1 This is ld.info, produced by makeinfo version 4.8 from
2 /home/jingyu/projects/gcc/android-toolchainsrc/build/../binutils/binutils-2.19/ld/ld.texinfo.
5 * Ld: (ld). The GNU linker.
8 This file documents the GNU linker LD (GNU Binutils) version 2.19.
10 Copyright (C) 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
11 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
13 Permission is granted to copy, distribute and/or modify this document
14 under the terms of the GNU Free Documentation License, Version 1.1 or
15 any later version published by the Free Software Foundation; with no
16 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
17 Texts. A copy of the license is included in the section entitled "GNU
18 Free Documentation License".
21 File: ld.info, Node: Top, Next: Overview, Up: (dir)
26 This file documents the GNU linker ld (GNU Binutils) version 2.19.
28 This document is distributed under the terms of the GNU Free
29 Documentation License. A copy of the license is included in the
30 section entitled "GNU Free Documentation License".
35 * Invocation:: Invocation
36 * Scripts:: Linker Scripts
38 * Machine Dependent:: Machine Dependent Features
42 * Reporting Bugs:: Reporting Bugs
43 * MRI:: MRI Compatible Script Files
44 * GNU Free Documentation License:: GNU Free Documentation License
48 File: ld.info, Node: Overview, Next: Invocation, Prev: Top, Up: Top
53 `ld' combines a number of object and archive files, relocates their
54 data and ties up symbol references. Usually the last step in compiling
55 a program is to run `ld'.
57 `ld' accepts Linker Command Language files written in a superset of
58 AT&T's Link Editor Command Language syntax, to provide explicit and
59 total control over the linking process.
61 This version of `ld' uses the general purpose BFD libraries to
62 operate on object files. This allows `ld' to read, combine, and write
63 object files in many different formats--for example, COFF or `a.out'.
64 Different formats may be linked together to produce any available kind
65 of object file. *Note BFD::, for more information.
67 Aside from its flexibility, the GNU linker is more helpful than other
68 linkers in providing diagnostic information. Many linkers abandon
69 execution immediately upon encountering an error; whenever possible,
70 `ld' continues executing, allowing you to identify other errors (or, in
71 some cases, to get an output file in spite of the error).
74 File: ld.info, Node: Invocation, Next: Scripts, Prev: Overview, Up: Top
79 The GNU linker `ld' is meant to cover a broad range of situations, and
80 to be as compatible as possible with other linkers. As a result, you
81 have many choices to control its behavior.
85 * Options:: Command Line Options
86 * Environment:: Environment Variables
89 File: ld.info, Node: Options, Next: Environment, Up: Invocation
91 2.1 Command Line Options
92 ========================
94 The linker supports a plethora of command-line options, but in actual
95 practice few of them are used in any particular context. For instance,
96 a frequent use of `ld' is to link standard Unix object files on a
97 standard, supported Unix system. On such a system, to link a file
100 ld -o OUTPUT /lib/crt0.o hello.o -lc
102 This tells `ld' to produce a file called OUTPUT as the result of
103 linking the file `/lib/crt0.o' with `hello.o' and the library `libc.a',
104 which will come from the standard search directories. (See the
105 discussion of the `-l' option below.)
107 Some of the command-line options to `ld' may be specified at any
108 point in the command line. However, options which refer to files, such
109 as `-l' or `-T', cause the file to be read at the point at which the
110 option appears in the command line, relative to the object files and
111 other file options. Repeating non-file options with a different
112 argument will either have no further effect, or override prior
113 occurrences (those further to the left on the command line) of that
114 option. Options which may be meaningfully specified more than once are
115 noted in the descriptions below.
117 Non-option arguments are object files or archives which are to be
118 linked together. They may follow, precede, or be mixed in with
119 command-line options, except that an object file argument may not be
120 placed between an option and its argument.
122 Usually the linker is invoked with at least one object file, but you
123 can specify other forms of binary input files using `-l', `-R', and the
124 script command language. If _no_ binary input files at all are
125 specified, the linker does not produce any output, and issues the
126 message `No input files'.
128 If the linker cannot recognize the format of an object file, it will
129 assume that it is a linker script. A script specified in this way
130 augments the main linker script used for the link (either the default
131 linker script or the one specified by using `-T'). This feature
132 permits the linker to link against a file which appears to be an object
133 or an archive, but actually merely defines some symbol values, or uses
134 `INPUT' or `GROUP' to load other objects. Specifying a script in this
135 way merely augments the main linker script, with the extra commands
136 placed after the main script; use the `-T' option to replace the
137 default linker script entirely, but note the effect of the `INSERT'
138 command. *Note Scripts::.
140 For options whose names are a single letter, option arguments must
141 either follow the option letter without intervening whitespace, or be
142 given as separate arguments immediately following the option that
145 For options whose names are multiple letters, either one dash or two
146 can precede the option name; for example, `-trace-symbol' and
147 `--trace-symbol' are equivalent. Note--there is one exception to this
148 rule. Multiple letter options that start with a lower case 'o' can
149 only be preceded by two dashes. This is to reduce confusion with the
150 `-o' option. So for example `-omagic' sets the output file name to
151 `magic' whereas `--omagic' sets the NMAGIC flag on the output.
153 Arguments to multiple-letter options must either be separated from
154 the option name by an equals sign, or be given as separate arguments
155 immediately following the option that requires them. For example,
156 `--trace-symbol foo' and `--trace-symbol=foo' are equivalent. Unique
157 abbreviations of the names of multiple-letter options are accepted.
159 Note--if the linker is being invoked indirectly, via a compiler
160 driver (e.g. `gcc') then all the linker command line options should be
161 prefixed by `-Wl,' (or whatever is appropriate for the particular
162 compiler driver) like this:
164 gcc -Wl,--startgroup foo.o bar.o -Wl,--endgroup
166 This is important, because otherwise the compiler driver program may
167 silently drop the linker options, resulting in a bad link.
169 Here is a table of the generic command line switches accepted by the
173 Read command-line options from FILE. The options read are
174 inserted in place of the original @FILE option. If FILE does not
175 exist, or cannot be read, then the option will be treated
176 literally, and not removed.
178 Options in FILE are separated by whitespace. A whitespace
179 character may be included in an option by surrounding the entire
180 option in either single or double quotes. Any character
181 (including a backslash) may be included by prefixing the character
182 to be included with a backslash. The FILE may itself contain
183 additional @FILE options; any such options will be processed
187 This option is supported for HP/UX compatibility. The KEYWORD
188 argument must be one of the strings `archive', `shared', or
189 `default'. `-aarchive' is functionally equivalent to `-Bstatic',
190 and the other two keywords are functionally equivalent to
191 `-Bdynamic'. This option may be used any number of times.
194 `--architecture=ARCHITECTURE'
195 In the current release of `ld', this option is useful only for the
196 Intel 960 family of architectures. In that `ld' configuration, the
197 ARCHITECTURE argument identifies the particular architecture in
198 the 960 family, enabling some safeguards and modifying the
199 archive-library search path. *Note `ld' and the Intel 960 family:
202 Future releases of `ld' may support similar functionality for
203 other architecture families.
206 `--format=INPUT-FORMAT'
207 `ld' may be configured to support more than one kind of object
208 file. If your `ld' is configured this way, you can use the `-b'
209 option to specify the binary format for input object files that
210 follow this option on the command line. Even when `ld' is
211 configured to support alternative object formats, you don't
212 usually need to specify this, as `ld' should be configured to
213 expect as a default input format the most usual format on each
214 machine. INPUT-FORMAT is a text string, the name of a particular
215 format supported by the BFD libraries. (You can list the
216 available binary formats with `objdump -i'.) *Note BFD::.
218 You may want to use this option if you are linking files with an
219 unusual binary format. You can also use `-b' to switch formats
220 explicitly (when linking object files of different formats), by
221 including `-b INPUT-FORMAT' before each group of object files in a
224 The default format is taken from the environment variable
225 `GNUTARGET'. *Note Environment::. You can also define the input
226 format from a script, using the command `TARGET'; see *Note Format
230 `--mri-script=MRI-COMMANDFILE'
231 For compatibility with linkers produced by MRI, `ld' accepts script
232 files written in an alternate, restricted command language,
233 described in *Note MRI Compatible Script Files: MRI. Introduce
234 MRI script files with the option `-c'; use the `-T' option to run
235 linker scripts written in the general-purpose `ld' scripting
236 language. If MRI-CMDFILE does not exist, `ld' looks for it in the
237 directories specified by any `-L' options.
242 These three options are equivalent; multiple forms are supported
243 for compatibility with other linkers. They assign space to common
244 symbols even if a relocatable output file is specified (with
245 `-r'). The script command `FORCE_COMMON_ALLOCATION' has the same
246 effect. *Note Miscellaneous Commands::.
250 Use ENTRY as the explicit symbol for beginning execution of your
251 program, rather than the default entry point. If there is no
252 symbol named ENTRY, the linker will try to parse ENTRY as a number,
253 and use that as the entry address (the number will be interpreted
254 in base 10; you may use a leading `0x' for base 16, or a leading
255 `0' for base 8). *Note Entry Point::, for a discussion of defaults
256 and other ways of specifying the entry point.
258 `--exclude-libs LIB,LIB,...'
259 Specifies a list of archive libraries from which symbols should
260 not be automatically exported. The library names may be delimited
261 by commas or colons. Specifying `--exclude-libs ALL' excludes
262 symbols in all archive libraries from automatic export. This
263 option is available only for the i386 PE targeted port of the
264 linker and for ELF targeted ports. For i386 PE, symbols
265 explicitly listed in a .def file are still exported, regardless of
266 this option. For ELF targeted ports, symbols affected by this
267 option will be treated as hidden.
271 When creating a dynamically linked executable, add all symbols to
272 the dynamic symbol table. The dynamic symbol table is the set of
273 symbols which are visible from dynamic objects at run time.
275 If you do not use this option, the dynamic symbol table will
276 normally contain only those symbols which are referenced by some
277 dynamic object mentioned in the link.
279 If you use `dlopen' to load a dynamic object which needs to refer
280 back to the symbols defined by the program, rather than some other
281 dynamic object, then you will probably need to use this option when
282 linking the program itself.
284 You can also use the dynamic list to control what symbols should
285 be added to the dynamic symbol table if the output format supports
286 it. See the description of `--dynamic-list'.
289 Link big-endian objects. This affects the default output format.
292 Link little-endian objects. This affects the default output
297 When creating an ELF shared object, set the internal DT_AUXILIARY
298 field to the specified name. This tells the dynamic linker that
299 the symbol table of the shared object should be used as an
300 auxiliary filter on the symbol table of the shared object NAME.
302 If you later link a program against this filter object, then, when
303 you run the program, the dynamic linker will see the DT_AUXILIARY
304 field. If the dynamic linker resolves any symbols from the filter
305 object, it will first check whether there is a definition in the
306 shared object NAME. If there is one, it will be used instead of
307 the definition in the filter object. The shared object NAME need
308 not exist. Thus the shared object NAME may be used to provide an
309 alternative implementation of certain functions, perhaps for
310 debugging or for machine specific performance.
312 This option may be specified more than once. The DT_AUXILIARY
313 entries will be created in the order in which they appear on the
318 When creating an ELF shared object, set the internal DT_FILTER
319 field to the specified name. This tells the dynamic linker that
320 the symbol table of the shared object which is being created
321 should be used as a filter on the symbol table of the shared
324 If you later link a program against this filter object, then, when
325 you run the program, the dynamic linker will see the DT_FILTER
326 field. The dynamic linker will resolve symbols according to the
327 symbol table of the filter object as usual, but it will actually
328 link to the definitions found in the shared object NAME. Thus the
329 filter object can be used to select a subset of the symbols
330 provided by the object NAME.
332 Some older linkers used the `-F' option throughout a compilation
333 toolchain for specifying object-file format for both input and
334 output object files. The GNU linker uses other mechanisms for
335 this purpose: the `-b', `--format', `--oformat' options, the
336 `TARGET' command in linker scripts, and the `GNUTARGET'
337 environment variable. The GNU linker will ignore the `-F' option
338 when not creating an ELF shared object.
341 When creating an ELF executable or shared object, call NAME when
342 the executable or shared object is unloaded, by setting DT_FINI to
343 the address of the function. By default, the linker uses `_fini'
344 as the function to call.
347 Ignored. Provided for compatibility with other tools.
351 Set the maximum size of objects to be optimized using the GP
352 register to SIZE. This is only meaningful for object file formats
353 such as MIPS ECOFF which supports putting large and small objects
354 into different sections. This is ignored for other object file
359 When creating an ELF shared object, set the internal DT_SONAME
360 field to the specified name. When an executable is linked with a
361 shared object which has a DT_SONAME field, then when the
362 executable is run the dynamic linker will attempt to load the
363 shared object specified by the DT_SONAME field rather than the
364 using the file name given to the linker.
367 Perform an incremental link (same as option `-r').
370 When creating an ELF executable or shared object, call NAME when
371 the executable or shared object is loaded, by setting DT_INIT to
372 the address of the function. By default, the linker uses `_init'
373 as the function to call.
377 Add the archive or object file specified by NAMESPEC to the list
378 of files to link. This option may be used any number of times.
379 If NAMESPEC is of the form `:FILENAME', `ld' will search the
380 library path for a file called FILENAME, otherise it will search
381 the library path for a file called `libNAMESPEC.a'.
383 On systems which support shared libraries, `ld' may also search for
384 files other than `libNAMESPEC.a'. Specifically, on ELF and SunOS
385 systems, `ld' will search a directory for a library called
386 `libNAMESPEC.so' before searching for one called `libNAMESPEC.a'.
387 (By convention, a `.so' extension indicates a shared library.)
388 Note that this behavior does not apply to `:FILENAME', which
389 always specifies a file called FILENAME.
391 The linker will search an archive only once, at the location where
392 it is specified on the command line. If the archive defines a
393 symbol which was undefined in some object which appeared before
394 the archive on the command line, the linker will include the
395 appropriate file(s) from the archive. However, an undefined
396 symbol in an object appearing later on the command line will not
397 cause the linker to search the archive again.
399 See the `-(' option for a way to force the linker to search
400 archives multiple times.
402 You may list the same archive multiple times on the command line.
404 This type of archive searching is standard for Unix linkers.
405 However, if you are using `ld' on AIX, note that it is different
406 from the behaviour of the AIX linker.
409 `--library-path=SEARCHDIR'
410 Add path SEARCHDIR to the list of paths that `ld' will search for
411 archive libraries and `ld' control scripts. You may use this
412 option any number of times. The directories are searched in the
413 order in which they are specified on the command line.
414 Directories specified on the command line are searched before the
415 default directories. All `-L' options apply to all `-l' options,
416 regardless of the order in which the options appear.
418 If SEARCHDIR begins with `=', then the `=' will be replaced by the
419 "sysroot prefix", a path specified when the linker is configured.
421 The default set of paths searched (without being specified with
422 `-L') depends on which emulation mode `ld' is using, and in some
423 cases also on how it was configured. *Note Environment::.
425 The paths can also be specified in a link script with the
426 `SEARCH_DIR' command. Directories specified this way are searched
427 at the point in which the linker script appears in the command
431 Emulate the EMULATION linker. You can list the available
432 emulations with the `--verbose' or `-V' options.
434 If the `-m' option is not used, the emulation is taken from the
435 `LDEMULATION' environment variable, if that is defined.
437 Otherwise, the default emulation depends upon how the linker was
442 Print a link map to the standard output. A link map provides
443 information about the link, including the following:
445 * Where object files are mapped into memory.
447 * How common symbols are allocated.
449 * All archive members included in the link, with a mention of
450 the symbol which caused the archive member to be brought in.
452 * The values assigned to symbols.
454 Note - symbols whose values are computed by an expression
455 which involves a reference to a previous value of the same
456 symbol may not have correct result displayed in the link map.
457 This is because the linker discards intermediate results and
458 only retains the final value of an expression. Under such
459 circumstances the linker will display the final value
460 enclosed by square brackets. Thus for example a linker
467 will produce the following output in the link map if the `-M'
471 [0x0000000c] foo = (foo * 0x4)
472 [0x0000000c] foo = (foo + 0x8)
474 See *Note Expressions:: for more information about
475 expressions in linker scripts.
479 Turn off page alignment of sections, and mark the output as
480 `NMAGIC' if possible.
484 Set the text and data sections to be readable and writable. Also,
485 do not page-align the data segment, and disable linking against
486 shared libraries. If the output format supports Unix style magic
487 numbers, mark the output as `OMAGIC'. Note: Although a writable
488 text section is allowed for PE-COFF targets, it does not conform
489 to the format specification published by Microsoft.
492 This option negates most of the effects of the `-N' option. It
493 sets the text section to be read-only, and forces the data segment
494 to be page-aligned. Note - this option does not enable linking
495 against shared libraries. Use `-Bdynamic' for this.
499 Use OUTPUT as the name for the program produced by `ld'; if this
500 option is not specified, the name `a.out' is used by default. The
501 script command `OUTPUT' can also specify the output file name.
504 If LEVEL is a numeric values greater than zero `ld' optimizes the
505 output. This might take significantly longer and therefore
506 probably should only be enabled for the final binary. At the
507 moment this option only affects ELF shared library generation.
508 Future releases of the linker may make more use of this option.
509 Also currently there is no difference in the linker's behaviour
510 for different non-zero values of this option. Again this may
511 change with future releases.
515 Leave relocation sections and contents in fully linked executables.
516 Post link analysis and optimization tools may need this
517 information in order to perform correct modifications of
518 executables. This results in larger executables.
520 This option is currently only supported on ELF platforms.
523 Force the output file to have dynamic sections. This option is
524 specific to VxWorks targets.
528 Generate relocatable output--i.e., generate an output file that
529 can in turn serve as input to `ld'. This is often called "partial
530 linking". As a side effect, in environments that support standard
531 Unix magic numbers, this option also sets the output file's magic
532 number to `OMAGIC'. If this option is not specified, an absolute
533 file is produced. When linking C++ programs, this option _will
534 not_ resolve references to constructors; to do that, use `-Ur'.
536 When an input file does not have the same format as the output
537 file, partial linking is only supported if that input file does
538 not contain any relocations. Different output formats can have
539 further restrictions; for example some `a.out'-based formats do
540 not support partial linking with input files in other formats at
543 This option does the same thing as `-i'.
546 `--just-symbols=FILENAME'
547 Read symbol names and their addresses from FILENAME, but do not
548 relocate it or include it in the output. This allows your output
549 file to refer symbolically to absolute locations of memory defined
550 in other programs. You may use this option more than once.
552 For compatibility with other ELF linkers, if the `-R' option is
553 followed by a directory name, rather than a file name, it is
554 treated as the `-rpath' option.
558 Omit all symbol information from the output file.
562 Omit debugger symbol information (but not all symbols) from the
567 Print the names of the input files as `ld' processes them.
570 `--script=SCRIPTFILE'
571 Use SCRIPTFILE as the linker script. This script replaces `ld''s
572 default linker script (rather than adding to it), so COMMANDFILE
573 must specify everything necessary to describe the output file.
574 *Note Scripts::. If SCRIPTFILE does not exist in the current
575 directory, `ld' looks for it in the directories specified by any
576 preceding `-L' options. Multiple `-T' options accumulate.
579 `--default-script=SCRIPTFILE'
580 Use SCRIPTFILE as the default linker script. *Note Scripts::.
582 This option is similar to the `--script' option except that
583 processing of the script is delayed until after the rest of the
584 command line has been processed. This allows options placed after
585 the `--default-script' option on the command line to affect the
586 behaviour of the linker script, which can be important when the
587 linker command line cannot be directly controlled by the user.
588 (eg because the command line is being constructed by another tool,
593 Force SYMBOL to be entered in the output file as an undefined
594 symbol. Doing this may, for example, trigger linking of additional
595 modules from standard libraries. `-u' may be repeated with
596 different option arguments to enter additional undefined symbols.
597 This option is equivalent to the `EXTERN' linker script command.
600 For anything other than C++ programs, this option is equivalent to
601 `-r': it generates relocatable output--i.e., an output file that
602 can in turn serve as input to `ld'. When linking C++ programs,
603 `-Ur' _does_ resolve references to constructors, unlike `-r'. It
604 does not work to use `-Ur' on files that were themselves linked
605 with `-Ur'; once the constructor table has been built, it cannot
606 be added to. Use `-Ur' only for the last partial link, and `-r'
610 Creates a separate output section for every input section matching
611 SECTION, or if the optional wildcard SECTION argument is missing,
612 for every orphan input section. An orphan section is one not
613 specifically mentioned in a linker script. You may use this option
614 multiple times on the command line; It prevents the normal
615 merging of input sections with the same name, overriding output
616 section assignments in a linker script.
621 Display the version number for `ld'. The `-V' option also lists
622 the supported emulations.
626 Delete all local symbols.
630 Delete all temporary local symbols. (These symbols start with
631 system-specific local label prefixes, typically `.L' for ELF
632 systems or `L' for traditional a.out systems.)
635 `--trace-symbol=SYMBOL'
636 Print the name of each linked file in which SYMBOL appears. This
637 option may be given any number of times. On many systems it is
638 necessary to prepend an underscore.
640 This option is useful when you have an undefined symbol in your
641 link but don't know where the reference is coming from.
644 Add PATH to the default library search path. This option exists
645 for Solaris compatibility.
648 The recognized keywords are:
650 Combines multiple reloc sections and sorts them to make
651 dynamic symbol lookup caching possible.
654 Disallows undefined symbols in object files. Undefined
655 symbols in shared libraries are still allowed.
658 Marks the object as requiring executable stack.
661 This option is only meaningful when building a shared object.
662 It marks the object so that its runtime initialization will
663 occur before the runtime initialization of any other objects
664 brought into the process at the same time. Similarly the
665 runtime finalization of the object will occur after the
666 runtime finalization of any other objects.
669 Marks the object that its symbol table interposes before all
670 symbols but the primary executable.
673 When generating an executable or shared library, mark it to
674 tell the dynamic linker to defer function call resolution to
675 the point when the function is called (lazy binding), rather
676 than at load time. Lazy binding is the default.
679 Marks the object that its filters be processed immediately at
683 Allows multiple definitions.
686 Disables multiple reloc sections combining.
689 Disables production of copy relocs.
692 Marks the object that the search for dependencies of this
693 object will ignore any default library search paths.
696 Marks the object shouldn't be unloaded at runtime.
699 Marks the object not available to `dlopen'.
702 Marks the object can not be dumped by `dldump'.
705 Marks the object as not requiring executable stack.
708 Don't create an ELF `PT_GNU_RELRO' segment header in the
712 When generating an executable or shared library, mark it to
713 tell the dynamic linker to resolve all symbols when the
714 program is started, or when the shared library is linked to
715 using dlopen, instead of deferring function call resolution
716 to the point when the function is first called.
719 Marks the object may contain $ORIGIN.
722 Create an ELF `PT_GNU_RELRO' segment header in the object.
724 `max-page-size=VALUE'
725 Set the emulation maximum page size to VALUE.
727 `common-page-size=VALUE'
728 Set the emulation common page size to VALUE.
731 Other keywords are ignored for Solaris compatibility.
734 `--start-group ARCHIVES --end-group'
735 The ARCHIVES should be a list of archive files. They may be
736 either explicit file names, or `-l' options.
738 The specified archives are searched repeatedly until no new
739 undefined references are created. Normally, an archive is
740 searched only once in the order that it is specified on the
741 command line. If a symbol in that archive is needed to resolve an
742 undefined symbol referred to by an object in an archive that
743 appears later on the command line, the linker would not be able to
744 resolve that reference. By grouping the archives, they all be
745 searched repeatedly until all possible references are resolved.
747 Using this option has a significant performance cost. It is best
748 to use it only when there are unavoidable circular references
749 between two or more archives.
751 `--accept-unknown-input-arch'
752 `--no-accept-unknown-input-arch'
753 Tells the linker to accept input files whose architecture cannot be
754 recognised. The assumption is that the user knows what they are
755 doing and deliberately wants to link in these unknown input files.
756 This was the default behaviour of the linker, before release
757 2.14. The default behaviour from release 2.14 onwards is to
758 reject such input files, and so the `--accept-unknown-input-arch'
759 option has been added to restore the old behaviour.
763 This option affects ELF DT_NEEDED tags for dynamic libraries
764 mentioned on the command line after the `--as-needed' option.
765 Normally, the linker will add a DT_NEEDED tag for each dynamic
766 library mentioned on the command line, regardless of whether the
767 library is actually needed. `--as-needed' causes DT_NEEDED tags
768 to only be emitted for libraries that satisfy some symbol
769 reference from regular objects which is undefined at the point
770 that the library was linked. `--no-as-needed' restores the
775 This option affects the treatment of dynamic libraries from ELF
776 DT_NEEDED tags in dynamic libraries mentioned on the command line
777 after the `--no-add-needed' option. Normally, the linker will add
778 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
779 `--no-add-needed' causes DT_NEEDED tags will never be emitted for
780 those libraries from DT_NEEDED tags. `--add-needed' restores the
784 This option is ignored for SunOS compatibility.
789 Link against dynamic libraries. This is only meaningful on
790 platforms for which shared libraries are supported. This option
791 is normally the default on such platforms. The different variants
792 of this option are for compatibility with various systems. You
793 may use this option multiple times on the command line: it affects
794 library searching for `-l' options which follow it.
797 Set the `DF_1_GROUP' flag in the `DT_FLAGS_1' entry in the dynamic
798 section. This causes the runtime linker to handle lookups in this
799 object and its dependencies to be performed only inside the group.
800 `--unresolved-symbols=report-all' is implied. This option is only
801 meaningful on ELF platforms which support shared libraries.
807 Do not link against shared libraries. This is only meaningful on
808 platforms for which shared libraries are supported. The different
809 variants of this option are for compatibility with various
810 systems. You may use this option multiple times on the command
811 line: it affects library searching for `-l' options which follow
812 it. This option also implies `--unresolved-symbols=report-all'.
813 This option can be used with `-shared'. Doing so means that a
814 shared library is being created but that all of the library's
815 external references must be resolved by pulling in entries from
819 When creating a shared library, bind references to global symbols
820 to the definition within the shared library, if any. Normally, it
821 is possible for a program linked against a shared library to
822 override the definition within the shared library. This option is
823 only meaningful on ELF platforms which support shared libraries.
825 `-Bsymbolic-functions'
826 When creating a shared library, bind references to global function
827 symbols to the definition within the shared library, if any. This
828 option is only meaningful on ELF platforms which support shared
831 `--dynamic-list=DYNAMIC-LIST-FILE'
832 Specify the name of a dynamic list file to the linker. This is
833 typically used when creating shared libraries to specify a list of
834 global symbols whose references shouldn't be bound to the
835 definition within the shared library, or creating dynamically
836 linked executables to specify a list of symbols which should be
837 added to the symbol table in the executable. This option is only
838 meaningful on ELF platforms which support shared libraries.
840 The format of the dynamic list is the same as the version node
841 without scope and node name. See *Note VERSION:: for more
844 `--dynamic-list-data'
845 Include all global data symbols to the dynamic list.
847 `--dynamic-list-cpp-new'
848 Provide the builtin dynamic list for C++ operator new and delete.
849 It is mainly useful for building shared libstdc++.
851 `--dynamic-list-cpp-typeinfo'
852 Provide the builtin dynamic list for C++ runtime type
856 `--no-check-sections'
857 Asks the linker _not_ to check section addresses after they have
858 been assigned to see if there are any overlaps. Normally the
859 linker will perform this check, and if it finds any overlaps it
860 will produce suitable error messages. The linker does know about,
861 and does make allowances for sections in overlays. The default
862 behaviour can be restored by using the command line switch
866 Output a cross reference table. If a linker map file is being
867 generated, the cross reference table is printed to the map file.
868 Otherwise, it is printed on the standard output.
870 The format of the table is intentionally simple, so that it may be
871 easily processed by a script if necessary. The symbols are
872 printed out, sorted by name. For each symbol, a list of file
873 names is given. If the symbol is defined, the first file listed
874 is the location of the definition. The remaining files contain
875 references to the symbol.
878 This option inhibits the assignment of addresses to common symbols.
879 The script command `INHIBIT_COMMON_ALLOCATION' has the same effect.
880 *Note Miscellaneous Commands::.
882 The `--no-define-common' option allows decoupling the decision to
883 assign addresses to Common symbols from the choice of the output
884 file type; otherwise a non-Relocatable output type forces
885 assigning addresses to Common symbols. Using `--no-define-common'
886 allows Common symbols that are referenced from a shared library to
887 be assigned addresses only in the main program. This eliminates
888 the unused duplicate space in the shared library, and also
889 prevents any possible confusion over resolving to the wrong
890 duplicate when there are many dynamic modules with specialized
891 search paths for runtime symbol resolution.
893 `--defsym SYMBOL=EXPRESSION'
894 Create a global symbol in the output file, containing the absolute
895 address given by EXPRESSION. You may use this option as many
896 times as necessary to define multiple symbols in the command line.
897 A limited form of arithmetic is supported for the EXPRESSION in
898 this context: you may give a hexadecimal constant or the name of
899 an existing symbol, or use `+' and `-' to add or subtract
900 hexadecimal constants or symbols. If you need more elaborate
901 expressions, consider using the linker command language from a
902 script (*note Assignment: Symbol Definitions: Assignments.).
903 _Note:_ there should be no white space between SYMBOL, the equals
904 sign ("<=>"), and EXPRESSION.
908 These options control whether to demangle symbol names in error
909 messages and other output. When the linker is told to demangle,
910 it tries to present symbol names in a readable fashion: it strips
911 leading underscores if they are used by the object file format,
912 and converts C++ mangled symbol names into user readable names.
913 Different compilers have different mangling styles. The optional
914 demangling style argument can be used to choose an appropriate
915 demangling style for your compiler. The linker will demangle by
916 default unless the environment variable `COLLECT_NO_DEMANGLE' is
917 set. These options may be used to override the default.
919 `--dynamic-linker FILE'
920 Set the name of the dynamic linker. This is only meaningful when
921 generating dynamically linked ELF executables. The default dynamic
922 linker is normally correct; don't use this unless you know what
926 `--no-fatal-warnings'
927 Treat all warnings as errors. The default behaviour can be
928 restored with the option `--no-fatal-warnings'.
931 Make sure that an output file has a .exe suffix.
933 If a successfully built fully linked output file does not have a
934 `.exe' or `.dll' suffix, this option forces the linker to copy the
935 output file to one of the same name with a `.exe' suffix. This
936 option is useful when using unmodified Unix makefiles on a
937 Microsoft Windows host, since some versions of Windows won't run
938 an image unless it ends in a `.exe' suffix.
942 Enable garbage collection of unused input sections. It is ignored
943 on targets that do not support this option. The default behaviour
944 (of not performing this garbage collection) can be restored by
945 specifying `--no-gc-sections' on the command line.
947 `--gc-sections' decides which input sections are used by examining
948 symbols and relocations. The section containing the entry symbol
949 and all sections containing symbols undefined on the command-line
950 will be kept, as will sections containing symbols referenced by
951 dynamic objects. Note that when building shared libraries, the
952 linker must assume that any visible symbol is referenced. Once
953 this initial set of sections has been determined, the linker
954 recursively marks as used any section referenced by their
955 relocations. See `--entry' and `--undefined'.
957 This option can be set when doing a partial link (enabled with
958 option `-r'). In this case the root of symbols kept must be
959 explicitely specified either by an `--entry' or `--undefined'
960 option or by a `ENTRY' command in the linker script.
962 `--print-gc-sections'
963 `--no-print-gc-sections'
964 List all sections removed by garbage collection. The listing is
965 printed on stderr. This option is only effective if garbage
966 collection has been enabled via the `--gc-sections') option. The
967 default behaviour (of not listing the sections that are removed)
968 can be restored by specifying `--no-print-gc-sections' on the
972 Print a summary of the command-line options on the standard output
976 Print a summary of all target specific options on the standard
980 Print a link map to the file MAPFILE. See the description of the
984 `ld' normally optimizes for speed over memory usage by caching the
985 symbol tables of input files in memory. This option tells `ld' to
986 instead optimize for memory usage, by rereading the symbol tables
987 as necessary. This may be required if `ld' runs out of memory
988 space while linking a large executable.
992 Report unresolved symbol references from regular object files.
993 This is done even if the linker is creating a non-symbolic shared
994 library. The switch `--[no-]allow-shlib-undefined' controls the
995 behaviour for reporting unresolved references found in shared
996 libraries being linked in.
998 `--allow-multiple-definition'
1000 Normally when a symbol is defined multiple times, the linker will
1001 report a fatal error. These options allow multiple definitions and
1002 the first definition will be used.
1004 `--allow-shlib-undefined'
1005 `--no-allow-shlib-undefined'
1006 Allows (the default) or disallows undefined symbols in shared
1007 libraries. This switch is similar to `--no-undefined' except that
1008 it determines the behaviour when the undefined symbols are in a
1009 shared library rather than a regular object file. It does not
1010 affect how undefined symbols in regular object files are handled.
1012 The reason that `--allow-shlib-undefined' is the default is that
1013 the shared library being specified at link time may not be the
1014 same as the one that is available at load time, so the symbols
1015 might actually be resolvable at load time. Plus there are some
1016 systems, (eg BeOS) where undefined symbols in shared libraries is
1017 normal. (The kernel patches them at load time to select which
1018 function is most appropriate for the current architecture. This
1019 is used for example to dynamically select an appropriate memset
1020 function). Apparently it is also normal for HPPA shared libraries
1021 to have undefined symbols.
1023 `--no-undefined-version'
1024 Normally when a symbol has an undefined version, the linker will
1025 ignore it. This option disallows symbols with undefined version
1026 and a fatal error will be issued instead.
