3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
6 @include configdoc.texi
7 @c (configdoc.texi is generated by the Makefile)
12 @macro gcctabopt{body}
18 @c Configure for the generation of man pages
55 * Ld: (ld). The GNU linker.
61 This file documents the @sc{gnu} linker LD version @value{VERSION}.
63 Copyright (C) 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
64 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
68 Permission is granted to copy, distribute and/or modify this document
69 under the terms of the GNU Free Documentation License, Version 1.1
70 or any later version published by the Free Software Foundation;
71 with no Invariant Sections, with no Front-Cover Texts, and with no
72 Back-Cover Texts. A copy of the license is included in the
73 section entitled ``GNU Free Documentation License''.
75 Permission is granted to process this file through Tex and print the
76 results, provided the printed document carries copying permission
77 notice identical to this one except for the removal of this paragraph
78 (this paragraph not being relevant to the printed manual).
84 @setchapternewpage odd
85 @settitle Using LD, the GNU linker
88 @subtitle The GNU linker
90 @subtitle @code{ld} version 2
91 @subtitle Version @value{VERSION}
92 @author Steve Chamberlain
93 @author Ian Lance Taylor
98 \hfill Red Hat Inc\par
99 \hfill nickc\@credhat.com, doc\@redhat.com\par
100 \hfill {\it Using LD, the GNU linker}\par
101 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
103 \global\parindent=0pt % Steve likes it this way.
106 @vskip 0pt plus 1filll
107 @c man begin COPYRIGHT
108 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
109 2002, 2003, 2004 Free Software Foundation, Inc.
111 Permission is granted to copy, distribute and/or modify this document
112 under the terms of the GNU Free Documentation License, Version 1.1
113 or any later version published by the Free Software Foundation;
114 with no Invariant Sections, with no Front-Cover Texts, and with no
115 Back-Cover Texts. A copy of the license is included in the
116 section entitled ``GNU Free Documentation License''.
121 @c FIXME: Talk about importance of *order* of args, cmds to linker!
126 This file documents the @sc{gnu} linker ld version @value{VERSION}.
128 This document is distributed under the terms of the GNU Free
129 Documentation License. A copy of the license is included in the
130 section entitled ``GNU Free Documentation License''.
133 * Overview:: Overview
134 * Invocation:: Invocation
135 * Scripts:: Linker Scripts
137 * Machine Dependent:: Machine Dependent Features
141 * H8/300:: ld and the H8/300
144 * Renesas:: ld and other Renesas micros
147 * i960:: ld and the Intel 960 family
150 * ARM:: ld and the ARM family
153 * HPPA ELF32:: ld and HPPA 32-bit ELF
156 * TI COFF:: ld and the TI COFF
159 * Win32:: ld and WIN32 (cygwin/mingw)
162 * Xtensa:: ld and Xtensa Processors
165 @ifclear SingleFormat
168 @c Following blank line required for remaining bug in makeinfo conds/menus
170 * Reporting Bugs:: Reporting Bugs
171 * MRI:: MRI Compatible Script Files
172 * GNU Free Documentation License:: GNU Free Documentation License
180 @cindex @sc{gnu} linker
181 @cindex what is this?
184 @c man begin SYNOPSIS
185 ld [@b{options}] @var{objfile} @dots{}
189 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
190 the Info entries for @file{binutils} and
195 @c man begin DESCRIPTION
197 @command{ld} combines a number of object and archive files, relocates
198 their data and ties up symbol references. Usually the last step in
199 compiling a program is to run @command{ld}.
201 @command{ld} accepts Linker Command Language files written in
202 a superset of AT&T's Link Editor Command Language syntax,
203 to provide explicit and total control over the linking process.
207 This man page does not describe the command language; see the
208 @command{ld} entry in @code{info}, or the manual
209 ld: the GNU linker, for full details on the command language and
210 on other aspects of the GNU linker.
213 @ifclear SingleFormat
214 This version of @command{ld} uses the general purpose BFD libraries
215 to operate on object files. This allows @command{ld} to read, combine, and
216 write object files in many different formats---for example, COFF or
217 @code{a.out}. Different formats may be linked together to produce any
218 available kind of object file. @xref{BFD}, for more information.
221 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
222 linkers in providing diagnostic information. Many linkers abandon
223 execution immediately upon encountering an error; whenever possible,
224 @command{ld} continues executing, allowing you to identify other errors
225 (or, in some cases, to get an output file in spite of the error).
