1 This is ld.info, produced by makeinfo version 4.11 from ../.././../binutils/binutils-2.19/ld/ld.texinfo.
4 * Ld: (ld). The GNU linker.
7 This file documents the GNU linker LD (GNU Binutils) version 2.19.
9 Copyright (C) 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
10 2002, 2003, 2004, 2005, 2006, 2007 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 1.1 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 2.19.
27 This document is distributed under the terms of the GNU Free
28 Documentation License. A copy of the license is included in the
29 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,--startgroup foo.o bar.o -Wl,--endgroup
165 This is important, because otherwise the compiler driver program may
166 silently drop the linker options, resulting in a bad link.
168 Here is a table of the generic command line switches accepted by the
172 Read command-line options from FILE. The options read are
173 inserted in place of the original @FILE option. If FILE does not
174 exist, or cannot be read, then the option will be treated
175 literally, and not removed.
177 Options in FILE are separated by whitespace. A whitespace
178 character may be included in an option by surrounding the entire
179 option in either single or double quotes. Any character
180 (including a backslash) may be included by prefixing the character
181 to be included with a backslash. The FILE may itself contain
182 additional @FILE options; any such options will be processed
186 This option is supported for HP/UX compatibility. The KEYWORD
187 argument must be one of the strings `archive', `shared', or
188 `default'. `-aarchive' is functionally equivalent to `-Bstatic',
189 and the other two keywords are functionally equivalent to
190 `-Bdynamic'. This option may be used any number of times.
193 `--architecture=ARCHITECTURE'
194 In the current release of `ld', this option is useful only for the
195 Intel 960 family of architectures. In that `ld' configuration, the
196 ARCHITECTURE argument identifies the particular architecture in
197 the 960 family, enabling some safeguards and modifying the
198 archive-library search path. *Note `ld' and the Intel 960 family:
201 Future releases of `ld' may support similar functionality for
202 other architecture families.
205 `--format=INPUT-FORMAT'
206 `ld' may be configured to support more than one kind of object
207 file. If your `ld' is configured this way, you can use the `-b'
208 option to specify the binary format for input object files that
209 follow this option on the command line. Even when `ld' is
210 configured to support alternative object formats, you don't
211 usually need to specify this, as `ld' should be configured to
212 expect as a default input format the most usual format on each
213 machine. INPUT-FORMAT is a text string, the name of a particular
214 format supported by the BFD libraries. (You can list the
215 available binary formats with `objdump -i'.) *Note BFD::.
217 You may want to use this option if you are linking files with an
218 unusual binary format. You can also use `-b' to switch formats
219 explicitly (when linking object files of different formats), by
220 including `-b INPUT-FORMAT' before each group of object files in a
223 The default format is taken from the environment variable
224 `GNUTARGET'. *Note Environment::. You can also define the input
225 format from a script, using the command `TARGET'; see *note Format
229 `--mri-script=MRI-COMMANDFILE'
230 For compatibility with linkers produced by MRI, `ld' accepts script
231 files written in an alternate, restricted command language,
232 described in *note MRI Compatible Script Files: MRI. Introduce
233 MRI script files with the option `-c'; use the `-T' option to run
234 linker scripts written in the general-purpose `ld' scripting
235 language. If MRI-CMDFILE does not exist, `ld' looks for it in the
236 directories specified by any `-L' options.
241 These three options are equivalent; multiple forms are supported
242 for compatibility with other linkers. They assign space to common
243 symbols even if a relocatable output file is specified (with
244 `-r'). The script command `FORCE_COMMON_ALLOCATION' has the same
245 effect. *Note Miscellaneous Commands::.
249 Use ENTRY as the explicit symbol for beginning execution of your
250 program, rather than the default entry point. If there is no
251 symbol named ENTRY, the linker will try to parse ENTRY as a number,
252 and use that as the entry address (the number will be interpreted
253 in base 10; you may use a leading `0x' for base 16, or a leading
254 `0' for base 8). *Note Entry Point::, for a discussion of defaults
255 and other ways of specifying the entry point.
257 `--exclude-libs LIB,LIB,...'
258 Specifies a list of archive libraries from which symbols should
259 not be automatically exported. The library names may be delimited
260 by commas or colons. Specifying `--exclude-libs ALL' excludes
261 symbols in all archive libraries from automatic export. This
262 option is available only for the i386 PE targeted port of the
263 linker and for ELF targeted ports. For i386 PE, symbols
264 explicitly listed in a .def file are still exported, regardless of
265 this option. For ELF targeted ports, symbols affected by this
266 option will be treated as hidden.
270 When creating a dynamically linked executable, add all symbols to
271 the dynamic symbol table. The dynamic symbol table is the set of
272 symbols which are visible from dynamic objects at run time.
274 If you do not use this option, the dynamic symbol table will
275 normally contain only those symbols which are referenced by some
276 dynamic object mentioned in the link.
278 If you use `dlopen' to load a dynamic object which needs to refer
279 back to the symbols defined by the program, rather than some other
280 dynamic object, then you will probably need to use this option when
281 linking the program itself.
283 You can also use the dynamic list to control what symbols should
284 be added to the dynamic symbol table if the output format supports
285 it. See the description of `--dynamic-list'.
288 Link big-endian objects. This affects the default output format.
291 Link little-endian objects. This affects the default output
296 When creating an ELF shared object, set the internal DT_AUXILIARY
297 field to the specified name. This tells the dynamic linker that
298 the symbol table of the shared object should be used as an
299 auxiliary filter on the symbol table of the shared object NAME.
301 If you later link a program against this filter object, then, when
302 you run the program, the dynamic linker will see the DT_AUXILIARY
303 field. If the dynamic linker resolves any symbols from the filter
304 object, it will first check whether there is a definition in the
305 shared object NAME. If there is one, it will be used instead of
306 the definition in the filter object. The shared object NAME need
307 not exist. Thus the shared object NAME may be used to provide an
308 alternative implementation of certain functions, perhaps for
309 debugging or for machine specific performance.
311 This option may be specified more than once. The DT_AUXILIARY
312 entries will be created in the order in which they appear on the
317 When creating an ELF shared object, set the internal DT_FILTER
318 field to the specified name. This tells the dynamic linker that
319 the symbol table of the shared object which is being created
320 should be used as a filter on the symbol table of the shared
323 If you later link a program against this filter object, then, when
324 you run the program, the dynamic linker will see the DT_FILTER
325 field. The dynamic linker will resolve symbols according to the
326 symbol table of the filter object as usual, but it will actually
327 link to the definitions found in the shared object NAME. Thus the
328 filter object can be used to select a subset of the symbols
329 provided by the object NAME.
331 Some older linkers used the `-F' option throughout a compilation
332 toolchain for specifying object-file format for both input and
333 output object files. The GNU linker uses other mechanisms for
334 this purpose: the `-b', `--format', `--oformat' options, the
335 `TARGET' command in linker scripts, and the `GNUTARGET'
336 environment variable. The GNU linker will ignore the `-F' option
337 when not creating an ELF shared object.
340 When creating an ELF executable or shared object, call NAME when
341 the executable or shared object is unloaded, by setting DT_FINI to
342 the address of the function. By default, the linker uses `_fini'
343 as the function to call.
346 Ignored. Provided for compatibility with other tools.
350 Set the maximum size of objects to be optimized using the GP
351 register to SIZE. This is only meaningful for object file formats
352 such as MIPS ECOFF which supports putting large and small objects
353 into different sections. This is ignored for other object file
358 When creating an ELF shared object, set the internal DT_SONAME
359 field to the specified name. When an executable is linked with a
360 shared object which has a DT_SONAME field, then when the
361 executable is run the dynamic linker will attempt to load the
362 shared object specified by the DT_SONAME field rather than the
363 using the file name given to the linker.
366 Perform an incremental link (same as option `-r').
369 When creating an ELF executable or shared object, call NAME when
370 the executable or shared object is loaded, by setting DT_INIT to
371 the address of the function. By default, the linker uses `_init'
372 as the function to call.
376 Add the archive or object file specified by NAMESPEC to the list
377 of files to link. This option may be used any number of times.
378 If NAMESPEC is of the form `:FILENAME', `ld' will search the
379 library path for a file called FILENAME, otherise it will search
380 the library path for a file called `libNAMESPEC.a'.
382 On systems which support shared libraries, `ld' may also search for
383 files other than `libNAMESPEC.a'. Specifically, on ELF and SunOS
384 systems, `ld' will search a directory for a library called
385 `libNAMESPEC.so' before searching for one called `libNAMESPEC.a'.
386 (By convention, a `.so' extension indicates a shared library.)
387 Note that this behavior does not apply to `:FILENAME', which
388 always specifies a file called FILENAME.
390 The linker will search an archive only once, at the location where
391 it is specified on the command line. If the archive defines a
392 symbol which was undefined in some object which appeared before
393 the archive on the command line, the linker will include the
394 appropriate file(s) from the archive. However, an undefined
395 symbol in an object appearing later on the command line will not
396 cause the linker to search the archive again.
398 See the `-(' option for a way to force the linker to search
399 archives multiple times.
401 You may list the same archive multiple times on the command line.
403 This type of archive searching is standard for Unix linkers.
404 However, if you are using `ld' on AIX, note that it is different
405 from the behaviour of the AIX linker.
408 `--library-path=SEARCHDIR'
409 Add path SEARCHDIR to the list of paths that `ld' will search for
410 archive libraries and `ld' control scripts. You may use this
411 option any number of times. The directories are searched in the
412 order in which they are specified on the command line.
413 Directories specified on the command line are searched before the
414 default directories. All `-L' options apply to all `-l' options,
415 regardless of the order in which the options appear.
417 If SEARCHDIR begins with `=', then the `=' will be replaced by the
418 "sysroot prefix", a path specified when the linker is configured.
420 The default set of paths searched (without being specified with
421 `-L') depends on which emulation mode `ld' is using, and in some
422 cases also on how it was configured. *Note Environment::.
424 The paths can also be specified in a link script with the
425 `SEARCH_DIR' command. Directories specified this way are searched
426 at the point in which the linker script appears in the command
430 Emulate the EMULATION linker. You can list the available
431 emulations with the `--verbose' or `-V' options.
433 If the `-m' option is not used, the emulation is taken from the
434 `LDEMULATION' environment variable, if that is defined.
436 Otherwise, the default emulation depends upon how the linker was
441 Print a link map to the standard output. A link map provides
442 information about the link, including the following:
444 * Where object files are mapped into memory.
446 * How common symbols are allocated.
448 * All archive members included in the link, with a mention of
449 the symbol which caused the archive member to be brought in.
451 * The values assigned to symbols.
453 Note - symbols whose values are computed by an expression
454 which involves a reference to a previous value of the same
455 symbol may not have correct result displayed in the link map.
456 This is because the linker discards intermediate results and
457 only retains the final value of an expression. Under such
458 circumstances the linker will display the final value
459 enclosed by square brackets. Thus for example a linker
466 will produce the following output in the link map if the `-M'
470 [0x0000000c] foo = (foo * 0x4)
471 [0x0000000c] foo = (foo + 0x8)
473 See *note Expressions:: for more information about
474 expressions in linker scripts.
478 Turn off page alignment of sections, and mark the output as
479 `NMAGIC' if possible.
483 Set the text and data sections to be readable and writable. Also,
484 do not page-align the data segment, and disable linking against
485 shared libraries. If the output format supports Unix style magic
486 numbers, mark the output as `OMAGIC'. Note: Although a writable
487 text section is allowed for PE-COFF targets, it does not conform
488 to the format specification published by Microsoft.
491 This option negates most of the effects of the `-N' option. It
492 sets the text section to be read-only, and forces the data segment
493 to be page-aligned. Note - this option does not enable linking
494 against shared libraries. Use `-Bdynamic' for this.
498 Use OUTPUT as the name for the program produced by `ld'; if this
499 option is not specified, the name `a.out' is used by default. The
500 script command `OUTPUT' can also specify the output file name.
503 If LEVEL is a numeric values greater than zero `ld' optimizes the
504 output. This might take significantly longer and therefore
505 probably should only be enabled for the final binary. At the
506 moment this option only affects ELF shared library generation.
507 Future releases of the linker may make more use of this option.
508 Also currently there is no difference in the linker's behaviour
509 for different non-zero values of this option. Again this may
510 change with future releases.
514 Leave relocation sections and contents in fully linked executables.
515 Post link analysis and optimization tools may need this
516 information in order to perform correct modifications of
517 executables. This results in larger executables.
519 This option is currently only supported on ELF platforms.
522 Force the output file to have dynamic sections. This option is
523 specific to VxWorks targets.
527 Generate relocatable output--i.e., generate an output file that
528 can in turn serve as input to `ld'. This is often called "partial
529 linking". As a side effect, in environments that support standard
530 Unix magic numbers, this option also sets the output file's magic
531 number to `OMAGIC'. If this option is not specified, an absolute
532 file is produced. When linking C++ programs, this option _will
533 not_ resolve references to constructors; to do that, use `-Ur'.
535 When an input file does not have the same format as the output
536 file, partial linking is only supported if that input file does
537 not contain any relocations. Different output formats can have
538 further restrictions; for example some `a.out'-based formats do
539 not support partial linking with input files in other formats at
542 This option does the same thing as `-i'.
545 `--just-symbols=FILENAME'
546 Read symbol names and their addresses from FILENAME, but do not
547 relocate it or include it in the output. This allows your output
548 file to refer symbolically to absolute locations of memory defined
549 in other programs. You may use this option more than once.
551 For compatibility with other ELF linkers, if the `-R' option is
552 followed by a directory name, rather than a file name, it is
553 treated as the `-rpath' option.
557 Omit all symbol information from the output file.
561 Omit debugger symbol information (but not all symbols) from the
566 Print the names of the input files as `ld' processes them.
569 `--script=SCRIPTFILE'
570 Use SCRIPTFILE as the linker script. This script replaces `ld''s
571 default linker script (rather than adding to it), so COMMANDFILE
572 must specify everything necessary to describe the output file.
573 *Note Scripts::. If SCRIPTFILE does not exist in the current
574 directory, `ld' looks for it in the directories specified by any
575 preceding `-L' options. Multiple `-T' options accumulate.
578 `--default-script=SCRIPTFILE'
579 Use SCRIPTFILE as the default linker script. *Note Scripts::.
581 This option is similar to the `--script' option except that
582 processing of the script is delayed until after the rest of the
583 command line has been processed. This allows options placed after
584 the `--default-script' option on the command line to affect the
585 behaviour of the linker script, which can be important when the
586 linker command line cannot be directly controlled by the user.
587 (eg because the command line is being constructed by another tool,
592 Force SYMBOL to be entered in the output file as an undefined
593 symbol. Doing this may, for example, trigger linking of additional
594 modules from standard libraries. `-u' may be repeated with
595 different option arguments to enter additional undefined symbols.
596 This option is equivalent to the `EXTERN' linker script command.
599 For anything other than C++ programs, this option is equivalent to
600 `-r': it generates relocatable output--i.e., an output file that
601 can in turn serve as input to `ld'. When linking C++ programs,
602 `-Ur' _does_ resolve references to constructors, unlike `-r'. It
603 does not work to use `-Ur' on files that were themselves linked
604 with `-Ur'; once the constructor table has been built, it cannot
605 be added to. Use `-Ur' only for the last partial link, and `-r'
609 Creates a separate output section for every input section matching
610 SECTION, or if the optional wildcard SECTION argument is missing,
611 for every orphan input section. An orphan section is one not
612 specifically mentioned in a linker script. You may use this option
613 multiple times on the command line; It prevents the normal
614 merging of input sections with the same name, overriding output
615 section assignments in a linker script.
620 Display the version number for `ld'. The `-V' option also lists
621 the supported emulations.
625 Delete all local symbols.
629 Delete all temporary local symbols. (These symbols start with
630 system-specific local label prefixes, typically `.L' for ELF
631 systems or `L' for traditional a.out systems.)
634 `--trace-symbol=SYMBOL'
635 Print the name of each linked file in which SYMBOL appears. This
636 option may be given any number of times. On many systems it is
637 necessary to prepend an underscore.
639 This option is useful when you have an undefined symbol in your
640 link but don't know where the reference is coming from.
643 Add PATH to the default library search path. This option exists
644 for Solaris compatibility.
647 The recognized keywords are:
649 Combines multiple reloc sections and sorts them to make
650 dynamic symbol lookup caching possible.
653 Disallows undefined symbols in object files. Undefined
654 symbols in shared libraries are still allowed.
657 Marks the object as requiring executable stack.
660 This option is only meaningful when building a shared object.
661 It marks the object so that its runtime initialization will
662 occur before the runtime initialization of any other objects
663 brought into the process at the same time. Similarly the
664 runtime finalization of the object will occur after the
665 runtime finalization of any other objects.
668 Marks the object that its symbol table interposes before all
669 symbols but the primary executable.
672 When generating an executable or shared library, mark it to
673 tell the dynamic linker to defer function call resolution to
674 the point when the function is called (lazy binding), rather
675 than at load time. Lazy binding is the default.
678 Marks the object that its filters be processed immediately at
682 Allows multiple definitions.
685 Disables multiple reloc sections combining.
688 Disables production of copy relocs.
691 Marks the object that the search for dependencies of this
692 object will ignore any default library search paths.
695 Marks the object shouldn't be unloaded at runtime.
698 Marks the object not available to `dlopen'.
701 Marks the object can not be dumped by `dldump'.
704 Marks the object as not requiring executable stack.
707 Don't create an ELF `PT_GNU_RELRO' segment header in the
711 When generating an executable or shared library, mark it to
712 tell the dynamic linker to resolve all symbols when the
713 program is started, or when the shared library is linked to
714 using dlopen, instead of deferring function call resolution
715 to the point when the function is first called.
718 Marks the object may contain $ORIGIN.
721 Create an ELF `PT_GNU_RELRO' segment header in the object.
723 `max-page-size=VALUE'
724 Set the emulation maximum page size to VALUE.
726 `common-page-size=VALUE'
727 Set the emulation common page size to VALUE.
730 Other keywords are ignored for Solaris compatibility.
733 `--start-group ARCHIVES --end-group'
734 The ARCHIVES should be a list of archive files. They may be
735 either explicit file names, or `-l' options.
737 The specified archives are searched repeatedly until no new
738 undefined references are created. Normally, an archive is
739 searched only once in the order that it is specified on the
740 command line. If a symbol in that archive is needed to resolve an
741 undefined symbol referred to by an object in an archive that
742 appears later on the command line, the linker would not be able to
743 resolve that reference. By grouping the archives, they all be
744 searched repeatedly until all possible references are resolved.
746 Using this option has a significant performance cost. It is best
747 to use it only when there are unavoidable circular references
748 between two or more archives.
750 `--accept-unknown-input-arch'
751 `--no-accept-unknown-input-arch'
752 Tells the linker to accept input files whose architecture cannot be
753 recognised. The assumption is that the user knows what they are
754 doing and deliberately wants to link in these unknown input files.
755 This was the default behaviour of the linker, before release 2.14.
756 The default behaviour from release 2.14 onwards is to reject such
757 input files, and so the `--accept-unknown-input-arch' option has
758 been added to restore the old behaviour.
762 This option affects ELF DT_NEEDED tags for dynamic libraries
763 mentioned on the command line after the `--as-needed' option.
764 Normally, the linker will add a DT_NEEDED tag for each dynamic
765 library mentioned on the command line, regardless of whether the
766 library is actually needed. `--as-needed' causes DT_NEEDED tags
767 to only be emitted for libraries that satisfy some symbol
768 reference from regular objects which is undefined at the point
769 that the library was linked. `--no-as-needed' restores the
774 This option affects the treatment of dynamic libraries from ELF
775 DT_NEEDED tags in dynamic libraries mentioned on the command line
776 after the `--no-add-needed' option. Normally, the linker will add
777 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
778 `--no-add-needed' causes DT_NEEDED tags will never be emitted for
779 those libraries from DT_NEEDED tags. `--add-needed' restores the
783 This option is ignored for SunOS compatibility.
788 Link against dynamic libraries. This is only meaningful on
789 platforms for which shared libraries are supported. This option
790 is normally the default on such platforms. The different variants
791 of this option are for compatibility with various systems. You
792 may use this option multiple times on the command line: it affects
793 library searching for `-l' options which follow it.
796 Set the `DF_1_GROUP' flag in the `DT_FLAGS_1' entry in the dynamic
797 section. This causes the runtime linker to handle lookups in this
798 object and its dependencies to be performed only inside the group.
799 `--unresolved-symbols=report-all' is implied. This option is only
800 meaningful on ELF platforms which support shared libraries.
806 Do not link against shared libraries. This is only meaningful on
807 platforms for which shared libraries are supported. The different
808 variants of this option are for compatibility with various
809 systems. You may use this option multiple times on the command
810 line: it affects library searching for `-l' options which follow
811 it. This option also implies `--unresolved-symbols=report-all'.
812 This option can be used with `-shared'. Doing so means that a
813 shared library is being created but that all of the library's
814 external references must be resolved by pulling in entries from
818 When creating a shared library, bind references to global symbols
819 to the definition within the shared library, if any. Normally, it
820 is possible for a program linked against a shared library to
821 override the definition within the shared library. This option is
822 only meaningful on ELF platforms which support shared libraries.
824 `-Bsymbolic-functions'
825 When creating a shared library, bind references to global function
826 symbols to the definition within the shared library, if any. This
827 option is only meaningful on ELF platforms which support shared
830 `--dynamic-list=DYNAMIC-LIST-FILE'
831 Specify the name of a dynamic list file to the linker. This is
832 typically used when creating shared libraries to specify a list of
833 global symbols whose references shouldn't be bound to the
834 definition within the shared library, or creating dynamically
835 linked executables to specify a list of symbols which should be
836 added to the symbol table in the executable. This option is only
837 meaningful on ELF platforms which support shared libraries.
839 The format of the dynamic list is the same as the version node
840 without scope and node name. See *note VERSION:: for more
843 `--dynamic-list-data'
844 Include all global data symbols to the dynamic list.
846 `--dynamic-list-cpp-new'
847 Provide the builtin dynamic list for C++ operator new and delete.
848 It is mainly useful for building shared libstdc++.
850 `--dynamic-list-cpp-typeinfo'
851 Provide the builtin dynamic list for C++ runtime type
855 `--no-check-sections'
856 Asks the linker _not_ to check section addresses after they have
857 been assigned to see if there are any overlaps. Normally the
858 linker will perform this check, and if it finds any overlaps it
859 will produce suitable error messages. The linker does know about,
860 and does make allowances for sections in overlays. The default
861 behaviour can be restored by using the command line switch
865 Output a cross reference table. If a linker map file is being
866 generated, the cross reference table is printed to the map file.
867 Otherwise, it is printed on the standard output.
869 The format of the table is intentionally simple, so that it may be
870 easily processed by a script if necessary. The symbols are
871 printed out, sorted by name. For each symbol, a list of file
872 names is given. If the symbol is defined, the first file listed
873 is the location of the definition. The remaining files contain
874 references to the symbol.
877 This option inhibits the assignment of addresses to common symbols.
878 The script command `INHIBIT_COMMON_ALLOCATION' has the same effect.
879 *Note Miscellaneous Commands::.
881 The `--no-define-common' option allows decoupling the decision to
882 assign addresses to Common symbols from the choice of the output
883 file type; otherwise a non-Relocatable output type forces
884 assigning addresses to Common symbols. Using `--no-define-common'
885 allows Common symbols that are referenced from a shared library to
886 be assigned addresses only in the main program. This eliminates
887 the unused duplicate space in the shared library, and also
888 prevents any possible confusion over resolving to the wrong
889 duplicate when there are many dynamic modules with specialized
890 search paths for runtime symbol resolution.
892 `--defsym SYMBOL=EXPRESSION'
893 Create a global symbol in the output file, containing the absolute
894 address given by EXPRESSION. You may use this option as many
895 times as necessary to define multiple symbols in the command line.
896 A limited form of arithmetic is supported for the EXPRESSION in
897 this context: you may give a hexadecimal constant or the name of
898 an existing symbol, or use `+' and `-' to add or subtract
899 hexadecimal constants or symbols. If you need more elaborate
900 expressions, consider using the linker command language from a
901 script (*note Assignment: Symbol Definitions: Assignments.).
902 _Note:_ there should be no white space between SYMBOL, the equals
903 sign ("<=>"), and EXPRESSION.
907 These options control whether to demangle symbol names in error
908 messages and other output. When the linker is told to demangle,
909 it tries to present symbol names in a readable fashion: it strips
910 leading underscores if they are used by the object file format,
911 and converts C++ mangled symbol names into user readable names.
912 Different compilers have different mangling styles. The optional
913 demangling style argument can be used to choose an appropriate
914 demangling style for your compiler. The linker will demangle by
915 default unless the environment variable `COLLECT_NO_DEMANGLE' is
916 set. These options may be used to override the default.
918 `--dynamic-linker FILE'
919 Set the name of the dynamic linker. This is only meaningful when
920 generating dynamically linked ELF executables. The default dynamic
921 linker is normally correct; don't use this unless you know what
925 `--no-fatal-warnings'
926 Treat all warnings as errors. The default behaviour can be
927 restored with the option `--no-fatal-warnings'.
930 Make sure that an output file has a .exe suffix.
932 If a successfully built fully linked output file does not have a
933 `.exe' or `.dll' suffix, this option forces the linker to copy the
934 output file to one of the same name with a `.exe' suffix. This
935 option is useful when using unmodified Unix makefiles on a
936 Microsoft Windows host, since some versions of Windows won't run
937 an image unless it ends in a `.exe' suffix.
941 Enable garbage collection of unused input sections. It is ignored
942 on targets that do not support this option. The default behaviour
943 (of not performing this garbage collection) can be restored by
944 specifying `--no-gc-sections' on the command line.
946 `--gc-sections' decides which input sections are used by examining
947 symbols and relocations. The section containing the entry symbol
948 and all sections containing symbols undefined on the command-line
949 will be kept, as will sections containing symbols referenced by
950 dynamic objects. Note that when building shared libraries, the
951 linker must assume that any visible symbol is referenced. Once
952 this initial set of sections has been determined, the linker
953 recursively marks as used any section referenced by their
954 relocations. See `--entry' and `--undefined'.
956 This option can be set when doing a partial link (enabled with
957 option `-r'). In this case the root of symbols kept must be
958 explicitely specified either by an `--entry' or `--undefined'
959 option or by a `ENTRY' command in the linker script.
961 `--print-gc-sections'
962 `--no-print-gc-sections'
963 List all sections removed by garbage collection. The listing is
964 printed on stderr. This option is only effective if garbage
965 collection has been enabled via the `--gc-sections') option. The
966 default behaviour (of not listing the sections that are removed)
967 can be restored by specifying `--no-print-gc-sections' on the
971 Print a summary of the command-line options on the standard output
975 Print a summary of all target specific options on the standard
979 Print a link map to the file MAPFILE. See the description of the
983 `ld' normally optimizes for speed over memory usage by caching the
984 symbol tables of input files in memory. This option tells `ld' to
985 instead optimize for memory usage, by rereading the symbol tables
986 as necessary. This may be required if `ld' runs out of memory
987 space while linking a large executable.
991 Report unresolved symbol references from regular object files.
992 This is done even if the linker is creating a non-symbolic shared
993 library. The switch `--[no-]allow-shlib-undefined' controls the
994 behaviour for reporting unresolved references found in shared
995 libraries being linked in.
997 `--allow-multiple-definition'
999 Normally when a symbol is defined multiple times, the linker will
1000 report a fatal error. These options allow multiple definitions and
1001 the first definition will be used.
1003 `--allow-shlib-undefined'
1004 `--no-allow-shlib-undefined'
1005 Allows (the default) or disallows undefined symbols in shared
1006 libraries. This switch is similar to `--no-undefined' except that
1007 it determines the behaviour when the undefined symbols are in a
1008 shared library rather than a regular object file. It does not
1009 affect how undefined symbols in regular object files are handled.
1011 The reason that `--allow-shlib-undefined' is the default is that
1012 the shared library being specified at link time may not be the
1013 same as the one that is available at load time, so the symbols
1014 might actually be resolvable at load time. Plus there are some
1015 systems, (eg BeOS) where undefined symbols in shared libraries is
1016 normal. (The kernel patches them at load time to select which
1017 function is most appropriate for the current architecture. This
1018 is used for example to dynamically select an appropriate memset
1019 function). Apparently it is also normal for HPPA shared libraries
1020 to have undefined symbols.
1022 `--no-undefined-version'
1023 Normally when a symbol has an undefined version, the linker will
1024 ignore it. This option disallows symbols with undefined version
1025 and a fatal error will be issued instead.
1028 Create and use a default symbol version (the soname) for
1029 unversioned exported symbols.
1031 `--default-imported-symver'
1032 Create and use a default symbol version (the soname) for
1033 unversioned imported symbols.
1035 `--no-warn-mismatch'
1036 Normally `ld' will give an error if you try to link together input
1037 files that are mismatched for some reason, perhaps because they
1038 have been compiled for different processors or for different
1039 endiannesses. This option tells `ld' that it should silently
1040 permit such possible errors. This option should only be used with
1041 care, in cases when you have taken some special action that
1042 ensures that the linker errors are inappropriate.
1044 `--no-warn-search-mismatch'
1045 Normally `ld' will give a warning if it finds an incompatible
1046 library during a library search. This option silences the warning.
1048 `--no-whole-archive'
1049 Turn off the effect of the `--whole-archive' option for subsequent
1053 Retain the executable output file whenever it is still usable.
1054 Normally, the linker will not produce an output file if it
1055 encounters errors during the link process; it exits without
1056 writing an output file when it issues any error whatsoever.
1059 Only search library directories explicitly specified on the
1060 command line. Library directories specified in linker scripts
1061 (including linker scripts specified on the command line) are
1064 `--oformat OUTPUT-FORMAT'
1065 `ld' may be configured to support more than one kind of object
1066 file. If your `ld' is configured this way, you can use the
1067 `--oformat' option to specify the binary format for the output
1068 object file. Even when `ld' is configured to support alternative
1069 object formats, you don't usually need to specify this, as `ld'
1070 should be configured to produce as a default output format the most
1071 usual format on each machine. OUTPUT-FORMAT is a text string, the
1072 name of a particular format supported by the BFD libraries. (You
1073 can list the available binary formats with `objdump -i'.) The
1074 script command `OUTPUT_FORMAT' can also specify the output format,
1075 but this option overrides it. *Note BFD::.
1079 Create a position independent executable. This is currently only
1080 supported on ELF platforms. Position independent executables are
1081 similar to shared libraries in that they are relocated by the
1082 dynamic linker to the virtual address the OS chooses for them
1083 (which can vary between invocations). Like normal dynamically
1084 linked executables they can be executed and symbols defined in the
1085 executable cannot be overridden by shared libraries.
1088 This option is ignored for Linux compatibility.
1091 This option is ignored for SVR4 compatibility.
1094 An option with machine dependent effects. This option is only
1095 supported on a few targets. *Note `ld' and the H8/300: H8/300.
1096 *Note `ld' and the Intel 960 family: i960. *Note `ld' and Xtensa
1097 Processors: Xtensa. *Note `ld' and the 68HC11 and 68HC12:
1098 M68HC11/68HC12. *Note `ld' and PowerPC 32-bit ELF Support:
1101 On some platforms, the `--relax' option performs global
1102 optimizations that become possible when the linker resolves
1103 addressing in the program, such as relaxing address modes and
1104 synthesizing new instructions in the output object file.
1106 On some platforms these link time global optimizations may make
1107 symbolic debugging of the resulting executable impossible. This
1108 is known to be the case for the Matsushita MN10200 and MN10300
1109 family of processors.
1111 On platforms where this is not supported, `--relax' is accepted,
1114 `--retain-symbols-file FILENAME'
1115 Retain _only_ the symbols listed in the file FILENAME, discarding
1116 all others. FILENAME is simply a flat file, with one symbol name
1117 per line. This option is especially useful in environments (such
1118 as VxWorks) where a large global symbol table is accumulated
1119 gradually, to conserve run-time memory.
1121 `--retain-symbols-file' does _not_ discard undefined symbols, or
1122 symbols needed for relocations.
1124 You may only specify `--retain-symbols-file' once in the command
1125 line. It overrides `-s' and `-S'.
1128 Add a directory to the runtime library search path. This is used
1129 when linking an ELF executable with shared objects. All `-rpath'
1130 arguments are concatenated and passed to the runtime linker, which
1131 uses them to locate shared objects at runtime. The `-rpath'
1132 option is also used when locating shared objects which are needed
1133 by shared objects explicitly included in the link; see the
1134 description of the `-rpath-link' option. If `-rpath' is not used
1135 when linking an ELF executable, the contents of the environment
1136 variable `LD_RUN_PATH' will be used if it is defined.
1138 The `-rpath' option may also be used on SunOS. By default, on
1139 SunOS, the linker will form a runtime search patch out of all the
1140 `-L' options it is given. If a `-rpath' option is used, the
1141 runtime search path will be formed exclusively using the `-rpath'
1142 options, ignoring the `-L' options. This can be useful when using
1143 gcc, which adds many `-L' options which may be on NFS mounted file
1146 For compatibility with other ELF linkers, if the `-R' option is
1147 followed by a directory name, rather than a file name, it is
1148 treated as the `-rpath' option.
