1 This is ld.info, produced by makeinfo version 4.8 from ld.texinfo.
3 INFO-DIR-SECTION Software development
5 * Ld: (ld). The GNU linker.
8 This file documents the GNU linker LD (GNU Binutils) version 2.21.1.
10 Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
11 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 Free
12 Software Foundation, Inc.
14 Permission is granted to copy, distribute and/or modify this document
15 under the terms of the GNU Free Documentation License, Version 1.3 or
16 any later version published by the Free Software Foundation; with no
17 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
18 Texts. A copy of the license is included in the section entitled "GNU
19 Free Documentation License".
22 File: ld.info, Node: Top, Next: Overview, Up: (dir)
27 This file documents the GNU linker ld (GNU Binutils) version 2.21.1.
29 This document is distributed under the terms of the GNU Free
30 Documentation License version 1.3. A copy of the license is included
31 in the section entitled "GNU Free Documentation License".
36 * Invocation:: Invocation
37 * Scripts:: Linker Scripts
39 * Machine Dependent:: Machine Dependent Features
43 * Reporting Bugs:: Reporting Bugs
44 * MRI:: MRI Compatible Script Files
45 * GNU Free Documentation License:: GNU Free Documentation License
49 File: ld.info, Node: Overview, Next: Invocation, Prev: Top, Up: Top
54 `ld' combines a number of object and archive files, relocates their
55 data and ties up symbol references. Usually the last step in compiling
56 a program is to run `ld'.
58 `ld' accepts Linker Command Language files written in a superset of
59 AT&T's Link Editor Command Language syntax, to provide explicit and
60 total control over the linking process.
62 This version of `ld' uses the general purpose BFD libraries to
63 operate on object files. This allows `ld' to read, combine, and write
64 object files in many different formats--for example, COFF or `a.out'.
65 Different formats may be linked together to produce any available kind
66 of object file. *Note BFD::, for more information.
68 Aside from its flexibility, the GNU linker is more helpful than other
69 linkers in providing diagnostic information. Many linkers abandon
70 execution immediately upon encountering an error; whenever possible,
71 `ld' continues executing, allowing you to identify other errors (or, in
72 some cases, to get an output file in spite of the error).
75 File: ld.info, Node: Invocation, Next: Scripts, Prev: Overview, Up: Top
80 The GNU linker `ld' is meant to cover a broad range of situations, and
81 to be as compatible as possible with other linkers. As a result, you
82 have many choices to control its behavior.
86 * Options:: Command Line Options
87 * Environment:: Environment Variables
90 File: ld.info, Node: Options, Next: Environment, Up: Invocation
92 2.1 Command Line Options
93 ========================
95 The linker supports a plethora of command-line options, but in actual
96 practice few of them are used in any particular context. For instance,
97 a frequent use of `ld' is to link standard Unix object files on a
98 standard, supported Unix system. On such a system, to link a file
101 ld -o OUTPUT /lib/crt0.o hello.o -lc
103 This tells `ld' to produce a file called OUTPUT as the result of
104 linking the file `/lib/crt0.o' with `hello.o' and the library `libc.a',
105 which will come from the standard search directories. (See the
106 discussion of the `-l' option below.)
108 Some of the command-line options to `ld' may be specified at any
109 point in the command line. However, options which refer to files, such
110 as `-l' or `-T', cause the file to be read at the point at which the
111 option appears in the command line, relative to the object files and
112 other file options. Repeating non-file options with a different
113 argument will either have no further effect, or override prior
114 occurrences (those further to the left on the command line) of that
115 option. Options which may be meaningfully specified more than once are
116 noted in the descriptions below.
118 Non-option arguments are object files or archives which are to be
119 linked together. They may follow, precede, or be mixed in with
120 command-line options, except that an object file argument may not be
121 placed between an option and its argument.
123 Usually the linker is invoked with at least one object file, but you
124 can specify other forms of binary input files using `-l', `-R', and the
125 script command language. If _no_ binary input files at all are
126 specified, the linker does not produce any output, and issues the
127 message `No input files'.
129 If the linker cannot recognize the format of an object file, it will
130 assume that it is a linker script. A script specified in this way
131 augments the main linker script used for the link (either the default
132 linker script or the one specified by using `-T'). This feature
133 permits the linker to link against a file which appears to be an object
134 or an archive, but actually merely defines some symbol values, or uses
135 `INPUT' or `GROUP' to load other objects. Specifying a script in this
136 way merely augments the main linker script, with the extra commands
137 placed after the main script; use the `-T' option to replace the
138 default linker script entirely, but note the effect of the `INSERT'
139 command. *Note Scripts::.
141 For options whose names are a single letter, option arguments must
142 either follow the option letter without intervening whitespace, or be
143 given as separate arguments immediately following the option that
146 For options whose names are multiple letters, either one dash or two
147 can precede the option name; for example, `-trace-symbol' and
148 `--trace-symbol' are equivalent. Note--there is one exception to this
149 rule. Multiple letter options that start with a lower case 'o' can
150 only be preceded by two dashes. This is to reduce confusion with the
151 `-o' option. So for example `-omagic' sets the output file name to
152 `magic' whereas `--omagic' sets the NMAGIC flag on the output.
154 Arguments to multiple-letter options must either be separated from
155 the option name by an equals sign, or be given as separate arguments
156 immediately following the option that requires them. For example,
157 `--trace-symbol foo' and `--trace-symbol=foo' are equivalent. Unique
158 abbreviations of the names of multiple-letter options are accepted.
160 Note--if the linker is being invoked indirectly, via a compiler
161 driver (e.g. `gcc') then all the linker command line options should be
162 prefixed by `-Wl,' (or whatever is appropriate for the particular
163 compiler driver) like this:
165 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
167 This is important, because otherwise the compiler driver program may
168 silently drop the linker options, resulting in a bad link. Confusion
169 may also arise when passing options that require values through a
170 driver, as the use of a space between option and argument acts as a
171 separator, and causes the driver to pass only the option to the linker
172 and the argument to the compiler. In this case, it is simplest to use
173 the joined forms of both single- and multiple-letter options, such as:
175 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
177 Here is a table of the generic command line switches accepted by the
181 Read command-line options from FILE. The options read are
182 inserted in place of the original @FILE option. If FILE does not
183 exist, or cannot be read, then the option will be treated
184 literally, and not removed.
186 Options in FILE are separated by whitespace. A whitespace
187 character may be included in an option by surrounding the entire
188 option in either single or double quotes. Any character
189 (including a backslash) may be included by prefixing the character
190 to be included with a backslash. The FILE may itself contain
191 additional @FILE options; any such options will be processed
195 This option is supported for HP/UX compatibility. The KEYWORD
196 argument must be one of the strings `archive', `shared', or
197 `default'. `-aarchive' is functionally equivalent to `-Bstatic',
198 and the other two keywords are functionally equivalent to
199 `-Bdynamic'. This option may be used any number of times.
202 Adds AUDITLIB to the `DT_AUDIT' entry of the dynamic section.
203 AUDITLIB is not checked for existence, nor will it use the
204 DT_SONAME specified in the library. If specified multiple times
205 `DT_AUDIT' will contain a colon separated list of audit interfaces
206 to use. If the linker finds an object with an audit entry while
207 searching for shared libraries, it will add a corresponding
208 `DT_DEPAUDIT' entry in the output file. This option is only
209 meaningful on ELF platforms supporting the rtld-audit interface.
212 `--architecture=ARCHITECTURE'
213 In the current release of `ld', this option is useful only for the
214 Intel 960 family of architectures. In that `ld' configuration, the
215 ARCHITECTURE argument identifies the particular architecture in
216 the 960 family, enabling some safeguards and modifying the
217 archive-library search path. *Note `ld' and the Intel 960 family:
220 Future releases of `ld' may support similar functionality for
221 other architecture families.
224 `--format=INPUT-FORMAT'
225 `ld' may be configured to support more than one kind of object
226 file. If your `ld' is configured this way, you can use the `-b'
227 option to specify the binary format for input object files that
228 follow this option on the command line. Even when `ld' is
229 configured to support alternative object formats, you don't
230 usually need to specify this, as `ld' should be configured to
231 expect as a default input format the most usual format on each
232 machine. INPUT-FORMAT is a text string, the name of a particular
233 format supported by the BFD libraries. (You can list the
234 available binary formats with `objdump -i'.) *Note BFD::.
236 You may want to use this option if you are linking files with an
237 unusual binary format. You can also use `-b' to switch formats
238 explicitly (when linking object files of different formats), by
239 including `-b INPUT-FORMAT' before each group of object files in a
242 The default format is taken from the environment variable
243 `GNUTARGET'. *Note Environment::. You can also define the input
244 format from a script, using the command `TARGET'; see *Note Format
248 `--mri-script=MRI-COMMANDFILE'
249 For compatibility with linkers produced by MRI, `ld' accepts script
250 files written in an alternate, restricted command language,
251 described in *Note MRI Compatible Script Files: MRI. Introduce
252 MRI script files with the option `-c'; use the `-T' option to run
253 linker scripts written in the general-purpose `ld' scripting
254 language. If MRI-CMDFILE does not exist, `ld' looks for it in the
255 directories specified by any `-L' options.
260 These three options are equivalent; multiple forms are supported
261 for compatibility with other linkers. They assign space to common
262 symbols even if a relocatable output file is specified (with
263 `-r'). The script command `FORCE_COMMON_ALLOCATION' has the same
264 effect. *Note Miscellaneous Commands::.
266 `--depaudit AUDITLIB'
268 Adds AUDITLIB to the `DT_DEPAUDIT' entry of the dynamic section.
269 AUDITLIB is not checked for existence, nor will it use the
270 DT_SONAME specified in the library. If specified multiple times
271 `DT_DEPAUDIT' will contain a colon separated list of audit
272 interfaces to use. This option is only meaningful on ELF
273 platforms supporting the rtld-audit interface. The -P option is
274 provided for Solaris compatibility.
278 Use ENTRY as the explicit symbol for beginning execution of your
279 program, rather than the default entry point. If there is no
280 symbol named ENTRY, the linker will try to parse ENTRY as a number,
281 and use that as the entry address (the number will be interpreted
282 in base 10; you may use a leading `0x' for base 16, or a leading
283 `0' for base 8). *Note Entry Point::, for a discussion of defaults
284 and other ways of specifying the entry point.
286 `--exclude-libs LIB,LIB,...'
287 Specifies a list of archive libraries from which symbols should
288 not be automatically exported. The library names may be delimited
289 by commas or colons. Specifying `--exclude-libs ALL' excludes
290 symbols in all archive libraries from automatic export. This
291 option is available only for the i386 PE targeted port of the
292 linker and for ELF targeted ports. For i386 PE, symbols
293 explicitly listed in a .def file are still exported, regardless of
294 this option. For ELF targeted ports, symbols affected by this
295 option will be treated as hidden.
297 `--exclude-modules-for-implib MODULE,MODULE,...'
298 Specifies a list of object files or archive members, from which
299 symbols should not be automatically exported, but which should be
300 copied wholesale into the import library being generated during
301 the link. The module names may be delimited by commas or colons,
302 and must match exactly the filenames used by `ld' to open the
303 files; for archive members, this is simply the member name, but
304 for object files the name listed must include and match precisely
305 any path used to specify the input file on the linker's
306 command-line. This option is available only for the i386 PE
307 targeted port of the linker. Symbols explicitly listed in a .def
308 file are still exported, regardless of this option.
312 `--no-export-dynamic'
313 When creating a dynamically linked executable, using the `-E'
314 option or the `--export-dynamic' option causes the linker to add
315 all symbols to the dynamic symbol table. The dynamic symbol table
316 is the set of symbols which are visible from dynamic objects at
319 If you do not use either of these options (or use the
320 `--no-export-dynamic' option to restore the default behavior), the
321 dynamic symbol table will normally contain only those symbols
322 which are referenced by some dynamic object mentioned in the link.
324 If you use `dlopen' to load a dynamic object which needs to refer
325 back to the symbols defined by the program, rather than some other
326 dynamic object, then you will probably need to use this option when
327 linking the program itself.
329 You can also use the dynamic list to control what symbols should
330 be added to the dynamic symbol table if the output format supports
331 it. See the description of `--dynamic-list'.
333 Note that this option is specific to ELF targeted ports. PE
334 targets support a similar function to export all symbols from a
335 DLL or EXE; see the description of `--export-all-symbols' below.
338 Link big-endian objects. This affects the default output format.
341 Link little-endian objects. This affects the default output
346 When creating an ELF shared object, set the internal DT_AUXILIARY
347 field to the specified name. This tells the dynamic linker that
348 the symbol table of the shared object should be used as an
349 auxiliary filter on the symbol table of the shared object NAME.
351 If you later link a program against this filter object, then, when
352 you run the program, the dynamic linker will see the DT_AUXILIARY
353 field. If the dynamic linker resolves any symbols from the filter
354 object, it will first check whether there is a definition in the
355 shared object NAME. If there is one, it will be used instead of
356 the definition in the filter object. The shared object NAME need
357 not exist. Thus the shared object NAME may be used to provide an
358 alternative implementation of certain functions, perhaps for
359 debugging or for machine specific performance.
361 This option may be specified more than once. The DT_AUXILIARY
362 entries will be created in the order in which they appear on the
367 When creating an ELF shared object, set the internal DT_FILTER
368 field to the specified name. This tells the dynamic linker that
369 the symbol table of the shared object which is being created
370 should be used as a filter on the symbol table of the shared
373 If you later link a program against this filter object, then, when
374 you run the program, the dynamic linker will see the DT_FILTER
375 field. The dynamic linker will resolve symbols according to the
376 symbol table of the filter object as usual, but it will actually
377 link to the definitions found in the shared object NAME. Thus the
378 filter object can be used to select a subset of the symbols
379 provided by the object NAME.
381 Some older linkers used the `-F' option throughout a compilation
382 toolchain for specifying object-file format for both input and
383 output object files. The GNU linker uses other mechanisms for
384 this purpose: the `-b', `--format', `--oformat' options, the
385 `TARGET' command in linker scripts, and the `GNUTARGET'
386 environment variable. The GNU linker will ignore the `-F' option
387 when not creating an ELF shared object.
390 When creating an ELF executable or shared object, call NAME when
391 the executable or shared object is unloaded, by setting DT_FINI to
392 the address of the function. By default, the linker uses `_fini'
393 as the function to call.
396 Ignored. Provided for compatibility with other tools.
400 Set the maximum size of objects to be optimized using the GP
401 register to SIZE. This is only meaningful for object file formats
402 such as MIPS ECOFF which supports putting large and small objects
403 into different sections. This is ignored for other object file
408 When creating an ELF shared object, set the internal DT_SONAME
409 field to the specified name. When an executable is linked with a
410 shared object which has a DT_SONAME field, then when the
411 executable is run the dynamic linker will attempt to load the
412 shared object specified by the DT_SONAME field rather than the
413 using the file name given to the linker.
416 Perform an incremental link (same as option `-r').
419 When creating an ELF executable or shared object, call NAME when
420 the executable or shared object is loaded, by setting DT_INIT to
421 the address of the function. By default, the linker uses `_init'
422 as the function to call.
426 Add the archive or object file specified by NAMESPEC to the list
427 of files to link. This option may be used any number of times.
428 If NAMESPEC is of the form `:FILENAME', `ld' will search the
429 library path for a file called FILENAME, otherwise it will search
430 the library path for a file called `libNAMESPEC.a'.
432 On systems which support shared libraries, `ld' may also search for
433 files other than `libNAMESPEC.a'. Specifically, on ELF and SunOS
434 systems, `ld' will search a directory for a library called
435 `libNAMESPEC.so' before searching for one called `libNAMESPEC.a'.
436 (By convention, a `.so' extension indicates a shared library.)
437 Note that this behavior does not apply to `:FILENAME', which
438 always specifies a file called FILENAME.
440 The linker will search an archive only once, at the location where
441 it is specified on the command line. If the archive defines a
442 symbol which was undefined in some object which appeared before
443 the archive on the command line, the linker will include the
444 appropriate file(s) from the archive. However, an undefined
445 symbol in an object appearing later on the command line will not
446 cause the linker to search the archive again.
448 See the `-(' option for a way to force the linker to search
449 archives multiple times.
451 You may list the same archive multiple times on the command line.
453 This type of archive searching is standard for Unix linkers.
454 However, if you are using `ld' on AIX, note that it is different
455 from the behaviour of the AIX linker.
458 `--library-path=SEARCHDIR'
459 Add path SEARCHDIR to the list of paths that `ld' will search for
460 archive libraries and `ld' control scripts. You may use this
461 option any number of times. The directories are searched in the
462 order in which they are specified on the command line.
463 Directories specified on the command line are searched before the
464 default directories. All `-L' options apply to all `-l' options,
465 regardless of the order in which the options appear. `-L' options
466 do not affect how `ld' searches for a linker script unless `-T'
469 If SEARCHDIR begins with `=', then the `=' will be replaced by the
470 "sysroot prefix", a path specified when the linker is configured.
472 The default set of paths searched (without being specified with
473 `-L') depends on which emulation mode `ld' is using, and in some
474 cases also on how it was configured. *Note Environment::.
476 The paths can also be specified in a link script with the
477 `SEARCH_DIR' command. Directories specified this way are searched
478 at the point in which the linker script appears in the command
482 Emulate the EMULATION linker. You can list the available
483 emulations with the `--verbose' or `-V' options.
485 If the `-m' option is not used, the emulation is taken from the
486 `LDEMULATION' environment variable, if that is defined.
488 Otherwise, the default emulation depends upon how the linker was
493 Print a link map to the standard output. A link map provides
494 information about the link, including the following:
496 * Where object files are mapped into memory.
498 * How common symbols are allocated.
500 * All archive members included in the link, with a mention of
501 the symbol which caused the archive member to be brought in.
503 * The values assigned to symbols.
505 Note - symbols whose values are computed by an expression
506 which involves a reference to a previous value of the same
507 symbol may not have correct result displayed in the link map.
508 This is because the linker discards intermediate results and
509 only retains the final value of an expression. Under such
510 circumstances the linker will display the final value
511 enclosed by square brackets. Thus for example a linker
518 will produce the following output in the link map if the `-M'
522 [0x0000000c] foo = (foo * 0x4)
523 [0x0000000c] foo = (foo + 0x8)
525 See *Note Expressions:: for more information about
526 expressions in linker scripts.
530 Turn off page alignment of sections, and disable linking against
531 shared libraries. If the output format supports Unix style magic
532 numbers, mark the output as `NMAGIC'.
536 Set the text and data sections to be readable and writable. Also,
537 do not page-align the data segment, and disable linking against
538 shared libraries. If the output format supports Unix style magic
539 numbers, mark the output as `OMAGIC'. Note: Although a writable
540 text section is allowed for PE-COFF targets, it does not conform
541 to the format specification published by Microsoft.
544 This option negates most of the effects of the `-N' option. It
545 sets the text section to be read-only, and forces the data segment
546 to be page-aligned. Note - this option does not enable linking
547 against shared libraries. Use `-Bdynamic' for this.
551 Use OUTPUT as the name for the program produced by `ld'; if this
552 option is not specified, the name `a.out' is used by default. The
553 script command `OUTPUT' can also specify the output file name.
556 If LEVEL is a numeric values greater than zero `ld' optimizes the
557 output. This might take significantly longer and therefore
558 probably should only be enabled for the final binary. At the
559 moment this option only affects ELF shared library generation.
560 Future releases of the linker may make more use of this option.
561 Also currently there is no difference in the linker's behaviour
562 for different non-zero values of this option. Again this may
563 change with future releases.
567 Leave relocation sections and contents in fully linked executables.
568 Post link analysis and optimization tools may need this
569 information in order to perform correct modifications of
570 executables. This results in larger executables.
572 This option is currently only supported on ELF platforms.
575 Force the output file to have dynamic sections. This option is
576 specific to VxWorks targets.
580 Generate relocatable output--i.e., generate an output file that
581 can in turn serve as input to `ld'. This is often called "partial
582 linking". As a side effect, in environments that support standard
583 Unix magic numbers, this option also sets the output file's magic
584 number to `OMAGIC'. If this option is not specified, an absolute
585 file is produced. When linking C++ programs, this option _will
586 not_ resolve references to constructors; to do that, use `-Ur'.
588 When an input file does not have the same format as the output
589 file, partial linking is only supported if that input file does
590 not contain any relocations. Different output formats can have
591 further restrictions; for example some `a.out'-based formats do
592 not support partial linking with input files in other formats at
595 This option does the same thing as `-i'.
598 `--just-symbols=FILENAME'
599 Read symbol names and their addresses from FILENAME, but do not
600 relocate it or include it in the output. This allows your output
601 file to refer symbolically to absolute locations of memory defined
602 in other programs. You may use this option more than once.
604 For compatibility with other ELF linkers, if the `-R' option is
605 followed by a directory name, rather than a file name, it is
606 treated as the `-rpath' option.
610 Omit all symbol information from the output file.
614 Omit debugger symbol information (but not all symbols) from the
619 Print the names of the input files as `ld' processes them.
622 `--script=SCRIPTFILE'
623 Use SCRIPTFILE as the linker script. This script replaces `ld''s
624 default linker script (rather than adding to it), so COMMANDFILE
625 must specify everything necessary to describe the output file.
626 *Note Scripts::. If SCRIPTFILE does not exist in the current
627 directory, `ld' looks for it in the directories specified by any
628 preceding `-L' options. Multiple `-T' options accumulate.
631 `--default-script=SCRIPTFILE'
632 Use SCRIPTFILE as the default linker script. *Note Scripts::.
634 This option is similar to the `--script' option except that
635 processing of the script is delayed until after the rest of the
636 command line has been processed. This allows options placed after
637 the `--default-script' option on the command line to affect the
638 behaviour of the linker script, which can be important when the
639 linker command line cannot be directly controlled by the user.
640 (eg because the command line is being constructed by another tool,
645 Force SYMBOL to be entered in the output file as an undefined
646 symbol. Doing this may, for example, trigger linking of additional
647 modules from standard libraries. `-u' may be repeated with
648 different option arguments to enter additional undefined symbols.
649 This option is equivalent to the `EXTERN' linker script command.
652 For anything other than C++ programs, this option is equivalent to
653 `-r': it generates relocatable output--i.e., an output file that
654 can in turn serve as input to `ld'. When linking C++ programs,
655 `-Ur' _does_ resolve references to constructors, unlike `-r'. It
656 does not work to use `-Ur' on files that were themselves linked
657 with `-Ur'; once the constructor table has been built, it cannot
658 be added to. Use `-Ur' only for the last partial link, and `-r'
662 Creates a separate output section for every input section matching
663 SECTION, or if the optional wildcard SECTION argument is missing,
664 for every orphan input section. An orphan section is one not
665 specifically mentioned in a linker script. You may use this option
666 multiple times on the command line; It prevents the normal
667 merging of input sections with the same name, overriding output
668 section assignments in a linker script.
673 Display the version number for `ld'. The `-V' option also lists
674 the supported emulations.
678 Delete all local symbols.
682 Delete all temporary local symbols. (These symbols start with
683 system-specific local label prefixes, typically `.L' for ELF
684 systems or `L' for traditional a.out systems.)
687 `--trace-symbol=SYMBOL'
688 Print the name of each linked file in which SYMBOL appears. This
689 option may be given any number of times. On many systems it is
690 necessary to prepend an underscore.
692 This option is useful when you have an undefined symbol in your
693 link but don't know where the reference is coming from.
696 Add PATH to the default library search path. This option exists
697 for Solaris compatibility.
700 The recognized keywords are:
702 Combines multiple reloc sections and sorts them to make
703 dynamic symbol lookup caching possible.
706 Disallows undefined symbols in object files. Undefined
707 symbols in shared libraries are still allowed.
710 Marks the object as requiring executable stack.
713 This option is only meaningful when building a shared object.
714 It marks the object so that its runtime initialization will
715 occur before the runtime initialization of any other objects
716 brought into the process at the same time. Similarly the
717 runtime finalization of the object will occur after the
718 runtime finalization of any other objects.
721 Marks the object that its symbol table interposes before all
722 symbols but the primary executable.
725 When generating an executable or shared library, mark it to
726 tell the dynamic linker to defer function call resolution to
727 the point when the function is called (lazy binding), rather
728 than at load time. Lazy binding is the default.
731 Marks the object that its filters be processed immediately at
735 Allows multiple definitions.
738 Disables multiple reloc sections combining.
741 Disables production of copy relocs.
744 Marks the object that the search for dependencies of this
745 object will ignore any default library search paths.
748 Marks the object shouldn't be unloaded at runtime.
751 Marks the object not available to `dlopen'.
754 Marks the object can not be dumped by `dldump'.
757 Marks the object as not requiring executable stack.
760 Don't create an ELF `PT_GNU_RELRO' segment header in the
764 When generating an executable or shared library, mark it to
765 tell the dynamic linker to resolve all symbols when the
766 program is started, or when the shared library is linked to
767 using dlopen, instead of deferring function call resolution
768 to the point when the function is first called.
771 Marks the object may contain $ORIGIN.
774 Create an ELF `PT_GNU_RELRO' segment header in the object.
776 `max-page-size=VALUE'
777 Set the emulation maximum page size to VALUE.
779 `common-page-size=VALUE'
780 Set the emulation common page size to VALUE.
783 Other keywords are ignored for Solaris compatibility.
786 `--start-group ARCHIVES --end-group'
787 The ARCHIVES should be a list of archive files. They may be
788 either explicit file names, or `-l' options.
790 The specified archives are searched repeatedly until no new
791 undefined references are created. Normally, an archive is
792 searched only once in the order that it is specified on the
793 command line. If a symbol in that archive is needed to resolve an
794 undefined symbol referred to by an object in an archive that
795 appears later on the command line, the linker would not be able to
796 resolve that reference. By grouping the archives, they all be
797 searched repeatedly until all possible references are resolved.
799 Using this option has a significant performance cost. It is best
800 to use it only when there are unavoidable circular references
801 between two or more archives.
803 `--accept-unknown-input-arch'
804 `--no-accept-unknown-input-arch'
805 Tells the linker to accept input files whose architecture cannot be
806 recognised. The assumption is that the user knows what they are
807 doing and deliberately wants to link in these unknown input files.
808 This was the default behaviour of the linker, before release
809 2.14. The default behaviour from release 2.14 onwards is to
810 reject such input files, and so the `--accept-unknown-input-arch'
811 option has been added to restore the old behaviour.
815 This option affects ELF DT_NEEDED tags for dynamic libraries
816 mentioned on the command line after the `--as-needed' option.
817 Normally the linker will add a DT_NEEDED tag for each dynamic
818 library mentioned on the command line, regardless of whether the
819 library is actually needed or not. `--as-needed' causes a
820 DT_NEEDED tag to only be emitted for a library that satisfies an
821 undefined symbol reference from a regular object file or, if the
822 library is not found in the DT_NEEDED lists of other libraries
823 linked up to that point, an undefined symbol reference from
824 another dynamic library. `--no-as-needed' restores the default
829 These two options have been deprecated because of the similarity of
830 their names to the `--as-needed' and `--no-as-needed' options.
831 They have been replaced by `--copy-dt-needed-entries' and
832 `--no-copy-dt-needed-entries'.
835 This option is ignored for SunOS compatibility.
840 Link against dynamic libraries. This is only meaningful on
841 platforms for which shared libraries are supported. This option
842 is normally the default on such platforms. The different variants
843 of this option are for compatibility with various systems. You
844 may use this option multiple times on the command line: it affects
845 library searching for `-l' options which follow it.
848 Set the `DF_1_GROUP' flag in the `DT_FLAGS_1' entry in the dynamic
849 section. This causes the runtime linker to handle lookups in this
850 object and its dependencies to be performed only inside the group.
851 `--unresolved-symbols=report-all' is implied. This option is only
852 meaningful on ELF platforms which support shared libraries.
858 Do not link against shared libraries. This is only meaningful on
859 platforms for which shared libraries are supported. The different
860 variants of this option are for compatibility with various
861 systems. You may use this option multiple times on the command
862 line: it affects library searching for `-l' options which follow
863 it. This option also implies `--unresolved-symbols=report-all'.
864 This option can be used with `-shared'. Doing so means that a
865 shared library is being created but that all of the library's
866 external references must be resolved by pulling in entries from
870 When creating a shared library, bind references to global symbols
871 to the definition within the shared library, if any. Normally, it
872 is possible for a program linked against a shared library to
873 override the definition within the shared library. This option is
874 only meaningful on ELF platforms which support shared libraries.
876 `-Bsymbolic-functions'
877 When creating a shared library, bind references to global function
878 symbols to the definition within the shared library, if any. This
879 option is only meaningful on ELF platforms which support shared
882 `--dynamic-list=DYNAMIC-LIST-FILE'
883 Specify the name of a dynamic list file to the linker. This is
884 typically used when creating shared libraries to specify a list of
885 global symbols whose references shouldn't be bound to the
886 definition within the shared library, or creating dynamically
887 linked executables to specify a list of symbols which should be
888 added to the symbol table in the executable. This option is only
889 meaningful on ELF platforms which support shared libraries.
891 The format of the dynamic list is the same as the version node
892 without scope and node name. See *Note VERSION:: for more
895 `--dynamic-list-data'
896 Include all global data symbols to the dynamic list.
898 `--dynamic-list-cpp-new'
899 Provide the builtin dynamic list for C++ operator new and delete.
900 It is mainly useful for building shared libstdc++.
902 `--dynamic-list-cpp-typeinfo'
903 Provide the builtin dynamic list for C++ runtime type
907 `--no-check-sections'
908 Asks the linker _not_ to check section addresses after they have
909 been assigned to see if there are any overlaps. Normally the
910 linker will perform this check, and if it finds any overlaps it
911 will produce suitable error messages. The linker does know about,
912 and does make allowances for sections in overlays. The default
913 behaviour can be restored by using the command line switch
914 `--check-sections'. Section overlap is not usually checked for
915 relocatable links. You can force checking in that case by using
916 the `--check-sections' option.
918 `--copy-dt-needed-entries'
919 `--no-copy-dt-needed-entries'
920 This option affects the treatment of dynamic libraries referred to
921 by DT_NEEDED tags _inside_ ELF dynamic libraries mentioned on the
922 command line. Normally the linker will add a DT_NEEDED tag to the
923 output binary for each library mentioned in a DT_NEEDED tag in an
924 input dynamic library. With `--no-copy-dt-needed-entries'
925 specified on the command line however any dynamic libraries that
926 follow it will have their DT_NEEDED entries ignored. The default
927 behaviour can be restored with `--copy-dt-needed-entries'.
929 This option also has an effect on the resolution of symbols in
930 dynamic libraries. With the default setting dynamic libraries
931 mentioned on the command line will be recursively searched,
932 following their DT_NEEDED tags to other libraries, in order to
933 resolve symbols required by the output binary. With
934 `--no-copy-dt-needed-entries' specified however the searching of
935 dynamic libraries that follow it will stop with the dynamic
936 library itself. No DT_NEEDED links will be traversed to resolve
940 Output a cross reference table. If a linker map file is being
941 generated, the cross reference table is printed to the map file.
942 Otherwise, it is printed on the standard output.
944 The format of the table is intentionally simple, so that it may be
945 easily processed by a script if necessary. The symbols are
946 printed out, sorted by name. For each symbol, a list of file
947 names is given. If the symbol is defined, the first file listed
948 is the location of the definition. The remaining files contain
949 references to the symbol.
952 This option inhibits the assignment of addresses to common symbols.
953 The script command `INHIBIT_COMMON_ALLOCATION' has the same effect.
954 *Note Miscellaneous Commands::.
956 The `--no-define-common' option allows decoupling the decision to
957 assign addresses to Common symbols from the choice of the output
958 file type; otherwise a non-Relocatable output type forces
959 assigning addresses to Common symbols. Using `--no-define-common'
960 allows Common symbols that are referenced from a shared library to
961 be assigned addresses only in the main program. This eliminates
962 the unused duplicate space in the shared library, and also
963 prevents any possible confusion over resolving to the wrong
964 duplicate when there are many dynamic modules with specialized
965 search paths for runtime symbol resolution.
967 `--defsym=SYMBOL=EXPRESSION'
968 Create a global symbol in the output file, containing the absolute
969 address given by EXPRESSION. You may use this option as many
970 times as necessary to define multiple symbols in the command line.
971 A limited form of arithmetic is supported for the EXPRESSION in
972 this context: you may give a hexadecimal constant or the name of
973 an existing symbol, or use `+' and `-' to add or subtract
974 hexadecimal constants or symbols. If you need more elaborate
975 expressions, consider using the linker command language from a
976 script (*note Assignment: Symbol Definitions: Assignments.).
977 _Note:_ there should be no white space between SYMBOL, the equals
978 sign ("<=>"), and EXPRESSION.
982 These options control whether to demangle symbol names in error
983 messages and other output. When the linker is told to demangle,
984 it tries to present symbol names in a readable fashion: it strips
985 leading underscores if they are used by the object file format,
986 and converts C++ mangled symbol names into user readable names.
987 Different compilers have different mangling styles. The optional
988 demangling style argument can be used to choose an appropriate
989 demangling style for your compiler. The linker will demangle by
990 default unless the environment variable `COLLECT_NO_DEMANGLE' is
991 set. These options may be used to override the default.
994 `--dynamic-linker=FILE'
995 Set the name of the dynamic linker. This is only meaningful when
996 generating dynamically linked ELF executables. The default dynamic
997 linker is normally correct; don't use this unless you know what
1001 `--no-fatal-warnings'
1002 Treat all warnings as errors. The default behaviour can be
1003 restored with the option `--no-fatal-warnings'.
1005 `--force-exe-suffix'
1006 Make sure that an output file has a .exe suffix.
1008 If a successfully built fully linked output file does not have a
1009 `.exe' or `.dll' suffix, this option forces the linker to copy the
1010 output file to one of the same name with a `.exe' suffix. This
1011 option is useful when using unmodified Unix makefiles on a
1012 Microsoft Windows host, since some versions of Windows won't run
1013 an image unless it ends in a `.exe' suffix.
1017 Enable garbage collection of unused input sections. It is ignored
1018 on targets that do not support this option. The default behaviour
1019 (of not performing this garbage collection) can be restored by
1020 specifying `--no-gc-sections' on the command line.
1022 `--gc-sections' decides which input sections are used by examining
1023 symbols and relocations. The section containing the entry symbol
1024 and all sections containing symbols undefined on the command-line
1025 will be kept, as will sections containing symbols referenced by
1026 dynamic objects. Note that when building shared libraries, the
1027 linker must assume that any visible symbol is referenced. Once
1028 this initial set of sections has been determined, the linker
1029 recursively marks as used any section referenced by their
1030 relocations. See `--entry' and `--undefined'.
1032 This option can be set when doing a partial link (enabled with
1033 option `-r'). In this case the root of symbols kept must be
1034 explicitly specified either by an `--entry' or `--undefined'
1035 option or by a `ENTRY' command in the linker script.
1037 `--print-gc-sections'
1038 `--no-print-gc-sections'
1039 List all sections removed by garbage collection. The listing is
1040 printed on stderr. This option is only effective if garbage
1041 collection has been enabled via the `--gc-sections') option. The
1042 default behaviour (of not listing the sections that are removed)
1043 can be restored by specifying `--no-print-gc-sections' on the
1047 Print a summary of the command-line options on the standard output
1051 Print a summary of all target specific options on the standard
1055 Print a link map to the file MAPFILE. See the description of the
1059 `ld' normally optimizes for speed over memory usage by caching the
1060 symbol tables of input files in memory. This option tells `ld' to
1061 instead optimize for memory usage, by rereading the symbol tables
1062 as necessary. This may be required if `ld' runs out of memory
1063 space while linking a large executable.
1067 Report unresolved symbol references from regular object files.
1068 This is done even if the linker is creating a non-symbolic shared
1069 library. The switch `--[no-]allow-shlib-undefined' controls the
1070 behaviour for reporting unresolved references found in shared
1071 libraries being linked in.
1073 `--allow-multiple-definition'
1075 Normally when a symbol is defined multiple times, the linker will
1076 report a fatal error. These options allow multiple definitions and
1077 the first definition will be used.
1079 `--allow-shlib-undefined'
1080 `--no-allow-shlib-undefined'
1081 Allows or disallows undefined symbols in shared libraries. This
1082 switch is similar to `--no-undefined' except that it determines
1083 the behaviour when the undefined symbols are in a shared library
1084 rather than a regular object file. It does not affect how
1085 undefined symbols in regular object files are handled.
1087 The default behaviour is to report errors for any undefined symbols
1088 referenced in shared libraries if the linker is being used to
1089 create an executable, but to allow them if the linker is being
1090 used to create a shared library.
