbol
 329                                                    pointers */
 330 #define S_SYMBOL_STUBS                  0x8     /* section with only symbol
 331                                                    stubs, byte size of stub in
 332                                                    the reserved2 field */
 333 #define S_MOD_INIT_FUNC_POINTERS        0x9     /* section with only function
 334                                                    pointers for initialization*/
 335 /*
 336  * Constants for the section attributes part of the flags field of a section
 337  * structure.
 338  */
 339 #define SECTION_ATTRIBUTES_USR   0xff000000     /* User setable attributes */
 340 #define S_ATTR_PURE_INSTRUCTIONS 0x80000000     /* section contains only true
 341                                                    machine instructions */
 342 #define SECTION_ATTRIBUTES_SYS   0x00ffff00     /* system setable attributes */
 343 #define S_ATTR_SOME_INSTRUCTIONS 0x00000400     /* section contains some
 344                                                    machine instructions */
 345 #define S_ATTR_EXT_RELOC         0x00000200     /* section has external
 346                                                    relocation entries */
 347 #define S_ATTR_LOC_RELOC         0x00000100     /* section has local
 348                                                    relocation entries */
 349
 350
 351 /*
 352  * The names of segments and sections in them are mostly meaningless to the
 353  * link-editor.  But there are few things to support traditional UNIX
 354  * executables that require the link-editor and assembler to use some names
 355  * agreed upon by convention.
 356  *
 357  * The initial protection of the "__TEXT" segment has write protection turned
 358  * off (not writeable).
 359  *
 360  * The link-editor will allocate common symbols at the end of the "__common"
 361  * section in the "__DATA" segment.  It will create the section and segment
 362  * if needed.
 363  */
 364
 365 /* The currently known segment names and the section names in those segments */
 366
 367 #define SEG_PAGEZERO    "__PAGEZERO"    /* the pagezero segment which has no */
 368                                         /* protections and catches NULL */
 369                                         /* references for MH_EXECUTE files */
 370
 371
 372 #define SEG_TEXT        "__TEXT"        /* the tradition UNIX text segment */
 373 #define SECT_TEXT       "__text"        /* the real text part of the text */
 374                                         /* section no headers, and no padding */
 375 #define SECT_FVMLIB_INIT0 "__fvmlib_init0"      /* the fvmlib initialization */
 376                                                 /*  section */
 377 #define SECT_FVMLIB_INIT1 "__fvmlib_init1"      /* the section following the */
 378                                                 /*  fvmlib initialization */
 379                                                 /*  section */
 380
 381 #define SEG_DATA        "__DATA"        /* the tradition UNIX data segment */
 382 #define SECT_DATA       "__data"        /* the real initialized data section */
 383                                         /* no padding, no bss overlap */
 384 #define SECT_BSS        "__bss"         /* the real uninitialized data section*/
 385                                         /* no padding */
 386 #define SECT_COMMON     "__common"      /* the section common symbols are */
 387                                         /* allocated in by the link editor */
 388
 389 #define SEG_OBJC        "__OBJC"        /* objective-C runtime segment */
 390 #define SECT_OBJC_SYMBOLS "__symbol_table"      /* symbol table */
 391 #define SECT_OBJC_MODULES "__module_info"       /* module information */
 392 #define SECT_OBJC_STRINGS "__selector_strs"     /* string table */
 393 #define SECT_OBJC_REFS "__selector_refs"        /* string table */
 394
 395 #define SEG_ICON         "__ICON"       /* the NeXT icon segment */
 396 #define SECT_ICON_HEADER "__header"     /* the icon headers */
 397 #define SECT_ICON_TIFF   "__tiff"       /* the icons in tiff format */
 398
 399 #define SEG_LINKEDIT    "__LINKEDIT"    /* the segment containing all structs */
 400                                         /* created and maintained by the link */
 401                                         /* editor.  Created with -seglinkedit */
 402                                         /* option to ld(1) for MH_EXECUTE and */
 403                                         /* FVMLIB file types only */
 404
 405 #define SEG_UNIXSTACK   "__UNIXSTACK"   /* the unix stack segment */
 406
 407 /*
 408  * Fixed virtual memory shared libraries are identified by two things.  The
 409  * target pathname (the name of the library as found for execution), and the
 410  * minor version number.  The address of where the headers are loaded is in
 411  * header_addr.
 412  */
 413 struct fvmlib {
 414         union lc_str    name;           /* library's target pathname */
 415         unsigned long   minor_version;  /* library's minor version number */
 416         unsigned long   header_addr;    /* library's header address */
 417 };
 418
 419 /*
 420  * A fixed virtual shared library (filetype == MH_FVMLIB in the mach header)
 421  * contains a fvmlib_command (cmd == LC_IDFVMLIB) to identify the library.
