1.\" Copyright (c) 1991, 1993 2.\" The Regents of the University of California. All rights reserved. 3.\" 4.\" This manual page is derived from documentation contributed to Berkeley by 5.\" Donn Seeley at UUNET Technologies, Inc. 6.\" 7.\" Redistribution and use in source and binary forms, with or without 8.\" modification, are permitted provided that the following conditions 9.\" are met: 10.\" 1. Redistributions of source code must retain the above copyright 11.\" notice, this list of conditions and the following disclaimer. 12.\" 2. Redistributions in binary form must reproduce the above copyright 13.\" notice, this list of conditions and the following disclaimer in the 14.\" documentation and/or other materials provided with the distribution. 15.\" 3. Neither the name of the University nor the names of its contributors 16.\" may be used to endorse or promote products derived from this software 17.\" without specific prior written permission. 18.\" 19.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29.\" SUCH DAMAGE. 30.\" 31.\" @(#)a.out.5 8.1 (Berkeley) 6/5/93 32.\" 33.Dd June 10, 2010 34.Dt A.OUT 5 35.Os 36.Sh NAME 37.Nm a.out 38.Nd format of executable binary files 39.Sh SYNOPSIS 40.In a.out.h 41.Sh DESCRIPTION 42The include file 43.In a.out.h 44declares three structures and several macros. 45The structures describe the format of 46executable machine code files 47.Pq Sq binaries 48on the system. 49.Pp 50A binary file consists of up to 7 sections. 51In order, these sections are: 52.Bl -tag -width "text relocations" 53.It exec header 54Contains parameters used by the kernel 55to load a binary file into memory and execute it, 56and by the link editor 57.Xr ld 1 58to combine a binary file with other binary files. 59This section is the only mandatory one. 60.It text segment 61Contains machine code and related data 62that are loaded into memory when a program executes. 63May be loaded read-only. 64.It data segment 65Contains initialized data; always loaded into writable memory. 66.It text relocations 67Contains records used by the link editor 68to update pointers in the text segment when combining binary files. 69.It data relocations 70Like the text relocation section, but for data segment pointers. 71.It symbol table 72Contains records used by the link editor 73to cross reference the addresses of named variables and functions 74.Pq Sq symbols 75between binary files. 76.It string table 77Contains the character strings corresponding to the symbol names. 78.El 79.Pp 80Every binary file begins with an 81.Fa exec 82structure: 83.Bd -literal -offset indent 84struct exec { 85 unsigned long a_midmag; 86 unsigned long a_text; 87 unsigned long a_data; 88 unsigned long a_bss; 89 unsigned long a_syms; 90 unsigned long a_entry; 91 unsigned long a_trsize; 92 unsigned long a_drsize; 93}; 94.Ed 95.Pp 96The fields have the following functions: 97.Bl -tag -width a_trsize 98.It Fa a_midmag 99This field is stored in host byte-order. 100It has a number of sub-components accessed by the macros 101.Fn N_GETFLAG , 102.Fn N_GETMID , 103and 104.Fn N_GETMAGIC , 105and set by the macro 106.Fn N_SETMAGIC . 107.Pp 108The macro 109.Fn N_GETFLAG 110returns a few flags: 111.Bl -tag -width EX_DYNAMIC 112.It Dv EX_DYNAMIC 113indicates that the executable requires the services of the run-time link editor. 114.It Dv EX_PIC 115indicates that the object contains position independent code. 116This flag is 117set by 118.Xr as 1 119when given the 120.Sq -k 121flag and is preserved by 122.Xr ld 1 123if necessary. 124.El 125.Pp 126If both EX_DYNAMIC and EX_PIC are set, the object file is a position independent 127executable image (e.g.\& a shared library), which is to be loaded into the 128process address space by the run-time link editor. 129.Pp 130The macro 131.Fn N_GETMID 132returns the machine-id. 133This indicates which machine(s) the binary is intended to run on. 134.Pp 135.Fn N_GETMAGIC 136specifies the magic number, which uniquely identifies binary files 137and distinguishes different loading conventions. 138The field must contain one of the following values: 139.Bl -tag -width ZMAGIC 140.It Dv OMAGIC 141The text and data segments immediately follow the header 142and are contiguous. 143The kernel loads both text and data segments into writable memory. 144.It Dv NMAGIC 145As with 146.Dv OMAGIC , 147text and data segments immediately follow the header and are contiguous. 148However, the kernel loads the text into read-only memory 149and loads the data into writable memory at the next 150page boundary after the text. 151.It Dv ZMAGIC 152The kernel loads individual pages on demand from the binary. 153The header, text segment and data segment are all 154padded by the link editor to a multiple of the page size. 155Pages that the kernel loads from the text segment are read-only, 156while pages from the data segment are writable. 157.El 158.It Fa a_text 159Contains the size of the text segment in bytes. 