3 @c Copyright 1991-2013 Free Software Foundation, Inc.
6 @include configdoc.texi
7 @c (configdoc.texi is generated by the Makefile)
13 @macro gcctabopt{body}
19 @c Configure for the generation of man pages
42 @dircategory Software development
44 * Ld: (ld). The GNU linker.
49 This file documents the @sc{gnu} linker LD
50 @ifset VERSION_PACKAGE
51 @value{VERSION_PACKAGE}
53 version @value{VERSION}.
55 Copyright @copyright{} 1991-2013 Free Software Foundation, Inc.
57 Permission is granted to copy, distribute and/or modify this document
58 under the terms of the GNU Free Documentation License, Version 1.3
59 or any later version published by the Free Software Foundation;
60 with no Invariant Sections, with no Front-Cover Texts, and with no
61 Back-Cover Texts. A copy of the license is included in the
62 section entitled ``GNU Free Documentation License''.
66 @setchapternewpage odd
67 @settitle The GNU linker
72 @ifset VERSION_PACKAGE
73 @subtitle @value{VERSION_PACKAGE}
75 @subtitle Version @value{VERSION}
76 @author Steve Chamberlain
77 @author Ian Lance Taylor
82 \hfill Red Hat Inc\par
83 \hfill nickc\@credhat.com, doc\@redhat.com\par
84 \hfill {\it The GNU linker}\par
85 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
87 \global\parindent=0pt % Steve likes it this way.
90 @vskip 0pt plus 1filll
91 @c man begin COPYRIGHT
92 Copyright @copyright{} 1991-2013 Free Software Foundation, Inc.
94 Permission is granted to copy, distribute and/or modify this document
95 under the terms of the GNU Free Documentation License, Version 1.3
96 or any later version published by the Free Software Foundation;
97 with no Invariant Sections, with no Front-Cover Texts, and with no
98 Back-Cover Texts. A copy of the license is included in the
99 section entitled ``GNU Free Documentation License''.
105 @c FIXME: Talk about importance of *order* of args, cmds to linker!
110 This file documents the @sc{gnu} linker ld
111 @ifset VERSION_PACKAGE
112 @value{VERSION_PACKAGE}
114 version @value{VERSION}.
116 This document is distributed under the terms of the GNU Free
117 Documentation License version 1.3. A copy of the license is included
118 in the section entitled ``GNU Free Documentation License''.
121 * Overview:: Overview
122 * Invocation:: Invocation
123 * Scripts:: Linker Scripts
125 * Machine Dependent:: Machine Dependent Features
129 * H8/300:: ld and the H8/300
132 * Renesas:: ld and other Renesas micros
135 * i960:: ld and the Intel 960 family
138 * ARM:: ld and the ARM family
141 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
144 * HPPA ELF32:: ld and HPPA 32-bit ELF
147 * M68K:: ld and Motorola 68K family
150 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
153 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
156 * SPU ELF:: ld and SPU ELF Support
159 * TI COFF:: ld and the TI COFF
162 * Win32:: ld and WIN32 (cygwin/mingw)
165 * Xtensa:: ld and Xtensa Processors
168 @ifclear SingleFormat
171 @c Following blank line required for remaining bug in makeinfo conds/menus
173 * Reporting Bugs:: Reporting Bugs
174 * MRI:: MRI Compatible Script Files
175 * GNU Free Documentation License:: GNU Free Documentation License
176 * LD Index:: LD Index
183 @cindex @sc{gnu} linker
184 @cindex what is this?
187 @c man begin SYNOPSIS
188 ld [@b{options}] @var{objfile} @dots{}
192 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
193 the Info entries for @file{binutils} and
198 @c man begin DESCRIPTION
200 @command{ld} combines a number of object and archive files, relocates
201 their data and ties up symbol references. Usually the last step in
202 compiling a program is to run @command{ld}.
204 @command{ld} accepts Linker Command Language files written in
205 a superset of AT&T's Link Editor Command Language syntax,
206 to provide explicit and total control over the linking process.
210 This man page does not describe the command language; see the
211 @command{ld} entry in @code{info} for full details on the command
212 language and on other aspects of the GNU linker.
215 @ifclear SingleFormat
216 This version of @command{ld} uses the general purpose BFD libraries
217 to operate on object files. This allows @command{ld} to read, combine, and
218 write object files in many different formats---for example, COFF or
219 @code{a.out}. Different formats may be linked together to produce any
220 available kind of object file. @xref{BFD}, for more information.
223 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
224 linkers in providing diagnostic information. Many linkers abandon
225 execution immediately upon encountering an error; whenever possible,
226 @command{ld} continues executing, allowing you to identify other errors
227 (or, in some cases, to get an output file in spite of the error).
234 @c man begin DESCRIPTION
236 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
237 and to be as compatible as possible with other linkers. As a result,
238 you have many choices to control its behavior.
244 * Options:: Command Line Options
245 * Environment:: Environment Variables
249 @section Command Line Options
257 The linker supports a plethora of command-line options, but in actual
258 practice few of them are used in any particular context.
259 @cindex standard Unix system
260 For instance, a frequent use of @command{ld} is to link standard Unix
261 object files on a standard, supported Unix system. On such a system, to
262 link a file @code{hello.o}:
265 ld -o @var{output} /lib/crt0.o hello.o -lc
268 This tells @command{ld} to produce a file called @var{output} as the
269 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
270 the library @code{libc.a}, which will come from the standard search
271 directories. (See the discussion of the @samp{-l} option below.)
273 Some of the command-line options to @command{ld} may be specified at any
274 point in the command line. However, options which refer to files, such
275 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
276 which the option appears in the command line, relative to the object
277 files and other file options. Repeating non-file options with a
278 different argument will either have no further effect, or override prior
279 occurrences (those further to the left on the command line) of that
280 option. Options which may be meaningfully specified more than once are
281 noted in the descriptions below.
284 Non-option arguments are object files or archives which are to be linked
285 together. They may follow, precede, or be mixed in with command-line
286 options, except that an object file argument may not be placed between
287 an option and its argument.
289 Usually the linker is invoked with at least one object file, but you can
290 specify other forms of binary input files using @samp{-l}, @samp{-R},
291 and the script command language. If @emph{no} binary input files at all
292 are specified, the linker does not produce any output, and issues the
293 message @samp{No input files}.
295 If the linker cannot recognize the format of an object file, it will
296 assume that it is a linker script. A script specified in this way
297 augments the main linker script used for the link (either the default
298 linker script or the one specified by using @samp{-T}). This feature
299 permits the linker to link against a file which appears to be an object
300 or an archive, but actually merely defines some symbol values, or uses
301 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
302 script in this way merely augments the main linker script, with the
303 extra commands placed after the main script; use the @samp{-T} option
304 to replace the default linker script entirely, but note the effect of
305 the @code{INSERT} command. @xref{Scripts}.
307 For options whose names are a single letter,
308 option arguments must either follow the option letter without intervening
309 whitespace, or be given as separate arguments immediately following the
310 option that requires them.
312 For options whose names are multiple letters, either one dash or two can
313 precede the option name; for example, @samp{-trace-symbol} and
314 @samp{--trace-symbol} are equivalent. Note---there is one exception to
315 this rule. Multiple letter options that start with a lower case 'o' can
316 only be preceded by two dashes. This is to reduce confusion with the
317 @samp{-o} option. So for example @samp{-omagic} sets the output file
318 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
321 Arguments to multiple-letter options must either be separated from the
322 option name by an equals sign, or be given as separate arguments
323 immediately following the option that requires them. For example,
324 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
325 Unique abbreviations of the names of multiple-letter options are
328 Note---if the linker is being invoked indirectly, via a compiler driver
329 (e.g. @samp{gcc}) then all the linker command line options should be
330 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
331 compiler driver) like this:
334 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
337 This is important, because otherwise the compiler driver program may
338 silently drop the linker options, resulting in a bad link. Confusion
339 may also arise when passing options that require values through a
340 driver, as the use of a space between option and argument acts as
341 a separator, and causes the driver to pass only the option to the linker
342 and the argument to the compiler. In this case, it is simplest to use
343 the joined forms of both single- and multiple-letter options, such as:
346 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
349 Here is a table of the generic command line switches accepted by the GNU
353 @include at-file.texi
355 @kindex -a @var{keyword}
356 @item -a @var{keyword}
357 This option is supported for HP/UX compatibility. The @var{keyword}
358 argument must be one of the strings @samp{archive}, @samp{shared}, or
359 @samp{default}. @samp{-aarchive} is functionally equivalent to
360 @samp{-Bstatic}, and the other two keywords are functionally equivalent
361 to @samp{-Bdynamic}. This option may be used any number of times.
363 @kindex --audit @var{AUDITLIB}
364 @item --audit @var{AUDITLIB}
365 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
366 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
367 specified in the library. If specified multiple times @code{DT_AUDIT}
368 will contain a colon separated list of audit interfaces to use. If the linker
369 finds an object with an audit entry while searching for shared libraries,
370 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
371 This option is only meaningful on ELF platforms supporting the rtld-audit
375 @cindex architectures
376 @kindex -A @var{arch}
377 @item -A @var{architecture}
378 @kindex --architecture=@var{arch}
379 @itemx --architecture=@var{architecture}
380 In the current release of @command{ld}, this option is useful only for the
381 Intel 960 family of architectures. In that @command{ld} configuration, the
382 @var{architecture} argument identifies the particular architecture in
383 the 960 family, enabling some safeguards and modifying the
384 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
385 family}, for details.
387 Future releases of @command{ld} may support similar functionality for
388 other architecture families.
391 @ifclear SingleFormat
392 @cindex binary input format
393 @kindex -b @var{format}
394 @kindex --format=@var{format}
397 @item -b @var{input-format}
398 @itemx --format=@var{input-format}
399 @command{ld} may be configured to support more than one kind of object
400 file. If your @command{ld} is configured this way, you can use the
401 @samp{-b} option to specify the binary format for input object files
402 that follow this option on the command line. Even when @command{ld} is
403 configured to support alternative object formats, you don't usually need
404 to specify this, as @command{ld} should be configured to expect as a
405 default input format the most usual format on each machine.
406 @var{input-format} is a text string, the name of a particular format
407 supported by the BFD libraries. (You can list the available binary
408 formats with @samp{objdump -i}.)
411 You may want to use this option if you are linking files with an unusual
412 binary format. You can also use @samp{-b} to switch formats explicitly (when
413 linking object files of different formats), by including
414 @samp{-b @var{input-format}} before each group of object files in a
417 The default format is taken from the environment variable
422 You can also define the input format from a script, using the command
425 see @ref{Format Commands}.
429 @kindex -c @var{MRI-cmdfile}
430 @kindex --mri-script=@var{MRI-cmdfile}
431 @cindex compatibility, MRI
432 @item -c @var{MRI-commandfile}
433 @itemx --mri-script=@var{MRI-commandfile}
434 For compatibility with linkers produced by MRI, @command{ld} accepts script
435 files written in an alternate, restricted command language, described in
437 @ref{MRI,,MRI Compatible Script Files}.
440 the MRI Compatible Script Files section of GNU ld documentation.
442 Introduce MRI script files with
443 the option @samp{-c}; use the @samp{-T} option to run linker
444 scripts written in the general-purpose @command{ld} scripting language.
445 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
446 specified by any @samp{-L} options.
448 @cindex common allocation
455 These three options are equivalent; multiple forms are supported for
456 compatibility with other linkers. They assign space to common symbols
457 even if a relocatable output file is specified (with @samp{-r}). The
458 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
459 @xref{Miscellaneous Commands}.
461 @kindex --depaudit @var{AUDITLIB}
462 @kindex -P @var{AUDITLIB}
463 @item --depaudit @var{AUDITLIB}
464 @itemx -P @var{AUDITLIB}
465 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
466 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
467 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
468 will contain a colon separated list of audit interfaces to use. This
469 option is only meaningful on ELF platforms supporting the rtld-audit interface.
470 The -P option is provided for Solaris compatibility.
472 @cindex entry point, from command line
473 @kindex -e @var{entry}
474 @kindex --entry=@var{entry}
476 @itemx --entry=@var{entry}
477 Use @var{entry} as the explicit symbol for beginning execution of your
478 program, rather than the default entry point. If there is no symbol
479 named @var{entry}, the linker will try to parse @var{entry} as a number,
480 and use that as the entry address (the number will be interpreted in
481 base 10; you may use a leading @samp{0x} for base 16, or a leading
482 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
483 and other ways of specifying the entry point.
485 @kindex --exclude-libs
486 @item --exclude-libs @var{lib},@var{lib},...
487 Specifies a list of archive libraries from which symbols should not be automatically
488 exported. The library names may be delimited by commas or colons. Specifying
489 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
490 automatic export. This option is available only for the i386 PE targeted
491 port of the linker and for ELF targeted ports. For i386 PE, symbols
492 explicitly listed in a .def file are still exported, regardless of this
493 option. For ELF targeted ports, symbols affected by this option will
494 be treated as hidden.
496 @kindex --exclude-modules-for-implib
497 @item --exclude-modules-for-implib @var{module},@var{module},...
498 Specifies a list of object files or archive members, from which symbols
499 should not be automatically exported, but which should be copied wholesale
500 into the import library being generated during the link. The module names
501 may be delimited by commas or colons, and must match exactly the filenames
502 used by @command{ld} to open the files; for archive members, this is simply
503 the member name, but for object files the name listed must include and
504 match precisely any path used to specify the input file on the linker's
505 command-line. This option is available only for the i386 PE targeted port
506 of the linker. Symbols explicitly listed in a .def file are still exported,
507 regardless of this option.
509 @cindex dynamic symbol table
511 @kindex --export-dynamic
512 @kindex --no-export-dynamic
514 @itemx --export-dynamic
515 @itemx --no-export-dynamic
516 When creating a dynamically linked executable, using the @option{-E}
517 option or the @option{--export-dynamic} option causes the linker to add
518 all symbols to the dynamic symbol table. The dynamic symbol table is the
519 set of symbols which are visible from dynamic objects at run time.
521 If you do not use either of these options (or use the
522 @option{--no-export-dynamic} option to restore the default behavior), the
523 dynamic symbol table will normally contain only those symbols which are
524 referenced by some dynamic object mentioned in the link.
526 If you use @code{dlopen} to load a dynamic object which needs to refer
527 back to the symbols defined by the program, rather than some other
528 dynamic object, then you will probably need to use this option when
529 linking the program itself.
531 You can also use the dynamic list to control what symbols should
532 be added to the dynamic symbol table if the output format supports it.
533 See the description of @samp{--dynamic-list}.
535 Note that this option is specific to ELF targeted ports. PE targets
536 support a similar function to export all symbols from a DLL or EXE; see
537 the description of @samp{--export-all-symbols} below.
539 @ifclear SingleFormat
540 @cindex big-endian objects
544 Link big-endian objects. This affects the default output format.
546 @cindex little-endian objects
549 Link little-endian objects. This affects the default output format.
552 @kindex -f @var{name}
553 @kindex --auxiliary=@var{name}
555 @itemx --auxiliary=@var{name}
556 When creating an ELF shared object, set the internal DT_AUXILIARY field
557 to the specified name. This tells the dynamic linker that the symbol
558 table of the shared object should be used as an auxiliary filter on the
559 symbol table of the shared object @var{name}.
561 If you later link a program against this filter object, then, when you
562 run the program, the dynamic linker will see the DT_AUXILIARY field. If
563 the dynamic linker resolves any symbols from the filter object, it will
564 first check whether there is a definition in the shared object
565 @var{name}. If there is one, it will be used instead of the definition
566 in the filter object. The shared object @var{name} need not exist.
567 Thus the shared object @var{name} may be used to provide an alternative
568 implementation of certain functions, perhaps for debugging or for
569 machine specific performance.
571 This option may be specified more than once. The DT_AUXILIARY entries
572 will be created in the order in which they appear on the command line.
574 @kindex -F @var{name}
575 @kindex --filter=@var{name}
577 @itemx --filter=@var{name}
578 When creating an ELF shared object, set the internal DT_FILTER field to
579 the specified name. This tells the dynamic linker that the symbol table
580 of the shared object which is being created should be used as a filter
581 on the symbol table of the shared object @var{name}.
583 If you later link a program against this filter object, then, when you
584 run the program, the dynamic linker will see the DT_FILTER field. The
585 dynamic linker will resolve symbols according to the symbol table of the
586 filter object as usual, but it will actually link to the definitions
587 found in the shared object @var{name}. Thus the filter object can be
588 used to select a subset of the symbols provided by the object
591 Some older linkers used the @option{-F} option throughout a compilation
592 toolchain for specifying object-file format for both input and output
594 @ifclear SingleFormat
595 The @sc{gnu} linker uses other mechanisms for this purpose: the
596 @option{-b}, @option{--format}, @option{--oformat} options, the
597 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
598 environment variable.
600 The @sc{gnu} linker will ignore the @option{-F} option when not
601 creating an ELF shared object.
603 @cindex finalization function
604 @kindex -fini=@var{name}
605 @item -fini=@var{name}
606 When creating an ELF executable or shared object, call NAME when the
607 executable or shared object is unloaded, by setting DT_FINI to the
608 address of the function. By default, the linker uses @code{_fini} as
609 the function to call.
613 Ignored. Provided for compatibility with other tools.
615 @kindex -G @var{value}
616 @kindex --gpsize=@var{value}
619 @itemx --gpsize=@var{value}
620 Set the maximum size of objects to be optimized using the GP register to
621 @var{size}. This is only meaningful for object file formats such as
622 MIPS ECOFF which supports putting large and small objects into different
623 sections. This is ignored for other object file formats.
625 @cindex runtime library name
626 @kindex -h @var{name}
627 @kindex -soname=@var{name}
629 @itemx -soname=@var{name}
630 When creating an ELF shared object, set the internal DT_SONAME field to
631 the specified name. When an executable is linked with a shared object
632 which has a DT_SONAME field, then when the executable is run the dynamic
633 linker will attempt to load the shared object specified by the DT_SONAME
634 field rather than the using the file name given to the linker.
637 @cindex incremental link
639 Perform an incremental link (same as option @samp{-r}).
641 @cindex initialization function
642 @kindex -init=@var{name}
643 @item -init=@var{name}
644 When creating an ELF executable or shared object, call NAME when the
645 executable or shared object is loaded, by setting DT_INIT to the address
646 of the function. By default, the linker uses @code{_init} as the
649 @cindex archive files, from cmd line
650 @kindex -l @var{namespec}
651 @kindex --library=@var{namespec}
652 @item -l @var{namespec}
653 @itemx --library=@var{namespec}
654 Add the archive or object file specified by @var{namespec} to the
655 list of files to link. This option may be used any number of times.
656 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
657 will search the library path for a file called @var{filename}, otherwise it
658 will search the library path for a file called @file{lib@var{namespec}.a}.
660 On systems which support shared libraries, @command{ld} may also search for
661 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
662 and SunOS systems, @command{ld} will search a directory for a library
663 called @file{lib@var{namespec}.so} before searching for one called
664 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
665 indicates a shared library.) Note that this behavior does not apply
666 to @file{:@var{filename}}, which always specifies a file called
669 The linker will search an archive only once, at the location where it is
670 specified on the command line. If the archive defines a symbol which
671 was undefined in some object which appeared before the archive on the
672 command line, the linker will include the appropriate file(s) from the
673 archive. However, an undefined symbol in an object appearing later on
674 the command line will not cause the linker to search the archive again.
676 See the @option{-(} option for a way to force the linker to search
677 archives multiple times.
679 You may list the same archive multiple times on the command line.
682 This type of archive searching is standard for Unix linkers. However,
683 if you are using @command{ld} on AIX, note that it is different from the
684 behaviour of the AIX linker.
687 @cindex search directory, from cmd line
689 @kindex --library-path=@var{dir}
690 @item -L @var{searchdir}
691 @itemx --library-path=@var{searchdir}
692 Add path @var{searchdir} to the list of paths that @command{ld} will search
693 for archive libraries and @command{ld} control scripts. You may use this
694 option any number of times. The directories are searched in the order
695 in which they are specified on the command line. Directories specified
696 on the command line are searched before the default directories. All
697 @option{-L} options apply to all @option{-l} options, regardless of the
698 order in which the options appear. @option{-L} options do not affect
699 how @command{ld} searches for a linker script unless @option{-T}
702 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
703 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
706 The default set of paths searched (without being specified with
707 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
708 some cases also on how it was configured. @xref{Environment}.
711 The paths can also be specified in a link script with the
712 @code{SEARCH_DIR} command. Directories specified this way are searched
713 at the point in which the linker script appears in the command line.
716 @kindex -m @var{emulation}
717 @item -m @var{emulation}
718 Emulate the @var{emulation} linker. You can list the available
719 emulations with the @samp{--verbose} or @samp{-V} options.
721 If the @samp{-m} option is not used, the emulation is taken from the
722 @code{LDEMULATION} environment variable, if that is defined.
724 Otherwise, the default emulation depends upon how the linker was
732 Print a link map to the standard output. A link map provides
733 information about the link, including the following:
737 Where object files are mapped into memory.
739 How common symbols are allocated.
741 All archive members included in the link, with a mention of the symbol
742 which caused the archive member to be brought in.
744 The values assigned to symbols.
746 Note - symbols whose values are computed by an expression which
747 involves a reference to a previous value of the same symbol may not
748 have correct result displayed in the link map. This is because the
749 linker discards intermediate results and only retains the final value
750 of an expression. Under such circumstances the linker will display
751 the final value enclosed by square brackets. Thus for example a
752 linker script containing:
760 will produce the following output in the link map if the @option{-M}
765 [0x0000000c] foo = (foo * 0x4)
766 [0x0000000c] foo = (foo + 0x8)
769 See @ref{Expressions} for more information about expressions in linker
774 @cindex read-only text
779 Turn off page alignment of sections, and disable linking against shared
780 libraries. If the output format supports Unix style magic numbers,
781 mark the output as @code{NMAGIC}.
785 @cindex read/write from cmd line
789 Set the text and data sections to be readable and writable. Also, do
790 not page-align the data segment, and disable linking against shared
791 libraries. If the output format supports Unix style magic numbers,
792 mark the output as @code{OMAGIC}. Note: Although a writable text section
793 is allowed for PE-COFF targets, it does not conform to the format
794 specification published by Microsoft.
799 This option negates most of the effects of the @option{-N} option. It
800 sets the text section to be read-only, and forces the data segment to
801 be page-aligned. Note - this option does not enable linking against
802 shared libraries. Use @option{-Bdynamic} for this.
804 @kindex -o @var{output}
805 @kindex --output=@var{output}
806 @cindex naming the output file
807 @item -o @var{output}
808 @itemx --output=@var{output}
809 Use @var{output} as the name for the program produced by @command{ld}; if this
810 option is not specified, the name @file{a.out} is used by default. The
811 script command @code{OUTPUT} can also specify the output file name.
813 @kindex -O @var{level}
814 @cindex generating optimized output
816 If @var{level} is a numeric values greater than zero @command{ld} optimizes
817 the output. This might take significantly longer and therefore probably
818 should only be enabled for the final binary. At the moment this
819 option only affects ELF shared library generation. Future releases of
820 the linker may make more use of this option. Also currently there is
821 no difference in the linker's behaviour for different non-zero values
822 of this option. Again this may change with future releases.
825 @kindex --emit-relocs
826 @cindex retain relocations in final executable
829 Leave relocation sections and contents in fully linked executables.
830 Post link analysis and optimization tools may need this information in
831 order to perform correct modifications of executables. This results
832 in larger executables.
834 This option is currently only supported on ELF platforms.
836 @kindex --force-dynamic
837 @cindex forcing the creation of dynamic sections
838 @item --force-dynamic
839 Force the output file to have dynamic sections. This option is specific
843 @cindex relocatable output
845 @kindex --relocatable
848 Generate relocatable output---i.e., generate an output file that can in
849 turn serve as input to @command{ld}. This is often called @dfn{partial
850 linking}. As a side effect, in environments that support standard Unix
851 magic numbers, this option also sets the output file's magic number to
853 @c ; see @option{-N}.
854 If this option is not specified, an absolute file is produced. When
855 linking C++ programs, this option @emph{will not} resolve references to
856 constructors; to do that, use @samp{-Ur}.
858 When an input file does not have the same format as the output file,
859 partial linking is only supported if that input file does not contain any
860 relocations. Different output formats can have further restrictions; for
861 example some @code{a.out}-based formats do not support partial linking
862 with input files in other formats at all.
864 This option does the same thing as @samp{-i}.
866 @kindex -R @var{file}
867 @kindex --just-symbols=@var{file}
868 @cindex symbol-only input
869 @item -R @var{filename}
870 @itemx --just-symbols=@var{filename}
871 Read symbol names and their addresses from @var{filename}, but do not
872 relocate it or include it in the output. This allows your output file
873 to refer symbolically to absolute locations of memory defined in other
874 programs. You may use this option more than once.
876 For compatibility with other ELF linkers, if the @option{-R} option is
877 followed by a directory name, rather than a file name, it is treated as
878 the @option{-rpath} option.
882 @cindex strip all symbols
885 Omit all symbol information from the output file.
888 @kindex --strip-debug
889 @cindex strip debugger symbols
892 Omit debugger symbol information (but not all symbols) from the output file.
896 @cindex input files, displaying
899 Print the names of the input files as @command{ld} processes them.
901 @kindex -T @var{script}
902 @kindex --script=@var{script}
904 @item -T @var{scriptfile}
905 @itemx --script=@var{scriptfile}
906 Use @var{scriptfile} as the linker script. This script replaces
907 @command{ld}'s default linker script (rather than adding to it), so
908 @var{commandfile} must specify everything necessary to describe the
909 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
910 the current directory, @code{ld} looks for it in the directories
911 specified by any preceding @samp{-L} options. Multiple @samp{-T}
914 @kindex -dT @var{script}
915 @kindex --default-script=@var{script}
917 @item -dT @var{scriptfile}
918 @itemx --default-script=@var{scriptfile}
919 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
921 This option is similar to the @option{--script} option except that
922 processing of the script is delayed until after the rest of the
923 command line has been processed. This allows options placed after the
924 @option{--default-script} option on the command line to affect the
925 behaviour of the linker script, which can be important when the linker
926 command line cannot be directly controlled by the user. (eg because
927 the command line is being constructed by another tool, such as
930 @kindex -u @var{symbol}
931 @kindex --undefined=@var{symbol}
932 @cindex undefined symbol
933 @item -u @var{symbol}
934 @itemx --undefined=@var{symbol}
935 Force @var{symbol} to be entered in the output file as an undefined
936 symbol. Doing this may, for example, trigger linking of additional
937 modules from standard libraries. @samp{-u} may be repeated with
938 different option arguments to enter additional undefined symbols. This
939 option is equivalent to the @code{EXTERN} linker script command.
944 For anything other than C++ programs, this option is equivalent to
945 @samp{-r}: it generates relocatable output---i.e., an output file that can in
946 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
947 @emph{does} resolve references to constructors, unlike @samp{-r}.
948 It does not work to use @samp{-Ur} on files that were themselves linked
949 with @samp{-Ur}; once the constructor table has been built, it cannot
950 be added to. Use @samp{-Ur} only for the last partial link, and
951 @samp{-r} for the others.
953 @kindex --unique[=@var{SECTION}]
954 @item --unique[=@var{SECTION}]
955 Creates a separate output section for every input section matching
956 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
957 missing, for every orphan input section. An orphan section is one not
958 specifically mentioned in a linker script. You may use this option
959 multiple times on the command line; It prevents the normal merging of
960 input sections with the same name, overriding output section assignments
970 Display the version number for @command{ld}. The @option{-V} option also
971 lists the supported emulations.
974 @kindex --discard-all
975 @cindex deleting local symbols
978 Delete all local symbols.
981 @kindex --discard-locals
982 @cindex local symbols, deleting
984 @itemx --discard-locals
985 Delete all temporary local symbols. (These symbols start with
986 system-specific local label prefixes, typically @samp{.L} for ELF systems
987 or @samp{L} for traditional a.out systems.)
989 @kindex -y @var{symbol}
990 @kindex --trace-symbol=@var{symbol}
991 @cindex symbol tracing
992 @item -y @var{symbol}
993 @itemx --trace-symbol=@var{symbol}
994 Print the name of each linked file in which @var{symbol} appears. This
995 option may be given any number of times. On many systems it is necessary
996 to prepend an underscore.
998 This option is useful when you have an undefined symbol in your link but
999 don't know where the reference is coming from.
1001 @kindex -Y @var{path}
1003 Add @var{path} to the default library search path. This option exists
1004 for Solaris compatibility.
1006 @kindex -z @var{keyword}
1007 @item -z @var{keyword}
1008 The recognized keywords are:
1012 Combines multiple reloc sections and sorts them to make dynamic symbol
1013 lookup caching possible.
1016 Disallows undefined symbols in object files. Undefined symbols in
1017 shared libraries are still allowed.
1020 Marks the object as requiring executable stack.
1023 This option is only meaningful when building a shared object. It makes
1024 the symbols defined by this shared object available for symbol resolution
1025 of subsequently loaded libraries.
1028 This option is only meaningful when building a shared object.
1029 It marks the object so that its runtime initialization will occur
1030 before the runtime initialization of any other objects brought into
1031 the process at the same time. Similarly the runtime finalization of
1032 the object will occur after the runtime finalization of any other
1036 Marks the object that its symbol table interposes before all symbols
1037 but the primary executable.
1040 When generating an executable or shared library, mark it to tell the
1041 dynamic linker to defer function call resolution to the point when
1042 the function is called (lazy binding), rather than at load time.
1043 Lazy binding is the default.
1046 Marks the object that its filters be processed immediately at
1050 Allows multiple definitions.
1053 Disables multiple reloc sections combining.
1056 Disables production of copy relocs.
1059 Marks the object that the search for dependencies of this object will
1060 ignore any default library search paths.