1151 When using ELF or SunOS, one shared library may require another.
1152 This happens when an `ld -shared' link includes a shared library
1153 as one of the input files.
1155 When the linker encounters such a dependency when doing a
1156 non-shared, non-relocatable link, it will automatically try to
1157 locate the required shared library and include it in the link, if
1158 it is not included explicitly. In such a case, the `-rpath-link'
1159 option specifies the first set of directories to search. The
1160 `-rpath-link' option may specify a sequence of directory names
1161 either by specifying a list of names separated by colons, or by
1162 appearing multiple times.
1164 This option should be used with caution as it overrides the search
1165 path that may have been hard compiled into a shared library. In
1166 such a case it is possible to use unintentionally a different
1167 search path than the runtime linker would do.
1169 The linker uses the following search paths to locate required
1171 1. Any directories specified by `-rpath-link' options.
1173 2. Any directories specified by `-rpath' options. The difference
1174 between `-rpath' and `-rpath-link' is that directories
1175 specified by `-rpath' options are included in the executable
1176 and used at runtime, whereas the `-rpath-link' option is only
1177 effective at link time. Searching `-rpath' in this way is
1178 only supported by native linkers and cross linkers which have
1179 been configured with the `--with-sysroot' option.
1181 3. On an ELF system, for native linkers, if the `-rpath' and
1182 `-rpath-link' options were not used, search the contents of
1183 the environment variable `LD_RUN_PATH'.
1185 4. On SunOS, if the `-rpath' option was not used, search any
1186 directories specified using `-L' options.
1188 5. For a native linker, the search the contents of the
1189 environment variable `LD_LIBRARY_PATH'.
1191 6. For a native ELF linker, the directories in `DT_RUNPATH' or
1192 `DT_RPATH' of a shared library are searched for shared
1193 libraries needed by it. The `DT_RPATH' entries are ignored if
1194 `DT_RUNPATH' entries exist.
1196 7. The default directories, normally `/lib' and `/usr/lib'.
1198 8. For a native linker on an ELF system, if the file
1199 `/etc/ld.so.conf' exists, the list of directories found in
1202 If the required shared library is not found, the linker will issue
1203 a warning and continue with the link.
1207 Create a shared library. This is currently only supported on ELF,
1208 XCOFF and SunOS platforms. On SunOS, the linker will
1209 automatically create a shared library if the `-e' option is not
1210 used and there are undefined symbols in the link.
1212 `--sort-common [= ascending | descending]'
1213 This option tells `ld' to sort the common symbols by alignment in
1214 ascending or descending order when it places them in the
1215 appropriate output sections. The symbol alignments considered are
1216 sixteen-byte or larger, eight-byte, four-byte, two-byte, and
1217 one-byte. This is to prevent gaps between symbols due to alignment
1218 constraints. If no sorting order is specified, then descending
1221 `--sort-section name'
1222 This option will apply `SORT_BY_NAME' to all wildcard section
1223 patterns in the linker script.
1225 `--sort-section alignment'
1226 This option will apply `SORT_BY_ALIGNMENT' to all wildcard section
1227 patterns in the linker script.
1229 `--split-by-file [SIZE]'
1230 Similar to `--split-by-reloc' but creates a new output section for
1231 each input file when SIZE is reached. SIZE defaults to a size of
1234 `--split-by-reloc [COUNT]'
1235 Tries to creates extra sections in the output file so that no
1236 single output section in the file contains more than COUNT
1237 relocations. This is useful when generating huge relocatable
1238 files for downloading into certain real time kernels with the COFF
1239 object file format; since COFF cannot represent more than 65535
1240 relocations in a single section. Note that this will fail to work
1241 with object file formats which do not support arbitrary sections.
1242 The linker will not split up individual input sections for
1243 redistribution, so if a single input section contains more than
1244 COUNT relocations one output section will contain that many
1245 relocations. COUNT defaults to a value of 32768.
1248 Compute and display statistics about the operation of the linker,
1249 such as execution time and memory usage.
1251 `--sysroot=DIRECTORY'
1252 Use DIRECTORY as the location of the sysroot, overriding the
1253 configure-time default. This option is only supported by linkers
1254 that were configured using `--with-sysroot'.
1256 `--traditional-format'
1257 For some targets, the output of `ld' is different in some ways from
1258 the output of some existing linker. This switch requests `ld' to
1259 use the traditional format instead.
1261 For example, on SunOS, `ld' combines duplicate entries in the
1262 symbol string table. This can reduce the size of an output file
1263 with full debugging information by over 30 percent.
1264 Unfortunately, the SunOS `dbx' program can not read the resulting
1265 program (`gdb' has no trouble). The `--traditional-format' switch
1266 tells `ld' to not combine duplicate entries.
1268 `--section-start SECTIONNAME=ORG'
1269 Locate a section in the output file at the absolute address given
1270 by ORG. You may use this option as many times as necessary to
1271 locate multiple sections in the command line. ORG must be a
1272 single hexadecimal integer; for compatibility with other linkers,
1273 you may omit the leading `0x' usually associated with hexadecimal
1274 values. _Note:_ there should be no white space between
1275 SECTIONNAME, the equals sign ("<=>"), and ORG.
1280 Same as -section-start, with `.bss', `.data' or `.text' as the
1283 `--unresolved-symbols=METHOD'
1284 Determine how to handle unresolved symbols. There are four
1285 possible values for `method':
1288 Do not report any unresolved symbols.
1291 Report all unresolved symbols. This is the default.
1293 `ignore-in-object-files'
1294 Report unresolved symbols that are contained in shared
1295 libraries, but ignore them if they come from regular object
1298 `ignore-in-shared-libs'
1299 Report unresolved symbols that come from regular object
1300 files, but ignore them if they come from shared libraries.
1301 This can be useful when creating a dynamic binary and it is
1302 known that all the shared libraries that it should be
1303 referencing are included on the linker's command line.
1305 The behaviour for shared libraries on their own can also be
1306 controlled by the `--[no-]allow-shlib-undefined' option.
1308 Normally the linker will generate an error message for each
1309 reported unresolved symbol but the option
1310 `--warn-unresolved-symbols' can change this to a warning.
1314 Display the version number for `ld' and list the linker emulations
1315 supported. Display which input files can and cannot be opened.
1316 Display the linker script being used by the linker.
1318 `--version-script=VERSION-SCRIPTFILE'
1319 Specify the name of a version script to the linker. This is
1320 typically used when creating shared libraries to specify
1321 additional information about the version hierarchy for the library
1322 being created. This option is only meaningful on ELF platforms
1323 which support shared libraries. *Note VERSION::.
1326 Warn when a common symbol is combined with another common symbol
1327 or with a symbol definition. Unix linkers allow this somewhat
1328 sloppy practise, but linkers on some other operating systems do
1329 not. This option allows you to find potential problems from
1330 combining global symbols. Unfortunately, some C libraries use
1331 this practise, so you may get some warnings about symbols in the
1332 libraries as well as in your programs.
1334 There are three kinds of global symbols, illustrated here by C
1338 A definition, which goes in the initialized data section of
1342 An undefined reference, which does not allocate space. There
1343 must be either a definition or a common symbol for the
1347 A common symbol. If there are only (one or more) common
1348 symbols for a variable, it goes in the uninitialized data
1349 area of the output file. The linker merges multiple common
1350 symbols for the same variable into a single symbol. If they
1351 are of different sizes, it picks the largest size. The
1352 linker turns a common symbol into a declaration, if there is
1353 a definition of the same variable.
1355 The `--warn-common' option can produce five kinds of warnings.
1356 Each warning consists of a pair of lines: the first describes the
1357 symbol just encountered, and the second describes the previous
1358 symbol encountered with the same name. One or both of the two
1359 symbols will be a common symbol.
1361 1. Turning a common symbol into a reference, because there is
1362 already a definition for the symbol.
1363 FILE(SECTION): warning: common of `SYMBOL'
1364 overridden by definition
1365 FILE(SECTION): warning: defined here
1367 2. Turning a common symbol into a reference, because a later
1368 definition for the symbol is encountered. This is the same
1369 as the previous case, except that the symbols are encountered
1370 in a different order.
1371 FILE(SECTION): warning: definition of `SYMBOL'
1373 FILE(SECTION): warning: common is here
1375 3. Merging a common symbol with a previous same-sized common
1377 FILE(SECTION): warning: multiple common
1379 FILE(SECTION): warning: previous common is here
1381 4. Merging a common symbol with a previous larger common symbol.
1382 FILE(SECTION): warning: common of `SYMBOL'
1383 overridden by larger common
1384 FILE(SECTION): warning: larger common is here
1386 5. Merging a common symbol with a previous smaller common
1387 symbol. This is the same as the previous case, except that
1388 the symbols are encountered in a different order.
1389 FILE(SECTION): warning: common of `SYMBOL'
1390 overriding smaller common
1391 FILE(SECTION): warning: smaller common is here
1393 `--warn-constructors'
1394 Warn if any global constructors are used. This is only useful for
1395 a few object file formats. For formats like COFF or ELF, the
1396 linker can not detect the use of global constructors.
1398 `--warn-multiple-gp'
1399 Warn if multiple global pointer values are required in the output
1400 file. This is only meaningful for certain processors, such as the
1401 Alpha. Specifically, some processors put large-valued constants
1402 in a special section. A special register (the global pointer)
1403 points into the middle of this section, so that constants can be
1404 loaded efficiently via a base-register relative addressing mode.
1405 Since the offset in base-register relative mode is fixed and
1406 relatively small (e.g., 16 bits), this limits the maximum size of
1407 the constant pool. Thus, in large programs, it is often necessary
1408 to use multiple global pointer values in order to be able to
1409 address all possible constants. This option causes a warning to
1410 be issued whenever this case occurs.
1413 Only warn once for each undefined symbol, rather than once per
1414 module which refers to it.
1416 `--warn-section-align'
1417 Warn if the address of an output section is changed because of
1418 alignment. Typically, the alignment will be set by an input
1419 section. The address will only be changed if it not explicitly
1420 specified; that is, if the `SECTIONS' command does not specify a
1421 start address for the section (*note SECTIONS::).
1423 `--warn-shared-textrel'
1424 Warn if the linker adds a DT_TEXTREL to a shared object.
1426 `--warn-unresolved-symbols'
1427 If the linker is going to report an unresolved symbol (see the
1428 option `--unresolved-symbols') it will normally generate an error.
1429 This option makes it generate a warning instead.
1431 `--error-unresolved-symbols'
1432 This restores the linker's default behaviour of generating errors
1433 when it is reporting unresolved symbols.
1436 For each archive mentioned on the command line after the
1437 `--whole-archive' option, include every object file in the archive
1438 in the link, rather than searching the archive for the required
1439 object files. This is normally used to turn an archive file into
1440 a shared library, forcing every object to be included in the
1441 resulting shared library. This option may be used more than once.
1443 Two notes when using this option from gcc: First, gcc doesn't know
1444 about this option, so you have to use `-Wl,-whole-archive'.
1445 Second, don't forget to use `-Wl,-no-whole-archive' after your
1446 list of archives, because gcc will add its own list of archives to
1447 your link and you may not want this flag to affect those as well.
1450 Use a wrapper function for SYMBOL. Any undefined reference to
1451 SYMBOL will be resolved to `__wrap_SYMBOL'. Any undefined
1452 reference to `__real_SYMBOL' will be resolved to SYMBOL.
1454 This can be used to provide a wrapper for a system function. The
1455 wrapper function should be called `__wrap_SYMBOL'. If it wishes
1456 to call the system function, it should call `__real_SYMBOL'.
1458 Here is a trivial example:
1461 __wrap_malloc (size_t c)
1463 printf ("malloc called with %zu\n", c);
1464 return __real_malloc (c);
1467 If you link other code with this file using `--wrap malloc', then
1468 all calls to `malloc' will call the function `__wrap_malloc'
1469 instead. The call to `__real_malloc' in `__wrap_malloc' will call
1470 the real `malloc' function.
1472 You may wish to provide a `__real_malloc' function as well, so that
1473 links without the `--wrap' option will succeed. If you do this,
1474 you should not put the definition of `__real_malloc' in the same
1475 file as `__wrap_malloc'; if you do, the assembler may resolve the
1476 call before the linker has a chance to wrap it to `malloc'.
1479 Request creation of `.eh_frame_hdr' section and ELF
1480 `PT_GNU_EH_FRAME' segment header.
1482 `--enable-new-dtags'
1483 `--disable-new-dtags'
1484 This linker can create the new dynamic tags in ELF. But the older
1485 ELF systems may not understand them. If you specify
1486 `--enable-new-dtags', the dynamic tags will be created as needed.
1487 If you specify `--disable-new-dtags', no new dynamic tags will be
1488 created. By default, the new dynamic tags are not created. Note
1489 that those options are only available for ELF systems.
1491 `--hash-size=NUMBER'
1492 Set the default size of the linker's hash tables to a prime number
1493 close to NUMBER. Increasing this value can reduce the length of
1494 time it takes the linker to perform its tasks, at the expense of
1495 increasing the linker's memory requirements. Similarly reducing
1496 this value can reduce the memory requirements at the expense of
1499 `--hash-style=STYLE'
1500 Set the type of linker's hash table(s). STYLE can be either
1501 `sysv' for classic ELF `.hash' section, `gnu' for new style GNU
1502 `.gnu.hash' section or `both' for both the classic ELF `.hash' and
1503 new style GNU `.gnu.hash' hash tables. The default is `sysv'.
1505 `--reduce-memory-overheads'
1506 This option reduces memory requirements at ld runtime, at the
1507 expense of linking speed. This was introduced to select the old
1508 O(n^2) algorithm for link map file generation, rather than the new
1509 O(n) algorithm which uses about 40% more memory for symbol storage.
1511 Another effect of the switch is to set the default hash table size
1512 to 1021, which again saves memory at the cost of lengthening the
1513 linker's run time. This is not done however if the `--hash-size'
1514 switch has been used.
1516 The `--reduce-memory-overheads' switch may be also be used to
1517 enable other tradeoffs in future versions of the linker.
1521 Request creation of `.note.gnu.build-id' ELF note section. The
1522 contents of the note are unique bits identifying this linked file.
1523 STYLE can be `uuid' to use 128 random bits, `sha1' to use a
1524 160-bit SHA1 hash on the normative parts of the output contents,
1525 `md5' to use a 128-bit MD5 hash on the normative parts of the
1526 output contents, or `0xHEXSTRING' to use a chosen bit string
1527 specified as an even number of hexadecimal digits (`-' and `:'
1528 characters between digit pairs are ignored). If STYLE is omitted,
1531 The `md5' and `sha1' styles produces an identifier that is always
1532 the same in an identical output file, but will be unique among all
1533 nonidentical output files. It is not intended to be compared as a
1534 checksum for the file's contents. A linked file may be changed
1535 later by other tools, but the build ID bit string identifying the
1536 original linked file does not change.
1538 Passing `none' for STYLE disables the setting from any
1539 `--build-id' options earlier on the command line.
1541 2.1.1 Options Specific to i386 PE Targets
1542 -----------------------------------------
1544 The i386 PE linker supports the `-shared' option, which causes the
1545 output to be a dynamically linked library (DLL) instead of a normal
1546 executable. You should name the output `*.dll' when you use this
1547 option. In addition, the linker fully supports the standard `*.def'
1548 files, which may be specified on the linker command line like an object
1549 file (in fact, it should precede archives it exports symbols from, to
1550 ensure that they get linked in, just like a normal object file).
1552 In addition to the options common to all targets, the i386 PE linker
1553 support additional command line options that are specific to the i386
1554 PE target. Options that take values may be separated from their values
1555 by either a space or an equals sign.
1557 `--add-stdcall-alias'
1558 If given, symbols with a stdcall suffix (@NN) will be exported
1559 as-is and also with the suffix stripped. [This option is specific
1560 to the i386 PE targeted port of the linker]
1563 Use FILE as the name of a file in which to save the base addresses
1564 of all the relocations needed for generating DLLs with `dlltool'.
1565 [This is an i386 PE specific option]
1568 Create a DLL instead of a regular executable. You may also use
1569 `-shared' or specify a `LIBRARY' in a given `.def' file. [This
1570 option is specific to the i386 PE targeted port of the linker]
1572 `--enable-stdcall-fixup'
1573 `--disable-stdcall-fixup'
1574 If the link finds a symbol that it cannot resolve, it will attempt
1575 to do "fuzzy linking" by looking for another defined symbol that
1576 differs only in the format of the symbol name (cdecl vs stdcall)
1577 and will resolve that symbol by linking to the match. For
1578 example, the undefined symbol `_foo' might be linked to the
1579 function `_foo@12', or the undefined symbol `_bar@16' might be
1580 linked to the function `_bar'. When the linker does this, it
1581 prints a warning, since it normally should have failed to link,
1582 but sometimes import libraries generated from third-party dlls may
1583 need this feature to be usable. If you specify
1584 `--enable-stdcall-fixup', this feature is fully enabled and
1585 warnings are not printed. If you specify
1586 `--disable-stdcall-fixup', this feature is disabled and such
1587 mismatches are considered to be errors. [This option is specific
1588 to the i386 PE targeted port of the linker]
1590 `--export-all-symbols'
1591 If given, all global symbols in the objects used to build a DLL
1592 will be exported by the DLL. Note that this is the default if
1593 there otherwise wouldn't be any exported symbols. When symbols are
1594 explicitly exported via DEF files or implicitly exported via
1595 function attributes, the default is to not export anything else
1596 unless this option is given. Note that the symbols `DllMain@12',
1597 `DllEntryPoint@0', `DllMainCRTStartup@12', and `impure_ptr' will
1598 not be automatically exported. Also, symbols imported from other
1599 DLLs will not be re-exported, nor will symbols specifying the
1600 DLL's internal layout such as those beginning with `_head_' or
1601 ending with `_iname'. In addition, no symbols from `libgcc',
1602 `libstd++', `libmingw32', or `crtX.o' will be exported. Symbols
1603 whose names begin with `__rtti_' or `__builtin_' will not be
1604 exported, to help with C++ DLLs. Finally, there is an extensive
1605 list of cygwin-private symbols that are not exported (obviously,
1606 this applies on when building DLLs for cygwin targets). These
1607 cygwin-excludes are: `_cygwin_dll_entry@12',
1608 `_cygwin_crt0_common@8', `_cygwin_noncygwin_dll_entry@12',
1609 `_fmode', `_impure_ptr', `cygwin_attach_dll', `cygwin_premain0',
1610 `cygwin_premain1', `cygwin_premain2', `cygwin_premain3', and
1611 `environ'. [This option is specific to the i386 PE targeted port
1614 `--exclude-symbols SYMBOL,SYMBOL,...'
1615 Specifies a list of symbols which should not be automatically
1616 exported. The symbol names may be delimited by commas or colons.
1617 [This option is specific to the i386 PE targeted port of the
1621 Specify the file alignment. Sections in the file will always
1622 begin at file offsets which are multiples of this number. This
1623 defaults to 512. [This option is specific to the i386 PE targeted
1627 `--heap RESERVE,COMMIT'
1628 Specify the number of bytes of memory to reserve (and optionally
1629 commit) to be used as heap for this program. The default is 1Mb
1630 reserved, 4K committed. [This option is specific to the i386 PE
1631 targeted port of the linker]
1633 `--image-base VALUE'
1634 Use VALUE as the base address of your program or dll. This is the
1635 lowest memory location that will be used when your program or dll
1636 is loaded. To reduce the need to relocate and improve performance
1637 of your dlls, each should have a unique base address and not
1638 overlap any other dlls. The default is 0x400000 for executables,
1639 and 0x10000000 for dlls. [This option is specific to the i386 PE
1640 targeted port of the linker]
1643 If given, the stdcall suffixes (@NN) will be stripped from symbols
1644 before they are exported. [This option is specific to the i386 PE
1645 targeted port of the linker]
1647 `--large-address-aware'
1648 If given, the appropriate bit in the "Characteristics" field of
1649 the COFF header is set to indicate that this executable supports
1650 virtual addresses greater than 2 gigabytes. This should be used
1651 in conjunction with the /3GB or /USERVA=VALUE megabytes switch in
1652 the "[operating systems]" section of the BOOT.INI. Otherwise,
1653 this bit has no effect. [This option is specific to PE targeted
1654 ports of the linker]
1656 `--major-image-version VALUE'
1657 Sets the major number of the "image version". Defaults to 1.
1658 [This option is specific to the i386 PE targeted port of the
1661 `--major-os-version VALUE'
1662 Sets the major number of the "os version". Defaults to 4. [This
1663 option is specific to the i386 PE targeted port of the linker]
1665 `--major-subsystem-version VALUE'
1666 Sets the major number of the "subsystem version". Defaults to 4.
1667 [This option is specific to the i386 PE targeted port of the
1670 `--minor-image-version VALUE'
1671 Sets the minor number of the "image version". Defaults to 0.
1672 [This option is specific to the i386 PE targeted port of the
1675 `--minor-os-version VALUE'
1676 Sets the minor number of the "os version". Defaults to 0. [This
1677 option is specific to the i386 PE targeted port of the linker]
1679 `--minor-subsystem-version VALUE'
1680 Sets the minor number of the "subsystem version". Defaults to 0.
1681 [This option is specific to the i386 PE targeted port of the
1685 The linker will create the file FILE which will contain a DEF file
1686 corresponding to the DLL the linker is generating. This DEF file
1687 (which should be called `*.def') may be used to create an import
1688 library with `dlltool' or may be used as a reference to
1689 automatically or implicitly exported symbols. [This option is
1690 specific to the i386 PE targeted port of the linker]
1693 The linker will create the file FILE which will contain an import
1694 lib corresponding to the DLL the linker is generating. This import
1695 lib (which should be called `*.dll.a' or `*.a' may be used to link
1696 clients against the generated DLL; this behaviour makes it
1697 possible to skip a separate `dlltool' import library creation step.
1698 [This option is specific to the i386 PE targeted port of the
1701 `--enable-auto-image-base'
1702 Automatically choose the image base for DLLs, unless one is
1703 specified using the `--image-base' argument. By using a hash
1704 generated from the dllname to create unique image bases for each
1705 DLL, in-memory collisions and relocations which can delay program
1706 execution are avoided. [This option is specific to the i386 PE
1707 targeted port of the linker]
1709 `--disable-auto-image-base'
1710 Do not automatically generate a unique image base. If there is no
1711 user-specified image base (`--image-base') then use the platform
1712 default. [This option is specific to the i386 PE targeted port of
1715 `--dll-search-prefix STRING'
1716 When linking dynamically to a dll without an import library,
1717 search for `<string><basename>.dll' in preference to
1718 `lib<basename>.dll'. This behaviour allows easy distinction
1719 between DLLs built for the various "subplatforms": native, cygwin,
1720 uwin, pw, etc. For instance, cygwin DLLs typically use
1721 `--dll-search-prefix=cyg'. [This option is specific to the i386
1722 PE targeted port of the linker]
1724 `--enable-auto-import'
1725 Do sophisticated linking of `_symbol' to `__imp__symbol' for DATA
1726 imports from DLLs, and create the necessary thunking symbols when
1727 building the import libraries with those DATA exports. Note: Use
1728 of the 'auto-import' extension will cause the text section of the
1729 image file to be made writable. This does not conform to the
1730 PE-COFF format specification published by Microsoft.
1732 Note - use of the 'auto-import' extension will also cause read only
1733 data which would normally be placed into the .rdata section to be
1734 placed into the .data section instead. This is in order to work
1735 around a problem with consts that is described here:
1736 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
1738 Using 'auto-import' generally will 'just work' - but sometimes you
1739 may see this message:
1741 "variable '<var>' can't be auto-imported. Please read the
1742 documentation for ld's `--enable-auto-import' for details."
1744 This message occurs when some (sub)expression accesses an address
1745 ultimately given by the sum of two constants (Win32 import tables
1746 only allow one). Instances where this may occur include accesses
1747 to member fields of struct variables imported from a DLL, as well
1748 as using a constant index into an array variable imported from a
1749 DLL. Any multiword variable (arrays, structs, long long, etc) may
1750 trigger this error condition. However, regardless of the exact
1751 data type of the offending exported variable, ld will always
1752 detect it, issue the warning, and exit.
1754 There are several ways to address this difficulty, regardless of
1755 the data type of the exported variable:
1757 One way is to use -enable-runtime-pseudo-reloc switch. This leaves
1758 the task of adjusting references in your client code for runtime
1759 environment, so this method works only when runtime environment
1760 supports this feature.
1762 A second solution is to force one of the 'constants' to be a
1763 variable - that is, unknown and un-optimizable at compile time.
1764 For arrays, there are two possibilities: a) make the indexee (the
1765 array's address) a variable, or b) make the 'constant' index a
1768 extern type extern_array[];
1770 { volatile type *t=extern_array; t[1] }
1774 extern type extern_array[];
1776 { volatile int t=1; extern_array[t] }
1778 For structs (and most other multiword data types) the only option
1779 is to make the struct itself (or the long long, or the ...)
1782 extern struct s extern_struct;
1783 extern_struct.field -->
1784 { volatile struct s *t=&extern_struct; t->field }
1788 extern long long extern_ll;
1790 { volatile long long * local_ll=&extern_ll; *local_ll }
1792 A third method of dealing with this difficulty is to abandon
1793 'auto-import' for the offending symbol and mark it with
1794 `__declspec(dllimport)'. However, in practise that requires using
1795 compile-time #defines to indicate whether you are building a DLL,
1796 building client code that will link to the DLL, or merely
1797 building/linking to a static library. In making the choice
1798 between the various methods of resolving the 'direct address with
1799 constant offset' problem, you should consider typical real-world
1807 void main(int argc, char **argv){
1808 printf("%d\n",arr[1]);
1816 void main(int argc, char **argv){
1817 /* This workaround is for win32 and cygwin; do not "optimize" */
1818 volatile int *parr = arr;
1819 printf("%d\n",parr[1]);
1824 /* Note: auto-export is assumed (no __declspec(dllexport)) */
1825 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
1826 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
1827 #define FOO_IMPORT __declspec(dllimport)
1831 extern FOO_IMPORT int arr[];
1834 void main(int argc, char **argv){
1835 printf("%d\n",arr[1]);
1838 A fourth way to avoid this problem is to re-code your library to
1839 use a functional interface rather than a data interface for the
1840 offending variables (e.g. set_foo() and get_foo() accessor
1841 functions). [This option is specific to the i386 PE targeted port
1844 `--disable-auto-import'
1845 Do not attempt to do sophisticated linking of `_symbol' to
1846 `__imp__symbol' for DATA imports from DLLs. [This option is
1847 specific to the i386 PE targeted port of the linker]
1849 `--enable-runtime-pseudo-reloc'
1850 If your code contains expressions described in -enable-auto-import
1851 section, that is, DATA imports from DLL with non-zero offset, this
1852 switch will create a vector of 'runtime pseudo relocations' which
1853 can be used by runtime environment to adjust references to such
1854 data in your client code. [This option is specific to the i386 PE
1855 targeted port of the linker]
1857 `--disable-runtime-pseudo-reloc'
1858 Do not create pseudo relocations for non-zero offset DATA imports
1859 from DLLs. This is the default. [This option is specific to the
1860 i386 PE targeted port of the linker]
1862 `--enable-extra-pe-debug'
1863 Show additional debug info related to auto-import symbol thunking.
1864 [This option is specific to the i386 PE targeted port of the
1867 `--section-alignment'
1868 Sets the section alignment. Sections in memory will always begin
1869 at addresses which are a multiple of this number. Defaults to
1870 0x1000. [This option is specific to the i386 PE targeted port of
1874 `--stack RESERVE,COMMIT'
1875 Specify the number of bytes of memory to reserve (and optionally
1876 commit) to be used as stack for this program. The default is 2Mb
1877 reserved, 4K committed. [This option is specific to the i386 PE
1878 targeted port of the linker]
1881 `--subsystem WHICH:MAJOR'
1882 `--subsystem WHICH:MAJOR.MINOR'
1883 Specifies the subsystem under which your program will execute. The
1884 legal values for WHICH are `native', `windows', `console',
1885 `posix', and `xbox'. You may optionally set the subsystem version
1886 also. Numeric values are also accepted for WHICH. [This option
1887 is specific to the i386 PE targeted port of the linker]
1890 2.1.2 Options specific to Motorola 68HC11 and 68HC12 targets
1891 ------------------------------------------------------------
1893 The 68HC11 and 68HC12 linkers support specific options to control the
1894 memory bank switching mapping and trampoline code generation.
1897 This option disables the generation of trampoline. By default a
1898 trampoline is generated for each far function which is called
1899 using a `jsr' instruction (this happens when a pointer to a far
1902 `--bank-window NAME'
1903 This option indicates to the linker the name of the memory region
1904 in the `MEMORY' specification that describes the memory bank
1905 window. The definition of such region is then used by the linker
1906 to compute paging and addresses within the memory window.
1909 2.1.3 Options specific to Motorola 68K target
1910 ---------------------------------------------
1912 The following options are supported to control handling of GOT
1913 generation when linking for 68K targets.
1916 This option tells the linker which GOT generation scheme to use.
1917 TYPE should be one of `single', `negative', `multigot' or
1918 `target'. For more information refer to the Info entry for `ld'.
1922 File: ld.info, Node: Environment, Prev: Options, Up: Invocation
1924 2.2 Environment Variables
1925 =========================
1927 You can change the behaviour of `ld' with the environment variables
1928 `GNUTARGET', `LDEMULATION' and `COLLECT_NO_DEMANGLE'.
1930 `GNUTARGET' determines the input-file object format if you don't use
1931 `-b' (or its synonym `--format'). Its value should be one of the BFD
1932 names for an input format (*note BFD::). If there is no `GNUTARGET' in
1933 the environment, `ld' uses the natural format of the target. If
1934 `GNUTARGET' is set to `default' then BFD attempts to discover the input
1935 format by examining binary input files; this method often succeeds, but
1936 there are potential ambiguities, since there is no method of ensuring
1937 that the magic number used to specify object-file formats is unique.
1938 However, the configuration procedure for BFD on each system places the
1939 conventional format for that system first in the search-list, so
1940 ambiguities are resolved in favor of convention.
1942 `LDEMULATION' determines the default emulation if you don't use the
1943 `-m' option. The emulation can affect various aspects of linker
1944 behaviour, particularly the default linker script. You can list the
1945 available emulations with the `--verbose' or `-V' options. If the `-m'
1946 option is not used, and the `LDEMULATION' environment variable is not
1947 defined, the default emulation depends upon how the linker was
1950 Normally, the linker will default to demangling symbols. However, if
1951 `COLLECT_NO_DEMANGLE' is set in the environment, then it will default
1952 to not demangling symbols. This environment variable is used in a
1953 similar fashion by the `gcc' linker wrapper program. The default may
1954 be overridden by the `--demangle' and `--no-demangle' options.
1957 File: ld.info, Node: Scripts, Next: Machine Dependent, Prev: Invocation, Up: Top
1962 Every link is controlled by a "linker script". This script is written
1963 in the linker command language.
1965 The main purpose of the linker script is to describe how the
1966 sections in the input files should be mapped into the output file, and
1967 to control the memory layout of the output file. Most linker scripts
1968 do nothing more than this. However, when necessary, the linker script
1969 can also direct the linker to perform many other operations, using the
1970 commands described below.
1972 The linker always uses a linker script. If you do not supply one
1973 yourself, the linker will use a default script that is compiled into the
1974 linker executable. You can use the `--verbose' command line option to
1975 display the default linker script. Certain command line options, such
1976 as `-r' or `-N', will affect the default linker script.
1978 You may supply your own linker script by using the `-T' command line
1979 option. When you do this, your linker script will replace the default
1982 You may also use linker scripts implicitly by naming them as input
1983 files to the linker, as though they were files to be linked. *Note
1984 Implicit Linker Scripts::.
1988 * Basic Script Concepts:: Basic Linker Script Concepts
1989 * Script Format:: Linker Script Format
1990 * Simple Example:: Simple Linker Script Example
1991 * Simple Commands:: Simple Linker Script Commands
1992 * Assignments:: Assigning Values to Symbols
1993 * SECTIONS:: SECTIONS Command
1994 * MEMORY:: MEMORY Command
1995 * PHDRS:: PHDRS Command
1996 * VERSION:: VERSION Command
1997 * Expressions:: Expressions in Linker Scripts
1998 * Implicit Linker Scripts:: Implicit Linker Scripts
2001 File: ld.info, Node: Basic Script Concepts, Next: Script Format, Up: Scripts
2003 3.1 Basic Linker Script Concepts
2004 ================================
2006 We need to define some basic concepts and vocabulary in order to
2007 describe the linker script language.