160.It Fa a_data 161Contains the size of the data segment in bytes. 162.It Fa a_bss 163Contains the number of bytes in the 164.Sq bss segment 165and is used by the kernel to set the initial break 166.Pq Xr brk 2 167after the data segment. 168The kernel loads the program so that this amount of writable memory 169appears to follow the data segment and initially reads as zeroes. 170.Em ( bss 171= block started by symbol) 172.It Fa a_syms 173Contains the size in bytes of the symbol table section. 174.It Fa a_entry 175Contains the address in memory of the entry point 176of the program after the kernel has loaded it; 177the kernel starts the execution of the program 178from the machine instruction at this address. 179.It Fa a_trsize 180Contains the size in bytes of the text relocation table. 181.It Fa a_drsize 182Contains the size in bytes of the data relocation table. 183.El 184.Pp 185The 186.In a.out.h 187include file defines several macros which use an 188.Fa exec 189structure to test consistency or to locate section offsets in the binary file. 190.Bl -tag -width N_BADMAG(exec) 191.It Fn N_BADMAG exec 192Nonzero if the 193.Fa a_magic 194field does not contain a recognized value. 195.It Fn N_TXTOFF exec 196The byte offset in the binary file of the beginning of the text segment. 197.It Fn N_SYMOFF exec 198The byte offset of the beginning of the symbol table. 199.It Fn N_STROFF exec 200The byte offset of the beginning of the string table. 201.El 202.Pp 203Relocation records have a standard format which 204is described by the 205.Fa relocation_info 206structure: 207.Bd -literal -offset indent 208struct relocation_info { 209 int r_address; 210 unsigned int r_symbolnum : 24, 211 r_pcrel : 1, 212 r_length : 2, 213 r_extern : 1, 214 r_baserel : 1, 215 r_jmptable : 1, 216 r_relative : 1, 217 r_copy : 1; 218}; 219.Ed 220.Pp 221The 222.Fa relocation_info 223fields are used as follows: 224.Bl -tag -width r_symbolnum 225.It Fa r_address 226Contains the byte offset of a pointer that needs to be link-edited. 227Text relocation offsets are reckoned from the start of the text segment, 228and data relocation offsets from the start of the data segment. 229The link editor adds the value that is already stored at this offset 230into the new value that it computes using this relocation record. 231.It Fa r_symbolnum 232Contains the ordinal number of a symbol structure 233in the symbol table (it is 234.Em not 235a byte offset). 236After the link editor resolves the absolute address for this symbol, 237it adds that address to the pointer that is undergoing relocation. 238(If the 239.Fa r_extern 240bit is clear, the situation is different; see below.) 241.It Fa r_pcrel 242If this is set, 243the link editor assumes that it is updating a pointer 244that is part of a machine code instruction using pc-relative addressing. 245The address of the relocated pointer is implicitly added 246to its value when the running program uses it. 247.It Fa r_length 248Contains the log base 2 of the length of the pointer in bytes; 2490 for 1-byte displacements, 1 for 2-byte displacements, 2502 for 4-byte displacements. 251.It Fa r_extern 252Set if this relocation requires an external reference; 253the link editor must use a symbol address to update the pointer. 254When the 255.Fa r_extern 256bit is clear, the relocation is 257.Sq local ; 258the link editor updates the pointer to reflect 259changes in the load addresses of the various segments, 260rather than changes in the value of a symbol (except when 261.Fa r_baserel 262is also set (see below). 263In this case, the content of the 264.Fa r_symbolnum 265field is an 266.Fa n_type 267value (see below); 268this type field tells the link editor 269what segment the relocated pointer points into. 270.It Fa r_baserel 271If set, the symbol, as identified by the 272.Fa r_symbolnum 273field, is to be relocated to an offset into the Global Offset Table. 274At run-time, the entry in the Global Offset Table at this offset is set to 275be the address of the symbol. 276.It Fa r_jmptable 277If set, the symbol, as identified by the 278.Fa r_symbolnum 279field, is to be relocated to an offset into the Procedure Linkage Table. 280.It Fa r_relative 281If set, this relocation is relative to the (run-time) load address of the 282image this object file is going to be a part of. 283This type of relocation 284only occurs in shared objects. 285.It Fa r_copy 286If set, this relocation record identifies a symbol whose contents should 287be copied to the location given in 288.Fa r_address . 289The copying is done by the run-time link-editor from a suitable data 290item in a shared object. 291.El 292.Pp 293Symbols map names to addresses (or more generally, strings to values). 294Since the link-editor adjusts addresses, 295a symbol's name must be used to stand for its address 296until an absolute value has been assigned. 