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 2.25.
10 Copyright (C) 1991-2014 Free Software Foundation, Inc.
12 Permission is granted to copy, distribute and/or modify this document
13 under the terms of the GNU Free Documentation License, Version 1.3 or
14 any later version published by the Free Software Foundation; with no
15 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
16 Texts. A copy of the license is included in the section entitled "GNU
17 Free Documentation License".
20 File: ld.info, Node: Top, Next: Overview, Up: (dir)
25 This file documents the GNU linker ld (GNU Binutils) version 2.25.
27 This document is distributed under the terms of the GNU Free
28 Documentation License version 1.3. A copy of the license is included
29 in the section entitled "GNU Free Documentation License".
34 * Invocation:: Invocation
35 * Scripts:: Linker Scripts
37 * Machine Dependent:: Machine Dependent Features
41 * Reporting Bugs:: Reporting Bugs
42 * MRI:: MRI Compatible Script Files
43 * GNU Free Documentation License:: GNU Free Documentation License
47 File: ld.info, Node: Overview, Next: Invocation, Prev: Top, Up: Top
52 `ld' combines a number of object and archive files, relocates their
53 data and ties up symbol references. Usually the last step in compiling
54 a program is to run `ld'.
56 `ld' accepts Linker Command Language files written in a superset of
57 AT&T's Link Editor Command Language syntax, to provide explicit and
58 total control over the linking process.
60 This version of `ld' uses the general purpose BFD libraries to
61 operate on object files. This allows `ld' to read, combine, and write
62 object files in many different formats--for example, COFF or `a.out'.
63 Different formats may be linked together to produce any available kind
64 of object file. *Note BFD::, for more information.
66 Aside from its flexibility, the GNU linker is more helpful than other
67 linkers in providing diagnostic information. Many linkers abandon
68 execution immediately upon encountering an error; whenever possible,
69 `ld' continues executing, allowing you to identify other errors (or, in
70 some cases, to get an output file in spite of the error).
73 File: ld.info, Node: Invocation, Next: Scripts, Prev: Overview, Up: Top
78 The GNU linker `ld' is meant to cover a broad range of situations, and
79 to be as compatible as possible with other linkers. As a result, you
80 have many choices to control its behavior.
84 * Options:: Command Line Options
85 * Environment:: Environment Variables
88 File: ld.info, Node: Options, Next: Environment, Up: Invocation
90 2.1 Command Line Options
91 ========================
93 The linker supports a plethora of command-line options, but in actual
94 practice few of them are used in any particular context. For instance,
95 a frequent use of `ld' is to link standard Unix object files on a
96 standard, supported Unix system. On such a system, to link a file
99 ld -o OUTPUT /lib/crt0.o hello.o -lc
101 This tells `ld' to produce a file called OUTPUT as the result of
102 linking the file `/lib/crt0.o' with `hello.o' and the library `libc.a',
103 which will come from the standard search directories. (See the
104 discussion of the `-l' option below.)
106 Some of the command-line options to `ld' may be specified at any
107 point in the command line. However, options which refer to files, such
108 as `-l' or `-T', cause the file to be read at the point at which the
109 option appears in the command line, relative to the object files and
110 other file options. Repeating non-file options with a different
111 argument will either have no further effect, or override prior
112 occurrences (those further to the left on the command line) of that
113 option. Options which may be meaningfully specified more than once are
114 noted in the descriptions below.
116 Non-option arguments are object files or archives which are to be
117 linked together. They may follow, precede, or be mixed in with
118 command-line options, except that an object file argument may not be
119 placed between an option and its argument.
121 Usually the linker is invoked with at least one object file, but you
122 can specify other forms of binary input files using `-l', `-R', and the
123 script command language. If _no_ binary input files at all are
124 specified, the linker does not produce any output, and issues the
125 message `No input files'.
127 If the linker cannot recognize the format of an object file, it will
128 assume that it is a linker script. A script specified in this way
129 augments the main linker script used for the link (either the default
130 linker script or the one specified by using `-T'). This feature
131 permits the linker to link against a file which appears to be an object
132 or an archive, but actually merely defines some symbol values, or uses
133 `INPUT' or `GROUP' to load other objects. Specifying a script in this
134 way merely augments the main linker script, with the extra commands
135 placed after the main script; use the `-T' option to replace the
136 default linker script entirely, but note the effect of the `INSERT'
137 command. *Note Scripts::.
139 For options whose names are a single letter, option arguments must
140 either follow the option letter without intervening whitespace, or be
141 given as separate arguments immediately following the option that
144 For options whose names are multiple letters, either one dash or two
145 can precede the option name; for example, `-trace-symbol' and
146 `--trace-symbol' are equivalent. Note--there is one exception to this
147 rule. Multiple letter options that start with a lower case 'o' can
148 only be preceded by two dashes. This is to reduce confusion with the
149 `-o' option. So for example `-omagic' sets the output file name to
150 `magic' whereas `--omagic' sets the NMAGIC flag on the output.
152 Arguments to multiple-letter options must either be separated from
153 the option name by an equals sign, or be given as separate arguments
154 immediately following the option that requires them. For example,
155 `--trace-symbol foo' and `--trace-symbol=foo' are equivalent. Unique
156 abbreviations of the names of multiple-letter options are accepted.
158 Note--if the linker is being invoked indirectly, via a compiler
159 driver (e.g. `gcc') then all the linker command line options should be
160 prefixed by `-Wl,' (or whatever is appropriate for the particular
161 compiler driver) like this:
163 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
165 This is important, because otherwise the compiler driver program may
166 silently drop the linker options, resulting in a bad link. Confusion
167 may also arise when passing options that require values through a
168 driver, as the use of a space between option and argument acts as a
169 separator, and causes the driver to pass only the option to the linker
170 and the argument to the compiler. In this case, it is simplest to use
171 the joined forms of both single- and multiple-letter options, such as:
173 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
175 Here is a table of the generic command line switches accepted by the
179 Read command-line options from FILE. The options read are
180 inserted in place of the original @FILE option. If FILE does not
181 exist, or cannot be read, then the option will be treated
182 literally, and not removed.
184 Options in FILE are separated by whitespace. A whitespace
185 character may be included in an option by surrounding the entire
186 option in either single or double quotes. Any character
187 (including a backslash) may be included by prefixing the character
188 to be included with a backslash. The FILE may itself contain
189 additional @FILE options; any such options will be processed
193 This option is supported for HP/UX compatibility. The KEYWORD
194 argument must be one of the strings `archive', `shared', or
195 `default'. `-aarchive' is functionally equivalent to `-Bstatic',
196 and the other two keywords are functionally equivalent to
197 `-Bdynamic'. This option may be used any number of times.
200 Adds AUDITLIB to the `DT_AUDIT' entry of the dynamic section.
201 AUDITLIB is not checked for existence, nor will it use the
202 DT_SONAME specified in the library. If specified multiple times
203 `DT_AUDIT' will contain a colon separated list of audit interfaces
204 to use. If the linker finds an object with an audit entry while
205 searching for shared libraries, it will add a corresponding
206 `DT_DEPAUDIT' entry in the output file. This option is only
207 meaningful on ELF platforms supporting the rtld-audit interface.
210 `--architecture=ARCHITECTURE'
211 In the current release of `ld', this option is useful only for the
212 Intel 960 family of architectures. In that `ld' configuration, the
213 ARCHITECTURE argument identifies the particular architecture in
214 the 960 family, enabling some safeguards and modifying the
215 archive-library search path. *Note `ld' and the Intel 960 family:
218 Future releases of `ld' may support similar functionality for
219 other architecture families.
222 `--format=INPUT-FORMAT'
223 `ld' may be configured to support more than one kind of object
224 file. If your `ld' is configured this way, you can use the `-b'
225 option to specify the binary format for input object files that
226 follow this option on the command line. Even when `ld' is
227 configured to support alternative object formats, you don't
228 usually need to specify this, as `ld' should be configured to
229 expect as a default input format the most usual format on each
230 machine. INPUT-FORMAT is a text string, the name of a particular
231 format supported by the BFD libraries. (You can list the
232 available binary formats with `objdump -i'.) *Note BFD::.
234 You may want to use this option if you are linking files with an
235 unusual binary format. You can also use `-b' to switch formats
236 explicitly (when linking object files of different formats), by
237 including `-b INPUT-FORMAT' before each group of object files in a
240 The default format is taken from the environment variable
241 `GNUTARGET'. *Note Environment::. You can also define the input
242 format from a script, using the command `TARGET'; see *Note Format
246 `--mri-script=MRI-COMMANDFILE'
247 For compatibility with linkers produced by MRI, `ld' accepts script
248 files written in an alternate, restricted command language,
249 described in *Note MRI Compatible Script Files: MRI. Introduce
250 MRI script files with the option `-c'; use the `-T' option to run
251 linker scripts written in the general-purpose `ld' scripting
252 language. If MRI-CMDFILE does not exist, `ld' looks for it in the
253 directories specified by any `-L' options.
258 These three options are equivalent; multiple forms are supported
259 for compatibility with other linkers. They assign space to common
260 symbols even if a relocatable output file is specified (with
261 `-r'). The script command `FORCE_COMMON_ALLOCATION' has the same
262 effect. *Note Miscellaneous Commands::.
264 `--depaudit AUDITLIB'
266 Adds AUDITLIB to the `DT_DEPAUDIT' entry of the dynamic section.
267 AUDITLIB is not checked for existence, nor will it use the
268 DT_SONAME specified in the library. If specified multiple times
269 `DT_DEPAUDIT' will contain a colon separated list of audit
270 interfaces to use. This option is only meaningful on ELF
271 platforms supporting the rtld-audit interface. The -P option is
272 provided for Solaris compatibility.
276 Use ENTRY as the explicit symbol for beginning execution of your
277 program, rather than the default entry point. If there is no
278 symbol named ENTRY, the linker will try to parse ENTRY as a number,
279 and use that as the entry address (the number will be interpreted
280 in base 10; you may use a leading `0x' for base 16, or a leading
281 `0' for base 8). *Note Entry Point::, for a discussion of defaults
282 and other ways of specifying the entry point.
284 `--exclude-libs LIB,LIB,...'
285 Specifies a list of archive libraries from which symbols should
286 not be automatically exported. The library names may be delimited
287 by commas or colons. Specifying `--exclude-libs ALL' excludes
288 symbols in all archive libraries from automatic export. This
289 option is available only for the i386 PE targeted port of the
290 linker and for ELF targeted ports. For i386 PE, symbols
291 explicitly listed in a .def file are still exported, regardless of
292 this option. For ELF targeted ports, symbols affected by this
293 option will be treated as hidden.
295 `--exclude-modules-for-implib MODULE,MODULE,...'
296 Specifies a list of object files or archive members, from which
297 symbols should not be automatically exported, but which should be
298 copied wholesale into the import library being generated during
299 the link. The module names may be delimited by commas or colons,
300 and must match exactly the filenames used by `ld' to open the
301 files; for archive members, this is simply the member name, but
302 for object files the name listed must include and match precisely
303 any path used to specify the input file on the linker's
304 command-line. This option is available only for the i386 PE
305 targeted port of the linker. Symbols explicitly listed in a .def
306 file are still exported, regardless of this option.
310 `--no-export-dynamic'
311 When creating a dynamically linked executable, using the `-E'
312 option or the `--export-dynamic' option causes the linker to add
313 all symbols to the dynamic symbol table. The dynamic symbol table
314 is the set of symbols which are visible from dynamic objects at
317 If you do not use either of these options (or use the
318 `--no-export-dynamic' option to restore the default behavior), the
319 dynamic symbol table will normally contain only those symbols
320 which are referenced by some dynamic object mentioned in the link.
322 If you use `dlopen' to load a dynamic object which needs to refer
323 back to the symbols defined by the program, rather than some other
324 dynamic object, then you will probably need to use this option when
325 linking the program itself.
327 You can also use the dynamic list to control what symbols should
328 be added to the dynamic symbol table if the output format supports
329 it. See the description of `--dynamic-list'.
331 Note that this option is specific to ELF targeted ports. PE
332 targets support a similar function to export all symbols from a
333 DLL or EXE; see the description of `--export-all-symbols' below.
336 Link big-endian objects. This affects the default output format.
339 Link little-endian objects. This affects the default output
344 When creating an ELF shared object, set the internal DT_AUXILIARY
345 field to the specified name. This tells the dynamic linker that
346 the symbol table of the shared object should be used as an
347 auxiliary filter on the symbol table of the shared object NAME.
349 If you later link a program against this filter object, then, when
350 you run the program, the dynamic linker will see the DT_AUXILIARY
351 field. If the dynamic linker resolves any symbols from the filter
352 object, it will first check whether there is a definition in the
353 shared object NAME. If there is one, it will be used instead of
354 the definition in the filter object. The shared object NAME need
355 not exist. Thus the shared object NAME may be used to provide an
356 alternative implementation of certain functions, perhaps for
357 debugging or for machine specific performance.
359 This option may be specified more than once. The DT_AUXILIARY
360 entries will be created in the order in which they appear on the
365 When creating an ELF shared object, set the internal DT_FILTER
366 field to the specified name. This tells the dynamic linker that
367 the symbol table of the shared object which is being created
368 should be used as a filter on the symbol table of the shared
371 If you later link a program against this filter object, then, when
372 you run the program, the dynamic linker will see the DT_FILTER
373 field. The dynamic linker will resolve symbols according to the
374 symbol table of the filter object as usual, but it will actually
375 link to the definitions found in the shared object NAME. Thus the
376 filter object can be used to select a subset of the symbols
377 provided by the object NAME.
379 Some older linkers used the `-F' option throughout a compilation
380 toolchain for specifying object-file format for both input and
381 output object files. The GNU linker uses other mechanisms for
382 this purpose: the `-b', `--format', `--oformat' options, the
383 `TARGET' command in linker scripts, and the `GNUTARGET'
384 environment variable. The GNU linker will ignore the `-F' option
385 when not creating an ELF shared object.
388 When creating an ELF executable or shared object, call NAME when
389 the executable or shared object is unloaded, by setting DT_FINI to
390 the address of the function. By default, the linker uses `_fini'
391 as the function to call.
394 Ignored. Provided for compatibility with other tools.
398 Set the maximum size of objects to be optimized using the GP
399 register to SIZE. This is only meaningful for object file formats
400 such as MIPS ELF that support putting large and small objects into
401 different sections. This is ignored for other object file formats.
405 When creating an ELF shared object, set the internal DT_SONAME
406 field to the specified name. When an executable is linked with a
407 shared object which has a DT_SONAME field, then when the
408 executable is run the dynamic linker will attempt to load the
409 shared object specified by the DT_SONAME field rather than the
410 using the file name given to the linker.
413 Perform an incremental link (same as option `-r').
416 When creating an ELF executable or shared object, call NAME when
417 the executable or shared object is loaded, by setting DT_INIT to
418 the address of the function. By default, the linker uses `_init'
419 as the function to call.
423 Add the archive or object file specified by NAMESPEC to the list
424 of files to link. This option may be used any number of times.
425 If NAMESPEC is of the form `:FILENAME', `ld' will search the
426 library path for a file called FILENAME, otherwise it will search
427 the library path for a file called `libNAMESPEC.a'.
429 On systems which support shared libraries, `ld' may also search for
430 files other than `libNAMESPEC.a'. Specifically, on ELF and SunOS
431 systems, `ld' will search a directory for a library called
432 `libNAMESPEC.so' before searching for one called `libNAMESPEC.a'.
433 (By convention, a `.so' extension indicates a shared library.)
434 Note that this behavior does not apply to `:FILENAME', which
435 always specifies a file called FILENAME.
437 The linker will search an archive only once, at the location where
438 it is specified on the command line. If the archive defines a
439 symbol which was undefined in some object which appeared before
440 the archive on the command line, the linker will include the
441 appropriate file(s) from the archive. However, an undefined
442 symbol in an object appearing later on the command line will not
443 cause the linker to search the archive again.
445 See the `-(' option for a way to force the linker to search
446 archives multiple times.
448 You may list the same archive multiple times on the command line.
450 This type of archive searching is standard for Unix linkers.
451 However, if you are using `ld' on AIX, note that it is different
452 from the behaviour of the AIX linker.
455 `--library-path=SEARCHDIR'
456 Add path SEARCHDIR to the list of paths that `ld' will search for
457 archive libraries and `ld' control scripts. You may use this
458 option any number of times. The directories are searched in the
459 order in which they are specified on the command line.
460 Directories specified on the command line are searched before the
461 default directories. All `-L' options apply to all `-l' options,
462 regardless of the order in which the options appear. `-L' options
463 do not affect how `ld' searches for a linker script unless `-T'
466 If SEARCHDIR begins with `=', then the `=' will be replaced by the
467 "sysroot prefix", controlled by the `--sysroot' option, or
468 specified when the linker is configured.
470 The default set of paths searched (without being specified with
471 `-L') depends on which emulation mode `ld' is using, and in some
472 cases also on how it was configured. *Note Environment::.
474 The paths can also be specified in a link script with the
475 `SEARCH_DIR' command. Directories specified this way are searched
476 at the point in which the linker script appears in the command
480 Emulate the EMULATION linker. You can list the available
481 emulations with the `--verbose' or `-V' options.
483 If the `-m' option is not used, the emulation is taken from the
484 `LDEMULATION' environment variable, if that is defined.
486 Otherwise, the default emulation depends upon how the linker was
491 Print a link map to the standard output. A link map provides
492 information about the link, including the following:
494 * Where object files are mapped into memory.
496 * How common symbols are allocated.
498 * All archive members included in the link, with a mention of
499 the symbol which caused the archive member to be brought in.
501 * The values assigned to symbols.
503 Note - symbols whose values are computed by an expression
504 which involves a reference to a previous value of the same
505 symbol may not have correct result displayed in the link map.
506 This is because the linker discards intermediate results and
507 only retains the final value of an expression. Under such
508 circumstances the linker will display the final value
509 enclosed by square brackets. Thus for example a linker
516 will produce the following output in the link map if the `-M'
520 [0x0000000c] foo = (foo * 0x4)
521 [0x0000000c] foo = (foo + 0x8)
523 See *Note Expressions:: for more information about
524 expressions in linker scripts.
528 Turn off page alignment of sections, and disable linking against
529 shared libraries. If the output format supports Unix style magic
530 numbers, mark the output as `NMAGIC'.
534 Set the text and data sections to be readable and writable. Also,
535 do not page-align the data segment, and disable linking against
536 shared libraries. If the output format supports Unix style magic
537 numbers, mark the output as `OMAGIC'. Note: Although a writable
538 text section is allowed for PE-COFF targets, it does not conform
539 to the format specification published by Microsoft.
542 This option negates most of the effects of the `-N' option. It
543 sets the text section to be read-only, and forces the data segment
544 to be page-aligned. Note - this option does not enable linking
545 against shared libraries. Use `-Bdynamic' for this.
549 Use OUTPUT as the name for the program produced by `ld'; if this
550 option is not specified, the name `a.out' is used by default. The
551 script command `OUTPUT' can also specify the output file name.
554 If LEVEL is a numeric values greater than zero `ld' optimizes the
555 output. This might take significantly longer and therefore
556 probably should only be enabled for the final binary. At the
557 moment this option only affects ELF shared library generation.
558 Future releases of the linker may make more use of this option.
559 Also currently there is no difference in the linker's behaviour
560 for different non-zero values of this option. Again this may
561 change with future releases.
564 The `--push-state' allows to preserve the current state of the
565 flags which govern the input file handling so that they can all be
566 restored with one corresponding `--pop-state' option.
568 The option which are covered are: `-Bdynamic', `-Bstatic', `-dn',
569 `-dy', `-call_shared', `-non_shared', `-static', `-N', `-n',
570 `--whole-archive', `--no-whole-archive', `-r', `-Ur',
571 `--copy-dt-needed-entries', `--no-copy-dt-needed-entries',
572 `--as-needed', `--no-as-needed', and `-a'.
574 One target for this option are specifications for `pkg-config'.
575 When used with the `--libs' option all possibly needed libraries
576 are listed and then possibly linked with all the time. It is
577 better to return something as follows:
579 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
581 Undoes the effect of -push-state, restores the previous values of
582 the flags governing input file handling.
586 Leave relocation sections and contents in fully linked executables.
587 Post link analysis and optimization tools may need this
588 information in order to perform correct modifications of
589 executables. This results in larger executables.
591 This option is currently only supported on ELF platforms.
594 Force the output file to have dynamic sections. This option is
595 specific to VxWorks targets.
599 Generate relocatable output--i.e., generate an output file that
600 can in turn serve as input to `ld'. This is often called "partial
601 linking". As a side effect, in environments that support standard
602 Unix magic numbers, this option also sets the output file's magic
603 number to `OMAGIC'. If this option is not specified, an absolute
604 file is produced. When linking C++ programs, this option _will
605 not_ resolve references to constructors; to do that, use `-Ur'.
607 When an input file does not have the same format as the output
608 file, partial linking is only supported if that input file does
609 not contain any relocations. Different output formats can have
610 further restrictions; for example some `a.out'-based formats do
611 not support partial linking with input files in other formats at
614 This option does the same thing as `-i'.
617 `--just-symbols=FILENAME'
618 Read symbol names and their addresses from FILENAME, but do not
619 relocate it or include it in the output. This allows your output
620 file to refer symbolically to absolute locations of memory defined
621 in other programs. You may use this option more than once.
623 For compatibility with other ELF linkers, if the `-R' option is
624 followed by a directory name, rather than a file name, it is
625 treated as the `-rpath' option.
629 Omit all symbol information from the output file.
633 Omit debugger symbol information (but not all symbols) from the
638 Print the names of the input files as `ld' processes them.
641 `--script=SCRIPTFILE'
642 Use SCRIPTFILE as the linker script. This script replaces `ld''s
643 default linker script (rather than adding to it), so COMMANDFILE
644 must specify everything necessary to describe the output file.
645 *Note Scripts::. If SCRIPTFILE does not exist in the current
646 directory, `ld' looks for it in the directories specified by any
647 preceding `-L' options. Multiple `-T' options accumulate.
650 `--default-script=SCRIPTFILE'
651 Use SCRIPTFILE as the default linker script. *Note Scripts::.
653 This option is similar to the `--script' option except that
654 processing of the script is delayed until after the rest of the
655 command line has been processed. This allows options placed after
656 the `--default-script' option on the command line to affect the
657 behaviour of the linker script, which can be important when the
658 linker command line cannot be directly controlled by the user.
659 (eg because the command line is being constructed by another tool,
664 Force SYMBOL to be entered in the output file as an undefined
665 symbol. Doing this may, for example, trigger linking of additional
666 modules from standard libraries. `-u' may be repeated with
667 different option arguments to enter additional undefined symbols.
668 This option is equivalent to the `EXTERN' linker script command.
671 For anything other than C++ programs, this option is equivalent to
672 `-r': it generates relocatable output--i.e., an output file that
673 can in turn serve as input to `ld'. When linking C++ programs,
674 `-Ur' _does_ resolve references to constructors, unlike `-r'. It
675 does not work to use `-Ur' on files that were themselves linked
676 with `-Ur'; once the constructor table has been built, it cannot
677 be added to. Use `-Ur' only for the last partial link, and `-r'
681 Creates a separate output section for every input section matching
682 SECTION, or if the optional wildcard SECTION argument is missing,
683 for every orphan input section. An orphan section is one not
684 specifically mentioned in a linker script. You may use this option
685 multiple times on the command line; It prevents the normal
686 merging of input sections with the same name, overriding output
687 section assignments in a linker script.
692 Display the version number for `ld'. The `-V' option also lists
693 the supported emulations.
697 Delete all local symbols.
701 Delete all temporary local symbols. (These symbols start with
702 system-specific local label prefixes, typically `.L' for ELF
703 systems or `L' for traditional a.out systems.)
706 `--trace-symbol=SYMBOL'
707 Print the name of each linked file in which SYMBOL appears. This
708 option may be given any number of times. On many systems it is
709 necessary to prepend an underscore.
711 This option is useful when you have an undefined symbol in your
712 link but don't know where the reference is coming from.
715 Add PATH to the default library search path. This option exists
716 for Solaris compatibility.
719 The recognized keywords are:
721 Combines multiple reloc sections and sorts them to make
722 dynamic symbol lookup caching possible.
725 Disallows undefined symbols in object files. Undefined
726 symbols in shared libraries are still allowed.
729 Marks the object as requiring executable stack.
732 This option is only meaningful when building a shared object.
733 It makes the symbols defined by this shared object available
734 for symbol resolution of subsequently loaded libraries.
737 This option is only meaningful when building a shared object.
738 It marks the object so that its runtime initialization will
739 occur before the runtime initialization of any other objects
740 brought into the process at the same time. Similarly the
741 runtime finalization of the object will occur after the
742 runtime finalization of any other objects.
745 Marks the object that its symbol table interposes before all
746 symbols but the primary executable.
749 When generating an executable or shared library, mark it to
750 tell the dynamic linker to defer function call resolution to
751 the point when the function is called (lazy binding), rather
752 than at load time. Lazy binding is the default.
755 Marks the object that its filters be processed immediately at
759 Allows multiple definitions.
762 Disables multiple reloc sections combining.
765 Disables production of copy relocs.
768 Marks the object that the search for dependencies of this
769 object will ignore any default library search paths.
772 Marks the object shouldn't be unloaded at runtime.
775 Marks the object not available to `dlopen'.
778 Marks the object can not be dumped by `dldump'.
781 Marks the object as not requiring executable stack.
784 Don't create an ELF `PT_GNU_RELRO' segment header in the
788 When generating an executable or shared library, mark it to
789 tell the dynamic linker to resolve all symbols when the
790 program is started, or when the shared library is linked to
791 using dlopen, instead of deferring function call resolution
792 to the point when the function is first called.
795 Marks the object may contain $ORIGIN.
798 Create an ELF `PT_GNU_RELRO' segment header in the object.
800 `max-page-size=VALUE'
801 Set the emulation maximum page size to VALUE.
803 `common-page-size=VALUE'
804 Set the emulation common page size to VALUE.
807 Specify a stack size for in an ELF `PT_GNU_STACK' segment.
808 Specifying zero will override any default non-zero sized
809 `PT_GNU_STACK' segment creation.
812 Always generate BND prefix in PLT entries. Supported for
816 Other keywords are ignored for Solaris compatibility.
819 `--start-group ARCHIVES --end-group'
820 The ARCHIVES should be a list of archive files. They may be
821 either explicit file names, or `-l' options.
823 The specified archives are searched repeatedly until no new
824 undefined references are created. Normally, an archive is
825 searched only once in the order that it is specified on the
826 command line. If a symbol in that archive is needed to resolve an
827 undefined symbol referred to by an object in an archive that
828 appears later on the command line, the linker would not be able to
829 resolve that reference. By grouping the archives, they all be
830 searched repeatedly until all possible references are resolved.
832 Using this option has a significant performance cost. It is best
833 to use it only when there are unavoidable circular references
834 between two or more archives.
836 `--accept-unknown-input-arch'
837 `--no-accept-unknown-input-arch'
838 Tells the linker to accept input files whose architecture cannot be
839 recognised. The assumption is that the user knows what they are
840 doing and deliberately wants to link in these unknown input files.
841 This was the default behaviour of the linker, before release
842 2.14. The default behaviour from release 2.14 onwards is to
843 reject such input files, and so the `--accept-unknown-input-arch'
844 option has been added to restore the old behaviour.
848 This option affects ELF DT_NEEDED tags for dynamic libraries
849 mentioned on the command line after the `--as-needed' option.
850 Normally the linker will add a DT_NEEDED tag for each dynamic
851 library mentioned on the command line, regardless of whether the
852 library is actually needed or not. `--as-needed' causes a
853 DT_NEEDED tag to only be emitted for a library that _at that point
854 in the link_ satisfies a non-weak undefined symbol reference from
855 a regular object file or, if the library is not found in the
856 DT_NEEDED lists of other needed libraries, a non-weak undefined
857 symbol reference from another needed dynamic library. Object
858 files or libraries appearing on the command line _after_ the
859 library in question do not affect whether the library is seen as
860 needed. This is similar to the rules for extraction of object
861 files from archives. `--no-as-needed' restores the default
866 These two options have been deprecated because of the similarity of
867 their names to the `--as-needed' and `--no-as-needed' options.
868 They have been replaced by `--copy-dt-needed-entries' and
869 `--no-copy-dt-needed-entries'.
872 This option is ignored for SunOS compatibility.
877 Link against dynamic libraries. This is only meaningful on
878 platforms for which shared libraries are supported. This option
879 is normally the default on such platforms. The different variants
880 of this option are for compatibility with various systems. You
881 may use this option multiple times on the command line: it affects
882 library searching for `-l' options which follow it.
885 Set the `DF_1_GROUP' flag in the `DT_FLAGS_1' entry in the dynamic
886 section. This causes the runtime linker to handle lookups in this
887 object and its dependencies to be performed only inside the group.
888 `--unresolved-symbols=report-all' is implied. This option is only
889 meaningful on ELF platforms which support shared libraries.
895 Do not link against shared libraries. This is only meaningful on
896 platforms for which shared libraries are supported. The different
897 variants of this option are for compatibility with various
898 systems. You may use this option multiple times on the command
899 line: it affects library searching for `-l' options which follow
900 it. This option also implies `--unresolved-symbols=report-all'.
901 This option can be used with `-shared'. Doing so means that a
902 shared library is being created but that all of the library's
903 external references must be resolved by pulling in entries from
907 When creating a shared library, bind references to global symbols
908 to the definition within the shared library, if any. Normally, it
909 is possible for a program linked against a shared library to
910 override the definition within the shared library. This option is
911 only meaningful on ELF platforms which support shared libraries.
913 `-Bsymbolic-functions'
914 When creating a shared library, bind references to global function
915 symbols to the definition within the shared library, if any. This
916 option is only meaningful on ELF platforms which support shared
919 `--dynamic-list=DYNAMIC-LIST-FILE'
920 Specify the name of a dynamic list file to the linker. This is
921 typically used when creating shared libraries to specify a list of
922 global symbols whose references shouldn't be bound to the
923 definition within the shared library, or creating dynamically
924 linked executables to specify a list of symbols which should be
925 added to the symbol table in the executable. This option is only
926 meaningful on ELF platforms which support shared libraries.
928 The format of the dynamic list is the same as the version node
929 without scope and node name. See *Note VERSION:: for more
932 `--dynamic-list-data'
933 Include all global data symbols to the dynamic list.
935 `--dynamic-list-cpp-new'
936 Provide the builtin dynamic list for C++ operator new and delete.
937 It is mainly useful for building shared libstdc++.
939 `--dynamic-list-cpp-typeinfo'
940 Provide the builtin dynamic list for C++ runtime type
944 `--no-check-sections'
945 Asks the linker _not_ to check section addresses after they have
946 been assigned to see if there are any overlaps. Normally the
947 linker will perform this check, and if it finds any overlaps it
948 will produce suitable error messages. The linker does know about,
949 and does make allowances for sections in overlays. The default
950 behaviour can be restored by using the command line switch
951 `--check-sections'. Section overlap is not usually checked for
952 relocatable links. You can force checking in that case by using
953 the `--check-sections' option.
955 `--copy-dt-needed-entries'
956 `--no-copy-dt-needed-entries'
957 This option affects the treatment of dynamic libraries referred to
958 by DT_NEEDED tags _inside_ ELF dynamic libraries mentioned on the
959 command line. Normally the linker won't add a DT_NEEDED tag to the
960 output binary for each library mentioned in a DT_NEEDED tag in an
961 input dynamic library. With `--copy-dt-needed-entries' specified
962 on the command line however any dynamic libraries that follow it
963 will have their DT_NEEDED entries added. The default behaviour
964 can be restored with `--no-copy-dt-needed-entries'.
966 This option also has an effect on the resolution of symbols in
967 dynamic libraries. With `--copy-dt-needed-entries' dynamic
968 libraries mentioned on the command line will be recursively
969 searched, following their DT_NEEDED tags to other libraries, in
970 order to resolve symbols required by the output binary. With the
971 default setting however the searching of dynamic libraries that
972 follow it will stop with the dynamic library itself. No DT_NEEDED
973 links will be traversed to resolve symbols.
976 Output a cross reference table. If a linker map file is being
977 generated, the cross reference table is printed to the map file.
978 Otherwise, it is printed on the standard output.
980 The format of the table is intentionally simple, so that it may be
981 easily processed by a script if necessary. The symbols are
982 printed out, sorted by name. For each symbol, a list of file
983 names is given. If the symbol is defined, the first file listed
984 is the location of the definition. If the symbol is defined as a
985 common value then any files where this happens appear next.
986 Finally any files that reference the symbol are listed.
989 This option inhibits the assignment of addresses to common symbols.
990 The script command `INHIBIT_COMMON_ALLOCATION' has the same effect.
991 *Note Miscellaneous Commands::.
993 The `--no-define-common' option allows decoupling the decision to
994 assign addresses to Common symbols from the choice of the output
995 file type; otherwise a non-Relocatable output type forces
996 assigning addresses to Common symbols. Using `--no-define-common'
997 allows Common symbols that are referenced from a shared library to
998 be assigned addresses only in the main program. This eliminates
999 the unused duplicate space in the shared library, and also
1000 prevents any possible confusion over resolving to the wrong
1001 duplicate when there are many dynamic modules with specialized
1002 search paths for runtime symbol resolution.
1004 `--defsym=SYMBOL=EXPRESSION'
1005 Create a global symbol in the output file, containing the absolute
1006 address given by EXPRESSION. You may use this option as many
1007 times as necessary to define multiple symbols in the command line.
1008 A limited form of arithmetic is supported for the EXPRESSION in
1009 this context: you may give a hexadecimal constant or the name of
1010 an existing symbol, or use `+' and `-' to add or subtract
1011 hexadecimal constants or symbols. If you need more elaborate
1012 expressions, consider using the linker command language from a
1013 script (*note Assignment: Symbol Definitions: Assignments.).
1014 _Note:_ there should be no white space between SYMBOL, the equals
1015 sign ("<=>"), and EXPRESSION.
1017 `--demangle[=STYLE]'
1019 These options control whether to demangle symbol names in error
1020 messages and other output. When the linker is told to demangle,
1021 it tries to present symbol names in a readable fashion: it strips
1022 leading underscores if they are used by the object file format,
1023 and converts C++ mangled symbol names into user readable names.
1024 Different compilers have different mangling styles. The optional
1025 demangling style argument can be used to choose an appropriate
1026 demangling style for your compiler. The linker will demangle by
1027 default unless the environment variable `COLLECT_NO_DEMANGLE' is
1028 set. These options may be used to override the default.
1031 `--dynamic-linker=FILE'
1032 Set the name of the dynamic linker. This is only meaningful when
1033 generating dynamically linked ELF executables. The default dynamic
1034 linker is normally correct; don't use this unless you know what
1038 `--no-fatal-warnings'
1039 Treat all warnings as errors. The default behaviour can be
1040 restored with the option `--no-fatal-warnings'.
1042 `--force-exe-suffix'
1043 Make sure that an output file has a .exe suffix.
1045 If a successfully built fully linked output file does not have a
1046 `.exe' or `.dll' suffix, this option forces the linker to copy the
1047 output file to one of the same name with a `.exe' suffix. This
1048 option is useful when using unmodified Unix makefiles on a
1049 Microsoft Windows host, since some versions of Windows won't run
1050 an image unless it ends in a `.exe' suffix.
1054 Enable garbage collection of unused input sections. It is ignored
1055 on targets that do not support this option. The default behaviour
1056 (of not performing this garbage collection) can be restored by
1057 specifying `--no-gc-sections' on the command line.
1059 `--gc-sections' decides which input sections are used by examining
1060 symbols and relocations. The section containing the entry symbol
1061 and all sections containing symbols undefined on the command-line
1062 will be kept, as will sections containing symbols referenced by
1063 dynamic objects. Note that when building shared libraries, the
1064 linker must assume that any visible symbol is referenced. Once
1065 this initial set of sections has been determined, the linker
1066 recursively marks as used any section referenced by their
1067 relocations. See `--entry' and `--undefined'.
1069 This option can be set when doing a partial link (enabled with
1070 option `-r'). In this case the root of symbols kept must be
1071 explicitly specified either by an `--entry' or `--undefined'
1072 option or by a `ENTRY' command in the linker script.
1074 `--print-gc-sections'
1075 `--no-print-gc-sections'
1076 List all sections removed by garbage collection. The listing is
1077 printed on stderr. This option is only effective if garbage
1078 collection has been enabled via the `--gc-sections') option. The
1079 default behaviour (of not listing the sections that are removed)
1080 can be restored by specifying `--no-print-gc-sections' on the
1083 `--print-output-format'
1084 Print the name of the default output format (perhaps influenced by
1085 other command-line options). This is the string that would appear
1086 in an `OUTPUT_FORMAT' linker script command (*note File
1090 Print a summary of the command-line options on the standard output
1094 Print a summary of all target specific options on the standard
1098 Print a link map to the file MAPFILE. See the description of the
1102 `ld' normally optimizes for speed over memory usage by caching the
1103 symbol tables of input files in memory. This option tells `ld' to
1104 instead optimize for memory usage, by rereading the symbol tables
1105 as necessary. This may be required if `ld' runs out of memory
1106 space while linking a large executable.
1110 Report unresolved symbol references from regular object files.
1111 This is done even if the linker is creating a non-symbolic shared
1112 library. The switch `--[no-]allow-shlib-undefined' controls the
1113 behaviour for reporting unresolved references found in shared
1114 libraries being linked in.
1116 `--allow-multiple-definition'
1118 Normally when a symbol is defined multiple times, the linker will
1119 report a fatal error. These options allow multiple definitions and
1120 the first definition will be used.
1122 `--allow-shlib-undefined'
1123 `--no-allow-shlib-undefined'
1124 Allows or disallows undefined symbols in shared libraries. This
1125 switch is similar to `--no-undefined' except that it determines
1126 the behaviour when the undefined symbols are in a shared library
1127 rather than a regular object file. It does not affect how
1128 undefined symbols in regular object files are handled.
1130 The default behaviour is to report errors for any undefined symbols
1131 referenced in shared libraries if the linker is being used to
1132 create an executable, but to allow them if the linker is being
1133 used to create a shared library.
1135 The reasons for allowing undefined symbol references in shared
1136 libraries specified at link time are that:
1138 * A shared library specified at link time may not be the same
1139 as the one that is available at load time, so the symbol
1140 might actually be resolvable at load time.
1142 * There are some operating systems, eg BeOS and HPPA, where
1143 undefined symbols in shared libraries are normal.
1145 The BeOS kernel for example patches shared libraries at load
1146 time to select whichever function is most appropriate for the
1147 current architecture. This is used, for example, to
1148 dynamically select an appropriate memset function.