1063 Marks the object shouldn't be unloaded at runtime.
1066 Marks the object not available to @code{dlopen}.
1069 Marks the object can not be dumped by @code{dldump}.
1072 Marks the object as not requiring executable stack.
1075 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1078 When generating an executable or shared library, mark it to tell the
1079 dynamic linker to resolve all symbols when the program is started, or
1080 when the shared library is linked to using dlopen, instead of
1081 deferring function call resolution to the point when the function is
1085 Marks the object may contain $ORIGIN.
1088 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1090 @item max-page-size=@var{value}
1091 Set the emulation maximum page size to @var{value}.
1093 @item common-page-size=@var{value}
1094 Set the emulation common page size to @var{value}.
1096 @item stack-size=@var{value}
1097 Specify a stack size for in an ELF @code{PT_GNU_STACK} segment.
1098 Specifying zero will override any default non-zero sized
1099 @code{PT_GNU_STACK} segment creation.
1103 Other keywords are ignored for Solaris compatibility.
1106 @cindex groups of archives
1107 @item -( @var{archives} -)
1108 @itemx --start-group @var{archives} --end-group
1109 The @var{archives} should be a list of archive files. They may be
1110 either explicit file names, or @samp{-l} options.
1112 The specified archives are searched repeatedly until no new undefined
1113 references are created. Normally, an archive is searched only once in
1114 the order that it is specified on the command line. If a symbol in that
1115 archive is needed to resolve an undefined symbol referred to by an
1116 object in an archive that appears later on the command line, the linker
1117 would not be able to resolve that reference. By grouping the archives,
1118 they all be searched repeatedly until all possible references are
1121 Using this option has a significant performance cost. It is best to use
1122 it only when there are unavoidable circular references between two or
1125 @kindex --accept-unknown-input-arch
1126 @kindex --no-accept-unknown-input-arch
1127 @item --accept-unknown-input-arch
1128 @itemx --no-accept-unknown-input-arch
1129 Tells the linker to accept input files whose architecture cannot be
1130 recognised. The assumption is that the user knows what they are doing
1131 and deliberately wants to link in these unknown input files. This was
1132 the default behaviour of the linker, before release 2.14. The default
1133 behaviour from release 2.14 onwards is to reject such input files, and
1134 so the @samp{--accept-unknown-input-arch} option has been added to
1135 restore the old behaviour.
1138 @kindex --no-as-needed
1140 @itemx --no-as-needed
1141 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1142 on the command line after the @option{--as-needed} option. Normally
1143 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1144 on the command line, regardless of whether the library is actually
1145 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1146 emitted for a library that @emph{at that point in the link} satisfies a
1147 non-weak undefined symbol reference from a regular object file or, if
1148 the library is not found in the DT_NEEDED lists of other libraries, a
1149 non-weak undefined symbol reference from another dynamic library.
1150 Object files or libraries appearing on the command line @emph{after}
1151 the library in question do not affect whether the library is seen as
1152 needed. This is similar to the rules for extraction of object files
1153 from archives. @option{--no-as-needed} restores the default behaviour.
1155 @kindex --add-needed
1156 @kindex --no-add-needed
1158 @itemx --no-add-needed
1159 These two options have been deprecated because of the similarity of
1160 their names to the @option{--as-needed} and @option{--no-as-needed}
1161 options. They have been replaced by @option{--copy-dt-needed-entries}
1162 and @option{--no-copy-dt-needed-entries}.
1164 @kindex -assert @var{keyword}
1165 @item -assert @var{keyword}
1166 This option is ignored for SunOS compatibility.
1170 @kindex -call_shared
1174 Link against dynamic libraries. This is only meaningful on platforms
1175 for which shared libraries are supported. This option is normally the
1176 default on such platforms. The different variants of this option are
1177 for compatibility with various systems. You may use this option
1178 multiple times on the command line: it affects library searching for
1179 @option{-l} options which follow it.
1183 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1184 section. This causes the runtime linker to handle lookups in this
1185 object and its dependencies to be performed only inside the group.
1186 @option{--unresolved-symbols=report-all} is implied. This option is
1187 only meaningful on ELF platforms which support shared libraries.
1197 Do not link against shared libraries. This is only meaningful on
1198 platforms for which shared libraries are supported. The different
1199 variants of this option are for compatibility with various systems. You
1200 may use this option multiple times on the command line: it affects
1201 library searching for @option{-l} options which follow it. This
1202 option also implies @option{--unresolved-symbols=report-all}. This
1203 option can be used with @option{-shared}. Doing so means that a
1204 shared library is being created but that all of the library's external
1205 references must be resolved by pulling in entries from static
1210 When creating a shared library, bind references to global symbols to the
1211 definition within the shared library, if any. Normally, it is possible
1212 for a program linked against a shared library to override the definition
1213 within the shared library. This option is only meaningful on ELF
1214 platforms which support shared libraries.
1216 @kindex -Bsymbolic-functions
1217 @item -Bsymbolic-functions
1218 When creating a shared library, bind references to global function
1219 symbols to the definition within the shared library, if any.
1220 This option is only meaningful on ELF platforms which support shared
1223 @kindex --dynamic-list=@var{dynamic-list-file}
1224 @item --dynamic-list=@var{dynamic-list-file}
1225 Specify the name of a dynamic list file to the linker. This is
1226 typically used when creating shared libraries to specify a list of
1227 global symbols whose references shouldn't be bound to the definition
1228 within the shared library, or creating dynamically linked executables
1229 to specify a list of symbols which should be added to the symbol table
1230 in the executable. This option is only meaningful on ELF platforms
1231 which support shared libraries.
1233 The format of the dynamic list is the same as the version node without
1234 scope and node name. See @ref{VERSION} for more information.
1236 @kindex --dynamic-list-data
1237 @item --dynamic-list-data
1238 Include all global data symbols to the dynamic list.
1240 @kindex --dynamic-list-cpp-new
1241 @item --dynamic-list-cpp-new
1242 Provide the builtin dynamic list for C++ operator new and delete. It
1243 is mainly useful for building shared libstdc++.
1245 @kindex --dynamic-list-cpp-typeinfo
1246 @item --dynamic-list-cpp-typeinfo
1247 Provide the builtin dynamic list for C++ runtime type identification.
1249 @kindex --check-sections
1250 @kindex --no-check-sections
1251 @item --check-sections
1252 @itemx --no-check-sections
1253 Asks the linker @emph{not} to check section addresses after they have
1254 been assigned to see if there are any overlaps. Normally the linker will
1255 perform this check, and if it finds any overlaps it will produce
1256 suitable error messages. The linker does know about, and does make
1257 allowances for sections in overlays. The default behaviour can be
1258 restored by using the command line switch @option{--check-sections}.
1259 Section overlap is not usually checked for relocatable links. You can
1260 force checking in that case by using the @option{--check-sections}
1263 @kindex --copy-dt-needed-entries
1264 @kindex --no-copy-dt-needed-entries
1265 @item --copy-dt-needed-entries
1266 @itemx --no-copy-dt-needed-entries
1267 This option affects the treatment of dynamic libraries referred to
1268 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1269 command line. Normally the linker won't add a DT_NEEDED tag to the
1270 output binary for each library mentioned in a DT_NEEDED tag in an
1271 input dynamic library. With @option{--copy-dt-needed-entries}
1272 specified on the command line however any dynamic libraries that
1273 follow it will have their DT_NEEDED entries added. The default
1274 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1276 This option also has an effect on the resolution of symbols in dynamic
1277 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1278 mentioned on the command line will be recursively searched, following
1279 their DT_NEEDED tags to other libraries, in order to resolve symbols
1280 required by the output binary. With the default setting however
1281 the searching of dynamic libraries that follow it will stop with the
1282 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1285 @cindex cross reference table
1288 Output a cross reference table. If a linker map file is being
1289 generated, the cross reference table is printed to the map file.
1290 Otherwise, it is printed on the standard output.
1292 The format of the table is intentionally simple, so that it may be
1293 easily processed by a script if necessary. The symbols are printed out,
1294 sorted by name. For each symbol, a list of file names is given. If the
1295 symbol is defined, the first file listed is the location of the
1296 definition. The remaining files contain references to the symbol.
1298 @cindex common allocation
1299 @kindex --no-define-common
1300 @item --no-define-common
1301 This option inhibits the assignment of addresses to common symbols.
1302 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1303 @xref{Miscellaneous Commands}.
1305 The @samp{--no-define-common} option allows decoupling
1306 the decision to assign addresses to Common symbols from the choice
1307 of the output file type; otherwise a non-Relocatable output type
1308 forces assigning addresses to Common symbols.
1309 Using @samp{--no-define-common} allows Common symbols that are referenced
1310 from a shared library to be assigned addresses only in the main program.
1311 This eliminates the unused duplicate space in the shared library,
1312 and also prevents any possible confusion over resolving to the wrong
1313 duplicate when there are many dynamic modules with specialized search
1314 paths for runtime symbol resolution.
1316 @cindex symbols, from command line
1317 @kindex --defsym=@var{symbol}=@var{exp}
1318 @item --defsym=@var{symbol}=@var{expression}
1319 Create a global symbol in the output file, containing the absolute
1320 address given by @var{expression}. You may use this option as many
1321 times as necessary to define multiple symbols in the command line. A
1322 limited form of arithmetic is supported for the @var{expression} in this
1323 context: you may give a hexadecimal constant or the name of an existing
1324 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1325 constants or symbols. If you need more elaborate expressions, consider
1326 using the linker command language from a script (@pxref{Assignments,,
1327 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1328 space between @var{symbol}, the equals sign (``@key{=}''), and
1331 @cindex demangling, from command line
1332 @kindex --demangle[=@var{style}]
1333 @kindex --no-demangle
1334 @item --demangle[=@var{style}]
1335 @itemx --no-demangle
1336 These options control whether to demangle symbol names in error messages
1337 and other output. When the linker is told to demangle, it tries to
1338 present symbol names in a readable fashion: it strips leading
1339 underscores if they are used by the object file format, and converts C++
1340 mangled symbol names into user readable names. Different compilers have
1341 different mangling styles. The optional demangling style argument can be used
1342 to choose an appropriate demangling style for your compiler. The linker will
1343 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1344 is set. These options may be used to override the default.
1346 @cindex dynamic linker, from command line
1347 @kindex -I@var{file}
1348 @kindex --dynamic-linker=@var{file}
1350 @itemx --dynamic-linker=@var{file}
1351 Set the name of the dynamic linker. This is only meaningful when
1352 generating dynamically linked ELF executables. The default dynamic
1353 linker is normally correct; don't use this unless you know what you are
1356 @kindex --fatal-warnings
1357 @kindex --no-fatal-warnings
1358 @item --fatal-warnings
1359 @itemx --no-fatal-warnings
1360 Treat all warnings as errors. The default behaviour can be restored
1361 with the option @option{--no-fatal-warnings}.
1363 @kindex --force-exe-suffix
1364 @item --force-exe-suffix
1365 Make sure that an output file has a .exe suffix.
1367 If a successfully built fully linked output file does not have a
1368 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1369 the output file to one of the same name with a @code{.exe} suffix. This
1370 option is useful when using unmodified Unix makefiles on a Microsoft
1371 Windows host, since some versions of Windows won't run an image unless
1372 it ends in a @code{.exe} suffix.
1374 @kindex --gc-sections
1375 @kindex --no-gc-sections
1376 @cindex garbage collection
1378 @itemx --no-gc-sections
1379 Enable garbage collection of unused input sections. It is ignored on
1380 targets that do not support this option. The default behaviour (of not
1381 performing this garbage collection) can be restored by specifying
1382 @samp{--no-gc-sections} on the command line.
1384 @samp{--gc-sections} decides which input sections are used by
1385 examining symbols and relocations. The section containing the entry
1386 symbol and all sections containing symbols undefined on the
1387 command-line will be kept, as will sections containing symbols
1388 referenced by dynamic objects. Note that when building shared
1389 libraries, the linker must assume that any visible symbol is
1390 referenced. Once this initial set of sections has been determined,
1391 the linker recursively marks as used any section referenced by their
1392 relocations. See @samp{--entry} and @samp{--undefined}.
1394 This option can be set when doing a partial link (enabled with option
1395 @samp{-r}). In this case the root of symbols kept must be explicitly
1396 specified either by an @samp{--entry} or @samp{--undefined} option or by
1397 a @code{ENTRY} command in the linker script.
1399 @kindex --print-gc-sections
1400 @kindex --no-print-gc-sections
1401 @cindex garbage collection
1402 @item --print-gc-sections
1403 @itemx --no-print-gc-sections
1404 List all sections removed by garbage collection. The listing is
1405 printed on stderr. This option is only effective if garbage
1406 collection has been enabled via the @samp{--gc-sections}) option. The
1407 default behaviour (of not listing the sections that are removed) can
1408 be restored by specifying @samp{--no-print-gc-sections} on the command
1411 @kindex --print-output-format
1412 @cindex output format
1413 @item --print-output-format
1414 Print the name of the default output format (perhaps influenced by
1415 other command-line options). This is the string that would appear
1416 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1422 Print a summary of the command-line options on the standard output and exit.
1424 @kindex --target-help
1426 Print a summary of all target specific options on the standard output and exit.
1428 @kindex -Map=@var{mapfile}
1429 @item -Map=@var{mapfile}
1430 Print a link map to the file @var{mapfile}. See the description of the
1431 @option{-M} option, above.
1433 @cindex memory usage
1434 @kindex --no-keep-memory
1435 @item --no-keep-memory
1436 @command{ld} normally optimizes for speed over memory usage by caching the
1437 symbol tables of input files in memory. This option tells @command{ld} to
1438 instead optimize for memory usage, by rereading the symbol tables as
1439 necessary. This may be required if @command{ld} runs out of memory space
1440 while linking a large executable.
1442 @kindex --no-undefined
1444 @item --no-undefined
1446 Report unresolved symbol references from regular object files. This
1447 is done even if the linker is creating a non-symbolic shared library.
1448 The switch @option{--[no-]allow-shlib-undefined} controls the
1449 behaviour for reporting unresolved references found in shared
1450 libraries being linked in.
1452 @kindex --allow-multiple-definition
1454 @item --allow-multiple-definition
1456 Normally when a symbol is defined multiple times, the linker will
1457 report a fatal error. These options allow multiple definitions and the
1458 first definition will be used.
1460 @kindex --allow-shlib-undefined
1461 @kindex --no-allow-shlib-undefined
1462 @item --allow-shlib-undefined
1463 @itemx --no-allow-shlib-undefined
1464 Allows or disallows undefined symbols in shared libraries.
1465 This switch is similar to @option{--no-undefined} except that it
1466 determines the behaviour when the undefined symbols are in a
1467 shared library rather than a regular object file. It does not affect
1468 how undefined symbols in regular object files are handled.
1470 The default behaviour is to report errors for any undefined symbols
1471 referenced in shared libraries if the linker is being used to create
1472 an executable, but to allow them if the linker is being used to create
1475 The reasons for allowing undefined symbol references in shared
1476 libraries specified at link time are that:
1480 A shared library specified at link time may not be the same as the one
1481 that is available at load time, so the symbol might actually be
1482 resolvable at load time.
1484 There are some operating systems, eg BeOS and HPPA, where undefined
1485 symbols in shared libraries are normal.
1487 The BeOS kernel for example patches shared libraries at load time to
1488 select whichever function is most appropriate for the current
1489 architecture. This is used, for example, to dynamically select an
1490 appropriate memset function.
1493 @kindex --no-undefined-version
1494 @item --no-undefined-version
1495 Normally when a symbol has an undefined version, the linker will ignore
1496 it. This option disallows symbols with undefined version and a fatal error
1497 will be issued instead.
1499 @kindex --default-symver
1500 @item --default-symver
1501 Create and use a default symbol version (the soname) for unversioned
1504 @kindex --default-imported-symver
1505 @item --default-imported-symver
1506 Create and use a default symbol version (the soname) for unversioned
1509 @kindex --no-warn-mismatch
1510 @item --no-warn-mismatch
1511 Normally @command{ld} will give an error if you try to link together input
1512 files that are mismatched for some reason, perhaps because they have
1513 been compiled for different processors or for different endiannesses.
1514 This option tells @command{ld} that it should silently permit such possible
1515 errors. This option should only be used with care, in cases when you
1516 have taken some special action that ensures that the linker errors are
1519 @kindex --no-warn-search-mismatch
1520 @item --no-warn-search-mismatch
1521 Normally @command{ld} will give a warning if it finds an incompatible
1522 library during a library search. This option silences the warning.
1524 @kindex --no-whole-archive
1525 @item --no-whole-archive
1526 Turn off the effect of the @option{--whole-archive} option for subsequent
1529 @cindex output file after errors
1530 @kindex --noinhibit-exec
1531 @item --noinhibit-exec
1532 Retain the executable output file whenever it is still usable.
1533 Normally, the linker will not produce an output file if it encounters
1534 errors during the link process; it exits without writing an output file
1535 when it issues any error whatsoever.
1539 Only search library directories explicitly specified on the
1540 command line. Library directories specified in linker scripts
1541 (including linker scripts specified on the command line) are ignored.
1543 @ifclear SingleFormat
1544 @kindex --oformat=@var{output-format}
1545 @item --oformat=@var{output-format}
1546 @command{ld} may be configured to support more than one kind of object
1547 file. If your @command{ld} is configured this way, you can use the
1548 @samp{--oformat} option to specify the binary format for the output
1549 object file. Even when @command{ld} is configured to support alternative
1550 object formats, you don't usually need to specify this, as @command{ld}
1551 should be configured to produce as a default output format the most
1552 usual format on each machine. @var{output-format} is a text string, the
1553 name of a particular format supported by the BFD libraries. (You can
1554 list the available binary formats with @samp{objdump -i}.) The script
1555 command @code{OUTPUT_FORMAT} can also specify the output format, but
1556 this option overrides it. @xref{BFD}.
1560 @kindex --pic-executable
1562 @itemx --pic-executable
1563 @cindex position independent executables
1564 Create a position independent executable. This is currently only supported on
1565 ELF platforms. Position independent executables are similar to shared
1566 libraries in that they are relocated by the dynamic linker to the virtual
1567 address the OS chooses for them (which can vary between invocations). Like
1568 normal dynamically linked executables they can be executed and symbols
1569 defined in the executable cannot be overridden by shared libraries.
1573 This option is ignored for Linux compatibility.
1577 This option is ignored for SVR4 compatibility.
1580 @cindex synthesizing linker
1581 @cindex relaxing addressing modes
1585 An option with machine dependent effects.
1587 This option is only supported on a few targets.
1590 @xref{H8/300,,@command{ld} and the H8/300}.
1593 @xref{i960,, @command{ld} and the Intel 960 family}.
1596 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1599 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1602 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1605 On some platforms the @samp{--relax} option performs target specific,
1606 global optimizations that become possible when the linker resolves
1607 addressing in the program, such as relaxing address modes,
1608 synthesizing new instructions, selecting shorter version of current
1609 instructions, and combining constant values.
1611 On some platforms these link time global optimizations may make symbolic
1612 debugging of the resulting executable impossible.
1614 This is known to be the case for the Matsushita MN10200 and MN10300
1615 family of processors.
1619 On platforms where this is not supported, @samp{--relax} is accepted,
1623 On platforms where @samp{--relax} is accepted the option
1624 @samp{--no-relax} can be used to disable the feature.
1626 @cindex retaining specified symbols
1627 @cindex stripping all but some symbols
1628 @cindex symbols, retaining selectively
1629 @kindex --retain-symbols-file=@var{filename}
1630 @item --retain-symbols-file=@var{filename}
1631 Retain @emph{only} the symbols listed in the file @var{filename},
1632 discarding all others. @var{filename} is simply a flat file, with one
1633 symbol name per line. This option is especially useful in environments
1637 where a large global symbol table is accumulated gradually, to conserve
1640 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1641 or symbols needed for relocations.
1643 You may only specify @samp{--retain-symbols-file} once in the command
1644 line. It overrides @samp{-s} and @samp{-S}.
1647 @item -rpath=@var{dir}
1648 @cindex runtime library search path
1649 @kindex -rpath=@var{dir}
1650 Add a directory to the runtime library search path. This is used when
1651 linking an ELF executable with shared objects. All @option{-rpath}
1652 arguments are concatenated and passed to the runtime linker, which uses
1653 them to locate shared objects at runtime. The @option{-rpath} option is
1654 also used when locating shared objects which are needed by shared
1655 objects explicitly included in the link; see the description of the
1656 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1657 ELF executable, the contents of the environment variable
1658 @code{LD_RUN_PATH} will be used if it is defined.
1660 The @option{-rpath} option may also be used on SunOS. By default, on
1661 SunOS, the linker will form a runtime search patch out of all the
1662 @option{-L} options it is given. If a @option{-rpath} option is used, the
1663 runtime search path will be formed exclusively using the @option{-rpath}
1664 options, ignoring the @option{-L} options. This can be useful when using
1665 gcc, which adds many @option{-L} options which may be on NFS mounted
1668 For compatibility with other ELF linkers, if the @option{-R} option is
1669 followed by a directory name, rather than a file name, it is treated as
1670 the @option{-rpath} option.
1674 @cindex link-time runtime library search path
1675 @kindex -rpath-link=@var{dir}
1676 @item -rpath-link=@var{dir}
1677 When using ELF or SunOS, one shared library may require another. This
1678 happens when an @code{ld -shared} link includes a shared library as one
1681 When the linker encounters such a dependency when doing a non-shared,
1682 non-relocatable link, it will automatically try to locate the required
1683 shared library and include it in the link, if it is not included
1684 explicitly. In such a case, the @option{-rpath-link} option
1685 specifies the first set of directories to search. The
1686 @option{-rpath-link} option may specify a sequence of directory names
1687 either by specifying a list of names separated by colons, or by
1688 appearing multiple times.
1690 This option should be used with caution as it overrides the search path
1691 that may have been hard compiled into a shared library. In such a case it
1692 is possible to use unintentionally a different search path than the
1693 runtime linker would do.
1695 The linker uses the following search paths to locate required shared
1699 Any directories specified by @option{-rpath-link} options.
1701 Any directories specified by @option{-rpath} options. The difference
1702 between @option{-rpath} and @option{-rpath-link} is that directories
1703 specified by @option{-rpath} options are included in the executable and
1704 used at runtime, whereas the @option{-rpath-link} option is only effective
1705 at link time. Searching @option{-rpath} in this way is only supported
1706 by native linkers and cross linkers which have been configured with
1707 the @option{--with-sysroot} option.
1709 On an ELF system, for native linkers, if the @option{-rpath} and
1710 @option{-rpath-link} options were not used, search the contents of the
1711 environment variable @code{LD_RUN_PATH}.
1713 On SunOS, if the @option{-rpath} option was not used, search any
1714 directories specified using @option{-L} options.
1716 For a native linker, search the contents of the environment
1717 variable @code{LD_LIBRARY_PATH}.
1719 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1720 @code{DT_RPATH} of a shared library are searched for shared
1721 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1722 @code{DT_RUNPATH} entries exist.
1724 The default directories, normally @file{/lib} and @file{/usr/lib}.
1726 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1727 exists, the list of directories found in that file.
1730 If the required shared library is not found, the linker will issue a
1731 warning and continue with the link.
1738 @cindex shared libraries
1739 Create a shared library. This is currently only supported on ELF, XCOFF
1740 and SunOS platforms. On SunOS, the linker will automatically create a
1741 shared library if the @option{-e} option is not used and there are
1742 undefined symbols in the link.
1744 @kindex --sort-common
1746 @itemx --sort-common=ascending
1747 @itemx --sort-common=descending
1748 This option tells @command{ld} to sort the common symbols by alignment in
1749 ascending or descending order when it places them in the appropriate output
1750 sections. The symbol alignments considered are sixteen-byte or larger,
1751 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1752 between symbols due to alignment constraints. If no sorting order is
1753 specified, then descending order is assumed.
1755 @kindex --sort-section=name
1756 @item --sort-section=name
1757 This option will apply @code{SORT_BY_NAME} to all wildcard section
1758 patterns in the linker script.
1760 @kindex --sort-section=alignment
1761 @item --sort-section=alignment
1762 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1763 patterns in the linker script.
1765 @kindex --split-by-file
1766 @item --split-by-file[=@var{size}]
1767 Similar to @option{--split-by-reloc} but creates a new output section for
1768 each input file when @var{size} is reached. @var{size} defaults to a
1769 size of 1 if not given.
1771 @kindex --split-by-reloc
1772 @item --split-by-reloc[=@var{count}]
1773 Tries to creates extra sections in the output file so that no single
1774 output section in the file contains more than @var{count} relocations.
1775 This is useful when generating huge relocatable files for downloading into
1776 certain real time kernels with the COFF object file format; since COFF
1777 cannot represent more than 65535 relocations in a single section. Note
1778 that this will fail to work with object file formats which do not
1779 support arbitrary sections. The linker will not split up individual
1780 input sections for redistribution, so if a single input section contains
1781 more than @var{count} relocations one output section will contain that
1782 many relocations. @var{count} defaults to a value of 32768.
1786 Compute and display statistics about the operation of the linker, such
1787 as execution time and memory usage.
1789 @kindex --sysroot=@var{directory}
1790 @item --sysroot=@var{directory}
1791 Use @var{directory} as the location of the sysroot, overriding the
1792 configure-time default. This option is only supported by linkers
1793 that were configured using @option{--with-sysroot}.
1795 @kindex --traditional-format
1796 @cindex traditional format
1797 @item --traditional-format
1798 For some targets, the output of @command{ld} is different in some ways from
1799 the output of some existing linker. This switch requests @command{ld} to
1800 use the traditional format instead.
1803 For example, on SunOS, @command{ld} combines duplicate entries in the
1804 symbol string table. This can reduce the size of an output file with
1805 full debugging information by over 30 percent. Unfortunately, the SunOS
1806 @code{dbx} program can not read the resulting program (@code{gdb} has no
1807 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1808 combine duplicate entries.
1810 @kindex --section-start=@var{sectionname}=@var{org}
1811 @item --section-start=@var{sectionname}=@var{org}
1812 Locate a section in the output file at the absolute
1813 address given by @var{org}. You may use this option as many
1814 times as necessary to locate multiple sections in the command
1816 @var{org} must be a single hexadecimal integer;
1817 for compatibility with other linkers, you may omit the leading
1818 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1819 should be no white space between @var{sectionname}, the equals
1820 sign (``@key{=}''), and @var{org}.
1822 @kindex -Tbss=@var{org}
1823 @kindex -Tdata=@var{org}
1824 @kindex -Ttext=@var{org}
1825 @cindex segment origins, cmd line
1826 @item -Tbss=@var{org}
1827 @itemx -Tdata=@var{org}
1828 @itemx -Ttext=@var{org}
1829 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
1830 @code{.text} as the @var{sectionname}.
1832 @kindex -Ttext-segment=@var{org}
1833 @item -Ttext-segment=@var{org}
1834 @cindex text segment origin, cmd line
1835 When creating an ELF executable or shared object, it will set the address
1836 of the first byte of the text segment.
1838 @kindex -Trodata-segment=@var{org}
1839 @item -Trodata-segment=@var{org}
1840 @cindex rodata segment origin, cmd line
1841 When creating an ELF executable or shared object for a target where
1842 the read-only data is in its own segment separate from the executable
1843 text, it will set the address of the first byte of the read-only data segment.
1845 @kindex -Tldata-segment=@var{org}
1846 @item -Tldata-segment=@var{org}
1847 @cindex ldata segment origin, cmd line
1848 When creating an ELF executable or shared object for x86-64 medium memory
1849 model, it will set the address of the first byte of the ldata segment.
1851 @kindex --unresolved-symbols
1852 @item --unresolved-symbols=@var{method}
1853 Determine how to handle unresolved symbols. There are four possible
1854 values for @samp{method}:
1858 Do not report any unresolved symbols.
1861 Report all unresolved symbols. This is the default.
1863 @item ignore-in-object-files
1864 Report unresolved symbols that are contained in shared libraries, but
1865 ignore them if they come from regular object files.
1867 @item ignore-in-shared-libs
1868 Report unresolved symbols that come from regular object files, but
1869 ignore them if they come from shared libraries. This can be useful
1870 when creating a dynamic binary and it is known that all the shared
1871 libraries that it should be referencing are included on the linker's
1875 The behaviour for shared libraries on their own can also be controlled
1876 by the @option{--[no-]allow-shlib-undefined} option.
1878 Normally the linker will generate an error message for each reported
1879 unresolved symbol but the option @option{--warn-unresolved-symbols}
1880 can change this to a warning.
1882 @kindex --verbose[=@var{NUMBER}]
1883 @cindex verbose[=@var{NUMBER}]
1885 @itemx --verbose[=@var{NUMBER}]
1886 Display the version number for @command{ld} and list the linker emulations
1887 supported. Display which input files can and cannot be opened. Display
1888 the linker script being used by the linker. If the optional @var{NUMBER}
1889 argument > 1, plugin symbol status will also be displayed.
1891 @kindex --version-script=@var{version-scriptfile}
1892 @cindex version script, symbol versions
1893 @item --version-script=@var{version-scriptfile}
1894 Specify the name of a version script to the linker. This is typically
1895 used when creating shared libraries to specify additional information
1896 about the version hierarchy for the library being created. This option
1897 is only fully supported on ELF platforms which support shared libraries;
1898 see @ref{VERSION}. It is partially supported on PE platforms, which can
1899 use version scripts to filter symbol visibility in auto-export mode: any
1900 symbols marked @samp{local} in the version script will not be exported.
1903 @kindex --warn-common
1904 @cindex warnings, on combining symbols
1905 @cindex combining symbols, warnings on
1907 Warn when a common symbol is combined with another common symbol or with
1908 a symbol definition. Unix linkers allow this somewhat sloppy practice,
1909 but linkers on some other operating systems do not. This option allows
1910 you to find potential problems from combining global symbols.