1029 Create and use a default symbol version (the soname) for
1030 unversioned exported symbols.
1032 `--default-imported-symver'
1033 Create and use a default symbol version (the soname) for
1034 unversioned imported symbols.
1036 `--no-warn-mismatch'
1037 Normally `ld' will give an error if you try to link together input
1038 files that are mismatched for some reason, perhaps because they
1039 have been compiled for different processors or for different
1040 endiannesses. This option tells `ld' that it should silently
1041 permit such possible errors. This option should only be used with
1042 care, in cases when you have taken some special action that
1043 ensures that the linker errors are inappropriate.
1045 `--no-warn-search-mismatch'
1046 Normally `ld' will give a warning if it finds an incompatible
1047 library during a library search. This option silences the warning.
1049 `--no-whole-archive'
1050 Turn off the effect of the `--whole-archive' option for subsequent
1054 Retain the executable output file whenever it is still usable.
1055 Normally, the linker will not produce an output file if it
1056 encounters errors during the link process; it exits without
1057 writing an output file when it issues any error whatsoever.
1060 Only search library directories explicitly specified on the
1061 command line. Library directories specified in linker scripts
1062 (including linker scripts specified on the command line) are
1065 `--oformat OUTPUT-FORMAT'
1066 `ld' may be configured to support more than one kind of object
1067 file. If your `ld' is configured this way, you can use the
1068 `--oformat' option to specify the binary format for the output
1069 object file. Even when `ld' is configured to support alternative
1070 object formats, you don't usually need to specify this, as `ld'
1071 should be configured to produce as a default output format the most
1072 usual format on each machine. OUTPUT-FORMAT is a text string, the
1073 name of a particular format supported by the BFD libraries. (You
1074 can list the available binary formats with `objdump -i'.) The
1075 script command `OUTPUT_FORMAT' can also specify the output format,
1076 but this option overrides it. *Note BFD::.
1080 Create a position independent executable. This is currently only
1081 supported on ELF platforms. Position independent executables are
1082 similar to shared libraries in that they are relocated by the
1083 dynamic linker to the virtual address the OS chooses for them
1084 (which can vary between invocations). Like normal dynamically
1085 linked executables they can be executed and symbols defined in the
1086 executable cannot be overridden by shared libraries.
1089 This option is ignored for Linux compatibility.
1092 This option is ignored for SVR4 compatibility.
1095 An option with machine dependent effects. This option is only
1096 supported on a few targets. *Note `ld' and the H8/300: H8/300.
1097 *Note `ld' and the Intel 960 family: i960. *Note `ld' and Xtensa
1098 Processors: Xtensa. *Note `ld' and the 68HC11 and 68HC12:
1099 M68HC11/68HC12. *Note `ld' and PowerPC 32-bit ELF Support:
1102 On some platforms, the `--relax' option performs global
1103 optimizations that become possible when the linker resolves
1104 addressing in the program, such as relaxing address modes and
1105 synthesizing new instructions in the output object file.
1107 On some platforms these link time global optimizations may make
1108 symbolic debugging of the resulting executable impossible. This
1109 is known to be the case for the Matsushita MN10200 and MN10300
1110 family of processors.
1112 On platforms where this is not supported, `--relax' is accepted,
1115 `--retain-symbols-file FILENAME'
1116 Retain _only_ the symbols listed in the file FILENAME, discarding
1117 all others. FILENAME is simply a flat file, with one symbol name
1118 per line. This option is especially useful in environments (such
1119 as VxWorks) where a large global symbol table is accumulated
1120 gradually, to conserve run-time memory.
1122 `--retain-symbols-file' does _not_ discard undefined symbols, or
1123 symbols needed for relocations.
1125 You may only specify `--retain-symbols-file' once in the command
1126 line. It overrides `-s' and `-S'.
1129 Add a directory to the runtime library search path. This is used
1130 when linking an ELF executable with shared objects. All `-rpath'
1131 arguments are concatenated and passed to the runtime linker, which
1132 uses them to locate shared objects at runtime. The `-rpath'
1133 option is also used when locating shared objects which are needed
1134 by shared objects explicitly included in the link; see the
1135 description of the `-rpath-link' option. If `-rpath' is not used
1136 when linking an ELF executable, the contents of the environment
1137 variable `LD_RUN_PATH' will be used if it is defined.
1139 The `-rpath' option may also be used on SunOS. By default, on
1140 SunOS, the linker will form a runtime search patch out of all the
1141 `-L' options it is given. If a `-rpath' option is used, the
1142 runtime search path will be formed exclusively using the `-rpath'
1143 options, ignoring the `-L' options. This can be useful when using
1144 gcc, which adds many `-L' options which may be on NFS mounted file
1147 For compatibility with other ELF linkers, if the `-R' option is
1148 followed by a directory name, rather than a file name, it is
1149 treated as the `-rpath' option.
1152 When using ELF or SunOS, one shared library may require another.
1153 This happens when an `ld -shared' link includes a shared library
1154 as one of the input files.
1156 When the linker encounters such a dependency when doing a
1157 non-shared, non-relocatable link, it will automatically try to
1158 locate the required shared library and include it in the link, if
1159 it is not included explicitly. In such a case, the `-rpath-link'
1160 option specifies the first set of directories to search. The
1161 `-rpath-link' option may specify a sequence of directory names
1162 either by specifying a list of names separated by colons, or by
1163 appearing multiple times.
1165 This option should be used with caution as it overrides the search
1166 path that may have been hard compiled into a shared library. In
1167 such a case it is possible to use unintentionally a different
1168 search path than the runtime linker would do.
1170 The linker uses the following search paths to locate required
1172 1. Any directories specified by `-rpath-link' options.
1174 2. Any directories specified by `-rpath' options. The difference
1175 between `-rpath' and `-rpath-link' is that directories
1176 specified by `-rpath' options are included in the executable
1177 and used at runtime, whereas the `-rpath-link' option is only
1178 effective at link time. Searching `-rpath' in this way is
1179 only supported by native linkers and cross linkers which have
1180 been configured with the `--with-sysroot' option.
1182 3. On an ELF system, for native linkers, if the `-rpath' and
1183 `-rpath-link' options were not used, search the contents of
1184 the environment variable `LD_RUN_PATH'.
1186 4. On SunOS, if the `-rpath' option was not used, search any
1187 directories specified using `-L' options.
1189 5. For a native linker, the search the contents of the
1190 environment variable `LD_LIBRARY_PATH'.
1192 6. For a native ELF linker, the directories in `DT_RUNPATH' or
1193 `DT_RPATH' of a shared library are searched for shared
1194 libraries needed by it. The `DT_RPATH' entries are ignored if
1195 `DT_RUNPATH' entries exist.
1197 7. The default directories, normally `/lib' and `/usr/lib'.
1199 8. For a native linker on an ELF system, if the file
1200 `/etc/ld.so.conf' exists, the list of directories found in
1203 If the required shared library is not found, the linker will issue
1204 a warning and continue with the link.
1208 Create a shared library. This is currently only supported on ELF,
1209 XCOFF and SunOS platforms. On SunOS, the linker will
1210 automatically create a shared library if the `-e' option is not
1211 used and there are undefined symbols in the link.
1213 `--sort-common [= ascending | descending]'
1214 This option tells `ld' to sort the common symbols by alignment in
1215 ascending or descending order when it places them in the
1216 appropriate output sections. The symbol alignments considered are
1217 sixteen-byte or larger, eight-byte, four-byte, two-byte, and
1218 one-byte. This is to prevent gaps between symbols due to alignment
1219 constraints. If no sorting order is specified, then descending
1222 `--sort-section name'
1223 This option will apply `SORT_BY_NAME' to all wildcard section
1224 patterns in the linker script.
1226 `--sort-section alignment'
1227 This option will apply `SORT_BY_ALIGNMENT' to all wildcard section
1228 patterns in the linker script.
1230 `--split-by-file [SIZE]'
1231 Similar to `--split-by-reloc' but creates a new output section for
1232 each input file when SIZE is reached. SIZE defaults to a size of
1235 `--split-by-reloc [COUNT]'
1236 Tries to creates extra sections in the output file so that no
1237 single output section in the file contains more than COUNT
1238 relocations. This is useful when generating huge relocatable
1239 files for downloading into certain real time kernels with the COFF
1240 object file format; since COFF cannot represent more than 65535
1241 relocations in a single section. Note that this will fail to work
1242 with object file formats which do not support arbitrary sections.
1243 The linker will not split up individual input sections for
1244 redistribution, so if a single input section contains more than
1245 COUNT relocations one output section will contain that many
1246 relocations. COUNT defaults to a value of 32768.
1249 Compute and display statistics about the operation of the linker,
1250 such as execution time and memory usage.
1252 `--sysroot=DIRECTORY'
1253 Use DIRECTORY as the location of the sysroot, overriding the
1254 configure-time default. This option is only supported by linkers
1255 that were configured using `--with-sysroot'.
1257 `--traditional-format'
1258 For some targets, the output of `ld' is different in some ways from
1259 the output of some existing linker. This switch requests `ld' to
1260 use the traditional format instead.
1262 For example, on SunOS, `ld' combines duplicate entries in the
1263 symbol string table. This can reduce the size of an output file
1264 with full debugging information by over 30 percent.
1265 Unfortunately, the SunOS `dbx' program can not read the resulting
1266 program (`gdb' has no trouble). The `--traditional-format' switch
1267 tells `ld' to not combine duplicate entries.
1269 `--section-start SECTIONNAME=ORG'
1270 Locate a section in the output file at the absolute address given
1271 by ORG. You may use this option as many times as necessary to
1272 locate multiple sections in the command line. ORG must be a
1273 single hexadecimal integer; for compatibility with other linkers,
1274 you may omit the leading `0x' usually associated with hexadecimal
1275 values. _Note:_ there should be no white space between
1276 SECTIONNAME, the equals sign ("<=>"), and ORG.
1281 Same as -section-start, with `.bss', `.data' or `.text' as the
1284 `--unresolved-symbols=METHOD'
1285 Determine how to handle unresolved symbols. There are four
1286 possible values for `method':
1289 Do not report any unresolved symbols.
1292 Report all unresolved symbols. This is the default.
1294 `ignore-in-object-files'
1295 Report unresolved symbols that are contained in shared
1296 libraries, but ignore them if they come from regular object
1299 `ignore-in-shared-libs'
1300 Report unresolved symbols that come from regular object
1301 files, but ignore them if they come from shared libraries.
1302 This can be useful when creating a dynamic binary and it is
1303 known that all the shared libraries that it should be
1304 referencing are included on the linker's command line.
1306 The behaviour for shared libraries on their own can also be
1307 controlled by the `--[no-]allow-shlib-undefined' option.
1309 Normally the linker will generate an error message for each
1310 reported unresolved symbol but the option
1311 `--warn-unresolved-symbols' can change this to a warning.
1315 Display the version number for `ld' and list the linker emulations
1316 supported. Display which input files can and cannot be opened.
1317 Display the linker script being used by the linker.
1319 `--version-script=VERSION-SCRIPTFILE'
1320 Specify the name of a version script to the linker. This is
1321 typically used when creating shared libraries to specify
1322 additional information about the version hierarchy for the library
1323 being created. This option is only meaningful on ELF platforms
1324 which support shared libraries. *Note VERSION::.
1327 Warn when a common symbol is combined with another common symbol
1328 or with a symbol definition. Unix linkers allow this somewhat
1329 sloppy practise, but linkers on some other operating systems do
1330 not. This option allows you to find potential problems from
1331 combining global symbols. Unfortunately, some C libraries use
1332 this practise, so you may get some warnings about symbols in the
1333 libraries as well as in your programs.
1335 There are three kinds of global symbols, illustrated here by C
1339 A definition, which goes in the initialized data section of
1343 An undefined reference, which does not allocate space. There
1344 must be either a definition or a common symbol for the
1348 A common symbol. If there are only (one or more) common
1349 symbols for a variable, it goes in the uninitialized data
1350 area of the output file. The linker merges multiple common
1351 symbols for the same variable into a single symbol. If they
1352 are of different sizes, it picks the largest size. The
1353 linker turns a common symbol into a declaration, if there is
1354 a definition of the same variable.
1356 The `--warn-common' option can produce five kinds of warnings.
1357 Each warning consists of a pair of lines: the first describes the
1358 symbol just encountered, and the second describes the previous
1359 symbol encountered with the same name. One or both of the two
1360 symbols will be a common symbol.
1362 1. Turning a common symbol into a reference, because there is
1363 already a definition for the symbol.
1364 FILE(SECTION): warning: common of `SYMBOL'
1365 overridden by definition
1366 FILE(SECTION): warning: defined here
1368 2. Turning a common symbol into a reference, because a later
1369 definition for the symbol is encountered. This is the same
1370 as the previous case, except that the symbols are encountered
1371 in a different order.
1372 FILE(SECTION): warning: definition of `SYMBOL'
1374 FILE(SECTION): warning: common is here
1376 3. Merging a common symbol with a previous same-sized common
1378 FILE(SECTION): warning: multiple common
1380 FILE(SECTION): warning: previous common is here
1382 4. Merging a common symbol with a previous larger common symbol.
1383 FILE(SECTION): warning: common of `SYMBOL'
1384 overridden by larger common
1385 FILE(SECTION): warning: larger common is here
1387 5. Merging a common symbol with a previous smaller common
1388 symbol. This is the same as the previous case, except that
1389 the symbols are encountered in a different order.
1390 FILE(SECTION): warning: common of `SYMBOL'
1391 overriding smaller common
1392 FILE(SECTION): warning: smaller common is here
1394 `--warn-constructors'
1395 Warn if any global constructors are used. This is only useful for
1396 a few object file formats. For formats like COFF or ELF, the
1397 linker can not detect the use of global constructors.
1399 `--warn-multiple-gp'
1400 Warn if multiple global pointer values are required in the output
1401 file. This is only meaningful for certain processors, such as the
1402 Alpha. Specifically, some processors put large-valued constants
1403 in a special section. A special register (the global pointer)
1404 points into the middle of this section, so that constants can be
1405 loaded efficiently via a base-register relative addressing mode.
1406 Since the offset in base-register relative mode is fixed and
1407 relatively small (e.g., 16 bits), this limits the maximum size of
1408 the constant pool. Thus, in large programs, it is often necessary
1409 to use multiple global pointer values in order to be able to
1410 address all possible constants. This option causes a warning to
1411 be issued whenever this case occurs.
1414 Only warn once for each undefined symbol, rather than once per
1415 module which refers to it.
1417 `--warn-section-align'
1418 Warn if the address of an output section is changed because of
1419 alignment. Typically, the alignment will be set by an input
1420 section. The address will only be changed if it not explicitly
1421 specified; that is, if the `SECTIONS' command does not specify a
1422 start address for the section (*note SECTIONS::).
1424 `--warn-shared-textrel'
1425 Warn if the linker adds a DT_TEXTREL to a shared object.
1427 `--warn-unresolved-symbols'
1428 If the linker is going to report an unresolved symbol (see the
1429 option `--unresolved-symbols') it will normally generate an error.
1430 This option makes it generate a warning instead.
1432 `--error-unresolved-symbols'
1433 This restores the linker's default behaviour of generating errors
1434 when it is reporting unresolved symbols.
1437 For each archive mentioned on the command line after the
1438 `--whole-archive' option, include every object file in the archive
1439 in the link, rather than searching the archive for the required
1440 object files. This is normally used to turn an archive file into
1441 a shared library, forcing every object to be included in the
1442 resulting shared library. This option may be used more than once.
1444 Two notes when using this option from gcc: First, gcc doesn't know
1445 about this option, so you have to use `-Wl,-whole-archive'.
1446 Second, don't forget to use `-Wl,-no-whole-archive' after your
1447 list of archives, because gcc will add its own list of archives to
1448 your link and you may not want this flag to affect those as well.
1451 Use a wrapper function for SYMBOL. Any undefined reference to
1452 SYMBOL will be resolved to `__wrap_SYMBOL'. Any undefined
1453 reference to `__real_SYMBOL' will be resolved to SYMBOL.
1455 This can be used to provide a wrapper for a system function. The
1456 wrapper function should be called `__wrap_SYMBOL'. If it wishes
1457 to call the system function, it should call `__real_SYMBOL'.
1459 Here is a trivial example:
1462 __wrap_malloc (size_t c)
1464 printf ("malloc called with %zu\n", c);
1465 return __real_malloc (c);
1468 If you link other code with this file using `--wrap malloc', then
1469 all calls to `malloc' will call the function `__wrap_malloc'
1470 instead. The call to `__real_malloc' in `__wrap_malloc' will call
1471 the real `malloc' function.
1473 You may wish to provide a `__real_malloc' function as well, so that
1474 links without the `--wrap' option will succeed. If you do this,
1475 you should not put the definition of `__real_malloc' in the same
1476 file as `__wrap_malloc'; if you do, the assembler may resolve the
1477 call before the linker has a chance to wrap it to `malloc'.
1480 Request creation of `.eh_frame_hdr' section and ELF
1481 `PT_GNU_EH_FRAME' segment header.
1483 `--enable-new-dtags'
1484 `--disable-new-dtags'
1485 This linker can create the new dynamic tags in ELF. But the older
1486 ELF systems may not understand them. If you specify
1487 `--enable-new-dtags', the dynamic tags will be created as needed.
1488 If you specify `--disable-new-dtags', no new dynamic tags will be
1489 created. By default, the new dynamic tags are not created. Note
1490 that those options are only available for ELF systems.
1492 `--hash-size=NUMBER'
1493 Set the default size of the linker's hash tables to a prime number
1494 close to NUMBER. Increasing this value can reduce the length of
1495 time it takes the linker to perform its tasks, at the expense of
1496 increasing the linker's memory requirements. Similarly reducing
1497 this value can reduce the memory requirements at the expense of
1500 `--hash-style=STYLE'
1501 Set the type of linker's hash table(s). STYLE can be either
1502 `sysv' for classic ELF `.hash' section, `gnu' for new style GNU
1503 `.gnu.hash' section or `both' for both the classic ELF `.hash' and
1504 new style GNU `.gnu.hash' hash tables. The default is `sysv'.
1506 `--reduce-memory-overheads'
1507 This option reduces memory requirements at ld runtime, at the
1508 expense of linking speed. This was introduced to select the old
1509 O(n^2) algorithm for link map file generation, rather than the new
1510 O(n) algorithm which uses about 40% more memory for symbol storage.
1512 Another effect of the switch is to set the default hash table size
1513 to 1021, which again saves memory at the cost of lengthening the
1514 linker's run time. This is not done however if the `--hash-size'
1515 switch has been used.
1517 The `--reduce-memory-overheads' switch may be also be used to
1518 enable other tradeoffs in future versions of the linker.
1522 Request creation of `.note.gnu.build-id' ELF note section. The
1523 contents of the note are unique bits identifying this linked file.
1524 STYLE can be `uuid' to use 128 random bits, `sha1' to use a
1525 160-bit SHA1 hash on the normative parts of the output contents,
1526 `md5' to use a 128-bit MD5 hash on the normative parts of the
1527 output contents, or `0xHEXSTRING' to use a chosen bit string
1528 specified as an even number of hexadecimal digits (`-' and `:'
1529 characters between digit pairs are ignored). If STYLE is omitted,
1532 The `md5' and `sha1' styles produces an identifier that is always
1533 the same in an identical output file, but will be unique among all
1534 nonidentical output files. It is not intended to be compared as a
1535 checksum for the file's contents. A linked file may be changed
1536 later by other tools, but the build ID bit string identifying the
1537 original linked file does not change.
1539 Passing `none' for STYLE disables the setting from any
1540 `--build-id' options earlier on the command line.
1542 2.1.1 Options Specific to i386 PE Targets
1543 -----------------------------------------
1545 The i386 PE linker supports the `-shared' option, which causes the
1546 output to be a dynamically linked library (DLL) instead of a normal
1547 executable. You should name the output `*.dll' when you use this
1548 option. In addition, the linker fully supports the standard `*.def'
1549 files, which may be specified on the linker command line like an object
1550 file (in fact, it should precede archives it exports symbols from, to
1551 ensure that they get linked in, just like a normal object file).
1553 In addition to the options common to all targets, the i386 PE linker
1554 support additional command line options that are specific to the i386
1555 PE target. Options that take values may be separated from their values
1556 by either a space or an equals sign.
1558 `--add-stdcall-alias'
1559 If given, symbols with a stdcall suffix (@NN) will be exported
1560 as-is and also with the suffix stripped. [This option is specific
1561 to the i386 PE targeted port of the linker]
1564 Use FILE as the name of a file in which to save the base addresses
1565 of all the relocations needed for generating DLLs with `dlltool'.
1566 [This is an i386 PE specific option]
1569 Create a DLL instead of a regular executable. You may also use
1570 `-shared' or specify a `LIBRARY' in a given `.def' file. [This
1571 option is specific to the i386 PE targeted port of the linker]
1573 `--enable-stdcall-fixup'
1574 `--disable-stdcall-fixup'
1575 If the link finds a symbol that it cannot resolve, it will attempt
1576 to do "fuzzy linking" by looking for another defined symbol that
1577 differs only in the format of the symbol name (cdecl vs stdcall)
1578 and will resolve that symbol by linking to the match. For
1579 example, the undefined symbol `_foo' might be linked to the
1580 function `_foo@12', or the undefined symbol `_bar@16' might be
1581 linked to the function `_bar'. When the linker does this, it
1582 prints a warning, since it normally should have failed to link,
1583 but sometimes import libraries generated from third-party dlls may
1584 need this feature to be usable. If you specify
1585 `--enable-stdcall-fixup', this feature is fully enabled and
1586 warnings are not printed. If you specify
1587 `--disable-stdcall-fixup', this feature is disabled and such
1588 mismatches are considered to be errors. [This option is specific
1589 to the i386 PE targeted port of the linker]
1591 `--export-all-symbols'
1592 If given, all global symbols in the objects used to build a DLL
1593 will be exported by the DLL. Note that this is the default if
1594 there otherwise wouldn't be any exported symbols. When symbols are
1595 explicitly exported via DEF files or implicitly exported via
1596 function attributes, the default is to not export anything else
1597 unless this option is given. Note that the symbols `DllMain@12',
1598 `DllEntryPoint@0', `DllMainCRTStartup@12', and `impure_ptr' will
1599 not be automatically exported. Also, symbols imported from other
1600 DLLs will not be re-exported, nor will symbols specifying the
1601 DLL's internal layout such as those beginning with `_head_' or
1602 ending with `_iname'. In addition, no symbols from `libgcc',
1603 `libstd++', `libmingw32', or `crtX.o' will be exported. Symbols
1604 whose names begin with `__rtti_' or `__builtin_' will not be
1605 exported, to help with C++ DLLs. Finally, there is an extensive
1606 list of cygwin-private symbols that are not exported (obviously,
1607 this applies on when building DLLs for cygwin targets). These
1608 cygwin-excludes are: `_cygwin_dll_entry@12',
1609 `_cygwin_crt0_common@8', `_cygwin_noncygwin_dll_entry@12',
1610 `_fmode', `_impure_ptr', `cygwin_attach_dll', `cygwin_premain0',
1611 `cygwin_premain1', `cygwin_premain2', `cygwin_premain3', and
1612 `environ'. [This option is specific to the i386 PE targeted port
1615 `--exclude-symbols SYMBOL,SYMBOL,...'
1616 Specifies a list of symbols which should not be automatically
1617 exported. The symbol names may be delimited by commas or colons.
1618 [This option is specific to the i386 PE targeted port of the
1622 Specify the file alignment. Sections in the file will always
1623 begin at file offsets which are multiples of this number. This
1624 defaults to 512. [This option is specific to the i386 PE targeted
1628 `--heap RESERVE,COMMIT'
1629 Specify the number of bytes of memory to reserve (and optionally
1630 commit) to be used as heap for this program. The default is 1Mb
1631 reserved, 4K committed. [This option is specific to the i386 PE
1632 targeted port of the linker]
1634 `--image-base VALUE'
1635 Use VALUE as the base address of your program or dll. This is the
1636 lowest memory location that will be used when your program or dll
1637 is loaded. To reduce the need to relocate and improve performance
1638 of your dlls, each should have a unique base address and not
1639 overlap any other dlls. The default is 0x400000 for executables,
1640 and 0x10000000 for dlls. [This option is specific to the i386 PE
1641 targeted port of the linker]
1644 If given, the stdcall suffixes (@NN) will be stripped from symbols
1645 before they are exported. [This option is specific to the i386 PE
1646 targeted port of the linker]
1648 `--large-address-aware'
1649 If given, the appropriate bit in the "Characteristics" field of
1650 the COFF header is set to indicate that this executable supports
1651 virtual addresses greater than 2 gigabytes. This should be used
1652 in conjunction with the /3GB or /USERVA=VALUE megabytes switch in
1653 the "[operating systems]" section of the BOOT.INI. Otherwise,
1654 this bit has no effect. [This option is specific to PE targeted
1655 ports of the linker]
1657 `--major-image-version VALUE'
1658 Sets the major number of the "image version". Defaults to 1.
1659 [This option is specific to the i386 PE targeted port of the
1662 `--major-os-version VALUE'
1663 Sets the major number of the "os version". Defaults to 4. [This
1664 option is specific to the i386 PE targeted port of the linker]
1666 `--major-subsystem-version VALUE'
1667 Sets the major number of the "subsystem version". Defaults to 4.
1668 [This option is specific to the i386 PE targeted port of the
1671 `--minor-image-version VALUE'
1672 Sets the minor number of the "image version". Defaults to 0.
1673 [This option is specific to the i386 PE targeted port of the
1676 `--minor-os-version VALUE'
1677 Sets the minor number of the "os version". Defaults to 0. [This
1678 option is specific to the i386 PE targeted port of the linker]
1680 `--minor-subsystem-version VALUE'
1681 Sets the minor number of the "subsystem version". Defaults to 0.
1682 [This option is specific to the i386 PE targeted port of the
1686 The linker will create the file FILE which will contain a DEF file
1687 corresponding to the DLL the linker is generating. This DEF file
1688 (which should be called `*.def') may be used to create an import
1689 library with `dlltool' or may be used as a reference to
1690 automatically or implicitly exported symbols. [This option is
1691 specific to the i386 PE targeted port of the linker]
1694 The linker will create the file FILE which will contain an import
1695 lib corresponding to the DLL the linker is generating. This import
1696 lib (which should be called `*.dll.a' or `*.a' may be used to link
1697 clients against the generated DLL; this behaviour makes it
1698 possible to skip a separate `dlltool' import library creation step.
1699 [This option is specific to the i386 PE targeted port of the
1702 `--enable-auto-image-base'
1703 Automatically choose the image base for DLLs, unless one is
1704 specified using the `--image-base' argument. By using a hash
1705 generated from the dllname to create unique image bases for each
1706 DLL, in-memory collisions and relocations which can delay program
1707 execution are avoided. [This option is specific to the i386 PE
1708 targeted port of the linker]
1710 `--disable-auto-image-base'
1711 Do not automatically generate a unique image base. If there is no
1712 user-specified image base (`--image-base') then use the platform
1713 default. [This option is specific to the i386 PE targeted port of
1716 `--dll-search-prefix STRING'
1717 When linking dynamically to a dll without an import library,
1718 search for `<string><basename>.dll' in preference to
1719 `lib<basename>.dll'. This behaviour allows easy distinction
1720 between DLLs built for the various "subplatforms": native, cygwin,
1721 uwin, pw, etc. For instance, cygwin DLLs typically use
1722 `--dll-search-prefix=cyg'. [This option is specific to the i386
1723 PE targeted port of the linker]
1725 `--enable-auto-import'
1726 Do sophisticated linking of `_symbol' to `__imp__symbol' for DATA
1727 imports from DLLs, and create the necessary thunking symbols when
1728 building the import libraries with those DATA exports. Note: Use
1729 of the 'auto-import' extension will cause the text section of the
1730 image file to be made writable. This does not conform to the
1731 PE-COFF format specification published by Microsoft.
1733 Note - use of the 'auto-import' extension will also cause read only
1734 data which would normally be placed into the .rdata section to be
1735 placed into the .data section instead. This is in order to work
1736 around a problem with consts that is described here:
1737 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
1739 Using 'auto-import' generally will 'just work' - but sometimes you
1740 may see this message:
1742 "variable '<var>' can't be auto-imported. Please read the
1743 documentation for ld's `--enable-auto-import' for details."
1745 This message occurs when some (sub)expression accesses an address
1746 ultimately given by the sum of two constants (Win32 import tables
1747 only allow one). Instances where this may occur include accesses
1748 to member fields of struct variables imported from a DLL, as well
1749 as using a constant index into an array variable imported from a
1750 DLL. Any multiword variable (arrays, structs, long long, etc) may
1751 trigger this error condition. However, regardless of the exact
1752 data type of the offending exported variable, ld will always
1753 detect it, issue the warning, and exit.
1755 There are several ways to address this difficulty, regardless of
1756 the data type of the exported variable:
1758 One way is to use -enable-runtime-pseudo-reloc switch. This leaves
1759 the task of adjusting references in your client code for runtime
1760 environment, so this method works only when runtime environment
1761 supports this feature.
1763 A second solution is to force one of the 'constants' to be a
1764 variable - that is, unknown and un-optimizable at compile time.
1765 For arrays, there are two possibilities: a) make the indexee (the
1766 array's address) a variable, or b) make the 'constant' index a
1769 extern type extern_array[];
1771 { volatile type *t=extern_array; t[1] }
1775 extern type extern_array[];
1777 { volatile int t=1; extern_array[t] }
1779 For structs (and most other multiword data types) the only option
1780 is to make the struct itself (or the long long, or the ...)
1783 extern struct s extern_struct;
1784 extern_struct.field -->
1785 { volatile struct s *t=&extern_struct; t->field }
1789 extern long long extern_ll;
1791 { volatile long long * local_ll=&extern_ll; *local_ll }
1793 A third method of dealing with this difficulty is to abandon
1794 'auto-import' for the offending symbol and mark it with
1795 `__declspec(dllimport)'. However, in practise that requires using
1796 compile-time #defines to indicate whether you are building a DLL,
1797 building client code that will link to the DLL, or merely
1798 building/linking to a static library. In making the choice
1799 between the various methods of resolving the 'direct address with
1800 constant offset' problem, you should consider typical real-world
1808 void main(int argc, char **argv){
1809 printf("%d\n",arr[1]);
1817 void main(int argc, char **argv){
1818 /* This workaround is for win32 and cygwin; do not "optimize" */
1819 volatile int *parr = arr;
1820 printf("%d\n",parr[1]);
1825 /* Note: auto-export is assumed (no __declspec(dllexport)) */
1826 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
1827 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
1828 #define FOO_IMPORT __declspec(dllimport)
1832 extern FOO_IMPORT int arr[];
1835 void main(int argc, char **argv){
1836 printf("%d\n",arr[1]);
1839 A fourth way to avoid this problem is to re-code your library to
1840 use a functional interface rather than a data interface for the
1841 offending variables (e.g. set_foo() and get_foo() accessor
1842 functions). [This option is specific to the i386 PE targeted port
1845 `--disable-auto-import'
1846 Do not attempt to do sophisticated linking of `_symbol' to
1847 `__imp__symbol' for DATA imports from DLLs. [This option is
1848 specific to the i386 PE targeted port of the linker]
1850 `--enable-runtime-pseudo-reloc'
1851 If your code contains expressions described in -enable-auto-import
1852 section, that is, DATA imports from DLL with non-zero offset, this
1853 switch will create a vector of 'runtime pseudo relocations' which
1854 can be used by runtime environment to adjust references to such
1855 data in your client code. [This option is specific to the i386 PE
1856 targeted port of the linker]
1858 `--disable-runtime-pseudo-reloc'
1859 Do not create pseudo relocations for non-zero offset DATA imports
1860 from DLLs. This is the default. [This option is specific to the
1861 i386 PE targeted port of the linker]
1863 `--enable-extra-pe-debug'
1864 Show additional debug info related to auto-import symbol thunking.
1865 [This option is specific to the i386 PE targeted port of the
1868 `--section-alignment'
1869 Sets the section alignment. Sections in memory will always begin
1870 at addresses which are a multiple of this number. Defaults to
1871 0x1000. [This option is specific to the i386 PE targeted port of
1875 `--stack RESERVE,COMMIT'
1876 Specify the number of bytes of memory to reserve (and optionally
1877 commit) to be used as stack for this program. The default is 2Mb
1878 reserved, 4K committed. [This option is specific to the i386 PE
1879 targeted port of the linker]
1882 `--subsystem WHICH:MAJOR'
1883 `--subsystem WHICH:MAJOR.MINOR'
1884 Specifies the subsystem under which your program will execute. The
1885 legal values for WHICH are `native', `windows', `console',
1886 `posix', and `xbox'. You may optionally set the subsystem version
1887 also. Numeric values are also accepted for WHICH. [This option
1888 is specific to the i386 PE targeted port of the linker]
1891 2.1.2 Options specific to Motorola 68HC11 and 68HC12 targets
1892 ------------------------------------------------------------
1894 The 68HC11 and 68HC12 linkers support specific options to control the
1895 memory bank switching mapping and trampoline code generation.
1898 This option disables the generation of trampoline. By default a
1899 trampoline is generated for each far function which is called
1900 using a `jsr' instruction (this happens when a pointer to a far
1903 `--bank-window NAME'
1904 This option indicates to the linker the name of the memory region
1905 in the `MEMORY' specification that describes the memory bank
1906 window. The definition of such region is then used by the linker
1907 to compute paging and addresses within the memory window.
1910 2.1.3 Options specific to Motorola 68K target
1911 ---------------------------------------------
1913 The following options are supported to control handling of GOT
1914 generation when linking for 68K targets.