1150 `--no-undefined-version'
1151 Normally when a symbol has an undefined version, the linker will
1152 ignore it. This option disallows symbols with undefined version
1153 and a fatal error will be issued instead.
1156 Create and use a default symbol version (the soname) for
1157 unversioned exported symbols.
1159 `--default-imported-symver'
1160 Create and use a default symbol version (the soname) for
1161 unversioned imported symbols.
1163 `--no-warn-mismatch'
1164 Normally `ld' will give an error if you try to link together input
1165 files that are mismatched for some reason, perhaps because they
1166 have been compiled for different processors or for different
1167 endiannesses. This option tells `ld' that it should silently
1168 permit such possible errors. This option should only be used with
1169 care, in cases when you have taken some special action that
1170 ensures that the linker errors are inappropriate.
1172 `--no-warn-search-mismatch'
1173 Normally `ld' will give a warning if it finds an incompatible
1174 library during a library search. This option silences the warning.
1176 `--no-whole-archive'
1177 Turn off the effect of the `--whole-archive' option for subsequent
1181 Retain the executable output file whenever it is still usable.
1182 Normally, the linker will not produce an output file if it
1183 encounters errors during the link process; it exits without
1184 writing an output file when it issues any error whatsoever.
1187 Only search library directories explicitly specified on the
1188 command line. Library directories specified in linker scripts
1189 (including linker scripts specified on the command line) are
1192 `--oformat=OUTPUT-FORMAT'
1193 `ld' may be configured to support more than one kind of object
1194 file. If your `ld' is configured this way, you can use the
1195 `--oformat' option to specify the binary format for the output
1196 object file. Even when `ld' is configured to support alternative
1197 object formats, you don't usually need to specify this, as `ld'
1198 should be configured to produce as a default output format the most
1199 usual format on each machine. OUTPUT-FORMAT is a text string, the
1200 name of a particular format supported by the BFD libraries. (You
1201 can list the available binary formats with `objdump -i'.) The
1202 script command `OUTPUT_FORMAT' can also specify the output format,
1203 but this option overrides it. *Note BFD::.
1207 Create a position independent executable. This is currently only
1208 supported on ELF platforms. Position independent executables are
1209 similar to shared libraries in that they are relocated by the
1210 dynamic linker to the virtual address the OS chooses for them
1211 (which can vary between invocations). Like normal dynamically
1212 linked executables they can be executed and symbols defined in the
1213 executable cannot be overridden by shared libraries.
1216 This option is ignored for Linux compatibility.
1219 This option is ignored for SVR4 compatibility.
1223 An option with machine dependent effects. This option is only
1224 supported on a few targets. *Note `ld' and the H8/300: H8/300.
1225 *Note `ld' and the Intel 960 family: i960. *Note `ld' and Xtensa
1226 Processors: Xtensa. *Note `ld' and the 68HC11 and 68HC12:
1227 M68HC11/68HC12. *Note `ld' and the Altera Nios II: Nios II.
1228 *Note `ld' and PowerPC 32-bit ELF Support: PowerPC ELF32.
1230 On some platforms the `--relax' option performs target specific,
1231 global optimizations that become possible when the linker resolves
1232 addressing in the program, such as relaxing address modes,
1233 synthesizing new instructions, selecting shorter version of current
1234 instructions, and combining constant values.
1236 On some platforms these link time global optimizations may make
1237 symbolic debugging of the resulting executable impossible. This
1238 is known to be the case for the Matsushita MN10200 and MN10300
1239 family of processors.
1241 On platforms where this is not supported, `--relax' is accepted,
1244 On platforms where `--relax' is accepted the option `--no-relax'
1245 can be used to disable the feature.
1247 `--retain-symbols-file=FILENAME'
1248 Retain _only_ the symbols listed in the file FILENAME, discarding
1249 all others. FILENAME is simply a flat file, with one symbol name
1250 per line. This option is especially useful in environments (such
1251 as VxWorks) where a large global symbol table is accumulated
1252 gradually, to conserve run-time memory.
1254 `--retain-symbols-file' does _not_ discard undefined symbols, or
1255 symbols needed for relocations.
1257 You may only specify `--retain-symbols-file' once in the command
1258 line. It overrides `-s' and `-S'.
1261 Add a directory to the runtime library search path. This is used
1262 when linking an ELF executable with shared objects. All `-rpath'
1263 arguments are concatenated and passed to the runtime linker, which
1264 uses them to locate shared objects at runtime. The `-rpath'
1265 option is also used when locating shared objects which are needed
1266 by shared objects explicitly included in the link; see the
1267 description of the `-rpath-link' option. If `-rpath' is not used
1268 when linking an ELF executable, the contents of the environment
1269 variable `LD_RUN_PATH' will be used if it is defined.
1271 The `-rpath' option may also be used on SunOS. By default, on
1272 SunOS, the linker will form a runtime search patch out of all the
1273 `-L' options it is given. If a `-rpath' option is used, the
1274 runtime search path will be formed exclusively using the `-rpath'
1275 options, ignoring the `-L' options. This can be useful when using
1276 gcc, which adds many `-L' options which may be on NFS mounted file
1279 For compatibility with other ELF linkers, if the `-R' option is
1280 followed by a directory name, rather than a file name, it is
1281 treated as the `-rpath' option.
1284 When using ELF or SunOS, one shared library may require another.
1285 This happens when an `ld -shared' link includes a shared library
1286 as one of the input files.
1288 When the linker encounters such a dependency when doing a
1289 non-shared, non-relocatable link, it will automatically try to
1290 locate the required shared library and include it in the link, if
1291 it is not included explicitly. In such a case, the `-rpath-link'
1292 option specifies the first set of directories to search. The
1293 `-rpath-link' option may specify a sequence of directory names
1294 either by specifying a list of names separated by colons, or by
1295 appearing multiple times.
1297 This option should be used with caution as it overrides the search
1298 path that may have been hard compiled into a shared library. In
1299 such a case it is possible to use unintentionally a different
1300 search path than the runtime linker would do.
1302 The linker uses the following search paths to locate required
1304 1. Any directories specified by `-rpath-link' options.
1306 2. Any directories specified by `-rpath' options. The difference
1307 between `-rpath' and `-rpath-link' is that directories
1308 specified by `-rpath' options are included in the executable
1309 and used at runtime, whereas the `-rpath-link' option is only
1310 effective at link time. Searching `-rpath' in this way is
1311 only supported by native linkers and cross linkers which have
1312 been configured with the `--with-sysroot' option.
1314 3. On an ELF system, for native linkers, if the `-rpath' and
1315 `-rpath-link' options were not used, search the contents of
1316 the environment variable `LD_RUN_PATH'.
1318 4. On SunOS, if the `-rpath' option was not used, search any
1319 directories specified using `-L' options.
1321 5. For a native linker, search the contents of the environment
1322 variable `LD_LIBRARY_PATH'.
1324 6. For a native ELF linker, the directories in `DT_RUNPATH' or
1325 `DT_RPATH' of a shared library are searched for shared
1326 libraries needed by it. The `DT_RPATH' entries are ignored if
1327 `DT_RUNPATH' entries exist.
1329 7. The default directories, normally `/lib' and `/usr/lib'.
1331 8. For a native linker on an ELF system, if the file
1332 `/etc/ld.so.conf' exists, the list of directories found in
1335 If the required shared library is not found, the linker will issue
1336 a warning and continue with the link.
1340 Create a shared library. This is currently only supported on ELF,
1341 XCOFF and SunOS platforms. On SunOS, the linker will
1342 automatically create a shared library if the `-e' option is not
1343 used and there are undefined symbols in the link.
1346 `--sort-common=ascending'
1347 `--sort-common=descending'
1348 This option tells `ld' to sort the common symbols by alignment in
1349 ascending or descending order when it places them in the
1350 appropriate output sections. The symbol alignments considered are
1351 sixteen-byte or larger, eight-byte, four-byte, two-byte, and
1352 one-byte. This is to prevent gaps between symbols due to alignment
1353 constraints. If no sorting order is specified, then descending
1356 `--sort-section=name'
1357 This option will apply `SORT_BY_NAME' to all wildcard section
1358 patterns in the linker script.
1360 `--sort-section=alignment'
1361 This option will apply `SORT_BY_ALIGNMENT' to all wildcard section
1362 patterns in the linker script.
1364 `--split-by-file[=SIZE]'
1365 Similar to `--split-by-reloc' but creates a new output section for
1366 each input file when SIZE is reached. SIZE defaults to a size of
1369 `--split-by-reloc[=COUNT]'
1370 Tries to creates extra sections in the output file so that no
1371 single output section in the file contains more than COUNT
1372 relocations. This is useful when generating huge relocatable
1373 files for downloading into certain real time kernels with the COFF
1374 object file format; since COFF cannot represent more than 65535
1375 relocations in a single section. Note that this will fail to work
1376 with object file formats which do not support arbitrary sections.
1377 The linker will not split up individual input sections for
1378 redistribution, so if a single input section contains more than
1379 COUNT relocations one output section will contain that many
1380 relocations. COUNT defaults to a value of 32768.
1383 Compute and display statistics about the operation of the linker,
1384 such as execution time and memory usage.
1386 `--sysroot=DIRECTORY'
1387 Use DIRECTORY as the location of the sysroot, overriding the
1388 configure-time default. This option is only supported by linkers
1389 that were configured using `--with-sysroot'.
1391 `--traditional-format'
1392 For some targets, the output of `ld' is different in some ways from
1393 the output of some existing linker. This switch requests `ld' to
1394 use the traditional format instead.
1396 For example, on SunOS, `ld' combines duplicate entries in the
1397 symbol string table. This can reduce the size of an output file
1398 with full debugging information by over 30 percent.
1399 Unfortunately, the SunOS `dbx' program can not read the resulting
1400 program (`gdb' has no trouble). The `--traditional-format' switch
1401 tells `ld' to not combine duplicate entries.
1403 `--section-start=SECTIONNAME=ORG'
1404 Locate a section in the output file at the absolute address given
1405 by ORG. You may use this option as many times as necessary to
1406 locate multiple sections in the command line. ORG must be a
1407 single hexadecimal integer; for compatibility with other linkers,
1408 you may omit the leading `0x' usually associated with hexadecimal
1409 values. _Note:_ there should be no white space between
1410 SECTIONNAME, the equals sign ("<=>"), and ORG.
1415 Same as `--section-start', with `.bss', `.data' or `.text' as the
1418 `-Ttext-segment=ORG'
1419 When creating an ELF executable, it will set the address of the
1420 first byte of the text segment.
1422 `-Trodata-segment=ORG'
1423 When creating an ELF executable or shared object for a target where
1424 the read-only data is in its own segment separate from the
1425 executable text, it will set the address of the first byte of the
1426 read-only data segment.
1428 `-Tldata-segment=ORG'
1429 When creating an ELF executable or shared object for x86-64 medium
1430 memory model, it will set the address of the first byte of the
1433 `--unresolved-symbols=METHOD'
1434 Determine how to handle unresolved symbols. There are four
1435 possible values for `method':
1438 Do not report any unresolved symbols.
1441 Report all unresolved symbols. This is the default.
1443 `ignore-in-object-files'
1444 Report unresolved symbols that are contained in shared
1445 libraries, but ignore them if they come from regular object
1448 `ignore-in-shared-libs'
1449 Report unresolved symbols that come from regular object
1450 files, but ignore them if they come from shared libraries.
1451 This can be useful when creating a dynamic binary and it is
1452 known that all the shared libraries that it should be
1453 referencing are included on the linker's command line.
1455 The behaviour for shared libraries on their own can also be
1456 controlled by the `--[no-]allow-shlib-undefined' option.
1458 Normally the linker will generate an error message for each
1459 reported unresolved symbol but the option
1460 `--warn-unresolved-symbols' can change this to a warning.
1463 `--verbose[=NUMBER]'
1464 Display the version number for `ld' and list the linker emulations
1465 supported. Display which input files can and cannot be opened.
1466 Display the linker script being used by the linker. If the
1467 optional NUMBER argument > 1, plugin symbol status will also be
1470 `--version-script=VERSION-SCRIPTFILE'
1471 Specify the name of a version script to the linker. This is
1472 typically used when creating shared libraries to specify
1473 additional information about the version hierarchy for the library
1474 being created. This option is only fully supported on ELF
1475 platforms which support shared libraries; see *Note VERSION::. It
1476 is partially supported on PE platforms, which can use version
1477 scripts to filter symbol visibility in auto-export mode: any
1478 symbols marked `local' in the version script will not be exported.
1482 Warn when a common symbol is combined with another common symbol
1483 or with a symbol definition. Unix linkers allow this somewhat
1484 sloppy practice, but linkers on some other operating systems do
1485 not. This option allows you to find potential problems from
1486 combining global symbols. Unfortunately, some C libraries use
1487 this practice, so you may get some warnings about symbols in the
1488 libraries as well as in your programs.
1490 There are three kinds of global symbols, illustrated here by C
1494 A definition, which goes in the initialized data section of
1498 An undefined reference, which does not allocate space. There
1499 must be either a definition or a common symbol for the
1503 A common symbol. If there are only (one or more) common
1504 symbols for a variable, it goes in the uninitialized data
1505 area of the output file. The linker merges multiple common
1506 symbols for the same variable into a single symbol. If they
1507 are of different sizes, it picks the largest size. The
1508 linker turns a common symbol into a declaration, if there is
1509 a definition of the same variable.
1511 The `--warn-common' option can produce five kinds of warnings.
1512 Each warning consists of a pair of lines: the first describes the
1513 symbol just encountered, and the second describes the previous
1514 symbol encountered with the same name. One or both of the two
1515 symbols will be a common symbol.
1517 1. Turning a common symbol into a reference, because there is
1518 already a definition for the symbol.
1519 FILE(SECTION): warning: common of `SYMBOL'
1520 overridden by definition
1521 FILE(SECTION): warning: defined here
1523 2. Turning a common symbol into a reference, because a later
1524 definition for the symbol is encountered. This is the same
1525 as the previous case, except that the symbols are encountered
1526 in a different order.
1527 FILE(SECTION): warning: definition of `SYMBOL'
1529 FILE(SECTION): warning: common is here
1531 3. Merging a common symbol with a previous same-sized common
1533 FILE(SECTION): warning: multiple common
1535 FILE(SECTION): warning: previous common is here
1537 4. Merging a common symbol with a previous larger common symbol.
1538 FILE(SECTION): warning: common of `SYMBOL'
1539 overridden by larger common
1540 FILE(SECTION): warning: larger common is here
1542 5. Merging a common symbol with a previous smaller common
1543 symbol. This is the same as the previous case, except that
1544 the symbols are encountered in a different order.
1545 FILE(SECTION): warning: common of `SYMBOL'
1546 overriding smaller common
1547 FILE(SECTION): warning: smaller common is here
1549 `--warn-constructors'
1550 Warn if any global constructors are used. This is only useful for
1551 a few object file formats. For formats like COFF or ELF, the
1552 linker can not detect the use of global constructors.
1554 `--warn-multiple-gp'
1555 Warn if multiple global pointer values are required in the output
1556 file. This is only meaningful for certain processors, such as the
1557 Alpha. Specifically, some processors put large-valued constants
1558 in a special section. A special register (the global pointer)
1559 points into the middle of this section, so that constants can be
1560 loaded efficiently via a base-register relative addressing mode.
1561 Since the offset in base-register relative mode is fixed and
1562 relatively small (e.g., 16 bits), this limits the maximum size of
1563 the constant pool. Thus, in large programs, it is often necessary
1564 to use multiple global pointer values in order to be able to
1565 address all possible constants. This option causes a warning to
1566 be issued whenever this case occurs.
1569 Only warn once for each undefined symbol, rather than once per
1570 module which refers to it.
1572 `--warn-section-align'
1573 Warn if the address of an output section is changed because of
1574 alignment. Typically, the alignment will be set by an input
1575 section. The address will only be changed if it not explicitly
1576 specified; that is, if the `SECTIONS' command does not specify a
1577 start address for the section (*note SECTIONS::).
1579 `--warn-shared-textrel'
1580 Warn if the linker adds a DT_TEXTREL to a shared object.
1582 `--warn-alternate-em'
1583 Warn if an object has alternate ELF machine code.
1585 `--warn-unresolved-symbols'
1586 If the linker is going to report an unresolved symbol (see the
1587 option `--unresolved-symbols') it will normally generate an error.
1588 This option makes it generate a warning instead.
1590 `--error-unresolved-symbols'
1591 This restores the linker's default behaviour of generating errors
1592 when it is reporting unresolved symbols.
1595 For each archive mentioned on the command line after the
1596 `--whole-archive' option, include every object file in the archive
1597 in the link, rather than searching the archive for the required
1598 object files. This is normally used to turn an archive file into
1599 a shared library, forcing every object to be included in the
1600 resulting shared library. This option may be used more than once.
1602 Two notes when using this option from gcc: First, gcc doesn't know
1603 about this option, so you have to use `-Wl,-whole-archive'.
1604 Second, don't forget to use `-Wl,-no-whole-archive' after your
1605 list of archives, because gcc will add its own list of archives to
1606 your link and you may not want this flag to affect those as well.
1609 Use a wrapper function for SYMBOL. Any undefined reference to
1610 SYMBOL will be resolved to `__wrap_SYMBOL'. Any undefined
1611 reference to `__real_SYMBOL' will be resolved to SYMBOL.
1613 This can be used to provide a wrapper for a system function. The
1614 wrapper function should be called `__wrap_SYMBOL'. If it wishes
1615 to call the system function, it should call `__real_SYMBOL'.
1617 Here is a trivial example:
1620 __wrap_malloc (size_t c)
1622 printf ("malloc called with %zu\n", c);
1623 return __real_malloc (c);
1626 If you link other code with this file using `--wrap malloc', then
1627 all calls to `malloc' will call the function `__wrap_malloc'
1628 instead. The call to `__real_malloc' in `__wrap_malloc' will call
1629 the real `malloc' function.
1631 You may wish to provide a `__real_malloc' function as well, so that
1632 links without the `--wrap' option will succeed. If you do this,
1633 you should not put the definition of `__real_malloc' in the same
1634 file as `__wrap_malloc'; if you do, the assembler may resolve the
1635 call before the linker has a chance to wrap it to `malloc'.
1638 Request creation of `.eh_frame_hdr' section and ELF
1639 `PT_GNU_EH_FRAME' segment header.
1641 `--no-ld-generated-unwind-info'
1642 Request creation of `.eh_frame' unwind info for linker generated
1643 code sections like PLT. This option is on by default if linker
1644 generated unwind info is supported.
1646 `--enable-new-dtags'
1647 `--disable-new-dtags'
1648 This linker can create the new dynamic tags in ELF. But the older
1649 ELF systems may not understand them. If you specify
1650 `--enable-new-dtags', the new dynamic tags will be created as
1651 needed and older dynamic tags will be omitted. If you specify
1652 `--disable-new-dtags', no new dynamic tags will be created. By
1653 default, the new dynamic tags are not created. Note that those
1654 options are only available for ELF systems.
1656 `--hash-size=NUMBER'
1657 Set the default size of the linker's hash tables to a prime number
1658 close to NUMBER. Increasing this value can reduce the length of
1659 time it takes the linker to perform its tasks, at the expense of
1660 increasing the linker's memory requirements. Similarly reducing
1661 this value can reduce the memory requirements at the expense of
1664 `--hash-style=STYLE'
1665 Set the type of linker's hash table(s). STYLE can be either
1666 `sysv' for classic ELF `.hash' section, `gnu' for new style GNU
1667 `.gnu.hash' section or `both' for both the classic ELF `.hash' and
1668 new style GNU `.gnu.hash' hash tables. The default is `sysv'.
1670 `--reduce-memory-overheads'
1671 This option reduces memory requirements at ld runtime, at the
1672 expense of linking speed. This was introduced to select the old
1673 O(n^2) algorithm for link map file generation, rather than the new
1674 O(n) algorithm which uses about 40% more memory for symbol storage.
1676 Another effect of the switch is to set the default hash table size
1677 to 1021, which again saves memory at the cost of lengthening the
1678 linker's run time. This is not done however if the `--hash-size'
1679 switch has been used.
1681 The `--reduce-memory-overheads' switch may be also be used to
1682 enable other tradeoffs in future versions of the linker.
1686 Request the creation of a `.note.gnu.build-id' ELF note section or
1687 a `.build-id' COFF section. The contents of the note are unique
1688 bits identifying this linked file. STYLE can be `uuid' to use 128
1689 random bits, `sha1' to use a 160-bit SHA1 hash on the normative
1690 parts of the output contents, `md5' to use a 128-bit MD5 hash on
1691 the normative parts of the output contents, or `0xHEXSTRING' to
1692 use a chosen bit string specified as an even number of hexadecimal
1693 digits (`-' and `:' characters between digit pairs are ignored).
1694 If STYLE is omitted, `sha1' is used.
1696 The `md5' and `sha1' styles produces an identifier that is always
1697 the same in an identical output file, but will be unique among all
1698 nonidentical output files. It is not intended to be compared as a
1699 checksum for the file's contents. A linked file may be changed
1700 later by other tools, but the build ID bit string identifying the
1701 original linked file does not change.
1703 Passing `none' for STYLE disables the setting from any
1704 `--build-id' options earlier on the command line.
1706 2.1.1 Options Specific to i386 PE Targets
1707 -----------------------------------------
1709 The i386 PE linker supports the `-shared' option, which causes the
1710 output to be a dynamically linked library (DLL) instead of a normal
1711 executable. You should name the output `*.dll' when you use this
1712 option. In addition, the linker fully supports the standard `*.def'
1713 files, which may be specified on the linker command line like an object
1714 file (in fact, it should precede archives it exports symbols from, to
1715 ensure that they get linked in, just like a normal object file).
1717 In addition to the options common to all targets, the i386 PE linker
1718 support additional command line options that are specific to the i386
1719 PE target. Options that take values may be separated from their values
1720 by either a space or an equals sign.
1722 `--add-stdcall-alias'
1723 If given, symbols with a stdcall suffix (@NN) will be exported
1724 as-is and also with the suffix stripped. [This option is specific
1725 to the i386 PE targeted port of the linker]
1728 Use FILE as the name of a file in which to save the base addresses
1729 of all the relocations needed for generating DLLs with `dlltool'.
1730 [This is an i386 PE specific option]
1733 Create a DLL instead of a regular executable. You may also use
1734 `-shared' or specify a `LIBRARY' in a given `.def' file. [This
1735 option is specific to the i386 PE targeted port of the linker]
1737 `--enable-long-section-names'
1738 `--disable-long-section-names'
1739 The PE variants of the Coff object format add an extension that
1740 permits the use of section names longer than eight characters, the
1741 normal limit for Coff. By default, these names are only allowed
1742 in object files, as fully-linked executable images do not carry
1743 the Coff string table required to support the longer names. As a
1744 GNU extension, it is possible to allow their use in executable
1745 images as well, or to (probably pointlessly!) disallow it in
1746 object files, by using these two options. Executable images
1747 generated with these long section names are slightly non-standard,
1748 carrying as they do a string table, and may generate confusing
1749 output when examined with non-GNU PE-aware tools, such as file
1750 viewers and dumpers. However, GDB relies on the use of PE long
1751 section names to find Dwarf-2 debug information sections in an
1752 executable image at runtime, and so if neither option is specified
1753 on the command-line, `ld' will enable long section names,
1754 overriding the default and technically correct behaviour, when it
1755 finds the presence of debug information while linking an executable
1756 image and not stripping symbols. [This option is valid for all PE
1757 targeted ports of the linker]
1759 `--enable-stdcall-fixup'
1760 `--disable-stdcall-fixup'
1761 If the link finds a symbol that it cannot resolve, it will attempt
1762 to do "fuzzy linking" by looking for another defined symbol that
1763 differs only in the format of the symbol name (cdecl vs stdcall)
1764 and will resolve that symbol by linking to the match. For
1765 example, the undefined symbol `_foo' might be linked to the
1766 function `_foo@12', or the undefined symbol `_bar@16' might be
1767 linked to the function `_bar'. When the linker does this, it
1768 prints a warning, since it normally should have failed to link,
1769 but sometimes import libraries generated from third-party dlls may
1770 need this feature to be usable. If you specify
1771 `--enable-stdcall-fixup', this feature is fully enabled and
1772 warnings are not printed. If you specify
1773 `--disable-stdcall-fixup', this feature is disabled and such
1774 mismatches are considered to be errors. [This option is specific
1775 to the i386 PE targeted port of the linker]
1777 `--leading-underscore'
1778 `--no-leading-underscore'
1779 For most targets default symbol-prefix is an underscore and is
1780 defined in target's description. By this option it is possible to
1781 disable/enable the default underscore symbol-prefix.
1783 `--export-all-symbols'
1784 If given, all global symbols in the objects used to build a DLL
1785 will be exported by the DLL. Note that this is the default if
1786 there otherwise wouldn't be any exported symbols. When symbols are
1787 explicitly exported via DEF files or implicitly exported via
1788 function attributes, the default is to not export anything else
1789 unless this option is given. Note that the symbols `DllMain@12',
1790 `DllEntryPoint@0', `DllMainCRTStartup@12', and `impure_ptr' will
1791 not be automatically exported. Also, symbols imported from other
1792 DLLs will not be re-exported, nor will symbols specifying the
1793 DLL's internal layout such as those beginning with `_head_' or
1794 ending with `_iname'. In addition, no symbols from `libgcc',
1795 `libstd++', `libmingw32', or `crtX.o' will be exported. Symbols
1796 whose names begin with `__rtti_' or `__builtin_' will not be
1797 exported, to help with C++ DLLs. Finally, there is an extensive
1798 list of cygwin-private symbols that are not exported (obviously,
1799 this applies on when building DLLs for cygwin targets). These
1800 cygwin-excludes are: `_cygwin_dll_entry@12',
1801 `_cygwin_crt0_common@8', `_cygwin_noncygwin_dll_entry@12',
1802 `_fmode', `_impure_ptr', `cygwin_attach_dll', `cygwin_premain0',
1803 `cygwin_premain1', `cygwin_premain2', `cygwin_premain3', and
1804 `environ'. [This option is specific to the i386 PE targeted port
1807 `--exclude-symbols SYMBOL,SYMBOL,...'
1808 Specifies a list of symbols which should not be automatically
1809 exported. The symbol names may be delimited by commas or colons.
1810 [This option is specific to the i386 PE targeted port of the
1813 `--exclude-all-symbols'
1814 Specifies no symbols should be automatically exported. [This
1815 option is specific to the i386 PE targeted port of the linker]
1818 Specify the file alignment. Sections in the file will always
1819 begin at file offsets which are multiples of this number. This
1820 defaults to 512. [This option is specific to the i386 PE targeted
1824 `--heap RESERVE,COMMIT'
1825 Specify the number of bytes of memory to reserve (and optionally
1826 commit) to be used as heap for this program. The default is 1MB
1827 reserved, 4K committed. [This option is specific to the i386 PE
1828 targeted port of the linker]
1830 `--image-base VALUE'
1831 Use VALUE as the base address of your program or dll. This is the
1832 lowest memory location that will be used when your program or dll
1833 is loaded. To reduce the need to relocate and improve performance
1834 of your dlls, each should have a unique base address and not
1835 overlap any other dlls. The default is 0x400000 for executables,
1836 and 0x10000000 for dlls. [This option is specific to the i386 PE
1837 targeted port of the linker]
1840 If given, the stdcall suffixes (@NN) will be stripped from symbols
1841 before they are exported. [This option is specific to the i386 PE
1842 targeted port of the linker]
1844 `--large-address-aware'
1845 If given, the appropriate bit in the "Characteristics" field of
1846 the COFF header is set to indicate that this executable supports
1847 virtual addresses greater than 2 gigabytes. This should be used
1848 in conjunction with the /3GB or /USERVA=VALUE megabytes switch in
1849 the "[operating systems]" section of the BOOT.INI. Otherwise,
1850 this bit has no effect. [This option is specific to PE targeted
1851 ports of the linker]
1853 `--disable-large-address-aware'
1854 Reverts the effect of a previous `--large-address-aware' option.
1855 This is useful if `--large-address-aware' is always set by the
1856 compiler driver (e.g. Cygwin gcc) and the executable does not
1857 support virtual addresses greater than 2 gigabytes. [This option
1858 is specific to PE targeted ports of the linker]
1860 `--major-image-version VALUE'
1861 Sets the major number of the "image version". Defaults to 1.
1862 [This option is specific to the i386 PE targeted port of the
1865 `--major-os-version VALUE'
1866 Sets the major number of the "os version". Defaults to 4. [This
1867 option is specific to the i386 PE targeted port of the linker]
1869 `--major-subsystem-version VALUE'
1870 Sets the major number of the "subsystem version". Defaults to 4.
1871 [This option is specific to the i386 PE targeted port of the
1874 `--minor-image-version VALUE'
1875 Sets the minor number of the "image version". Defaults to 0.
1876 [This option is specific to the i386 PE targeted port of the
1879 `--minor-os-version VALUE'
1880 Sets the minor number of the "os version". Defaults to 0. [This
1881 option is specific to the i386 PE targeted port of the linker]
1883 `--minor-subsystem-version VALUE'
1884 Sets the minor number of the "subsystem version". Defaults to 0.
1885 [This option is specific to the i386 PE targeted port of the
1889 The linker will create the file FILE which will contain a DEF file
1890 corresponding to the DLL the linker is generating. This DEF file
1891 (which should be called `*.def') may be used to create an import
1892 library with `dlltool' or may be used as a reference to
1893 automatically or implicitly exported symbols. [This option is
1894 specific to the i386 PE targeted port of the linker]
1897 The linker will create the file FILE which will contain an import
1898 lib corresponding to the DLL the linker is generating. This import
1899 lib (which should be called `*.dll.a' or `*.a' may be used to link
1900 clients against the generated DLL; this behaviour makes it
1901 possible to skip a separate `dlltool' import library creation step.
1902 [This option is specific to the i386 PE targeted port of the
1905 `--enable-auto-image-base'
1906 `--enable-auto-image-base=VALUE'
1907 Automatically choose the image base for DLLs, optionally starting
1908 with base VALUE, unless one is specified using the `--image-base'
1909 argument. By using a hash generated from the dllname to create
1910 unique image bases for each DLL, in-memory collisions and
1911 relocations which can delay program execution are avoided. [This
1912 option is specific to the i386 PE targeted port of the linker]
1914 `--disable-auto-image-base'
1915 Do not automatically generate a unique image base. If there is no
1916 user-specified image base (`--image-base') then use the platform
1917 default. [This option is specific to the i386 PE targeted port of
1920 `--dll-search-prefix STRING'
1921 When linking dynamically to a dll without an import library,
1922 search for `<string><basename>.dll' in preference to
1923 `lib<basename>.dll'. This behaviour allows easy distinction
1924 between DLLs built for the various "subplatforms": native, cygwin,
1925 uwin, pw, etc. For instance, cygwin DLLs typically use
1926 `--dll-search-prefix=cyg'. [This option is specific to the i386
1927 PE targeted port of the linker]
1929 `--enable-auto-import'
1930 Do sophisticated linking of `_symbol' to `__imp__symbol' for DATA
1931 imports from DLLs, and create the necessary thunking symbols when
1932 building the import libraries with those DATA exports. Note: Use
1933 of the 'auto-import' extension will cause the text section of the
1934 image file to be made writable. This does not conform to the
1935 PE-COFF format specification published by Microsoft.
1937 Note - use of the 'auto-import' extension will also cause read only
1938 data which would normally be placed into the .rdata section to be
1939 placed into the .data section instead. This is in order to work
1940 around a problem with consts that is described here:
1941 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
1943 Using 'auto-import' generally will 'just work' - but sometimes you
1944 may see this message:
1946 "variable '<var>' can't be auto-imported. Please read the
1947 documentation for ld's `--enable-auto-import' for details."
1949 This message occurs when some (sub)expression accesses an address
1950 ultimately given by the sum of two constants (Win32 import tables
1951 only allow one). Instances where this may occur include accesses
1952 to member fields of struct variables imported from a DLL, as well
1953 as using a constant index into an array variable imported from a
1954 DLL. Any multiword variable (arrays, structs, long long, etc) may
1955 trigger this error condition. However, regardless of the exact
1956 data type of the offending exported variable, ld will always
1957 detect it, issue the warning, and exit.
1959 There are several ways to address this difficulty, regardless of
1960 the data type of the exported variable:
1962 One way is to use -enable-runtime-pseudo-reloc switch. This leaves
1963 the task of adjusting references in your client code for runtime
1964 environment, so this method works only when runtime environment
1965 supports this feature.
1967 A second solution is to force one of the 'constants' to be a
1968 variable - that is, unknown and un-optimizable at compile time.
1969 For arrays, there are two possibilities: a) make the indexee (the
1970 array's address) a variable, or b) make the 'constant' index a
1973 extern type extern_array[];
1975 { volatile type *t=extern_array; t[1] }
1979 extern type extern_array[];
1981 { volatile int t=1; extern_array[t] }
1983 For structs (and most other multiword data types) the only option
1984 is to make the struct itself (or the long long, or the ...)
1987 extern struct s extern_struct;
1988 extern_struct.field -->
1989 { volatile struct s *t=&extern_struct; t->field }
1993 extern long long extern_ll;
1995 { volatile long long * local_ll=&extern_ll; *local_ll }
1997 A third method of dealing with this difficulty is to abandon
1998 'auto-import' for the offending symbol and mark it with
1999 `__declspec(dllimport)'. However, in practice that requires using
2000 compile-time #defines to indicate whether you are building a DLL,
2001 building client code that will link to the DLL, or merely
2002 building/linking to a static library. In making the choice
2003 between the various methods of resolving the 'direct address with
2004 constant offset' problem, you should consider typical real-world
2012 void main(int argc, char **argv){
2013 printf("%d\n",arr[1]);
2021 void main(int argc, char **argv){
2022 /* This workaround is for win32 and cygwin; do not "optimize" */
2023 volatile int *parr = arr;
2024 printf("%d\n",parr[1]);
2029 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2030 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2031 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2032 #define FOO_IMPORT __declspec(dllimport)
2036 extern FOO_IMPORT int arr[];
2039 void main(int argc, char **argv){
2040 printf("%d\n",arr[1]);
2043 A fourth way to avoid this problem is to re-code your library to
2044 use a functional interface rather than a data interface for the
2045 offending variables (e.g. set_foo() and get_foo() accessor
2046 functions). [This option is specific to the i386 PE targeted port
2049 `--disable-auto-import'
2050 Do not attempt to do sophisticated linking of `_symbol' to
2051 `__imp__symbol' for DATA imports from DLLs. [This option is
2052 specific to the i386 PE targeted port of the linker]
2054 `--enable-runtime-pseudo-reloc'
2055 If your code contains expressions described in -enable-auto-import
2056 section, that is, DATA imports from DLL with non-zero offset, this
2057 switch will create a vector of 'runtime pseudo relocations' which
2058 can be used by runtime environment to adjust references to such
2059 data in your client code. [This option is specific to the i386 PE
2060 targeted port of the linker]
2062 `--disable-runtime-pseudo-reloc'
2063 Do not create pseudo relocations for non-zero offset DATA imports
2064 from DLLs. [This option is specific to the i386 PE targeted port
2067 `--enable-extra-pe-debug'
2068 Show additional debug info related to auto-import symbol thunking.
2069 [This option is specific to the i386 PE targeted port of the
2072 `--section-alignment'
2073 Sets the section alignment. Sections in memory will always begin
2074 at addresses which are a multiple of this number. Defaults to
2075 0x1000. [This option is specific to the i386 PE targeted port of
2079 `--stack RESERVE,COMMIT'
2080 Specify the number of bytes of memory to reserve (and optionally
2081 commit) to be used as stack for this program. The default is 2MB
2082 reserved, 4K committed. [This option is specific to the i386 PE
2083 targeted port of the linker]
2086 `--subsystem WHICH:MAJOR'
2087 `--subsystem WHICH:MAJOR.MINOR'
2088 Specifies the subsystem under which your program will execute. The
2089 legal values for WHICH are `native', `windows', `console',
2090 `posix', and `xbox'. You may optionally set the subsystem version
2091 also. Numeric values are also accepted for WHICH. [This option
2092 is specific to the i386 PE targeted port of the linker]
2094 The following options set flags in the `DllCharacteristics' field
2095 of the PE file header: [These options are specific to PE targeted
2096 ports of the linker]
2099 Image is compatible with 64-bit address space layout randomization
2103 The image base address may be relocated using address space layout
2104 randomization (ASLR). This feature was introduced with MS Windows
2105 Vista for i386 PE targets.
2108 Code integrity checks are enforced.
2111 The image is compatible with the Data Execution Prevention. This
2112 feature was introduced with MS Windows XP SP2 for i386 PE targets.
2115 Although the image understands isolation, do not isolate the image.
2118 The image does not use SEH. No SE handler may be called from this
2122 Do not bind this image.
2125 The driver uses the MS Windows Driver Model.
2128 The image is Terminal Server aware.
2130 `--insert-timestamp'
2131 `--no-insert-timestamp'
2132 Insert a real timestamp into the image. This is the default
2133 behaviour as it matches legacy code and it means that the image
2134 will work with other, proprietary tools. The problem with this
2135 default is that it will result in slightly different images being
2136 produced each tiem the same sources are linked. The option
2137 `--no-insert-timestamp' can be used to insert a zero value for the
2138 timestamp, this ensuring that binaries produced from indentical
2139 sources will compare identically.
2141 2.1.2 Options specific to C6X uClinux targets
2142 ---------------------------------------------
2144 The C6X uClinux target uses a binary format called DSBT to support
2145 shared libraries. Each shared library in the system needs to have a
2146 unique index; all executables use an index of 0.
2149 This option sets the number of entires in the DSBT of the current
2150 executable or shared library to SIZE. The default is to create a
2151 table with 64 entries.
2153 `--dsbt-index INDEX'
2154 This option sets the DSBT index of the current executable or
2155 shared library to INDEX. The default is 0, which is appropriate
2156 for generating executables. If a shared library is generated with
2157 a DSBT index of 0, the `R_C6000_DSBT_INDEX' relocs are copied into
2160 The `--no-merge-exidx-entries' switch disables the merging of
2161 adjacent exidx entries in frame unwind info.
2164 2.1.3 Options specific to Motorola 68HC11 and 68HC12 targets
2165 ------------------------------------------------------------
2167 The 68HC11 and 68HC12 linkers support specific options to control the
2168 memory bank switching mapping and trampoline code generation.
2171 This option disables the generation of trampoline. By default a
2172 trampoline is generated for each far function which is called
2173 using a `jsr' instruction (this happens when a pointer to a far
2176 `--bank-window NAME'
2177 This option indicates to the linker the name of the memory region
2178 in the `MEMORY' specification that describes the memory bank
2179 window. The definition of such region is then used by the linker
2180 to compute paging and addresses within the memory window.