1911 Unfortunately, some C libraries use this practice, so you may get some
1912 warnings about symbols in the libraries as well as in your programs.
1914 There are three kinds of global symbols, illustrated here by C examples:
1918 A definition, which goes in the initialized data section of the output
1922 An undefined reference, which does not allocate space.
1923 There must be either a definition or a common symbol for the
1927 A common symbol. If there are only (one or more) common symbols for a
1928 variable, it goes in the uninitialized data area of the output file.
1929 The linker merges multiple common symbols for the same variable into a
1930 single symbol. If they are of different sizes, it picks the largest
1931 size. The linker turns a common symbol into a declaration, if there is
1932 a definition of the same variable.
1935 The @samp{--warn-common} option can produce five kinds of warnings.
1936 Each warning consists of a pair of lines: the first describes the symbol
1937 just encountered, and the second describes the previous symbol
1938 encountered with the same name. One or both of the two symbols will be
1943 Turning a common symbol into a reference, because there is already a
1944 definition for the symbol.
1946 @var{file}(@var{section}): warning: common of `@var{symbol}'
1947 overridden by definition
1948 @var{file}(@var{section}): warning: defined here
1952 Turning a common symbol into a reference, because a later definition for
1953 the symbol is encountered. This is the same as the previous case,
1954 except that the symbols are encountered in a different order.
1956 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1958 @var{file}(@var{section}): warning: common is here
1962 Merging a common symbol with a previous same-sized common symbol.
1964 @var{file}(@var{section}): warning: multiple common
1966 @var{file}(@var{section}): warning: previous common is here
1970 Merging a common symbol with a previous larger common symbol.
1972 @var{file}(@var{section}): warning: common of `@var{symbol}'
1973 overridden by larger common
1974 @var{file}(@var{section}): warning: larger common is here
1978 Merging a common symbol with a previous smaller common symbol. This is
1979 the same as the previous case, except that the symbols are
1980 encountered in a different order.
1982 @var{file}(@var{section}): warning: common of `@var{symbol}'
1983 overriding smaller common
1984 @var{file}(@var{section}): warning: smaller common is here
1988 @kindex --warn-constructors
1989 @item --warn-constructors
1990 Warn if any global constructors are used. This is only useful for a few
1991 object file formats. For formats like COFF or ELF, the linker can not
1992 detect the use of global constructors.
1994 @kindex --warn-multiple-gp
1995 @item --warn-multiple-gp
1996 Warn if multiple global pointer values are required in the output file.
1997 This is only meaningful for certain processors, such as the Alpha.
1998 Specifically, some processors put large-valued constants in a special
1999 section. A special register (the global pointer) points into the middle
2000 of this section, so that constants can be loaded efficiently via a
2001 base-register relative addressing mode. Since the offset in
2002 base-register relative mode is fixed and relatively small (e.g., 16
2003 bits), this limits the maximum size of the constant pool. Thus, in
2004 large programs, it is often necessary to use multiple global pointer
2005 values in order to be able to address all possible constants. This
2006 option causes a warning to be issued whenever this case occurs.
2009 @cindex warnings, on undefined symbols
2010 @cindex undefined symbols, warnings on
2012 Only warn once for each undefined symbol, rather than once per module
2015 @kindex --warn-section-align
2016 @cindex warnings, on section alignment
2017 @cindex section alignment, warnings on
2018 @item --warn-section-align
2019 Warn if the address of an output section is changed because of
2020 alignment. Typically, the alignment will be set by an input section.
2021 The address will only be changed if it not explicitly specified; that
2022 is, if the @code{SECTIONS} command does not specify a start address for
2023 the section (@pxref{SECTIONS}).
2025 @kindex --warn-shared-textrel
2026 @item --warn-shared-textrel
2027 Warn if the linker adds a DT_TEXTREL to a shared object.
2029 @kindex --warn-alternate-em
2030 @item --warn-alternate-em
2031 Warn if an object has alternate ELF machine code.
2033 @kindex --warn-unresolved-symbols
2034 @item --warn-unresolved-symbols
2035 If the linker is going to report an unresolved symbol (see the option
2036 @option{--unresolved-symbols}) it will normally generate an error.
2037 This option makes it generate a warning instead.
2039 @kindex --error-unresolved-symbols
2040 @item --error-unresolved-symbols
2041 This restores the linker's default behaviour of generating errors when
2042 it is reporting unresolved symbols.
2044 @kindex --whole-archive
2045 @cindex including an entire archive
2046 @item --whole-archive
2047 For each archive mentioned on the command line after the
2048 @option{--whole-archive} option, include every object file in the archive
2049 in the link, rather than searching the archive for the required object
2050 files. This is normally used to turn an archive file into a shared
2051 library, forcing every object to be included in the resulting shared
2052 library. This option may be used more than once.
2054 Two notes when using this option from gcc: First, gcc doesn't know
2055 about this option, so you have to use @option{-Wl,-whole-archive}.
2056 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2057 list of archives, because gcc will add its own list of archives to
2058 your link and you may not want this flag to affect those as well.
2060 @kindex --wrap=@var{symbol}
2061 @item --wrap=@var{symbol}
2062 Use a wrapper function for @var{symbol}. Any undefined reference to
2063 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2064 undefined reference to @code{__real_@var{symbol}} will be resolved to
2067 This can be used to provide a wrapper for a system function. The
2068 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2069 wishes to call the system function, it should call
2070 @code{__real_@var{symbol}}.
2072 Here is a trivial example:
2076 __wrap_malloc (size_t c)
2078 printf ("malloc called with %zu\n", c);
2079 return __real_malloc (c);
2083 If you link other code with this file using @option{--wrap malloc}, then
2084 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2085 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2086 call the real @code{malloc} function.
2088 You may wish to provide a @code{__real_malloc} function as well, so that
2089 links without the @option{--wrap} option will succeed. If you do this,
2090 you should not put the definition of @code{__real_malloc} in the same
2091 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2092 call before the linker has a chance to wrap it to @code{malloc}.
2094 @kindex --eh-frame-hdr
2095 @item --eh-frame-hdr
2096 Request creation of @code{.eh_frame_hdr} section and ELF
2097 @code{PT_GNU_EH_FRAME} segment header.
2099 @kindex --ld-generated-unwind-info
2100 @item --no-ld-generated-unwind-info
2101 Request creation of @code{.eh_frame} unwind info for linker
2102 generated code sections like PLT. This option is on by default
2103 if linker generated unwind info is supported.
2105 @kindex --enable-new-dtags
2106 @kindex --disable-new-dtags
2107 @item --enable-new-dtags
2108 @itemx --disable-new-dtags
2109 This linker can create the new dynamic tags in ELF. But the older ELF
2110 systems may not understand them. If you specify
2111 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2112 and older dynamic tags will be omitted.
2113 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2114 created. By default, the new dynamic tags are not created. Note that
2115 those options are only available for ELF systems.
2117 @kindex --hash-size=@var{number}
2118 @item --hash-size=@var{number}
2119 Set the default size of the linker's hash tables to a prime number
2120 close to @var{number}. Increasing this value can reduce the length of
2121 time it takes the linker to perform its tasks, at the expense of
2122 increasing the linker's memory requirements. Similarly reducing this
2123 value can reduce the memory requirements at the expense of speed.
2125 @kindex --hash-style=@var{style}
2126 @item --hash-style=@var{style}
2127 Set the type of linker's hash table(s). @var{style} can be either
2128 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2129 new style GNU @code{.gnu.hash} section or @code{both} for both
2130 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2131 hash tables. The default is @code{sysv}.
2133 @kindex --reduce-memory-overheads
2134 @item --reduce-memory-overheads
2135 This option reduces memory requirements at ld runtime, at the expense of
2136 linking speed. This was introduced to select the old O(n^2) algorithm
2137 for link map file generation, rather than the new O(n) algorithm which uses
2138 about 40% more memory for symbol storage.
2140 Another effect of the switch is to set the default hash table size to
2141 1021, which again saves memory at the cost of lengthening the linker's
2142 run time. This is not done however if the @option{--hash-size} switch
2145 The @option{--reduce-memory-overheads} switch may be also be used to
2146 enable other tradeoffs in future versions of the linker.
2149 @kindex --build-id=@var{style}
2151 @itemx --build-id=@var{style}
2152 Request creation of @code{.note.gnu.build-id} ELF note section.
2153 The contents of the note are unique bits identifying this linked
2154 file. @var{style} can be @code{uuid} to use 128 random bits,
2155 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
2156 parts of the output contents, @code{md5} to use a 128-bit
2157 @sc{MD5} hash on the normative parts of the output contents, or
2158 @code{0x@var{hexstring}} to use a chosen bit string specified as
2159 an even number of hexadecimal digits (@code{-} and @code{:}
2160 characters between digit pairs are ignored). If @var{style} is
2161 omitted, @code{sha1} is used.
2163 The @code{md5} and @code{sha1} styles produces an identifier
2164 that is always the same in an identical output file, but will be
2165 unique among all nonidentical output files. It is not intended
2166 to be compared as a checksum for the file's contents. A linked
2167 file may be changed later by other tools, but the build ID bit
2168 string identifying the original linked file does not change.
2170 Passing @code{none} for @var{style} disables the setting from any
2171 @code{--build-id} options earlier on the command line.
2176 @subsection Options Specific to i386 PE Targets
2178 @c man begin OPTIONS
2180 The i386 PE linker supports the @option{-shared} option, which causes
2181 the output to be a dynamically linked library (DLL) instead of a
2182 normal executable. You should name the output @code{*.dll} when you
2183 use this option. In addition, the linker fully supports the standard
2184 @code{*.def} files, which may be specified on the linker command line
2185 like an object file (in fact, it should precede archives it exports
2186 symbols from, to ensure that they get linked in, just like a normal
2189 In addition to the options common to all targets, the i386 PE linker
2190 support additional command line options that are specific to the i386
2191 PE target. Options that take values may be separated from their
2192 values by either a space or an equals sign.
2196 @kindex --add-stdcall-alias
2197 @item --add-stdcall-alias
2198 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2199 as-is and also with the suffix stripped.
2200 [This option is specific to the i386 PE targeted port of the linker]
2203 @item --base-file @var{file}
2204 Use @var{file} as the name of a file in which to save the base
2205 addresses of all the relocations needed for generating DLLs with
2207 [This is an i386 PE specific option]
2211 Create a DLL instead of a regular executable. You may also use
2212 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2214 [This option is specific to the i386 PE targeted port of the linker]
2216 @kindex --enable-long-section-names
2217 @kindex --disable-long-section-names
2218 @item --enable-long-section-names
2219 @itemx --disable-long-section-names
2220 The PE variants of the Coff object format add an extension that permits
2221 the use of section names longer than eight characters, the normal limit
2222 for Coff. By default, these names are only allowed in object files, as
2223 fully-linked executable images do not carry the Coff string table required
2224 to support the longer names. As a GNU extension, it is possible to
2225 allow their use in executable images as well, or to (probably pointlessly!)
2226 disallow it in object files, by using these two options. Executable images
2227 generated with these long section names are slightly non-standard, carrying
2228 as they do a string table, and may generate confusing output when examined
2229 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2230 GDB relies on the use of PE long section names to find Dwarf-2 debug
2231 information sections in an executable image at runtime, and so if neither
2232 option is specified on the command-line, @command{ld} will enable long
2233 section names, overriding the default and technically correct behaviour,
2234 when it finds the presence of debug information while linking an executable
2235 image and not stripping symbols.
2236 [This option is valid for all PE targeted ports of the linker]
2238 @kindex --enable-stdcall-fixup
2239 @kindex --disable-stdcall-fixup
2240 @item --enable-stdcall-fixup
2241 @itemx --disable-stdcall-fixup
2242 If the link finds a symbol that it cannot resolve, it will attempt to
2243 do ``fuzzy linking'' by looking for another defined symbol that differs
2244 only in the format of the symbol name (cdecl vs stdcall) and will
2245 resolve that symbol by linking to the match. For example, the
2246 undefined symbol @code{_foo} might be linked to the function
2247 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2248 to the function @code{_bar}. When the linker does this, it prints a
2249 warning, since it normally should have failed to link, but sometimes
2250 import libraries generated from third-party dlls may need this feature
2251 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2252 feature is fully enabled and warnings are not printed. If you specify
2253 @option{--disable-stdcall-fixup}, this feature is disabled and such
2254 mismatches are considered to be errors.
2255 [This option is specific to the i386 PE targeted port of the linker]
2257 @kindex --leading-underscore
2258 @kindex --no-leading-underscore
2259 @item --leading-underscore
2260 @itemx --no-leading-underscore
2261 For most targets default symbol-prefix is an underscore and is defined
2262 in target's description. By this option it is possible to
2263 disable/enable the default underscore symbol-prefix.
2265 @cindex DLLs, creating
2266 @kindex --export-all-symbols
2267 @item --export-all-symbols
2268 If given, all global symbols in the objects used to build a DLL will
2269 be exported by the DLL. Note that this is the default if there
2270 otherwise wouldn't be any exported symbols. When symbols are
2271 explicitly exported via DEF files or implicitly exported via function
2272 attributes, the default is to not export anything else unless this
2273 option is given. Note that the symbols @code{DllMain@@12},
2274 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2275 @code{impure_ptr} will not be automatically
2276 exported. Also, symbols imported from other DLLs will not be
2277 re-exported, nor will symbols specifying the DLL's internal layout
2278 such as those beginning with @code{_head_} or ending with
2279 @code{_iname}. In addition, no symbols from @code{libgcc},
2280 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2281 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2282 not be exported, to help with C++ DLLs. Finally, there is an
2283 extensive list of cygwin-private symbols that are not exported
2284 (obviously, this applies on when building DLLs for cygwin targets).
2285 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2286 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2287 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2288 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2289 @code{cygwin_premain3}, and @code{environ}.
2290 [This option is specific to the i386 PE targeted port of the linker]
2292 @kindex --exclude-symbols
2293 @item --exclude-symbols @var{symbol},@var{symbol},...
2294 Specifies a list of symbols which should not be automatically
2295 exported. The symbol names may be delimited by commas or colons.
2296 [This option is specific to the i386 PE targeted port of the linker]
2298 @kindex --exclude-all-symbols
2299 @item --exclude-all-symbols
2300 Specifies no symbols should be automatically exported.
2301 [This option is specific to the i386 PE targeted port of the linker]
2303 @kindex --file-alignment
2304 @item --file-alignment
2305 Specify the file alignment. Sections in the file will always begin at
2306 file offsets which are multiples of this number. This defaults to
2308 [This option is specific to the i386 PE targeted port of the linker]
2312 @item --heap @var{reserve}
2313 @itemx --heap @var{reserve},@var{commit}
2314 Specify the number of bytes of memory to reserve (and optionally commit)
2315 to be used as heap for this program. The default is 1Mb reserved, 4K
2317 [This option is specific to the i386 PE targeted port of the linker]
2320 @kindex --image-base
2321 @item --image-base @var{value}
2322 Use @var{value} as the base address of your program or dll. This is
2323 the lowest memory location that will be used when your program or dll
2324 is loaded. To reduce the need to relocate and improve performance of
2325 your dlls, each should have a unique base address and not overlap any
2326 other dlls. The default is 0x400000 for executables, and 0x10000000
2328 [This option is specific to the i386 PE targeted port of the linker]
2332 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2333 symbols before they are exported.
2334 [This option is specific to the i386 PE targeted port of the linker]
2336 @kindex --large-address-aware
2337 @item --large-address-aware
2338 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2339 header is set to indicate that this executable supports virtual addresses
2340 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2341 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2342 section of the BOOT.INI. Otherwise, this bit has no effect.
2343 [This option is specific to PE targeted ports of the linker]
2345 @kindex --major-image-version
2346 @item --major-image-version @var{value}
2347 Sets the major number of the ``image version''. Defaults to 1.
2348 [This option is specific to the i386 PE targeted port of the linker]
2350 @kindex --major-os-version
2351 @item --major-os-version @var{value}
2352 Sets the major number of the ``os version''. Defaults to 4.
2353 [This option is specific to the i386 PE targeted port of the linker]
2355 @kindex --major-subsystem-version
2356 @item --major-subsystem-version @var{value}
2357 Sets the major number of the ``subsystem version''. Defaults to 4.
2358 [This option is specific to the i386 PE targeted port of the linker]
2360 @kindex --minor-image-version
2361 @item --minor-image-version @var{value}
2362 Sets the minor number of the ``image version''. Defaults to 0.
2363 [This option is specific to the i386 PE targeted port of the linker]
2365 @kindex --minor-os-version
2366 @item --minor-os-version @var{value}
2367 Sets the minor number of the ``os version''. Defaults to 0.
2368 [This option is specific to the i386 PE targeted port of the linker]
2370 @kindex --minor-subsystem-version
2371 @item --minor-subsystem-version @var{value}
2372 Sets the minor number of the ``subsystem version''. Defaults to 0.
2373 [This option is specific to the i386 PE targeted port of the linker]
2375 @cindex DEF files, creating
2376 @cindex DLLs, creating
2377 @kindex --output-def
2378 @item --output-def @var{file}
2379 The linker will create the file @var{file} which will contain a DEF
2380 file corresponding to the DLL the linker is generating. This DEF file
2381 (which should be called @code{*.def}) may be used to create an import
2382 library with @code{dlltool} or may be used as a reference to
2383 automatically or implicitly exported symbols.
2384 [This option is specific to the i386 PE targeted port of the linker]
2386 @cindex DLLs, creating
2387 @kindex --out-implib
2388 @item --out-implib @var{file}
2389 The linker will create the file @var{file} which will contain an
2390 import lib corresponding to the DLL the linker is generating. This
2391 import lib (which should be called @code{*.dll.a} or @code{*.a}
2392 may be used to link clients against the generated DLL; this behaviour
2393 makes it possible to skip a separate @code{dlltool} import library
2395 [This option is specific to the i386 PE targeted port of the linker]
2397 @kindex --enable-auto-image-base
2398 @item --enable-auto-image-base
2399 Automatically choose the image base for DLLs, unless one is specified
2400 using the @code{--image-base} argument. By using a hash generated
2401 from the dllname to create unique image bases for each DLL, in-memory
2402 collisions and relocations which can delay program execution are
2404 [This option is specific to the i386 PE targeted port of the linker]
2406 @kindex --disable-auto-image-base
2407 @item --disable-auto-image-base
2408 Do not automatically generate a unique image base. If there is no
2409 user-specified image base (@code{--image-base}) then use the platform
2411 [This option is specific to the i386 PE targeted port of the linker]
2413 @cindex DLLs, linking to
2414 @kindex --dll-search-prefix
2415 @item --dll-search-prefix @var{string}
2416 When linking dynamically to a dll without an import library,
2417 search for @code{<string><basename>.dll} in preference to
2418 @code{lib<basename>.dll}. This behaviour allows easy distinction
2419 between DLLs built for the various "subplatforms": native, cygwin,
2420 uwin, pw, etc. For instance, cygwin DLLs typically use
2421 @code{--dll-search-prefix=cyg}.
2422 [This option is specific to the i386 PE targeted port of the linker]
2424 @kindex --enable-auto-import
2425 @item --enable-auto-import
2426 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2427 DATA imports from DLLs, and create the necessary thunking symbols when
2428 building the import libraries with those DATA exports. Note: Use of the
2429 'auto-import' extension will cause the text section of the image file
2430 to be made writable. This does not conform to the PE-COFF format
2431 specification published by Microsoft.
2433 Note - use of the 'auto-import' extension will also cause read only
2434 data which would normally be placed into the .rdata section to be
2435 placed into the .data section instead. This is in order to work
2436 around a problem with consts that is described here:
2437 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2439 Using 'auto-import' generally will 'just work' -- but sometimes you may
2442 "variable '<var>' can't be auto-imported. Please read the
2443 documentation for ld's @code{--enable-auto-import} for details."
2445 This message occurs when some (sub)expression accesses an address
2446 ultimately given by the sum of two constants (Win32 import tables only
2447 allow one). Instances where this may occur include accesses to member
2448 fields of struct variables imported from a DLL, as well as using a
2449 constant index into an array variable imported from a DLL. Any
2450 multiword variable (arrays, structs, long long, etc) may trigger
2451 this error condition. However, regardless of the exact data type
2452 of the offending exported variable, ld will always detect it, issue
2453 the warning, and exit.
2455 There are several ways to address this difficulty, regardless of the
2456 data type of the exported variable:
2458 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2459 of adjusting references in your client code for runtime environment, so
2460 this method works only when runtime environment supports this feature.
2462 A second solution is to force one of the 'constants' to be a variable --
2463 that is, unknown and un-optimizable at compile time. For arrays,
2464 there are two possibilities: a) make the indexee (the array's address)
2465 a variable, or b) make the 'constant' index a variable. Thus:
2468 extern type extern_array[];
2470 @{ volatile type *t=extern_array; t[1] @}
2476 extern type extern_array[];
2478 @{ volatile int t=1; extern_array[t] @}
2481 For structs (and most other multiword data types) the only option
2482 is to make the struct itself (or the long long, or the ...) variable:
2485 extern struct s extern_struct;
2486 extern_struct.field -->
2487 @{ volatile struct s *t=&extern_struct; t->field @}
2493 extern long long extern_ll;
2495 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2498 A third method of dealing with this difficulty is to abandon
2499 'auto-import' for the offending symbol and mark it with
2500 @code{__declspec(dllimport)}. However, in practice that
2501 requires using compile-time #defines to indicate whether you are
2502 building a DLL, building client code that will link to the DLL, or
2503 merely building/linking to a static library. In making the choice
2504 between the various methods of resolving the 'direct address with
2505 constant offset' problem, you should consider typical real-world usage:
2513 void main(int argc, char **argv)@{
2514 printf("%d\n",arr[1]);
2524 void main(int argc, char **argv)@{
2525 /* This workaround is for win32 and cygwin; do not "optimize" */
2526 volatile int *parr = arr;
2527 printf("%d\n",parr[1]);
2534 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2535 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2536 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2537 #define FOO_IMPORT __declspec(dllimport)
2541 extern FOO_IMPORT int arr[];
2544 void main(int argc, char **argv)@{
2545 printf("%d\n",arr[1]);
2549 A fourth way to avoid this problem is to re-code your
2550 library to use a functional interface rather than a data interface
2551 for the offending variables (e.g. set_foo() and get_foo() accessor
2553 [This option is specific to the i386 PE targeted port of the linker]
2555 @kindex --disable-auto-import
2556 @item --disable-auto-import
2557 Do not attempt to do sophisticated linking of @code{_symbol} to
2558 @code{__imp__symbol} for DATA imports from DLLs.
2559 [This option is specific to the i386 PE targeted port of the linker]
2561 @kindex --enable-runtime-pseudo-reloc
2562 @item --enable-runtime-pseudo-reloc
2563 If your code contains expressions described in --enable-auto-import section,
2564 that is, DATA imports from DLL with non-zero offset, this switch will create
2565 a vector of 'runtime pseudo relocations' which can be used by runtime
2566 environment to adjust references to such data in your client code.
2567 [This option is specific to the i386 PE targeted port of the linker]
2569 @kindex --disable-runtime-pseudo-reloc
2570 @item --disable-runtime-pseudo-reloc
2571 Do not create pseudo relocations for non-zero offset DATA imports from
2573 [This option is specific to the i386 PE targeted port of the linker]
2575 @kindex --enable-extra-pe-debug
2576 @item --enable-extra-pe-debug
2577 Show additional debug info related to auto-import symbol thunking.
2578 [This option is specific to the i386 PE targeted port of the linker]
2580 @kindex --section-alignment
2581 @item --section-alignment
2582 Sets the section alignment. Sections in memory will always begin at
2583 addresses which are a multiple of this number. Defaults to 0x1000.
2584 [This option is specific to the i386 PE targeted port of the linker]
2588 @item --stack @var{reserve}
2589 @itemx --stack @var{reserve},@var{commit}
2590 Specify the number of bytes of memory to reserve (and optionally commit)
2591 to be used as stack for this program. The default is 2Mb reserved, 4K
2593 [This option is specific to the i386 PE targeted port of the linker]
2596 @item --subsystem @var{which}
2597 @itemx --subsystem @var{which}:@var{major}
2598 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2599 Specifies the subsystem under which your program will execute. The
2600 legal values for @var{which} are @code{native}, @code{windows},
2601 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2602 the subsystem version also. Numeric values are also accepted for
2604 [This option is specific to the i386 PE targeted port of the linker]
2606 The following options set flags in the @code{DllCharacteristics} field
2607 of the PE file header:
2608 [These options are specific to PE targeted ports of the linker]
2610 @kindex --dynamicbase
2612 The image base address may be relocated using address space layout
2613 randomization (ASLR). This feature was introduced with MS Windows
2614 Vista for i386 PE targets.
2616 @kindex --forceinteg
2618 Code integrity checks are enforced.
2622 The image is compatible with the Data Execution Prevention.
2623 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2625 @kindex --no-isolation
2626 @item --no-isolation
2627 Although the image understands isolation, do not isolate the image.
2631 The image does not use SEH. No SE handler may be called from
2636 Do not bind this image.
2640 The driver uses the MS Windows Driver Model.
2644 The image is Terminal Server aware.
2651 @subsection Options specific to C6X uClinux targets
2653 @c man begin OPTIONS
2655 The C6X uClinux target uses a binary format called DSBT to support shared
2656 libraries. Each shared library in the system needs to have a unique index;
2657 all executables use an index of 0.
2662 @item --dsbt-size @var{size}
2663 This option sets the number of entires in the DSBT of the current executable
2664 or shared library to @var{size}. The default is to create a table with 64
2667 @kindex --dsbt-index
2668 @item --dsbt-index @var{index}
2669 This option sets the DSBT index of the current executable or shared library
2670 to @var{index}. The default is 0, which is appropriate for generating
2671 executables. If a shared library is generated with a DSBT index of 0, the
2672 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
2674 @kindex --no-merge-exidx-entries
2675 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
2676 exidx entries in frame unwind info.
2684 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2686 @c man begin OPTIONS
2688 The 68HC11 and 68HC12 linkers support specific options to control the
2689 memory bank switching mapping and trampoline code generation.
2693 @kindex --no-trampoline
2694 @item --no-trampoline
2695 This option disables the generation of trampoline. By default a trampoline
2696 is generated for each far function which is called using a @code{jsr}
2697 instruction (this happens when a pointer to a far function is taken).
2699 @kindex --bank-window
2700 @item --bank-window @var{name}
2701 This option indicates to the linker the name of the memory region in
2702 the @samp{MEMORY} specification that describes the memory bank window.
2703 The definition of such region is then used by the linker to compute
2704 paging and addresses within the memory window.
2712 @subsection Options specific to Motorola 68K target
2714 @c man begin OPTIONS
2716 The following options are supported to control handling of GOT generation
2717 when linking for 68K targets.
2722 @item --got=@var{type}
2723 This option tells the linker which GOT generation scheme to use.
2724 @var{type} should be one of @samp{single}, @samp{negative},
2725 @samp{multigot} or @samp{target}. For more information refer to the
2726 Info entry for @file{ld}.
2735 @section Environment Variables
2737 @c man begin ENVIRONMENT
2739 You can change the behaviour of @command{ld} with the environment variables
2740 @ifclear SingleFormat
2743 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2745 @ifclear SingleFormat
2747 @cindex default input format
2748 @code{GNUTARGET} determines the input-file object format if you don't
2749 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2750 of the BFD names for an input format (@pxref{BFD}). If there is no
2751 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2752 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2753 attempts to discover the input format by examining binary input files;
2754 this method often succeeds, but there are potential ambiguities, since
2755 there is no method of ensuring that the magic number used to specify
2756 object-file formats is unique. However, the configuration procedure for
2757 BFD on each system places the conventional format for that system first
2758 in the search-list, so ambiguities are resolved in favor of convention.
2762 @cindex default emulation
2763 @cindex emulation, default
2764 @code{LDEMULATION} determines the default emulation if you don't use the
2765 @samp{-m} option. The emulation can affect various aspects of linker
2766 behaviour, particularly the default linker script. You can list the
2767 available emulations with the @samp{--verbose} or @samp{-V} options. If
2768 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2769 variable is not defined, the default emulation depends upon how the
2770 linker was configured.
2772 @kindex COLLECT_NO_DEMANGLE
2773 @cindex demangling, default
2774 Normally, the linker will default to demangling symbols. However, if
2775 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2776 default to not demangling symbols. This environment variable is used in
2777 a similar fashion by the @code{gcc} linker wrapper program. The default
2778 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2785 @chapter Linker Scripts
2788 @cindex linker scripts
2789 @cindex command files
2790 Every link is controlled by a @dfn{linker script}. This script is
2791 written in the linker command language.
2793 The main purpose of the linker script is to describe how the sections in
2794 the input files should be mapped into the output file, and to control
2795 the memory layout of the output file. Most linker scripts do nothing
2796 more than this. However, when necessary, the linker script can also
2797 direct the linker to perform many other operations, using the commands
2800 The linker always uses a linker script. If you do not supply one
2801 yourself, the linker will use a default script that is compiled into the
2802 linker executable. You can use the @samp{--verbose} command line option
2803 to display the default linker script. Certain command line options,
2804 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2806 You may supply your own linker script by using the @samp{-T} command
2807 line option. When you do this, your linker script will replace the
2808 default linker script.