1917 This option tells the linker which GOT generation scheme to use.
1918 TYPE should be one of `single', `negative', `multigot' or
1919 `target'. For more information refer to the Info entry for `ld'.
1923 File: ld.info, Node: Environment, Prev: Options, Up: Invocation
1925 2.2 Environment Variables
1926 =========================
1928 You can change the behaviour of `ld' with the environment variables
1929 `GNUTARGET', `LDEMULATION' and `COLLECT_NO_DEMANGLE'.
1931 `GNUTARGET' determines the input-file object format if you don't use
1932 `-b' (or its synonym `--format'). Its value should be one of the BFD
1933 names for an input format (*note BFD::). If there is no `GNUTARGET' in
1934 the environment, `ld' uses the natural format of the target. If
1935 `GNUTARGET' is set to `default' then BFD attempts to discover the input
1936 format by examining binary input files; this method often succeeds, but
1937 there are potential ambiguities, since there is no method of ensuring
1938 that the magic number used to specify object-file formats is unique.
1939 However, the configuration procedure for BFD on each system places the
1940 conventional format for that system first in the search-list, so
1941 ambiguities are resolved in favor of convention.
1943 `LDEMULATION' determines the default emulation if you don't use the
1944 `-m' option. The emulation can affect various aspects of linker
1945 behaviour, particularly the default linker script. You can list the
1946 available emulations with the `--verbose' or `-V' options. If the `-m'
1947 option is not used, and the `LDEMULATION' environment variable is not
1948 defined, the default emulation depends upon how the linker was
1951 Normally, the linker will default to demangling symbols. However, if
1952 `COLLECT_NO_DEMANGLE' is set in the environment, then it will default
1953 to not demangling symbols. This environment variable is used in a
1954 similar fashion by the `gcc' linker wrapper program. The default may
1955 be overridden by the `--demangle' and `--no-demangle' options.
1958 File: ld.info, Node: Scripts, Next: Machine Dependent, Prev: Invocation, Up: Top
1963 Every link is controlled by a "linker script". This script is written
1964 in the linker command language.
1966 The main purpose of the linker script is to describe how the
1967 sections in the input files should be mapped into the output file, and
1968 to control the memory layout of the output file. Most linker scripts
1969 do nothing more than this. However, when necessary, the linker script
1970 can also direct the linker to perform many other operations, using the
1971 commands described below.
1973 The linker always uses a linker script. If you do not supply one
1974 yourself, the linker will use a default script that is compiled into the
1975 linker executable. You can use the `--verbose' command line option to
1976 display the default linker script. Certain command line options, such
1977 as `-r' or `-N', will affect the default linker script.
1979 You may supply your own linker script by using the `-T' command line
1980 option. When you do this, your linker script will replace the default
1983 You may also use linker scripts implicitly by naming them as input
1984 files to the linker, as though they were files to be linked. *Note
1985 Implicit Linker Scripts::.
1989 * Basic Script Concepts:: Basic Linker Script Concepts
1990 * Script Format:: Linker Script Format
1991 * Simple Example:: Simple Linker Script Example
1992 * Simple Commands:: Simple Linker Script Commands
1993 * Assignments:: Assigning Values to Symbols
1994 * SECTIONS:: SECTIONS Command
1995 * MEMORY:: MEMORY Command
1996 * PHDRS:: PHDRS Command
1997 * VERSION:: VERSION Command
1998 * Expressions:: Expressions in Linker Scripts
1999 * Implicit Linker Scripts:: Implicit Linker Scripts
2002 File: ld.info, Node: Basic Script Concepts, Next: Script Format, Up: Scripts
2004 3.1 Basic Linker Script Concepts
2005 ================================
2007 We need to define some basic concepts and vocabulary in order to
2008 describe the linker script language.
2010 The linker combines input files into a single output file. The
2011 output file and each input file are in a special data format known as an
2012 "object file format". Each file is called an "object file". The
2013 output file is often called an "executable", but for our purposes we
2014 will also call it an object file. Each object file has, among other
2015 things, a list of "sections". We sometimes refer to a section in an
2016 input file as an "input section"; similarly, a section in the output
2017 file is an "output section".
2019 Each section in an object file has a name and a size. Most sections
2020 also have an associated block of data, known as the "section contents".
2021 A section may be marked as "loadable", which mean that the contents
2022 should be loaded into memory when the output file is run. A section
2023 with no contents may be "allocatable", which means that an area in
2024 memory should be set aside, but nothing in particular should be loaded
2025 there (in some cases this memory must be zeroed out). A section which
2026 is neither loadable nor allocatable typically contains some sort of
2027 debugging information.
2029 Every loadable or allocatable output section has two addresses. The
2030 first is the "VMA", or virtual memory address. This is the address the
2031 section will have when the output file is run. The second is the
2032 "LMA", or load memory address. This is the address at which the
2033 section will be loaded. In most cases the two addresses will be the
2034 same. An example of when they might be different is when a data section
2035 is loaded into ROM, and then copied into RAM when the program starts up
2036 (this technique is often used to initialize global variables in a ROM
2037 based system). In this case the ROM address would be the LMA, and the
2038 RAM address would be the VMA.
2040 You can see the sections in an object file by using the `objdump'
2041 program with the `-h' option.
2043 Every object file also has a list of "symbols", known as the "symbol
2044 table". A symbol may be defined or undefined. Each symbol has a name,
2045 and each defined symbol has an address, among other information. If
2046 you compile a C or C++ program into an object file, you will get a
2047 defined symbol for every defined function and global or static
2048 variable. Every undefined function or global variable which is
2049 referenced in the input file will become an undefined symbol.
2051 You can see the symbols in an object file by using the `nm' program,
2052 or by using the `objdump' program with the `-t' option.
2055 File: ld.info, Node: Script Format, Next: Simple Example, Prev: Basic Script Concepts, Up: Scripts
2057 3.2 Linker Script Format
2058 ========================
2060 Linker scripts are text files.
2062 You write a linker script as a series of commands. Each command is
2063 either a keyword, possibly followed by arguments, or an assignment to a
2064 symbol. You may separate commands using semicolons. Whitespace is
2067 Strings such as file or format names can normally be entered
2068 directly. If the file name contains a character such as a comma which
2069 would otherwise serve to separate file names, you may put the file name
2070 in double quotes. There is no way to use a double quote character in a
2073 You may include comments in linker scripts just as in C, delimited by
2074 `/*' and `*/'. As in C, comments are syntactically equivalent to
2078 File: ld.info, Node: Simple Example, Next: Simple Commands, Prev: Script Format, Up: Scripts
2080 3.3 Simple Linker Script Example
2081 ================================
2083 Many linker scripts are fairly simple.
2085 The simplest possible linker script has just one command:
2086 `SECTIONS'. You use the `SECTIONS' command to describe the memory
2087 layout of the output file.
2089 The `SECTIONS' command is a powerful command. Here we will describe
2090 a simple use of it. Let's assume your program consists only of code,
2091 initialized data, and uninitialized data. These will be in the
2092 `.text', `.data', and `.bss' sections, respectively. Let's assume
2093 further that these are the only sections which appear in your input
2096 For this example, let's say that the code should be loaded at address
2097 0x10000, and that the data should start at address 0x8000000. Here is a
2098 linker script which will do that:
2102 .text : { *(.text) }
2104 .data : { *(.data) }
2108 You write the `SECTIONS' command as the keyword `SECTIONS', followed
2109 by a series of symbol assignments and output section descriptions
2110 enclosed in curly braces.
2112 The first line inside the `SECTIONS' command of the above example
2113 sets the value of the special symbol `.', which is the location
2114 counter. If you do not specify the address of an output section in some
2115 other way (other ways are described later), the address is set from the
2116 current value of the location counter. The location counter is then
2117 incremented by the size of the output section. At the start of the
2118 `SECTIONS' command, the location counter has the value `0'.
2120 The second line defines an output section, `.text'. The colon is
2121 required syntax which may be ignored for now. Within the curly braces
2122 after the output section name, you list the names of the input sections
2123 which should be placed into this output section. The `*' is a wildcard
2124 which matches any file name. The expression `*(.text)' means all
2125 `.text' input sections in all input files.
2127 Since the location counter is `0x10000' when the output section
2128 `.text' is defined, the linker will set the address of the `.text'
2129 section in the output file to be `0x10000'.
2131 The remaining lines define the `.data' and `.bss' sections in the
2132 output file. The linker will place the `.data' output section at
2133 address `0x8000000'. After the linker places the `.data' output
2134 section, the value of the location counter will be `0x8000000' plus the
2135 size of the `.data' output section. The effect is that the linker will
2136 place the `.bss' output section immediately after the `.data' output
2139 The linker will ensure that each output section has the required
2140 alignment, by increasing the location counter if necessary. In this
2141 example, the specified addresses for the `.text' and `.data' sections
2142 will probably satisfy any alignment constraints, but the linker may
2143 have to create a small gap between the `.data' and `.bss' sections.
2145 That's it! That's a simple and complete linker script.
2148 File: ld.info, Node: Simple Commands, Next: Assignments, Prev: Simple Example, Up: Scripts
2150 3.4 Simple Linker Script Commands
2151 =================================
2153 In this section we describe the simple linker script commands.
2157 * Entry Point:: Setting the entry point
2158 * File Commands:: Commands dealing with files
2160 * Format Commands:: Commands dealing with object file formats
2162 * Miscellaneous Commands:: Other linker script commands
2165 File: ld.info, Node: Entry Point, Next: File Commands, Up: Simple Commands
2167 3.4.1 Setting the Entry Point
2168 -----------------------------
2170 The first instruction to execute in a program is called the "entry
2171 point". You can use the `ENTRY' linker script command to set the entry
2172 point. The argument is a symbol name:
2175 There are several ways to set the entry point. The linker will set
2176 the entry point by trying each of the following methods in order, and
2177 stopping when one of them succeeds:
2178 * the `-e' ENTRY command-line option;
2180 * the `ENTRY(SYMBOL)' command in a linker script;
2182 * the value of the symbol `start', if defined;
2184 * the address of the first byte of the `.text' section, if present;
2189 File: ld.info, Node: File Commands, Next: Format Commands, Prev: Entry Point, Up: Simple Commands
2191 3.4.2 Commands Dealing with Files
2192 ---------------------------------
2194 Several linker script commands deal with files.
2197 Include the linker script FILENAME at this point. The file will
2198 be searched for in the current directory, and in any directory
2199 specified with the `-L' option. You can nest calls to `INCLUDE'
2200 up to 10 levels deep.
2202 You can place `INCLUDE' directives at the top level, in `MEMORY' or
2203 `SECTIONS' commands, or in output section descriptions.
2205 `INPUT(FILE, FILE, ...)'
2206 `INPUT(FILE FILE ...)'
2207 The `INPUT' command directs the linker to include the named files
2208 in the link, as though they were named on the command line.
2210 For example, if you always want to include `subr.o' any time you do
2211 a link, but you can't be bothered to put it on every link command
2212 line, then you can put `INPUT (subr.o)' in your linker script.
2214 In fact, if you like, you can list all of your input files in the
2215 linker script, and then invoke the linker with nothing but a `-T'
2218 In case a "sysroot prefix" is configured, and the filename starts
2219 with the `/' character, and the script being processed was located
2220 inside the "sysroot prefix", the filename will be looked for in
2221 the "sysroot prefix". Otherwise, the linker will try to open the
2222 file in the current directory. If it is not found, the linker
2223 will search through the archive library search path. See the
2224 description of `-L' in *Note Command Line Options: Options.
2226 If you use `INPUT (-lFILE)', `ld' will transform the name to
2227 `libFILE.a', as with the command line argument `-l'.
2229 When you use the `INPUT' command in an implicit linker script, the
2230 files will be included in the link at the point at which the linker
2231 script file is included. This can affect archive searching.
2233 `GROUP(FILE, FILE, ...)'
2234 `GROUP(FILE FILE ...)'
2235 The `GROUP' command is like `INPUT', except that the named files
2236 should all be archives, and they are searched repeatedly until no
2237 new undefined references are created. See the description of `-('
2238 in *Note Command Line Options: Options.
2240 `AS_NEEDED(FILE, FILE, ...)'
2241 `AS_NEEDED(FILE FILE ...)'
2242 This construct can appear only inside of the `INPUT' or `GROUP'
2243 commands, among other filenames. The files listed will be handled
2244 as if they appear directly in the `INPUT' or `GROUP' commands,
2245 with the exception of ELF shared libraries, that will be added only
2246 when they are actually needed. This construct essentially enables
2247 `--as-needed' option for all the files listed inside of it and
2248 restores previous `--as-needed' resp. `--no-as-needed' setting
2252 The `OUTPUT' command names the output file. Using
2253 `OUTPUT(FILENAME)' in the linker script is exactly like using `-o
2254 FILENAME' on the command line (*note Command Line Options:
2255 Options.). If both are used, the command line option takes
2258 You can use the `OUTPUT' command to define a default name for the
2259 output file other than the usual default of `a.out'.
2262 The `SEARCH_DIR' command adds PATH to the list of paths where `ld'
2263 looks for archive libraries. Using `SEARCH_DIR(PATH)' is exactly
2264 like using `-L PATH' on the command line (*note Command Line
2265 Options: Options.). If both are used, then the linker will search
2266 both paths. Paths specified using the command line option are
2270 The `STARTUP' command is just like the `INPUT' command, except
2271 that FILENAME will become the first input file to be linked, as
2272 though it were specified first on the command line. This may be
2273 useful when using a system in which the entry point is always the
2274 start of the first file.
2277 File: ld.info, Node: Format Commands, Next: Miscellaneous Commands, Prev: File Commands, Up: Simple Commands
2279 3.4.3 Commands Dealing with Object File Formats
2280 -----------------------------------------------
2282 A couple of linker script commands deal with object file formats.
2284 `OUTPUT_FORMAT(BFDNAME)'
2285 `OUTPUT_FORMAT(DEFAULT, BIG, LITTLE)'
2286 The `OUTPUT_FORMAT' command names the BFD format to use for the
2287 output file (*note BFD::). Using `OUTPUT_FORMAT(BFDNAME)' is
2288 exactly like using `--oformat BFDNAME' on the command line (*note
2289 Command Line Options: Options.). If both are used, the command
2290 line option takes precedence.
2292 You can use `OUTPUT_FORMAT' with three arguments to use different
2293 formats based on the `-EB' and `-EL' command line options. This
2294 permits the linker script to set the output format based on the
2297 If neither `-EB' nor `-EL' are used, then the output format will
2298 be the first argument, DEFAULT. If `-EB' is used, the output
2299 format will be the second argument, BIG. If `-EL' is used, the
2300 output format will be the third argument, LITTLE.
2302 For example, the default linker script for the MIPS ELF target
2304 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2305 This says that the default format for the output file is
2306 `elf32-bigmips', but if the user uses the `-EL' command line
2307 option, the output file will be created in the `elf32-littlemips'
2311 The `TARGET' command names the BFD format to use when reading input
2312 files. It affects subsequent `INPUT' and `GROUP' commands. This
2313 command is like using `-b BFDNAME' on the command line (*note
2314 Command Line Options: Options.). If the `TARGET' command is used
2315 but `OUTPUT_FORMAT' is not, then the last `TARGET' command is also
2316 used to set the format for the output file. *Note BFD::.
2319 File: ld.info, Node: Miscellaneous Commands, Prev: Format Commands, Up: Simple Commands
2321 3.4.4 Other Linker Script Commands
2322 ----------------------------------
2324 There are a few other linker scripts commands.
2326 `ASSERT(EXP, MESSAGE)'
2327 Ensure that EXP is non-zero. If it is zero, then exit the linker
2328 with an error code, and print MESSAGE.
2330 `EXTERN(SYMBOL SYMBOL ...)'
2331 Force SYMBOL to be entered in the output file as an undefined
2332 symbol. Doing this may, for example, trigger linking of additional
2333 modules from standard libraries. You may list several SYMBOLs for
2334 each `EXTERN', and you may use `EXTERN' multiple times. This
2335 command has the same effect as the `-u' command-line option.
2337 `FORCE_COMMON_ALLOCATION'
2338 This command has the same effect as the `-d' command-line option:
2339 to make `ld' assign space to common symbols even if a relocatable
2340 output file is specified (`-r').
2342 `INHIBIT_COMMON_ALLOCATION'
2343 This command has the same effect as the `--no-define-common'
2344 command-line option: to make `ld' omit the assignment of addresses
2345 to common symbols even for a non-relocatable output file.
2347 `INSERT [ AFTER | BEFORE ] OUTPUT_SECTION'
2348 This command is typically used in a script specified by `-T' to
2349 augment the default `SECTIONS' with, for example, overlays. It
2350 inserts all prior linker script statements after (or before)
2351 OUTPUT_SECTION, and also causes `-T' to not override the default
2352 linker script. The exact insertion point is as for orphan
2353 sections. *Note Location Counter::. The insertion happens after
2354 the linker has mapped input sections to output sections. Prior to
2355 the insertion, since `-T' scripts are parsed before the default
2356 linker script, statements in the `-T' script occur before the
2357 default linker script statements in the internal linker
2358 representation of the script. In particular, input section
2359 assignments will be made to `-T' output sections before those in
2360 the default script. Here is an example of how a `-T' script using
2361 `INSERT' might look:
2367 .ov1 { ov1*(.text) }
2368 .ov2 { ov2*(.text) }
2373 `NOCROSSREFS(SECTION SECTION ...)'
2374 This command may be used to tell `ld' to issue an error about any
2375 references among certain output sections.
2377 In certain types of programs, particularly on embedded systems when
2378 using overlays, when one section is loaded into memory, another
2379 section will not be. Any direct references between the two
2380 sections would be errors. For example, it would be an error if
2381 code in one section called a function defined in the other section.
2383 The `NOCROSSREFS' command takes a list of output section names. If
2384 `ld' detects any cross references between the sections, it reports
2385 an error and returns a non-zero exit status. Note that the
2386 `NOCROSSREFS' command uses output section names, not input section
2389 `OUTPUT_ARCH(BFDARCH)'
2390 Specify a particular output machine architecture. The argument is
2391 one of the names used by the BFD library (*note BFD::). You can
2392 see the architecture of an object file by using the `objdump'
2393 program with the `-f' option.
2396 File: ld.info, Node: Assignments, Next: SECTIONS, Prev: Simple Commands, Up: Scripts
2398 3.5 Assigning Values to Symbols
2399 ===============================
2401 You may assign a value to a symbol in a linker script. This will define
2402 the symbol and place it into the symbol table with a global scope.
2406 * Simple Assignments:: Simple Assignments
2408 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2409 * Source Code Reference:: How to use a linker script defined symbol in source code
2412 File: ld.info, Node: Simple Assignments, Next: PROVIDE, Up: Assignments
2414 3.5.1 Simple Assignments
2415 ------------------------
2417 You may assign to a symbol using any of the C assignment operators:
2419 `SYMBOL = EXPRESSION ;'
2420 `SYMBOL += EXPRESSION ;'
2421 `SYMBOL -= EXPRESSION ;'
2422 `SYMBOL *= EXPRESSION ;'
2423 `SYMBOL /= EXPRESSION ;'
2424 `SYMBOL <<= EXPRESSION ;'
2425 `SYMBOL >>= EXPRESSION ;'
2426 `SYMBOL &= EXPRESSION ;'
2427 `SYMBOL |= EXPRESSION ;'
2429 The first case will define SYMBOL to the value of EXPRESSION. In
2430 the other cases, SYMBOL must already be defined, and the value will be
2431 adjusted accordingly.
2433 The special symbol name `.' indicates the location counter. You may
2434 only use this within a `SECTIONS' command. *Note Location Counter::.
2436 The semicolon after EXPRESSION is required.
2438 Expressions are defined below; see *Note Expressions::.
2440 You may write symbol assignments as commands in their own right, or
2441 as statements within a `SECTIONS' command, or as part of an output
2442 section description in a `SECTIONS' command.
2444 The section of the symbol will be set from the section of the
2445 expression; for more information, see *Note Expression Section::.
2447 Here is an example showing the three different places that symbol
2448 assignments may be used:
2458 _bdata = (. + 3) & ~ 3;
2459 .data : { *(.data) }
2461 In this example, the symbol `floating_point' will be defined as
2462 zero. The symbol `_etext' will be defined as the address following the
2463 last `.text' input section. The symbol `_bdata' will be defined as the
2464 address following the `.text' output section aligned upward to a 4 byte
2468 File: ld.info, Node: PROVIDE, Next: PROVIDE_HIDDEN, Prev: Simple Assignments, Up: Assignments
2473 In some cases, it is desirable for a linker script to define a symbol
2474 only if it is referenced and is not defined by any object included in
2475 the link. For example, traditional linkers defined the symbol `etext'.
2476 However, ANSI C requires that the user be able to use `etext' as a
2477 function name without encountering an error. The `PROVIDE' keyword may
2478 be used to define a symbol, such as `etext', only if it is referenced
2479 but not defined. The syntax is `PROVIDE(SYMBOL = EXPRESSION)'.
2481 Here is an example of using `PROVIDE' to define `etext':
2492 In this example, if the program defines `_etext' (with a leading
2493 underscore), the linker will give a multiple definition error. If, on
2494 the other hand, the program defines `etext' (with no leading
2495 underscore), the linker will silently use the definition in the program.
2496 If the program references `etext' but does not define it, the linker
2497 will use the definition in the linker script.
2500 File: ld.info, Node: PROVIDE_HIDDEN, Next: Source Code Reference, Prev: PROVIDE, Up: Assignments
2502 3.5.3 PROVIDE_HIDDEN
2503 --------------------
2505 Similar to `PROVIDE'. For ELF targeted ports, the symbol will be
2506 hidden and won't be exported.
2509 File: ld.info, Node: Source Code Reference, Prev: PROVIDE_HIDDEN, Up: Assignments
2511 3.5.4 Source Code Reference
2512 ---------------------------
2514 Accessing a linker script defined variable from source code is not
2515 intuitive. In particular a linker script symbol is not equivalent to a
2516 variable declaration in a high level language, it is instead a symbol
2517 that does not have a value.
2519 Before going further, it is important to note that compilers often
2520 transform names in the source code into different names when they are
2521 stored in the symbol table. For example, Fortran compilers commonly
2522 prepend or append an underscore, and C++ performs extensive `name
2523 mangling'. Therefore there might be a discrepancy between the name of
2524 a variable as it is used in source code and the name of the same
2525 variable as it is defined in a linker script. For example in C a
2526 linker script variable might be referred to as:
2530 But in the linker script it might be defined as:
2534 In the remaining examples however it is assumed that no name
2535 transformation has taken place.
2537 When a symbol is declared in a high level language such as C, two
2538 things happen. The first is that the compiler reserves enough space in
2539 the program's memory to hold the _value_ of the symbol. The second is
2540 that the compiler creates an entry in the program's symbol table which
2541 holds the symbol's _address_. ie the symbol table contains the address
2542 of the block of memory holding the symbol's value. So for example the
2543 following C declaration, at file scope:
2547 creates a entry called `foo' in the symbol table. This entry holds
2548 the address of an `int' sized block of memory where the number 1000 is
2551 When a program references a symbol the compiler generates code that
2552 first accesses the symbol table to find the address of the symbol's
2553 memory block and then code to read the value from that memory block.
2558 looks up the symbol `foo' in the symbol table, gets the address
2559 associated with this symbol and then writes the value 1 into that
2564 looks up the symbol `foo' in the symbol table, gets it address and
2565 then copies this address into the block of memory associated with the
2568 Linker scripts symbol declarations, by contrast, create an entry in
2569 the symbol table but do not assign any memory to them. Thus they are
2570 an address without a value. So for example the linker script
2575 creates an entry in the symbol table called `foo' which holds the
2576 address of memory location 1000, but nothing special is stored at
2577 address 1000. This means that you cannot access the _value_ of a
2578 linker script defined symbol - it has no value - all you can do is
2579 access the _address_ of a linker script defined symbol.
2581 Hence when you are using a linker script defined symbol in source
2582 code you should always take the address of the symbol, and never
2583 attempt to use its value. For example suppose you want to copy the
2584 contents of a section of memory called .ROM into a section called
2585 .FLASH and the linker script contains these declarations:
2587 start_of_ROM = .ROM;
2588 end_of_ROM = .ROM + sizeof (.ROM) - 1;
2589 start_of_FLASH = .FLASH;
2591 Then the C source code to perform the copy would be:
2593 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
2595 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
2597 Note the use of the `&' operators. These are correct.
2600 File: ld.info, Node: SECTIONS, Next: MEMORY, Prev: Assignments, Up: Scripts
2602 3.6 SECTIONS Command
2603 ====================
2605 The `SECTIONS' command tells the linker how to map input sections into
2606 output sections, and how to place the output sections in memory.
2608 The format of the `SECTIONS' command is:
2616 Each SECTIONS-COMMAND may of be one of the following:
2618 * an `ENTRY' command (*note Entry command: Entry Point.)
2620 * a symbol assignment (*note Assignments::)
2622 * an output section description
2624 * an overlay description
2626 The `ENTRY' command and symbol assignments are permitted inside the
2627 `SECTIONS' command for convenience in using the location counter in
2628 those commands. This can also make the linker script easier to
2629 understand because you can use those commands at meaningful points in
2630 the layout of the output file.
2632 Output section descriptions and overlay descriptions are described
2635 If you do not use a `SECTIONS' command in your linker script, the
2636 linker will place each input section into an identically named output
2637 section in the order that the sections are first encountered in the
2638 input files. If all input sections are present in the first file, for
2639 example, the order of sections in the output file will match the order
2640 in the first input file. The first section will be at address zero.
2644 * Output Section Description:: Output section description
2645 * Output Section Name:: Output section name
2646 * Output Section Address:: Output section address
2647 * Input Section:: Input section description
2648 * Output Section Data:: Output section data
2649 * Output Section Keywords:: Output section keywords
2650 * Output Section Discarding:: Output section discarding
2651 * Output Section Attributes:: Output section attributes
2652 * Overlay Description:: Overlay description
2655 File: ld.info, Node: Output Section Description, Next: Output Section Name, Up: SECTIONS
2657 3.6.1 Output Section Description
2658 --------------------------------
2660 The full description of an output section looks like this:
2661 SECTION [ADDRESS] [(TYPE)] :
2662 [AT(LMA)] [ALIGN(SECTION_ALIGN)] [SUBALIGN(SUBSECTION_ALIGN)]
2664 OUTPUT-SECTION-COMMAND
2665 OUTPUT-SECTION-COMMAND
2667 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
2669 Most output sections do not use most of the optional section
2672 The whitespace around SECTION is required, so that the section name
2673 is unambiguous. The colon and the curly braces are also required. The
2674 line breaks and other white space are optional.
2676 Each OUTPUT-SECTION-COMMAND may be one of the following:
2678 * a symbol assignment (*note Assignments::)
2680 * an input section description (*note Input Section::)
2682 * data values to include directly (*note Output Section Data::)
2684 * a special output section keyword (*note Output Section Keywords::)
2687 File: ld.info, Node: Output Section Name, Next: Output Section Address, Prev: Output Section Description, Up: SECTIONS
2689 3.6.2 Output Section Name
2690 -------------------------
2692 The name of the output section is SECTION. SECTION must meet the
2693 constraints of your output format. In formats which only support a
2694 limited number of sections, such as `a.out', the name must be one of
2695 the names supported by the format (`a.out', for example, allows only
2696 `.text', `.data' or `.bss'). If the output format supports any number
2697 of sections, but with numbers and not names (as is the case for Oasys),
2698 the name should be supplied as a quoted numeric string. A section name
2699 may consist of any sequence of characters, but a name which contains
2700 any unusual characters such as commas must be quoted.
2702 The output section name `/DISCARD/' is special; *Note Output Section
2706 File: ld.info, Node: Output Section Address, Next: Input Section, Prev: Output Section Name, Up: SECTIONS
2708 3.6.3 Output Section Address
2709 ----------------------------
2711 The ADDRESS is an expression for the VMA (the virtual memory address)
2712 of the output section. If you do not provide ADDRESS, the linker will
2713 set it based on REGION if present, or otherwise based on the current
2714 value of the location counter.
2716 If you provide ADDRESS, the address of the output section will be
2717 set to precisely that. If you provide neither ADDRESS nor REGION, then
2718 the address of the output section will be set to the current value of
2719 the location counter aligned to the alignment requirements of the
2720 output section. The alignment requirement of the output section is the
2721 strictest alignment of any input section contained within the output
2725 .text . : { *(.text) }
2727 .text : { *(.text) }
2728 are subtly different. The first will set the address of the `.text'
2729 output section to the current value of the location counter. The
2730 second will set it to the current value of the location counter aligned
2731 to the strictest alignment of a `.text' input section.
2733 The ADDRESS may be an arbitrary expression; *Note Expressions::.
2734 For example, if you want to align the section on a 0x10 byte boundary,
2735 so that the lowest four bits of the section address are zero, you could
2736 do something like this:
2737 .text ALIGN(0x10) : { *(.text) }
2738 This works because `ALIGN' returns the current location counter
2739 aligned upward to the specified value.
2741 Specifying ADDRESS for a section will change the value of the
2745 File: ld.info, Node: Input Section, Next: Output Section Data, Prev: Output Section Address, Up: SECTIONS
2747 3.6.4 Input Section Description
2748 -------------------------------
2750 The most common output section command is an input section description.
2752 The input section description is the most basic linker script
2753 operation. You use output sections to tell the linker how to lay out
2754 your program in memory. You use input section descriptions to tell the
2755 linker how to map the input files into your memory layout.
2759 * Input Section Basics:: Input section basics
2760 * Input Section Wildcards:: Input section wildcard patterns
2761 * Input Section Common:: Input section for common symbols
2762 * Input Section Keep:: Input section and garbage collection
2763 * Input Section Example:: Input section example
2766 File: ld.info, Node: Input Section Basics, Next: Input Section Wildcards, Up: Input Section
2768 3.6.4.1 Input Section Basics
2769 ............................
2771 An input section description consists of a file name optionally followed
2772 by a list of section names in parentheses.
2774 The file name and the section name may be wildcard patterns, which we
2775 describe further below (*note Input Section Wildcards::).
2777 The most common input section description is to include all input
2778 sections with a particular name in the output section. For example, to
2779 include all input `.text' sections, you would write:
2781 Here the `*' is a wildcard which matches any file name. To exclude
2782 a list of files from matching the file name wildcard, EXCLUDE_FILE may
2783 be used to match all files except the ones specified in the
2784 EXCLUDE_FILE list. For example:
2785 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
2786 will cause all .ctors sections from all files except `crtend.o' and
2787 `otherfile.o' to be included.
2789 There are two ways to include more than one section:
2792 The difference between these is the order in which the `.text' and
2793 `.rdata' input sections will appear in the output section. In the
2794 first example, they will be intermingled, appearing in the same order as
2795 they are found in the linker input. In the second example, all `.text'
2796 input sections will appear first, followed by all `.rdata' input
2799 You can specify a file name to include sections from a particular
2800 file. You would do this if one or more of your files contain special
2801 data that needs to be at a particular location in memory. For example:
2804 You can also specify files within archives by writing a pattern
2805 matching the archive, a colon, then the pattern matching the file, with
2806 no whitespace around the colon.
2809 matches file within archive
2812 matches the whole archive
2815 matches file but not one in an archive
2817 Either one or both of `archive' and `file' can contain shell
2818 wildcards. On DOS based file systems, the linker will assume that a
2819 single letter followed by a colon is a drive specifier, so `c:myfile.o'
2820 is a simple file specification, not `myfile.o' within an archive called
2821 `c'. `archive:file' filespecs may also be used within an
2822 `EXCLUDE_FILE' list, but may not appear in other linker script
2823 contexts. For instance, you cannot extract a file from an archive by
2824 using `archive:file' in an `INPUT' command.
2826 If you use a file name without a list of sections, then all sections
2827 in the input file will be included in the output section. This is not
2828 commonly done, but it may by useful on occasion. For example:
2831 When you use a file name which is not an `archive:file' specifier
2832 and does not contain any wild card characters, the linker will first
2833 see if you also specified the file name on the linker command line or
2834 in an `INPUT' command. If you did not, the linker will attempt to open
2835 the file as an input file, as though it appeared on the command line.
2836 Note that this differs from an `INPUT' command, because the linker will
2837 not search for the file in the archive search path.
2840 File: ld.info, Node: Input Section Wildcards, Next: Input Section Common, Prev: Input Section Basics, Up: Input Section
2842 3.6.4.2 Input Section Wildcard Patterns
2843 .......................................
2845 In an input section description, either the file name or the section
2846 name or both may be wildcard patterns.
2848 The file name of `*' seen in many examples is a simple wildcard
2849 pattern for the file name.
2851 The wildcard patterns are like those used by the Unix shell.
2854 matches any number of characters
2857 matches any single character
2860 matches a single instance of any of the CHARS; the `-' character
2861 may be used to specify a range of characters, as in `[a-z]' to
2862 match any lower case letter
2865 quotes the following character
2867 When a file name is matched with a wildcard, the wildcard characters
2868 will not match a `/' character (used to separate directory names on
2869 Unix). A pattern consisting of a single `*' character is an exception;
2870 it will always match any file name, whether it contains a `/' or not.
2871 In a section name, the wildcard characters will match a `/' character.
2873 File name wildcard patterns only match files which are explicitly
2874 specified on the command line or in an `INPUT' command. The linker
2875 does not search directories to expand wildcards.