1092 The reasons for allowing undefined symbol references in shared
1093 libraries specified at link time are that:
1095 * A shared library specified at link time may not be the same
1096 as the one that is available at load time, so the symbol
1097 might actually be resolvable at load time.
1099 * There are some operating systems, eg BeOS and HPPA, where
1100 undefined symbols in shared libraries are normal.
1102 The BeOS kernel for example patches shared libraries at load
1103 time to select whichever function is most appropriate for the
1104 current architecture. This is used, for example, to
1105 dynamically select an appropriate memset function.
1107 `--no-undefined-version'
1108 Normally when a symbol has an undefined version, the linker will
1109 ignore it. This option disallows symbols with undefined version
1110 and a fatal error will be issued instead.
1113 Create and use a default symbol version (the soname) for
1114 unversioned exported symbols.
1116 `--default-imported-symver'
1117 Create and use a default symbol version (the soname) for
1118 unversioned imported symbols.
1120 `--no-warn-mismatch'
1121 Normally `ld' will give an error if you try to link together input
1122 files that are mismatched for some reason, perhaps because they
1123 have been compiled for different processors or for different
1124 endiannesses. This option tells `ld' that it should silently
1125 permit such possible errors. This option should only be used with
1126 care, in cases when you have taken some special action that
1127 ensures that the linker errors are inappropriate.
1129 `--no-warn-search-mismatch'
1130 Normally `ld' will give a warning if it finds an incompatible
1131 library during a library search. This option silences the warning.
1133 `--no-whole-archive'
1134 Turn off the effect of the `--whole-archive' option for subsequent
1138 Retain the executable output file whenever it is still usable.
1139 Normally, the linker will not produce an output file if it
1140 encounters errors during the link process; it exits without
1141 writing an output file when it issues any error whatsoever.
1144 Only search library directories explicitly specified on the
1145 command line. Library directories specified in linker scripts
1146 (including linker scripts specified on the command line) are
1149 `--oformat=OUTPUT-FORMAT'
1150 `ld' may be configured to support more than one kind of object
1151 file. If your `ld' is configured this way, you can use the
1152 `--oformat' option to specify the binary format for the output
1153 object file. Even when `ld' is configured to support alternative
1154 object formats, you don't usually need to specify this, as `ld'
1155 should be configured to produce as a default output format the most
1156 usual format on each machine. OUTPUT-FORMAT is a text string, the
1157 name of a particular format supported by the BFD libraries. (You
1158 can list the available binary formats with `objdump -i'.) The
1159 script command `OUTPUT_FORMAT' can also specify the output format,
1160 but this option overrides it. *Note BFD::.
1164 Create a position independent executable. This is currently only
1165 supported on ELF platforms. Position independent executables are
1166 similar to shared libraries in that they are relocated by the
1167 dynamic linker to the virtual address the OS chooses for them
1168 (which can vary between invocations). Like normal dynamically
1169 linked executables they can be executed and symbols defined in the
1170 executable cannot be overridden by shared libraries.
1173 This option is ignored for Linux compatibility.
1176 This option is ignored for SVR4 compatibility.
1180 An option with machine dependent effects. This option is only
1181 supported on a few targets. *Note `ld' and the H8/300: H8/300.
1182 *Note `ld' and the Intel 960 family: i960. *Note `ld' and Xtensa
1183 Processors: Xtensa. *Note `ld' and the 68HC11 and 68HC12:
1184 M68HC11/68HC12. *Note `ld' and PowerPC 32-bit ELF Support:
1187 On some platforms the `--relax' option performs target specific,
1188 global optimizations that become possible when the linker resolves
1189 addressing in the program, such as relaxing address modes,
1190 synthesizing new instructions, selecting shorter version of current
1191 instructions, and combinig constant values.
1193 On some platforms these link time global optimizations may make
1194 symbolic debugging of the resulting executable impossible. This
1195 is known to be the case for the Matsushita MN10200 and MN10300
1196 family of processors.
1198 On platforms where this is not supported, `--relax' is accepted,
1201 On platforms where `--relax' is accepted the option `--no-relax'
1202 can be used to disable the feature.
1204 `--retain-symbols-file=FILENAME'
1205 Retain _only_ the symbols listed in the file FILENAME, discarding
1206 all others. FILENAME is simply a flat file, with one symbol name
1207 per line. This option is especially useful in environments (such
1208 as VxWorks) where a large global symbol table is accumulated
1209 gradually, to conserve run-time memory.
1211 `--retain-symbols-file' does _not_ discard undefined symbols, or
1212 symbols needed for relocations.
1214 You may only specify `--retain-symbols-file' once in the command
1215 line. It overrides `-s' and `-S'.
1218 Add a directory to the runtime library search path. This is used
1219 when linking an ELF executable with shared objects. All `-rpath'
1220 arguments are concatenated and passed to the runtime linker, which
1221 uses them to locate shared objects at runtime. The `-rpath'
1222 option is also used when locating shared objects which are needed
1223 by shared objects explicitly included in the link; see the
1224 description of the `-rpath-link' option. If `-rpath' is not used
1225 when linking an ELF executable, the contents of the environment
1226 variable `LD_RUN_PATH' will be used if it is defined.
1228 The `-rpath' option may also be used on SunOS. By default, on
1229 SunOS, the linker will form a runtime search patch out of all the
1230 `-L' options it is given. If a `-rpath' option is used, the
1231 runtime search path will be formed exclusively using the `-rpath'
1232 options, ignoring the `-L' options. This can be useful when using
1233 gcc, which adds many `-L' options which may be on NFS mounted file
1236 For compatibility with other ELF linkers, if the `-R' option is
1237 followed by a directory name, rather than a file name, it is
1238 treated as the `-rpath' option.
1241 When using ELF or SunOS, one shared library may require another.
1242 This happens when an `ld -shared' link includes a shared library
1243 as one of the input files.
1245 When the linker encounters such a dependency when doing a
1246 non-shared, non-relocatable link, it will automatically try to
1247 locate the required shared library and include it in the link, if
1248 it is not included explicitly. In such a case, the `-rpath-link'
1249 option specifies the first set of directories to search. The
1250 `-rpath-link' option may specify a sequence of directory names
1251 either by specifying a list of names separated by colons, or by
1252 appearing multiple times.
1254 This option should be used with caution as it overrides the search
1255 path that may have been hard compiled into a shared library. In
1256 such a case it is possible to use unintentionally a different
1257 search path than the runtime linker would do.
1259 The linker uses the following search paths to locate required
1261 1. Any directories specified by `-rpath-link' options.
1263 2. Any directories specified by `-rpath' options. The difference
1264 between `-rpath' and `-rpath-link' is that directories
1265 specified by `-rpath' options are included in the executable
1266 and used at runtime, whereas the `-rpath-link' option is only
1267 effective at link time. Searching `-rpath' in this way is
1268 only supported by native linkers and cross linkers which have
1269 been configured with the `--with-sysroot' option.
1271 3. On an ELF system, for native linkers, if the `-rpath' and
1272 `-rpath-link' options were not used, search the contents of
1273 the environment variable `LD_RUN_PATH'.
1275 4. On SunOS, if the `-rpath' option was not used, search any
1276 directories specified using `-L' options.
1278 5. For a native linker, the search the contents of the
1279 environment variable `LD_LIBRARY_PATH'.
1281 6. For a native ELF linker, the directories in `DT_RUNPATH' or
1282 `DT_RPATH' of a shared library are searched for shared
1283 libraries needed by it. The `DT_RPATH' entries are ignored if
1284 `DT_RUNPATH' entries exist.
1286 7. The default directories, normally `/lib' and `/usr/lib'.
1288 8. For a native linker on an ELF system, if the file
1289 `/etc/ld.so.conf' exists, the list of directories found in
1292 If the required shared library is not found, the linker will issue
1293 a warning and continue with the link.
1297 Create a shared library. This is currently only supported on ELF,
1298 XCOFF and SunOS platforms. On SunOS, the linker will
1299 automatically create a shared library if the `-e' option is not
1300 used and there are undefined symbols in the link.
1303 `--sort-common=ascending'
1304 `--sort-common=descending'
1305 This option tells `ld' to sort the common symbols by alignment in
1306 ascending or descending order when it places them in the
1307 appropriate output sections. The symbol alignments considered are
1308 sixteen-byte or larger, eight-byte, four-byte, two-byte, and
1309 one-byte. This is to prevent gaps between symbols due to alignment
1310 constraints. If no sorting order is specified, then descending
1313 `--sort-section=name'
1314 This option will apply `SORT_BY_NAME' to all wildcard section
1315 patterns in the linker script.
1317 `--sort-section=alignment'
1318 This option will apply `SORT_BY_ALIGNMENT' to all wildcard section
1319 patterns in the linker script.
1321 `--split-by-file[=SIZE]'
1322 Similar to `--split-by-reloc' but creates a new output section for
1323 each input file when SIZE is reached. SIZE defaults to a size of
1326 `--split-by-reloc[=COUNT]'
1327 Tries to creates extra sections in the output file so that no
1328 single output section in the file contains more than COUNT
1329 relocations. This is useful when generating huge relocatable
1330 files for downloading into certain real time kernels with the COFF
1331 object file format; since COFF cannot represent more than 65535
1332 relocations in a single section. Note that this will fail to work
1333 with object file formats which do not support arbitrary sections.
1334 The linker will not split up individual input sections for
1335 redistribution, so if a single input section contains more than
1336 COUNT relocations one output section will contain that many
1337 relocations. COUNT defaults to a value of 32768.
1340 Compute and display statistics about the operation of the linker,
1341 such as execution time and memory usage.
1343 `--sysroot=DIRECTORY'
1344 Use DIRECTORY as the location of the sysroot, overriding the
1345 configure-time default. This option is only supported by linkers
1346 that were configured using `--with-sysroot'.
1348 `--traditional-format'
1349 For some targets, the output of `ld' is different in some ways from
1350 the output of some existing linker. This switch requests `ld' to
1351 use the traditional format instead.
1353 For example, on SunOS, `ld' combines duplicate entries in the
1354 symbol string table. This can reduce the size of an output file
1355 with full debugging information by over 30 percent.
1356 Unfortunately, the SunOS `dbx' program can not read the resulting
1357 program (`gdb' has no trouble). The `--traditional-format' switch
1358 tells `ld' to not combine duplicate entries.
1360 `--section-start=SECTIONNAME=ORG'
1361 Locate a section in the output file at the absolute address given
1362 by ORG. You may use this option as many times as necessary to
1363 locate multiple sections in the command line. ORG must be a
1364 single hexadecimal integer; for compatibility with other linkers,
1365 you may omit the leading `0x' usually associated with hexadecimal
1366 values. _Note:_ there should be no white space between
1367 SECTIONNAME, the equals sign ("<=>"), and ORG.
1372 Same as `--section-start', with `.bss', `.data' or `.text' as the
1375 `-Ttext-segment=ORG'
1376 When creating an ELF executable or shared object, it will set the
1377 address of the first byte of the text segment.
1379 `--unresolved-symbols=METHOD'
1380 Determine how to handle unresolved symbols. There are four
1381 possible values for `method':
1384 Do not report any unresolved symbols.
1387 Report all unresolved symbols. This is the default.
1389 `ignore-in-object-files'
1390 Report unresolved symbols that are contained in shared
1391 libraries, but ignore them if they come from regular object
1394 `ignore-in-shared-libs'
1395 Report unresolved symbols that come from regular object
1396 files, but ignore them if they come from shared libraries.
1397 This can be useful when creating a dynamic binary and it is
1398 known that all the shared libraries that it should be
1399 referencing are included on the linker's command line.
1401 The behaviour for shared libraries on their own can also be
1402 controlled by the `--[no-]allow-shlib-undefined' option.
1404 Normally the linker will generate an error message for each
1405 reported unresolved symbol but the option
1406 `--warn-unresolved-symbols' can change this to a warning.
1409 `--verbose[=NUMBER]'
1410 Display the version number for `ld' and list the linker emulations
1411 supported. Display which input files can and cannot be opened.
1412 Display the linker script being used by the linker. If the
1413 optional NUMBER argument > 1, plugin symbol status will also be
1416 `--version-script=VERSION-SCRIPTFILE'
1417 Specify the name of a version script to the linker. This is
1418 typically used when creating shared libraries to specify
1419 additional information about the version hierarchy for the library
1420 being created. This option is only fully supported on ELF
1421 platforms which support shared libraries; see *Note VERSION::. It
1422 is partially supported on PE platforms, which can use version
1423 scripts to filter symbol visibility in auto-export mode: any
1424 symbols marked `local' in the version script will not be exported.
1428 Warn when a common symbol is combined with another common symbol
1429 or with a symbol definition. Unix linkers allow this somewhat
1430 sloppy practise, but linkers on some other operating systems do
1431 not. This option allows you to find potential problems from
1432 combining global symbols. Unfortunately, some C libraries use
1433 this practise, so you may get some warnings about symbols in the
1434 libraries as well as in your programs.
1436 There are three kinds of global symbols, illustrated here by C
1440 A definition, which goes in the initialized data section of
1444 An undefined reference, which does not allocate space. There
1445 must be either a definition or a common symbol for the
1449 A common symbol. If there are only (one or more) common
1450 symbols for a variable, it goes in the uninitialized data
1451 area of the output file. The linker merges multiple common
1452 symbols for the same variable into a single symbol. If they
1453 are of different sizes, it picks the largest size. The
1454 linker turns a common symbol into a declaration, if there is
1455 a definition of the same variable.
1457 The `--warn-common' option can produce five kinds of warnings.
1458 Each warning consists of a pair of lines: the first describes the
1459 symbol just encountered, and the second describes the previous
1460 symbol encountered with the same name. One or both of the two
1461 symbols will be a common symbol.
1463 1. Turning a common symbol into a reference, because there is
1464 already a definition for the symbol.
1465 FILE(SECTION): warning: common of `SYMBOL'
1466 overridden by definition
1467 FILE(SECTION): warning: defined here
1469 2. Turning a common symbol into a reference, because a later
1470 definition for the symbol is encountered. This is the same
1471 as the previous case, except that the symbols are encountered
1472 in a different order.
1473 FILE(SECTION): warning: definition of `SYMBOL'
1475 FILE(SECTION): warning: common is here
1477 3. Merging a common symbol with a previous same-sized common
1479 FILE(SECTION): warning: multiple common
1481 FILE(SECTION): warning: previous common is here
1483 4. Merging a common symbol with a previous larger common symbol.
1484 FILE(SECTION): warning: common of `SYMBOL'
1485 overridden by larger common
1486 FILE(SECTION): warning: larger common is here
1488 5. Merging a common symbol with a previous smaller common
1489 symbol. This is the same as the previous case, except that
1490 the symbols are encountered in a different order.
1491 FILE(SECTION): warning: common of `SYMBOL'
1492 overriding smaller common
1493 FILE(SECTION): warning: smaller common is here
1495 `--warn-constructors'
1496 Warn if any global constructors are used. This is only useful for
1497 a few object file formats. For formats like COFF or ELF, the
1498 linker can not detect the use of global constructors.
1500 `--warn-multiple-gp'
1501 Warn if multiple global pointer values are required in the output
1502 file. This is only meaningful for certain processors, such as the
1503 Alpha. Specifically, some processors put large-valued constants
1504 in a special section. A special register (the global pointer)
1505 points into the middle of this section, so that constants can be
1506 loaded efficiently via a base-register relative addressing mode.
1507 Since the offset in base-register relative mode is fixed and
1508 relatively small (e.g., 16 bits), this limits the maximum size of
1509 the constant pool. Thus, in large programs, it is often necessary
1510 to use multiple global pointer values in order to be able to
1511 address all possible constants. This option causes a warning to
1512 be issued whenever this case occurs.
1515 Only warn once for each undefined symbol, rather than once per
1516 module which refers to it.
1518 `--warn-section-align'
1519 Warn if the address of an output section is changed because of
1520 alignment. Typically, the alignment will be set by an input
1521 section. The address will only be changed if it not explicitly
1522 specified; that is, if the `SECTIONS' command does not specify a
1523 start address for the section (*note SECTIONS::).
1525 `--warn-shared-textrel'
1526 Warn if the linker adds a DT_TEXTREL to a shared object.
1528 `--warn-alternate-em'
1529 Warn if an object has alternate ELF machine code.
1531 `--warn-unresolved-symbols'
1532 If the linker is going to report an unresolved symbol (see the
1533 option `--unresolved-symbols') it will normally generate an error.
1534 This option makes it generate a warning instead.
1536 `--error-unresolved-symbols'
1537 This restores the linker's default behaviour of generating errors
1538 when it is reporting unresolved symbols.
1541 For each archive mentioned on the command line after the
1542 `--whole-archive' option, include every object file in the archive
1543 in the link, rather than searching the archive for the required
1544 object files. This is normally used to turn an archive file into
1545 a shared library, forcing every object to be included in the
1546 resulting shared library. This option may be used more than once.
1548 Two notes when using this option from gcc: First, gcc doesn't know
1549 about this option, so you have to use `-Wl,-whole-archive'.
1550 Second, don't forget to use `-Wl,-no-whole-archive' after your
1551 list of archives, because gcc will add its own list of archives to
1552 your link and you may not want this flag to affect those as well.
1555 Use a wrapper function for SYMBOL. Any undefined reference to
1556 SYMBOL will be resolved to `__wrap_SYMBOL'. Any undefined
1557 reference to `__real_SYMBOL' will be resolved to SYMBOL.
1559 This can be used to provide a wrapper for a system function. The
1560 wrapper function should be called `__wrap_SYMBOL'. If it wishes
1561 to call the system function, it should call `__real_SYMBOL'.
1563 Here is a trivial example:
1566 __wrap_malloc (size_t c)
1568 printf ("malloc called with %zu\n", c);
1569 return __real_malloc (c);
1572 If you link other code with this file using `--wrap malloc', then
1573 all calls to `malloc' will call the function `__wrap_malloc'
1574 instead. The call to `__real_malloc' in `__wrap_malloc' will call
1575 the real `malloc' function.
1577 You may wish to provide a `__real_malloc' function as well, so that
1578 links without the `--wrap' option will succeed. If you do this,
1579 you should not put the definition of `__real_malloc' in the same
1580 file as `__wrap_malloc'; if you do, the assembler may resolve the
1581 call before the linker has a chance to wrap it to `malloc'.
1584 Request creation of `.eh_frame_hdr' section and ELF
1585 `PT_GNU_EH_FRAME' segment header.
1587 `--enable-new-dtags'
1588 `--disable-new-dtags'
1589 This linker can create the new dynamic tags in ELF. But the older
1590 ELF systems may not understand them. If you specify
1591 `--enable-new-dtags', the dynamic tags will be created as needed.
1592 If you specify `--disable-new-dtags', no new dynamic tags will be
1593 created. By default, the new dynamic tags are not created. Note
1594 that those options are only available for ELF systems.
1596 `--hash-size=NUMBER'
1597 Set the default size of the linker's hash tables to a prime number
1598 close to NUMBER. Increasing this value can reduce the length of
1599 time it takes the linker to perform its tasks, at the expense of
1600 increasing the linker's memory requirements. Similarly reducing
1601 this value can reduce the memory requirements at the expense of
1604 `--hash-style=STYLE'
1605 Set the type of linker's hash table(s). STYLE can be either
1606 `sysv' for classic ELF `.hash' section, `gnu' for new style GNU
1607 `.gnu.hash' section or `both' for both the classic ELF `.hash' and
1608 new style GNU `.gnu.hash' hash tables. The default is `sysv'.
1610 `--reduce-memory-overheads'
1611 This option reduces memory requirements at ld runtime, at the
1612 expense of linking speed. This was introduced to select the old
1613 O(n^2) algorithm for link map file generation, rather than the new
1614 O(n) algorithm which uses about 40% more memory for symbol storage.
1616 Another effect of the switch is to set the default hash table size
1617 to 1021, which again saves memory at the cost of lengthening the
1618 linker's run time. This is not done however if the `--hash-size'
1619 switch has been used.
1621 The `--reduce-memory-overheads' switch may be also be used to
1622 enable other tradeoffs in future versions of the linker.
1626 Request creation of `.note.gnu.build-id' ELF note section. The
1627 contents of the note are unique bits identifying this linked file.
1628 STYLE can be `uuid' to use 128 random bits, `sha1' to use a
1629 160-bit SHA1 hash on the normative parts of the output contents,
1630 `md5' to use a 128-bit MD5 hash on the normative parts of the
1631 output contents, or `0xHEXSTRING' to use a chosen bit string
1632 specified as an even number of hexadecimal digits (`-' and `:'
1633 characters between digit pairs are ignored). If STYLE is omitted,
1636 The `md5' and `sha1' styles produces an identifier that is always
1637 the same in an identical output file, but will be unique among all
1638 nonidentical output files. It is not intended to be compared as a
1639 checksum for the file's contents. A linked file may be changed
1640 later by other tools, but the build ID bit string identifying the
1641 original linked file does not change.
1643 Passing `none' for STYLE disables the setting from any
1644 `--build-id' options earlier on the command line.
1646 2.1.1 Options Specific to i386 PE Targets
1647 -----------------------------------------
1649 The i386 PE linker supports the `-shared' option, which causes the
1650 output to be a dynamically linked library (DLL) instead of a normal
1651 executable. You should name the output `*.dll' when you use this
1652 option. In addition, the linker fully supports the standard `*.def'
1653 files, which may be specified on the linker command line like an object
1654 file (in fact, it should precede archives it exports symbols from, to
1655 ensure that they get linked in, just like a normal object file).
1657 In addition to the options common to all targets, the i386 PE linker
1658 support additional command line options that are specific to the i386
1659 PE target. Options that take values may be separated from their values
1660 by either a space or an equals sign.
1662 `--add-stdcall-alias'
1663 If given, symbols with a stdcall suffix (@NN) will be exported
1664 as-is and also with the suffix stripped. [This option is specific
1665 to the i386 PE targeted port of the linker]
1668 Use FILE as the name of a file in which to save the base addresses
1669 of all the relocations needed for generating DLLs with `dlltool'.
1670 [This is an i386 PE specific option]
1673 Create a DLL instead of a regular executable. You may also use
1674 `-shared' or specify a `LIBRARY' in a given `.def' file. [This
1675 option is specific to the i386 PE targeted port of the linker]
1677 `--enable-long-section-names'
1678 `--disable-long-section-names'
1679 The PE variants of the Coff object format add an extension that
1680 permits the use of section names longer than eight characters, the
1681 normal limit for Coff. By default, these names are only allowed
1682 in object files, as fully-linked executable images do not carry
1683 the Coff string table required to support the longer names. As a
1684 GNU extension, it is possible to allow their use in executable
1685 images as well, or to (probably pointlessly!) disallow it in
1686 object files, by using these two options. Executable images
1687 generated with these long section names are slightly non-standard,
1688 carrying as they do a string table, and may generate confusing
1689 output when examined with non-GNU PE-aware tools, such as file
1690 viewers and dumpers. However, GDB relies on the use of PE long
1691 section names to find Dwarf-2 debug information sections in an
1692 executable image at runtime, and so if neither option is specified
1693 on the command-line, `ld' will enable long section names,
1694 overriding the default and technically correct behaviour, when it
1695 finds the presence of debug information while linking an executable
1696 image and not stripping symbols. [This option is valid for all PE
1697 targeted ports of the linker]
1699 `--enable-stdcall-fixup'
1700 `--disable-stdcall-fixup'
1701 If the link finds a symbol that it cannot resolve, it will attempt
1702 to do "fuzzy linking" by looking for another defined symbol that
1703 differs only in the format of the symbol name (cdecl vs stdcall)
1704 and will resolve that symbol by linking to the match. For
1705 example, the undefined symbol `_foo' might be linked to the
1706 function `_foo@12', or the undefined symbol `_bar@16' might be
1707 linked to the function `_bar'. When the linker does this, it
1708 prints a warning, since it normally should have failed to link,
1709 but sometimes import libraries generated from third-party dlls may
1710 need this feature to be usable. If you specify
1711 `--enable-stdcall-fixup', this feature is fully enabled and
1712 warnings are not printed. If you specify
1713 `--disable-stdcall-fixup', this feature is disabled and such
1714 mismatches are considered to be errors. [This option is specific
1715 to the i386 PE targeted port of the linker]
1717 `--leading-underscore'
1718 `--no-leading-underscore'
1719 For most targets default symbol-prefix is an underscore and is
1720 defined in target's description. By this option it is possible to
1721 disable/enable the default underscore symbol-prefix.
1723 `--export-all-symbols'
1724 If given, all global symbols in the objects used to build a DLL
1725 will be exported by the DLL. Note that this is the default if
1726 there otherwise wouldn't be any exported symbols. When symbols are
1727 explicitly exported via DEF files or implicitly exported via
1728 function attributes, the default is to not export anything else
1729 unless this option is given. Note that the symbols `DllMain@12',
1730 `DllEntryPoint@0', `DllMainCRTStartup@12', and `impure_ptr' will
1731 not be automatically exported. Also, symbols imported from other
1732 DLLs will not be re-exported, nor will symbols specifying the
1733 DLL's internal layout such as those beginning with `_head_' or
1734 ending with `_iname'. In addition, no symbols from `libgcc',
1735 `libstd++', `libmingw32', or `crtX.o' will be exported. Symbols
1736 whose names begin with `__rtti_' or `__builtin_' will not be
1737 exported, to help with C++ DLLs. Finally, there is an extensive
1738 list of cygwin-private symbols that are not exported (obviously,
1739 this applies on when building DLLs for cygwin targets). These
1740 cygwin-excludes are: `_cygwin_dll_entry@12',
1741 `_cygwin_crt0_common@8', `_cygwin_noncygwin_dll_entry@12',
1742 `_fmode', `_impure_ptr', `cygwin_attach_dll', `cygwin_premain0',
1743 `cygwin_premain1', `cygwin_premain2', `cygwin_premain3', and
1744 `environ'. [This option is specific to the i386 PE targeted port
1747 `--exclude-symbols SYMBOL,SYMBOL,...'
1748 Specifies a list of symbols which should not be automatically
1749 exported. The symbol names may be delimited by commas or colons.
1750 [This option is specific to the i386 PE targeted port of the
1753 `--exclude-all-symbols'
1754 Specifies no symbols should be automatically exported. [This
1755 option is specific to the i386 PE targeted port of the linker]
1758 Specify the file alignment. Sections in the file will always
1759 begin at file offsets which are multiples of this number. This
1760 defaults to 512. [This option is specific to the i386 PE targeted
1764 `--heap RESERVE,COMMIT'
1765 Specify the number of bytes of memory to reserve (and optionally
1766 commit) to be used as heap for this program. The default is 1Mb
1767 reserved, 4K committed. [This option is specific to the i386 PE
1768 targeted port of the linker]
1770 `--image-base VALUE'
1771 Use VALUE as the base address of your program or dll. This is the
1772 lowest memory location that will be used when your program or dll
1773 is loaded. To reduce the need to relocate and improve performance
1774 of your dlls, each should have a unique base address and not
1775 overlap any other dlls. The default is 0x400000 for executables,
1776 and 0x10000000 for dlls. [This option is specific to the i386 PE
1777 targeted port of the linker]
1780 If given, the stdcall suffixes (@NN) will be stripped from symbols
1781 before they are exported. [This option is specific to the i386 PE
1782 targeted port of the linker]
1784 `--large-address-aware'
1785 If given, the appropriate bit in the "Characteristics" field of
1786 the COFF header is set to indicate that this executable supports
1787 virtual addresses greater than 2 gigabytes. This should be used
1788 in conjunction with the /3GB or /USERVA=VALUE megabytes switch in
1789 the "[operating systems]" section of the BOOT.INI. Otherwise,
1790 this bit has no effect. [This option is specific to PE targeted
1791 ports of the linker]
1793 `--major-image-version VALUE'
1794 Sets the major number of the "image version". Defaults to 1.
1795 [This option is specific to the i386 PE targeted port of the
1798 `--major-os-version VALUE'
1799 Sets the major number of the "os version". Defaults to 4. [This
1800 option is specific to the i386 PE targeted port of the linker]
1802 `--major-subsystem-version VALUE'
1803 Sets the major number of the "subsystem version". Defaults to 4.
1804 [This option is specific to the i386 PE targeted port of the
1807 `--minor-image-version VALUE'
1808 Sets the minor number of the "image version". Defaults to 0.
1809 [This option is specific to the i386 PE targeted port of the
1812 `--minor-os-version VALUE'
1813 Sets the minor number of the "os version". Defaults to 0. [This
1814 option is specific to the i386 PE targeted port of the linker]
1816 `--minor-subsystem-version VALUE'
1817 Sets the minor number of the "subsystem version". Defaults to 0.
1818 [This option is specific to the i386 PE targeted port of the
1822 The linker will create the file FILE which will contain a DEF file
1823 corresponding to the DLL the linker is generating. This DEF file
1824 (which should be called `*.def') may be used to create an import
1825 library with `dlltool' or may be used as a reference to
1826 automatically or implicitly exported symbols. [This option is
1827 specific to the i386 PE targeted port of the linker]
1830 The linker will create the file FILE which will contain an import
1831 lib corresponding to the DLL the linker is generating. This import
1832 lib (which should be called `*.dll.a' or `*.a' may be used to link
1833 clients against the generated DLL; this behaviour makes it
1834 possible to skip a separate `dlltool' import library creation step.
1835 [This option is specific to the i386 PE targeted port of the
1838 `--enable-auto-image-base'
1839 Automatically choose the image base for DLLs, unless one is
1840 specified using the `--image-base' argument. By using a hash
1841 generated from the dllname to create unique image bases for each
1842 DLL, in-memory collisions and relocations which can delay program
1843 execution are avoided. [This option is specific to the i386 PE
1844 targeted port of the linker]
1846 `--disable-auto-image-base'
1847 Do not automatically generate a unique image base. If there is no
1848 user-specified image base (`--image-base') then use the platform
1849 default. [This option is specific to the i386 PE targeted port of
1852 `--dll-search-prefix STRING'
1853 When linking dynamically to a dll without an import library,
1854 search for `<string><basename>.dll' in preference to
1855 `lib<basename>.dll'. This behaviour allows easy distinction
1856 between DLLs built for the various "subplatforms": native, cygwin,
1857 uwin, pw, etc. For instance, cygwin DLLs typically use
1858 `--dll-search-prefix=cyg'. [This option is specific to the i386
1859 PE targeted port of the linker]
1861 `--enable-auto-import'
1862 Do sophisticated linking of `_symbol' to `__imp__symbol' for DATA
1863 imports from DLLs, and create the necessary thunking symbols when
1864 building the import libraries with those DATA exports. Note: Use
1865 of the 'auto-import' extension will cause the text section of the
1866 image file to be made writable. This does not conform to the
1867 PE-COFF format specification published by Microsoft.
1869 Note - use of the 'auto-import' extension will also cause read only
1870 data which would normally be placed into the .rdata section to be
1871 placed into the .data section instead. This is in order to work
1872 around a problem with consts that is described here:
1873 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
1875 Using 'auto-import' generally will 'just work' - but sometimes you
1876 may see this message:
1878 "variable '<var>' can't be auto-imported. Please read the
1879 documentation for ld's `--enable-auto-import' for details."
1881 This message occurs when some (sub)expression accesses an address
1882 ultimately given by the sum of two constants (Win32 import tables
1883 only allow one). Instances where this may occur include accesses
1884 to member fields of struct variables imported from a DLL, as well
1885 as using a constant index into an array variable imported from a
1886 DLL. Any multiword variable (arrays, structs, long long, etc) may
1887 trigger this error condition. However, regardless of the exact
1888 data type of the offending exported variable, ld will always
1889 detect it, issue the warning, and exit.
1891 There are several ways to address this difficulty, regardless of
1892 the data type of the exported variable:
1894 One way is to use -enable-runtime-pseudo-reloc switch. This leaves
1895 the task of adjusting references in your client code for runtime
1896 environment, so this method works only when runtime environment
1897 supports this feature.
1899 A second solution is to force one of the 'constants' to be a
1900 variable - that is, unknown and un-optimizable at compile time.
1901 For arrays, there are two possibilities: a) make the indexee (the
1902 array's address) a variable, or b) make the 'constant' index a
1905 extern type extern_array[];
1907 { volatile type *t=extern_array; t[1] }
1911 extern type extern_array[];
1913 { volatile int t=1; extern_array[t] }
1915 For structs (and most other multiword data types) the only option
1916 is to make the struct itself (or the long long, or the ...)
1919 extern struct s extern_struct;
1920 extern_struct.field -->
1921 { volatile struct s *t=&extern_struct; t->field }
1925 extern long long extern_ll;
1927 { volatile long long * local_ll=&extern_ll; *local_ll }
1929 A third method of dealing with this difficulty is to abandon
1930 'auto-import' for the offending symbol and mark it with
1931 `__declspec(dllimport)'. However, in practise that requires using
1932 compile-time #defines to indicate whether you are building a DLL,
1933 building client code that will link to the DLL, or merely
1934 building/linking to a static library. In making the choice
1935 between the various methods of resolving the 'direct address with
1936 constant offset' problem, you should consider typical real-world
1944 void main(int argc, char **argv){
1945 printf("%d\n",arr[1]);
1953 void main(int argc, char **argv){
1954 /* This workaround is for win32 and cygwin; do not "optimize" */
1955 volatile int *parr = arr;
1956 printf("%d\n",parr[1]);
1961 /* Note: auto-export is assumed (no __declspec(dllexport)) */
1962 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
1963 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
1964 #define FOO_IMPORT __declspec(dllimport)
1968 extern FOO_IMPORT int arr[];
1971 void main(int argc, char **argv){
1972 printf("%d\n",arr[1]);
1975 A fourth way to avoid this problem is to re-code your library to
1976 use a functional interface rather than a data interface for the
1977 offending variables (e.g. set_foo() and get_foo() accessor
1978 functions). [This option is specific to the i386 PE targeted port
1981 `--disable-auto-import'
1982 Do not attempt to do sophisticated linking of `_symbol' to
1983 `__imp__symbol' for DATA imports from DLLs. [This option is
1984 specific to the i386 PE targeted port of the linker]
1986 `--enable-runtime-pseudo-reloc'
1987 If your code contains expressions described in -enable-auto-import
1988 section, that is, DATA imports from DLL with non-zero offset, this
1989 switch will create a vector of 'runtime pseudo relocations' which
1990 can be used by runtime environment to adjust references to such
1991 data in your client code. [This option is specific to the i386 PE
1992 targeted port of the linker]
1994 `--disable-runtime-pseudo-reloc'
1995 Do not create pseudo relocations for non-zero offset DATA imports
1996 from DLLs. This is the default. [This option is specific to the
1997 i386 PE targeted port of the linker]
1999 `--enable-extra-pe-debug'
2000 Show additional debug info related to auto-import symbol thunking.
2001 [This option is specific to the i386 PE targeted port of the
2004 `--section-alignment'
2005 Sets the section alignment. Sections in memory will always begin
2006 at addresses which are a multiple of this number. Defaults to
2007 0x1000. [This option is specific to the i386 PE targeted port of
2011 `--stack RESERVE,COMMIT'
2012 Specify the number of bytes of memory to reserve (and optionally
2013 commit) to be used as stack for this program. The default is 2Mb
2014 reserved, 4K committed. [This option is specific to the i386 PE
2015 targeted port of the linker]
2018 `--subsystem WHICH:MAJOR'
2019 `--subsystem WHICH:MAJOR.MINOR'
2020 Specifies the subsystem under which your program will execute. The
2021 legal values for WHICH are `native', `windows', `console',
2022 `posix', and `xbox'. You may optionally set the subsystem version
2023 also. Numeric values are also accepted for WHICH. [This option
2024 is specific to the i386 PE targeted port of the linker]
2026 The following options set flags in the `DllCharacteristics' field
2027 of the PE file header: [These options are specific to PE targeted
2028 ports of the linker]
2031 The image base address may be relocated using address space layout
2032 randomization (ASLR). This feature was introduced with MS Windows
2033 Vista for i386 PE targets.
2036 Code integrity checks are enforced.
2039 The image is compatible with the Data Execution Prevention. This
2040 feature was introduced with MS Windows XP SP2 for i386 PE targets.
2043 Although the image understands isolation, do not isolate the image.
2046 The image does not use SEH. No SE handler may be called from this
2050 Do not bind this image.
2053 The driver uses the MS Windows Driver Model.
2056 The image is Terminal Server aware.
2059 2.1.2 Options specific to Motorola 68HC11 and 68HC12 targets
2060 ------------------------------------------------------------
2062 The 68HC11 and 68HC12 linkers support specific options to control the
2063 memory bank switching mapping and trampoline code generation.