2810 You may also use linker scripts implicitly by naming them as input files
2811 to the linker, as though they were files to be linked. @xref{Implicit
2815 * Basic Script Concepts:: Basic Linker Script Concepts
2816 * Script Format:: Linker Script Format
2817 * Simple Example:: Simple Linker Script Example
2818 * Simple Commands:: Simple Linker Script Commands
2819 * Assignments:: Assigning Values to Symbols
2820 * SECTIONS:: SECTIONS Command
2821 * MEMORY:: MEMORY Command
2822 * PHDRS:: PHDRS Command
2823 * VERSION:: VERSION Command
2824 * Expressions:: Expressions in Linker Scripts
2825 * Implicit Linker Scripts:: Implicit Linker Scripts
2828 @node Basic Script Concepts
2829 @section Basic Linker Script Concepts
2830 @cindex linker script concepts
2831 We need to define some basic concepts and vocabulary in order to
2832 describe the linker script language.
2834 The linker combines input files into a single output file. The output
2835 file and each input file are in a special data format known as an
2836 @dfn{object file format}. Each file is called an @dfn{object file}.
2837 The output file is often called an @dfn{executable}, but for our
2838 purposes we will also call it an object file. Each object file has,
2839 among other things, a list of @dfn{sections}. We sometimes refer to a
2840 section in an input file as an @dfn{input section}; similarly, a section
2841 in the output file is an @dfn{output section}.
2843 Each section in an object file has a name and a size. Most sections
2844 also have an associated block of data, known as the @dfn{section
2845 contents}. A section may be marked as @dfn{loadable}, which means that
2846 the contents should be loaded into memory when the output file is run.
2847 A section with no contents may be @dfn{allocatable}, which means that an
2848 area in memory should be set aside, but nothing in particular should be
2849 loaded there (in some cases this memory must be zeroed out). A section
2850 which is neither loadable nor allocatable typically contains some sort
2851 of debugging information.
2853 Every loadable or allocatable output section has two addresses. The
2854 first is the @dfn{VMA}, or virtual memory address. This is the address
2855 the section will have when the output file is run. The second is the
2856 @dfn{LMA}, or load memory address. This is the address at which the
2857 section will be loaded. In most cases the two addresses will be the
2858 same. An example of when they might be different is when a data section
2859 is loaded into ROM, and then copied into RAM when the program starts up
2860 (this technique is often used to initialize global variables in a ROM
2861 based system). In this case the ROM address would be the LMA, and the
2862 RAM address would be the VMA.
2864 You can see the sections in an object file by using the @code{objdump}
2865 program with the @samp{-h} option.
2867 Every object file also has a list of @dfn{symbols}, known as the
2868 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2869 has a name, and each defined symbol has an address, among other
2870 information. If you compile a C or C++ program into an object file, you
2871 will get a defined symbol for every defined function and global or
2872 static variable. Every undefined function or global variable which is
2873 referenced in the input file will become an undefined symbol.
2875 You can see the symbols in an object file by using the @code{nm}
2876 program, or by using the @code{objdump} program with the @samp{-t}
2880 @section Linker Script Format
2881 @cindex linker script format
2882 Linker scripts are text files.
2884 You write a linker script as a series of commands. Each command is
2885 either a keyword, possibly followed by arguments, or an assignment to a
2886 symbol. You may separate commands using semicolons. Whitespace is
2889 Strings such as file or format names can normally be entered directly.
2890 If the file name contains a character such as a comma which would
2891 otherwise serve to separate file names, you may put the file name in
2892 double quotes. There is no way to use a double quote character in a
2895 You may include comments in linker scripts just as in C, delimited by
2896 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2899 @node Simple Example
2900 @section Simple Linker Script Example
2901 @cindex linker script example
2902 @cindex example of linker script
2903 Many linker scripts are fairly simple.
2905 The simplest possible linker script has just one command:
2906 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2907 memory layout of the output file.
2909 The @samp{SECTIONS} command is a powerful command. Here we will
2910 describe a simple use of it. Let's assume your program consists only of
2911 code, initialized data, and uninitialized data. These will be in the
2912 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2913 Let's assume further that these are the only sections which appear in
2916 For this example, let's say that the code should be loaded at address
2917 0x10000, and that the data should start at address 0x8000000. Here is a
2918 linker script which will do that:
2923 .text : @{ *(.text) @}
2925 .data : @{ *(.data) @}
2926 .bss : @{ *(.bss) @}
2930 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2931 followed by a series of symbol assignments and output section
2932 descriptions enclosed in curly braces.
2934 The first line inside the @samp{SECTIONS} command of the above example
2935 sets the value of the special symbol @samp{.}, which is the location
2936 counter. If you do not specify the address of an output section in some
2937 other way (other ways are described later), the address is set from the
2938 current value of the location counter. The location counter is then
2939 incremented by the size of the output section. At the start of the
2940 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2942 The second line defines an output section, @samp{.text}. The colon is
2943 required syntax which may be ignored for now. Within the curly braces
2944 after the output section name, you list the names of the input sections
2945 which should be placed into this output section. The @samp{*} is a
2946 wildcard which matches any file name. The expression @samp{*(.text)}
2947 means all @samp{.text} input sections in all input files.
2949 Since the location counter is @samp{0x10000} when the output section
2950 @samp{.text} is defined, the linker will set the address of the
2951 @samp{.text} section in the output file to be @samp{0x10000}.
2953 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2954 the output file. The linker will place the @samp{.data} output section
2955 at address @samp{0x8000000}. After the linker places the @samp{.data}
2956 output section, the value of the location counter will be
2957 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2958 effect is that the linker will place the @samp{.bss} output section
2959 immediately after the @samp{.data} output section in memory.
2961 The linker will ensure that each output section has the required
2962 alignment, by increasing the location counter if necessary. In this
2963 example, the specified addresses for the @samp{.text} and @samp{.data}
2964 sections will probably satisfy any alignment constraints, but the linker
2965 may have to create a small gap between the @samp{.data} and @samp{.bss}
2968 That's it! That's a simple and complete linker script.
2970 @node Simple Commands
2971 @section Simple Linker Script Commands
2972 @cindex linker script simple commands
2973 In this section we describe the simple linker script commands.
2976 * Entry Point:: Setting the entry point
2977 * File Commands:: Commands dealing with files
2978 @ifclear SingleFormat
2979 * Format Commands:: Commands dealing with object file formats
2982 * REGION_ALIAS:: Assign alias names to memory regions
2983 * Miscellaneous Commands:: Other linker script commands
2987 @subsection Setting the Entry Point
2988 @kindex ENTRY(@var{symbol})
2989 @cindex start of execution
2990 @cindex first instruction
2992 The first instruction to execute in a program is called the @dfn{entry
2993 point}. You can use the @code{ENTRY} linker script command to set the
2994 entry point. The argument is a symbol name:
2999 There are several ways to set the entry point. The linker will set the
3000 entry point by trying each of the following methods in order, and
3001 stopping when one of them succeeds:
3004 the @samp{-e} @var{entry} command-line option;
3006 the @code{ENTRY(@var{symbol})} command in a linker script;
3008 the value of a target specific symbol, if it is defined; For many
3009 targets this is @code{start}, but PE and BeOS based systems for example
3010 check a list of possible entry symbols, matching the first one found.
3012 the address of the first byte of the @samp{.text} section, if present;
3014 The address @code{0}.
3018 @subsection Commands Dealing with Files
3019 @cindex linker script file commands
3020 Several linker script commands deal with files.
3023 @item INCLUDE @var{filename}
3024 @kindex INCLUDE @var{filename}
3025 @cindex including a linker script
3026 Include the linker script @var{filename} at this point. The file will
3027 be searched for in the current directory, and in any directory specified
3028 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3031 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3032 @code{SECTIONS} commands, or in output section descriptions.
3034 @item INPUT(@var{file}, @var{file}, @dots{})
3035 @itemx INPUT(@var{file} @var{file} @dots{})
3036 @kindex INPUT(@var{files})
3037 @cindex input files in linker scripts
3038 @cindex input object files in linker scripts
3039 @cindex linker script input object files
3040 The @code{INPUT} command directs the linker to include the named files
3041 in the link, as though they were named on the command line.
3043 For example, if you always want to include @file{subr.o} any time you do
3044 a link, but you can't be bothered to put it on every link command line,
3045 then you can put @samp{INPUT (subr.o)} in your linker script.
3047 In fact, if you like, you can list all of your input files in the linker
3048 script, and then invoke the linker with nothing but a @samp{-T} option.
3050 In case a @dfn{sysroot prefix} is configured, and the filename starts
3051 with the @samp{/} character, and the script being processed was
3052 located inside the @dfn{sysroot prefix}, the filename will be looked
3053 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3054 open the file in the current directory. If it is not found, the
3055 linker will search through the archive library search path. See the
3056 description of @samp{-L} in @ref{Options,,Command Line Options}.
3058 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3059 name to @code{lib@var{file}.a}, as with the command line argument
3062 When you use the @code{INPUT} command in an implicit linker script, the
3063 files will be included in the link at the point at which the linker
3064 script file is included. This can affect archive searching.
3066 @item GROUP(@var{file}, @var{file}, @dots{})
3067 @itemx GROUP(@var{file} @var{file} @dots{})
3068 @kindex GROUP(@var{files})
3069 @cindex grouping input files
3070 The @code{GROUP} command is like @code{INPUT}, except that the named
3071 files should all be archives, and they are searched repeatedly until no
3072 new undefined references are created. See the description of @samp{-(}
3073 in @ref{Options,,Command Line Options}.
3075 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3076 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3077 @kindex AS_NEEDED(@var{files})
3078 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3079 commands, among other filenames. The files listed will be handled
3080 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3081 with the exception of ELF shared libraries, that will be added only
3082 when they are actually needed. This construct essentially enables
3083 @option{--as-needed} option for all the files listed inside of it
3084 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3087 @item OUTPUT(@var{filename})
3088 @kindex OUTPUT(@var{filename})
3089 @cindex output file name in linker script
3090 The @code{OUTPUT} command names the output file. Using
3091 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3092 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3093 Line Options}). If both are used, the command line option takes
3096 You can use the @code{OUTPUT} command to define a default name for the
3097 output file other than the usual default of @file{a.out}.
3099 @item SEARCH_DIR(@var{path})
3100 @kindex SEARCH_DIR(@var{path})
3101 @cindex library search path in linker script
3102 @cindex archive search path in linker script
3103 @cindex search path in linker script
3104 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3105 @command{ld} looks for archive libraries. Using
3106 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3107 on the command line (@pxref{Options,,Command Line Options}). If both
3108 are used, then the linker will search both paths. Paths specified using
3109 the command line option are searched first.
3111 @item STARTUP(@var{filename})
3112 @kindex STARTUP(@var{filename})
3113 @cindex first input file
3114 The @code{STARTUP} command is just like the @code{INPUT} command, except
3115 that @var{filename} will become the first input file to be linked, as
3116 though it were specified first on the command line. This may be useful
3117 when using a system in which the entry point is always the start of the
3121 @ifclear SingleFormat
3122 @node Format Commands
3123 @subsection Commands Dealing with Object File Formats
3124 A couple of linker script commands deal with object file formats.
3127 @item OUTPUT_FORMAT(@var{bfdname})
3128 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3129 @kindex OUTPUT_FORMAT(@var{bfdname})
3130 @cindex output file format in linker script
3131 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3132 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3133 exactly like using @samp{--oformat @var{bfdname}} on the command line
3134 (@pxref{Options,,Command Line Options}). If both are used, the command
3135 line option takes precedence.
3137 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3138 formats based on the @samp{-EB} and @samp{-EL} command line options.
3139 This permits the linker script to set the output format based on the
3142 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3143 will be the first argument, @var{default}. If @samp{-EB} is used, the
3144 output format will be the second argument, @var{big}. If @samp{-EL} is
3145 used, the output format will be the third argument, @var{little}.
3147 For example, the default linker script for the MIPS ELF target uses this
3150 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3152 This says that the default format for the output file is
3153 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3154 option, the output file will be created in the @samp{elf32-littlemips}
3157 @item TARGET(@var{bfdname})
3158 @kindex TARGET(@var{bfdname})
3159 @cindex input file format in linker script
3160 The @code{TARGET} command names the BFD format to use when reading input
3161 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3162 This command is like using @samp{-b @var{bfdname}} on the command line
3163 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3164 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3165 command is also used to set the format for the output file. @xref{BFD}.
3170 @subsection Assign alias names to memory regions
3171 @kindex REGION_ALIAS(@var{alias}, @var{region})
3172 @cindex region alias
3173 @cindex region names
3175 Alias names can be added to existing memory regions created with the
3176 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3179 REGION_ALIAS(@var{alias}, @var{region})
3182 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3183 memory region @var{region}. This allows a flexible mapping of output sections
3184 to memory regions. An example follows.
3186 Suppose we have an application for embedded systems which come with various
3187 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3188 that allows code execution or data storage. Some may have a read-only,
3189 non-volatile memory @code{ROM} that allows code execution and read-only data
3190 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3191 read-only data access and no code execution capability. We have four output
3196 @code{.text} program code;
3198 @code{.rodata} read-only data;
3200 @code{.data} read-write initialized data;
3202 @code{.bss} read-write zero initialized data.
3205 The goal is to provide a linker command file that contains a system independent
3206 part defining the output sections and a system dependent part mapping the
3207 output sections to the memory regions available on the system. Our embedded
3208 systems come with three different memory setups @code{A}, @code{B} and
3210 @multitable @columnfractions .25 .25 .25 .25
3211 @item Section @tab Variant A @tab Variant B @tab Variant C
3212 @item .text @tab RAM @tab ROM @tab ROM
3213 @item .rodata @tab RAM @tab ROM @tab ROM2
3214 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3215 @item .bss @tab RAM @tab RAM @tab RAM
3217 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3218 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3219 the load address of the @code{.data} section starts in all three variants at
3220 the end of the @code{.rodata} section.
3222 The base linker script that deals with the output sections follows. It
3223 includes the system dependent @code{linkcmds.memory} file that describes the
3226 INCLUDE linkcmds.memory
3239 .data : AT (rodata_end)
3244 data_size = SIZEOF(.data);
3245 data_load_start = LOADADDR(.data);
3253 Now we need three different @code{linkcmds.memory} files to define memory
3254 regions and alias names. The content of @code{linkcmds.memory} for the three
3255 variants @code{A}, @code{B} and @code{C}:
3258 Here everything goes into the @code{RAM}.
3262 RAM : ORIGIN = 0, LENGTH = 4M
3265 REGION_ALIAS("REGION_TEXT", RAM);
3266 REGION_ALIAS("REGION_RODATA", RAM);
3267 REGION_ALIAS("REGION_DATA", RAM);
3268 REGION_ALIAS("REGION_BSS", RAM);
3271 Program code and read-only data go into the @code{ROM}. Read-write data goes
3272 into the @code{RAM}. An image of the initialized data is loaded into the
3273 @code{ROM} and will be copied during system start into the @code{RAM}.
3277 ROM : ORIGIN = 0, LENGTH = 3M
3278 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3281 REGION_ALIAS("REGION_TEXT", ROM);
3282 REGION_ALIAS("REGION_RODATA", ROM);
3283 REGION_ALIAS("REGION_DATA", RAM);
3284 REGION_ALIAS("REGION_BSS", RAM);
3287 Program code goes into the @code{ROM}. Read-only data goes into the
3288 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3289 initialized data is loaded into the @code{ROM2} and will be copied during
3290 system start into the @code{RAM}.
3294 ROM : ORIGIN = 0, LENGTH = 2M
3295 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3296 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3299 REGION_ALIAS("REGION_TEXT", ROM);
3300 REGION_ALIAS("REGION_RODATA", ROM2);
3301 REGION_ALIAS("REGION_DATA", RAM);
3302 REGION_ALIAS("REGION_BSS", RAM);
3306 It is possible to write a common system initialization routine to copy the
3307 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3312 extern char data_start [];
3313 extern char data_size [];
3314 extern char data_load_start [];
3316 void copy_data(void)
3318 if (data_start != data_load_start)
3320 memcpy(data_start, data_load_start, (size_t) data_size);
3325 @node Miscellaneous Commands
3326 @subsection Other Linker Script Commands
3327 There are a few other linker scripts commands.
3330 @item ASSERT(@var{exp}, @var{message})
3332 @cindex assertion in linker script
3333 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3334 with an error code, and print @var{message}.
3336 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3338 @cindex undefined symbol in linker script
3339 Force @var{symbol} to be entered in the output file as an undefined
3340 symbol. Doing this may, for example, trigger linking of additional
3341 modules from standard libraries. You may list several @var{symbol}s for
3342 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3343 command has the same effect as the @samp{-u} command-line option.
3345 @item FORCE_COMMON_ALLOCATION
3346 @kindex FORCE_COMMON_ALLOCATION
3347 @cindex common allocation in linker script
3348 This command has the same effect as the @samp{-d} command-line option:
3349 to make @command{ld} assign space to common symbols even if a relocatable
3350 output file is specified (@samp{-r}).
3352 @item INHIBIT_COMMON_ALLOCATION
3353 @kindex INHIBIT_COMMON_ALLOCATION
3354 @cindex common allocation in linker script
3355 This command has the same effect as the @samp{--no-define-common}
3356 command-line option: to make @code{ld} omit the assignment of addresses
3357 to common symbols even for a non-relocatable output file.
3359 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3361 @cindex insert user script into default script
3362 This command is typically used in a script specified by @samp{-T} to
3363 augment the default @code{SECTIONS} with, for example, overlays. It
3364 inserts all prior linker script statements after (or before)
3365 @var{output_section}, and also causes @samp{-T} to not override the
3366 default linker script. The exact insertion point is as for orphan
3367 sections. @xref{Location Counter}. The insertion happens after the
3368 linker has mapped input sections to output sections. Prior to the
3369 insertion, since @samp{-T} scripts are parsed before the default
3370 linker script, statements in the @samp{-T} script occur before the
3371 default linker script statements in the internal linker representation
3372 of the script. In particular, input section assignments will be made
3373 to @samp{-T} output sections before those in the default script. Here
3374 is an example of how a @samp{-T} script using @code{INSERT} might look:
3381 .ov1 @{ ov1*(.text) @}
3382 .ov2 @{ ov2*(.text) @}
3388 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3389 @kindex NOCROSSREFS(@var{sections})
3390 @cindex cross references
3391 This command may be used to tell @command{ld} to issue an error about any
3392 references among certain output sections.
3394 In certain types of programs, particularly on embedded systems when
3395 using overlays, when one section is loaded into memory, another section
3396 will not be. Any direct references between the two sections would be
3397 errors. For example, it would be an error if code in one section called
3398 a function defined in the other section.
3400 The @code{NOCROSSREFS} command takes a list of output section names. If
3401 @command{ld} detects any cross references between the sections, it reports
3402 an error and returns a non-zero exit status. Note that the
3403 @code{NOCROSSREFS} command uses output section names, not input section
3406 @ifclear SingleFormat
3407 @item OUTPUT_ARCH(@var{bfdarch})
3408 @kindex OUTPUT_ARCH(@var{bfdarch})
3409 @cindex machine architecture
3410 @cindex architecture
3411 Specify a particular output machine architecture. The argument is one
3412 of the names used by the BFD library (@pxref{BFD}). You can see the
3413 architecture of an object file by using the @code{objdump} program with
3414 the @samp{-f} option.
3417 @item LD_FEATURE(@var{string})
3418 @kindex LD_FEATURE(@var{string})
3419 This command may be used to modify @command{ld} behavior. If
3420 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3421 in a script are simply treated as numbers everywhere.
3422 @xref{Expression Section}.
3426 @section Assigning Values to Symbols
3427 @cindex assignment in scripts
3428 @cindex symbol definition, scripts
3429 @cindex variables, defining
3430 You may assign a value to a symbol in a linker script. This will define
3431 the symbol and place it into the symbol table with a global scope.
3434 * Simple Assignments:: Simple Assignments
3437 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3438 * Source Code Reference:: How to use a linker script defined symbol in source code
3441 @node Simple Assignments
3442 @subsection Simple Assignments
3444 You may assign to a symbol using any of the C assignment operators:
3447 @item @var{symbol} = @var{expression} ;
3448 @itemx @var{symbol} += @var{expression} ;
3449 @itemx @var{symbol} -= @var{expression} ;
3450 @itemx @var{symbol} *= @var{expression} ;
3451 @itemx @var{symbol} /= @var{expression} ;
3452 @itemx @var{symbol} <<= @var{expression} ;
3453 @itemx @var{symbol} >>= @var{expression} ;
3454 @itemx @var{symbol} &= @var{expression} ;
3455 @itemx @var{symbol} |= @var{expression} ;
3458 The first case will define @var{symbol} to the value of
3459 @var{expression}. In the other cases, @var{symbol} must already be
3460 defined, and the value will be adjusted accordingly.
3462 The special symbol name @samp{.} indicates the location counter. You
3463 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3465 The semicolon after @var{expression} is required.
3467 Expressions are defined below; see @ref{Expressions}.
3469 You may write symbol assignments as commands in their own right, or as
3470 statements within a @code{SECTIONS} command, or as part of an output
3471 section description in a @code{SECTIONS} command.
3473 The section of the symbol will be set from the section of the
3474 expression; for more information, see @ref{Expression Section}.
3476 Here is an example showing the three different places that symbol
3477 assignments may be used:
3488 _bdata = (. + 3) & ~ 3;
3489 .data : @{ *(.data) @}
3493 In this example, the symbol @samp{floating_point} will be defined as
3494 zero. The symbol @samp{_etext} will be defined as the address following
3495 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3496 defined as the address following the @samp{.text} output section aligned
3497 upward to a 4 byte boundary.
3502 For ELF targeted ports, define a symbol that will be hidden and won't be
3503 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
3505 Here is the example from @ref{Simple Assignments}, rewritten to use
3509 HIDDEN(floating_point = 0);
3517 HIDDEN(_bdata = (. + 3) & ~ 3);
3518 .data : @{ *(.data) @}
3522 In this case none of the three symbols will be visible outside this module.
3527 In some cases, it is desirable for a linker script to define a symbol
3528 only if it is referenced and is not defined by any object included in
3529 the link. For example, traditional linkers defined the symbol
3530 @samp{etext}. However, ANSI C requires that the user be able to use
3531 @samp{etext} as a function name without encountering an error. The
3532 @code{PROVIDE} keyword may be used to define a symbol, such as
3533 @samp{etext}, only if it is referenced but not defined. The syntax is
3534 @code{PROVIDE(@var{symbol} = @var{expression})}.
3536 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3549 In this example, if the program defines @samp{_etext} (with a leading
3550 underscore), the linker will give a multiple definition error. If, on
3551 the other hand, the program defines @samp{etext} (with no leading
3552 underscore), the linker will silently use the definition in the program.
3553 If the program references @samp{etext} but does not define it, the
3554 linker will use the definition in the linker script.
3556 @node PROVIDE_HIDDEN
3557 @subsection PROVIDE_HIDDEN
3558 @cindex PROVIDE_HIDDEN
3559 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3560 hidden and won't be exported.
3562 @node Source Code Reference
3563 @subsection Source Code Reference
3565 Accessing a linker script defined variable from source code is not
3566 intuitive. In particular a linker script symbol is not equivalent to
3567 a variable declaration in a high level language, it is instead a
3568 symbol that does not have a value.
3570 Before going further, it is important to note that compilers often
3571 transform names in the source code into different names when they are
3572 stored in the symbol table. For example, Fortran compilers commonly
3573 prepend or append an underscore, and C++ performs extensive @samp{name
3574 mangling}. Therefore there might be a discrepancy between the name
3575 of a variable as it is used in source code and the name of the same
3576 variable as it is defined in a linker script. For example in C a
3577 linker script variable might be referred to as:
3583 But in the linker script it might be defined as:
3589 In the remaining examples however it is assumed that no name
3590 transformation has taken place.
3592 When a symbol is declared in a high level language such as C, two
3593 things happen. The first is that the compiler reserves enough space
3594 in the program's memory to hold the @emph{value} of the symbol. The
3595 second is that the compiler creates an entry in the program's symbol
3596 table which holds the symbol's @emph{address}. ie the symbol table
3597 contains the address of the block of memory holding the symbol's
3598 value. So for example the following C declaration, at file scope:
3604 creates a entry called @samp{foo} in the symbol table. This entry
3605 holds the address of an @samp{int} sized block of memory where the
3606 number 1000 is initially stored.
3608 When a program references a symbol the compiler generates code that
3609 first accesses the symbol table to find the address of the symbol's
3610 memory block and then code to read the value from that memory block.
3617 looks up the symbol @samp{foo} in the symbol table, gets the address
3618 associated with this symbol and then writes the value 1 into that
3625 looks up the symbol @samp{foo} in the symbol table, gets it address
3626 and then copies this address into the block of memory associated with
3627 the variable @samp{a}.
3629 Linker scripts symbol declarations, by contrast, create an entry in
3630 the symbol table but do not assign any memory to them. Thus they are
3631 an address without a value. So for example the linker script definition:
3637 creates an entry in the symbol table called @samp{foo} which holds
3638 the address of memory location 1000, but nothing special is stored at
3639 address 1000. This means that you cannot access the @emph{value} of a
3640 linker script defined symbol - it has no value - all you can do is
3641 access the @emph{address} of a linker script defined symbol.
3643 Hence when you are using a linker script defined symbol in source code
3644 you should always take the address of the symbol, and never attempt to
3645 use its value. For example suppose you want to copy the contents of a
3646 section of memory called .ROM into a section called .FLASH and the
3647 linker script contains these declarations:
3651 start_of_ROM = .ROM;
3652 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3653 start_of_FLASH = .FLASH;
3657 Then the C source code to perform the copy would be:
3661 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3663 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3667 Note the use of the @samp{&} operators. These are correct.
3670 @section SECTIONS Command
3672 The @code{SECTIONS} command tells the linker how to map input sections
3673 into output sections, and how to place the output sections in memory.
3675 The format of the @code{SECTIONS} command is:
3679 @var{sections-command}
3680 @var{sections-command}
3685 Each @var{sections-command} may of be one of the following:
3689 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3691 a symbol assignment (@pxref{Assignments})
3693 an output section description
3695 an overlay description
3698 The @code{ENTRY} command and symbol assignments are permitted inside the
3699 @code{SECTIONS} command for convenience in using the location counter in
3700 those commands. This can also make the linker script easier to
3701 understand because you can use those commands at meaningful points in
3702 the layout of the output file.
3704 Output section descriptions and overlay descriptions are described
3707 If you do not use a @code{SECTIONS} command in your linker script, the
3708 linker will place each input section into an identically named output
3709 section in the order that the sections are first encountered in the
3710 input files. If all input sections are present in the first file, for
3711 example, the order of sections in the output file will match the order
3712 in the first input file. The first section will be at address zero.
3715 * Output Section Description:: Output section description
3716 * Output Section Name:: Output section name
3717 * Output Section Address:: Output section address
3718 * Input Section:: Input section description
3719 * Output Section Data:: Output section data
3720 * Output Section Keywords:: Output section keywords
3721 * Output Section Discarding:: Output section discarding
3722 * Output Section Attributes:: Output section attributes
3723 * Overlay Description:: Overlay description
3726 @node Output Section Description
3727 @subsection Output Section Description
3728 The full description of an output section looks like this:
3731 @var{section} [@var{address}] [(@var{type})] :
3733 [ALIGN(@var{section_align})]
3734 [SUBALIGN(@var{subsection_align})]
3737 @var{output-section-command}
3738 @var{output-section-command}
3740 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3744 Most output sections do not use most of the optional section attributes.
3746 The whitespace around @var{section} is required, so that the section
3747 name is unambiguous. The colon and the curly braces are also required.
3748 The line breaks and other white space are optional.
3750 Each @var{output-section-command} may be one of the following:
3754 a symbol assignment (@pxref{Assignments})
3756 an input section description (@pxref{Input Section})
3758 data values to include directly (@pxref{Output Section Data})
3760 a special output section keyword (@pxref{Output Section Keywords})
3763 @node Output Section Name
3764 @subsection Output Section Name
3765 @cindex name, section
3766 @cindex section name
3767 The name of the output section is @var{section}. @var{section} must
3768 meet the constraints of your output format. In formats which only
3769 support a limited number of sections, such as @code{a.out}, the name
3770 must be one of the names supported by the format (@code{a.out}, for
3771 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3772 output format supports any number of sections, but with numbers and not
3773 names (as is the case for Oasys), the name should be supplied as a
3774 quoted numeric string. A section name may consist of any sequence of
3775 characters, but a name which contains any unusual characters such as
3776 commas must be quoted.
3778 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3781 @node Output Section Address
3782 @subsection Output Section Address
3783 @cindex address, section
3784 @cindex section address
3785 The @var{address} is an expression for the VMA (the virtual memory
3786 address) of the output section. This address is optional, but if it
3787 is provided then the output address will be set exactly as specified.
3789 If the output address is not specified then one will be chosen for the
3790 section, based on the heuristic below. This address will be adjusted
3791 to fit the alignment requirement of the output section. The
3792 alignment requirement is the strictest alignment of any input section
3793 contained within the output section.
3795 The output section address heuristic is as follows:
3799 If an output memory @var{region} is set for the section then it
3800 is added to this region and its address will be the next free address
3804 If the MEMORY command has been used to create a list of memory
3805 regions then the first region which has attributes compatible with the
3806 section is selected to contain it. The section's output address will
3807 be the next free address in that region; @ref{MEMORY}.
3810 If no memory regions were specified, or none match the section then
3811 the output address will be based on the current value of the location
3819 .text . : @{ *(.text) @}
3826 .text : @{ *(.text) @}
3830 are subtly different. The first will set the address of the
3831 @samp{.text} output section to the current value of the location
3832 counter. The second will set it to the current value of the location
3833 counter aligned to the strictest alignment of any of the @samp{.text}
3836 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3837 For example, if you want to align the section on a 0x10 byte boundary,
3838 so that the lowest four bits of the section address are zero, you could
3839 do something like this:
3841 .text ALIGN(0x10) : @{ *(.text) @}
3844 This works because @code{ALIGN} returns the current location counter
3845 aligned upward to the specified value.