2877 If a file name matches more than one wildcard pattern, or if a file
2878 name appears explicitly and is also matched by a wildcard pattern, the
2879 linker will use the first match in the linker script. For example, this
2880 sequence of input section descriptions is probably in error, because the
2881 `data.o' rule will not be used:
2882 .data : { *(.data) }
2883 .data1 : { data.o(.data) }
2885 Normally, the linker will place files and sections matched by
2886 wildcards in the order in which they are seen during the link. You can
2887 change this by using the `SORT_BY_NAME' keyword, which appears before a
2888 wildcard pattern in parentheses (e.g., `SORT_BY_NAME(.text*)'). When
2889 the `SORT_BY_NAME' keyword is used, the linker will sort the files or
2890 sections into ascending order by name before placing them in the output
2893 `SORT_BY_ALIGNMENT' is very similar to `SORT_BY_NAME'. The
2894 difference is `SORT_BY_ALIGNMENT' will sort sections into ascending
2895 order by alignment before placing them in the output file.
2897 `SORT' is an alias for `SORT_BY_NAME'.
2899 When there are nested section sorting commands in linker script,
2900 there can be at most 1 level of nesting for section sorting commands.
2902 1. `SORT_BY_NAME' (`SORT_BY_ALIGNMENT' (wildcard section pattern)).
2903 It will sort the input sections by name first, then by alignment
2904 if 2 sections have the same name.
2906 2. `SORT_BY_ALIGNMENT' (`SORT_BY_NAME' (wildcard section pattern)).
2907 It will sort the input sections by alignment first, then by name
2908 if 2 sections have the same alignment.
2910 3. `SORT_BY_NAME' (`SORT_BY_NAME' (wildcard section pattern)) is
2911 treated the same as `SORT_BY_NAME' (wildcard section pattern).
2913 4. `SORT_BY_ALIGNMENT' (`SORT_BY_ALIGNMENT' (wildcard section
2914 pattern)) is treated the same as `SORT_BY_ALIGNMENT' (wildcard
2917 5. All other nested section sorting commands are invalid.
2919 When both command line section sorting option and linker script
2920 section sorting command are used, section sorting command always takes
2921 precedence over the command line option.
2923 If the section sorting command in linker script isn't nested, the
2924 command line option will make the section sorting command to be treated
2925 as nested sorting command.
2927 1. `SORT_BY_NAME' (wildcard section pattern ) with `--sort-sections
2928 alignment' is equivalent to `SORT_BY_NAME' (`SORT_BY_ALIGNMENT'
2929 (wildcard section pattern)).
2931 2. `SORT_BY_ALIGNMENT' (wildcard section pattern) with
2932 `--sort-section name' is equivalent to `SORT_BY_ALIGNMENT'
2933 (`SORT_BY_NAME' (wildcard section pattern)).
2935 If the section sorting command in linker script is nested, the
2936 command line option will be ignored.
2938 If you ever get confused about where input sections are going, use
2939 the `-M' linker option to generate a map file. The map file shows
2940 precisely how input sections are mapped to output sections.
2942 This example shows how wildcard patterns might be used to partition
2943 files. This linker script directs the linker to place all `.text'
2944 sections in `.text' and all `.bss' sections in `.bss'. The linker will
2945 place the `.data' section from all files beginning with an upper case
2946 character in `.DATA'; for all other files, the linker will place the
2947 `.data' section in `.data'.
2949 .text : { *(.text) }
2950 .DATA : { [A-Z]*(.data) }
2951 .data : { *(.data) }
2956 File: ld.info, Node: Input Section Common, Next: Input Section Keep, Prev: Input Section Wildcards, Up: Input Section
2958 3.6.4.3 Input Section for Common Symbols
2959 ........................................
2961 A special notation is needed for common symbols, because in many object
2962 file formats common symbols do not have a particular input section. The
2963 linker treats common symbols as though they are in an input section
2966 You may use file names with the `COMMON' section just as with any
2967 other input sections. You can use this to place common symbols from a
2968 particular input file in one section while common symbols from other
2969 input files are placed in another section.
2971 In most cases, common symbols in input files will be placed in the
2972 `.bss' section in the output file. For example:
2973 .bss { *(.bss) *(COMMON) }
2975 Some object file formats have more than one type of common symbol.
2976 For example, the MIPS ELF object file format distinguishes standard
2977 common symbols and small common symbols. In this case, the linker will
2978 use a different special section name for other types of common symbols.
2979 In the case of MIPS ELF, the linker uses `COMMON' for standard common
2980 symbols and `.scommon' for small common symbols. This permits you to
2981 map the different types of common symbols into memory at different
2984 You will sometimes see `[COMMON]' in old linker scripts. This
2985 notation is now considered obsolete. It is equivalent to `*(COMMON)'.
2988 File: ld.info, Node: Input Section Keep, Next: Input Section Example, Prev: Input Section Common, Up: Input Section
2990 3.6.4.4 Input Section and Garbage Collection
2991 ............................................
2993 When link-time garbage collection is in use (`--gc-sections'), it is
2994 often useful to mark sections that should not be eliminated. This is
2995 accomplished by surrounding an input section's wildcard entry with
2996 `KEEP()', as in `KEEP(*(.init))' or `KEEP(SORT_BY_NAME(*)(.ctors))'.
2999 File: ld.info, Node: Input Section Example, Prev: Input Section Keep, Up: Input Section
3001 3.6.4.5 Input Section Example
3002 .............................
3004 The following example is a complete linker script. It tells the linker
3005 to read all of the sections from file `all.o' and place them at the
3006 start of output section `outputa' which starts at location `0x10000'.
3007 All of section `.input1' from file `foo.o' follows immediately, in the
3008 same output section. All of section `.input2' from `foo.o' goes into
3009 output section `outputb', followed by section `.input1' from `foo1.o'.
3010 All of the remaining `.input1' and `.input2' sections from any files
3011 are written to output section `outputc'.
3032 File: ld.info, Node: Output Section Data, Next: Output Section Keywords, Prev: Input Section, Up: SECTIONS
3034 3.6.5 Output Section Data
3035 -------------------------
3037 You can include explicit bytes of data in an output section by using
3038 `BYTE', `SHORT', `LONG', `QUAD', or `SQUAD' as an output section
3039 command. Each keyword is followed by an expression in parentheses
3040 providing the value to store (*note Expressions::). The value of the
3041 expression is stored at the current value of the location counter.
3043 The `BYTE', `SHORT', `LONG', and `QUAD' commands store one, two,
3044 four, and eight bytes (respectively). After storing the bytes, the
3045 location counter is incremented by the number of bytes stored.
3047 For example, this will store the byte 1 followed by the four byte
3048 value of the symbol `addr':
3052 When using a 64 bit host or target, `QUAD' and `SQUAD' are the same;
3053 they both store an 8 byte, or 64 bit, value. When both host and target
3054 are 32 bits, an expression is computed as 32 bits. In this case `QUAD'
3055 stores a 32 bit value zero extended to 64 bits, and `SQUAD' stores a 32
3056 bit value sign extended to 64 bits.
3058 If the object file format of the output file has an explicit
3059 endianness, which is the normal case, the value will be stored in that
3060 endianness. When the object file format does not have an explicit
3061 endianness, as is true of, for example, S-records, the value will be
3062 stored in the endianness of the first input object file.
3064 Note--these commands only work inside a section description and not
3065 between them, so the following will produce an error from the linker:
3066 SECTIONS { .text : { *(.text) } LONG(1) .data : { *(.data) } }
3067 whereas this will work:
3068 SECTIONS { .text : { *(.text) ; LONG(1) } .data : { *(.data) } }
3070 You may use the `FILL' command to set the fill pattern for the
3071 current section. It is followed by an expression in parentheses. Any
3072 otherwise unspecified regions of memory within the section (for example,
3073 gaps left due to the required alignment of input sections) are filled
3074 with the value of the expression, repeated as necessary. A `FILL'
3075 statement covers memory locations after the point at which it occurs in
3076 the section definition; by including more than one `FILL' statement,
3077 you can have different fill patterns in different parts of an output
3080 This example shows how to fill unspecified regions of memory with the
3084 The `FILL' command is similar to the `=FILLEXP' output section
3085 attribute, but it only affects the part of the section following the
3086 `FILL' command, rather than the entire section. If both are used, the
3087 `FILL' command takes precedence. *Note Output Section Fill::, for
3088 details on the fill expression.
3091 File: ld.info, Node: Output Section Keywords, Next: Output Section Discarding, Prev: Output Section Data, Up: SECTIONS
3093 3.6.6 Output Section Keywords
3094 -----------------------------
3096 There are a couple of keywords which can appear as output section
3099 `CREATE_OBJECT_SYMBOLS'
3100 The command tells the linker to create a symbol for each input
3101 file. The name of each symbol will be the name of the
3102 corresponding input file. The section of each symbol will be the
3103 output section in which the `CREATE_OBJECT_SYMBOLS' command
3106 This is conventional for the a.out object file format. It is not
3107 normally used for any other object file format.
3110 When linking using the a.out object file format, the linker uses an
3111 unusual set construct to support C++ global constructors and
3112 destructors. When linking object file formats which do not support
3113 arbitrary sections, such as ECOFF and XCOFF, the linker will
3114 automatically recognize C++ global constructors and destructors by
3115 name. For these object file formats, the `CONSTRUCTORS' command
3116 tells the linker to place constructor information in the output
3117 section where the `CONSTRUCTORS' command appears. The
3118 `CONSTRUCTORS' command is ignored for other object file formats.
3120 The symbol `__CTOR_LIST__' marks the start of the global
3121 constructors, and the symbol `__CTOR_END__' marks the end.
3122 Similarly, `__DTOR_LIST__' and `__DTOR_END__' mark the start and
3123 end of the global destructors. The first word in the list is the
3124 number of entries, followed by the address of each constructor or
3125 destructor, followed by a zero word. The compiler must arrange to
3126 actually run the code. For these object file formats GNU C++
3127 normally calls constructors from a subroutine `__main'; a call to
3128 `__main' is automatically inserted into the startup code for
3129 `main'. GNU C++ normally runs destructors either by using
3130 `atexit', or directly from the function `exit'.
3132 For object file formats such as `COFF' or `ELF' which support
3133 arbitrary section names, GNU C++ will normally arrange to put the
3134 addresses of global constructors and destructors into the `.ctors'
3135 and `.dtors' sections. Placing the following sequence into your
3136 linker script will build the sort of table which the GNU C++
3137 runtime code expects to see.
3140 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3145 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3150 If you are using the GNU C++ support for initialization priority,
3151 which provides some control over the order in which global
3152 constructors are run, you must sort the constructors at link time
3153 to ensure that they are executed in the correct order. When using
3154 the `CONSTRUCTORS' command, use `SORT_BY_NAME(CONSTRUCTORS)'
3155 instead. When using the `.ctors' and `.dtors' sections, use
3156 `*(SORT_BY_NAME(.ctors))' and `*(SORT_BY_NAME(.dtors))' instead of
3157 just `*(.ctors)' and `*(.dtors)'.
3159 Normally the compiler and linker will handle these issues
3160 automatically, and you will not need to concern yourself with
3161 them. However, you may need to consider this if you are using C++
3162 and writing your own linker scripts.
3166 File: ld.info, Node: Output Section Discarding, Next: Output Section Attributes, Prev: Output Section Keywords, Up: SECTIONS
3168 3.6.7 Output Section Discarding
3169 -------------------------------
3171 The linker will not create output sections with no contents. This is
3172 for convenience when referring to input sections that may or may not be
3173 present in any of the input files. For example:
3175 will only create a `.foo' section in the output file if there is a
3176 `.foo' section in at least one input file, and if the input sections
3177 are not all empty. Other link script directives that allocate space in
3178 an output section will also create the output section.
3180 The linker will ignore address assignments (*note Output Section
3181 Address::) on discarded output sections, except when the linker script
3182 defines symbols in the output section. In that case the linker will
3183 obey the address assignments, possibly advancing dot even though the
3184 section is discarded.
3186 The special output section name `/DISCARD/' may be used to discard
3187 input sections. Any input sections which are assigned to an output
3188 section named `/DISCARD/' are not included in the output file.
3191 File: ld.info, Node: Output Section Attributes, Next: Overlay Description, Prev: Output Section Discarding, Up: SECTIONS
3193 3.6.8 Output Section Attributes
3194 -------------------------------
3196 We showed above that the full description of an output section looked
3198 SECTION [ADDRESS] [(TYPE)] :
3199 [AT(LMA)] [ALIGN(SECTION_ALIGN)] [SUBALIGN(SUBSECTION_ALIGN)]
3201 OUTPUT-SECTION-COMMAND
3202 OUTPUT-SECTION-COMMAND
3204 } [>REGION] [AT>LMA_REGION] [:PHDR :PHDR ...] [=FILLEXP]
3205 We've already described SECTION, ADDRESS, and
3206 OUTPUT-SECTION-COMMAND. In this section we will describe the remaining
3211 * Output Section Type:: Output section type
3212 * Output Section LMA:: Output section LMA
3213 * Forced Output Alignment:: Forced Output Alignment
3214 * Forced Input Alignment:: Forced Input Alignment
3215 * Output Section Region:: Output section region
3216 * Output Section Phdr:: Output section phdr
3217 * Output Section Fill:: Output section fill
3220 File: ld.info, Node: Output Section Type, Next: Output Section LMA, Up: Output Section Attributes
3222 3.6.8.1 Output Section Type
3223 ...........................
3225 Each output section may have a type. The type is a keyword in
3226 parentheses. The following types are defined:
3229 The section should be marked as not loadable, so that it will not
3230 be loaded into memory when the program is run.
3236 These type names are supported for backward compatibility, and are
3237 rarely used. They all have the same effect: the section should be
3238 marked as not allocatable, so that no memory is allocated for the
3239 section when the program is run.
3241 The linker normally sets the attributes of an output section based on
3242 the input sections which map into it. You can override this by using
3243 the section type. For example, in the script sample below, the `ROM'
3244 section is addressed at memory location `0' and does not need to be
3245 loaded when the program is run. The contents of the `ROM' section will
3246 appear in the linker output file as usual.
3248 ROM 0 (NOLOAD) : { ... }
3253 File: ld.info, Node: Output Section LMA, Next: Forced Output Alignment, Prev: Output Section Type, Up: Output Section Attributes
3255 3.6.8.2 Output Section LMA
3256 ..........................
3258 Every section has a virtual address (VMA) and a load address (LMA); see
3259 *Note Basic Script Concepts::. The address expression which may appear
3260 in an output section description sets the VMA (*note Output Section
3263 The expression LMA that follows the `AT' keyword specifies the load
3264 address of the section.
3266 Alternatively, with `AT>LMA_REGION' expression, you may specify a
3267 memory region for the section's load address. *Note MEMORY::. Note
3268 that if the section has not had a VMA assigned to it then the linker
3269 will use the LMA_REGION as the VMA region as well.
3271 If neither `AT' nor `AT>' is specified for an allocatable section,
3272 the linker will set the LMA such that the difference between VMA and
3273 LMA for the section is the same as the preceding output section in the
3274 same region. If there is no preceding output section or the section is
3275 not allocatable, the linker will set the LMA equal to the VMA. *Note
3276 Output Section Region::.
3278 This feature is designed to make it easy to build a ROM image. For
3279 example, the following linker script creates three output sections: one
3280 called `.text', which starts at `0x1000', one called `.mdata', which is
3281 loaded at the end of the `.text' section even though its VMA is
3282 `0x2000', and one called `.bss' to hold uninitialized data at address
3283 `0x3000'. The symbol `_data' is defined with the value `0x2000', which
3284 shows that the location counter holds the VMA value, not the LMA value.
3288 .text 0x1000 : { *(.text) _etext = . ; }
3290 AT ( ADDR (.text) + SIZEOF (.text) )
3291 { _data = . ; *(.data); _edata = . ; }
3293 { _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;}
3296 The run-time initialization code for use with a program generated
3297 with this linker script would include something like the following, to
3298 copy the initialized data from the ROM image to its runtime address.
3299 Notice how this code takes advantage of the symbols defined by the
3302 extern char _etext, _data, _edata, _bstart, _bend;
3303 char *src = &_etext;
3306 /* ROM has data at end of text; copy it. */
3307 while (dst < &_edata) {
3312 for (dst = &_bstart; dst< &_bend; dst++)
3316 File: ld.info, Node: Forced Output Alignment, Next: Forced Input Alignment, Prev: Output Section LMA, Up: Output Section Attributes
3318 3.6.8.3 Forced Output Alignment
3319 ...............................
3321 You can increase an output section's alignment by using ALIGN.
3324 File: ld.info, Node: Forced Input Alignment, Next: Output Section Region, Prev: Forced Output Alignment, Up: Output Section Attributes
3326 3.6.8.4 Forced Input Alignment
3327 ..............................
3329 You can force input section alignment within an output section by using
3330 SUBALIGN. The value specified overrides any alignment given by input
3331 sections, whether larger or smaller.
3334 File: ld.info, Node: Output Section Region, Next: Output Section Phdr, Prev: Forced Input Alignment, Up: Output Section Attributes
3336 3.6.8.5 Output Section Region
3337 .............................
3339 You can assign a section to a previously defined region of memory by
3340 using `>REGION'. *Note MEMORY::.
3342 Here is a simple example:
3343 MEMORY { rom : ORIGIN = 0x1000, LENGTH = 0x1000 }
3344 SECTIONS { ROM : { *(.text) } >rom }
3347 File: ld.info, Node: Output Section Phdr, Next: Output Section Fill, Prev: Output Section Region, Up: Output Section Attributes
3349 3.6.8.6 Output Section Phdr
3350 ...........................
3352 You can assign a section to a previously defined program segment by
3353 using `:PHDR'. *Note PHDRS::. If a section is assigned to one or more
3354 segments, then all subsequent allocated sections will be assigned to
3355 those segments as well, unless they use an explicitly `:PHDR' modifier.
3356 You can use `:NONE' to tell the linker to not put the section in any
3359 Here is a simple example:
3360 PHDRS { text PT_LOAD ; }
3361 SECTIONS { .text : { *(.text) } :text }
3364 File: ld.info, Node: Output Section Fill, Prev: Output Section Phdr, Up: Output Section Attributes
3366 3.6.8.7 Output Section Fill
3367 ...........................
3369 You can set the fill pattern for an entire section by using `=FILLEXP'.
3370 FILLEXP is an expression (*note Expressions::). Any otherwise
3371 unspecified regions of memory within the output section (for example,
3372 gaps left due to the required alignment of input sections) will be
3373 filled with the value, repeated as necessary. If the fill expression
3374 is a simple hex number, ie. a string of hex digit starting with `0x'
3375 and without a trailing `k' or `M', then an arbitrarily long sequence of
3376 hex digits can be used to specify the fill pattern; Leading zeros
3377 become part of the pattern too. For all other cases, including extra
3378 parentheses or a unary `+', the fill pattern is the four least
3379 significant bytes of the value of the expression. In all cases, the
3380 number is big-endian.
3382 You can also change the fill value with a `FILL' command in the
3383 output section commands; (*note Output Section Data::).
3385 Here is a simple example:
3386 SECTIONS { .text : { *(.text) } =0x90909090 }
3389 File: ld.info, Node: Overlay Description, Prev: Output Section Attributes, Up: SECTIONS
3391 3.6.9 Overlay Description
3392 -------------------------
3394 An overlay description provides an easy way to describe sections which
3395 are to be loaded as part of a single memory image but are to be run at
3396 the same memory address. At run time, some sort of overlay manager will
3397 copy the overlaid sections in and out of the runtime memory address as
3398 required, perhaps by simply manipulating addressing bits. This approach
3399 can be useful, for example, when a certain region of memory is faster
3402 Overlays are described using the `OVERLAY' command. The `OVERLAY'
3403 command is used within a `SECTIONS' command, like an output section
3404 description. The full syntax of the `OVERLAY' command is as follows:
3405 OVERLAY [START] : [NOCROSSREFS] [AT ( LDADDR )]
3409 OUTPUT-SECTION-COMMAND
3410 OUTPUT-SECTION-COMMAND
3412 } [:PHDR...] [=FILL]
3415 OUTPUT-SECTION-COMMAND
3416 OUTPUT-SECTION-COMMAND
3418 } [:PHDR...] [=FILL]
3420 } [>REGION] [:PHDR...] [=FILL]
3422 Everything is optional except `OVERLAY' (a keyword), and each
3423 section must have a name (SECNAME1 and SECNAME2 above). The section
3424 definitions within the `OVERLAY' construct are identical to those
3425 within the general `SECTIONS' contruct (*note SECTIONS::), except that
3426 no addresses and no memory regions may be defined for sections within
3429 The sections are all defined with the same starting address. The
3430 load addresses of the sections are arranged such that they are
3431 consecutive in memory starting at the load address used for the
3432 `OVERLAY' as a whole (as with normal section definitions, the load
3433 address is optional, and defaults to the start address; the start
3434 address is also optional, and defaults to the current value of the
3437 If the `NOCROSSREFS' keyword is used, and there any references among
3438 the sections, the linker will report an error. Since the sections all
3439 run at the same address, it normally does not make sense for one
3440 section to refer directly to another. *Note NOCROSSREFS: Miscellaneous
3443 For each section within the `OVERLAY', the linker automatically
3444 provides two symbols. The symbol `__load_start_SECNAME' is defined as
3445 the starting load address of the section. The symbol
3446 `__load_stop_SECNAME' is defined as the final load address of the
3447 section. Any characters within SECNAME which are not legal within C
3448 identifiers are removed. C (or assembler) code may use these symbols
3449 to move the overlaid sections around as necessary.
3451 At the end of the overlay, the value of the location counter is set
3452 to the start address of the overlay plus the size of the largest
3455 Here is an example. Remember that this would appear inside a
3456 `SECTIONS' construct.
3457 OVERLAY 0x1000 : AT (0x4000)
3459 .text0 { o1/*.o(.text) }
3460 .text1 { o2/*.o(.text) }
3462 This will define both `.text0' and `.text1' to start at address
3463 0x1000. `.text0' will be loaded at address 0x4000, and `.text1' will
3464 be loaded immediately after `.text0'. The following symbols will be
3465 defined if referenced: `__load_start_text0', `__load_stop_text0',
3466 `__load_start_text1', `__load_stop_text1'.
3468 C code to copy overlay `.text1' into the overlay area might look
3471 extern char __load_start_text1, __load_stop_text1;
3472 memcpy ((char *) 0x1000, &__load_start_text1,
3473 &__load_stop_text1 - &__load_start_text1);
3475 Note that the `OVERLAY' command is just syntactic sugar, since
3476 everything it does can be done using the more basic commands. The above
3477 example could have been written identically as follows.
3479 .text0 0x1000 : AT (0x4000) { o1/*.o(.text) }
3480 PROVIDE (__load_start_text0 = LOADADDR (.text0));
3481 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
3482 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) { o2/*.o(.text) }
3483 PROVIDE (__load_start_text1 = LOADADDR (.text1));
3484 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
3485 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
3488 File: ld.info, Node: MEMORY, Next: PHDRS, Prev: SECTIONS, Up: Scripts
3493 The linker's default configuration permits allocation of all available
3494 memory. You can override this by using the `MEMORY' command.
3496 The `MEMORY' command describes the location and size of blocks of
3497 memory in the target. You can use it to describe which memory regions
3498 may be used by the linker, and which memory regions it must avoid. You
3499 can then assign sections to particular memory regions. The linker will
3500 set section addresses based on the memory regions, and will warn about
3501 regions that become too full. The linker will not shuffle sections
3502 around to fit into the available regions.
3504 A linker script may contain at most one use of the `MEMORY' command.
3505 However, you can define as many blocks of memory within it as you
3506 wish. The syntax is:
3509 NAME [(ATTR)] : ORIGIN = ORIGIN, LENGTH = LEN
3513 The NAME is a name used in the linker script to refer to the region.
3514 The region name has no meaning outside of the linker script. Region
3515 names are stored in a separate name space, and will not conflict with
3516 symbol names, file names, or section names. Each memory region must
3517 have a distinct name.
3519 The ATTR string is an optional list of attributes that specify
3520 whether to use a particular memory region for an input section which is
3521 not explicitly mapped in the linker script. As described in *Note
3522 SECTIONS::, if you do not specify an output section for some input
3523 section, the linker will create an output section with the same name as
3524 the input section. If you define region attributes, the linker will use
3525 them to select the memory region for the output section that it creates.
3527 The ATTR string must consist only of the following characters:
3547 Invert the sense of any of the preceding attributes
3549 If a unmapped section matches any of the listed attributes other than
3550 `!', it will be placed in the memory region. The `!' attribute
3551 reverses this test, so that an unmapped section will be placed in the
3552 memory region only if it does not match any of the listed attributes.
3554 The ORIGIN is an numerical expression for the start address of the
3555 memory region. The expression must evaluate to a constant and it
3556 cannot involve any symbols. The keyword `ORIGIN' may be abbreviated to
3557 `org' or `o' (but not, for example, `ORG').
3559 The LEN is an expression for the size in bytes of the memory region.
3560 As with the ORIGIN expression, the expression must be numerical only
3561 and must evaluate to a constant. The keyword `LENGTH' may be
3562 abbreviated to `len' or `l'.
3564 In the following example, we specify that there are two memory
3565 regions available for allocation: one starting at `0' for 256 kilobytes,
3566 and the other starting at `0x40000000' for four megabytes. The linker
3567 will place into the `rom' memory region every section which is not
3568 explicitly mapped into a memory region, and is either read-only or
3569 executable. The linker will place other sections which are not
3570 explicitly mapped into a memory region into the `ram' memory region.
3574 rom (rx) : ORIGIN = 0, LENGTH = 256K
3575 ram (!rx) : org = 0x40000000, l = 4M
3578 Once you define a memory region, you can direct the linker to place
3579 specific output sections into that memory region by using the `>REGION'
3580 output section attribute. For example, if you have a memory region
3581 named `mem', you would use `>mem' in the output section definition.
3582 *Note Output Section Region::. If no address was specified for the
3583 output section, the linker will set the address to the next available
3584 address within the memory region. If the combined output sections
3585 directed to a memory region are too large for the region, the linker
3586 will issue an error message.
3588 It is possible to access the origin and length of a memory in an
3589 expression via the `ORIGIN(MEMORY)' and `LENGTH(MEMORY)' functions:
3591 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
3594 File: ld.info, Node: PHDRS, Next: VERSION, Prev: MEMORY, Up: Scripts
3599 The ELF object file format uses "program headers", also knows as
3600 "segments". The program headers describe how the program should be
3601 loaded into memory. You can print them out by using the `objdump'
3602 program with the `-p' option.
3604 When you run an ELF program on a native ELF system, the system loader
3605 reads the program headers in order to figure out how to load the
3606 program. This will only work if the program headers are set correctly.
3607 This manual does not describe the details of how the system loader
3608 interprets program headers; for more information, see the ELF ABI.
3610 The linker will create reasonable program headers by default.
3611 However, in some cases, you may need to specify the program headers more
3612 precisely. You may use the `PHDRS' command for this purpose. When the
3613 linker sees the `PHDRS' command in the linker script, it will not
3614 create any program headers other than the ones specified.
3616 The linker only pays attention to the `PHDRS' command when
3617 generating an ELF output file. In other cases, the linker will simply
3620 This is the syntax of the `PHDRS' command. The words `PHDRS',
3621 `FILEHDR', `AT', and `FLAGS' are keywords.
3625 NAME TYPE [ FILEHDR ] [ PHDRS ] [ AT ( ADDRESS ) ]
3626 [ FLAGS ( FLAGS ) ] ;
3629 The NAME is used only for reference in the `SECTIONS' command of the
3630 linker script. It is not put into the output file. Program header
3631 names are stored in a separate name space, and will not conflict with
3632 symbol names, file names, or section names. Each program header must
3633 have a distinct name.
3635 Certain program header types describe segments of memory which the
3636 system loader will load from the file. In the linker script, you
3637 specify the contents of these segments by placing allocatable output
3638 sections in the segments. You use the `:PHDR' output section attribute
3639 to place a section in a particular segment. *Note Output Section
3642 It is normal to put certain sections in more than one segment. This
3643 merely implies that one segment of memory contains another. You may
3644 repeat `:PHDR', using it once for each segment which should contain the
3647 If you place a section in one or more segments using `:PHDR', then
3648 the linker will place all subsequent allocatable sections which do not
3649 specify `:PHDR' in the same segments. This is for convenience, since
3650 generally a whole set of contiguous sections will be placed in a single
3651 segment. You can use `:NONE' to override the default segment and tell
3652 the linker to not put the section in any segment at all.
3654 You may use the `FILEHDR' and `PHDRS' keywords appear after the
3655 program header type to further describe the contents of the segment.
3656 The `FILEHDR' keyword means that the segment should include the ELF
3657 file header. The `PHDRS' keyword means that the segment should include
3658 the ELF program headers themselves.
3660 The TYPE may be one of the following. The numbers indicate the
3661 value of the keyword.
3664 Indicates an unused program header.
3667 Indicates that this program header describes a segment to be
3668 loaded from the file.
3671 Indicates a segment where dynamic linking information can be found.
3674 Indicates a segment where the name of the program interpreter may
3678 Indicates a segment holding note information.
3681 A reserved program header type, defined but not specified by the
3685 Indicates a segment where the program headers may be found.
3688 An expression giving the numeric type of the program header. This
3689 may be used for types not defined above.
3691 You can specify that a segment should be loaded at a particular
3692 address in memory by using an `AT' expression. This is identical to the
3693 `AT' command used as an output section attribute (*note Output Section
3694 LMA::). The `AT' command for a program header overrides the output
3697 The linker will normally set the segment flags based on the sections
3698 which comprise the segment. You may use the `FLAGS' keyword to
3699 explicitly specify the segment flags. The value of FLAGS must be an
3700 integer. It is used to set the `p_flags' field of the program header.
3702 Here is an example of `PHDRS'. This shows a typical set of program
3703 headers used on a native ELF system.
3707 headers PT_PHDR PHDRS ;
3709 text PT_LOAD FILEHDR PHDRS ;
3711 dynamic PT_DYNAMIC ;
3717 .interp : { *(.interp) } :text :interp
3718 .text : { *(.text) } :text
3719 .rodata : { *(.rodata) } /* defaults to :text */
3721 . = . + 0x1000; /* move to a new page in memory */
3722 .data : { *(.data) } :data
3723 .dynamic : { *(.dynamic) } :data :dynamic
3728 File: ld.info, Node: VERSION, Next: Expressions, Prev: PHDRS, Up: Scripts
3733 The linker supports symbol versions when using ELF. Symbol versions are
3734 only useful when using shared libraries. The dynamic linker can use
3735 symbol versions to select a specific version of a function when it runs
3736 a program that may have been linked against an earlier version of the
3739 You can include a version script directly in the main linker script,
3740 or you can supply the version script as an implicit linker script. You
3741 can also use the `--version-script' linker option.
3743 The syntax of the `VERSION' command is simply
3744 VERSION { version-script-commands }
3746 The format of the version script commands is identical to that used
3747 by Sun's linker in Solaris 2.5. The version script defines a tree of
3748 version nodes. You specify the node names and interdependencies in the
3749 version script. You can specify which symbols are bound to which
3750 version nodes, and you can reduce a specified set of symbols to local
3751 scope so that they are not globally visible outside of the shared
3754 The easiest way to demonstrate the version script language is with a
3774 "int f(int, double)";
3778 This example version script defines three version nodes. The first
3779 version node defined is `VERS_1.1'; it has no other dependencies. The
3780 script binds the symbol `foo1' to `VERS_1.1'. It reduces a number of
3781 symbols to local scope so that they are not visible outside of the
3782 shared library; this is done using wildcard patterns, so that any
3783 symbol whose name begins with `old', `original', or `new' is matched.
3784 The wildcard patterns available are the same as those used in the shell
3785 when matching filenames (also known as "globbing"). However, if you
3786 specify the symbol name inside double quotes, then the name is treated
3787 as literal, rather than as a glob pattern.
3789 Next, the version script defines node `VERS_1.2'. This node depends
3790 upon `VERS_1.1'. The script binds the symbol `foo2' to the version
3793 Finally, the version script defines node `VERS_2.0'. This node
3794 depends upon `VERS_1.2'. The scripts binds the symbols `bar1' and
3795 `bar2' are bound to the version node `VERS_2.0'.
3797 When the linker finds a symbol defined in a library which is not
3798 specifically bound to a version node, it will effectively bind it to an
3799 unspecified base version of the library. You can bind all otherwise
3800 unspecified symbols to a given version node by using `global: *;'
3801 somewhere in the version script.
3803 The names of the version nodes have no specific meaning other than
3804 what they might suggest to the person reading them. The `2.0' version
3805 could just as well have appeared in between `1.1' and `1.2'. However,
3806 this would be a confusing way to write a version script.
3808 Node name can be omitted, provided it is the only version node in
3809 the version script. Such version script doesn't assign any versions to
3810 symbols, only selects which symbols will be globally visible out and
3813 { global: foo; bar; local: *; };
3815 When you link an application against a shared library that has
3816 versioned symbols, the application itself knows which version of each
3817 symbol it requires, and it also knows which version nodes it needs from
3818 each shared library it is linked against. Thus at runtime, the dynamic
3819 loader can make a quick check to make sure that the libraries you have
3820 linked against do in fact supply all of the version nodes that the
3821 application will need to resolve all of the dynamic symbols. In this
3822 way it is possible for the dynamic linker to know with certainty that
3823 all external symbols that it needs will be resolvable without having to
3824 search for each symbol reference.
3826 The symbol versioning is in effect a much more sophisticated way of
3827 doing minor version checking that SunOS does. The fundamental problem
3828 that is being addressed here is that typically references to external
3829 functions are bound on an as-needed basis, and are not all bound when
3830 the application starts up. If a shared library is out of date, a
3831 required interface may be missing; when the application tries to use
3832 that interface, it may suddenly and unexpectedly fail. With symbol
3833 versioning, the user will get a warning when they start their program if
3834 the libraries being used with the application are too old.