2009 The linker combines input files into a single output file. The
2010 output file and each input file are in a special data format known as an
2011 "object file format". Each file is called an "object file". The
2012 output file is often called an "executable", but for our purposes we
2013 will also call it an object file. Each object file has, among other
2014 things, a list of "sections". We sometimes refer to a section in an
2015 input file as an "input section"; similarly, a section in the output
2016 file is an "output section".
2018 Each section in an object file has a name and a size. Most sections
2019 also have an associated block of data, known as the "section contents".
2020 A section may be marked as "loadable", which mean that the contents
2021 should be loaded into memory when the output file is run. A section
2022 with no contents may be "allocatable", which means that an area in
2023 memory should be set aside, but nothing in particular should be loaded
2024 there (in some cases this memory must be zeroed out). A section which
2025 is neither loadable nor allocatable typically contains some sort of
2026 debugging information.
2028 Every loadable or allocatable output section has two addresses. The
2029 first is the "VMA", or virtual memory address. This is the address the
2030 section will have when the output file is run. The second is the
2031 "LMA", or load memory address. This is the address at which the
2032 section will be loaded. In most cases the two addresses will be the
2033 same. An example of when they might be different is when a data section
2034 is loaded into ROM, and then copied into RAM when the program starts up
2035 (this technique is often used to initialize global variables in a ROM
2036 based system). In this case the ROM address would be the LMA, and the
2037 RAM address would be the VMA.
2039 You can see the sections in an object file by using the `objdump'
2040 program with the `-h' option.
2042 Every object file also has a list of "symbols", known as the "symbol
2043 table". A symbol may be defined or undefined. Each symbol has a name,
2044 and each defined symbol has an address, among other information. If
2045 you compile a C or C++ program into an object file, you will get a
2046 defined symbol for every defined function and global or static
2047 variable. Every undefined function or global variable which is
2048 referenced in the input file will become an undefined symbol.
2050 You can see the symbols in an object file by using the `nm' program,
2051 or by using the `objdump' program with the `-t' option.
2054 File: ld.info, Node: Script Format, Next: Simple Example, Prev: Basic Script Concepts, Up: Scripts
2056 3.2 Linker Script Format
2057 ========================
2059 Linker scripts are text files.
2061 You write a linker script as a series of commands. Each command is
2062 either a keyword, possibly followed by arguments, or an assignment to a
2063 symbol. You may separate commands using semicolons. Whitespace is
2066 Strings such as file or format names can normally be entered
2067 directly. If the file name contains a character such as a comma which
2068 would otherwise serve to separate file names, you may put the file name
2069 in double quotes. There is no way to use a double quote character in a
2072 You may include comments in linker scripts just as in C, delimited by
2073 `/*' and `*/'. As in C, comments are syntactically equivalent to
2077 File: ld.info, Node: Simple Example, Next: Simple Commands, Prev: Script Format, Up: Scripts
2079 3.3 Simple Linker Script Example
2080 ================================
2082 Many linker scripts are fairly simple.
2084 The simplest possible linker script has just one command:
2085 `SECTIONS'. You use the `SECTIONS' command to describe the memory
2086 layout of the output file.
2088 The `SECTIONS' command is a powerful command. Here we will describe
2089 a simple use of it. Let's assume your program consists only of code,
2090 initialized data, and uninitialized data. These will be in the
2091 `.text', `.data', and `.bss' sections, respectively. Let's assume
2092 further that these are the only sections which appear in your input
2095 For this example, let's say that the code should be loaded at address
2096 0x10000, and that the data should start at address 0x8000000. Here is a
2097 linker script which will do that:
2101 .text : { *(.text) }
2103 .data : { *(.data) }
2107 You write the `SECTIONS' command as the keyword `SECTIONS', followed
2108 by a series of symbol assignments and output section descriptions
2109 enclosed in curly braces.
2111 The first line inside the `SECTIONS' command of the above example
2112 sets the value of the special symbol `.', which is the location
2113 counter. If you do not specify the address of an output section in some
2114 other way (other ways are described later), the address is set from the
2115 current value of the location counter. The location counter is then
2116 incremented by the size of the output section. At the start of the
2117 `SECTIONS' command, the location counter has the value `0'.
2119 The second line defines an output section, `.text'. The colon is
2120 required syntax which may be ignored for now. Within the curly braces
2121 after the output section name, you list the names of the input sections
2122 which should be placed into this output section. The `*' is a wildcard
2123 which matches any file name. The expression `*(.text)' means all
2124 `.text' input sections in all input files.
2126 Since the location counter is `0x10000' when the output section
2127 `.text' is defined, the linker will set the address of the `.text'
2128 section in the output file to be `0x10000'.
2130 The remaining lines define the `.data' and `.bss' sections in the
2131 output file. The linker will place the `.data' output section at
2132 address `0x8000000'. After the linker places the `.data' output
2133 section, the value of the location counter will be `0x8000000' plus the
2134 size of the `.data' output section. The effect is that the linker will
2135 place the `.bss' output section immediately after the `.data' output
2138 The linker will ensure that each output section has the required
2139 alignment, by increasing the location counter if necessary. In this
2140 example, the specified addresses for the `.text' and `.data' sections
2141 will probably satisfy any alignment constraints, but the linker may
2142 have to create a small gap between the `.data' and `.bss' sections.
2144 That's it! That's a simple and complete linker script.
2147 File: ld.info, Node: Simple Commands, Next: Assignments, Prev: Simple Example, Up: Scripts
2149 3.4 Simple Linker Script Commands
2150 =================================
2152 In this section we describe the simple linker script commands.
2156 * Entry Point:: Setting the entry point
2157 * File Commands:: Commands dealing with files
2159 * Format Commands:: Commands dealing with object file formats
2161 * Miscellaneous Commands:: Other linker script commands
2164 File: ld.info, Node: Entry Point, Next: File Commands, Up: Simple Commands
2166 3.4.1 Setting the Entry Point
2167 -----------------------------
2169 The first instruction to execute in a program is called the "entry
2170 point". You can use the `ENTRY' linker script command to set the entry
2171 point. The argument is a symbol name:
2174 There are several ways to set the entry point. The linker will set
2175 the entry point by trying each of the following methods in order, and
2176 stopping when one of them succeeds:
2177 * the `-e' ENTRY command-line option;
2179 * the `ENTRY(SYMBOL)' command in a linker script;
2181 * the value of the symbol `start', if defined;
2183 * the address of the first byte of the `.text' section, if present;
2188 File: ld.info, Node: File Commands, Next: Format Commands, Prev: Entry Point, Up: Simple Commands
2190 3.4.2 Commands Dealing with Files
2191 ---------------------------------
2193 Several linker script commands deal with files.
2196 Include the linker script FILENAME at this point. The file will
2197 be searched for in the current directory, and in any directory
2198 specified with the `-L' option. You can nest calls to `INCLUDE'
2199 up to 10 levels deep.
2201 You can place `INCLUDE' directives at the top level, in `MEMORY' or
2202 `SECTIONS' commands, or in output section descriptions.
2204 `INPUT(FILE, FILE, ...)'
2205 `INPUT(FILE FILE ...)'
2206 The `INPUT' command directs the linker to include the named files
2207 in the link, as though they were named on the command line.
2209 For example, if you always want to include `subr.o' any time you do
2210 a link, but you can't be bothered to put it on every link command
2211 line, then you can put `INPUT (subr.o)' in your linker script.
2213 In fact, if you like, you can list all of your input files in the
2214 linker script, and then invoke the linker with nothing but a `-T'
2217 In case a "sysroot prefix" is configured, and the filename starts
2218 with the `/' character, and the script being processed was located
2219 inside the "sysroot prefix", the filename will be looked for in
2220 the "sysroot prefix". Otherwise, the linker will try to open the
2221 file in the current directory. If it is not found, the linker
2222 will search through the archive library search path. See the
2223 description of `-L' in *note Command Line Options: Options.
2225 If you use `INPUT (-lFILE)', `ld' will transform the name to
2226 `libFILE.a', as with the command line argument `-l'.
2228 When you use the `INPUT' command in an implicit linker script, the
2229 files will be included in the link at the point at which the linker
2230 script file is included. This can affect archive searching.
2232 `GROUP(FILE, FILE, ...)'
2233 `GROUP(FILE FILE ...)'
2234 The `GROUP' command is like `INPUT', except that the named files
2235 should all be archives, and they are searched repeatedly until no
2236 new undefined references are created. See the description of `-('
2237 in *note Command Line Options: Options.
2239 `AS_NEEDED(FILE, FILE, ...)'
2240 `AS_NEEDED(FILE FILE ...)'
2241 This construct can appear only inside of the `INPUT' or `GROUP'
2242 commands, among other filenames. The files listed will be handled
2243 as if they appear directly in the `INPUT' or `GROUP' commands,
2244 with the exception of ELF shared libraries, that will be added only
2245 when they are actually needed. This construct essentially enables
2246 `--as-needed' option for all the files listed inside of it and
2247 restores previous `--as-needed' resp. `--no-as-needed' setting
2251 The `OUTPUT' command names the output file. Using
2252 `OUTPUT(FILENAME)' in the linker script is exactly like using `-o
2253 FILENAME' on the command line (*note Command Line Options:
2254 Options.). If both are used, the command line option takes
2257 You can use the `OUTPUT' command to define a default name for the
2258 output file other than the usual default of `a.out'.
2261 The `SEARCH_DIR' command adds PATH to the list of paths where `ld'
2262 looks for archive libraries. Using `SEARCH_DIR(PATH)' is exactly
2263 like using `-L PATH' on the command line (*note Command Line
2264 Options: Options.). If both are used, then the linker will search
2265 both paths. Paths specified using the command line option are
2269 The `STARTUP' command is just like the `INPUT' command, except
2270 that FILENAME will become the first input file to be linked, as
2271 though it were specified first on the command line. This may be
2272 useful when using a system in which the entry point is always the
2273 start of the first file.
2276 File: ld.info, Node: Format Commands, Next: Miscellaneous Commands, Prev: File Commands, Up: Simple Commands
2278 3.4.3 Commands Dealing with Object File Formats
2279 -----------------------------------------------
2281 A couple of linker script commands deal with object file formats.
2283 `OUTPUT_FORMAT(BFDNAME)'
2284 `OUTPUT_FORMAT(DEFAULT, BIG, LITTLE)'
2285 The `OUTPUT_FORMAT' command names the BFD format to use for the
2286 output file (*note BFD::). Using `OUTPUT_FORMAT(BFDNAME)' is
2287 exactly like using `--oformat BFDNAME' on the command line (*note
2288 Command Line Options: Options.). If both are used, the command
2289 line option takes precedence.
2291 You can use `OUTPUT_FORMAT' with three arguments to use different
2292 formats based on the `-EB' and `-EL' command line options. This
2293 permits the linker script to set the output format based on the
2296 If neither `-EB' nor `-EL' are used, then the output format will
2297 be the first argument, DEFAULT. If `-EB' is used, the output
2298 format will be the second argument, BIG. If `-EL' is used, the
2299 output format will be the third argument, LITTLE.
2301 For example, the default linker script for the MIPS ELF target
2303 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2304 This says that the default format for the output file is
2305 `elf32-bigmips', but if the user uses the `-EL' command line
2306 option, the output file will be created in the `elf32-littlemips'
2310 The `TARGET' command names the BFD format to use when reading input
2311 files. It affects subsequent `INPUT' and `GROUP' commands. This
2312 command is like using `-b BFDNAME' on the command line (*note
2313 Command Line Options: Options.). If the `TARGET' command is used
2314 but `OUTPUT_FORMAT' is not, then the last `TARGET' command is also
2315 used to set the format for the output file. *Note BFD::.
2318 File: ld.info, Node: Miscellaneous Commands, Prev: Format Commands, Up: Simple Commands
2320 3.4.4 Other Linker Script Commands
2321 ----------------------------------
2323 There are a few other linker scripts commands.
2325 `ASSERT(EXP, MESSAGE)'
2326 Ensure that EXP is non-zero. If it is zero, then exit the linker
2327 with an error code, and print MESSAGE.
2329 `EXTERN(SYMBOL SYMBOL ...)'
2330 Force SYMBOL to be entered in the output file as an undefined
2331 symbol. Doing this may, for example, trigger linking of additional
2332 modules from standard libraries. You may list several SYMBOLs for
2333 each `EXTERN', and you may use `EXTERN' multiple times. This
2334 command has the same effect as the `-u' command-line option.
2336 `FORCE_COMMON_ALLOCATION'
2337 This command has the same effect as the `-d' command-line option:
2338 to make `ld' assign space to common symbols even if a relocatable
2339 output file is specified (`-r').
2341 `INHIBIT_COMMON_ALLOCATION'
2342 This command has the same effect as the `--no-define-common'
2343 command-line option: to make `ld' omit the assignment of addresses
2344 to common symbols even for a non-relocatable output file.
2346 `INSERT [ AFTER | BEFORE ] OUTPUT_SECTION'
2347 This command is typically used in a script specified by `-T' to
2348 augment the default `SECTIONS' with, for example, overlays. It
2349 inserts all prior linker script statements after (or before)
2350 OUTPUT_SECTION, and also causes `-T' to not override the default
2351 linker script. The exact insertion point is as for orphan
2352 sections. *Note Location Counter::. The insertion happens after
2353 the linker has mapped input sections to output sections. Prior to
2354 the insertion, since `-T' scripts are parsed before the default
2355 linker script, statements in the `-T' script occur before the
2356 default linker script statements in the internal linker
2357 representation of the script. In particular, input section
2358 assignments will be made to `-T' output sections before those in
2359 the default script. Here is an example of how a `-T' script using
2360 `INSERT' might look:
2366 .ov1 { ov1*(.text) }
2367 .ov2 { ov2*(.text) }
2372 `NOCROSSREFS(SECTION SECTION ...)'
2373 This command may be used to tell `ld' to issue an error about any
2374 references among certain output sections.
2376 In certain types of programs, particularly on embedded systems when
2377 using overlays, when one section is loaded into memory, another
2378 section will not be. Any direct references between the two
2379 sections would be errors. For example, it would be an error if
2380 code in one section called a function defined in the other section.
2382 The `NOCROSSREFS' command takes a list of output section names. If
2383 `ld' detects any cross references between the sections, it reports
2384 an error and returns a non-zero exit status. Note that the
2385 `NOCROSSREFS' command uses output section names, not input section
2388 `OUTPUT_ARCH(BFDARCH)'
2389 Specify a particular output machine architecture. The argument is
2390 one of the names used by the BFD library (*note BFD::). You can
2391 see the architecture of an object file by using the `objdump'
2392 program with the `-f' option.
2395 File: ld.info, Node: Assignments, Next: SECTIONS, Prev: Simple Commands, Up: Scripts
2397 3.5 Assigning Values to Symbols
2398 ===============================
2400 You may assign a value to a symbol in a linker script. This will define
2401 the symbol and place it into the symbol table with a global scope.
2405 * Simple Assignments:: Simple Assignments
2407 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2408 * Source Code Reference:: How to use a linker script defined symbol in source code
2411 File: ld.info, Node: Simple Assignments, Next: PROVIDE, Up: Assignments
2413 3.5.1 Simple Assignments
2414 ------------------------
2416 You may assign to a symbol using any of the C assignment operators:
2418 `SYMBOL = EXPRESSION ;'
2419 `SYMBOL += EXPRESSION ;'
2420 `SYMBOL -= EXPRESSION ;'
2421 `SYMBOL *= EXPRESSION ;'
2422 `SYMBOL /= EXPRESSION ;'
2423 `SYMBOL <<= EXPRESSION ;'
2424 `SYMBOL >>= EXPRESSION ;'
2425 `SYMBOL &= EXPRESSION ;'
2426 `SYMBOL |= EXPRESSION ;'
2428 The first case will define SYMBOL to the value of EXPRESSION. In
2429 the other cases, SYMBOL must already be defined, and the value will be
2430 adjusted accordingly.
2432 The special symbol name `.' indicates the location counter. You may
2433 only use this within a `SECTIONS' command. *Note Location Counter::.
2435 The semicolon after EXPRESSION is required.
2437 Expressions are defined below; see *note Expressions::.
2439 You may write symbol assignments as commands in their own right, or
2440 as statements within a `SECTIONS' command, or as part of an output
2441 section description in a `SECTIONS' command.
2443 The section of the symbol will be set from the section of the
2444 expression; for more information, see *note Expression Section::.
2446 Here is an example showing the three different places that symbol
2447 assignments may be used:
2457 _bdata = (. + 3) & ~ 3;
2458 .data : { *(.data) }
2460 In this example, the symbol `floating_point' will be defined as
2461 zero. The symbol `_etext' will be defined as the address following the
2462 last `.text' input section. The symbol `_bdata' will be defined as the
2463 address following the `.text' output section aligned upward to a 4 byte
2467 File: ld.info, Node: PROVIDE, Next: PROVIDE_HIDDEN, Prev: Simple Assignments, Up: Assignments
2472 In some cases, it is desirable for a linker script to define a symbol
2473 only if it is referenced and is not defined by any object included in
2474 the link. For example, traditional linkers defined the symbol `etext'.
2475 However, ANSI C requires that the user be able to use `etext' as a
2476 function name without encountering an error. The `PROVIDE' keyword may
2477 be used to define a symbol, such as `etext', only if it is referenced
2478 but not defined. The syntax is `PROVIDE(SYMBOL = EXPRESSION)'.
2480 Here is an example of using `PROVIDE' to define `etext':
2491 In this example, if the program defines `_etext' (with a leading
2492 underscore), the linker will give a multiple definition error. If, on
2493 the other hand, the program defines `etext' (with no leading
2494 underscore), the linker will silently use the definition in the program.
2495 If the program references `etext' but does not define it, the linker
2496 will use the definition in the linker script.
2499 File: ld.info, Node: PROVIDE_HIDDEN, Next: Source Code Reference, Prev: PROVIDE, Up: Assignments
2501 3.5.3 PROVIDE_HIDDEN
2502 --------------------
2504 Similar to `PROVIDE'. For ELF targeted ports, the symbol will be
2505 hidden and won't be exported.
2508 File: ld.info, Node: Source Code Reference, Prev: PROVIDE_HIDDEN, Up: Assignments
2510 3.5.4 Source Code Reference
2511 ---------------------------
2513 Accessing a linker script defined variable from source code is not
2514 intuitive. In particular a linker script symbol is not equivalent to a
2515 variable declaration in a high level language, it is instead a symbol
2516 that does not have a value.
2518 Before going further, it is important to note that compilers often
2519 transform names in the source code into different names when they are
2520 stored in the symbol table. For example, Fortran compilers commonly
2521 prepend or append an underscore, and C++ performs extensive `name
2522 mangling'. Therefore there might be a discrepancy between the name of
2523 a variable as it is used in source code and the name of the same
2524 variable as it is defined in a linker script. For example in C a
2525 linker script variable might be referred to as:
2529 But in the linker script it might be defined as:
2533 In the remaining examples however it is assumed that no name
2534 transformation has taken place.
2536 When a symbol is declared in a high level language such as C, two
2537 things happen. The first is that the compiler reserves enough space in
2538 the program's memory to hold the _value_ of the symbol. The second is
2539 that the compiler creates an entry in the program's symbol table which
2540 holds the symbol's _address_. ie the symbol table contains the address
2541 of the block of memory holding the symbol's value. So for example the
2542 following C declaration, at file scope:
2546 creates a entry called `foo' in the symbol table. This entry holds
2547 the address of an `int' sized block of memory where the number 1000 is
2550 When a program references a symbol the compiler generates code that
2551 first accesses the symbol table to find the address of the symbol's
2552 memory block and then code to read the value from that memory block.
2557 looks up the symbol `foo' in the symbol table, gets the address
2558 associated with this symbol and then writes the value 1 into that
2563 looks up the symbol `foo' in the symbol table, gets it address and
2564 then copies this address into the block of memory associated with the
2567 Linker scripts symbol declarations, by contrast, create an entry in
2568 the symbol table but do not assign any memory to them. Thus they are
2569 an address without a value. So for example the linker script
2574 creates an entry in the symbol table called `foo' which holds the
2575 address of memory location 1000, but nothing special is stored at
2576 address 1000. This means that you cannot access the _value_ of a
2577 linker script defined symbol - it has no value - all you can do is
2578 access the _address_ of a linker script defined symbol.
2580 Hence when you are using a linker script defined symbol in source
2581 code you should always take the address of the symbol, and never
2582 attempt to use its value. For example suppose you want to copy the
2583 contents of a section of memory called .ROM into a section called
2584 .FLASH and the linker script contains these declarations:
2586 start_of_ROM = .ROM;
2587 end_of_ROM = .ROM + sizeof (.ROM) - 1;
2588 start_of_FLASH = .FLASH;
2590 Then the C source code to perform the copy would be:
2592 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
2594 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
2596 Note the use of the `&' operators. These are correct.
2599 File: ld.info, Node: SECTIONS, Next: MEMORY, Prev: Assignments, Up: Scripts
2601 3.6 SECTIONS Command
2602 ====================
2604 The `SECTIONS' command tells the linker how to map input sections into
2605 output sections, and how to place the output sections in memory.
2607 The format of the `SECTIONS' command is:
2615 Each SECTIONS-COMMAND may of be one of the following:
2617 * an `ENTRY' command (*note Entry command: Entry Point.)
2619 * a symbol assignment (*note Assignments::)
2621 * an output section description
2623 * an overlay description
2625 The `ENTRY' command and symbol assignments are permitted inside the
2626 `SECTIONS' command for convenience in using the location counter in
2627 those commands. This can also make the linker script easier to
2628 understand because you can use those commands at meaningful points in
2629 the layout of the output file.
2631 Output section descriptions and overlay descriptions are described
2634 If you do not use a `SECTIONS' command in your linker script, the
2635 linker will place each input section into an identically named output
2636 section in the order that the sections are first encountered in the
2637 input files. If all input sections are present in the first file, for
2638 example, the order of sections in the output file will match the order
2639 in the first input file. The first section will be at address zero.
2643 * Output Section Description:: Output section description
2644 * Output Section Name:: Output section name
2645 * Output Section Address:: Output section address
2646 * Input Section:: Input section description
2647 * Output Section Data:: Output section data
2648 * Output Section Keywords:: Output section keywords
2649 * Output Section Discarding:: Output section discarding
2650 * Output Section Attributes:: Output section attributes
2651 * Overlay Description:: Overlay description
2654 File: ld.info, Node: Output Section Description, Next: Output Section Name, Up: SECTIONS
2656 3.6.1 Output Section Description
2657 --------------------------------
2659 The full description of an output section looks like this:
2660 SECTION [ADDRESS] [(TYPE)] :
2661 [AT(LMA)] [ALIGN(SECTION_ALIGN)] [SUBALIGN(SUBSECTION_ALIGN)]
2663 OUTPUT-SECTION-COMMAND
2664 OUTPUT-SECTION-COMMAND
2666 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
2668 Most output sections do not use most of the optional section
2671 The whitespace around SECTION is required, so that the section name
2672 is unambiguous. The colon and the curly braces are also required. The
2673 line breaks and other white space are optional.
2675 Each OUTPUT-SECTION-COMMAND may be one of the following:
2677 * a symbol assignment (*note Assignments::)
2679 * an input section description (*note Input Section::)
2681 * data values to include directly (*note Output Section Data::)
2683 * a special output section keyword (*note Output Section Keywords::)
2686 File: ld.info, Node: Output Section Name, Next: Output Section Address, Prev: Output Section Description, Up: SECTIONS
2688 3.6.2 Output Section Name
2689 -------------------------
2691 The name of the output section is SECTION. SECTION must meet the
2692 constraints of your output format. In formats which only support a
2693 limited number of sections, such as `a.out', the name must be one of
2694 the names supported by the format (`a.out', for example, allows only
2695 `.text', `.data' or `.bss'). If the output format supports any number
2696 of sections, but with numbers and not names (as is the case for Oasys),
2697 the name should be supplied as a quoted numeric string. A section name
2698 may consist of any sequence of characters, but a name which contains
2699 any unusual characters such as commas must be quoted.
2701 The output section name `/DISCARD/' is special; *note Output Section
2705 File: ld.info, Node: Output Section Address, Next: Input Section, Prev: Output Section Name, Up: SECTIONS
2707 3.6.3 Output Section Address
2708 ----------------------------
2710 The ADDRESS is an expression for the VMA (the virtual memory address)
2711 of the output section. If you do not provide ADDRESS, the linker will
2712 set it based on REGION if present, or otherwise based on the current
2713 value of the location counter.
2715 If you provide ADDRESS, the address of the output section will be
2716 set to precisely that. If you provide neither ADDRESS nor REGION, then
2717 the address of the output section will be set to the current value of
2718 the location counter aligned to the alignment requirements of the
2719 output section. The alignment requirement of the output section is the
2720 strictest alignment of any input section contained within the output
2724 .text . : { *(.text) }
2726 .text : { *(.text) }
2727 are subtly different. The first will set the address of the `.text'
2728 output section to the current value of the location counter. The
2729 second will set it to the current value of the location counter aligned
2730 to the strictest alignment of a `.text' input section.
2732 The ADDRESS may be an arbitrary expression; *note Expressions::.
2733 For example, if you want to align the section on a 0x10 byte boundary,
2734 so that the lowest four bits of the section address are zero, you could
2735 do something like this:
2736 .text ALIGN(0x10) : { *(.text) }
2737 This works because `ALIGN' returns the current location counter
2738 aligned upward to the specified value.
2740 Specifying ADDRESS for a section will change the value of the
2744 File: ld.info, Node: Input Section, Next: Output Section Data, Prev: Output Section Address, Up: SECTIONS
2746 3.6.4 Input Section Description
2747 -------------------------------
2749 The most common output section command is an input section description.
2751 The input section description is the most basic linker script
2752 operation. You use output sections to tell the linker how to lay out
2753 your program in memory. You use input section descriptions to tell the
2754 linker how to map the input files into your memory layout.
2758 * Input Section Basics:: Input section basics
2759 * Input Section Wildcards:: Input section wildcard patterns
2760 * Input Section Common:: Input section for common symbols
2761 * Input Section Keep:: Input section and garbage collection
2762 * Input Section Example:: Input section example
2765 File: ld.info, Node: Input Section Basics, Next: Input Section Wildcards, Up: Input Section
2767 3.6.4.1 Input Section Basics
2768 ............................
2770 An input section description consists of a file name optionally followed
2771 by a list of section names in parentheses.
2773 The file name and the section name may be wildcard patterns, which we
2774 describe further below (*note Input Section Wildcards::).
2776 The most common input section description is to include all input
2777 sections with a particular name in the output section. For example, to
2778 include all input `.text' sections, you would write:
2780 Here the `*' is a wildcard which matches any file name. To exclude
2781 a list of files from matching the file name wildcard, EXCLUDE_FILE may
2782 be used to match all files except the ones specified in the
2783 EXCLUDE_FILE list. For example:
2784 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
2785 will cause all .ctors sections from all files except `crtend.o' and
2786 `otherfile.o' to be included.
2788 There are two ways to include more than one section:
2791 The difference between these is the order in which the `.text' and
2792 `.rdata' input sections will appear in the output section. In the
2793 first example, they will be intermingled, appearing in the same order as
2794 they are found in the linker input. In the second example, all `.text'
2795 input sections will appear first, followed by all `.rdata' input
2798 You can specify a file name to include sections from a particular
2799 file. You would do this if one or more of your files contain special
2800 data that needs to be at a particular location in memory. For example:
2803 You can also specify files within archives by writing a pattern
2804 matching the archive, a colon, then the pattern matching the file, with
2805 no whitespace around the colon.
2808 matches file within archive
2811 matches the whole archive
2814 matches file but not one in an archive
2816 Either one or both of `archive' and `file' can contain shell
2817 wildcards. On DOS based file systems, the linker will assume that a
2818 single letter followed by a colon is a drive specifier, so `c:myfile.o'
2819 is a simple file specification, not `myfile.o' within an archive called
2820 `c'. `archive:file' filespecs may also be used within an
2821 `EXCLUDE_FILE' list, but may not appear in other linker script
2822 contexts. For instance, you cannot extract a file from an archive by
2823 using `archive:file' in an `INPUT' command.
2825 If you use a file name without a list of sections, then all sections
2826 in the input file will be included in the output section. This is not
2827 commonly done, but it may by useful on occasion. For example:
2830 When you use a file name which is not an `archive:file' specifier
2831 and does not contain any wild card characters, the linker will first
2832 see if you also specified the file name on the linker command line or
2833 in an `INPUT' command. If you did not, the linker will attempt to open
2834 the file as an input file, as though it appeared on the command line.
2835 Note that this differs from an `INPUT' command, because the linker will
2836 not search for the file in the archive search path.
2839 File: ld.info, Node: Input Section Wildcards, Next: Input Section Common, Prev: Input Section Basics, Up: Input Section
2841 3.6.4.2 Input Section Wildcard Patterns
2842 .......................................
2844 In an input section description, either the file name or the section
2845 name or both may be wildcard patterns.
2847 The file name of `*' seen in many examples is a simple wildcard
2848 pattern for the file name.
2850 The wildcard patterns are like those used by the Unix shell.
2853 matches any number of characters
2856 matches any single character
2859 matches a single instance of any of the CHARS; the `-' character
2860 may be used to specify a range of characters, as in `[a-z]' to
2861 match any lower case letter
2864 quotes the following character
2866 When a file name is matched with a wildcard, the wildcard characters
2867 will not match a `/' character (used to separate directory names on
2868 Unix). A pattern consisting of a single `*' character is an exception;
2869 it will always match any file name, whether it contains a `/' or not.
2870 In a section name, the wildcard characters will match a `/' character.
2872 File name wildcard patterns only match files which are explicitly
2873 specified on the command line or in an `INPUT' command. The linker
2874 does not search directories to expand wildcards.
2876 If a file name matches more than one wildcard pattern, or if a file
2877 name appears explicitly and is also matched by a wildcard pattern, the
2878 linker will use the first match in the linker script. For example, this
2879 sequence of input section descriptions is probably in error, because the
2880 `data.o' rule will not be used:
2881 .data : { *(.data) }
2882 .data1 : { data.o(.data) }
2884 Normally, the linker will place files and sections matched by
2885 wildcards in the order in which they are seen during the link. You can
2886 change this by using the `SORT_BY_NAME' keyword, which appears before a
2887 wildcard pattern in parentheses (e.g., `SORT_BY_NAME(.text*)'). When
2888 the `SORT_BY_NAME' keyword is used, the linker will sort the files or
2889 sections into ascending order by name before placing them in the output
2892 `SORT_BY_ALIGNMENT' is very similar to `SORT_BY_NAME'. The
2893 difference is `SORT_BY_ALIGNMENT' will sort sections into ascending
2894 order by alignment before placing them in the output file.
2896 `SORT' is an alias for `SORT_BY_NAME'.
2898 When there are nested section sorting commands in linker script,
2899 there can be at most 1 level of nesting for section sorting commands.
2901 1. `SORT_BY_NAME' (`SORT_BY_ALIGNMENT' (wildcard section pattern)).
2902 It will sort the input sections by name first, then by alignment
2903 if 2 sections have the same name.
2905 2. `SORT_BY_ALIGNMENT' (`SORT_BY_NAME' (wildcard section pattern)).
2906 It will sort the input sections by alignment first, then by name
2907 if 2 sections have the same alignment.
2909 3. `SORT_BY_NAME' (`SORT_BY_NAME' (wildcard section pattern)) is
2910 treated the same as `SORT_BY_NAME' (wildcard section pattern).
2912 4. `SORT_BY_ALIGNMENT' (`SORT_BY_ALIGNMENT' (wildcard section
2913 pattern)) is treated the same as `SORT_BY_ALIGNMENT' (wildcard
2916 5. All other nested section sorting commands are invalid.
2918 When both command line section sorting option and linker script
2919 section sorting command are used, section sorting command always takes
2920 precedence over the command line option.
2922 If the section sorting command in linker script isn't nested, the
2923 command line option will make the section sorting command to be treated
2924 as nested sorting command.
2926 1. `SORT_BY_NAME' (wildcard section pattern ) with `--sort-sections
2927 alignment' is equivalent to `SORT_BY_NAME' (`SORT_BY_ALIGNMENT'
2928 (wildcard section pattern)).
2930 2. `SORT_BY_ALIGNMENT' (wildcard section pattern) with
2931 `--sort-section name' is equivalent to `SORT_BY_ALIGNMENT'
2932 (`SORT_BY_NAME' (wildcard section pattern)).
2934 If the section sorting command in linker script is nested, the
2935 command line option will be ignored.
2937 If you ever get confused about where input sections are going, use
2938 the `-M' linker option to generate a map file. The map file shows
2939 precisely how input sections are mapped to output sections.
2941 This example shows how wildcard patterns might be used to partition
2942 files. This linker script directs the linker to place all `.text'
2943 sections in `.text' and all `.bss' sections in `.bss'. The linker will
2944 place the `.data' section from all files beginning with an upper case
2945 character in `.DATA'; for all other files, the linker will place the
2946 `.data' section in `.data'.