3847 Specifying @var{address} for a section will change the value of the
3848 location counter, provided that the section is non-empty. (Empty
3849 sections are ignored).
3852 @subsection Input Section Description
3853 @cindex input sections
3854 @cindex mapping input sections to output sections
3855 The most common output section command is an input section description.
3857 The input section description is the most basic linker script operation.
3858 You use output sections to tell the linker how to lay out your program
3859 in memory. You use input section descriptions to tell the linker how to
3860 map the input files into your memory layout.
3863 * Input Section Basics:: Input section basics
3864 * Input Section Wildcards:: Input section wildcard patterns
3865 * Input Section Common:: Input section for common symbols
3866 * Input Section Keep:: Input section and garbage collection
3867 * Input Section Example:: Input section example
3870 @node Input Section Basics
3871 @subsubsection Input Section Basics
3872 @cindex input section basics
3873 An input section description consists of a file name optionally followed
3874 by a list of section names in parentheses.
3876 The file name and the section name may be wildcard patterns, which we
3877 describe further below (@pxref{Input Section Wildcards}).
3879 The most common input section description is to include all input
3880 sections with a particular name in the output section. For example, to
3881 include all input @samp{.text} sections, you would write:
3886 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3887 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3888 match all files except the ones specified in the EXCLUDE_FILE list. For
3891 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3893 will cause all .ctors sections from all files except @file{crtend.o} and
3894 @file{otherfile.o} to be included.
3896 There are two ways to include more than one section:
3902 The difference between these is the order in which the @samp{.text} and
3903 @samp{.rdata} input sections will appear in the output section. In the
3904 first example, they will be intermingled, appearing in the same order as
3905 they are found in the linker input. In the second example, all
3906 @samp{.text} input sections will appear first, followed by all
3907 @samp{.rdata} input sections.
3909 You can specify a file name to include sections from a particular file.
3910 You would do this if one or more of your files contain special data that
3911 needs to be at a particular location in memory. For example:
3916 To refine the sections that are included based on the section flags
3917 of an input section, INPUT_SECTION_FLAGS may be used.
3919 Here is a simple example for using Section header flags for ELF sections:
3924 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
3925 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
3930 In this example, the output section @samp{.text} will be comprised of any
3931 input section matching the name *(.text) whose section header flags
3932 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
3933 @samp{.text2} will be comprised of any input section matching the name *(.text)
3934 whose section header flag @code{SHF_WRITE} is clear.
3936 You can also specify files within archives by writing a pattern
3937 matching the archive, a colon, then the pattern matching the file,
3938 with no whitespace around the colon.
3942 matches file within archive
3944 matches the whole archive
3946 matches file but not one in an archive
3949 Either one or both of @samp{archive} and @samp{file} can contain shell
3950 wildcards. On DOS based file systems, the linker will assume that a
3951 single letter followed by a colon is a drive specifier, so
3952 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
3953 within an archive called @samp{c}. @samp{archive:file} filespecs may
3954 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
3955 other linker script contexts. For instance, you cannot extract a file
3956 from an archive by using @samp{archive:file} in an @code{INPUT}
3959 If you use a file name without a list of sections, then all sections in
3960 the input file will be included in the output section. This is not
3961 commonly done, but it may by useful on occasion. For example:
3966 When you use a file name which is not an @samp{archive:file} specifier
3967 and does not contain any wild card
3968 characters, the linker will first see if you also specified the file
3969 name on the linker command line or in an @code{INPUT} command. If you
3970 did not, the linker will attempt to open the file as an input file, as
3971 though it appeared on the command line. Note that this differs from an
3972 @code{INPUT} command, because the linker will not search for the file in
3973 the archive search path.
3975 @node Input Section Wildcards
3976 @subsubsection Input Section Wildcard Patterns
3977 @cindex input section wildcards
3978 @cindex wildcard file name patterns
3979 @cindex file name wildcard patterns
3980 @cindex section name wildcard patterns
3981 In an input section description, either the file name or the section
3982 name or both may be wildcard patterns.
3984 The file name of @samp{*} seen in many examples is a simple wildcard
3985 pattern for the file name.
3987 The wildcard patterns are like those used by the Unix shell.
3991 matches any number of characters
3993 matches any single character
3995 matches a single instance of any of the @var{chars}; the @samp{-}
3996 character may be used to specify a range of characters, as in
3997 @samp{[a-z]} to match any lower case letter
3999 quotes the following character
4002 When a file name is matched with a wildcard, the wildcard characters
4003 will not match a @samp{/} character (used to separate directory names on
4004 Unix). A pattern consisting of a single @samp{*} character is an
4005 exception; it will always match any file name, whether it contains a
4006 @samp{/} or not. In a section name, the wildcard characters will match
4007 a @samp{/} character.
4009 File name wildcard patterns only match files which are explicitly
4010 specified on the command line or in an @code{INPUT} command. The linker
4011 does not search directories to expand wildcards.
4013 If a file name matches more than one wildcard pattern, or if a file name
4014 appears explicitly and is also matched by a wildcard pattern, the linker
4015 will use the first match in the linker script. For example, this
4016 sequence of input section descriptions is probably in error, because the
4017 @file{data.o} rule will not be used:
4019 .data : @{ *(.data) @}
4020 .data1 : @{ data.o(.data) @}
4023 @cindex SORT_BY_NAME
4024 Normally, the linker will place files and sections matched by wildcards
4025 in the order in which they are seen during the link. You can change
4026 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4027 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4028 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4029 into ascending order by name before placing them in the output file.
4031 @cindex SORT_BY_ALIGNMENT
4032 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4033 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4034 descending order by alignment before placing them in the output file.
4035 Larger alignments are placed before smaller alignments in order to
4036 reduce the amount of padding necessary.
4038 @cindex SORT_BY_INIT_PRIORITY
4039 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4040 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4041 ascending order by numerical value of the GCC init_priority attribute
4042 encoded in the section name before placing them in the output file.
4045 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4047 When there are nested section sorting commands in linker script, there
4048 can be at most 1 level of nesting for section sorting commands.
4052 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4053 It will sort the input sections by name first, then by alignment if two
4054 sections have the same name.
4056 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4057 It will sort the input sections by alignment first, then by name if two
4058 sections have the same alignment.
4060 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4061 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4063 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4064 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4066 All other nested section sorting commands are invalid.
4069 When both command line section sorting option and linker script
4070 section sorting command are used, section sorting command always
4071 takes precedence over the command line option.
4073 If the section sorting command in linker script isn't nested, the
4074 command line option will make the section sorting command to be
4075 treated as nested sorting command.
4079 @code{SORT_BY_NAME} (wildcard section pattern ) with
4080 @option{--sort-sections alignment} is equivalent to
4081 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4083 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4084 @option{--sort-section name} is equivalent to
4085 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4088 If the section sorting command in linker script is nested, the
4089 command line option will be ignored.
4092 @code{SORT_NONE} disables section sorting by ignoring the command line
4093 section sorting option.
4095 If you ever get confused about where input sections are going, use the
4096 @samp{-M} linker option to generate a map file. The map file shows
4097 precisely how input sections are mapped to output sections.
4099 This example shows how wildcard patterns might be used to partition
4100 files. This linker script directs the linker to place all @samp{.text}
4101 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4102 The linker will place the @samp{.data} section from all files beginning
4103 with an upper case character in @samp{.DATA}; for all other files, the
4104 linker will place the @samp{.data} section in @samp{.data}.
4108 .text : @{ *(.text) @}
4109 .DATA : @{ [A-Z]*(.data) @}
4110 .data : @{ *(.data) @}
4111 .bss : @{ *(.bss) @}
4116 @node Input Section Common
4117 @subsubsection Input Section for Common Symbols
4118 @cindex common symbol placement
4119 @cindex uninitialized data placement
4120 A special notation is needed for common symbols, because in many object
4121 file formats common symbols do not have a particular input section. The
4122 linker treats common symbols as though they are in an input section
4123 named @samp{COMMON}.
4125 You may use file names with the @samp{COMMON} section just as with any
4126 other input sections. You can use this to place common symbols from a
4127 particular input file in one section while common symbols from other
4128 input files are placed in another section.
4130 In most cases, common symbols in input files will be placed in the
4131 @samp{.bss} section in the output file. For example:
4133 .bss @{ *(.bss) *(COMMON) @}
4136 @cindex scommon section
4137 @cindex small common symbols
4138 Some object file formats have more than one type of common symbol. For
4139 example, the MIPS ELF object file format distinguishes standard common
4140 symbols and small common symbols. In this case, the linker will use a
4141 different special section name for other types of common symbols. In
4142 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4143 symbols and @samp{.scommon} for small common symbols. This permits you
4144 to map the different types of common symbols into memory at different
4148 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4149 notation is now considered obsolete. It is equivalent to
4152 @node Input Section Keep
4153 @subsubsection Input Section and Garbage Collection
4155 @cindex garbage collection
4156 When link-time garbage collection is in use (@samp{--gc-sections}),
4157 it is often useful to mark sections that should not be eliminated.
4158 This is accomplished by surrounding an input section's wildcard entry
4159 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4160 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4162 @node Input Section Example
4163 @subsubsection Input Section Example
4164 The following example is a complete linker script. It tells the linker
4165 to read all of the sections from file @file{all.o} and place them at the
4166 start of output section @samp{outputa} which starts at location
4167 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4168 follows immediately, in the same output section. All of section
4169 @samp{.input2} from @file{foo.o} goes into output section
4170 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4171 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4172 files are written to output section @samp{outputc}.
4200 @node Output Section Data
4201 @subsection Output Section Data
4203 @cindex section data
4204 @cindex output section data
4205 @kindex BYTE(@var{expression})
4206 @kindex SHORT(@var{expression})
4207 @kindex LONG(@var{expression})
4208 @kindex QUAD(@var{expression})
4209 @kindex SQUAD(@var{expression})
4210 You can include explicit bytes of data in an output section by using
4211 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4212 an output section command. Each keyword is followed by an expression in
4213 parentheses providing the value to store (@pxref{Expressions}). The
4214 value of the expression is stored at the current value of the location
4217 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4218 store one, two, four, and eight bytes (respectively). After storing the
4219 bytes, the location counter is incremented by the number of bytes
4222 For example, this will store the byte 1 followed by the four byte value
4223 of the symbol @samp{addr}:
4229 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4230 same; they both store an 8 byte, or 64 bit, value. When both host and
4231 target are 32 bits, an expression is computed as 32 bits. In this case
4232 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4233 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4235 If the object file format of the output file has an explicit endianness,
4236 which is the normal case, the value will be stored in that endianness.
4237 When the object file format does not have an explicit endianness, as is
4238 true of, for example, S-records, the value will be stored in the
4239 endianness of the first input object file.
4241 Note---these commands only work inside a section description and not
4242 between them, so the following will produce an error from the linker:
4244 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4246 whereas this will work:
4248 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4251 @kindex FILL(@var{expression})
4252 @cindex holes, filling
4253 @cindex unspecified memory
4254 You may use the @code{FILL} command to set the fill pattern for the
4255 current section. It is followed by an expression in parentheses. Any
4256 otherwise unspecified regions of memory within the section (for example,
4257 gaps left due to the required alignment of input sections) are filled
4258 with the value of the expression, repeated as
4259 necessary. A @code{FILL} statement covers memory locations after the
4260 point at which it occurs in the section definition; by including more
4261 than one @code{FILL} statement, you can have different fill patterns in
4262 different parts of an output section.
4264 This example shows how to fill unspecified regions of memory with the
4270 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4271 section attribute, but it only affects the
4272 part of the section following the @code{FILL} command, rather than the
4273 entire section. If both are used, the @code{FILL} command takes
4274 precedence. @xref{Output Section Fill}, for details on the fill
4277 @node Output Section Keywords
4278 @subsection Output Section Keywords
4279 There are a couple of keywords which can appear as output section
4283 @kindex CREATE_OBJECT_SYMBOLS
4284 @cindex input filename symbols
4285 @cindex filename symbols
4286 @item CREATE_OBJECT_SYMBOLS
4287 The command tells the linker to create a symbol for each input file.
4288 The name of each symbol will be the name of the corresponding input
4289 file. The section of each symbol will be the output section in which
4290 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4292 This is conventional for the a.out object file format. It is not
4293 normally used for any other object file format.
4295 @kindex CONSTRUCTORS
4296 @cindex C++ constructors, arranging in link
4297 @cindex constructors, arranging in link
4299 When linking using the a.out object file format, the linker uses an
4300 unusual set construct to support C++ global constructors and
4301 destructors. When linking object file formats which do not support
4302 arbitrary sections, such as ECOFF and XCOFF, the linker will
4303 automatically recognize C++ global constructors and destructors by name.
4304 For these object file formats, the @code{CONSTRUCTORS} command tells the
4305 linker to place constructor information in the output section where the
4306 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4307 ignored for other object file formats.
4309 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4310 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4311 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4312 the start and end of the global destructors. The
4313 first word in the list is the number of entries, followed by the address
4314 of each constructor or destructor, followed by a zero word. The
4315 compiler must arrange to actually run the code. For these object file
4316 formats @sc{gnu} C++ normally calls constructors from a subroutine
4317 @code{__main}; a call to @code{__main} is automatically inserted into
4318 the startup code for @code{main}. @sc{gnu} C++ normally runs
4319 destructors either by using @code{atexit}, or directly from the function
4322 For object file formats such as @code{COFF} or @code{ELF} which support
4323 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4324 addresses of global constructors and destructors into the @code{.ctors}
4325 and @code{.dtors} sections. Placing the following sequence into your
4326 linker script will build the sort of table which the @sc{gnu} C++
4327 runtime code expects to see.
4331 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4336 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4342 If you are using the @sc{gnu} C++ support for initialization priority,
4343 which provides some control over the order in which global constructors
4344 are run, you must sort the constructors at link time to ensure that they
4345 are executed in the correct order. When using the @code{CONSTRUCTORS}
4346 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4347 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4348 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4351 Normally the compiler and linker will handle these issues automatically,
4352 and you will not need to concern yourself with them. However, you may
4353 need to consider this if you are using C++ and writing your own linker
4358 @node Output Section Discarding
4359 @subsection Output Section Discarding
4360 @cindex discarding sections
4361 @cindex sections, discarding
4362 @cindex removing sections
4363 The linker will not create output sections with no contents. This is
4364 for convenience when referring to input sections that may or may not
4365 be present in any of the input files. For example:
4367 .foo : @{ *(.foo) @}
4370 will only create a @samp{.foo} section in the output file if there is a
4371 @samp{.foo} section in at least one input file, and if the input
4372 sections are not all empty. Other link script directives that allocate
4373 space in an output section will also create the output section.
4375 The linker will ignore address assignments (@pxref{Output Section Address})
4376 on discarded output sections, except when the linker script defines
4377 symbols in the output section. In that case the linker will obey
4378 the address assignments, possibly advancing dot even though the
4379 section is discarded.
4382 The special output section name @samp{/DISCARD/} may be used to discard
4383 input sections. Any input sections which are assigned to an output
4384 section named @samp{/DISCARD/} are not included in the output file.
4386 @node Output Section Attributes
4387 @subsection Output Section Attributes
4388 @cindex output section attributes
4389 We showed above that the full description of an output section looked
4394 @var{section} [@var{address}] [(@var{type})] :
4396 [ALIGN(@var{section_align})]
4397 [SUBALIGN(@var{subsection_align})]
4400 @var{output-section-command}
4401 @var{output-section-command}
4403 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4407 We've already described @var{section}, @var{address}, and
4408 @var{output-section-command}. In this section we will describe the
4409 remaining section attributes.
4412 * Output Section Type:: Output section type
4413 * Output Section LMA:: Output section LMA
4414 * Forced Output Alignment:: Forced Output Alignment
4415 * Forced Input Alignment:: Forced Input Alignment
4416 * Output Section Constraint:: Output section constraint
4417 * Output Section Region:: Output section region
4418 * Output Section Phdr:: Output section phdr
4419 * Output Section Fill:: Output section fill
4422 @node Output Section Type
4423 @subsubsection Output Section Type
4424 Each output section may have a type. The type is a keyword in
4425 parentheses. The following types are defined:
4429 The section should be marked as not loadable, so that it will not be
4430 loaded into memory when the program is run.
4435 These type names are supported for backward compatibility, and are
4436 rarely used. They all have the same effect: the section should be
4437 marked as not allocatable, so that no memory is allocated for the
4438 section when the program is run.
4442 @cindex prevent unnecessary loading
4443 @cindex loading, preventing
4444 The linker normally sets the attributes of an output section based on
4445 the input sections which map into it. You can override this by using
4446 the section type. For example, in the script sample below, the
4447 @samp{ROM} section is addressed at memory location @samp{0} and does not
4448 need to be loaded when the program is run.
4452 ROM 0 (NOLOAD) : @{ @dots{} @}
4458 @node Output Section LMA
4459 @subsubsection Output Section LMA
4460 @kindex AT>@var{lma_region}
4461 @kindex AT(@var{lma})
4462 @cindex load address
4463 @cindex section load address
4464 Every section has a virtual address (VMA) and a load address (LMA); see
4465 @ref{Basic Script Concepts}. The virtual address is specified by the
4466 @pxref{Output Section Address} described earlier. The load address is
4467 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
4468 address is optional.
4470 The @code{AT} keyword takes an expression as an argument. This
4471 specifies the exact load address of the section. The @code{AT>} keyword
4472 takes the name of a memory region as an argument. @xref{MEMORY}. The
4473 load address of the section is set to the next free address in the
4474 region, aligned to the section's alignment requirements.
4476 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4477 section, the linker will use the following heuristic to determine the
4482 If the section has a specific VMA address, then this is used as
4483 the LMA address as well.
4486 If the section is not allocatable then its LMA is set to its VMA.
4489 Otherwise if a memory region can be found that is compatible
4490 with the current section, and this region contains at least one
4491 section, then the LMA is set so the difference between the
4492 VMA and LMA is the same as the difference between the VMA and LMA of
4493 the last section in the located region.
4496 If no memory regions have been declared then a default region
4497 that covers the entire address space is used in the previous step.
4500 If no suitable region could be found, or there was no previous
4501 section then the LMA is set equal to the VMA.
4504 @cindex ROM initialized data
4505 @cindex initialized data in ROM
4506 This feature is designed to make it easy to build a ROM image. For
4507 example, the following linker script creates three output sections: one
4508 called @samp{.text}, which starts at @code{0x1000}, one called
4509 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
4510 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
4511 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
4512 defined with the value @code{0x2000}, which shows that the location
4513 counter holds the VMA value, not the LMA value.
4519 .text 0x1000 : @{ *(.text) _etext = . ; @}
4521 AT ( ADDR (.text) + SIZEOF (.text) )
4522 @{ _data = . ; *(.data); _edata = . ; @}
4524 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
4529 The run-time initialization code for use with a program generated with
4530 this linker script would include something like the following, to copy
4531 the initialized data from the ROM image to its runtime address. Notice
4532 how this code takes advantage of the symbols defined by the linker
4537 extern char _etext, _data, _edata, _bstart, _bend;
4538 char *src = &_etext;
4541 /* ROM has data at end of text; copy it. */
4542 while (dst < &_edata)
4546 for (dst = &_bstart; dst< &_bend; dst++)
4551 @node Forced Output Alignment
4552 @subsubsection Forced Output Alignment
4553 @kindex ALIGN(@var{section_align})
4554 @cindex forcing output section alignment
4555 @cindex output section alignment
4556 You can increase an output section's alignment by using ALIGN.
4558 @node Forced Input Alignment
4559 @subsubsection Forced Input Alignment
4560 @kindex SUBALIGN(@var{subsection_align})
4561 @cindex forcing input section alignment
4562 @cindex input section alignment
4563 You can force input section alignment within an output section by using
4564 SUBALIGN. The value specified overrides any alignment given by input
4565 sections, whether larger or smaller.
4567 @node Output Section Constraint
4568 @subsubsection Output Section Constraint
4571 @cindex constraints on output sections
4572 You can specify that an output section should only be created if all
4573 of its input sections are read-only or all of its input sections are
4574 read-write by using the keyword @code{ONLY_IF_RO} and
4575 @code{ONLY_IF_RW} respectively.
4577 @node Output Section Region
4578 @subsubsection Output Section Region
4579 @kindex >@var{region}
4580 @cindex section, assigning to memory region
4581 @cindex memory regions and sections
4582 You can assign a section to a previously defined region of memory by
4583 using @samp{>@var{region}}. @xref{MEMORY}.
4585 Here is a simple example:
4588 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4589 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4593 @node Output Section Phdr
4594 @subsubsection Output Section Phdr
4596 @cindex section, assigning to program header
4597 @cindex program headers and sections
4598 You can assign a section to a previously defined program segment by
4599 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4600 one or more segments, then all subsequent allocated sections will be
4601 assigned to those segments as well, unless they use an explicitly
4602 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4603 linker to not put the section in any segment at all.
4605 Here is a simple example:
4608 PHDRS @{ text PT_LOAD ; @}
4609 SECTIONS @{ .text : @{ *(.text) @} :text @}
4613 @node Output Section Fill
4614 @subsubsection Output Section Fill
4615 @kindex =@var{fillexp}
4616 @cindex section fill pattern
4617 @cindex fill pattern, entire section
4618 You can set the fill pattern for an entire section by using
4619 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4620 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4621 within the output section (for example, gaps left due to the required
4622 alignment of input sections) will be filled with the value, repeated as
4623 necessary. If the fill expression is a simple hex number, ie. a string
4624 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4625 an arbitrarily long sequence of hex digits can be used to specify the
4626 fill pattern; Leading zeros become part of the pattern too. For all
4627 other cases, including extra parentheses or a unary @code{+}, the fill
4628 pattern is the four least significant bytes of the value of the
4629 expression. In all cases, the number is big-endian.
4631 You can also change the fill value with a @code{FILL} command in the
4632 output section commands; (@pxref{Output Section Data}).
4634 Here is a simple example:
4637 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4641 @node Overlay Description
4642 @subsection Overlay Description
4645 An overlay description provides an easy way to describe sections which
4646 are to be loaded as part of a single memory image but are to be run at
4647 the same memory address. At run time, some sort of overlay manager will
4648 copy the overlaid sections in and out of the runtime memory address as
4649 required, perhaps by simply manipulating addressing bits. This approach
4650 can be useful, for example, when a certain region of memory is faster
4653 Overlays are described using the @code{OVERLAY} command. The
4654 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4655 output section description. The full syntax of the @code{OVERLAY}
4656 command is as follows:
4659 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4663 @var{output-section-command}
4664 @var{output-section-command}
4666 @} [:@var{phdr}@dots{}] [=@var{fill}]
4669 @var{output-section-command}
4670 @var{output-section-command}
4672 @} [:@var{phdr}@dots{}] [=@var{fill}]
4674 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4678 Everything is optional except @code{OVERLAY} (a keyword), and each
4679 section must have a name (@var{secname1} and @var{secname2} above). The
4680 section definitions within the @code{OVERLAY} construct are identical to
4681 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
4682 except that no addresses and no memory regions may be defined for
4683 sections within an @code{OVERLAY}.
4685 The sections are all defined with the same starting address. The load
4686 addresses of the sections are arranged such that they are consecutive in
4687 memory starting at the load address used for the @code{OVERLAY} as a
4688 whole (as with normal section definitions, the load address is optional,
4689 and defaults to the start address; the start address is also optional,
4690 and defaults to the current value of the location counter).
4692 If the @code{NOCROSSREFS} keyword is used, and there are any
4693 references among the sections, the linker will report an error. Since
4694 the sections all run at the same address, it normally does not make
4695 sense for one section to refer directly to another.
4696 @xref{Miscellaneous Commands, NOCROSSREFS}.
4698 For each section within the @code{OVERLAY}, the linker automatically
4699 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4700 defined as the starting load address of the section. The symbol
4701 @code{__load_stop_@var{secname}} is defined as the final load address of
4702 the section. Any characters within @var{secname} which are not legal
4703 within C identifiers are removed. C (or assembler) code may use these
4704 symbols to move the overlaid sections around as necessary.
4706 At the end of the overlay, the value of the location counter is set to
4707 the start address of the overlay plus the size of the largest section.
4709 Here is an example. Remember that this would appear inside a
4710 @code{SECTIONS} construct.
4713 OVERLAY 0x1000 : AT (0x4000)
4715 .text0 @{ o1/*.o(.text) @}
4716 .text1 @{ o2/*.o(.text) @}
4721 This will define both @samp{.text0} and @samp{.text1} to start at
4722 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4723 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4724 following symbols will be defined if referenced: @code{__load_start_text0},
4725 @code{__load_stop_text0}, @code{__load_start_text1},
4726 @code{__load_stop_text1}.
4728 C code to copy overlay @code{.text1} into the overlay area might look
4733 extern char __load_start_text1, __load_stop_text1;
4734 memcpy ((char *) 0x1000, &__load_start_text1,
4735 &__load_stop_text1 - &__load_start_text1);
4739 Note that the @code{OVERLAY} command is just syntactic sugar, since
4740 everything it does can be done using the more basic commands. The above
4741 example could have been written identically as follows.
4745 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4746 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4747 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4748 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4749 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4750 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4751 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4756 @section MEMORY Command
4758 @cindex memory regions
4759 @cindex regions of memory
4760 @cindex allocating memory
4761 @cindex discontinuous memory
4762 The linker's default configuration permits allocation of all available
4763 memory. You can override this by using the @code{MEMORY} command.
4765 The @code{MEMORY} command describes the location and size of blocks of
4766 memory in the target. You can use it to describe which memory regions
4767 may be used by the linker, and which memory regions it must avoid. You
4768 can then assign sections to particular memory regions. The linker will
4769 set section addresses based on the memory regions, and will warn about
4770 regions that become too full. The linker will not shuffle sections
4771 around to fit into the available regions.
4773 A linker script may contain at most one use of the @code{MEMORY}
4774 command. However, you can define as many blocks of memory within it as
4775 you wish. The syntax is:
4780 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4786 The @var{name} is a name used in the linker script to refer to the
4787 region. The region name has no meaning outside of the linker script.
4788 Region names are stored in a separate name space, and will not conflict
4789 with symbol names, file names, or section names. Each memory region
4790 must have a distinct name within the @code{MEMORY} command. However you can
4791 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
4794 @cindex memory region attributes
4795 The @var{attr} string is an optional list of attributes that specify
4796 whether to use a particular memory region for an input section which is
4797 not explicitly mapped in the linker script. As described in
4798 @ref{SECTIONS}, if you do not specify an output section for some input
4799 section, the linker will create an output section with the same name as
4800 the input section. If you define region attributes, the linker will use
4801 them to select the memory region for the output section that it creates.
4803 The @var{attr} string must consist only of the following characters:
4818 Invert the sense of any of the attributes that follow
4821 If a unmapped section matches any of the listed attributes other than
4822 @samp{!}, it will be placed in the memory region. The @samp{!}
4823 attribute reverses this test, so that an unmapped section will be placed
4824 in the memory region only if it does not match any of the listed
4830 The @var{origin} is an numerical expression for the start address of
4831 the memory region. The expression must evaluate to a constant and it
4832 cannot involve any symbols. The keyword @code{ORIGIN} may be
4833 abbreviated to @code{org} or @code{o} (but not, for example,
4839 The @var{len} is an expression for the size in bytes of the memory
4840 region. As with the @var{origin} expression, the expression must
4841 be numerical only and must evaluate to a constant. The keyword
4842 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4844 In the following example, we specify that there are two memory regions
4845 available for allocation: one starting at @samp{0} for 256 kilobytes,
4846 and the other starting at @samp{0x40000000} for four megabytes. The
4847 linker will place into the @samp{rom} memory region every section which
4848 is not explicitly mapped into a memory region, and is either read-only
4849 or executable. The linker will place other sections which are not
4850 explicitly mapped into a memory region into the @samp{ram} memory
4857 rom (rx) : ORIGIN = 0, LENGTH = 256K
4858 ram (!rx) : org = 0x40000000, l = 4M
4863 Once you define a memory region, you can direct the linker to place
4864 specific output sections into that memory region by using the
4865 @samp{>@var{region}} output section attribute. For example, if you have
4866 a memory region named @samp{mem}, you would use @samp{>mem} in the
4867 output section definition. @xref{Output Section Region}. If no address
4868 was specified for the output section, the linker will set the address to
4869 the next available address within the memory region. If the combined
4870 output sections directed to a memory region are too large for the
4871 region, the linker will issue an error message.
4873 It is possible to access the origin and length of a memory in an
4874 expression via the @code{ORIGIN(@var{memory})} and
4875 @code{LENGTH(@var{memory})} functions:
4879 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4884 @section PHDRS Command
4886 @cindex program headers
4887 @cindex ELF program headers
4888 @cindex program segments
4889 @cindex segments, ELF
4890 The ELF object file format uses @dfn{program headers}, also knows as
4891 @dfn{segments}. The program headers describe how the program should be
4892 loaded into memory. You can print them out by using the @code{objdump}
4893 program with the @samp{-p} option.
4895 When you run an ELF program on a native ELF system, the system loader
4896 reads the program headers in order to figure out how to load the
4897 program. This will only work if the program headers are set correctly.
4898 This manual does not describe the details of how the system loader
4899 interprets program headers; for more information, see the ELF ABI.
4901 The linker will create reasonable program headers by default. However,
4902 in some cases, you may need to specify the program headers more
4903 precisely. You may use the @code{PHDRS} command for this purpose. When
4904 the linker sees the @code{PHDRS} command in the linker script, it will
4905 not create any program headers other than the ones specified.
4907 The linker only pays attention to the @code{PHDRS} command when
4908 generating an ELF output file. In other cases, the linker will simply
4909 ignore @code{PHDRS}.