3836 There are several GNU extensions to Sun's versioning approach. The
3837 first of these is the ability to bind a symbol to a version node in the
3838 source file where the symbol is defined instead of in the versioning
3839 script. This was done mainly to reduce the burden on the library
3840 maintainer. You can do this by putting something like:
3841 __asm__(".symver original_foo,foo@VERS_1.1");
3842 in the C source file. This renames the function `original_foo' to
3843 be an alias for `foo' bound to the version node `VERS_1.1'. The
3844 `local:' directive can be used to prevent the symbol `original_foo'
3845 from being exported. A `.symver' directive takes precedence over a
3848 The second GNU extension is to allow multiple versions of the same
3849 function to appear in a given shared library. In this way you can make
3850 an incompatible change to an interface without increasing the major
3851 version number of the shared library, while still allowing applications
3852 linked against the old interface to continue to function.
3854 To do this, you must use multiple `.symver' directives in the source
3855 file. Here is an example:
3857 __asm__(".symver original_foo,foo@");
3858 __asm__(".symver old_foo,foo@VERS_1.1");
3859 __asm__(".symver old_foo1,foo@VERS_1.2");
3860 __asm__(".symver new_foo,foo@@VERS_2.0");
3862 In this example, `foo@' represents the symbol `foo' bound to the
3863 unspecified base version of the symbol. The source file that contains
3864 this example would define 4 C functions: `original_foo', `old_foo',
3865 `old_foo1', and `new_foo'.
3867 When you have multiple definitions of a given symbol, there needs to
3868 be some way to specify a default version to which external references to
3869 this symbol will be bound. You can do this with the `foo@@VERS_2.0'
3870 type of `.symver' directive. You can only declare one version of a
3871 symbol as the default in this manner; otherwise you would effectively
3872 have multiple definitions of the same symbol.
3874 If you wish to bind a reference to a specific version of the symbol
3875 within the shared library, you can use the aliases of convenience
3876 (i.e., `old_foo'), or you can use the `.symver' directive to
3877 specifically bind to an external version of the function in question.
3879 You can also specify the language in the version script:
3881 VERSION extern "lang" { version-script-commands }
3883 The supported `lang's are `C', `C++', and `Java'. The linker will
3884 iterate over the list of symbols at the link time and demangle them
3885 according to `lang' before matching them to the patterns specified in
3886 `version-script-commands'.
3888 Demangled names may contains spaces and other special characters. As
3889 described above, you can use a glob pattern to match demangled names,
3890 or you can use a double-quoted string to match the string exactly. In
3891 the latter case, be aware that minor differences (such as differing
3892 whitespace) between the version script and the demangler output will
3893 cause a mismatch. As the exact string generated by the demangler might
3894 change in the future, even if the mangled name does not, you should
3895 check that all of your version directives are behaving as you expect
3899 File: ld.info, Node: Expressions, Next: Implicit Linker Scripts, Prev: VERSION, Up: Scripts
3901 3.10 Expressions in Linker Scripts
3902 ==================================
3904 The syntax for expressions in the linker script language is identical to
3905 that of C expressions. All expressions are evaluated as integers. All
3906 expressions are evaluated in the same size, which is 32 bits if both the
3907 host and target are 32 bits, and is otherwise 64 bits.
3909 You can use and set symbol values in expressions.
3911 The linker defines several special purpose builtin functions for use
3916 * Constants:: Constants
3917 * Symbols:: Symbol Names
3918 * Orphan Sections:: Orphan Sections
3919 * Location Counter:: The Location Counter
3920 * Operators:: Operators
3921 * Evaluation:: Evaluation
3922 * Expression Section:: The Section of an Expression
3923 * Builtin Functions:: Builtin Functions
3926 File: ld.info, Node: Constants, Next: Symbols, Up: Expressions
3931 All constants are integers.
3933 As in C, the linker considers an integer beginning with `0' to be
3934 octal, and an integer beginning with `0x' or `0X' to be hexadecimal.
3935 The linker considers other integers to be decimal.
3937 In addition, you can use the suffixes `K' and `M' to scale a
3938 constant by `1024' or `1024*1024' respectively. For example, the
3939 following all refer to the same quantity:
3945 File: ld.info, Node: Symbols, Next: Orphan Sections, Prev: Constants, Up: Expressions
3950 Unless quoted, symbol names start with a letter, underscore, or period
3951 and may include letters, digits, underscores, periods, and hyphens.
3952 Unquoted symbol names must not conflict with any keywords. You can
3953 specify a symbol which contains odd characters or has the same name as a
3954 keyword by surrounding the symbol name in double quotes:
3956 "with a space" = "also with a space" + 10;
3958 Since symbols can contain many non-alphabetic characters, it is
3959 safest to delimit symbols with spaces. For example, `A-B' is one
3960 symbol, whereas `A - B' is an expression involving subtraction.
3963 File: ld.info, Node: Orphan Sections, Next: Location Counter, Prev: Symbols, Up: Expressions
3965 3.10.3 Orphan Sections
3966 ----------------------
3968 Orphan sections are sections present in the input files which are not
3969 explicitly placed into the output file by the linker script. The
3970 linker will still copy these sections into the output file, but it has
3971 to guess as to where they should be placed. The linker uses a simple
3972 heuristic to do this. It attempts to place orphan sections after
3973 non-orphan sections of the same attribute, such as code vs data,
3974 loadable vs non-loadable, etc. If there is not enough room to do this
3975 then it places at the end of the file.
3977 For ELF targets, the attribute of the section includes section type
3978 as well as section flag.
3980 If an orphaned section's name is representable as a C identifier then
3981 the linker will automatically *note PROVIDE:: two symbols:
3982 __start_SECNAME and __end_SECNAME, where SECNAME is the name of the
3983 section. These indicate the start address and end address of the
3984 orphaned section respectively. Note: most section names are not
3985 representable as C identifiers because they contain a `.' character.
3988 File: ld.info, Node: Location Counter, Next: Operators, Prev: Orphan Sections, Up: Expressions
3990 3.10.4 The Location Counter
3991 ---------------------------
3993 The special linker variable "dot" `.' always contains the current
3994 output location counter. Since the `.' always refers to a location in
3995 an output section, it may only appear in an expression within a
3996 `SECTIONS' command. The `.' symbol may appear anywhere that an
3997 ordinary symbol is allowed in an expression.
3999 Assigning a value to `.' will cause the location counter to be
4000 moved. This may be used to create holes in the output section. The
4001 location counter may not be moved backwards inside an output section,
4002 and may not be moved backwards outside of an output section if so doing
4003 creates areas with overlapping LMAs.
4016 In the previous example, the `.text' section from `file1' is located
4017 at the beginning of the output section `output'. It is followed by a
4018 1000 byte gap. Then the `.text' section from `file2' appears, also
4019 with a 1000 byte gap following before the `.text' section from `file3'.
4020 The notation `= 0x12345678' specifies what data to write in the gaps
4021 (*note Output Section Fill::).
4023 Note: `.' actually refers to the byte offset from the start of the
4024 current containing object. Normally this is the `SECTIONS' statement,
4025 whose start address is 0, hence `.' can be used as an absolute address.
4026 If `.' is used inside a section description however, it refers to the
4027 byte offset from the start of that section, not an absolute address.
4028 Thus in a script like this:
4044 The `.text' section will be assigned a starting address of 0x100 and
4045 a size of exactly 0x200 bytes, even if there is not enough data in the
4046 `.text' input sections to fill this area. (If there is too much data,
4047 an error will be produced because this would be an attempt to move `.'
4048 backwards). The `.data' section will start at 0x500 and it will have
4049 an extra 0x600 bytes worth of space after the end of the values from
4050 the `.data' input sections and before the end of the `.data' output
4053 Setting symbols to the value of the location counter outside of an
4054 output section statement can result in unexpected values if the linker
4055 needs to place orphan sections. For example, given the following:
4068 If the linker needs to place some input section, e.g. `.rodata', not
4069 mentioned in the script, it might choose to place that section between
4070 `.text' and `.data'. You might think the linker should place `.rodata'
4071 on the blank line in the above script, but blank lines are of no
4072 particular significance to the linker. As well, the linker doesn't
4073 associate the above symbol names with their sections. Instead, it
4074 assumes that all assignments or other statements belong to the previous
4075 output section, except for the special case of an assignment to `.'.
4076 I.e., the linker will place the orphan `.rodata' section as if the
4077 script was written as follows:
4086 .rodata: { *(.rodata) }
4091 This may or may not be the script author's intention for the value of
4092 `start_of_data'. One way to influence the orphan section placement is
4093 to assign the location counter to itself, as the linker assumes that an
4094 assignment to `.' is setting the start address of a following output
4095 section and thus should be grouped with that section. So you could
4110 Now, the orphan `.rodata' section will be placed between
4111 `end_of_text' and `start_of_data'.
4114 File: ld.info, Node: Operators, Next: Evaluation, Prev: Location Counter, Up: Expressions
4119 The linker recognizes the standard C set of arithmetic operators, with
4120 the standard bindings and precedence levels:
4121 precedence associativity Operators Notes
4127 5 left == != > < <= >=
4133 11 right &= += -= *= /= (2)
4135 Notes: (1) Prefix operators (2) *Note Assignments::.
4138 File: ld.info, Node: Evaluation, Next: Expression Section, Prev: Operators, Up: Expressions
4143 The linker evaluates expressions lazily. It only computes the value of
4144 an expression when absolutely necessary.
4146 The linker needs some information, such as the value of the start
4147 address of the first section, and the origins and lengths of memory
4148 regions, in order to do any linking at all. These values are computed
4149 as soon as possible when the linker reads in the linker script.
4151 However, other values (such as symbol values) are not known or needed
4152 until after storage allocation. Such values are evaluated later, when
4153 other information (such as the sizes of output sections) is available
4154 for use in the symbol assignment expression.
4156 The sizes of sections cannot be known until after allocation, so
4157 assignments dependent upon these are not performed until after
4160 Some expressions, such as those depending upon the location counter
4161 `.', must be evaluated during section allocation.
4163 If the result of an expression is required, but the value is not
4164 available, then an error results. For example, a script like the
4168 .text 9+this_isnt_constant :
4171 will cause the error message `non constant expression for initial
4175 File: ld.info, Node: Expression Section, Next: Builtin Functions, Prev: Evaluation, Up: Expressions
4177 3.10.7 The Section of an Expression
4178 -----------------------------------
4180 When the linker evaluates an expression, the result is either absolute
4181 or relative to some section. A relative expression is expressed as a
4182 fixed offset from the base of a section.
4184 The position of the expression within the linker script determines
4185 whether it is absolute or relative. An expression which appears within
4186 an output section definition is relative to the base of the output
4187 section. An expression which appears elsewhere will be absolute.
4189 A symbol set to a relative expression will be relocatable if you
4190 request relocatable output using the `-r' option. That means that a
4191 further link operation may change the value of the symbol. The symbol's
4192 section will be the section of the relative expression.
4194 A symbol set to an absolute expression will retain the same value
4195 through any further link operation. The symbol will be absolute, and
4196 will not have any particular associated section.
4198 You can use the builtin function `ABSOLUTE' to force an expression
4199 to be absolute when it would otherwise be relative. For example, to
4200 create an absolute symbol set to the address of the end of the output
4204 .data : { *(.data) _edata = ABSOLUTE(.); }
4206 If `ABSOLUTE' were not used, `_edata' would be relative to the
4210 File: ld.info, Node: Builtin Functions, Prev: Expression Section, Up: Expressions
4212 3.10.8 Builtin Functions
4213 ------------------------
4215 The linker script language includes a number of builtin functions for
4216 use in linker script expressions.
4219 Return the absolute (non-relocatable, as opposed to non-negative)
4220 value of the expression EXP. Primarily useful to assign an
4221 absolute value to a symbol within a section definition, where
4222 symbol values are normally section relative. *Note Expression
4226 Return the absolute address (the VMA) of the named SECTION. Your
4227 script must previously have defined the location of that section.
4228 In the following example, `symbol_1' and `symbol_2' are assigned
4233 start_of_output_1 = ABSOLUTE(.);
4238 symbol_1 = ADDR(.output1);
4239 symbol_2 = start_of_output_1;
4245 Return the location counter (`.') or arbitrary expression aligned
4246 to the next ALIGN boundary. The single operand `ALIGN' doesn't
4247 change the value of the location counter--it just does arithmetic
4248 on it. The two operand `ALIGN' allows an arbitrary expression to
4249 be aligned upwards (`ALIGN(ALIGN)' is equivalent to `ALIGN(.,
4252 Here is an example which aligns the output `.data' section to the
4253 next `0x2000' byte boundary after the preceding section and sets a
4254 variable within the section to the next `0x8000' boundary after the
4257 .data ALIGN(0x2000): {
4259 variable = ALIGN(0x8000);
4262 The first use of `ALIGN' in this example specifies the
4263 location of a section because it is used as the optional ADDRESS
4264 attribute of a section definition (*note Output Section
4265 Address::). The second use of `ALIGN' is used to defines the
4268 The builtin function `NEXT' is closely related to `ALIGN'.
4271 Return the alignment in bytes of the named SECTION, if that
4272 section has been allocated. If the section has not been allocated
4273 when this is evaluated, the linker will report an error. In the
4274 following example, the alignment of the `.output' section is
4275 stored as the first value in that section.
4278 LONG (ALIGNOF (.output))
4284 This is a synonym for `ALIGN', for compatibility with older linker
4285 scripts. It is most often seen when setting the address of an
4288 `DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE)'
4289 This is equivalent to either
4290 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - 1)))
4292 (ALIGN(MAXPAGESIZE) + (. & (MAXPAGESIZE - COMMONPAGESIZE)))
4293 depending on whether the latter uses fewer COMMONPAGESIZE sized
4294 pages for the data segment (area between the result of this
4295 expression and `DATA_SEGMENT_END') than the former or not. If the
4296 latter form is used, it means COMMONPAGESIZE bytes of runtime
4297 memory will be saved at the expense of up to COMMONPAGESIZE wasted
4298 bytes in the on-disk file.
4300 This expression can only be used directly in `SECTIONS' commands,
4301 not in any output section descriptions and only once in the linker
4302 script. COMMONPAGESIZE should be less or equal to MAXPAGESIZE and
4303 should be the system page size the object wants to be optimized
4304 for (while still working on system page sizes up to MAXPAGESIZE).
4307 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
4309 `DATA_SEGMENT_END(EXP)'
4310 This defines the end of data segment for `DATA_SEGMENT_ALIGN'
4311 evaluation purposes.
4313 . = DATA_SEGMENT_END(.);
4315 `DATA_SEGMENT_RELRO_END(OFFSET, EXP)'
4316 This defines the end of the `PT_GNU_RELRO' segment when `-z relro'
4317 option is used. Second argument is returned. When `-z relro'
4318 option is not present, `DATA_SEGMENT_RELRO_END' does nothing,
4319 otherwise `DATA_SEGMENT_ALIGN' is padded so that EXP + OFFSET is
4320 aligned to the most commonly used page boundary for particular
4321 target. If present in the linker script, it must always come in
4322 between `DATA_SEGMENT_ALIGN' and `DATA_SEGMENT_END'.
4324 . = DATA_SEGMENT_RELRO_END(24, .);
4327 Return 1 if SYMBOL is in the linker global symbol table and is
4328 defined before the statement using DEFINED in the script, otherwise
4329 return 0. You can use this function to provide default values for
4330 symbols. For example, the following script fragment shows how to
4331 set a global symbol `begin' to the first location in the `.text'
4332 section--but if a symbol called `begin' already existed, its value
4337 begin = DEFINED(begin) ? begin : . ;
4344 Return the length of the memory region named MEMORY.
4347 Return the absolute LMA of the named SECTION. This is normally
4348 the same as `ADDR', but it may be different if the `AT' attribute
4349 is used in the output section definition (*note Output Section
4353 Returns the maximum of EXP1 and EXP2.
4356 Returns the minimum of EXP1 and EXP2.
4359 Return the next unallocated address that is a multiple of EXP.
4360 This function is closely related to `ALIGN(EXP)'; unless you use
4361 the `MEMORY' command to define discontinuous memory for the output
4362 file, the two functions are equivalent.
4365 Return the origin of the memory region named MEMORY.
4367 `SEGMENT_START(SEGMENT, DEFAULT)'
4368 Return the base address of the named SEGMENT. If an explicit
4369 value has been given for this segment (with a command-line `-T'
4370 option) that value will be returned; otherwise the value will be
4371 DEFAULT. At present, the `-T' command-line option can only be
4372 used to set the base address for the "text", "data", and "bss"
4373 sections, but you use `SEGMENT_START' with any segment name.
4376 Return the size in bytes of the named SECTION, if that section has
4377 been allocated. If the section has not been allocated when this is
4378 evaluated, the linker will report an error. In the following
4379 example, `symbol_1' and `symbol_2' are assigned identical values:
4386 symbol_1 = .end - .start ;
4387 symbol_2 = SIZEOF(.output);
4392 Return the size in bytes of the output file's headers. This is
4393 information which appears at the start of the output file. You
4394 can use this number when setting the start address of the first
4395 section, if you choose, to facilitate paging.
4397 When producing an ELF output file, if the linker script uses the
4398 `SIZEOF_HEADERS' builtin function, the linker must compute the
4399 number of program headers before it has determined all the section
4400 addresses and sizes. If the linker later discovers that it needs
4401 additional program headers, it will report an error `not enough
4402 room for program headers'. To avoid this error, you must avoid
4403 using the `SIZEOF_HEADERS' function, or you must rework your linker
4404 script to avoid forcing the linker to use additional program
4405 headers, or you must define the program headers yourself using the
4406 `PHDRS' command (*note PHDRS::).
4409 File: ld.info, Node: Implicit Linker Scripts, Prev: Expressions, Up: Scripts
4411 3.11 Implicit Linker Scripts
4412 ============================
4414 If you specify a linker input file which the linker can not recognize as
4415 an object file or an archive file, it will try to read the file as a
4416 linker script. If the file can not be parsed as a linker script, the
4417 linker will report an error.
4419 An implicit linker script will not replace the default linker script.
4421 Typically an implicit linker script would contain only symbol
4422 assignments, or the `INPUT', `GROUP', or `VERSION' commands.
4424 Any input files read because of an implicit linker script will be
4425 read at the position in the command line where the implicit linker
4426 script was read. This can affect archive searching.
4429 File: ld.info, Node: Machine Dependent, Next: BFD, Prev: Scripts, Up: Top
4431 4 Machine Dependent Features
4432 ****************************
4434 `ld' has additional features on some platforms; the following sections
4435 describe them. Machines where `ld' has no additional functionality are
4441 * H8/300:: `ld' and the H8/300
4443 * i960:: `ld' and the Intel 960 family
4445 * ARM:: `ld' and the ARM family
4447 * HPPA ELF32:: `ld' and HPPA 32-bit ELF
4449 * M68K:: `ld' and the Motorola 68K family
4451 * MMIX:: `ld' and MMIX
4453 * MSP430:: `ld' and MSP430
4455 * M68HC11/68HC12:: `ld' and the Motorola 68HC11 and 68HC12 families
4457 * PowerPC ELF32:: `ld' and PowerPC 32-bit ELF Support
4459 * PowerPC64 ELF64:: `ld' and PowerPC64 64-bit ELF Support
4461 * SPU ELF:: `ld' and SPU ELF Support
4463 * TI COFF:: `ld' and TI COFF
4465 * WIN32:: `ld' and WIN32 (cygwin/mingw)
4467 * Xtensa:: `ld' and Xtensa Processors
4470 File: ld.info, Node: H8/300, Next: i960, Up: Machine Dependent
4472 4.1 `ld' and the H8/300
4473 =======================
4475 For the H8/300, `ld' can perform these global optimizations when you
4476 specify the `--relax' command-line option.
4478 _relaxing address modes_
4479 `ld' finds all `jsr' and `jmp' instructions whose targets are
4480 within eight bits, and turns them into eight-bit program-counter
4481 relative `bsr' and `bra' instructions, respectively.
4483 _synthesizing instructions_
4484 `ld' finds all `mov.b' instructions which use the sixteen-bit
4485 absolute address form, but refer to the top page of memory, and
4486 changes them to use the eight-bit address form. (That is: the
4487 linker turns `mov.b `@'AA:16' into `mov.b `@'AA:8' whenever the
4488 address AA is in the top page of memory).
4490 _bit manipulation instructions_
4491 `ld' finds all bit manipulation instructions like `band, bclr,
4492 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst,
4493 bxor' which use 32 bit and 16 bit absolute address form, but refer
4494 to the top page of memory, and changes them to use the 8 bit
4495 address form. (That is: the linker turns `bset #xx:3,`@'AA:32'
4496 into `bset #xx:3,`@'AA:8' whenever the address AA is in the top
4499 _system control instructions_
4500 `ld' finds all `ldc.w, stc.w' instructions which use the 32 bit
4501 absolute address form, but refer to the top page of memory, and
4502 changes them to use 16 bit address form. (That is: the linker
4503 turns `ldc.w `@'AA:32,ccr' into `ldc.w `@'AA:16,ccr' whenever the
4504 address AA is in the top page of memory).
4507 File: ld.info, Node: i960, Next: ARM, Prev: H8/300, Up: Machine Dependent
4509 4.2 `ld' and the Intel 960 Family
4510 =================================
4512 You can use the `-AARCHITECTURE' command line option to specify one of
4513 the two-letter names identifying members of the 960 family; the option
4514 specifies the desired output target, and warns of any incompatible
4515 instructions in the input files. It also modifies the linker's search
4516 strategy for archive libraries, to support the use of libraries
4517 specific to each particular architecture, by including in the search
4518 loop names suffixed with the string identifying the architecture.
4520 For example, if your `ld' command line included `-ACA' as well as
4521 `-ltry', the linker would look (in its built-in search paths, and in
4522 any paths you specify with `-L') for a library with the names
4529 The first two possibilities would be considered in any event; the last
4530 two are due to the use of `-ACA'.
4532 You can meaningfully use `-A' more than once on a command line, since
4533 the 960 architecture family allows combination of target architectures;
4534 each use will add another pair of name variants to search for when `-l'
4535 specifies a library.
4537 `ld' supports the `--relax' option for the i960 family. If you
4538 specify `--relax', `ld' finds all `balx' and `calx' instructions whose
4539 targets are within 24 bits, and turns them into 24-bit program-counter
4540 relative `bal' and `cal' instructions, respectively. `ld' also turns
4541 `cal' instructions into `bal' instructions when it determines that the
4542 target subroutine is a leaf routine (that is, the target subroutine does
4543 not itself call any subroutines).
4545 The `--fix-cortex-a8' switch enables a link-time workaround for an
4546 erratum in certain Cortex-A8 processors. The workaround is enabled by
4547 default if you are targeting the ARM v7-A architecture profile. It can
4548 be enabled otherwise by specifying `--fix-cortex-a8', or disabled
4549 unconditionally by specifying `--no-fix-cortex-a8'.
4551 The erratum only affects Thumb-2 code. Please contact ARM for
4555 File: ld.info, Node: M68HC11/68HC12, Next: PowerPC ELF32, Prev: MSP430, Up: Machine Dependent
4557 4.3 `ld' and the Motorola 68HC11 and 68HC12 families
4558 ====================================================
4560 4.3.1 Linker Relaxation
4561 -----------------------
4563 For the Motorola 68HC11, `ld' can perform these global optimizations
4564 when you specify the `--relax' command-line option.
4566 _relaxing address modes_
4567 `ld' finds all `jsr' and `jmp' instructions whose targets are
4568 within eight bits, and turns them into eight-bit program-counter
4569 relative `bsr' and `bra' instructions, respectively.
4571 `ld' also looks at all 16-bit extended addressing modes and
4572 transforms them in a direct addressing mode when the address is in
4573 page 0 (between 0 and 0x0ff).
4575 _relaxing gcc instruction group_
4576 When `gcc' is called with `-mrelax', it can emit group of
4577 instructions that the linker can optimize to use a 68HC11 direct
4578 addressing mode. These instructions consists of `bclr' or `bset'
4582 4.3.2 Trampoline Generation
4583 ---------------------------
4585 For 68HC11 and 68HC12, `ld' can generate trampoline code to call a far
4586 function using a normal `jsr' instruction. The linker will also change
4587 the relocation to some far function to use the trampoline address
4588 instead of the function address. This is typically the case when a
4589 pointer to a function is taken. The pointer will in fact point to the
4590 function trampoline.
4593 File: ld.info, Node: ARM, Next: HPPA ELF32, Prev: i960, Up: Machine Dependent
4595 4.4 `ld' and the ARM family
4596 ===========================
4598 For the ARM, `ld' will generate code stubs to allow functions calls
4599 between ARM and Thumb code. These stubs only work with code that has
4600 been compiled and assembled with the `-mthumb-interwork' command line
4601 option. If it is necessary to link with old ARM object files or
4602 libraries, which have not been compiled with the -mthumb-interwork
4603 option then the `--support-old-code' command line switch should be
4604 given to the linker. This will make it generate larger stub functions
4605 which will work with non-interworking aware ARM code. Note, however,
4606 the linker does not support generating stubs for function calls to
4607 non-interworking aware Thumb code.
4609 The `--thumb-entry' switch is a duplicate of the generic `--entry'
4610 switch, in that it sets the program's starting address. But it also
4611 sets the bottom bit of the address, so that it can be branched to using
4612 a BX instruction, and the program will start executing in Thumb mode
4615 The `--be8' switch instructs `ld' to generate BE8 format
4616 executables. This option is only valid when linking big-endian objects.
4617 The resulting image will contain big-endian data and little-endian code.
4619 The `R_ARM_TARGET1' relocation is typically used for entries in the
4620 `.init_array' section. It is interpreted as either `R_ARM_REL32' or
4621 `R_ARM_ABS32', depending on the target. The `--target1-rel' and
4622 `--target1-abs' switches override the default.
4624 The `--target2=type' switch overrides the default definition of the
4625 `R_ARM_TARGET2' relocation. Valid values for `type', their meanings,
4626 and target defaults are as follows:
4628 `R_ARM_REL32' (arm*-*-elf, arm*-*-eabi)
4631 `R_ARM_ABS32' (arm*-*-symbianelf)
4634 `R_ARM_GOT_PREL' (arm*-*-linux, arm*-*-*bsd)
4636 The `R_ARM_V4BX' relocation (defined by the ARM AAELF specification)
4637 enables objects compiled for the ARMv4 architecture to be
4638 interworking-safe when linked with other objects compiled for ARMv4t,
4639 but also allows pure ARMv4 binaries to be built from the same ARMv4
4642 In the latter case, the switch `--fix-v4bx' must be passed to the
4643 linker, which causes v4t `BX rM' instructions to be rewritten as `MOV
4644 PC,rM', since v4 processors do not have a `BX' instruction.
4646 In the former case, the switch should not be used, and `R_ARM_V4BX'
4647 relocations are ignored.
4649 Replace `BX rM' instructions identified by `R_ARM_V4BX' relocations
4650 with a branch to the following veneer:
4656 This allows generation of libraries/applications that work on ARMv4
4657 cores and are still interworking safe. Note that the above veneer
4658 clobbers the condition flags, so may cause incorrect progrm behavior in
4661 The `--use-blx' switch enables the linker to use ARM/Thumb BLX
4662 instructions (available on ARMv5t and above) in various situations.
4663 Currently it is used to perform calls via the PLT from Thumb code using
4664 BLX rather than using BX and a mode-switching stub before each PLT
4665 entry. This should lead to such calls executing slightly faster.
4667 This option is enabled implicitly for SymbianOS, so there is no need
4668 to specify it if you are using that target.
4670 The `--vfp11-denorm-fix' switch enables a link-time workaround for a
4671 bug in certain VFP11 coprocessor hardware, which sometimes allows
4672 instructions with denorm operands (which must be handled by support
4673 code) to have those operands overwritten by subsequent instructions
4674 before the support code can read the intended values.
4676 The bug may be avoided in scalar mode if you allow at least one
4677 intervening instruction between a VFP11 instruction which uses a
4678 register and another instruction which writes to the same register, or
4679 at least two intervening instructions if vector mode is in use. The bug
4680 only affects full-compliance floating-point mode: you do not need this
4681 workaround if you are using "runfast" mode. Please contact ARM for
4684 If you know you are using buggy VFP11 hardware, you can enable this
4685 workaround by specifying the linker option `--vfp-denorm-fix=scalar' if
4686 you are using the VFP11 scalar mode only, or `--vfp-denorm-fix=vector'
4687 if you are using vector mode (the latter also works for scalar code).
4688 The default is `--vfp-denorm-fix=none'.
4690 If the workaround is enabled, instructions are scanned for
4691 potentially-troublesome sequences, and a veneer is created for each
4692 such sequence which may trigger the erratum. The veneer consists of the
4693 first instruction of the sequence and a branch back to the subsequent
4694 instruction. The original instruction is then replaced with a branch to
4695 the veneer. The extra cycles required to call and return from the veneer
4696 are sufficient to avoid the erratum in both the scalar and vector cases.
4698 The `--no-enum-size-warning' switch prevents the linker from warning
4699 when linking object files that specify incompatible EABI enumeration
4700 size attributes. For example, with this switch enabled, linking of an
4701 object file using 32-bit enumeration values with another using
4702 enumeration values fitted into the smallest possible space will not be
4705 The `--no-wchar-size-warning' switch prevents the linker from
4706 warning when linking object files that specify incompatible EABI
4707 `wchar_t' size attributes. For example, with this switch enabled,
4708 linking of an object file using 32-bit `wchar_t' values with another
4709 using 16-bit `wchar_t' values will not be diagnosed.
4711 The `--pic-veneer' switch makes the linker use PIC sequences for
4712 ARM/Thumb interworking veneers, even if the rest of the binary is not
4713 PIC. This avoids problems on uClinux targets where `--emit-relocs' is
4714 used to generate relocatable binaries.
4716 The linker will automatically generate and insert small sequences of
4717 code into a linked ARM ELF executable whenever an attempt is made to
4718 perform a function call to a symbol that is too far away. The
4719 placement of these sequences of instructions - called stubs - is
4720 controlled by the command line option `--stub-group-size=N'. The
4721 placement is important because a poor choice can create a need for
4722 duplicate stubs, increasing the code sizw. The linker will try to
4723 group stubs together in order to reduce interruptions to the flow of
4724 code, but it needs guidance as to how big these groups should be and
4725 where they should be placed.
4727 The value of `N', the parameter to the `--stub-group-size=' option
4728 controls where the stub groups are placed. If it is negative then all
4729 stubs are placed after the first branch that needs them. If it is
4730 positive then the stubs can be placed either before or after the
4731 branches that need them. If the value of `N' is 1 (either +1 or -1)
4732 then the linker will choose exactly where to place groups of stubs,
4733 using its built in heuristics. A value of `N' greater than 1 (or
4734 smaller than -1) tells the linker that a single group of stubs can
4735 service at most `N' bytes from the input sections.
4737 The default, if `--stub-group-size=' is not specified, is `N = +1'.
4739 Farcalls stubs insertion is fully supported for the ARM-EABI target
4740 only, because it relies on object files properties not present
4744 File: ld.info, Node: HPPA ELF32, Next: M68K, Prev: ARM, Up: Machine Dependent
4746 4.5 `ld' and HPPA 32-bit ELF Support
4747 ====================================
4749 When generating a shared library, `ld' will by default generate import
4750 stubs suitable for use with a single sub-space application. The
4751 `--multi-subspace' switch causes `ld' to generate export stubs, and
4752 different (larger) import stubs suitable for use with multiple
4755 Long branch stubs and import/export stubs are placed by `ld' in stub
4756 sections located between groups of input sections. `--stub-group-size'
4757 specifies the maximum size of a group of input sections handled by one
4758 stub section. Since branch offsets are signed, a stub section may
4759 serve two groups of input sections, one group before the stub section,
4760 and one group after it. However, when using conditional branches that
4761 require stubs, it may be better (for branch prediction) that stub
4762 sections only serve one group of input sections. A negative value for
4763 `N' chooses this scheme, ensuring that branches to stubs always use a
4764 negative offset. Two special values of `N' are recognized, `1' and
4765 `-1'. These both instruct `ld' to automatically size input section
4766 groups for the branch types detected, with the same behaviour regarding
4767 stub placement as other positive or negative values of `N' respectively.
4769 Note that `--stub-group-size' does not split input sections. A
4770 single input section larger than the group size specified will of course
4771 create a larger group (of one section). If input sections are too
4772 large, it may not be possible for a branch to reach its stub.
4775 File: ld.info, Node: M68K, Next: MMIX, Prev: HPPA ELF32, Up: Machine Dependent
4777 4.6 `ld' and the Motorola 68K family
4778 ====================================
4780 The `--got=TYPE' option lets you choose the GOT generation scheme. The
4781 choices are `single', `negative', `multigot' and `target'. When
4782 `target' is selected the linker chooses the default GOT generation
4783 scheme for the current target. `single' tells the linker to generate a
4784 single GOT with entries only at non-negative offsets. `negative'
4785 instructs the linker to generate a single GOT with entries at both
4786 negative and positive offsets. Not all environments support such GOTs.
4787 `multigot' allows the linker to generate several GOTs in the output
4788 file. All GOT references from a single input object file access the
4789 same GOT, but references from different input object files might access
4790 different GOTs. Not all environments support such GOTs.
4793 File: ld.info, Node: MMIX, Next: MSP430, Prev: M68K, Up: Machine Dependent
4798 For MMIX, there is a choice of generating `ELF' object files or `mmo'
4799 object files when linking. The simulator `mmix' understands the `mmo'
4800 format. The binutils `objcopy' utility can translate between the two
4803 There is one special section, the `.MMIX.reg_contents' section.
4804 Contents in this section is assumed to correspond to that of global
4805 registers, and symbols referring to it are translated to special
4806 symbols, equal to registers. In a final link, the start address of the
4807 `.MMIX.reg_contents' section corresponds to the first allocated global
4808 register multiplied by 8. Register `$255' is not included in this
4809 section; it is always set to the program entry, which is at the symbol
4810 `Main' for `mmo' files.