 422  * An object that uses a fixed virtual shared library also contains a
 423  * fvmlib_command (cmd == LC_LOADFVMLIB) for each library it uses.
 424  */
 425 struct fvmlib_command {
 426         unsigned long   cmd;            /* LC_IDFVMLIB or LC_LOADFVMLIB */
 427         unsigned long   cmdsize;        /* includes pathname string */
 428         struct fvmlib   fvmlib;         /* the library identification */
 429 };
 430
 431 /*
 432  * Dynamicly linked shared libraries are identified by two things.  The
 433  * pathname (the name of the library as found for execution), and the
 434  * compatibility version number.  The pathname must match and the compatibility
 435  * number in the user of the library must be greater than or equal to the
 436  * library being used.  The time stamp is used to record the time a library was
 437  * built and copied into user so it can be use to determined if the library used
 438  * at runtime is exactly the same as used to built the program.
 439  */
 440 struct dylib {
 441     union lc_str  name;                 /* library's path name */
 442     unsigned long timestamp;            /* library's build time stamp */
 443     unsigned long current_version;      /* library's current version number */
 444     unsigned long compatibility_version;/* library's compatibility vers number*/
 445 };
 446
 447 /*
 448  * A dynamicly linked shared library (filetype == MH_DYLIB in the mach header)
 449  * contains a dylib_command (cmd == LC_ID_DYLIB) to identify the library.
 450  * An object that uses a dynamicly linked shared library also contains a
 451  * dylib_command (cmd == LC_LOAD_DYLIB) for each library it uses.
 452  */
 453 struct dylib_command {
 454         unsigned long   cmd;            /* LC_ID_DYLIB or LC_LOAD_DYLIB */
 455         unsigned long   cmdsize;        /* includes pathname string */
 456         struct dylib    dylib;          /* the library identification */
 457 };
 458
 459 /*
 460  * A program (filetype == MH_EXECUTE) or bundle (filetype == MH_BUNDLE) that is
 461  * prebound to it's dynamic libraries has one of these for each library that
 462  * the static linker used in prebinding.  It contains a bit vector for the
 463  * modules in the library.  The bits indicate which modules are bound (1) and
 464  * which are not (0) from the library.  The bit for module 0 is the low bit
 465  * of the first byte.  So the bit for the Nth module is:
 466  * (linked_modules[N/8] >> N%8) & 1
 467  */
 468 struct prebound_dylib_command {
 469         unsigned long   cmd;            /* LC_PREBOUND_DYLIB */
 470         unsigned long   cmdsize;        /* includes strings */
 471         union lc_str    name;           /* library's path name */
 472         unsigned long   nmodules;       /* number of modules in library */
 473         union lc_str    linked_modules; /* bit vector of linked modules */
 474 };
 475
 476 /*
 477  * A program that uses a dynamic linker contains a dylinker_command to identify
 478  * the name of the dynamic linker (LC_LOAD_DYLINKER).  And a dynamic linker
 479  * contains a dylinker_command to identify the dynamic linker (LC_ID_DYLINKER).
 480  * A file can have at most one of these.
 481  */
 482 struct dylinker_command {
 483         unsigned long   cmd;            /* LC_ID_DYLINKER or LC_LOAD_DYLINKER */
 484         unsigned long   cmdsize;        /* includes pathname string */
 485         union lc_str    name;           /* dynamic linker's path name */
 486 };
 487
 488 /*
 489  * Thread commands contain machine-specific data structures suitable for
 490  * use in the thread state primitives.  The machine specific data structures
 491  * follow the struct thread_command as follows.
 492  * Each flavor of machine specific data structure is preceded by an unsigned
 493  * long constant for the flavor of that data structure, an unsigned long
 494  * that is the count of longs of the size of the state data structure and then
 495  * the state data structure follows.  This triple may be repeated for many
 496  * flavors.  The constants for the flavors, counts and state data structure
 497  * definitions are expected to be in the header file <machine/thread_status.h>.
 498  * These machine specific data structures sizes must be multiples of
 499  * sizeof(long).  The cmdsize reflects the total size of the thread_command
 500  * and all of the sizes of the constants for the flavors, counts and state
 501  * data structures.
 502  *
 503  * For executable objects that are unix processes there will be one
 504  * thread_command (cmd == LC_UNIXTHREAD) created for it by the link-editor.
 505  * This is the same as a LC_THREAD, except that a stack is automatically
 506  * created (based on the shell's limit for the stack size).  Command arguments
 507  * and environment variables are copied onto that stack.