2066 This option disables the generation of trampoline. By default a
2067 trampoline is generated for each far function which is called
2068 using a `jsr' instruction (this happens when a pointer to a far
2071 `--bank-window NAME'
2072 This option indicates to the linker the name of the memory region
2073 in the `MEMORY' specification that describes the memory bank
2074 window. The definition of such region is then used by the linker
2075 to compute paging and addresses within the memory window.
2078 2.1.3 Options specific to Motorola 68K target
2079 ---------------------------------------------
2081 The following options are supported to control handling of GOT
2082 generation when linking for 68K targets.
2085 This option tells the linker which GOT generation scheme to use.
2086 TYPE should be one of `single', `negative', `multigot' or
2087 `target'. For more information refer to the Info entry for `ld'.
2091 File: ld.info, Node: Environment, Prev: Options, Up: Invocation
2093 2.2 Environment Variables
2094 =========================
2096 You can change the behaviour of `ld' with the environment variables
2097 `GNUTARGET', `LDEMULATION' and `COLLECT_NO_DEMANGLE'.
2099 `GNUTARGET' determines the input-file object format if you don't use
2100 `-b' (or its synonym `--format'). Its value should be one of the BFD
2101 names for an input format (*note BFD::). If there is no `GNUTARGET' in
2102 the environment, `ld' uses the natural format of the target. If
2103 `GNUTARGET' is set to `default' then BFD attempts to discover the input
2104 format by examining binary input files; this method often succeeds, but
2105 there are potential ambiguities, since there is no method of ensuring
2106 that the magic number used to specify object-file formats is unique.
2107 However, the configuration procedure for BFD on each system places the
2108 conventional format for that system first in the search-list, so
2109 ambiguities are resolved in favor of convention.
2111 `LDEMULATION' determines the default emulation if you don't use the
2112 `-m' option. The emulation can affect various aspects of linker
2113 behaviour, particularly the default linker script. You can list the
2114 available emulations with the `--verbose' or `-V' options. If the `-m'
2115 option is not used, and the `LDEMULATION' environment variable is not
2116 defined, the default emulation depends upon how the linker was
2119 Normally, the linker will default to demangling symbols. However, if
2120 `COLLECT_NO_DEMANGLE' is set in the environment, then it will default
2121 to not demangling symbols. This environment variable is used in a
2122 similar fashion by the `gcc' linker wrapper program. The default may
2123 be overridden by the `--demangle' and `--no-demangle' options.
2126 File: ld.info, Node: Scripts, Next: Machine Dependent, Prev: Invocation, Up: Top
2131 Every link is controlled by a "linker script". This script is written
2132 in the linker command language.
2134 The main purpose of the linker script is to describe how the
2135 sections in the input files should be mapped into the output file, and
2136 to control the memory layout of the output file. Most linker scripts
2137 do nothing more than this. However, when necessary, the linker script
2138 can also direct the linker to perform many other operations, using the
2139 commands described below.
2141 The linker always uses a linker script. If you do not supply one
2142 yourself, the linker will use a default script that is compiled into the
2143 linker executable. You can use the `--verbose' command line option to
2144 display the default linker script. Certain command line options, such
2145 as `-r' or `-N', will affect the default linker script.
2147 You may supply your own linker script by using the `-T' command line
2148 option. When you do this, your linker script will replace the default
2151 You may also use linker scripts implicitly by naming them as input
2152 files to the linker, as though they were files to be linked. *Note
2153 Implicit Linker Scripts::.
2157 * Basic Script Concepts:: Basic Linker Script Concepts
2158 * Script Format:: Linker Script Format
2159 * Simple Example:: Simple Linker Script Example
2160 * Simple Commands:: Simple Linker Script Commands
2161 * Assignments:: Assigning Values to Symbols
2162 * SECTIONS:: SECTIONS Command
2163 * MEMORY:: MEMORY Command
2164 * PHDRS:: PHDRS Command
2165 * VERSION:: VERSION Command
2166 * Expressions:: Expressions in Linker Scripts
2167 * Implicit Linker Scripts:: Implicit Linker Scripts
2170 File: ld.info, Node: Basic Script Concepts, Next: Script Format, Up: Scripts
2172 3.1 Basic Linker Script Concepts
2173 ================================
2175 We need to define some basic concepts and vocabulary in order to
2176 describe the linker script language.
2178 The linker combines input files into a single output file. The
2179 output file and each input file are in a special data format known as an
2180 "object file format". Each file is called an "object file". The
2181 output file is often called an "executable", but for our purposes we
2182 will also call it an object file. Each object file has, among other
2183 things, a list of "sections". We sometimes refer to a section in an
2184 input file as an "input section"; similarly, a section in the output
2185 file is an "output section".
2187 Each section in an object file has a name and a size. Most sections
2188 also have an associated block of data, known as the "section contents".
2189 A section may be marked as "loadable", which mean that the contents
2190 should be loaded into memory when the output file is run. A section
2191 with no contents may be "allocatable", which means that an area in
2192 memory should be set aside, but nothing in particular should be loaded
2193 there (in some cases this memory must be zeroed out). A section which
2194 is neither loadable nor allocatable typically contains some sort of
2195 debugging information.
2197 Every loadable or allocatable output section has two addresses. The
2198 first is the "VMA", or virtual memory address. This is the address the
2199 section will have when the output file is run. The second is the
2200 "LMA", or load memory address. This is the address at which the
2201 section will be loaded. In most cases the two addresses will be the
2202 same. An example of when they might be different is when a data section
2203 is loaded into ROM, and then copied into RAM when the program starts up
2204 (this technique is often used to initialize global variables in a ROM
2205 based system). In this case the ROM address would be the LMA, and the
2206 RAM address would be the VMA.
2208 You can see the sections in an object file by using the `objdump'
2209 program with the `-h' option.
2211 Every object file also has a list of "symbols", known as the "symbol
2212 table". A symbol may be defined or undefined. Each symbol has a name,
2213 and each defined symbol has an address, among other information. If
2214 you compile a C or C++ program into an object file, you will get a
2215 defined symbol for every defined function and global or static
2216 variable. Every undefined function or global variable which is
2217 referenced in the input file will become an undefined symbol.
2219 You can see the symbols in an object file by using the `nm' program,
2220 or by using the `objdump' program with the `-t' option.
2223 File: ld.info, Node: Script Format, Next: Simple Example, Prev: Basic Script Concepts, Up: Scripts
2225 3.2 Linker Script Format
2226 ========================
2228 Linker scripts are text files.
2230 You write a linker script as a series of commands. Each command is
2231 either a keyword, possibly followed by arguments, or an assignment to a
2232 symbol. You may separate commands using semicolons. Whitespace is
2235 Strings such as file or format names can normally be entered
2236 directly. If the file name contains a character such as a comma which
2237 would otherwise serve to separate file names, you may put the file name
2238 in double quotes. There is no way to use a double quote character in a
2241 You may include comments in linker scripts just as in C, delimited by
2242 `/*' and `*/'. As in C, comments are syntactically equivalent to
2246 File: ld.info, Node: Simple Example, Next: Simple Commands, Prev: Script Format, Up: Scripts
2248 3.3 Simple Linker Script Example
2249 ================================
2251 Many linker scripts are fairly simple.
2253 The simplest possible linker script has just one command:
2254 `SECTIONS'. You use the `SECTIONS' command to describe the memory
2255 layout of the output file.
2257 The `SECTIONS' command is a powerful command. Here we will describe
2258 a simple use of it. Let's assume your program consists only of code,
2259 initialized data, and uninitialized data. These will be in the
2260 `.text', `.data', and `.bss' sections, respectively. Let's assume
2261 further that these are the only sections which appear in your input
2264 For this example, let's say that the code should be loaded at address
2265 0x10000, and that the data should start at address 0x8000000. Here is a
2266 linker script which will do that:
2270 .text : { *(.text) }
2272 .data : { *(.data) }
2276 You write the `SECTIONS' command as the keyword `SECTIONS', followed
2277 by a series of symbol assignments and output section descriptions
2278 enclosed in curly braces.
2280 The first line inside the `SECTIONS' command of the above example
2281 sets the value of the special symbol `.', which is the location
2282 counter. If you do not specify the address of an output section in some
2283 other way (other ways are described later), the address is set from the
2284 current value of the location counter. The location counter is then
2285 incremented by the size of the output section. At the start of the
2286 `SECTIONS' command, the location counter has the value `0'.
2288 The second line defines an output section, `.text'. The colon is
2289 required syntax which may be ignored for now. Within the curly braces
2290 after the output section name, you list the names of the input sections
2291 which should be placed into this output section. The `*' is a wildcard
2292 which matches any file name. The expression `*(.text)' means all
2293 `.text' input sections in all input files.
2295 Since the location counter is `0x10000' when the output section
2296 `.text' is defined, the linker will set the address of the `.text'
2297 section in the output file to be `0x10000'.
2299 The remaining lines define the `.data' and `.bss' sections in the
2300 output file. The linker will place the `.data' output section at
2301 address `0x8000000'. After the linker places the `.data' output
2302 section, the value of the location counter will be `0x8000000' plus the
2303 size of the `.data' output section. The effect is that the linker will
2304 place the `.bss' output section immediately after the `.data' output
2307 The linker will ensure that each output section has the required
2308 alignment, by increasing the location counter if necessary. In this
2309 example, the specified addresses for the `.text' and `.data' sections
2310 will probably satisfy any alignment constraints, but the linker may
2311 have to create a small gap between the `.data' and `.bss' sections.
2313 That's it! That's a simple and complete linker script.
2316 File: ld.info, Node: Simple Commands, Next: Assignments, Prev: Simple Example, Up: Scripts
2318 3.4 Simple Linker Script Commands
2319 =================================
2321 In this section we describe the simple linker script commands.
2325 * Entry Point:: Setting the entry point
2326 * File Commands:: Commands dealing with files
2328 * Format Commands:: Commands dealing with object file formats
2330 * REGION_ALIAS:: Assign alias names to memory regions
2331 * Miscellaneous Commands:: Other linker script commands
2334 File: ld.info, Node: Entry Point, Next: File Commands, Up: Simple Commands
2336 3.4.1 Setting the Entry Point
2337 -----------------------------
2339 The first instruction to execute in a program is called the "entry
2340 point". You can use the `ENTRY' linker script command to set the entry
2341 point. The argument is a symbol name:
2344 There are several ways to set the entry point. The linker will set
2345 the entry point by trying each of the following methods in order, and
2346 stopping when one of them succeeds:
2347 * the `-e' ENTRY command-line option;
2349 * the `ENTRY(SYMBOL)' command in a linker script;
2351 * the value of a target specific symbol, if it is defined; For many
2352 targets this is `start', but PE and BeOS based systems for example
2353 check a list of possible entry symbols, matching the first one
2356 * the address of the first byte of the `.text' section, if present;
2361 File: ld.info, Node: File Commands, Next: Format Commands, Prev: Entry Point, Up: Simple Commands
2363 3.4.2 Commands Dealing with Files
2364 ---------------------------------
2366 Several linker script commands deal with files.
2369 Include the linker script FILENAME at this point. The file will
2370 be searched for in the current directory, and in any directory
2371 specified with the `-L' option. You can nest calls to `INCLUDE'
2372 up to 10 levels deep.
2374 You can place `INCLUDE' directives at the top level, in `MEMORY' or
2375 `SECTIONS' commands, or in output section descriptions.
2377 `INPUT(FILE, FILE, ...)'
2378 `INPUT(FILE FILE ...)'
2379 The `INPUT' command directs the linker to include the named files
2380 in the link, as though they were named on the command line.
2382 For example, if you always want to include `subr.o' any time you do
2383 a link, but you can't be bothered to put it on every link command
2384 line, then you can put `INPUT (subr.o)' in your linker script.
2386 In fact, if you like, you can list all of your input files in the
2387 linker script, and then invoke the linker with nothing but a `-T'
2390 In case a "sysroot prefix" is configured, and the filename starts
2391 with the `/' character, and the script being processed was located
2392 inside the "sysroot prefix", the filename will be looked for in
2393 the "sysroot prefix". Otherwise, the linker will try to open the
2394 file in the current directory. If it is not found, the linker
2395 will search through the archive library search path. See the
2396 description of `-L' in *Note Command Line Options: Options.
2398 If you use `INPUT (-lFILE)', `ld' will transform the name to
2399 `libFILE.a', as with the command line argument `-l'.
2401 When you use the `INPUT' command in an implicit linker script, the
2402 files will be included in the link at the point at which the linker
2403 script file is included. This can affect archive searching.
2405 `GROUP(FILE, FILE, ...)'
2406 `GROUP(FILE FILE ...)'
2407 The `GROUP' command is like `INPUT', except that the named files
2408 should all be archives, and they are searched repeatedly until no
2409 new undefined references are created. See the description of `-('
2410 in *Note Command Line Options: Options.
2412 `AS_NEEDED(FILE, FILE, ...)'
2413 `AS_NEEDED(FILE FILE ...)'
2414 This construct can appear only inside of the `INPUT' or `GROUP'
2415 commands, among other filenames. The files listed will be handled
2416 as if they appear directly in the `INPUT' or `GROUP' commands,
2417 with the exception of ELF shared libraries, that will be added only
2418 when they are actually needed. This construct essentially enables
2419 `--as-needed' option for all the files listed inside of it and
2420 restores previous `--as-needed' resp. `--no-as-needed' setting
2424 The `OUTPUT' command names the output file. Using
2425 `OUTPUT(FILENAME)' in the linker script is exactly like using `-o
2426 FILENAME' on the command line (*note Command Line Options:
2427 Options.). If both are used, the command line option takes
2430 You can use the `OUTPUT' command to define a default name for the
2431 output file other than the usual default of `a.out'.
2434 The `SEARCH_DIR' command adds PATH to the list of paths where `ld'
2435 looks for archive libraries. Using `SEARCH_DIR(PATH)' is exactly
2436 like using `-L PATH' on the command line (*note Command Line
2437 Options: Options.). If both are used, then the linker will search
2438 both paths. Paths specified using the command line option are
2442 The `STARTUP' command is just like the `INPUT' command, except
2443 that FILENAME will become the first input file to be linked, as
2444 though it were specified first on the command line. This may be
2445 useful when using a system in which the entry point is always the
2446 start of the first file.
2449 File: ld.info, Node: Format Commands, Next: REGION_ALIAS, Prev: File Commands, Up: Simple Commands
2451 3.4.3 Commands Dealing with Object File Formats
2452 -----------------------------------------------
2454 A couple of linker script commands deal with object file formats.
2456 `OUTPUT_FORMAT(BFDNAME)'
2457 `OUTPUT_FORMAT(DEFAULT, BIG, LITTLE)'
2458 The `OUTPUT_FORMAT' command names the BFD format to use for the
2459 output file (*note BFD::). Using `OUTPUT_FORMAT(BFDNAME)' is
2460 exactly like using `--oformat BFDNAME' on the command line (*note
2461 Command Line Options: Options.). If both are used, the command
2462 line option takes precedence.
2464 You can use `OUTPUT_FORMAT' with three arguments to use different
2465 formats based on the `-EB' and `-EL' command line options. This
2466 permits the linker script to set the output format based on the
2469 If neither `-EB' nor `-EL' are used, then the output format will
2470 be the first argument, DEFAULT. If `-EB' is used, the output
2471 format will be the second argument, BIG. If `-EL' is used, the
2472 output format will be the third argument, LITTLE.
2474 For example, the default linker script for the MIPS ELF target
2476 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2477 This says that the default format for the output file is
2478 `elf32-bigmips', but if the user uses the `-EL' command line
2479 option, the output file will be created in the `elf32-littlemips'
2483 The `TARGET' command names the BFD format to use when reading input
2484 files. It affects subsequent `INPUT' and `GROUP' commands. This
2485 command is like using `-b BFDNAME' on the command line (*note
2486 Command Line Options: Options.). If the `TARGET' command is used
2487 but `OUTPUT_FORMAT' is not, then the last `TARGET' command is also
2488 used to set the format for the output file. *Note BFD::.
2491 File: ld.info, Node: REGION_ALIAS, Next: Miscellaneous Commands, Prev: Format Commands, Up: Simple Commands
2493 3.4.4 Assign alias names to memory regions
2494 ------------------------------------------
2496 Alias names can be added to existing memory regions created with the
2497 *Note MEMORY:: command. Each name corresponds to at most one memory
2500 REGION_ALIAS(ALIAS, REGION)
2502 The `REGION_ALIAS' function creates an alias name ALIAS for the
2503 memory region REGION. This allows a flexible mapping of output sections
2504 to memory regions. An example follows.
2506 Suppose we have an application for embedded systems which come with
2507 various memory storage devices. All have a general purpose, volatile
2508 memory `RAM' that allows code execution or data storage. Some may have
2509 a read-only, non-volatile memory `ROM' that allows code execution and
2510 read-only data access. The last variant is a read-only, non-volatile
2511 memory `ROM2' with read-only data access and no code execution
2512 capability. We have four output sections:
2514 * `.text' program code;
2516 * `.rodata' read-only data;
2518 * `.data' read-write initialized data;
2520 * `.bss' read-write zero initialized data.
2522 The goal is to provide a linker command file that contains a system
2523 independent part defining the output sections and a system dependent
2524 part mapping the output sections to the memory regions available on the
2525 system. Our embedded systems come with three different memory setups
2527 Section Variant A Variant B Variant C
2529 .rodata RAM ROM ROM2
2530 .data RAM RAM/ROM RAM/ROM2
2532 The notation `RAM/ROM' or `RAM/ROM2' means that this section is
2533 loaded into region `ROM' or `ROM2' respectively. Please note that the
2534 load address of the `.data' section starts in all three variants at the
2535 end of the `.rodata' section.
2537 The base linker script that deals with the output sections follows.
2538 It includes the system dependent `linkcmds.memory' file that describes
2540 INCLUDE linkcmds.memory
2553 .data : AT (rodata_end)
2558 data_size = SIZEOF(.data);
2559 data_load_start = LOADADDR(.data);
2566 Now we need three different `linkcmds.memory' files to define memory
2567 regions and alias names. The content of `linkcmds.memory' for the three
2568 variants `A', `B' and `C':
2570 Here everything goes into the `RAM'.
2573 RAM : ORIGIN = 0, LENGTH = 4M
2576 REGION_ALIAS("REGION_TEXT", RAM);
2577 REGION_ALIAS("REGION_RODATA", RAM);
2578 REGION_ALIAS("REGION_DATA", RAM);
2579 REGION_ALIAS("REGION_BSS", RAM);
2582 Program code and read-only data go into the `ROM'. Read-write
2583 data goes into the `RAM'. An image of the initialized data is
2584 loaded into the `ROM' and will be copied during system start into
2588 ROM : ORIGIN = 0, LENGTH = 3M
2589 RAM : ORIGIN = 0x10000000, LENGTH = 1M
2592 REGION_ALIAS("REGION_TEXT", ROM);
2593 REGION_ALIAS("REGION_RODATA", ROM);
2594 REGION_ALIAS("REGION_DATA", RAM);
2595 REGION_ALIAS("REGION_BSS", RAM);
2598 Program code goes into the `ROM'. Read-only data goes into the
2599 `ROM2'. Read-write data goes into the `RAM'. An image of the
2600 initialized data is loaded into the `ROM2' and will be copied
2601 during system start into the `RAM'.
2604 ROM : ORIGIN = 0, LENGTH = 2M
2605 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
2606 RAM : ORIGIN = 0x20000000, LENGTH = 1M
2609 REGION_ALIAS("REGION_TEXT", ROM);
2610 REGION_ALIAS("REGION_RODATA", ROM2);
2611 REGION_ALIAS("REGION_DATA", RAM);
2612 REGION_ALIAS("REGION_BSS", RAM);
2614 It is possible to write a common system initialization routine to
2615 copy the `.data' section from `ROM' or `ROM2' into the `RAM' if
2619 extern char data_start [];
2620 extern char data_size [];
2621 extern char data_load_start [];
2623 void copy_data(void)
2625 if (data_start != data_load_start)
2627 memcpy(data_start, data_load_start, (size_t) data_size);
2632 File: ld.info, Node: Miscellaneous Commands, Prev: REGION_ALIAS, Up: Simple Commands
2634 3.4.5 Other Linker Script Commands
2635 ----------------------------------
2637 There are a few other linker scripts commands.
2639 `ASSERT(EXP, MESSAGE)'
2640 Ensure that EXP is non-zero. If it is zero, then exit the linker
2641 with an error code, and print MESSAGE.
2643 `EXTERN(SYMBOL SYMBOL ...)'
2644 Force SYMBOL to be entered in the output file as an undefined
2645 symbol. Doing this may, for example, trigger linking of additional
2646 modules from standard libraries. You may list several SYMBOLs for
2647 each `EXTERN', and you may use `EXTERN' multiple times. This
2648 command has the same effect as the `-u' command-line option.
2650 `FORCE_COMMON_ALLOCATION'
2651 This command has the same effect as the `-d' command-line option:
2652 to make `ld' assign space to common symbols even if a relocatable
2653 output file is specified (`-r').
2655 `INHIBIT_COMMON_ALLOCATION'
2656 This command has the same effect as the `--no-define-common'
2657 command-line option: to make `ld' omit the assignment of addresses
2658 to common symbols even for a non-relocatable output file.
2660 `INSERT [ AFTER | BEFORE ] OUTPUT_SECTION'
2661 This command is typically used in a script specified by `-T' to
2662 augment the default `SECTIONS' with, for example, overlays. It
2663 inserts all prior linker script statements after (or before)
2664 OUTPUT_SECTION, and also causes `-T' to not override the default
2665 linker script. The exact insertion point is as for orphan
2666 sections. *Note Location Counter::. The insertion happens after
2667 the linker has mapped input sections to output sections. Prior to
2668 the insertion, since `-T' scripts are parsed before the default
2669 linker script, statements in the `-T' script occur before the
2670 default linker script statements in the internal linker
2671 representation of the script. In particular, input section
2672 assignments will be made to `-T' output sections before those in
2673 the default script. Here is an example of how a `-T' script using
2674 `INSERT' might look:
2680 .ov1 { ov1*(.text) }
2681 .ov2 { ov2*(.text) }
2686 `NOCROSSREFS(SECTION SECTION ...)'
2687 This command may be used to tell `ld' to issue an error about any
2688 references among certain output sections.
2690 In certain types of programs, particularly on embedded systems when
2691 using overlays, when one section is loaded into memory, another
2692 section will not be. Any direct references between the two
2693 sections would be errors. For example, it would be an error if
2694 code in one section called a function defined in the other section.
2696 The `NOCROSSREFS' command takes a list of output section names. If
2697 `ld' detects any cross references between the sections, it reports
2698 an error and returns a non-zero exit status. Note that the
2699 `NOCROSSREFS' command uses output section names, not input section
2702 `OUTPUT_ARCH(BFDARCH)'
2703 Specify a particular output machine architecture. The argument is
2704 one of the names used by the BFD library (*note BFD::). You can
2705 see the architecture of an object file by using the `objdump'
2706 program with the `-f' option.
2708 `LD_FEATURE(STRING)'
2709 This command may be used to modify `ld' behavior. If STRING is
2710 `"SANE_EXPR"' then absolute symbols and numbers in a script are
2711 simply treated as numbers everywhere. *Note Expression Section::.
2714 File: ld.info, Node: Assignments, Next: SECTIONS, Prev: Simple Commands, Up: Scripts
2716 3.5 Assigning Values to Symbols
2717 ===============================
2719 You may assign a value to a symbol in a linker script. This will define
2720 the symbol and place it into the symbol table with a global scope.
2724 * Simple Assignments:: Simple Assignments
2726 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2727 * Source Code Reference:: How to use a linker script defined symbol in source code
2730 File: ld.info, Node: Simple Assignments, Next: PROVIDE, Up: Assignments
2732 3.5.1 Simple Assignments
2733 ------------------------
2735 You may assign to a symbol using any of the C assignment operators:
2737 `SYMBOL = EXPRESSION ;'
2738 `SYMBOL += EXPRESSION ;'
2739 `SYMBOL -= EXPRESSION ;'
2740 `SYMBOL *= EXPRESSION ;'
2741 `SYMBOL /= EXPRESSION ;'
2742 `SYMBOL <<= EXPRESSION ;'
2743 `SYMBOL >>= EXPRESSION ;'
2744 `SYMBOL &= EXPRESSION ;'
2745 `SYMBOL |= EXPRESSION ;'
2747 The first case will define SYMBOL to the value of EXPRESSION. In
2748 the other cases, SYMBOL must already be defined, and the value will be
2749 adjusted accordingly.
2751 The special symbol name `.' indicates the location counter. You may
2752 only use this within a `SECTIONS' command. *Note Location Counter::.
2754 The semicolon after EXPRESSION is required.
2756 Expressions are defined below; see *Note Expressions::.
2758 You may write symbol assignments as commands in their own right, or
2759 as statements within a `SECTIONS' command, or as part of an output
2760 section description in a `SECTIONS' command.
2762 The section of the symbol will be set from the section of the
2763 expression; for more information, see *Note Expression Section::.
2765 Here is an example showing the three different places that symbol
2766 assignments may be used:
2776 _bdata = (. + 3) & ~ 3;
2777 .data : { *(.data) }
2779 In this example, the symbol `floating_point' will be defined as
2780 zero. The symbol `_etext' will be defined as the address following the
2781 last `.text' input section. The symbol `_bdata' will be defined as the
2782 address following the `.text' output section aligned upward to a 4 byte
2786 File: ld.info, Node: PROVIDE, Next: PROVIDE_HIDDEN, Prev: Simple Assignments, Up: Assignments
2791 In some cases, it is desirable for a linker script to define a symbol
2792 only if it is referenced and is not defined by any object included in
2793 the link. For example, traditional linkers defined the symbol `etext'.
2794 However, ANSI C requires that the user be able to use `etext' as a
2795 function name without encountering an error. The `PROVIDE' keyword may
2796 be used to define a symbol, such as `etext', only if it is referenced
2797 but not defined. The syntax is `PROVIDE(SYMBOL = EXPRESSION)'.
2799 Here is an example of using `PROVIDE' to define `etext':
2810 In this example, if the program defines `_etext' (with a leading
2811 underscore), the linker will give a multiple definition error. If, on
2812 the other hand, the program defines `etext' (with no leading
2813 underscore), the linker will silently use the definition in the program.
2814 If the program references `etext' but does not define it, the linker
2815 will use the definition in the linker script.
2818 File: ld.info, Node: PROVIDE_HIDDEN, Next: Source Code Reference, Prev: PROVIDE, Up: Assignments
2820 3.5.3 PROVIDE_HIDDEN
2821 --------------------
2823 Similar to `PROVIDE'. For ELF targeted ports, the symbol will be
2824 hidden and won't be exported.
2827 File: ld.info, Node: Source Code Reference, Prev: PROVIDE_HIDDEN, Up: Assignments
2829 3.5.4 Source Code Reference
2830 ---------------------------
2832 Accessing a linker script defined variable from source code is not
2833 intuitive. In particular a linker script symbol is not equivalent to a
2834 variable declaration in a high level language, it is instead a symbol
2835 that does not have a value.
2837 Before going further, it is important to note that compilers often
2838 transform names in the source code into different names when they are
2839 stored in the symbol table. For example, Fortran compilers commonly
2840 prepend or append an underscore, and C++ performs extensive `name
2841 mangling'. Therefore there might be a discrepancy between the name of
2842 a variable as it is used in source code and the name of the same
2843 variable as it is defined in a linker script. For example in C a
2844 linker script variable might be referred to as:
2848 But in the linker script it might be defined as:
2852 In the remaining examples however it is assumed that no name
2853 transformation has taken place.
2855 When a symbol is declared in a high level language such as C, two
2856 things happen. The first is that the compiler reserves enough space in
2857 the program's memory to hold the _value_ of the symbol. The second is
2858 that the compiler creates an entry in the program's symbol table which
2859 holds the symbol's _address_. ie the symbol table contains the address
2860 of the block of memory holding the symbol's value. So for example the
2861 following C declaration, at file scope:
2865 creates a entry called `foo' in the symbol table. This entry holds
2866 the address of an `int' sized block of memory where the number 1000 is
2869 When a program references a symbol the compiler generates code that
2870 first accesses the symbol table to find the address of the symbol's
2871 memory block and then code to read the value from that memory block.
2876 looks up the symbol `foo' in the symbol table, gets the address
2877 associated with this symbol and then writes the value 1 into that
2882 looks up the symbol `foo' in the symbol table, gets it address and
2883 then copies this address into the block of memory associated with the
2886 Linker scripts symbol declarations, by contrast, create an entry in
2887 the symbol table but do not assign any memory to them. Thus they are
2888 an address without a value. So for example the linker script
2893 creates an entry in the symbol table called `foo' which holds the
2894 address of memory location 1000, but nothing special is stored at
2895 address 1000. This means that you cannot access the _value_ of a
2896 linker script defined symbol - it has no value - all you can do is
2897 access the _address_ of a linker script defined symbol.
2899 Hence when you are using a linker script defined symbol in source
2900 code you should always take the address of the symbol, and never
2901 attempt to use its value. For example suppose you want to copy the
2902 contents of a section of memory called .ROM into a section called
2903 .FLASH and the linker script contains these declarations:
2905 start_of_ROM = .ROM;
2906 end_of_ROM = .ROM + sizeof (.ROM) - 1;
2907 start_of_FLASH = .FLASH;
2909 Then the C source code to perform the copy would be:
2911 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
2913 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
2915 Note the use of the `&' operators. These are correct.
2918 File: ld.info, Node: SECTIONS, Next: MEMORY, Prev: Assignments, Up: Scripts
2920 3.6 SECTIONS Command
2921 ====================
2923 The `SECTIONS' command tells the linker how to map input sections into
2924 output sections, and how to place the output sections in memory.
2926 The format of the `SECTIONS' command is:
2934 Each SECTIONS-COMMAND may of be one of the following:
2936 * an `ENTRY' command (*note Entry command: Entry Point.)
2938 * a symbol assignment (*note Assignments::)
2940 * an output section description
2942 * an overlay description
2944 The `ENTRY' command and symbol assignments are permitted inside the
2945 `SECTIONS' command for convenience in using the location counter in
2946 those commands. This can also make the linker script easier to
2947 understand because you can use those commands at meaningful points in
2948 the layout of the output file.
2950 Output section descriptions and overlay descriptions are described
2953 If you do not use a `SECTIONS' command in your linker script, the
2954 linker will place each input section into an identically named output
2955 section in the order that the sections are first encountered in the
2956 input files. If all input sections are present in the first file, for
2957 example, the order of sections in the output file will match the order
2958 in the first input file. The first section will be at address zero.
2962 * Output Section Description:: Output section description
2963 * Output Section Name:: Output section name
2964 * Output Section Address:: Output section address
2965 * Input Section:: Input section description
2966 * Output Section Data:: Output section data
2967 * Output Section Keywords:: Output section keywords
2968 * Output Section Discarding:: Output section discarding
2969 * Output Section Attributes:: Output section attributes
2970 * Overlay Description:: Overlay description
2973 File: ld.info, Node: Output Section Description, Next: Output Section Name, Up: SECTIONS
2975 3.6.1 Output Section Description
2976 --------------------------------
2978 The full description of an output section looks like this:
2979 SECTION [ADDRESS] [(TYPE)] :
2981 [ALIGN(SECTION_ALIGN)]
2982 [SUBALIGN(SUBSECTION_ALIGN)]
2985 OUTPUT-SECTION-COMMAND
2986 OUTPUT-SECTION-COMMAND
2988 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
2990 Most output sections do not use most of the optional section
2993 The whitespace around SECTION is required, so that the section name
2994 is unambiguous. The colon and the curly braces are also required. The
2995 line breaks and other white space are optional.
2997 Each OUTPUT-SECTION-COMMAND may be one of the following:
2999 * a symbol assignment (*note Assignments::)
3001 * an input section description (*note Input Section::)
3003 * data values to include directly (*note Output Section Data::)
3005 * a special output section keyword (*note Output Section Keywords::)
3008 File: ld.info, Node: Output Section Name, Next: Output Section Address, Prev: Output Section Description, Up: SECTIONS
3010 3.6.2 Output Section Name
3011 -------------------------
3013 The name of the output section is SECTION. SECTION must meet the
3014 constraints of your output format. In formats which only support a
3015 limited number of sections, such as `a.out', the name must be one of
3016 the names supported by the format (`a.out', for example, allows only
3017 `.text', `.data' or `.bss'). If the output format supports any number
3018 of sections, but with numbers and not names (as is the case for Oasys),
3019 the name should be supplied as a quoted numeric string. A section name
3020 may consist of any sequence of characters, but a name which contains
3021 any unusual characters such as commas must be quoted.
3023 The output section name `/DISCARD/' is special; *Note Output Section
3027 File: ld.info, Node: Output Section Address, Next: Input Section, Prev: Output Section Name, Up: SECTIONS
3029 3.6.3 Output Section Address
3030 ----------------------------
3032 The ADDRESS is an expression for the VMA (the virtual memory address)
3033 of the output section. This address is optional, but if it is provided
3034 then the output address will be set exactly as specified.
3036 If the output address is not specified then one will be chosen for
3037 the section, based on the heuristic below. This address will be
3038 adjusted to fit the alignment requirement of the output section. The
3039 alignment requirement is the strictest alignment of any input section
3040 contained within the output section.
3042 The output section address heuristic is as follows:
3044 * If an output memory REGION is set for the section then it is added
3045 to this region and its address will be the next free address in
3048 * If the MEMORY command has been used to create a list of memory
3049 regions then the first region which has attributes compatible with
3050 the section is selected to contain it. The section's output
3051 address will be the next free address in that region; *Note
3054 * If no memory regions were specified, or none match the section then
3055 the output address will be based on the current value of the
3060 .text . : { *(.text) }
3064 .text : { *(.text) }
3066 are subtly different. The first will set the address of the `.text'
3067 output section to the current value of the location counter. The
3068 second will set it to the current value of the location counter aligned
3069 to the strictest alignment of any of the `.text' input sections.
3071 The ADDRESS may be an arbitrary expression; *Note Expressions::.
3072 For example, if you want to align the section on a 0x10 byte boundary,
3073 so that the lowest four bits of the section address are zero, you could
3074 do something like this:
3075 .text ALIGN(0x10) : { *(.text) }
3076 This works because `ALIGN' returns the current location counter
3077 aligned upward to the specified value.
3079 Specifying ADDRESS for a section will change the value of the
3080 location counter, provided that the section is non-empty. (Empty
3081 sections are ignored).
3084 File: ld.info, Node: Input Section, Next: Output Section Data, Prev: Output Section Address, Up: SECTIONS
3086 3.6.4 Input Section Description
3087 -------------------------------
3089 The most common output section command is an input section description.
3091 The input section description is the most basic linker script
3092 operation. You use output sections to tell the linker how to lay out
3093 your program in memory. You use input section descriptions to tell the
3094 linker how to map the input files into your memory layout.
3098 * Input Section Basics:: Input section basics
3099 * Input Section Wildcards:: Input section wildcard patterns
3100 * Input Section Common:: Input section for common symbols
3101 * Input Section Keep:: Input section and garbage collection
3102 * Input Section Example:: Input section example
3105 File: ld.info, Node: Input Section Basics, Next: Input Section Wildcards, Up: Input Section
3107 3.6.4.1 Input Section Basics
3108 ............................
3110 An input section description consists of a file name optionally followed
3111 by a list of section names in parentheses.
3113 The file name and the section name may be wildcard patterns, which we
3114 describe further below (*note Input Section Wildcards::).
3116 The most common input section description is to include all input
3117 sections with a particular name in the output section. For example, to
3118 include all input `.text' sections, you would write:
3120 Here the `*' is a wildcard which matches any file name. To exclude
3121 a list of files from matching the file name wildcard, EXCLUDE_FILE may
3122 be used to match all files except the ones specified in the
3123 EXCLUDE_FILE list. For example:
3124 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3125 will cause all .ctors sections from all files except `crtend.o' and
3126 `otherfile.o' to be included.
3128 There are two ways to include more than one section:
3131 The difference between these is the order in which the `.text' and
3132 `.rdata' input sections will appear in the output section. In the
3133 first example, they will be intermingled, appearing in the same order as
3134 they are found in the linker input. In the second example, all `.text'
3135 input sections will appear first, followed by all `.rdata' input
3138 You can specify a file name to include sections from a particular
3139 file. You would do this if one or more of your files contain special
3140 data that needs to be at a particular location in memory. For example:
3143 You can also specify files within archives by writing a pattern
3144 matching the archive, a colon, then the pattern matching the file, with
3145 no whitespace around the colon.
3148 matches file within archive
3151 matches the whole archive
3154 matches file but not one in an archive
3156 Either one or both of `archive' and `file' can contain shell
3157 wildcards. On DOS based file systems, the linker will assume that a
3158 single letter followed by a colon is a drive specifier, so `c:myfile.o'
3159 is a simple file specification, not `myfile.o' within an archive called
3160 `c'. `archive:file' filespecs may also be used within an
3161 `EXCLUDE_FILE' list, but may not appear in other linker script
3162 contexts. For instance, you cannot extract a file from an archive by
3163 using `archive:file' in an `INPUT' command.