232 @c man begin DESCRIPTION
234 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
235 and to be as compatible as possible with other linkers. As a result,
236 you have many choices to control its behavior.
242 * Options:: Command Line Options
243 * Environment:: Environment Variables
247 @section Command Line Options
255 The linker supports a plethora of command-line options, but in actual
256 practice few of them are used in any particular context.
257 @cindex standard Unix system
258 For instance, a frequent use of @command{ld} is to link standard Unix
259 object files on a standard, supported Unix system. On such a system, to
260 link a file @code{hello.o}:
263 ld -o @var{output} /lib/crt0.o hello.o -lc
266 This tells @command{ld} to produce a file called @var{output} as the
267 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
268 the library @code{libc.a}, which will come from the standard search
269 directories. (See the discussion of the @samp{-l} option below.)
271 Some of the command-line options to @command{ld} may be specified at any
272 point in the command line. However, options which refer to files, such
273 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
274 which the option appears in the command line, relative to the object
275 files and other file options. Repeating non-file options with a
276 different argument will either have no further effect, or override prior
277 occurrences (those further to the left on the command line) of that
278 option. Options which may be meaningfully specified more than once are
279 noted in the descriptions below.
282 Non-option arguments are object files or archives which are to be linked
283 together. They may follow, precede, or be mixed in with command-line
284 options, except that an object file argument may not be placed between
285 an option and its argument.
287 Usually the linker is invoked with at least one object file, but you can
288 specify other forms of binary input files using @samp{-l}, @samp{-R},
289 and the script command language. If @emph{no} binary input files at all
290 are specified, the linker does not produce any output, and issues the
291 message @samp{No input files}.
293 If the linker cannot recognize the format of an object file, it will
294 assume that it is a linker script. A script specified in this way
295 augments the main linker script used for the link (either the default
296 linker script or the one specified by using @samp{-T}). This feature
297 permits the linker to link against a file which appears to be an object
298 or an archive, but actually merely defines some symbol values, or uses
299 @code{INPUT} or @code{GROUP} to load other objects. Note that
300 specifying a script in this way merely augments the main linker script;
301 use the @samp{-T} option to replace the default linker script entirely.
304 For options whose names are a single letter,
305 option arguments must either follow the option letter without intervening
306 whitespace, or be given as separate arguments immediately following the
307 option that requires them.
309 For options whose names are multiple letters, either one dash or two can
310 precede the option name; for example, @samp{-trace-symbol} and
311 @samp{--trace-symbol} are equivalent. Note---there is one exception to
312 this rule. Multiple letter options that start with a lower case 'o' can
313 only be preceeded by two dashes. This is to reduce confusion with the
314 @samp{-o} option. So for example @samp{-omagic} sets the output file
315 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
318 Arguments to multiple-letter options must either be separated from the
319 option name by an equals sign, or be given as separate arguments
320 immediately following the option that requires them. For example,
321 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
322 Unique abbreviations of the names of multiple-letter options are
325 Note---if the linker is being invoked indirectly, via a compiler driver
326 (e.g. @samp{gcc}) then all the linker command line options should be
327 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
328 compiler driver) like this:
331 gcc -Wl,--startgroup foo.o bar.o -Wl,--endgroup
334 This is important, because otherwise the compiler driver program may
335 silently drop the linker options, resulting in a bad link.
337 Here is a table of the generic command line switches accepted by the GNU
341 @kindex -a@var{keyword}
342 @item -a@var{keyword}
343 This option is supported for HP/UX compatibility. The @var{keyword}
344 argument must be one of the strings @samp{archive}, @samp{shared}, or
345 @samp{default}. @samp{-aarchive} is functionally equivalent to
346 @samp{-Bstatic}, and the other two keywords are functionally equivalent
347 to @samp{-Bdynamic}. This option may be used any number of times.
350 @cindex architectures
352 @item -A@var{architecture}
353 @kindex --architecture=@var{arch}
354 @itemx --architecture=@var{architecture}
355 In the current release of @command{ld}, this option is useful only for the
356 Intel 960 family of architectures. In that @command{ld} configuration, the
357 @var{architecture} argument identifies the particular architecture in
358 the 960 family, enabling some safeguards and modifying the
359 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
360 family}, for details.
362 Future releases of @command{ld} may support similar functionality for
363 other architecture families.