4812 Global symbols with the prefix `__.MMIX.start.', for example
4813 `__.MMIX.start..text' and `__.MMIX.start..data' are special. The
4814 default linker script uses these to set the default start address of a
4817 Initial and trailing multiples of zero-valued 32-bit words in a
4818 section, are left out from an mmo file.
4821 File: ld.info, Node: MSP430, Next: M68HC11/68HC12, Prev: MMIX, Up: Machine Dependent
4826 For the MSP430 it is possible to select the MPU architecture. The flag
4827 `-m [mpu type]' will select an appropriate linker script for selected
4828 MPU type. (To get a list of known MPUs just pass `-m help' option to
4831 The linker will recognize some extra sections which are MSP430
4835 Defines a portion of ROM where interrupt vectors located.
4838 Defines the bootloader portion of the ROM (if applicable). Any
4839 code in this section will be uploaded to the MPU.
4842 Defines an information memory section (if applicable). Any code in
4843 this section will be uploaded to the MPU.
4846 This is the same as the `.infomem' section except that any code in
4847 this section will not be uploaded to the MPU.
4850 Denotes a portion of RAM located above `.bss' section.
4852 The last two sections are used by gcc.
4855 File: ld.info, Node: PowerPC ELF32, Next: PowerPC64 ELF64, Prev: M68HC11/68HC12, Up: Machine Dependent
4857 4.9 `ld' and PowerPC 32-bit ELF Support
4858 =======================================
4860 Branches on PowerPC processors are limited to a signed 26-bit
4861 displacement, which may result in `ld' giving `relocation truncated to
4862 fit' errors with very large programs. `--relax' enables the generation
4863 of trampolines that can access the entire 32-bit address space. These
4864 trampolines are inserted at section boundaries, so may not themselves
4865 be reachable if an input section exceeds 33M in size.
4868 Current PowerPC GCC accepts a `-msecure-plt' option that generates
4869 code capable of using a newer PLT and GOT layout that has the
4870 security advantage of no executable section ever needing to be
4871 writable and no writable section ever being executable. PowerPC
4872 `ld' will generate this layout, including stubs to access the PLT,
4873 if all input files (including startup and static libraries) were
4874 compiled with `-msecure-plt'. `--bss-plt' forces the old BSS PLT
4875 (and GOT layout) which can give slightly better performance.
4878 `ld' will use the new PLT and GOT layout if it is linking new
4879 `-fpic' or `-fPIC' code, but does not do so automatically when
4880 linking non-PIC code. This option requests the new PLT and GOT
4881 layout. A warning will be given if some object file requires the
4885 The new secure PLT and GOT are placed differently relative to other
4886 sections compared to older BSS PLT and GOT placement. The
4887 location of `.plt' must change because the new secure PLT is an
4888 initialized section while the old PLT is uninitialized. The
4889 reason for the `.got' change is more subtle: The new placement
4890 allows `.got' to be read-only in applications linked with `-z
4891 relro -z now'. However, this placement means that `.sdata' cannot
4892 always be used in shared libraries, because the PowerPC ABI
4893 accesses `.sdata' in shared libraries from the GOT pointer.
4894 `--sdata-got' forces the old GOT placement. PowerPC GCC doesn't
4895 use `.sdata' in shared libraries, so this option is really only
4896 useful for other compilers that may do so.
4899 This option causes `ld' to label linker stubs with a local symbol
4900 that encodes the stub type and destination.
4903 PowerPC `ld' normally performs some optimization of code sequences
4904 used to access Thread-Local Storage. Use this option to disable
4908 File: ld.info, Node: PowerPC64 ELF64, Next: SPU ELF, Prev: PowerPC ELF32, Up: Machine Dependent
4910 4.10 `ld' and PowerPC64 64-bit ELF Support
4911 ==========================================
4914 Long branch stubs, PLT call stubs and TOC adjusting stubs are
4915 placed by `ld' in stub sections located between groups of input
4916 sections. `--stub-group-size' specifies the maximum size of a
4917 group of input sections handled by one stub section. Since branch
4918 offsets are signed, a stub section may serve two groups of input
4919 sections, one group before the stub section, and one group after
4920 it. However, when using conditional branches that require stubs,
4921 it may be better (for branch prediction) that stub sections only
4922 serve one group of input sections. A negative value for `N'
4923 chooses this scheme, ensuring that branches to stubs always use a
4924 negative offset. Two special values of `N' are recognized, `1'
4925 and `-1'. These both instruct `ld' to automatically size input
4926 section groups for the branch types detected, with the same
4927 behaviour regarding stub placement as other positive or negative
4928 values of `N' respectively.
4930 Note that `--stub-group-size' does not split input sections. A
4931 single input section larger than the group size specified will of
4932 course create a larger group (of one section). If input sections
4933 are too large, it may not be possible for a branch to reach its
4937 This option causes `ld' to label linker stubs with a local symbol
4938 that encodes the stub type and destination.
4940 `--dotsyms, --no-dotsyms'
4941 These two options control how `ld' interprets version patterns in
4942 a version script. Older PowerPC64 compilers emitted both a
4943 function descriptor symbol with the same name as the function, and
4944 a code entry symbol with the name prefixed by a dot (`.'). To
4945 properly version a function `foo', the version script thus needs
4946 to control both `foo' and `.foo'. The option `--dotsyms', on by
4947 default, automatically adds the required dot-prefixed patterns.
4948 Use `--no-dotsyms' to disable this feature.
4951 PowerPC64 `ld' normally performs some optimization of code
4952 sequences used to access Thread-Local Storage. Use this option to
4953 disable the optimization.
4956 PowerPC64 `ld' normally removes `.opd' section entries
4957 corresponding to deleted link-once functions, or functions removed
4958 by the action of `--gc-sections' or linker script `/DISCARD/'.
4959 Use this option to disable `.opd' optimization.
4961 `--non-overlapping-opd'
4962 Some PowerPC64 compilers have an option to generate compressed
4963 `.opd' entries spaced 16 bytes apart, overlapping the third word,
4964 the static chain pointer (unused in C) with the first word of the
4965 next entry. This option expands such entries to the full 24 bytes.
4968 PowerPC64 `ld' normally removes unused `.toc' section entries.
4969 Such entries are detected by examining relocations that reference
4970 the TOC in code sections. A reloc in a deleted code section marks
4971 a TOC word as unneeded, while a reloc in a kept code section marks
4972 a TOC word as needed. Since the TOC may reference itself, TOC
4973 relocs are also examined. TOC words marked as both needed and
4974 unneeded will of course be kept. TOC words without any referencing
4975 reloc are assumed to be part of a multi-word entry, and are kept or
4976 discarded as per the nearest marked preceding word. This works
4977 reliably for compiler generated code, but may be incorrect if
4978 assembly code is used to insert TOC entries. Use this option to
4979 disable the optimization.
4982 By default, PowerPC64 GCC generates code for a TOC model where TOC
4983 entries are accessed with a 16-bit offset from r2. This limits the
4984 total TOC size to 64K. PowerPC64 `ld' extends this limit by
4985 grouping code sections such that each group uses less than 64K for
4986 its TOC entries, then inserts r2 adjusting stubs between
4987 inter-group calls. `ld' does not split apart input sections, so
4988 cannot help if a single input file has a `.toc' section that
4989 exceeds 64K, most likely from linking multiple files with `ld -r'.
4990 Use this option to turn off this feature.
4993 File: ld.info, Node: SPU ELF, Next: TI COFF, Prev: PowerPC64 ELF64, Up: Machine Dependent
4995 4.11 `ld' and SPU ELF Support
4996 =============================
4999 This option marks an executable as a PIC plugin module.
5002 Normally, `ld' recognizes calls to functions within overlay
5003 regions, and redirects such calls to an overlay manager via a stub.
5004 `ld' also provides a built-in overlay manager. This option turns
5005 off all this special overlay handling.
5008 This option causes `ld' to label overlay stubs with a local symbol
5009 that encodes the stub type and destination.
5011 `--extra-overlay-stubs'
5012 This option causes `ld' to add overlay call stubs on all function
5013 calls out of overlay regions. Normally stubs are not added on
5014 calls to non-overlay regions.
5016 `--local-store=lo:hi'
5017 `ld' usually checks that a final executable for SPU fits in the
5018 address range 0 to 256k. This option may be used to change the
5019 range. Disable the check entirely with `--local-store=0:0'.
5022 SPU local store space is limited. Over-allocation of stack space
5023 unnecessarily limits space available for code and data, while
5024 under-allocation results in runtime failures. If given this
5025 option, `ld' will provide an estimate of maximum stack usage.
5026 `ld' does this by examining symbols in code sections to determine
5027 the extents of functions, and looking at function prologues for
5028 stack adjusting instructions. A call-graph is created by looking
5029 for relocations on branch instructions. The graph is then searched
5030 for the maximum stack usage path. Note that this analysis does not
5031 find calls made via function pointers, and does not handle
5032 recursion and other cycles in the call graph. Stack usage may be
5033 under-estimated if your code makes such calls. Also, stack usage
5034 for dynamic allocation, e.g. alloca, will not be detected. If a
5035 link map is requested, detailed information about each function's
5036 stack usage and calls will be given.
5039 This option, if given along with `--stack-analysis' will result in
5040 `ld' emitting stack sizing symbols for each function. These take
5041 the form `__stack_<function_name>' for global functions, and
5042 `__stack_<number>_<function_name>' for static functions.
5043 `<number>' is the section id in hex. The value of such symbols is
5044 the stack requirement for the corresponding function. The symbol
5045 size will be zero, type `STT_NOTYPE', binding `STB_LOCAL', and
5049 File: ld.info, Node: TI COFF, Next: WIN32, Prev: SPU ELF, Up: Machine Dependent
5051 4.12 `ld''s Support for Various TI COFF Versions
5052 ================================================
5054 The `--format' switch allows selection of one of the various TI COFF
5055 versions. The latest of this writing is 2; versions 0 and 1 are also
5056 supported. The TI COFF versions also vary in header byte-order format;
5057 `ld' will read any version or byte order, but the output header format
5058 depends on the default specified by the specific target.
5061 File: ld.info, Node: WIN32, Next: Xtensa, Prev: TI COFF, Up: Machine Dependent
5063 4.13 `ld' and WIN32 (cygwin/mingw)
5064 ==================================
5066 This section describes some of the win32 specific `ld' issues. See
5067 *Note Command Line Options: Options. for detailed description of the
5068 command line options mentioned here.
5071 The standard Windows linker creates and uses so-called import
5072 libraries, which contains information for linking to dll's. They
5073 are regular static archives and are handled as any other static
5074 archive. The cygwin and mingw ports of `ld' have specific support
5075 for creating such libraries provided with the `--out-implib'
5076 command line option.
5078 _exporting DLL symbols_
5079 The cygwin/mingw `ld' has several ways to export symbols for dll's.
5081 _using auto-export functionality_
5082 By default `ld' exports symbols with the auto-export
5083 functionality, which is controlled by the following command
5086 * -export-all-symbols [This is the default]
5092 If, however, `--export-all-symbols' is not given explicitly
5093 on the command line, then the default auto-export behavior
5094 will be _disabled_ if either of the following are true:
5096 * A DEF file is used.
5098 * Any symbol in any object file was marked with the
5099 __declspec(dllexport) attribute.
5102 Another way of exporting symbols is using a DEF file. A DEF
5103 file is an ASCII file containing definitions of symbols which
5104 should be exported when a dll is created. Usually it is
5105 named `<dll name>.def' and is added as any other object file
5106 to the linker's command line. The file's name must end in
5109 gcc -o <output> <objectfiles> <dll name>.def
5111 Using a DEF file turns off the normal auto-export behavior,
5112 unless the `--export-all-symbols' option is also used.
5114 Here is an example of a DEF file for a shared library called
5117 LIBRARY "xyz.dll" BASE=0x20000000
5123 another_foo = abc.dll.afoo
5126 This example defines a DLL with a non-default base address
5127 and five symbols in the export table. The third exported
5128 symbol `_bar' is an alias for the second. The fourth symbol,
5129 `another_foo' is resolved by "forwarding" to another module
5130 and treating it as an alias for `afoo' exported from the DLL
5131 `abc.dll'. The final symbol `var1' is declared to be a data
5134 The optional `LIBRARY <name>' command indicates the _internal_
5135 name of the output DLL. If `<name>' does not include a suffix,
5136 the default library suffix, `.DLL' is appended.
5138 When the .DEF file is used to build an application, rather
5139 than a library, the `NAME <name>' command should be used
5140 instead of `LIBRARY'. If `<name>' does not include a suffix,
5141 the default executable suffix, `.EXE' is appended.
5143 With either `LIBRARY <name>' or `NAME <name>' the optional
5144 specification `BASE = <number>' may be used to specify a
5145 non-default base address for the image.
5147 If neither `LIBRARY <name>' nor `NAME <name>' is specified,
5148 or they specify an empty string, the internal name is the
5149 same as the filename specified on the command line.
5151 The complete specification of an export symbol is:
5154 ( ( ( <name1> [ = <name2> ] )
5155 | ( <name1> = <module-name> . <external-name>))
5156 [ @ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
5158 Declares `<name1>' as an exported symbol from the DLL, or
5159 declares `<name1>' as an exported alias for `<name2>'; or
5160 declares `<name1>' as a "forward" alias for the symbol
5161 `<external-name>' in the DLL `<module-name>'. Optionally,
5162 the symbol may be exported by the specified ordinal
5165 The optional keywords that follow the declaration indicate:
5167 `NONAME': Do not put the symbol name in the DLL's export
5168 table. It will still be exported by its ordinal alias
5169 (either the value specified by the .def specification or,
5170 otherwise, the value assigned by the linker). The symbol
5171 name, however, does remain visible in the import library (if
5172 any), unless `PRIVATE' is also specified.
5174 `DATA': The symbol is a variable or object, rather than a
5175 function. The import lib will export only an indirect
5176 reference to `foo' as the symbol `_imp__foo' (ie, `foo' must
5177 be resolved as `*_imp__foo').
5179 `CONSTANT': Like `DATA', but put the undecorated `foo' as
5180 well as `_imp__foo' into the import library. Both refer to the
5181 read-only import address table's pointer to the variable, not
5182 to the variable itself. This can be dangerous. If the user
5183 code fails to add the `dllimport' attribute and also fails to
5184 explicitly add the extra indirection that the use of the
5185 attribute enforces, the application will behave unexpectedly.
5187 `PRIVATE': Put the symbol in the DLL's export table, but do
5188 not put it into the static import library used to resolve
5189 imports at link time. The symbol can still be imported using
5190 the `LoadLibrary/GetProcAddress' API at runtime or by by
5191 using the GNU ld extension of linking directly to the DLL
5192 without an import library.
5194 See ld/deffilep.y in the binutils sources for the full
5195 specification of other DEF file statements
5197 While linking a shared dll, `ld' is able to create a DEF file
5198 with the `--output-def <file>' command line option.
5201 Another way of marking symbols for export is to modify the
5202 source code itself, so that when building the DLL each symbol
5203 to be exported is declared as:
5205 __declspec(dllexport) int a_variable
5206 __declspec(dllexport) void a_function(int with_args)
5208 All such symbols will be exported from the DLL. If, however,
5209 any of the object files in the DLL contain symbols decorated
5210 in this way, then the normal auto-export behavior is
5211 disabled, unless the `--export-all-symbols' option is also
5214 Note that object files that wish to access these symbols must
5215 _not_ decorate them with dllexport. Instead, they should use
5218 __declspec(dllimport) int a_variable
5219 __declspec(dllimport) void a_function(int with_args)
5221 This complicates the structure of library header files,
5222 because when included by the library itself the header must
5223 declare the variables and functions as dllexport, but when
5224 included by client code the header must declare them as
5225 dllimport. There are a number of idioms that are typically
5226 used to do this; often client code can omit the __declspec()
5227 declaration completely. See `--enable-auto-import' and
5228 `automatic data imports' for more information.
5230 _automatic data imports_
5231 The standard Windows dll format supports data imports from dlls
5232 only by adding special decorations (dllimport/dllexport), which
5233 let the compiler produce specific assembler instructions to deal
5234 with this issue. This increases the effort necessary to port
5235 existing Un*x code to these platforms, especially for large c++
5236 libraries and applications. The auto-import feature, which was
5237 initially provided by Paul Sokolovsky, allows one to omit the
5238 decorations to achieve a behavior that conforms to that on
5239 POSIX/Un*x platforms. This feature is enabled with the
5240 `--enable-auto-import' command-line option, although it is enabled
5241 by default on cygwin/mingw. The `--enable-auto-import' option
5242 itself now serves mainly to suppress any warnings that are
5243 ordinarily emitted when linked objects trigger the feature's use.
5245 auto-import of variables does not always work flawlessly without
5246 additional assistance. Sometimes, you will see this message
5248 "variable '<var>' can't be auto-imported. Please read the
5249 documentation for ld's `--enable-auto-import' for details."
5251 The `--enable-auto-import' documentation explains why this error
5252 occurs, and several methods that can be used to overcome this
5253 difficulty. One of these methods is the _runtime pseudo-relocs_
5254 feature, described below.
5256 For complex variables imported from DLLs (such as structs or
5257 classes), object files typically contain a base address for the
5258 variable and an offset (_addend_) within the variable-to specify a
5259 particular field or public member, for instance. Unfortunately,
5260 the runtime loader used in win32 environments is incapable of
5261 fixing these references at runtime without the additional
5262 information supplied by dllimport/dllexport decorations. The
5263 standard auto-import feature described above is unable to resolve
5266 The `--enable-runtime-pseudo-relocs' switch allows these
5267 references to be resolved without error, while leaving the task of
5268 adjusting the references themselves (with their non-zero addends)
5269 to specialized code provided by the runtime environment. Recent
5270 versions of the cygwin and mingw environments and compilers
5271 provide this runtime support; older versions do not. However, the
5272 support is only necessary on the developer's platform; the
5273 compiled result will run without error on an older system.
5275 `--enable-runtime-pseudo-relocs' is not the default; it must be
5276 explicitly enabled as needed.
5278 _direct linking to a dll_
5279 The cygwin/mingw ports of `ld' support the direct linking,
5280 including data symbols, to a dll without the usage of any import
5281 libraries. This is much faster and uses much less memory than
5282 does the traditional import library method, especially when
5283 linking large libraries or applications. When `ld' creates an
5284 import lib, each function or variable exported from the dll is
5285 stored in its own bfd, even though a single bfd could contain many
5286 exports. The overhead involved in storing, loading, and
5287 processing so many bfd's is quite large, and explains the
5288 tremendous time, memory, and storage needed to link against
5289 particularly large or complex libraries when using import libs.
5291 Linking directly to a dll uses no extra command-line switches
5292 other than `-L' and `-l', because `ld' already searches for a
5293 number of names to match each library. All that is needed from
5294 the developer's perspective is an understanding of this search, in
5295 order to force ld to select the dll instead of an import library.
5297 For instance, when ld is called with the argument `-lxxx' it will
5298 attempt to find, in the first directory of its search path,
5308 before moving on to the next directory in the search path.
5310 (*) Actually, this is not `cygxxx.dll' but in fact is
5311 `<prefix>xxx.dll', where `<prefix>' is set by the `ld' option
5312 `--dll-search-prefix=<prefix>'. In the case of cygwin, the
5313 standard gcc spec file includes `--dll-search-prefix=cyg', so in
5314 effect we actually search for `cygxxx.dll'.
5316 Other win32-based unix environments, such as mingw or pw32, may
5317 use other `<prefix>'es, although at present only cygwin makes use
5318 of this feature. It was originally intended to help avoid name
5319 conflicts among dll's built for the various win32/un*x
5320 environments, so that (for example) two versions of a zlib dll
5321 could coexist on the same machine.
5323 The generic cygwin/mingw path layout uses a `bin' directory for
5324 applications and dll's and a `lib' directory for the import
5325 libraries (using cygwin nomenclature):
5330 libxxx.dll.a (in case of dll's)
5331 libxxx.a (in case of static archive)
5333 Linking directly to a dll without using the import library can be
5336 1. Use the dll directly by adding the `bin' path to the link line
5337 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
5339 However, as the dll's often have version numbers appended to their
5340 names (`cygncurses-5.dll') this will often fail, unless one
5341 specifies `-L../bin -lncurses-5' to include the version. Import
5342 libs are generally not versioned, and do not have this difficulty.
5344 2. Create a symbolic link from the dll to a file in the `lib'
5345 directory according to the above mentioned search pattern. This
5346 should be used to avoid unwanted changes in the tools needed for
5349 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
5351 Then you can link without any make environment changes.
5353 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
5355 This technique also avoids the version number problems, because
5356 the following is perfectly legal
5361 libxxx.dll.a -> ../bin/cygxxx-5.dll
5363 Linking directly to a dll without using an import lib will work
5364 even when auto-import features are exercised, and even when
5365 `--enable-runtime-pseudo-relocs' is used.
5367 Given the improvements in speed and memory usage, one might
5368 justifiably wonder why import libraries are used at all. There
5371 1. Until recently, the link-directly-to-dll functionality did _not_
5372 work with auto-imported data.
5374 2. Sometimes it is necessary to include pure static objects within
5375 the import library (which otherwise contains only bfd's for
5376 indirection symbols that point to the exports of a dll). Again,
5377 the import lib for the cygwin kernel makes use of this ability,
5378 and it is not possible to do this without an import lib.
5380 3. Symbol aliases can only be resolved using an import lib. This
5381 is critical when linking against OS-supplied dll's (eg, the win32
5382 API) in which symbols are usually exported as undecorated aliases
5383 of their stdcall-decorated assembly names.
5385 So, import libs are not going away. But the ability to replace
5386 true import libs with a simple symbolic link to (or a copy of) a
5387 dll, in many cases, is a useful addition to the suite of tools
5388 binutils makes available to the win32 developer. Given the
5389 massive improvements in memory requirements during linking, storage
5390 requirements, and linking speed, we expect that many developers
5391 will soon begin to use this feature whenever possible.
5395 _adding additional names_
5396 Sometimes, it is useful to export symbols with additional
5397 names. A symbol `foo' will be exported as `foo', but it can
5398 also be exported as `_foo' by using special directives in the
5399 DEF file when creating the dll. This will affect also the
5400 optional created import library. Consider the following DEF
5403 LIBRARY "xyz.dll" BASE=0x61000000
5409 The line `_foo = foo' maps the symbol `foo' to `_foo'.
5411 Another method for creating a symbol alias is to create it in
5412 the source code using the "weak" attribute:
5414 void foo () { /* Do something. */; }
5415 void _foo () __attribute__ ((weak, alias ("foo")));
5417 See the gcc manual for more information about attributes and
5421 Sometimes it is useful to rename exports. For instance, the
5422 cygwin kernel does this regularly. A symbol `_foo' can be
5423 exported as `foo' but not as `_foo' by using special
5424 directives in the DEF file. (This will also affect the import
5425 library, if it is created). In the following example:
5427 LIBRARY "xyz.dll" BASE=0x61000000
5432 The line `_foo = foo' maps the exported symbol `foo' to
5435 Note: using a DEF file disables the default auto-export behavior,
5436 unless the `--export-all-symbols' command line option is used.
5437 If, however, you are trying to rename symbols, then you should list
5438 _all_ desired exports in the DEF file, including the symbols that
5439 are not being renamed, and do _not_ use the `--export-all-symbols'
5440 option. If you list only the renamed symbols in the DEF file, and
5441 use `--export-all-symbols' to handle the other symbols, then the
5442 both the new names _and_ the original names for the renamed
5443 symbols will be exported. In effect, you'd be aliasing those
5444 symbols, not renaming them, which is probably not what you wanted.
5447 The Windows object format, PE, specifies a form of weak symbols
5448 called weak externals. When a weak symbol is linked and the
5449 symbol is not defined, the weak symbol becomes an alias for some
5450 other symbol. There are three variants of weak externals:
5451 * Definition is searched for in objects and libraries,
5452 historically called lazy externals.
5454 * Definition is searched for only in other objects, not in
5455 libraries. This form is not presently implemented.
5457 * No search; the symbol is an alias. This form is not presently
5459 As a GNU extension, weak symbols that do not specify an alternate
5460 symbol are supported. If the symbol is undefined when linking,
5461 the symbol uses a default value.
5464 File: ld.info, Node: Xtensa, Prev: WIN32, Up: Machine Dependent
5466 4.14 `ld' and Xtensa Processors
5467 ===============================
5469 The default `ld' behavior for Xtensa processors is to interpret
5470 `SECTIONS' commands so that lists of explicitly named sections in a
5471 specification with a wildcard file will be interleaved when necessary to
5472 keep literal pools within the range of PC-relative load offsets. For
5473 example, with the command:
5482 `ld' may interleave some of the `.literal' and `.text' sections from
5483 different object files to ensure that the literal pools are within the
5484 range of PC-relative load offsets. A valid interleaving might place
5485 the `.literal' sections from an initial group of files followed by the
5486 `.text' sections of that group of files. Then, the `.literal' sections
5487 from the rest of the files and the `.text' sections from the rest of
5488 the files would follow.
5490 Relaxation is enabled by default for the Xtensa version of `ld' and
5491 provides two important link-time optimizations. The first optimization
5492 is to combine identical literal values to reduce code size. A redundant
5493 literal will be removed and all the `L32R' instructions that use it
5494 will be changed to reference an identical literal, as long as the
5495 location of the replacement literal is within the offset range of all
5496 the `L32R' instructions. The second optimization is to remove
5497 unnecessary overhead from assembler-generated "longcall" sequences of
5498 `L32R'/`CALLXN' when the target functions are within range of direct
5499 `CALLN' instructions.
5501 For each of these cases where an indirect call sequence can be
5502 optimized to a direct call, the linker will change the `CALLXN'
5503 instruction to a `CALLN' instruction, remove the `L32R' instruction,
5504 and remove the literal referenced by the `L32R' instruction if it is
5505 not used for anything else. Removing the `L32R' instruction always
5506 reduces code size but can potentially hurt performance by changing the
5507 alignment of subsequent branch targets. By default, the linker will
5508 always preserve alignments, either by switching some instructions
5509 between 24-bit encodings and the equivalent density instructions or by
5510 inserting a no-op in place of the `L32R' instruction that was removed.
5511 If code size is more important than performance, the `--size-opt'
5512 option can be used to prevent the linker from widening density
5513 instructions or inserting no-ops, except in a few cases where no-ops
5514 are required for correctness.
5516 The following Xtensa-specific command-line options can be used to
5520 Since the Xtensa version of `ld' enables the `--relax' option by
5521 default, the `--no-relax' option is provided to disable relaxation.
5524 When optimizing indirect calls to direct calls, optimize for code
5525 size more than performance. With this option, the linker will not
5526 insert no-ops or widen density instructions to preserve branch
5527 target alignment. There may still be some cases where no-ops are
5528 required to preserve the correctness of the code.
5531 File: ld.info, Node: BFD, Next: Reporting Bugs, Prev: Machine Dependent, Up: Top
5536 The linker accesses object and archive files using the BFD libraries.
5537 These libraries allow the linker to use the same routines to operate on
5538 object files whatever the object file format. A different object file
5539 format can be supported simply by creating a new BFD back end and adding
5540 it to the library. To conserve runtime memory, however, the linker and
5541 associated tools are usually configured to support only a subset of the
5542 object file formats available. You can use `objdump -i' (*note
5543 objdump: (binutils.info)objdump.) to list all the formats available for
5546 As with most implementations, BFD is a compromise between several
5547 conflicting requirements. The major factor influencing BFD design was
5548 efficiency: any time used converting between formats is time which
5549 would not have been spent had BFD not been involved. This is partly
5550 offset by abstraction payback; since BFD simplifies applications and
5551 back ends, more time and care may be spent optimizing algorithms for a
5554 One minor artifact of the BFD solution which you should bear in mind
5555 is the potential for information loss. There are two places where
5556 useful information can be lost using the BFD mechanism: during
5557 conversion and during output. *Note BFD information loss::.
5561 * BFD outline:: How it works: an outline of BFD
5564 File: ld.info, Node: BFD outline, Up: BFD
5566 5.1 How It Works: An Outline of BFD
5567 ===================================
5569 When an object file is opened, BFD subroutines automatically determine
5570 the format of the input object file. They then build a descriptor in
5571 memory with pointers to routines that will be used to access elements of
5572 the object file's data structures.
5574 As different information from the object files is required, BFD
5575 reads from different sections of the file and processes them. For
5576 example, a very common operation for the linker is processing symbol
5577 tables. Each BFD back end provides a routine for converting between
5578 the object file's representation of symbols and an internal canonical
5579 format. When the linker asks for the symbol table of an object file, it
5580 calls through a memory pointer to the routine from the relevant BFD
5581 back end which reads and converts the table into a canonical form. The
5582 linker then operates upon the canonical form. When the link is finished
5583 and the linker writes the output file's symbol table, another BFD back
5584 end routine is called to take the newly created symbol table and
5585 convert it into the chosen output format.
5589 * BFD information loss:: Information Loss
5590 * Canonical format:: The BFD canonical object-file format
5593 File: ld.info, Node: BFD information loss, Next: Canonical format, Up: BFD outline
5595 5.1.1 Information Loss
5596 ----------------------
5598 _Information can be lost during output._ The output formats supported
5599 by BFD do not provide identical facilities, and information which can
5600 be described in one form has nowhere to go in another format. One
5601 example of this is alignment information in `b.out'. There is nowhere
5602 in an `a.out' format file to store alignment information on the
5603 contained data, so when a file is linked from `b.out' and an `a.out'
5604 image is produced, alignment information will not propagate to the
5605 output file. (The linker will still use the alignment information
5606 internally, so the link is performed correctly).
5608 Another example is COFF section names. COFF files may contain an
5609 unlimited number of sections, each one with a textual section name. If
5610 the target of the link is a format which does not have many sections
5611 (e.g., `a.out') or has sections without names (e.g., the Oasys format),
5612 the link cannot be done simply. You can circumvent this problem by
5613 describing the desired input-to-output section mapping with the linker
5616 _Information can be lost during canonicalization._ The BFD internal
5617 canonical form of the external formats is not exhaustive; there are
5618 structures in input formats for which there is no direct representation
5619 internally. This means that the BFD back ends cannot maintain all
5620 possible data richness through the transformation between external to
5621 internal and back to external formats.
5623 This limitation is only a problem when an application reads one
5624 format and writes another. Each BFD back end is responsible for
5625 maintaining as much data as possible, and the internal BFD canonical
5626 form has structures which are opaque to the BFD core, and exported only
5627 to the back ends. When a file is read in one format, the canonical form
5628 is generated for BFD and the application. At the same time, the back
5629 end saves away any information which may otherwise be lost. If the data
5630 is then written back in the same format, the back end routine will be
5631 able to use the canonical form provided by the BFD core as well as the
5632 information it prepared earlier. Since there is a great deal of
5633 commonality between back ends, there is no information lost when
5634 linking or copying big endian COFF to little endian COFF, or `a.out' to
5635 `b.out'. When a mixture of formats is linked, the information is only
5636 lost from the files whose format differs from the destination.
5639 File: ld.info, Node: Canonical format, Prev: BFD information loss, Up: BFD outline
5641 5.1.2 The BFD canonical object-file format
5642 ------------------------------------------
5644 The greatest potential for loss of information occurs when there is the
5645 least overlap between the information provided by the source format,
5646 that stored by the canonical format, and that needed by the destination
5647 format. A brief description of the canonical form may help you
5648 understand which kinds of data you can count on preserving across
5652 Information stored on a per-file basis includes target machine
5653 architecture, particular implementation format type, a demand
5654 pageable bit, and a write protected bit. Information like Unix
5655 magic numbers is not stored here--only the magic numbers' meaning,
5656 so a `ZMAGIC' file would have both the demand pageable bit and the
5657 write protected text bit set. The byte order of the target is
5658 stored on a per-file basis, so that big- and little-endian object
5659 files may be used with one another.
5662 Each section in the input file contains the name of the section,
5663 the section's original address in the object file, size and
5664 alignment information, various flags, and pointers into other BFD
5668 Each symbol contains a pointer to the information for the object
5669 file which originally defined it, its name, its value, and various
5670 flag bits. When a BFD back end reads in a symbol table, it
5671 relocates all symbols to make them relative to the base of the
5672 section where they were defined. Doing this ensures that each
5673 symbol points to its containing section. Each symbol also has a
5674 varying amount of hidden private data for the BFD back end. Since
5675 the symbol points to the original file, the private data format
5676 for that symbol is accessible. `ld' can operate on a collection
5677 of symbols of wildly different formats without problems.
5679 Normal global and simple local symbols are maintained on output,
5680 so an output file (no matter its format) will retain symbols
5681 pointing to functions and to global, static, and common variables.
5682 Some symbol information is not worth retaining; in `a.out', type
5683 information is stored in the symbol table as long symbol names.
5684 This information would be useless to most COFF debuggers; the
5685 linker has command line switches to allow users to throw it away.
5687 There is one word of type information within the symbol, so if the
5688 format supports symbol type information within symbols (for
5689 example, COFF, IEEE, Oasys) and the type is simple enough to fit
5690 within one word (nearly everything but aggregates), the
5691 information will be preserved.