3165 If you use a file name without a list of sections, then all sections
3166 in the input file will be included in the output section. This is not
3167 commonly done, but it may by useful on occasion. For example:
3170 When you use a file name which is not an `archive:file' specifier
3171 and does not contain any wild card characters, the linker will first
3172 see if you also specified the file name on the linker command line or
3173 in an `INPUT' command. If you did not, the linker will attempt to open
3174 the file as an input file, as though it appeared on the command line.
3175 Note that this differs from an `INPUT' command, because the linker will
3176 not search for the file in the archive search path.
3179 File: ld.info, Node: Input Section Wildcards, Next: Input Section Common, Prev: Input Section Basics, Up: Input Section
3181 3.6.4.2 Input Section Wildcard Patterns
3182 .......................................
3184 In an input section description, either the file name or the section
3185 name or both may be wildcard patterns.
3187 The file name of `*' seen in many examples is a simple wildcard
3188 pattern for the file name.
3190 The wildcard patterns are like those used by the Unix shell.
3193 matches any number of characters
3196 matches any single character
3199 matches a single instance of any of the CHARS; the `-' character
3200 may be used to specify a range of characters, as in `[a-z]' to
3201 match any lower case letter
3204 quotes the following character
3206 When a file name is matched with a wildcard, the wildcard characters
3207 will not match a `/' character (used to separate directory names on
3208 Unix). A pattern consisting of a single `*' character is an exception;
3209 it will always match any file name, whether it contains a `/' or not.
3210 In a section name, the wildcard characters will match a `/' character.
3212 File name wildcard patterns only match files which are explicitly
3213 specified on the command line or in an `INPUT' command. The linker
3214 does not search directories to expand wildcards.
3216 If a file name matches more than one wildcard pattern, or if a file
3217 name appears explicitly and is also matched by a wildcard pattern, the
3218 linker will use the first match in the linker script. For example, this
3219 sequence of input section descriptions is probably in error, because the
3220 `data.o' rule will not be used:
3221 .data : { *(.data) }
3222 .data1 : { data.o(.data) }
3224 Normally, the linker will place files and sections matched by
3225 wildcards in the order in which they are seen during the link. You can
3226 change this by using the `SORT_BY_NAME' keyword, which appears before a
3227 wildcard pattern in parentheses (e.g., `SORT_BY_NAME(.text*)'). When
3228 the `SORT_BY_NAME' keyword is used, the linker will sort the files or
3229 sections into ascending order by name before placing them in the output
3232 `SORT_BY_ALIGNMENT' is very similar to `SORT_BY_NAME'. The
3233 difference is `SORT_BY_ALIGNMENT' will sort sections into ascending
3234 order by alignment before placing them in the output file.
3236 `SORT' is an alias for `SORT_BY_NAME'.
3238 When there are nested section sorting commands in linker script,
3239 there can be at most 1 level of nesting for section sorting commands.
3241 1. `SORT_BY_NAME' (`SORT_BY_ALIGNMENT' (wildcard section pattern)).
3242 It will sort the input sections by name first, then by alignment
3243 if 2 sections have the same name.
3245 2. `SORT_BY_ALIGNMENT' (`SORT_BY_NAME' (wildcard section pattern)).
3246 It will sort the input sections by alignment first, then by name
3247 if 2 sections have the same alignment.
3249 3. `SORT_BY_NAME' (`SORT_BY_NAME' (wildcard section pattern)) is
3250 treated the same as `SORT_BY_NAME' (wildcard section pattern).
3252 4. `SORT_BY_ALIGNMENT' (`SORT_BY_ALIGNMENT' (wildcard section
3253 pattern)) is treated the same as `SORT_BY_ALIGNMENT' (wildcard
3256 5. All other nested section sorting commands are invalid.
3258 When both command line section sorting option and linker script
3259 section sorting command are used, section sorting command always takes
3260 precedence over the command line option.
3262 If the section sorting command in linker script isn't nested, the
3263 command line option will make the section sorting command to be treated
3264 as nested sorting command.
3266 1. `SORT_BY_NAME' (wildcard section pattern ) with `--sort-sections
3267 alignment' is equivalent to `SORT_BY_NAME' (`SORT_BY_ALIGNMENT'
3268 (wildcard section pattern)).
3270 2. `SORT_BY_ALIGNMENT' (wildcard section pattern) with
3271 `--sort-section name' is equivalent to `SORT_BY_ALIGNMENT'
3272 (`SORT_BY_NAME' (wildcard section pattern)).
3274 If the section sorting command in linker script is nested, the
3275 command line option will be ignored.
3277 If you ever get confused about where input sections are going, use
3278 the `-M' linker option to generate a map file. The map file shows
3279 precisely how input sections are mapped to output sections.
3281 This example shows how wildcard patterns might be used to partition
3282 files. This linker script directs the linker to place all `.text'
3283 sections in `.text' and all `.bss' sections in `.bss'. The linker will
3284 place the `.data' section from all files beginning with an upper case
3285 character in `.DATA'; for all other files, the linker will place the
3286 `.data' section in `.data'.
3288 .text : { *(.text) }
3289 .DATA : { [A-Z]*(.data) }
3290 .data : { *(.data) }
3295 File: ld.info, Node: Input Section Common, Next: Input Section Keep, Prev: Input Section Wildcards, Up: Input Section
3297 3.6.4.3 Input Section for Common Symbols
3298 ........................................
3300 A special notation is needed for common symbols, because in many object
3301 file formats common symbols do not have a particular input section. The
3302 linker treats common symbols as though they are in an input section
3305 You may use file names with the `COMMON' section just as with any
3306 other input sections. You can use this to place common symbols from a
3307 particular input file in one section while common symbols from other
3308 input files are placed in another section.
3310 In most cases, common symbols in input files will be placed in the
3311 `.bss' section in the output file. For example:
3312 .bss { *(.bss) *(COMMON) }
3314 Some object file formats have more than one type of common symbol.
3315 For example, the MIPS ELF object file format distinguishes standard
3316 common symbols and small common symbols. In this case, the linker will
3317 use a different special section name for other types of common symbols.
3318 In the case of MIPS ELF, the linker uses `COMMON' for standard common
3319 symbols and `.scommon' for small common symbols. This permits you to
3320 map the different types of common symbols into memory at different
3323 You will sometimes see `[COMMON]' in old linker scripts. This
3324 notation is now considered obsolete. It is equivalent to `*(COMMON)'.
3327 File: ld.info, Node: Input Section Keep, Next: Input Section Example, Prev: Input Section Common, Up: Input Section
3329 3.6.4.4 Input Section and Garbage Collection
3330 ............................................
3332 When link-time garbage collection is in use (`--gc-sections'), it is
3333 often useful to mark sections that should not be eliminated. This is
3334 accomplished by surrounding an input section's wildcard entry with
3335 `KEEP()', as in `KEEP(*(.init))' or `KEEP(SORT_BY_NAME(*)(.ctors))'.
3338 File: ld.info, Node: Input Section Example, Prev: Input Section Keep, Up: Input Section
3340 3.6.4.5 Input Section Example
3341 .............................
3343 The following example is a complete linker script. It tells the linker
3344 to read all of the sections from file `all.o' and place them at the
3345 start of output section `outputa' which starts at location `0x10000'.
3346 All of section `.input1' from file `foo.o' follows immediately, in the
3347 same output section. All of section `.input2' from `foo.o' goes into
3348 output section `outputb', followed by section `.input1' from `foo1.o'.
3349 All of the remaining `.input1' and `.input2' sections from any files
3350 are written to output section `outputc'.
3371 File: ld.info, Node: Output Section Data, Next: Output Section Keywords, Prev: Input Section, Up: SECTIONS
3373 3.6.5 Output Section Data
3374 -------------------------
3376 You can include explicit bytes of data in an output section by using
3377 `BYTE', `SHORT', `LONG', `QUAD', or `SQUAD' as an output section
3378 command. Each keyword is followed by an expression in parentheses
3379 providing the value to store (*note Expressions::). The value of the
3380 expression is stored at the current value of the location counter.
3382 The `BYTE', `SHORT', `LONG', and `QUAD' commands store one, two,
3383 four, and eight bytes (respectively). After storing the bytes, the
3384 location counter is incremented by the number of bytes stored.
3386 For example, this will store the byte 1 followed by the four byte
3387 value of the symbol `addr':
3391 When using a 64 bit host or target, `QUAD' and `SQUAD' are the same;
3392 they both store an 8 byte, or 64 bit, value. When both host and target
3393 are 32 bits, an expression is computed as 32 bits. In this case `QUAD'
3394 stores a 32 bit value zero extended to 64 bits, and `SQUAD' stores a 32
3395 bit value sign extended to 64 bits.
3397 If the object file format of the output file has an explicit
3398 endianness, which is the normal case, the value will be stored in that
3399 endianness. When the object file format does not have an explicit
3400 endianness, as is true of, for example, S-records, the value will be
3401 stored in the endianness of the first input object file.
3403 Note--these commands only work inside a section description and not
3404 between them, so the following will produce an error from the linker:
3405 SECTIONS { .text : { *(.text) } LONG(1) .data : { *(.data) } }
3406 whereas this will work:
3407 SECTIONS { .text : { *(.text) ; LONG(1) } .data : { *(.data) } }
3409 You may use the `FILL' command to set the fill pattern for the
3410 current section. It is followed by an expression in parentheses. Any
3411 otherwise unspecified regions of memory within the section (for example,
3412 gaps left due to the required alignment of input sections) are filled
3413 with the value of the expression, repeated as necessary. A `FILL'
3414 statement covers memory locations after the point at which it occurs in
3415 the section definition; by including more than one `FILL' statement,
3416 you can have different fill patterns in different parts of an output
3419 This example shows how to fill unspecified regions of memory with the
3423 The `FILL' command is similar to the `=FILLEXP' output section
3424 attribute, but it only affects the part of the section following the
3425 `FILL' command, rather than the entire section. If both are used, the
3426 `FILL' command takes precedence. *Note Output Section Fill::, for
3427 details on the fill expression.
3430 File: ld.info, Node: Output Section Keywords, Next: Output Section Discarding, Prev: Output Section Data, Up: SECTIONS
3432 3.6.6 Output Section Keywords
3433 -----------------------------
3435 There are a couple of keywords which can appear as output section
3438 `CREATE_OBJECT_SYMBOLS'
3439 The command tells the linker to create a symbol for each input
3440 file. The name of each symbol will be the name of the
3441 corresponding input file. The section of each symbol will be the
3442 output section in which the `CREATE_OBJECT_SYMBOLS' command
3445 This is conventional for the a.out object file format. It is not
3446 normally used for any other object file format.
3449 When linking using the a.out object file format, the linker uses an
3450 unusual set construct to support C++ global constructors and
3451 destructors. When linking object file formats which do not support
3452 arbitrary sections, such as ECOFF and XCOFF, the linker will
3453 automatically recognize C++ global constructors and destructors by
3454 name. For these object file formats, the `CONSTRUCTORS' command
3455 tells the linker to place constructor information in the output
3456 section where the `CONSTRUCTORS' command appears. The
3457 `CONSTRUCTORS' command is ignored for other object file formats.
3459 The symbol `__CTOR_LIST__' marks the start of the global
3460 constructors, and the symbol `__CTOR_END__' marks the end.
3461 Similarly, `__DTOR_LIST__' and `__DTOR_END__' mark the start and
3462 end of the global destructors. The first word in the list is the
3463 number of entries, followed by the address of each constructor or
3464 destructor, followed by a zero word. The compiler must arrange to
3465 actually run the code. For these object file formats GNU C++
3466 normally calls constructors from a subroutine `__main'; a call to
3467 `__main' is automatically inserted into the startup code for
3468 `main'. GNU C++ normally runs destructors either by using
3469 `atexit', or directly from the function `exit'.
3471 For object file formats such as `COFF' or `ELF' which support
3472 arbitrary section names, GNU C++ will normally arrange to put the
3473 addresses of global constructors and destructors into the `.ctors'
3474 and `.dtors' sections. Placing the following sequence into your
3475 linker script will build the sort of table which the GNU C++
3476 runtime code expects to see.
3479 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3484 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3489 If you are using the GNU C++ support for initialization priority,
3490 which provides some control over the order in which global
3491 constructors are run, you must sort the constructors at link time
3492 to ensure that they are executed in the correct order. When using
3493 the `CONSTRUCTORS' command, use `SORT_BY_NAME(CONSTRUCTORS)'
3494 instead. When using the `.ctors' and `.dtors' sections, use
3495 `*(SORT_BY_NAME(.ctors))' and `*(SORT_BY_NAME(.dtors))' instead of
3496 just `*(.ctors)' and `*(.dtors)'.
3498 Normally the compiler and linker will handle these issues
3499 automatically, and you will not need to concern yourself with
3500 them. However, you may need to consider this if you are using C++
3501 and writing your own linker scripts.
3505 File: ld.info, Node: Output Section Discarding, Next: Output Section Attributes, Prev: Output Section Keywords, Up: SECTIONS
3507 3.6.7 Output Section Discarding
3508 -------------------------------
3510 The linker will not create output sections with no contents. This is
3511 for convenience when referring to input sections that may or may not be
3512 present in any of the input files. For example:
3514 will only create a `.foo' section in the output file if there is a
3515 `.foo' section in at least one input file, and if the input sections
3516 are not all empty. Other link script directives that allocate space in
3517 an output section will also create the output section.
3519 The linker will ignore address assignments (*note Output Section
3520 Address::) on discarded output sections, except when the linker script
3521 defines symbols in the output section. In that case the linker will
3522 obey the address assignments, possibly advancing dot even though the
3523 section is discarded.
3525 The special output section name `/DISCARD/' may be used to discard
3526 input sections. Any input sections which are assigned to an output
3527 section named `/DISCARD/' are not included in the output file.
3530 File: ld.info, Node: Output Section Attributes, Next: Overlay Description, Prev: Output Section Discarding, Up: SECTIONS
3532 3.6.8 Output Section Attributes
3533 -------------------------------
3535 We showed above that the full description of an output section looked
3538 SECTION [ADDRESS] [(TYPE)] :
3540 [ALIGN(SECTION_ALIGN)]
3541 [SUBALIGN(SUBSECTION_ALIGN)]
3544 OUTPUT-SECTION-COMMAND
3545 OUTPUT-SECTION-COMMAND
3547 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
3549 We've already described SECTION, ADDRESS, and
3550 OUTPUT-SECTION-COMMAND. In this section we will describe the remaining
3555 * Output Section Type:: Output section type
3556 * Output Section LMA:: Output section LMA
3557 * Forced Output Alignment:: Forced Output Alignment
3558 * Forced Input Alignment:: Forced Input Alignment
3559 * Output Section Constraint:: Output section constraint
3560 * Output Section Region:: Output section region
3561 * Output Section Phdr:: Output section phdr
3562 * Output Section Fill:: Output section fill
3565 File: ld.info, Node: Output Section Type, Next: Output Section LMA, Up: Output Section Attributes
3567 3.6.8.1 Output Section Type
3568 ...........................
3570 Each output section may have a type. The type is a keyword in
3571 parentheses. The following types are defined:
3574 The section should be marked as not loadable, so that it will not
3575 be loaded into memory when the program is run.
3581 These type names are supported for backward compatibility, and are
3582 rarely used. They all have the same effect: the section should be
3583 marked as not allocatable, so that no memory is allocated for the
3584 section when the program is run.
3586 The linker normally sets the attributes of an output section based on
3587 the input sections which map into it. You can override this by using
3588 the section type. For example, in the script sample below, the `ROM'
3589 section is addressed at memory location `0' and does not need to be
3590 loaded when the program is run.
3592 ROM 0 (NOLOAD) : { ... }
3597 File: ld.info, Node: Output Section LMA, Next: Forced Output Alignment, Prev: Output Section Type, Up: Output Section Attributes
3599 3.6.8.2 Output Section LMA
3600 ..........................
3602 Every section has a virtual address (VMA) and a load address (LMA); see
3603 *Note Basic Script Concepts::. The virtual address is specified by the
3604 *note Output Section Address:: described earlier. The load address is
3605 specified by the `AT' or `AT>' keywords. Specifying a load address is
3608 The `AT' keyword takes an expression as an argument. This specifies
3609 the exact load address of the section. The `AT>' keyword takes the
3610 name of a memory region as an argument. *Note MEMORY::. The load
3611 address of the section is set to the next free address in the region,
3612 aligned to the section's alignment requirements.
3614 If neither `AT' nor `AT>' is specified for an allocatable section,
3615 the linker will use the following heuristic to determine the load
3618 * If the section has a specific VMA address, then this is used as
3619 the LMA address as well.
3621 * If the section is not allocatable then its LMA is set to its VMA.
3623 * Otherwise if a memory region can be found that is compatible with
3624 the current section, and this region contains at least one
3625 section, then the LMA is set so the difference between the VMA and
3626 LMA is the same as the difference between the VMA and LMA of the
3627 last section in the located region.
3629 * If no memory regions have been declared then a default region that
3630 covers the entire address space is used in the previous step.
3632 * If no suitable region could be found, or there was no previous
3633 section then the LMA is set equal to the VMA.
3635 This feature is designed to make it easy to build a ROM image. For
3636 example, the following linker script creates three output sections: one
3637 called `.text', which starts at `0x1000', one called `.mdata', which is
3638 loaded at the end of the `.text' section even though its VMA is
3639 `0x2000', and one called `.bss' to hold uninitialized data at address
3640 `0x3000'. The symbol `_data' is defined with the value `0x2000', which
3641 shows that the location counter holds the VMA value, not the LMA value.
3645 .text 0x1000 : { *(.text) _etext = . ; }
3647 AT ( ADDR (.text) + SIZEOF (.text) )
3648 { _data = . ; *(.data); _edata = . ; }
3650 { _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;}
3653 The run-time initialization code for use with a program generated
3654 with this linker script would include something like the following, to
3655 copy the initialized data from the ROM image to its runtime address.
3656 Notice how this code takes advantage of the symbols defined by the
3659 extern char _etext, _data, _edata, _bstart, _bend;
3660 char *src = &_etext;
3663 /* ROM has data at end of text; copy it. */
3664 while (dst < &_edata)
3668 for (dst = &_bstart; dst< &_bend; dst++)
3672 File: ld.info, Node: Forced Output Alignment, Next: Forced Input Alignment, Prev: Output Section LMA, Up: Output Section Attributes
3674 3.6.8.3 Forced Output Alignment
3675 ...............................
3677 You can increase an output section's alignment by using ALIGN.
3680 File: ld.info, Node: Forced Input Alignment, Next: Output Section Constraint, Prev: Forced Output Alignment, Up: Output Section Attributes
3682 3.6.8.4 Forced Input Alignment
3683 ..............................
3685 You can force input section alignment within an output section by using
3686 SUBALIGN. The value specified overrides any alignment given by input
3687 sections, whether larger or smaller.
3690 File: ld.info, Node: Output Section Constraint, Next: Output Section Region, Prev: Forced Input Alignment, Up: Output Section Attributes
3692 3.6.8.5 Output Section Constraint
3693 .................................
3695 You can specify that an output section should only be created if all of
3696 its input sections are read-only or all of its input sections are
3697 read-write by using the keyword `ONLY_IF_RO' and `ONLY_IF_RW'
3701 File: ld.info, Node: Output Section Region, Next: Output Section Phdr, Prev: Output Section Constraint, Up: Output Section Attributes
3703 3.6.8.6 Output Section Region
3704 .............................
3706 You can assign a section to a previously defined region of memory by
3707 using `>REGION'. *Note MEMORY::.
3709 Here is a simple example:
3710 MEMORY { rom : ORIGIN = 0x1000, LENGTH = 0x1000 }
3711 SECTIONS { ROM : { *(.text) } >rom }
3714 File: ld.info, Node: Output Section Phdr, Next: Output Section Fill, Prev: Output Section Region, Up: Output Section Attributes
3716 3.6.8.7 Output Section Phdr
3717 ...........................
3719 You can assign a section to a previously defined program segment by
3720 using `:PHDR'. *Note PHDRS::. If a section is assigned to one or more
3721 segments, then all subsequent allocated sections will be assigned to
3722 those segments as well, unless they use an explicitly `:PHDR' modifier.
3723 You can use `:NONE' to tell the linker to not put the section in any
3726 Here is a simple example:
3727 PHDRS { text PT_LOAD ; }
3728 SECTIONS { .text : { *(.text) } :text }
3731 File: ld.info, Node: Output Section Fill, Prev: Output Section Phdr, Up: Output Section Attributes
3733 3.6.8.8 Output Section Fill
3734 ...........................
3736 You can set the fill pattern for an entire section by using `=FILLEXP'.
3737 FILLEXP is an expression (*note Expressions::). Any otherwise
3738 unspecified regions of memory within the output section (for example,
3739 gaps left due to the required alignment of input sections) will be
3740 filled with the value, repeated as necessary. If the fill expression
3741 is a simple hex number, ie. a string of hex digit starting with `0x'
3742 and without a trailing `k' or `M', then an arbitrarily long sequence of
3743 hex digits can be used to specify the fill pattern; Leading zeros
3744 become part of the pattern too. For all other cases, including extra
3745 parentheses or a unary `+', the fill pattern is the four least
3746 significant bytes of the value of the expression. In all cases, the
3747 number is big-endian.
3749 You can also change the fill value with a `FILL' command in the
3750 output section commands; (*note Output Section Data::).
3752 Here is a simple example:
3753 SECTIONS { .text : { *(.text) } =0x90909090 }
3756 File: ld.info, Node: Overlay Description, Prev: Output Section Attributes, Up: SECTIONS
3758 3.6.9 Overlay Description
3759 -------------------------
3761 An overlay description provides an easy way to describe sections which
3762 are to be loaded as part of a single memory image but are to be run at
3763 the same memory address. At run time, some sort of overlay manager will
3764 copy the overlaid sections in and out of the runtime memory address as
3765 required, perhaps by simply manipulating addressing bits. This approach
3766 can be useful, for example, when a certain region of memory is faster
3769 Overlays are described using the `OVERLAY' command. The `OVERLAY'
3770 command is used within a `SECTIONS' command, like an output section
3771 description. The full syntax of the `OVERLAY' command is as follows:
3772 OVERLAY [START] : [NOCROSSREFS] [AT ( LDADDR )]
3776 OUTPUT-SECTION-COMMAND
3777 OUTPUT-SECTION-COMMAND
3779 } [:PHDR...] [=FILL]
3782 OUTPUT-SECTION-COMMAND
3783 OUTPUT-SECTION-COMMAND
3785 } [:PHDR...] [=FILL]
3787 } [>REGION] [:PHDR...] [=FILL]
3789 Everything is optional except `OVERLAY' (a keyword), and each
3790 section must have a name (SECNAME1 and SECNAME2 above). The section
3791 definitions within the `OVERLAY' construct are identical to those
3792 within the general `SECTIONS' contruct (*note SECTIONS::), except that
3793 no addresses and no memory regions may be defined for sections within
3796 The sections are all defined with the same starting address. The
3797 load addresses of the sections are arranged such that they are
3798 consecutive in memory starting at the load address used for the
3799 `OVERLAY' as a whole (as with normal section definitions, the load
3800 address is optional, and defaults to the start address; the start
3801 address is also optional, and defaults to the current value of the
3804 If the `NOCROSSREFS' keyword is used, and there any references among
3805 the sections, the linker will report an error. Since the sections all
3806 run at the same address, it normally does not make sense for one
3807 section to refer directly to another. *Note NOCROSSREFS: Miscellaneous
3810 For each section within the `OVERLAY', the linker automatically
3811 provides two symbols. The symbol `__load_start_SECNAME' is defined as
3812 the starting load address of the section. The symbol
3813 `__load_stop_SECNAME' is defined as the final load address of the
3814 section. Any characters within SECNAME which are not legal within C
3815 identifiers are removed. C (or assembler) code may use these symbols
3816 to move the overlaid sections around as necessary.
3818 At the end of the overlay, the value of the location counter is set
3819 to the start address of the overlay plus the size of the largest
3822 Here is an example. Remember that this would appear inside a
3823 `SECTIONS' construct.
3824 OVERLAY 0x1000 : AT (0x4000)
3826 .text0 { o1/*.o(.text) }
3827 .text1 { o2/*.o(.text) }
3829 This will define both `.text0' and `.text1' to start at address
3830 0x1000. `.text0' will be loaded at address 0x4000, and `.text1' will
3831 be loaded immediately after `.text0'. The following symbols will be
3832 defined if referenced: `__load_start_text0', `__load_stop_text0',
3833 `__load_start_text1', `__load_stop_text1'.
3835 C code to copy overlay `.text1' into the overlay area might look
3838 extern char __load_start_text1, __load_stop_text1;
3839 memcpy ((char *) 0x1000, &__load_start_text1,
3840 &__load_stop_text1 - &__load_start_text1);
3842 Note that the `OVERLAY' command is just syntactic sugar, since
3843 everything it does can be done using the more basic commands. The above
3844 example could have been written identically as follows.
3846 .text0 0x1000 : AT (0x4000) { o1/*.o(.text) }
3847 PROVIDE (__load_start_text0 = LOADADDR (.text0));
3848 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
3849 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) { o2/*.o(.text) }
3850 PROVIDE (__load_start_text1 = LOADADDR (.text1));
3851 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
3852 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
3855 File: ld.info, Node: MEMORY, Next: PHDRS, Prev: SECTIONS, Up: Scripts
3860 The linker's default configuration permits allocation of all available
3861 memory. You can override this by using the `MEMORY' command.
3863 The `MEMORY' command describes the location and size of blocks of
3864 memory in the target. You can use it to describe which memory regions
3865 may be used by the linker, and which memory regions it must avoid. You
3866 can then assign sections to particular memory regions. The linker will
3867 set section addresses based on the memory regions, and will warn about
3868 regions that become too full. The linker will not shuffle sections
3869 around to fit into the available regions.
3871 A linker script may contain at most one use of the `MEMORY' command.
3872 However, you can define as many blocks of memory within it as you
3873 wish. The syntax is:
3876 NAME [(ATTR)] : ORIGIN = ORIGIN, LENGTH = LEN
3880 The NAME is a name used in the linker script to refer to the region.
3881 The region name has no meaning outside of the linker script. Region
3882 names are stored in a separate name space, and will not conflict with
3883 symbol names, file names, or section names. Each memory region must
3884 have a distinct name within the `MEMORY' command. However you can add
3885 later alias names to existing memory regions with the *Note
3886 REGION_ALIAS:: command.
3888 The ATTR string is an optional list of attributes that specify
3889 whether to use a particular memory region for an input section which is
3890 not explicitly mapped in the linker script. As described in *Note
3891 SECTIONS::, if you do not specify an output section for some input
3892 section, the linker will create an output section with the same name as
3893 the input section. If you define region attributes, the linker will use
3894 them to select the memory region for the output section that it creates.
3896 The ATTR string must consist only of the following characters:
3916 Invert the sense of any of the attributes that follow
3918 If a unmapped section matches any of the listed attributes other than
3919 `!', it will be placed in the memory region. The `!' attribute
3920 reverses this test, so that an unmapped section will be placed in the
3921 memory region only if it does not match any of the listed attributes.
3923 The ORIGIN is an numerical expression for the start address of the
3924 memory region. The expression must evaluate to a constant and it
3925 cannot involve any symbols. The keyword `ORIGIN' may be abbreviated to
3926 `org' or `o' (but not, for example, `ORG').
3928 The LEN is an expression for the size in bytes of the memory region.
3929 As with the ORIGIN expression, the expression must be numerical only
3930 and must evaluate to a constant. The keyword `LENGTH' may be
3931 abbreviated to `len' or `l'.
3933 In the following example, we specify that there are two memory
3934 regions available for allocation: one starting at `0' for 256 kilobytes,
3935 and the other starting at `0x40000000' for four megabytes. The linker
3936 will place into the `rom' memory region every section which is not
3937 explicitly mapped into a memory region, and is either read-only or
3938 executable. The linker will place other sections which are not
3939 explicitly mapped into a memory region into the `ram' memory region.
3943 rom (rx) : ORIGIN = 0, LENGTH = 256K
3944 ram (!rx) : org = 0x40000000, l = 4M
3947 Once you define a memory region, you can direct the linker to place
3948 specific output sections into that memory region by using the `>REGION'
3949 output section attribute. For example, if you have a memory region
3950 named `mem', you would use `>mem' in the output section definition.
3951 *Note Output Section Region::. If no address was specified for the
3952 output section, the linker will set the address to the next available
3953 address within the memory region. If the combined output sections
3954 directed to a memory region are too large for the region, the linker
3955 will issue an error message.
3957 It is possible to access the origin and length of a memory in an
3958 expression via the `ORIGIN(MEMORY)' and `LENGTH(MEMORY)' functions:
3960 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
3963 File: ld.info, Node: PHDRS, Next: VERSION, Prev: MEMORY, Up: Scripts
3968 The ELF object file format uses "program headers", also knows as
3969 "segments". The program headers describe how the program should be
3970 loaded into memory. You can print them out by using the `objdump'
3971 program with the `-p' option.
3973 When you run an ELF program on a native ELF system, the system loader
3974 reads the program headers in order to figure out how to load the
3975 program. This will only work if the program headers are set correctly.
3976 This manual does not describe the details of how the system loader
3977 interprets program headers; for more information, see the ELF ABI.
3979 The linker will create reasonable program headers by default.
3980 However, in some cases, you may need to specify the program headers more
3981 precisely. You may use the `PHDRS' command for this purpose. When the
3982 linker sees the `PHDRS' command in the linker script, it will not
3983 create any program headers other than the ones specified.
3985 The linker only pays attention to the `PHDRS' command when
3986 generating an ELF output file. In other cases, the linker will simply
3989 This is the syntax of the `PHDRS' command. The words `PHDRS',
3990 `FILEHDR', `AT', and `FLAGS' are keywords.
3994 NAME TYPE [ FILEHDR ] [ PHDRS ] [ AT ( ADDRESS ) ]
3995 [ FLAGS ( FLAGS ) ] ;
3998 The NAME is used only for reference in the `SECTIONS' command of the
3999 linker script. It is not put into the output file. Program header
4000 names are stored in a separate name space, and will not conflict with
4001 symbol names, file names, or section names. Each program header must
4002 have a distinct name. The headers are processed in order and it is
4003 usual for them to map to sections in ascending load address order.
4005 Certain program header types describe segments of memory which the
4006 system loader will load from the file. In the linker script, you
4007 specify the contents of these segments by placing allocatable output
4008 sections in the segments. You use the `:PHDR' output section attribute
4009 to place a section in a particular segment. *Note Output Section
4012 It is normal to put certain sections in more than one segment. This
4013 merely implies that one segment of memory contains another. You may
4014 repeat `:PHDR', using it once for each segment which should contain the
4017 If you place a section in one or more segments using `:PHDR', then
4018 the linker will place all subsequent allocatable sections which do not
4019 specify `:PHDR' in the same segments. This is for convenience, since
4020 generally a whole set of contiguous sections will be placed in a single
4021 segment. You can use `:NONE' to override the default segment and tell
4022 the linker to not put the section in any segment at all.
4024 You may use the `FILEHDR' and `PHDRS' keywords after the program
4025 header type to further describe the contents of the segment. The
4026 `FILEHDR' keyword means that the segment should include the ELF file
4027 header. The `PHDRS' keyword means that the segment should include the
4028 ELF program headers themselves. If applied to a loadable segment
4029 (`PT_LOAD'), all prior loadable segments must have one of these
4032 The TYPE may be one of the following. The numbers indicate the
4033 value of the keyword.
4036 Indicates an unused program header.
4039 Indicates that this program header describes a segment to be
4040 loaded from the file.
4043 Indicates a segment where dynamic linking information can be found.
4046 Indicates a segment where the name of the program interpreter may
4050 Indicates a segment holding note information.
4053 A reserved program header type, defined but not specified by the
4057 Indicates a segment where the program headers may be found.
4060 An expression giving the numeric type of the program header. This
4061 may be used for types not defined above.
4063 You can specify that a segment should be loaded at a particular
4064 address in memory by using an `AT' expression. This is identical to the
4065 `AT' command used as an output section attribute (*note Output Section
4066 LMA::). The `AT' command for a program header overrides the output
4069 The linker will normally set the segment flags based on the sections
4070 which comprise the segment. You may use the `FLAGS' keyword to
4071 explicitly specify the segment flags. The value of FLAGS must be an
4072 integer. It is used to set the `p_flags' field of the program header.
4074 Here is an example of `PHDRS'. This shows a typical set of program
4075 headers used on a native ELF system.
4079 headers PT_PHDR PHDRS ;
4081 text PT_LOAD FILEHDR PHDRS ;
4083 dynamic PT_DYNAMIC ;
4089 .interp : { *(.interp) } :text :interp
4090 .text : { *(.text) } :text
4091 .rodata : { *(.rodata) } /* defaults to :text */
4093 . = . + 0x1000; /* move to a new page in memory */
4094 .data : { *(.data) } :data
4095 .dynamic : { *(.dynamic) } :data :dynamic
4100 File: ld.info, Node: VERSION, Next: Expressions, Prev: PHDRS, Up: Scripts
4105 The linker supports symbol versions when using ELF. Symbol versions are
4106 only useful when using shared libraries. The dynamic linker can use
4107 symbol versions to select a specific version of a function when it runs
4108 a program that may have been linked against an earlier version of the
4111 You can include a version script directly in the main linker script,
4112 or you can supply the version script as an implicit linker script. You
4113 can also use the `--version-script' linker option.
4115 The syntax of the `VERSION' command is simply
4116 VERSION { version-script-commands }
4118 The format of the version script commands is identical to that used
4119 by Sun's linker in Solaris 2.5. The version script defines a tree of
4120 version nodes. You specify the node names and interdependencies in the
4121 version script. You can specify which symbols are bound to which
4122 version nodes, and you can reduce a specified set of symbols to local
4123 scope so that they are not globally visible outside of the shared
4126 The easiest way to demonstrate the version script language is with a
4150 This example version script defines three version nodes. The first
4151 version node defined is `VERS_1.1'; it has no other dependencies. The
4152 script binds the symbol `foo1' to `VERS_1.1'. It reduces a number of
4153 symbols to local scope so that they are not visible outside of the
4154 shared library; this is done using wildcard patterns, so that any
4155 symbol whose name begins with `old', `original', or `new' is matched.
4156 The wildcard patterns available are the same as those used in the shell
4157 when matching filenames (also known as "globbing"). However, if you
4158 specify the symbol name inside double quotes, then the name is treated
4159 as literal, rather than as a glob pattern.
4161 Next, the version script defines node `VERS_1.2'. This node depends
4162 upon `VERS_1.1'. The script binds the symbol `foo2' to the version
4165 Finally, the version script defines node `VERS_2.0'. This node
4166 depends upon `VERS_1.2'. The scripts binds the symbols `bar1' and
4167 `bar2' are bound to the version node `VERS_2.0'.
4169 When the linker finds a symbol defined in a library which is not
4170 specifically bound to a version node, it will effectively bind it to an
4171 unspecified base version of the library. You can bind all otherwise
4172 unspecified symbols to a given version node by using `global: *;'
4173 somewhere in the version script. Note that it's slightly crazy to use
4174 wildcards in a global spec except on the last version node. Global
4175 wildcards elsewhere run the risk of accidentally adding symbols to the
4176 set exported for an old version. That's wrong since older versions
4177 ought to have a fixed set of symbols.
4179 The names of the version nodes have no specific meaning other than
4180 what they might suggest to the person reading them. The `2.0' version
4181 could just as well have appeared in between `1.1' and `1.2'. However,
4182 this would be a confusing way to write a version script.
4184 Node name can be omitted, provided it is the only version node in
4185 the version script. Such version script doesn't assign any versions to
4186 symbols, only selects which symbols will be globally visible out and
4189 { global: foo; bar; local: *; };
4191 When you link an application against a shared library that has
4192 versioned symbols, the application itself knows which version of each
4193 symbol it requires, and it also knows which version nodes it needs from
4194 each shared library it is linked against. Thus at runtime, the dynamic
4195 loader can make a quick check to make sure that the libraries you have
4196 linked against do in fact supply all of the version nodes that the
4197 application will need to resolve all of the dynamic symbols. In this
4198 way it is possible for the dynamic linker to know with certainty that
4199 all external symbols that it needs will be resolvable without having to
4200 search for each symbol reference.