2948 .text : { *(.text) }
2949 .DATA : { [A-Z]*(.data) }
2950 .data : { *(.data) }
2955 File: ld.info, Node: Input Section Common, Next: Input Section Keep, Prev: Input Section Wildcards, Up: Input Section
2957 3.6.4.3 Input Section for Common Symbols
2958 ........................................
2960 A special notation is needed for common symbols, because in many object
2961 file formats common symbols do not have a particular input section. The
2962 linker treats common symbols as though they are in an input section
2965 You may use file names with the `COMMON' section just as with any
2966 other input sections. You can use this to place common symbols from a
2967 particular input file in one section while common symbols from other
2968 input files are placed in another section.
2970 In most cases, common symbols in input files will be placed in the
2971 `.bss' section in the output file. For example:
2972 .bss { *(.bss) *(COMMON) }
2974 Some object file formats have more than one type of common symbol.
2975 For example, the MIPS ELF object file format distinguishes standard
2976 common symbols and small common symbols. In this case, the linker will
2977 use a different special section name for other types of common symbols.
2978 In the case of MIPS ELF, the linker uses `COMMON' for standard common
2979 symbols and `.scommon' for small common symbols. This permits you to
2980 map the different types of common symbols into memory at different
2983 You will sometimes see `[COMMON]' in old linker scripts. This
2984 notation is now considered obsolete. It is equivalent to `*(COMMON)'.
2987 File: ld.info, Node: Input Section Keep, Next: Input Section Example, Prev: Input Section Common, Up: Input Section
2989 3.6.4.4 Input Section and Garbage Collection
2990 ............................................
2992 When link-time garbage collection is in use (`--gc-sections'), it is
2993 often useful to mark sections that should not be eliminated. This is
2994 accomplished by surrounding an input section's wildcard entry with
2995 `KEEP()', as in `KEEP(*(.init))' or `KEEP(SORT_BY_NAME(*)(.ctors))'.
2998 File: ld.info, Node: Input Section Example, Prev: Input Section Keep, Up: Input Section
3000 3.6.4.5 Input Section Example
3001 .............................
3003 The following example is a complete linker script. It tells the linker
3004 to read all of the sections from file `all.o' and place them at the
3005 start of output section `outputa' which starts at location `0x10000'.
3006 All of section `.input1' from file `foo.o' follows immediately, in the
3007 same output section. All of section `.input2' from `foo.o' goes into
3008 output section `outputb', followed by section `.input1' from `foo1.o'.
3009 All of the remaining `.input1' and `.input2' sections from any files
3010 are written to output section `outputc'.
3031 File: ld.info, Node: Output Section Data, Next: Output Section Keywords, Prev: Input Section, Up: SECTIONS
3033 3.6.5 Output Section Data
3034 -------------------------
3036 You can include explicit bytes of data in an output section by using
3037 `BYTE', `SHORT', `LONG', `QUAD', or `SQUAD' as an output section
3038 command. Each keyword is followed by an expression in parentheses
3039 providing the value to store (*note Expressions::). The value of the
3040 expression is stored at the current value of the location counter.
3042 The `BYTE', `SHORT', `LONG', and `QUAD' commands store one, two,
3043 four, and eight bytes (respectively). After storing the bytes, the
3044 location counter is incremented by the number of bytes stored.
3046 For example, this will store the byte 1 followed by the four byte
3047 value of the symbol `addr':
3051 When using a 64 bit host or target, `QUAD' and `SQUAD' are the same;
3052 they both store an 8 byte, or 64 bit, value. When both host and target
3053 are 32 bits, an expression is computed as 32 bits. In this case `QUAD'
3054 stores a 32 bit value zero extended to 64 bits, and `SQUAD' stores a 32
3055 bit value sign extended to 64 bits.
3057 If the object file format of the output file has an explicit
3058 endianness, which is the normal case, the value will be stored in that
3059 endianness. When the object file format does not have an explicit
3060 endianness, as is true of, for example, S-records, the value will be
3061 stored in the endianness of the first input object file.
3063 Note--these commands only work inside a section description and not
3064 between them, so the following will produce an error from the linker:
3065 SECTIONS { .text : { *(.text) } LONG(1) .data : { *(.data) } }
3066 whereas this will work:
3067 SECTIONS { .text : { *(.text) ; LONG(1) } .data : { *(.data) } }
3069 You may use the `FILL' command to set the fill pattern for the
3070 current section. It is followed by an expression in parentheses. Any
3071 otherwise unspecified regions of memory within the section (for example,
3072 gaps left due to the required alignment of input sections) are filled
3073 with the value of the expression, repeated as necessary. A `FILL'
3074 statement covers memory locations after the point at which it occurs in
3075 the section definition; by including more than one `FILL' statement,
3076 you can have different fill patterns in different parts of an output
3079 This example shows how to fill unspecified regions of memory with the
3083 The `FILL' command is similar to the `=FILLEXP' output section
3084 attribute, but it only affects the part of the section following the
3085 `FILL' command, rather than the entire section. If both are used, the
3086 `FILL' command takes precedence. *Note Output Section Fill::, for
3087 details on the fill expression.
3090 File: ld.info, Node: Output Section Keywords, Next: Output Section Discarding, Prev: Output Section Data, Up: SECTIONS
3092 3.6.6 Output Section Keywords
3093 -----------------------------
3095 There are a couple of keywords which can appear as output section
3098 `CREATE_OBJECT_SYMBOLS'
3099 The command tells the linker to create a symbol for each input
3100 file. The name of each symbol will be the name of the
3101 corresponding input file. The section of each symbol will be the
3102 output section in which the `CREATE_OBJECT_SYMBOLS' command
3105 This is conventional for the a.out object file format. It is not
3106 normally used for any other object file format.
3109 When linking using the a.out object file format, the linker uses an
3110 unusual set construct to support C++ global constructors and
3111 destructors. When linking object file formats which do not support
3112 arbitrary sections, such as ECOFF and XCOFF, the linker will
3113 automatically recognize C++ global constructors and destructors by
3114 name. For these object file formats, the `CONSTRUCTORS' command
3115 tells the linker to place constructor information in the output
3116 section where the `CONSTRUCTORS' command appears. The
3117 `CONSTRUCTORS' command is ignored for other object file formats.
3119 The symbol `__CTOR_LIST__' marks the start of the global
3120 constructors, and the symbol `__CTOR_END__' marks the end.
3121 Similarly, `__DTOR_LIST__' and `__DTOR_END__' mark the start and
3122 end of the global destructors. The first word in the list is the
3123 number of entries, followed by the address of each constructor or
3124 destructor, followed by a zero word. The compiler must arrange to
3125 actually run the code. For these object file formats GNU C++
3126 normally calls constructors from a subroutine `__main'; a call to
3127 `__main' is automatically inserted into the startup code for
3128 `main'. GNU C++ normally runs destructors either by using
3129 `atexit', or directly from the function `exit'.
3131 For object file formats such as `COFF' or `ELF' which support
3132 arbitrary section names, GNU C++ will normally arrange to put the
3133 addresses of global constructors and destructors into the `.ctors'
3134 and `.dtors' sections. Placing the following sequence into your
3135 linker script will build the sort of table which the GNU C++
3136 runtime code expects to see.
3139 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3144 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3149 If you are using the GNU C++ support for initialization priority,
3150 which provides some control over the order in which global
3151 constructors are run, you must sort the constructors at link time
3152 to ensure that they are executed in the correct order. When using
3153 the `CONSTRUCTORS' command, use `SORT_BY_NAME(CONSTRUCTORS)'
3154 instead. When using the `.ctors' and `.dtors' sections, use
3155 `*(SORT_BY_NAME(.ctors))' and `*(SORT_BY_NAME(.dtors))' instead of
3156 just `*(.ctors)' and `*(.dtors)'.
3158 Normally the compiler and linker will handle these issues
3159 automatically, and you will not need to concern yourself with
3160 them. However, you may need to consider this if you are using C++
3161 and writing your own linker scripts.
3165 File: ld.info, Node: Output Section Discarding, Next: Output Section Attributes, Prev: Output Section Keywords, Up: SECTIONS
3167 3.6.7 Output Section Discarding
3168 -------------------------------
3170 The linker will not create output sections with no contents. This is
3171 for convenience when referring to input sections that may or may not be
3172 present in any of the input files. For example:
3174 will only create a `.foo' section in the output file if there is a
3175 `.foo' section in at least one input file, and if the input sections
3176 are not all empty. Other link script directives that allocate space in
3177 an output section will also create the output section.
3179 The linker will ignore address assignments (*note Output Section
3180 Address::) on discarded output sections, except when the linker script
3181 defines symbols in the output section. In that case the linker will
3182 obey the address assignments, possibly advancing dot even though the
3183 section is discarded.
3185 The special output section name `/DISCARD/' may be used to discard
3186 input sections. Any input sections which are assigned to an output
3187 section named `/DISCARD/' are not included in the output file.
3190 File: ld.info, Node: Output Section Attributes, Next: Overlay Description, Prev: Output Section Discarding, Up: SECTIONS
3192 3.6.8 Output Section Attributes
3193 -------------------------------
3195 We showed above that the full description of an output section looked
3197 SECTION [ADDRESS] [(TYPE)] :
3198 [AT(LMA)] [ALIGN(SECTION_ALIGN)] [SUBALIGN(SUBSECTION_ALIGN)]
3200 OUTPUT-SECTION-COMMAND
3201 OUTPUT-SECTION-COMMAND
3203 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
3204 We've already described SECTION, ADDRESS, and OUTPUT-SECTION-COMMAND.
3205 In this section we will describe the remaining section attributes.
3209 * Output Section Type:: Output section type
3210 * Output Section LMA:: Output section LMA
3211 * Forced Output Alignment:: Forced Output Alignment
3212 * Forced Input Alignment:: Forced Input Alignment
3213 * Output Section Region:: Output section region
3214 * Output Section Phdr:: Output section phdr
3215 * Output Section Fill:: Output section fill
3218 File: ld.info, Node: Output Section Type, Next: Output Section LMA, Up: Output Section Attributes
3220 3.6.8.1 Output Section Type
3221 ...........................
3223 Each output section may have a type. The type is a keyword in
3224 parentheses. The following types are defined:
3227 The section should be marked as not loadable, so that it will not
3228 be loaded into memory when the program is run.
3234 These type names are supported for backward compatibility, and are
3235 rarely used. They all have the same effect: the section should be
3236 marked as not allocatable, so that no memory is allocated for the
3237 section when the program is run.
3239 The linker normally sets the attributes of an output section based on
3240 the input sections which map into it. You can override this by using
3241 the section type. For example, in the script sample below, the `ROM'
3242 section is addressed at memory location `0' and does not need to be
3243 loaded when the program is run. The contents of the `ROM' section will
3244 appear in the linker output file as usual.
3246 ROM 0 (NOLOAD) : { ... }
3251 File: ld.info, Node: Output Section LMA, Next: Forced Output Alignment, Prev: Output Section Type, Up: Output Section Attributes
3253 3.6.8.2 Output Section LMA
3254 ..........................
3256 Every section has a virtual address (VMA) and a load address (LMA); see
3257 *note Basic Script Concepts::. The address expression which may appear
3258 in an output section description sets the VMA (*note Output Section
3261 The expression LMA that follows the `AT' keyword specifies the load
3262 address of the section.
3264 Alternatively, with `AT>LMA_REGION' expression, you may specify a
3265 memory region for the section's load address. *Note MEMORY::. Note
3266 that if the section has not had a VMA assigned to it then the linker
3267 will use the LMA_REGION as the VMA region as well.
3269 If neither `AT' nor `AT>' is specified for an allocatable section,
3270 the linker will set the LMA such that the difference between VMA and
3271 LMA for the section is the same as the preceding output section in the
3272 same region. If there is no preceding output section or the section is
3273 not allocatable, the linker will set the LMA equal to the VMA. *Note
3274 Output Section Region::.
3276 This feature is designed to make it easy to build a ROM image. For
3277 example, the following linker script creates three output sections: one
3278 called `.text', which starts at `0x1000', one called `.mdata', which is
3279 loaded at the end of the `.text' section even though its VMA is
3280 `0x2000', and one called `.bss' to hold uninitialized data at address
3281 `0x3000'. The symbol `_data' is defined with the value `0x2000', which
3282 shows that the location counter holds the VMA value, not the LMA value.
3286 .text 0x1000 : { *(.text) _etext = . ; }
3288 AT ( ADDR (.text) + SIZEOF (.text) )
3289 { _data = . ; *(.data); _edata = . ; }
3291 { _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;}
3294 The run-time initialization code for use with a program generated
3295 with this linker script would include something like the following, to
3296 copy the initialized data from the ROM image to its runtime address.
3297 Notice how this code takes advantage of the symbols defined by the
3300 extern char _etext, _data, _edata, _bstart, _bend;
3301 char *src = &_etext;
3304 /* ROM has data at end of text; copy it. */
3305 while (dst < &_edata) {
3310 for (dst = &_bstart; dst< &_bend; dst++)
3314 File: ld.info, Node: Forced Output Alignment, Next: Forced Input Alignment, Prev: Output Section LMA, Up: Output Section Attributes
3316 3.6.8.3 Forced Output Alignment
3317 ...............................
3319 You can increase an output section's alignment by using ALIGN.
3322 File: ld.info, Node: Forced Input Alignment, Next: Output Section Region, Prev: Forced Output Alignment, Up: Output Section Attributes
3324 3.6.8.4 Forced Input Alignment
3325 ..............................
3327 You can force input section alignment within an output section by using
3328 SUBALIGN. The value specified overrides any alignment given by input
3329 sections, whether larger or smaller.
3332 File: ld.info, Node: Output Section Region, Next: Output Section Phdr, Prev: Forced Input Alignment, Up: Output Section Attributes
3334 3.6.8.5 Output Section Region
3335 .............................
3337 You can assign a section to a previously defined region of memory by
3338 using `>REGION'. *Note MEMORY::.
3340 Here is a simple example:
3341 MEMORY { rom : ORIGIN = 0x1000, LENGTH = 0x1000 }
3342 SECTIONS { ROM : { *(.text) } >rom }
3345 File: ld.info, Node: Output Section Phdr, Next: Output Section Fill, Prev: Output Section Region, Up: Output Section Attributes
3347 3.6.8.6 Output Section Phdr
3348 ...........................
3350 You can assign a section to a previously defined program segment by
3351 using `:PHDR'. *Note PHDRS::. If a section is assigned to one or more
3352 segments, then all subsequent allocated sections will be assigned to
3353 those segments as well, unless they use an explicitly `:PHDR' modifier.
3354 You can use `:NONE' to tell the linker to not put the section in any
3357 Here is a simple example:
3358 PHDRS { text PT_LOAD ; }
3359 SECTIONS { .text : { *(.text) } :text }
3362 File: ld.info, Node: Output Section Fill, Prev: Output Section Phdr, Up: Output Section Attributes
3364 3.6.8.7 Output Section Fill
3365 ...........................
3367 You can set the fill pattern for an entire section by using `=FILLEXP'.
3368 FILLEXP is an expression (*note Expressions::). Any otherwise
3369 unspecified regions of memory within the output section (for example,
3370 gaps left due to the required alignment of input sections) will be
3371 filled with the value, repeated as necessary. If the fill expression
3372 is a simple hex number, ie. a string of hex digit starting with `0x'
3373 and without a trailing `k' or `M', then an arbitrarily long sequence of
3374 hex digits can be used to specify the fill pattern; Leading zeros
3375 become part of the pattern too. For all other cases, including extra
3376 parentheses or a unary `+', the fill pattern is the four least
3377 significant bytes of the value of the expression. In all cases, the
3378 number is big-endian.
3380 You can also change the fill value with a `FILL' command in the
3381 output section commands; (*note Output Section Data::).
3383 Here is a simple example:
3384 SECTIONS { .text : { *(.text) } =0x90909090 }
3387 File: ld.info, Node: Overlay Description, Prev: Output Section Attributes, Up: SECTIONS
3389 3.6.9 Overlay Description
3390 -------------------------
3392 An overlay description provides an easy way to describe sections which
3393 are to be loaded as part of a single memory image but are to be run at
3394 the same memory address. At run time, some sort of overlay manager will
3395 copy the overlaid sections in and out of the runtime memory address as
3396 required, perhaps by simply manipulating addressing bits. This approach
3397 can be useful, for example, when a certain region of memory is faster
3400 Overlays are described using the `OVERLAY' command. The `OVERLAY'
3401 command is used within a `SECTIONS' command, like an output section
3402 description. The full syntax of the `OVERLAY' command is as follows:
3403 OVERLAY [START] : [NOCROSSREFS] [AT ( LDADDR )]
3407 OUTPUT-SECTION-COMMAND
3408 OUTPUT-SECTION-COMMAND
3410 } [:PHDR...] [=FILL]
3413 OUTPUT-SECTION-COMMAND
3414 OUTPUT-SECTION-COMMAND
3416 } [:PHDR...] [=FILL]
3418 } [>REGION] [:PHDR...] [=FILL]
3420 Everything is optional except `OVERLAY' (a keyword), and each
3421 section must have a name (SECNAME1 and SECNAME2 above). The section
3422 definitions within the `OVERLAY' construct are identical to those
3423 within the general `SECTIONS' contruct (*note SECTIONS::), except that
3424 no addresses and no memory regions may be defined for sections within
3427 The sections are all defined with the same starting address. The
3428 load addresses of the sections are arranged such that they are
3429 consecutive in memory starting at the load address used for the
3430 `OVERLAY' as a whole (as with normal section definitions, the load
3431 address is optional, and defaults to the start address; the start
3432 address is also optional, and defaults to the current value of the
3435 If the `NOCROSSREFS' keyword is used, and there any references among
3436 the sections, the linker will report an error. Since the sections all
3437 run at the same address, it normally does not make sense for one
3438 section to refer directly to another. *Note NOCROSSREFS: Miscellaneous
3441 For each section within the `OVERLAY', the linker automatically
3442 provides two symbols. The symbol `__load_start_SECNAME' is defined as
3443 the starting load address of the section. The symbol
3444 `__load_stop_SECNAME' is defined as the final load address of the
3445 section. Any characters within SECNAME which are not legal within C
3446 identifiers are removed. C (or assembler) code may use these symbols
3447 to move the overlaid sections around as necessary.
3449 At the end of the overlay, the value of the location counter is set
3450 to the start address of the overlay plus the size of the largest
3453 Here is an example. Remember that this would appear inside a
3454 `SECTIONS' construct.
3455 OVERLAY 0x1000 : AT (0x4000)
3457 .text0 { o1/*.o(.text) }
3458 .text1 { o2/*.o(.text) }
3460 This will define both `.text0' and `.text1' to start at address 0x1000.
3461 `.text0' will be loaded at address 0x4000, and `.text1' will be loaded
3462 immediately after `.text0'. The following symbols will be defined if
3463 referenced: `__load_start_text0', `__load_stop_text0',
3464 `__load_start_text1', `__load_stop_text1'.
3466 C code to copy overlay `.text1' into the overlay area might look
3469 extern char __load_start_text1, __load_stop_text1;
3470 memcpy ((char *) 0x1000, &__load_start_text1,
3471 &__load_stop_text1 - &__load_start_text1);
3473 Note that the `OVERLAY' command is just syntactic sugar, since
3474 everything it does can be done using the more basic commands. The above
3475 example could have been written identically as follows.
3477 .text0 0x1000 : AT (0x4000) { o1/*.o(.text) }
3478 PROVIDE (__load_start_text0 = LOADADDR (.text0));
3479 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
3480 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) { o2/*.o(.text) }
3481 PROVIDE (__load_start_text1 = LOADADDR (.text1));
3482 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
3483 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
3486 File: ld.info, Node: MEMORY, Next: PHDRS, Prev: SECTIONS, Up: Scripts
3491 The linker's default configuration permits allocation of all available
3492 memory. You can override this by using the `MEMORY' command.
3494 The `MEMORY' command describes the location and size of blocks of
3495 memory in the target. You can use it to describe which memory regions
3496 may be used by the linker, and which memory regions it must avoid. You
3497 can then assign sections to particular memory regions. The linker will
3498 set section addresses based on the memory regions, and will warn about
3499 regions that become too full. The linker will not shuffle sections
3500 around to fit into the available regions.
3502 A linker script may contain at most one use of the `MEMORY' command.
3503 However, you can define as many blocks of memory within it as you wish.
3507 NAME [(ATTR)] : ORIGIN = ORIGIN, LENGTH = LEN
3511 The NAME is a name used in the linker script to refer to the region.
3512 The region name has no meaning outside of the linker script. Region
3513 names are stored in a separate name space, and will not conflict with
3514 symbol names, file names, or section names. Each memory region must
3515 have a distinct name.
3517 The ATTR string is an optional list of attributes that specify
3518 whether to use a particular memory region for an input section which is
3519 not explicitly mapped in the linker script. As described in *note
3520 SECTIONS::, if you do not specify an output section for some input
3521 section, the linker will create an output section with the same name as
3522 the input section. If you define region attributes, the linker will use
3523 them to select the memory region for the output section that it creates.
3525 The ATTR string must consist only of the following characters:
3545 Invert the sense of any of the preceding attributes
3547 If a unmapped section matches any of the listed attributes other than
3548 `!', it will be placed in the memory region. The `!' attribute
3549 reverses this test, so that an unmapped section will be placed in the
3550 memory region only if it does not match any of the listed attributes.
3552 The ORIGIN is an numerical expression for the start address of the
3553 memory region. The expression must evaluate to a constant and it
3554 cannot involve any symbols. The keyword `ORIGIN' may be abbreviated to
3555 `org' or `o' (but not, for example, `ORG').
3557 The LEN is an expression for the size in bytes of the memory region.
3558 As with the ORIGIN expression, the expression must be numerical only
3559 and must evaluate to a constant. The keyword `LENGTH' may be
3560 abbreviated to `len' or `l'.
3562 In the following example, we specify that there are two memory
3563 regions available for allocation: one starting at `0' for 256 kilobytes,
3564 and the other starting at `0x40000000' for four megabytes. The linker
3565 will place into the `rom' memory region every section which is not
3566 explicitly mapped into a memory region, and is either read-only or
3567 executable. The linker will place other sections which are not
3568 explicitly mapped into a memory region into the `ram' memory region.
3572 rom (rx) : ORIGIN = 0, LENGTH = 256K
3573 ram (!rx) : org = 0x40000000, l = 4M
3576 Once you define a memory region, you can direct the linker to place
3577 specific output sections into that memory region by using the `>REGION'
3578 output section attribute. For example, if you have a memory region
3579 named `mem', you would use `>mem' in the output section definition.
3580 *Note Output Section Region::. If no address was specified for the
3581 output section, the linker will set the address to the next available
3582 address within the memory region. If the combined output sections
3583 directed to a memory region are too large for the region, the linker
3584 will issue an error message.
3586 It is possible to access the origin and length of a memory in an
3587 expression via the `ORIGIN(MEMORY)' and `LENGTH(MEMORY)' functions:
3589 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
3592 File: ld.info, Node: PHDRS, Next: VERSION, Prev: MEMORY, Up: Scripts
3597 The ELF object file format uses "program headers", also knows as
3598 "segments". The program headers describe how the program should be
3599 loaded into memory. You can print them out by using the `objdump'
3600 program with the `-p' option.
3602 When you run an ELF program on a native ELF system, the system loader
3603 reads the program headers in order to figure out how to load the
3604 program. This will only work if the program headers are set correctly.
3605 This manual does not describe the details of how the system loader
3606 interprets program headers; for more information, see the ELF ABI.
3608 The linker will create reasonable program headers by default.
3609 However, in some cases, you may need to specify the program headers more
3610 precisely. You may use the `PHDRS' command for this purpose. When the
3611 linker sees the `PHDRS' command in the linker script, it will not
3612 create any program headers other than the ones specified.
3614 The linker only pays attention to the `PHDRS' command when
3615 generating an ELF output file. In other cases, the linker will simply
3618 This is the syntax of the `PHDRS' command. The words `PHDRS',
3619 `FILEHDR', `AT', and `FLAGS' are keywords.
3623 NAME TYPE [ FILEHDR ] [ PHDRS ] [ AT ( ADDRESS ) ]
3624 [ FLAGS ( FLAGS ) ] ;
3627 The NAME is used only for reference in the `SECTIONS' command of the
3628 linker script. It is not put into the output file. Program header
3629 names are stored in a separate name space, and will not conflict with
3630 symbol names, file names, or section names. Each program header must
3631 have a distinct name.
3633 Certain program header types describe segments of memory which the
3634 system loader will load from the file. In the linker script, you
3635 specify the contents of these segments by placing allocatable output
3636 sections in the segments. You use the `:PHDR' output section attribute
3637 to place a section in a particular segment. *Note Output Section
3640 It is normal to put certain sections in more than one segment. This
3641 merely implies that one segment of memory contains another. You may
3642 repeat `:PHDR', using it once for each segment which should contain the
3645 If you place a section in one or more segments using `:PHDR', then
3646 the linker will place all subsequent allocatable sections which do not
3647 specify `:PHDR' in the same segments. This is for convenience, since
3648 generally a whole set of contiguous sections will be placed in a single
3649 segment. You can use `:NONE' to override the default segment and tell
3650 the linker to not put the section in any segment at all.
3652 You may use the `FILEHDR' and `PHDRS' keywords appear after the
3653 program header type to further describe the contents of the segment.
3654 The `FILEHDR' keyword means that the segment should include the ELF
3655 file header. The `PHDRS' keyword means that the segment should include
3656 the ELF program headers themselves.
3658 The TYPE may be one of the following. The numbers indicate the
3659 value of the keyword.
3662 Indicates an unused program header.
3665 Indicates that this program header describes a segment to be
3666 loaded from the file.
3669 Indicates a segment where dynamic linking information can be found.
3672 Indicates a segment where the name of the program interpreter may
3676 Indicates a segment holding note information.
3679 A reserved program header type, defined but not specified by the
3683 Indicates a segment where the program headers may be found.
3686 An expression giving the numeric type of the program header. This
3687 may be used for types not defined above.
3689 You can specify that a segment should be loaded at a particular
3690 address in memory by using an `AT' expression. This is identical to the
3691 `AT' command used as an output section attribute (*note Output Section
3692 LMA::). The `AT' command for a program header overrides the output
3695 The linker will normally set the segment flags based on the sections
3696 which comprise the segment. You may use the `FLAGS' keyword to
3697 explicitly specify the segment flags. The value of FLAGS must be an
3698 integer. It is used to set the `p_flags' field of the program header.
3700 Here is an example of `PHDRS'. This shows a typical set of program
3701 headers used on a native ELF system.
3705 headers PT_PHDR PHDRS ;
3707 text PT_LOAD FILEHDR PHDRS ;
3709 dynamic PT_DYNAMIC ;
3715 .interp : { *(.interp) } :text :interp
3716 .text : { *(.text) } :text
3717 .rodata : { *(.rodata) } /* defaults to :text */
3719 . = . + 0x1000; /* move to a new page in memory */
3720 .data : { *(.data) } :data
3721 .dynamic : { *(.dynamic) } :data :dynamic
3726 File: ld.info, Node: VERSION, Next: Expressions, Prev: PHDRS, Up: Scripts
3731 The linker supports symbol versions when using ELF. Symbol versions are
3732 only useful when using shared libraries. The dynamic linker can use
3733 symbol versions to select a specific version of a function when it runs
3734 a program that may have been linked against an earlier version of the
3737 You can include a version script directly in the main linker script,
3738 or you can supply the version script as an implicit linker script. You
3739 can also use the `--version-script' linker option.
3741 The syntax of the `VERSION' command is simply
3742 VERSION { version-script-commands }
3744 The format of the version script commands is identical to that used
3745 by Sun's linker in Solaris 2.5. The version script defines a tree of
3746 version nodes. You specify the node names and interdependencies in the
3747 version script. You can specify which symbols are bound to which
3748 version nodes, and you can reduce a specified set of symbols to local
3749 scope so that they are not globally visible outside of the shared
3752 The easiest way to demonstrate the version script language is with a
3772 "int f(int, double)";
3776 This example version script defines three version nodes. The first
3777 version node defined is `VERS_1.1'; it has no other dependencies. The
3778 script binds the symbol `foo1' to `VERS_1.1'. It reduces a number of
3779 symbols to local scope so that they are not visible outside of the
3780 shared library; this is done using wildcard patterns, so that any
3781 symbol whose name begins with `old', `original', or `new' is matched.
3782 The wildcard patterns available are the same as those used in the shell
3783 when matching filenames (also known as "globbing"). However, if you
3784 specify the symbol name inside double quotes, then the name is treated
3785 as literal, rather than as a glob pattern.
3787 Next, the version script defines node `VERS_1.2'. This node depends
3788 upon `VERS_1.1'. The script binds the symbol `foo2' to the version
3791 Finally, the version script defines node `VERS_2.0'. This node
3792 depends upon `VERS_1.2'. The scripts binds the symbols `bar1' and
3793 `bar2' are bound to the version node `VERS_2.0'.
3795 When the linker finds a symbol defined in a library which is not
3796 specifically bound to a version node, it will effectively bind it to an
3797 unspecified base version of the library. You can bind all otherwise
3798 unspecified symbols to a given version node by using `global: *;'
3799 somewhere in the version script.
3801 The names of the version nodes have no specific meaning other than
3802 what they might suggest to the person reading them. The `2.0' version
3803 could just as well have appeared in between `1.1' and `1.2'. However,
3804 this would be a confusing way to write a version script.
3806 Node name can be omitted, provided it is the only version node in
3807 the version script. Such version script doesn't assign any versions to
3808 symbols, only selects which symbols will be globally visible out and
3811 { global: foo; bar; local: *; };
3813 When you link an application against a shared library that has
3814 versioned symbols, the application itself knows which version of each
3815 symbol it requires, and it also knows which version nodes it needs from
3816 each shared library it is linked against. Thus at runtime, the dynamic
3817 loader can make a quick check to make sure that the libraries you have
3818 linked against do in fact supply all of the version nodes that the
3819 application will need to resolve all of the dynamic symbols. In this
3820 way it is possible for the dynamic linker to know with certainty that
3821 all external symbols that it needs will be resolvable without having to
3822 search for each symbol reference.
3824 The symbol versioning is in effect a much more sophisticated way of
3825 doing minor version checking that SunOS does. The fundamental problem
3826 that is being addressed here is that typically references to external
3827 functions are bound on an as-needed basis, and are not all bound when
3828 the application starts up. If a shared library is out of date, a
3829 required interface may be missing; when the application tries to use
3830 that interface, it may suddenly and unexpectedly fail. With symbol
3831 versioning, the user will get a warning when they start their program if
3832 the libraries being used with the application are too old.
3834 There are several GNU extensions to Sun's versioning approach. The
3835 first of these is the ability to bind a symbol to a version node in the
3836 source file where the symbol is defined instead of in the versioning
3837 script. This was done mainly to reduce the burden on the library
3838 maintainer. You can do this by putting something like:
3839 __asm__(".symver original_foo,foo@VERS_1.1");
3840 in the C source file. This renames the function `original_foo' to
3841 be an alias for `foo' bound to the version node `VERS_1.1'. The
3842 `local:' directive can be used to prevent the symbol `original_foo'
3843 from being exported. A `.symver' directive takes precedence over a
3846 The second GNU extension is to allow multiple versions of the same
3847 function to appear in a given shared library. In this way you can make
3848 an incompatible change to an interface without increasing the major
3849 version number of the shared library, while still allowing applications
3850 linked against the old interface to continue to function.
3852 To do this, you must use multiple `.symver' directives in the source
3853 file. Here is an example:
3855 __asm__(".symver original_foo,foo@");
3856 __asm__(".symver old_foo,foo@VERS_1.1");
3857 __asm__(".symver old_foo1,foo@VERS_1.2");
3858 __asm__(".symver new_foo,foo@@VERS_2.0");
3860 In this example, `foo@' represents the symbol `foo' bound to the
3861 unspecified base version of the symbol. The source file that contains
3862 this example would define 4 C functions: `original_foo', `old_foo',
3863 `old_foo1', and `new_foo'.
3865 When you have multiple definitions of a given symbol, there needs to
3866 be some way to specify a default version to which external references to
3867 this symbol will be bound. You can do this with the `foo@@VERS_2.0'
3868 type of `.symver' directive. You can only declare one version of a
3869 symbol as the default in this manner; otherwise you would effectively
3870 have multiple definitions of the same symbol.
3872 If you wish to bind a reference to a specific version of the symbol
3873 within the shared library, you can use the aliases of convenience
3874 (i.e., `old_foo'), or you can use the `.symver' directive to
3875 specifically bind to an external version of the function in question.
3877 You can also specify the language in the version script:
3879 VERSION extern "lang" { version-script-commands }
3881 The supported `lang's are `C', `C++', and `Java'. The linker will
3882 iterate over the list of symbols at the link time and demangle them
3883 according to `lang' before matching them to the patterns specified in
3884 `version-script-commands'.