4911 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4912 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4918 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4919 [ FLAGS ( @var{flags} ) ] ;
4924 The @var{name} is used only for reference in the @code{SECTIONS} command
4925 of the linker script. It is not put into the output file. Program
4926 header names are stored in a separate name space, and will not conflict
4927 with symbol names, file names, or section names. Each program header
4928 must have a distinct name. The headers are processed in order and it
4929 is usual for them to map to sections in ascending load address order.
4931 Certain program header types describe segments of memory which the
4932 system loader will load from the file. In the linker script, you
4933 specify the contents of these segments by placing allocatable output
4934 sections in the segments. You use the @samp{:@var{phdr}} output section
4935 attribute to place a section in a particular segment. @xref{Output
4938 It is normal to put certain sections in more than one segment. This
4939 merely implies that one segment of memory contains another. You may
4940 repeat @samp{:@var{phdr}}, using it once for each segment which should
4941 contain the section.
4943 If you place a section in one or more segments using @samp{:@var{phdr}},
4944 then the linker will place all subsequent allocatable sections which do
4945 not specify @samp{:@var{phdr}} in the same segments. This is for
4946 convenience, since generally a whole set of contiguous sections will be
4947 placed in a single segment. You can use @code{:NONE} to override the
4948 default segment and tell the linker to not put the section in any
4953 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
4954 the program header type to further describe the contents of the segment.
4955 The @code{FILEHDR} keyword means that the segment should include the ELF
4956 file header. The @code{PHDRS} keyword means that the segment should
4957 include the ELF program headers themselves. If applied to a loadable
4958 segment (@code{PT_LOAD}), all prior loadable segments must have one of
4961 The @var{type} may be one of the following. The numbers indicate the
4962 value of the keyword.
4965 @item @code{PT_NULL} (0)
4966 Indicates an unused program header.
4968 @item @code{PT_LOAD} (1)
4969 Indicates that this program header describes a segment to be loaded from
4972 @item @code{PT_DYNAMIC} (2)
4973 Indicates a segment where dynamic linking information can be found.
4975 @item @code{PT_INTERP} (3)
4976 Indicates a segment where the name of the program interpreter may be
4979 @item @code{PT_NOTE} (4)
4980 Indicates a segment holding note information.
4982 @item @code{PT_SHLIB} (5)
4983 A reserved program header type, defined but not specified by the ELF
4986 @item @code{PT_PHDR} (6)
4987 Indicates a segment where the program headers may be found.
4989 @item @var{expression}
4990 An expression giving the numeric type of the program header. This may
4991 be used for types not defined above.
4994 You can specify that a segment should be loaded at a particular address
4995 in memory by using an @code{AT} expression. This is identical to the
4996 @code{AT} command used as an output section attribute (@pxref{Output
4997 Section LMA}). The @code{AT} command for a program header overrides the
4998 output section attribute.
5000 The linker will normally set the segment flags based on the sections
5001 which comprise the segment. You may use the @code{FLAGS} keyword to
5002 explicitly specify the segment flags. The value of @var{flags} must be
5003 an integer. It is used to set the @code{p_flags} field of the program
5006 Here is an example of @code{PHDRS}. This shows a typical set of program
5007 headers used on a native ELF system.
5013 headers PT_PHDR PHDRS ;
5015 text PT_LOAD FILEHDR PHDRS ;
5017 dynamic PT_DYNAMIC ;
5023 .interp : @{ *(.interp) @} :text :interp
5024 .text : @{ *(.text) @} :text
5025 .rodata : @{ *(.rodata) @} /* defaults to :text */
5027 . = . + 0x1000; /* move to a new page in memory */
5028 .data : @{ *(.data) @} :data
5029 .dynamic : @{ *(.dynamic) @} :data :dynamic
5036 @section VERSION Command
5037 @kindex VERSION @{script text@}
5038 @cindex symbol versions
5039 @cindex version script
5040 @cindex versions of symbols
5041 The linker supports symbol versions when using ELF. Symbol versions are
5042 only useful when using shared libraries. The dynamic linker can use
5043 symbol versions to select a specific version of a function when it runs
5044 a program that may have been linked against an earlier version of the
5047 You can include a version script directly in the main linker script, or
5048 you can supply the version script as an implicit linker script. You can
5049 also use the @samp{--version-script} linker option.
5051 The syntax of the @code{VERSION} command is simply
5053 VERSION @{ version-script-commands @}
5056 The format of the version script commands is identical to that used by
5057 Sun's linker in Solaris 2.5. The version script defines a tree of
5058 version nodes. You specify the node names and interdependencies in the
5059 version script. You can specify which symbols are bound to which
5060 version nodes, and you can reduce a specified set of symbols to local
5061 scope so that they are not globally visible outside of the shared
5064 The easiest way to demonstrate the version script language is with a few
5090 This example version script defines three version nodes. The first
5091 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5092 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5093 a number of symbols to local scope so that they are not visible outside
5094 of the shared library; this is done using wildcard patterns, so that any
5095 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5096 is matched. The wildcard patterns available are the same as those used
5097 in the shell when matching filenames (also known as ``globbing'').
5098 However, if you specify the symbol name inside double quotes, then the
5099 name is treated as literal, rather than as a glob pattern.
5101 Next, the version script defines node @samp{VERS_1.2}. This node
5102 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5103 to the version node @samp{VERS_1.2}.
5105 Finally, the version script defines node @samp{VERS_2.0}. This node
5106 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5107 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5109 When the linker finds a symbol defined in a library which is not
5110 specifically bound to a version node, it will effectively bind it to an
5111 unspecified base version of the library. You can bind all otherwise
5112 unspecified symbols to a given version node by using @samp{global: *;}
5113 somewhere in the version script. Note that it's slightly crazy to use
5114 wildcards in a global spec except on the last version node. Global
5115 wildcards elsewhere run the risk of accidentally adding symbols to the
5116 set exported for an old version. That's wrong since older versions
5117 ought to have a fixed set of symbols.
5119 The names of the version nodes have no specific meaning other than what
5120 they might suggest to the person reading them. The @samp{2.0} version
5121 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5122 However, this would be a confusing way to write a version script.
5124 Node name can be omitted, provided it is the only version node
5125 in the version script. Such version script doesn't assign any versions to
5126 symbols, only selects which symbols will be globally visible out and which
5130 @{ global: foo; bar; local: *; @};
5133 When you link an application against a shared library that has versioned
5134 symbols, the application itself knows which version of each symbol it
5135 requires, and it also knows which version nodes it needs from each
5136 shared library it is linked against. Thus at runtime, the dynamic
5137 loader can make a quick check to make sure that the libraries you have
5138 linked against do in fact supply all of the version nodes that the
5139 application will need to resolve all of the dynamic symbols. In this
5140 way it is possible for the dynamic linker to know with certainty that
5141 all external symbols that it needs will be resolvable without having to
5142 search for each symbol reference.
5144 The symbol versioning is in effect a much more sophisticated way of
5145 doing minor version checking that SunOS does. The fundamental problem
5146 that is being addressed here is that typically references to external
5147 functions are bound on an as-needed basis, and are not all bound when
5148 the application starts up. If a shared library is out of date, a
5149 required interface may be missing; when the application tries to use
5150 that interface, it may suddenly and unexpectedly fail. With symbol
5151 versioning, the user will get a warning when they start their program if
5152 the libraries being used with the application are too old.
5154 There are several GNU extensions to Sun's versioning approach. The
5155 first of these is the ability to bind a symbol to a version node in the
5156 source file where the symbol is defined instead of in the versioning
5157 script. This was done mainly to reduce the burden on the library
5158 maintainer. You can do this by putting something like:
5160 __asm__(".symver original_foo,foo@@VERS_1.1");
5163 in the C source file. This renames the function @samp{original_foo} to
5164 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5165 The @samp{local:} directive can be used to prevent the symbol
5166 @samp{original_foo} from being exported. A @samp{.symver} directive
5167 takes precedence over a version script.
5169 The second GNU extension is to allow multiple versions of the same
5170 function to appear in a given shared library. In this way you can make
5171 an incompatible change to an interface without increasing the major
5172 version number of the shared library, while still allowing applications
5173 linked against the old interface to continue to function.
5175 To do this, you must use multiple @samp{.symver} directives in the
5176 source file. Here is an example:
5179 __asm__(".symver original_foo,foo@@");
5180 __asm__(".symver old_foo,foo@@VERS_1.1");
5181 __asm__(".symver old_foo1,foo@@VERS_1.2");
5182 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5185 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5186 unspecified base version of the symbol. The source file that contains this
5187 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5188 @samp{old_foo1}, and @samp{new_foo}.
5190 When you have multiple definitions of a given symbol, there needs to be
5191 some way to specify a default version to which external references to
5192 this symbol will be bound. You can do this with the
5193 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5194 declare one version of a symbol as the default in this manner; otherwise
5195 you would effectively have multiple definitions of the same symbol.
5197 If you wish to bind a reference to a specific version of the symbol
5198 within the shared library, you can use the aliases of convenience
5199 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5200 specifically bind to an external version of the function in question.
5202 You can also specify the language in the version script:
5205 VERSION extern "lang" @{ version-script-commands @}
5208 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5209 The linker will iterate over the list of symbols at the link time and
5210 demangle them according to @samp{lang} before matching them to the
5211 patterns specified in @samp{version-script-commands}. The default
5212 @samp{lang} is @samp{C}.
5214 Demangled names may contains spaces and other special characters. As
5215 described above, you can use a glob pattern to match demangled names,
5216 or you can use a double-quoted string to match the string exactly. In
5217 the latter case, be aware that minor differences (such as differing
5218 whitespace) between the version script and the demangler output will
5219 cause a mismatch. As the exact string generated by the demangler
5220 might change in the future, even if the mangled name does not, you
5221 should check that all of your version directives are behaving as you
5222 expect when you upgrade.
5225 @section Expressions in Linker Scripts
5228 The syntax for expressions in the linker script language is identical to
5229 that of C expressions. All expressions are evaluated as integers. All
5230 expressions are evaluated in the same size, which is 32 bits if both the
5231 host and target are 32 bits, and is otherwise 64 bits.
5233 You can use and set symbol values in expressions.
5235 The linker defines several special purpose builtin functions for use in
5239 * Constants:: Constants
5240 * Symbolic Constants:: Symbolic constants
5241 * Symbols:: Symbol Names
5242 * Orphan Sections:: Orphan Sections
5243 * Location Counter:: The Location Counter
5244 * Operators:: Operators
5245 * Evaluation:: Evaluation
5246 * Expression Section:: The Section of an Expression
5247 * Builtin Functions:: Builtin Functions
5251 @subsection Constants
5252 @cindex integer notation
5253 @cindex constants in linker scripts
5254 All constants are integers.
5256 As in C, the linker considers an integer beginning with @samp{0} to be
5257 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5258 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5259 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5260 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5261 value without a prefix or a suffix is considered to be decimal.
5263 @cindex scaled integers
5264 @cindex K and M integer suffixes
5265 @cindex M and K integer suffixes
5266 @cindex suffixes for integers
5267 @cindex integer suffixes
5268 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5272 @c END TEXI2ROFF-KILL
5273 @code{1024} or @code{1024*1024}
5277 ${\rm 1024}$ or ${\rm 1024}^2$
5279 @c END TEXI2ROFF-KILL
5280 respectively. For example, the following
5281 all refer to the same quantity:
5290 Note - the @code{K} and @code{M} suffixes cannot be used in
5291 conjunction with the base suffixes mentioned above.
5293 @node Symbolic Constants
5294 @subsection Symbolic Constants
5295 @cindex symbolic constants
5297 It is possible to refer to target specific constants via the use of
5298 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5303 The target's maximum page size.
5305 @item COMMONPAGESIZE
5306 @kindex COMMONPAGESIZE
5307 The target's default page size.
5313 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5316 will create a text section aligned to the largest page boundary
5317 supported by the target.
5320 @subsection Symbol Names
5321 @cindex symbol names
5323 @cindex quoted symbol names
5325 Unless quoted, symbol names start with a letter, underscore, or period
5326 and may include letters, digits, underscores, periods, and hyphens.
5327 Unquoted symbol names must not conflict with any keywords. You can
5328 specify a symbol which contains odd characters or has the same name as a
5329 keyword by surrounding the symbol name in double quotes:
5332 "with a space" = "also with a space" + 10;
5335 Since symbols can contain many non-alphabetic characters, it is safest
5336 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5337 whereas @samp{A - B} is an expression involving subtraction.
5339 @node Orphan Sections
5340 @subsection Orphan Sections
5342 Orphan sections are sections present in the input files which
5343 are not explicitly placed into the output file by the linker
5344 script. The linker will still copy these sections into the
5345 output file, but it has to guess as to where they should be
5346 placed. The linker uses a simple heuristic to do this. It
5347 attempts to place orphan sections after non-orphan sections of the
5348 same attribute, such as code vs data, loadable vs non-loadable, etc.
5349 If there is not enough room to do this then it places
5350 at the end of the file.
5352 For ELF targets, the attribute of the section includes section type as
5353 well as section flag.
5355 If an orphaned section's name is representable as a C identifier then
5356 the linker will automatically @pxref{PROVIDE} two symbols:
5357 __start_SECNAME and __stop_SECNAME, where SECNAME is the name of the
5358 section. These indicate the start address and end address of the
5359 orphaned section respectively. Note: most section names are not
5360 representable as C identifiers because they contain a @samp{.}
5363 @node Location Counter
5364 @subsection The Location Counter
5367 @cindex location counter
5368 @cindex current output location
5369 The special linker variable @dfn{dot} @samp{.} always contains the
5370 current output location counter. Since the @code{.} always refers to a
5371 location in an output section, it may only appear in an expression
5372 within a @code{SECTIONS} command. The @code{.} symbol may appear
5373 anywhere that an ordinary symbol is allowed in an expression.
5376 Assigning a value to @code{.} will cause the location counter to be
5377 moved. This may be used to create holes in the output section. The
5378 location counter may not be moved backwards inside an output section,
5379 and may not be moved backwards outside of an output section if so
5380 doing creates areas with overlapping LMAs.
5396 In the previous example, the @samp{.text} section from @file{file1} is
5397 located at the beginning of the output section @samp{output}. It is
5398 followed by a 1000 byte gap. Then the @samp{.text} section from
5399 @file{file2} appears, also with a 1000 byte gap following before the
5400 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5401 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5403 @cindex dot inside sections
5404 Note: @code{.} actually refers to the byte offset from the start of the
5405 current containing object. Normally this is the @code{SECTIONS}
5406 statement, whose start address is 0, hence @code{.} can be used as an
5407 absolute address. If @code{.} is used inside a section description
5408 however, it refers to the byte offset from the start of that section,
5409 not an absolute address. Thus in a script like this:
5427 The @samp{.text} section will be assigned a starting address of 0x100
5428 and a size of exactly 0x200 bytes, even if there is not enough data in
5429 the @samp{.text} input sections to fill this area. (If there is too
5430 much data, an error will be produced because this would be an attempt to
5431 move @code{.} backwards). The @samp{.data} section will start at 0x500
5432 and it will have an extra 0x600 bytes worth of space after the end of
5433 the values from the @samp{.data} input sections and before the end of
5434 the @samp{.data} output section itself.
5436 @cindex dot outside sections
5437 Setting symbols to the value of the location counter outside of an
5438 output section statement can result in unexpected values if the linker
5439 needs to place orphan sections. For example, given the following:
5445 .text: @{ *(.text) @}
5449 .data: @{ *(.data) @}
5454 If the linker needs to place some input section, e.g. @code{.rodata},
5455 not mentioned in the script, it might choose to place that section
5456 between @code{.text} and @code{.data}. You might think the linker
5457 should place @code{.rodata} on the blank line in the above script, but
5458 blank lines are of no particular significance to the linker. As well,
5459 the linker doesn't associate the above symbol names with their
5460 sections. Instead, it assumes that all assignments or other
5461 statements belong to the previous output section, except for the
5462 special case of an assignment to @code{.}. I.e., the linker will
5463 place the orphan @code{.rodata} section as if the script was written
5470 .text: @{ *(.text) @}
5474 .rodata: @{ *(.rodata) @}
5475 .data: @{ *(.data) @}
5480 This may or may not be the script author's intention for the value of
5481 @code{start_of_data}. One way to influence the orphan section
5482 placement is to assign the location counter to itself, as the linker
5483 assumes that an assignment to @code{.} is setting the start address of
5484 a following output section and thus should be grouped with that
5485 section. So you could write:
5491 .text: @{ *(.text) @}
5496 .data: @{ *(.data) @}
5501 Now, the orphan @code{.rodata} section will be placed between
5502 @code{end_of_text} and @code{start_of_data}.
5506 @subsection Operators
5507 @cindex operators for arithmetic
5508 @cindex arithmetic operators
5509 @cindex precedence in expressions
5510 The linker recognizes the standard C set of arithmetic operators, with
5511 the standard bindings and precedence levels:
5514 @c END TEXI2ROFF-KILL
5516 precedence associativity Operators Notes
5522 5 left == != > < <= >=
5528 11 right &= += -= *= /= (2)
5532 (1) Prefix operators
5533 (2) @xref{Assignments}.
5537 \vskip \baselineskip
5538 %"lispnarrowing" is the extra indent used generally for smallexample
5539 \hskip\lispnarrowing\vbox{\offinterlineskip
5542 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
5543 height2pt&\omit&&\omit&&\omit&\cr
5544 &Precedence&& Associativity &&{\rm Operators}&\cr
5545 height2pt&\omit&&\omit&&\omit&\cr
5547 height2pt&\omit&&\omit&&\omit&\cr
5549 % '176 is tilde, '~' in tt font
5550 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
5551 &2&&left&&* / \%&\cr
5554 &5&&left&&== != > < <= >=&\cr
5557 &8&&left&&{\&\&}&\cr
5560 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
5562 height2pt&\omit&&\omit&&\omit&\cr}
5567 @obeylines@parskip=0pt@parindent=0pt
5568 @dag@quad Prefix operators.
5569 @ddag@quad @xref{Assignments}.
5572 @c END TEXI2ROFF-KILL
5575 @subsection Evaluation
5576 @cindex lazy evaluation
5577 @cindex expression evaluation order
5578 The linker evaluates expressions lazily. It only computes the value of
5579 an expression when absolutely necessary.
5581 The linker needs some information, such as the value of the start
5582 address of the first section, and the origins and lengths of memory
5583 regions, in order to do any linking at all. These values are computed
5584 as soon as possible when the linker reads in the linker script.
5586 However, other values (such as symbol values) are not known or needed
5587 until after storage allocation. Such values are evaluated later, when
5588 other information (such as the sizes of output sections) is available
5589 for use in the symbol assignment expression.
5591 The sizes of sections cannot be known until after allocation, so
5592 assignments dependent upon these are not performed until after
5595 Some expressions, such as those depending upon the location counter
5596 @samp{.}, must be evaluated during section allocation.
5598 If the result of an expression is required, but the value is not
5599 available, then an error results. For example, a script like the
5605 .text 9+this_isnt_constant :
5611 will cause the error message @samp{non constant expression for initial
5614 @node Expression Section
5615 @subsection The Section of an Expression
5616 @cindex expression sections
5617 @cindex absolute expressions
5618 @cindex relative expressions
5619 @cindex absolute and relocatable symbols
5620 @cindex relocatable and absolute symbols
5621 @cindex symbols, relocatable and absolute
5622 Addresses and symbols may be section relative, or absolute. A section
5623 relative symbol is relocatable. If you request relocatable output
5624 using the @samp{-r} option, a further link operation may change the
5625 value of a section relative symbol. On the other hand, an absolute
5626 symbol will retain the same value throughout any further link
5629 Some terms in linker expressions are addresses. This is true of
5630 section relative symbols and for builtin functions that return an
5631 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
5632 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
5633 functions that return a non-address value, such as @code{LENGTH}.
5634 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
5635 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
5636 differently depending on their location, for compatibility with older
5637 versions of @code{ld}. Expressions appearing outside an output
5638 section definition treat all numbers as absolute addresses.
5639 Expressions appearing inside an output section definition treat
5640 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
5641 given, then absolute symbols and numbers are simply treated as numbers
5644 In the following simple example,
5651 __executable_start = 0x100;
5655 __data_start = 0x10;
5663 both @code{.} and @code{__executable_start} are set to the absolute
5664 address 0x100 in the first two assignments, then both @code{.} and
5665 @code{__data_start} are set to 0x10 relative to the @code{.data}
5666 section in the second two assignments.
5668 For expressions involving numbers, relative addresses and absolute
5669 addresses, ld follows these rules to evaluate terms:
5673 Unary operations on an absolute address or number, and binary
5674 operations on two absolute addresses or two numbers, or between one
5675 absolute address and a number, apply the operator to the value(s).
5677 Unary operations on a relative address, and binary operations on two
5678 relative addresses in the same section or between one relative address
5679 and a number, apply the operator to the offset part of the address(es).
5681 Other binary operations, that is, between two relative addresses not
5682 in the same section, or between a relative address and an absolute
5683 address, first convert any non-absolute term to an absolute address
5684 before applying the operator.
5687 The result section of each sub-expression is as follows:
5691 An operation involving only numbers results in a number.
5693 The result of comparisons, @samp{&&} and @samp{||} is also a number.
5695 The result of other binary arithmetic and logical operations on two
5696 relative addresses in the same section or two absolute addresses
5697 (after above conversions) is also a number.
5699 The result of other operations on relative addresses or one
5700 relative address and a number, is a relative address in the same
5701 section as the relative operand(s).
5703 The result of other operations on absolute addresses (after above
5704 conversions) is an absolute address.
5707 You can use the builtin function @code{ABSOLUTE} to force an expression
5708 to be absolute when it would otherwise be relative. For example, to
5709 create an absolute symbol set to the address of the end of the output
5710 section @samp{.data}:
5714 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
5718 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
5719 @samp{.data} section.
5721 Using @code{LOADADDR} also forces an expression absolute, since this
5722 particular builtin function returns an absolute address.
5724 @node Builtin Functions
5725 @subsection Builtin Functions
5726 @cindex functions in expressions
5727 The linker script language includes a number of builtin functions for
5728 use in linker script expressions.
5731 @item ABSOLUTE(@var{exp})
5732 @kindex ABSOLUTE(@var{exp})
5733 @cindex expression, absolute
5734 Return the absolute (non-relocatable, as opposed to non-negative) value
5735 of the expression @var{exp}. Primarily useful to assign an absolute
5736 value to a symbol within a section definition, where symbol values are
5737 normally section relative. @xref{Expression Section}.
5739 @item ADDR(@var{section})
5740 @kindex ADDR(@var{section})
5741 @cindex section address in expression
5742 Return the address (VMA) of the named @var{section}. Your
5743 script must previously have defined the location of that section. In
5744 the following example, @code{start_of_output_1}, @code{symbol_1} and
5745 @code{symbol_2} are assigned equivalent values, except that
5746 @code{symbol_1} will be relative to the @code{.output1} section while
5747 the other two will be absolute:
5753 start_of_output_1 = ABSOLUTE(.);
5758 symbol_1 = ADDR(.output1);
5759 symbol_2 = start_of_output_1;
5765 @item ALIGN(@var{align})
5766 @itemx ALIGN(@var{exp},@var{align})
5767 @kindex ALIGN(@var{align})
5768 @kindex ALIGN(@var{exp},@var{align})
5769 @cindex round up location counter
5770 @cindex align location counter
5771 @cindex round up expression
5772 @cindex align expression
5773 Return the location counter (@code{.}) or arbitrary expression aligned
5774 to the next @var{align} boundary. The single operand @code{ALIGN}
5775 doesn't change the value of the location counter---it just does
5776 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5777 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5778 equivalent to @code{ALIGN(., @var{align})}).
5780 Here is an example which aligns the output @code{.data} section to the
5781 next @code{0x2000} byte boundary after the preceding section and sets a
5782 variable within the section to the next @code{0x8000} boundary after the
5787 .data ALIGN(0x2000): @{
5789 variable = ALIGN(0x8000);
5795 The first use of @code{ALIGN} in this example specifies the location of
5796 a section because it is used as the optional @var{address} attribute of
5797 a section definition (@pxref{Output Section Address}). The second use
5798 of @code{ALIGN} is used to defines the value of a symbol.
5800 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5802 @item ALIGNOF(@var{section})
5803 @kindex ALIGNOF(@var{section})
5804 @cindex section alignment
5805 Return the alignment in bytes of the named @var{section}, if that section has
5806 been allocated. If the section has not been allocated when this is
5807 evaluated, the linker will report an error. In the following example,
5808 the alignment of the @code{.output} section is stored as the first
5809 value in that section.
5814 LONG (ALIGNOF (.output))
5821 @item BLOCK(@var{exp})
5822 @kindex BLOCK(@var{exp})
5823 This is a synonym for @code{ALIGN}, for compatibility with older linker
5824 scripts. It is most often seen when setting the address of an output
5827 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5828 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5829 This is equivalent to either
5831 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5835 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5838 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5839 for the data segment (area between the result of this expression and
5840 @code{DATA_SEGMENT_END}) than the former or not.
5841 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5842 memory will be saved at the expense of up to @var{commonpagesize} wasted
5843 bytes in the on-disk file.
5845 This expression can only be used directly in @code{SECTIONS} commands, not in
5846 any output section descriptions and only once in the linker script.
5847 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5848 be the system page size the object wants to be optimized for (while still
5849 working on system page sizes up to @var{maxpagesize}).
5854 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5857 @item DATA_SEGMENT_END(@var{exp})
5858 @kindex DATA_SEGMENT_END(@var{exp})
5859 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5860 evaluation purposes.
5863 . = DATA_SEGMENT_END(.);
5866 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5867 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5868 This defines the end of the @code{PT_GNU_RELRO} segment when
5869 @samp{-z relro} option is used. Second argument is returned.
5870 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5871 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5872 @var{exp} + @var{offset} is aligned to the most commonly used page
5873 boundary for particular target. If present in the linker script,
5874 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5875 @code{DATA_SEGMENT_END}.
5878 . = DATA_SEGMENT_RELRO_END(24, .);
5881 @item DEFINED(@var{symbol})
5882 @kindex DEFINED(@var{symbol})
5883 @cindex symbol defaults
5884 Return 1 if @var{symbol} is in the linker global symbol table and is
5885 defined before the statement using DEFINED in the script, otherwise
5886 return 0. You can use this function to provide
5887 default values for symbols. For example, the following script fragment
5888 shows how to set a global symbol @samp{begin} to the first location in
5889 the @samp{.text} section---but if a symbol called @samp{begin} already
5890 existed, its value is preserved:
5896 begin = DEFINED(begin) ? begin : . ;
5904 @item LENGTH(@var{memory})
5905 @kindex LENGTH(@var{memory})
5906 Return the length of the memory region named @var{memory}.
5908 @item LOADADDR(@var{section})
5909 @kindex LOADADDR(@var{section})
5910 @cindex section load address in expression
5911 Return the absolute LMA of the named @var{section}. (@pxref{Output
5915 @item MAX(@var{exp1}, @var{exp2})
5916 Returns the maximum of @var{exp1} and @var{exp2}.
5919 @item MIN(@var{exp1}, @var{exp2})
5920 Returns the minimum of @var{exp1} and @var{exp2}.
5922 @item NEXT(@var{exp})
5923 @kindex NEXT(@var{exp})
5924 @cindex unallocated address, next
5925 Return the next unallocated address that is a multiple of @var{exp}.
5926 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5927 use the @code{MEMORY} command to define discontinuous memory for the
5928 output file, the two functions are equivalent.
5930 @item ORIGIN(@var{memory})
5931 @kindex ORIGIN(@var{memory})
5932 Return the origin of the memory region named @var{memory}.
5934 @item SEGMENT_START(@var{segment}, @var{default})
5935 @kindex SEGMENT_START(@var{segment}, @var{default})
5936 Return the base address of the named @var{segment}. If an explicit
5937 value has been given for this segment (with a command-line @samp{-T}
5938 option) that value will be returned; otherwise the value will be
5939 @var{default}. At present, the @samp{-T} command-line option can only
5940 be used to set the base address for the ``text'', ``data'', and
5941 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
5944 @item SIZEOF(@var{section})
5945 @kindex SIZEOF(@var{section})
5946 @cindex section size
5947 Return the size in bytes of the named @var{section}, if that section has
5948 been allocated. If the section has not been allocated when this is
5949 evaluated, the linker will report an error. In the following example,
5950 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5959 symbol_1 = .end - .start ;
5960 symbol_2 = SIZEOF(.output);
5965 @item SIZEOF_HEADERS
5966 @itemx sizeof_headers
5967 @kindex SIZEOF_HEADERS
5969 Return the size in bytes of the output file's headers. This is
5970 information which appears at the start of the output file. You can use
5971 this number when setting the start address of the first section, if you
5972 choose, to facilitate paging.
5974 @cindex not enough room for program headers
5975 @cindex program headers, not enough room
5976 When producing an ELF output file, if the linker script uses the
5977 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5978 number of program headers before it has determined all the section
5979 addresses and sizes. If the linker later discovers that it needs
5980 additional program headers, it will report an error @samp{not enough
5981 room for program headers}. To avoid this error, you must avoid using
5982 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5983 script to avoid forcing the linker to use additional program headers, or
5984 you must define the program headers yourself using the @code{PHDRS}
5985 command (@pxref{PHDRS}).
5988 @node Implicit Linker Scripts
5989 @section Implicit Linker Scripts
5990 @cindex implicit linker scripts
5991 If you specify a linker input file which the linker can not recognize as
5992 an object file or an archive file, it will try to read the file as a
5993 linker script. If the file can not be parsed as a linker script, the
5994 linker will report an error.
5996 An implicit linker script will not replace the default linker script.
5998 Typically an implicit linker script would contain only symbol
5999 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6002 Any input files read because of an implicit linker script will be read
6003 at the position in the command line where the implicit linker script was
6004 read. This can affect archive searching.