4202 The symbol versioning is in effect a much more sophisticated way of
4203 doing minor version checking that SunOS does. The fundamental problem
4204 that is being addressed here is that typically references to external
4205 functions are bound on an as-needed basis, and are not all bound when
4206 the application starts up. If a shared library is out of date, a
4207 required interface may be missing; when the application tries to use
4208 that interface, it may suddenly and unexpectedly fail. With symbol
4209 versioning, the user will get a warning when they start their program if
4210 the libraries being used with the application are too old.
4212 There are several GNU extensions to Sun's versioning approach. The
4213 first of these is the ability to bind a symbol to a version node in the
4214 source file where the symbol is defined instead of in the versioning
4215 script. This was done mainly to reduce the burden on the library
4216 maintainer. You can do this by putting something like:
4217 __asm__(".symver original_foo,foo@VERS_1.1");
4218 in the C source file. This renames the function `original_foo' to
4219 be an alias for `foo' bound to the version node `VERS_1.1'. The
4220 `local:' directive can be used to prevent the symbol `original_foo'
4221 from being exported. A `.symver' directive takes precedence over a
4224 The second GNU extension is to allow multiple versions of the same
4225 function to appear in a given shared library. In this way you can make
4226 an incompatible change to an interface without increasing the major
4227 version number of the shared library, while still allowing applications
4228 linked against the old interface to continue to function.
4230 To do this, you must use multiple `.symver' directives in the source
4231 file. Here is an example:
4233 __asm__(".symver original_foo,foo@");
4234 __asm__(".symver old_foo,foo@VERS_1.1");
4235 __asm__(".symver old_foo1,foo@VERS_1.2");
4236 __asm__(".symver new_foo,foo@@VERS_2.0");
4238 In this example, `foo@' represents the symbol `foo' bound to the
4239 unspecified base version of the symbol. The source file that contains
4240 this example would define 4 C functions: `original_foo', `old_foo',
4241 `old_foo1', and `new_foo'.
4243 When you have multiple definitions of a given symbol, there needs to
4244 be some way to specify a default version to which external references to
4245 this symbol will be bound. You can do this with the `foo@@VERS_2.0'
4246 type of `.symver' directive. You can only declare one version of a
4247 symbol as the default in this manner; otherwise you would effectively
4248 have multiple definitions of the same symbol.
4250 If you wish to bind a reference to a specific version of the symbol
4251 within the shared library, you can use the aliases of convenience
4252 (i.e., `old_foo'), or you can use the `.symver' directive to
4253 specifically bind to an external version of the function in question.
4255 You can also specify the language in the version script:
4257 VERSION extern "lang" { version-script-commands }
4259 The supported `lang's are `C', `C++', and `Java'. The linker will
4260 iterate over the list of symbols at the link time and demangle them
4261 according to `lang' before matching them to the patterns specified in
4262 `version-script-commands'. The default `lang' is `C'.
4264 Demangled names may contains spaces and other special characters. As
4265 described above, you can use a glob pattern to match demangled names,
4266 or you can use a double-quoted string to match the string exactly. In
4267 the latter case, be aware that minor differences (such as differing
4268 whitespace) between the version script and the demangler output will
4269 cause a mismatch. As the exact string generated by the demangler might
4270 change in the future, even if the mangled name does not, you should
4271 check that all of your version directives are behaving as you expect
4275 File: ld.info, Node: Expressions, Next: Implicit Linker Scripts, Prev: VERSION, Up: Scripts
4277 3.10 Expressions in Linker Scripts
4278 ==================================
4280 The syntax for expressions in the linker script language is identical to
4281 that of C expressions. All expressions are evaluated as integers. All
4282 expressions are evaluated in the same size, which is 32 bits if both the
4283 host and target are 32 bits, and is otherwise 64 bits.
4285 You can use and set symbol values in expressions.
4287 The linker defines several special purpose builtin functions for use
4292 * Constants:: Constants
4293 * Symbolic Constants:: Symbolic constants
4294 * Symbols:: Symbol Names
4295 * Orphan Sections:: Orphan Sections
4296 * Location Counter:: The Location Counter
4297 * Operators:: Operators
4298 * Evaluation:: Evaluation
4299 * Expression Section:: The Section of an Expression
4300 * Builtin Functions:: Builtin Functions
4303 File: ld.info, Node: Constants, Next: Symbolic Constants, Up: Expressions
4308 All constants are integers.
4310 As in C, the linker considers an integer beginning with `0' to be
4311 octal, and an integer beginning with `0x' or `0X' to be hexadecimal.
4312 Alternatively the linker accepts suffixes of `h' or `H' for
4313 hexadeciaml, `o' or `O' for octal, `b' or `B' for binary and `d' or `D'
4314 for decimal. Any integer value without a prefix or a suffix is
4315 considered to be decimal.
4317 In addition, you can use the suffixes `K' and `M' to scale a
4318 constant by `1024' or `1024*1024' respectively. For example, the
4319 following all refer to the same quantity:
4326 Note - the `K' and `M' suffixes cannot be used in conjunction with
4327 the base suffixes mentioned above.
4330 File: ld.info, Node: Symbolic Constants, Next: Symbols, Prev: Constants, Up: Expressions
4332 3.10.2 Symbolic Constants
4333 -------------------------
4335 It is possible to refer to target specific constants via the use of the
4336 `CONSTANT(NAME)' operator, where NAME is one of:
4339 The target's maximum page size.
4342 The target's default page size.
4346 .text ALIGN (CONSTANT (MAXPAGESIZE)) : { *(.text) }
4348 will create a text section aligned to the largest page boundary
4349 supported by the target.
4352 File: ld.info, Node: Symbols, Next: Orphan Sections, Prev: Symbolic Constants, Up: Expressions
4357 Unless quoted, symbol names start with a letter, underscore, or period
4358 and may include letters, digits, underscores, periods, and hyphens.
4359 Unquoted symbol names must not conflict with any keywords. You can
4360 specify a symbol which contains odd characters or has the same name as a
4361 keyword by surrounding the symbol name in double quotes:
4363 "with a space" = "also with a space" + 10;
4365 Since symbols can contain many non-alphabetic characters, it is
4366 safest to delimit symbols with spaces. For example, `A-B' is one
4367 symbol, whereas `A - B' is an expression involving subtraction.
4370 File: ld.info, Node: Orphan Sections, Next: Location Counter, Prev: Symbols, Up: Expressions
4372 3.10.4 Orphan Sections
4373 ----------------------
4375 Orphan sections are sections present in the input files which are not
4376 explicitly placed into the output file by the linker script. The
4377 linker will still copy these sections into the output file, but it has
4378 to guess as to where they should be placed. The linker uses a simple
4379 heuristic to do this. It attempts to place orphan sections after
4380 non-orphan sections of the same attribute, such as code vs data,
4381 loadable vs non-loadable, etc. If there is not enough room to do this
4382 then it places at the end of the file.
4384 For ELF targets, the attribute of the section includes section type
4385 as well as section flag.
4387 If an orphaned section's name is representable as a C identifier then
4388 the linker will automatically *note PROVIDE:: two symbols:
4389 __start_SECNAME and __end_SECNAME, where SECNAME is the name of the
4390 section. These indicate the start address and end address of the
4391 orphaned section respectively. Note: most section names are not
4392 representable as C identifiers because they contain a `.' character.
4395 File: ld.info, Node: Location Counter, Next: Operators, Prev: Orphan Sections, Up: Expressions
4397 3.10.5 The Location Counter
4398 ---------------------------
4400 The special linker variable "dot" `.' always contains the current
4401 output location counter. Since the `.' always refers to a location in
4402 an output section, it may only appear in an expression within a
4403 `SECTIONS' command. The `.' symbol may appear anywhere that an
4404 ordinary symbol is allowed in an expression.
4406 Assigning a value to `.' will cause the location counter to be
4407 moved. This may be used to create holes in the output section. The
4408 location counter may not be moved backwards inside an output section,
4409 and may not be moved backwards outside of an output section if so doing
4410 creates areas with overlapping LMAs.
4423 In the previous example, the `.text' section from `file1' is located
4424 at the beginning of the output section `output'. It is followed by a
4425 1000 byte gap. Then the `.text' section from `file2' appears, also
4426 with a 1000 byte gap following before the `.text' section from `file3'.
4427 The notation `= 0x12345678' specifies what data to write in the gaps
4428 (*note Output Section Fill::).
4430 Note: `.' actually refers to the byte offset from the start of the
4431 current containing object. Normally this is the `SECTIONS' statement,
4432 whose start address is 0, hence `.' can be used as an absolute address.
4433 If `.' is used inside a section description however, it refers to the
4434 byte offset from the start of that section, not an absolute address.
4435 Thus in a script like this:
4451 The `.text' section will be assigned a starting address of 0x100 and
4452 a size of exactly 0x200 bytes, even if there is not enough data in the
4453 `.text' input sections to fill this area. (If there is too much data,
4454 an error will be produced because this would be an attempt to move `.'
4455 backwards). The `.data' section will start at 0x500 and it will have
4456 an extra 0x600 bytes worth of space after the end of the values from
4457 the `.data' input sections and before the end of the `.data' output
4460 Setting symbols to the value of the location counter outside of an
4461 output section statement can result in unexpected values if the linker
4462 needs to place orphan sections. For example, given the following:
4475 If the linker needs to place some input section, e.g. `.rodata', not
4476 mentioned in the script, it might choose to place that section between
4477 `.text' and `.data'. You might think the linker should place `.rodata'
4478 on the blank line in the above script, but blank lines are of no
4479 particular significance to the linker. As well, the linker doesn't
4480 associate the above symbol names with their sections. Instead, it
4481 assumes that all assignments or other statements belong to the previous
4482 output section, except for the special case of an assignment to `.'.
4483 I.e., the linker will place the orphan `.rodata' section as if the
4484 script was written as follows:
4493 .rodata: { *(.rodata) }
4498 This may or may not be the script author's intention for the value of
4499 `start_of_data'. One way to influence the orphan section placement is
4500 to assign the location counter to itself, as the linker assumes that an
4501 assignment to `.' is setting the start address of a following output
4502 section and thus should be grouped with that section. So you could
4517 Now, the orphan `.rodata' section will be placed between
4518 `end_of_text' and `start_of_data'.
4521 File: ld.info, Node: Operators, Next: Evaluation, Prev: Location Counter, Up: Expressions
4526 The linker recognizes the standard C set of arithmetic operators, with
4527 the standard bindings and precedence levels:
4528 precedence associativity Operators Notes
4534 5 left == != > < <= >=
4540 11 right &= += -= *= /= (2)
4542 Notes: (1) Prefix operators (2) *Note Assignments::.
4545 File: ld.info, Node: Evaluation, Next: Expression Section, Prev: Operators, Up: Expressions
4550 The linker evaluates expressions lazily. It only computes the value of
4551 an expression when absolutely necessary.
4553 The linker needs some information, such as the value of the start
4554 address of the first section, and the origins and lengths of memory
4555 regions, in order to do any linking at all. These values are computed
4556 as soon as possible when the linker reads in the linker script.
4558 However, other values (such as symbol values) are not known or needed
4559 until after storage allocation. Such values are evaluated later, when
4560 other information (such as the sizes of output sections) is available
4561 for use in the symbol assignment expression.
4563 The sizes of sections cannot be known until after allocation, so
4564 assignments dependent upon these are not performed until after
4567 Some expressions, such as those depending upon the location counter
4568 `.', must be evaluated during section allocation.
4570 If the result of an expression is required, but the value is not
4571 available, then an error results. For example, a script like the
4575 .text 9+this_isnt_constant :
4578 will cause the error message `non constant expression for initial
4582 File: ld.info, Node: Expression Section, Next: Builtin Functions, Prev: Evaluation, Up: Expressions
4584 3.10.8 The Section of an Expression
4585 -----------------------------------
4587 Addresses and symbols may be section relative, or absolute. A section
4588 relative symbol is relocatable. If you request relocatable output
4589 using the `-r' option, a further link operation may change the value of
4590 a section relative symbol. On the other hand, an absolute symbol will
4591 retain the same value throughout any further link operations.
4593 Some terms in linker expressions are addresses. This is true of
4594 section relative symbols and for builtin functions that return an
4595 address, such as `ADDR', `LOADADDR', `ORIGIN' and `SEGMENT_START'.
4596 Other terms are simply numbers, or are builtin functions that return a
4597 non-address value, such as `LENGTH'. One complication is that unless
4598 you set `LD_FEATURE ("SANE_EXPR")' (*note Miscellaneous Commands::),
4599 numbers and absolute symbols are treated differently depending on their
4600 location, for compatibility with older versions of `ld'. Expressions
4601 appearing outside an output section definition treat all numbers as
4602 absolute addresses. Expressions appearing inside an output section
4603 definition treat absolute symbols as numbers. If `LD_FEATURE
4604 ("SANE_EXPR")' is given, then absolute symbols and numbers are simply
4605 treated as numbers everywhere.
4607 In the following simple example,
4612 __executable_start = 0x100;
4616 __data_start = 0x10;
4622 both `.' and `__executable_start' are set to the absolute address
4623 0x100 in the first two assignments, then both `.' and `__data_start'
4624 are set to 0x10 relative to the `.data' section in the second two
4627 For expressions involving numbers, relative addresses and absolute
4628 addresses, ld follows these rules to evaluate terms:
4630 * Unary operations on a relative address, and binary operations on
4631 two relative addresses in the same section or between one relative
4632 address and a number, apply the operator to the offset part of the
4635 * Unary operations on an absolute address, and binary operations on
4636 one or more absolute addresses or on two relative addresses not in
4637 the same section, first convert any non-absolute term to an
4638 absolute address before applying the operator.
4640 The result section of each sub-expression is as follows:
4642 * An operation involving only numbers results in a number.
4644 * The result of comparisons, `&&' and `||' is also a number.
4646 * The result of other binary arithmetic and logical operations on two
4647 relative addresses in the same section or two absolute addresess
4648 (after above conversions) is also a number.
4650 * The result of other operations on relative addresses or one
4651 relative address and a number, is a relative address in the same
4652 section as the relative operand(s).
4654 * The result of other operations on absolute addresses (after above
4655 conversions) is an absolute address.
4657 You can use the builtin function `ABSOLUTE' to force an expression
4658 to be absolute when it would otherwise be relative. For example, to
4659 create an absolute symbol set to the address of the end of the output
4663 .data : { *(.data) _edata = ABSOLUTE(.); }
4665 If `ABSOLUTE' were not used, `_edata' would be relative to the
4668 Using `LOADADDR' also forces an expression absolute, since this
4669 particular builtin function returns an absolute address.
4672 File: ld.info, Node: Builtin Functions, Prev: Expression Section, Up: Expressions
4674 3.10.9 Builtin Functions
4675 ------------------------
4677 The linker script language includes a number of builtin functions for
4678 use in linker script expressions.
4681 Return the absolute (non-relocatable, as opposed to non-negative)
4682 value of the expression EXP. Primarily useful to assign an
4683 absolute value to a symbol within a section definition, where
4684 symbol values are normally section relative. *Note Expression
4688 Return the address (VMA) of the named SECTION. Your script must
4689 previously have defined the location of that section. In the
4690 following example, `start_of_output_1', `symbol_1' and `symbol_2'
4691 are assigned equivalent values, except that `symbol_1' will be
4692 relative to the `.output1' section while the other two will be
4697 start_of_output_1 = ABSOLUTE(.);
4702 symbol_1 = ADDR(.output1);
4703 symbol_2 = start_of_output_1;
4709 Return the location counter (`.') or arbitrary expression aligned
4710 to the next ALIGN boundary. The single operand `ALIGN' doesn't
4711 change the value of the location counter--it just does arithmetic
4712 on it. The two operand `ALIGN' allows an arbitrary expression to
4713 be aligned upwards (`ALIGN(ALIGN)' is equivalent to `ALIGN(.,
4716 Here is an example which aligns the output `.data' section to the
4717 next `0x2000' byte boundary after the preceding section and sets a
4718 variable within the section to the next `0x8000' boundary after the
4721 .data ALIGN(0x2000): {
4723 variable = ALIGN(0x8000);
4726 The first use of `ALIGN' in this example specifies the
4727 location of a section because it is used as the optional ADDRESS
4728 attribute of a section definition (*note Output Section
4729 Address::). The second use of `ALIGN' is used to defines the
4732 The builtin function `NEXT' is closely related to `ALIGN'.
4735 Return the alignment in bytes of the named SECTION, if that
4736 section has been allocated. If the section has not been allocated
4737 when this is evaluated, the linker will report an error. In the
4738 following example, the alignment of the `.output' section is
4739 stored as the first value in that section.
4742 LONG (ALIGNOF (.output))
4748 This is a synonym for `ALIGN', for compatibility with older linker
4749 scripts. It is most often seen when setting the address of an
4752 `DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE)'
4753 This is equivalent to either
4754 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - 1)))
4756 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - COMMONPAGESIZE)))
4757 depending on whether the latter uses fewer COMMONPAGESIZE sized
4758 pages for the data segment (area between the result of this
4759 expression and `DATA_SEGMENT_END') than the former or not. If the
4760 latter form is used, it means COMMONPAGESIZE bytes of runtime
4761 memory will be saved at the expense of up to COMMONPAGESIZE wasted
4762 bytes in the on-disk file.
4764 This expression can only be used directly in `SECTIONS' commands,
4765 not in any output section descriptions and only once in the linker
4766 script. COMMONPAGESIZE should be less or equal to MAXPAGESIZE and
4767 should be the system page size the object wants to be optimized
4768 for (while still working on system page sizes up to MAXPAGESIZE).
4771 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
4773 `DATA_SEGMENT_END(EXP)'
4774 This defines the end of data segment for `DATA_SEGMENT_ALIGN'
4775 evaluation purposes.
4777 . = DATA_SEGMENT_END(.);
4779 `DATA_SEGMENT_RELRO_END(OFFSET, EXP)'
4780 This defines the end of the `PT_GNU_RELRO' segment when `-z relro'
4781 option is used. Second argument is returned. When `-z relro'
4782 option is not present, `DATA_SEGMENT_RELRO_END' does nothing,
4783 otherwise `DATA_SEGMENT_ALIGN' is padded so that EXP + OFFSET is
4784 aligned to the most commonly used page boundary for particular
4785 target. If present in the linker script, it must always come in
4786 between `DATA_SEGMENT_ALIGN' and `DATA_SEGMENT_END'.
4788 . = DATA_SEGMENT_RELRO_END(24, .);
4791 Return 1 if SYMBOL is in the linker global symbol table and is
4792 defined before the statement using DEFINED in the script, otherwise
4793 return 0. You can use this function to provide default values for
4794 symbols. For example, the following script fragment shows how to
4795 set a global symbol `begin' to the first location in the `.text'
4796 section--but if a symbol called `begin' already existed, its value
4801 begin = DEFINED(begin) ? begin : . ;
4808 Return the length of the memory region named MEMORY.
4811 Return the absolute LMA of the named SECTION. (*note Output
4815 Returns the maximum of EXP1 and EXP2.
4818 Returns the minimum of EXP1 and EXP2.
4821 Return the next unallocated address that is a multiple of EXP.
4822 This function is closely related to `ALIGN(EXP)'; unless you use
4823 the `MEMORY' command to define discontinuous memory for the output
4824 file, the two functions are equivalent.
4827 Return the origin of the memory region named MEMORY.
4829 `SEGMENT_START(SEGMENT, DEFAULT)'
4830 Return the base address of the named SEGMENT. If an explicit
4831 value has been given for this segment (with a command-line `-T'
4832 option) that value will be returned; otherwise the value will be
4833 DEFAULT. At present, the `-T' command-line option can only be
4834 used to set the base address for the "text", "data", and "bss"
4835 sections, but you can use `SEGMENT_START' with any segment name.
4838 Return the size in bytes of the named SECTION, if that section has
4839 been allocated. If the section has not been allocated when this is
4840 evaluated, the linker will report an error. In the following
4841 example, `symbol_1' and `symbol_2' are assigned identical values:
4848 symbol_1 = .end - .start ;
4849 symbol_2 = SIZEOF(.output);
4854 Return the size in bytes of the output file's headers. This is
4855 information which appears at the start of the output file. You
4856 can use this number when setting the start address of the first
4857 section, if you choose, to facilitate paging.
4859 When producing an ELF output file, if the linker script uses the
4860 `SIZEOF_HEADERS' builtin function, the linker must compute the
4861 number of program headers before it has determined all the section
4862 addresses and sizes. If the linker later discovers that it needs
4863 additional program headers, it will report an error `not enough
4864 room for program headers'. To avoid this error, you must avoid
4865 using the `SIZEOF_HEADERS' function, or you must rework your linker
4866 script to avoid forcing the linker to use additional program
4867 headers, or you must define the program headers yourself using the
4868 `PHDRS' command (*note PHDRS::).
4871 File: ld.info, Node: Implicit Linker Scripts, Prev: Expressions, Up: Scripts
4873 3.11 Implicit Linker Scripts
4874 ============================
4876 If you specify a linker input file which the linker can not recognize as
4877 an object file or an archive file, it will try to read the file as a
4878 linker script. If the file can not be parsed as a linker script, the
4879 linker will report an error.
4881 An implicit linker script will not replace the default linker script.
4883 Typically an implicit linker script would contain only symbol
4884 assignments, or the `INPUT', `GROUP', or `VERSION' commands.
4886 Any input files read because of an implicit linker script will be
4887 read at the position in the command line where the implicit linker
4888 script was read. This can affect archive searching.
4891 File: ld.info, Node: Machine Dependent, Next: BFD, Prev: Scripts, Up: Top
4893 4 Machine Dependent Features
4894 ****************************
4896 `ld' has additional features on some platforms; the following sections
4897 describe them. Machines where `ld' has no additional functionality are
4903 * H8/300:: `ld' and the H8/300
4905 * i960:: `ld' and the Intel 960 family
4907 * ARM:: `ld' and the ARM family
4909 * HPPA ELF32:: `ld' and HPPA 32-bit ELF
4911 * M68K:: `ld' and the Motorola 68K family
4913 * MMIX:: `ld' and MMIX
4915 * MSP430:: `ld' and MSP430
4917 * M68HC11/68HC12:: `ld' and the Motorola 68HC11 and 68HC12 families
4919 * PowerPC ELF32:: `ld' and PowerPC 32-bit ELF Support
4921 * PowerPC64 ELF64:: `ld' and PowerPC64 64-bit ELF Support
4923 * SPU ELF:: `ld' and SPU ELF Support
4925 * TI COFF:: `ld' and TI COFF
4927 * WIN32:: `ld' and WIN32 (cygwin/mingw)
4929 * Xtensa:: `ld' and Xtensa Processors
4932 File: ld.info, Node: H8/300, Next: i960, Up: Machine Dependent
4934 4.1 `ld' and the H8/300
4935 =======================
4937 For the H8/300, `ld' can perform these global optimizations when you
4938 specify the `--relax' command-line option.
4940 _relaxing address modes_
4941 `ld' finds all `jsr' and `jmp' instructions whose targets are
4942 within eight bits, and turns them into eight-bit program-counter
4943 relative `bsr' and `bra' instructions, respectively.
4945 _synthesizing instructions_
4946 `ld' finds all `mov.b' instructions which use the sixteen-bit
4947 absolute address form, but refer to the top page of memory, and
4948 changes them to use the eight-bit address form. (That is: the
4949 linker turns `mov.b `@'AA:16' into `mov.b `@'AA:8' whenever the
4950 address AA is in the top page of memory).
4952 _bit manipulation instructions_
4953 `ld' finds all bit manipulation instructions like `band, bclr,
4954 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst,
4955 bxor' which use 32 bit and 16 bit absolute address form, but refer
4956 to the top page of memory, and changes them to use the 8 bit
4957 address form. (That is: the linker turns `bset #xx:3,`@'AA:32'
4958 into `bset #xx:3,`@'AA:8' whenever the address AA is in the top
4961 _system control instructions_
4962 `ld' finds all `ldc.w, stc.w' instructions which use the 32 bit
4963 absolute address form, but refer to the top page of memory, and
4964 changes them to use 16 bit address form. (That is: the linker
4965 turns `ldc.w `@'AA:32,ccr' into `ldc.w `@'AA:16,ccr' whenever the
4966 address AA is in the top page of memory).
4969 File: ld.info, Node: i960, Next: ARM, Prev: H8/300, Up: Machine Dependent
4971 4.2 `ld' and the Intel 960 Family
4972 =================================
4974 You can use the `-AARCHITECTURE' command line option to specify one of
4975 the two-letter names identifying members of the 960 family; the option
4976 specifies the desired output target, and warns of any incompatible
4977 instructions in the input files. It also modifies the linker's search
4978 strategy for archive libraries, to support the use of libraries
4979 specific to each particular architecture, by including in the search
4980 loop names suffixed with the string identifying the architecture.
4982 For example, if your `ld' command line included `-ACA' as well as
4983 `-ltry', the linker would look (in its built-in search paths, and in
4984 any paths you specify with `-L') for a library with the names
4991 The first two possibilities would be considered in any event; the last
4992 two are due to the use of `-ACA'.
4994 You can meaningfully use `-A' more than once on a command line, since
4995 the 960 architecture family allows combination of target architectures;
4996 each use will add another pair of name variants to search for when `-l'
4997 specifies a library.
4999 `ld' supports the `--relax' option for the i960 family. If you
5000 specify `--relax', `ld' finds all `balx' and `calx' instructions whose
5001 targets are within 24 bits, and turns them into 24-bit program-counter
5002 relative `bal' and `cal' instructions, respectively. `ld' also turns
5003 `cal' instructions into `bal' instructions when it determines that the
5004 target subroutine is a leaf routine (that is, the target subroutine does
5005 not itself call any subroutines).
5007 The `--fix-cortex-a8' switch enables a link-time workaround for an
5008 erratum in certain Cortex-A8 processors. The workaround is enabled by
5009 default if you are targeting the ARM v7-A architecture profile. It can
5010 be enabled otherwise by specifying `--fix-cortex-a8', or disabled
5011 unconditionally by specifying `--no-fix-cortex-a8'.
5013 The erratum only affects Thumb-2 code. Please contact ARM for
5016 The `--no-merge-exidx-entries' switch disables the merging of
5017 adjacent exidx entries in debuginfo.
5020 File: ld.info, Node: M68HC11/68HC12, Next: PowerPC ELF32, Prev: MSP430, Up: Machine Dependent
5022 4.3 `ld' and the Motorola 68HC11 and 68HC12 families
5023 ====================================================
5025 4.3.1 Linker Relaxation
5026 -----------------------
5028 For the Motorola 68HC11, `ld' can perform these global optimizations
5029 when you specify the `--relax' command-line option.
5031 _relaxing address modes_
5032 `ld' finds all `jsr' and `jmp' instructions whose targets are
5033 within eight bits, and turns them into eight-bit program-counter
5034 relative `bsr' and `bra' instructions, respectively.
5036 `ld' also looks at all 16-bit extended addressing modes and
5037 transforms them in a direct addressing mode when the address is in
5038 page 0 (between 0 and 0x0ff).
5040 _relaxing gcc instruction group_
5041 When `gcc' is called with `-mrelax', it can emit group of
5042 instructions that the linker can optimize to use a 68HC11 direct
5043 addressing mode. These instructions consists of `bclr' or `bset'
5047 4.3.2 Trampoline Generation
5048 ---------------------------
5050 For 68HC11 and 68HC12, `ld' can generate trampoline code to call a far
5051 function using a normal `jsr' instruction. The linker will also change
5052 the relocation to some far function to use the trampoline address
5053 instead of the function address. This is typically the case when a
5054 pointer to a function is taken. The pointer will in fact point to the
5055 function trampoline.
5058 File: ld.info, Node: ARM, Next: HPPA ELF32, Prev: i960, Up: Machine Dependent
5060 4.4 `ld' and the ARM family
5061 ===========================
5063 For the ARM, `ld' will generate code stubs to allow functions calls
5064 between ARM and Thumb code. These stubs only work with code that has
5065 been compiled and assembled with the `-mthumb-interwork' command line
5066 option. If it is necessary to link with old ARM object files or
5067 libraries, which have not been compiled with the -mthumb-interwork
5068 option then the `--support-old-code' command line switch should be
5069 given to the linker. This will make it generate larger stub functions
5070 which will work with non-interworking aware ARM code. Note, however,
5071 the linker does not support generating stubs for function calls to
5072 non-interworking aware Thumb code.
5074 The `--thumb-entry' switch is a duplicate of the generic `--entry'
5075 switch, in that it sets the program's starting address. But it also
5076 sets the bottom bit of the address, so that it can be branched to using
5077 a BX instruction, and the program will start executing in Thumb mode
5080 The `--use-nul-prefixed-import-tables' switch is specifying, that
5081 the import tables idata4 and idata5 have to be generated with a zero
5082 elememt prefix for import libraries. This is the old style to generate
5083 import tables. By default this option is turned off.
5085 The `--be8' switch instructs `ld' to generate BE8 format
5086 executables. This option is only valid when linking big-endian objects.
5087 The resulting image will contain big-endian data and little-endian code.
5089 The `R_ARM_TARGET1' relocation is typically used for entries in the
5090 `.init_array' section. It is interpreted as either `R_ARM_REL32' or
5091 `R_ARM_ABS32', depending on the target. The `--target1-rel' and
5092 `--target1-abs' switches override the default.
5094 The `--target2=type' switch overrides the default definition of the
5095 `R_ARM_TARGET2' relocation. Valid values for `type', their meanings,
5096 and target defaults are as follows:
5098 `R_ARM_REL32' (arm*-*-elf, arm*-*-eabi)
5101 `R_ARM_ABS32' (arm*-*-symbianelf)
5104 `R_ARM_GOT_PREL' (arm*-*-linux, arm*-*-*bsd)
5106 The `R_ARM_V4BX' relocation (defined by the ARM AAELF specification)
5107 enables objects compiled for the ARMv4 architecture to be
5108 interworking-safe when linked with other objects compiled for ARMv4t,
5109 but also allows pure ARMv4 binaries to be built from the same ARMv4
5112 In the latter case, the switch `--fix-v4bx' must be passed to the
5113 linker, which causes v4t `BX rM' instructions to be rewritten as `MOV
5114 PC,rM', since v4 processors do not have a `BX' instruction.
5116 In the former case, the switch should not be used, and `R_ARM_V4BX'
5117 relocations are ignored.
5119 Replace `BX rM' instructions identified by `R_ARM_V4BX' relocations
5120 with a branch to the following veneer:
5126 This allows generation of libraries/applications that work on ARMv4
5127 cores and are still interworking safe. Note that the above veneer
5128 clobbers the condition flags, so may cause incorrect progrm behavior in
5131 The `--use-blx' switch enables the linker to use ARM/Thumb BLX
5132 instructions (available on ARMv5t and above) in various situations.
5133 Currently it is used to perform calls via the PLT from Thumb code using
5134 BLX rather than using BX and a mode-switching stub before each PLT
5135 entry. This should lead to such calls executing slightly faster.
5137 This option is enabled implicitly for SymbianOS, so there is no need
5138 to specify it if you are using that target.
5140 The `--vfp11-denorm-fix' switch enables a link-time workaround for a
5141 bug in certain VFP11 coprocessor hardware, which sometimes allows
5142 instructions with denorm operands (which must be handled by support
5143 code) to have those operands overwritten by subsequent instructions
5144 before the support code can read the intended values.
5146 The bug may be avoided in scalar mode if you allow at least one
5147 intervening instruction between a VFP11 instruction which uses a
5148 register and another instruction which writes to the same register, or
5149 at least two intervening instructions if vector mode is in use. The bug
5150 only affects full-compliance floating-point mode: you do not need this
5151 workaround if you are using "runfast" mode. Please contact ARM for
5154 If you know you are using buggy VFP11 hardware, you can enable this
5155 workaround by specifying the linker option `--vfp-denorm-fix=scalar' if
5156 you are using the VFP11 scalar mode only, or `--vfp-denorm-fix=vector'
5157 if you are using vector mode (the latter also works for scalar code).
5158 The default is `--vfp-denorm-fix=none'.
5160 If the workaround is enabled, instructions are scanned for
5161 potentially-troublesome sequences, and a veneer is created for each
5162 such sequence which may trigger the erratum. The veneer consists of the
5163 first instruction of the sequence and a branch back to the subsequent
5164 instruction. The original instruction is then replaced with a branch to
5165 the veneer. The extra cycles required to call and return from the veneer
5166 are sufficient to avoid the erratum in both the scalar and vector cases.
5168 The `--no-enum-size-warning' switch prevents the linker from warning
5169 when linking object files that specify incompatible EABI enumeration
5170 size attributes. For example, with this switch enabled, linking of an
5171 object file using 32-bit enumeration values with another using
5172 enumeration values fitted into the smallest possible space will not be
5175 The `--no-wchar-size-warning' switch prevents the linker from
5176 warning when linking object files that specify incompatible EABI
5177 `wchar_t' size attributes. For example, with this switch enabled,
5178 linking of an object file using 32-bit `wchar_t' values with another
5179 using 16-bit `wchar_t' values will not be diagnosed.
5181 The `--pic-veneer' switch makes the linker use PIC sequences for
5182 ARM/Thumb interworking veneers, even if the rest of the binary is not
5183 PIC. This avoids problems on uClinux targets where `--emit-relocs' is
5184 used to generate relocatable binaries.
5186 The linker will automatically generate and insert small sequences of
5187 code into a linked ARM ELF executable whenever an attempt is made to
5188 perform a function call to a symbol that is too far away. The
5189 placement of these sequences of instructions - called stubs - is
5190 controlled by the command line option `--stub-group-size=N'. The
5191 placement is important because a poor choice can create a need for
5192 duplicate stubs, increasing the code sizw. The linker will try to
5193 group stubs together in order to reduce interruptions to the flow of
5194 code, but it needs guidance as to how big these groups should be and
5195 where they should be placed.
5197 The value of `N', the parameter to the `--stub-group-size=' option
5198 controls where the stub groups are placed. If it is negative then all
5199 stubs are placed after the first branch that needs them. If it is
5200 positive then the stubs can be placed either before or after the
5201 branches that need them. If the value of `N' is 1 (either +1 or -1)
5202 then the linker will choose exactly where to place groups of stubs,
5203 using its built in heuristics. A value of `N' greater than 1 (or
5204 smaller than -1) tells the linker that a single group of stubs can
5205 service at most `N' bytes from the input sections.
5207 The default, if `--stub-group-size=' is not specified, is `N = +1'.
5209 Farcalls stubs insertion is fully supported for the ARM-EABI target
5210 only, because it relies on object files properties not present
5214 File: ld.info, Node: HPPA ELF32, Next: M68K, Prev: ARM, Up: Machine Dependent
5216 4.5 `ld' and HPPA 32-bit ELF Support
5217 ====================================
5219 When generating a shared library, `ld' will by default generate import
5220 stubs suitable for use with a single sub-space application. The
5221 `--multi-subspace' switch causes `ld' to generate export stubs, and
5222 different (larger) import stubs suitable for use with multiple
5225 Long branch stubs and import/export stubs are placed by `ld' in stub
5226 sections located between groups of input sections. `--stub-group-size'
5227 specifies the maximum size of a group of input sections handled by one
5228 stub section. Since branch offsets are signed, a stub section may
5229 serve two groups of input sections, one group before the stub section,
5230 and one group after it. However, when using conditional branches that
5231 require stubs, it may be better (for branch prediction) that stub
5232 sections only serve one group of input sections. A negative value for
5233 `N' chooses this scheme, ensuring that branches to stubs always use a
5234 negative offset. Two special values of `N' are recognized, `1' and
5235 `-1'. These both instruct `ld' to automatically size input section
5236 groups for the branch types detected, with the same behaviour regarding
5237 stub placement as other positive or negative values of `N' respectively.
5239 Note that `--stub-group-size' does not split input sections. A
5240 single input section larger than the group size specified will of course
5241 create a larger group (of one section). If input sections are too
5242 large, it may not be possible for a branch to reach its stub.
5245 File: ld.info, Node: M68K, Next: MMIX, Prev: HPPA ELF32, Up: Machine Dependent
5247 4.6 `ld' and the Motorola 68K family
5248 ====================================
5250 The `--got=TYPE' option lets you choose the GOT generation scheme. The
5251 choices are `single', `negative', `multigot' and `target'. When
5252 `target' is selected the linker chooses the default GOT generation
5253 scheme for the current target. `single' tells the linker to generate a
5254 single GOT with entries only at non-negative offsets. `negative'
5255 instructs the linker to generate a single GOT with entries at both
5256 negative and positive offsets. Not all environments support such GOTs.
5257 `multigot' allows the linker to generate several GOTs in the output
5258 file. All GOT references from a single input object file access the
5259 same GOT, but references from different input object files might access
5260 different GOTs. Not all environments support such GOTs.