3886 Demangled names may contains spaces and other special characters. As
3887 described above, you can use a glob pattern to match demangled names,
3888 or you can use a double-quoted string to match the string exactly. In
3889 the latter case, be aware that minor differences (such as differing
3890 whitespace) between the version script and the demangler output will
3891 cause a mismatch. As the exact string generated by the demangler might
3892 change in the future, even if the mangled name does not, you should
3893 check that all of your version directives are behaving as you expect
3897 File: ld.info, Node: Expressions, Next: Implicit Linker Scripts, Prev: VERSION, Up: Scripts
3899 3.10 Expressions in Linker Scripts
3900 ==================================
3902 The syntax for expressions in the linker script language is identical to
3903 that of C expressions. All expressions are evaluated as integers. All
3904 expressions are evaluated in the same size, which is 32 bits if both the
3905 host and target are 32 bits, and is otherwise 64 bits.
3907 You can use and set symbol values in expressions.
3909 The linker defines several special purpose builtin functions for use
3914 * Constants:: Constants
3915 * Symbols:: Symbol Names
3916 * Orphan Sections:: Orphan Sections
3917 * Location Counter:: The Location Counter
3918 * Operators:: Operators
3919 * Evaluation:: Evaluation
3920 * Expression Section:: The Section of an Expression
3921 * Builtin Functions:: Builtin Functions
3924 File: ld.info, Node: Constants, Next: Symbols, Up: Expressions
3929 All constants are integers.
3931 As in C, the linker considers an integer beginning with `0' to be
3932 octal, and an integer beginning with `0x' or `0X' to be hexadecimal.
3933 The linker considers other integers to be decimal.
3935 In addition, you can use the suffixes `K' and `M' to scale a
3936 constant by `1024' or `1024*1024' respectively. For example, the
3937 following all refer to the same quantity:
3943 File: ld.info, Node: Symbols, Next: Orphan Sections, Prev: Constants, Up: Expressions
3948 Unless quoted, symbol names start with a letter, underscore, or period
3949 and may include letters, digits, underscores, periods, and hyphens.
3950 Unquoted symbol names must not conflict with any keywords. You can
3951 specify a symbol which contains odd characters or has the same name as a
3952 keyword by surrounding the symbol name in double quotes:
3954 "with a space" = "also with a space" + 10;
3956 Since symbols can contain many non-alphabetic characters, it is
3957 safest to delimit symbols with spaces. For example, `A-B' is one
3958 symbol, whereas `A - B' is an expression involving subtraction.
3961 File: ld.info, Node: Orphan Sections, Next: Location Counter, Prev: Symbols, Up: Expressions
3963 3.10.3 Orphan Sections
3964 ----------------------
3966 Orphan sections are sections present in the input files which are not
3967 explicitly placed into the output file by the linker script. The
3968 linker will still copy these sections into the output file, but it has
3969 to guess as to where they should be placed. The linker uses a simple
3970 heuristic to do this. It attempts to place orphan sections after
3971 non-orphan sections of the same attribute, such as code vs data,
3972 loadable vs non-loadable, etc. If there is not enough room to do this
3973 then it places at the end of the file.
3975 For ELF targets, the attribute of the section includes section type
3976 as well as section flag.
3978 If an orphaned section's name is representable as a C identifier then
3979 the linker will automatically *note PROVIDE:: two symbols:
3980 __start_SECNAME and __end_SECNAME, where SECNAME is the name of the
3981 section. These indicate the start address and end address of the
3982 orphaned section respectively. Note: most section names are not
3983 representable as C identifiers because they contain a `.' character.
3986 File: ld.info, Node: Location Counter, Next: Operators, Prev: Orphan Sections, Up: Expressions
3988 3.10.4 The Location Counter
3989 ---------------------------
3991 The special linker variable "dot" `.' always contains the current
3992 output location counter. Since the `.' always refers to a location in
3993 an output section, it may only appear in an expression within a
3994 `SECTIONS' command. The `.' symbol may appear anywhere that an
3995 ordinary symbol is allowed in an expression.
3997 Assigning a value to `.' will cause the location counter to be
3998 moved. This may be used to create holes in the output section. The
3999 location counter may not be moved backwards inside an output section,
4000 and may not be moved backwards outside of an output section if so doing
4001 creates areas with overlapping LMAs.
4014 In the previous example, the `.text' section from `file1' is located
4015 at the beginning of the output section `output'. It is followed by a
4016 1000 byte gap. Then the `.text' section from `file2' appears, also
4017 with a 1000 byte gap following before the `.text' section from `file3'.
4018 The notation `= 0x12345678' specifies what data to write in the gaps
4019 (*note Output Section Fill::).
4021 Note: `.' actually refers to the byte offset from the start of the
4022 current containing object. Normally this is the `SECTIONS' statement,
4023 whose start address is 0, hence `.' can be used as an absolute address.
4024 If `.' is used inside a section description however, it refers to the
4025 byte offset from the start of that section, not an absolute address.
4026 Thus in a script like this:
4042 The `.text' section will be assigned a starting address of 0x100 and
4043 a size of exactly 0x200 bytes, even if there is not enough data in the
4044 `.text' input sections to fill this area. (If there is too much data,
4045 an error will be produced because this would be an attempt to move `.'
4046 backwards). The `.data' section will start at 0x500 and it will have
4047 an extra 0x600 bytes worth of space after the end of the values from
4048 the `.data' input sections and before the end of the `.data' output
4051 Setting symbols to the value of the location counter outside of an
4052 output section statement can result in unexpected values if the linker
4053 needs to place orphan sections. For example, given the following:
4066 If the linker needs to place some input section, e.g. `.rodata', not
4067 mentioned in the script, it might choose to place that section between
4068 `.text' and `.data'. You might think the linker should place `.rodata'
4069 on the blank line in the above script, but blank lines are of no
4070 particular significance to the linker. As well, the linker doesn't
4071 associate the above symbol names with their sections. Instead, it
4072 assumes that all assignments or other statements belong to the previous
4073 output section, except for the special case of an assignment to `.'.
4074 I.e., the linker will place the orphan `.rodata' section as if the
4075 script was written as follows:
4084 .rodata: { *(.rodata) }
4089 This may or may not be the script author's intention for the value of
4090 `start_of_data'. One way to influence the orphan section placement is
4091 to assign the location counter to itself, as the linker assumes that an
4092 assignment to `.' is setting the start address of a following output
4093 section and thus should be grouped with that section. So you could
4108 Now, the orphan `.rodata' section will be placed between
4109 `end_of_text' and `start_of_data'.
4112 File: ld.info, Node: Operators, Next: Evaluation, Prev: Location Counter, Up: Expressions
4117 The linker recognizes the standard C set of arithmetic operators, with
4118 the standard bindings and precedence levels:
4119 precedence associativity Operators Notes
4125 5 left == != > < <= >=
4131 11 right &= += -= *= /= (2)
4133 Notes: (1) Prefix operators (2) *Note Assignments::.
4136 File: ld.info, Node: Evaluation, Next: Expression Section, Prev: Operators, Up: Expressions
4141 The linker evaluates expressions lazily. It only computes the value of
4142 an expression when absolutely necessary.
4144 The linker needs some information, such as the value of the start
4145 address of the first section, and the origins and lengths of memory
4146 regions, in order to do any linking at all. These values are computed
4147 as soon as possible when the linker reads in the linker script.
4149 However, other values (such as symbol values) are not known or needed
4150 until after storage allocation. Such values are evaluated later, when
4151 other information (such as the sizes of output sections) is available
4152 for use in the symbol assignment expression.
4154 The sizes of sections cannot be known until after allocation, so
4155 assignments dependent upon these are not performed until after
4158 Some expressions, such as those depending upon the location counter
4159 `.', must be evaluated during section allocation.
4161 If the result of an expression is required, but the value is not
4162 available, then an error results. For example, a script like the
4166 .text 9+this_isnt_constant :
4169 will cause the error message `non constant expression for initial
4173 File: ld.info, Node: Expression Section, Next: Builtin Functions, Prev: Evaluation, Up: Expressions
4175 3.10.7 The Section of an Expression
4176 -----------------------------------
4178 When the linker evaluates an expression, the result is either absolute
4179 or relative to some section. A relative expression is expressed as a
4180 fixed offset from the base of a section.
4182 The position of the expression within the linker script determines
4183 whether it is absolute or relative. An expression which appears within
4184 an output section definition is relative to the base of the output
4185 section. An expression which appears elsewhere will be absolute.
4187 A symbol set to a relative expression will be relocatable if you
4188 request relocatable output using the `-r' option. That means that a
4189 further link operation may change the value of the symbol. The symbol's
4190 section will be the section of the relative expression.
4192 A symbol set to an absolute expression will retain the same value
4193 through any further link operation. The symbol will be absolute, and
4194 will not have any particular associated section.
4196 You can use the builtin function `ABSOLUTE' to force an expression
4197 to be absolute when it would otherwise be relative. For example, to
4198 create an absolute symbol set to the address of the end of the output
4202 .data : { *(.data) _edata = ABSOLUTE(.); }
4204 If `ABSOLUTE' were not used, `_edata' would be relative to the
4208 File: ld.info, Node: Builtin Functions, Prev: Expression Section, Up: Expressions
4210 3.10.8 Builtin Functions
4211 ------------------------
4213 The linker script language includes a number of builtin functions for
4214 use in linker script expressions.
4217 Return the absolute (non-relocatable, as opposed to non-negative)
4218 value of the expression EXP. Primarily useful to assign an
4219 absolute value to a symbol within a section definition, where
4220 symbol values are normally section relative. *Note Expression
4224 Return the absolute address (the VMA) of the named SECTION. Your
4225 script must previously have defined the location of that section.
4226 In the following example, `symbol_1' and `symbol_2' are assigned
4231 start_of_output_1 = ABSOLUTE(.);
4236 symbol_1 = ADDR(.output1);
4237 symbol_2 = start_of_output_1;
4243 Return the location counter (`.') or arbitrary expression aligned
4244 to the next ALIGN boundary. The single operand `ALIGN' doesn't
4245 change the value of the location counter--it just does arithmetic
4246 on it. The two operand `ALIGN' allows an arbitrary expression to
4247 be aligned upwards (`ALIGN(ALIGN)' is equivalent to `ALIGN(.,
4250 Here is an example which aligns the output `.data' section to the
4251 next `0x2000' byte boundary after the preceding section and sets a
4252 variable within the section to the next `0x8000' boundary after the
4255 .data ALIGN(0x2000): {
4257 variable = ALIGN(0x8000);
4260 The first use of `ALIGN' in this example specifies the location of
4261 a section because it is used as the optional ADDRESS attribute of
4262 a section definition (*note Output Section Address::). The second
4263 use of `ALIGN' is used to defines the value of a symbol.
4265 The builtin function `NEXT' is closely related to `ALIGN'.
4268 Return the alignment in bytes of the named SECTION, if that
4269 section has been allocated. If the section has not been allocated
4270 when this is evaluated, the linker will report an error. In the
4271 following example, the alignment of the `.output' section is
4272 stored as the first value in that section.
4275 LONG (ALIGNOF (.output))
4281 This is a synonym for `ALIGN', for compatibility with older linker
4282 scripts. It is most often seen when setting the address of an
4285 `DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE)'
4286 This is equivalent to either
4287 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - 1)))
4289 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - COMMONPAGESIZE)))
4290 depending on whether the latter uses fewer COMMONPAGESIZE sized
4291 pages for the data segment (area between the result of this
4292 expression and `DATA_SEGMENT_END') than the former or not. If the
4293 latter form is used, it means COMMONPAGESIZE bytes of runtime
4294 memory will be saved at the expense of up to COMMONPAGESIZE wasted
4295 bytes in the on-disk file.
4297 This expression can only be used directly in `SECTIONS' commands,
4298 not in any output section descriptions and only once in the linker
4299 script. COMMONPAGESIZE should be less or equal to MAXPAGESIZE and
4300 should be the system page size the object wants to be optimized
4301 for (while still working on system page sizes up to MAXPAGESIZE).
4304 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
4306 `DATA_SEGMENT_END(EXP)'
4307 This defines the end of data segment for `DATA_SEGMENT_ALIGN'
4308 evaluation purposes.
4310 . = DATA_SEGMENT_END(.);
4312 `DATA_SEGMENT_RELRO_END(OFFSET, EXP)'
4313 This defines the end of the `PT_GNU_RELRO' segment when `-z relro'
4314 option is used. Second argument is returned. When `-z relro'
4315 option is not present, `DATA_SEGMENT_RELRO_END' does nothing,
4316 otherwise `DATA_SEGMENT_ALIGN' is padded so that EXP + OFFSET is
4317 aligned to the most commonly used page boundary for particular
4318 target. If present in the linker script, it must always come in
4319 between `DATA_SEGMENT_ALIGN' and `DATA_SEGMENT_END'.
4321 . = DATA_SEGMENT_RELRO_END(24, .);
4324 Return 1 if SYMBOL is in the linker global symbol table and is
4325 defined before the statement using DEFINED in the script, otherwise
4326 return 0. You can use this function to provide default values for
4327 symbols. For example, the following script fragment shows how to
4328 set a global symbol `begin' to the first location in the `.text'
4329 section--but if a symbol called `begin' already existed, its value
4334 begin = DEFINED(begin) ? begin : . ;
4341 Return the length of the memory region named MEMORY.
4344 Return the absolute LMA of the named SECTION. This is normally
4345 the same as `ADDR', but it may be different if the `AT' attribute
4346 is used in the output section definition (*note Output Section
4350 Returns the maximum of EXP1 and EXP2.
4353 Returns the minimum of EXP1 and EXP2.
4356 Return the next unallocated address that is a multiple of EXP.
4357 This function is closely related to `ALIGN(EXP)'; unless you use
4358 the `MEMORY' command to define discontinuous memory for the output
4359 file, the two functions are equivalent.
4362 Return the origin of the memory region named MEMORY.
4364 `SEGMENT_START(SEGMENT, DEFAULT)'
4365 Return the base address of the named SEGMENT. If an explicit
4366 value has been given for this segment (with a command-line `-T'
4367 option) that value will be returned; otherwise the value will be
4368 DEFAULT. At present, the `-T' command-line option can only be
4369 used to set the base address for the "text", "data", and "bss"
4370 sections, but you use `SEGMENT_START' with any segment name.
4373 Return the size in bytes of the named SECTION, if that section has
4374 been allocated. If the section has not been allocated when this is
4375 evaluated, the linker will report an error. In the following
4376 example, `symbol_1' and `symbol_2' are assigned identical values:
4383 symbol_1 = .end - .start ;
4384 symbol_2 = SIZEOF(.output);
4389 Return the size in bytes of the output file's headers. This is
4390 information which appears at the start of the output file. You
4391 can use this number when setting the start address of the first
4392 section, if you choose, to facilitate paging.
4394 When producing an ELF output file, if the linker script uses the
4395 `SIZEOF_HEADERS' builtin function, the linker must compute the
4396 number of program headers before it has determined all the section
4397 addresses and sizes. If the linker later discovers that it needs
4398 additional program headers, it will report an error `not enough
4399 room for program headers'. To avoid this error, you must avoid
4400 using the `SIZEOF_HEADERS' function, or you must rework your linker
4401 script to avoid forcing the linker to use additional program
4402 headers, or you must define the program headers yourself using the
4403 `PHDRS' command (*note PHDRS::).
4406 File: ld.info, Node: Implicit Linker Scripts, Prev: Expressions, Up: Scripts
4408 3.11 Implicit Linker Scripts
4409 ============================
4411 If you specify a linker input file which the linker can not recognize as
4412 an object file or an archive file, it will try to read the file as a
4413 linker script. If the file can not be parsed as a linker script, the
4414 linker will report an error.
4416 An implicit linker script will not replace the default linker script.
4418 Typically an implicit linker script would contain only symbol
4419 assignments, or the `INPUT', `GROUP', or `VERSION' commands.
4421 Any input files read because of an implicit linker script will be
4422 read at the position in the command line where the implicit linker
4423 script was read. This can affect archive searching.
4426 File: ld.info, Node: Machine Dependent, Next: BFD, Prev: Scripts, Up: Top
4428 4 Machine Dependent Features
4429 ****************************
4431 `ld' has additional features on some platforms; the following sections
4432 describe them. Machines where `ld' has no additional functionality are
4438 * H8/300:: `ld' and the H8/300
4440 * i960:: `ld' and the Intel 960 family
4442 * ARM:: `ld' and the ARM family
4444 * HPPA ELF32:: `ld' and HPPA 32-bit ELF
4446 * M68K:: `ld' and the Motorola 68K family
4448 * MMIX:: `ld' and MMIX
4450 * MSP430:: `ld' and MSP430
4452 * M68HC11/68HC12:: `ld' and the Motorola 68HC11 and 68HC12 families
4454 * PowerPC ELF32:: `ld' and PowerPC 32-bit ELF Support
4456 * PowerPC64 ELF64:: `ld' and PowerPC64 64-bit ELF Support
4458 * SPU ELF:: `ld' and SPU ELF Support
4460 * TI COFF:: `ld' and TI COFF
4462 * WIN32:: `ld' and WIN32 (cygwin/mingw)
4464 * Xtensa:: `ld' and Xtensa Processors
4467 File: ld.info, Node: H8/300, Next: i960, Up: Machine Dependent
4469 4.1 `ld' and the H8/300
4470 =======================
4472 For the H8/300, `ld' can perform these global optimizations when you
4473 specify the `--relax' command-line option.
4475 _relaxing address modes_
4476 `ld' finds all `jsr' and `jmp' instructions whose targets are
4477 within eight bits, and turns them into eight-bit program-counter
4478 relative `bsr' and `bra' instructions, respectively.
4480 _synthesizing instructions_
4481 `ld' finds all `mov.b' instructions which use the sixteen-bit
4482 absolute address form, but refer to the top page of memory, and
4483 changes them to use the eight-bit address form. (That is: the
4484 linker turns `mov.b `@'AA:16' into `mov.b `@'AA:8' whenever the
4485 address AA is in the top page of memory).
4487 _bit manipulation instructions_
4488 `ld' finds all bit manipulation instructions like `band, bclr,
4489 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst,
4490 bxor' which use 32 bit and 16 bit absolute address form, but refer
4491 to the top page of memory, and changes them to use the 8 bit
4492 address form. (That is: the linker turns `bset #xx:3,`@'AA:32'
4493 into `bset #xx:3,`@'AA:8' whenever the address AA is in the top
4496 _system control instructions_
4497 `ld' finds all `ldc.w, stc.w' instructions which use the 32 bit
4498 absolute address form, but refer to the top page of memory, and
4499 changes them to use 16 bit address form. (That is: the linker
4500 turns `ldc.w `@'AA:32,ccr' into `ldc.w `@'AA:16,ccr' whenever the
4501 address AA is in the top page of memory).
4504 File: ld.info, Node: i960, Next: ARM, Prev: H8/300, Up: Machine Dependent
4506 4.2 `ld' and the Intel 960 Family
4507 =================================
4509 You can use the `-AARCHITECTURE' command line option to specify one of
4510 the two-letter names identifying members of the 960 family; the option
4511 specifies the desired output target, and warns of any incompatible
4512 instructions in the input files. It also modifies the linker's search
4513 strategy for archive libraries, to support the use of libraries
4514 specific to each particular architecture, by including in the search
4515 loop names suffixed with the string identifying the architecture.
4517 For example, if your `ld' command line included `-ACA' as well as
4518 `-ltry', the linker would look (in its built-in search paths, and in
4519 any paths you specify with `-L') for a library with the names
4526 The first two possibilities would be considered in any event; the last
4527 two are due to the use of `-ACA'.
4529 You can meaningfully use `-A' more than once on a command line, since
4530 the 960 architecture family allows combination of target architectures;
4531 each use will add another pair of name variants to search for when `-l'
4532 specifies a library.
4534 `ld' supports the `--relax' option for the i960 family. If you
4535 specify `--relax', `ld' finds all `balx' and `calx' instructions whose
4536 targets are within 24 bits, and turns them into 24-bit program-counter
4537 relative `bal' and `cal' instructions, respectively. `ld' also turns
4538 `cal' instructions into `bal' instructions when it determines that the
4539 target subroutine is a leaf routine (that is, the target subroutine does
4540 not itself call any subroutines).
4542 The `--fix-cortex-a8' switch enables a link-time workaround for an
4543 erratum in certain Cortex-A8 processors. The workaround is enabled by
4544 default if you are targeting the ARM v7-A architecture profile. It can
4545 be enabled otherwise by specifying `--fix-cortex-a8', or disabled
4546 unconditionally by specifying `--no-fix-cortex-a8'.
4548 The erratum only affects Thumb-2 code. Please contact ARM for
4552 File: ld.info, Node: M68HC11/68HC12, Next: PowerPC ELF32, Prev: MSP430, Up: Machine Dependent
4554 4.3 `ld' and the Motorola 68HC11 and 68HC12 families
4555 ====================================================
4557 4.3.1 Linker Relaxation
4558 -----------------------
4560 For the Motorola 68HC11, `ld' can perform these global optimizations
4561 when you specify the `--relax' command-line option.
4563 _relaxing address modes_
4564 `ld' finds all `jsr' and `jmp' instructions whose targets are
4565 within eight bits, and turns them into eight-bit program-counter
4566 relative `bsr' and `bra' instructions, respectively.
4568 `ld' also looks at all 16-bit extended addressing modes and
4569 transforms them in a direct addressing mode when the address is in
4570 page 0 (between 0 and 0x0ff).
4572 _relaxing gcc instruction group_
4573 When `gcc' is called with `-mrelax', it can emit group of
4574 instructions that the linker can optimize to use a 68HC11 direct
4575 addressing mode. These instructions consists of `bclr' or `bset'
4579 4.3.2 Trampoline Generation
4580 ---------------------------
4582 For 68HC11 and 68HC12, `ld' can generate trampoline code to call a far
4583 function using a normal `jsr' instruction. The linker will also change
4584 the relocation to some far function to use the trampoline address
4585 instead of the function address. This is typically the case when a
4586 pointer to a function is taken. The pointer will in fact point to the
4587 function trampoline.
4590 File: ld.info, Node: ARM, Next: HPPA ELF32, Prev: i960, Up: Machine Dependent
4592 4.4 `ld' and the ARM family
4593 ===========================
4595 For the ARM, `ld' will generate code stubs to allow functions calls
4596 between ARM and Thumb code. These stubs only work with code that has
4597 been compiled and assembled with the `-mthumb-interwork' command line
4598 option. If it is necessary to link with old ARM object files or
4599 libraries, which have not been compiled with the -mthumb-interwork
4600 option then the `--support-old-code' command line switch should be
4601 given to the linker. This will make it generate larger stub functions
4602 which will work with non-interworking aware ARM code. Note, however,
4603 the linker does not support generating stubs for function calls to
4604 non-interworking aware Thumb code.
4606 The `--thumb-entry' switch is a duplicate of the generic `--entry'
4607 switch, in that it sets the program's starting address. But it also
4608 sets the bottom bit of the address, so that it can be branched to using
4609 a BX instruction, and the program will start executing in Thumb mode
4612 The `--be8' switch instructs `ld' to generate BE8 format
4613 executables. This option is only valid when linking big-endian objects.
4614 The resulting image will contain big-endian data and little-endian code.
4616 The `R_ARM_TARGET1' relocation is typically used for entries in the
4617 `.init_array' section. It is interpreted as either `R_ARM_REL32' or
4618 `R_ARM_ABS32', depending on the target. The `--target1-rel' and
4619 `--target1-abs' switches override the default.
4621 The `--target2=type' switch overrides the default definition of the
4622 `R_ARM_TARGET2' relocation. Valid values for `type', their meanings,
4623 and target defaults are as follows:
4625 `R_ARM_REL32' (arm*-*-elf, arm*-*-eabi)
4628 `R_ARM_ABS32' (arm*-*-symbianelf)
4631 `R_ARM_GOT_PREL' (arm*-*-linux, arm*-*-*bsd)
4633 The `R_ARM_V4BX' relocation (defined by the ARM AAELF specification)
4634 enables objects compiled for the ARMv4 architecture to be
4635 interworking-safe when linked with other objects compiled for ARMv4t,
4636 but also allows pure ARMv4 binaries to be built from the same ARMv4
4639 In the latter case, the switch `--fix-v4bx' must be passed to the
4640 linker, which causes v4t `BX rM' instructions to be rewritten as `MOV
4641 PC,rM', since v4 processors do not have a `BX' instruction.
4643 In the former case, the switch should not be used, and `R_ARM_V4BX'
4644 relocations are ignored.
4646 Replace `BX rM' instructions identified by `R_ARM_V4BX' relocations
4647 with a branch to the following veneer:
4653 This allows generation of libraries/applications that work on ARMv4
4654 cores and are still interworking safe. Note that the above veneer
4655 clobbers the condition flags, so may cause incorrect progrm behavior in
4658 The `--use-blx' switch enables the linker to use ARM/Thumb BLX
4659 instructions (available on ARMv5t and above) in various situations.
4660 Currently it is used to perform calls via the PLT from Thumb code using
4661 BLX rather than using BX and a mode-switching stub before each PLT
4662 entry. This should lead to such calls executing slightly faster.
4664 This option is enabled implicitly for SymbianOS, so there is no need
4665 to specify it if you are using that target.
4667 The `--vfp11-denorm-fix' switch enables a link-time workaround for a
4668 bug in certain VFP11 coprocessor hardware, which sometimes allows
4669 instructions with denorm operands (which must be handled by support
4670 code) to have those operands overwritten by subsequent instructions
4671 before the support code can read the intended values.
4673 The bug may be avoided in scalar mode if you allow at least one
4674 intervening instruction between a VFP11 instruction which uses a
4675 register and another instruction which writes to the same register, or
4676 at least two intervening instructions if vector mode is in use. The bug
4677 only affects full-compliance floating-point mode: you do not need this
4678 workaround if you are using "runfast" mode. Please contact ARM for
4681 If you know you are using buggy VFP11 hardware, you can enable this
4682 workaround by specifying the linker option `--vfp-denorm-fix=scalar' if
4683 you are using the VFP11 scalar mode only, or `--vfp-denorm-fix=vector'
4684 if you are using vector mode (the latter also works for scalar code).
4685 The default is `--vfp-denorm-fix=none'.
4687 If the workaround is enabled, instructions are scanned for
4688 potentially-troublesome sequences, and a veneer is created for each
4689 such sequence which may trigger the erratum. The veneer consists of the
4690 first instruction of the sequence and a branch back to the subsequent
4691 instruction. The original instruction is then replaced with a branch to
4692 the veneer. The extra cycles required to call and return from the veneer
4693 are sufficient to avoid the erratum in both the scalar and vector cases.
4695 The `--no-enum-size-warning' switch prevents the linker from warning
4696 when linking object files that specify incompatible EABI enumeration
4697 size attributes. For example, with this switch enabled, linking of an
4698 object file using 32-bit enumeration values with another using
4699 enumeration values fitted into the smallest possible space will not be
4702 The `--no-wchar-size-warning' switch prevents the linker from
4703 warning when linking object files that specify incompatible EABI
4704 `wchar_t' size attributes. For example, with this switch enabled,
4705 linking of an object file using 32-bit `wchar_t' values with another
4706 using 16-bit `wchar_t' values will not be diagnosed.
4708 The `--pic-veneer' switch makes the linker use PIC sequences for
4709 ARM/Thumb interworking veneers, even if the rest of the binary is not
4710 PIC. This avoids problems on uClinux targets where `--emit-relocs' is
4711 used to generate relocatable binaries.
4713 The linker will automatically generate and insert small sequences of
4714 code into a linked ARM ELF executable whenever an attempt is made to
4715 perform a function call to a symbol that is too far away. The
4716 placement of these sequences of instructions - called stubs - is
4717 controlled by the command line option `--stub-group-size=N'. The
4718 placement is important because a poor choice can create a need for
4719 duplicate stubs, increasing the code sizw. The linker will try to
4720 group stubs together in order to reduce interruptions to the flow of
4721 code, but it needs guidance as to how big these groups should be and
4722 where they should be placed.
4724 The value of `N', the parameter to the `--stub-group-size=' option
4725 controls where the stub groups are placed. If it is negative then all
4726 stubs are placed after the first branch that needs them. If it is
4727 positive then the stubs can be placed either before or after the
4728 branches that need them. If the value of `N' is 1 (either +1 or -1)
4729 then the linker will choose exactly where to place groups of stubs,
4730 using its built in heuristics. A value of `N' greater than 1 (or
4731 smaller than -1) tells the linker that a single group of stubs can
4732 service at most `N' bytes from the input sections.
4734 The default, if `--stub-group-size=' is not specified, is `N = +1'.
4736 Farcalls stubs insertion is fully supported for the ARM-EABI target
4737 only, because it relies on object files properties not present
4741 File: ld.info, Node: HPPA ELF32, Next: M68K, Prev: ARM, Up: Machine Dependent
4743 4.5 `ld' and HPPA 32-bit ELF Support
4744 ====================================
4746 When generating a shared library, `ld' will by default generate import
4747 stubs suitable for use with a single sub-space application. The
4748 `--multi-subspace' switch causes `ld' to generate export stubs, and
4749 different (larger) import stubs suitable for use with multiple
4752 Long branch stubs and import/export stubs are placed by `ld' in stub
4753 sections located between groups of input sections. `--stub-group-size'
4754 specifies the maximum size of a group of input sections handled by one
4755 stub section. Since branch offsets are signed, a stub section may
4756 serve two groups of input sections, one group before the stub section,
4757 and one group after it. However, when using conditional branches that
4758 require stubs, it may be better (for branch prediction) that stub
4759 sections only serve one group of input sections. A negative value for
4760 `N' chooses this scheme, ensuring that branches to stubs always use a
4761 negative offset. Two special values of `N' are recognized, `1' and
4762 `-1'. These both instruct `ld' to automatically size input section
4763 groups for the branch types detected, with the same behaviour regarding
4764 stub placement as other positive or negative values of `N' respectively.
4766 Note that `--stub-group-size' does not split input sections. A
4767 single input section larger than the group size specified will of course
4768 create a larger group (of one section). If input sections are too
4769 large, it may not be possible for a branch to reach its stub.
4772 File: ld.info, Node: M68K, Next: MMIX, Prev: HPPA ELF32, Up: Machine Dependent
4774 4.6 `ld' and the Motorola 68K family
4775 ====================================
4777 The `--got=TYPE' option lets you choose the GOT generation scheme. The
4778 choices are `single', `negative', `multigot' and `target'. When
4779 `target' is selected the linker chooses the default GOT generation
4780 scheme for the current target. `single' tells the linker to generate a
4781 single GOT with entries only at non-negative offsets. `negative'
4782 instructs the linker to generate a single GOT with entries at both
4783 negative and positive offsets. Not all environments support such GOTs.
4784 `multigot' allows the linker to generate several GOTs in the output
4785 file. All GOT references from a single input object file access the
4786 same GOT, but references from different input object files might access
4787 different GOTs. Not all environments support such GOTs.
4790 File: ld.info, Node: MMIX, Next: MSP430, Prev: M68K, Up: Machine Dependent
4795 For MMIX, there is a choice of generating `ELF' object files or `mmo'
4796 object files when linking. The simulator `mmix' understands the `mmo'
4797 format. The binutils `objcopy' utility can translate between the two
4800 There is one special section, the `.MMIX.reg_contents' section.
4801 Contents in this section is assumed to correspond to that of global
4802 registers, and symbols referring to it are translated to special
4803 symbols, equal to registers. In a final link, the start address of the
4804 `.MMIX.reg_contents' section corresponds to the first allocated global
4805 register multiplied by 8. Register `$255' is not included in this
4806 section; it is always set to the program entry, which is at the symbol
4807 `Main' for `mmo' files.
4809 Global symbols with the prefix `__.MMIX.start.', for example
4810 `__.MMIX.start..text' and `__.MMIX.start..data' are special. The
4811 default linker script uses these to set the default start address of a
4814 Initial and trailing multiples of zero-valued 32-bit words in a
4815 section, are left out from an mmo file.
4818 File: ld.info, Node: MSP430, Next: M68HC11/68HC12, Prev: MMIX, Up: Machine Dependent
4823 For the MSP430 it is possible to select the MPU architecture. The flag
4824 `-m [mpu type]' will select an appropriate linker script for selected
4825 MPU type. (To get a list of known MPUs just pass `-m help' option to
4828 The linker will recognize some extra sections which are MSP430
4832 Defines a portion of ROM where interrupt vectors located.
4835 Defines the bootloader portion of the ROM (if applicable). Any
4836 code in this section will be uploaded to the MPU.
4839 Defines an information memory section (if applicable). Any code in
4840 this section will be uploaded to the MPU.
4843 This is the same as the `.infomem' section except that any code in
4844 this section will not be uploaded to the MPU.
4847 Denotes a portion of RAM located above `.bss' section.
4849 The last two sections are used by gcc.
4852 File: ld.info, Node: PowerPC ELF32, Next: PowerPC64 ELF64, Prev: M68HC11/68HC12, Up: Machine Dependent
4854 4.9 `ld' and PowerPC 32-bit ELF Support
4855 =======================================
4857 Branches on PowerPC processors are limited to a signed 26-bit
4858 displacement, which may result in `ld' giving `relocation truncated to
4859 fit' errors with very large programs. `--relax' enables the generation
4860 of trampolines that can access the entire 32-bit address space. These
4861 trampolines are inserted at section boundaries, so may not themselves
4862 be reachable if an input section exceeds 33M in size.