6007 @node Machine Dependent
6008 @chapter Machine Dependent Features
6010 @cindex machine dependencies
6011 @command{ld} has additional features on some platforms; the following
6012 sections describe them. Machines where @command{ld} has no additional
6013 functionality are not listed.
6017 * H8/300:: @command{ld} and the H8/300
6020 * i960:: @command{ld} and the Intel 960 family
6023 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6026 * ARM:: @command{ld} and the ARM family
6029 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6032 * M68K:: @command{ld} and the Motorola 68K family
6035 * MMIX:: @command{ld} and MMIX
6038 * MSP430:: @command{ld} and MSP430
6041 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6044 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6047 * SPU ELF:: @command{ld} and SPU ELF Support
6050 * TI COFF:: @command{ld} and TI COFF
6053 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6056 * Xtensa:: @command{ld} and Xtensa Processors
6067 @section @command{ld} and the H8/300
6069 @cindex H8/300 support
6070 For the H8/300, @command{ld} can perform these global optimizations when
6071 you specify the @samp{--relax} command-line option.
6074 @cindex relaxing on H8/300
6075 @item relaxing address modes
6076 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6077 targets are within eight bits, and turns them into eight-bit
6078 program-counter relative @code{bsr} and @code{bra} instructions,
6081 @cindex synthesizing on H8/300
6082 @item synthesizing instructions
6083 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6084 @command{ld} finds all @code{mov.b} instructions which use the
6085 sixteen-bit absolute address form, but refer to the top
6086 page of memory, and changes them to use the eight-bit address form.
6087 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6088 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6089 top page of memory).
6091 @command{ld} finds all @code{mov} instructions which use the register
6092 indirect with 32-bit displacement addressing mode, but use a small
6093 displacement inside 16-bit displacement range, and changes them to use
6094 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6095 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6096 whenever the displacement @var{d} is in the 16 bit signed integer
6097 range. Only implemented in ELF-format ld).
6099 @item bit manipulation instructions
6100 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6101 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6102 which use 32 bit and 16 bit absolute address form, but refer to the top
6103 page of memory, and changes them to use the 8 bit address form.
6104 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6105 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6106 the top page of memory).
6108 @item system control instructions
6109 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6110 32 bit absolute address form, but refer to the top page of memory, and
6111 changes them to use 16 bit address form.
6112 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6113 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6114 the top page of memory).
6124 @c This stuff is pointless to say unless you're especially concerned
6125 @c with Renesas chips; don't enable it for generic case, please.
6127 @chapter @command{ld} and Other Renesas Chips
6129 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6130 H8/500, and SH chips. No special features, commands, or command-line
6131 options are required for these chips.
6141 @section @command{ld} and the Intel 960 Family
6143 @cindex i960 support
6145 You can use the @samp{-A@var{architecture}} command line option to
6146 specify one of the two-letter names identifying members of the 960
6147 family; the option specifies the desired output target, and warns of any
6148 incompatible instructions in the input files. It also modifies the
6149 linker's search strategy for archive libraries, to support the use of
6150 libraries specific to each particular architecture, by including in the
6151 search loop names suffixed with the string identifying the architecture.
6153 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
6154 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
6155 paths, and in any paths you specify with @samp{-L}) for a library with
6168 The first two possibilities would be considered in any event; the last
6169 two are due to the use of @w{@samp{-ACA}}.
6171 You can meaningfully use @samp{-A} more than once on a command line, since
6172 the 960 architecture family allows combination of target architectures; each
6173 use will add another pair of name variants to search for when @w{@samp{-l}}
6174 specifies a library.
6176 @cindex @option{--relax} on i960
6177 @cindex relaxing on i960
6178 @command{ld} supports the @samp{--relax} option for the i960 family. If
6179 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
6180 @code{calx} instructions whose targets are within 24 bits, and turns
6181 them into 24-bit program-counter relative @code{bal} and @code{cal}
6182 instructions, respectively. @command{ld} also turns @code{cal}
6183 instructions into @code{bal} instructions when it determines that the
6184 target subroutine is a leaf routine (that is, the target subroutine does
6185 not itself call any subroutines).
6187 @cindex Cortex-A8 erratum workaround
6188 @kindex --fix-cortex-a8
6189 @kindex --no-fix-cortex-a8
6190 The @samp{--fix-cortex-a8} switch enables a link-time workaround for an erratum in certain Cortex-A8 processors. The workaround is enabled by default if you are targeting the ARM v7-A architecture profile. It can be enabled otherwise by specifying @samp{--fix-cortex-a8}, or disabled unconditionally by specifying @samp{--no-fix-cortex-a8}.
6192 The erratum only affects Thumb-2 code. Please contact ARM for further details.
6194 @kindex --merge-exidx-entries
6195 @kindex --no-merge-exidx-entries
6196 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
6213 @node M68HC11/68HC12
6214 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6216 @cindex M68HC11 and 68HC12 support
6218 @subsection Linker Relaxation
6220 For the Motorola 68HC11, @command{ld} can perform these global
6221 optimizations when you specify the @samp{--relax} command-line option.
6224 @cindex relaxing on M68HC11
6225 @item relaxing address modes
6226 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6227 targets are within eight bits, and turns them into eight-bit
6228 program-counter relative @code{bsr} and @code{bra} instructions,
6231 @command{ld} also looks at all 16-bit extended addressing modes and
6232 transforms them in a direct addressing mode when the address is in
6233 page 0 (between 0 and 0x0ff).
6235 @item relaxing gcc instruction group
6236 When @command{gcc} is called with @option{-mrelax}, it can emit group
6237 of instructions that the linker can optimize to use a 68HC11 direct
6238 addressing mode. These instructions consists of @code{bclr} or
6239 @code{bset} instructions.
6243 @subsection Trampoline Generation
6245 @cindex trampoline generation on M68HC11
6246 @cindex trampoline generation on M68HC12
6247 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6248 call a far function using a normal @code{jsr} instruction. The linker
6249 will also change the relocation to some far function to use the
6250 trampoline address instead of the function address. This is typically the
6251 case when a pointer to a function is taken. The pointer will in fact
6252 point to the function trampoline.
6260 @section @command{ld} and the ARM family
6262 @cindex ARM interworking support
6263 @kindex --support-old-code
6264 For the ARM, @command{ld} will generate code stubs to allow functions calls
6265 between ARM and Thumb code. These stubs only work with code that has
6266 been compiled and assembled with the @samp{-mthumb-interwork} command
6267 line option. If it is necessary to link with old ARM object files or
6268 libraries, which have not been compiled with the -mthumb-interwork
6269 option then the @samp{--support-old-code} command line switch should be
6270 given to the linker. This will make it generate larger stub functions
6271 which will work with non-interworking aware ARM code. Note, however,
6272 the linker does not support generating stubs for function calls to
6273 non-interworking aware Thumb code.
6275 @cindex thumb entry point
6276 @cindex entry point, thumb
6277 @kindex --thumb-entry=@var{entry}
6278 The @samp{--thumb-entry} switch is a duplicate of the generic
6279 @samp{--entry} switch, in that it sets the program's starting address.
6280 But it also sets the bottom bit of the address, so that it can be
6281 branched to using a BX instruction, and the program will start
6282 executing in Thumb mode straight away.
6284 @cindex PE import table prefixing
6285 @kindex --use-nul-prefixed-import-tables
6286 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6287 the import tables idata4 and idata5 have to be generated with a zero
6288 element prefix for import libraries. This is the old style to generate
6289 import tables. By default this option is turned off.
6293 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6294 executables. This option is only valid when linking big-endian objects.
6295 The resulting image will contain big-endian data and little-endian code.
6298 @kindex --target1-rel
6299 @kindex --target1-abs
6300 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6301 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6302 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6303 and @samp{--target1-abs} switches override the default.
6306 @kindex --target2=@var{type}
6307 The @samp{--target2=type} switch overrides the default definition of the
6308 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6309 meanings, and target defaults are as follows:
6312 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6314 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6316 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6321 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6322 specification) enables objects compiled for the ARMv4 architecture to be
6323 interworking-safe when linked with other objects compiled for ARMv4t, but
6324 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6326 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6327 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6328 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6330 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6331 relocations are ignored.
6333 @cindex FIX_V4BX_INTERWORKING
6334 @kindex --fix-v4bx-interworking
6335 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6336 relocations with a branch to the following veneer:
6344 This allows generation of libraries/applications that work on ARMv4 cores
6345 and are still interworking safe. Note that the above veneer clobbers the
6346 condition flags, so may cause incorrect program behavior in rare cases.
6350 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6351 BLX instructions (available on ARMv5t and above) in various
6352 situations. Currently it is used to perform calls via the PLT from Thumb
6353 code using BLX rather than using BX and a mode-switching stub before
6354 each PLT entry. This should lead to such calls executing slightly faster.
6356 This option is enabled implicitly for SymbianOS, so there is no need to
6357 specify it if you are using that target.
6359 @cindex VFP11_DENORM_FIX
6360 @kindex --vfp11-denorm-fix
6361 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6362 bug in certain VFP11 coprocessor hardware, which sometimes allows
6363 instructions with denorm operands (which must be handled by support code)
6364 to have those operands overwritten by subsequent instructions before
6365 the support code can read the intended values.
6367 The bug may be avoided in scalar mode if you allow at least one
6368 intervening instruction between a VFP11 instruction which uses a register
6369 and another instruction which writes to the same register, or at least two
6370 intervening instructions if vector mode is in use. The bug only affects
6371 full-compliance floating-point mode: you do not need this workaround if
6372 you are using "runfast" mode. Please contact ARM for further details.
6374 If you know you are using buggy VFP11 hardware, you can
6375 enable this workaround by specifying the linker option
6376 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6377 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6378 vector mode (the latter also works for scalar code). The default is
6379 @samp{--vfp-denorm-fix=none}.
6381 If the workaround is enabled, instructions are scanned for
6382 potentially-troublesome sequences, and a veneer is created for each
6383 such sequence which may trigger the erratum. The veneer consists of the
6384 first instruction of the sequence and a branch back to the subsequent
6385 instruction. The original instruction is then replaced with a branch to
6386 the veneer. The extra cycles required to call and return from the veneer
6387 are sufficient to avoid the erratum in both the scalar and vector cases.
6389 @cindex ARM1176 erratum workaround
6390 @kindex --fix-arm1176
6391 @kindex --no-fix-arm1176
6392 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6393 in certain ARM1176 processors. The workaround is enabled by default if you
6394 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6395 unconditionally by specifying @samp{--no-fix-arm1176}.
6397 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6398 Programmer Advice Notice'' available on the ARM documentation website at:
6399 http://infocenter.arm.com/.
6401 @cindex NO_ENUM_SIZE_WARNING
6402 @kindex --no-enum-size-warning
6403 The @option{--no-enum-size-warning} switch prevents the linker from
6404 warning when linking object files that specify incompatible EABI
6405 enumeration size attributes. For example, with this switch enabled,
6406 linking of an object file using 32-bit enumeration values with another
6407 using enumeration values fitted into the smallest possible space will
6410 @cindex NO_WCHAR_SIZE_WARNING
6411 @kindex --no-wchar-size-warning
6412 The @option{--no-wchar-size-warning} switch prevents the linker from
6413 warning when linking object files that specify incompatible EABI
6414 @code{wchar_t} size attributes. For example, with this switch enabled,
6415 linking of an object file using 32-bit @code{wchar_t} values with another
6416 using 16-bit @code{wchar_t} values will not be diagnosed.
6419 @kindex --pic-veneer
6420 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6421 ARM/Thumb interworking veneers, even if the rest of the binary
6422 is not PIC. This avoids problems on uClinux targets where
6423 @samp{--emit-relocs} is used to generate relocatable binaries.
6425 @cindex STUB_GROUP_SIZE
6426 @kindex --stub-group-size=@var{N}
6427 The linker will automatically generate and insert small sequences of
6428 code into a linked ARM ELF executable whenever an attempt is made to
6429 perform a function call to a symbol that is too far away. The
6430 placement of these sequences of instructions - called stubs - is
6431 controlled by the command line option @option{--stub-group-size=N}.
6432 The placement is important because a poor choice can create a need for
6433 duplicate stubs, increasing the code size. The linker will try to
6434 group stubs together in order to reduce interruptions to the flow of
6435 code, but it needs guidance as to how big these groups should be and
6436 where they should be placed.
6438 The value of @samp{N}, the parameter to the
6439 @option{--stub-group-size=} option controls where the stub groups are
6440 placed. If it is negative then all stubs are placed after the first
6441 branch that needs them. If it is positive then the stubs can be
6442 placed either before or after the branches that need them. If the
6443 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6444 exactly where to place groups of stubs, using its built in heuristics.
6445 A value of @samp{N} greater than 1 (or smaller than -1) tells the
6446 linker that a single group of stubs can service at most @samp{N} bytes
6447 from the input sections.
6449 The default, if @option{--stub-group-size=} is not specified, is
6452 Farcalls stubs insertion is fully supported for the ARM-EABI target
6453 only, because it relies on object files properties not present
6467 @section @command{ld} and HPPA 32-bit ELF Support
6468 @cindex HPPA multiple sub-space stubs
6469 @kindex --multi-subspace
6470 When generating a shared library, @command{ld} will by default generate
6471 import stubs suitable for use with a single sub-space application.
6472 The @samp{--multi-subspace} switch causes @command{ld} to generate export
6473 stubs, and different (larger) import stubs suitable for use with
6474 multiple sub-spaces.
6476 @cindex HPPA stub grouping
6477 @kindex --stub-group-size=@var{N}
6478 Long branch stubs and import/export stubs are placed by @command{ld} in
6479 stub sections located between groups of input sections.
6480 @samp{--stub-group-size} specifies the maximum size of a group of input
6481 sections handled by one stub section. Since branch offsets are signed,
6482 a stub section may serve two groups of input sections, one group before
6483 the stub section, and one group after it. However, when using
6484 conditional branches that require stubs, it may be better (for branch
6485 prediction) that stub sections only serve one group of input sections.
6486 A negative value for @samp{N} chooses this scheme, ensuring that
6487 branches to stubs always use a negative offset. Two special values of
6488 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6489 @command{ld} to automatically size input section groups for the branch types
6490 detected, with the same behaviour regarding stub placement as other
6491 positive or negative values of @samp{N} respectively.
6493 Note that @samp{--stub-group-size} does not split input sections. A
6494 single input section larger than the group size specified will of course
6495 create a larger group (of one section). If input sections are too
6496 large, it may not be possible for a branch to reach its stub.
6509 @section @command{ld} and the Motorola 68K family
6511 @cindex Motorola 68K GOT generation
6512 @kindex --got=@var{type}
6513 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
6514 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
6515 @samp{target}. When @samp{target} is selected the linker chooses
6516 the default GOT generation scheme for the current target.
6517 @samp{single} tells the linker to generate a single GOT with
6518 entries only at non-negative offsets.
6519 @samp{negative} instructs the linker to generate a single GOT with
6520 entries at both negative and positive offsets. Not all environments
6522 @samp{multigot} allows the linker to generate several GOTs in the
6523 output file. All GOT references from a single input object
6524 file access the same GOT, but references from different input object
6525 files might access different GOTs. Not all environments support such GOTs.
6538 @section @code{ld} and MMIX
6539 For MMIX, there is a choice of generating @code{ELF} object files or
6540 @code{mmo} object files when linking. The simulator @code{mmix}
6541 understands the @code{mmo} format. The binutils @code{objcopy} utility
6542 can translate between the two formats.
6544 There is one special section, the @samp{.MMIX.reg_contents} section.
6545 Contents in this section is assumed to correspond to that of global
6546 registers, and symbols referring to it are translated to special symbols,
6547 equal to registers. In a final link, the start address of the
6548 @samp{.MMIX.reg_contents} section corresponds to the first allocated
6549 global register multiplied by 8. Register @code{$255} is not included in
6550 this section; it is always set to the program entry, which is at the
6551 symbol @code{Main} for @code{mmo} files.
6553 Global symbols with the prefix @code{__.MMIX.start.}, for example
6554 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
6555 The default linker script uses these to set the default start address
6558 Initial and trailing multiples of zero-valued 32-bit words in a section,
6559 are left out from an mmo file.
6572 @section @code{ld} and MSP430
6573 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
6574 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
6575 just pass @samp{-m help} option to the linker).
6577 @cindex MSP430 extra sections
6578 The linker will recognize some extra sections which are MSP430 specific:
6581 @item @samp{.vectors}
6582 Defines a portion of ROM where interrupt vectors located.
6584 @item @samp{.bootloader}
6585 Defines the bootloader portion of the ROM (if applicable). Any code
6586 in this section will be uploaded to the MPU.
6588 @item @samp{.infomem}
6589 Defines an information memory section (if applicable). Any code in
6590 this section will be uploaded to the MPU.
6592 @item @samp{.infomemnobits}
6593 This is the same as the @samp{.infomem} section except that any code
6594 in this section will not be uploaded to the MPU.
6596 @item @samp{.noinit}
6597 Denotes a portion of RAM located above @samp{.bss} section.
6599 The last two sections are used by gcc.
6613 @section @command{ld} and PowerPC 32-bit ELF Support
6614 @cindex PowerPC long branches
6615 @kindex --relax on PowerPC
6616 Branches on PowerPC processors are limited to a signed 26-bit
6617 displacement, which may result in @command{ld} giving
6618 @samp{relocation truncated to fit} errors with very large programs.
6619 @samp{--relax} enables the generation of trampolines that can access
6620 the entire 32-bit address space. These trampolines are inserted at
6621 section boundaries, so may not themselves be reachable if an input
6622 section exceeds 33M in size. You may combine @samp{-r} and
6623 @samp{--relax} to add trampolines in a partial link. In that case
6624 both branches to undefined symbols and inter-section branches are also
6625 considered potentially out of range, and trampolines inserted.
6627 @cindex PowerPC ELF32 options
6632 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
6633 generates code capable of using a newer PLT and GOT layout that has
6634 the security advantage of no executable section ever needing to be
6635 writable and no writable section ever being executable. PowerPC
6636 @command{ld} will generate this layout, including stubs to access the
6637 PLT, if all input files (including startup and static libraries) were
6638 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
6639 BSS PLT (and GOT layout) which can give slightly better performance.
6641 @kindex --secure-plt
6643 @command{ld} will use the new PLT and GOT layout if it is linking new
6644 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
6645 when linking non-PIC code. This option requests the new PLT and GOT
6646 layout. A warning will be given if some object file requires the old
6652 The new secure PLT and GOT are placed differently relative to other
6653 sections compared to older BSS PLT and GOT placement. The location of
6654 @code{.plt} must change because the new secure PLT is an initialized
6655 section while the old PLT is uninitialized. The reason for the
6656 @code{.got} change is more subtle: The new placement allows
6657 @code{.got} to be read-only in applications linked with
6658 @samp{-z relro -z now}. However, this placement means that
6659 @code{.sdata} cannot always be used in shared libraries, because the
6660 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
6661 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
6662 GCC doesn't use @code{.sdata} in shared libraries, so this option is
6663 really only useful for other compilers that may do so.
6665 @cindex PowerPC stub symbols
6666 @kindex --emit-stub-syms
6667 @item --emit-stub-syms
6668 This option causes @command{ld} to label linker stubs with a local
6669 symbol that encodes the stub type and destination.
6671 @cindex PowerPC TLS optimization
6672 @kindex --no-tls-optimize
6673 @item --no-tls-optimize
6674 PowerPC @command{ld} normally performs some optimization of code
6675 sequences used to access Thread-Local Storage. Use this option to
6676 disable the optimization.
6689 @node PowerPC64 ELF64
6690 @section @command{ld} and PowerPC64 64-bit ELF Support
6692 @cindex PowerPC64 ELF64 options
6694 @cindex PowerPC64 stub grouping
6695 @kindex --stub-group-size
6696 @item --stub-group-size
6697 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
6698 by @command{ld} in stub sections located between groups of input sections.
6699 @samp{--stub-group-size} specifies the maximum size of a group of input
6700 sections handled by one stub section. Since branch offsets are signed,
6701 a stub section may serve two groups of input sections, one group before
6702 the stub section, and one group after it. However, when using
6703 conditional branches that require stubs, it may be better (for branch
6704 prediction) that stub sections only serve one group of input sections.
6705 A negative value for @samp{N} chooses this scheme, ensuring that
6706 branches to stubs always use a negative offset. Two special values of
6707 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6708 @command{ld} to automatically size input section groups for the branch types
6709 detected, with the same behaviour regarding stub placement as other
6710 positive or negative values of @samp{N} respectively.
6712 Note that @samp{--stub-group-size} does not split input sections. A
6713 single input section larger than the group size specified will of course
6714 create a larger group (of one section). If input sections are too
6715 large, it may not be possible for a branch to reach its stub.
6717 @cindex PowerPC64 stub symbols
6718 @kindex --emit-stub-syms
6719 @item --emit-stub-syms
6720 This option causes @command{ld} to label linker stubs with a local
6721 symbol that encodes the stub type and destination.
6723 @cindex PowerPC64 dot symbols
6725 @kindex --no-dotsyms
6726 @item --dotsyms, --no-dotsyms
6727 These two options control how @command{ld} interprets version patterns
6728 in a version script. Older PowerPC64 compilers emitted both a
6729 function descriptor symbol with the same name as the function, and a
6730 code entry symbol with the name prefixed by a dot (@samp{.}). To
6731 properly version a function @samp{foo}, the version script thus needs
6732 to control both @samp{foo} and @samp{.foo}. The option
6733 @samp{--dotsyms}, on by default, automatically adds the required
6734 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
6737 @cindex PowerPC64 TLS optimization
6738 @kindex --no-tls-optimize
6739 @item --no-tls-optimize
6740 PowerPC64 @command{ld} normally performs some optimization of code
6741 sequences used to access Thread-Local Storage. Use this option to
6742 disable the optimization.
6744 @cindex PowerPC64 OPD optimization
6745 @kindex --no-opd-optimize
6746 @item --no-opd-optimize
6747 PowerPC64 @command{ld} normally removes @code{.opd} section entries
6748 corresponding to deleted link-once functions, or functions removed by
6749 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
6750 Use this option to disable @code{.opd} optimization.
6752 @cindex PowerPC64 OPD spacing
6753 @kindex --non-overlapping-opd
6754 @item --non-overlapping-opd
6755 Some PowerPC64 compilers have an option to generate compressed
6756 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
6757 the static chain pointer (unused in C) with the first word of the next
6758 entry. This option expands such entries to the full 24 bytes.
6760 @cindex PowerPC64 TOC optimization
6761 @kindex --no-toc-optimize
6762 @item --no-toc-optimize
6763 PowerPC64 @command{ld} normally removes unused @code{.toc} section
6764 entries. Such entries are detected by examining relocations that
6765 reference the TOC in code sections. A reloc in a deleted code section
6766 marks a TOC word as unneeded, while a reloc in a kept code section
6767 marks a TOC word as needed. Since the TOC may reference itself, TOC
6768 relocs are also examined. TOC words marked as both needed and
6769 unneeded will of course be kept. TOC words without any referencing
6770 reloc are assumed to be part of a multi-word entry, and are kept or
6771 discarded as per the nearest marked preceding word. This works
6772 reliably for compiler generated code, but may be incorrect if assembly
6773 code is used to insert TOC entries. Use this option to disable the
6776 @cindex PowerPC64 multi-TOC
6777 @kindex --no-multi-toc
6778 @item --no-multi-toc
6779 If given any toc option besides @code{-mcmodel=medium} or
6780 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
6782 entries are accessed with a 16-bit offset from r2. This limits the
6783 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
6784 grouping code sections such that each group uses less than 64K for its
6785 TOC entries, then inserts r2 adjusting stubs between inter-group
6786 calls. @command{ld} does not split apart input sections, so cannot
6787 help if a single input file has a @code{.toc} section that exceeds
6788 64K, most likely from linking multiple files with @command{ld -r}.
6789 Use this option to turn off this feature.
6791 @cindex PowerPC64 TOC sorting
6792 @kindex --no-toc-sort
6794 By default, @command{ld} sorts TOC sections so that those whose file
6795 happens to have a section called @code{.init} or @code{.fini} are
6796 placed first, followed by TOC sections referenced by code generated
6797 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
6798 referenced only by code generated with PowerPC64 gcc's
6799 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
6800 results in better TOC grouping for multi-TOC. Use this option to turn
6803 @cindex PowerPC64 PLT stub alignment
6805 @kindex --no-plt-align
6807 @itemx --no-plt-align
6808 Use these options to control whether individual PLT call stubs are
6809 aligned to a 32-byte boundary, or to the specified power of two
6810 boundary when using @code{--plt-align=}. By default PLT call stubs
6813 @cindex PowerPC64 PLT call stub static chain
6814 @kindex --plt-static-chain
6815 @kindex --no-plt-static-chain
6816 @item --plt-static-chain
6817 @itemx --no-plt-static-chain
6818 Use these options to control whether PLT call stubs load the static
6819 chain pointer (r11). @code{ld} defaults to not loading the static
6820 chain since there is never any need to do so on a PLT call.
6822 @cindex PowerPC64 PLT call stub thread safety
6823 @kindex --plt-thread-safe
6824 @kindex --no-plt-thread-safe
6825 @item --plt-thread-safe
6826 @itemx --no-thread-safe
6827 With power7's weakly ordered memory model, it is possible when using
6828 lazy binding for ld.so to update a plt entry in one thread and have
6829 another thread see the individual plt entry words update in the wrong
6830 order, despite ld.so carefully writing in the correct order and using
6831 memory write barriers. To avoid this we need some sort of read
6832 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
6833 looks for calls to commonly used functions that create threads, and if
6834 seen, adds the necessary barriers. Use these options to change the
6849 @section @command{ld} and SPU ELF Support
6851 @cindex SPU ELF options
6857 This option marks an executable as a PIC plugin module.
6859 @cindex SPU overlays
6860 @kindex --no-overlays
6862 Normally, @command{ld} recognizes calls to functions within overlay
6863 regions, and redirects such calls to an overlay manager via a stub.
6864 @command{ld} also provides a built-in overlay manager. This option
6865 turns off all this special overlay handling.
6867 @cindex SPU overlay stub symbols
6868 @kindex --emit-stub-syms
6869 @item --emit-stub-syms
6870 This option causes @command{ld} to label overlay stubs with a local
6871 symbol that encodes the stub type and destination.
6873 @cindex SPU extra overlay stubs
6874 @kindex --extra-overlay-stubs
6875 @item --extra-overlay-stubs
6876 This option causes @command{ld} to add overlay call stubs on all
6877 function calls out of overlay regions. Normally stubs are not added
6878 on calls to non-overlay regions.
6880 @cindex SPU local store size
6881 @kindex --local-store=lo:hi
6882 @item --local-store=lo:hi
6883 @command{ld} usually checks that a final executable for SPU fits in
6884 the address range 0 to 256k. This option may be used to change the
6885 range. Disable the check entirely with @option{--local-store=0:0}.
6888 @kindex --stack-analysis
6889 @item --stack-analysis
6890 SPU local store space is limited. Over-allocation of stack space
6891 unnecessarily limits space available for code and data, while
6892 under-allocation results in runtime failures. If given this option,
6893 @command{ld} will provide an estimate of maximum stack usage.
6894 @command{ld} does this by examining symbols in code sections to
6895 determine the extents of functions, and looking at function prologues
6896 for stack adjusting instructions. A call-graph is created by looking
6897 for relocations on branch instructions. The graph is then searched
6898 for the maximum stack usage path. Note that this analysis does not
6899 find calls made via function pointers, and does not handle recursion
6900 and other cycles in the call graph. Stack usage may be
6901 under-estimated if your code makes such calls. Also, stack usage for
6902 dynamic allocation, e.g. alloca, will not be detected. If a link map
6903 is requested, detailed information about each function's stack usage
6904 and calls will be given.
6907 @kindex --emit-stack-syms
6908 @item --emit-stack-syms
6909 This option, if given along with @option{--stack-analysis} will result
6910 in @command{ld} emitting stack sizing symbols for each function.
6911 These take the form @code{__stack_<function_name>} for global
6912 functions, and @code{__stack_<number>_<function_name>} for static
6913 functions. @code{<number>} is the section id in hex. The value of
6914 such symbols is the stack requirement for the corresponding function.
6915 The symbol size will be zero, type @code{STT_NOTYPE}, binding
6916 @code{STB_LOCAL}, and section @code{SHN_ABS}.
6930 @section @command{ld}'s Support for Various TI COFF Versions
6931 @cindex TI COFF versions
6932 @kindex --format=@var{version}
6933 The @samp{--format} switch allows selection of one of the various
6934 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
6935 also supported. The TI COFF versions also vary in header byte-order
6936 format; @command{ld} will read any version or byte order, but the output
6937 header format depends on the default specified by the specific target.
6950 @section @command{ld} and WIN32 (cygwin/mingw)
6952 This section describes some of the win32 specific @command{ld} issues.
6953 See @ref{Options,,Command Line Options} for detailed description of the
6954 command line options mentioned here.
6957 @cindex import libraries
6958 @item import libraries
6959 The standard Windows linker creates and uses so-called import
6960 libraries, which contains information for linking to dll's. They are
6961 regular static archives and are handled as any other static
6962 archive. The cygwin and mingw ports of @command{ld} have specific
6963 support for creating such libraries provided with the
6964 @samp{--out-implib} command line option.
6966 @item exporting DLL symbols
6967 @cindex exporting DLL symbols
6968 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6971 @item using auto-export functionality
6972 @cindex using auto-export functionality
6973 By default @command{ld} exports symbols with the auto-export functionality,
6974 which is controlled by the following command line options:
6977 @item --export-all-symbols [This is the default]
6978 @item --exclude-symbols
6979 @item --exclude-libs
6980 @item --exclude-modules-for-implib
6981 @item --version-script
6984 When auto-export is in operation, @command{ld} will export all the non-local
6985 (global and common) symbols it finds in a DLL, with the exception of a few
6986 symbols known to belong to the system's runtime and libraries. As it will
6987 often not be desirable to export all of a DLL's symbols, which may include
6988 private functions that are not part of any public interface, the command-line
6989 options listed above may be used to filter symbols out from the list for
6990 exporting. The @samp{--output-def} option can be used in order to see the
6991 final list of exported symbols with all exclusions taken into effect.