2183 2.1.4 Options specific to Motorola 68K target
2184 ---------------------------------------------
2186 The following options are supported to control handling of GOT
2187 generation when linking for 68K targets.
2190 This option tells the linker which GOT generation scheme to use.
2191 TYPE should be one of `single', `negative', `multigot' or
2192 `target'. For more information refer to the Info entry for `ld'.
2195 2.1.5 Options specific to MIPS targets
2196 --------------------------------------
2198 The following options are supported to control microMIPS instruction
2199 generation when linking for MIPS targets.
2203 These options control the choice of microMIPS instructions used in
2204 code generated by the linker, such as that in the PLT or lazy
2205 binding stubs, or in relaxation. If `--insn32' is used, then the
2206 linker only uses 32-bit instruction encodings. By default or if
2207 `--no-insn32' is used, all instruction encodings are used,
2208 including 16-bit ones where possible.
2212 File: ld.info, Node: Environment, Prev: Options, Up: Invocation
2214 2.2 Environment Variables
2215 =========================
2217 You can change the behaviour of `ld' with the environment variables
2218 `GNUTARGET', `LDEMULATION' and `COLLECT_NO_DEMANGLE'.
2220 `GNUTARGET' determines the input-file object format if you don't use
2221 `-b' (or its synonym `--format'). Its value should be one of the BFD
2222 names for an input format (*note BFD::). If there is no `GNUTARGET' in
2223 the environment, `ld' uses the natural format of the target. If
2224 `GNUTARGET' is set to `default' then BFD attempts to discover the input
2225 format by examining binary input files; this method often succeeds, but
2226 there are potential ambiguities, since there is no method of ensuring
2227 that the magic number used to specify object-file formats is unique.
2228 However, the configuration procedure for BFD on each system places the
2229 conventional format for that system first in the search-list, so
2230 ambiguities are resolved in favor of convention.
2232 `LDEMULATION' determines the default emulation if you don't use the
2233 `-m' option. The emulation can affect various aspects of linker
2234 behaviour, particularly the default linker script. You can list the
2235 available emulations with the `--verbose' or `-V' options. If the `-m'
2236 option is not used, and the `LDEMULATION' environment variable is not
2237 defined, the default emulation depends upon how the linker was
2240 Normally, the linker will default to demangling symbols. However, if
2241 `COLLECT_NO_DEMANGLE' is set in the environment, then it will default
2242 to not demangling symbols. This environment variable is used in a
2243 similar fashion by the `gcc' linker wrapper program. The default may
2244 be overridden by the `--demangle' and `--no-demangle' options.
2247 File: ld.info, Node: Scripts, Next: Machine Dependent, Prev: Invocation, Up: Top
2252 Every link is controlled by a "linker script". This script is written
2253 in the linker command language.
2255 The main purpose of the linker script is to describe how the
2256 sections in the input files should be mapped into the output file, and
2257 to control the memory layout of the output file. Most linker scripts
2258 do nothing more than this. However, when necessary, the linker script
2259 can also direct the linker to perform many other operations, using the
2260 commands described below.
2262 The linker always uses a linker script. If you do not supply one
2263 yourself, the linker will use a default script that is compiled into the
2264 linker executable. You can use the `--verbose' command line option to
2265 display the default linker script. Certain command line options, such
2266 as `-r' or `-N', will affect the default linker script.
2268 You may supply your own linker script by using the `-T' command line
2269 option. When you do this, your linker script will replace the default
2272 You may also use linker scripts implicitly by naming them as input
2273 files to the linker, as though they were files to be linked. *Note
2274 Implicit Linker Scripts::.
2278 * Basic Script Concepts:: Basic Linker Script Concepts
2279 * Script Format:: Linker Script Format
2280 * Simple Example:: Simple Linker Script Example
2281 * Simple Commands:: Simple Linker Script Commands
2282 * Assignments:: Assigning Values to Symbols
2283 * SECTIONS:: SECTIONS Command
2284 * MEMORY:: MEMORY Command
2285 * PHDRS:: PHDRS Command
2286 * VERSION:: VERSION Command
2287 * Expressions:: Expressions in Linker Scripts
2288 * Implicit Linker Scripts:: Implicit Linker Scripts
2291 File: ld.info, Node: Basic Script Concepts, Next: Script Format, Up: Scripts
2293 3.1 Basic Linker Script Concepts
2294 ================================
2296 We need to define some basic concepts and vocabulary in order to
2297 describe the linker script language.
2299 The linker combines input files into a single output file. The
2300 output file and each input file are in a special data format known as an
2301 "object file format". Each file is called an "object file". The
2302 output file is often called an "executable", but for our purposes we
2303 will also call it an object file. Each object file has, among other
2304 things, a list of "sections". We sometimes refer to a section in an
2305 input file as an "input section"; similarly, a section in the output
2306 file is an "output section".
2308 Each section in an object file has a name and a size. Most sections
2309 also have an associated block of data, known as the "section contents".
2310 A section may be marked as "loadable", which means that the contents
2311 should be loaded into memory when the output file is run. A section
2312 with no contents may be "allocatable", which means that an area in
2313 memory should be set aside, but nothing in particular should be loaded
2314 there (in some cases this memory must be zeroed out). A section which
2315 is neither loadable nor allocatable typically contains some sort of
2316 debugging information.
2318 Every loadable or allocatable output section has two addresses. The
2319 first is the "VMA", or virtual memory address. This is the address the
2320 section will have when the output file is run. The second is the
2321 "LMA", or load memory address. This is the address at which the
2322 section will be loaded. In most cases the two addresses will be the
2323 same. An example of when they might be different is when a data section
2324 is loaded into ROM, and then copied into RAM when the program starts up
2325 (this technique is often used to initialize global variables in a ROM
2326 based system). In this case the ROM address would be the LMA, and the
2327 RAM address would be the VMA.
2329 You can see the sections in an object file by using the `objdump'
2330 program with the `-h' option.
2332 Every object file also has a list of "symbols", known as the "symbol
2333 table". A symbol may be defined or undefined. Each symbol has a name,
2334 and each defined symbol has an address, among other information. If
2335 you compile a C or C++ program into an object file, you will get a
2336 defined symbol for every defined function and global or static
2337 variable. Every undefined function or global variable which is
2338 referenced in the input file will become an undefined symbol.
2340 You can see the symbols in an object file by using the `nm' program,
2341 or by using the `objdump' program with the `-t' option.
2344 File: ld.info, Node: Script Format, Next: Simple Example, Prev: Basic Script Concepts, Up: Scripts
2346 3.2 Linker Script Format
2347 ========================
2349 Linker scripts are text files.
2351 You write a linker script as a series of commands. Each command is
2352 either a keyword, possibly followed by arguments, or an assignment to a
2353 symbol. You may separate commands using semicolons. Whitespace is
2356 Strings such as file or format names can normally be entered
2357 directly. If the file name contains a character such as a comma which
2358 would otherwise serve to separate file names, you may put the file name
2359 in double quotes. There is no way to use a double quote character in a
2362 You may include comments in linker scripts just as in C, delimited by
2363 `/*' and `*/'. As in C, comments are syntactically equivalent to
2367 File: ld.info, Node: Simple Example, Next: Simple Commands, Prev: Script Format, Up: Scripts
2369 3.3 Simple Linker Script Example
2370 ================================
2372 Many linker scripts are fairly simple.
2374 The simplest possible linker script has just one command:
2375 `SECTIONS'. You use the `SECTIONS' command to describe the memory
2376 layout of the output file.
2378 The `SECTIONS' command is a powerful command. Here we will describe
2379 a simple use of it. Let's assume your program consists only of code,
2380 initialized data, and uninitialized data. These will be in the
2381 `.text', `.data', and `.bss' sections, respectively. Let's assume
2382 further that these are the only sections which appear in your input
2385 For this example, let's say that the code should be loaded at address
2386 0x10000, and that the data should start at address 0x8000000. Here is a
2387 linker script which will do that:
2391 .text : { *(.text) }
2393 .data : { *(.data) }
2397 You write the `SECTIONS' command as the keyword `SECTIONS', followed
2398 by a series of symbol assignments and output section descriptions
2399 enclosed in curly braces.
2401 The first line inside the `SECTIONS' command of the above example
2402 sets the value of the special symbol `.', which is the location
2403 counter. If you do not specify the address of an output section in some
2404 other way (other ways are described later), the address is set from the
2405 current value of the location counter. The location counter is then
2406 incremented by the size of the output section. At the start of the
2407 `SECTIONS' command, the location counter has the value `0'.
2409 The second line defines an output section, `.text'. The colon is
2410 required syntax which may be ignored for now. Within the curly braces
2411 after the output section name, you list the names of the input sections
2412 which should be placed into this output section. The `*' is a wildcard
2413 which matches any file name. The expression `*(.text)' means all
2414 `.text' input sections in all input files.
2416 Since the location counter is `0x10000' when the output section
2417 `.text' is defined, the linker will set the address of the `.text'
2418 section in the output file to be `0x10000'.
2420 The remaining lines define the `.data' and `.bss' sections in the
2421 output file. The linker will place the `.data' output section at
2422 address `0x8000000'. After the linker places the `.data' output
2423 section, the value of the location counter will be `0x8000000' plus the
2424 size of the `.data' output section. The effect is that the linker will
2425 place the `.bss' output section immediately after the `.data' output
2428 The linker will ensure that each output section has the required
2429 alignment, by increasing the location counter if necessary. In this
2430 example, the specified addresses for the `.text' and `.data' sections
2431 will probably satisfy any alignment constraints, but the linker may
2432 have to create a small gap between the `.data' and `.bss' sections.
2434 That's it! That's a simple and complete linker script.
2437 File: ld.info, Node: Simple Commands, Next: Assignments, Prev: Simple Example, Up: Scripts
2439 3.4 Simple Linker Script Commands
2440 =================================
2442 In this section we describe the simple linker script commands.
2446 * Entry Point:: Setting the entry point
2447 * File Commands:: Commands dealing with files
2449 * Format Commands:: Commands dealing with object file formats
2451 * REGION_ALIAS:: Assign alias names to memory regions
2452 * Miscellaneous Commands:: Other linker script commands
2455 File: ld.info, Node: Entry Point, Next: File Commands, Up: Simple Commands
2457 3.4.1 Setting the Entry Point
2458 -----------------------------
2460 The first instruction to execute in a program is called the "entry
2461 point". You can use the `ENTRY' linker script command to set the entry
2462 point. The argument is a symbol name:
2465 There are several ways to set the entry point. The linker will set
2466 the entry point by trying each of the following methods in order, and
2467 stopping when one of them succeeds:
2468 * the `-e' ENTRY command-line option;
2470 * the `ENTRY(SYMBOL)' command in a linker script;
2472 * the value of a target specific symbol, if it is defined; For many
2473 targets this is `start', but PE and BeOS based systems for example
2474 check a list of possible entry symbols, matching the first one
2477 * the address of the first byte of the `.text' section, if present;
2482 File: ld.info, Node: File Commands, Next: Format Commands, Prev: Entry Point, Up: Simple Commands
2484 3.4.2 Commands Dealing with Files
2485 ---------------------------------
2487 Several linker script commands deal with files.
2490 Include the linker script FILENAME at this point. The file will
2491 be searched for in the current directory, and in any directory
2492 specified with the `-L' option. You can nest calls to `INCLUDE'
2493 up to 10 levels deep.
2495 You can place `INCLUDE' directives at the top level, in `MEMORY' or
2496 `SECTIONS' commands, or in output section descriptions.
2498 `INPUT(FILE, FILE, ...)'
2499 `INPUT(FILE FILE ...)'
2500 The `INPUT' command directs the linker to include the named files
2501 in the link, as though they were named on the command line.
2503 For example, if you always want to include `subr.o' any time you do
2504 a link, but you can't be bothered to put it on every link command
2505 line, then you can put `INPUT (subr.o)' in your linker script.
2507 In fact, if you like, you can list all of your input files in the
2508 linker script, and then invoke the linker with nothing but a `-T'
2511 In case a "sysroot prefix" is configured, and the filename starts
2512 with the `/' character, and the script being processed was located
2513 inside the "sysroot prefix", the filename will be looked for in
2514 the "sysroot prefix". Otherwise, the linker will try to open the
2515 file in the current directory. If it is not found, the linker
2516 will search through the archive library search path. The "sysroot
2517 prefix" can also be forced by specifying `=' as the first
2518 character in the filename path. See also the description of `-L'
2519 in *Note Command Line Options: Options.
2521 If you use `INPUT (-lFILE)', `ld' will transform the name to
2522 `libFILE.a', as with the command line argument `-l'.
2524 When you use the `INPUT' command in an implicit linker script, the
2525 files will be included in the link at the point at which the linker
2526 script file is included. This can affect archive searching.
2528 `GROUP(FILE, FILE, ...)'
2529 `GROUP(FILE FILE ...)'
2530 The `GROUP' command is like `INPUT', except that the named files
2531 should all be archives, and they are searched repeatedly until no
2532 new undefined references are created. See the description of `-('
2533 in *Note Command Line Options: Options.
2535 `AS_NEEDED(FILE, FILE, ...)'
2536 `AS_NEEDED(FILE FILE ...)'
2537 This construct can appear only inside of the `INPUT' or `GROUP'
2538 commands, among other filenames. The files listed will be handled
2539 as if they appear directly in the `INPUT' or `GROUP' commands,
2540 with the exception of ELF shared libraries, that will be added only
2541 when they are actually needed. This construct essentially enables
2542 `--as-needed' option for all the files listed inside of it and
2543 restores previous `--as-needed' resp. `--no-as-needed' setting
2547 The `OUTPUT' command names the output file. Using
2548 `OUTPUT(FILENAME)' in the linker script is exactly like using `-o
2549 FILENAME' on the command line (*note Command Line Options:
2550 Options.). If both are used, the command line option takes
2553 You can use the `OUTPUT' command to define a default name for the
2554 output file other than the usual default of `a.out'.
2557 The `SEARCH_DIR' command adds PATH to the list of paths where `ld'
2558 looks for archive libraries. Using `SEARCH_DIR(PATH)' is exactly
2559 like using `-L PATH' on the command line (*note Command Line
2560 Options: Options.). If both are used, then the linker will search
2561 both paths. Paths specified using the command line option are
2565 The `STARTUP' command is just like the `INPUT' command, except
2566 that FILENAME will become the first input file to be linked, as
2567 though it were specified first on the command line. This may be
2568 useful when using a system in which the entry point is always the
2569 start of the first file.
2572 File: ld.info, Node: Format Commands, Next: REGION_ALIAS, Prev: File Commands, Up: Simple Commands
2574 3.4.3 Commands Dealing with Object File Formats
2575 -----------------------------------------------
2577 A couple of linker script commands deal with object file formats.
2579 `OUTPUT_FORMAT(BFDNAME)'
2580 `OUTPUT_FORMAT(DEFAULT, BIG, LITTLE)'
2581 The `OUTPUT_FORMAT' command names the BFD format to use for the
2582 output file (*note BFD::). Using `OUTPUT_FORMAT(BFDNAME)' is
2583 exactly like using `--oformat BFDNAME' on the command line (*note
2584 Command Line Options: Options.). If both are used, the command
2585 line option takes precedence.
2587 You can use `OUTPUT_FORMAT' with three arguments to use different
2588 formats based on the `-EB' and `-EL' command line options. This
2589 permits the linker script to set the output format based on the
2592 If neither `-EB' nor `-EL' are used, then the output format will
2593 be the first argument, DEFAULT. If `-EB' is used, the output
2594 format will be the second argument, BIG. If `-EL' is used, the
2595 output format will be the third argument, LITTLE.
2597 For example, the default linker script for the MIPS ELF target
2599 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2600 This says that the default format for the output file is
2601 `elf32-bigmips', but if the user uses the `-EL' command line
2602 option, the output file will be created in the `elf32-littlemips'
2606 The `TARGET' command names the BFD format to use when reading input
2607 files. It affects subsequent `INPUT' and `GROUP' commands. This
2608 command is like using `-b BFDNAME' on the command line (*note
2609 Command Line Options: Options.). If the `TARGET' command is used
2610 but `OUTPUT_FORMAT' is not, then the last `TARGET' command is also
2611 used to set the format for the output file. *Note BFD::.
2614 File: ld.info, Node: REGION_ALIAS, Next: Miscellaneous Commands, Prev: Format Commands, Up: Simple Commands
2616 3.4.4 Assign alias names to memory regions
2617 ------------------------------------------
2619 Alias names can be added to existing memory regions created with the
2620 *Note MEMORY:: command. Each name corresponds to at most one memory
2623 REGION_ALIAS(ALIAS, REGION)
2625 The `REGION_ALIAS' function creates an alias name ALIAS for the
2626 memory region REGION. This allows a flexible mapping of output sections
2627 to memory regions. An example follows.
2629 Suppose we have an application for embedded systems which come with
2630 various memory storage devices. All have a general purpose, volatile
2631 memory `RAM' that allows code execution or data storage. Some may have
2632 a read-only, non-volatile memory `ROM' that allows code execution and
2633 read-only data access. The last variant is a read-only, non-volatile
2634 memory `ROM2' with read-only data access and no code execution
2635 capability. We have four output sections:
2637 * `.text' program code;
2639 * `.rodata' read-only data;
2641 * `.data' read-write initialized data;
2643 * `.bss' read-write zero initialized data.
2645 The goal is to provide a linker command file that contains a system
2646 independent part defining the output sections and a system dependent
2647 part mapping the output sections to the memory regions available on the
2648 system. Our embedded systems come with three different memory setups
2650 Section Variant A Variant B Variant C
2652 .rodata RAM ROM ROM2
2653 .data RAM RAM/ROM RAM/ROM2
2655 The notation `RAM/ROM' or `RAM/ROM2' means that this section is
2656 loaded into region `ROM' or `ROM2' respectively. Please note that the
2657 load address of the `.data' section starts in all three variants at the
2658 end of the `.rodata' section.
2660 The base linker script that deals with the output sections follows.
2661 It includes the system dependent `linkcmds.memory' file that describes
2663 INCLUDE linkcmds.memory
2676 .data : AT (rodata_end)
2681 data_size = SIZEOF(.data);
2682 data_load_start = LOADADDR(.data);
2689 Now we need three different `linkcmds.memory' files to define memory
2690 regions and alias names. The content of `linkcmds.memory' for the three
2691 variants `A', `B' and `C':
2693 Here everything goes into the `RAM'.
2696 RAM : ORIGIN = 0, LENGTH = 4M
2699 REGION_ALIAS("REGION_TEXT", RAM);
2700 REGION_ALIAS("REGION_RODATA", RAM);
2701 REGION_ALIAS("REGION_DATA", RAM);
2702 REGION_ALIAS("REGION_BSS", RAM);
2705 Program code and read-only data go into the `ROM'. Read-write
2706 data goes into the `RAM'. An image of the initialized data is
2707 loaded into the `ROM' and will be copied during system start into
2711 ROM : ORIGIN = 0, LENGTH = 3M
2712 RAM : ORIGIN = 0x10000000, LENGTH = 1M
2715 REGION_ALIAS("REGION_TEXT", ROM);
2716 REGION_ALIAS("REGION_RODATA", ROM);
2717 REGION_ALIAS("REGION_DATA", RAM);
2718 REGION_ALIAS("REGION_BSS", RAM);
2721 Program code goes into the `ROM'. Read-only data goes into the
2722 `ROM2'. Read-write data goes into the `RAM'. An image of the
2723 initialized data is loaded into the `ROM2' and will be copied
2724 during system start into the `RAM'.
2727 ROM : ORIGIN = 0, LENGTH = 2M
2728 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
2729 RAM : ORIGIN = 0x20000000, LENGTH = 1M
2732 REGION_ALIAS("REGION_TEXT", ROM);
2733 REGION_ALIAS("REGION_RODATA", ROM2);
2734 REGION_ALIAS("REGION_DATA", RAM);
2735 REGION_ALIAS("REGION_BSS", RAM);
2737 It is possible to write a common system initialization routine to
2738 copy the `.data' section from `ROM' or `ROM2' into the `RAM' if
2742 extern char data_start [];
2743 extern char data_size [];
2744 extern char data_load_start [];
2746 void copy_data(void)
2748 if (data_start != data_load_start)
2750 memcpy(data_start, data_load_start, (size_t) data_size);
2755 File: ld.info, Node: Miscellaneous Commands, Prev: REGION_ALIAS, Up: Simple Commands
2757 3.4.5 Other Linker Script Commands
2758 ----------------------------------
2760 There are a few other linker scripts commands.
2762 `ASSERT(EXP, MESSAGE)'
2763 Ensure that EXP is non-zero. If it is zero, then exit the linker
2764 with an error code, and print MESSAGE.
2766 `EXTERN(SYMBOL SYMBOL ...)'
2767 Force SYMBOL to be entered in the output file as an undefined
2768 symbol. Doing this may, for example, trigger linking of additional
2769 modules from standard libraries. You may list several SYMBOLs for
2770 each `EXTERN', and you may use `EXTERN' multiple times. This
2771 command has the same effect as the `-u' command-line option.
2773 `FORCE_COMMON_ALLOCATION'
2774 This command has the same effect as the `-d' command-line option:
2775 to make `ld' assign space to common symbols even if a relocatable
2776 output file is specified (`-r').
2778 `INHIBIT_COMMON_ALLOCATION'
2779 This command has the same effect as the `--no-define-common'
2780 command-line option: to make `ld' omit the assignment of addresses
2781 to common symbols even for a non-relocatable output file.
2783 `INSERT [ AFTER | BEFORE ] OUTPUT_SECTION'
2784 This command is typically used in a script specified by `-T' to
2785 augment the default `SECTIONS' with, for example, overlays. It
2786 inserts all prior linker script statements after (or before)
2787 OUTPUT_SECTION, and also causes `-T' to not override the default
2788 linker script. The exact insertion point is as for orphan
2789 sections. *Note Location Counter::. The insertion happens after
2790 the linker has mapped input sections to output sections. Prior to
2791 the insertion, since `-T' scripts are parsed before the default
2792 linker script, statements in the `-T' script occur before the
2793 default linker script statements in the internal linker
2794 representation of the script. In particular, input section
2795 assignments will be made to `-T' output sections before those in
2796 the default script. Here is an example of how a `-T' script using
2797 `INSERT' might look:
2803 .ov1 { ov1*(.text) }
2804 .ov2 { ov2*(.text) }
2809 `NOCROSSREFS(SECTION SECTION ...)'
2810 This command may be used to tell `ld' to issue an error about any
2811 references among certain output sections.
2813 In certain types of programs, particularly on embedded systems when
2814 using overlays, when one section is loaded into memory, another
2815 section will not be. Any direct references between the two
2816 sections would be errors. For example, it would be an error if
2817 code in one section called a function defined in the other section.
2819 The `NOCROSSREFS' command takes a list of output section names. If
2820 `ld' detects any cross references between the sections, it reports
2821 an error and returns a non-zero exit status. Note that the
2822 `NOCROSSREFS' command uses output section names, not input section
2825 `OUTPUT_ARCH(BFDARCH)'
2826 Specify a particular output machine architecture. The argument is
2827 one of the names used by the BFD library (*note BFD::). You can
2828 see the architecture of an object file by using the `objdump'
2829 program with the `-f' option.
2831 `LD_FEATURE(STRING)'
2832 This command may be used to modify `ld' behavior. If STRING is
2833 `"SANE_EXPR"' then absolute symbols and numbers in a script are
2834 simply treated as numbers everywhere. *Note Expression Section::.
2837 File: ld.info, Node: Assignments, Next: SECTIONS, Prev: Simple Commands, Up: Scripts
2839 3.5 Assigning Values to Symbols
2840 ===============================
2842 You may assign a value to a symbol in a linker script. This will define
2843 the symbol and place it into the symbol table with a global scope.
2847 * Simple Assignments:: Simple Assignments
2850 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2851 * Source Code Reference:: How to use a linker script defined symbol in source code
2854 File: ld.info, Node: Simple Assignments, Next: HIDDEN, Up: Assignments
2856 3.5.1 Simple Assignments
2857 ------------------------
2859 You may assign to a symbol using any of the C assignment operators:
2861 `SYMBOL = EXPRESSION ;'
2862 `SYMBOL += EXPRESSION ;'
2863 `SYMBOL -= EXPRESSION ;'
2864 `SYMBOL *= EXPRESSION ;'
2865 `SYMBOL /= EXPRESSION ;'
2866 `SYMBOL <<= EXPRESSION ;'
2867 `SYMBOL >>= EXPRESSION ;'
2868 `SYMBOL &= EXPRESSION ;'
2869 `SYMBOL |= EXPRESSION ;'
2871 The first case will define SYMBOL to the value of EXPRESSION. In
2872 the other cases, SYMBOL must already be defined, and the value will be
2873 adjusted accordingly.
2875 The special symbol name `.' indicates the location counter. You may
2876 only use this within a `SECTIONS' command. *Note Location Counter::.
2878 The semicolon after EXPRESSION is required.
2880 Expressions are defined below; see *Note Expressions::.
2882 You may write symbol assignments as commands in their own right, or
2883 as statements within a `SECTIONS' command, or as part of an output
2884 section description in a `SECTIONS' command.
2886 The section of the symbol will be set from the section of the
2887 expression; for more information, see *Note Expression Section::.
2889 Here is an example showing the three different places that symbol
2890 assignments may be used:
2900 _bdata = (. + 3) & ~ 3;
2901 .data : { *(.data) }
2903 In this example, the symbol `floating_point' will be defined as
2904 zero. The symbol `_etext' will be defined as the address following the
2905 last `.text' input section. The symbol `_bdata' will be defined as the
2906 address following the `.text' output section aligned upward to a 4 byte
2910 File: ld.info, Node: HIDDEN, Next: PROVIDE, Prev: Simple Assignments, Up: Assignments
2915 For ELF targeted ports, define a symbol that will be hidden and won't be
2916 exported. The syntax is `HIDDEN(SYMBOL = EXPRESSION)'.
2918 Here is the example from *Note Simple Assignments::, rewritten to use
2921 HIDDEN(floating_point = 0);
2929 HIDDEN(_bdata = (. + 3) & ~ 3);
2930 .data : { *(.data) }
2932 In this case none of the three symbols will be visible outside this
2936 File: ld.info, Node: PROVIDE, Next: PROVIDE_HIDDEN, Prev: HIDDEN, Up: Assignments
2941 In some cases, it is desirable for a linker script to define a symbol
2942 only if it is referenced and is not defined by any object included in
2943 the link. For example, traditional linkers defined the symbol `etext'.
2944 However, ANSI C requires that the user be able to use `etext' as a
2945 function name without encountering an error. The `PROVIDE' keyword may
2946 be used to define a symbol, such as `etext', only if it is referenced
2947 but not defined. The syntax is `PROVIDE(SYMBOL = EXPRESSION)'.
2949 Here is an example of using `PROVIDE' to define `etext':
2960 In this example, if the program defines `_etext' (with a leading
2961 underscore), the linker will give a multiple definition error. If, on
2962 the other hand, the program defines `etext' (with no leading
2963 underscore), the linker will silently use the definition in the program.
2964 If the program references `etext' but does not define it, the linker
2965 will use the definition in the linker script.
2968 File: ld.info, Node: PROVIDE_HIDDEN, Next: Source Code Reference, Prev: PROVIDE, Up: Assignments
2970 3.5.4 PROVIDE_HIDDEN
2971 --------------------
2973 Similar to `PROVIDE'. For ELF targeted ports, the symbol will be
2974 hidden and won't be exported.
2977 File: ld.info, Node: Source Code Reference, Prev: PROVIDE_HIDDEN, Up: Assignments
2979 3.5.5 Source Code Reference
2980 ---------------------------
2982 Accessing a linker script defined variable from source code is not
2983 intuitive. In particular a linker script symbol is not equivalent to a
2984 variable declaration in a high level language, it is instead a symbol
2985 that does not have a value.
2987 Before going further, it is important to note that compilers often
2988 transform names in the source code into different names when they are
2989 stored in the symbol table. For example, Fortran compilers commonly
2990 prepend or append an underscore, and C++ performs extensive `name
2991 mangling'. Therefore there might be a discrepancy between the name of
2992 a variable as it is used in source code and the name of the same
2993 variable as it is defined in a linker script. For example in C a
2994 linker script variable might be referred to as:
2998 But in the linker script it might be defined as:
3002 In the remaining examples however it is assumed that no name
3003 transformation has taken place.
3005 When a symbol is declared in a high level language such as C, two
3006 things happen. The first is that the compiler reserves enough space in
3007 the program's memory to hold the _value_ of the symbol. The second is
3008 that the compiler creates an entry in the program's symbol table which
3009 holds the symbol's _address_. ie the symbol table contains the address
3010 of the block of memory holding the symbol's value. So for example the
3011 following C declaration, at file scope:
3015 creates an entry called `foo' in the symbol table. This entry holds
3016 the address of an `int' sized block of memory where the number 1000 is
3019 When a program references a symbol the compiler generates code that
3020 first accesses the symbol table to find the address of the symbol's
3021 memory block and then code to read the value from that memory block.
3026 looks up the symbol `foo' in the symbol table, gets the address
3027 associated with this symbol and then writes the value 1 into that
3032 looks up the symbol `foo' in the symbol table, gets its address and
3033 then copies this address into the block of memory associated with the
3036 Linker scripts symbol declarations, by contrast, create an entry in
3037 the symbol table but do not assign any memory to them. Thus they are
3038 an address without a value. So for example the linker script
3043 creates an entry in the symbol table called `foo' which holds the
3044 address of memory location 1000, but nothing special is stored at
3045 address 1000. This means that you cannot access the _value_ of a
3046 linker script defined symbol - it has no value - all you can do is
3047 access the _address_ of a linker script defined symbol.
3049 Hence when you are using a linker script defined symbol in source
3050 code you should always take the address of the symbol, and never
3051 attempt to use its value. For example suppose you want to copy the
3052 contents of a section of memory called .ROM into a section called
3053 .FLASH and the linker script contains these declarations:
3055 start_of_ROM = .ROM;
3056 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3057 start_of_FLASH = .FLASH;
3059 Then the C source code to perform the copy would be:
3061 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3063 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3065 Note the use of the `&' operators. These are correct.
3068 File: ld.info, Node: SECTIONS, Next: MEMORY, Prev: Assignments, Up: Scripts
3070 3.6 SECTIONS Command
3071 ====================
3073 The `SECTIONS' command tells the linker how to map input sections into
3074 output sections, and how to place the output sections in memory.
3076 The format of the `SECTIONS' command is:
3084 Each SECTIONS-COMMAND may of be one of the following:
3086 * an `ENTRY' command (*note Entry command: Entry Point.)
3088 * a symbol assignment (*note Assignments::)
3090 * an output section description
3092 * an overlay description
3094 The `ENTRY' command and symbol assignments are permitted inside the
3095 `SECTIONS' command for convenience in using the location counter in
3096 those commands. This can also make the linker script easier to
3097 understand because you can use those commands at meaningful points in
3098 the layout of the output file.
3100 Output section descriptions and overlay descriptions are described
3103 If you do not use a `SECTIONS' command in your linker script, the
3104 linker will place each input section into an identically named output
3105 section in the order that the sections are first encountered in the
3106 input files. If all input sections are present in the first file, for
3107 example, the order of sections in the output file will match the order
3108 in the first input file. The first section will be at address zero.
3112 * Output Section Description:: Output section description
3113 * Output Section Name:: Output section name
3114 * Output Section Address:: Output section address
3115 * Input Section:: Input section description
3116 * Output Section Data:: Output section data
3117 * Output Section Keywords:: Output section keywords
3118 * Output Section Discarding:: Output section discarding
3119 * Output Section Attributes:: Output section attributes
3120 * Overlay Description:: Overlay description
3123 File: ld.info, Node: Output Section Description, Next: Output Section Name, Up: SECTIONS
3125 3.6.1 Output Section Description
3126 --------------------------------
3128 The full description of an output section looks like this:
3129 SECTION [ADDRESS] [(TYPE)] :
3131 [ALIGN(SECTION_ALIGN) | ALIGN_WITH_INPUT]
3132 [SUBALIGN(SUBSECTION_ALIGN)]
3135 OUTPUT-SECTION-COMMAND
3136 OUTPUT-SECTION-COMMAND
3138 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP] [,]
3140 Most output sections do not use most of the optional section
3143 The whitespace around SECTION is required, so that the section name
3144 is unambiguous. The colon and the curly braces are also required. The
3145 comma at the end may be required if a FILLEXP is used and the next
3146 SECTIONS-COMMAND looks like a continuation of the expression. The line
3147 breaks and other white space are optional.
3149 Each OUTPUT-SECTION-COMMAND may be one of the following:
3151 * a symbol assignment (*note Assignments::)
3153 * an input section description (*note Input Section::)
3155 * data values to include directly (*note Output Section Data::)
3157 * a special output section keyword (*note Output Section Keywords::)
3160 File: ld.info, Node: Output Section Name, Next: Output Section Address, Prev: Output Section Description, Up: SECTIONS
3162 3.6.2 Output Section Name
3163 -------------------------
3165 The name of the output section is SECTION. SECTION must meet the
3166 constraints of your output format. In formats which only support a
3167 limited number of sections, such as `a.out', the name must be one of
3168 the names supported by the format (`a.out', for example, allows only
3169 `.text', `.data' or `.bss'). If the output format supports any number
3170 of sections, but with numbers and not names (as is the case for Oasys),
3171 the name should be supplied as a quoted numeric string. A section name
3172 may consist of any sequence of characters, but a name which contains
3173 any unusual characters such as commas must be quoted.
3175 The output section name `/DISCARD/' is special; *Note Output Section
3179 File: ld.info, Node: Output Section Address, Next: Input Section, Prev: Output Section Name, Up: SECTIONS
3181 3.6.3 Output Section Address
3182 ----------------------------
3184 The ADDRESS is an expression for the VMA (the virtual memory address)
3185 of the output section. This address is optional, but if it is provided
3186 then the output address will be set exactly as specified.
3188 If the output address is not specified then one will be chosen for
3189 the section, based on the heuristic below. This address will be
3190 adjusted to fit the alignment requirement of the output section. The
3191 alignment requirement is the strictest alignment of any input section
3192 contained within the output section.
3194 The output section address heuristic is as follows:
3196 * If an output memory REGION is set for the section then it is added
3197 to this region and its address will be the next free address in
3200 * If the MEMORY command has been used to create a list of memory
3201 regions then the first region which has attributes compatible with
3202 the section is selected to contain it. The section's output
3203 address will be the next free address in that region; *Note
3206 * If no memory regions were specified, or none match the section then
3207 the output address will be based on the current value of the
3212 .text . : { *(.text) }
3216 .text : { *(.text) }
3218 are subtly different. The first will set the address of the `.text'
3219 output section to the current value of the location counter. The
3220 second will set it to the current value of the location counter aligned
3221 to the strictest alignment of any of the `.text' input sections.
3223 The ADDRESS may be an arbitrary expression; *Note Expressions::.
3224 For example, if you want to align the section on a 0x10 byte boundary,
3225 so that the lowest four bits of the section address are zero, you could
3226 do something like this:
3227 .text ALIGN(0x10) : { *(.text) }
3228 This works because `ALIGN' returns the current location counter
3229 aligned upward to the specified value.
3231 Specifying ADDRESS for a section will change the value of the
3232 location counter, provided that the section is non-empty. (Empty
3233 sections are ignored).
3236 File: ld.info, Node: Input Section, Next: Output Section Data, Prev: Output Section Address, Up: SECTIONS
3238 3.6.4 Input Section Description
3239 -------------------------------
3241 The most common output section command is an input section description.
3243 The input section description is the most basic linker script
3244 operation. You use output sections to tell the linker how to lay out
3245 your program in memory. You use input section descriptions to tell the
3246 linker how to map the input files into your memory layout.
3250 * Input Section Basics:: Input section basics
3251 * Input Section Wildcards:: Input section wildcard patterns
3252 * Input Section Common:: Input section for common symbols
3253 * Input Section Keep:: Input section and garbage collection
3254 * Input Section Example:: Input section example
3257 File: ld.info, Node: Input Section Basics, Next: Input Section Wildcards, Up: Input Section
3259 3.6.4.1 Input Section Basics
3260 ............................
3262 An input section description consists of a file name optionally followed
3263 by a list of section names in parentheses.
3265 The file name and the section name may be wildcard patterns, which we
3266 describe further below (*note Input Section Wildcards::).
3268 The most common input section description is to include all input
3269 sections with a particular name in the output section. For example, to
3270 include all input `.text' sections, you would write:
3272 Here the `*' is a wildcard which matches any file name. To exclude
3273 a list of files from matching the file name wildcard, EXCLUDE_FILE may
3274 be used to match all files except the ones specified in the
3275 EXCLUDE_FILE list. For example:
3276 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3277 will cause all .ctors sections from all files except `crtend.o' and
3278 `otherfile.o' to be included.
3280 There are two ways to include more than one section:
3283 The difference between these is the order in which the `.text' and
3284 `.rdata' input sections will appear in the output section. In the
3285 first example, they will be intermingled, appearing in the same order as
3286 they are found in the linker input. In the second example, all `.text'
3287 input sections will appear first, followed by all `.rdata' input
3290 You can specify a file name to include sections from a particular
3291 file. You would do this if one or more of your files contain special
3292 data that needs to be at a particular location in memory. For example:
3295 To refine the sections that are included based on the section flags
3296 of an input section, INPUT_SECTION_FLAGS may be used.
3298 Here is a simple example for using Section header flags for ELF
3302 .text : { INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) }
3303 .text2 : { INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) }
3306 In this example, the output section `.text' will be comprised of any
3307 input section matching the name *(.text) whose section header flags
3308 `SHF_MERGE' and `SHF_STRINGS' are set. The output section `.text2'
3309 will be comprised of any input section matching the name *(.text) whose
3310 section header flag `SHF_WRITE' is clear.
3312 You can also specify files within archives by writing a pattern
3313 matching the archive, a colon, then the pattern matching the file, with
3314 no whitespace around the colon.
3317 matches file within archive
3320 matches the whole archive
3323 matches file but not one in an archive
3325 Either one or both of `archive' and `file' can contain shell
3326 wildcards. On DOS based file systems, the linker will assume that a
3327 single letter followed by a colon is a drive specifier, so `c:myfile.o'
3328 is a simple file specification, not `myfile.o' within an archive called
3329 `c'. `archive:file' filespecs may also be used within an
3330 `EXCLUDE_FILE' list, but may not appear in other linker script
3331 contexts. For instance, you cannot extract a file from an archive by
3332 using `archive:file' in an `INPUT' command.