 508  */
 509 struct thread_command {
 510         unsigned long   cmd;            /* LC_THREAD or  LC_UNIXTHREAD */
 511         unsigned long   cmdsize;        /* total size of this command */
 512         /* unsigned long flavor            flavor of thread state */
 513         /* unsigned long count             count of longs in thread state */
 514         /* struct XXX_thread_state state   thread state for this flavor */
 515         /* ... */
 516 };
 517
 518 /*
 519  * The symtab_command contains the offsets and sizes of the link-edit 4.3BSD
 520  * "stab" style symbol table information as described in the header files
 521  * <nlist.h> and <stab.h>.
 522  */
 523 struct symtab_command {
 524         unsigned long   cmd;            /* LC_SYMTAB */
 525         unsigned long   cmdsize;        /* sizeof(struct symtab_command) */
 526         unsigned long   symoff;         /* symbol table offset */
 527         unsigned long   nsyms;          /* number of symbol table entries */
 528         unsigned long   stroff;         /* string table offset */
 529         unsigned long   strsize;        /* string table size in bytes */
 530 };
 531
 532 /*
 533  * This is the second set of the symbolic information which is used to support
 534  * the data structures for the dynamicly link editor.
 535  *
 536  * The original set of symbolic information in the symtab_command which contains
 537  * the symbol and string tables must also be present when this load command is
 538  * present.  When this load command is present the symbol table is organized
 539  * into three groups of symbols:
 540  *      local symbols (static and debugging symbols) - grouped by module
 541  *      defined external symbols - grouped by module (sorted by name if not lib)
 542  *      undefined external symbols (sorted by name)
 543  * In this load command there are offsets and counts to each of the three groups
 544  * of symbols.
 545  *
 546  * This load command contains a the offsets and sizes of the following new
 547  * symbolic information tables:
 548  *      table of contents
 549  *      module table
 550  *      reference symbol table
 551  *      indirect symbol table
 552  * The first three tables above (the table of contents, module table and
 553  * reference symbol table) are only present if the file is a dynamicly linked
 554  * shared library.  For executable and object modules, which are files
 555  * containing only one module, the information that would be in these three
 556  * tables is determined as follows:
 557  *      table of contents - the defined external symbols are sorted by name
 558  *      module table - the file contains only one module so everything in the
 559  *                     file is part of the module.
 560  *      reference symbol table - is the defined and undefined external symbols
 561  *
 562  * For dynamicly linked shared library files this load command also contains
 563  * offsets and sizes to the pool of relocation entries for all sections
 564  * separated into two groups:
 565  *      external relocation entries
 566  *      local relocation entries
 567  * For executable and object modules the relocation entries continue to hang
 568  * off the section structures.
 569  */
 570 struct dysymtab_command {
 571     unsigned long cmd;          /* LC_DYSYMTAB */
 572     unsigned long cmdsize;      /* sizeof(struct dysymtab_command) */
 573
 574     /*
 575      * The symbols indicated by symoff and nsyms of the LC_SYMTAB load command
 576      * are grouped into the following three groups:
 577      *    local symbols (further grouped by the module they are from)
 578      *    defined external symbols (further grouped by the module they are from)
 579      *    undefined symbols
 580      *
 581      * The local symbols are used only for debugging.  The dynamic binding
 582      * process may have to use them to indicate to the debugger the local
 583      * symbols for a module that is being bound.
 584      *
 585      * The last two groups are used by the dynamic binding process to do the
 586      * binding (indirectly through the module table and the reference symbol
 587      * table when this is a dynamicly linked shared library file).
 588      */
 589     unsigned long ilocalsym;    /* index to local symbols */
 590     unsigned long nlocalsym;    /* number of local symbols */
 591
 592     unsigned long iextdefsym;   /* index to externally defined symbols */
 593     unsigned long nextdefsym;   /* number of externally defined symbols */
 594
 595     unsigned long iundefsym;    /* index to undefined symbols */
 596     unsigned long nundefsym;    /* number of undefined symbols */
 597
 598     /*
 599      * For the for the dynamic binding process to find which module a symbol
 600      * is defined in the table of contents is used (analogous to the ranlib
 601      * structure in an archive) which maps defined external symbols to modules
 602      * they are defined in.  This exists only in a dynamicly linked shared
 603      * library file.  For executable and object modules the defined external
 604      * symbols are sorted by name and is use as the table of contents.