4865 Current PowerPC GCC accepts a `-msecure-plt' option that generates
4866 code capable of using a newer PLT and GOT layout that has the
4867 security advantage of no executable section ever needing to be
4868 writable and no writable section ever being executable. PowerPC
4869 `ld' will generate this layout, including stubs to access the PLT,
4870 if all input files (including startup and static libraries) were
4871 compiled with `-msecure-plt'. `--bss-plt' forces the old BSS PLT
4872 (and GOT layout) which can give slightly better performance.
4875 `ld' will use the new PLT and GOT layout if it is linking new
4876 `-fpic' or `-fPIC' code, but does not do so automatically when
4877 linking non-PIC code. This option requests the new PLT and GOT
4878 layout. A warning will be given if some object file requires the
4882 The new secure PLT and GOT are placed differently relative to other
4883 sections compared to older BSS PLT and GOT placement. The
4884 location of `.plt' must change because the new secure PLT is an
4885 initialized section while the old PLT is uninitialized. The
4886 reason for the `.got' change is more subtle: The new placement
4887 allows `.got' to be read-only in applications linked with `-z
4888 relro -z now'. However, this placement means that `.sdata' cannot
4889 always be used in shared libraries, because the PowerPC ABI
4890 accesses `.sdata' in shared libraries from the GOT pointer.
4891 `--sdata-got' forces the old GOT placement. PowerPC GCC doesn't
4892 use `.sdata' in shared libraries, so this option is really only
4893 useful for other compilers that may do so.
4896 This option causes `ld' to label linker stubs with a local symbol
4897 that encodes the stub type and destination.
4900 PowerPC `ld' normally performs some optimization of code sequences
4901 used to access Thread-Local Storage. Use this option to disable
4905 File: ld.info, Node: PowerPC64 ELF64, Next: SPU ELF, Prev: PowerPC ELF32, Up: Machine Dependent
4907 4.10 `ld' and PowerPC64 64-bit ELF Support
4908 ==========================================
4911 Long branch stubs, PLT call stubs and TOC adjusting stubs are
4912 placed by `ld' in stub sections located between groups of input
4913 sections. `--stub-group-size' specifies the maximum size of a
4914 group of input sections handled by one stub section. Since branch
4915 offsets are signed, a stub section may serve two groups of input
4916 sections, one group before the stub section, and one group after
4917 it. However, when using conditional branches that require stubs,
4918 it may be better (for branch prediction) that stub sections only
4919 serve one group of input sections. A negative value for `N'
4920 chooses this scheme, ensuring that branches to stubs always use a
4921 negative offset. Two special values of `N' are recognized, `1'
4922 and `-1'. These both instruct `ld' to automatically size input
4923 section groups for the branch types detected, with the same
4924 behaviour regarding stub placement as other positive or negative
4925 values of `N' respectively.
4927 Note that `--stub-group-size' does not split input sections. A
4928 single input section larger than the group size specified will of
4929 course create a larger group (of one section). If input sections
4930 are too large, it may not be possible for a branch to reach its
4934 This option causes `ld' to label linker stubs with a local symbol
4935 that encodes the stub type and destination.
4937 `--dotsyms, --no-dotsyms'
4938 These two options control how `ld' interprets version patterns in
4939 a version script. Older PowerPC64 compilers emitted both a
4940 function descriptor symbol with the same name as the function, and
4941 a code entry symbol with the name prefixed by a dot (`.'). To
4942 properly version a function `foo', the version script thus needs
4943 to control both `foo' and `.foo'. The option `--dotsyms', on by
4944 default, automatically adds the required dot-prefixed patterns.
4945 Use `--no-dotsyms' to disable this feature.
4948 PowerPC64 `ld' normally performs some optimization of code
4949 sequences used to access Thread-Local Storage. Use this option to
4950 disable the optimization.
4953 PowerPC64 `ld' normally removes `.opd' section entries
4954 corresponding to deleted link-once functions, or functions removed
4955 by the action of `--gc-sections' or linker script `/DISCARD/'.
4956 Use this option to disable `.opd' optimization.
4958 `--non-overlapping-opd'
4959 Some PowerPC64 compilers have an option to generate compressed
4960 `.opd' entries spaced 16 bytes apart, overlapping the third word,
4961 the static chain pointer (unused in C) with the first word of the
4962 next entry. This option expands such entries to the full 24 bytes.
4965 PowerPC64 `ld' normally removes unused `.toc' section entries.
4966 Such entries are detected by examining relocations that reference
4967 the TOC in code sections. A reloc in a deleted code section marks
4968 a TOC word as unneeded, while a reloc in a kept code section marks
4969 a TOC word as needed. Since the TOC may reference itself, TOC
4970 relocs are also examined. TOC words marked as both needed and
4971 unneeded will of course be kept. TOC words without any referencing
4972 reloc are assumed to be part of a multi-word entry, and are kept or
4973 discarded as per the nearest marked preceding word. This works
4974 reliably for compiler generated code, but may be incorrect if
4975 assembly code is used to insert TOC entries. Use this option to
4976 disable the optimization.
4979 By default, PowerPC64 GCC generates code for a TOC model where TOC
4980 entries are accessed with a 16-bit offset from r2. This limits the
4981 total TOC size to 64K. PowerPC64 `ld' extends this limit by
4982 grouping code sections such that each group uses less than 64K for
4983 its TOC entries, then inserts r2 adjusting stubs between
4984 inter-group calls. `ld' does not split apart input sections, so
4985 cannot help if a single input file has a `.toc' section that
4986 exceeds 64K, most likely from linking multiple files with `ld -r'.
4987 Use this option to turn off this feature.
4990 File: ld.info, Node: SPU ELF, Next: TI COFF, Prev: PowerPC64 ELF64, Up: Machine Dependent
4992 4.11 `ld' and SPU ELF Support
4993 =============================
4996 This option marks an executable as a PIC plugin module.
4999 Normally, `ld' recognizes calls to functions within overlay
5000 regions, and redirects such calls to an overlay manager via a stub.
5001 `ld' also provides a built-in overlay manager. This option turns
5002 off all this special overlay handling.
5005 This option causes `ld' to label overlay stubs with a local symbol
5006 that encodes the stub type and destination.
5008 `--extra-overlay-stubs'
5009 This option causes `ld' to add overlay call stubs on all function
5010 calls out of overlay regions. Normally stubs are not added on
5011 calls to non-overlay regions.
5013 `--local-store=lo:hi'
5014 `ld' usually checks that a final executable for SPU fits in the
5015 address range 0 to 256k. This option may be used to change the
5016 range. Disable the check entirely with `--local-store=0:0'.
5019 SPU local store space is limited. Over-allocation of stack space
5020 unnecessarily limits space available for code and data, while
5021 under-allocation results in runtime failures. If given this
5022 option, `ld' will provide an estimate of maximum stack usage.
5023 `ld' does this by examining symbols in code sections to determine
5024 the extents of functions, and looking at function prologues for
5025 stack adjusting instructions. A call-graph is created by looking
5026 for relocations on branch instructions. The graph is then searched
5027 for the maximum stack usage path. Note that this analysis does not
5028 find calls made via function pointers, and does not handle
5029 recursion and other cycles in the call graph. Stack usage may be
5030 under-estimated if your code makes such calls. Also, stack usage
5031 for dynamic allocation, e.g. alloca, will not be detected. If a
5032 link map is requested, detailed information about each function's
5033 stack usage and calls will be given.
5036 This option, if given along with `--stack-analysis' will result in
5037 `ld' emitting stack sizing symbols for each function. These take
5038 the form `__stack_<function_name>' for global functions, and
5039 `__stack_<number>_<function_name>' for static functions.
5040 `<number>' is the section id in hex. The value of such symbols is
5041 the stack requirement for the corresponding function. The symbol
5042 size will be zero, type `STT_NOTYPE', binding `STB_LOCAL', and
5046 File: ld.info, Node: TI COFF, Next: WIN32, Prev: SPU ELF, Up: Machine Dependent
5048 4.12 `ld''s Support for Various TI COFF Versions
5049 ================================================
5051 The `--format' switch allows selection of one of the various TI COFF
5052 versions. The latest of this writing is 2; versions 0 and 1 are also
5053 supported. The TI COFF versions also vary in header byte-order format;
5054 `ld' will read any version or byte order, but the output header format
5055 depends on the default specified by the specific target.
5058 File: ld.info, Node: WIN32, Next: Xtensa, Prev: TI COFF, Up: Machine Dependent
5060 4.13 `ld' and WIN32 (cygwin/mingw)
5061 ==================================
5063 This section describes some of the win32 specific `ld' issues. See
5064 *note Command Line Options: Options. for detailed description of the
5065 command line options mentioned here.
5068 The standard Windows linker creates and uses so-called import
5069 libraries, which contains information for linking to dll's. They
5070 are regular static archives and are handled as any other static
5071 archive. The cygwin and mingw ports of `ld' have specific support
5072 for creating such libraries provided with the `--out-implib'
5073 command line option.
5075 _exporting DLL symbols_
5076 The cygwin/mingw `ld' has several ways to export symbols for dll's.
5078 _using auto-export functionality_
5079 By default `ld' exports symbols with the auto-export
5080 functionality, which is controlled by the following command
5083 * -export-all-symbols [This is the default]
5089 If, however, `--export-all-symbols' is not given explicitly
5090 on the command line, then the default auto-export behavior
5091 will be _disabled_ if either of the following are true:
5093 * A DEF file is used.
5095 * Any symbol in any object file was marked with the
5096 __declspec(dllexport) attribute.
5099 Another way of exporting symbols is using a DEF file. A DEF
5100 file is an ASCII file containing definitions of symbols which
5101 should be exported when a dll is created. Usually it is
5102 named `<dll name>.def' and is added as any other object file
5103 to the linker's command line. The file's name must end in
5106 gcc -o <output> <objectfiles> <dll name>.def
5108 Using a DEF file turns off the normal auto-export behavior,
5109 unless the `--export-all-symbols' option is also used.
5111 Here is an example of a DEF file for a shared library called
5114 LIBRARY "xyz.dll" BASE=0x20000000
5120 another_foo = abc.dll.afoo
5123 This example defines a DLL with a non-default base address
5124 and five symbols in the export table. The third exported
5125 symbol `_bar' is an alias for the second. The fourth symbol,
5126 `another_foo' is resolved by "forwarding" to another module
5127 and treating it as an alias for `afoo' exported from the DLL
5128 `abc.dll'. The final symbol `var1' is declared to be a data
5131 The optional `LIBRARY <name>' command indicates the _internal_
5132 name of the output DLL. If `<name>' does not include a suffix,
5133 the default library suffix, `.DLL' is appended.
5135 When the .DEF file is used to build an application, rather
5136 than a library, the `NAME <name>' command should be used
5137 instead of `LIBRARY'. If `<name>' does not include a suffix,
5138 the default executable suffix, `.EXE' is appended.
5140 With either `LIBRARY <name>' or `NAME <name>' the optional
5141 specification `BASE = <number>' may be used to specify a
5142 non-default base address for the image.
5144 If neither `LIBRARY <name>' nor `NAME <name>' is specified,
5145 or they specify an empty string, the internal name is the
5146 same as the filename specified on the command line.
5148 The complete specification of an export symbol is:
5151 ( ( ( <name1> [ = <name2> ] )
5152 | ( <name1> = <module-name> . <external-name>))
5153 [ @ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
5155 Declares `<name1>' as an exported symbol from the DLL, or
5156 declares `<name1>' as an exported alias for `<name2>'; or
5157 declares `<name1>' as a "forward" alias for the symbol
5158 `<external-name>' in the DLL `<module-name>'. Optionally,
5159 the symbol may be exported by the specified ordinal
5162 The optional keywords that follow the declaration indicate:
5164 `NONAME': Do not put the symbol name in the DLL's export
5165 table. It will still be exported by its ordinal alias
5166 (either the value specified by the .def specification or,
5167 otherwise, the value assigned by the linker). The symbol
5168 name, however, does remain visible in the import library (if
5169 any), unless `PRIVATE' is also specified.
5171 `DATA': The symbol is a variable or object, rather than a
5172 function. The import lib will export only an indirect
5173 reference to `foo' as the symbol `_imp__foo' (ie, `foo' must
5174 be resolved as `*_imp__foo').
5176 `CONSTANT': Like `DATA', but put the undecorated `foo' as
5177 well as `_imp__foo' into the import library. Both refer to the
5178 read-only import address table's pointer to the variable, not
5179 to the variable itself. This can be dangerous. If the user
5180 code fails to add the `dllimport' attribute and also fails to
5181 explicitly add the extra indirection that the use of the
5182 attribute enforces, the application will behave unexpectedly.
5184 `PRIVATE': Put the symbol in the DLL's export table, but do
5185 not put it into the static import library used to resolve
5186 imports at link time. The symbol can still be imported using
5187 the `LoadLibrary/GetProcAddress' API at runtime or by by
5188 using the GNU ld extension of linking directly to the DLL
5189 without an import library.
5191 See ld/deffilep.y in the binutils sources for the full
5192 specification of other DEF file statements
5194 While linking a shared dll, `ld' is able to create a DEF file
5195 with the `--output-def <file>' command line option.
5198 Another way of marking symbols for export is to modify the
5199 source code itself, so that when building the DLL each symbol
5200 to be exported is declared as:
5202 __declspec(dllexport) int a_variable
5203 __declspec(dllexport) void a_function(int with_args)
5205 All such symbols will be exported from the DLL. If, however,
5206 any of the object files in the DLL contain symbols decorated
5207 in this way, then the normal auto-export behavior is
5208 disabled, unless the `--export-all-symbols' option is also
5211 Note that object files that wish to access these symbols must
5212 _not_ decorate them with dllexport. Instead, they should use
5215 __declspec(dllimport) int a_variable
5216 __declspec(dllimport) void a_function(int with_args)
5218 This complicates the structure of library header files,
5219 because when included by the library itself the header must
5220 declare the variables and functions as dllexport, but when
5221 included by client code the header must declare them as
5222 dllimport. There are a number of idioms that are typically
5223 used to do this; often client code can omit the __declspec()
5224 declaration completely. See `--enable-auto-import' and
5225 `automatic data imports' for more information.
5227 _automatic data imports_
5228 The standard Windows dll format supports data imports from dlls
5229 only by adding special decorations (dllimport/dllexport), which
5230 let the compiler produce specific assembler instructions to deal
5231 with this issue. This increases the effort necessary to port
5232 existing Un*x code to these platforms, especially for large c++
5233 libraries and applications. The auto-import feature, which was
5234 initially provided by Paul Sokolovsky, allows one to omit the
5235 decorations to achieve a behavior that conforms to that on
5236 POSIX/Un*x platforms. This feature is enabled with the
5237 `--enable-auto-import' command-line option, although it is enabled
5238 by default on cygwin/mingw. The `--enable-auto-import' option
5239 itself now serves mainly to suppress any warnings that are
5240 ordinarily emitted when linked objects trigger the feature's use.
5242 auto-import of variables does not always work flawlessly without
5243 additional assistance. Sometimes, you will see this message
5245 "variable '<var>' can't be auto-imported. Please read the
5246 documentation for ld's `--enable-auto-import' for details."
5248 The `--enable-auto-import' documentation explains why this error
5249 occurs, and several methods that can be used to overcome this
5250 difficulty. One of these methods is the _runtime pseudo-relocs_
5251 feature, described below.
5253 For complex variables imported from DLLs (such as structs or
5254 classes), object files typically contain a base address for the
5255 variable and an offset (_addend_) within the variable-to specify a
5256 particular field or public member, for instance. Unfortunately,
5257 the runtime loader used in win32 environments is incapable of
5258 fixing these references at runtime without the additional
5259 information supplied by dllimport/dllexport decorations. The
5260 standard auto-import feature described above is unable to resolve
5263 The `--enable-runtime-pseudo-relocs' switch allows these
5264 references to be resolved without error, while leaving the task of
5265 adjusting the references themselves (with their non-zero addends)
5266 to specialized code provided by the runtime environment. Recent
5267 versions of the cygwin and mingw environments and compilers
5268 provide this runtime support; older versions do not. However, the
5269 support is only necessary on the developer's platform; the
5270 compiled result will run without error on an older system.
5272 `--enable-runtime-pseudo-relocs' is not the default; it must be
5273 explicitly enabled as needed.
5275 _direct linking to a dll_
5276 The cygwin/mingw ports of `ld' support the direct linking,
5277 including data symbols, to a dll without the usage of any import
5278 libraries. This is much faster and uses much less memory than
5279 does the traditional import library method, especially when
5280 linking large libraries or applications. When `ld' creates an
5281 import lib, each function or variable exported from the dll is
5282 stored in its own bfd, even though a single bfd could contain many
5283 exports. The overhead involved in storing, loading, and
5284 processing so many bfd's is quite large, and explains the
5285 tremendous time, memory, and storage needed to link against
5286 particularly large or complex libraries when using import libs.
5288 Linking directly to a dll uses no extra command-line switches
5289 other than `-L' and `-l', because `ld' already searches for a
5290 number of names to match each library. All that is needed from
5291 the developer's perspective is an understanding of this search, in
5292 order to force ld to select the dll instead of an import library.
5294 For instance, when ld is called with the argument `-lxxx' it will
5295 attempt to find, in the first directory of its search path,
5305 before moving on to the next directory in the search path.
5307 (*) Actually, this is not `cygxxx.dll' but in fact is
5308 `<prefix>xxx.dll', where `<prefix>' is set by the `ld' option
5309 `--dll-search-prefix=<prefix>'. In the case of cygwin, the
5310 standard gcc spec file includes `--dll-search-prefix=cyg', so in
5311 effect we actually search for `cygxxx.dll'.
5313 Other win32-based unix environments, such as mingw or pw32, may
5314 use other `<prefix>'es, although at present only cygwin makes use
5315 of this feature. It was originally intended to help avoid name
5316 conflicts among dll's built for the various win32/un*x
5317 environments, so that (for example) two versions of a zlib dll
5318 could coexist on the same machine.
5320 The generic cygwin/mingw path layout uses a `bin' directory for
5321 applications and dll's and a `lib' directory for the import
5322 libraries (using cygwin nomenclature):
5327 libxxx.dll.a (in case of dll's)
5328 libxxx.a (in case of static archive)
5330 Linking directly to a dll without using the import library can be
5333 1. Use the dll directly by adding the `bin' path to the link line
5334 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
5336 However, as the dll's often have version numbers appended to their
5337 names (`cygncurses-5.dll') this will often fail, unless one
5338 specifies `-L../bin -lncurses-5' to include the version. Import
5339 libs are generally not versioned, and do not have this difficulty.
5341 2. Create a symbolic link from the dll to a file in the `lib'
5342 directory according to the above mentioned search pattern. This
5343 should be used to avoid unwanted changes in the tools needed for
5346 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
5348 Then you can link without any make environment changes.
5350 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
5352 This technique also avoids the version number problems, because
5353 the following is perfectly legal
5358 libxxx.dll.a -> ../bin/cygxxx-5.dll
5360 Linking directly to a dll without using an import lib will work
5361 even when auto-import features are exercised, and even when
5362 `--enable-runtime-pseudo-relocs' is used.
5364 Given the improvements in speed and memory usage, one might
5365 justifiably wonder why import libraries are used at all. There
5368 1. Until recently, the link-directly-to-dll functionality did _not_
5369 work with auto-imported data.
5371 2. Sometimes it is necessary to include pure static objects within
5372 the import library (which otherwise contains only bfd's for
5373 indirection symbols that point to the exports of a dll). Again,
5374 the import lib for the cygwin kernel makes use of this ability,
5375 and it is not possible to do this without an import lib.
5377 3. Symbol aliases can only be resolved using an import lib. This
5378 is critical when linking against OS-supplied dll's (eg, the win32
5379 API) in which symbols are usually exported as undecorated aliases
5380 of their stdcall-decorated assembly names.
5382 So, import libs are not going away. But the ability to replace
5383 true import libs with a simple symbolic link to (or a copy of) a
5384 dll, in many cases, is a useful addition to the suite of tools
5385 binutils makes available to the win32 developer. Given the
5386 massive improvements in memory requirements during linking, storage
5387 requirements, and linking speed, we expect that many developers
5388 will soon begin to use this feature whenever possible.
5392 _adding additional names_
5393 Sometimes, it is useful to export symbols with additional
5394 names. A symbol `foo' will be exported as `foo', but it can
5395 also be exported as `_foo' by using special directives in the
5396 DEF file when creating the dll. This will affect also the
5397 optional created import library. Consider the following DEF
5400 LIBRARY "xyz.dll" BASE=0x61000000
5406 The line `_foo = foo' maps the symbol `foo' to `_foo'.
5408 Another method for creating a symbol alias is to create it in
5409 the source code using the "weak" attribute:
5411 void foo () { /* Do something. */; }
5412 void _foo () __attribute__ ((weak, alias ("foo")));
5414 See the gcc manual for more information about attributes and
5418 Sometimes it is useful to rename exports. For instance, the
5419 cygwin kernel does this regularly. A symbol `_foo' can be
5420 exported as `foo' but not as `_foo' by using special
5421 directives in the DEF file. (This will also affect the import
5422 library, if it is created). In the following example:
5424 LIBRARY "xyz.dll" BASE=0x61000000
5429 The line `_foo = foo' maps the exported symbol `foo' to
5432 Note: using a DEF file disables the default auto-export behavior,
5433 unless the `--export-all-symbols' command line option is used.
5434 If, however, you are trying to rename symbols, then you should list
5435 _all_ desired exports in the DEF file, including the symbols that
5436 are not being renamed, and do _not_ use the `--export-all-symbols'
5437 option. If you list only the renamed symbols in the DEF file, and
5438 use `--export-all-symbols' to handle the other symbols, then the
5439 both the new names _and_ the original names for the renamed
5440 symbols will be exported. In effect, you'd be aliasing those
5441 symbols, not renaming them, which is probably not what you wanted.
5444 The Windows object format, PE, specifies a form of weak symbols
5445 called weak externals. When a weak symbol is linked and the
5446 symbol is not defined, the weak symbol becomes an alias for some
5447 other symbol. There are three variants of weak externals:
5448 * Definition is searched for in objects and libraries,
5449 historically called lazy externals.
5451 * Definition is searched for only in other objects, not in
5452 libraries. This form is not presently implemented.
5454 * No search; the symbol is an alias. This form is not presently
5456 As a GNU extension, weak symbols that do not specify an alternate
5457 symbol are supported. If the symbol is undefined when linking,
5458 the symbol uses a default value.
5461 File: ld.info, Node: Xtensa, Prev: WIN32, Up: Machine Dependent
5463 4.14 `ld' and Xtensa Processors
5464 ===============================
5466 The default `ld' behavior for Xtensa processors is to interpret
5467 `SECTIONS' commands so that lists of explicitly named sections in a
5468 specification with a wildcard file will be interleaved when necessary to
5469 keep literal pools within the range of PC-relative load offsets. For
5470 example, with the command:
5479 `ld' may interleave some of the `.literal' and `.text' sections from
5480 different object files to ensure that the literal pools are within the
5481 range of PC-relative load offsets. A valid interleaving might place
5482 the `.literal' sections from an initial group of files followed by the
5483 `.text' sections of that group of files. Then, the `.literal' sections
5484 from the rest of the files and the `.text' sections from the rest of
5485 the files would follow.
5487 Relaxation is enabled by default for the Xtensa version of `ld' and
5488 provides two important link-time optimizations. The first optimization
5489 is to combine identical literal values to reduce code size. A redundant
5490 literal will be removed and all the `L32R' instructions that use it
5491 will be changed to reference an identical literal, as long as the
5492 location of the replacement literal is within the offset range of all
5493 the `L32R' instructions. The second optimization is to remove
5494 unnecessary overhead from assembler-generated "longcall" sequences of
5495 `L32R'/`CALLXN' when the target functions are within range of direct
5496 `CALLN' instructions.
5498 For each of these cases where an indirect call sequence can be
5499 optimized to a direct call, the linker will change the `CALLXN'
5500 instruction to a `CALLN' instruction, remove the `L32R' instruction,
5501 and remove the literal referenced by the `L32R' instruction if it is
5502 not used for anything else. Removing the `L32R' instruction always
5503 reduces code size but can potentially hurt performance by changing the
5504 alignment of subsequent branch targets. By default, the linker will
5505 always preserve alignments, either by switching some instructions
5506 between 24-bit encodings and the equivalent density instructions or by
5507 inserting a no-op in place of the `L32R' instruction that was removed.
5508 If code size is more important than performance, the `--size-opt'
5509 option can be used to prevent the linker from widening density
5510 instructions or inserting no-ops, except in a few cases where no-ops
5511 are required for correctness.
5513 The following Xtensa-specific command-line options can be used to
5517 Since the Xtensa version of `ld' enables the `--relax' option by
5518 default, the `--no-relax' option is provided to disable relaxation.
5521 When optimizing indirect calls to direct calls, optimize for code
5522 size more than performance. With this option, the linker will not
5523 insert no-ops or widen density instructions to preserve branch
5524 target alignment. There may still be some cases where no-ops are
5525 required to preserve the correctness of the code.
5528 File: ld.info, Node: BFD, Next: Reporting Bugs, Prev: Machine Dependent, Up: Top
5533 The linker accesses object and archive files using the BFD libraries.
5534 These libraries allow the linker to use the same routines to operate on
5535 object files whatever the object file format. A different object file
5536 format can be supported simply by creating a new BFD back end and adding
5537 it to the library. To conserve runtime memory, however, the linker and
5538 associated tools are usually configured to support only a subset of the
5539 object file formats available. You can use `objdump -i' (*note
5540 objdump: (binutils.info)objdump.) to list all the formats available for
5543 As with most implementations, BFD is a compromise between several
5544 conflicting requirements. The major factor influencing BFD design was
5545 efficiency: any time used converting between formats is time which
5546 would not have been spent had BFD not been involved. This is partly
5547 offset by abstraction payback; since BFD simplifies applications and
5548 back ends, more time and care may be spent optimizing algorithms for a
5551 One minor artifact of the BFD solution which you should bear in mind
5552 is the potential for information loss. There are two places where
5553 useful information can be lost using the BFD mechanism: during
5554 conversion and during output. *Note BFD information loss::.
5558 * BFD outline:: How it works: an outline of BFD
5561 File: ld.info, Node: BFD outline, Up: BFD
5563 5.1 How It Works: An Outline of BFD
5564 ===================================
5566 When an object file is opened, BFD subroutines automatically determine
5567 the format of the input object file. They then build a descriptor in
5568 memory with pointers to routines that will be used to access elements of
5569 the object file's data structures.
5571 As different information from the object files is required, BFD
5572 reads from different sections of the file and processes them. For
5573 example, a very common operation for the linker is processing symbol
5574 tables. Each BFD back end provides a routine for converting between
5575 the object file's representation of symbols and an internal canonical
5576 format. When the linker asks for the symbol table of an object file, it
5577 calls through a memory pointer to the routine from the relevant BFD
5578 back end which reads and converts the table into a canonical form. The
5579 linker then operates upon the canonical form. When the link is finished
5580 and the linker writes the output file's symbol table, another BFD back
5581 end routine is called to take the newly created symbol table and
5582 convert it into the chosen output format.
5586 * BFD information loss:: Information Loss
5587 * Canonical format:: The BFD canonical object-file format
5590 File: ld.info, Node: BFD information loss, Next: Canonical format, Up: BFD outline
5592 5.1.1 Information Loss
5593 ----------------------
5595 _Information can be lost during output._ The output formats supported
5596 by BFD do not provide identical facilities, and information which can
5597 be described in one form has nowhere to go in another format. One
5598 example of this is alignment information in `b.out'. There is nowhere
5599 in an `a.out' format file to store alignment information on the
5600 contained data, so when a file is linked from `b.out' and an `a.out'
5601 image is produced, alignment information will not propagate to the
5602 output file. (The linker will still use the alignment information
5603 internally, so the link is performed correctly).
5605 Another example is COFF section names. COFF files may contain an
5606 unlimited number of sections, each one with a textual section name. If
5607 the target of the link is a format which does not have many sections
5608 (e.g., `a.out') or has sections without names (e.g., the Oasys format),
5609 the link cannot be done simply. You can circumvent this problem by
5610 describing the desired input-to-output section mapping with the linker
5613 _Information can be lost during canonicalization._ The BFD internal
5614 canonical form of the external formats is not exhaustive; there are
5615 structures in input formats for which there is no direct representation
5616 internally. This means that the BFD back ends cannot maintain all
5617 possible data richness through the transformation between external to
5618 internal and back to external formats.
5620 This limitation is only a problem when an application reads one
5621 format and writes another. Each BFD back end is responsible for
5622 maintaining as much data as possible, and the internal BFD canonical
5623 form has structures which are opaque to the BFD core, and exported only
5624 to the back ends. When a file is read in one format, the canonical form
5625 is generated for BFD and the application. At the same time, the back
5626 end saves away any information which may otherwise be lost. If the data
5627 is then written back in the same format, the back end routine will be
5628 able to use the canonical form provided by the BFD core as well as the
5629 information it prepared earlier. Since there is a great deal of
5630 commonality between back ends, there is no information lost when
5631 linking or copying big endian COFF to little endian COFF, or `a.out' to
5632 `b.out'. When a mixture of formats is linked, the information is only
5633 lost from the files whose format differs from the destination.
5636 File: ld.info, Node: Canonical format, Prev: BFD information loss, Up: BFD outline
5638 5.1.2 The BFD canonical object-file format
5639 ------------------------------------------
5641 The greatest potential for loss of information occurs when there is the
5642 least overlap between the information provided by the source format,
5643 that stored by the canonical format, and that needed by the destination
5644 format. A brief description of the canonical form may help you
5645 understand which kinds of data you can count on preserving across
5649 Information stored on a per-file basis includes target machine
5650 architecture, particular implementation format type, a demand
5651 pageable bit, and a write protected bit. Information like Unix
5652 magic numbers is not stored here--only the magic numbers' meaning,
5653 so a `ZMAGIC' file would have both the demand pageable bit and the
5654 write protected text bit set. The byte order of the target is
5655 stored on a per-file basis, so that big- and little-endian object
5656 files may be used with one another.
5659 Each section in the input file contains the name of the section,
5660 the section's original address in the object file, size and
5661 alignment information, various flags, and pointers into other BFD
5665 Each symbol contains a pointer to the information for the object
5666 file which originally defined it, its name, its value, and various
5667 flag bits. When a BFD back end reads in a symbol table, it
5668 relocates all symbols to make them relative to the base of the
5669 section where they were defined. Doing this ensures that each
5670 symbol points to its containing section. Each symbol also has a
5671 varying amount of hidden private data for the BFD back end. Since
5672 the symbol points to the original file, the private data format
5673 for that symbol is accessible. `ld' can operate on a collection
5674 of symbols of wildly different formats without problems.
5676 Normal global and simple local symbols are maintained on output,
5677 so an output file (no matter its format) will retain symbols
5678 pointing to functions and to global, static, and common variables.
5679 Some symbol information is not worth retaining; in `a.out', type
5680 information is stored in the symbol table as long symbol names.
5681 This information would be useless to most COFF debuggers; the
5682 linker has command line switches to allow users to throw it away.
5684 There is one word of type information within the symbol, so if the
5685 format supports symbol type information within symbols (for
5686 example, COFF, IEEE, Oasys) and the type is simple enough to fit
5687 within one word (nearly everything but aggregates), the
5688 information will be preserved.
5691 Each canonical BFD relocation record contains a pointer to the
5692 symbol to relocate to, the offset of the data to relocate, the
5693 section the data is in, and a pointer to a relocation type
5694 descriptor. Relocation is performed by passing messages through
5695 the relocation type descriptor and the symbol pointer. Therefore,
5696 relocations can be performed on output data using a relocation
5697 method that is only available in one of the input formats. For
5698 instance, Oasys provides a byte relocation format. A relocation
5699 record requesting this relocation type would point indirectly to a
5700 routine to perform this, so the relocation may be performed on a
5701 byte being written to a 68k COFF file, even though 68k COFF has no
5702 such relocation type.
5705 Object formats can contain, for debugging purposes, some form of
5706 mapping between symbols, source line numbers, and addresses in the
5707 output file. These addresses have to be relocated along with the
5708 symbol information. Each symbol with an associated list of line
5709 number records points to the first record of the list. The head
5710 of a line number list consists of a pointer to the symbol, which
5711 allows finding out the address of the function whose line number
5712 is being described. The rest of the list is made up of pairs:
5713 offsets into the section and line numbers. Any format which can
5714 simply derive this information can pass it successfully between
5715 formats (COFF, IEEE and Oasys).
5718 File: ld.info, Node: Reporting Bugs, Next: MRI, Prev: BFD, Up: Top
5723 Your bug reports play an essential role in making `ld' reliable.
5725 Reporting a bug may help you by bringing a solution to your problem,
5726 or it may not. But in any case the principal function of a bug report
5727 is to help the entire community by making the next version of `ld' work
5728 better. Bug reports are your contribution to the maintenance of `ld'.