3334 If you use a file name without a list of sections, then all sections
3335 in the input file will be included in the output section. This is not
3336 commonly done, but it may by useful on occasion. For example:
3339 When you use a file name which is not an `archive:file' specifier
3340 and does not contain any wild card characters, the linker will first
3341 see if you also specified the file name on the linker command line or
3342 in an `INPUT' command. If you did not, the linker will attempt to open
3343 the file as an input file, as though it appeared on the command line.
3344 Note that this differs from an `INPUT' command, because the linker will
3345 not search for the file in the archive search path.
3348 File: ld.info, Node: Input Section Wildcards, Next: Input Section Common, Prev: Input Section Basics, Up: Input Section
3350 3.6.4.2 Input Section Wildcard Patterns
3351 .......................................
3353 In an input section description, either the file name or the section
3354 name or both may be wildcard patterns.
3356 The file name of `*' seen in many examples is a simple wildcard
3357 pattern for the file name.
3359 The wildcard patterns are like those used by the Unix shell.
3362 matches any number of characters
3365 matches any single character
3368 matches a single instance of any of the CHARS; the `-' character
3369 may be used to specify a range of characters, as in `[a-z]' to
3370 match any lower case letter
3373 quotes the following character
3375 When a file name is matched with a wildcard, the wildcard characters
3376 will not match a `/' character (used to separate directory names on
3377 Unix). A pattern consisting of a single `*' character is an exception;
3378 it will always match any file name, whether it contains a `/' or not.
3379 In a section name, the wildcard characters will match a `/' character.
3381 File name wildcard patterns only match files which are explicitly
3382 specified on the command line or in an `INPUT' command. The linker
3383 does not search directories to expand wildcards.
3385 If a file name matches more than one wildcard pattern, or if a file
3386 name appears explicitly and is also matched by a wildcard pattern, the
3387 linker will use the first match in the linker script. For example, this
3388 sequence of input section descriptions is probably in error, because the
3389 `data.o' rule will not be used:
3390 .data : { *(.data) }
3391 .data1 : { data.o(.data) }
3393 Normally, the linker will place files and sections matched by
3394 wildcards in the order in which they are seen during the link. You can
3395 change this by using the `SORT_BY_NAME' keyword, which appears before a
3396 wildcard pattern in parentheses (e.g., `SORT_BY_NAME(.text*)'). When
3397 the `SORT_BY_NAME' keyword is used, the linker will sort the files or
3398 sections into ascending order by name before placing them in the output
3401 `SORT_BY_ALIGNMENT' is very similar to `SORT_BY_NAME'. The
3402 difference is `SORT_BY_ALIGNMENT' will sort sections into descending
3403 order by alignment before placing them in the output file. Larger
3404 alignments are placed before smaller alignments in order to reduce the
3405 amount of padding necessary.
3407 `SORT_BY_INIT_PRIORITY' is very similar to `SORT_BY_NAME'. The
3408 difference is `SORT_BY_INIT_PRIORITY' will sort sections into ascending
3409 order by numerical value of the GCC init_priority attribute encoded in
3410 the section name before placing them in the output file.
3412 `SORT' is an alias for `SORT_BY_NAME'.
3414 When there are nested section sorting commands in linker script,
3415 there can be at most 1 level of nesting for section sorting commands.
3417 1. `SORT_BY_NAME' (`SORT_BY_ALIGNMENT' (wildcard section pattern)).
3418 It will sort the input sections by name first, then by alignment
3419 if two sections have the same name.
3421 2. `SORT_BY_ALIGNMENT' (`SORT_BY_NAME' (wildcard section pattern)).
3422 It will sort the input sections by alignment first, then by name
3423 if two sections have the same alignment.
3425 3. `SORT_BY_NAME' (`SORT_BY_NAME' (wildcard section pattern)) is
3426 treated the same as `SORT_BY_NAME' (wildcard section pattern).
3428 4. `SORT_BY_ALIGNMENT' (`SORT_BY_ALIGNMENT' (wildcard section
3429 pattern)) is treated the same as `SORT_BY_ALIGNMENT' (wildcard
3432 5. All other nested section sorting commands are invalid.
3434 When both command line section sorting option and linker script
3435 section sorting command are used, section sorting command always takes
3436 precedence over the command line option.
3438 If the section sorting command in linker script isn't nested, the
3439 command line option will make the section sorting command to be treated
3440 as nested sorting command.
3442 1. `SORT_BY_NAME' (wildcard section pattern ) with `--sort-sections
3443 alignment' is equivalent to `SORT_BY_NAME' (`SORT_BY_ALIGNMENT'
3444 (wildcard section pattern)).
3446 2. `SORT_BY_ALIGNMENT' (wildcard section pattern) with
3447 `--sort-section name' is equivalent to `SORT_BY_ALIGNMENT'
3448 (`SORT_BY_NAME' (wildcard section pattern)).
3450 If the section sorting command in linker script is nested, the
3451 command line option will be ignored.
3453 `SORT_NONE' disables section sorting by ignoring the command line
3454 section sorting option.
3456 If you ever get confused about where input sections are going, use
3457 the `-M' linker option to generate a map file. The map file shows
3458 precisely how input sections are mapped to output sections.
3460 This example shows how wildcard patterns might be used to partition
3461 files. This linker script directs the linker to place all `.text'
3462 sections in `.text' and all `.bss' sections in `.bss'. The linker will
3463 place the `.data' section from all files beginning with an upper case
3464 character in `.DATA'; for all other files, the linker will place the
3465 `.data' section in `.data'.
3467 .text : { *(.text) }
3468 .DATA : { [A-Z]*(.data) }
3469 .data : { *(.data) }
3474 File: ld.info, Node: Input Section Common, Next: Input Section Keep, Prev: Input Section Wildcards, Up: Input Section
3476 3.6.4.3 Input Section for Common Symbols
3477 ........................................
3479 A special notation is needed for common symbols, because in many object
3480 file formats common symbols do not have a particular input section. The
3481 linker treats common symbols as though they are in an input section
3484 You may use file names with the `COMMON' section just as with any
3485 other input sections. You can use this to place common symbols from a
3486 particular input file in one section while common symbols from other
3487 input files are placed in another section.
3489 In most cases, common symbols in input files will be placed in the
3490 `.bss' section in the output file. For example:
3491 .bss { *(.bss) *(COMMON) }
3493 Some object file formats have more than one type of common symbol.
3494 For example, the MIPS ELF object file format distinguishes standard
3495 common symbols and small common symbols. In this case, the linker will
3496 use a different special section name for other types of common symbols.
3497 In the case of MIPS ELF, the linker uses `COMMON' for standard common
3498 symbols and `.scommon' for small common symbols. This permits you to
3499 map the different types of common symbols into memory at different
3502 You will sometimes see `[COMMON]' in old linker scripts. This
3503 notation is now considered obsolete. It is equivalent to `*(COMMON)'.
3506 File: ld.info, Node: Input Section Keep, Next: Input Section Example, Prev: Input Section Common, Up: Input Section
3508 3.6.4.4 Input Section and Garbage Collection
3509 ............................................
3511 When link-time garbage collection is in use (`--gc-sections'), it is
3512 often useful to mark sections that should not be eliminated. This is
3513 accomplished by surrounding an input section's wildcard entry with
3514 `KEEP()', as in `KEEP(*(.init))' or `KEEP(SORT_BY_NAME(*)(.ctors))'.
3517 File: ld.info, Node: Input Section Example, Prev: Input Section Keep, Up: Input Section
3519 3.6.4.5 Input Section Example
3520 .............................
3522 The following example is a complete linker script. It tells the linker
3523 to read all of the sections from file `all.o' and place them at the
3524 start of output section `outputa' which starts at location `0x10000'.
3525 All of section `.input1' from file `foo.o' follows immediately, in the
3526 same output section. All of section `.input2' from `foo.o' goes into
3527 output section `outputb', followed by section `.input1' from `foo1.o'.
3528 All of the remaining `.input1' and `.input2' sections from any files
3529 are written to output section `outputc'.
3550 File: ld.info, Node: Output Section Data, Next: Output Section Keywords, Prev: Input Section, Up: SECTIONS
3552 3.6.5 Output Section Data
3553 -------------------------
3555 You can include explicit bytes of data in an output section by using
3556 `BYTE', `SHORT', `LONG', `QUAD', or `SQUAD' as an output section
3557 command. Each keyword is followed by an expression in parentheses
3558 providing the value to store (*note Expressions::). The value of the
3559 expression is stored at the current value of the location counter.
3561 The `BYTE', `SHORT', `LONG', and `QUAD' commands store one, two,
3562 four, and eight bytes (respectively). After storing the bytes, the
3563 location counter is incremented by the number of bytes stored.
3565 For example, this will store the byte 1 followed by the four byte
3566 value of the symbol `addr':
3570 When using a 64 bit host or target, `QUAD' and `SQUAD' are the same;
3571 they both store an 8 byte, or 64 bit, value. When both host and target
3572 are 32 bits, an expression is computed as 32 bits. In this case `QUAD'
3573 stores a 32 bit value zero extended to 64 bits, and `SQUAD' stores a 32
3574 bit value sign extended to 64 bits.
3576 If the object file format of the output file has an explicit
3577 endianness, which is the normal case, the value will be stored in that
3578 endianness. When the object file format does not have an explicit
3579 endianness, as is true of, for example, S-records, the value will be
3580 stored in the endianness of the first input object file.
3582 Note--these commands only work inside a section description and not
3583 between them, so the following will produce an error from the linker:
3584 SECTIONS { .text : { *(.text) } LONG(1) .data : { *(.data) } }
3585 whereas this will work:
3586 SECTIONS { .text : { *(.text) ; LONG(1) } .data : { *(.data) } }
3588 You may use the `FILL' command to set the fill pattern for the
3589 current section. It is followed by an expression in parentheses. Any
3590 otherwise unspecified regions of memory within the section (for example,
3591 gaps left due to the required alignment of input sections) are filled
3592 with the value of the expression, repeated as necessary. A `FILL'
3593 statement covers memory locations after the point at which it occurs in
3594 the section definition; by including more than one `FILL' statement,
3595 you can have different fill patterns in different parts of an output
3598 This example shows how to fill unspecified regions of memory with the
3602 The `FILL' command is similar to the `=FILLEXP' output section
3603 attribute, but it only affects the part of the section following the
3604 `FILL' command, rather than the entire section. If both are used, the
3605 `FILL' command takes precedence. *Note Output Section Fill::, for
3606 details on the fill expression.
3609 File: ld.info, Node: Output Section Keywords, Next: Output Section Discarding, Prev: Output Section Data, Up: SECTIONS
3611 3.6.6 Output Section Keywords
3612 -----------------------------
3614 There are a couple of keywords which can appear as output section
3617 `CREATE_OBJECT_SYMBOLS'
3618 The command tells the linker to create a symbol for each input
3619 file. The name of each symbol will be the name of the
3620 corresponding input file. The section of each symbol will be the
3621 output section in which the `CREATE_OBJECT_SYMBOLS' command
3624 This is conventional for the a.out object file format. It is not
3625 normally used for any other object file format.
3628 When linking using the a.out object file format, the linker uses an
3629 unusual set construct to support C++ global constructors and
3630 destructors. When linking object file formats which do not support
3631 arbitrary sections, such as ECOFF and XCOFF, the linker will
3632 automatically recognize C++ global constructors and destructors by
3633 name. For these object file formats, the `CONSTRUCTORS' command
3634 tells the linker to place constructor information in the output
3635 section where the `CONSTRUCTORS' command appears. The
3636 `CONSTRUCTORS' command is ignored for other object file formats.
3638 The symbol `__CTOR_LIST__' marks the start of the global
3639 constructors, and the symbol `__CTOR_END__' marks the end.
3640 Similarly, `__DTOR_LIST__' and `__DTOR_END__' mark the start and
3641 end of the global destructors. The first word in the list is the
3642 number of entries, followed by the address of each constructor or
3643 destructor, followed by a zero word. The compiler must arrange to
3644 actually run the code. For these object file formats GNU C++
3645 normally calls constructors from a subroutine `__main'; a call to
3646 `__main' is automatically inserted into the startup code for
3647 `main'. GNU C++ normally runs destructors either by using
3648 `atexit', or directly from the function `exit'.
3650 For object file formats such as `COFF' or `ELF' which support
3651 arbitrary section names, GNU C++ will normally arrange to put the
3652 addresses of global constructors and destructors into the `.ctors'
3653 and `.dtors' sections. Placing the following sequence into your
3654 linker script will build the sort of table which the GNU C++
3655 runtime code expects to see.
3658 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3663 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3668 If you are using the GNU C++ support for initialization priority,
3669 which provides some control over the order in which global
3670 constructors are run, you must sort the constructors at link time
3671 to ensure that they are executed in the correct order. When using
3672 the `CONSTRUCTORS' command, use `SORT_BY_NAME(CONSTRUCTORS)'
3673 instead. When using the `.ctors' and `.dtors' sections, use
3674 `*(SORT_BY_NAME(.ctors))' and `*(SORT_BY_NAME(.dtors))' instead of
3675 just `*(.ctors)' and `*(.dtors)'.
3677 Normally the compiler and linker will handle these issues
3678 automatically, and you will not need to concern yourself with
3679 them. However, you may need to consider this if you are using C++
3680 and writing your own linker scripts.
3684 File: ld.info, Node: Output Section Discarding, Next: Output Section Attributes, Prev: Output Section Keywords, Up: SECTIONS
3686 3.6.7 Output Section Discarding
3687 -------------------------------
3689 The linker will not normally create output sections with no contents.
3690 This is for convenience when referring to input sections that may or
3691 may not be present in any of the input files. For example:
3693 will only create a `.foo' section in the output file if there is a
3694 `.foo' section in at least one input file, and if the input sections
3695 are not all empty. Other link script directives that allocate space in
3696 an output section will also create the output section. So too will
3697 assignments to dot even if the assignment does not create space, except
3698 for `. = 0', `. = . + 0', `. = sym', `. = . + sym' and `. = ALIGN (. !=
3699 0, expr, 1)' when `sym' is an absolute symbol of value 0 defined in the
3700 script. This allows you to force output of an empty section with `. =
3703 The linker will ignore address assignments (*note Output Section
3704 Address::) on discarded output sections, except when the linker script
3705 defines symbols in the output section. In that case the linker will
3706 obey the address assignments, possibly advancing dot even though the
3707 section is discarded.
3709 The special output section name `/DISCARD/' may be used to discard
3710 input sections. Any input sections which are assigned to an output
3711 section named `/DISCARD/' are not included in the output file.
3714 File: ld.info, Node: Output Section Attributes, Next: Overlay Description, Prev: Output Section Discarding, Up: SECTIONS
3716 3.6.8 Output Section Attributes
3717 -------------------------------
3719 We showed above that the full description of an output section looked
3722 SECTION [ADDRESS] [(TYPE)] :
3724 [ALIGN(SECTION_ALIGN)]
3725 [SUBALIGN(SUBSECTION_ALIGN)]
3728 OUTPUT-SECTION-COMMAND
3729 OUTPUT-SECTION-COMMAND
3731 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
3733 We've already described SECTION, ADDRESS, and
3734 OUTPUT-SECTION-COMMAND. In this section we will describe the remaining
3739 * Output Section Type:: Output section type
3740 * Output Section LMA:: Output section LMA
3741 * Forced Output Alignment:: Forced Output Alignment
3742 * Forced Input Alignment:: Forced Input Alignment
3743 * Output Section Constraint:: Output section constraint
3744 * Output Section Region:: Output section region
3745 * Output Section Phdr:: Output section phdr
3746 * Output Section Fill:: Output section fill
3749 File: ld.info, Node: Output Section Type, Next: Output Section LMA, Up: Output Section Attributes
3751 3.6.8.1 Output Section Type
3752 ...........................
3754 Each output section may have a type. The type is a keyword in
3755 parentheses. The following types are defined:
3758 The section should be marked as not loadable, so that it will not
3759 be loaded into memory when the program is run.
3765 These type names are supported for backward compatibility, and are
3766 rarely used. They all have the same effect: the section should be
3767 marked as not allocatable, so that no memory is allocated for the
3768 section when the program is run.
3770 The linker normally sets the attributes of an output section based on
3771 the input sections which map into it. You can override this by using
3772 the section type. For example, in the script sample below, the `ROM'
3773 section is addressed at memory location `0' and does not need to be
3774 loaded when the program is run.
3776 ROM 0 (NOLOAD) : { ... }
3781 File: ld.info, Node: Output Section LMA, Next: Forced Output Alignment, Prev: Output Section Type, Up: Output Section Attributes
3783 3.6.8.2 Output Section LMA
3784 ..........................
3786 Every section has a virtual address (VMA) and a load address (LMA); see
3787 *Note Basic Script Concepts::. The virtual address is specified by the
3788 *note Output Section Address:: described earlier. The load address is
3789 specified by the `AT' or `AT>' keywords. Specifying a load address is
3792 The `AT' keyword takes an expression as an argument. This specifies
3793 the exact load address of the section. The `AT>' keyword takes the
3794 name of a memory region as an argument. *Note MEMORY::. The load
3795 address of the section is set to the next free address in the region,
3796 aligned to the section's alignment requirements.
3798 If neither `AT' nor `AT>' is specified for an allocatable section,
3799 the linker will use the following heuristic to determine the load
3802 * If the section has a specific VMA address, then this is used as
3803 the LMA address as well.
3805 * If the section is not allocatable then its LMA is set to its VMA.
3807 * Otherwise if a memory region can be found that is compatible with
3808 the current section, and this region contains at least one
3809 section, then the LMA is set so the difference between the VMA and
3810 LMA is the same as the difference between the VMA and LMA of the
3811 last section in the located region.
3813 * If no memory regions have been declared then a default region that
3814 covers the entire address space is used in the previous step.
3816 * If no suitable region could be found, or there was no previous
3817 section then the LMA is set equal to the VMA.
3819 This feature is designed to make it easy to build a ROM image. For
3820 example, the following linker script creates three output sections: one
3821 called `.text', which starts at `0x1000', one called `.mdata', which is
3822 loaded at the end of the `.text' section even though its VMA is
3823 `0x2000', and one called `.bss' to hold uninitialized data at address
3824 `0x3000'. The symbol `_data' is defined with the value `0x2000', which
3825 shows that the location counter holds the VMA value, not the LMA value.
3829 .text 0x1000 : { *(.text) _etext = . ; }
3831 AT ( ADDR (.text) + SIZEOF (.text) )
3832 { _data = . ; *(.data); _edata = . ; }
3834 { _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;}
3837 The run-time initialization code for use with a program generated
3838 with this linker script would include something like the following, to
3839 copy the initialized data from the ROM image to its runtime address.
3840 Notice how this code takes advantage of the symbols defined by the
3843 extern char _etext, _data, _edata, _bstart, _bend;
3844 char *src = &_etext;
3847 /* ROM has data at end of text; copy it. */
3848 while (dst < &_edata)
3852 for (dst = &_bstart; dst< &_bend; dst++)
3856 File: ld.info, Node: Forced Output Alignment, Next: Forced Input Alignment, Prev: Output Section LMA, Up: Output Section Attributes
3858 3.6.8.3 Forced Output Alignment
3859 ...............................
3861 You can increase an output section's alignment by using ALIGN. As an
3862 alternative you can enforce that the difference between the VMA and LMA
3863 remains intact throughout this output section with the ALIGN_WITH_INPUT
3867 File: ld.info, Node: Forced Input Alignment, Next: Output Section Constraint, Prev: Forced Output Alignment, Up: Output Section Attributes
3869 3.6.8.4 Forced Input Alignment
3870 ..............................
3872 You can force input section alignment within an output section by using
3873 SUBALIGN. The value specified overrides any alignment given by input
3874 sections, whether larger or smaller.
3877 File: ld.info, Node: Output Section Constraint, Next: Output Section Region, Prev: Forced Input Alignment, Up: Output Section Attributes
3879 3.6.8.5 Output Section Constraint
3880 .................................
3882 You can specify that an output section should only be created if all of
3883 its input sections are read-only or all of its input sections are
3884 read-write by using the keyword `ONLY_IF_RO' and `ONLY_IF_RW'
3888 File: ld.info, Node: Output Section Region, Next: Output Section Phdr, Prev: Output Section Constraint, Up: Output Section Attributes
3890 3.6.8.6 Output Section Region
3891 .............................
3893 You can assign a section to a previously defined region of memory by
3894 using `>REGION'. *Note MEMORY::.
3896 Here is a simple example:
3897 MEMORY { rom : ORIGIN = 0x1000, LENGTH = 0x1000 }
3898 SECTIONS { ROM : { *(.text) } >rom }
3901 File: ld.info, Node: Output Section Phdr, Next: Output Section Fill, Prev: Output Section Region, Up: Output Section Attributes
3903 3.6.8.7 Output Section Phdr
3904 ...........................
3906 You can assign a section to a previously defined program segment by
3907 using `:PHDR'. *Note PHDRS::. If a section is assigned to one or more
3908 segments, then all subsequent allocated sections will be assigned to
3909 those segments as well, unless they use an explicitly `:PHDR' modifier.
3910 You can use `:NONE' to tell the linker to not put the section in any
3913 Here is a simple example:
3914 PHDRS { text PT_LOAD ; }
3915 SECTIONS { .text : { *(.text) } :text }
3918 File: ld.info, Node: Output Section Fill, Prev: Output Section Phdr, Up: Output Section Attributes
3920 3.6.8.8 Output Section Fill
3921 ...........................
3923 You can set the fill pattern for an entire section by using `=FILLEXP'.
3924 FILLEXP is an expression (*note Expressions::). Any otherwise
3925 unspecified regions of memory within the output section (for example,
3926 gaps left due to the required alignment of input sections) will be
3927 filled with the value, repeated as necessary. If the fill expression
3928 is a simple hex number, ie. a string of hex digit starting with `0x'
3929 and without a trailing `k' or `M', then an arbitrarily long sequence of
3930 hex digits can be used to specify the fill pattern; Leading zeros
3931 become part of the pattern too. For all other cases, including extra
3932 parentheses or a unary `+', the fill pattern is the four least
3933 significant bytes of the value of the expression. In all cases, the
3934 number is big-endian.
3936 You can also change the fill value with a `FILL' command in the
3937 output section commands; (*note Output Section Data::).
3939 Here is a simple example:
3940 SECTIONS { .text : { *(.text) } =0x90909090 }
3943 File: ld.info, Node: Overlay Description, Prev: Output Section Attributes, Up: SECTIONS
3945 3.6.9 Overlay Description
3946 -------------------------
3948 An overlay description provides an easy way to describe sections which
3949 are to be loaded as part of a single memory image but are to be run at
3950 the same memory address. At run time, some sort of overlay manager will
3951 copy the overlaid sections in and out of the runtime memory address as
3952 required, perhaps by simply manipulating addressing bits. This approach
3953 can be useful, for example, when a certain region of memory is faster
3956 Overlays are described using the `OVERLAY' command. The `OVERLAY'
3957 command is used within a `SECTIONS' command, like an output section
3958 description. The full syntax of the `OVERLAY' command is as follows:
3959 OVERLAY [START] : [NOCROSSREFS] [AT ( LDADDR )]
3963 OUTPUT-SECTION-COMMAND
3964 OUTPUT-SECTION-COMMAND
3966 } [:PHDR...] [=FILL]
3969 OUTPUT-SECTION-COMMAND
3970 OUTPUT-SECTION-COMMAND
3972 } [:PHDR...] [=FILL]
3974 } [>REGION] [:PHDR...] [=FILL] [,]
3976 Everything is optional except `OVERLAY' (a keyword), and each
3977 section must have a name (SECNAME1 and SECNAME2 above). The section
3978 definitions within the `OVERLAY' construct are identical to those
3979 within the general `SECTIONS' construct (*note SECTIONS::), except that
3980 no addresses and no memory regions may be defined for sections within
3983 The comma at the end may be required if a FILL is used and the next
3984 SECTIONS-COMMAND looks like a continuation of the expression.
3986 The sections are all defined with the same starting address. The
3987 load addresses of the sections are arranged such that they are
3988 consecutive in memory starting at the load address used for the
3989 `OVERLAY' as a whole (as with normal section definitions, the load
3990 address is optional, and defaults to the start address; the start
3991 address is also optional, and defaults to the current value of the
3994 If the `NOCROSSREFS' keyword is used, and there are any references
3995 among the sections, the linker will report an error. Since the
3996 sections all run at the same address, it normally does not make sense
3997 for one section to refer directly to another. *Note NOCROSSREFS:
3998 Miscellaneous Commands.
4000 For each section within the `OVERLAY', the linker automatically
4001 provides two symbols. The symbol `__load_start_SECNAME' is defined as
4002 the starting load address of the section. The symbol
4003 `__load_stop_SECNAME' is defined as the final load address of the
4004 section. Any characters within SECNAME which are not legal within C
4005 identifiers are removed. C (or assembler) code may use these symbols
4006 to move the overlaid sections around as necessary.
4008 At the end of the overlay, the value of the location counter is set
4009 to the start address of the overlay plus the size of the largest
4012 Here is an example. Remember that this would appear inside a
4013 `SECTIONS' construct.
4014 OVERLAY 0x1000 : AT (0x4000)
4016 .text0 { o1/*.o(.text) }
4017 .text1 { o2/*.o(.text) }
4019 This will define both `.text0' and `.text1' to start at address
4020 0x1000. `.text0' will be loaded at address 0x4000, and `.text1' will
4021 be loaded immediately after `.text0'. The following symbols will be
4022 defined if referenced: `__load_start_text0', `__load_stop_text0',
4023 `__load_start_text1', `__load_stop_text1'.
4025 C code to copy overlay `.text1' into the overlay area might look
4028 extern char __load_start_text1, __load_stop_text1;
4029 memcpy ((char *) 0x1000, &__load_start_text1,
4030 &__load_stop_text1 - &__load_start_text1);
4032 Note that the `OVERLAY' command is just syntactic sugar, since
4033 everything it does can be done using the more basic commands. The above
4034 example could have been written identically as follows.
4036 .text0 0x1000 : AT (0x4000) { o1/*.o(.text) }
4037 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4038 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4039 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) { o2/*.o(.text) }
4040 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4041 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4042 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4045 File: ld.info, Node: MEMORY, Next: PHDRS, Prev: SECTIONS, Up: Scripts
4050 The linker's default configuration permits allocation of all available
4051 memory. You can override this by using the `MEMORY' command.
4053 The `MEMORY' command describes the location and size of blocks of
4054 memory in the target. You can use it to describe which memory regions
4055 may be used by the linker, and which memory regions it must avoid. You
4056 can then assign sections to particular memory regions. The linker will
4057 set section addresses based on the memory regions, and will warn about
4058 regions that become too full. The linker will not shuffle sections
4059 around to fit into the available regions.
4061 A linker script may contain at most one use of the `MEMORY' command.
4062 However, you can define as many blocks of memory within it as you
4063 wish. The syntax is:
4066 NAME [(ATTR)] : ORIGIN = ORIGIN, LENGTH = LEN
4070 The NAME is a name used in the linker script to refer to the region.
4071 The region name has no meaning outside of the linker script. Region
4072 names are stored in a separate name space, and will not conflict with
4073 symbol names, file names, or section names. Each memory region must
4074 have a distinct name within the `MEMORY' command. However you can add
4075 later alias names to existing memory regions with the *Note
4076 REGION_ALIAS:: command.
4078 The ATTR string is an optional list of attributes that specify
4079 whether to use a particular memory region for an input section which is
4080 not explicitly mapped in the linker script. As described in *Note
4081 SECTIONS::, if you do not specify an output section for some input
4082 section, the linker will create an output section with the same name as
4083 the input section. If you define region attributes, the linker will use
4084 them to select the memory region for the output section that it creates.
4086 The ATTR string must consist only of the following characters:
4106 Invert the sense of any of the attributes that follow
4108 If a unmapped section matches any of the listed attributes other than
4109 `!', it will be placed in the memory region. The `!' attribute
4110 reverses this test, so that an unmapped section will be placed in the
4111 memory region only if it does not match any of the listed attributes.
4113 The ORIGIN is an numerical expression for the start address of the
4114 memory region. The expression must evaluate to a constant and it
4115 cannot involve any symbols. The keyword `ORIGIN' may be abbreviated to
4116 `org' or `o' (but not, for example, `ORG').
4118 The LEN is an expression for the size in bytes of the memory region.
4119 As with the ORIGIN expression, the expression must be numerical only
4120 and must evaluate to a constant. The keyword `LENGTH' may be
4121 abbreviated to `len' or `l'.
4123 In the following example, we specify that there are two memory
4124 regions available for allocation: one starting at `0' for 256 kilobytes,
4125 and the other starting at `0x40000000' for four megabytes. The linker
4126 will place into the `rom' memory region every section which is not
4127 explicitly mapped into a memory region, and is either read-only or
4128 executable. The linker will place other sections which are not
4129 explicitly mapped into a memory region into the `ram' memory region.
4133 rom (rx) : ORIGIN = 0, LENGTH = 256K
4134 ram (!rx) : org = 0x40000000, l = 4M
4137 Once you define a memory region, you can direct the linker to place
4138 specific output sections into that memory region by using the `>REGION'
4139 output section attribute. For example, if you have a memory region
4140 named `mem', you would use `>mem' in the output section definition.
4141 *Note Output Section Region::. If no address was specified for the
4142 output section, the linker will set the address to the next available
4143 address within the memory region. If the combined output sections
4144 directed to a memory region are too large for the region, the linker
4145 will issue an error message.
4147 It is possible to access the origin and length of a memory in an
4148 expression via the `ORIGIN(MEMORY)' and `LENGTH(MEMORY)' functions:
4150 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4153 File: ld.info, Node: PHDRS, Next: VERSION, Prev: MEMORY, Up: Scripts
4158 The ELF object file format uses "program headers", also knows as
4159 "segments". The program headers describe how the program should be
4160 loaded into memory. You can print them out by using the `objdump'
4161 program with the `-p' option.
4163 When you run an ELF program on a native ELF system, the system loader
4164 reads the program headers in order to figure out how to load the
4165 program. This will only work if the program headers are set correctly.
4166 This manual does not describe the details of how the system loader
4167 interprets program headers; for more information, see the ELF ABI.
4169 The linker will create reasonable program headers by default.
4170 However, in some cases, you may need to specify the program headers more
4171 precisely. You may use the `PHDRS' command for this purpose. When the
4172 linker sees the `PHDRS' command in the linker script, it will not
4173 create any program headers other than the ones specified.
4175 The linker only pays attention to the `PHDRS' command when
4176 generating an ELF output file. In other cases, the linker will simply
4179 This is the syntax of the `PHDRS' command. The words `PHDRS',
4180 `FILEHDR', `AT', and `FLAGS' are keywords.
4184 NAME TYPE [ FILEHDR ] [ PHDRS ] [ AT ( ADDRESS ) ]
4185 [ FLAGS ( FLAGS ) ] ;
4188 The NAME is used only for reference in the `SECTIONS' command of the
4189 linker script. It is not put into the output file. Program header
4190 names are stored in a separate name space, and will not conflict with
4191 symbol names, file names, or section names. Each program header must
4192 have a distinct name. The headers are processed in order and it is
4193 usual for them to map to sections in ascending load address order.
4195 Certain program header types describe segments of memory which the
4196 system loader will load from the file. In the linker script, you
4197 specify the contents of these segments by placing allocatable output
4198 sections in the segments. You use the `:PHDR' output section attribute
4199 to place a section in a particular segment. *Note Output Section
4202 It is normal to put certain sections in more than one segment. This
4203 merely implies that one segment of memory contains another. You may
4204 repeat `:PHDR', using it once for each segment which should contain the
4207 If you place a section in one or more segments using `:PHDR', then
4208 the linker will place all subsequent allocatable sections which do not
4209 specify `:PHDR' in the same segments. This is for convenience, since
4210 generally a whole set of contiguous sections will be placed in a single
4211 segment. You can use `:NONE' to override the default segment and tell
4212 the linker to not put the section in any segment at all.
4214 You may use the `FILEHDR' and `PHDRS' keywords after the program
4215 header type to further describe the contents of the segment. The
4216 `FILEHDR' keyword means that the segment should include the ELF file
4217 header. The `PHDRS' keyword means that the segment should include the
4218 ELF program headers themselves. If applied to a loadable segment
4219 (`PT_LOAD'), all prior loadable segments must have one of these
4222 The TYPE may be one of the following. The numbers indicate the
4223 value of the keyword.
4226 Indicates an unused program header.
4229 Indicates that this program header describes a segment to be
4230 loaded from the file.
4233 Indicates a segment where dynamic linking information can be found.
4236 Indicates a segment where the name of the program interpreter may
4240 Indicates a segment holding note information.
4243 A reserved program header type, defined but not specified by the
4247 Indicates a segment where the program headers may be found.
4250 An expression giving the numeric type of the program header. This
4251 may be used for types not defined above.
4253 You can specify that a segment should be loaded at a particular
4254 address in memory by using an `AT' expression. This is identical to the
4255 `AT' command used as an output section attribute (*note Output Section
4256 LMA::). The `AT' command for a program header overrides the output
4259 The linker will normally set the segment flags based on the sections
4260 which comprise the segment. You may use the `FLAGS' keyword to
4261 explicitly specify the segment flags. The value of FLAGS must be an
4262 integer. It is used to set the `p_flags' field of the program header.
4264 Here is an example of `PHDRS'. This shows a typical set of program
4265 headers used on a native ELF system.
4269 headers PT_PHDR PHDRS ;
4271 text PT_LOAD FILEHDR PHDRS ;
4273 dynamic PT_DYNAMIC ;
4279 .interp : { *(.interp) } :text :interp
4280 .text : { *(.text) } :text
4281 .rodata : { *(.rodata) } /* defaults to :text */
4283 . = . + 0x1000; /* move to a new page in memory */
4284 .data : { *(.data) } :data
4285 .dynamic : { *(.dynamic) } :data :dynamic
4290 File: ld.info, Node: VERSION, Next: Expressions, Prev: PHDRS, Up: Scripts
4295 The linker supports symbol versions when using ELF. Symbol versions are
4296 only useful when using shared libraries. The dynamic linker can use
4297 symbol versions to select a specific version of a function when it runs
4298 a program that may have been linked against an earlier version of the
4301 You can include a version script directly in the main linker script,
4302 or you can supply the version script as an implicit linker script. You
4303 can also use the `--version-script' linker option.
4305 The syntax of the `VERSION' command is simply
4306 VERSION { version-script-commands }
4308 The format of the version script commands is identical to that used
4309 by Sun's linker in Solaris 2.5. The version script defines a tree of
4310 version nodes. You specify the node names and interdependencies in the
4311 version script. You can specify which symbols are bound to which
4312 version nodes, and you can reduce a specified set of symbols to local
4313 scope so that they are not globally visible outside of the shared
4316 The easiest way to demonstrate the version script language is with a
4340 This example version script defines three version nodes. The first
4341 version node defined is `VERS_1.1'; it has no other dependencies. The
4342 script binds the symbol `foo1' to `VERS_1.1'. It reduces a number of
4343 symbols to local scope so that they are not visible outside of the
4344 shared library; this is done using wildcard patterns, so that any
4345 symbol whose name begins with `old', `original', or `new' is matched.
4346 The wildcard patterns available are the same as those used in the shell
4347 when matching filenames (also known as "globbing"). However, if you
4348 specify the symbol name inside double quotes, then the name is treated
4349 as literal, rather than as a glob pattern.
4351 Next, the version script defines node `VERS_1.2'. This node depends
4352 upon `VERS_1.1'. The script binds the symbol `foo2' to the version
4355 Finally, the version script defines node `VERS_2.0'. This node
4356 depends upon `VERS_1.2'. The scripts binds the symbols `bar1' and
4357 `bar2' are bound to the version node `VERS_2.0'.
4359 When the linker finds a symbol defined in a library which is not
4360 specifically bound to a version node, it will effectively bind it to an
4361 unspecified base version of the library. You can bind all otherwise
4362 unspecified symbols to a given version node by using `global: *;'
4363 somewhere in the version script. Note that it's slightly crazy to use
4364 wildcards in a global spec except on the last version node. Global
4365 wildcards elsewhere run the risk of accidentally adding symbols to the
4366 set exported for an old version. That's wrong since older versions
4367 ought to have a fixed set of symbols.
4369 The names of the version nodes have no specific meaning other than
4370 what they might suggest to the person reading them. The `2.0' version
4371 could just as well have appeared in between `1.1' and `1.2'. However,
4372 this would be a confusing way to write a version script.
4374 Node name can be omitted, provided it is the only version node in
4375 the version script. Such version script doesn't assign any versions to
4376 symbols, only selects which symbols will be globally visible out and
4379 { global: foo; bar; local: *; };
4381 When you link an application against a shared library that has
4382 versioned symbols, the application itself knows which version of each
4383 symbol it requires, and it also knows which version nodes it needs from
4384 each shared library it is linked against. Thus at runtime, the dynamic
4385 loader can make a quick check to make sure that the libraries you have
4386 linked against do in fact supply all of the version nodes that the
4387 application will need to resolve all of the dynamic symbols. In this
4388 way it is possible for the dynamic linker to know with certainty that
4389 all external symbols that it needs will be resolvable without having to
4390 search for each symbol reference.
4392 The symbol versioning is in effect a much more sophisticated way of
4393 doing minor version checking that SunOS does. The fundamental problem
4394 that is being addressed here is that typically references to external
4395 functions are bound on an as-needed basis, and are not all bound when
4396 the application starts up. If a shared library is out of date, a
4397 required interface may be missing; when the application tries to use
4398 that interface, it may suddenly and unexpectedly fail. With symbol
4399 versioning, the user will get a warning when they start their program if
4400 the libraries being used with the application are too old.
4402 There are several GNU extensions to Sun's versioning approach. The
4403 first of these is the ability to bind a symbol to a version node in the
4404 source file where the symbol is defined instead of in the versioning
4405 script. This was done mainly to reduce the burden on the library
4406 maintainer. You can do this by putting something like:
4407 __asm__(".symver original_foo,foo@VERS_1.1");
4408 in the C source file. This renames the function `original_foo' to
4409 be an alias for `foo' bound to the version node `VERS_1.1'. The
4410 `local:' directive can be used to prevent the symbol `original_foo'
4411 from being exported. A `.symver' directive takes precedence over a
4414 The second GNU extension is to allow multiple versions of the same
4415 function to appear in a given shared library. In this way you can make
4416 an incompatible change to an interface without increasing the major
4417 version number of the shared library, while still allowing applications
4418 linked against the old interface to continue to function.