5263 File: ld.info, Node: MMIX, Next: MSP430, Prev: M68K, Up: Machine Dependent
5268 For MMIX, there is a choice of generating `ELF' object files or `mmo'
5269 object files when linking. The simulator `mmix' understands the `mmo'
5270 format. The binutils `objcopy' utility can translate between the two
5273 There is one special section, the `.MMIX.reg_contents' section.
5274 Contents in this section is assumed to correspond to that of global
5275 registers, and symbols referring to it are translated to special
5276 symbols, equal to registers. In a final link, the start address of the
5277 `.MMIX.reg_contents' section corresponds to the first allocated global
5278 register multiplied by 8. Register `$255' is not included in this
5279 section; it is always set to the program entry, which is at the symbol
5280 `Main' for `mmo' files.
5282 Global symbols with the prefix `__.MMIX.start.', for example
5283 `__.MMIX.start..text' and `__.MMIX.start..data' are special. The
5284 default linker script uses these to set the default start address of a
5287 Initial and trailing multiples of zero-valued 32-bit words in a
5288 section, are left out from an mmo file.
5291 File: ld.info, Node: MSP430, Next: M68HC11/68HC12, Prev: MMIX, Up: Machine Dependent
5296 For the MSP430 it is possible to select the MPU architecture. The flag
5297 `-m [mpu type]' will select an appropriate linker script for selected
5298 MPU type. (To get a list of known MPUs just pass `-m help' option to
5301 The linker will recognize some extra sections which are MSP430
5305 Defines a portion of ROM where interrupt vectors located.
5308 Defines the bootloader portion of the ROM (if applicable). Any
5309 code in this section will be uploaded to the MPU.
5312 Defines an information memory section (if applicable). Any code in
5313 this section will be uploaded to the MPU.
5316 This is the same as the `.infomem' section except that any code in
5317 this section will not be uploaded to the MPU.
5320 Denotes a portion of RAM located above `.bss' section.
5322 The last two sections are used by gcc.
5325 File: ld.info, Node: PowerPC ELF32, Next: PowerPC64 ELF64, Prev: M68HC11/68HC12, Up: Machine Dependent
5327 4.9 `ld' and PowerPC 32-bit ELF Support
5328 =======================================
5330 Branches on PowerPC processors are limited to a signed 26-bit
5331 displacement, which may result in `ld' giving `relocation truncated to
5332 fit' errors with very large programs. `--relax' enables the generation
5333 of trampolines that can access the entire 32-bit address space. These
5334 trampolines are inserted at section boundaries, so may not themselves
5335 be reachable if an input section exceeds 33M in size. You may combine
5336 `-r' and `--relax' to add trampolines in a partial link. In that case
5337 both branches to undefined symbols and inter-section branches are also
5338 considered potentially out of range, and trampolines inserted.
5341 Current PowerPC GCC accepts a `-msecure-plt' option that generates
5342 code capable of using a newer PLT and GOT layout that has the
5343 security advantage of no executable section ever needing to be
5344 writable and no writable section ever being executable. PowerPC
5345 `ld' will generate this layout, including stubs to access the PLT,
5346 if all input files (including startup and static libraries) were
5347 compiled with `-msecure-plt'. `--bss-plt' forces the old BSS PLT
5348 (and GOT layout) which can give slightly better performance.
5351 `ld' will use the new PLT and GOT layout if it is linking new
5352 `-fpic' or `-fPIC' code, but does not do so automatically when
5353 linking non-PIC code. This option requests the new PLT and GOT
5354 layout. A warning will be given if some object file requires the
5358 The new secure PLT and GOT are placed differently relative to other
5359 sections compared to older BSS PLT and GOT placement. The
5360 location of `.plt' must change because the new secure PLT is an
5361 initialized section while the old PLT is uninitialized. The
5362 reason for the `.got' change is more subtle: The new placement
5363 allows `.got' to be read-only in applications linked with `-z
5364 relro -z now'. However, this placement means that `.sdata' cannot
5365 always be used in shared libraries, because the PowerPC ABI
5366 accesses `.sdata' in shared libraries from the GOT pointer.
5367 `--sdata-got' forces the old GOT placement. PowerPC GCC doesn't
5368 use `.sdata' in shared libraries, so this option is really only
5369 useful for other compilers that may do so.
5372 This option causes `ld' to label linker stubs with a local symbol
5373 that encodes the stub type and destination.
5376 PowerPC `ld' normally performs some optimization of code sequences
5377 used to access Thread-Local Storage. Use this option to disable
5381 File: ld.info, Node: PowerPC64 ELF64, Next: SPU ELF, Prev: PowerPC ELF32, Up: Machine Dependent
5383 4.10 `ld' and PowerPC64 64-bit ELF Support
5384 ==========================================
5387 Long branch stubs, PLT call stubs and TOC adjusting stubs are
5388 placed by `ld' in stub sections located between groups of input
5389 sections. `--stub-group-size' specifies the maximum size of a
5390 group of input sections handled by one stub section. Since branch
5391 offsets are signed, a stub section may serve two groups of input
5392 sections, one group before the stub section, and one group after
5393 it. However, when using conditional branches that require stubs,
5394 it may be better (for branch prediction) that stub sections only
5395 serve one group of input sections. A negative value for `N'
5396 chooses this scheme, ensuring that branches to stubs always use a
5397 negative offset. Two special values of `N' are recognized, `1'
5398 and `-1'. These both instruct `ld' to automatically size input
5399 section groups for the branch types detected, with the same
5400 behaviour regarding stub placement as other positive or negative
5401 values of `N' respectively.
5403 Note that `--stub-group-size' does not split input sections. A
5404 single input section larger than the group size specified will of
5405 course create a larger group (of one section). If input sections
5406 are too large, it may not be possible for a branch to reach its
5410 This option causes `ld' to label linker stubs with a local symbol
5411 that encodes the stub type and destination.
5413 `--dotsyms, --no-dotsyms'
5414 These two options control how `ld' interprets version patterns in
5415 a version script. Older PowerPC64 compilers emitted both a
5416 function descriptor symbol with the same name as the function, and
5417 a code entry symbol with the name prefixed by a dot (`.'). To
5418 properly version a function `foo', the version script thus needs
5419 to control both `foo' and `.foo'. The option `--dotsyms', on by
5420 default, automatically adds the required dot-prefixed patterns.
5421 Use `--no-dotsyms' to disable this feature.
5424 PowerPC64 `ld' normally performs some optimization of code
5425 sequences used to access Thread-Local Storage. Use this option to
5426 disable the optimization.
5429 PowerPC64 `ld' normally removes `.opd' section entries
5430 corresponding to deleted link-once functions, or functions removed
5431 by the action of `--gc-sections' or linker script `/DISCARD/'.
5432 Use this option to disable `.opd' optimization.
5434 `--non-overlapping-opd'
5435 Some PowerPC64 compilers have an option to generate compressed
5436 `.opd' entries spaced 16 bytes apart, overlapping the third word,
5437 the static chain pointer (unused in C) with the first word of the
5438 next entry. This option expands such entries to the full 24 bytes.
5441 PowerPC64 `ld' normally removes unused `.toc' section entries.
5442 Such entries are detected by examining relocations that reference
5443 the TOC in code sections. A reloc in a deleted code section marks
5444 a TOC word as unneeded, while a reloc in a kept code section marks
5445 a TOC word as needed. Since the TOC may reference itself, TOC
5446 relocs are also examined. TOC words marked as both needed and
5447 unneeded will of course be kept. TOC words without any referencing
5448 reloc are assumed to be part of a multi-word entry, and are kept or
5449 discarded as per the nearest marked preceding word. This works
5450 reliably for compiler generated code, but may be incorrect if
5451 assembly code is used to insert TOC entries. Use this option to
5452 disable the optimization.
5455 By default, PowerPC64 GCC generates code for a TOC model where TOC
5456 entries are accessed with a 16-bit offset from r2. This limits the
5457 total TOC size to 64K. PowerPC64 `ld' extends this limit by
5458 grouping code sections such that each group uses less than 64K for
5459 its TOC entries, then inserts r2 adjusting stubs between
5460 inter-group calls. `ld' does not split apart input sections, so
5461 cannot help if a single input file has a `.toc' section that
5462 exceeds 64K, most likely from linking multiple files with `ld -r'.
5463 Use this option to turn off this feature.
5466 File: ld.info, Node: SPU ELF, Next: TI COFF, Prev: PowerPC64 ELF64, Up: Machine Dependent
5468 4.11 `ld' and SPU ELF Support
5469 =============================
5472 This option marks an executable as a PIC plugin module.
5475 Normally, `ld' recognizes calls to functions within overlay
5476 regions, and redirects such calls to an overlay manager via a stub.
5477 `ld' also provides a built-in overlay manager. This option turns
5478 off all this special overlay handling.
5481 This option causes `ld' to label overlay stubs with a local symbol
5482 that encodes the stub type and destination.
5484 `--extra-overlay-stubs'
5485 This option causes `ld' to add overlay call stubs on all function
5486 calls out of overlay regions. Normally stubs are not added on
5487 calls to non-overlay regions.
5489 `--local-store=lo:hi'
5490 `ld' usually checks that a final executable for SPU fits in the
5491 address range 0 to 256k. This option may be used to change the
5492 range. Disable the check entirely with `--local-store=0:0'.
5495 SPU local store space is limited. Over-allocation of stack space
5496 unnecessarily limits space available for code and data, while
5497 under-allocation results in runtime failures. If given this
5498 option, `ld' will provide an estimate of maximum stack usage.
5499 `ld' does this by examining symbols in code sections to determine
5500 the extents of functions, and looking at function prologues for
5501 stack adjusting instructions. A call-graph is created by looking
5502 for relocations on branch instructions. The graph is then searched
5503 for the maximum stack usage path. Note that this analysis does not
5504 find calls made via function pointers, and does not handle
5505 recursion and other cycles in the call graph. Stack usage may be
5506 under-estimated if your code makes such calls. Also, stack usage
5507 for dynamic allocation, e.g. alloca, will not be detected. If a
5508 link map is requested, detailed information about each function's
5509 stack usage and calls will be given.
5512 This option, if given along with `--stack-analysis' will result in
5513 `ld' emitting stack sizing symbols for each function. These take
5514 the form `__stack_<function_name>' for global functions, and
5515 `__stack_<number>_<function_name>' for static functions.
5516 `<number>' is the section id in hex. The value of such symbols is
5517 the stack requirement for the corresponding function. The symbol
5518 size will be zero, type `STT_NOTYPE', binding `STB_LOCAL', and
5522 File: ld.info, Node: TI COFF, Next: WIN32, Prev: SPU ELF, Up: Machine Dependent
5524 4.12 `ld''s Support for Various TI COFF Versions
5525 ================================================
5527 The `--format' switch allows selection of one of the various TI COFF
5528 versions. The latest of this writing is 2; versions 0 and 1 are also
5529 supported. The TI COFF versions also vary in header byte-order format;
5530 `ld' will read any version or byte order, but the output header format
5531 depends on the default specified by the specific target.
5534 File: ld.info, Node: WIN32, Next: Xtensa, Prev: TI COFF, Up: Machine Dependent
5536 4.13 `ld' and WIN32 (cygwin/mingw)
5537 ==================================
5539 This section describes some of the win32 specific `ld' issues. See
5540 *Note Command Line Options: Options. for detailed description of the
5541 command line options mentioned here.
5544 The standard Windows linker creates and uses so-called import
5545 libraries, which contains information for linking to dll's. They
5546 are regular static archives and are handled as any other static
5547 archive. The cygwin and mingw ports of `ld' have specific support
5548 for creating such libraries provided with the `--out-implib'
5549 command line option.
5551 _exporting DLL symbols_
5552 The cygwin/mingw `ld' has several ways to export symbols for dll's.
5554 _using auto-export functionality_
5555 By default `ld' exports symbols with the auto-export
5556 functionality, which is controlled by the following command
5559 * -export-all-symbols [This is the default]
5565 * -exclude-modules-for-implib
5569 When auto-export is in operation, `ld' will export all the
5570 non-local (global and common) symbols it finds in a DLL, with
5571 the exception of a few symbols known to belong to the
5572 system's runtime and libraries. As it will often not be
5573 desirable to export all of a DLL's symbols, which may include
5574 private functions that are not part of any public interface,
5575 the command-line options listed above may be used to filter
5576 symbols out from the list for exporting. The `--output-def'
5577 option can be used in order to see the final list of exported
5578 symbols with all exclusions taken into effect.
5580 If `--export-all-symbols' is not given explicitly on the
5581 command line, then the default auto-export behavior will be
5582 _disabled_ if either of the following are true:
5584 * A DEF file is used.
5586 * Any symbol in any object file was marked with the
5587 __declspec(dllexport) attribute.
5590 Another way of exporting symbols is using a DEF file. A DEF
5591 file is an ASCII file containing definitions of symbols which
5592 should be exported when a dll is created. Usually it is
5593 named `<dll name>.def' and is added as any other object file
5594 to the linker's command line. The file's name must end in
5597 gcc -o <output> <objectfiles> <dll name>.def
5599 Using a DEF file turns off the normal auto-export behavior,
5600 unless the `--export-all-symbols' option is also used.
5602 Here is an example of a DEF file for a shared library called
5605 LIBRARY "xyz.dll" BASE=0x20000000
5611 another_foo = abc.dll.afoo
5616 This example defines a DLL with a non-default base address
5617 and seven symbols in the export table. The third exported
5618 symbol `_bar' is an alias for the second. The fourth symbol,
5619 `another_foo' is resolved by "forwarding" to another module
5620 and treating it as an alias for `afoo' exported from the DLL
5621 `abc.dll'. The final symbol `var1' is declared to be a data
5622 object. The `doo' symbol in export library is an alias of
5623 `foo', which gets the string name in export table `foo2'. The
5624 `eoo' symbol is an data export symbol, which gets in export
5625 table the name `var1'.
5627 The optional `LIBRARY <name>' command indicates the _internal_
5628 name of the output DLL. If `<name>' does not include a suffix,
5629 the default library suffix, `.DLL' is appended.
5631 When the .DEF file is used to build an application, rather
5632 than a library, the `NAME <name>' command should be used
5633 instead of `LIBRARY'. If `<name>' does not include a suffix,
5634 the default executable suffix, `.EXE' is appended.
5636 With either `LIBRARY <name>' or `NAME <name>' the optional
5637 specification `BASE = <number>' may be used to specify a
5638 non-default base address for the image.
5640 If neither `LIBRARY <name>' nor `NAME <name>' is specified,
5641 or they specify an empty string, the internal name is the
5642 same as the filename specified on the command line.
5644 The complete specification of an export symbol is:
5647 ( ( ( <name1> [ = <name2> ] )
5648 | ( <name1> = <module-name> . <external-name>))
5649 [ @ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
5651 Declares `<name1>' as an exported symbol from the DLL, or
5652 declares `<name1>' as an exported alias for `<name2>'; or
5653 declares `<name1>' as a "forward" alias for the symbol
5654 `<external-name>' in the DLL `<module-name>'. Optionally,
5655 the symbol may be exported by the specified ordinal
5656 `<integer>' alias. The optional `<name3>' is the to be used
5657 string in import/export table for the symbol.
5659 The optional keywords that follow the declaration indicate:
5661 `NONAME': Do not put the symbol name in the DLL's export
5662 table. It will still be exported by its ordinal alias
5663 (either the value specified by the .def specification or,
5664 otherwise, the value assigned by the linker). The symbol
5665 name, however, does remain visible in the import library (if
5666 any), unless `PRIVATE' is also specified.
5668 `DATA': The symbol is a variable or object, rather than a
5669 function. The import lib will export only an indirect
5670 reference to `foo' as the symbol `_imp__foo' (ie, `foo' must
5671 be resolved as `*_imp__foo').
5673 `CONSTANT': Like `DATA', but put the undecorated `foo' as
5674 well as `_imp__foo' into the import library. Both refer to the
5675 read-only import address table's pointer to the variable, not
5676 to the variable itself. This can be dangerous. If the user
5677 code fails to add the `dllimport' attribute and also fails to
5678 explicitly add the extra indirection that the use of the
5679 attribute enforces, the application will behave unexpectedly.
5681 `PRIVATE': Put the symbol in the DLL's export table, but do
5682 not put it into the static import library used to resolve
5683 imports at link time. The symbol can still be imported using
5684 the `LoadLibrary/GetProcAddress' API at runtime or by by
5685 using the GNU ld extension of linking directly to the DLL
5686 without an import library.
5688 See ld/deffilep.y in the binutils sources for the full
5689 specification of other DEF file statements
5691 While linking a shared dll, `ld' is able to create a DEF file
5692 with the `--output-def <file>' command line option.
5695 Another way of marking symbols for export is to modify the
5696 source code itself, so that when building the DLL each symbol
5697 to be exported is declared as:
5699 __declspec(dllexport) int a_variable
5700 __declspec(dllexport) void a_function(int with_args)
5702 All such symbols will be exported from the DLL. If, however,
5703 any of the object files in the DLL contain symbols decorated
5704 in this way, then the normal auto-export behavior is
5705 disabled, unless the `--export-all-symbols' option is also
5708 Note that object files that wish to access these symbols must
5709 _not_ decorate them with dllexport. Instead, they should use
5712 __declspec(dllimport) int a_variable
5713 __declspec(dllimport) void a_function(int with_args)
5715 This complicates the structure of library header files,
5716 because when included by the library itself the header must
5717 declare the variables and functions as dllexport, but when
5718 included by client code the header must declare them as
5719 dllimport. There are a number of idioms that are typically
5720 used to do this; often client code can omit the __declspec()
5721 declaration completely. See `--enable-auto-import' and
5722 `automatic data imports' for more information.
5724 _automatic data imports_
5725 The standard Windows dll format supports data imports from dlls
5726 only by adding special decorations (dllimport/dllexport), which
5727 let the compiler produce specific assembler instructions to deal
5728 with this issue. This increases the effort necessary to port
5729 existing Un*x code to these platforms, especially for large c++
5730 libraries and applications. The auto-import feature, which was
5731 initially provided by Paul Sokolovsky, allows one to omit the
5732 decorations to achieve a behavior that conforms to that on
5733 POSIX/Un*x platforms. This feature is enabled with the
5734 `--enable-auto-import' command-line option, although it is enabled
5735 by default on cygwin/mingw. The `--enable-auto-import' option
5736 itself now serves mainly to suppress any warnings that are
5737 ordinarily emitted when linked objects trigger the feature's use.
5739 auto-import of variables does not always work flawlessly without
5740 additional assistance. Sometimes, you will see this message
5742 "variable '<var>' can't be auto-imported. Please read the
5743 documentation for ld's `--enable-auto-import' for details."
5745 The `--enable-auto-import' documentation explains why this error
5746 occurs, and several methods that can be used to overcome this
5747 difficulty. One of these methods is the _runtime pseudo-relocs_
5748 feature, described below.
5750 For complex variables imported from DLLs (such as structs or
5751 classes), object files typically contain a base address for the
5752 variable and an offset (_addend_) within the variable-to specify a
5753 particular field or public member, for instance. Unfortunately,
5754 the runtime loader used in win32 environments is incapable of
5755 fixing these references at runtime without the additional
5756 information supplied by dllimport/dllexport decorations. The
5757 standard auto-import feature described above is unable to resolve
5760 The `--enable-runtime-pseudo-relocs' switch allows these
5761 references to be resolved without error, while leaving the task of
5762 adjusting the references themselves (with their non-zero addends)
5763 to specialized code provided by the runtime environment. Recent
5764 versions of the cygwin and mingw environments and compilers
5765 provide this runtime support; older versions do not. However, the
5766 support is only necessary on the developer's platform; the
5767 compiled result will run without error on an older system.
5769 `--enable-runtime-pseudo-relocs' is not the default; it must be
5770 explicitly enabled as needed.
5772 _direct linking to a dll_
5773 The cygwin/mingw ports of `ld' support the direct linking,
5774 including data symbols, to a dll without the usage of any import
5775 libraries. This is much faster and uses much less memory than
5776 does the traditional import library method, especially when
5777 linking large libraries or applications. When `ld' creates an
5778 import lib, each function or variable exported from the dll is
5779 stored in its own bfd, even though a single bfd could contain many
5780 exports. The overhead involved in storing, loading, and
5781 processing so many bfd's is quite large, and explains the
5782 tremendous time, memory, and storage needed to link against
5783 particularly large or complex libraries when using import libs.
5785 Linking directly to a dll uses no extra command-line switches
5786 other than `-L' and `-l', because `ld' already searches for a
5787 number of names to match each library. All that is needed from
5788 the developer's perspective is an understanding of this search, in
5789 order to force ld to select the dll instead of an import library.
5791 For instance, when ld is called with the argument `-lxxx' it will
5792 attempt to find, in the first directory of its search path,
5802 before moving on to the next directory in the search path.
5804 (*) Actually, this is not `cygxxx.dll' but in fact is
5805 `<prefix>xxx.dll', where `<prefix>' is set by the `ld' option
5806 `--dll-search-prefix=<prefix>'. In the case of cygwin, the
5807 standard gcc spec file includes `--dll-search-prefix=cyg', so in
5808 effect we actually search for `cygxxx.dll'.
5810 Other win32-based unix environments, such as mingw or pw32, may
5811 use other `<prefix>'es, although at present only cygwin makes use
5812 of this feature. It was originally intended to help avoid name
5813 conflicts among dll's built for the various win32/un*x
5814 environments, so that (for example) two versions of a zlib dll
5815 could coexist on the same machine.
5817 The generic cygwin/mingw path layout uses a `bin' directory for
5818 applications and dll's and a `lib' directory for the import
5819 libraries (using cygwin nomenclature):
5824 libxxx.dll.a (in case of dll's)
5825 libxxx.a (in case of static archive)
5827 Linking directly to a dll without using the import library can be
5830 1. Use the dll directly by adding the `bin' path to the link line
5831 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
5833 However, as the dll's often have version numbers appended to their
5834 names (`cygncurses-5.dll') this will often fail, unless one
5835 specifies `-L../bin -lncurses-5' to include the version. Import
5836 libs are generally not versioned, and do not have this difficulty.
5838 2. Create a symbolic link from the dll to a file in the `lib'
5839 directory according to the above mentioned search pattern. This
5840 should be used to avoid unwanted changes in the tools needed for
5843 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
5845 Then you can link without any make environment changes.
5847 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
5849 This technique also avoids the version number problems, because
5850 the following is perfectly legal
5855 libxxx.dll.a -> ../bin/cygxxx-5.dll
5857 Linking directly to a dll without using an import lib will work
5858 even when auto-import features are exercised, and even when
5859 `--enable-runtime-pseudo-relocs' is used.
5861 Given the improvements in speed and memory usage, one might
5862 justifiably wonder why import libraries are used at all. There
5865 1. Until recently, the link-directly-to-dll functionality did _not_
5866 work with auto-imported data.
5868 2. Sometimes it is necessary to include pure static objects within
5869 the import library (which otherwise contains only bfd's for
5870 indirection symbols that point to the exports of a dll). Again,
5871 the import lib for the cygwin kernel makes use of this ability,
5872 and it is not possible to do this without an import lib.
5874 3. Symbol aliases can only be resolved using an import lib. This
5875 is critical when linking against OS-supplied dll's (eg, the win32
5876 API) in which symbols are usually exported as undecorated aliases
5877 of their stdcall-decorated assembly names.
5879 So, import libs are not going away. But the ability to replace
5880 true import libs with a simple symbolic link to (or a copy of) a
5881 dll, in many cases, is a useful addition to the suite of tools
5882 binutils makes available to the win32 developer. Given the
5883 massive improvements in memory requirements during linking, storage
5884 requirements, and linking speed, we expect that many developers
5885 will soon begin to use this feature whenever possible.
5889 _adding additional names_
5890 Sometimes, it is useful to export symbols with additional
5891 names. A symbol `foo' will be exported as `foo', but it can
5892 also be exported as `_foo' by using special directives in the
5893 DEF file when creating the dll. This will affect also the
5894 optional created import library. Consider the following DEF
5897 LIBRARY "xyz.dll" BASE=0x61000000
5903 The line `_foo = foo' maps the symbol `foo' to `_foo'.
5905 Another method for creating a symbol alias is to create it in
5906 the source code using the "weak" attribute:
5908 void foo () { /* Do something. */; }
5909 void _foo () __attribute__ ((weak, alias ("foo")));
5911 See the gcc manual for more information about attributes and
5915 Sometimes it is useful to rename exports. For instance, the
5916 cygwin kernel does this regularly. A symbol `_foo' can be
5917 exported as `foo' but not as `_foo' by using special
5918 directives in the DEF file. (This will also affect the import
5919 library, if it is created). In the following example:
5921 LIBRARY "xyz.dll" BASE=0x61000000
5926 The line `_foo = foo' maps the exported symbol `foo' to
5929 Note: using a DEF file disables the default auto-export behavior,
5930 unless the `--export-all-symbols' command line option is used.
5931 If, however, you are trying to rename symbols, then you should list
5932 _all_ desired exports in the DEF file, including the symbols that
5933 are not being renamed, and do _not_ use the `--export-all-symbols'
5934 option. If you list only the renamed symbols in the DEF file, and
5935 use `--export-all-symbols' to handle the other symbols, then the
5936 both the new names _and_ the original names for the renamed
5937 symbols will be exported. In effect, you'd be aliasing those
5938 symbols, not renaming them, which is probably not what you wanted.
5941 The Windows object format, PE, specifies a form of weak symbols
5942 called weak externals. When a weak symbol is linked and the
5943 symbol is not defined, the weak symbol becomes an alias for some
5944 other symbol. There are three variants of weak externals:
5945 * Definition is searched for in objects and libraries,
5946 historically called lazy externals.
5948 * Definition is searched for only in other objects, not in
5949 libraries. This form is not presently implemented.
5951 * No search; the symbol is an alias. This form is not presently
5953 As a GNU extension, weak symbols that do not specify an alternate
5954 symbol are supported. If the symbol is undefined when linking,
5955 the symbol uses a default value.
5957 _aligned common symbols_
5958 As a GNU extension to the PE file format, it is possible to
5959 specify the desired alignment for a common symbol. This
5960 information is conveyed from the assembler or compiler to the
5961 linker by means of GNU-specific commands carried in the object
5962 file's `.drectve' section, which are recognized by `ld' and
5963 respected when laying out the common symbols. Native tools will
5964 be able to process object files employing this GNU extension, but
5965 will fail to respect the alignment instructions, and may issue
5966 noisy warnings about unknown linker directives.
5969 File: ld.info, Node: Xtensa, Prev: WIN32, Up: Machine Dependent
5971 4.14 `ld' and Xtensa Processors
5972 ===============================
5974 The default `ld' behavior for Xtensa processors is to interpret
5975 `SECTIONS' commands so that lists of explicitly named sections in a
5976 specification with a wildcard file will be interleaved when necessary to
5977 keep literal pools within the range of PC-relative load offsets. For
5978 example, with the command:
5987 `ld' may interleave some of the `.literal' and `.text' sections from
5988 different object files to ensure that the literal pools are within the
5989 range of PC-relative load offsets. A valid interleaving might place
5990 the `.literal' sections from an initial group of files followed by the
5991 `.text' sections of that group of files. Then, the `.literal' sections
5992 from the rest of the files and the `.text' sections from the rest of
5993 the files would follow.
5995 Relaxation is enabled by default for the Xtensa version of `ld' and
5996 provides two important link-time optimizations. The first optimization
5997 is to combine identical literal values to reduce code size. A redundant
5998 literal will be removed and all the `L32R' instructions that use it
5999 will be changed to reference an identical literal, as long as the
6000 location of the replacement literal is within the offset range of all
6001 the `L32R' instructions. The second optimization is to remove
6002 unnecessary overhead from assembler-generated "longcall" sequences of
6003 `L32R'/`CALLXN' when the target functions are within range of direct
6004 `CALLN' instructions.
6006 For each of these cases where an indirect call sequence can be
6007 optimized to a direct call, the linker will change the `CALLXN'
6008 instruction to a `CALLN' instruction, remove the `L32R' instruction,
6009 and remove the literal referenced by the `L32R' instruction if it is
6010 not used for anything else. Removing the `L32R' instruction always
6011 reduces code size but can potentially hurt performance by changing the
6012 alignment of subsequent branch targets. By default, the linker will
6013 always preserve alignments, either by switching some instructions
6014 between 24-bit encodings and the equivalent density instructions or by
6015 inserting a no-op in place of the `L32R' instruction that was removed.
6016 If code size is more important than performance, the `--size-opt'
6017 option can be used to prevent the linker from widening density
6018 instructions or inserting no-ops, except in a few cases where no-ops
6019 are required for correctness.
6021 The following Xtensa-specific command-line options can be used to
6025 When optimizing indirect calls to direct calls, optimize for code
6026 size more than performance. With this option, the linker will not
6027 insert no-ops or widen density instructions to preserve branch
6028 target alignment. There may still be some cases where no-ops are
6029 required to preserve the correctness of the code.
6032 File: ld.info, Node: BFD, Next: Reporting Bugs, Prev: Machine Dependent, Up: Top
6037 The linker accesses object and archive files using the BFD libraries.
6038 These libraries allow the linker to use the same routines to operate on
6039 object files whatever the object file format. A different object file
6040 format can be supported simply by creating a new BFD back end and adding
6041 it to the library. To conserve runtime memory, however, the linker and
6042 associated tools are usually configured to support only a subset of the
6043 object file formats available. You can use `objdump -i' (*note
6044 objdump: (binutils.info)objdump.) to list all the formats available for
6047 As with most implementations, BFD is a compromise between several
6048 conflicting requirements. The major factor influencing BFD design was
6049 efficiency: any time used converting between formats is time which
6050 would not have been spent had BFD not been involved. This is partly
6051 offset by abstraction payback; since BFD simplifies applications and
6052 back ends, more time and care may be spent optimizing algorithms for a
6055 One minor artifact of the BFD solution which you should bear in mind
6056 is the potential for information loss. There are two places where
6057 useful information can be lost using the BFD mechanism: during
6058 conversion and during output. *Note BFD information loss::.
6062 * BFD outline:: How it works: an outline of BFD
6065 File: ld.info, Node: BFD outline, Up: BFD
6067 5.1 How It Works: An Outline of BFD
6068 ===================================
6070 When an object file is opened, BFD subroutines automatically determine
6071 the format of the input object file. They then build a descriptor in
6072 memory with pointers to routines that will be used to access elements of
6073 the object file's data structures.
6075 As different information from the object files is required, BFD
6076 reads from different sections of the file and processes them. For
6077 example, a very common operation for the linker is processing symbol
6078 tables. Each BFD back end provides a routine for converting between
6079 the object file's representation of symbols and an internal canonical
6080 format. When the linker asks for the symbol table of an object file, it
6081 calls through a memory pointer to the routine from the relevant BFD
6082 back end which reads and converts the table into a canonical form. The
6083 linker then operates upon the canonical form. When the link is finished
6084 and the linker writes the output file's symbol table, another BFD back
6085 end routine is called to take the newly created symbol table and
6086 convert it into the chosen output format.
6090 * BFD information loss:: Information Loss
6091 * Canonical format:: The BFD canonical object-file format
6094 File: ld.info, Node: BFD information loss, Next: Canonical format, Up: BFD outline
6096 5.1.1 Information Loss
6097 ----------------------
6099 _Information can be lost during output._ The output formats supported
6100 by BFD do not provide identical facilities, and information which can
6101 be described in one form has nowhere to go in another format. One
6102 example of this is alignment information in `b.out'. There is nowhere
6103 in an `a.out' format file to store alignment information on the
6104 contained data, so when a file is linked from `b.out' and an `a.out'
6105 image is produced, alignment information will not propagate to the
6106 output file. (The linker will still use the alignment information
6107 internally, so the link is performed correctly).
6109 Another example is COFF section names. COFF files may contain an
6110 unlimited number of sections, each one with a textual section name. If
6111 the target of the link is a format which does not have many sections
6112 (e.g., `a.out') or has sections without names (e.g., the Oasys format),
6113 the link cannot be done simply. You can circumvent this problem by
6114 describing the desired input-to-output section mapping with the linker
6117 _Information can be lost during canonicalization._ The BFD internal
6118 canonical form of the external formats is not exhaustive; there are
6119 structures in input formats for which there is no direct representation
6120 internally. This means that the BFD back ends cannot maintain all
6121 possible data richness through the transformation between external to
6122 internal and back to external formats.
6124 This limitation is only a problem when an application reads one
6125 format and writes another. Each BFD back end is responsible for
6126 maintaining as much data as possible, and the internal BFD canonical
6127 form has structures which are opaque to the BFD core, and exported only
6128 to the back ends. When a file is read in one format, the canonical form
6129 is generated for BFD and the application. At the same time, the back
6130 end saves away any information which may otherwise be lost. If the data
6131 is then written back in the same format, the back end routine will be
6132 able to use the canonical form provided by the BFD core as well as the
6133 information it prepared earlier. Since there is a great deal of
6134 commonality between back ends, there is no information lost when
6135 linking or copying big endian COFF to little endian COFF, or `a.out' to
6136 `b.out'. When a mixture of formats is linked, the information is only
6137 lost from the files whose format differs from the destination.
6140 File: ld.info, Node: Canonical format, Prev: BFD information loss, Up: BFD outline
6142 5.1.2 The BFD canonical object-file format
6143 ------------------------------------------
6145 The greatest potential for loss of information occurs when there is the
6146 least overlap between the information provided by the source format,
6147 that stored by the canonical format, and that needed by the destination
6148 format. A brief description of the canonical form may help you
6149 understand which kinds of data you can count on preserving across
6153 Information stored on a per-file basis includes target machine
6154 architecture, particular implementation format type, a demand
6155 pageable bit, and a write protected bit. Information like Unix
6156 magic numbers is not stored here--only the magic numbers' meaning,
6157 so a `ZMAGIC' file would have both the demand pageable bit and the
6158 write protected text bit set. The byte order of the target is
6159 stored on a per-file basis, so that big- and little-endian object
6160 files may be used with one another.
6163 Each section in the input file contains the name of the section,
6164 the section's original address in the object file, size and
6165 alignment information, various flags, and pointers into other BFD
6169 Each symbol contains a pointer to the information for the object
6170 file which originally defined it, its name, its value, and various
6171 flag bits. When a BFD back end reads in a symbol table, it
6172 relocates all symbols to make them relative to the base of the
6173 section where they were defined. Doing this ensures that each
6174 symbol points to its containing section. Each symbol also has a
6175 varying amount of hidden private data for the BFD back end. Since
6176 the symbol points to the original file, the private data format
6177 for that symbol is accessible. `ld' can operate on a collection
6178 of symbols of wildly different formats without problems.
6180 Normal global and simple local symbols are maintained on output,
6181 so an output file (no matter its format) will retain symbols
6182 pointing to functions and to global, static, and common variables.
6183 Some symbol information is not worth retaining; in `a.out', type
6184 information is stored in the symbol table as long symbol names.
6185 This information would be useless to most COFF debuggers; the
6186 linker has command line switches to allow users to throw it away.
6188 There is one word of type information within the symbol, so if the
6189 format supports symbol type information within symbols (for
6190 example, COFF, IEEE, Oasys) and the type is simple enough to fit
6191 within one word (nearly everything but aggregates), the
6192 information will be preserved.
6195 Each canonical BFD relocation record contains a pointer to the
6196 symbol to relocate to, the offset of the data to relocate, the
6197 section the data is in, and a pointer to a relocation type
6198 descriptor. Relocation is performed by passing messages through
6199 the relocation type descriptor and the symbol pointer. Therefore,
6200 relocations can be performed on output data using a relocation
6201 method that is only available in one of the input formats. For
6202 instance, Oasys provides a byte relocation format. A relocation
6203 record requesting this relocation type would point indirectly to a
6204 routine to perform this, so the relocation may be performed on a
6205 byte being written to a 68k COFF file, even though 68k COFF has no
6206 such relocation type.
6209 Object formats can contain, for debugging purposes, some form of
6210 mapping between symbols, source line numbers, and addresses in the
6211 output file. These addresses have to be relocated along with the
6212 symbol information. Each symbol with an associated list of line
6213 number records points to the first record of the list. The head
6214 of a line number list consists of a pointer to the symbol, which
6215 allows finding out the address of the function whose line number
6216 is being described. The rest of the list is made up of pairs:
6217 offsets into the section and line numbers. Any format which can
6218 simply derive this information can pass it successfully between
6219 formats (COFF, IEEE and Oasys).
6222 File: ld.info, Node: Reporting Bugs, Next: MRI, Prev: BFD, Up: Top
6227 Your bug reports play an essential role in making `ld' reliable.
6229 Reporting a bug may help you by bringing a solution to your problem,
6230 or it may not. But in any case the principal function of a bug report
6231 is to help the entire community by making the next version of `ld' work
6232 better. Bug reports are your contribution to the maintenance of `ld'.