297Symbols consist of a fixed-length record in the symbol table 298and a variable-length name in the string table. 299The symbol table is an array of 300.Fa nlist 301structures: 302.Bd -literal -offset indent 303struct nlist { 304 union { 305 const char *n_name; 306 long n_strx; 307 } n_un; 308 unsigned char n_type; 309 char n_other; 310 short n_desc; 311 unsigned long n_value; 312}; 313.Ed 314.Pp 315The fields are used as follows: 316.Bl -tag -width n_un.n_strx 317.It Fa n_un.n_strx 318Contains a byte offset into the string table 319for the name of this symbol. 320When a program accesses a symbol table with the 321.Xr nlist 3 322function, 323this field is replaced with the 324.Fa n_un.n_name 325field, which is a pointer to the string in memory. 326.It Fa n_type 327Used by the link editor to determine 328how to update the symbol's value. 329The 330.Fa n_type 331field is broken down into three sub-fields using bitmasks. 332The link editor treats symbols with the 333.Dv N_EXT 334type bit set as 335.Sq external 336symbols and permits references to them from other binary files. 337The 338.Dv N_TYPE 339mask selects bits of interest to the link editor: 340.Bl -tag -width N_TEXT 341.It Dv N_UNDF 342An undefined symbol. 343The link editor must locate an external symbol with the same name 344in another binary file to determine the absolute value of this symbol. 345As a special case, if the 346.Fa n_value 347field is nonzero and no binary file in the link-edit defines this symbol, 348the link-editor will resolve this symbol to an address 349in the bss segment, 350reserving an amount of bytes equal to 351.Fa n_value . 352If this symbol is undefined in more than one binary file 353and the binary files do not agree on the size, 354the link editor chooses the greatest size found across all binaries. 355.It Dv N_ABS 356An absolute symbol. 357The link editor does not update an absolute symbol. 358.It Dv N_TEXT 359A text symbol. 360This symbol's value is a text address and 361the link editor will update it when it merges binary files. 362.It Dv N_DATA 363A data symbol; similar to 364.Dv N_TEXT 365but for data addresses. 366The values for text and data symbols are not file offsets but 367addresses; to recover the file offsets, it is necessary 368to identify the loaded address of the beginning of the corresponding 369section and subtract it, then add the offset of the section. 370.It Dv N_BSS 371A bss symbol; like text or data symbols but 372has no corresponding offset in the binary file. 373.It Dv N_FN 374A filename symbol. 375The link editor inserts this symbol before 376the other symbols from a binary file when 377merging binary files. 378The name of the symbol is the filename given to the link editor, 379and its value is the first text address from that binary file. 380Filename symbols are not needed for link-editing or loading, 381but are useful for debuggers. 382.El 383.Pp 384The 385.Dv N_STAB 386mask selects bits of interest to symbolic debuggers 387such as 388.Xr gdb 1 Pq Pa ports/devel/gdb ; 389the values are described in 390.Xr stab 5 . 391.It Fa n_other 392This field provides information on the nature of the symbol independent of 393the symbol's location in terms of segments as determined by the 394.Fa n_type 395field. 396Currently, the lower 4 bits of the 397.Fa n_other 398field hold one of two values: 399.Dv AUX_FUNC 400and 401.Dv AUX_OBJECT 402(see 403.In link.h 404for their definitions). 405.Dv AUX_FUNC 406associates the symbol with a callable function, while 407.Dv AUX_OBJECT 408associates the symbol with data, irrespective of their locations in 409either the text or the data segment. 410This field is intended to be used by 411.Xr ld 1 412for the construction of dynamic executables. 413.It Fa n_desc 414Reserved for use by debuggers; passed untouched by the link editor. 415Different debuggers use this field for different purposes. 416.It Fa n_value 417Contains the value of the symbol. 418For text, data and bss symbols, this is an address; 419for other symbols (such as debugger symbols), 420the value may be arbitrary. 421.El 422.Pp 423The string table consists of an 424.Em unsigned long 425length followed by null-terminated symbol strings. 426The length represents the size of the entire table in bytes, 427so its minimum value (or the offset of the first string) 428is always 4 on 32-bit machines. 429.Sh SEE ALSO 430.Xr as 1 , 431.Xr gdb 1 Pq Pa ports/devel/gdb , 432.Xr ld 1 , 433.Xr brk 2 , 434.Xr execve 2 , 435.Xr nlist 3 , 436.Xr core 5 , 437.Xr elf 5 , 438.Xr link 5 , 439.Xr stab 5 440.Sh HISTORY 441The 442.In a.out.h 443include file appeared in 444.At v7 . 445.Sh BUGS 446Since not all of the supported architectures use the 447.Fa a_midmag 448field, 449it can be difficult to determine what 450architecture a binary will execute on 451without examining its actual machine code. 452Even with a machine identifier, 453the byte order of the 454.Fa exec 455header is machine-dependent. 456