4420 To do this, you must use multiple `.symver' directives in the source
4421 file. Here is an example:
4423 __asm__(".symver original_foo,foo@");
4424 __asm__(".symver old_foo,foo@VERS_1.1");
4425 __asm__(".symver old_foo1,foo@VERS_1.2");
4426 __asm__(".symver new_foo,foo@@VERS_2.0");
4428 In this example, `foo@' represents the symbol `foo' bound to the
4429 unspecified base version of the symbol. The source file that contains
4430 this example would define 4 C functions: `original_foo', `old_foo',
4431 `old_foo1', and `new_foo'.
4433 When you have multiple definitions of a given symbol, there needs to
4434 be some way to specify a default version to which external references to
4435 this symbol will be bound. You can do this with the `foo@@VERS_2.0'
4436 type of `.symver' directive. You can only declare one version of a
4437 symbol as the default in this manner; otherwise you would effectively
4438 have multiple definitions of the same symbol.
4440 If you wish to bind a reference to a specific version of the symbol
4441 within the shared library, you can use the aliases of convenience
4442 (i.e., `old_foo'), or you can use the `.symver' directive to
4443 specifically bind to an external version of the function in question.
4445 You can also specify the language in the version script:
4447 VERSION extern "lang" { version-script-commands }
4449 The supported `lang's are `C', `C++', and `Java'. The linker will
4450 iterate over the list of symbols at the link time and demangle them
4451 according to `lang' before matching them to the patterns specified in
4452 `version-script-commands'. The default `lang' is `C'.
4454 Demangled names may contains spaces and other special characters. As
4455 described above, you can use a glob pattern to match demangled names,
4456 or you can use a double-quoted string to match the string exactly. In
4457 the latter case, be aware that minor differences (such as differing
4458 whitespace) between the version script and the demangler output will
4459 cause a mismatch. As the exact string generated by the demangler might
4460 change in the future, even if the mangled name does not, you should
4461 check that all of your version directives are behaving as you expect
4465 File: ld.info, Node: Expressions, Next: Implicit Linker Scripts, Prev: VERSION, Up: Scripts
4467 3.10 Expressions in Linker Scripts
4468 ==================================
4470 The syntax for expressions in the linker script language is identical to
4471 that of C expressions. All expressions are evaluated as integers. All
4472 expressions are evaluated in the same size, which is 32 bits if both the
4473 host and target are 32 bits, and is otherwise 64 bits.
4475 You can use and set symbol values in expressions.
4477 The linker defines several special purpose builtin functions for use
4482 * Constants:: Constants
4483 * Symbolic Constants:: Symbolic constants
4484 * Symbols:: Symbol Names
4485 * Orphan Sections:: Orphan Sections
4486 * Location Counter:: The Location Counter
4487 * Operators:: Operators
4488 * Evaluation:: Evaluation
4489 * Expression Section:: The Section of an Expression
4490 * Builtin Functions:: Builtin Functions
4493 File: ld.info, Node: Constants, Next: Symbolic Constants, Up: Expressions
4498 All constants are integers.
4500 As in C, the linker considers an integer beginning with `0' to be
4501 octal, and an integer beginning with `0x' or `0X' to be hexadecimal.
4502 Alternatively the linker accepts suffixes of `h' or `H' for
4503 hexadecimal, `o' or `O' for octal, `b' or `B' for binary and `d' or `D'
4504 for decimal. Any integer value without a prefix or a suffix is
4505 considered to be decimal.
4507 In addition, you can use the suffixes `K' and `M' to scale a
4508 constant by `1024' or `1024*1024' respectively. For example, the
4509 following all refer to the same quantity:
4516 Note - the `K' and `M' suffixes cannot be used in conjunction with
4517 the base suffixes mentioned above.
4520 File: ld.info, Node: Symbolic Constants, Next: Symbols, Prev: Constants, Up: Expressions
4522 3.10.2 Symbolic Constants
4523 -------------------------
4525 It is possible to refer to target specific constants via the use of the
4526 `CONSTANT(NAME)' operator, where NAME is one of:
4529 The target's maximum page size.
4532 The target's default page size.
4536 .text ALIGN (CONSTANT (MAXPAGESIZE)) : { *(.text) }
4538 will create a text section aligned to the largest page boundary
4539 supported by the target.
4542 File: ld.info, Node: Symbols, Next: Orphan Sections, Prev: Symbolic Constants, Up: Expressions
4547 Unless quoted, symbol names start with a letter, underscore, or period
4548 and may include letters, digits, underscores, periods, and hyphens.
4549 Unquoted symbol names must not conflict with any keywords. You can
4550 specify a symbol which contains odd characters or has the same name as a
4551 keyword by surrounding the symbol name in double quotes:
4553 "with a space" = "also with a space" + 10;
4555 Since symbols can contain many non-alphabetic characters, it is
4556 safest to delimit symbols with spaces. For example, `A-B' is one
4557 symbol, whereas `A - B' is an expression involving subtraction.
4560 File: ld.info, Node: Orphan Sections, Next: Location Counter, Prev: Symbols, Up: Expressions
4562 3.10.4 Orphan Sections
4563 ----------------------
4565 Orphan sections are sections present in the input files which are not
4566 explicitly placed into the output file by the linker script. The
4567 linker will still copy these sections into the output file, but it has
4568 to guess as to where they should be placed. The linker uses a simple
4569 heuristic to do this. It attempts to place orphan sections after
4570 non-orphan sections of the same attribute, such as code vs data,
4571 loadable vs non-loadable, etc. If there is not enough room to do this
4572 then it places at the end of the file.
4574 For ELF targets, the attribute of the section includes section type
4575 as well as section flag.
4577 If an orphaned section's name is representable as a C identifier then
4578 the linker will automatically *note PROVIDE:: two symbols:
4579 __start_SECNAME and __stop_SECNAME, where SECNAME is the name of the
4580 section. These indicate the start address and end address of the
4581 orphaned section respectively. Note: most section names are not
4582 representable as C identifiers because they contain a `.' character.
4585 File: ld.info, Node: Location Counter, Next: Operators, Prev: Orphan Sections, Up: Expressions
4587 3.10.5 The Location Counter
4588 ---------------------------
4590 The special linker variable "dot" `.' always contains the current
4591 output location counter. Since the `.' always refers to a location in
4592 an output section, it may only appear in an expression within a
4593 `SECTIONS' command. The `.' symbol may appear anywhere that an
4594 ordinary symbol is allowed in an expression.
4596 Assigning a value to `.' will cause the location counter to be
4597 moved. This may be used to create holes in the output section. The
4598 location counter may not be moved backwards inside an output section,
4599 and may not be moved backwards outside of an output section if so doing
4600 creates areas with overlapping LMAs.
4613 In the previous example, the `.text' section from `file1' is located
4614 at the beginning of the output section `output'. It is followed by a
4615 1000 byte gap. Then the `.text' section from `file2' appears, also
4616 with a 1000 byte gap following before the `.text' section from `file3'.
4617 The notation `= 0x12345678' specifies what data to write in the gaps
4618 (*note Output Section Fill::).
4620 Note: `.' actually refers to the byte offset from the start of the
4621 current containing object. Normally this is the `SECTIONS' statement,
4622 whose start address is 0, hence `.' can be used as an absolute address.
4623 If `.' is used inside a section description however, it refers to the
4624 byte offset from the start of that section, not an absolute address.
4625 Thus in a script like this:
4641 The `.text' section will be assigned a starting address of 0x100 and
4642 a size of exactly 0x200 bytes, even if there is not enough data in the
4643 `.text' input sections to fill this area. (If there is too much data,
4644 an error will be produced because this would be an attempt to move `.'
4645 backwards). The `.data' section will start at 0x500 and it will have
4646 an extra 0x600 bytes worth of space after the end of the values from
4647 the `.data' input sections and before the end of the `.data' output
4650 Setting symbols to the value of the location counter outside of an
4651 output section statement can result in unexpected values if the linker
4652 needs to place orphan sections. For example, given the following:
4665 If the linker needs to place some input section, e.g. `.rodata', not
4666 mentioned in the script, it might choose to place that section between
4667 `.text' and `.data'. You might think the linker should place `.rodata'
4668 on the blank line in the above script, but blank lines are of no
4669 particular significance to the linker. As well, the linker doesn't
4670 associate the above symbol names with their sections. Instead, it
4671 assumes that all assignments or other statements belong to the previous
4672 output section, except for the special case of an assignment to `.'.
4673 I.e., the linker will place the orphan `.rodata' section as if the
4674 script was written as follows:
4683 .rodata: { *(.rodata) }
4688 This may or may not be the script author's intention for the value of
4689 `start_of_data'. One way to influence the orphan section placement is
4690 to assign the location counter to itself, as the linker assumes that an
4691 assignment to `.' is setting the start address of a following output
4692 section and thus should be grouped with that section. So you could
4707 Now, the orphan `.rodata' section will be placed between
4708 `end_of_text' and `start_of_data'.
4711 File: ld.info, Node: Operators, Next: Evaluation, Prev: Location Counter, Up: Expressions
4716 The linker recognizes the standard C set of arithmetic operators, with
4717 the standard bindings and precedence levels:
4718 precedence associativity Operators Notes
4724 5 left == != > < <= >=
4730 11 right &= += -= *= /= (2)
4732 Notes: (1) Prefix operators (2) *Note Assignments::.
4735 File: ld.info, Node: Evaluation, Next: Expression Section, Prev: Operators, Up: Expressions
4740 The linker evaluates expressions lazily. It only computes the value of
4741 an expression when absolutely necessary.
4743 The linker needs some information, such as the value of the start
4744 address of the first section, and the origins and lengths of memory
4745 regions, in order to do any linking at all. These values are computed
4746 as soon as possible when the linker reads in the linker script.
4748 However, other values (such as symbol values) are not known or needed
4749 until after storage allocation. Such values are evaluated later, when
4750 other information (such as the sizes of output sections) is available
4751 for use in the symbol assignment expression.
4753 The sizes of sections cannot be known until after allocation, so
4754 assignments dependent upon these are not performed until after
4757 Some expressions, such as those depending upon the location counter
4758 `.', must be evaluated during section allocation.
4760 If the result of an expression is required, but the value is not
4761 available, then an error results. For example, a script like the
4765 .text 9+this_isnt_constant :
4768 will cause the error message `non constant expression for initial
4772 File: ld.info, Node: Expression Section, Next: Builtin Functions, Prev: Evaluation, Up: Expressions
4774 3.10.8 The Section of an Expression
4775 -----------------------------------
4777 Addresses and symbols may be section relative, or absolute. A section
4778 relative symbol is relocatable. If you request relocatable output
4779 using the `-r' option, a further link operation may change the value of
4780 a section relative symbol. On the other hand, an absolute symbol will
4781 retain the same value throughout any further link operations.
4783 Some terms in linker expressions are addresses. This is true of
4784 section relative symbols and for builtin functions that return an
4785 address, such as `ADDR', `LOADADDR', `ORIGIN' and `SEGMENT_START'.
4786 Other terms are simply numbers, or are builtin functions that return a
4787 non-address value, such as `LENGTH'. One complication is that unless
4788 you set `LD_FEATURE ("SANE_EXPR")' (*note Miscellaneous Commands::),
4789 numbers and absolute symbols are treated differently depending on their
4790 location, for compatibility with older versions of `ld'. Expressions
4791 appearing outside an output section definition treat all numbers as
4792 absolute addresses. Expressions appearing inside an output section
4793 definition treat absolute symbols as numbers. If `LD_FEATURE
4794 ("SANE_EXPR")' is given, then absolute symbols and numbers are simply
4795 treated as numbers everywhere.
4797 In the following simple example,
4802 __executable_start = 0x100;
4806 __data_start = 0x10;
4812 both `.' and `__executable_start' are set to the absolute address
4813 0x100 in the first two assignments, then both `.' and `__data_start'
4814 are set to 0x10 relative to the `.data' section in the second two
4817 For expressions involving numbers, relative addresses and absolute
4818 addresses, ld follows these rules to evaluate terms:
4820 * Unary operations on an absolute address or number, and binary
4821 operations on two absolute addresses or two numbers, or between one
4822 absolute address and a number, apply the operator to the value(s).
4824 * Unary operations on a relative address, and binary operations on
4825 two relative addresses in the same section or between one relative
4826 address and a number, apply the operator to the offset part of the
4829 * Other binary operations, that is, between two relative addresses
4830 not in the same section, or between a relative address and an
4831 absolute address, first convert any non-absolute term to an
4832 absolute address before applying the operator.
4834 The result section of each sub-expression is as follows:
4836 * An operation involving only numbers results in a number.
4838 * The result of comparisons, `&&' and `||' is also a number.
4840 * The result of other binary arithmetic and logical operations on two
4841 relative addresses in the same section or two absolute addresses
4842 (after above conversions) is also a number.
4844 * The result of other operations on relative addresses or one
4845 relative address and a number, is a relative address in the same
4846 section as the relative operand(s).
4848 * The result of other operations on absolute addresses (after above
4849 conversions) is an absolute address.
4851 You can use the builtin function `ABSOLUTE' to force an expression
4852 to be absolute when it would otherwise be relative. For example, to
4853 create an absolute symbol set to the address of the end of the output
4857 .data : { *(.data) _edata = ABSOLUTE(.); }
4859 If `ABSOLUTE' were not used, `_edata' would be relative to the
4862 Using `LOADADDR' also forces an expression absolute, since this
4863 particular builtin function returns an absolute address.
4866 File: ld.info, Node: Builtin Functions, Prev: Expression Section, Up: Expressions
4868 3.10.9 Builtin Functions
4869 ------------------------
4871 The linker script language includes a number of builtin functions for
4872 use in linker script expressions.
4875 Return the absolute (non-relocatable, as opposed to non-negative)
4876 value of the expression EXP. Primarily useful to assign an
4877 absolute value to a symbol within a section definition, where
4878 symbol values are normally section relative. *Note Expression
4882 Return the address (VMA) of the named SECTION. Your script must
4883 previously have defined the location of that section. In the
4884 following example, `start_of_output_1', `symbol_1' and `symbol_2'
4885 are assigned equivalent values, except that `symbol_1' will be
4886 relative to the `.output1' section while the other two will be
4891 start_of_output_1 = ABSOLUTE(.);
4896 symbol_1 = ADDR(.output1);
4897 symbol_2 = start_of_output_1;
4903 Return the location counter (`.') or arbitrary expression aligned
4904 to the next ALIGN boundary. The single operand `ALIGN' doesn't
4905 change the value of the location counter--it just does arithmetic
4906 on it. The two operand `ALIGN' allows an arbitrary expression to
4907 be aligned upwards (`ALIGN(ALIGN)' is equivalent to `ALIGN(.,
4910 Here is an example which aligns the output `.data' section to the
4911 next `0x2000' byte boundary after the preceding section and sets a
4912 variable within the section to the next `0x8000' boundary after the
4915 .data ALIGN(0x2000): {
4917 variable = ALIGN(0x8000);
4920 The first use of `ALIGN' in this example specifies the
4921 location of a section because it is used as the optional ADDRESS
4922 attribute of a section definition (*note Output Section
4923 Address::). The second use of `ALIGN' is used to defines the
4926 The builtin function `NEXT' is closely related to `ALIGN'.
4929 Return the alignment in bytes of the named SECTION, if that
4930 section has been allocated. If the section has not been allocated
4931 when this is evaluated, the linker will report an error. In the
4932 following example, the alignment of the `.output' section is
4933 stored as the first value in that section.
4936 LONG (ALIGNOF (.output))
4942 This is a synonym for `ALIGN', for compatibility with older linker
4943 scripts. It is most often seen when setting the address of an
4946 `DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE)'
4947 This is equivalent to either
4948 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - 1)))
4950 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - COMMONPAGESIZE)))
4951 depending on whether the latter uses fewer COMMONPAGESIZE sized
4952 pages for the data segment (area between the result of this
4953 expression and `DATA_SEGMENT_END') than the former or not. If the
4954 latter form is used, it means COMMONPAGESIZE bytes of runtime
4955 memory will be saved at the expense of up to COMMONPAGESIZE wasted
4956 bytes in the on-disk file.
4958 This expression can only be used directly in `SECTIONS' commands,
4959 not in any output section descriptions and only once in the linker
4960 script. COMMONPAGESIZE should be less or equal to MAXPAGESIZE and
4961 should be the system page size the object wants to be optimized
4962 for (while still working on system page sizes up to MAXPAGESIZE).
4965 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
4967 `DATA_SEGMENT_END(EXP)'
4968 This defines the end of data segment for `DATA_SEGMENT_ALIGN'
4969 evaluation purposes.
4971 . = DATA_SEGMENT_END(.);
4973 `DATA_SEGMENT_RELRO_END(OFFSET, EXP)'
4974 This defines the end of the `PT_GNU_RELRO' segment when `-z relro'
4975 option is used. When `-z relro' option is not present,
4976 `DATA_SEGMENT_RELRO_END' does nothing, otherwise
4977 `DATA_SEGMENT_ALIGN' is padded so that EXP + OFFSET is aligned to
4978 the most commonly used page boundary for particular target. If
4979 present in the linker script, it must always come in between
4980 `DATA_SEGMENT_ALIGN' and `DATA_SEGMENT_END'. Evaluates to the
4981 second argument plus any padding needed at the end of the
4982 `PT_GNU_RELRO' segment due to section alignment.
4984 . = DATA_SEGMENT_RELRO_END(24, .);
4987 Return 1 if SYMBOL is in the linker global symbol table and is
4988 defined before the statement using DEFINED in the script, otherwise
4989 return 0. You can use this function to provide default values for
4990 symbols. For example, the following script fragment shows how to
4991 set a global symbol `begin' to the first location in the `.text'
4992 section--but if a symbol called `begin' already existed, its value
4997 begin = DEFINED(begin) ? begin : . ;
5004 Return the length of the memory region named MEMORY.
5007 Return the absolute LMA of the named SECTION. (*note Output
5011 Return the binary logarithm of EXP rounded towards infinity.
5012 `LOG2CEIL(0)' returns 0.
5015 Returns the maximum of EXP1 and EXP2.
5018 Returns the minimum of EXP1 and EXP2.
5021 Return the next unallocated address that is a multiple of EXP.
5022 This function is closely related to `ALIGN(EXP)'; unless you use
5023 the `MEMORY' command to define discontinuous memory for the output
5024 file, the two functions are equivalent.
5027 Return the origin of the memory region named MEMORY.
5029 `SEGMENT_START(SEGMENT, DEFAULT)'
5030 Return the base address of the named SEGMENT. If an explicit
5031 value has already been given for this segment (with a command-line
5032 `-T' option) then that value will be returned otherwise the value
5033 will be DEFAULT. At present, the `-T' command-line option can
5034 only be used to set the base address for the "text", "data", and
5035 "bss" sections, but you can use `SEGMENT_START' with any segment
5039 Return the size in bytes of the named SECTION, if that section has
5040 been allocated. If the section has not been allocated when this is
5041 evaluated, the linker will report an error. In the following
5042 example, `symbol_1' and `symbol_2' are assigned identical values:
5049 symbol_1 = .end - .start ;
5050 symbol_2 = SIZEOF(.output);
5055 Return the size in bytes of the output file's headers. This is
5056 information which appears at the start of the output file. You
5057 can use this number when setting the start address of the first
5058 section, if you choose, to facilitate paging.
5060 When producing an ELF output file, if the linker script uses the
5061 `SIZEOF_HEADERS' builtin function, the linker must compute the
5062 number of program headers before it has determined all the section
5063 addresses and sizes. If the linker later discovers that it needs
5064 additional program headers, it will report an error `not enough
5065 room for program headers'. To avoid this error, you must avoid
5066 using the `SIZEOF_HEADERS' function, or you must rework your linker
5067 script to avoid forcing the linker to use additional program
5068 headers, or you must define the program headers yourself using the
5069 `PHDRS' command (*note PHDRS::).
5072 File: ld.info, Node: Implicit Linker Scripts, Prev: Expressions, Up: Scripts
5074 3.11 Implicit Linker Scripts
5075 ============================
5077 If you specify a linker input file which the linker can not recognize as
5078 an object file or an archive file, it will try to read the file as a
5079 linker script. If the file can not be parsed as a linker script, the
5080 linker will report an error.
5082 An implicit linker script will not replace the default linker script.
5084 Typically an implicit linker script would contain only symbol
5085 assignments, or the `INPUT', `GROUP', or `VERSION' commands.
5087 Any input files read because of an implicit linker script will be
5088 read at the position in the command line where the implicit linker
5089 script was read. This can affect archive searching.
5092 File: ld.info, Node: Machine Dependent, Next: BFD, Prev: Scripts, Up: Top
5094 4 Machine Dependent Features
5095 ****************************
5097 `ld' has additional features on some platforms; the following sections
5098 describe them. Machines where `ld' has no additional functionality are
5104 * H8/300:: `ld' and the H8/300
5106 * i960:: `ld' and the Intel 960 family
5108 * M68HC11/68HC12:: `ld' and the Motorola 68HC11 and 68HC12 families
5110 * ARM:: `ld' and the ARM family
5112 * HPPA ELF32:: `ld' and HPPA 32-bit ELF
5114 * M68K:: `ld' and the Motorola 68K family
5116 * MIPS:: `ld' and the MIPS family
5118 * MMIX:: `ld' and MMIX
5120 * MSP430:: `ld' and MSP430
5122 * NDS32:: `ld' and NDS32
5124 * Nios II:: `ld' and the Altera Nios II
5126 * PowerPC ELF32:: `ld' and PowerPC 32-bit ELF Support
5128 * PowerPC64 ELF64:: `ld' and PowerPC64 64-bit ELF Support
5130 * SPU ELF:: `ld' and SPU ELF Support
5132 * TI COFF:: `ld' and TI COFF
5134 * WIN32:: `ld' and WIN32 (cygwin/mingw)
5136 * Xtensa:: `ld' and Xtensa Processors
5139 File: ld.info, Node: H8/300, Next: i960, Up: Machine Dependent
5141 4.1 `ld' and the H8/300
5142 =======================
5144 For the H8/300, `ld' can perform these global optimizations when you
5145 specify the `--relax' command-line option.
5147 _relaxing address modes_
5148 `ld' finds all `jsr' and `jmp' instructions whose targets are
5149 within eight bits, and turns them into eight-bit program-counter
5150 relative `bsr' and `bra' instructions, respectively.
5152 _synthesizing instructions_
5153 `ld' finds all `mov.b' instructions which use the sixteen-bit
5154 absolute address form, but refer to the top page of memory, and
5155 changes them to use the eight-bit address form. (That is: the
5156 linker turns `mov.b `@'AA:16' into `mov.b `@'AA:8' whenever the
5157 address AA is in the top page of memory).
5159 `ld' finds all `mov' instructions which use the register indirect
5160 with 32-bit displacement addressing mode, but use a small
5161 displacement inside 16-bit displacement range, and changes them to
5162 use the 16-bit displacement form. (That is: the linker turns
5163 `mov.b `@'D:32,ERx' into `mov.b `@'D:16,ERx' whenever the
5164 displacement D is in the 16 bit signed integer range. Only
5165 implemented in ELF-format ld).
5167 _bit manipulation instructions_
5168 `ld' finds all bit manipulation instructions like `band, bclr,
5169 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst,
5170 bxor' which use 32 bit and 16 bit absolute address form, but refer
5171 to the top page of memory, and changes them to use the 8 bit
5172 address form. (That is: the linker turns `bset #xx:3,`@'AA:32'
5173 into `bset #xx:3,`@'AA:8' whenever the address AA is in the top
5176 _system control instructions_
5177 `ld' finds all `ldc.w, stc.w' instructions which use the 32 bit
5178 absolute address form, but refer to the top page of memory, and
5179 changes them to use 16 bit address form. (That is: the linker
5180 turns `ldc.w `@'AA:32,ccr' into `ldc.w `@'AA:16,ccr' whenever the
5181 address AA is in the top page of memory).
5184 File: ld.info, Node: i960, Next: M68HC11/68HC12, Prev: H8/300, Up: Machine Dependent
5186 4.2 `ld' and the Intel 960 Family
5187 =================================
5189 You can use the `-AARCHITECTURE' command line option to specify one of
5190 the two-letter names identifying members of the 960 family; the option
5191 specifies the desired output target, and warns of any incompatible
5192 instructions in the input files. It also modifies the linker's search
5193 strategy for archive libraries, to support the use of libraries
5194 specific to each particular architecture, by including in the search
5195 loop names suffixed with the string identifying the architecture.
5197 For example, if your `ld' command line included `-ACA' as well as
5198 `-ltry', the linker would look (in its built-in search paths, and in
5199 any paths you specify with `-L') for a library with the names
5206 The first two possibilities would be considered in any event; the last
5207 two are due to the use of `-ACA'.
5209 You can meaningfully use `-A' more than once on a command line, since
5210 the 960 architecture family allows combination of target architectures;
5211 each use will add another pair of name variants to search for when `-l'
5212 specifies a library.
5214 `ld' supports the `--relax' option for the i960 family. If you
5215 specify `--relax', `ld' finds all `balx' and `calx' instructions whose
5216 targets are within 24 bits, and turns them into 24-bit program-counter
5217 relative `bal' and `cal' instructions, respectively. `ld' also turns
5218 `cal' instructions into `bal' instructions when it determines that the
5219 target subroutine is a leaf routine (that is, the target subroutine does
5220 not itself call any subroutines).
5223 File: ld.info, Node: M68HC11/68HC12, Next: ARM, Prev: i960, Up: Machine Dependent
5225 4.3 `ld' and the Motorola 68HC11 and 68HC12 families
5226 ====================================================
5228 4.3.1 Linker Relaxation
5229 -----------------------
5231 For the Motorola 68HC11, `ld' can perform these global optimizations
5232 when you specify the `--relax' command-line option.
5234 _relaxing address modes_
5235 `ld' finds all `jsr' and `jmp' instructions whose targets are
5236 within eight bits, and turns them into eight-bit program-counter
5237 relative `bsr' and `bra' instructions, respectively.
5239 `ld' also looks at all 16-bit extended addressing modes and
5240 transforms them in a direct addressing mode when the address is in
5241 page 0 (between 0 and 0x0ff).
5243 _relaxing gcc instruction group_
5244 When `gcc' is called with `-mrelax', it can emit group of
5245 instructions that the linker can optimize to use a 68HC11 direct
5246 addressing mode. These instructions consists of `bclr' or `bset'
5250 4.3.2 Trampoline Generation
5251 ---------------------------
5253 For 68HC11 and 68HC12, `ld' can generate trampoline code to call a far
5254 function using a normal `jsr' instruction. The linker will also change
5255 the relocation to some far function to use the trampoline address
5256 instead of the function address. This is typically the case when a
5257 pointer to a function is taken. The pointer will in fact point to the
5258 function trampoline.
5261 File: ld.info, Node: ARM, Next: HPPA ELF32, Prev: M68HC11/68HC12, Up: Machine Dependent
5263 4.4 `ld' and the ARM family
5264 ===========================
5266 For the ARM, `ld' will generate code stubs to allow functions calls
5267 between ARM and Thumb code. These stubs only work with code that has
5268 been compiled and assembled with the `-mthumb-interwork' command line
5269 option. If it is necessary to link with old ARM object files or
5270 libraries, which have not been compiled with the -mthumb-interwork
5271 option then the `--support-old-code' command line switch should be
5272 given to the linker. This will make it generate larger stub functions
5273 which will work with non-interworking aware ARM code. Note, however,
5274 the linker does not support generating stubs for function calls to
5275 non-interworking aware Thumb code.
5277 The `--thumb-entry' switch is a duplicate of the generic `--entry'
5278 switch, in that it sets the program's starting address. But it also
5279 sets the bottom bit of the address, so that it can be branched to using
5280 a BX instruction, and the program will start executing in Thumb mode
5283 The `--use-nul-prefixed-import-tables' switch is specifying, that
5284 the import tables idata4 and idata5 have to be generated with a zero
5285 element prefix for import libraries. This is the old style to generate
5286 import tables. By default this option is turned off.
5288 The `--be8' switch instructs `ld' to generate BE8 format
5289 executables. This option is only valid when linking big-endian objects.
5290 The resulting image will contain big-endian data and little-endian code.
5292 The `R_ARM_TARGET1' relocation is typically used for entries in the
5293 `.init_array' section. It is interpreted as either `R_ARM_REL32' or
5294 `R_ARM_ABS32', depending on the target. The `--target1-rel' and
5295 `--target1-abs' switches override the default.
5297 The `--target2=type' switch overrides the default definition of the
5298 `R_ARM_TARGET2' relocation. Valid values for `type', their meanings,
5299 and target defaults are as follows:
5301 `R_ARM_REL32' (arm*-*-elf, arm*-*-eabi)
5304 `R_ARM_ABS32' (arm*-*-symbianelf)
5307 `R_ARM_GOT_PREL' (arm*-*-linux, arm*-*-*bsd)
5309 The `R_ARM_V4BX' relocation (defined by the ARM AAELF specification)
5310 enables objects compiled for the ARMv4 architecture to be
5311 interworking-safe when linked with other objects compiled for ARMv4t,
5312 but also allows pure ARMv4 binaries to be built from the same ARMv4
5315 In the latter case, the switch `--fix-v4bx' must be passed to the
5316 linker, which causes v4t `BX rM' instructions to be rewritten as `MOV
5317 PC,rM', since v4 processors do not have a `BX' instruction.
5319 In the former case, the switch should not be used, and `R_ARM_V4BX'
5320 relocations are ignored.
5322 Replace `BX rM' instructions identified by `R_ARM_V4BX' relocations
5323 with a branch to the following veneer:
5329 This allows generation of libraries/applications that work on ARMv4
5330 cores and are still interworking safe. Note that the above veneer
5331 clobbers the condition flags, so may cause incorrect program behavior
5334 The `--use-blx' switch enables the linker to use ARM/Thumb BLX
5335 instructions (available on ARMv5t and above) in various situations.
5336 Currently it is used to perform calls via the PLT from Thumb code using
5337 BLX rather than using BX and a mode-switching stub before each PLT
5338 entry. This should lead to such calls executing slightly faster.
5340 This option is enabled implicitly for SymbianOS, so there is no need
5341 to specify it if you are using that target.
5343 The `--vfp11-denorm-fix' switch enables a link-time workaround for a
5344 bug in certain VFP11 coprocessor hardware, which sometimes allows
5345 instructions with denorm operands (which must be handled by support
5346 code) to have those operands overwritten by subsequent instructions
5347 before the support code can read the intended values.
5349 The bug may be avoided in scalar mode if you allow at least one
5350 intervening instruction between a VFP11 instruction which uses a
5351 register and another instruction which writes to the same register, or
5352 at least two intervening instructions if vector mode is in use. The bug
5353 only affects full-compliance floating-point mode: you do not need this
5354 workaround if you are using "runfast" mode. Please contact ARM for
5357 If you know you are using buggy VFP11 hardware, you can enable this
5358 workaround by specifying the linker option `--vfp-denorm-fix=scalar' if
5359 you are using the VFP11 scalar mode only, or `--vfp-denorm-fix=vector'
5360 if you are using vector mode (the latter also works for scalar code).
5361 The default is `--vfp-denorm-fix=none'.
5363 If the workaround is enabled, instructions are scanned for
5364 potentially-troublesome sequences, and a veneer is created for each
5365 such sequence which may trigger the erratum. The veneer consists of the
5366 first instruction of the sequence and a branch back to the subsequent
5367 instruction. The original instruction is then replaced with a branch to
5368 the veneer. The extra cycles required to call and return from the veneer
5369 are sufficient to avoid the erratum in both the scalar and vector cases.
5371 The `--fix-arm1176' switch enables a link-time workaround for an
5372 erratum in certain ARM1176 processors. The workaround is enabled by
5373 default if you are targeting ARM v6 (excluding ARM v6T2) or earlier.
5374 It can be disabled unconditionally by specifying `--no-fix-arm1176'.
5376 Further information is available in the "ARM1176JZ-S and ARM1176JZF-S
5377 Programmer Advice Notice" available on the ARM documentation website at:
5378 http://infocenter.arm.com/.
5380 The `--no-enum-size-warning' switch prevents the linker from warning
5381 when linking object files that specify incompatible EABI enumeration
5382 size attributes. For example, with this switch enabled, linking of an
5383 object file using 32-bit enumeration values with another using
5384 enumeration values fitted into the smallest possible space will not be
5387 The `--no-wchar-size-warning' switch prevents the linker from
5388 warning when linking object files that specify incompatible EABI
5389 `wchar_t' size attributes. For example, with this switch enabled,
5390 linking of an object file using 32-bit `wchar_t' values with another
5391 using 16-bit `wchar_t' values will not be diagnosed.
5393 The `--pic-veneer' switch makes the linker use PIC sequences for
5394 ARM/Thumb interworking veneers, even if the rest of the binary is not
5395 PIC. This avoids problems on uClinux targets where `--emit-relocs' is
5396 used to generate relocatable binaries.
5398 The linker will automatically generate and insert small sequences of
5399 code into a linked ARM ELF executable whenever an attempt is made to
5400 perform a function call to a symbol that is too far away. The
5401 placement of these sequences of instructions - called stubs - is
5402 controlled by the command line option `--stub-group-size=N'. The
5403 placement is important because a poor choice can create a need for
5404 duplicate stubs, increasing the code size. The linker will try to
5405 group stubs together in order to reduce interruptions to the flow of
5406 code, but it needs guidance as to how big these groups should be and
5407 where they should be placed.
5409 The value of `N', the parameter to the `--stub-group-size=' option
5410 controls where the stub groups are placed. If it is negative then all
5411 stubs are placed after the first branch that needs them. If it is
5412 positive then the stubs can be placed either before or after the
5413 branches that need them. If the value of `N' is 1 (either +1 or -1)
5414 then the linker will choose exactly where to place groups of stubs,
5415 using its built in heuristics. A value of `N' greater than 1 (or
5416 smaller than -1) tells the linker that a single group of stubs can
5417 service at most `N' bytes from the input sections.
5419 The default, if `--stub-group-size=' is not specified, is `N = +1'.
5421 Farcalls stubs insertion is fully supported for the ARM-EABI target
5422 only, because it relies on object files properties not present
5425 The `--fix-cortex-a8' switch enables a link-time workaround for an
5426 erratum in certain Cortex-A8 processors. The workaround is enabled by
5427 default if you are targeting the ARM v7-A architecture profile. It can
5428 be enabled otherwise by specifying `--fix-cortex-a8', or disabled
5429 unconditionally by specifying `--no-fix-cortex-a8'.
5431 The erratum only affects Thumb-2 code. Please contact ARM for
5434 The `--fix-cortex-a53-835769' switch enables a link-time workaround
5435 for erratum 835769 present on certain early revisions of Cortex-A53
5436 processors. The workaround is disabled by default. It can be enabled
5437 by specifying `--fix-cortex-a53-835769', or disabled unconditionally by
5438 specifying `--no-fix-cortex-a53-835769'.
5440 Please contact ARM for further details.
5442 The `--no-merge-exidx-entries' switch disables the merging of
5443 adjacent exidx entries in debuginfo.
5445 The `--long-plt' option enables the use of 16 byte PLT entries which
5446 support up to 4Gb of code. The default is to use 12 byte PLT entries
5447 which only support 512Mb of code.
5450 File: ld.info, Node: HPPA ELF32, Next: M68K, Prev: ARM, Up: Machine Dependent
5452 4.5 `ld' and HPPA 32-bit ELF Support
5453 ====================================
5455 When generating a shared library, `ld' will by default generate import
5456 stubs suitable for use with a single sub-space application. The
5457 `--multi-subspace' switch causes `ld' to generate export stubs, and
5458 different (larger) import stubs suitable for use with multiple
5461 Long branch stubs and import/export stubs are placed by `ld' in stub
5462 sections located between groups of input sections. `--stub-group-size'
5463 specifies the maximum size of a group of input sections handled by one
5464 stub section. Since branch offsets are signed, a stub section may
5465 serve two groups of input sections, one group before the stub section,
5466 and one group after it. However, when using conditional branches that
5467 require stubs, it may be better (for branch prediction) that stub
5468 sections only serve one group of input sections. A negative value for
5469 `N' chooses this scheme, ensuring that branches to stubs always use a
5470 negative offset. Two special values of `N' are recognized, `1' and
5471 `-1'. These both instruct `ld' to automatically size input section
5472 groups for the branch types detected, with the same behaviour regarding
5473 stub placement as other positive or negative values of `N' respectively.
5475 Note that `--stub-group-size' does not split input sections. A
5476 single input section larger than the group size specified will of course
5477 create a larger group (of one section). If input sections are too
5478 large, it may not be possible for a branch to reach its stub.
5481 File: ld.info, Node: M68K, Next: MIPS, Prev: HPPA ELF32, Up: Machine Dependent
5483 4.6 `ld' and the Motorola 68K family
5484 ====================================
5486 The `--got=TYPE' option lets you choose the GOT generation scheme. The
5487 choices are `single', `negative', `multigot' and `target'. When
5488 `target' is selected the linker chooses the default GOT generation
5489 scheme for the current target. `single' tells the linker to generate a
5490 single GOT with entries only at non-negative offsets. `negative'
5491 instructs the linker to generate a single GOT with entries at both
5492 negative and positive offsets. Not all environments support such GOTs.
5493 `multigot' allows the linker to generate several GOTs in the output
5494 file. All GOT references from a single input object file access the
5495 same GOT, but references from different input object files might access
5496 different GOTs. Not all environments support such GOTs.
5499 File: ld.info, Node: MIPS, Next: MMIX, Prev: M68K, Up: Machine Dependent
5501 4.7 `ld' and the MIPS family
5502 ============================
5504 The `--insn32' and `--no-insn32' options control the choice of
5505 microMIPS instructions used in code generated by the linker, such as
5506 that in the PLT or lazy binding stubs, or in relaxation. If `--insn32'
5507 is used, then the linker only uses 32-bit instruction encodings. By
5508 default or if `--no-insn32' is used, all instruction encodings are used,
5509 including 16-bit ones where possible.
5512 File: ld.info, Node: MMIX, Next: MSP430, Prev: MIPS, Up: Machine Dependent
5517 For MMIX, there is a choice of generating `ELF' object files or `mmo'
5518 object files when linking. The simulator `mmix' understands the `mmo'
5519 format. The binutils `objcopy' utility can translate between the two
5522 There is one special section, the `.MMIX.reg_contents' section.
5523 Contents in this section is assumed to correspond to that of global
5524 registers, and symbols referring to it are translated to special
5525 symbols, equal to registers. In a final link, the start address of the
5526 `.MMIX.reg_contents' section corresponds to the first allocated global
5527 register multiplied by 8. Register `$255' is not included in this
5528 section; it is always set to the program entry, which is at the symbol
5529 `Main' for `mmo' files.