 605      */
 606     unsigned long tocoff;       /* file offset to table of contents */
 607     unsigned long ntoc;         /* number of entries in table of contents */
 608
 609     /*
 610      * To support dynamic binding of "modules" (whole object files) the symbol
 611      * table must reflect the modules that the file was created from.  This is
 612      * done by having a module table that has indexes and counts into the merged
 613      * tables for each module.  The module structure that these two entries
 614      * refer to is described below.  This exists only in a dynamicly linked
 615      * shared library file.  For executable and object modules the file only
 616      * contains one module so everything in the file belongs to the module.
 617      */
 618     unsigned long modtaboff;    /* file offset to module table */
 619     unsigned long nmodtab;      /* number of module table entries */
 620
 621     /*
 622      * To support dynamic module binding the module structure for each module
 623      * indicates the external references (defined and undefined) each module
 624      * makes.  For each module there is an offset and a count into the
 625      * reference symbol table for the symbols that the module references.
 626      * This exists only in a dynamicly linked shared library file.  For
 627      * executable and object modules the defined external symbols and the
 628      * undefined external symbols indicates the external references.
 629      */
 630     unsigned long extrefsymoff;  /* offset to referenced symbol table */
 631     unsigned long nextrefsyms;   /* number of referenced symbol table entries */
 632
 633     /*
 634      * The sections that contain "symbol pointers" and "routine stubs" have
 635      * indexes and (implied counts based on the size of the section and fixed
 636      * size of the entry) into the "indirect symbol" table for each pointer
 637      * and stub.  For every section of these two types the index into the
 638      * indirect symbol table is stored in the section header in the field
 639      * reserved1.  An indirect symbol table entry is simply a 32bit index into
 640      * the symbol table to the symbol that the pointer or stub is referring to.
 641      * The indirect symbol table is ordered to match the entries in the section.
 642      */
 643     unsigned long indirectsymoff; /* file offset to the indirect symbol table */
 644     unsigned long nindirectsyms;  /* number of indirect symbol table entries */
 645
 646     /*
 647      * To support relocating an individual module in a library file quickly the
 648      * external relocation entries for each module in the library need to be
 649      * accessed efficiently.  Since the relocation entries can't be accessed
 650      * through the section headers for a library file they are separated into
 651      * groups of local and external entries further grouped by module.  In this
 652      * case the presents of this load command who's extreloff, nextrel,
 653      * locreloff and nlocrel fields are non-zero indicates that the relocation
 654      * entries of non-merged sections are not referenced through the section
 655      * structures (and the reloff and nreloc fields in the section headers are
 656      * set to zero).
 657      *
 658      * Since the relocation entries are not accessed through the section headers
 659      * this requires the r_address field to be something other than a section
 660      * offset to identify the item to be relocated.  In this case r_address is
 661      * set to the offset from the vmaddr of the first LC_SEGMENT command.
 662      *
 663      * The relocation entries are grouped by module and the module table
 664      * entries have indexes and counts into them for the group of external
 665      * relocation entries for that the module.
 666      *
 667      * For sections that are merged across modules there must not be any
 668      * remaining external relocation entries for them (for merged sections
 669      * remaining relocation entries must be local).
 670      */
 671     unsigned long extreloff;    /* offset to external relocation entries */
 672     unsigned long nextrel;      /* number of external relocation entries */
 673
 674     /*
 675      * All the local relocation entries are grouped together (they are not
 676      * grouped by their module since they are only used if the object is moved
 677      * from it staticly link edited address).
 678      */
 679     unsigned long locreloff;    /* offset to local relocation entries */
 680     unsigned long nlocrel;      /* number of local relocation entries */
 681
 682 };
 683
 684 /*
 685  * An indirect symbol table entry is simply a 32bit index into the symbol table
 686  * to the symbol that the pointer or stub is refering to.  Unless it is for a
 687  * non-lazy symbol pointer section for a defined symbol which strip(1) as
 688  * removed.  In which case it has the value INDIRECT_SYMBOL_LOCAL.  If the
 689  * symbol was also absolute INDIRECT_SYMBOL_ABS is or'ed with that.