5694 Each canonical BFD relocation record contains a pointer to the
5695 symbol to relocate to, the offset of the data to relocate, the
5696 section the data is in, and a pointer to a relocation type
5697 descriptor. Relocation is performed by passing messages through
5698 the relocation type descriptor and the symbol pointer. Therefore,
5699 relocations can be performed on output data using a relocation
5700 method that is only available in one of the input formats. For
5701 instance, Oasys provides a byte relocation format. A relocation
5702 record requesting this relocation type would point indirectly to a
5703 routine to perform this, so the relocation may be performed on a
5704 byte being written to a 68k COFF file, even though 68k COFF has no
5705 such relocation type.
5708 Object formats can contain, for debugging purposes, some form of
5709 mapping between symbols, source line numbers, and addresses in the
5710 output file. These addresses have to be relocated along with the
5711 symbol information. Each symbol with an associated list of line
5712 number records points to the first record of the list. The head
5713 of a line number list consists of a pointer to the symbol, which
5714 allows finding out the address of the function whose line number
5715 is being described. The rest of the list is made up of pairs:
5716 offsets into the section and line numbers. Any format which can
5717 simply derive this information can pass it successfully between
5718 formats (COFF, IEEE and Oasys).
5721 File: ld.info, Node: Reporting Bugs, Next: MRI, Prev: BFD, Up: Top
5726 Your bug reports play an essential role in making `ld' reliable.
5728 Reporting a bug may help you by bringing a solution to your problem,
5729 or it may not. But in any case the principal function of a bug report
5730 is to help the entire community by making the next version of `ld' work
5731 better. Bug reports are your contribution to the maintenance of `ld'.
5733 In order for a bug report to serve its purpose, you must include the
5734 information that enables us to fix the bug.
5738 * Bug Criteria:: Have you found a bug?
5739 * Bug Reporting:: How to report bugs
5742 File: ld.info, Node: Bug Criteria, Next: Bug Reporting, Up: Reporting Bugs
5744 6.1 Have You Found a Bug?
5745 =========================
5747 If you are not sure whether you have found a bug, here are some
5750 * If the linker gets a fatal signal, for any input whatever, that is
5751 a `ld' bug. Reliable linkers never crash.
5753 * If `ld' produces an error message for valid input, that is a bug.
5755 * If `ld' does not produce an error message for invalid input, that
5756 may be a bug. In the general case, the linker can not verify that
5757 object files are correct.
5759 * If you are an experienced user of linkers, your suggestions for
5760 improvement of `ld' are welcome in any case.
5763 File: ld.info, Node: Bug Reporting, Prev: Bug Criteria, Up: Reporting Bugs
5765 6.2 How to Report Bugs
5766 ======================
5768 A number of companies and individuals offer support for GNU products.
5769 If you obtained `ld' from a support organization, we recommend you
5770 contact that organization first.
5772 You can find contact information for many support companies and
5773 individuals in the file `etc/SERVICE' in the GNU Emacs distribution.
5775 Otherwise, send bug reports for `ld' to
5776 `http://www.sourceware.org/bugzilla/'.
5778 The fundamental principle of reporting bugs usefully is this:
5779 *report all the facts*. If you are not sure whether to state a fact or
5780 leave it out, state it!
5782 Often people omit facts because they think they know what causes the
5783 problem and assume that some details do not matter. Thus, you might
5784 assume that the name of a symbol you use in an example does not matter.
5785 Well, probably it does not, but one cannot be sure. Perhaps the bug
5786 is a stray memory reference which happens to fetch from the location
5787 where that name is stored in memory; perhaps, if the name were
5788 different, the contents of that location would fool the linker into
5789 doing the right thing despite the bug. Play it safe and give a
5790 specific, complete example. That is the easiest thing for you to do,
5791 and the most helpful.
5793 Keep in mind that the purpose of a bug report is to enable us to fix
5794 the bug if it is new to us. Therefore, always write your bug reports
5795 on the assumption that the bug has not been reported previously.
5797 Sometimes people give a few sketchy facts and ask, "Does this ring a
5798 bell?" This cannot help us fix a bug, so it is basically useless. We
5799 respond by asking for enough details to enable us to investigate. You
5800 might as well expedite matters by sending them to begin with.
5802 To enable us to fix the bug, you should include all these things:
5804 * The version of `ld'. `ld' announces it if you start it with the
5805 `--version' argument.
5807 Without this, we will not know whether there is any point in
5808 looking for the bug in the current version of `ld'.
5810 * Any patches you may have applied to the `ld' source, including any
5811 patches made to the `BFD' library.
5813 * The type of machine you are using, and the operating system name
5816 * What compiler (and its version) was used to compile `ld'--e.g.
5819 * The command arguments you gave the linker to link your example and
5820 observe the bug. To guarantee you will not omit something
5821 important, list them all. A copy of the Makefile (or the output
5822 from make) is sufficient.
5824 If we were to try to guess the arguments, we would probably guess
5825 wrong and then we might not encounter the bug.
5827 * A complete input file, or set of input files, that will reproduce
5828 the bug. It is generally most helpful to send the actual object
5829 files provided that they are reasonably small. Say no more than
5830 10K. For bigger files you can either make them available by FTP
5831 or HTTP or else state that you are willing to send the object
5832 file(s) to whomever requests them. (Note - your email will be
5833 going to a mailing list, so we do not want to clog it up with
5834 large attachments). But small attachments are best.
5836 If the source files were assembled using `gas' or compiled using
5837 `gcc', then it may be OK to send the source files rather than the
5838 object files. In this case, be sure to say exactly what version of
5839 `gas' or `gcc' was used to produce the object files. Also say how
5840 `gas' or `gcc' were configured.
5842 * A description of what behavior you observe that you believe is
5843 incorrect. For example, "It gets a fatal signal."
5845 Of course, if the bug is that `ld' gets a fatal signal, then we
5846 will certainly notice it. But if the bug is incorrect output, we
5847 might not notice unless it is glaringly wrong. You might as well
5848 not give us a chance to make a mistake.
5850 Even if the problem you experience is a fatal signal, you should
5851 still say so explicitly. Suppose something strange is going on,
5852 such as, your copy of `ld' is out of sync, or you have encountered
5853 a bug in the C library on your system. (This has happened!) Your
5854 copy might crash and ours would not. If you told us to expect a
5855 crash, then when ours fails to crash, we would know that the bug
5856 was not happening for us. If you had not told us to expect a
5857 crash, then we would not be able to draw any conclusion from our
5860 * If you wish to suggest changes to the `ld' source, send us context
5861 diffs, as generated by `diff' with the `-u', `-c', or `-p' option.
5862 Always send diffs from the old file to the new file. If you even
5863 discuss something in the `ld' source, refer to it by context, not
5866 The line numbers in our development sources will not match those
5867 in your sources. Your line numbers would convey no useful
5870 Here are some things that are not necessary:
5872 * A description of the envelope of the bug.
5874 Often people who encounter a bug spend a lot of time investigating
5875 which changes to the input file will make the bug go away and which
5876 changes will not affect it.
5878 This is often time consuming and not very useful, because the way
5879 we will find the bug is by running a single example under the
5880 debugger with breakpoints, not by pure deduction from a series of
5881 examples. We recommend that you save your time for something else.
5883 Of course, if you can find a simpler example to report _instead_
5884 of the original one, that is a convenience for us. Errors in the
5885 output will be easier to spot, running under the debugger will take
5886 less time, and so on.
5888 However, simplification is not vital; if you do not want to do
5889 this, report the bug anyway and send us the entire test case you
5892 * A patch for the bug.
5894 A patch for the bug does help us if it is a good one. But do not
5895 omit the necessary information, such as the test case, on the
5896 assumption that a patch is all we need. We might see problems
5897 with your patch and decide to fix the problem another way, or we
5898 might not understand it at all.
5900 Sometimes with a program as complicated as `ld' it is very hard to
5901 construct an example that will make the program follow a certain
5902 path through the code. If you do not send us the example, we will
5903 not be able to construct one, so we will not be able to verify
5904 that the bug is fixed.
5906 And if we cannot understand what bug you are trying to fix, or why
5907 your patch should be an improvement, we will not install it. A
5908 test case will help us to understand.
5910 * A guess about what the bug is or what it depends on.
5912 Such guesses are usually wrong. Even we cannot guess right about
5913 such things without first using the debugger to find the facts.
5916 File: ld.info, Node: MRI, Next: GNU Free Documentation License, Prev: Reporting Bugs, Up: Top
5918 Appendix A MRI Compatible Script Files
5919 **************************************
5921 To aid users making the transition to GNU `ld' from the MRI linker,
5922 `ld' can use MRI compatible linker scripts as an alternative to the
5923 more general-purpose linker scripting language described in *Note
5924 Scripts::. MRI compatible linker scripts have a much simpler command
5925 set than the scripting language otherwise used with `ld'. GNU `ld'
5926 supports the most commonly used MRI linker commands; these commands are
5929 In general, MRI scripts aren't of much use with the `a.out' object
5930 file format, since it only has three sections and MRI scripts lack some
5931 features to make use of them.
5933 You can specify a file containing an MRI-compatible script using the
5934 `-c' command-line option.
5936 Each command in an MRI-compatible script occupies its own line; each
5937 command line starts with the keyword that identifies the command (though
5938 blank lines are also allowed for punctuation). If a line of an
5939 MRI-compatible script begins with an unrecognized keyword, `ld' issues
5940 a warning message, but continues processing the script.
5942 Lines beginning with `*' are comments.
5944 You can write these commands using all upper-case letters, or all
5945 lower case; for example, `chip' is the same as `CHIP'. The following
5946 list shows only the upper-case form of each command.
5949 `ABSOLUTE SECNAME, SECNAME, ... SECNAME'
5950 Normally, `ld' includes in the output file all sections from all
5951 the input files. However, in an MRI-compatible script, you can
5952 use the `ABSOLUTE' command to restrict the sections that will be
5953 present in your output program. If the `ABSOLUTE' command is used
5954 at all in a script, then only the sections named explicitly in
5955 `ABSOLUTE' commands will appear in the linker output. You can
5956 still use other input sections (whatever you select on the command
5957 line, or using `LOAD') to resolve addresses in the output file.
5959 `ALIAS OUT-SECNAME, IN-SECNAME'
5960 Use this command to place the data from input section IN-SECNAME
5961 in a section called OUT-SECNAME in the linker output file.
5963 IN-SECNAME may be an integer.
5965 `ALIGN SECNAME = EXPRESSION'
5966 Align the section called SECNAME to EXPRESSION. The EXPRESSION
5967 should be a power of two.
5970 Use the value of EXPRESSION as the lowest address (other than
5971 absolute addresses) in the output file.
5974 `CHIP EXPRESSION, EXPRESSION'
5975 This command does nothing; it is accepted only for compatibility.
5978 This command does nothing whatever; it's only accepted for
5981 `FORMAT OUTPUT-FORMAT'
5982 Similar to the `OUTPUT_FORMAT' command in the more general linker
5983 language, but restricted to one of these output formats:
5985 1. S-records, if OUTPUT-FORMAT is `S'
5987 2. IEEE, if OUTPUT-FORMAT is `IEEE'
5989 3. COFF (the `coff-m68k' variant in BFD), if OUTPUT-FORMAT is
5993 Print (to the standard output file) a link map, as produced by the
5994 `ld' command-line option `-M'.
5996 The keyword `LIST' may be followed by anything on the same line,
5997 with no change in its effect.
6000 `LOAD FILENAME, FILENAME, ... FILENAME'
6001 Include one or more object file FILENAME in the link; this has the
6002 same effect as specifying FILENAME directly on the `ld' command
6006 OUTPUT-NAME is the name for the program produced by `ld'; the
6007 MRI-compatible command `NAME' is equivalent to the command-line
6008 option `-o' or the general script language command `OUTPUT'.
6010 `ORDER SECNAME, SECNAME, ... SECNAME'
6011 `ORDER SECNAME SECNAME SECNAME'
6012 Normally, `ld' orders the sections in its output file in the order
6013 in which they first appear in the input files. In an
6014 MRI-compatible script, you can override this ordering with the
6015 `ORDER' command. The sections you list with `ORDER' will appear
6016 first in your output file, in the order specified.
6018 `PUBLIC NAME=EXPRESSION'
6019 `PUBLIC NAME,EXPRESSION'
6020 `PUBLIC NAME EXPRESSION'
6021 Supply a value (EXPRESSION) for external symbol NAME used in the
6024 `SECT SECNAME, EXPRESSION'
6025 `SECT SECNAME=EXPRESSION'
6026 `SECT SECNAME EXPRESSION'
6027 You can use any of these three forms of the `SECT' command to
6028 specify the start address (EXPRESSION) for section SECNAME. If
6029 you have more than one `SECT' statement for the same SECNAME, only
6030 the _first_ sets the start address.
6033 File: ld.info, Node: GNU Free Documentation License, Next: LD Index, Prev: MRI, Up: Top
6035 Appendix B GNU Free Documentation License
6036 *****************************************
6038 Version 1.1, March 2000
6040 Copyright (C) 2000, 2003 Free Software Foundation, Inc.
6041 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
6043 Everyone is permitted to copy and distribute verbatim copies
6044 of this license document, but changing it is not allowed.
6049 The purpose of this License is to make a manual, textbook, or other
6050 written document "free" in the sense of freedom: to assure everyone
6051 the effective freedom to copy and redistribute it, with or without
6052 modifying it, either commercially or noncommercially. Secondarily,
6053 this License preserves for the author and publisher a way to get
6054 credit for their work, while not being considered responsible for
6055 modifications made by others.
6057 This License is a kind of "copyleft", which means that derivative
6058 works of the document must themselves be free in the same sense.
6059 It complements the GNU General Public License, which is a copyleft
6060 license designed for free software.
6062 We have designed this License in order to use it for manuals for
6063 free software, because free software needs free documentation: a
6064 free program should come with manuals providing the same freedoms
6065 that the software does. But this License is not limited to
6066 software manuals; it can be used for any textual work, regardless
6067 of subject matter or whether it is published as a printed book.
6068 We recommend this License principally for works whose purpose is
6069 instruction or reference.
6072 1. APPLICABILITY AND DEFINITIONS
6074 This License applies to any manual or other work that contains a
6075 notice placed by the copyright holder saying it can be distributed
6076 under the terms of this License. The "Document", below, refers to
6077 any such manual or work. Any member of the public is a licensee,
6078 and is addressed as "you."
6080 A "Modified Version" of the Document means any work containing the
6081 Document or a portion of it, either copied verbatim, or with
6082 modifications and/or translated into another language.
6084 A "Secondary Section" is a named appendix or a front-matter
6085 section of the Document that deals exclusively with the
6086 relationship of the publishers or authors of the Document to the
6087 Document's overall subject (or to related matters) and contains
6088 nothing that could fall directly within that overall subject.
6089 (For example, if the Document is in part a textbook of
6090 mathematics, a Secondary Section may not explain any mathematics.)
6091 The relationship could be a matter of historical connection with
6092 the subject or with related matters, or of legal, commercial,
6093 philosophical, ethical or political position regarding them.
6095 The "Invariant Sections" are certain Secondary Sections whose
6096 titles are designated, as being those of Invariant Sections, in
6097 the notice that says that the Document is released under this
6100 The "Cover Texts" are certain short passages of text that are
6101 listed, as Front-Cover Texts or Back-Cover Texts, in the notice
6102 that says that the Document is released under this License.
6104 A "Transparent" copy of the Document means a machine-readable copy,
6105 represented in a format whose specification is available to the
6106 general public, whose contents can be viewed and edited directly
6107 and straightforwardly with generic text editors or (for images
6108 composed of pixels) generic paint programs or (for drawings) some
6109 widely available drawing editor, and that is suitable for input to
6110 text formatters or for automatic translation to a variety of
6111 formats suitable for input to text formatters. A copy made in an
6112 otherwise Transparent file format whose markup has been designed
6113 to thwart or discourage subsequent modification by readers is not
6114 Transparent. A copy that is not "Transparent" is called "Opaque."
6116 Examples of suitable formats for Transparent copies include plain
6117 ASCII without markup, Texinfo input format, LaTeX input format,
6118 SGML or XML using a publicly available DTD, and
6119 standard-conforming simple HTML designed for human modification.
6120 Opaque formats include PostScript, PDF, proprietary formats that
6121 can be read and edited only by proprietary word processors, SGML
6122 or XML for which the DTD and/or processing tools are not generally
6123 available, and the machine-generated HTML produced by some word
6124 processors for output purposes only.
6126 The "Title Page" means, for a printed book, the title page itself,
6127 plus such following pages as are needed to hold, legibly, the
6128 material this License requires to appear in the title page. For
6129 works in formats which do not have any title page as such, "Title
6130 Page" means the text near the most prominent appearance of the
6131 work's title, preceding the beginning of the body of the text.
6135 You may copy and distribute the Document in any medium, either
6136 commercially or noncommercially, provided that this License, the
6137 copyright notices, and the license notice saying this License
6138 applies to the Document are reproduced in all copies, and that you
6139 add no other conditions whatsoever to those of this License. You
6140 may not use technical measures to obstruct or control the reading
6141 or further copying of the copies you make or distribute. However,
6142 you may accept compensation in exchange for copies. If you
6143 distribute a large enough number of copies you must also follow
6144 the conditions in section 3.
6146 You may also lend copies, under the same conditions stated above,
6147 and you may publicly display copies.
6149 3. COPYING IN QUANTITY
6151 If you publish printed copies of the Document numbering more than
6152 100, and the Document's license notice requires Cover Texts, you
6153 must enclose the copies in covers that carry, clearly and legibly,
6154 all these Cover Texts: Front-Cover Texts on the front cover, and
6155 Back-Cover Texts on the back cover. Both covers must also clearly
6156 and legibly identify you as the publisher of these copies. The
6157 front cover must present the full title with all words of the
6158 title equally prominent and visible. You may add other material
6159 on the covers in addition. Copying with changes limited to the
6160 covers, as long as they preserve the title of the Document and
6161 satisfy these conditions, can be treated as verbatim copying in
6164 If the required texts for either cover are too voluminous to fit
6165 legibly, you should put the first ones listed (as many as fit
6166 reasonably) on the actual cover, and continue the rest onto
6169 If you publish or distribute Opaque copies of the Document
6170 numbering more than 100, you must either include a
6171 machine-readable Transparent copy along with each Opaque copy, or
6172 state in or with each Opaque copy a publicly-accessible
6173 computer-network location containing a complete Transparent copy
6174 of the Document, free of added material, which the general
6175 network-using public has access to download anonymously at no
6176 charge using public-standard network protocols. If you use the
6177 latter option, you must take reasonably prudent steps, when you
6178 begin distribution of Opaque copies in quantity, to ensure that
6179 this Transparent copy will remain thus accessible at the stated
6180 location until at least one year after the last time you
6181 distribute an Opaque copy (directly or through your agents or
6182 retailers) of that edition to the public.
6184 It is requested, but not required, that you contact the authors of
6185 the Document well before redistributing any large number of
6186 copies, to give them a chance to provide you with an updated
6187 version of the Document.
6191 You may copy and distribute a Modified Version of the Document
6192 under the conditions of sections 2 and 3 above, provided that you
6193 release the Modified Version under precisely this License, with
6194 the Modified Version filling the role of the Document, thus
6195 licensing distribution and modification of the Modified Version to
6196 whoever possesses a copy of it. In addition, you must do these
6197 things in the Modified Version:
6199 A. Use in the Title Page (and on the covers, if any) a title
6200 distinct from that of the Document, and from those of previous
6201 versions (which should, if there were any, be listed in the
6202 History section of the Document). You may use the same title
6203 as a previous version if the original publisher of that version
6205 B. List on the Title Page, as authors, one or more persons or
6206 entities responsible for authorship of the modifications in the
6207 Modified Version, together with at least five of the principal
6208 authors of the Document (all of its principal authors, if it
6209 has less than five).
6210 C. State on the Title page the name of the publisher of the
6211 Modified Version, as the publisher.
6212 D. Preserve all the copyright notices of the Document.
6213 E. Add an appropriate copyright notice for your modifications
6214 adjacent to the other copyright notices.
6215 F. Include, immediately after the copyright notices, a license
6216 notice giving the public permission to use the Modified Version
6217 under the terms of this License, in the form shown in the
6219 G. Preserve in that license notice the full lists of Invariant
6220 Sections and required Cover Texts given in the Document's
6222 H. Include an unaltered copy of this License.
6223 I. Preserve the section entitled "History", and its title, and add
6224 to it an item stating at least the title, year, new authors, and
6225 publisher of the Modified Version as given on the Title Page.
6226 If there is no section entitled "History" in the Document,
6227 create one stating the title, year, authors, and publisher of
6228 the Document as given on its Title Page, then add an item
6229 describing the Modified Version as stated in the previous
6231 J. Preserve the network location, if any, given in the Document for
6232 public access to a Transparent copy of the Document, and
6233 likewise the network locations given in the Document for
6234 previous versions it was based on. These may be placed in the
6235 "History" section. You may omit a network location for a work
6236 that was published at least four years before the Document
6237 itself, or if the original publisher of the version it refers
6238 to gives permission.
6239 K. In any section entitled "Acknowledgements" or "Dedications",
6240 preserve the section's title, and preserve in the section all the
6241 substance and tone of each of the contributor acknowledgements
6242 and/or dedications given therein.
6243 L. Preserve all the Invariant Sections of the Document,
6244 unaltered in their text and in their titles. Section numbers
6245 or the equivalent are not considered part of the section titles.
6246 M. Delete any section entitled "Endorsements." Such a section
6247 may not be included in the Modified Version.
6248 N. Do not retitle any existing section as "Endorsements" or to
6249 conflict in title with any Invariant Section.
6251 If the Modified Version includes new front-matter sections or
6252 appendices that qualify as Secondary Sections and contain no
6253 material copied from the Document, you may at your option
6254 designate some or all of these sections as invariant. To do this,
6255 add their titles to the list of Invariant Sections in the Modified
6256 Version's license notice. These titles must be distinct from any
6257 other section titles.
6259 You may add a section entitled "Endorsements", provided it contains
6260 nothing but endorsements of your Modified Version by various
6261 parties-for example, statements of peer review or that the text has
6262 been approved by an organization as the authoritative definition
6265 You may add a passage of up to five words as a Front-Cover Text,
6266 and a passage of up to 25 words as a Back-Cover Text, to the end
6267 of the list of Cover Texts in the Modified Version. Only one
6268 passage of Front-Cover Text and one of Back-Cover Text may be
6269 added by (or through arrangements made by) any one entity. If the
6270 Document already includes a cover text for the same cover,
6271 previously added by you or by arrangement made by the same entity
6272 you are acting on behalf of, you may not add another; but you may
6273 replace the old one, on explicit permission from the previous
6274 publisher that added the old one.
6276 The author(s) and publisher(s) of the Document do not by this
6277 License give permission to use their names for publicity for or to
6278 assert or imply endorsement of any Modified Version.
6280 5. COMBINING DOCUMENTS
6282 You may combine the Document with other documents released under
6283 this License, under the terms defined in section 4 above for
6284 modified versions, provided that you include in the combination
6285 all of the Invariant Sections of all of the original documents,
6286 unmodified, and list them all as Invariant Sections of your
6287 combined work in its license notice.
6289 The combined work need only contain one copy of this License, and
6290 multiple identical Invariant Sections may be replaced with a single
6291 copy. If there are multiple Invariant Sections with the same name
6292 but different contents, make the title of each such section unique
6293 by adding at the end of it, in parentheses, the name of the
6294 original author or publisher of that section if known, or else a
6295 unique number. Make the same adjustment to the section titles in
6296 the list of Invariant Sections in the license notice of the
6299 In the combination, you must combine any sections entitled
6300 "History" in the various original documents, forming one section
6301 entitled "History"; likewise combine any sections entitled
6302 "Acknowledgements", and any sections entitled "Dedications." You
6303 must delete all sections entitled "Endorsements."
6305 6. COLLECTIONS OF DOCUMENTS
6307 You may make a collection consisting of the Document and other
6308 documents released under this License, and replace the individual
6309 copies of this License in the various documents with a single copy
6310 that is included in the collection, provided that you follow the
6311 rules of this License for verbatim copying of each of the
6312 documents in all other respects.
6314 You may extract a single document from such a collection, and
6315 distribute it individually under this License, provided you insert
6316 a copy of this License into the extracted document, and follow
6317 this License in all other respects regarding verbatim copying of
6320 7. AGGREGATION WITH INDEPENDENT WORKS
6322 A compilation of the Document or its derivatives with other
6323 separate and independent documents or works, in or on a volume of
6324 a storage or distribution medium, does not as a whole count as a
6325 Modified Version of the Document, provided no compilation
6326 copyright is claimed for the compilation. Such a compilation is
6327 called an "aggregate", and this License does not apply to the
6328 other self-contained works thus compiled with the Document, on
6329 account of their being thus compiled, if they are not themselves
6330 derivative works of the Document.
6332 If the Cover Text requirement of section 3 is applicable to these
6333 copies of the Document, then if the Document is less than one
6334 quarter of the entire aggregate, the Document's Cover Texts may be
6335 placed on covers that surround only the Document within the
6336 aggregate. Otherwise they must appear on covers around the whole
6341 Translation is considered a kind of modification, so you may
6342 distribute translations of the Document under the terms of section
6343 4. Replacing Invariant Sections with translations requires special
6344 permission from their copyright holders, but you may include
6345 translations of some or all Invariant Sections in addition to the
6346 original versions of these Invariant Sections. You may include a
6347 translation of this License provided that you also include the
6348 original English version of this License. In case of a
6349 disagreement between the translation and the original English
6350 version of this License, the original English version will prevail.
6354 You may not copy, modify, sublicense, or distribute the Document
6355 except as expressly provided for under this License. Any other
6356 attempt to copy, modify, sublicense or distribute the Document is
6357 void, and will automatically terminate your rights under this
6358 License. However, parties who have received copies, or rights,
6359 from you under this License will not have their licenses
6360 terminated so long as such parties remain in full compliance.
6362 10. FUTURE REVISIONS OF THIS LICENSE
6364 The Free Software Foundation may publish new, revised versions of
6365 the GNU Free Documentation License from time to time. Such new
6366 versions will be similar in spirit to the present version, but may
6367 differ in detail to address new problems or concerns. See
6368 http://www.gnu.org/copyleft/.
6370 Each version of the License is given a distinguishing version
6371 number. If the Document specifies that a particular numbered
6372 version of this License "or any later version" applies to it, you
6373 have the option of following the terms and conditions either of
6374 that specified version or of any later version that has been
6375 published (not as a draft) by the Free Software Foundation. If
6376 the Document does not specify a version number of this License,
6377 you may choose any version ever published (not as a draft) by the
6378 Free Software Foundation.
6381 ADDENDUM: How to use this License for your documents
6382 ====================================================
6384 To use this License in a document you have written, include a copy of
6385 the License in the document and put the following copyright and license
6386 notices just after the title page:
6388 Copyright (C) YEAR YOUR NAME.
6389 Permission is granted to copy, distribute and/or modify this document
6390 under the terms of the GNU Free Documentation License, Version 1.1
6391 or any later version published by the Free Software Foundation;
6392 with the Invariant Sections being LIST THEIR TITLES, with the
6393 Front-Cover Texts being LIST, and with the Back-Cover Texts being LIST.
6394 A copy of the license is included in the section entitled "GNU
6395 Free Documentation License."
6397 If you have no Invariant Sections, write "with no Invariant Sections"
6398 instead of saying which ones are invariant. If you have no Front-Cover
6399 Texts, write "no Front-Cover Texts" instead of "Front-Cover Texts being
6400 LIST"; likewise for Back-Cover Texts.
6402 If your document contains nontrivial examples of program code, we
6403 recommend releasing these examples in parallel under your choice of
6404 free software license, such as the GNU General Public License, to
6405 permit their use in free software.