5531 Global symbols with the prefix `__.MMIX.start.', for example
5532 `__.MMIX.start..text' and `__.MMIX.start..data' are special. The
5533 default linker script uses these to set the default start address of a
5536 Initial and trailing multiples of zero-valued 32-bit words in a
5537 section, are left out from an mmo file.
5540 File: ld.info, Node: MSP430, Next: NDS32, Prev: MMIX, Up: Machine Dependent
5545 For the MSP430 it is possible to select the MPU architecture. The flag
5546 `-m [mpu type]' will select an appropriate linker script for selected
5547 MPU type. (To get a list of known MPUs just pass `-m help' option to
5550 The linker will recognize some extra sections which are MSP430
5554 Defines a portion of ROM where interrupt vectors located.
5557 Defines the bootloader portion of the ROM (if applicable). Any
5558 code in this section will be uploaded to the MPU.
5561 Defines an information memory section (if applicable). Any code in
5562 this section will be uploaded to the MPU.
5565 This is the same as the `.infomem' section except that any code in
5566 this section will not be uploaded to the MPU.
5569 Denotes a portion of RAM located above `.bss' section.
5571 The last two sections are used by gcc.
5574 File: ld.info, Node: NDS32, Next: Nios II, Prev: MSP430, Up: Machine Dependent
5579 For NDS32, there are some options to select relaxation behavior. The
5580 linker relaxes objects according to these options.
5582 ``--m[no-]fp-as-gp''
5583 Disable/enable fp-as-gp relaxation.
5585 ``--mexport-symbols=FILE''
5586 Exporting symbols and their address into FILE as linker script.
5589 Disable/enable link-time EX9 relaxation.
5591 ``--mexport-ex9=FILE''
5592 Export the EX9 table after linking.
5594 ``--mimport-ex9=FILE''
5595 Import the Ex9 table for EX9 relaxation.
5598 Update the existing EX9 table.
5600 ``--mex9-limit=NUM''
5601 Maximum number of entries in the ex9 table.
5603 ``--mex9-loop-aware''
5604 Avoid generating the EX9 instruction inside the loop.
5607 Disable/enable the link-time IFC optimization.
5609 ``--mifc-loop-aware''
5610 Avoid generating the IFC instruction inside the loop.
5613 File: ld.info, Node: Nios II, Next: PowerPC ELF32, Prev: NDS32, Up: Machine Dependent
5615 4.11 `ld' and the Altera Nios II
5616 ================================
5618 Call and immediate jump instructions on Nios II processors are limited
5619 to transferring control to addresses in the same 256MB memory segment,
5620 which may result in `ld' giving `relocation truncated to fit' errors
5621 with very large programs. The command-line option `--relax' enables
5622 the generation of trampolines that can access the entire 32-bit address
5623 space for calls outside the normal `call' and `jmpi' address range.
5624 These trampolines are inserted at section boundaries, so may not
5625 themselves be reachable if an input section and its associated call
5626 trampolines are larger than 256MB.
5628 The `--relax' option is enabled by default unless `-r' is also
5629 specified. You can disable trampoline generation by using the
5630 `--no-relax' linker option. You can also disable this optimization
5631 locally by using the `set .noat' directive in assembly-language source
5632 files, as the linker-inserted trampolines use the `at' register as a
5635 Note that the linker `--relax' option is independent of assembler
5636 relaxation options, and that using the GNU assembler's `-relax-all'
5637 option interferes with the linker's more selective call instruction
5641 File: ld.info, Node: PowerPC ELF32, Next: PowerPC64 ELF64, Prev: Nios II, Up: Machine Dependent
5643 4.12 `ld' and PowerPC 32-bit ELF Support
5644 ========================================
5646 Branches on PowerPC processors are limited to a signed 26-bit
5647 displacement, which may result in `ld' giving `relocation truncated to
5648 fit' errors with very large programs. `--relax' enables the generation
5649 of trampolines that can access the entire 32-bit address space. These
5650 trampolines are inserted at section boundaries, so may not themselves
5651 be reachable if an input section exceeds 33M in size. You may combine
5652 `-r' and `--relax' to add trampolines in a partial link. In that case
5653 both branches to undefined symbols and inter-section branches are also
5654 considered potentially out of range, and trampolines inserted.
5657 Current PowerPC GCC accepts a `-msecure-plt' option that generates
5658 code capable of using a newer PLT and GOT layout that has the
5659 security advantage of no executable section ever needing to be
5660 writable and no writable section ever being executable. PowerPC
5661 `ld' will generate this layout, including stubs to access the PLT,
5662 if all input files (including startup and static libraries) were
5663 compiled with `-msecure-plt'. `--bss-plt' forces the old BSS PLT
5664 (and GOT layout) which can give slightly better performance.
5667 `ld' will use the new PLT and GOT layout if it is linking new
5668 `-fpic' or `-fPIC' code, but does not do so automatically when
5669 linking non-PIC code. This option requests the new PLT and GOT
5670 layout. A warning will be given if some object file requires the
5674 The new secure PLT and GOT are placed differently relative to other
5675 sections compared to older BSS PLT and GOT placement. The
5676 location of `.plt' must change because the new secure PLT is an
5677 initialized section while the old PLT is uninitialized. The
5678 reason for the `.got' change is more subtle: The new placement
5679 allows `.got' to be read-only in applications linked with `-z
5680 relro -z now'. However, this placement means that `.sdata' cannot
5681 always be used in shared libraries, because the PowerPC ABI
5682 accesses `.sdata' in shared libraries from the GOT pointer.
5683 `--sdata-got' forces the old GOT placement. PowerPC GCC doesn't
5684 use `.sdata' in shared libraries, so this option is really only
5685 useful for other compilers that may do so.
5688 This option causes `ld' to label linker stubs with a local symbol
5689 that encodes the stub type and destination.
5692 PowerPC `ld' normally performs some optimization of code sequences
5693 used to access Thread-Local Storage. Use this option to disable
5697 File: ld.info, Node: PowerPC64 ELF64, Next: SPU ELF, Prev: PowerPC ELF32, Up: Machine Dependent
5699 4.13 `ld' and PowerPC64 64-bit ELF Support
5700 ==========================================
5703 Long branch stubs, PLT call stubs and TOC adjusting stubs are
5704 placed by `ld' in stub sections located between groups of input
5705 sections. `--stub-group-size' specifies the maximum size of a
5706 group of input sections handled by one stub section. Since branch
5707 offsets are signed, a stub section may serve two groups of input
5708 sections, one group before the stub section, and one group after
5709 it. However, when using conditional branches that require stubs,
5710 it may be better (for branch prediction) that stub sections only
5711 serve one group of input sections. A negative value for `N'
5712 chooses this scheme, ensuring that branches to stubs always use a
5713 negative offset. Two special values of `N' are recognized, `1'
5714 and `-1'. These both instruct `ld' to automatically size input
5715 section groups for the branch types detected, with the same
5716 behaviour regarding stub placement as other positive or negative
5717 values of `N' respectively.
5719 Note that `--stub-group-size' does not split input sections. A
5720 single input section larger than the group size specified will of
5721 course create a larger group (of one section). If input sections
5722 are too large, it may not be possible for a branch to reach its
5726 This option causes `ld' to label linker stubs with a local symbol
5727 that encodes the stub type and destination.
5729 `--dotsyms, --no-dotsyms'
5730 These two options control how `ld' interprets version patterns in
5731 a version script. Older PowerPC64 compilers emitted both a
5732 function descriptor symbol with the same name as the function, and
5733 a code entry symbol with the name prefixed by a dot (`.'). To
5734 properly version a function `foo', the version script thus needs
5735 to control both `foo' and `.foo'. The option `--dotsyms', on by
5736 default, automatically adds the required dot-prefixed patterns.
5737 Use `--no-dotsyms' to disable this feature.
5740 PowerPC64 `ld' normally performs some optimization of code
5741 sequences used to access Thread-Local Storage. Use this option to
5742 disable the optimization.
5745 PowerPC64 `ld' normally removes `.opd' section entries
5746 corresponding to deleted link-once functions, or functions removed
5747 by the action of `--gc-sections' or linker script `/DISCARD/'.
5748 Use this option to disable `.opd' optimization.
5750 `--non-overlapping-opd'
5751 Some PowerPC64 compilers have an option to generate compressed
5752 `.opd' entries spaced 16 bytes apart, overlapping the third word,
5753 the static chain pointer (unused in C) with the first word of the
5754 next entry. This option expands such entries to the full 24 bytes.
5757 PowerPC64 `ld' normally removes unused `.toc' section entries.
5758 Such entries are detected by examining relocations that reference
5759 the TOC in code sections. A reloc in a deleted code section marks
5760 a TOC word as unneeded, while a reloc in a kept code section marks
5761 a TOC word as needed. Since the TOC may reference itself, TOC
5762 relocs are also examined. TOC words marked as both needed and
5763 unneeded will of course be kept. TOC words without any referencing
5764 reloc are assumed to be part of a multi-word entry, and are kept or
5765 discarded as per the nearest marked preceding word. This works
5766 reliably for compiler generated code, but may be incorrect if
5767 assembly code is used to insert TOC entries. Use this option to
5768 disable the optimization.
5771 If given any toc option besides `-mcmodel=medium' or
5772 `-mcmodel=large', PowerPC64 GCC generates code for a TOC model
5773 where TOC entries are accessed with a 16-bit offset from r2. This
5774 limits the total TOC size to 64K. PowerPC64 `ld' extends this
5775 limit by grouping code sections such that each group uses less
5776 than 64K for its TOC entries, then inserts r2 adjusting stubs
5777 between inter-group calls. `ld' does not split apart input
5778 sections, so cannot help if a single input file has a `.toc'
5779 section that exceeds 64K, most likely from linking multiple files
5780 with `ld -r'. Use this option to turn off this feature.
5783 By default, `ld' sorts TOC sections so that those whose file
5784 happens to have a section called `.init' or `.fini' are placed
5785 first, followed by TOC sections referenced by code generated with
5786 PowerPC64 gcc's `-mcmodel=small', and lastly TOC sections
5787 referenced only by code generated with PowerPC64 gcc's
5788 `-mcmodel=medium' or `-mcmodel=large' options. Doing this results
5789 in better TOC grouping for multi-TOC. Use this option to turn off
5794 Use these options to control whether individual PLT call stubs are
5795 padded so that they don't cross a 32-byte boundary, or to the
5796 specified power of two boundary when using `--plt-align='. Note
5797 that this isn't alignment in the usual sense. By default PLT call
5798 stubs are packed tightly.
5800 `--plt-static-chain'
5801 `--no-plt-static-chain'
5802 Use these options to control whether PLT call stubs load the static
5803 chain pointer (r11). `ld' defaults to not loading the static
5804 chain since there is never any need to do so on a PLT call.
5808 With power7's weakly ordered memory model, it is possible when
5809 using lazy binding for ld.so to update a plt entry in one thread
5810 and have another thread see the individual plt entry words update
5811 in the wrong order, despite ld.so carefully writing in the correct
5812 order and using memory write barriers. To avoid this we need some
5813 sort of read barrier in the call stub, or use LD_BIND_NOW=1. By
5814 default, `ld' looks for calls to commonly used functions that
5815 create threads, and if seen, adds the necessary barriers. Use
5816 these options to change the default behaviour.
5819 File: ld.info, Node: SPU ELF, Next: TI COFF, Prev: PowerPC64 ELF64, Up: Machine Dependent
5821 4.14 `ld' and SPU ELF Support
5822 =============================
5825 This option marks an executable as a PIC plugin module.
5828 Normally, `ld' recognizes calls to functions within overlay
5829 regions, and redirects such calls to an overlay manager via a stub.
5830 `ld' also provides a built-in overlay manager. This option turns
5831 off all this special overlay handling.
5834 This option causes `ld' to label overlay stubs with a local symbol
5835 that encodes the stub type and destination.
5837 `--extra-overlay-stubs'
5838 This option causes `ld' to add overlay call stubs on all function
5839 calls out of overlay regions. Normally stubs are not added on
5840 calls to non-overlay regions.
5842 `--local-store=lo:hi'
5843 `ld' usually checks that a final executable for SPU fits in the
5844 address range 0 to 256k. This option may be used to change the
5845 range. Disable the check entirely with `--local-store=0:0'.
5848 SPU local store space is limited. Over-allocation of stack space
5849 unnecessarily limits space available for code and data, while
5850 under-allocation results in runtime failures. If given this
5851 option, `ld' will provide an estimate of maximum stack usage.
5852 `ld' does this by examining symbols in code sections to determine
5853 the extents of functions, and looking at function prologues for
5854 stack adjusting instructions. A call-graph is created by looking
5855 for relocations on branch instructions. The graph is then searched
5856 for the maximum stack usage path. Note that this analysis does not
5857 find calls made via function pointers, and does not handle
5858 recursion and other cycles in the call graph. Stack usage may be
5859 under-estimated if your code makes such calls. Also, stack usage
5860 for dynamic allocation, e.g. alloca, will not be detected. If a
5861 link map is requested, detailed information about each function's
5862 stack usage and calls will be given.
5865 This option, if given along with `--stack-analysis' will result in
5866 `ld' emitting stack sizing symbols for each function. These take
5867 the form `__stack_<function_name>' for global functions, and
5868 `__stack_<number>_<function_name>' for static functions.
5869 `<number>' is the section id in hex. The value of such symbols is
5870 the stack requirement for the corresponding function. The symbol
5871 size will be zero, type `STT_NOTYPE', binding `STB_LOCAL', and
5875 File: ld.info, Node: TI COFF, Next: WIN32, Prev: SPU ELF, Up: Machine Dependent
5877 4.15 `ld''s Support for Various TI COFF Versions
5878 ================================================
5880 The `--format' switch allows selection of one of the various TI COFF
5881 versions. The latest of this writing is 2; versions 0 and 1 are also
5882 supported. The TI COFF versions also vary in header byte-order format;
5883 `ld' will read any version or byte order, but the output header format
5884 depends on the default specified by the specific target.
5887 File: ld.info, Node: WIN32, Next: Xtensa, Prev: TI COFF, Up: Machine Dependent
5889 4.16 `ld' and WIN32 (cygwin/mingw)
5890 ==================================
5892 This section describes some of the win32 specific `ld' issues. See
5893 *Note Command Line Options: Options. for detailed description of the
5894 command line options mentioned here.
5897 The standard Windows linker creates and uses so-called import
5898 libraries, which contains information for linking to dll's. They
5899 are regular static archives and are handled as any other static
5900 archive. The cygwin and mingw ports of `ld' have specific support
5901 for creating such libraries provided with the `--out-implib'
5902 command line option.
5904 _exporting DLL symbols_
5905 The cygwin/mingw `ld' has several ways to export symbols for dll's.
5907 _using auto-export functionality_
5908 By default `ld' exports symbols with the auto-export
5909 functionality, which is controlled by the following command
5912 * -export-all-symbols [This is the default]
5918 * -exclude-modules-for-implib
5922 When auto-export is in operation, `ld' will export all the
5923 non-local (global and common) symbols it finds in a DLL, with
5924 the exception of a few symbols known to belong to the
5925 system's runtime and libraries. As it will often not be
5926 desirable to export all of a DLL's symbols, which may include
5927 private functions that are not part of any public interface,
5928 the command-line options listed above may be used to filter
5929 symbols out from the list for exporting. The `--output-def'
5930 option can be used in order to see the final list of exported
5931 symbols with all exclusions taken into effect.
5933 If `--export-all-symbols' is not given explicitly on the
5934 command line, then the default auto-export behavior will be
5935 _disabled_ if either of the following are true:
5937 * A DEF file is used.
5939 * Any symbol in any object file was marked with the
5940 __declspec(dllexport) attribute.
5943 Another way of exporting symbols is using a DEF file. A DEF
5944 file is an ASCII file containing definitions of symbols which
5945 should be exported when a dll is created. Usually it is
5946 named `<dll name>.def' and is added as any other object file
5947 to the linker's command line. The file's name must end in
5950 gcc -o <output> <objectfiles> <dll name>.def
5952 Using a DEF file turns off the normal auto-export behavior,
5953 unless the `--export-all-symbols' option is also used.
5955 Here is an example of a DEF file for a shared library called
5958 LIBRARY "xyz.dll" BASE=0x20000000
5964 another_foo = abc.dll.afoo
5969 This example defines a DLL with a non-default base address
5970 and seven symbols in the export table. The third exported
5971 symbol `_bar' is an alias for the second. The fourth symbol,
5972 `another_foo' is resolved by "forwarding" to another module
5973 and treating it as an alias for `afoo' exported from the DLL
5974 `abc.dll'. The final symbol `var1' is declared to be a data
5975 object. The `doo' symbol in export library is an alias of
5976 `foo', which gets the string name in export table `foo2'. The
5977 `eoo' symbol is an data export symbol, which gets in export
5978 table the name `var1'.
5980 The optional `LIBRARY <name>' command indicates the _internal_
5981 name of the output DLL. If `<name>' does not include a suffix,
5982 the default library suffix, `.DLL' is appended.
5984 When the .DEF file is used to build an application, rather
5985 than a library, the `NAME <name>' command should be used
5986 instead of `LIBRARY'. If `<name>' does not include a suffix,
5987 the default executable suffix, `.EXE' is appended.
5989 With either `LIBRARY <name>' or `NAME <name>' the optional
5990 specification `BASE = <number>' may be used to specify a
5991 non-default base address for the image.
5993 If neither `LIBRARY <name>' nor `NAME <name>' is specified,
5994 or they specify an empty string, the internal name is the
5995 same as the filename specified on the command line.
5997 The complete specification of an export symbol is:
6000 ( ( ( <name1> [ = <name2> ] )
6001 | ( <name1> = <module-name> . <external-name>))
6002 [ @ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
6004 Declares `<name1>' as an exported symbol from the DLL, or
6005 declares `<name1>' as an exported alias for `<name2>'; or
6006 declares `<name1>' as a "forward" alias for the symbol
6007 `<external-name>' in the DLL `<module-name>'. Optionally,
6008 the symbol may be exported by the specified ordinal
6009 `<integer>' alias. The optional `<name3>' is the to be used
6010 string in import/export table for the symbol.
6012 The optional keywords that follow the declaration indicate:
6014 `NONAME': Do not put the symbol name in the DLL's export
6015 table. It will still be exported by its ordinal alias
6016 (either the value specified by the .def specification or,
6017 otherwise, the value assigned by the linker). The symbol
6018 name, however, does remain visible in the import library (if
6019 any), unless `PRIVATE' is also specified.
6021 `DATA': The symbol is a variable or object, rather than a
6022 function. The import lib will export only an indirect
6023 reference to `foo' as the symbol `_imp__foo' (ie, `foo' must
6024 be resolved as `*_imp__foo').
6026 `CONSTANT': Like `DATA', but put the undecorated `foo' as
6027 well as `_imp__foo' into the import library. Both refer to the
6028 read-only import address table's pointer to the variable, not
6029 to the variable itself. This can be dangerous. If the user
6030 code fails to add the `dllimport' attribute and also fails to
6031 explicitly add the extra indirection that the use of the
6032 attribute enforces, the application will behave unexpectedly.
6034 `PRIVATE': Put the symbol in the DLL's export table, but do
6035 not put it into the static import library used to resolve
6036 imports at link time. The symbol can still be imported using
6037 the `LoadLibrary/GetProcAddress' API at runtime or by by
6038 using the GNU ld extension of linking directly to the DLL
6039 without an import library.
6041 See ld/deffilep.y in the binutils sources for the full
6042 specification of other DEF file statements
6044 While linking a shared dll, `ld' is able to create a DEF file
6045 with the `--output-def <file>' command line option.
6048 Another way of marking symbols for export is to modify the
6049 source code itself, so that when building the DLL each symbol
6050 to be exported is declared as:
6052 __declspec(dllexport) int a_variable
6053 __declspec(dllexport) void a_function(int with_args)
6055 All such symbols will be exported from the DLL. If, however,
6056 any of the object files in the DLL contain symbols decorated
6057 in this way, then the normal auto-export behavior is
6058 disabled, unless the `--export-all-symbols' option is also
6061 Note that object files that wish to access these symbols must
6062 _not_ decorate them with dllexport. Instead, they should use
6065 __declspec(dllimport) int a_variable
6066 __declspec(dllimport) void a_function(int with_args)
6068 This complicates the structure of library header files,
6069 because when included by the library itself the header must
6070 declare the variables and functions as dllexport, but when
6071 included by client code the header must declare them as
6072 dllimport. There are a number of idioms that are typically
6073 used to do this; often client code can omit the __declspec()
6074 declaration completely. See `--enable-auto-import' and
6075 `automatic data imports' for more information.
6077 _automatic data imports_
6078 The standard Windows dll format supports data imports from dlls
6079 only by adding special decorations (dllimport/dllexport), which
6080 let the compiler produce specific assembler instructions to deal
6081 with this issue. This increases the effort necessary to port
6082 existing Un*x code to these platforms, especially for large c++
6083 libraries and applications. The auto-import feature, which was
6084 initially provided by Paul Sokolovsky, allows one to omit the
6085 decorations to achieve a behavior that conforms to that on
6086 POSIX/Un*x platforms. This feature is enabled with the
6087 `--enable-auto-import' command-line option, although it is enabled
6088 by default on cygwin/mingw. The `--enable-auto-import' option
6089 itself now serves mainly to suppress any warnings that are
6090 ordinarily emitted when linked objects trigger the feature's use.
6092 auto-import of variables does not always work flawlessly without
6093 additional assistance. Sometimes, you will see this message
6095 "variable '<var>' can't be auto-imported. Please read the
6096 documentation for ld's `--enable-auto-import' for details."
6098 The `--enable-auto-import' documentation explains why this error
6099 occurs, and several methods that can be used to overcome this
6100 difficulty. One of these methods is the _runtime pseudo-relocs_
6101 feature, described below.
6103 For complex variables imported from DLLs (such as structs or
6104 classes), object files typically contain a base address for the
6105 variable and an offset (_addend_) within the variable-to specify a
6106 particular field or public member, for instance. Unfortunately,
6107 the runtime loader used in win32 environments is incapable of
6108 fixing these references at runtime without the additional
6109 information supplied by dllimport/dllexport decorations. The
6110 standard auto-import feature described above is unable to resolve
6113 The `--enable-runtime-pseudo-relocs' switch allows these
6114 references to be resolved without error, while leaving the task of
6115 adjusting the references themselves (with their non-zero addends)
6116 to specialized code provided by the runtime environment. Recent
6117 versions of the cygwin and mingw environments and compilers
6118 provide this runtime support; older versions do not. However, the
6119 support is only necessary on the developer's platform; the
6120 compiled result will run without error on an older system.
6122 `--enable-runtime-pseudo-relocs' is not the default; it must be
6123 explicitly enabled as needed.
6125 _direct linking to a dll_
6126 The cygwin/mingw ports of `ld' support the direct linking,
6127 including data symbols, to a dll without the usage of any import
6128 libraries. This is much faster and uses much less memory than
6129 does the traditional import library method, especially when
6130 linking large libraries or applications. When `ld' creates an
6131 import lib, each function or variable exported from the dll is
6132 stored in its own bfd, even though a single bfd could contain many
6133 exports. The overhead involved in storing, loading, and
6134 processing so many bfd's is quite large, and explains the
6135 tremendous time, memory, and storage needed to link against
6136 particularly large or complex libraries when using import libs.
6138 Linking directly to a dll uses no extra command-line switches
6139 other than `-L' and `-l', because `ld' already searches for a
6140 number of names to match each library. All that is needed from
6141 the developer's perspective is an understanding of this search, in
6142 order to force ld to select the dll instead of an import library.
6144 For instance, when ld is called with the argument `-lxxx' it will
6145 attempt to find, in the first directory of its search path,
6155 before moving on to the next directory in the search path.
6157 (*) Actually, this is not `cygxxx.dll' but in fact is
6158 `<prefix>xxx.dll', where `<prefix>' is set by the `ld' option
6159 `--dll-search-prefix=<prefix>'. In the case of cygwin, the
6160 standard gcc spec file includes `--dll-search-prefix=cyg', so in
6161 effect we actually search for `cygxxx.dll'.
6163 Other win32-based unix environments, such as mingw or pw32, may
6164 use other `<prefix>'es, although at present only cygwin makes use
6165 of this feature. It was originally intended to help avoid name
6166 conflicts among dll's built for the various win32/un*x
6167 environments, so that (for example) two versions of a zlib dll
6168 could coexist on the same machine.
6170 The generic cygwin/mingw path layout uses a `bin' directory for
6171 applications and dll's and a `lib' directory for the import
6172 libraries (using cygwin nomenclature):
6177 libxxx.dll.a (in case of dll's)
6178 libxxx.a (in case of static archive)
6180 Linking directly to a dll without using the import library can be
6183 1. Use the dll directly by adding the `bin' path to the link line
6184 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6186 However, as the dll's often have version numbers appended to their
6187 names (`cygncurses-5.dll') this will often fail, unless one
6188 specifies `-L../bin -lncurses-5' to include the version. Import
6189 libs are generally not versioned, and do not have this difficulty.
6191 2. Create a symbolic link from the dll to a file in the `lib'
6192 directory according to the above mentioned search pattern. This
6193 should be used to avoid unwanted changes in the tools needed for
6196 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6198 Then you can link without any make environment changes.
6200 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6202 This technique also avoids the version number problems, because
6203 the following is perfectly legal
6208 libxxx.dll.a -> ../bin/cygxxx-5.dll
6210 Linking directly to a dll without using an import lib will work
6211 even when auto-import features are exercised, and even when
6212 `--enable-runtime-pseudo-relocs' is used.
6214 Given the improvements in speed and memory usage, one might
6215 justifiably wonder why import libraries are used at all. There
6218 1. Until recently, the link-directly-to-dll functionality did _not_
6219 work with auto-imported data.
6221 2. Sometimes it is necessary to include pure static objects within
6222 the import library (which otherwise contains only bfd's for
6223 indirection symbols that point to the exports of a dll). Again,
6224 the import lib for the cygwin kernel makes use of this ability,
6225 and it is not possible to do this without an import lib.
6227 3. Symbol aliases can only be resolved using an import lib. This
6228 is critical when linking against OS-supplied dll's (eg, the win32
6229 API) in which symbols are usually exported as undecorated aliases
6230 of their stdcall-decorated assembly names.
6232 So, import libs are not going away. But the ability to replace
6233 true import libs with a simple symbolic link to (or a copy of) a
6234 dll, in many cases, is a useful addition to the suite of tools
6235 binutils makes available to the win32 developer. Given the
6236 massive improvements in memory requirements during linking, storage
6237 requirements, and linking speed, we expect that many developers
6238 will soon begin to use this feature whenever possible.
6242 _adding additional names_
6243 Sometimes, it is useful to export symbols with additional
6244 names. A symbol `foo' will be exported as `foo', but it can
6245 also be exported as `_foo' by using special directives in the
6246 DEF file when creating the dll. This will affect also the
6247 optional created import library. Consider the following DEF
6250 LIBRARY "xyz.dll" BASE=0x61000000
6256 The line `_foo = foo' maps the symbol `foo' to `_foo'.
6258 Another method for creating a symbol alias is to create it in
6259 the source code using the "weak" attribute:
6261 void foo () { /* Do something. */; }
6262 void _foo () __attribute__ ((weak, alias ("foo")));
6264 See the gcc manual for more information about attributes and
6268 Sometimes it is useful to rename exports. For instance, the
6269 cygwin kernel does this regularly. A symbol `_foo' can be
6270 exported as `foo' but not as `_foo' by using special
6271 directives in the DEF file. (This will also affect the import
6272 library, if it is created). In the following example:
6274 LIBRARY "xyz.dll" BASE=0x61000000
6279 The line `_foo = foo' maps the exported symbol `foo' to
6282 Note: using a DEF file disables the default auto-export behavior,
6283 unless the `--export-all-symbols' command line option is used.
6284 If, however, you are trying to rename symbols, then you should list
6285 _all_ desired exports in the DEF file, including the symbols that
6286 are not being renamed, and do _not_ use the `--export-all-symbols'
6287 option. If you list only the renamed symbols in the DEF file, and
6288 use `--export-all-symbols' to handle the other symbols, then the
6289 both the new names _and_ the original names for the renamed
6290 symbols will be exported. In effect, you'd be aliasing those
6291 symbols, not renaming them, which is probably not what you wanted.
6294 The Windows object format, PE, specifies a form of weak symbols
6295 called weak externals. When a weak symbol is linked and the
6296 symbol is not defined, the weak symbol becomes an alias for some
6297 other symbol. There are three variants of weak externals:
6298 * Definition is searched for in objects and libraries,
6299 historically called lazy externals.
6301 * Definition is searched for only in other objects, not in
6302 libraries. This form is not presently implemented.
6304 * No search; the symbol is an alias. This form is not presently
6306 As a GNU extension, weak symbols that do not specify an alternate
6307 symbol are supported. If the symbol is undefined when linking,
6308 the symbol uses a default value.
6310 _aligned common symbols_
6311 As a GNU extension to the PE file format, it is possible to
6312 specify the desired alignment for a common symbol. This
6313 information is conveyed from the assembler or compiler to the
6314 linker by means of GNU-specific commands carried in the object
6315 file's `.drectve' section, which are recognized by `ld' and
6316 respected when laying out the common symbols. Native tools will
6317 be able to process object files employing this GNU extension, but
6318 will fail to respect the alignment instructions, and may issue
6319 noisy warnings about unknown linker directives.
6323 File: ld.info, Node: Xtensa, Prev: WIN32, Up: Machine Dependent
6325 4.17 `ld' and Xtensa Processors
6326 ===============================
6328 The default `ld' behavior for Xtensa processors is to interpret
6329 `SECTIONS' commands so that lists of explicitly named sections in a
6330 specification with a wildcard file will be interleaved when necessary to
6331 keep literal pools within the range of PC-relative load offsets. For
6332 example, with the command:
6341 `ld' may interleave some of the `.literal' and `.text' sections from
6342 different object files to ensure that the literal pools are within the
6343 range of PC-relative load offsets. A valid interleaving might place
6344 the `.literal' sections from an initial group of files followed by the
6345 `.text' sections of that group of files. Then, the `.literal' sections
6346 from the rest of the files and the `.text' sections from the rest of
6347 the files would follow.
6349 Relaxation is enabled by default for the Xtensa version of `ld' and
6350 provides two important link-time optimizations. The first optimization
6351 is to combine identical literal values to reduce code size. A redundant
6352 literal will be removed and all the `L32R' instructions that use it
6353 will be changed to reference an identical literal, as long as the
6354 location of the replacement literal is within the offset range of all
6355 the `L32R' instructions. The second optimization is to remove
6356 unnecessary overhead from assembler-generated "longcall" sequences of
6357 `L32R'/`CALLXN' when the target functions are within range of direct
6358 `CALLN' instructions.
6360 For each of these cases where an indirect call sequence can be
6361 optimized to a direct call, the linker will change the `CALLXN'
6362 instruction to a `CALLN' instruction, remove the `L32R' instruction,
6363 and remove the literal referenced by the `L32R' instruction if it is
6364 not used for anything else. Removing the `L32R' instruction always
6365 reduces code size but can potentially hurt performance by changing the
6366 alignment of subsequent branch targets. By default, the linker will
6367 always preserve alignments, either by switching some instructions
6368 between 24-bit encodings and the equivalent density instructions or by
6369 inserting a no-op in place of the `L32R' instruction that was removed.
6370 If code size is more important than performance, the `--size-opt'
6371 option can be used to prevent the linker from widening density
6372 instructions or inserting no-ops, except in a few cases where no-ops
6373 are required for correctness.
6375 The following Xtensa-specific command-line options can be used to
6379 When optimizing indirect calls to direct calls, optimize for code
6380 size more than performance. With this option, the linker will not
6381 insert no-ops or widen density instructions to preserve branch
6382 target alignment. There may still be some cases where no-ops are
6383 required to preserve the correctness of the code.
6386 File: ld.info, Node: BFD, Next: Reporting Bugs, Prev: Machine Dependent, Up: Top
6391 The linker accesses object and archive files using the BFD libraries.
6392 These libraries allow the linker to use the same routines to operate on
6393 object files whatever the object file format. A different object file
6394 format can be supported simply by creating a new BFD back end and adding
6395 it to the library. To conserve runtime memory, however, the linker and
6396 associated tools are usually configured to support only a subset of the
6397 object file formats available. You can use `objdump -i' (*note
6398 objdump: (binutils.info)objdump.) to list all the formats available for
6401 As with most implementations, BFD is a compromise between several
6402 conflicting requirements. The major factor influencing BFD design was
6403 efficiency: any time used converting between formats is time which
6404 would not have been spent had BFD not been involved. This is partly
6405 offset by abstraction payback; since BFD simplifies applications and
6406 back ends, more time and care may be spent optimizing algorithms for a
6409 One minor artifact of the BFD solution which you should bear in mind
6410 is the potential for information loss. There are two places where
6411 useful information can be lost using the BFD mechanism: during
6412 conversion and during output. *Note BFD information loss::.
6416 * BFD outline:: How it works: an outline of BFD
6419 File: ld.info, Node: BFD outline, Up: BFD
6421 5.1 How It Works: An Outline of BFD
6422 ===================================
6424 When an object file is opened, BFD subroutines automatically determine
6425 the format of the input object file. They then build a descriptor in
6426 memory with pointers to routines that will be used to access elements of
6427 the object file's data structures.
6429 As different information from the object files is required, BFD
6430 reads from different sections of the file and processes them. For
6431 example, a very common operation for the linker is processing symbol
6432 tables. Each BFD back end provides a routine for converting between
6433 the object file's representation of symbols and an internal canonical
6434 format. When the linker asks for the symbol table of an object file, it
6435 calls through a memory pointer to the routine from the relevant BFD
6436 back end which reads and converts the table into a canonical form. The
6437 linker then operates upon the canonical form. When the link is finished
6438 and the linker writes the output file's symbol table, another BFD back
6439 end routine is called to take the newly created symbol table and
6440 convert it into the chosen output format.
6444 * BFD information loss:: Information Loss
6445 * Canonical format:: The BFD canonical object-file format
6448 File: ld.info, Node: BFD information loss, Next: Canonical format, Up: BFD outline
6450 5.1.1 Information Loss
6451 ----------------------
6453 _Information can be lost during output._ The output formats supported
6454 by BFD do not provide identical facilities, and information which can
6455 be described in one form has nowhere to go in another format. One
6456 example of this is alignment information in `b.out'. There is nowhere
6457 in an `a.out' format file to store alignment information on the
6458 contained data, so when a file is linked from `b.out' and an `a.out'
6459 image is produced, alignment information will not propagate to the
6460 output file. (The linker will still use the alignment information
6461 internally, so the link is performed correctly).
6463 Another example is COFF section names. COFF files may contain an
6464 unlimited number of sections, each one with a textual section name. If
6465 the target of the link is a format which does not have many sections
6466 (e.g., `a.out') or has sections without names (e.g., the Oasys format),
6467 the link cannot be done simply. You can circumvent this problem by
6468 describing the desired input-to-output section mapping with the linker
6471 _Information can be lost during canonicalization._ The BFD internal
6472 canonical form of the external formats is not exhaustive; there are
6473 structures in input formats for which there is no direct representation
6474 internally. This means that the BFD back ends cannot maintain all
6475 possible data richness through the transformation between external to
6476 internal and back to external formats.
6478 This limitation is only a problem when an application reads one
6479 format and writes another. Each BFD back end is responsible for
6480 maintaining as much data as possible, and the internal BFD canonical
6481 form has structures which are opaque to the BFD core, and exported only
6482 to the back ends. When a file is read in one format, the canonical form
6483 is generated for BFD and the application. At the same time, the back
6484 end saves away any information which may otherwise be lost. If the data
6485 is then written back in the same format, the back end routine will be
6486 able to use the canonical form provided by the BFD core as well as the
6487 information it prepared earlier. Since there is a great deal of
6488 commonality between back ends, there is no information lost when
6489 linking or copying big endian COFF to little endian COFF, or `a.out' to
6490 `b.out'. When a mixture of formats is linked, the information is only
6491 lost from the files whose format differs from the destination.
6494 File: ld.info, Node: Canonical format, Prev: BFD information loss, Up: BFD outline
6496 5.1.2 The BFD canonical object-file format
6497 ------------------------------------------
6499 The greatest potential for loss of information occurs when there is the
6500 least overlap between the information provided by the source format,
6501 that stored by the canonical format, and that needed by the destination
6502 format. A brief description of the canonical form may help you
6503 understand which kinds of data you can count on preserving across
6507 Information stored on a per-file basis includes target machine
6508 architecture, particular implementation format type, a demand
6509 pageable bit, and a write protected bit. Information like Unix
6510 magic numbers is not stored here--only the magic numbers' meaning,
6511 so a `ZMAGIC' file would have both the demand pageable bit and the
6512 write protected text bit set. The byte order of the target is
6513 stored on a per-file basis, so that big- and little-endian object
6514 files may be used with one another.
6517 Each section in the input file contains the name of the section,
6518 the section's original address in the object file, size and
6519 alignment information, various flags, and pointers into other BFD
6523 Each symbol contains a pointer to the information for the object
6524 file which originally defined it, its name, its value, and various
6525 flag bits. When a BFD back end reads in a symbol table, it
6526 relocates all symbols to make them relative to the base of the
6527 section where they were defined. Doing this ensures that each
6528 symbol points to its containing section. Each symbol also has a
6529 varying amount of hidden private data for the BFD back end. Since
6530 the symbol points to the original file, the private data format
6531 for that symbol is accessible. `ld' can operate on a collection
6532 of symbols of wildly different formats without problems.
6534 Normal global and simple local symbols are maintained on output,
6535 so an output file (no matter its format) will retain symbols
6536 pointing to functions and to global, static, and common variables.
6537 Some symbol information is not worth retaining; in `a.out', type
6538 information is stored in the symbol table as long symbol names.
6539 This information would be useless to most COFF debuggers; the
6540 linker has command line switches to allow users to throw it away.
6542 There is one word of type information within the symbol, so if the
6543 format supports symbol type information within symbols (for
6544 example, COFF, IEEE, Oasys) and the type is simple enough to fit
6545 within one word (nearly everything but aggregates), the
6546 information will be preserved.
6549 Each canonical BFD relocation record contains a pointer to the
6550 symbol to relocate to, the offset of the data to relocate, the
6551 section the data is in, and a pointer to a relocation type
6552 descriptor. Relocation is performed by passing messages through
6553 the relocation type descriptor and the symbol pointer. Therefore,
6554 relocations can be performed on output data using a relocation
6555 method that is only available in one of the input formats. For
6556 instance, Oasys provides a byte relocation format. A relocation
6557 record requesting this relocation type would point indirectly to a
6558 routine to perform this, so the relocation may be performed on a
6559 byte being written to a 68k COFF file, even though 68k COFF has no
6560 such relocation type.