366 @ifclear SingleFormat
367 @cindex binary input format
368 @kindex -b @var{format}
369 @kindex --format=@var{format}
372 @item -b @var{input-format}
373 @itemx --format=@var{input-format}
374 @command{ld} may be configured to support more than one kind of object
375 file. If your @command{ld} is configured this way, you can use the
376 @samp{-b} option to specify the binary format for input object files
377 that follow this option on the command line. Even when @command{ld} is
378 configured to support alternative object formats, you don't usually need
379 to specify this, as @command{ld} should be configured to expect as a
380 default input format the most usual format on each machine.
381 @var{input-format} is a text string, the name of a particular format
382 supported by the BFD libraries. (You can list the available binary
383 formats with @samp{objdump -i}.)
386 You may want to use this option if you are linking files with an unusual
387 binary format. You can also use @samp{-b} to switch formats explicitly (when
388 linking object files of different formats), by including
389 @samp{-b @var{input-format}} before each group of object files in a
392 The default format is taken from the environment variable
397 You can also define the input format from a script, using the command
400 see @ref{Format Commands}.
404 @kindex -c @var{MRI-cmdfile}
405 @kindex --mri-script=@var{MRI-cmdfile}
406 @cindex compatibility, MRI
407 @item -c @var{MRI-commandfile}
408 @itemx --mri-script=@var{MRI-commandfile}
409 For compatibility with linkers produced by MRI, @command{ld} accepts script
410 files written in an alternate, restricted command language, described in
412 @ref{MRI,,MRI Compatible Script Files}.
415 the MRI Compatible Script Files section of GNU ld documentation.
417 Introduce MRI script files with
418 the option @samp{-c}; use the @samp{-T} option to run linker
419 scripts written in the general-purpose @command{ld} scripting language.
420 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
421 specified by any @samp{-L} options.
423 @cindex common allocation
430 These three options are equivalent; multiple forms are supported for
431 compatibility with other linkers. They assign space to common symbols
432 even if a relocatable output file is specified (with @samp{-r}). The
433 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
434 @xref{Miscellaneous Commands}.
436 @cindex entry point, from command line
437 @kindex -e @var{entry}
438 @kindex --entry=@var{entry}
440 @itemx --entry=@var{entry}
441 Use @var{entry} as the explicit symbol for beginning execution of your
442 program, rather than the default entry point. If there is no symbol
443 named @var{entry}, the linker will try to parse @var{entry} as a number,
444 and use that as the entry address (the number will be interpreted in
445 base 10; you may use a leading @samp{0x} for base 16, or a leading
446 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
447 and other ways of specifying the entry point.
449 @cindex dynamic symbol table
451 @kindex --export-dynamic
453 @itemx --export-dynamic
454 When creating a dynamically linked executable, add all symbols to the
455 dynamic symbol table. The dynamic symbol table is the set of symbols
456 which are visible from dynamic objects at run time.
458 If you do not use this option, the dynamic symbol table will normally
459 contain only those symbols which are referenced by some dynamic object
460 mentioned in the link.
462 If you use @code{dlopen} to load a dynamic object which needs to refer
463 back to the symbols defined by the program, rather than some other
464 dynamic object, then you will probably need to use this option when
465 linking the program itself.
467 You can also use the version script to control what symbols should
468 be added to the dynamic symbol table if the output format supports it.
469 See the description of @samp{--version-script} in @ref{VERSION}.
471 @ifclear SingleFormat
472 @cindex big-endian objects
476 Link big-endian objects. This affects the default output format.
478 @cindex little-endian objects
481 Link little-endian objects. This affects the default output format.
487 @itemx --auxiliary @var{name}
488 When creating an ELF shared object, set the internal DT_AUXILIARY field
489 to the specified name. This tells the dynamic linker that the symbol
490 table of the shared object should be used as an auxiliary filter on the
491 symbol table of the shared object @var{name}.
493 If you later link a program against this filter object, then, when you
494 run the program, the dynamic linker will see the DT_AUXILIARY field. If
495 the dynamic linker resolves any symbols from the filter object, it will
496 first check whether there is a definition in the shared object
497 @var{name}. If there is one, it will be used instead of the definition
498 in the filter object. The shared object @var{name} need not exist.
499 Thus the shared object @var{name} may be used to provide an alternative
500 implementation of certain functions, perhaps for debugging or for
501 machine specific performance.
503 This option may be specified more than once. The DT_AUXILIARY entries
504 will be created in the order in which they appear on the command line.
509 @itemx --filter @var{name}
510 When creating an ELF shared object, set the internal DT_FILTER field to
511 the specified name. This tells the dynamic linker that the symbol table
512 of the shared object which is being created should be used as a filter
513 on the symbol table of the shared object @var{name}.