 690  */
 691 #define INDIRECT_SYMBOL_LOCAL   0x80000000
 692 #define INDIRECT_SYMBOL_ABS     0x40000000
 693
 694
 695 /* a table of contents entry */
 696 struct dylib_table_of_contents {
 697     unsigned long symbol_index; /* the defined external symbol
 698                                    (index into the symbol table) */
 699     unsigned long module_index; /* index into the module table this symbol
 700                                    is defined in */
 701 };
 702
 703 /* a module table entry */
 704 struct dylib_module {
 705     unsigned long module_name;  /* the module name (index into string table) */
 706
 707     unsigned long iextdefsym;   /* index into externally defined symbols */
 708     unsigned long nextdefsym;   /* number of externally defined symbols */
 709     unsigned long irefsym;              /* index into reference symbol table */
 710     unsigned long nrefsym;      /* number of reference symbol table entries */
 711     unsigned long ilocalsym;    /* index into symbols for local symbols */
 712     unsigned long nlocalsym;    /* number of local symbols */
 713
 714     unsigned long iextrel;      /* index into external relocation entries */
 715     unsigned long nextrel;      /* number of external relocation entries */
 716
 717     unsigned long iinit;        /* index into the init section */
 718     unsigned long ninit;        /* number of init section entries */
 719
 720     unsigned long               /* for this module address of the start of */
 721         objc_module_info_addr;  /*  the (__OBJC,__module_info) section */
 722     unsigned long               /* for this module size of */
 723         objc_module_info_size;  /*  the (__OBJC,__module_info) section */
 724 };
 725
 726 /* a 64-bit module table entry */
 727 struct dylib_module_64 {
 728         uint32_t module_name;   /* the module name (index into string table) */
 729
 730         uint32_t iextdefsym;    /* index into externally defined symbols */
 731         uint32_t nextdefsym;    /* number of externally defined symbols */
 732         uint32_t irefsym;       /* index into reference symbol table */
 733         uint32_t nrefsym;       /* number of reference symbol table entries */
 734         uint32_t ilocalsym;     /* index into symbols for local symbols */
 735         uint32_t nlocalsym;     /* number of local symbols */
 736
 737         uint32_t iextrel;       /* index into external relocation entries */
 738         uint32_t nextrel;       /* number of external relocation entries */
 739
 740         uint32_t iinit_iterm;   /* low 16 bits are the index into the init
 741                                    section, high 16 bits are the index into
 742                                    the term section */
 743         uint32_t ninit_nterm;   /* low 16 bits are the number of init section
 744                                    entries, high 16 bits are the number of
 745                                    term section entries */
 746
 747         uint32_t                /* for this module size of the */
 748                 objc_module_info_size;  /* (__OBJC,__module_info) section */
 749         uint64_t                /* for this module address of the start of */
 750                 objc_module_info_addr;  /* the (__OBJC,__module_info) section */
 751 };
 752
 753
 754 /*
 755  * The entries in the reference symbol table are used when loading the module
 756  * (both by the static and dynamic link editors) and if the module is unloaded
 757  * or replaced.  Therefore all external symbols (defined and undefined) are
 758  * listed in the module's reference table.  The flags describe the type of
 759  * reference that is being made.  The constants for the flags are defined in
 760  * <mach-o/nlist.h> as they are also used for symbol table entries.
 761  */
 762 struct dylib_reference {
 763     unsigned long isym:24,      /* index into the symbol table */
 764                   flags:8;      /* flags to indicate the type of reference */
 765 };
 766
 767 /*
 768  * The symseg_command contains the offset and size of the GNU style
 769  * symbol table information as described in the header file <symseg.h>.
 770  * The symbol roots of the symbol segments must also be aligned properly
 771  * in the file.  So the requirement of keeping the offsets aligned to a
 772  * multiple of a sizeof(long) translates to the length field of the symbol
 773  * roots also being a multiple of a long.  Also the padding must again be
 774  * zeroed. (THIS IS OBSOLETE and no longer supported).
 775  */
 776 struct symseg_command {
 777         unsigned long   cmd;            /* LC_SYMSEG */
 778         unsigned long   cmdsize;        /* sizeof(struct symseg_command) */
 779         unsigned long   offset;         /* symbol segment offset */
 780         unsigned long   size;           /* symbol segment size in bytes */
 781 };
 782
 783 /*
 784  * The ident_command contains a free format string table following the
 785  * ident_command structure.  The strings are null terminated and the size of
 786  * the command is padded out with zero bytes to a multiple of sizeof(long).
 787  * (THIS IS OBSOLETE and no longer supported).
 788  */
 789 struct ident_command {
 790         unsigned long cmd;              /* LC_IDENT */
 791         unsigned long cmdsize;          /* strings that follow this command */
 792 };
 793
 794 /*
 795  * The fvmfile_command contains a reference to a file to be loaded at the
 796  * specified virtual address.  (Presently, this command is reserved for NeXT
 797  * internal use.  The kernel ignores this command when loading a program into
 798  * memory).
 799  */
 800 struct fvmfile_command {
 801         unsigned long cmd;              /* LC_FVMFILE */
 802         unsigned long cmdsize;          /* includes pathname string */
 803         union lc_str    name;           /* files pathname */
 804         unsigned long   header_addr;    /* files virtual address */
 805 };
 806
 807 #endif /*_MACHO_LOADER_H_*/