6563 Object formats can contain, for debugging purposes, some form of
6564 mapping between symbols, source line numbers, and addresses in the
6565 output file. These addresses have to be relocated along with the
6566 symbol information. Each symbol with an associated list of line
6567 number records points to the first record of the list. The head
6568 of a line number list consists of a pointer to the symbol, which
6569 allows finding out the address of the function whose line number
6570 is being described. The rest of the list is made up of pairs:
6571 offsets into the section and line numbers. Any format which can
6572 simply derive this information can pass it successfully between
6573 formats (COFF, IEEE and Oasys).
6576 File: ld.info, Node: Reporting Bugs, Next: MRI, Prev: BFD, Up: Top
6581 Your bug reports play an essential role in making `ld' reliable.
6583 Reporting a bug may help you by bringing a solution to your problem,
6584 or it may not. But in any case the principal function of a bug report
6585 is to help the entire community by making the next version of `ld' work
6586 better. Bug reports are your contribution to the maintenance of `ld'.
6588 In order for a bug report to serve its purpose, you must include the
6589 information that enables us to fix the bug.
6593 * Bug Criteria:: Have you found a bug?
6594 * Bug Reporting:: How to report bugs
6597 File: ld.info, Node: Bug Criteria, Next: Bug Reporting, Up: Reporting Bugs
6599 6.1 Have You Found a Bug?
6600 =========================
6602 If you are not sure whether you have found a bug, here are some
6605 * If the linker gets a fatal signal, for any input whatever, that is
6606 a `ld' bug. Reliable linkers never crash.
6608 * If `ld' produces an error message for valid input, that is a bug.
6610 * If `ld' does not produce an error message for invalid input, that
6611 may be a bug. In the general case, the linker can not verify that
6612 object files are correct.
6614 * If you are an experienced user of linkers, your suggestions for
6615 improvement of `ld' are welcome in any case.
6618 File: ld.info, Node: Bug Reporting, Prev: Bug Criteria, Up: Reporting Bugs
6620 6.2 How to Report Bugs
6621 ======================
6623 A number of companies and individuals offer support for GNU products.
6624 If you obtained `ld' from a support organization, we recommend you
6625 contact that organization first.
6627 You can find contact information for many support companies and
6628 individuals in the file `etc/SERVICE' in the GNU Emacs distribution.
6630 Otherwise, send bug reports for `ld' to
6631 `http://www.sourceware.org/bugzilla/'.
6633 The fundamental principle of reporting bugs usefully is this:
6634 *report all the facts*. If you are not sure whether to state a fact or
6635 leave it out, state it!
6637 Often people omit facts because they think they know what causes the
6638 problem and assume that some details do not matter. Thus, you might
6639 assume that the name of a symbol you use in an example does not matter.
6640 Well, probably it does not, but one cannot be sure. Perhaps the bug
6641 is a stray memory reference which happens to fetch from the location
6642 where that name is stored in memory; perhaps, if the name were
6643 different, the contents of that location would fool the linker into
6644 doing the right thing despite the bug. Play it safe and give a
6645 specific, complete example. That is the easiest thing for you to do,
6646 and the most helpful.
6648 Keep in mind that the purpose of a bug report is to enable us to fix
6649 the bug if it is new to us. Therefore, always write your bug reports
6650 on the assumption that the bug has not been reported previously.
6652 Sometimes people give a few sketchy facts and ask, "Does this ring a
6653 bell?" This cannot help us fix a bug, so it is basically useless. We
6654 respond by asking for enough details to enable us to investigate. You
6655 might as well expedite matters by sending them to begin with.
6657 To enable us to fix the bug, you should include all these things:
6659 * The version of `ld'. `ld' announces it if you start it with the
6660 `--version' argument.
6662 Without this, we will not know whether there is any point in
6663 looking for the bug in the current version of `ld'.
6665 * Any patches you may have applied to the `ld' source, including any
6666 patches made to the `BFD' library.
6668 * The type of machine you are using, and the operating system name
6671 * What compiler (and its version) was used to compile `ld'--e.g.
6674 * The command arguments you gave the linker to link your example and
6675 observe the bug. To guarantee you will not omit something
6676 important, list them all. A copy of the Makefile (or the output
6677 from make) is sufficient.
6679 If we were to try to guess the arguments, we would probably guess
6680 wrong and then we might not encounter the bug.
6682 * A complete input file, or set of input files, that will reproduce
6683 the bug. It is generally most helpful to send the actual object
6684 files provided that they are reasonably small. Say no more than
6685 10K. For bigger files you can either make them available by FTP
6686 or HTTP or else state that you are willing to send the object
6687 file(s) to whomever requests them. (Note - your email will be
6688 going to a mailing list, so we do not want to clog it up with
6689 large attachments). But small attachments are best.
6691 If the source files were assembled using `gas' or compiled using
6692 `gcc', then it may be OK to send the source files rather than the
6693 object files. In this case, be sure to say exactly what version of
6694 `gas' or `gcc' was used to produce the object files. Also say how
6695 `gas' or `gcc' were configured.
6697 * A description of what behavior you observe that you believe is
6698 incorrect. For example, "It gets a fatal signal."
6700 Of course, if the bug is that `ld' gets a fatal signal, then we
6701 will certainly notice it. But if the bug is incorrect output, we
6702 might not notice unless it is glaringly wrong. You might as well
6703 not give us a chance to make a mistake.
6705 Even if the problem you experience is a fatal signal, you should
6706 still say so explicitly. Suppose something strange is going on,
6707 such as, your copy of `ld' is out of sync, or you have encountered
6708 a bug in the C library on your system. (This has happened!) Your
6709 copy might crash and ours would not. If you told us to expect a
6710 crash, then when ours fails to crash, we would know that the bug
6711 was not happening for us. If you had not told us to expect a
6712 crash, then we would not be able to draw any conclusion from our
6715 * If you wish to suggest changes to the `ld' source, send us context
6716 diffs, as generated by `diff' with the `-u', `-c', or `-p' option.
6717 Always send diffs from the old file to the new file. If you even
6718 discuss something in the `ld' source, refer to it by context, not
6721 The line numbers in our development sources will not match those
6722 in your sources. Your line numbers would convey no useful
6725 Here are some things that are not necessary:
6727 * A description of the envelope of the bug.
6729 Often people who encounter a bug spend a lot of time investigating
6730 which changes to the input file will make the bug go away and which
6731 changes will not affect it.
6733 This is often time consuming and not very useful, because the way
6734 we will find the bug is by running a single example under the
6735 debugger with breakpoints, not by pure deduction from a series of
6736 examples. We recommend that you save your time for something else.
6738 Of course, if you can find a simpler example to report _instead_
6739 of the original one, that is a convenience for us. Errors in the
6740 output will be easier to spot, running under the debugger will take
6741 less time, and so on.
6743 However, simplification is not vital; if you do not want to do
6744 this, report the bug anyway and send us the entire test case you
6747 * A patch for the bug.
6749 A patch for the bug does help us if it is a good one. But do not
6750 omit the necessary information, such as the test case, on the
6751 assumption that a patch is all we need. We might see problems
6752 with your patch and decide to fix the problem another way, or we
6753 might not understand it at all.
6755 Sometimes with a program as complicated as `ld' it is very hard to
6756 construct an example that will make the program follow a certain
6757 path through the code. If you do not send us the example, we will
6758 not be able to construct one, so we will not be able to verify
6759 that the bug is fixed.
6761 And if we cannot understand what bug you are trying to fix, or why
6762 your patch should be an improvement, we will not install it. A
6763 test case will help us to understand.
6765 * A guess about what the bug is or what it depends on.
6767 Such guesses are usually wrong. Even we cannot guess right about
6768 such things without first using the debugger to find the facts.
6771 File: ld.info, Node: MRI, Next: GNU Free Documentation License, Prev: Reporting Bugs, Up: Top
6773 Appendix A MRI Compatible Script Files
6774 **************************************
6776 To aid users making the transition to GNU `ld' from the MRI linker,
6777 `ld' can use MRI compatible linker scripts as an alternative to the
6778 more general-purpose linker scripting language described in *Note
6779 Scripts::. MRI compatible linker scripts have a much simpler command
6780 set than the scripting language otherwise used with `ld'. GNU `ld'
6781 supports the most commonly used MRI linker commands; these commands are
6784 In general, MRI scripts aren't of much use with the `a.out' object
6785 file format, since it only has three sections and MRI scripts lack some
6786 features to make use of them.
6788 You can specify a file containing an MRI-compatible script using the
6789 `-c' command-line option.
6791 Each command in an MRI-compatible script occupies its own line; each
6792 command line starts with the keyword that identifies the command (though
6793 blank lines are also allowed for punctuation). If a line of an
6794 MRI-compatible script begins with an unrecognized keyword, `ld' issues
6795 a warning message, but continues processing the script.
6797 Lines beginning with `*' are comments.
6799 You can write these commands using all upper-case letters, or all
6800 lower case; for example, `chip' is the same as `CHIP'. The following
6801 list shows only the upper-case form of each command.
6804 `ABSOLUTE SECNAME, SECNAME, ... SECNAME'
6805 Normally, `ld' includes in the output file all sections from all
6806 the input files. However, in an MRI-compatible script, you can
6807 use the `ABSOLUTE' command to restrict the sections that will be
6808 present in your output program. If the `ABSOLUTE' command is used
6809 at all in a script, then only the sections named explicitly in
6810 `ABSOLUTE' commands will appear in the linker output. You can
6811 still use other input sections (whatever you select on the command
6812 line, or using `LOAD') to resolve addresses in the output file.
6814 `ALIAS OUT-SECNAME, IN-SECNAME'
6815 Use this command to place the data from input section IN-SECNAME
6816 in a section called OUT-SECNAME in the linker output file.
6818 IN-SECNAME may be an integer.
6820 `ALIGN SECNAME = EXPRESSION'
6821 Align the section called SECNAME to EXPRESSION. The EXPRESSION
6822 should be a power of two.
6825 Use the value of EXPRESSION as the lowest address (other than
6826 absolute addresses) in the output file.
6829 `CHIP EXPRESSION, EXPRESSION'
6830 This command does nothing; it is accepted only for compatibility.
6833 This command does nothing whatever; it's only accepted for
6836 `FORMAT OUTPUT-FORMAT'
6837 Similar to the `OUTPUT_FORMAT' command in the more general linker
6838 language, but restricted to one of these output formats:
6840 1. S-records, if OUTPUT-FORMAT is `S'
6842 2. IEEE, if OUTPUT-FORMAT is `IEEE'
6844 3. COFF (the `coff-m68k' variant in BFD), if OUTPUT-FORMAT is
6848 Print (to the standard output file) a link map, as produced by the
6849 `ld' command-line option `-M'.
6851 The keyword `LIST' may be followed by anything on the same line,
6852 with no change in its effect.
6855 `LOAD FILENAME, FILENAME, ... FILENAME'
6856 Include one or more object file FILENAME in the link; this has the
6857 same effect as specifying FILENAME directly on the `ld' command
6861 OUTPUT-NAME is the name for the program produced by `ld'; the
6862 MRI-compatible command `NAME' is equivalent to the command-line
6863 option `-o' or the general script language command `OUTPUT'.
6865 `ORDER SECNAME, SECNAME, ... SECNAME'
6866 `ORDER SECNAME SECNAME SECNAME'
6867 Normally, `ld' orders the sections in its output file in the order
6868 in which they first appear in the input files. In an
6869 MRI-compatible script, you can override this ordering with the
6870 `ORDER' command. The sections you list with `ORDER' will appear
6871 first in your output file, in the order specified.
6873 `PUBLIC NAME=EXPRESSION'
6874 `PUBLIC NAME,EXPRESSION'
6875 `PUBLIC NAME EXPRESSION'
6876 Supply a value (EXPRESSION) for external symbol NAME used in the
6879 `SECT SECNAME, EXPRESSION'
6880 `SECT SECNAME=EXPRESSION'
6881 `SECT SECNAME EXPRESSION'
6882 You can use any of these three forms of the `SECT' command to
6883 specify the start address (EXPRESSION) for section SECNAME. If
6884 you have more than one `SECT' statement for the same SECNAME, only
6885 the _first_ sets the start address.
6888 File: ld.info, Node: GNU Free Documentation License, Next: LD Index, Prev: MRI, Up: Top
6890 Appendix B GNU Free Documentation License
6891 *****************************************
6893 Version 1.3, 3 November 2008
6895 Copyright (C) 2000, 2001, 2002, 2007, 2008 Free Software Foundation, Inc.
6898 Everyone is permitted to copy and distribute verbatim copies
6899 of this license document, but changing it is not allowed.
6903 The purpose of this License is to make a manual, textbook, or other
6904 functional and useful document "free" in the sense of freedom: to
6905 assure everyone the effective freedom to copy and redistribute it,
6906 with or without modifying it, either commercially or
6907 noncommercially. Secondarily, this License preserves for the
6908 author and publisher a way to get credit for their work, while not
6909 being considered responsible for modifications made by others.
6911 This License is a kind of "copyleft", which means that derivative
6912 works of the document must themselves be free in the same sense.
6913 It complements the GNU General Public License, which is a copyleft
6914 license designed for free software.
6916 We have designed this License in order to use it for manuals for
6917 free software, because free software needs free documentation: a
6918 free program should come with manuals providing the same freedoms
6919 that the software does. But this License is not limited to
6920 software manuals; it can be used for any textual work, regardless
6921 of subject matter or whether it is published as a printed book.
6922 We recommend this License principally for works whose purpose is
6923 instruction or reference.
6925 1. APPLICABILITY AND DEFINITIONS
6927 This License applies to any manual or other work, in any medium,
6928 that contains a notice placed by the copyright holder saying it
6929 can be distributed under the terms of this License. Such a notice
6930 grants a world-wide, royalty-free license, unlimited in duration,
6931 to use that work under the conditions stated herein. The
6932 "Document", below, refers to any such manual or work. Any member
6933 of the public is a licensee, and is addressed as "you". You
6934 accept the license if you copy, modify or distribute the work in a
6935 way requiring permission under copyright law.
6937 A "Modified Version" of the Document means any work containing the
6938 Document or a portion of it, either copied verbatim, or with
6939 modifications and/or translated into another language.
6941 A "Secondary Section" is a named appendix or a front-matter section
6942 of the Document that deals exclusively with the relationship of the
6943 publishers or authors of the Document to the Document's overall
6944 subject (or to related matters) and contains nothing that could
6945 fall directly within that overall subject. (Thus, if the Document
6946 is in part a textbook of mathematics, a Secondary Section may not
6947 explain any mathematics.) The relationship could be a matter of
6948 historical connection with the subject or with related matters, or
6949 of legal, commercial, philosophical, ethical or political position
6952 The "Invariant Sections" are certain Secondary Sections whose
6953 titles are designated, as being those of Invariant Sections, in
6954 the notice that says that the Document is released under this
6955 License. If a section does not fit the above definition of
6956 Secondary then it is not allowed to be designated as Invariant.
6957 The Document may contain zero Invariant Sections. If the Document
6958 does not identify any Invariant Sections then there are none.
6960 The "Cover Texts" are certain short passages of text that are
6961 listed, as Front-Cover Texts or Back-Cover Texts, in the notice
6962 that says that the Document is released under this License. A
6963 Front-Cover Text may be at most 5 words, and a Back-Cover Text may
6964 be at most 25 words.
6966 A "Transparent" copy of the Document means a machine-readable copy,
6967 represented in a format whose specification is available to the
6968 general public, that is suitable for revising the document
6969 straightforwardly with generic text editors or (for images
6970 composed of pixels) generic paint programs or (for drawings) some
6971 widely available drawing editor, and that is suitable for input to
6972 text formatters or for automatic translation to a variety of
6973 formats suitable for input to text formatters. A copy made in an
6974 otherwise Transparent file format whose markup, or absence of
6975 markup, has been arranged to thwart or discourage subsequent
6976 modification by readers is not Transparent. An image format is
6977 not Transparent if used for any substantial amount of text. A
6978 copy that is not "Transparent" is called "Opaque".
6980 Examples of suitable formats for Transparent copies include plain
6981 ASCII without markup, Texinfo input format, LaTeX input format,
6982 SGML or XML using a publicly available DTD, and
6983 standard-conforming simple HTML, PostScript or PDF designed for
6984 human modification. Examples of transparent image formats include
6985 PNG, XCF and JPG. Opaque formats include proprietary formats that
6986 can be read and edited only by proprietary word processors, SGML or
6987 XML for which the DTD and/or processing tools are not generally
6988 available, and the machine-generated HTML, PostScript or PDF
6989 produced by some word processors for output purposes only.
6991 The "Title Page" means, for a printed book, the title page itself,
6992 plus such following pages as are needed to hold, legibly, the
6993 material this License requires to appear in the title page. For
6994 works in formats which do not have any title page as such, "Title
6995 Page" means the text near the most prominent appearance of the
6996 work's title, preceding the beginning of the body of the text.
6998 The "publisher" means any person or entity that distributes copies
6999 of the Document to the public.
7001 A section "Entitled XYZ" means a named subunit of the Document
7002 whose title either is precisely XYZ or contains XYZ in parentheses
7003 following text that translates XYZ in another language. (Here XYZ
7004 stands for a specific section name mentioned below, such as
7005 "Acknowledgements", "Dedications", "Endorsements", or "History".)
7006 To "Preserve the Title" of such a section when you modify the
7007 Document means that it remains a section "Entitled XYZ" according
7010 The Document may include Warranty Disclaimers next to the notice
7011 which states that this License applies to the Document. These
7012 Warranty Disclaimers are considered to be included by reference in
7013 this License, but only as regards disclaiming warranties: any other
7014 implication that these Warranty Disclaimers may have is void and
7015 has no effect on the meaning of this License.
7019 You may copy and distribute the Document in any medium, either
7020 commercially or noncommercially, provided that this License, the
7021 copyright notices, and the license notice saying this License
7022 applies to the Document are reproduced in all copies, and that you
7023 add no other conditions whatsoever to those of this License. You
7024 may not use technical measures to obstruct or control the reading
7025 or further copying of the copies you make or distribute. However,
7026 you may accept compensation in exchange for copies. If you
7027 distribute a large enough number of copies you must also follow
7028 the conditions in section 3.
7030 You may also lend copies, under the same conditions stated above,
7031 and you may publicly display copies.
7033 3. COPYING IN QUANTITY
7035 If you publish printed copies (or copies in media that commonly
7036 have printed covers) of the Document, numbering more than 100, and
7037 the Document's license notice requires Cover Texts, you must
7038 enclose the copies in covers that carry, clearly and legibly, all
7039 these Cover Texts: Front-Cover Texts on the front cover, and
7040 Back-Cover Texts on the back cover. Both covers must also clearly
7041 and legibly identify you as the publisher of these copies. The
7042 front cover must present the full title with all words of the
7043 title equally prominent and visible. You may add other material
7044 on the covers in addition. Copying with changes limited to the
7045 covers, as long as they preserve the title of the Document and
7046 satisfy these conditions, can be treated as verbatim copying in
7049 If the required texts for either cover are too voluminous to fit
7050 legibly, you should put the first ones listed (as many as fit
7051 reasonably) on the actual cover, and continue the rest onto
7054 If you publish or distribute Opaque copies of the Document
7055 numbering more than 100, you must either include a
7056 machine-readable Transparent copy along with each Opaque copy, or
7057 state in or with each Opaque copy a computer-network location from
7058 which the general network-using public has access to download
7059 using public-standard network protocols a complete Transparent
7060 copy of the Document, free of added material. If you use the
7061 latter option, you must take reasonably prudent steps, when you
7062 begin distribution of Opaque copies in quantity, to ensure that
7063 this Transparent copy will remain thus accessible at the stated
7064 location until at least one year after the last time you
7065 distribute an Opaque copy (directly or through your agents or
7066 retailers) of that edition to the public.
7068 It is requested, but not required, that you contact the authors of
7069 the Document well before redistributing any large number of
7070 copies, to give them a chance to provide you with an updated
7071 version of the Document.
7075 You may copy and distribute a Modified Version of the Document
7076 under the conditions of sections 2 and 3 above, provided that you
7077 release the Modified Version under precisely this License, with
7078 the Modified Version filling the role of the Document, thus
7079 licensing distribution and modification of the Modified Version to
7080 whoever possesses a copy of it. In addition, you must do these
7081 things in the Modified Version:
7083 A. Use in the Title Page (and on the covers, if any) a title
7084 distinct from that of the Document, and from those of
7085 previous versions (which should, if there were any, be listed
7086 in the History section of the Document). You may use the
7087 same title as a previous version if the original publisher of
7088 that version gives permission.
7090 B. List on the Title Page, as authors, one or more persons or
7091 entities responsible for authorship of the modifications in
7092 the Modified Version, together with at least five of the
7093 principal authors of the Document (all of its principal
7094 authors, if it has fewer than five), unless they release you
7095 from this requirement.
7097 C. State on the Title page the name of the publisher of the
7098 Modified Version, as the publisher.
7100 D. Preserve all the copyright notices of the Document.
7102 E. Add an appropriate copyright notice for your modifications
7103 adjacent to the other copyright notices.
7105 F. Include, immediately after the copyright notices, a license
7106 notice giving the public permission to use the Modified
7107 Version under the terms of this License, in the form shown in
7110 G. Preserve in that license notice the full lists of Invariant
7111 Sections and required Cover Texts given in the Document's
7114 H. Include an unaltered copy of this License.
7116 I. Preserve the section Entitled "History", Preserve its Title,
7117 and add to it an item stating at least the title, year, new
7118 authors, and publisher of the Modified Version as given on
7119 the Title Page. If there is no section Entitled "History" in
7120 the Document, create one stating the title, year, authors,
7121 and publisher of the Document as given on its Title Page,
7122 then add an item describing the Modified Version as stated in
7123 the previous sentence.
7125 J. Preserve the network location, if any, given in the Document
7126 for public access to a Transparent copy of the Document, and
7127 likewise the network locations given in the Document for
7128 previous versions it was based on. These may be placed in
7129 the "History" section. You may omit a network location for a
7130 work that was published at least four years before the
7131 Document itself, or if the original publisher of the version
7132 it refers to gives permission.
7134 K. For any section Entitled "Acknowledgements" or "Dedications",
7135 Preserve the Title of the section, and preserve in the
7136 section all the substance and tone of each of the contributor
7137 acknowledgements and/or dedications given therein.
7139 L. Preserve all the Invariant Sections of the Document,
7140 unaltered in their text and in their titles. Section numbers
7141 or the equivalent are not considered part of the section
7144 M. Delete any section Entitled "Endorsements". Such a section
7145 may not be included in the Modified Version.
7147 N. Do not retitle any existing section to be Entitled
7148 "Endorsements" or to conflict in title with any Invariant
7151 O. Preserve any Warranty Disclaimers.
7153 If the Modified Version includes new front-matter sections or
7154 appendices that qualify as Secondary Sections and contain no
7155 material copied from the Document, you may at your option
7156 designate some or all of these sections as invariant. To do this,
7157 add their titles to the list of Invariant Sections in the Modified
7158 Version's license notice. These titles must be distinct from any
7159 other section titles.
7161 You may add a section Entitled "Endorsements", provided it contains
7162 nothing but endorsements of your Modified Version by various
7163 parties--for example, statements of peer review or that the text
7164 has been approved by an organization as the authoritative
7165 definition of a standard.
7167 You may add a passage of up to five words as a Front-Cover Text,
7168 and a passage of up to 25 words as a Back-Cover Text, to the end
7169 of the list of Cover Texts in the Modified Version. Only one
7170 passage of Front-Cover Text and one of Back-Cover Text may be
7171 added by (or through arrangements made by) any one entity. If the
7172 Document already includes a cover text for the same cover,
7173 previously added by you or by arrangement made by the same entity
7174 you are acting on behalf of, you may not add another; but you may
7175 replace the old one, on explicit permission from the previous
7176 publisher that added the old one.
7178 The author(s) and publisher(s) of the Document do not by this
7179 License give permission to use their names for publicity for or to
7180 assert or imply endorsement of any Modified Version.
7182 5. COMBINING DOCUMENTS
7184 You may combine the Document with other documents released under
7185 this License, under the terms defined in section 4 above for
7186 modified versions, provided that you include in the combination
7187 all of the Invariant Sections of all of the original documents,
7188 unmodified, and list them all as Invariant Sections of your
7189 combined work in its license notice, and that you preserve all
7190 their Warranty Disclaimers.
7192 The combined work need only contain one copy of this License, and
7193 multiple identical Invariant Sections may be replaced with a single
7194 copy. If there are multiple Invariant Sections with the same name
7195 but different contents, make the title of each such section unique
7196 by adding at the end of it, in parentheses, the name of the
7197 original author or publisher of that section if known, or else a
7198 unique number. Make the same adjustment to the section titles in
7199 the list of Invariant Sections in the license notice of the
7202 In the combination, you must combine any sections Entitled
7203 "History" in the various original documents, forming one section
7204 Entitled "History"; likewise combine any sections Entitled
7205 "Acknowledgements", and any sections Entitled "Dedications". You
7206 must delete all sections Entitled "Endorsements."
7208 6. COLLECTIONS OF DOCUMENTS
7210 You may make a collection consisting of the Document and other
7211 documents released under this License, and replace the individual
7212 copies of this License in the various documents with a single copy
7213 that is included in the collection, provided that you follow the
7214 rules of this License for verbatim copying of each of the
7215 documents in all other respects.
7217 You may extract a single document from such a collection, and
7218 distribute it individually under this License, provided you insert
7219 a copy of this License into the extracted document, and follow
7220 this License in all other respects regarding verbatim copying of
7223 7. AGGREGATION WITH INDEPENDENT WORKS
7225 A compilation of the Document or its derivatives with other
7226 separate and independent documents or works, in or on a volume of
7227 a storage or distribution medium, is called an "aggregate" if the
7228 copyright resulting from the compilation is not used to limit the
7229 legal rights of the compilation's users beyond what the individual
7230 works permit. When the Document is included in an aggregate, this
7231 License does not apply to the other works in the aggregate which
7232 are not themselves derivative works of the Document.
7234 If the Cover Text requirement of section 3 is applicable to these
7235 copies of the Document, then if the Document is less than one half
7236 of the entire aggregate, the Document's Cover Texts may be placed
7237 on covers that bracket the Document within the aggregate, or the
7238 electronic equivalent of covers if the Document is in electronic
7239 form. Otherwise they must appear on printed covers that bracket
7240 the whole aggregate.
7244 Translation is considered a kind of modification, so you may
7245 distribute translations of the Document under the terms of section
7246 4. Replacing Invariant Sections with translations requires special
7247 permission from their copyright holders, but you may include
7248 translations of some or all Invariant Sections in addition to the
7249 original versions of these Invariant Sections. You may include a
7250 translation of this License, and all the license notices in the
7251 Document, and any Warranty Disclaimers, provided that you also
7252 include the original English version of this License and the
7253 original versions of those notices and disclaimers. In case of a
7254 disagreement between the translation and the original version of
7255 this License or a notice or disclaimer, the original version will
7258 If a section in the Document is Entitled "Acknowledgements",
7259 "Dedications", or "History", the requirement (section 4) to
7260 Preserve its Title (section 1) will typically require changing the
7265 You may not copy, modify, sublicense, or distribute the Document
7266 except as expressly provided under this License. Any attempt
7267 otherwise to copy, modify, sublicense, or distribute it is void,
7268 and will automatically terminate your rights under this License.
7270 However, if you cease all violation of this License, then your
7271 license from a particular copyright holder is reinstated (a)
7272 provisionally, unless and until the copyright holder explicitly
7273 and finally terminates your license, and (b) permanently, if the
7274 copyright holder fails to notify you of the violation by some
7275 reasonable means prior to 60 days after the cessation.
7277 Moreover, your license from a particular copyright holder is
7278 reinstated permanently if the copyright holder notifies you of the
7279 violation by some reasonable means, this is the first time you have
7280 received notice of violation of this License (for any work) from
7281 that copyright holder, and you cure the violation prior to 30 days
7282 after your receipt of the notice.
7284 Termination of your rights under this section does not terminate
7285 the licenses of parties who have received copies or rights from
7286 you under this License. If your rights have been terminated and
7287 not permanently reinstated, receipt of a copy of some or all of
7288 the same material does not give you any rights to use it.
7290 10. FUTURE REVISIONS OF THIS LICENSE
7292 The Free Software Foundation may publish new, revised versions of
7293 the GNU Free Documentation License from time to time. Such new
7294 versions will be similar in spirit to the present version, but may
7295 differ in detail to address new problems or concerns. See
7296 `http://www.gnu.org/copyleft/'.
7298 Each version of the License is given a distinguishing version
7299 number. If the Document specifies that a particular numbered
7300 version of this License "or any later version" applies to it, you
7301 have the option of following the terms and conditions either of
7302 that specified version or of any later version that has been
7303 published (not as a draft) by the Free Software Foundation. If
7304 the Document does not specify a version number of this License,
7305 you may choose any version ever published (not as a draft) by the
7306 Free Software Foundation. If the Document specifies that a proxy
7307 can decide which future versions of this License can be used, that
7308 proxy's public statement of acceptance of a version permanently
7309 authorizes you to choose that version for the Document.
7313 "Massive Multiauthor Collaboration Site" (or "MMC Site") means any
7314 World Wide Web server that publishes copyrightable works and also
7315 provides prominent facilities for anybody to edit those works. A
7316 public wiki that anybody can edit is an example of such a server.
7317 A "Massive Multiauthor Collaboration" (or "MMC") contained in the
7318 site means any set of copyrightable works thus published on the MMC
7321 "CC-BY-SA" means the Creative Commons Attribution-Share Alike 3.0
7322 license published by Creative Commons Corporation, a not-for-profit
7323 corporation with a principal place of business in San Francisco,
7324 California, as well as future copyleft versions of that license
7325 published by that same organization.
7327 "Incorporate" means to publish or republish a Document, in whole or
7328 in part, as part of another Document.
7330 An MMC is "eligible for relicensing" if it is licensed under this
7331 License, and if all works that were first published under this
7332 License somewhere other than this MMC, and subsequently
7333 incorporated in whole or in part into the MMC, (1) had no cover
7334 texts or invariant sections, and (2) were thus incorporated prior
7335 to November 1, 2008.
7337 The operator of an MMC Site may republish an MMC contained in the
7338 site under CC-BY-SA on the same site at any time before August 1,
7339 2009, provided the MMC is eligible for relicensing.
7342 ADDENDUM: How to use this License for your documents
7343 ====================================================
7345 To use this License in a document you have written, include a copy of
7346 the License in the document and put the following copyright and license
7347 notices just after the title page:
7349 Copyright (C) YEAR YOUR NAME.
7350 Permission is granted to copy, distribute and/or modify this document
7351 under the terms of the GNU Free Documentation License, Version 1.3
7352 or any later version published by the Free Software Foundation;
7353 with no Invariant Sections, no Front-Cover Texts, and no Back-Cover
7354 Texts. A copy of the license is included in the section entitled ``GNU
7355 Free Documentation License''.
7357 If you have Invariant Sections, Front-Cover Texts and Back-Cover
7358 Texts, replace the "with...Texts." line with this:
7360 with the Invariant Sections being LIST THEIR TITLES, with
7361 the Front-Cover Texts being LIST, and with the Back-Cover Texts
7364 If you have Invariant Sections without Cover Texts, or some other
7365 combination of the three, merge those two alternatives to suit the
7368 If your document contains nontrivial examples of program code, we
7369 recommend releasing these examples in parallel under your choice of
7370 free software license, such as the GNU General Public License, to
7371 permit their use in free software.