6234 In order for a bug report to serve its purpose, you must include the
6235 information that enables us to fix the bug.
6239 * Bug Criteria:: Have you found a bug?
6240 * Bug Reporting:: How to report bugs
6243 File: ld.info, Node: Bug Criteria, Next: Bug Reporting, Up: Reporting Bugs
6245 6.1 Have You Found a Bug?
6246 =========================
6248 If you are not sure whether you have found a bug, here are some
6251 * If the linker gets a fatal signal, for any input whatever, that is
6252 a `ld' bug. Reliable linkers never crash.
6254 * If `ld' produces an error message for valid input, that is a bug.
6256 * If `ld' does not produce an error message for invalid input, that
6257 may be a bug. In the general case, the linker can not verify that
6258 object files are correct.
6260 * If you are an experienced user of linkers, your suggestions for
6261 improvement of `ld' are welcome in any case.
6264 File: ld.info, Node: Bug Reporting, Prev: Bug Criteria, Up: Reporting Bugs
6266 6.2 How to Report Bugs
6267 ======================
6269 A number of companies and individuals offer support for GNU products.
6270 If you obtained `ld' from a support organization, we recommend you
6271 contact that organization first.
6273 You can find contact information for many support companies and
6274 individuals in the file `etc/SERVICE' in the GNU Emacs distribution.
6276 Otherwise, send bug reports for `ld' to
6277 `http://www.sourceware.org/bugzilla/'.
6279 The fundamental principle of reporting bugs usefully is this:
6280 *report all the facts*. If you are not sure whether to state a fact or
6281 leave it out, state it!
6283 Often people omit facts because they think they know what causes the
6284 problem and assume that some details do not matter. Thus, you might
6285 assume that the name of a symbol you use in an example does not matter.
6286 Well, probably it does not, but one cannot be sure. Perhaps the bug
6287 is a stray memory reference which happens to fetch from the location
6288 where that name is stored in memory; perhaps, if the name were
6289 different, the contents of that location would fool the linker into
6290 doing the right thing despite the bug. Play it safe and give a
6291 specific, complete example. That is the easiest thing for you to do,
6292 and the most helpful.
6294 Keep in mind that the purpose of a bug report is to enable us to fix
6295 the bug if it is new to us. Therefore, always write your bug reports
6296 on the assumption that the bug has not been reported previously.
6298 Sometimes people give a few sketchy facts and ask, "Does this ring a
6299 bell?" This cannot help us fix a bug, so it is basically useless. We
6300 respond by asking for enough details to enable us to investigate. You
6301 might as well expedite matters by sending them to begin with.
6303 To enable us to fix the bug, you should include all these things:
6305 * The version of `ld'. `ld' announces it if you start it with the
6306 `--version' argument.
6308 Without this, we will not know whether there is any point in
6309 looking for the bug in the current version of `ld'.
6311 * Any patches you may have applied to the `ld' source, including any
6312 patches made to the `BFD' library.
6314 * The type of machine you are using, and the operating system name
6317 * What compiler (and its version) was used to compile `ld'--e.g.
6320 * The command arguments you gave the linker to link your example and
6321 observe the bug. To guarantee you will not omit something
6322 important, list them all. A copy of the Makefile (or the output
6323 from make) is sufficient.
6325 If we were to try to guess the arguments, we would probably guess
6326 wrong and then we might not encounter the bug.
6328 * A complete input file, or set of input files, that will reproduce
6329 the bug. It is generally most helpful to send the actual object
6330 files provided that they are reasonably small. Say no more than
6331 10K. For bigger files you can either make them available by FTP
6332 or HTTP or else state that you are willing to send the object
6333 file(s) to whomever requests them. (Note - your email will be
6334 going to a mailing list, so we do not want to clog it up with
6335 large attachments). But small attachments are best.
6337 If the source files were assembled using `gas' or compiled using
6338 `gcc', then it may be OK to send the source files rather than the
6339 object files. In this case, be sure to say exactly what version of
6340 `gas' or `gcc' was used to produce the object files. Also say how
6341 `gas' or `gcc' were configured.
6343 * A description of what behavior you observe that you believe is
6344 incorrect. For example, "It gets a fatal signal."
6346 Of course, if the bug is that `ld' gets a fatal signal, then we
6347 will certainly notice it. But if the bug is incorrect output, we
6348 might not notice unless it is glaringly wrong. You might as well
6349 not give us a chance to make a mistake.
6351 Even if the problem you experience is a fatal signal, you should
6352 still say so explicitly. Suppose something strange is going on,
6353 such as, your copy of `ld' is out of sync, or you have encountered
6354 a bug in the C library on your system. (This has happened!) Your
6355 copy might crash and ours would not. If you told us to expect a
6356 crash, then when ours fails to crash, we would know that the bug
6357 was not happening for us. If you had not told us to expect a
6358 crash, then we would not be able to draw any conclusion from our
6361 * If you wish to suggest changes to the `ld' source, send us context
6362 diffs, as generated by `diff' with the `-u', `-c', or `-p' option.
6363 Always send diffs from the old file to the new file. If you even
6364 discuss something in the `ld' source, refer to it by context, not
6367 The line numbers in our development sources will not match those
6368 in your sources. Your line numbers would convey no useful
6371 Here are some things that are not necessary:
6373 * A description of the envelope of the bug.
6375 Often people who encounter a bug spend a lot of time investigating
6376 which changes to the input file will make the bug go away and which
6377 changes will not affect it.
6379 This is often time consuming and not very useful, because the way
6380 we will find the bug is by running a single example under the
6381 debugger with breakpoints, not by pure deduction from a series of
6382 examples. We recommend that you save your time for something else.
6384 Of course, if you can find a simpler example to report _instead_
6385 of the original one, that is a convenience for us. Errors in the
6386 output will be easier to spot, running under the debugger will take
6387 less time, and so on.
6389 However, simplification is not vital; if you do not want to do
6390 this, report the bug anyway and send us the entire test case you
6393 * A patch for the bug.
6395 A patch for the bug does help us if it is a good one. But do not
6396 omit the necessary information, such as the test case, on the
6397 assumption that a patch is all we need. We might see problems
6398 with your patch and decide to fix the problem another way, or we
6399 might not understand it at all.
6401 Sometimes with a program as complicated as `ld' it is very hard to
6402 construct an example that will make the program follow a certain
6403 path through the code. If you do not send us the example, we will
6404 not be able to construct one, so we will not be able to verify
6405 that the bug is fixed.
6407 And if we cannot understand what bug you are trying to fix, or why
6408 your patch should be an improvement, we will not install it. A
6409 test case will help us to understand.
6411 * A guess about what the bug is or what it depends on.
6413 Such guesses are usually wrong. Even we cannot guess right about
6414 such things without first using the debugger to find the facts.
6417 File: ld.info, Node: MRI, Next: GNU Free Documentation License, Prev: Reporting Bugs, Up: Top
6419 Appendix A MRI Compatible Script Files
6420 **************************************
6422 To aid users making the transition to GNU `ld' from the MRI linker,
6423 `ld' can use MRI compatible linker scripts as an alternative to the
6424 more general-purpose linker scripting language described in *Note
6425 Scripts::. MRI compatible linker scripts have a much simpler command
6426 set than the scripting language otherwise used with `ld'. GNU `ld'
6427 supports the most commonly used MRI linker commands; these commands are
6430 In general, MRI scripts aren't of much use with the `a.out' object
6431 file format, since it only has three sections and MRI scripts lack some
6432 features to make use of them.
6434 You can specify a file containing an MRI-compatible script using the
6435 `-c' command-line option.
6437 Each command in an MRI-compatible script occupies its own line; each
6438 command line starts with the keyword that identifies the command (though
6439 blank lines are also allowed for punctuation). If a line of an
6440 MRI-compatible script begins with an unrecognized keyword, `ld' issues
6441 a warning message, but continues processing the script.
6443 Lines beginning with `*' are comments.
6445 You can write these commands using all upper-case letters, or all
6446 lower case; for example, `chip' is the same as `CHIP'. The following
6447 list shows only the upper-case form of each command.
6450 `ABSOLUTE SECNAME, SECNAME, ... SECNAME'
6451 Normally, `ld' includes in the output file all sections from all
6452 the input files. However, in an MRI-compatible script, you can
6453 use the `ABSOLUTE' command to restrict the sections that will be
6454 present in your output program. If the `ABSOLUTE' command is used
6455 at all in a script, then only the sections named explicitly in
6456 `ABSOLUTE' commands will appear in the linker output. You can
6457 still use other input sections (whatever you select on the command
6458 line, or using `LOAD') to resolve addresses in the output file.
6460 `ALIAS OUT-SECNAME, IN-SECNAME'
6461 Use this command to place the data from input section IN-SECNAME
6462 in a section called OUT-SECNAME in the linker output file.
6464 IN-SECNAME may be an integer.
6466 `ALIGN SECNAME = EXPRESSION'
6467 Align the section called SECNAME to EXPRESSION. The EXPRESSION
6468 should be a power of two.
6471 Use the value of EXPRESSION as the lowest address (other than
6472 absolute addresses) in the output file.
6475 `CHIP EXPRESSION, EXPRESSION'
6476 This command does nothing; it is accepted only for compatibility.
6479 This command does nothing whatever; it's only accepted for
6482 `FORMAT OUTPUT-FORMAT'
6483 Similar to the `OUTPUT_FORMAT' command in the more general linker
6484 language, but restricted to one of these output formats:
6486 1. S-records, if OUTPUT-FORMAT is `S'
6488 2. IEEE, if OUTPUT-FORMAT is `IEEE'
6490 3. COFF (the `coff-m68k' variant in BFD), if OUTPUT-FORMAT is
6494 Print (to the standard output file) a link map, as produced by the
6495 `ld' command-line option `-M'.
6497 The keyword `LIST' may be followed by anything on the same line,
6498 with no change in its effect.
6501 `LOAD FILENAME, FILENAME, ... FILENAME'
6502 Include one or more object file FILENAME in the link; this has the
6503 same effect as specifying FILENAME directly on the `ld' command
6507 OUTPUT-NAME is the name for the program produced by `ld'; the
6508 MRI-compatible command `NAME' is equivalent to the command-line
6509 option `-o' or the general script language command `OUTPUT'.
6511 `ORDER SECNAME, SECNAME, ... SECNAME'
6512 `ORDER SECNAME SECNAME SECNAME'
6513 Normally, `ld' orders the sections in its output file in the order
6514 in which they first appear in the input files. In an
6515 MRI-compatible script, you can override this ordering with the
6516 `ORDER' command. The sections you list with `ORDER' will appear
6517 first in your output file, in the order specified.
6519 `PUBLIC NAME=EXPRESSION'
6520 `PUBLIC NAME,EXPRESSION'
6521 `PUBLIC NAME EXPRESSION'
6522 Supply a value (EXPRESSION) for external symbol NAME used in the
6525 `SECT SECNAME, EXPRESSION'
6526 `SECT SECNAME=EXPRESSION'
6527 `SECT SECNAME EXPRESSION'
6528 You can use any of these three forms of the `SECT' command to
6529 specify the start address (EXPRESSION) for section SECNAME. If
6530 you have more than one `SECT' statement for the same SECNAME, only
6531 the _first_ sets the start address.
6534 File: ld.info, Node: GNU Free Documentation License, Next: LD Index, Prev: MRI, Up: Top
6536 Appendix B GNU Free Documentation License
6537 *****************************************
6539 Version 1.3, 3 November 2008
6541 Copyright (C) 2000, 2001, 2002, 2007, 2008 Free Software Foundation, Inc.
6544 Everyone is permitted to copy and distribute verbatim copies
6545 of this license document, but changing it is not allowed.
6549 The purpose of this License is to make a manual, textbook, or other
6550 functional and useful document "free" in the sense of freedom: to
6551 assure everyone the effective freedom to copy and redistribute it,
6552 with or without modifying it, either commercially or
6553 noncommercially. Secondarily, this License preserves for the
6554 author and publisher a way to get credit for their work, while not
6555 being considered responsible for modifications made by others.
6557 This License is a kind of "copyleft", which means that derivative
6558 works of the document must themselves be free in the same sense.
6559 It complements the GNU General Public License, which is a copyleft
6560 license designed for free software.
6562 We have designed this License in order to use it for manuals for
6563 free software, because free software needs free documentation: a
6564 free program should come with manuals providing the same freedoms
6565 that the software does. But this License is not limited to
6566 software manuals; it can be used for any textual work, regardless
6567 of subject matter or whether it is published as a printed book.
6568 We recommend this License principally for works whose purpose is
6569 instruction or reference.
6571 1. APPLICABILITY AND DEFINITIONS
6573 This License applies to any manual or other work, in any medium,
6574 that contains a notice placed by the copyright holder saying it
6575 can be distributed under the terms of this License. Such a notice
6576 grants a world-wide, royalty-free license, unlimited in duration,
6577 to use that work under the conditions stated herein. The
6578 "Document", below, refers to any such manual or work. Any member
6579 of the public is a licensee, and is addressed as "you". You
6580 accept the license if you copy, modify or distribute the work in a
6581 way requiring permission under copyright law.
6583 A "Modified Version" of the Document means any work containing the
6584 Document or a portion of it, either copied verbatim, or with
6585 modifications and/or translated into another language.
6587 A "Secondary Section" is a named appendix or a front-matter section
6588 of the Document that deals exclusively with the relationship of the
6589 publishers or authors of the Document to the Document's overall
6590 subject (or to related matters) and contains nothing that could
6591 fall directly within that overall subject. (Thus, if the Document
6592 is in part a textbook of mathematics, a Secondary Section may not
6593 explain any mathematics.) The relationship could be a matter of
6594 historical connection with the subject or with related matters, or
6595 of legal, commercial, philosophical, ethical or political position
6598 The "Invariant Sections" are certain Secondary Sections whose
6599 titles are designated, as being those of Invariant Sections, in
6600 the notice that says that the Document is released under this
6601 License. If a section does not fit the above definition of
6602 Secondary then it is not allowed to be designated as Invariant.
6603 The Document may contain zero Invariant Sections. If the Document
6604 does not identify any Invariant Sections then there are none.
6606 The "Cover Texts" are certain short passages of text that are
6607 listed, as Front-Cover Texts or Back-Cover Texts, in the notice
6608 that says that the Document is released under this License. A
6609 Front-Cover Text may be at most 5 words, and a Back-Cover Text may
6610 be at most 25 words.
6612 A "Transparent" copy of the Document means a machine-readable copy,
6613 represented in a format whose specification is available to the
6614 general public, that is suitable for revising the document
6615 straightforwardly with generic text editors or (for images
6616 composed of pixels) generic paint programs or (for drawings) some
6617 widely available drawing editor, and that is suitable for input to
6618 text formatters or for automatic translation to a variety of
6619 formats suitable for input to text formatters. A copy made in an
6620 otherwise Transparent file format whose markup, or absence of
6621 markup, has been arranged to thwart or discourage subsequent
6622 modification by readers is not Transparent. An image format is
6623 not Transparent if used for any substantial amount of text. A
6624 copy that is not "Transparent" is called "Opaque".
6626 Examples of suitable formats for Transparent copies include plain
6627 ASCII without markup, Texinfo input format, LaTeX input format,
6628 SGML or XML using a publicly available DTD, and
6629 standard-conforming simple HTML, PostScript or PDF designed for
6630 human modification. Examples of transparent image formats include
6631 PNG, XCF and JPG. Opaque formats include proprietary formats that
6632 can be read and edited only by proprietary word processors, SGML or
6633 XML for which the DTD and/or processing tools are not generally
6634 available, and the machine-generated HTML, PostScript or PDF
6635 produced by some word processors for output purposes only.
6637 The "Title Page" means, for a printed book, the title page itself,
6638 plus such following pages as are needed to hold, legibly, the
6639 material this License requires to appear in the title page. For
6640 works in formats which do not have any title page as such, "Title
6641 Page" means the text near the most prominent appearance of the
6642 work's title, preceding the beginning of the body of the text.
6644 The "publisher" means any person or entity that distributes copies
6645 of the Document to the public.
6647 A section "Entitled XYZ" means a named subunit of the Document
6648 whose title either is precisely XYZ or contains XYZ in parentheses
6649 following text that translates XYZ in another language. (Here XYZ
6650 stands for a specific section name mentioned below, such as
6651 "Acknowledgements", "Dedications", "Endorsements", or "History".)
6652 To "Preserve the Title" of such a section when you modify the
6653 Document means that it remains a section "Entitled XYZ" according
6656 The Document may include Warranty Disclaimers next to the notice
6657 which states that this License applies to the Document. These
6658 Warranty Disclaimers are considered to be included by reference in
6659 this License, but only as regards disclaiming warranties: any other
6660 implication that these Warranty Disclaimers may have is void and
6661 has no effect on the meaning of this License.
6665 You may copy and distribute the Document in any medium, either
6666 commercially or noncommercially, provided that this License, the
6667 copyright notices, and the license notice saying this License
6668 applies to the Document are reproduced in all copies, and that you
6669 add no other conditions whatsoever to those of this License. You
6670 may not use technical measures to obstruct or control the reading
6671 or further copying of the copies you make or distribute. However,
6672 you may accept compensation in exchange for copies. If you
6673 distribute a large enough number of copies you must also follow
6674 the conditions in section 3.
6676 You may also lend copies, under the same conditions stated above,
6677 and you may publicly display copies.
6679 3. COPYING IN QUANTITY
6681 If you publish printed copies (or copies in media that commonly
6682 have printed covers) of the Document, numbering more than 100, and
6683 the Document's license notice requires Cover Texts, you must
6684 enclose the copies in covers that carry, clearly and legibly, all
6685 these Cover Texts: Front-Cover Texts on the front cover, and
6686 Back-Cover Texts on the back cover. Both covers must also clearly
6687 and legibly identify you as the publisher of these copies. The
6688 front cover must present the full title with all words of the
6689 title equally prominent and visible. You may add other material
6690 on the covers in addition. Copying with changes limited to the
6691 covers, as long as they preserve the title of the Document and
6692 satisfy these conditions, can be treated as verbatim copying in
6695 If the required texts for either cover are too voluminous to fit
6696 legibly, you should put the first ones listed (as many as fit
6697 reasonably) on the actual cover, and continue the rest onto
6700 If you publish or distribute Opaque copies of the Document
6701 numbering more than 100, you must either include a
6702 machine-readable Transparent copy along with each Opaque copy, or
6703 state in or with each Opaque copy a computer-network location from
6704 which the general network-using public has access to download
6705 using public-standard network protocols a complete Transparent
6706 copy of the Document, free of added material. If you use the
6707 latter option, you must take reasonably prudent steps, when you
6708 begin distribution of Opaque copies in quantity, to ensure that
6709 this Transparent copy will remain thus accessible at the stated
6710 location until at least one year after the last time you
6711 distribute an Opaque copy (directly or through your agents or
6712 retailers) of that edition to the public.
6714 It is requested, but not required, that you contact the authors of
6715 the Document well before redistributing any large number of
6716 copies, to give them a chance to provide you with an updated
6717 version of the Document.
6721 You may copy and distribute a Modified Version of the Document
6722 under the conditions of sections 2 and 3 above, provided that you
6723 release the Modified Version under precisely this License, with
6724 the Modified Version filling the role of the Document, thus
6725 licensing distribution and modification of the Modified Version to
6726 whoever possesses a copy of it. In addition, you must do these
6727 things in the Modified Version:
6729 A. Use in the Title Page (and on the covers, if any) a title
6730 distinct from that of the Document, and from those of
6731 previous versions (which should, if there were any, be listed
6732 in the History section of the Document). You may use the
6733 same title as a previous version if the original publisher of
6734 that version gives permission.
6736 B. List on the Title Page, as authors, one or more persons or
6737 entities responsible for authorship of the modifications in
6738 the Modified Version, together with at least five of the
6739 principal authors of the Document (all of its principal
6740 authors, if it has fewer than five), unless they release you
6741 from this requirement.
6743 C. State on the Title page the name of the publisher of the
6744 Modified Version, as the publisher.
6746 D. Preserve all the copyright notices of the Document.
6748 E. Add an appropriate copyright notice for your modifications
6749 adjacent to the other copyright notices.
6751 F. Include, immediately after the copyright notices, a license
6752 notice giving the public permission to use the Modified
6753 Version under the terms of this License, in the form shown in
6756 G. Preserve in that license notice the full lists of Invariant
6757 Sections and required Cover Texts given in the Document's
6760 H. Include an unaltered copy of this License.
6762 I. Preserve the section Entitled "History", Preserve its Title,
6763 and add to it an item stating at least the title, year, new
6764 authors, and publisher of the Modified Version as given on
6765 the Title Page. If there is no section Entitled "History" in
6766 the Document, create one stating the title, year, authors,
6767 and publisher of the Document as given on its Title Page,
6768 then add an item describing the Modified Version as stated in
6769 the previous sentence.
6771 J. Preserve the network location, if any, given in the Document
6772 for public access to a Transparent copy of the Document, and
6773 likewise the network locations given in the Document for
6774 previous versions it was based on. These may be placed in
6775 the "History" section. You may omit a network location for a
6776 work that was published at least four years before the
6777 Document itself, or if the original publisher of the version
6778 it refers to gives permission.
6780 K. For any section Entitled "Acknowledgements" or "Dedications",
6781 Preserve the Title of the section, and preserve in the
6782 section all the substance and tone of each of the contributor
6783 acknowledgements and/or dedications given therein.
6785 L. Preserve all the Invariant Sections of the Document,
6786 unaltered in their text and in their titles. Section numbers
6787 or the equivalent are not considered part of the section
6790 M. Delete any section Entitled "Endorsements". Such a section
6791 may not be included in the Modified Version.
6793 N. Do not retitle any existing section to be Entitled
6794 "Endorsements" or to conflict in title with any Invariant
6797 O. Preserve any Warranty Disclaimers.
6799 If the Modified Version includes new front-matter sections or
6800 appendices that qualify as Secondary Sections and contain no
6801 material copied from the Document, you may at your option
6802 designate some or all of these sections as invariant. To do this,
6803 add their titles to the list of Invariant Sections in the Modified
6804 Version's license notice. These titles must be distinct from any
6805 other section titles.
6807 You may add a section Entitled "Endorsements", provided it contains
6808 nothing but endorsements of your Modified Version by various
6809 parties--for example, statements of peer review or that the text
6810 has been approved by an organization as the authoritative
6811 definition of a standard.
6813 You may add a passage of up to five words as a Front-Cover Text,
6814 and a passage of up to 25 words as a Back-Cover Text, to the end
6815 of the list of Cover Texts in the Modified Version. Only one
6816 passage of Front-Cover Text and one of Back-Cover Text may be
6817 added by (or through arrangements made by) any one entity. If the
6818 Document already includes a cover text for the same cover,
6819 previously added by you or by arrangement made by the same entity
6820 you are acting on behalf of, you may not add another; but you may
6821 replace the old one, on explicit permission from the previous
6822 publisher that added the old one.
6824 The author(s) and publisher(s) of the Document do not by this
6825 License give permission to use their names for publicity for or to
6826 assert or imply endorsement of any Modified Version.
6828 5. COMBINING DOCUMENTS
6830 You may combine the Document with other documents released under
6831 this License, under the terms defined in section 4 above for
6832 modified versions, provided that you include in the combination
6833 all of the Invariant Sections of all of the original documents,
6834 unmodified, and list them all as Invariant Sections of your
6835 combined work in its license notice, and that you preserve all
6836 their Warranty Disclaimers.
6838 The combined work need only contain one copy of this License, and
6839 multiple identical Invariant Sections may be replaced with a single
6840 copy. If there are multiple Invariant Sections with the same name
6841 but different contents, make the title of each such section unique
6842 by adding at the end of it, in parentheses, the name of the
6843 original author or publisher of that section if known, or else a
6844 unique number. Make the same adjustment to the section titles in
6845 the list of Invariant Sections in the license notice of the
6848 In the combination, you must combine any sections Entitled
6849 "History" in the various original documents, forming one section
6850 Entitled "History"; likewise combine any sections Entitled
6851 "Acknowledgements", and any sections Entitled "Dedications". You
6852 must delete all sections Entitled "Endorsements."
6854 6. COLLECTIONS OF DOCUMENTS
6856 You may make a collection consisting of the Document and other
6857 documents released under this License, and replace the individual
6858 copies of this License in the various documents with a single copy
6859 that is included in the collection, provided that you follow the
6860 rules of this License for verbatim copying of each of the
6861 documents in all other respects.
6863 You may extract a single document from such a collection, and
6864 distribute it individually under this License, provided you insert
6865 a copy of this License into the extracted document, and follow
6866 this License in all other respects regarding verbatim copying of
6869 7. AGGREGATION WITH INDEPENDENT WORKS
6871 A compilation of the Document or its derivatives with other
6872 separate and independent documents or works, in or on a volume of
6873 a storage or distribution medium, is called an "aggregate" if the
6874 copyright resulting from the compilation is not used to limit the
6875 legal rights of the compilation's users beyond what the individual
6876 works permit. When the Document is included in an aggregate, this
6877 License does not apply to the other works in the aggregate which
6878 are not themselves derivative works of the Document.
6880 If the Cover Text requirement of section 3 is applicable to these
6881 copies of the Document, then if the Document is less than one half
6882 of the entire aggregate, the Document's Cover Texts may be placed
6883 on covers that bracket the Document within the aggregate, or the
6884 electronic equivalent of covers if the Document is in electronic
6885 form. Otherwise they must appear on printed covers that bracket
6886 the whole aggregate.
6890 Translation is considered a kind of modification, so you may
6891 distribute translations of the Document under the terms of section
6892 4. Replacing Invariant Sections with translations requires special
6893 permission from their copyright holders, but you may include
6894 translations of some or all Invariant Sections in addition to the
6895 original versions of these Invariant Sections. You may include a
6896 translation of this License, and all the license notices in the
6897 Document, and any Warranty Disclaimers, provided that you also
6898 include the original English version of this License and the
6899 original versions of those notices and disclaimers. In case of a
6900 disagreement between the translation and the original version of
6901 this License or a notice or disclaimer, the original version will
6904 If a section in the Document is Entitled "Acknowledgements",
6905 "Dedications", or "History", the requirement (section 4) to
6906 Preserve its Title (section 1) will typically require changing the
6911 You may not copy, modify, sublicense, or distribute the Document
6912 except as expressly provided under this License. Any attempt
6913 otherwise to copy, modify, sublicense, or distribute it is void,
6914 and will automatically terminate your rights under this License.
6916 However, if you cease all violation of this License, then your
6917 license from a particular copyright holder is reinstated (a)
6918 provisionally, unless and until the copyright holder explicitly
6919 and finally terminates your license, and (b) permanently, if the
6920 copyright holder fails to notify you of the violation by some
6921 reasonable means prior to 60 days after the cessation.
6923 Moreover, your license from a particular copyright holder is
6924 reinstated permanently if the copyright holder notifies you of the
6925 violation by some reasonable means, this is the first time you have
6926 received notice of violation of this License (for any work) from
6927 that copyright holder, and you cure the violation prior to 30 days
6928 after your receipt of the notice.
6930 Termination of your rights under this section does not terminate
6931 the licenses of parties who have received copies or rights from
6932 you under this License. If your rights have been terminated and
6933 not permanently reinstated, receipt of a copy of some or all of
6934 the same material does not give you any rights to use it.
6936 10. FUTURE REVISIONS OF THIS LICENSE
6938 The Free Software Foundation may publish new, revised versions of
6939 the GNU Free Documentation License from time to time. Such new
6940 versions will be similar in spirit to the present version, but may
6941 differ in detail to address new problems or concerns. See
6942 `http://www.gnu.org/copyleft/'.
6944 Each version of the License is given a distinguishing version
6945 number. If the Document specifies that a particular numbered
6946 version of this License "or any later version" applies to it, you
6947 have the option of following the terms and conditions either of
6948 that specified version or of any later version that has been
6949 published (not as a draft) by the Free Software Foundation. If
6950 the Document does not specify a version number of this License,
6951 you may choose any version ever published (not as a draft) by the
6952 Free Software Foundation. If the Document specifies that a proxy
6953 can decide which future versions of this License can be used, that
6954 proxy's public statement of acceptance of a version permanently
6955 authorizes you to choose that version for the Document.
6959 "Massive Multiauthor Collaboration Site" (or "MMC Site") means any
6960 World Wide Web server that publishes copyrightable works and also
6961 provides prominent facilities for anybody to edit those works. A
6962 public wiki that anybody can edit is an example of such a server.
6963 A "Massive Multiauthor Collaboration" (or "MMC") contained in the
6964 site means any set of copyrightable works thus published on the MMC
6967 "CC-BY-SA" means the Creative Commons Attribution-Share Alike 3.0
6968 license published by Creative Commons Corporation, a not-for-profit
6969 corporation with a principal place of business in San Francisco,
6970 California, as well as future copyleft versions of that license
6971 published by that same organization.
6973 "Incorporate" means to publish or republish a Document, in whole or
6974 in part, as part of another Document.
6976 An MMC is "eligible for relicensing" if it is licensed under this
6977 License, and if all works that were first published under this
6978 License somewhere other than this MMC, and subsequently
6979 incorporated in whole or in part into the MMC, (1) had no cover
6980 texts or invariant sections, and (2) were thus incorporated prior
6981 to November 1, 2008.
6983 The operator of an MMC Site may republish an MMC contained in the
6984 site under CC-BY-SA on the same site at any time before August 1,
6985 2009, provided the MMC is eligible for relicensing.
6988 ADDENDUM: How to use this License for your documents
6989 ====================================================
6991 To use this License in a document you have written, include a copy of
6992 the License in the document and put the following copyright and license
6993 notices just after the title page:
6995 Copyright (C) YEAR YOUR NAME.
6996 Permission is granted to copy, distribute and/or modify this document
6997 under the terms of the GNU Free Documentation License, Version 1.3
6998 or any later version published by the Free Software Foundation;
6999 with no Invariant Sections, no Front-Cover Texts, and no Back-Cover
7000 Texts. A copy of the license is included in the section entitled ``GNU
7001 Free Documentation License''.
7003 If you have Invariant Sections, Front-Cover Texts and Back-Cover
7004 Texts, replace the "with...Texts." line with this:
7006 with the Invariant Sections being LIST THEIR TITLES, with
7007 the Front-Cover Texts being LIST, and with the Back-Cover Texts
7010 If you have Invariant Sections without Cover Texts, or some other
7011 combination of the three, merge those two alternatives to suit the
7014 If your document contains nontrivial examples of program code, we
7015 recommend releasing these examples in parallel under your choice of
7016 free software license, such as the GNU General Public License, to
7017 permit their use in free software.