6993 If @samp{--export-all-symbols} is not given explicitly on the
6994 command line, then the default auto-export behavior will be @emph{disabled}
6995 if either of the following are true:
6998 @item A DEF file is used.
6999 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
7002 @item using a DEF file
7003 @cindex using a DEF file
7004 Another way of exporting symbols is using a DEF file. A DEF file is
7005 an ASCII file containing definitions of symbols which should be
7006 exported when a dll is created. Usually it is named @samp{<dll
7007 name>.def} and is added as any other object file to the linker's
7008 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7011 gcc -o <output> <objectfiles> <dll name>.def
7014 Using a DEF file turns off the normal auto-export behavior, unless the
7015 @samp{--export-all-symbols} option is also used.
7017 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7020 LIBRARY "xyz.dll" BASE=0x20000000
7026 another_foo = abc.dll.afoo
7032 This example defines a DLL with a non-default base address and seven
7033 symbols in the export table. The third exported symbol @code{_bar} is an
7034 alias for the second. The fourth symbol, @code{another_foo} is resolved
7035 by "forwarding" to another module and treating it as an alias for
7036 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7037 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7038 export library is an alias of @samp{foo}, which gets the string name
7039 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7040 symbol, which gets in export table the name @samp{var1}.
7042 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7043 name of the output DLL. If @samp{<name>} does not include a suffix,
7044 the default library suffix, @samp{.DLL} is appended.
7046 When the .DEF file is used to build an application, rather than a
7047 library, the @code{NAME <name>} command should be used instead of
7048 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7049 executable suffix, @samp{.EXE} is appended.
7051 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7052 specification @code{BASE = <number>} may be used to specify a
7053 non-default base address for the image.
7055 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7056 or they specify an empty string, the internal name is the same as the
7057 filename specified on the command line.
7059 The complete specification of an export symbol is:
7063 ( ( ( <name1> [ = <name2> ] )
7064 | ( <name1> = <module-name> . <external-name>))
7065 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7068 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7069 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7070 @samp{<name1>} as a "forward" alias for the symbol
7071 @samp{<external-name>} in the DLL @samp{<module-name>}.
7072 Optionally, the symbol may be exported by the specified ordinal
7073 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7074 string in import/export table for the symbol.
7076 The optional keywords that follow the declaration indicate:
7078 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7079 will still be exported by its ordinal alias (either the value specified
7080 by the .def specification or, otherwise, the value assigned by the
7081 linker). The symbol name, however, does remain visible in the import
7082 library (if any), unless @code{PRIVATE} is also specified.
7084 @code{DATA}: The symbol is a variable or object, rather than a function.
7085 The import lib will export only an indirect reference to @code{foo} as
7086 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7089 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7090 well as @code{_imp__foo} into the import library. Both refer to the
7091 read-only import address table's pointer to the variable, not to the
7092 variable itself. This can be dangerous. If the user code fails to add
7093 the @code{dllimport} attribute and also fails to explicitly add the
7094 extra indirection that the use of the attribute enforces, the
7095 application will behave unexpectedly.
7097 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7098 it into the static import library used to resolve imports at link time. The
7099 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7100 API at runtime or by by using the GNU ld extension of linking directly to
7101 the DLL without an import library.
7103 See ld/deffilep.y in the binutils sources for the full specification of
7104 other DEF file statements
7106 @cindex creating a DEF file
7107 While linking a shared dll, @command{ld} is able to create a DEF file
7108 with the @samp{--output-def <file>} command line option.
7110 @item Using decorations
7111 @cindex Using decorations
7112 Another way of marking symbols for export is to modify the source code
7113 itself, so that when building the DLL each symbol to be exported is
7117 __declspec(dllexport) int a_variable
7118 __declspec(dllexport) void a_function(int with_args)
7121 All such symbols will be exported from the DLL. If, however,
7122 any of the object files in the DLL contain symbols decorated in
7123 this way, then the normal auto-export behavior is disabled, unless
7124 the @samp{--export-all-symbols} option is also used.
7126 Note that object files that wish to access these symbols must @emph{not}
7127 decorate them with dllexport. Instead, they should use dllimport,
7131 __declspec(dllimport) int a_variable
7132 __declspec(dllimport) void a_function(int with_args)
7135 This complicates the structure of library header files, because
7136 when included by the library itself the header must declare the
7137 variables and functions as dllexport, but when included by client
7138 code the header must declare them as dllimport. There are a number
7139 of idioms that are typically used to do this; often client code can
7140 omit the __declspec() declaration completely. See
7141 @samp{--enable-auto-import} and @samp{automatic data imports} for more
7145 @cindex automatic data imports
7146 @item automatic data imports
7147 The standard Windows dll format supports data imports from dlls only
7148 by adding special decorations (dllimport/dllexport), which let the
7149 compiler produce specific assembler instructions to deal with this
7150 issue. This increases the effort necessary to port existing Un*x
7151 code to these platforms, especially for large
7152 c++ libraries and applications. The auto-import feature, which was
7153 initially provided by Paul Sokolovsky, allows one to omit the
7154 decorations to achieve a behavior that conforms to that on POSIX/Un*x
7155 platforms. This feature is enabled with the @samp{--enable-auto-import}
7156 command-line option, although it is enabled by default on cygwin/mingw.
7157 The @samp{--enable-auto-import} option itself now serves mainly to
7158 suppress any warnings that are ordinarily emitted when linked objects
7159 trigger the feature's use.
7161 auto-import of variables does not always work flawlessly without
7162 additional assistance. Sometimes, you will see this message
7164 "variable '<var>' can't be auto-imported. Please read the
7165 documentation for ld's @code{--enable-auto-import} for details."
7167 The @samp{--enable-auto-import} documentation explains why this error
7168 occurs, and several methods that can be used to overcome this difficulty.
7169 One of these methods is the @emph{runtime pseudo-relocs} feature, described
7172 @cindex runtime pseudo-relocation
7173 For complex variables imported from DLLs (such as structs or classes),
7174 object files typically contain a base address for the variable and an
7175 offset (@emph{addend}) within the variable--to specify a particular
7176 field or public member, for instance. Unfortunately, the runtime loader used
7177 in win32 environments is incapable of fixing these references at runtime
7178 without the additional information supplied by dllimport/dllexport decorations.
7179 The standard auto-import feature described above is unable to resolve these
7182 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
7183 be resolved without error, while leaving the task of adjusting the references
7184 themselves (with their non-zero addends) to specialized code provided by the
7185 runtime environment. Recent versions of the cygwin and mingw environments and
7186 compilers provide this runtime support; older versions do not. However, the
7187 support is only necessary on the developer's platform; the compiled result will
7188 run without error on an older system.
7190 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
7193 @cindex direct linking to a dll
7194 @item direct linking to a dll
7195 The cygwin/mingw ports of @command{ld} support the direct linking,
7196 including data symbols, to a dll without the usage of any import
7197 libraries. This is much faster and uses much less memory than does the
7198 traditional import library method, especially when linking large
7199 libraries or applications. When @command{ld} creates an import lib, each
7200 function or variable exported from the dll is stored in its own bfd, even
7201 though a single bfd could contain many exports. The overhead involved in
7202 storing, loading, and processing so many bfd's is quite large, and explains the
7203 tremendous time, memory, and storage needed to link against particularly
7204 large or complex libraries when using import libs.
7206 Linking directly to a dll uses no extra command-line switches other than
7207 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
7208 of names to match each library. All that is needed from the developer's
7209 perspective is an understanding of this search, in order to force ld to
7210 select the dll instead of an import library.
7213 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
7214 to find, in the first directory of its search path,
7226 before moving on to the next directory in the search path.
7228 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
7229 where @samp{<prefix>} is set by the @command{ld} option
7230 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
7231 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
7234 Other win32-based unix environments, such as mingw or pw32, may use other
7235 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
7236 was originally intended to help avoid name conflicts among dll's built for the
7237 various win32/un*x environments, so that (for example) two versions of a zlib dll
7238 could coexist on the same machine.
7240 The generic cygwin/mingw path layout uses a @samp{bin} directory for
7241 applications and dll's and a @samp{lib} directory for the import
7242 libraries (using cygwin nomenclature):
7248 libxxx.dll.a (in case of dll's)
7249 libxxx.a (in case of static archive)
7252 Linking directly to a dll without using the import library can be
7255 1. Use the dll directly by adding the @samp{bin} path to the link line
7257 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
7260 However, as the dll's often have version numbers appended to their names
7261 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
7262 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
7263 not versioned, and do not have this difficulty.
7265 2. Create a symbolic link from the dll to a file in the @samp{lib}
7266 directory according to the above mentioned search pattern. This
7267 should be used to avoid unwanted changes in the tools needed for
7271 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
7274 Then you can link without any make environment changes.
7277 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
7280 This technique also avoids the version number problems, because the following is
7287 libxxx.dll.a -> ../bin/cygxxx-5.dll
7290 Linking directly to a dll without using an import lib will work
7291 even when auto-import features are exercised, and even when
7292 @samp{--enable-runtime-pseudo-relocs} is used.
7294 Given the improvements in speed and memory usage, one might justifiably
7295 wonder why import libraries are used at all. There are three reasons:
7297 1. Until recently, the link-directly-to-dll functionality did @emph{not}
7298 work with auto-imported data.
7300 2. Sometimes it is necessary to include pure static objects within the
7301 import library (which otherwise contains only bfd's for indirection
7302 symbols that point to the exports of a dll). Again, the import lib
7303 for the cygwin kernel makes use of this ability, and it is not
7304 possible to do this without an import lib.
7306 3. Symbol aliases can only be resolved using an import lib. This is
7307 critical when linking against OS-supplied dll's (eg, the win32 API)
7308 in which symbols are usually exported as undecorated aliases of their
7309 stdcall-decorated assembly names.
7311 So, import libs are not going away. But the ability to replace
7312 true import libs with a simple symbolic link to (or a copy of)
7313 a dll, in many cases, is a useful addition to the suite of tools
7314 binutils makes available to the win32 developer. Given the
7315 massive improvements in memory requirements during linking, storage
7316 requirements, and linking speed, we expect that many developers
7317 will soon begin to use this feature whenever possible.
7319 @item symbol aliasing
7321 @item adding additional names
7322 Sometimes, it is useful to export symbols with additional names.
7323 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
7324 exported as @samp{_foo} by using special directives in the DEF file
7325 when creating the dll. This will affect also the optional created
7326 import library. Consider the following DEF file:
7329 LIBRARY "xyz.dll" BASE=0x61000000
7336 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
7338 Another method for creating a symbol alias is to create it in the
7339 source code using the "weak" attribute:
7342 void foo () @{ /* Do something. */; @}
7343 void _foo () __attribute__ ((weak, alias ("foo")));
7346 See the gcc manual for more information about attributes and weak
7349 @item renaming symbols
7350 Sometimes it is useful to rename exports. For instance, the cygwin
7351 kernel does this regularly. A symbol @samp{_foo} can be exported as
7352 @samp{foo} but not as @samp{_foo} by using special directives in the
7353 DEF file. (This will also affect the import library, if it is
7354 created). In the following example:
7357 LIBRARY "xyz.dll" BASE=0x61000000
7363 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
7367 Note: using a DEF file disables the default auto-export behavior,
7368 unless the @samp{--export-all-symbols} command line option is used.
7369 If, however, you are trying to rename symbols, then you should list
7370 @emph{all} desired exports in the DEF file, including the symbols
7371 that are not being renamed, and do @emph{not} use the
7372 @samp{--export-all-symbols} option. If you list only the
7373 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
7374 to handle the other symbols, then the both the new names @emph{and}
7375 the original names for the renamed symbols will be exported.
7376 In effect, you'd be aliasing those symbols, not renaming them,
7377 which is probably not what you wanted.
7379 @cindex weak externals
7380 @item weak externals
7381 The Windows object format, PE, specifies a form of weak symbols called
7382 weak externals. When a weak symbol is linked and the symbol is not
7383 defined, the weak symbol becomes an alias for some other symbol. There
7384 are three variants of weak externals:
7386 @item Definition is searched for in objects and libraries, historically
7387 called lazy externals.
7388 @item Definition is searched for only in other objects, not in libraries.
7389 This form is not presently implemented.
7390 @item No search; the symbol is an alias. This form is not presently
7393 As a GNU extension, weak symbols that do not specify an alternate symbol
7394 are supported. If the symbol is undefined when linking, the symbol
7395 uses a default value.
7397 @cindex aligned common symbols
7398 @item aligned common symbols
7399 As a GNU extension to the PE file format, it is possible to specify the
7400 desired alignment for a common symbol. This information is conveyed from
7401 the assembler or compiler to the linker by means of GNU-specific commands
7402 carried in the object file's @samp{.drectve} section, which are recognized
7403 by @command{ld} and respected when laying out the common symbols. Native
7404 tools will be able to process object files employing this GNU extension,
7405 but will fail to respect the alignment instructions, and may issue noisy
7406 warnings about unknown linker directives.
7420 @section @code{ld} and Xtensa Processors
7422 @cindex Xtensa processors
7423 The default @command{ld} behavior for Xtensa processors is to interpret
7424 @code{SECTIONS} commands so that lists of explicitly named sections in a
7425 specification with a wildcard file will be interleaved when necessary to
7426 keep literal pools within the range of PC-relative load offsets. For
7427 example, with the command:
7439 @command{ld} may interleave some of the @code{.literal}
7440 and @code{.text} sections from different object files to ensure that the
7441 literal pools are within the range of PC-relative load offsets. A valid
7442 interleaving might place the @code{.literal} sections from an initial
7443 group of files followed by the @code{.text} sections of that group of
7444 files. Then, the @code{.literal} sections from the rest of the files
7445 and the @code{.text} sections from the rest of the files would follow.
7447 @cindex @option{--relax} on Xtensa
7448 @cindex relaxing on Xtensa
7449 Relaxation is enabled by default for the Xtensa version of @command{ld} and
7450 provides two important link-time optimizations. The first optimization
7451 is to combine identical literal values to reduce code size. A redundant
7452 literal will be removed and all the @code{L32R} instructions that use it
7453 will be changed to reference an identical literal, as long as the
7454 location of the replacement literal is within the offset range of all
7455 the @code{L32R} instructions. The second optimization is to remove
7456 unnecessary overhead from assembler-generated ``longcall'' sequences of
7457 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
7458 range of direct @code{CALL@var{n}} instructions.
7460 For each of these cases where an indirect call sequence can be optimized
7461 to a direct call, the linker will change the @code{CALLX@var{n}}
7462 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
7463 instruction, and remove the literal referenced by the @code{L32R}
7464 instruction if it is not used for anything else. Removing the
7465 @code{L32R} instruction always reduces code size but can potentially
7466 hurt performance by changing the alignment of subsequent branch targets.
7467 By default, the linker will always preserve alignments, either by
7468 switching some instructions between 24-bit encodings and the equivalent
7469 density instructions or by inserting a no-op in place of the @code{L32R}
7470 instruction that was removed. If code size is more important than
7471 performance, the @option{--size-opt} option can be used to prevent the
7472 linker from widening density instructions or inserting no-ops, except in
7473 a few cases where no-ops are required for correctness.
7475 The following Xtensa-specific command-line options can be used to
7478 @cindex Xtensa options
7481 When optimizing indirect calls to direct calls, optimize for code size
7482 more than performance. With this option, the linker will not insert
7483 no-ops or widen density instructions to preserve branch target
7484 alignment. There may still be some cases where no-ops are required to
7485 preserve the correctness of the code.
7493 @ifclear SingleFormat
7498 @cindex object file management
7499 @cindex object formats available
7501 The linker accesses object and archive files using the BFD libraries.
7502 These libraries allow the linker to use the same routines to operate on
7503 object files whatever the object file format. A different object file
7504 format can be supported simply by creating a new BFD back end and adding
7505 it to the library. To conserve runtime memory, however, the linker and
7506 associated tools are usually configured to support only a subset of the
7507 object file formats available. You can use @code{objdump -i}
7508 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
7509 list all the formats available for your configuration.
7511 @cindex BFD requirements
7512 @cindex requirements for BFD
7513 As with most implementations, BFD is a compromise between
7514 several conflicting requirements. The major factor influencing
7515 BFD design was efficiency: any time used converting between
7516 formats is time which would not have been spent had BFD not
7517 been involved. This is partly offset by abstraction payback; since
7518 BFD simplifies applications and back ends, more time and care
7519 may be spent optimizing algorithms for a greater speed.
7521 One minor artifact of the BFD solution which you should bear in
7522 mind is the potential for information loss. There are two places where
7523 useful information can be lost using the BFD mechanism: during
7524 conversion and during output. @xref{BFD information loss}.
7527 * BFD outline:: How it works: an outline of BFD
7531 @section How It Works: An Outline of BFD
7532 @cindex opening object files
7533 @include bfdsumm.texi
7536 @node Reporting Bugs
7537 @chapter Reporting Bugs
7538 @cindex bugs in @command{ld}
7539 @cindex reporting bugs in @command{ld}
7541 Your bug reports play an essential role in making @command{ld} reliable.
7543 Reporting a bug may help you by bringing a solution to your problem, or
7544 it may not. But in any case the principal function of a bug report is
7545 to help the entire community by making the next version of @command{ld}
7546 work better. Bug reports are your contribution to the maintenance of
7549 In order for a bug report to serve its purpose, you must include the
7550 information that enables us to fix the bug.
7553 * Bug Criteria:: Have you found a bug?
7554 * Bug Reporting:: How to report bugs
7558 @section Have You Found a Bug?
7559 @cindex bug criteria
7561 If you are not sure whether you have found a bug, here are some guidelines:
7564 @cindex fatal signal
7565 @cindex linker crash
7566 @cindex crash of linker
7568 If the linker gets a fatal signal, for any input whatever, that is a
7569 @command{ld} bug. Reliable linkers never crash.
7571 @cindex error on valid input
7573 If @command{ld} produces an error message for valid input, that is a bug.
7575 @cindex invalid input
7577 If @command{ld} does not produce an error message for invalid input, that
7578 may be a bug. In the general case, the linker can not verify that
7579 object files are correct.
7582 If you are an experienced user of linkers, your suggestions for
7583 improvement of @command{ld} are welcome in any case.
7587 @section How to Report Bugs
7589 @cindex @command{ld} bugs, reporting
7591 A number of companies and individuals offer support for @sc{gnu}
7592 products. If you obtained @command{ld} from a support organization, we
7593 recommend you contact that organization first.
7595 You can find contact information for many support companies and
7596 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
7600 Otherwise, send bug reports for @command{ld} to
7604 The fundamental principle of reporting bugs usefully is this:
7605 @strong{report all the facts}. If you are not sure whether to state a
7606 fact or leave it out, state it!
7608 Often people omit facts because they think they know what causes the
7609 problem and assume that some details do not matter. Thus, you might
7610 assume that the name of a symbol you use in an example does not
7611 matter. Well, probably it does not, but one cannot be sure. Perhaps
7612 the bug is a stray memory reference which happens to fetch from the
7613 location where that name is stored in memory; perhaps, if the name
7614 were different, the contents of that location would fool the linker
7615 into doing the right thing despite the bug. Play it safe and give a
7616 specific, complete example. That is the easiest thing for you to do,
7617 and the most helpful.
7619 Keep in mind that the purpose of a bug report is to enable us to fix
7620 the bug if it is new to us. Therefore, always write your bug reports
7621 on the assumption that the bug has not been reported previously.
7623 Sometimes people give a few sketchy facts and ask, ``Does this ring a
7624 bell?'' This cannot help us fix a bug, so it is basically useless. We
7625 respond by asking for enough details to enable us to investigate.
7626 You might as well expedite matters by sending them to begin with.
7628 To enable us to fix the bug, you should include all these things:
7632 The version of @command{ld}. @command{ld} announces it if you start it with
7633 the @samp{--version} argument.
7635 Without this, we will not know whether there is any point in looking for
7636 the bug in the current version of @command{ld}.
7639 Any patches you may have applied to the @command{ld} source, including any
7640 patches made to the @code{BFD} library.
7643 The type of machine you are using, and the operating system name and
7647 What compiler (and its version) was used to compile @command{ld}---e.g.
7651 The command arguments you gave the linker to link your example and
7652 observe the bug. To guarantee you will not omit something important,
7653 list them all. A copy of the Makefile (or the output from make) is
7656 If we were to try to guess the arguments, we would probably guess wrong
7657 and then we might not encounter the bug.
7660 A complete input file, or set of input files, that will reproduce the
7661 bug. It is generally most helpful to send the actual object files
7662 provided that they are reasonably small. Say no more than 10K. For
7663 bigger files you can either make them available by FTP or HTTP or else
7664 state that you are willing to send the object file(s) to whomever
7665 requests them. (Note - your email will be going to a mailing list, so
7666 we do not want to clog it up with large attachments). But small
7667 attachments are best.
7669 If the source files were assembled using @code{gas} or compiled using
7670 @code{gcc}, then it may be OK to send the source files rather than the
7671 object files. In this case, be sure to say exactly what version of
7672 @code{gas} or @code{gcc} was used to produce the object files. Also say
7673 how @code{gas} or @code{gcc} were configured.
7676 A description of what behavior you observe that you believe is
7677 incorrect. For example, ``It gets a fatal signal.''
7679 Of course, if the bug is that @command{ld} gets a fatal signal, then we
7680 will certainly notice it. But if the bug is incorrect output, we might
7681 not notice unless it is glaringly wrong. You might as well not give us
7682 a chance to make a mistake.
7684 Even if the problem you experience is a fatal signal, you should still
7685 say so explicitly. Suppose something strange is going on, such as, your
7686 copy of @command{ld} is out of sync, or you have encountered a bug in the
7687 C library on your system. (This has happened!) Your copy might crash
7688 and ours would not. If you told us to expect a crash, then when ours
7689 fails to crash, we would know that the bug was not happening for us. If
7690 you had not told us to expect a crash, then we would not be able to draw
7691 any conclusion from our observations.
7694 If you wish to suggest changes to the @command{ld} source, send us context
7695 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
7696 @samp{-p} option. Always send diffs from the old file to the new file.
7697 If you even discuss something in the @command{ld} source, refer to it by
7698 context, not by line number.
7700 The line numbers in our development sources will not match those in your
7701 sources. Your line numbers would convey no useful information to us.
7704 Here are some things that are not necessary:
7708 A description of the envelope of the bug.
7710 Often people who encounter a bug spend a lot of time investigating
7711 which changes to the input file will make the bug go away and which
7712 changes will not affect it.
7714 This is often time consuming and not very useful, because the way we
7715 will find the bug is by running a single example under the debugger
7716 with breakpoints, not by pure deduction from a series of examples.
7717 We recommend that you save your time for something else.
7719 Of course, if you can find a simpler example to report @emph{instead}
7720 of the original one, that is a convenience for us. Errors in the
7721 output will be easier to spot, running under the debugger will take
7722 less time, and so on.
7724 However, simplification is not vital; if you do not want to do this,
7725 report the bug anyway and send us the entire test case you used.
7728 A patch for the bug.
7730 A patch for the bug does help us if it is a good one. But do not omit
7731 the necessary information, such as the test case, on the assumption that
7732 a patch is all we need. We might see problems with your patch and decide
7733 to fix the problem another way, or we might not understand it at all.
7735 Sometimes with a program as complicated as @command{ld} it is very hard to
7736 construct an example that will make the program follow a certain path
7737 through the code. If you do not send us the example, we will not be
7738 able to construct one, so we will not be able to verify that the bug is
7741 And if we cannot understand what bug you are trying to fix, or why your
7742 patch should be an improvement, we will not install it. A test case will
7743 help us to understand.
7746 A guess about what the bug is or what it depends on.
7748 Such guesses are usually wrong. Even we cannot guess right about such
7749 things without first using the debugger to find the facts.
7753 @appendix MRI Compatible Script Files
7754 @cindex MRI compatibility
7755 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
7756 linker, @command{ld} can use MRI compatible linker scripts as an
7757 alternative to the more general-purpose linker scripting language
7758 described in @ref{Scripts}. MRI compatible linker scripts have a much
7759 simpler command set than the scripting language otherwise used with
7760 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
7761 linker commands; these commands are described here.
7763 In general, MRI scripts aren't of much use with the @code{a.out} object
7764 file format, since it only has three sections and MRI scripts lack some
7765 features to make use of them.
7767 You can specify a file containing an MRI-compatible script using the
7768 @samp{-c} command-line option.
7770 Each command in an MRI-compatible script occupies its own line; each
7771 command line starts with the keyword that identifies the command (though
7772 blank lines are also allowed for punctuation). If a line of an
7773 MRI-compatible script begins with an unrecognized keyword, @command{ld}
7774 issues a warning message, but continues processing the script.
7776 Lines beginning with @samp{*} are comments.
7778 You can write these commands using all upper-case letters, or all
7779 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
7780 The following list shows only the upper-case form of each command.
7783 @cindex @code{ABSOLUTE} (MRI)
7784 @item ABSOLUTE @var{secname}
7785 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
7786 Normally, @command{ld} includes in the output file all sections from all
7787 the input files. However, in an MRI-compatible script, you can use the
7788 @code{ABSOLUTE} command to restrict the sections that will be present in
7789 your output program. If the @code{ABSOLUTE} command is used at all in a
7790 script, then only the sections named explicitly in @code{ABSOLUTE}
7791 commands will appear in the linker output. You can still use other
7792 input sections (whatever you select on the command line, or using
7793 @code{LOAD}) to resolve addresses in the output file.
7795 @cindex @code{ALIAS} (MRI)
7796 @item ALIAS @var{out-secname}, @var{in-secname}
7797 Use this command to place the data from input section @var{in-secname}
7798 in a section called @var{out-secname} in the linker output file.
7800 @var{in-secname} may be an integer.
7802 @cindex @code{ALIGN} (MRI)
7803 @item ALIGN @var{secname} = @var{expression}
7804 Align the section called @var{secname} to @var{expression}. The
7805 @var{expression} should be a power of two.
7807 @cindex @code{BASE} (MRI)
7808 @item BASE @var{expression}
7809 Use the value of @var{expression} as the lowest address (other than
7810 absolute addresses) in the output file.
7812 @cindex @code{CHIP} (MRI)
7813 @item CHIP @var{expression}
7814 @itemx CHIP @var{expression}, @var{expression}
7815 This command does nothing; it is accepted only for compatibility.
7817 @cindex @code{END} (MRI)
7819 This command does nothing whatever; it's only accepted for compatibility.
7821 @cindex @code{FORMAT} (MRI)
7822 @item FORMAT @var{output-format}
7823 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
7824 language, but restricted to one of these output formats:
7828 S-records, if @var{output-format} is @samp{S}
7831 IEEE, if @var{output-format} is @samp{IEEE}
7834 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
7838 @cindex @code{LIST} (MRI)
7839 @item LIST @var{anything}@dots{}
7840 Print (to the standard output file) a link map, as produced by the
7841 @command{ld} command-line option @samp{-M}.
7843 The keyword @code{LIST} may be followed by anything on the
7844 same line, with no change in its effect.
7846 @cindex @code{LOAD} (MRI)
7847 @item LOAD @var{filename}
7848 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
7849 Include one or more object file @var{filename} in the link; this has the
7850 same effect as specifying @var{filename} directly on the @command{ld}
7853 @cindex @code{NAME} (MRI)
7854 @item NAME @var{output-name}
7855 @var{output-name} is the name for the program produced by @command{ld}; the
7856 MRI-compatible command @code{NAME} is equivalent to the command-line
7857 option @samp{-o} or the general script language command @code{OUTPUT}.
7859 @cindex @code{ORDER} (MRI)
7860 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
7861 @itemx ORDER @var{secname} @var{secname} @var{secname}
7862 Normally, @command{ld} orders the sections in its output file in the
7863 order in which they first appear in the input files. In an MRI-compatible
7864 script, you can override this ordering with the @code{ORDER} command. The
7865 sections you list with @code{ORDER} will appear first in your output
7866 file, in the order specified.
7868 @cindex @code{PUBLIC} (MRI)
7869 @item PUBLIC @var{name}=@var{expression}
7870 @itemx PUBLIC @var{name},@var{expression}
7871 @itemx PUBLIC @var{name} @var{expression}
7872 Supply a value (@var{expression}) for external symbol
7873 @var{name} used in the linker input files.
7875 @cindex @code{SECT} (MRI)
7876 @item SECT @var{secname}, @var{expression}
7877 @itemx SECT @var{secname}=@var{expression}
7878 @itemx SECT @var{secname} @var{expression}
7879 You can use any of these three forms of the @code{SECT} command to
7880 specify the start address (@var{expression}) for section @var{secname}.
7881 If you have more than one @code{SECT} statement for the same
7882 @var{secname}, only the @emph{first} sets the start address.
7885 @node GNU Free Documentation License
7886 @appendix GNU Free Documentation License
7890 @unnumbered LD Index
7895 % I think something like @@colophon should be in texinfo. In the
7897 \long\def\colophon{\hbox to0pt{}\vfill
7898 \centerline{The body of this manual is set in}
7899 \centerline{\fontname\tenrm,}
7900 \centerline{with headings in {\bf\fontname\tenbf}}
7901 \centerline{and examples in {\tt\fontname\tentt}.}
7902 \centerline{{\it\fontname\tenit\/} and}
7903 \centerline{{\sl\fontname\tensl\/}}
7904 \centerline{are used for emphasis.}\vfill}
7906 % Blame: doc@@cygnus.com, 28mar91.