7374 File: ld.info, Node: LD Index, Prev: GNU Free Documentation License, Up: Top
7382 * ": Symbols. (line 6)
7383 * -(: Options. (line 731)
7384 * --accept-unknown-input-arch: Options. (line 749)
7385 * --add-needed: Options. (line 777)
7386 * --add-stdcall-alias: Options. (line 1635)
7387 * --allow-multiple-definition: Options. (line 1029)
7388 * --allow-shlib-undefined: Options. (line 1035)
7389 * --architecture=ARCH: Options. (line 123)
7390 * --as-needed: Options. (line 759)
7391 * --audit AUDITLIB: Options. (line 112)
7392 * --auxiliary=NAME: Options. (line 255)
7393 * --bank-window: Options. (line 2089)
7394 * --base-file: Options. (line 1640)
7395 * --be8: ARM. (line 28)
7396 * --bss-plt: PowerPC ELF32. (line 16)
7397 * --build-id: Options. (line 1597)
7398 * --build-id=STYLE: Options. (line 1597)
7399 * --check-sections: Options. (line 856)
7400 * --copy-dt-needed-entries: Options. (line 868)
7401 * --cref: Options. (line 888)
7402 * --default-imported-symver: Options. (line 1072)
7403 * --default-script=SCRIPT: Options. (line 562)
7404 * --default-symver: Options. (line 1068)
7405 * --defsym=SYMBOL=EXP: Options. (line 917)
7406 * --demangle[=STYLE]: Options. (line 930)
7407 * --depaudit AUDITLIB: Options. (line 177)
7408 * --disable-auto-image-base: Options. (line 1827)
7409 * --disable-auto-import: Options. (line 1962)
7410 * --disable-large-address-aware: Options. (line 1766)
7411 * --disable-long-section-names: Options. (line 1650)
7412 * --disable-new-dtags: Options. (line 1559)
7413 * --disable-runtime-pseudo-reloc: Options. (line 1975)
7414 * --disable-stdcall-fixup: Options. (line 1672)
7415 * --discard-all: Options. (line 608)
7416 * --discard-locals: Options. (line 612)
7417 * --dll: Options. (line 1645)
7418 * --dll-search-prefix: Options. (line 1833)
7419 * --dotsyms: PowerPC64 ELF64. (line 33)
7420 * --dsbt-index: Options. (line 2066)
7421 * --dsbt-size: Options. (line 2061)
7422 * --dynamic-linker=FILE: Options. (line 943)
7423 * --dynamic-list-cpp-new: Options. (line 848)
7424 * --dynamic-list-cpp-typeinfo: Options. (line 852)
7425 * --dynamic-list-data: Options. (line 845)
7426 * --dynamic-list=DYNAMIC-LIST-FILE: Options. (line 832)
7427 * --dynamicbase: Options. (line 2015)
7428 * --eh-frame-hdr: Options. (line 1550)
7429 * --emit-relocs: Options. (line 497)
7430 * --emit-stack-syms: SPU ELF. (line 46)
7431 * --emit-stub-syms <1>: SPU ELF. (line 15)
7432 * --emit-stub-syms <2>: PowerPC ELF32. (line 47)
7433 * --emit-stub-syms: PowerPC64 ELF64. (line 29)
7434 * --enable-auto-image-base: Options. (line 1818)
7435 * --enable-auto-import: Options. (line 1842)
7436 * --enable-extra-pe-debug: Options. (line 1980)
7437 * --enable-long-section-names: Options. (line 1650)
7438 * --enable-new-dtags: Options. (line 1559)
7439 * --enable-runtime-pseudo-reloc: Options. (line 1967)
7440 * --enable-stdcall-fixup: Options. (line 1672)
7441 * --entry=ENTRY: Options. (line 187)
7442 * --error-unresolved-symbols: Options. (line 1503)
7443 * --exclude-all-symbols: Options. (line 1726)
7444 * --exclude-libs: Options. (line 197)
7445 * --exclude-modules-for-implib: Options. (line 208)
7446 * --exclude-symbols: Options. (line 1720)
7447 * --export-all-symbols: Options. (line 1696)
7448 * --export-dynamic: Options. (line 221)
7449 * --extra-overlay-stubs: SPU ELF. (line 19)
7450 * --fatal-warnings: Options. (line 950)
7451 * --file-alignment: Options. (line 1730)
7452 * --filter=NAME: Options. (line 276)
7453 * --fix-arm1176: ARM. (line 111)
7454 * --fix-cortex-a53-835769: ARM. (line 174)
7455 * --fix-cortex-a8: ARM. (line 165)
7456 * --fix-v4bx: ARM. (line 49)
7457 * --fix-v4bx-interworking: ARM. (line 62)
7458 * --force-dynamic: Options. (line 506)
7459 * --force-exe-suffix: Options. (line 955)
7460 * --forceinteg: Options. (line 2020)
7461 * --format=FORMAT: Options. (line 134)
7462 * --format=VERSION: TI COFF. (line 6)
7463 * --gc-sections: Options. (line 965)
7464 * --got: Options. (line 2102)
7465 * --got=TYPE: M68K. (line 6)
7466 * --gpsize=VALUE: Options. (line 309)
7467 * --hash-size=NUMBER: Options. (line 1569)
7468 * --hash-style=STYLE: Options. (line 1577)
7469 * --heap: Options. (line 1736)
7470 * --help: Options. (line 1002)
7471 * --high-entropy-va: Options. (line 2011)
7472 * --image-base: Options. (line 1743)
7473 * --insert-timestamp: Options. (line 2043)
7474 * --insn32 <1>: MIPS. (line 6)
7475 * --insn32: Options. (line 2114)
7476 * --just-symbols=FILE: Options. (line 529)
7477 * --kill-at: Options. (line 1752)
7478 * --large-address-aware: Options. (line 1757)
7479 * --ld-generated-unwind-info: Options. (line 1554)
7480 * --leading-underscore: Options. (line 1690)
7481 * --library-path=DIR: Options. (line 367)
7482 * --library=NAMESPEC: Options. (line 334)
7483 * --local-store=lo:hi: SPU ELF. (line 24)
7484 * --long-plt: ARM. (line 185)
7485 * --major-image-version: Options. (line 1773)
7486 * --major-os-version: Options. (line 1778)
7487 * --major-subsystem-version: Options. (line 1782)
7488 * --merge-exidx-entries: ARM. (line 182)
7489 * --minor-image-version: Options. (line 1787)
7490 * --minor-os-version: Options. (line 1792)
7491 * --minor-subsystem-version: Options. (line 1796)
7492 * --mri-script=MRI-CMDFILE: Options. (line 158)
7493 * --multi-subspace: HPPA ELF32. (line 6)
7494 * --nmagic: Options. (line 439)
7495 * --no-accept-unknown-input-arch: Options. (line 749)
7496 * --no-add-needed: Options. (line 777)
7497 * --no-allow-shlib-undefined: Options. (line 1035)
7498 * --no-as-needed: Options. (line 759)
7499 * --no-bind: Options. (line 2034)
7500 * --no-check-sections: Options. (line 856)
7501 * --no-copy-dt-needed-entries: Options. (line 868)
7502 * --no-define-common: Options. (line 901)
7503 * --no-demangle: Options. (line 930)
7504 * --no-dotsyms: PowerPC64 ELF64. (line 33)
7505 * --no-enum-size-warning: ARM. (line 120)
7506 * --no-export-dynamic: Options. (line 221)
7507 * --no-fatal-warnings: Options. (line 950)
7508 * --no-fix-arm1176: ARM. (line 111)
7509 * --no-fix-cortex-a53-835769: ARM. (line 174)
7510 * --no-fix-cortex-a8: ARM. (line 165)
7511 * --no-gc-sections: Options. (line 965)
7512 * --no-insn32 <1>: MIPS. (line 6)
7513 * --no-insn32: Options. (line 2115)
7514 * --no-isolation: Options. (line 2027)
7515 * --no-keep-memory: Options. (line 1014)
7516 * --no-leading-underscore: Options. (line 1690)
7517 * --no-merge-exidx-entries <1>: ARM. (line 182)
7518 * --no-merge-exidx-entries: Options. (line 2073)
7519 * --no-multi-toc: PowerPC64 ELF64. (line 74)
7520 * --no-omagic: Options. (line 454)
7521 * --no-opd-optimize: PowerPC64 ELF64. (line 48)
7522 * --no-overlays: SPU ELF. (line 9)
7523 * --no-plt-align: PowerPC64 ELF64. (line 96)
7524 * --no-plt-static-chain: PowerPC64 ELF64. (line 104)
7525 * --no-plt-thread-safe: PowerPC64 ELF64. (line 110)
7526 * --no-print-gc-sections: Options. (line 987)
7527 * --no-seh: Options. (line 2030)
7528 * --no-tls-optimize <1>: PowerPC64 ELF64. (line 43)
7529 * --no-tls-optimize: PowerPC ELF32. (line 51)
7530 * --no-toc-optimize: PowerPC64 ELF64. (line 60)
7531 * --no-toc-sort: PowerPC64 ELF64. (line 86)
7532 * --no-trampoline: Options. (line 2083)
7533 * --no-undefined: Options. (line 1021)
7534 * --no-undefined-version: Options. (line 1063)
7535 * --no-warn-mismatch: Options. (line 1076)
7536 * --no-warn-search-mismatch: Options. (line 1085)
7537 * --no-wchar-size-warning: ARM. (line 127)
7538 * --no-whole-archive: Options. (line 1089)
7539 * --noinhibit-exec: Options. (line 1093)
7540 * --non-overlapping-opd: PowerPC64 ELF64. (line 54)
7541 * --nxcompat: Options. (line 2023)
7542 * --oformat=OUTPUT-FORMAT: Options. (line 1105)
7543 * --omagic: Options. (line 445)
7544 * --out-implib: Options. (line 1809)
7545 * --output-def: Options. (line 1801)
7546 * --output=OUTPUT: Options. (line 460)
7547 * --pic-executable: Options. (line 1118)
7548 * --pic-veneer: ARM. (line 133)
7549 * --plt-align: PowerPC64 ELF64. (line 96)
7550 * --plt-static-chain: PowerPC64 ELF64. (line 104)
7551 * --plt-thread-safe: PowerPC64 ELF64. (line 110)
7552 * --plugin: SPU ELF. (line 6)
7553 * --pop-state: Options. (line 494)
7554 * --print-gc-sections: Options. (line 987)
7555 * --print-map: Options. (line 402)
7556 * --print-output-format: Options. (line 996)
7557 * --push-state: Options. (line 476)
7558 * --reduce-memory-overheads: Options. (line 1583)
7559 * --relax: Options. (line 1134)
7560 * --relax on i960: i960. (line 31)
7561 * --relax on Nios II: Nios II. (line 6)
7562 * --relax on PowerPC: PowerPC ELF32. (line 6)
7563 * --relax on Xtensa: Xtensa. (line 27)
7564 * --relocatable: Options. (line 510)
7565 * --retain-symbols-file=FILENAME: Options. (line 1160)
7566 * --script=SCRIPT: Options. (line 553)
7567 * --sdata-got: PowerPC ELF32. (line 33)
7568 * --section-alignment: Options. (line 1985)
7569 * --section-start=SECTIONNAME=ORG: Options. (line 1316)
7570 * --secure-plt: PowerPC ELF32. (line 26)
7571 * --sort-common: Options. (line 1258)
7572 * --sort-section=alignment: Options. (line 1273)
7573 * --sort-section=name: Options. (line 1269)
7574 * --split-by-file: Options. (line 1277)
7575 * --split-by-reloc: Options. (line 1282)
7576 * --stack: Options. (line 1991)
7577 * --stack-analysis: SPU ELF. (line 29)
7578 * --stats: Options. (line 1295)
7579 * --strip-all: Options. (line 540)
7580 * --strip-debug: Options. (line 544)
7581 * --stub-group-size: PowerPC64 ELF64. (line 6)
7582 * --stub-group-size=N <1>: ARM. (line 138)
7583 * --stub-group-size=N: HPPA ELF32. (line 12)
7584 * --subsystem: Options. (line 1998)
7585 * --support-old-code: ARM. (line 6)
7586 * --sysroot=DIRECTORY: Options. (line 1299)
7587 * --target-help: Options. (line 1006)
7588 * --target1-abs: ARM. (line 32)
7589 * --target1-rel: ARM. (line 32)
7590 * --target2=TYPE: ARM. (line 37)
7591 * --thumb-entry=ENTRY: ARM. (line 17)
7592 * --trace: Options. (line 549)
7593 * --trace-symbol=SYMBOL: Options. (line 618)
7594 * --traditional-format: Options. (line 1304)
7595 * --tsaware: Options. (line 2040)
7596 * --undefined=SYMBOL: Options. (line 575)
7597 * --unique[=SECTION]: Options. (line 593)
7598 * --unresolved-symbols: Options. (line 1346)
7599 * --use-blx: ARM. (line 74)
7600 * --use-nul-prefixed-import-tables: ARM. (line 23)
7601 * --verbose[=NUMBER]: Options. (line 1375)
7602 * --version: Options. (line 602)
7603 * --version-script=VERSION-SCRIPTFILE: Options. (line 1383)
7604 * --vfp11-denorm-fix: ARM. (line 83)
7605 * --warn-alternate-em: Options. (line 1495)
7606 * --warn-common: Options. (line 1394)
7607 * --warn-constructors: Options. (line 1462)
7608 * --warn-multiple-gp: Options. (line 1467)
7609 * --warn-once: Options. (line 1481)
7610 * --warn-section-align: Options. (line 1485)
7611 * --warn-shared-textrel: Options. (line 1492)
7612 * --warn-unresolved-symbols: Options. (line 1498)
7613 * --wdmdriver: Options. (line 2037)
7614 * --whole-archive: Options. (line 1507)
7615 * --wrap=SYMBOL: Options. (line 1521)
7616 * -A ARCH: Options. (line 122)
7617 * -a KEYWORD: Options. (line 105)
7618 * -assert KEYWORD: Options. (line 784)
7619 * -b FORMAT: Options. (line 134)
7620 * -Bdynamic: Options. (line 787)
7621 * -Bgroup: Options. (line 797)
7622 * -Bshareable: Options. (line 1251)
7623 * -Bstatic: Options. (line 804)
7624 * -Bsymbolic: Options. (line 819)
7625 * -Bsymbolic-functions: Options. (line 826)
7626 * -c MRI-CMDFILE: Options. (line 158)
7627 * -call_shared: Options. (line 787)
7628 * -d: Options. (line 168)
7629 * -dc: Options. (line 168)
7630 * -dn: Options. (line 804)
7631 * -dp: Options. (line 168)
7632 * -dT SCRIPT: Options. (line 562)
7633 * -dy: Options. (line 787)
7634 * -E: Options. (line 221)
7635 * -e ENTRY: Options. (line 187)
7636 * -EB: Options. (line 248)
7637 * -EL: Options. (line 251)
7638 * -F NAME: Options. (line 276)
7639 * -f NAME: Options. (line 255)
7640 * -fini=NAME: Options. (line 300)
7641 * -g: Options. (line 306)
7642 * -G VALUE: Options. (line 309)
7643 * -h NAME: Options. (line 316)
7644 * -i: Options. (line 325)
7645 * -IFILE: Options. (line 943)
7646 * -init=NAME: Options. (line 328)
7647 * -L DIR: Options. (line 367)
7648 * -l NAMESPEC: Options. (line 334)
7649 * -M: Options. (line 402)
7650 * -m EMULATION: Options. (line 392)
7651 * -Map=MAPFILE: Options. (line 1010)
7652 * -N: Options. (line 445)
7653 * -n: Options. (line 439)
7654 * -no-relax: Options. (line 1134)
7655 * -non_shared: Options. (line 804)
7656 * -nostdlib: Options. (line 1099)
7657 * -O LEVEL: Options. (line 466)
7658 * -o OUTPUT: Options. (line 460)
7659 * -P AUDITLIB: Options. (line 177)
7660 * -pie: Options. (line 1118)
7661 * -q: Options. (line 497)
7662 * -qmagic: Options. (line 1128)
7663 * -Qy: Options. (line 1131)
7664 * -r: Options. (line 510)
7665 * -R FILE: Options. (line 529)
7666 * -rpath-link=DIR: Options. (line 1196)
7667 * -rpath=DIR: Options. (line 1174)
7668 * -S: Options. (line 544)
7669 * -s: Options. (line 540)
7670 * -shared: Options. (line 1251)
7671 * -soname=NAME: Options. (line 316)
7672 * -static: Options. (line 804)
7673 * -t: Options. (line 549)
7674 * -T SCRIPT: Options. (line 553)
7675 * -Tbss=ORG: Options. (line 1325)
7676 * -Tdata=ORG: Options. (line 1325)
7677 * -Tldata-segment=ORG: Options. (line 1341)
7678 * -Trodata-segment=ORG: Options. (line 1335)
7679 * -Ttext-segment=ORG: Options. (line 1331)
7680 * -Ttext=ORG: Options. (line 1325)
7681 * -u SYMBOL: Options. (line 575)
7682 * -Ur: Options. (line 583)
7683 * -v: Options. (line 602)
7684 * -V: Options. (line 602)
7685 * -x: Options. (line 608)
7686 * -X: Options. (line 612)
7687 * -Y PATH: Options. (line 627)
7688 * -y SYMBOL: Options. (line 618)
7689 * -z defs: Options. (line 1021)
7690 * -z KEYWORD: Options. (line 631)
7691 * -z muldefs: Options. (line 1029)
7692 * .: Location Counter. (line 6)
7693 * /DISCARD/: Output Section Discarding.
7695 * 32-bit PLT entries: ARM. (line 185)
7696 * :PHDR: Output Section Phdr.
7698 * =FILLEXP: Output Section Fill.
7700 * >REGION: Output Section Region.
7702 * [COMMON]: Input Section Common.
7704 * ABSOLUTE (MRI): MRI. (line 33)
7705 * absolute and relocatable symbols: Expression Section. (line 6)
7706 * absolute expressions: Expression Section. (line 6)
7707 * ABSOLUTE(EXP): Builtin Functions. (line 10)
7708 * ADDR(SECTION): Builtin Functions. (line 17)
7709 * address, section: Output Section Address.
7711 * ALIAS (MRI): MRI. (line 44)
7712 * ALIGN (MRI): MRI. (line 50)
7713 * align expression: Builtin Functions. (line 38)
7714 * align location counter: Builtin Functions. (line 38)
7715 * ALIGN(ALIGN): Builtin Functions. (line 38)
7716 * ALIGN(EXP,ALIGN): Builtin Functions. (line 38)
7717 * ALIGN(SECTION_ALIGN): Forced Output Alignment.
7719 * aligned common symbols: WIN32. (line 424)
7720 * ALIGNOF(SECTION): Builtin Functions. (line 64)
7721 * allocating memory: MEMORY. (line 6)
7722 * architecture: Miscellaneous Commands.
7724 * architectures: Options. (line 122)
7725 * archive files, from cmd line: Options. (line 334)
7726 * archive search path in linker script: File Commands. (line 76)
7727 * arithmetic: Expressions. (line 6)
7728 * arithmetic operators: Operators. (line 6)
7729 * ARM interworking support: ARM. (line 6)
7730 * ARM1176 erratum workaround: ARM. (line 111)
7731 * AS_NEEDED(FILES): File Commands. (line 56)
7732 * ASSERT: Miscellaneous Commands.
7734 * assertion in linker script: Miscellaneous Commands.
7736 * assignment in scripts: Assignments. (line 6)
7737 * AT(LMA): Output Section LMA. (line 6)
7738 * AT>LMA_REGION: Output Section LMA. (line 6)
7739 * automatic data imports: WIN32. (line 191)
7740 * back end: BFD. (line 6)
7741 * BASE (MRI): MRI. (line 54)
7742 * BE8: ARM. (line 28)
7743 * BFD canonical format: Canonical format. (line 11)
7744 * BFD requirements: BFD. (line 16)
7745 * big-endian objects: Options. (line 248)
7746 * binary input format: Options. (line 134)
7747 * BLOCK(EXP): Builtin Functions. (line 77)
7748 * bug criteria: Bug Criteria. (line 6)
7749 * bug reports: Bug Reporting. (line 6)
7750 * bugs in ld: Reporting Bugs. (line 6)
7751 * BYTE(EXPRESSION): Output Section Data.
7753 * C++ constructors, arranging in link: Output Section Keywords.
7755 * CHIP (MRI): MRI. (line 58)
7756 * COLLECT_NO_DEMANGLE: Environment. (line 29)
7757 * combining symbols, warnings on: Options. (line 1394)
7758 * command files: Scripts. (line 6)
7759 * command line: Options. (line 6)
7760 * common allocation: Options. (line 168)
7761 * common allocation in linker script: Miscellaneous Commands.
7763 * common symbol placement: Input Section Common.
7765 * COMMONPAGESIZE: Symbolic Constants. (line 13)
7766 * compatibility, MRI: Options. (line 158)
7767 * CONSTANT: Symbolic Constants. (line 6)
7768 * constants in linker scripts: Constants. (line 6)
7769 * constraints on output sections: Output Section Constraint.
7771 * constructors: Options. (line 583)
7772 * CONSTRUCTORS: Output Section Keywords.
7774 * constructors, arranging in link: Output Section Keywords.
7776 * Cortex-A53 erratum 835769 workaround: ARM. (line 174)
7777 * Cortex-A8 erratum workaround: ARM. (line 165)
7778 * crash of linker: Bug Criteria. (line 9)
7779 * CREATE_OBJECT_SYMBOLS: Output Section Keywords.
7781 * creating a DEF file: WIN32. (line 158)
7782 * cross reference table: Options. (line 888)
7783 * cross references: Miscellaneous Commands.
7785 * current output location: Location Counter. (line 6)
7786 * data: Output Section Data.
7788 * DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE): Builtin Functions.
7790 * DATA_SEGMENT_END(EXP): Builtin Functions. (line 103)
7791 * DATA_SEGMENT_RELRO_END(OFFSET, EXP): Builtin Functions. (line 109)
7792 * dbx: Options. (line 1309)
7793 * DEF files, creating: Options. (line 1801)
7794 * default emulation: Environment. (line 21)
7795 * default input format: Environment. (line 9)
7796 * DEFINED(SYMBOL): Builtin Functions. (line 122)
7797 * deleting local symbols: Options. (line 608)
7798 * demangling, default: Environment. (line 29)
7799 * demangling, from command line: Options. (line 930)
7800 * direct linking to a dll: WIN32. (line 239)
7801 * discarding sections: Output Section Discarding.
7803 * discontinuous memory: MEMORY. (line 6)
7804 * DLLs, creating: Options. (line 1809)
7805 * DLLs, linking to: Options. (line 1833)
7806 * dot: Location Counter. (line 6)
7807 * dot inside sections: Location Counter. (line 36)
7808 * dot outside sections: Location Counter. (line 66)
7809 * dynamic linker, from command line: Options. (line 943)
7810 * dynamic symbol table: Options. (line 221)
7811 * ELF program headers: PHDRS. (line 6)
7812 * emulation: Options. (line 392)
7813 * emulation, default: Environment. (line 21)
7814 * END (MRI): MRI. (line 62)
7815 * endianness: Options. (line 248)
7816 * entry point: Entry Point. (line 6)
7817 * entry point, from command line: Options. (line 187)
7818 * entry point, thumb: ARM. (line 17)
7819 * ENTRY(SYMBOL): Entry Point. (line 6)
7820 * error on valid input: Bug Criteria. (line 12)
7821 * example of linker script: Simple Example. (line 6)
7822 * exporting DLL symbols: WIN32. (line 19)
7823 * expression evaluation order: Evaluation. (line 6)
7824 * expression sections: Expression Section. (line 6)
7825 * expression, absolute: Builtin Functions. (line 10)
7826 * expressions: Expressions. (line 6)
7827 * EXTERN: Miscellaneous Commands.
7829 * fatal signal: Bug Criteria. (line 9)
7830 * file name wildcard patterns: Input Section Wildcards.
7832 * FILEHDR: PHDRS. (line 62)
7833 * filename symbols: Output Section Keywords.
7835 * fill pattern, entire section: Output Section Fill.
7837 * FILL(EXPRESSION): Output Section Data.
7839 * finalization function: Options. (line 300)
7840 * first input file: File Commands. (line 84)
7841 * first instruction: Entry Point. (line 6)
7842 * FIX_V4BX: ARM. (line 49)
7843 * FIX_V4BX_INTERWORKING: ARM. (line 62)
7844 * FORCE_COMMON_ALLOCATION: Miscellaneous Commands.
7846 * forcing input section alignment: Forced Input Alignment.
7848 * forcing output section alignment: Forced Output Alignment.
7850 * forcing the creation of dynamic sections: Options. (line 506)
7851 * FORMAT (MRI): MRI. (line 66)
7852 * functions in expressions: Builtin Functions. (line 6)
7853 * garbage collection <1>: Options. (line 965)
7854 * garbage collection <2>: Input Section Keep. (line 6)
7855 * garbage collection: Options. (line 987)
7856 * generating optimized output: Options. (line 466)
7857 * GNU linker: Overview. (line 6)
7858 * GNUTARGET: Environment. (line 9)
7859 * GROUP(FILES): File Commands. (line 49)
7860 * grouping input files: File Commands. (line 49)
7861 * groups of archives: Options. (line 731)
7862 * H8/300 support: H8/300. (line 6)
7863 * header size: Builtin Functions. (line 190)
7864 * heap size: Options. (line 1736)
7865 * help: Options. (line 1002)
7866 * HIDDEN: HIDDEN. (line 6)
7867 * holes: Location Counter. (line 12)
7868 * holes, filling: Output Section Data.
7870 * HPPA multiple sub-space stubs: HPPA ELF32. (line 6)
7871 * HPPA stub grouping: HPPA ELF32. (line 12)
7872 * i960 support: i960. (line 6)
7873 * image base: Options. (line 1743)
7874 * implicit linker scripts: Implicit Linker Scripts.
7876 * import libraries: WIN32. (line 10)
7877 * INCLUDE FILENAME: File Commands. (line 9)
7878 * including a linker script: File Commands. (line 9)
7879 * including an entire archive: Options. (line 1507)
7880 * incremental link: Options. (line 325)
7881 * INHIBIT_COMMON_ALLOCATION: Miscellaneous Commands.
7883 * initialization function: Options. (line 328)
7884 * initialized data in ROM: Output Section LMA. (line 39)
7885 * input file format in linker script: Format Commands. (line 35)
7886 * input filename symbols: Output Section Keywords.
7888 * input files in linker scripts: File Commands. (line 19)
7889 * input files, displaying: Options. (line 549)
7890 * input format: Options. (line 134)
7891 * input object files in linker scripts: File Commands. (line 19)
7892 * input section alignment: Forced Input Alignment.
7894 * input section basics: Input Section Basics.
7896 * input section wildcards: Input Section Wildcards.
7898 * input sections: Input Section. (line 6)
7899 * INPUT(FILES): File Commands. (line 19)
7900 * INSERT: Miscellaneous Commands.
7902 * insert user script into default script: Miscellaneous Commands.
7904 * integer notation: Constants. (line 6)
7905 * integer suffixes: Constants. (line 15)
7906 * internal object-file format: Canonical format. (line 11)
7907 * invalid input: Bug Criteria. (line 14)
7908 * K and M integer suffixes: Constants. (line 15)
7909 * KEEP: Input Section Keep. (line 6)
7910 * l =: MEMORY. (line 74)
7911 * lazy evaluation: Evaluation. (line 6)
7912 * ld bugs, reporting: Bug Reporting. (line 6)
7913 * LD_FEATURE(STRING): Miscellaneous Commands.
7915 * ldata segment origin, cmd line: Options. (line 1342)
7916 * LDEMULATION: Environment. (line 21)
7917 * len =: MEMORY. (line 74)
7918 * LENGTH =: MEMORY. (line 74)
7919 * LENGTH(MEMORY): Builtin Functions. (line 139)
7920 * library search path in linker script: File Commands. (line 76)
7921 * link map: Options. (line 402)
7922 * link-time runtime library search path: Options. (line 1196)
7923 * linker crash: Bug Criteria. (line 9)
7924 * linker script concepts: Basic Script Concepts.
7926 * linker script example: Simple Example. (line 6)
7927 * linker script file commands: File Commands. (line 6)
7928 * linker script format: Script Format. (line 6)
7929 * linker script input object files: File Commands. (line 19)
7930 * linker script simple commands: Simple Commands. (line 6)
7931 * linker scripts: Scripts. (line 6)
7932 * LIST (MRI): MRI. (line 77)
7933 * little-endian objects: Options. (line 251)
7934 * LOAD (MRI): MRI. (line 84)
7935 * load address: Output Section LMA. (line 6)
7936 * LOADADDR(SECTION): Builtin Functions. (line 142)
7937 * loading, preventing: Output Section Type.
7939 * local symbols, deleting: Options. (line 612)
7940 * location counter: Location Counter. (line 6)
7941 * LOG2CEIL(EXP): Builtin Functions. (line 146)
7942 * LONG(EXPRESSION): Output Section Data.
7944 * M and K integer suffixes: Constants. (line 15)
7945 * M68HC11 and 68HC12 support: M68HC11/68HC12. (line 6)
7946 * machine architecture: Miscellaneous Commands.
7948 * machine dependencies: Machine Dependent. (line 6)
7949 * mapping input sections to output sections: Input Section. (line 6)
7950 * MAX: Builtin Functions. (line 149)
7951 * MAXPAGESIZE: Symbolic Constants. (line 10)
7952 * MEMORY: MEMORY. (line 6)
7953 * memory region attributes: MEMORY. (line 34)
7954 * memory regions: MEMORY. (line 6)
7955 * memory regions and sections: Output Section Region.
7957 * memory usage: Options. (line 1014)
7958 * Merging exidx entries: ARM. (line 182)
7959 * MIN: Builtin Functions. (line 152)
7960 * MIPS microMIPS instruction choice selection: MIPS. (line 6)
7961 * Motorola 68K GOT generation: M68K. (line 6)
7962 * MRI compatibility: MRI. (line 6)
7963 * MSP430 extra sections: MSP430. (line 11)
7964 * NAME (MRI): MRI. (line 90)
7965 * name, section: Output Section Name.
7967 * names: Symbols. (line 6)
7968 * naming the output file: Options. (line 460)
7969 * NEXT(EXP): Builtin Functions. (line 156)
7970 * Nios II call relaxation: Nios II. (line 6)
7971 * NMAGIC: Options. (line 439)
7972 * NO_ENUM_SIZE_WARNING: ARM. (line 120)
7973 * NO_WCHAR_SIZE_WARNING: ARM. (line 127)
7974 * NOCROSSREFS(SECTIONS): Miscellaneous Commands.
7976 * NOLOAD: Output Section Type.
7978 * not enough room for program headers: Builtin Functions. (line 195)
7979 * o =: MEMORY. (line 69)
7980 * objdump -i: BFD. (line 6)
7981 * object file management: BFD. (line 6)
7982 * object files: Options. (line 29)
7983 * object formats available: BFD. (line 6)
7984 * object size: Options. (line 309)
7985 * OMAGIC: Options. (line 445)
7986 * ONLY_IF_RO: Output Section Constraint.
7988 * ONLY_IF_RW: Output Section Constraint.
7990 * opening object files: BFD outline. (line 6)
7991 * operators for arithmetic: Operators. (line 6)
7992 * options: Options. (line 6)
7993 * ORDER (MRI): MRI. (line 95)
7994 * org =: MEMORY. (line 69)
7995 * ORIGIN =: MEMORY. (line 69)
7996 * ORIGIN(MEMORY): Builtin Functions. (line 162)
7997 * orphan: Orphan Sections. (line 6)
7998 * output file after errors: Options. (line 1093)
7999 * output file format in linker script: Format Commands. (line 10)
8000 * output file name in linker script: File Commands. (line 66)
8001 * output format: Options. (line 996)
8002 * output section alignment: Forced Output Alignment.
8004 * output section attributes: Output Section Attributes.
8006 * output section data: Output Section Data.
8008 * OUTPUT(FILENAME): File Commands. (line 66)
8009 * OUTPUT_ARCH(BFDARCH): Miscellaneous Commands.
8011 * OUTPUT_FORMAT(BFDNAME): Format Commands. (line 10)
8012 * OVERLAY: Overlay Description.
8014 * overlays: Overlay Description.
8016 * partial link: Options. (line 510)
8017 * PE import table prefixing: ARM. (line 23)
8018 * PHDRS: PHDRS. (line 62)
8019 * PIC_VENEER: ARM. (line 133)
8020 * pop state governing input file handling: Options. (line 494)
8021 * position independent executables: Options. (line 1120)
8022 * PowerPC ELF32 options: PowerPC ELF32. (line 16)
8023 * PowerPC GOT: PowerPC ELF32. (line 33)
8024 * PowerPC long branches: PowerPC ELF32. (line 6)
8025 * PowerPC PLT: PowerPC ELF32. (line 16)
8026 * PowerPC stub symbols: PowerPC ELF32. (line 47)
8027 * PowerPC TLS optimization: PowerPC ELF32. (line 51)
8028 * PowerPC64 dot symbols: PowerPC64 ELF64. (line 33)
8029 * PowerPC64 ELF64 options: PowerPC64 ELF64. (line 6)
8030 * PowerPC64 multi-TOC: PowerPC64 ELF64. (line 74)
8031 * PowerPC64 OPD optimization: PowerPC64 ELF64. (line 48)
8032 * PowerPC64 OPD spacing: PowerPC64 ELF64. (line 54)
8033 * PowerPC64 PLT call stub static chain: PowerPC64 ELF64. (line 104)
8034 * PowerPC64 PLT call stub thread safety: PowerPC64 ELF64. (line 110)
8035 * PowerPC64 PLT stub alignment: PowerPC64 ELF64. (line 96)
8036 * PowerPC64 stub grouping: PowerPC64 ELF64. (line 6)
8037 * PowerPC64 stub symbols: PowerPC64 ELF64. (line 29)
8038 * PowerPC64 TLS optimization: PowerPC64 ELF64. (line 43)
8039 * PowerPC64 TOC optimization: PowerPC64 ELF64. (line 60)
8040 * PowerPC64 TOC sorting: PowerPC64 ELF64. (line 86)
8041 * precedence in expressions: Operators. (line 6)
8042 * prevent unnecessary loading: Output Section Type.
8044 * program headers: PHDRS. (line 6)
8045 * program headers and sections: Output Section Phdr.
8047 * program headers, not enough room: Builtin Functions. (line 195)
8048 * program segments: PHDRS. (line 6)
8049 * PROVIDE: PROVIDE. (line 6)
8050 * PROVIDE_HIDDEN: PROVIDE_HIDDEN. (line 6)
8051 * PUBLIC (MRI): MRI. (line 103)
8052 * push state governing input file handling: Options. (line 476)
8053 * QUAD(EXPRESSION): Output Section Data.
8055 * quoted symbol names: Symbols. (line 6)
8056 * read-only text: Options. (line 439)
8057 * read/write from cmd line: Options. (line 445)
8058 * region alias: REGION_ALIAS. (line 6)
8059 * region names: REGION_ALIAS. (line 6)
8060 * REGION_ALIAS(ALIAS, REGION): REGION_ALIAS. (line 6)
8061 * regions of memory: MEMORY. (line 6)
8062 * relative expressions: Expression Section. (line 6)
8063 * relaxing addressing modes: Options. (line 1134)
8064 * relaxing on H8/300: H8/300. (line 9)
8065 * relaxing on i960: i960. (line 31)
8066 * relaxing on M68HC11: M68HC11/68HC12. (line 12)
8067 * relaxing on NDS32: NDS32. (line 6)
8068 * relaxing on Xtensa: Xtensa. (line 27)
8069 * relocatable and absolute symbols: Expression Section. (line 6)
8070 * relocatable output: Options. (line 510)
8071 * removing sections: Output Section Discarding.
8073 * reporting bugs in ld: Reporting Bugs. (line 6)
8074 * requirements for BFD: BFD. (line 16)
8075 * retain relocations in final executable: Options. (line 497)
8076 * retaining specified symbols: Options. (line 1160)
8077 * rodata segment origin, cmd line: Options. (line 1336)
8078 * ROM initialized data: Output Section LMA. (line 39)
8079 * round up expression: Builtin Functions. (line 38)
8080 * round up location counter: Builtin Functions. (line 38)
8081 * runtime library name: Options. (line 316)
8082 * runtime library search path: Options. (line 1174)
8083 * runtime pseudo-relocation: WIN32. (line 217)
8084 * scaled integers: Constants. (line 15)
8085 * scommon section: Input Section Common.
8087 * script files: Options. (line 553)
8088 * scripts: Scripts. (line 6)
8089 * search directory, from cmd line: Options. (line 367)
8090 * search path in linker script: File Commands. (line 76)
8091 * SEARCH_DIR(PATH): File Commands. (line 76)
8092 * SECT (MRI): MRI. (line 109)
8093 * section address: Output Section Address.
8095 * section address in expression: Builtin Functions. (line 17)
8096 * section alignment: Builtin Functions. (line 64)
8097 * section alignment, warnings on: Options. (line 1485)
8098 * section data: Output Section Data.
8100 * section fill pattern: Output Section Fill.
8102 * section load address: Output Section LMA. (line 6)
8103 * section load address in expression: Builtin Functions. (line 142)
8104 * section name: Output Section Name.
8106 * section name wildcard patterns: Input Section Wildcards.
8108 * section size: Builtin Functions. (line 174)
8109 * section, assigning to memory region: Output Section Region.
8111 * section, assigning to program header: Output Section Phdr.
8113 * SECTIONS: SECTIONS. (line 6)
8114 * sections, discarding: Output Section Discarding.
8116 * segment origins, cmd line: Options. (line 1325)
8117 * SEGMENT_START(SEGMENT, DEFAULT): Builtin Functions. (line 165)
8118 * segments, ELF: PHDRS. (line 6)
8119 * shared libraries: Options. (line 1253)
8120 * SHORT(EXPRESSION): Output Section Data.
8122 * SIZEOF(SECTION): Builtin Functions. (line 174)
8123 * SIZEOF_HEADERS: Builtin Functions. (line 190)
8124 * small common symbols: Input Section Common.
8126 * SORT: Input Section Wildcards.
8128 * SORT_BY_ALIGNMENT: Input Section Wildcards.
8130 * SORT_BY_INIT_PRIORITY: Input Section Wildcards.
8132 * SORT_BY_NAME: Input Section Wildcards.
8134 * SORT_NONE: Input Section Wildcards.
8136 * SPU: SPU ELF. (line 29)
8137 * SPU ELF options: SPU ELF. (line 6)
8138 * SPU extra overlay stubs: SPU ELF. (line 19)
8139 * SPU local store size: SPU ELF. (line 24)
8140 * SPU overlay stub symbols: SPU ELF. (line 15)
8141 * SPU overlays: SPU ELF. (line 9)
8142 * SPU plugins: SPU ELF. (line 6)
8143 * SQUAD(EXPRESSION): Output Section Data.
8145 * stack size: Options. (line 1991)
8146 * standard Unix system: Options. (line 7)
8147 * start of execution: Entry Point. (line 6)
8148 * STARTUP(FILENAME): File Commands. (line 84)
8149 * strip all symbols: Options. (line 540)
8150 * strip debugger symbols: Options. (line 544)
8151 * stripping all but some symbols: Options. (line 1160)
8152 * STUB_GROUP_SIZE: ARM. (line 138)
8153 * SUBALIGN(SUBSECTION_ALIGN): Forced Input Alignment.
8155 * suffixes for integers: Constants. (line 15)
8156 * symbol defaults: Builtin Functions. (line 122)
8157 * symbol definition, scripts: Assignments. (line 6)
8158 * symbol names: Symbols. (line 6)
8159 * symbol tracing: Options. (line 618)
8160 * symbol versions: VERSION. (line 6)
8161 * symbol-only input: Options. (line 529)
8162 * symbolic constants: Symbolic Constants. (line 6)
8163 * symbols, from command line: Options. (line 917)
8164 * symbols, relocatable and absolute: Expression Section. (line 6)
8165 * symbols, retaining selectively: Options. (line 1160)
8166 * synthesizing linker: Options. (line 1134)
8167 * synthesizing on H8/300: H8/300. (line 14)
8168 * TARGET(BFDNAME): Format Commands. (line 35)
8169 * TARGET1: ARM. (line 32)
8170 * TARGET2: ARM. (line 37)
8171 * text segment origin, cmd line: Options. (line 1332)
8172 * thumb entry point: ARM. (line 17)
8173 * TI COFF versions: TI COFF. (line 6)
8174 * traditional format: Options. (line 1304)
8175 * trampoline generation on M68HC11: M68HC11/68HC12. (line 31)
8176 * trampoline generation on M68HC12: M68HC11/68HC12. (line 31)
8177 * unallocated address, next: Builtin Functions. (line 156)
8178 * undefined symbol: Options. (line 575)
8179 * undefined symbol in linker script: Miscellaneous Commands.
8181 * undefined symbols, warnings on: Options. (line 1481)
8182 * uninitialized data placement: Input Section Common.
8184 * unspecified memory: Output Section Data.
8186 * usage: Options. (line 1002)
8187 * USE_BLX: ARM. (line 74)
8188 * using a DEF file: WIN32. (line 57)
8189 * using auto-export functionality: WIN32. (line 22)
8190 * Using decorations: WIN32. (line 162)
8191 * variables, defining: Assignments. (line 6)
8192 * verbose[=NUMBER]: Options. (line 1375)
8193 * version: Options. (line 602)
8194 * version script: VERSION. (line 6)
8195 * version script, symbol versions: Options. (line 1383)
8196 * VERSION {script text}: VERSION. (line 6)
8197 * versions of symbols: VERSION. (line 6)
8198 * VFP11_DENORM_FIX: ARM. (line 83)
8199 * warnings, on combining symbols: Options. (line 1394)
8200 * warnings, on section alignment: Options. (line 1485)
8201 * warnings, on undefined symbols: Options. (line 1481)
8202 * weak externals: WIN32. (line 407)
8203 * what is this?: Overview. (line 6)
8204 * wildcard file name patterns: Input Section Wildcards.
8206 * Xtensa options: Xtensa. (line 56)
8207 * Xtensa processors: Xtensa. (line 6)
8213 Node: Overview
\7f1493
8214 Node: Invocation
\7f2607
8215 Node: Options
\7f3015
8216 Node: Environment
\7f98821
8217 Node: Scripts
\7f100581
8218 Node: Basic Script Concepts
\7f102315
8219 Node: Script Format
\7f105023
8220 Node: Simple Example
\7f105886
8221 Node: Simple Commands
\7f108982
8222 Node: Entry Point
\7f109488
8223 Node: File Commands
\7f110421
8224 Node: Format Commands
\7f114541
8225 Node: REGION_ALIAS
\7f116497
8226 Node: Miscellaneous Commands
\7f121329
8227 Node: Assignments
\7f124937
8228 Node: Simple Assignments
\7f125448
8229 Node: HIDDEN
\7f127183
8230 Node: PROVIDE
\7f127813
8231 Node: PROVIDE_HIDDEN
\7f129006
8232 Node: Source Code Reference
\7f129250
8233 Node: SECTIONS
\7f132832
8234 Node: Output Section Description
\7f134723
8235 Node: Output Section Name
\7f135967
8236 Node: Output Section Address
\7f136843
8237 Node: Input Section
\7f139078
8238 Node: Input Section Basics
\7f139879
8239 Node: Input Section Wildcards
\7f143785
8240 Node: Input Section Common
\7f148992
8241 Node: Input Section Keep
\7f150474
8242 Node: Input Section Example
\7f150964
8243 Node: Output Section Data
\7f151932
8244 Node: Output Section Keywords
\7f154709
8245 Node: Output Section Discarding
\7f158278
8246 Node: Output Section Attributes
\7f159771
8247 Node: Output Section Type
\7f160872
8248 Node: Output Section LMA
\7f161943
8249 Node: Forced Output Alignment
\7f165014
8250 Node: Forced Input Alignment
\7f165444
8251 Node: Output Section Constraint
\7f165833
8252 Node: Output Section Region
\7f166261
8253 Node: Output Section Phdr
\7f166694
8254 Node: Output Section Fill
\7f167358
8255 Node: Overlay Description
\7f168500
8256 Node: MEMORY
\7f172946
8257 Node: PHDRS
\7f177281
8258 Node: VERSION
\7f182535
8259 Node: Expressions
\7f190628
8260 Node: Constants
\7f191557
8261 Node: Symbolic Constants
\7f192432
8262 Node: Symbols
\7f192983
8263 Node: Orphan Sections
\7f193730
8264 Node: Location Counter
\7f194895
8265 Node: Operators
\7f199331
8266 Node: Evaluation
\7f200253
8267 Node: Expression Section
\7f201617
8268 Node: Builtin Functions
\7f205481
8269 Node: Implicit Linker Scripts
\7f213677
8270 Node: Machine Dependent
\7f214452
8271 Node: H8/300
\7f215576
8273 Node: M68HC11/68HC12
\7f219334
8275 Node: HPPA ELF32
\7f229814
8279 Node: MSP430
\7f234035
8280 Node: NDS32
\7f235075
8281 Node: Nios II
\7f236041
8282 Node: PowerPC ELF32
\7f237357
8283 Node: PowerPC64 ELF64
\7f240188
8284 Node: SPU ELF
\7f246422
8285 Node: TI COFF
\7f249054
8286 Node: WIN32
\7f249580
8287 Node: Xtensa
\7f269706
8289 Node: BFD outline
\7f274126
8290 Node: BFD information loss
\7f275412
8291 Node: Canonical format
\7f277929
8292 Node: Reporting Bugs
\7f282286
8293 Node: Bug Criteria
\7f282980
8294 Node: Bug Reporting
\7f283679
8296 Node: GNU Free Documentation License
\7f295361
8297 Node: LD Index
\7f320517