7020 File: ld.info, Node: LD Index, Prev: GNU Free Documentation License, Up: Top
7028 * ": Symbols. (line 6)
7029 * -(: Options. (line 696)
7030 * --accept-unknown-input-arch: Options. (line 714)
7031 * --add-needed: Options. (line 738)
7032 * --add-stdcall-alias: Options. (line 1573)
7033 * --allow-multiple-definition: Options. (line 984)
7034 * --allow-shlib-undefined: Options. (line 990)
7035 * --architecture=ARCH: Options. (line 123)
7036 * --as-needed: Options. (line 724)
7037 * --audit AUDITLIB: Options. (line 112)
7038 * --auxiliary=NAME: Options. (line 255)
7039 * --bank-window: Options. (line 1982)
7040 * --base-file: Options. (line 1578)
7041 * --be8: ARM. (line 28)
7042 * --bss-plt: PowerPC ELF32. (line 16)
7043 * --build-id: Options. (line 1535)
7044 * --build-id=STYLE: Options. (line 1535)
7045 * --check-sections: Options. (line 817)
7046 * --copy-dt-needed-entries: Options. (line 829)
7047 * --cref: Options. (line 850)
7048 * --default-imported-symver: Options. (line 1027)
7049 * --default-script=SCRIPT: Options. (line 541)
7050 * --default-symver: Options. (line 1023)
7051 * --defsym=SYMBOL=EXP: Options. (line 878)
7052 * --demangle[=STYLE]: Options. (line 891)
7053 * --depaudit AUDITLIB: Options. (line 177)
7054 * --disable-auto-image-base: Options. (line 1757)
7055 * --disable-auto-import: Options. (line 1892)
7056 * --disable-long-section-names: Options. (line 1588)
7057 * --disable-new-dtags: Options. (line 1498)
7058 * --disable-runtime-pseudo-reloc: Options. (line 1905)
7059 * --disable-stdcall-fixup: Options. (line 1610)
7060 * --discard-all: Options. (line 587)
7061 * --discard-locals: Options. (line 591)
7062 * --dll: Options. (line 1583)
7063 * --dll-search-prefix: Options. (line 1763)
7064 * --dotsyms: PowerPC64 ELF64. (line 33)
7065 * --dynamic-linker=FILE: Options. (line 904)
7066 * --dynamic-list-cpp-new: Options. (line 809)
7067 * --dynamic-list-cpp-typeinfo: Options. (line 813)
7068 * --dynamic-list-data: Options. (line 806)
7069 * --dynamic-list=DYNAMIC-LIST-FILE: Options. (line 793)
7070 * --dynamicbase: Options. (line 1941)
7071 * --eh-frame-hdr: Options. (line 1494)
7072 * --emit-relocs: Options. (line 476)
7073 * --emit-stack-syms: SPU ELF. (line 46)
7074 * --emit-stub-syms <1>: PowerPC ELF32. (line 47)
7075 * --emit-stub-syms <2>: SPU ELF. (line 15)
7076 * --emit-stub-syms: PowerPC64 ELF64. (line 29)
7077 * --enable-auto-image-base: Options. (line 1749)
7078 * --enable-auto-import: Options. (line 1772)
7079 * --enable-extra-pe-debug: Options. (line 1910)
7080 * --enable-long-section-names: Options. (line 1588)
7081 * --enable-new-dtags: Options. (line 1498)
7082 * --enable-runtime-pseudo-reloc: Options. (line 1897)
7083 * --enable-stdcall-fixup: Options. (line 1610)
7084 * --entry=ENTRY: Options. (line 187)
7085 * --error-unresolved-symbols: Options. (line 1447)
7086 * --exclude-all-symbols: Options. (line 1664)
7087 * --exclude-libs: Options. (line 197)
7088 * --exclude-modules-for-implib: Options. (line 208)
7089 * --exclude-symbols: Options. (line 1658)
7090 * --export-all-symbols: Options. (line 1634)
7091 * --export-dynamic: Options. (line 221)
7092 * --extra-overlay-stubs: SPU ELF. (line 19)
7093 * --fatal-warnings: Options. (line 911)
7094 * --file-alignment: Options. (line 1668)
7095 * --filter=NAME: Options. (line 276)
7096 * --fix-cortex-a8: i960. (line 39)
7097 * --fix-v4bx: ARM. (line 49)
7098 * --fix-v4bx-interworking: ARM. (line 62)
7099 * --force-dynamic: Options. (line 485)
7100 * --force-exe-suffix: Options. (line 916)
7101 * --forceinteg: Options. (line 1946)
7102 * --format=FORMAT: Options. (line 134)
7103 * --format=VERSION: TI COFF. (line 6)
7104 * --gc-sections: Options. (line 926)
7105 * --got: Options. (line 1995)
7106 * --got=TYPE: M68K. (line 6)
7107 * --gpsize=VALUE: Options. (line 309)
7108 * --hash-size=NUMBER: Options. (line 1507)
7109 * --hash-style=STYLE: Options. (line 1515)
7110 * --heap: Options. (line 1674)
7111 * --help: Options. (line 957)
7112 * --image-base: Options. (line 1681)
7113 * --just-symbols=FILE: Options. (line 508)
7114 * --kill-at: Options. (line 1690)
7115 * --large-address-aware: Options. (line 1695)
7116 * --leading-underscore: Options. (line 1628)
7117 * --library-path=DIR: Options. (line 368)
7118 * --library=NAMESPEC: Options. (line 335)
7119 * --local-store=lo:hi: SPU ELF. (line 24)
7120 * --major-image-version: Options. (line 1704)
7121 * --major-os-version: Options. (line 1709)
7122 * --major-subsystem-version: Options. (line 1713)
7123 * --merge-exidx-entries: i960. (line 48)
7124 * --minor-image-version: Options. (line 1718)
7125 * --minor-os-version: Options. (line 1723)
7126 * --minor-subsystem-version: Options. (line 1727)
7127 * --mri-script=MRI-CMDFILE: Options. (line 158)
7128 * --multi-subspace: HPPA ELF32. (line 6)
7129 * --nmagic: Options. (line 439)
7130 * --no-accept-unknown-input-arch: Options. (line 714)
7131 * --no-add-needed: Options. (line 738)
7132 * --no-allow-shlib-undefined: Options. (line 990)
7133 * --no-as-needed: Options. (line 724)
7134 * --no-bind: Options. (line 1960)
7135 * --no-check-sections: Options. (line 817)
7136 * --no-copy-dt-needed-entries: Options. (line 829)
7137 * --no-define-common: Options. (line 862)
7138 * --no-demangle: Options. (line 891)
7139 * --no-dotsyms: PowerPC64 ELF64. (line 33)
7140 * --no-enum-size-warning: ARM. (line 111)
7141 * --no-export-dynamic: Options. (line 221)
7142 * --no-fatal-warnings: Options. (line 911)
7143 * --no-fix-cortex-a8: i960. (line 39)
7144 * --no-gc-sections: Options. (line 926)
7145 * --no-isolation: Options. (line 1953)
7146 * --no-keep-memory: Options. (line 969)
7147 * --no-leading-underscore: Options. (line 1628)
7148 * --no-merge-exidx-entries: i960. (line 48)
7149 * --no-multi-toc: PowerPC64 ELF64. (line 74)
7150 * --no-omagic: Options. (line 454)
7151 * --no-opd-optimize: PowerPC64 ELF64. (line 48)
7152 * --no-overlays: SPU ELF. (line 9)
7153 * --no-print-gc-sections: Options. (line 948)
7154 * --no-seh: Options. (line 1956)
7155 * --no-tls-optimize <1>: PowerPC ELF32. (line 51)
7156 * --no-tls-optimize: PowerPC64 ELF64. (line 43)
7157 * --no-toc-optimize: PowerPC64 ELF64. (line 60)
7158 * --no-trampoline: Options. (line 1976)
7159 * --no-undefined: Options. (line 976)
7160 * --no-undefined-version: Options. (line 1018)
7161 * --no-warn-mismatch: Options. (line 1031)
7162 * --no-warn-search-mismatch: Options. (line 1040)
7163 * --no-wchar-size-warning: ARM. (line 118)
7164 * --no-whole-archive: Options. (line 1044)
7165 * --noinhibit-exec: Options. (line 1048)
7166 * --non-overlapping-opd: PowerPC64 ELF64. (line 54)
7167 * --nxcompat: Options. (line 1949)
7168 * --oformat=OUTPUT-FORMAT: Options. (line 1060)
7169 * --omagic: Options. (line 445)
7170 * --out-implib: Options. (line 1740)
7171 * --output-def: Options. (line 1732)
7172 * --output=OUTPUT: Options. (line 460)
7173 * --pic-executable: Options. (line 1073)
7174 * --pic-veneer: ARM. (line 124)
7175 * --plugin: SPU ELF. (line 6)
7176 * --print-gc-sections: Options. (line 948)
7177 * --print-map: Options. (line 402)
7178 * --reduce-memory-overheads: Options. (line 1521)
7179 * --relax: Options. (line 1089)
7180 * --relax on i960: i960. (line 31)
7181 * --relax on PowerPC: PowerPC ELF32. (line 6)
7182 * --relax on Xtensa: Xtensa. (line 27)
7183 * --relocatable: Options. (line 489)
7184 * --retain-symbols-file=FILENAME: Options. (line 1115)
7185 * --script=SCRIPT: Options. (line 532)
7186 * --sdata-got: PowerPC ELF32. (line 33)
7187 * --section-alignment: Options. (line 1915)
7188 * --section-start=SECTIONNAME=ORG: Options. (line 1271)
7189 * --secure-plt: PowerPC ELF32. (line 26)
7190 * --sort-common: Options. (line 1213)
7191 * --sort-section=alignment: Options. (line 1228)
7192 * --sort-section=name: Options. (line 1224)
7193 * --split-by-file: Options. (line 1232)
7194 * --split-by-reloc: Options. (line 1237)
7195 * --stack: Options. (line 1921)
7196 * --stack-analysis: SPU ELF. (line 29)
7197 * --stats: Options. (line 1250)
7198 * --strip-all: Options. (line 519)
7199 * --strip-debug: Options. (line 523)
7200 * --stub-group-size: PowerPC64 ELF64. (line 6)
7201 * --stub-group-size=N <1>: ARM. (line 129)
7202 * --stub-group-size=N: HPPA ELF32. (line 12)
7203 * --subsystem: Options. (line 1928)
7204 * --support-old-code: ARM. (line 6)
7205 * --sysroot=DIRECTORY: Options. (line 1254)
7206 * --target-help: Options. (line 961)
7207 * --target1-abs: ARM. (line 32)
7208 * --target1-rel: ARM. (line 32)
7209 * --target2=TYPE: ARM. (line 37)
7210 * --thumb-entry=ENTRY: ARM. (line 17)
7211 * --trace: Options. (line 528)
7212 * --trace-symbol=SYMBOL: Options. (line 597)
7213 * --traditional-format: Options. (line 1259)
7214 * --tsaware: Options. (line 1966)
7215 * --undefined=SYMBOL: Options. (line 554)
7216 * --unique[=SECTION]: Options. (line 572)
7217 * --unresolved-symbols: Options. (line 1290)
7218 * --use-blx: ARM. (line 74)
7219 * --use-nul-prefixed-import-tables: ARM. (line 23)
7220 * --verbose[=NUMBER]: Options. (line 1319)
7221 * --version: Options. (line 581)
7222 * --version-script=VERSION-SCRIPTFILE: Options. (line 1327)
7223 * --vfp11-denorm-fix: ARM. (line 83)
7224 * --warn-alternate-em: Options. (line 1439)
7225 * --warn-common: Options. (line 1338)
7226 * --warn-constructors: Options. (line 1406)
7227 * --warn-multiple-gp: Options. (line 1411)
7228 * --warn-once: Options. (line 1425)
7229 * --warn-section-align: Options. (line 1429)
7230 * --warn-shared-textrel: Options. (line 1436)
7231 * --warn-unresolved-symbols: Options. (line 1442)
7232 * --wdmdriver: Options. (line 1963)
7233 * --whole-archive: Options. (line 1451)
7234 * --wrap=SYMBOL: Options. (line 1465)
7235 * -A ARCH: Options. (line 122)
7236 * -a KEYWORD: Options. (line 105)
7237 * -assert KEYWORD: Options. (line 745)
7238 * -b FORMAT: Options. (line 134)
7239 * -Bdynamic: Options. (line 748)
7240 * -Bgroup: Options. (line 758)
7241 * -Bshareable: Options. (line 1206)
7242 * -Bstatic: Options. (line 765)
7243 * -Bsymbolic: Options. (line 780)
7244 * -Bsymbolic-functions: Options. (line 787)
7245 * -c MRI-CMDFILE: Options. (line 158)
7246 * -call_shared: Options. (line 748)
7247 * -d: Options. (line 168)
7248 * -dc: Options. (line 168)
7249 * -dn: Options. (line 765)
7250 * -dp: Options. (line 168)
7251 * -dT SCRIPT: Options. (line 541)
7252 * -dy: Options. (line 748)
7253 * -E: Options. (line 221)
7254 * -e ENTRY: Options. (line 187)
7255 * -EB: Options. (line 248)
7256 * -EL: Options. (line 251)
7257 * -f NAME: Options. (line 255)
7258 * -F NAME: Options. (line 276)
7259 * -fini=NAME: Options. (line 300)
7260 * -g: Options. (line 306)
7261 * -G VALUE: Options. (line 309)
7262 * -h NAME: Options. (line 317)
7263 * -i: Options. (line 326)
7264 * -IFILE: Options. (line 904)
7265 * -init=NAME: Options. (line 329)
7266 * -L DIR: Options. (line 368)
7267 * -l NAMESPEC: Options. (line 335)
7268 * -M: Options. (line 402)
7269 * -m EMULATION: Options. (line 392)
7270 * -Map=MAPFILE: Options. (line 965)
7271 * -n: Options. (line 439)
7272 * -N: Options. (line 445)
7273 * -no-relax: Options. (line 1089)
7274 * -non_shared: Options. (line 765)
7275 * -nostdlib: Options. (line 1054)
7276 * -O LEVEL: Options. (line 466)
7277 * -o OUTPUT: Options. (line 460)
7278 * -P AUDITLIB: Options. (line 177)
7279 * -pie: Options. (line 1073)
7280 * -q: Options. (line 476)
7281 * -qmagic: Options. (line 1083)
7282 * -Qy: Options. (line 1086)
7283 * -r: Options. (line 489)
7284 * -R FILE: Options. (line 508)
7285 * -rpath-link=DIR: Options. (line 1151)
7286 * -rpath=DIR: Options. (line 1129)
7287 * -s: Options. (line 519)
7288 * -S: Options. (line 523)
7289 * -shared: Options. (line 1206)
7290 * -soname=NAME: Options. (line 317)
7291 * -static: Options. (line 765)
7292 * -t: Options. (line 528)
7293 * -T SCRIPT: Options. (line 532)
7294 * -Tbss=ORG: Options. (line 1280)
7295 * -Tdata=ORG: Options. (line 1280)
7296 * -Ttext-segment=ORG: Options. (line 1286)
7297 * -Ttext=ORG: Options. (line 1280)
7298 * -u SYMBOL: Options. (line 554)
7299 * -Ur: Options. (line 562)
7300 * -V: Options. (line 581)
7301 * -v: Options. (line 581)
7302 * -X: Options. (line 591)
7303 * -x: Options. (line 587)
7304 * -Y PATH: Options. (line 606)
7305 * -y SYMBOL: Options. (line 597)
7306 * -z defs: Options. (line 976)
7307 * -z KEYWORD: Options. (line 610)
7308 * -z muldefs: Options. (line 984)
7309 * .: Location Counter. (line 6)
7310 * /DISCARD/: Output Section Discarding.
7312 * :PHDR: Output Section Phdr.
7314 * =FILLEXP: Output Section Fill.
7316 * >REGION: Output Section Region.
7318 * [COMMON]: Input Section Common.
7320 * ABSOLUTE (MRI): MRI. (line 33)
7321 * absolute and relocatable symbols: Expression Section. (line 6)
7322 * absolute expressions: Expression Section. (line 6)
7323 * ABSOLUTE(EXP): Builtin Functions. (line 10)
7324 * ADDR(SECTION): Builtin Functions. (line 17)
7325 * address, section: Output Section Address.
7327 * ALIAS (MRI): MRI. (line 44)
7328 * ALIGN (MRI): MRI. (line 50)
7329 * align expression: Builtin Functions. (line 38)
7330 * align location counter: Builtin Functions. (line 38)
7331 * ALIGN(ALIGN): Builtin Functions. (line 38)
7332 * ALIGN(EXP,ALIGN): Builtin Functions. (line 38)
7333 * ALIGN(SECTION_ALIGN): Forced Output Alignment.
7335 * aligned common symbols: WIN32. (line 424)
7336 * ALIGNOF(SECTION): Builtin Functions. (line 64)
7337 * allocating memory: MEMORY. (line 6)
7338 * architecture: Miscellaneous Commands.
7340 * architectures: Options. (line 122)
7341 * archive files, from cmd line: Options. (line 335)
7342 * archive search path in linker script: File Commands. (line 74)
7343 * arithmetic: Expressions. (line 6)
7344 * arithmetic operators: Operators. (line 6)
7345 * ARM interworking support: ARM. (line 6)
7346 * AS_NEEDED(FILES): File Commands. (line 54)
7347 * ASSERT: Miscellaneous Commands.
7349 * assertion in linker script: Miscellaneous Commands.
7351 * assignment in scripts: Assignments. (line 6)
7352 * AT(LMA): Output Section LMA. (line 6)
7353 * AT>LMA_REGION: Output Section LMA. (line 6)
7354 * automatic data imports: WIN32. (line 191)
7355 * back end: BFD. (line 6)
7356 * BASE (MRI): MRI. (line 54)
7357 * BE8: ARM. (line 28)
7358 * BFD canonical format: Canonical format. (line 11)
7359 * BFD requirements: BFD. (line 16)
7360 * big-endian objects: Options. (line 248)
7361 * binary input format: Options. (line 134)
7362 * BLOCK(EXP): Builtin Functions. (line 77)
7363 * bug criteria: Bug Criteria. (line 6)
7364 * bug reports: Bug Reporting. (line 6)
7365 * bugs in ld: Reporting Bugs. (line 6)
7366 * BYTE(EXPRESSION): Output Section Data.
7368 * C++ constructors, arranging in link: Output Section Keywords.
7370 * CHIP (MRI): MRI. (line 58)
7371 * COLLECT_NO_DEMANGLE: Environment. (line 29)
7372 * combining symbols, warnings on: Options. (line 1338)
7373 * command files: Scripts. (line 6)
7374 * command line: Options. (line 6)
7375 * common allocation: Options. (line 862)
7376 * common allocation in linker script: Miscellaneous Commands.
7378 * common symbol placement: Input Section Common.
7380 * COMMONPAGESIZE: Symbolic Constants. (line 13)
7381 * compatibility, MRI: Options. (line 158)
7382 * CONSTANT: Symbolic Constants. (line 6)
7383 * constants in linker scripts: Constants. (line 6)
7384 * constraints on output sections: Output Section Constraint.
7386 * CONSTRUCTORS: Output Section Keywords.
7388 * constructors: Options. (line 562)
7389 * constructors, arranging in link: Output Section Keywords.
7391 * Cortex-A8 erratum workaround: i960. (line 39)
7392 * crash of linker: Bug Criteria. (line 9)
7393 * CREATE_OBJECT_SYMBOLS: Output Section Keywords.
7395 * creating a DEF file: WIN32. (line 158)
7396 * cross reference table: Options. (line 850)
7397 * cross references: Miscellaneous Commands.
7399 * current output location: Location Counter. (line 6)
7400 * data: Output Section Data.
7402 * DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE): Builtin Functions.
7404 * DATA_SEGMENT_END(EXP): Builtin Functions. (line 103)
7405 * DATA_SEGMENT_RELRO_END(OFFSET, EXP): Builtin Functions. (line 109)
7406 * dbx: Options. (line 1264)
7407 * DEF files, creating: Options. (line 1732)
7408 * default emulation: Environment. (line 21)
7409 * default input format: Environment. (line 9)
7410 * DEFINED(SYMBOL): Builtin Functions. (line 120)
7411 * deleting local symbols: Options. (line 587)
7412 * demangling, default: Environment. (line 29)
7413 * demangling, from command line: Options. (line 891)
7414 * direct linking to a dll: WIN32. (line 239)
7415 * discarding sections: Output Section Discarding.
7417 * discontinuous memory: MEMORY. (line 6)
7418 * DLLs, creating: Options. (line 1634)
7419 * DLLs, linking to: Options. (line 1763)
7420 * dot: Location Counter. (line 6)
7421 * dot inside sections: Location Counter. (line 36)
7422 * dot outside sections: Location Counter. (line 66)
7423 * dynamic linker, from command line: Options. (line 904)
7424 * dynamic symbol table: Options. (line 221)
7425 * ELF program headers: PHDRS. (line 6)
7426 * emulation: Options. (line 392)
7427 * emulation, default: Environment. (line 21)
7428 * END (MRI): MRI. (line 62)
7429 * endianness: Options. (line 248)
7430 * entry point: Entry Point. (line 6)
7431 * entry point, from command line: Options. (line 187)
7432 * entry point, thumb: ARM. (line 17)
7433 * ENTRY(SYMBOL): Entry Point. (line 6)
7434 * error on valid input: Bug Criteria. (line 12)
7435 * example of linker script: Simple Example. (line 6)
7436 * exporting DLL symbols: WIN32. (line 19)
7437 * expression evaluation order: Evaluation. (line 6)
7438 * expression sections: Expression Section. (line 6)
7439 * expression, absolute: Builtin Functions. (line 10)
7440 * expressions: Expressions. (line 6)
7441 * EXTERN: Miscellaneous Commands.
7443 * fatal signal: Bug Criteria. (line 9)
7444 * file name wildcard patterns: Input Section Wildcards.
7446 * FILEHDR: PHDRS. (line 62)
7447 * filename symbols: Output Section Keywords.
7449 * fill pattern, entire section: Output Section Fill.
7451 * FILL(EXPRESSION): Output Section Data.
7453 * finalization function: Options. (line 300)
7454 * first input file: File Commands. (line 82)
7455 * first instruction: Entry Point. (line 6)
7456 * FIX_V4BX: ARM. (line 49)
7457 * FIX_V4BX_INTERWORKING: ARM. (line 62)
7458 * FORCE_COMMON_ALLOCATION: Miscellaneous Commands.
7460 * forcing input section alignment: Forced Input Alignment.
7462 * forcing output section alignment: Forced Output Alignment.
7464 * forcing the creation of dynamic sections: Options. (line 485)
7465 * FORMAT (MRI): MRI. (line 66)
7466 * functions in expressions: Builtin Functions. (line 6)
7467 * garbage collection <1>: Options. (line 948)
7468 * garbage collection: Input Section Keep. (line 6)
7469 * generating optimized output: Options. (line 466)
7470 * GNU linker: Overview. (line 6)
7471 * GNUTARGET: Environment. (line 9)
7472 * GROUP(FILES): File Commands. (line 47)
7473 * grouping input files: File Commands. (line 47)
7474 * groups of archives: Options. (line 696)
7475 * H8/300 support: H8/300. (line 6)
7476 * header size: Builtin Functions. (line 183)
7477 * heap size: Options. (line 1674)
7478 * help: Options. (line 957)
7479 * holes: Location Counter. (line 12)
7480 * holes, filling: Output Section Data.
7482 * HPPA multiple sub-space stubs: HPPA ELF32. (line 6)
7483 * HPPA stub grouping: HPPA ELF32. (line 12)
7484 * i960 support: i960. (line 6)
7485 * image base: Options. (line 1681)
7486 * implicit linker scripts: Implicit Linker Scripts.
7488 * import libraries: WIN32. (line 10)
7489 * INCLUDE FILENAME: File Commands. (line 9)
7490 * including a linker script: File Commands. (line 9)
7491 * including an entire archive: Options. (line 1451)
7492 * incremental link: Options. (line 326)
7493 * INHIBIT_COMMON_ALLOCATION: Miscellaneous Commands.
7495 * initialization function: Options. (line 329)
7496 * initialized data in ROM: Output Section LMA. (line 39)
7497 * input file format in linker script: Format Commands. (line 35)
7498 * input filename symbols: Output Section Keywords.
7500 * input files in linker scripts: File Commands. (line 19)
7501 * input files, displaying: Options. (line 528)
7502 * input format: Options. (line 134)
7503 * input object files in linker scripts: File Commands. (line 19)
7504 * input section alignment: Forced Input Alignment.
7506 * input section basics: Input Section Basics.
7508 * input section wildcards: Input Section Wildcards.
7510 * input sections: Input Section. (line 6)
7511 * INPUT(FILES): File Commands. (line 19)
7512 * INSERT: Miscellaneous Commands.
7514 * insert user script into default script: Miscellaneous Commands.
7516 * integer notation: Constants. (line 6)
7517 * integer suffixes: Constants. (line 15)
7518 * internal object-file format: Canonical format. (line 11)
7519 * invalid input: Bug Criteria. (line 14)
7520 * K and M integer suffixes: Constants. (line 15)
7521 * KEEP: Input Section Keep. (line 6)
7522 * l =: MEMORY. (line 74)
7523 * lazy evaluation: Evaluation. (line 6)
7524 * ld bugs, reporting: Bug Reporting. (line 6)
7525 * LD_FEATURE(STRING): Miscellaneous Commands.
7527 * LDEMULATION: Environment. (line 21)
7528 * len =: MEMORY. (line 74)
7529 * LENGTH =: MEMORY. (line 74)
7530 * LENGTH(MEMORY): Builtin Functions. (line 137)
7531 * library search path in linker script: File Commands. (line 74)
7532 * link map: Options. (line 402)
7533 * link-time runtime library search path: Options. (line 1151)
7534 * linker crash: Bug Criteria. (line 9)
7535 * linker script concepts: Basic Script Concepts.
7537 * linker script example: Simple Example. (line 6)
7538 * linker script file commands: File Commands. (line 6)
7539 * linker script format: Script Format. (line 6)
7540 * linker script input object files: File Commands. (line 19)
7541 * linker script simple commands: Simple Commands. (line 6)
7542 * linker scripts: Scripts. (line 6)
7543 * LIST (MRI): MRI. (line 77)
7544 * little-endian objects: Options. (line 251)
7545 * LOAD (MRI): MRI. (line 84)
7546 * load address: Output Section LMA. (line 6)
7547 * LOADADDR(SECTION): Builtin Functions. (line 140)
7548 * loading, preventing: Output Section Type.
7550 * local symbols, deleting: Options. (line 591)
7551 * location counter: Location Counter. (line 6)
7552 * LONG(EXPRESSION): Output Section Data.
7554 * M and K integer suffixes: Constants. (line 15)
7555 * M68HC11 and 68HC12 support: M68HC11/68HC12. (line 6)
7556 * machine architecture: Miscellaneous Commands.
7558 * machine dependencies: Machine Dependent. (line 6)
7559 * mapping input sections to output sections: Input Section. (line 6)
7560 * MAX: Builtin Functions. (line 143)
7561 * MAXPAGESIZE: Symbolic Constants. (line 10)
7562 * MEMORY: MEMORY. (line 6)
7563 * memory region attributes: MEMORY. (line 34)
7564 * memory regions: MEMORY. (line 6)
7565 * memory regions and sections: Output Section Region.
7567 * memory usage: Options. (line 969)
7568 * MIN: Builtin Functions. (line 146)
7569 * Motorola 68K GOT generation: M68K. (line 6)
7570 * MRI compatibility: MRI. (line 6)
7571 * MSP430 extra sections: MSP430. (line 11)
7572 * NAME (MRI): MRI. (line 90)
7573 * name, section: Output Section Name.
7575 * names: Symbols. (line 6)
7576 * naming the output file: Options. (line 460)
7577 * NEXT(EXP): Builtin Functions. (line 150)
7578 * NMAGIC: Options. (line 439)
7579 * NO_ENUM_SIZE_WARNING: ARM. (line 111)
7580 * NO_WCHAR_SIZE_WARNING: ARM. (line 118)
7581 * NOCROSSREFS(SECTIONS): Miscellaneous Commands.
7583 * NOLOAD: Output Section Type.
7585 * not enough room for program headers: Builtin Functions. (line 188)
7586 * o =: MEMORY. (line 69)
7587 * objdump -i: BFD. (line 6)
7588 * object file management: BFD. (line 6)
7589 * object files: Options. (line 29)
7590 * object formats available: BFD. (line 6)
7591 * object size: Options. (line 309)
7592 * OMAGIC: Options. (line 454)
7593 * ONLY_IF_RO: Output Section Constraint.
7595 * ONLY_IF_RW: Output Section Constraint.
7597 * opening object files: BFD outline. (line 6)
7598 * operators for arithmetic: Operators. (line 6)
7599 * options: Options. (line 6)
7600 * ORDER (MRI): MRI. (line 95)
7601 * org =: MEMORY. (line 69)
7602 * ORIGIN =: MEMORY. (line 69)
7603 * ORIGIN(MEMORY): Builtin Functions. (line 156)
7604 * orphan: Orphan Sections. (line 6)
7605 * output file after errors: Options. (line 1048)
7606 * output file format in linker script: Format Commands. (line 10)
7607 * output file name in linker script: File Commands. (line 64)
7608 * output section alignment: Forced Output Alignment.
7610 * output section attributes: Output Section Attributes.
7612 * output section data: Output Section Data.
7614 * OUTPUT(FILENAME): File Commands. (line 64)
7615 * OUTPUT_ARCH(BFDARCH): Miscellaneous Commands.
7617 * OUTPUT_FORMAT(BFDNAME): Format Commands. (line 10)
7618 * OVERLAY: Overlay Description.
7620 * overlays: Overlay Description.
7622 * partial link: Options. (line 489)
7623 * PE import table prefixing: ARM. (line 23)
7624 * PHDRS: PHDRS. (line 62)
7625 * PIC_VENEER: ARM. (line 124)
7626 * position independent executables: Options. (line 1075)
7627 * PowerPC ELF32 options: PowerPC ELF32. (line 16)
7628 * PowerPC GOT: PowerPC ELF32. (line 33)
7629 * PowerPC long branches: PowerPC ELF32. (line 6)
7630 * PowerPC PLT: PowerPC ELF32. (line 16)
7631 * PowerPC stub symbols: PowerPC ELF32. (line 47)
7632 * PowerPC TLS optimization: PowerPC ELF32. (line 51)
7633 * PowerPC64 dot symbols: PowerPC64 ELF64. (line 33)
7634 * PowerPC64 ELF64 options: PowerPC64 ELF64. (line 6)
7635 * PowerPC64 multi-TOC: PowerPC64 ELF64. (line 74)
7636 * PowerPC64 OPD optimization: PowerPC64 ELF64. (line 48)
7637 * PowerPC64 OPD spacing: PowerPC64 ELF64. (line 54)
7638 * PowerPC64 stub grouping: PowerPC64 ELF64. (line 6)
7639 * PowerPC64 stub symbols: PowerPC64 ELF64. (line 29)
7640 * PowerPC64 TLS optimization: PowerPC64 ELF64. (line 43)
7641 * PowerPC64 TOC optimization: PowerPC64 ELF64. (line 60)
7642 * precedence in expressions: Operators. (line 6)
7643 * prevent unnecessary loading: Output Section Type.
7645 * program headers: PHDRS. (line 6)
7646 * program headers and sections: Output Section Phdr.
7648 * program headers, not enough room: Builtin Functions. (line 188)
7649 * program segments: PHDRS. (line 6)
7650 * PROVIDE: PROVIDE. (line 6)
7651 * PROVIDE_HIDDEN: PROVIDE_HIDDEN. (line 6)
7652 * PUBLIC (MRI): MRI. (line 103)
7653 * QUAD(EXPRESSION): Output Section Data.
7655 * quoted symbol names: Symbols. (line 6)
7656 * read-only text: Options. (line 439)
7657 * read/write from cmd line: Options. (line 445)
7658 * region alias: REGION_ALIAS. (line 6)
7659 * region names: REGION_ALIAS. (line 6)
7660 * REGION_ALIAS(ALIAS, REGION): REGION_ALIAS. (line 6)
7661 * regions of memory: MEMORY. (line 6)
7662 * relative expressions: Expression Section. (line 6)
7663 * relaxing addressing modes: Options. (line 1089)
7664 * relaxing on H8/300: H8/300. (line 9)
7665 * relaxing on i960: i960. (line 31)
7666 * relaxing on M68HC11: M68HC11/68HC12. (line 12)
7667 * relaxing on Xtensa: Xtensa. (line 27)
7668 * relocatable and absolute symbols: Expression Section. (line 6)
7669 * relocatable output: Options. (line 489)
7670 * removing sections: Output Section Discarding.
7672 * reporting bugs in ld: Reporting Bugs. (line 6)
7673 * requirements for BFD: BFD. (line 16)
7674 * retain relocations in final executable: Options. (line 476)
7675 * retaining specified symbols: Options. (line 1115)
7676 * ROM initialized data: Output Section LMA. (line 39)
7677 * round up expression: Builtin Functions. (line 38)
7678 * round up location counter: Builtin Functions. (line 38)
7679 * runtime library name: Options. (line 317)
7680 * runtime library search path: Options. (line 1129)
7681 * runtime pseudo-relocation: WIN32. (line 217)
7682 * scaled integers: Constants. (line 15)
7683 * scommon section: Input Section Common.
7685 * script files: Options. (line 532)
7686 * scripts: Scripts. (line 6)
7687 * search directory, from cmd line: Options. (line 368)
7688 * search path in linker script: File Commands. (line 74)
7689 * SEARCH_DIR(PATH): File Commands. (line 74)
7690 * SECT (MRI): MRI. (line 109)
7691 * section address: Output Section Address.
7693 * section address in expression: Builtin Functions. (line 17)
7694 * section alignment: Builtin Functions. (line 64)
7695 * section alignment, warnings on: Options. (line 1429)
7696 * section data: Output Section Data.
7698 * section fill pattern: Output Section Fill.
7700 * section load address: Output Section LMA. (line 6)
7701 * section load address in expression: Builtin Functions. (line 140)
7702 * section name: Output Section Name.
7704 * section name wildcard patterns: Input Section Wildcards.
7706 * section size: Builtin Functions. (line 167)
7707 * section, assigning to memory region: Output Section Region.
7709 * section, assigning to program header: Output Section Phdr.
7711 * SECTIONS: SECTIONS. (line 6)
7712 * sections, discarding: Output Section Discarding.
7714 * segment origins, cmd line: Options. (line 1280)
7715 * SEGMENT_START(SEGMENT, DEFAULT): Builtin Functions. (line 159)
7716 * segments, ELF: PHDRS. (line 6)
7717 * shared libraries: Options. (line 1208)
7718 * SHORT(EXPRESSION): Output Section Data.
7720 * SIZEOF(SECTION): Builtin Functions. (line 167)
7721 * SIZEOF_HEADERS: Builtin Functions. (line 183)
7722 * small common symbols: Input Section Common.
7724 * SORT: Input Section Wildcards.
7726 * SORT_BY_ALIGNMENT: Input Section Wildcards.
7728 * SORT_BY_NAME: Input Section Wildcards.
7730 * SPU: SPU ELF. (line 29)
7731 * SPU ELF options: SPU ELF. (line 6)
7732 * SPU extra overlay stubs: SPU ELF. (line 19)
7733 * SPU local store size: SPU ELF. (line 24)
7734 * SPU overlay stub symbols: SPU ELF. (line 15)
7735 * SPU overlays: SPU ELF. (line 9)
7736 * SPU plugins: SPU ELF. (line 6)
7737 * SQUAD(EXPRESSION): Output Section Data.
7739 * stack size: Options. (line 1921)
7740 * standard Unix system: Options. (line 7)
7741 * start of execution: Entry Point. (line 6)
7742 * STARTUP(FILENAME): File Commands. (line 82)
7743 * strip all symbols: Options. (line 519)
7744 * strip debugger symbols: Options. (line 523)
7745 * stripping all but some symbols: Options. (line 1115)
7746 * STUB_GROUP_SIZE: ARM. (line 129)
7747 * SUBALIGN(SUBSECTION_ALIGN): Forced Input Alignment.
7749 * suffixes for integers: Constants. (line 15)
7750 * symbol defaults: Builtin Functions. (line 120)
7751 * symbol definition, scripts: Assignments. (line 6)
7752 * symbol names: Symbols. (line 6)
7753 * symbol tracing: Options. (line 597)
7754 * symbol versions: VERSION. (line 6)
7755 * symbol-only input: Options. (line 508)
7756 * symbolic constants: Symbolic Constants. (line 6)
7757 * symbols, from command line: Options. (line 878)
7758 * symbols, relocatable and absolute: Expression Section. (line 6)
7759 * symbols, retaining selectively: Options. (line 1115)
7760 * synthesizing linker: Options. (line 1089)
7761 * synthesizing on H8/300: H8/300. (line 14)
7762 * TARGET(BFDNAME): Format Commands. (line 35)
7763 * TARGET1: ARM. (line 32)
7764 * TARGET2: ARM. (line 37)
7765 * text segment origin, cmd line: Options. (line 1287)
7766 * thumb entry point: ARM. (line 17)
7767 * TI COFF versions: TI COFF. (line 6)
7768 * traditional format: Options. (line 1259)
7769 * trampoline generation on M68HC11: M68HC11/68HC12. (line 31)
7770 * trampoline generation on M68HC12: M68HC11/68HC12. (line 31)
7771 * unallocated address, next: Builtin Functions. (line 150)
7772 * undefined symbol: Options. (line 554)
7773 * undefined symbol in linker script: Miscellaneous Commands.
7775 * undefined symbols, warnings on: Options. (line 1425)
7776 * uninitialized data placement: Input Section Common.
7778 * unspecified memory: Output Section Data.
7780 * usage: Options. (line 957)
7781 * USE_BLX: ARM. (line 74)
7782 * using a DEF file: WIN32. (line 57)
7783 * using auto-export functionality: WIN32. (line 22)
7784 * Using decorations: WIN32. (line 162)
7785 * variables, defining: Assignments. (line 6)
7786 * verbose[=NUMBER]: Options. (line 1319)
7787 * version: Options. (line 581)
7788 * version script: VERSION. (line 6)
7789 * version script, symbol versions: Options. (line 1327)
7790 * VERSION {script text}: VERSION. (line 6)
7791 * versions of symbols: VERSION. (line 6)
7792 * VFP11_DENORM_FIX: ARM. (line 83)
7793 * warnings, on combining symbols: Options. (line 1338)
7794 * warnings, on section alignment: Options. (line 1429)
7795 * warnings, on undefined symbols: Options. (line 1425)
7796 * weak externals: WIN32. (line 407)
7797 * what is this?: Overview. (line 6)
7798 * wildcard file name patterns: Input Section Wildcards.
7800 * Xtensa options: Xtensa. (line 56)
7801 * Xtensa processors: Xtensa. (line 6)
7807 Node: Overview
\x7f1600
7808 Node: Invocation
\x7f2714
7809 Node: Options
\x7f3122
7810 Node: Environment
\x7f93440
7811 Node: Scripts
\x7f95200
7812 Node: Basic Script Concepts
\x7f96934
7813 Node: Script Format
\x7f99641
7814 Node: Simple Example
\x7f100504
7815 Node: Simple Commands
\x7f103600
7816 Node: Entry Point
\x7f104106
7817 Node: File Commands
\x7f105039
7818 Node: Format Commands
\x7f109040
7819 Node: REGION_ALIAS
\x7f110996
7820 Node: Miscellaneous Commands
\x7f115828
7821 Node: Assignments
\x7f119436
7822 Node: Simple Assignments
\x7f119927
7823 Node: PROVIDE
\x7f121663
7824 Node: PROVIDE_HIDDEN
\x7f122868
7825 Node: Source Code Reference
\x7f123112
7826 Node: SECTIONS
\x7f126692
7827 Node: Output Section Description
\x7f128583
7828 Node: Output Section Name
\x7f129670
7829 Node: Output Section Address
\x7f130546
7830 Node: Input Section
\x7f132781
7831 Node: Input Section Basics
\x7f133582
7832 Node: Input Section Wildcards
\x7f136800
7833 Node: Input Section Common
\x7f141533
7834 Node: Input Section Keep
\x7f143015
7835 Node: Input Section Example
\x7f143505
7836 Node: Output Section Data
\x7f144473
7837 Node: Output Section Keywords
\x7f147250
7838 Node: Output Section Discarding
\x7f150819
7839 Node: Output Section Attributes
\x7f152000
7840 Node: Output Section Type
\x7f153101
7841 Node: Output Section LMA
\x7f154172
7842 Node: Forced Output Alignment
\x7f157243
7843 Node: Forced Input Alignment
\x7f157511
7844 Node: Output Section Constraint
\x7f157900
7845 Node: Output Section Region
\x7f158328
7846 Node: Output Section Phdr
\x7f158761
7847 Node: Output Section Fill
\x7f159425
7848 Node: Overlay Description
\x7f160567
7849 Node: MEMORY
\x7f164870
7850 Node: PHDRS
\x7f169205
7851 Node: VERSION
\x7f174459
7852 Node: Expressions
\x7f182552
7853 Node: Constants
\x7f183481
7854 Node: Symbolic Constants
\x7f184356
7855 Node: Symbols
\x7f184907
7856 Node: Orphan Sections
\x7f185654
7857 Node: Location Counter
\x7f186818
7858 Node: Operators
\x7f191254
7859 Node: Evaluation
\x7f192176
7860 Node: Expression Section
\x7f193540
7861 Node: Builtin Functions
\x7f197197
7862 Node: Implicit Linker Scripts
\x7f205158
7863 Node: Machine Dependent
\x7f205933
7864 Node: H8/300
\x7f206949
7865 Node: i960
\x7f208574
7866 Node: M68HC11/68HC12
\x7f210778
7868 Node: HPPA ELF32
\x7f219744
7869 Node: M68K
\x7f221367
7870 Node: MMIX
\x7f222276
7871 Node: MSP430
\x7f223441
7872 Node: PowerPC ELF32
\x7f224490
7873 Node: PowerPC64 ELF64
\x7f227326
7874 Node: SPU ELF
\x7f231742
7875 Node: TI COFF
\x7f234374
7876 Node: WIN32
\x7f234900
7877 Node: Xtensa
\x7f255025
7879 Node: BFD outline
\x7f259445
7880 Node: BFD information loss
\x7f260731
7881 Node: Canonical format
\x7f263248
7882 Node: Reporting Bugs
\x7f267605
7883 Node: Bug Criteria
\x7f268299
7884 Node: Bug Reporting
\x7f268998
7886 Node: GNU Free Documentation License
\x7f280680
7887 Node: LD Index
\x7f305836