5730 In order for a bug report to serve its purpose, you must include the
5731 information that enables us to fix the bug.
5735 * Bug Criteria:: Have you found a bug?
5736 * Bug Reporting:: How to report bugs
5739 File: ld.info, Node: Bug Criteria, Next: Bug Reporting, Up: Reporting Bugs
5741 6.1 Have You Found a Bug?
5742 =========================
5744 If you are not sure whether you have found a bug, here are some
5747 * If the linker gets a fatal signal, for any input whatever, that is
5748 a `ld' bug. Reliable linkers never crash.
5750 * If `ld' produces an error message for valid input, that is a bug.
5752 * If `ld' does not produce an error message for invalid input, that
5753 may be a bug. In the general case, the linker can not verify that
5754 object files are correct.
5756 * If you are an experienced user of linkers, your suggestions for
5757 improvement of `ld' are welcome in any case.
5760 File: ld.info, Node: Bug Reporting, Prev: Bug Criteria, Up: Reporting Bugs
5762 6.2 How to Report Bugs
5763 ======================
5765 A number of companies and individuals offer support for GNU products.
5766 If you obtained `ld' from a support organization, we recommend you
5767 contact that organization first.
5769 You can find contact information for many support companies and
5770 individuals in the file `etc/SERVICE' in the GNU Emacs distribution.
5772 Otherwise, send bug reports for `ld' to
5773 `http://www.sourceware.org/bugzilla/'.
5775 The fundamental principle of reporting bugs usefully is this:
5776 *report all the facts*. If you are not sure whether to state a fact or
5777 leave it out, state it!
5779 Often people omit facts because they think they know what causes the
5780 problem and assume that some details do not matter. Thus, you might
5781 assume that the name of a symbol you use in an example does not matter.
5782 Well, probably it does not, but one cannot be sure. Perhaps the bug is
5783 a stray memory reference which happens to fetch from the location where
5784 that name is stored in memory; perhaps, if the name were different, the
5785 contents of that location would fool the linker into doing the right
5786 thing despite the bug. Play it safe and give a specific, complete
5787 example. That is the easiest thing for you to do, and the most helpful.
5789 Keep in mind that the purpose of a bug report is to enable us to fix
5790 the bug if it is new to us. Therefore, always write your bug reports
5791 on the assumption that the bug has not been reported previously.
5793 Sometimes people give a few sketchy facts and ask, "Does this ring a
5794 bell?" This cannot help us fix a bug, so it is basically useless. We
5795 respond by asking for enough details to enable us to investigate. You
5796 might as well expedite matters by sending them to begin with.
5798 To enable us to fix the bug, you should include all these things:
5800 * The version of `ld'. `ld' announces it if you start it with the
5801 `--version' argument.
5803 Without this, we will not know whether there is any point in
5804 looking for the bug in the current version of `ld'.
5806 * Any patches you may have applied to the `ld' source, including any
5807 patches made to the `BFD' library.
5809 * The type of machine you are using, and the operating system name
5812 * What compiler (and its version) was used to compile `ld'--e.g.
5815 * The command arguments you gave the linker to link your example and
5816 observe the bug. To guarantee you will not omit something
5817 important, list them all. A copy of the Makefile (or the output
5818 from make) is sufficient.
5820 If we were to try to guess the arguments, we would probably guess
5821 wrong and then we might not encounter the bug.
5823 * A complete input file, or set of input files, that will reproduce
5824 the bug. It is generally most helpful to send the actual object
5825 files provided that they are reasonably small. Say no more than
5826 10K. For bigger files you can either make them available by FTP
5827 or HTTP or else state that you are willing to send the object
5828 file(s) to whomever requests them. (Note - your email will be
5829 going to a mailing list, so we do not want to clog it up with
5830 large attachments). But small attachments are best.
5832 If the source files were assembled using `gas' or compiled using
5833 `gcc', then it may be OK to send the source files rather than the
5834 object files. In this case, be sure to say exactly what version of
5835 `gas' or `gcc' was used to produce the object files. Also say how
5836 `gas' or `gcc' were configured.
5838 * A description of what behavior you observe that you believe is
5839 incorrect. For example, "It gets a fatal signal."
5841 Of course, if the bug is that `ld' gets a fatal signal, then we
5842 will certainly notice it. But if the bug is incorrect output, we
5843 might not notice unless it is glaringly wrong. You might as well
5844 not give us a chance to make a mistake.
5846 Even if the problem you experience is a fatal signal, you should
5847 still say so explicitly. Suppose something strange is going on,
5848 such as, your copy of `ld' is out of sync, or you have encountered
5849 a bug in the C library on your system. (This has happened!) Your
5850 copy might crash and ours would not. If you told us to expect a
5851 crash, then when ours fails to crash, we would know that the bug
5852 was not happening for us. If you had not told us to expect a
5853 crash, then we would not be able to draw any conclusion from our
5856 * If you wish to suggest changes to the `ld' source, send us context
5857 diffs, as generated by `diff' with the `-u', `-c', or `-p' option.
5858 Always send diffs from the old file to the new file. If you even
5859 discuss something in the `ld' source, refer to it by context, not
5862 The line numbers in our development sources will not match those
5863 in your sources. Your line numbers would convey no useful
5866 Here are some things that are not necessary:
5868 * A description of the envelope of the bug.
5870 Often people who encounter a bug spend a lot of time investigating
5871 which changes to the input file will make the bug go away and which
5872 changes will not affect it.
5874 This is often time consuming and not very useful, because the way
5875 we will find the bug is by running a single example under the
5876 debugger with breakpoints, not by pure deduction from a series of
5877 examples. We recommend that you save your time for something else.
5879 Of course, if you can find a simpler example to report _instead_
5880 of the original one, that is a convenience for us. Errors in the
5881 output will be easier to spot, running under the debugger will take
5882 less time, and so on.
5884 However, simplification is not vital; if you do not want to do
5885 this, report the bug anyway and send us the entire test case you
5888 * A patch for the bug.
5890 A patch for the bug does help us if it is a good one. But do not
5891 omit the necessary information, such as the test case, on the
5892 assumption that a patch is all we need. We might see problems
5893 with your patch and decide to fix the problem another way, or we
5894 might not understand it at all.
5896 Sometimes with a program as complicated as `ld' it is very hard to
5897 construct an example that will make the program follow a certain
5898 path through the code. If you do not send us the example, we will
5899 not be able to construct one, so we will not be able to verify
5900 that the bug is fixed.
5902 And if we cannot understand what bug you are trying to fix, or why
5903 your patch should be an improvement, we will not install it. A
5904 test case will help us to understand.
5906 * A guess about what the bug is or what it depends on.
5908 Such guesses are usually wrong. Even we cannot guess right about
5909 such things without first using the debugger to find the facts.
5912 File: ld.info, Node: MRI, Next: GNU Free Documentation License, Prev: Reporting Bugs, Up: Top
5914 Appendix A MRI Compatible Script Files
5915 **************************************
5917 To aid users making the transition to GNU `ld' from the MRI linker,
5918 `ld' can use MRI compatible linker scripts as an alternative to the
5919 more general-purpose linker scripting language described in *note
5920 Scripts::. MRI compatible linker scripts have a much simpler command
5921 set than the scripting language otherwise used with `ld'. GNU `ld'
5922 supports the most commonly used MRI linker commands; these commands are
5925 In general, MRI scripts aren't of much use with the `a.out' object
5926 file format, since it only has three sections and MRI scripts lack some
5927 features to make use of them.
5929 You can specify a file containing an MRI-compatible script using the
5930 `-c' command-line option.
5932 Each command in an MRI-compatible script occupies its own line; each
5933 command line starts with the keyword that identifies the command (though
5934 blank lines are also allowed for punctuation). If a line of an
5935 MRI-compatible script begins with an unrecognized keyword, `ld' issues
5936 a warning message, but continues processing the script.
5938 Lines beginning with `*' are comments.
5940 You can write these commands using all upper-case letters, or all
5941 lower case; for example, `chip' is the same as `CHIP'. The following
5942 list shows only the upper-case form of each command.
5945 `ABSOLUTE SECNAME, SECNAME, ... SECNAME'
5946 Normally, `ld' includes in the output file all sections from all
5947 the input files. However, in an MRI-compatible script, you can
5948 use the `ABSOLUTE' command to restrict the sections that will be
5949 present in your output program. If the `ABSOLUTE' command is used
5950 at all in a script, then only the sections named explicitly in
5951 `ABSOLUTE' commands will appear in the linker output. You can
5952 still use other input sections (whatever you select on the command
5953 line, or using `LOAD') to resolve addresses in the output file.
5955 `ALIAS OUT-SECNAME, IN-SECNAME'
5956 Use this command to place the data from input section IN-SECNAME
5957 in a section called OUT-SECNAME in the linker output file.
5959 IN-SECNAME may be an integer.
5961 `ALIGN SECNAME = EXPRESSION'
5962 Align the section called SECNAME to EXPRESSION. The EXPRESSION
5963 should be a power of two.
5966 Use the value of EXPRESSION as the lowest address (other than
5967 absolute addresses) in the output file.
5970 `CHIP EXPRESSION, EXPRESSION'
5971 This command does nothing; it is accepted only for compatibility.
5974 This command does nothing whatever; it's only accepted for
5977 `FORMAT OUTPUT-FORMAT'
5978 Similar to the `OUTPUT_FORMAT' command in the more general linker
5979 language, but restricted to one of these output formats:
5981 1. S-records, if OUTPUT-FORMAT is `S'
5983 2. IEEE, if OUTPUT-FORMAT is `IEEE'
5985 3. COFF (the `coff-m68k' variant in BFD), if OUTPUT-FORMAT is
5989 Print (to the standard output file) a link map, as produced by the
5990 `ld' command-line option `-M'.
5992 The keyword `LIST' may be followed by anything on the same line,
5993 with no change in its effect.
5996 `LOAD FILENAME, FILENAME, ... FILENAME'
5997 Include one or more object file FILENAME in the link; this has the
5998 same effect as specifying FILENAME directly on the `ld' command
6002 OUTPUT-NAME is the name for the program produced by `ld'; the
6003 MRI-compatible command `NAME' is equivalent to the command-line
6004 option `-o' or the general script language command `OUTPUT'.
6006 `ORDER SECNAME, SECNAME, ... SECNAME'
6007 `ORDER SECNAME SECNAME SECNAME'
6008 Normally, `ld' orders the sections in its output file in the order
6009 in which they first appear in the input files. In an
6010 MRI-compatible script, you can override this ordering with the
6011 `ORDER' command. The sections you list with `ORDER' will appear
6012 first in your output file, in the order specified.
6014 `PUBLIC NAME=EXPRESSION'
6015 `PUBLIC NAME,EXPRESSION'
6016 `PUBLIC NAME EXPRESSION'
6017 Supply a value (EXPRESSION) for external symbol NAME used in the
6020 `SECT SECNAME, EXPRESSION'
6021 `SECT SECNAME=EXPRESSION'
6022 `SECT SECNAME EXPRESSION'
6023 You can use any of these three forms of the `SECT' command to
6024 specify the start address (EXPRESSION) for section SECNAME. If
6025 you have more than one `SECT' statement for the same SECNAME, only
6026 the _first_ sets the start address.
6029 File: ld.info, Node: GNU Free Documentation License, Next: LD Index, Prev: MRI, Up: Top
6031 Appendix B GNU Free Documentation License
6032 *****************************************
6034 Version 1.1, March 2000
6036 Copyright (C) 2000, 2003 Free Software Foundation, Inc.
6037 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
6039 Everyone is permitted to copy and distribute verbatim copies
6040 of this license document, but changing it is not allowed.
6045 The purpose of this License is to make a manual, textbook, or other
6046 written document "free" in the sense of freedom: to assure everyone
6047 the effective freedom to copy and redistribute it, with or without
6048 modifying it, either commercially or noncommercially. Secondarily,
6049 this License preserves for the author and publisher a way to get
6050 credit for their work, while not being considered responsible for
6051 modifications made by others.
6053 This License is a kind of "copyleft", which means that derivative
6054 works of the document must themselves be free in the same sense.
6055 It complements the GNU General Public License, which is a copyleft
6056 license designed for free software.
6058 We have designed this License in order to use it for manuals for
6059 free software, because free software needs free documentation: a
6060 free program should come with manuals providing the same freedoms
6061 that the software does. But this License is not limited to
6062 software manuals; it can be used for any textual work, regardless
6063 of subject matter or whether it is published as a printed book.
6064 We recommend this License principally for works whose purpose is
6065 instruction or reference.
6068 1. APPLICABILITY AND DEFINITIONS
6070 This License applies to any manual or other work that contains a
6071 notice placed by the copyright holder saying it can be distributed
6072 under the terms of this License. The "Document", below, refers to
6073 any such manual or work. Any member of the public is a licensee,
6074 and is addressed as "you."
6076 A "Modified Version" of the Document means any work containing the
6077 Document or a portion of it, either copied verbatim, or with
6078 modifications and/or translated into another language.
6080 A "Secondary Section" is a named appendix or a front-matter
6081 section of the Document that deals exclusively with the
6082 relationship of the publishers or authors of the Document to the
6083 Document's overall subject (or to related matters) and contains
6084 nothing that could fall directly within that overall subject.
6085 (For example, if the Document is in part a textbook of
6086 mathematics, a Secondary Section may not explain any mathematics.)
6087 The relationship could be a matter of historical connection with
6088 the subject or with related matters, or of legal, commercial,
6089 philosophical, ethical or political position regarding them.
6091 The "Invariant Sections" are certain Secondary Sections whose
6092 titles are designated, as being those of Invariant Sections, in
6093 the notice that says that the Document is released under this
6096 The "Cover Texts" are certain short passages of text that are
6097 listed, as Front-Cover Texts or Back-Cover Texts, in the notice
6098 that says that the Document is released under this License.
6100 A "Transparent" copy of the Document means a machine-readable copy,
6101 represented in a format whose specification is available to the
6102 general public, whose contents can be viewed and edited directly
6103 and straightforwardly with generic text editors or (for images
6104 composed of pixels) generic paint programs or (for drawings) some
6105 widely available drawing editor, and that is suitable for input to
6106 text formatters or for automatic translation to a variety of
6107 formats suitable for input to text formatters. A copy made in an
6108 otherwise Transparent file format whose markup has been designed
6109 to thwart or discourage subsequent modification by readers is not
6110 Transparent. A copy that is not "Transparent" is called "Opaque."
6112 Examples of suitable formats for Transparent copies include plain
6113 ASCII without markup, Texinfo input format, LaTeX input format,
6114 SGML or XML using a publicly available DTD, and
6115 standard-conforming simple HTML designed for human modification.
6116 Opaque formats include PostScript, PDF, proprietary formats that
6117 can be read and edited only by proprietary word processors, SGML
6118 or XML for which the DTD and/or processing tools are not generally
6119 available, and the machine-generated HTML produced by some word
6120 processors for output purposes only.
6122 The "Title Page" means, for a printed book, the title page itself,
6123 plus such following pages as are needed to hold, legibly, the
6124 material this License requires to appear in the title page. For
6125 works in formats which do not have any title page as such, "Title
6126 Page" means the text near the most prominent appearance of the
6127 work's title, preceding the beginning of the body of the text.
6131 You may copy and distribute the Document in any medium, either
6132 commercially or noncommercially, provided that this License, the
6133 copyright notices, and the license notice saying this License
6134 applies to the Document are reproduced in all copies, and that you
6135 add no other conditions whatsoever to those of this License. You
6136 may not use technical measures to obstruct or control the reading
6137 or further copying of the copies you make or distribute. However,
6138 you may accept compensation in exchange for copies. If you
6139 distribute a large enough number of copies you must also follow
6140 the conditions in section 3.
6142 You may also lend copies, under the same conditions stated above,
6143 and you may publicly display copies.
6145 3. COPYING IN QUANTITY
6147 If you publish printed copies of the Document numbering more than
6148 100, and the Document's license notice requires Cover Texts, you
6149 must enclose the copies in covers that carry, clearly and legibly,
6150 all these Cover Texts: Front-Cover Texts on the front cover, and
6151 Back-Cover Texts on the back cover. Both covers must also clearly
6152 and legibly identify you as the publisher of these copies. The
6153 front cover must present the full title with all words of the
6154 title equally prominent and visible. You may add other material
6155 on the covers in addition. Copying with changes limited to the
6156 covers, as long as they preserve the title of the Document and
6157 satisfy these conditions, can be treated as verbatim copying in
6160 If the required texts for either cover are too voluminous to fit
6161 legibly, you should put the first ones listed (as many as fit
6162 reasonably) on the actual cover, and continue the rest onto
6165 If you publish or distribute Opaque copies of the Document
6166 numbering more than 100, you must either include a
6167 machine-readable Transparent copy along with each Opaque copy, or
6168 state in or with each Opaque copy a publicly-accessible
6169 computer-network location containing a complete Transparent copy
6170 of the Document, free of added material, which the general
6171 network-using public has access to download anonymously at no
6172 charge using public-standard network protocols. If you use the
6173 latter option, you must take reasonably prudent steps, when you
6174 begin distribution of Opaque copies in quantity, to ensure that
6175 this Transparent copy will remain thus accessible at the stated
6176 location until at least one year after the last time you
6177 distribute an Opaque copy (directly or through your agents or
6178 retailers) of that edition to the public.
6180 It is requested, but not required, that you contact the authors of
6181 the Document well before redistributing any large number of
6182 copies, to give them a chance to provide you with an updated
6183 version of the Document.
6187 You may copy and distribute a Modified Version of the Document
6188 under the conditions of sections 2 and 3 above, provided that you
6189 release the Modified Version under precisely this License, with
6190 the Modified Version filling the role of the Document, thus
6191 licensing distribution and modification of the Modified Version to
6192 whoever possesses a copy of it. In addition, you must do these
6193 things in the Modified Version:
6195 A. Use in the Title Page (and on the covers, if any) a title
6196 distinct from that of the Document, and from those of previous
6197 versions (which should, if there were any, be listed in the
6198 History section of the Document). You may use the same title
6199 as a previous version if the original publisher of that version
6201 B. List on the Title Page, as authors, one or more persons or
6202 entities responsible for authorship of the modifications in the
6203 Modified Version, together with at least five of the principal
6204 authors of the Document (all of its principal authors, if it
6205 has less than five).
6206 C. State on the Title page the name of the publisher of the
6207 Modified Version, as the publisher.
6208 D. Preserve all the copyright notices of the Document.
6209 E. Add an appropriate copyright notice for your modifications
6210 adjacent to the other copyright notices.
6211 F. Include, immediately after the copyright notices, a license
6212 notice giving the public permission to use the Modified Version
6213 under the terms of this License, in the form shown in the
6215 G. Preserve in that license notice the full lists of Invariant
6216 Sections and required Cover Texts given in the Document's
6218 H. Include an unaltered copy of this License.
6219 I. Preserve the section entitled "History", and its title, and add
6220 to it an item stating at least the title, year, new authors, and
6221 publisher of the Modified Version as given on the Title Page. If
6222 there is no section entitled "History" in the Document, create
6223 one stating the title, year, authors, and publisher of the
6224 Document as given on its Title Page, then add an item
6225 describing the Modified Version as stated in the previous
6227 J. Preserve the network location, if any, given in the Document for
6228 public access to a Transparent copy of the Document, and likewise
6229 the network locations given in the Document for previous versions
6230 it was based on. These may be placed in the "History" section.
6231 You may omit a network location for a work that was published at
6232 least four years before the Document itself, or if the original
6233 publisher of the version it refers to gives permission.
6234 K. In any section entitled "Acknowledgements" or "Dedications",
6235 preserve the section's title, and preserve in the section all the
6236 substance and tone of each of the contributor acknowledgements
6237 and/or dedications given therein.
6238 L. Preserve all the Invariant Sections of the Document,
6239 unaltered in their text and in their titles. Section numbers
6240 or the equivalent are not considered part of the section titles.
6241 M. Delete any section entitled "Endorsements." Such a section
6242 may not be included in the Modified Version.
6243 N. Do not retitle any existing section as "Endorsements" or to
6244 conflict in title with any Invariant Section.
6246 If the Modified Version includes new front-matter sections or
6247 appendices that qualify as Secondary Sections and contain no
6248 material copied from the Document, you may at your option
6249 designate some or all of these sections as invariant. To do this,
6250 add their titles to the list of Invariant Sections in the Modified
6251 Version's license notice. These titles must be distinct from any
6252 other section titles.
6254 You may add a section entitled "Endorsements", provided it contains
6255 nothing but endorsements of your Modified Version by various
6256 parties-for example, statements of peer review or that the text has
6257 been approved by an organization as the authoritative definition
6260 You may add a passage of up to five words as a Front-Cover Text,
6261 and a passage of up to 25 words as a Back-Cover Text, to the end
6262 of the list of Cover Texts in the Modified Version. Only one
6263 passage of Front-Cover Text and one of Back-Cover Text may be
6264 added by (or through arrangements made by) any one entity. If the
6265 Document already includes a cover text for the same cover,
6266 previously added by you or by arrangement made by the same entity
6267 you are acting on behalf of, you may not add another; but you may
6268 replace the old one, on explicit permission from the previous
6269 publisher that added the old one.
6271 The author(s) and publisher(s) of the Document do not by this
6272 License give permission to use their names for publicity for or to
6273 assert or imply endorsement of any Modified Version.
6275 5. COMBINING DOCUMENTS
6277 You may combine the Document with other documents released under
6278 this License, under the terms defined in section 4 above for
6279 modified versions, provided that you include in the combination
6280 all of the Invariant Sections of all of the original documents,
6281 unmodified, and list them all as Invariant Sections of your
6282 combined work in its license notice.
6284 The combined work need only contain one copy of this License, and
6285 multiple identical Invariant Sections may be replaced with a single
6286 copy. If there are multiple Invariant Sections with the same name
6287 but different contents, make the title of each such section unique
6288 by adding at the end of it, in parentheses, the name of the
6289 original author or publisher of that section if known, or else a
6290 unique number. Make the same adjustment to the section titles in
6291 the list of Invariant Sections in the license notice of the
6294 In the combination, you must combine any sections entitled
6295 "History" in the various original documents, forming one section
6296 entitled "History"; likewise combine any sections entitled
6297 "Acknowledgements", and any sections entitled "Dedications." You
6298 must delete all sections entitled "Endorsements."
6300 6. COLLECTIONS OF DOCUMENTS
6302 You may make a collection consisting of the Document and other
6303 documents released under this License, and replace the individual
6304 copies of this License in the various documents with a single copy
6305 that is included in the collection, provided that you follow the
6306 rules of this License for verbatim copying of each of the
6307 documents in all other respects.
6309 You may extract a single document from such a collection, and
6310 distribute it individually under this License, provided you insert
6311 a copy of this License into the extracted document, and follow
6312 this License in all other respects regarding verbatim copying of
6315 7. AGGREGATION WITH INDEPENDENT WORKS
6317 A compilation of the Document or its derivatives with other
6318 separate and independent documents or works, in or on a volume of
6319 a storage or distribution medium, does not as a whole count as a
6320 Modified Version of the Document, provided no compilation
6321 copyright is claimed for the compilation. Such a compilation is
6322 called an "aggregate", and this License does not apply to the
6323 other self-contained works thus compiled with the Document, on
6324 account of their being thus compiled, if they are not themselves
6325 derivative works of the Document.
6327 If the Cover Text requirement of section 3 is applicable to these
6328 copies of the Document, then if the Document is less than one
6329 quarter of the entire aggregate, the Document's Cover Texts may be
6330 placed on covers that surround only the Document within the
6331 aggregate. Otherwise they must appear on covers around the whole
6336 Translation is considered a kind of modification, so you may
6337 distribute translations of the Document under the terms of section
6338 4. Replacing Invariant Sections with translations requires special
6339 permission from their copyright holders, but you may include
6340 translations of some or all Invariant Sections in addition to the
6341 original versions of these Invariant Sections. You may include a
6342 translation of this License provided that you also include the
6343 original English version of this License. In case of a
6344 disagreement between the translation and the original English
6345 version of this License, the original English version will prevail.
6349 You may not copy, modify, sublicense, or distribute the Document
6350 except as expressly provided for under this License. Any other
6351 attempt to copy, modify, sublicense or distribute the Document is
6352 void, and will automatically terminate your rights under this
6353 License. However, parties who have received copies, or rights,
6354 from you under this License will not have their licenses
6355 terminated so long as such parties remain in full compliance.
6357 10. FUTURE REVISIONS OF THIS LICENSE
6359 The Free Software Foundation may publish new, revised versions of
6360 the GNU Free Documentation License from time to time. Such new
6361 versions will be similar in spirit to the present version, but may
6362 differ in detail to address new problems or concerns. See
6363 http://www.gnu.org/copyleft/.
6365 Each version of the License is given a distinguishing version
6366 number. If the Document specifies that a particular numbered
6367 version of this License "or any later version" applies to it, you
6368 have the option of following the terms and conditions either of
6369 that specified version or of any later version that has been
6370 published (not as a draft) by the Free Software Foundation. If
6371 the Document does not specify a version number of this License,
6372 you may choose any version ever published (not as a draft) by the
6373 Free Software Foundation.
6376 ADDENDUM: How to use this License for your documents
6377 ====================================================
6379 To use this License in a document you have written, include a copy of
6380 the License in the document and put the following copyright and license
6381 notices just after the title page:
6383 Copyright (C) YEAR YOUR NAME.
6384 Permission is granted to copy, distribute and/or modify this document
6385 under the terms of the GNU Free Documentation License, Version 1.1
6386 or any later version published by the Free Software Foundation;
6387 with the Invariant Sections being LIST THEIR TITLES, with the
6388 Front-Cover Texts being LIST, and with the Back-Cover Texts being LIST.
6389 A copy of the license is included in the section entitled "GNU
6390 Free Documentation License."
6392 If you have no Invariant Sections, write "with no Invariant Sections"
6393 instead of saying which ones are invariant. If you have no Front-Cover
6394 Texts, write "no Front-Cover Texts" instead of "Front-Cover Texts being
6395 LIST"; likewise for Back-Cover Texts.
6397 If your document contains nontrivial examples of program code, we
6398 recommend releasing these examples in parallel under your choice of
6399 free software license, such as the GNU General Public License, to
6400 permit their use in free software.