6408 File: ld.info, Node: LD Index, Prev: GNU Free Documentation License, Up: Top
6416 * ": Symbols. (line 6)
6417 * -(: Options. (line 645)
6418 * --accept-unknown-input-arch: Options. (line 663)
6419 * --add-needed: Options. (line 685)
6420 * --add-stdcall-alias: Options. (line 1470)
6421 * --allow-multiple-definition: Options. (line 910)
6422 * --allow-shlib-undefined: Options. (line 916)
6423 * --architecture=ARCH: Options. (line 106)
6424 * --as-needed: Options. (line 673)
6425 * --auxiliary: Options. (line 207)
6426 * --bank-window: Options. (line 1815)
6427 * --base-file: Options. (line 1475)
6428 * --be8: ARM. (line 23)
6429 * --bss-plt: PowerPC ELF32. (line 13)
6430 * --build-id: Options. (line 1432)
6431 * --build-id=STYLE: Options. (line 1432)
6432 * --check-sections: Options. (line 767)
6433 * --cref: Options. (line 777)
6434 * --default-imported-symver: Options. (line 944)
6435 * --default-script=SCRIPT: Options. (line 490)
6436 * --default-symver: Options. (line 940)
6437 * --defsym SYMBOL=EXP: Options. (line 805)
6438 * --demangle[=STYLE]: Options. (line 818)
6439 * --disable-auto-image-base: Options. (line 1622)
6440 * --disable-auto-import: Options. (line 1757)
6441 * --disable-new-dtags: Options. (line 1395)
6442 * --disable-runtime-pseudo-reloc: Options. (line 1770)
6443 * --disable-stdcall-fixup: Options. (line 1485)
6444 * --discard-all: Options. (line 536)
6445 * --discard-locals: Options. (line 540)
6446 * --dll: Options. (line 1480)
6447 * --dll-search-prefix: Options. (line 1628)
6448 * --dotsyms: PowerPC64 ELF64. (line 33)
6449 * --dynamic-linker FILE: Options. (line 831)
6450 * --dynamic-list-cpp-new: Options. (line 759)
6451 * --dynamic-list-cpp-typeinfo: Options. (line 763)
6452 * --dynamic-list-data: Options. (line 756)
6453 * --dynamic-list=DYNAMIC-LIST-FILE: Options. (line 743)
6454 * --eh-frame-hdr: Options. (line 1391)
6455 * --emit-relocs: Options. (line 425)
6456 * --emit-stack-syms: SPU ELF. (line 46)
6457 * --emit-stub-syms <1>: SPU ELF. (line 15)
6458 * --emit-stub-syms <2>: PowerPC64 ELF64. (line 29)
6459 * --emit-stub-syms: PowerPC ELF32. (line 44)
6460 * --enable-auto-image-base: Options. (line 1614)
6461 * --enable-auto-import: Options. (line 1637)
6462 * --enable-extra-pe-debug: Options. (line 1775)
6463 * --enable-new-dtags: Options. (line 1395)
6464 * --enable-runtime-pseudo-reloc: Options. (line 1762)
6465 * --enable-stdcall-fixup: Options. (line 1485)
6466 * --entry=ENTRY: Options. (line 160)
6467 * --error-unresolved-symbols: Options. (line 1344)
6468 * --exclude-libs: Options. (line 170)
6469 * --exclude-symbols: Options. (line 1527)
6470 * --export-all-symbols: Options. (line 1503)
6471 * --export-dynamic: Options. (line 181)
6472 * --extra-overlay-stubs: SPU ELF. (line 19)
6473 * --fatal-warnings: Options. (line 837)
6474 * --file-alignment: Options. (line 1533)
6475 * --filter: Options. (line 228)
6476 * --fix-cortex-a8: i960. (line 39)
6477 * --fix-v4bx: ARM. (line 44)
6478 * --fix-v4bx-interworking: ARM. (line 57)
6479 * --force-dynamic: Options. (line 434)
6480 * --force-exe-suffix: Options. (line 842)
6481 * --format=FORMAT: Options. (line 117)
6482 * --format=VERSION: TI COFF. (line 6)
6483 * --gc-sections: Options. (line 852)
6484 * --got: Options. (line 1828)
6485 * --got=TYPE: M68K. (line 6)
6486 * --gpsize: Options. (line 261)
6487 * --hash-size=NUMBER: Options. (line 1404)
6488 * --hash-style=STYLE: Options. (line 1412)
6489 * --heap: Options. (line 1539)
6490 * --help: Options. (line 883)
6491 * --image-base: Options. (line 1546)
6492 * --just-symbols=FILE: Options. (line 457)
6493 * --kill-at: Options. (line 1555)
6494 * --large-address-aware: Options. (line 1560)
6495 * --library-path=DIR: Options. (line 320)
6496 * --library=NAMESPEC: Options. (line 287)
6497 * --local-store=lo:hi: SPU ELF. (line 24)
6498 * --major-image-version: Options. (line 1569)
6499 * --major-os-version: Options. (line 1574)
6500 * --major-subsystem-version: Options. (line 1578)
6501 * --minor-image-version: Options. (line 1583)
6502 * --minor-os-version: Options. (line 1588)
6503 * --minor-subsystem-version: Options. (line 1592)
6504 * --mri-script=MRI-CMDFILE: Options. (line 141)
6505 * --multi-subspace: HPPA ELF32. (line 6)
6506 * --nmagic: Options. (line 389)
6507 * --no-accept-unknown-input-arch: Options. (line 663)
6508 * --no-add-needed: Options. (line 685)
6509 * --no-allow-shlib-undefined: Options. (line 916)
6510 * --no-as-needed: Options. (line 673)
6511 * --no-check-sections: Options. (line 767)
6512 * --no-define-common: Options. (line 789)
6513 * --no-demangle: Options. (line 818)
6514 * --no-dotsyms: PowerPC64 ELF64. (line 33)
6515 * --no-enum-size-warning: ARM. (line 106)
6516 * --no-fatal-warnings: Options. (line 837)
6517 * --no-fix-cortex-a8: i960. (line 39)
6518 * --no-gc-sections: Options. (line 852)
6519 * --no-keep-memory: Options. (line 895)
6520 * --no-multi-toc: PowerPC64 ELF64. (line 74)
6521 * --no-omagic: Options. (line 403)
6522 * --no-opd-optimize: PowerPC64 ELF64. (line 48)
6523 * --no-overlays: SPU ELF. (line 9)
6524 * --no-print-gc-sections: Options. (line 874)
6525 * --no-relax: Xtensa. (line 56)
6526 * --no-tls-optimize <1>: PowerPC64 ELF64. (line 43)
6527 * --no-tls-optimize: PowerPC ELF32. (line 48)
6528 * --no-toc-optimize: PowerPC64 ELF64. (line 60)
6529 * --no-trampoline: Options. (line 1809)
6530 * --no-undefined: Options. (line 902)
6531 * --no-undefined-version: Options. (line 935)
6532 * --no-warn-mismatch: Options. (line 948)
6533 * --no-warn-search-mismatch: Options. (line 957)
6534 * --no-wchar-size-warning: ARM. (line 113)
6535 * --no-whole-archive: Options. (line 961)
6536 * --noinhibit-exec: Options. (line 965)
6537 * --non-overlapping-opd: PowerPC64 ELF64. (line 54)
6538 * --oformat: Options. (line 977)
6539 * --omagic: Options. (line 394)
6540 * --out-implib: Options. (line 1605)
6541 * --output-def: Options. (line 1597)
6542 * --output=OUTPUT: Options. (line 409)
6543 * --pic-executable: Options. (line 990)
6544 * --pic-veneer: ARM. (line 119)
6545 * --plugin: SPU ELF. (line 6)
6546 * --print-gc-sections: Options. (line 874)
6547 * --print-map: Options. (line 352)
6548 * --reduce-memory-overheads: Options. (line 1418)
6549 * --relax: Options. (line 1006)
6550 * --relax on i960: i960. (line 31)
6551 * --relax on PowerPC: PowerPC ELF32. (line 6)
6552 * --relax on Xtensa: Xtensa. (line 27)
6553 * --relocatable: Options. (line 438)
6554 * --script=SCRIPT: Options. (line 481)
6555 * --sdata-got: PowerPC ELF32. (line 30)
6556 * --section-alignment: Options. (line 1780)
6557 * --section-start SECTIONNAME=ORG: Options. (line 1181)
6558 * --secure-plt: PowerPC ELF32. (line 23)
6559 * --sort-common: Options. (line 1126)
6560 * --sort-section alignment: Options. (line 1138)
6561 * --sort-section name: Options. (line 1134)
6562 * --split-by-file: Options. (line 1142)
6563 * --split-by-reloc: Options. (line 1147)
6564 * --stack: Options. (line 1786)
6565 * --stack-analysis: SPU ELF. (line 29)
6566 * --stats: Options. (line 1160)
6567 * --strip-all: Options. (line 468)
6568 * --strip-debug: Options. (line 472)
6569 * --stub-group-size: PowerPC64 ELF64. (line 6)
6570 * --stub-group-size=N <1>: HPPA ELF32. (line 12)
6571 * --stub-group-size=N: ARM. (line 124)
6572 * --subsystem: Options. (line 1793)
6573 * --support-old-code: ARM. (line 6)
6574 * --sysroot: Options. (line 1164)
6575 * --target-help: Options. (line 887)
6576 * --target1-abs: ARM. (line 27)
6577 * --target1-rel: ARM. (line 27)
6578 * --target2=TYPE: ARM. (line 32)
6579 * --thumb-entry=ENTRY: ARM. (line 17)
6580 * --trace: Options. (line 477)
6581 * --trace-symbol=SYMBOL: Options. (line 546)
6582 * --traditional-format: Options. (line 1169)
6583 * --undefined=SYMBOL: Options. (line 503)
6584 * --unique[=SECTION]: Options. (line 521)
6585 * --unresolved-symbols: Options. (line 1196)
6586 * --use-blx: ARM. (line 69)
6587 * --verbose: Options. (line 1225)
6588 * --version: Options. (line 530)
6589 * --version-script=VERSION-SCRIPTFILE: Options. (line 1231)
6590 * --vfp11-denorm-fix: ARM. (line 78)
6591 * --warn-common: Options. (line 1238)
6592 * --warn-constructors: Options. (line 1306)
6593 * --warn-multiple-gp: Options. (line 1311)
6594 * --warn-once: Options. (line 1325)
6595 * --warn-section-align: Options. (line 1329)
6596 * --warn-shared-textrel: Options. (line 1336)
6597 * --warn-unresolved-symbols: Options. (line 1339)
6598 * --whole-archive: Options. (line 1348)
6599 * --wrap: Options. (line 1362)
6600 * -AARCH: Options. (line 105)
6601 * -aKEYWORD: Options. (line 98)
6602 * -assert KEYWORD: Options. (line 695)
6603 * -b FORMAT: Options. (line 117)
6604 * -Bdynamic: Options. (line 698)
6605 * -Bgroup: Options. (line 708)
6606 * -Bshareable: Options. (line 1118)
6607 * -Bstatic: Options. (line 715)
6608 * -Bsymbolic: Options. (line 730)
6609 * -Bsymbolic-functions: Options. (line 737)
6610 * -c MRI-CMDFILE: Options. (line 141)
6611 * -call_shared: Options. (line 698)
6612 * -d: Options. (line 151)
6613 * -dc: Options. (line 151)
6614 * -dn: Options. (line 715)
6615 * -dp: Options. (line 151)
6616 * -dT SCRIPT: Options. (line 490)
6617 * -dy: Options. (line 698)
6618 * -E: Options. (line 181)
6619 * -e ENTRY: Options. (line 160)
6620 * -EB: Options. (line 200)
6621 * -EL: Options. (line 203)
6622 * -F: Options. (line 228)
6623 * -f: Options. (line 207)
6624 * -fini: Options. (line 252)
6625 * -G: Options. (line 261)
6626 * -g: Options. (line 258)
6627 * -hNAME: Options. (line 269)
6628 * -i: Options. (line 278)
6629 * -IFILE: Options. (line 831)
6630 * -init: Options. (line 281)
6631 * -LDIR: Options. (line 320)
6632 * -lNAMESPEC: Options. (line 287)
6633 * -M: Options. (line 352)
6634 * -m EMULATION: Options. (line 342)
6635 * -Map: Options. (line 891)
6636 * -N: Options. (line 394)
6637 * -n: Options. (line 389)
6638 * -non_shared: Options. (line 715)
6639 * -nostdlib: Options. (line 971)
6640 * -O LEVEL: Options. (line 415)
6641 * -o OUTPUT: Options. (line 409)
6642 * -pie: Options. (line 990)
6643 * -q: Options. (line 425)
6644 * -qmagic: Options. (line 1000)
6645 * -Qy: Options. (line 1003)
6646 * -r: Options. (line 438)
6647 * -R FILE: Options. (line 457)
6648 * -rpath: Options. (line 1041)
6649 * -rpath-link: Options. (line 1063)
6650 * -S: Options. (line 472)
6651 * -s: Options. (line 468)
6652 * -shared: Options. (line 1118)
6653 * -soname=NAME: Options. (line 269)
6654 * -static: Options. (line 715)
6655 * -t: Options. (line 477)
6656 * -T SCRIPT: Options. (line 481)
6657 * -Tbss ORG: Options. (line 1190)
6658 * -Tdata ORG: Options. (line 1190)
6659 * -Ttext ORG: Options. (line 1190)
6660 * -u SYMBOL: Options. (line 503)
6661 * -Ur: Options. (line 511)
6662 * -V: Options. (line 530)
6663 * -v: Options. (line 530)
6664 * -X: Options. (line 540)
6665 * -x: Options. (line 536)
6666 * -Y PATH: Options. (line 555)
6667 * -y SYMBOL: Options. (line 546)
6668 * -z defs: Options. (line 902)
6669 * -z KEYWORD: Options. (line 559)
6670 * -z muldefs: Options. (line 910)
6671 * .: Location Counter. (line 6)
6672 * /DISCARD/: Output Section Discarding.
6674 * :PHDR: Output Section Phdr.
6676 * =FILLEXP: Output Section Fill.
6678 * >REGION: Output Section Region.
6680 * [COMMON]: Input Section Common.
6682 * ABSOLUTE (MRI): MRI. (line 33)
6683 * absolute and relocatable symbols: Expression Section. (line 6)
6684 * absolute expressions: Expression Section. (line 6)
6685 * ABSOLUTE(EXP): Builtin Functions. (line 10)
6686 * ADDR(SECTION): Builtin Functions. (line 17)
6687 * address, section: Output Section Address.
6689 * ALIAS (MRI): MRI. (line 44)
6690 * ALIGN (MRI): MRI. (line 50)
6691 * align expression: Builtin Functions. (line 36)
6692 * align location counter: Builtin Functions. (line 36)
6693 * ALIGN(ALIGN): Builtin Functions. (line 36)
6694 * ALIGN(EXP,ALIGN): Builtin Functions. (line 36)
6695 * ALIGN(SECTION_ALIGN): Forced Output Alignment.
6697 * ALIGNOF(SECTION): Builtin Functions. (line 62)
6698 * allocating memory: MEMORY. (line 6)
6699 * architecture: Miscellaneous Commands.
6701 * architectures: Options. (line 105)
6702 * archive files, from cmd line: Options. (line 287)
6703 * archive search path in linker script: File Commands. (line 74)
6704 * arithmetic: Expressions. (line 6)
6705 * arithmetic operators: Operators. (line 6)
6706 * ARM interworking support: ARM. (line 6)
6707 * AS_NEEDED(FILES): File Commands. (line 54)
6708 * ASSERT: Miscellaneous Commands.
6710 * assertion in linker script: Miscellaneous Commands.
6712 * assignment in scripts: Assignments. (line 6)
6713 * AT(LMA): Output Section LMA. (line 6)
6714 * AT>LMA_REGION: Output Section LMA. (line 6)
6715 * automatic data imports: WIN32. (line 170)
6716 * back end: BFD. (line 6)
6717 * BASE (MRI): MRI. (line 54)
6718 * BE8: ARM. (line 23)
6719 * BFD canonical format: Canonical format. (line 11)
6720 * BFD requirements: BFD. (line 16)
6721 * big-endian objects: Options. (line 200)
6722 * binary input format: Options. (line 117)
6723 * BLOCK(EXP): Builtin Functions. (line 75)
6724 * bug criteria: Bug Criteria. (line 6)
6725 * bug reports: Bug Reporting. (line 6)
6726 * bugs in ld: Reporting Bugs. (line 6)
6727 * BYTE(EXPRESSION): Output Section Data.
6729 * C++ constructors, arranging in link: Output Section Keywords.
6731 * CHIP (MRI): MRI. (line 58)
6732 * COLLECT_NO_DEMANGLE: Environment. (line 29)
6733 * combining symbols, warnings on: Options. (line 1238)
6734 * command files: Scripts. (line 6)
6735 * command line: Options. (line 6)
6736 * common allocation: Options. (line 151)
6737 * common allocation in linker script: Miscellaneous Commands.
6739 * common symbol placement: Input Section Common.
6741 * compatibility, MRI: Options. (line 141)
6742 * constants in linker scripts: Constants. (line 6)
6743 * CONSTRUCTORS: Output Section Keywords.
6745 * constructors: Options. (line 511)
6746 * constructors, arranging in link: Output Section Keywords.
6748 * Cortex-A8 erratum workaround: i960. (line 39)
6749 * crash of linker: Bug Criteria. (line 9)
6750 * CREATE_OBJECT_SYMBOLS: Output Section Keywords.
6752 * creating a DEF file: WIN32. (line 137)
6753 * cross reference table: Options. (line 777)
6754 * cross references: Miscellaneous Commands.
6756 * current output location: Location Counter. (line 6)
6757 * data: Output Section Data.
6759 * DATA_SEGMENT_ALIGN(MAXPAGESIZE, COMMONPAGESIZE): Builtin Functions.
6761 * DATA_SEGMENT_END(EXP): Builtin Functions. (line 101)
6762 * DATA_SEGMENT_RELRO_END(OFFSET, EXP): Builtin Functions. (line 107)
6763 * dbx: Options. (line 1174)
6764 * DEF files, creating: Options. (line 1597)
6765 * default emulation: Environment. (line 21)
6766 * default input format: Environment. (line 9)
6767 * DEFINED(SYMBOL): Builtin Functions. (line 118)
6768 * deleting local symbols: Options. (line 536)
6769 * demangling, default: Environment. (line 29)
6770 * demangling, from command line: Options. (line 818)
6771 * direct linking to a dll: WIN32. (line 218)
6772 * discarding sections: Output Section Discarding.
6774 * discontinuous memory: MEMORY. (line 6)
6775 * DLLs, creating: Options. (line 1503)
6776 * DLLs, linking to: Options. (line 1628)
6777 * dot: Location Counter. (line 6)
6778 * dot inside sections: Location Counter. (line 36)
6779 * dot outside sections: Location Counter. (line 66)
6780 * dynamic linker, from command line: Options. (line 831)
6781 * dynamic symbol table: Options. (line 181)
6782 * ELF program headers: PHDRS. (line 6)
6783 * emulation: Options. (line 342)
6784 * emulation, default: Environment. (line 21)
6785 * END (MRI): MRI. (line 62)
6786 * endianness: Options. (line 200)
6787 * entry point: Entry Point. (line 6)
6788 * entry point, from command line: Options. (line 160)
6789 * entry point, thumb: ARM. (line 17)
6790 * ENTRY(SYMBOL): Entry Point. (line 6)
6791 * error on valid input: Bug Criteria. (line 12)
6792 * example of linker script: Simple Example. (line 6)
6793 * exporting DLL symbols: WIN32. (line 19)
6794 * expression evaluation order: Evaluation. (line 6)
6795 * expression sections: Expression Section. (line 6)
6796 * expression, absolute: Builtin Functions. (line 10)
6797 * expressions: Expressions. (line 6)
6798 * EXTERN: Miscellaneous Commands.
6800 * fatal signal: Bug Criteria. (line 9)
6801 * file name wildcard patterns: Input Section Wildcards.
6803 * FILEHDR: PHDRS. (line 61)
6804 * filename symbols: Output Section Keywords.
6806 * fill pattern, entire section: Output Section Fill.
6808 * FILL(EXPRESSION): Output Section Data.
6810 * finalization function: Options. (line 252)
6811 * first input file: File Commands. (line 82)
6812 * first instruction: Entry Point. (line 6)
6813 * FIX_V4BX: ARM. (line 44)
6814 * FIX_V4BX_INTERWORKING: ARM. (line 57)
6815 * FORCE_COMMON_ALLOCATION: Miscellaneous Commands.
6817 * forcing input section alignment: Forced Input Alignment.
6819 * forcing output section alignment: Forced Output Alignment.
6821 * forcing the creation of dynamic sections: Options. (line 434)
6822 * FORMAT (MRI): MRI. (line 66)
6823 * functions in expressions: Builtin Functions. (line 6)
6824 * garbage collection <1>: Input Section Keep. (line 6)
6825 * garbage collection: Options. (line 852)
6826 * generating optimized output: Options. (line 415)
6827 * GNU linker: Overview. (line 6)
6828 * GNUTARGET: Environment. (line 9)
6829 * GROUP(FILES): File Commands. (line 47)
6830 * grouping input files: File Commands. (line 47)
6831 * groups of archives: Options. (line 645)
6832 * H8/300 support: H8/300. (line 6)
6833 * header size: Builtin Functions. (line 183)
6834 * heap size: Options. (line 1539)
6835 * help: Options. (line 883)
6836 * holes: Location Counter. (line 12)
6837 * holes, filling: Output Section Data.
6839 * HPPA multiple sub-space stubs: HPPA ELF32. (line 6)
6840 * HPPA stub grouping: HPPA ELF32. (line 12)
6841 * i960 support: i960. (line 6)
6842 * image base: Options. (line 1546)
6843 * implicit linker scripts: Implicit Linker Scripts.
6845 * import libraries: WIN32. (line 10)
6846 * INCLUDE FILENAME: File Commands. (line 9)
6847 * including a linker script: File Commands. (line 9)
6848 * including an entire archive: Options. (line 1348)
6849 * incremental link: Options. (line 278)
6850 * INHIBIT_COMMON_ALLOCATION: Miscellaneous Commands.
6852 * initialization function: Options. (line 281)
6853 * initialized data in ROM: Output Section LMA. (line 26)
6854 * input file format in linker script: Format Commands. (line 35)
6855 * input filename symbols: Output Section Keywords.
6857 * input files in linker scripts: File Commands. (line 19)
6858 * input files, displaying: Options. (line 477)
6859 * input format: Options. (line 117)
6860 * input object files in linker scripts: File Commands. (line 19)
6861 * input section alignment: Forced Input Alignment.
6863 * input section basics: Input Section Basics.
6865 * input section wildcards: Input Section Wildcards.
6867 * input sections: Input Section. (line 6)
6868 * INPUT(FILES): File Commands. (line 19)
6869 * INSERT: Miscellaneous Commands.
6871 * insert user script into default script: Miscellaneous Commands.
6873 * integer notation: Constants. (line 6)
6874 * integer suffixes: Constants. (line 12)
6875 * internal object-file format: Canonical format. (line 11)
6876 * invalid input: Bug Criteria. (line 14)
6877 * K and M integer suffixes: Constants. (line 12)
6878 * KEEP: Input Section Keep. (line 6)
6879 * l =: MEMORY. (line 72)
6880 * lazy evaluation: Evaluation. (line 6)
6881 * ld bugs, reporting: Bug Reporting. (line 6)
6882 * LDEMULATION: Environment. (line 21)
6883 * len =: MEMORY. (line 72)
6884 * LENGTH =: MEMORY. (line 72)
6885 * LENGTH(MEMORY): Builtin Functions. (line 135)
6886 * library search path in linker script: File Commands. (line 74)
6887 * link map: Options. (line 352)
6888 * link-time runtime library search path: Options. (line 1063)
6889 * linker crash: Bug Criteria. (line 9)
6890 * linker script concepts: Basic Script Concepts.
6892 * linker script example: Simple Example. (line 6)
6893 * linker script file commands: File Commands. (line 6)
6894 * linker script format: Script Format. (line 6)
6895 * linker script input object files: File Commands. (line 19)
6896 * linker script simple commands: Simple Commands. (line 6)
6897 * linker scripts: Scripts. (line 6)
6898 * LIST (MRI): MRI. (line 77)
6899 * little-endian objects: Options. (line 203)
6900 * LOAD (MRI): MRI. (line 84)
6901 * load address: Output Section LMA. (line 6)
6902 * LOADADDR(SECTION): Builtin Functions. (line 138)
6903 * loading, preventing: Output Section Type.
6905 * local symbols, deleting: Options. (line 540)
6906 * location counter: Location Counter. (line 6)
6907 * LONG(EXPRESSION): Output Section Data.
6909 * M and K integer suffixes: Constants. (line 12)
6910 * M68HC11 and 68HC12 support: M68HC11/68HC12. (line 6)
6911 * machine architecture: Miscellaneous Commands.
6913 * machine dependencies: Machine Dependent. (line 6)
6914 * mapping input sections to output sections: Input Section. (line 6)
6915 * MAX: Builtin Functions. (line 143)
6916 * MEMORY: MEMORY. (line 6)
6917 * memory region attributes: MEMORY. (line 32)
6918 * memory regions: MEMORY. (line 6)
6919 * memory regions and sections: Output Section Region.
6921 * memory usage: Options. (line 895)
6922 * MIN: Builtin Functions. (line 146)
6923 * Motorola 68K GOT generation: M68K. (line 6)
6924 * MRI compatibility: MRI. (line 6)
6925 * MSP430 extra sections: MSP430. (line 11)
6926 * NAME (MRI): MRI. (line 90)
6927 * name, section: Output Section Name.
6929 * names: Symbols. (line 6)
6930 * naming the output file: Options. (line 409)
6931 * NEXT(EXP): Builtin Functions. (line 150)
6932 * NMAGIC: Options. (line 389)
6933 * NO_ENUM_SIZE_WARNING: ARM. (line 106)
6934 * NO_WCHAR_SIZE_WARNING: ARM. (line 113)
6935 * NOCROSSREFS(SECTIONS): Miscellaneous Commands.
6937 * NOLOAD: Output Section Type.
6939 * not enough room for program headers: Builtin Functions. (line 188)
6940 * o =: MEMORY. (line 67)
6941 * objdump -i: BFD. (line 6)
6942 * object file management: BFD. (line 6)
6943 * object files: Options. (line 29)
6944 * object formats available: BFD. (line 6)
6945 * object size: Options. (line 261)
6946 * OMAGIC: Options. (line 394)
6947 * opening object files: BFD outline. (line 6)
6948 * operators for arithmetic: Operators. (line 6)
6949 * options: Options. (line 6)
6950 * ORDER (MRI): MRI. (line 95)
6951 * org =: MEMORY. (line 67)
6952 * ORIGIN =: MEMORY. (line 67)
6953 * ORIGIN(MEMORY): Builtin Functions. (line 156)
6954 * orphan: Orphan Sections. (line 6)
6955 * output file after errors: Options. (line 965)
6956 * output file format in linker script: Format Commands. (line 10)
6957 * output file name in linker script: File Commands. (line 64)
6958 * output section alignment: Forced Output Alignment.
6960 * output section attributes: Output Section Attributes.
6962 * output section data: Output Section Data.
6964 * OUTPUT(FILENAME): File Commands. (line 64)
6965 * OUTPUT_ARCH(BFDARCH): Miscellaneous Commands.
6967 * OUTPUT_FORMAT(BFDNAME): Format Commands. (line 10)
6968 * OVERLAY: Overlay Description.
6970 * overlays: Overlay Description.
6972 * partial link: Options. (line 438)
6973 * PHDRS: PHDRS. (line 6)
6974 * PIC_VENEER: ARM. (line 119)
6975 * position independent executables: Options. (line 992)
6976 * PowerPC ELF32 options: PowerPC ELF32. (line 13)
6977 * PowerPC GOT: PowerPC ELF32. (line 30)
6978 * PowerPC long branches: PowerPC ELF32. (line 6)
6979 * PowerPC PLT: PowerPC ELF32. (line 13)
6980 * PowerPC stub symbols: PowerPC ELF32. (line 44)
6981 * PowerPC TLS optimization: PowerPC ELF32. (line 48)
6982 * PowerPC64 dot symbols: PowerPC64 ELF64. (line 33)
6983 * PowerPC64 ELF64 options: PowerPC64 ELF64. (line 6)
6984 * PowerPC64 multi-TOC: PowerPC64 ELF64. (line 74)
6985 * PowerPC64 OPD optimization: PowerPC64 ELF64. (line 48)
6986 * PowerPC64 OPD spacing: PowerPC64 ELF64. (line 54)
6987 * PowerPC64 stub grouping: PowerPC64 ELF64. (line 6)
6988 * PowerPC64 stub symbols: PowerPC64 ELF64. (line 29)
6989 * PowerPC64 TLS optimization: PowerPC64 ELF64. (line 43)
6990 * PowerPC64 TOC optimization: PowerPC64 ELF64. (line 60)
6991 * precedence in expressions: Operators. (line 6)
6992 * prevent unnecessary loading: Output Section Type.
6994 * program headers: PHDRS. (line 6)
6995 * program headers and sections: Output Section Phdr.
6997 * program headers, not enough room: Builtin Functions. (line 188)
6998 * program segments: PHDRS. (line 6)
6999 * PROVIDE: PROVIDE. (line 6)
7000 * PROVIDE_HIDDEN: PROVIDE_HIDDEN. (line 6)
7001 * PUBLIC (MRI): MRI. (line 103)
7002 * QUAD(EXPRESSION): Output Section Data.
7004 * quoted symbol names: Symbols. (line 6)
7005 * read-only text: Options. (line 389)
7006 * read/write from cmd line: Options. (line 394)
7007 * regions of memory: MEMORY. (line 6)
7008 * relative expressions: Expression Section. (line 6)
7009 * relaxing addressing modes: Options. (line 1006)
7010 * relaxing on H8/300: H8/300. (line 9)
7011 * relaxing on i960: i960. (line 31)
7012 * relaxing on M68HC11: M68HC11/68HC12. (line 12)
7013 * relaxing on Xtensa: Xtensa. (line 27)
7014 * relocatable and absolute symbols: Expression Section. (line 6)
7015 * relocatable output: Options. (line 438)
7016 * removing sections: Output Section Discarding.
7018 * reporting bugs in ld: Reporting Bugs. (line 6)
7019 * requirements for BFD: BFD. (line 16)
7020 * retain relocations in final executable: Options. (line 425)
7021 * retaining specified symbols: Options. (line 1027)
7022 * ROM initialized data: Output Section LMA. (line 26)
7023 * round up expression: Builtin Functions. (line 36)
7024 * round up location counter: Builtin Functions. (line 36)
7025 * runtime library name: Options. (line 269)
7026 * runtime library search path: Options. (line 1041)
7027 * runtime pseudo-relocation: WIN32. (line 196)
7028 * scaled integers: Constants. (line 12)
7029 * scommon section: Input Section Common.
7031 * script files: Options. (line 481)
7032 * scripts: Scripts. (line 6)
7033 * search directory, from cmd line: Options. (line 320)
7034 * search path in linker script: File Commands. (line 74)
7035 * SEARCH_DIR(PATH): File Commands. (line 74)
7036 * SECT (MRI): MRI. (line 109)
7037 * section address: Output Section Address.
7039 * section address in expression: Builtin Functions. (line 17)
7040 * section alignment: Builtin Functions. (line 62)
7041 * section alignment, warnings on: Options. (line 1329)
7042 * section data: Output Section Data.
7044 * section fill pattern: Output Section Fill.
7046 * section load address: Output Section LMA. (line 6)
7047 * section load address in expression: Builtin Functions. (line 138)
7048 * section name: Output Section Name.
7050 * section name wildcard patterns: Input Section Wildcards.
7052 * section size: Builtin Functions. (line 167)
7053 * section, assigning to memory region: Output Section Region.
7055 * section, assigning to program header: Output Section Phdr.
7057 * SECTIONS: SECTIONS. (line 6)
7058 * sections, discarding: Output Section Discarding.
7060 * segment origins, cmd line: Options. (line 1190)
7061 * SEGMENT_START(SEGMENT, DEFAULT): Builtin Functions. (line 159)
7062 * segments, ELF: PHDRS. (line 6)
7063 * shared libraries: Options. (line 1120)
7064 * SHORT(EXPRESSION): Output Section Data.
7066 * SIZEOF(SECTION): Builtin Functions. (line 167)
7067 * SIZEOF_HEADERS: Builtin Functions. (line 183)
7068 * small common symbols: Input Section Common.
7070 * SORT: Input Section Wildcards.
7072 * SORT_BY_ALIGNMENT: Input Section Wildcards.
7074 * SORT_BY_NAME: Input Section Wildcards.
7076 * SPU: SPU ELF. (line 29)
7077 * SPU ELF options: SPU ELF. (line 6)
7078 * SPU extra overlay stubs: SPU ELF. (line 19)
7079 * SPU local store size: SPU ELF. (line 24)
7080 * SPU overlay stub symbols: SPU ELF. (line 15)
7081 * SPU overlays: SPU ELF. (line 9)
7082 * SPU plugins: SPU ELF. (line 6)
7083 * SQUAD(EXPRESSION): Output Section Data.
7085 * stack size: Options. (line 1786)
7086 * standard Unix system: Options. (line 7)
7087 * start of execution: Entry Point. (line 6)
7088 * STARTUP(FILENAME): File Commands. (line 82)
7089 * strip all symbols: Options. (line 468)
7090 * strip debugger symbols: Options. (line 472)
7091 * stripping all but some symbols: Options. (line 1027)
7092 * STUB_GROUP_SIZE: ARM. (line 124)
7093 * SUBALIGN(SUBSECTION_ALIGN): Forced Input Alignment.
7095 * suffixes for integers: Constants. (line 12)
7096 * symbol defaults: Builtin Functions. (line 118)
7097 * symbol definition, scripts: Assignments. (line 6)
7098 * symbol names: Symbols. (line 6)
7099 * symbol tracing: Options. (line 546)
7100 * symbol versions: VERSION. (line 6)
7101 * symbol-only input: Options. (line 457)
7102 * symbols, from command line: Options. (line 805)
7103 * symbols, relocatable and absolute: Expression Section. (line 6)
7104 * symbols, retaining selectively: Options. (line 1027)
7105 * synthesizing linker: Options. (line 1006)
7106 * synthesizing on H8/300: H8/300. (line 14)
7107 * TARGET(BFDNAME): Format Commands. (line 35)
7108 * TARGET1: ARM. (line 27)
7109 * TARGET2: ARM. (line 32)
7110 * thumb entry point: ARM. (line 17)
7111 * TI COFF versions: TI COFF. (line 6)
7112 * traditional format: Options. (line 1169)
7113 * trampoline generation on M68HC11: M68HC11/68HC12. (line 31)
7114 * trampoline generation on M68HC12: M68HC11/68HC12. (line 31)
7115 * unallocated address, next: Builtin Functions. (line 150)
7116 * undefined symbol: Options. (line 503)
7117 * undefined symbol in linker script: Miscellaneous Commands.
7119 * undefined symbols, warnings on: Options. (line 1325)
7120 * uninitialized data placement: Input Section Common.
7122 * unspecified memory: Output Section Data.
7124 * usage: Options. (line 883)
7125 * USE_BLX: ARM. (line 69)
7126 * using a DEF file: WIN32. (line 42)
7127 * using auto-export functionality: WIN32. (line 22)
7128 * Using decorations: WIN32. (line 141)
7129 * variables, defining: Assignments. (line 6)
7130 * verbose: Options. (line 1225)
7131 * version: Options. (line 530)
7132 * version script: VERSION. (line 6)
7133 * version script, symbol versions: Options. (line 1231)
7134 * VERSION {script text}: VERSION. (line 6)
7135 * versions of symbols: VERSION. (line 6)
7136 * VFP11_DENORM_FIX: ARM. (line 78)
7137 * warnings, on combining symbols: Options. (line 1238)
7138 * warnings, on section alignment: Options. (line 1329)
7139 * warnings, on undefined symbols: Options. (line 1325)
7140 * weak externals: WIN32. (line 386)
7141 * what is this?: Overview. (line 6)
7142 * wildcard file name patterns: Input Section Wildcards.
7144 * Xtensa options: Xtensa. (line 56)
7145 * Xtensa processors: Xtensa. (line 6)
7151 Node: Overview
\7f1600
7152 Node: Invocation
\7f2714
7153 Node: Options
\7f3122
7154 Node: Environment
\7f85587
7155 Node: Scripts
\7f87347
7156 Node: Basic Script Concepts
\7f89081
7157 Node: Script Format
\7f91788
7158 Node: Simple Example
\7f92651
7159 Node: Simple Commands
\7f95747
7160 Node: Entry Point
\7f96198
7161 Node: File Commands
\7f96957
7162 Node: Format Commands
\7f100958
7163 Node: Miscellaneous Commands
\7f102924
7164 Node: Assignments
\7f106303
7165 Node: Simple Assignments
\7f106794
7166 Node: PROVIDE
\7f108530
7167 Node: PROVIDE_HIDDEN
\7f109735
7168 Node: Source Code Reference
\7f109979
7169 Node: SECTIONS
\7f113559
7170 Node: Output Section Description
\7f115450
7171 Node: Output Section Name
\7f116503
7172 Node: Output Section Address
\7f117379
7173 Node: Input Section
\7f119028
7174 Node: Input Section Basics
\7f119829
7175 Node: Input Section Wildcards
\7f123047
7176 Node: Input Section Common
\7f127780
7177 Node: Input Section Keep
\7f129262
7178 Node: Input Section Example
\7f129752
7179 Node: Output Section Data
\7f130720
7180 Node: Output Section Keywords
\7f133497
7181 Node: Output Section Discarding
\7f137066
7182 Node: Output Section Attributes
\7f138247
7183 Node: Output Section Type
\7f139251
7184 Node: Output Section LMA
\7f140405
7185 Node: Forced Output Alignment
\7f142918
7186 Node: Forced Input Alignment
\7f143186
7187 Node: Output Section Region
\7f143571
7188 Node: Output Section Phdr
\7f144001
7189 Node: Output Section Fill
\7f144665
7190 Node: Overlay Description
\7f145807
7191 Node: MEMORY
\7f150110
7192 Node: PHDRS
\7f154310
7193 Node: VERSION
\7f159349
7194 Node: Expressions
\7f167141
7195 Node: Constants
\7f168019
7196 Node: Symbols
\7f168580
7197 Node: Orphan Sections
\7f169318
7198 Node: Location Counter
\7f170482
7199 Node: Operators
\7f174918
7200 Node: Evaluation
\7f175840
7201 Node: Expression Section
\7f177204
7202 Node: Builtin Functions
\7f178693
7203 Node: Implicit Linker Scripts
\7f186660
7204 Node: Machine Dependent
\7f187435
7205 Node: H8/300
\7f188451
7207 Node: M68HC11/68HC12
\7f192177
7209 Node: HPPA ELF32
\7f200880
7212 Node: MSP430
\7f204577
7213 Node: PowerPC ELF32
\7f205626
7214 Node: PowerPC64 ELF64
\7f208240
7215 Node: SPU ELF
\7f212656
7216 Node: TI COFF
\7f215288
7217 Node: WIN32
\7f215814
7218 Node: Xtensa
\7f234171
7220 Node: BFD outline
\7f238748
7221 Node: BFD information loss
\7f240034
7222 Node: Canonical format
\7f242551
7223 Node: Reporting Bugs
\7f246908
7224 Node: Bug Criteria
\7f247602
7225 Node: Bug Reporting
\7f248301
7227 Node: GNU Free Documentation License
\7f259983
7228 Node: LD Index
\7f279700