6403 File: ld.info, Node: LD Index, Prev: GNU Free Documentation License, Up: Top
6411 * ": Symbols. (line 6)
6412 * -(: Options. (line 645)
6413 * --accept-unknown-input-arch: Options. (line 663)
6414 * --add-needed: Options. (line 685)
6415 * --add-stdcall-alias: Options. (line 1470)
6416 * --allow-multiple-definition: Options. (line 910)
6417 * --allow-shlib-undefined: Options. (line 916)
6418 * --architecture=ARCH: Options. (line 106)
6419 * --as-needed: Options. (line 673)
6420 * --auxiliary: Options. (line 207)
6421 * --bank-window: Options. (line 1815)
6422 * --base-file: Options. (line 1475)
6423 * --be8: ARM. (line 23)
6424 * --bss-plt: PowerPC ELF32. (line 13)
6425 * --build-id: Options. (line 1432)
6426 * --build-id=STYLE: Options. (line 1432)
6427 * --check-sections: Options. (line 767)
6428 * --cref: Options. (line 777)
6429 * --default-imported-symver: Options. (line 944)
6430 * --default-script=SCRIPT: Options. (line 490)
6431 * --default-symver: Options. (line 940)
6432 * --defsym SYMBOL=EXP: Options. (line 805)
6433 * --demangle[=STYLE]: Options. (line 818)
6434 * --disable-auto-image-base: Options. (line 1622)
6435 * --disable-auto-import: Options. (line 1757)
6436 * --disable-new-dtags: Options. (line 1395)
6437 * --disable-runtime-pseudo-reloc: Options. (line 1770)
6438 * --disable-stdcall-fixup: Options. (line 1485)
6439 * --discard-all: Options. (line 536)
6440 * --discard-locals: Options. (line 540)
6441 * --dll: Options. (line 1480)
6442 * --dll-search-prefix: Options. (line 1628)
6443 * --dotsyms: PowerPC64 ELF64. (line 33)
6444 * --dynamic-linker FILE: Options. (line 831)
6445 * --dynamic-list-cpp-new: Options. (line 759)
6446 * --dynamic-list-cpp-typeinfo: Options. (line 763)
6447 * --dynamic-list-data: Options. (line 756)
6448 * --dynamic-list=DYNAMIC-LIST-FILE: Options. (line 743)
6449 * --eh-frame-hdr: Options. (line 1391)
6450 * --emit-relocs: Options. (line 425)
6451 * --emit-stack-syms: SPU ELF. (line 46)
6452 * --emit-stub-syms <1>: SPU ELF. (line 15)
6453 * --emit-stub-syms <2>: PowerPC64 ELF64. (line 29)
6454 * --emit-stub-syms: PowerPC ELF32. (line 44)
6455 * --enable-auto-image-base: Options. (line 1614)
6456 * --enable-auto-import: Options. (line 1637)
6457 * --enable-extra-pe-debug: Options. (line 1775)
6458 * --enable-new-dtags: Options. (line 1395)
6459 * --enable-runtime-pseudo-reloc: Options. (line 1762)
6460 * --enable-stdcall-fixup: Options. (line 1485)
6461 * --entry=ENTRY: Options. (line 160)
6462 * --error-unresolved-symbols: Options. (line 1344)
6463 * --exclude-libs: Options. (line 170)
6464 * --exclude-symbols: Options. (line 1527)
6465 * --export-all-symbols: Options. (line 1503)
6466 * --export-dynamic: Options. (line 181)
6467 * --extra-overlay-stubs: SPU ELF. (line 19)
6468 * --fatal-warnings: Options. (line 837)
6469 * --file-alignment: Options. (line 1533)
6470 * --filter: Options. (line 228)
6471 * --fix-cortex-a8: i960. (line 39)
6472 * --fix-v4bx: ARM. (line 44)
6473 * --fix-v4bx-interworking: ARM. (line 57)
6474 * --force-dynamic: Options. (line 434)
6475 * --force-exe-suffix: Options. (line 842)
6476 * --format=FORMAT: Options. (line 117)
6477 * --format=VERSION: TI COFF. (line 6)
6478 * --gc-sections: Options. (line 852)
6479 * --got: Options. (line 1828)
6480 * --got=TYPE: M68K. (line 6)
6481 * --gpsize: Options. (line 261)
6482 * --hash-size=NUMBER: Options. (line 1404)
6483 * --hash-style=STYLE: Options. (line 1412)
6484 * --heap: Options. (line 1539)
6485 * --help: Options. (line 883)
6486 * --image-base: Options. (line 1546)
6487 * --just-symbols=FILE: Options. (line 457)
6488 * --kill-at: Options. (line 1555)
6489 * --large-address-aware: Options. (line 1560)
6490 * --library-path=DIR: Options. (line 320)
6491 * --library=NAMESPEC: Options. (line 287)
6492 * --local-store=lo:hi: SPU ELF. (line 24)
6493 * --major-image-version: Options. (line 1569)
6494 * --major-os-version: Options. (line 1574)
6495 * --major-subsystem-version: Options. (line 1578)
6496 * --minor-image-version: Options. (line 1583)
6497 * --minor-os-version: Options. (line 1588)
6498 * --minor-subsystem-version: Options. (line 1592)
6499 * --mri-script=MRI-CMDFILE: Options. (line 141)
6500 * --multi-subspace: HPPA ELF32. (line 6)
6501 * --nmagic: Options. (line 389)
6502 * --no-accept-unknown-input-arch: Options. (line 663)
6503 * --no-add-needed: Options. (line 685)
6504 * --no-allow-shlib-undefined: Options. (line 916)
6505 * --no-as-needed: Options. (line 673)
6506 * --no-check-sections: Options. (line 767)
6507 * --no-define-common: Options. (line 789)
6508 * --no-demangle: Options. (line 818)
6509 * --no-dotsyms: PowerPC64 ELF64. (line 33)
6510 * --no-enum-size-warning: ARM. (line 106)
6511 * --no-fatal-warnings: Options. (line 837)
6512 * --no-fix-cortex-a8: i960. (line 39)
6513 * --no-gc-sections: Options. (line 852)
6514 * --no-keep-memory: Options. (line 895)
6515 * --no-multi-toc: PowerPC64 ELF64. (line 74)
6516 * --no-omagic: Options. (line 403)
6517 * --no-opd-optimize: PowerPC64 ELF64. (line 48)
6518 * --no-overlays: SPU ELF. (line 9)
6519 * --no-print-gc-sections: Options. (line 874)
6520 * --no-relax: Xtensa. (line 56)
6521 * --no-tls-optimize <1>: PowerPC64 ELF64. (line 43)
6522 * --no-tls-optimize: PowerPC ELF32. (line 48)
6523 * --no-toc-optimize: PowerPC64 ELF64. (line 60)
6524 * --no-trampoline: Options. (line 1809)
6525 * --no-undefined: Options. (line 902)
6526 * --no-undefined-version: Options. (line 935)
6527 * --no-warn-mismatch: Options. (line 948)
6528 * --no-warn-search-mismatch: Options. (line 957)
6529 * --no-wchar-size-warning: ARM. (line 113)
6530 * --no-whole-archive: Options. (line 961)
6531 * --noinhibit-exec: Options. (line 965)
6532 * --non-overlapping-opd: PowerPC64 ELF64. (line 54)
6533 * --oformat: Options. (line 977)
6534 * --omagic: Options. (line 394)
6535 * --out-implib: Options. (line 1605)
6536 * --output-def: Options. (line 1597)
6537 * --output=OUTPUT: Options. (line 409)
6538 * --pic-executable: Options. (line 990)
6539 * --pic-veneer: ARM. (line 119)
6540 * --plugin: SPU ELF. (line 6)
6541 * --print-gc-sections: Options. (line 874)
6542 * --print-map: Options. (line 352)
6543 * --reduce-memory-overheads: Options. (line 1418)
6544 * --relax: Options. (line 1006)
6545 * --relax on i960: i960. (line 31)
6546 * --relax on PowerPC: PowerPC ELF32. (line 6)
6547 * --relax on Xtensa: Xtensa. (line 27)
6548 * --relocatable: Options. (line 438)
6549 * --script=SCRIPT: Options. (line 481)
6550 * --sdata-got: PowerPC ELF32. (line 30)
6551 * --section-alignment: Options. (line 1780)
6552 * --section-start SECTIONNAME=ORG: Options. (line 1181)
6553 * --secure-plt: PowerPC ELF32. (line 23)
6554 * --sort-common: Options. (line 1126)
6555 * --sort-section alignment: Options. (line 1138)
6556 * --sort-section name: Options. (line 1134)
6557 * --split-by-file: Options. (line 1142)
6558 * --split-by-reloc: Options. (line 1147)
6559 * --stack: Options. (line 1786)
6560 * --stack-analysis: SPU ELF. (line 29)
6561 * --stats: Options. (line 1160)
6562 * --strip-all: Options. (line 468)
6563 * --strip-debug: Options. (line 472)
6564 * --stub-group-size: PowerPC64 ELF64. (line 6)
6565 * --stub-group-size=N <1>: HPPA ELF32. (line 12)
6566 * --stub-group-size=N: ARM. (line 124)
6567 * --subsystem: Options. (line 1793)
6568 * --support-old-code: ARM. (line 6)
6569 * --sysroot: Options. (line 1164)
6570 * --target-help: Options. (line 887)
6571 * --target1-abs: ARM. (line 27)
6572 * --target1-rel: ARM. (line 27)
6573 * --target2=TYPE: ARM. (line 32)
6574 * --thumb-entry=ENTRY: ARM. (line 17)
6575 * --trace: Options. (line 477)
6576 * --trace-symbol=SYMBOL: Options. (line 546)
6577 * --traditional-format: Options. (line 1169)
6578 * --undefined=SYMBOL: Options. (line 503)
6579 * --unique[=SECTION]: Options. (line 521)
6580 * --unresolved-symbols: Options. (line 1196)
6581 * --use-blx: ARM. (line 69)
6582 * --verbose: Options. (line 1225)
6583 * --version: Options. (line 530)
6584 * --version-script=VERSION-SCRIPTFILE: Options. (line 1231)
6585 * --vfp11-denorm-fix: ARM. (line 78)
6586 * --warn-common: Options. (line 1238)
6587 * --warn-constructors: Options. (line 1306)
6588 * --warn-multiple-gp: Options. (line 1311)
6589 * --warn-once: Options. (line 1325)
6590 * --warn-section-align: Options. (line 1329)
6591 * --warn-shared-textrel: Options. (line 1336)
6592 * --warn-unresolved-symbols: Options. (line 1339)
6593 * --whole-archive: Options. (line 1348)
6594 * --wrap: Options. (line 1362)
6595 * -AARCH: Options. (line 105)
6596 * -aKEYWORD: Options. (line 98)
6597 * -assert KEYWORD: Options. (line 695)
6598 * -b FORMAT: Options. (line 117)
6599 * -Bdynamic: Options. (line 698)
6600 * -Bgroup: Options. (line 708)
6601 * -Bshareable: Options. (line 1118)
6602 * -Bstatic: Options. (line 715)
6603 * -Bsymbolic: Options. (line 730)
6604 * -Bsymbolic-functions: Options. (line 737)
6605 * -c MRI-CMDFILE: Options. (line 141)
6606 * -call_shared: Options. (line 698)
6607 * -d: Options. (line 151)
6608 * -dc: Options. (line 151)
6609 * -dn: Options. (line 715)
6610 * -dp: Options. (line 151)
6611 * -dT SCRIPT: Options. (line 490)
6612 * -dy: Options. (line 698)
6613 * -E: Options. (line 181)
6614 * -e ENTRY: Options. (line 160)
6615 * -EB: Options. (line 200)
6616 * -EL: Options. (line 203)
6617 * -F: Options. (line 228)
6618 * -f: Options. (line 207)
6619 * -fini: Options. (line 252)
6620 * -G: Options. (line 261)
6621 * -g: Options. (line 258)
6622 * -hNAME: Options. (line 269)
6623 * -i: Options. (line 278)
6624 * -IFILE: Options. (line 831)
6625 * -init: Options. (line 281)
6626 * -LDIR: Options. (line 320)
6627 * -lNAMESPEC: Options. (line 287)
6628 * -M: Options. (line 352)
6629 * -m EMULATION: Options. (line 342)
6630 * -Map: Options. (line 891)
6631 * -N: Options. (line 394)
6632 * -n: Options. (line 389)
6633 * -non_shared: Options. (line 715)
6634 * -nostdlib: Options. (line 971)
6635 * -O LEVEL: Options. (line 415)
6636 * -o OUTPUT: Options. (line 409)
6637 * -pie: Options. (line 990)
6638 * -q: Options. (line 425)
6639 * -qmagic: Options. (line 1000)
6640 * -Qy: Options. (line 1003)
6641 * -r: Options. (line 438)
6642 * -R FILE: Options. (line 457)
6643 * -rpath: Options. (line 1041)
6644 * -rpath-link: Options. (line 1063)
6645 * -S: Options. (line 472)
6646 * -s: Options. (line 468)
6647 * -shared: Options. (line 1118)
6648 * -soname=NAME: Options. (line 269)
6649 * -static: Options. (line 715)
6650 * -t: Options. (line 477)
6651 * -T SCRIPT: Options. (line 481)
6652 * -Tbss ORG: Options. (line 1190)
6653 * -Tdata ORG: Options. (line 1190)
6654 * -Ttext ORG: Options. (line 1190)
6655 * -u SYMBOL: Options. (line 503)
6656 * -Ur: Options. (line 511)
6657 * -V: Options. (line 530)
6658 * -v: Options. (line 530)
6659 * -X: Options. (line 540)
6660 * -x: Options. (line 536)
6661 * -Y PATH: Options. (line 555)
6662 * -y SYMBOL: Options. (line 546)
6663 * -z defs: Options. (line 902)
6664 * -z KEYWORD: Options. (line 559)
6665 * -z muldefs: Options. (line 910)
6666 * .: Location Counter. (line 6)
6667 * /DISCARD/: Output Section Discarding.
6669 * :PHDR: Output Section Phdr.
6671 * =FILLEXP: Output Section Fill.
6673 * >REGION: Output Section Region.
6675 * [COMMON]: Input Section Common.
6677 * ABSOLUTE (MRI): MRI. (line 33)
6678 * absolute and relocatable symbols: Expression Section. (line 6)
6679 * absolute expressions: Expression Section. (line 6)
6680 * ABSOLUTE(EXP): Builtin Functions. (line 10)
6681 * ADDR(SECTION): Builtin Functions. (line 17)
6682 * address, section: Output Section Address.
6684 * ALIAS (MRI): MRI. (line 44)
6685 * ALIGN (MRI): MRI. (line 50)
6686 * align expression: Builtin Functions. (line 36)
6687 * align location counter: Builtin Functions. (line 36)
6688 * ALIGN(ALIGN): Builtin Functions. (line 36)
6689 * ALIGN(EXP,ALIGN): Builtin Functions. (line 36)
6690 * ALIGN(SECTION_ALIGN): Forced Output Alignment.
6692 * ALIGNOF(SECTION): Builtin Functions. (line 61)
6693 * allocating memory: MEMORY. (line 6)
6694 * architecture: Miscellaneous Commands.
6696 * architectures: Options. (line 105)
6697 * archive files, from cmd line: Options. (line 287)
6698 * archive search path in linker script: File Commands. (line 74)
6699 * arithmetic: Expressions. (line 6)
6700 * arithmetic operators: Operators. (line 6)
6701 * ARM interworking support: ARM. (line 6)
6702 * AS_NEEDED(FILES): File Commands. (line 54)
6703 * ASSERT: Miscellaneous Commands.
6705 * assertion in linker script: Miscellaneous Commands.
6707 * assignment in scripts: Assignments. (line 6)
6708 * AT(LMA): Output Section LMA. (line 6)
6709 * AT>LMA_REGION: Output Section LMA. (line 6)
6710 * automatic data imports: WIN32. (line 170)
6711 * back end: BFD. (line 6)
6712 * BASE (MRI): MRI. (line 54)
6713 * BE8: ARM. (line 23)
6714 * BFD canonical format: Canonical format. (line 11)
6715 * BFD requirements: BFD. (line 16)
6716 * big-endian objects: Options. (line 200)
6717 * binary input format: Options. (line 117)
6718 * BLOCK(EXP): Builtin Functions. (line 74)
6719 * bug criteria: Bug Criteria. (line 6)
6720 * bug reports: Bug Reporting. (line 6)
6721 * bugs in ld: Reporting Bugs. (line 6)
6722 * BYTE(EXPRESSION): Output Section Data.
6724 * C++ constructors, arranging in link: Output Section Keywords.
6726 * CHIP (MRI): MRI. (line 58)
6727 * COLLECT_NO_DEMANGLE: Environment. (line 29)
6728 * combining symbols, warnings on: Options. (line 1238)
6729 * command files: Scripts. (line 6)
6730 * command line: Options. (line 6)
6731 * common allocation: Options. (line 151)
6732 * common allocation in linker script: Miscellaneous Commands.
6734 * common symbol placement: Input Section Common.
6736 * compatibility, MRI: Options. (line 141)
6737 * constants in linker scripts: Constants. (line 6)
6738 * CONSTRUCTORS: Output Section Keywords.
6740 * constructors: Options. (line 511)
6741 * constructors, arranging in link: Output Section Keywords.
6743 * Cortex-A8 erratum workaround: i960. (line 39)
6744 * crash of linker: Bug Criteria. (line 9)
6745 * CREATE_OBJECT_SYMBOLS: Output Section Keywords.
6747 * creating a DEF file: WIN32. (line 137)
6748 * cross reference table: Options. (line 777)
6749 * cross references: Miscellaneous Commands.
6751 * current output location: Location Counter. (line 6)
6752 * data: Output Section Data.
6754 * DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE): Builtin Functions.
6756 * DATA_SEGMENT_END(EXP): Builtin Functions. (line 100)
6757 * DATA_SEGMENT_RELRO_END(OFFSET, EXP): Builtin Functions. (line 106)
6758 * dbx: Options. (line 1174)
6759 * DEF files, creating: Options. (line 1597)
6760 * default emulation: Environment. (line 21)
6761 * default input format: Environment. (line 9)
6762 * DEFINED(SYMBOL): Builtin Functions. (line 117)
6763 * deleting local symbols: Options. (line 536)
6764 * demangling, default: Environment. (line 29)
6765 * demangling, from command line: Options. (line 818)
6766 * direct linking to a dll: WIN32. (line 218)
6767 * discarding sections: Output Section Discarding.
6769 * discontinuous memory: MEMORY. (line 6)
6770 * DLLs, creating: Options. (line 1503)
6771 * DLLs, linking to: Options. (line 1628)
6772 * dot: Location Counter. (line 6)
6773 * dot inside sections: Location Counter. (line 36)
6774 * dot outside sections: Location Counter. (line 66)
6775 * dynamic linker, from command line: Options. (line 831)
6776 * dynamic symbol table: Options. (line 181)
6777 * ELF program headers: PHDRS. (line 6)
6778 * emulation: Options. (line 342)
6779 * emulation, default: Environment. (line 21)
6780 * END (MRI): MRI. (line 62)
6781 * endianness: Options. (line 200)
6782 * entry point: Entry Point. (line 6)
6783 * entry point, from command line: Options. (line 160)
6784 * entry point, thumb: ARM. (line 17)
6785 * ENTRY(SYMBOL): Entry Point. (line 6)
6786 * error on valid input: Bug Criteria. (line 12)
6787 * example of linker script: Simple Example. (line 6)
6788 * exporting DLL symbols: WIN32. (line 19)
6789 * expression evaluation order: Evaluation. (line 6)
6790 * expression sections: Expression Section. (line 6)
6791 * expression, absolute: Builtin Functions. (line 10)
6792 * expressions: Expressions. (line 6)
6793 * EXTERN: Miscellaneous Commands.
6795 * fatal signal: Bug Criteria. (line 9)
6796 * file name wildcard patterns: Input Section Wildcards.
6798 * FILEHDR: PHDRS. (line 61)
6799 * filename symbols: Output Section Keywords.
6801 * fill pattern, entire section: Output Section Fill.
6803 * FILL(EXPRESSION): Output Section Data.
6805 * finalization function: Options. (line 252)
6806 * first input file: File Commands. (line 82)
6807 * first instruction: Entry Point. (line 6)
6808 * FIX_V4BX: ARM. (line 44)
6809 * FIX_V4BX_INTERWORKING: ARM. (line 57)
6810 * FORCE_COMMON_ALLOCATION: Miscellaneous Commands.
6812 * forcing input section alignment: Forced Input Alignment.
6814 * forcing output section alignment: Forced Output Alignment.
6816 * forcing the creation of dynamic sections: Options. (line 434)
6817 * FORMAT (MRI): MRI. (line 66)
6818 * functions in expressions: Builtin Functions. (line 6)
6819 * garbage collection <1>: Input Section Keep. (line 6)
6820 * garbage collection: Options. (line 852)
6821 * generating optimized output: Options. (line 415)
6822 * GNU linker: Overview. (line 6)
6823 * GNUTARGET: Environment. (line 9)
6824 * GROUP(FILES): File Commands. (line 47)
6825 * grouping input files: File Commands. (line 47)
6826 * groups of archives: Options. (line 645)
6827 * H8/300 support: H8/300. (line 6)
6828 * header size: Builtin Functions. (line 182)
6829 * heap size: Options. (line 1539)
6830 * help: Options. (line 883)
6831 * holes: Location Counter. (line 12)
6832 * holes, filling: Output Section Data.
6834 * HPPA multiple sub-space stubs: HPPA ELF32. (line 6)
6835 * HPPA stub grouping: HPPA ELF32. (line 12)
6836 * i960 support: i960. (line 6)
6837 * image base: Options. (line 1546)
6838 * implicit linker scripts: Implicit Linker Scripts.
6840 * import libraries: WIN32. (line 10)
6841 * INCLUDE FILENAME: File Commands. (line 9)
6842 * including a linker script: File Commands. (line 9)
6843 * including an entire archive: Options. (line 1348)
6844 * incremental link: Options. (line 278)
6845 * INHIBIT_COMMON_ALLOCATION: Miscellaneous Commands.
6847 * initialization function: Options. (line 281)
6848 * initialized data in ROM: Output Section LMA. (line 26)
6849 * input file format in linker script: Format Commands. (line 35)
6850 * input filename symbols: Output Section Keywords.
6852 * input files in linker scripts: File Commands. (line 19)
6853 * input files, displaying: Options. (line 477)
6854 * input format: Options. (line 117)
6855 * input object files in linker scripts: File Commands. (line 19)
6856 * input section alignment: Forced Input Alignment.
6858 * input section basics: Input Section Basics.
6860 * input section wildcards: Input Section Wildcards.
6862 * input sections: Input Section. (line 6)
6863 * INPUT(FILES): File Commands. (line 19)
6864 * INSERT: Miscellaneous Commands.
6866 * insert user script into default script: Miscellaneous Commands.
6868 * integer notation: Constants. (line 6)
6869 * integer suffixes: Constants. (line 12)
6870 * internal object-file format: Canonical format. (line 11)
6871 * invalid input: Bug Criteria. (line 14)
6872 * K and M integer suffixes: Constants. (line 12)
6873 * KEEP: Input Section Keep. (line 6)
6874 * l =: MEMORY. (line 72)
6875 * lazy evaluation: Evaluation. (line 6)
6876 * ld bugs, reporting: Bug Reporting. (line 6)
6877 * LDEMULATION: Environment. (line 21)
6878 * len =: MEMORY. (line 72)
6879 * LENGTH =: MEMORY. (line 72)
6880 * LENGTH(MEMORY): Builtin Functions. (line 134)
6881 * library search path in linker script: File Commands. (line 74)
6882 * link map: Options. (line 352)
6883 * link-time runtime library search path: Options. (line 1063)
6884 * linker crash: Bug Criteria. (line 9)
6885 * linker script concepts: Basic Script Concepts.
6887 * linker script example: Simple Example. (line 6)
6888 * linker script file commands: File Commands. (line 6)
6889 * linker script format: Script Format. (line 6)
6890 * linker script input object files: File Commands. (line 19)
6891 * linker script simple commands: Simple Commands. (line 6)
6892 * linker scripts: Scripts. (line 6)
6893 * LIST (MRI): MRI. (line 77)
6894 * little-endian objects: Options. (line 203)
6895 * LOAD (MRI): MRI. (line 84)
6896 * load address: Output Section LMA. (line 6)
6897 * LOADADDR(SECTION): Builtin Functions. (line 137)
6898 * loading, preventing: Output Section Type.
6900 * local symbols, deleting: Options. (line 540)
6901 * location counter: Location Counter. (line 6)
6902 * LONG(EXPRESSION): Output Section Data.
6904 * M and K integer suffixes: Constants. (line 12)
6905 * M68HC11 and 68HC12 support: M68HC11/68HC12. (line 6)
6906 * machine architecture: Miscellaneous Commands.
6908 * machine dependencies: Machine Dependent. (line 6)
6909 * mapping input sections to output sections: Input Section. (line 6)
6910 * MAX: Builtin Functions. (line 142)
6911 * MEMORY: MEMORY. (line 6)
6912 * memory region attributes: MEMORY. (line 32)
6913 * memory regions: MEMORY. (line 6)
6914 * memory regions and sections: Output Section Region.
6916 * memory usage: Options. (line 895)
6917 * MIN: Builtin Functions. (line 145)
6918 * Motorola 68K GOT generation: M68K. (line 6)
6919 * MRI compatibility: MRI. (line 6)
6920 * MSP430 extra sections: MSP430. (line 11)
6921 * NAME (MRI): MRI. (line 90)
6922 * name, section: Output Section Name.
6924 * names: Symbols. (line 6)
6925 * naming the output file: Options. (line 409)
6926 * NEXT(EXP): Builtin Functions. (line 149)
6927 * NMAGIC: Options. (line 389)
6928 * NO_ENUM_SIZE_WARNING: ARM. (line 106)
6929 * NO_WCHAR_SIZE_WARNING: ARM. (line 113)
6930 * NOCROSSREFS(SECTIONS): Miscellaneous Commands.
6932 * NOLOAD: Output Section Type.
6934 * not enough room for program headers: Builtin Functions. (line 187)
6935 * o =: MEMORY. (line 67)
6936 * objdump -i: BFD. (line 6)
6937 * object file management: BFD. (line 6)
6938 * object files: Options. (line 29)
6939 * object formats available: BFD. (line 6)
6940 * object size: Options. (line 261)
6941 * OMAGIC: Options. (line 394)
6942 * opening object files: BFD outline. (line 6)
6943 * operators for arithmetic: Operators. (line 6)
6944 * options: Options. (line 6)
6945 * ORDER (MRI): MRI. (line 95)
6946 * org =: MEMORY. (line 67)
6947 * ORIGIN =: MEMORY. (line 67)
6948 * ORIGIN(MEMORY): Builtin Functions. (line 155)
6949 * orphan: Orphan Sections. (line 6)
6950 * output file after errors: Options. (line 965)
6951 * output file format in linker script: Format Commands. (line 10)
6952 * output file name in linker script: File Commands. (line 64)
6953 * output section alignment: Forced Output Alignment.
6955 * output section attributes: Output Section Attributes.
6957 * output section data: Output Section Data.
6959 * OUTPUT(FILENAME): File Commands. (line 64)
6960 * OUTPUT_ARCH(BFDARCH): Miscellaneous Commands.
6962 * OUTPUT_FORMAT(BFDNAME): Format Commands. (line 10)
6963 * OVERLAY: Overlay Description.
6965 * overlays: Overlay Description.
6967 * partial link: Options. (line 438)
6968 * PHDRS: PHDRS. (line 6)
6969 * PIC_VENEER: ARM. (line 119)
6970 * position independent executables: Options. (line 992)
6971 * PowerPC ELF32 options: PowerPC ELF32. (line 13)
6972 * PowerPC GOT: PowerPC ELF32. (line 30)
6973 * PowerPC long branches: PowerPC ELF32. (line 6)
6974 * PowerPC PLT: PowerPC ELF32. (line 13)
6975 * PowerPC stub symbols: PowerPC ELF32. (line 44)
6976 * PowerPC TLS optimization: PowerPC ELF32. (line 48)
6977 * PowerPC64 dot symbols: PowerPC64 ELF64. (line 33)
6978 * PowerPC64 ELF64 options: PowerPC64 ELF64. (line 6)
6979 * PowerPC64 multi-TOC: PowerPC64 ELF64. (line 74)
6980 * PowerPC64 OPD optimization: PowerPC64 ELF64. (line 48)
6981 * PowerPC64 OPD spacing: PowerPC64 ELF64. (line 54)
6982 * PowerPC64 stub grouping: PowerPC64 ELF64. (line 6)
6983 * PowerPC64 stub symbols: PowerPC64 ELF64. (line 29)
6984 * PowerPC64 TLS optimization: PowerPC64 ELF64. (line 43)
6985 * PowerPC64 TOC optimization: PowerPC64 ELF64. (line 60)
6986 * precedence in expressions: Operators. (line 6)
6987 * prevent unnecessary loading: Output Section Type.
6989 * program headers: PHDRS. (line 6)
6990 * program headers and sections: Output Section Phdr.
6992 * program headers, not enough room: Builtin Functions. (line 187)
6993 * program segments: PHDRS. (line 6)
6994 * PROVIDE: PROVIDE. (line 6)
6995 * PROVIDE_HIDDEN: PROVIDE_HIDDEN. (line 6)
6996 * PUBLIC (MRI): MRI. (line 103)
6997 * QUAD(EXPRESSION): Output Section Data.
6999 * quoted symbol names: Symbols. (line 6)
7000 * read-only text: Options. (line 389)
7001 * read/write from cmd line: Options. (line 394)
7002 * regions of memory: MEMORY. (line 6)
7003 * relative expressions: Expression Section. (line 6)
7004 * relaxing addressing modes: Options. (line 1006)
7005 * relaxing on H8/300: H8/300. (line 9)
7006 * relaxing on i960: i960. (line 31)
7007 * relaxing on M68HC11: M68HC11/68HC12. (line 12)
7008 * relaxing on Xtensa: Xtensa. (line 27)
7009 * relocatable and absolute symbols: Expression Section. (line 6)
7010 * relocatable output: Options. (line 438)
7011 * removing sections: Output Section Discarding.
7013 * reporting bugs in ld: Reporting Bugs. (line 6)
7014 * requirements for BFD: BFD. (line 16)
7015 * retain relocations in final executable: Options. (line 425)
7016 * retaining specified symbols: Options. (line 1027)
7017 * ROM initialized data: Output Section LMA. (line 26)
7018 * round up expression: Builtin Functions. (line 36)
7019 * round up location counter: Builtin Functions. (line 36)
7020 * runtime library name: Options. (line 269)
7021 * runtime library search path: Options. (line 1041)
7022 * runtime pseudo-relocation: WIN32. (line 196)
7023 * scaled integers: Constants. (line 12)
7024 * scommon section: Input Section Common.
7026 * script files: Options. (line 481)
7027 * scripts: Scripts. (line 6)
7028 * search directory, from cmd line: Options. (line 320)
7029 * search path in linker script: File Commands. (line 74)
7030 * SEARCH_DIR(PATH): File Commands. (line 74)
7031 * SECT (MRI): MRI. (line 109)
7032 * section address: Output Section Address.
7034 * section address in expression: Builtin Functions. (line 17)
7035 * section alignment: Builtin Functions. (line 61)
7036 * section alignment, warnings on: Options. (line 1329)
7037 * section data: Output Section Data.
7039 * section fill pattern: Output Section Fill.
7041 * section load address: Output Section LMA. (line 6)
7042 * section load address in expression: Builtin Functions. (line 137)
7043 * section name: Output Section Name.
7045 * section name wildcard patterns: Input Section Wildcards.
7047 * section size: Builtin Functions. (line 166)
7048 * section, assigning to memory region: Output Section Region.
7050 * section, assigning to program header: Output Section Phdr.
7052 * SECTIONS: SECTIONS. (line 6)
7053 * sections, discarding: Output Section Discarding.
7055 * segment origins, cmd line: Options. (line 1190)
7056 * SEGMENT_START(SEGMENT, DEFAULT): Builtin Functions. (line 158)
7057 * segments, ELF: PHDRS. (line 6)
7058 * shared libraries: Options. (line 1120)
7059 * SHORT(EXPRESSION): Output Section Data.
7061 * SIZEOF(SECTION): Builtin Functions. (line 166)
7062 * SIZEOF_HEADERS: Builtin Functions. (line 182)
7063 * small common symbols: Input Section Common.
7065 * SORT: Input Section Wildcards.
7067 * SORT_BY_ALIGNMENT: Input Section Wildcards.
7069 * SORT_BY_NAME: Input Section Wildcards.
7071 * SPU: SPU ELF. (line 29)
7072 * SPU ELF options: SPU ELF. (line 6)
7073 * SPU extra overlay stubs: SPU ELF. (line 19)
7074 * SPU local store size: SPU ELF. (line 24)
7075 * SPU overlay stub symbols: SPU ELF. (line 15)
7076 * SPU overlays: SPU ELF. (line 9)
7077 * SPU plugins: SPU ELF. (line 6)
7078 * SQUAD(EXPRESSION): Output Section Data.
7080 * stack size: Options. (line 1786)
7081 * standard Unix system: Options. (line 7)
7082 * start of execution: Entry Point. (line 6)
7083 * STARTUP(FILENAME): File Commands. (line 82)
7084 * strip all symbols: Options. (line 468)
7085 * strip debugger symbols: Options. (line 472)
7086 * stripping all but some symbols: Options. (line 1027)
7087 * STUB_GROUP_SIZE: ARM. (line 124)
7088 * SUBALIGN(SUBSECTION_ALIGN): Forced Input Alignment.
7090 * suffixes for integers: Constants. (line 12)
7091 * symbol defaults: Builtin Functions. (line 117)
7092 * symbol definition, scripts: Assignments. (line 6)
7093 * symbol names: Symbols. (line 6)
7094 * symbol tracing: Options. (line 546)
7095 * symbol versions: VERSION. (line 6)
7096 * symbol-only input: Options. (line 457)
7097 * symbols, from command line: Options. (line 805)
7098 * symbols, relocatable and absolute: Expression Section. (line 6)
7099 * symbols, retaining selectively: Options. (line 1027)
7100 * synthesizing linker: Options. (line 1006)
7101 * synthesizing on H8/300: H8/300. (line 14)
7102 * TARGET(BFDNAME): Format Commands. (line 35)
7103 * TARGET1: ARM. (line 27)
7104 * TARGET2: ARM. (line 32)
7105 * thumb entry point: ARM. (line 17)
7106 * TI COFF versions: TI COFF. (line 6)
7107 * traditional format: Options. (line 1169)
7108 * trampoline generation on M68HC11: M68HC11/68HC12. (line 31)
7109 * trampoline generation on M68HC12: M68HC11/68HC12. (line 31)
7110 * unallocated address, next: Builtin Functions. (line 149)
7111 * undefined symbol: Options. (line 503)
7112 * undefined symbol in linker script: Miscellaneous Commands.
7114 * undefined symbols, warnings on: Options. (line 1325)
7115 * uninitialized data placement: Input Section Common.
7117 * unspecified memory: Output Section Data.
7119 * usage: Options. (line 883)
7120 * USE_BLX: ARM. (line 69)
7121 * using a DEF file: WIN32. (line 42)
7122 * using auto-export functionality: WIN32. (line 22)
7123 * Using decorations: WIN32. (line 141)
7124 * variables, defining: Assignments. (line 6)
7125 * verbose: Options. (line 1225)
7126 * version: Options. (line 530)
7127 * version script: VERSION. (line 6)
7128 * version script, symbol versions: Options. (line 1231)
7129 * VERSION {script text}: VERSION. (line 6)
7130 * versions of symbols: VERSION. (line 6)
7131 * VFP11_DENORM_FIX: ARM. (line 78)
7132 * warnings, on combining symbols: Options. (line 1238)
7133 * warnings, on section alignment: Options. (line 1329)
7134 * warnings, on undefined symbols: Options. (line 1325)
7135 * weak externals: WIN32. (line 386)
7136 * what is this?: Overview. (line 6)
7137 * wildcard file name patterns: Input Section Wildcards.
7139 * Xtensa options: Xtensa. (line 56)
7140 * Xtensa processors: Xtensa. (line 6)
7146 Node: Overview
\7f1556
7147 Node: Invocation
\7f2670
7148 Node: Options
\7f3078
7149 Node: Environment
\7f85542
7150 Node: Scripts
\7f87302
7151 Node: Basic Script Concepts
\7f89036
7152 Node: Script Format
\7f91743
7153 Node: Simple Example
\7f92606
7154 Node: Simple Commands
\7f95702
7155 Node: Entry Point
\7f96153
7156 Node: File Commands
\7f96912
7157 Node: Format Commands
\7f100913
7158 Node: Miscellaneous Commands
\7f102879
7159 Node: Assignments
\7f106258
7160 Node: Simple Assignments
\7f106749
7161 Node: PROVIDE
\7f108485
7162 Node: PROVIDE_HIDDEN
\7f109690
7163 Node: Source Code Reference
\7f109934
7164 Node: SECTIONS
\7f113514
7165 Node: Output Section Description
\7f115405
7166 Node: Output Section Name
\7f116458
7167 Node: Output Section Address
\7f117334
7168 Node: Input Section
\7f118983
7169 Node: Input Section Basics
\7f119784
7170 Node: Input Section Wildcards
\7f123002
7171 Node: Input Section Common
\7f127735
7172 Node: Input Section Keep
\7f129217
7173 Node: Input Section Example
\7f129707
7174 Node: Output Section Data
\7f130675
7175 Node: Output Section Keywords
\7f133452
7176 Node: Output Section Discarding
\7f137021
7177 Node: Output Section Attributes
\7f138202
7178 Node: Output Section Type
\7f139205
7179 Node: Output Section LMA
\7f140359
7180 Node: Forced Output Alignment
\7f142872
7181 Node: Forced Input Alignment
\7f143140
7182 Node: Output Section Region
\7f143525
7183 Node: Output Section Phdr
\7f143955
7184 Node: Output Section Fill
\7f144619
7185 Node: Overlay Description
\7f145761
7186 Node: MEMORY
\7f150063
7187 Node: PHDRS
\7f154262
7188 Node: VERSION
\7f159301
7189 Node: Expressions
\7f167093
7190 Node: Constants
\7f167971
7191 Node: Symbols
\7f168532
7192 Node: Orphan Sections
\7f169270
7193 Node: Location Counter
\7f170434
7194 Node: Operators
\7f174870
7195 Node: Evaluation
\7f175792
7196 Node: Expression Section
\7f177156
7197 Node: Builtin Functions
\7f178645
7198 Node: Implicit Linker Scripts
\7f186607
7199 Node: Machine Dependent
\7f187382
7200 Node: H8/300
\7f188398
7202 Node: M68HC11/68HC12
\7f192124
7204 Node: HPPA ELF32
\7f200827
7207 Node: MSP430
\7f204524
7208 Node: PowerPC ELF32
\7f205573
7209 Node: PowerPC64 ELF64
\7f208187
7210 Node: SPU ELF
\7f212603
7211 Node: TI COFF
\7f215235
7212 Node: WIN32
\7f215761
7213 Node: Xtensa
\7f234118
7215 Node: BFD outline
\7f238695
7216 Node: BFD information loss
\7f239981
7217 Node: Canonical format
\7f242498
7218 Node: Reporting Bugs
\7f246855
7219 Node: Bug Criteria
\7f247549
7220 Node: Bug Reporting
\7f248248
7222 Node: GNU Free Documentation License
\7f259930
7223 Node: LD Index
\7f279635