515 If you later link a program against this filter object, then, when you
516 run the program, the dynamic linker will see the DT_FILTER field. The
517 dynamic linker will resolve symbols according to the symbol table of the
518 filter object as usual, but it will actually link to the definitions
519 found in the shared object @var{name}. Thus the filter object can be
520 used to select a subset of the symbols provided by the object
523 Some older linkers used the @option{-F} option throughout a compilation
524 toolchain for specifying object-file format for both input and output
526 @ifclear SingleFormat
527 The @sc{gnu} linker uses other mechanisms for this purpose: the
528 @option{-b}, @option{--format}, @option{--oformat} options, the
529 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
530 environment variable.
532 The @sc{gnu} linker will ignore the @option{-F} option when not
533 creating an ELF shared object.
535 @cindex finalization function
537 @item -fini @var{name}
538 When creating an ELF executable or shared object, call NAME when the
539 executable or shared object is unloaded, by setting DT_FINI to the
540 address of the function. By default, the linker uses @code{_fini} as
541 the function to call.
545 Ignored. Provided for compatibility with other tools.
551 @itemx --gpsize=@var{value}
552 Set the maximum size of objects to be optimized using the GP register to
553 @var{size}. This is only meaningful for object file formats such as
554 MIPS ECOFF which supports putting large and small objects into different
555 sections. This is ignored for other object file formats.
557 @cindex runtime library name
559 @kindex -soname=@var{name}
561 @itemx -soname=@var{name}
562 When creating an ELF shared object, set the internal DT_SONAME field to
563 the specified name. When an executable is linked with a shared object
564 which has a DT_SONAME field, then when the executable is run the dynamic
565 linker will attempt to load the shared object specified by the DT_SONAME
566 field rather than the using the file name given to the linker.
569 @cindex incremental link
571 Perform an incremental link (same as option @samp{-r}).
573 @cindex initialization function
575 @item -init @var{name}
576 When creating an ELF executable or shared object, call NAME when the
577 executable or shared object is loaded, by setting DT_INIT to the address
578 of the function. By default, the linker uses @code{_init} as the
581 @cindex archive files, from cmd line
582 @kindex -l@var{archive}
583 @kindex --library=@var{archive}
584 @item -l@var{archive}
585 @itemx --library=@var{archive}
586 Add archive file @var{archive} to the list of files to link. This
587 option may be used any number of times. @command{ld} will search its
588 path-list for occurrences of @code{lib@var{archive}.a} for every
589 @var{archive} specified.
591 On systems which support shared libraries, @command{ld} may also search for
592 libraries with extensions other than @code{.a}. Specifically, on ELF
593 and SunOS systems, @command{ld} will search a directory for a library with
594 an extension of @code{.so} before searching for one with an extension of
595 @code{.a}. By convention, a @code{.so} extension indicates a shared
598 The linker will search an archive only once, at the location where it is
599 specified on the command line. If the archive defines a symbol which
600 was undefined in some object which appeared before the archive on the
601 command line, the linker will include the appropriate file(s) from the
602 archive. However, an undefined symbol in an object appearing later on
603 the command line will not cause the linker to search the archive again.
605 See the @option{-(} option for a way to force the linker to search
606 archives multiple times.
608 You may list the same archive multiple times on the command line.
611 This type of archive searching is standard for Unix linkers. However,
612 if you are using @command{ld} on AIX, note that it is different from the
613 behaviour of the AIX linker.
616 @cindex search directory, from cmd line
618 @kindex --library-path=@var{dir}
619 @item -L@var{searchdir}
620 @itemx --library-path=@var{searchdir}
621 Add path @var{searchdir} to the list of paths that @command{ld} will search
622 for archive libraries and @command{ld} control scripts. You may use this
623 option any number of times. The directories are searched in the order
624 in which they are specified on the command line. Directories specified
625 on the command line are searched before the default directories. All
626 @option{-L} options apply to all @option{-l} options, regardless of the
627 order in which the options appear.
629 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
630 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
633 The default set of paths searched (without being specified with
634 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
635 some cases also on how it was configured. @xref{Environment}.
638 The paths can also be specified in a link script with the
639 @code{SEARCH_DIR} command. Directories specified this way are searched
640 at the point in which the linker script appears in the command line.
643 @kindex -m @var{emulation}
644 @item -m@var{emulation}
645 Emulate the @var{emulation} linker. You can list the available
646 emulations with the @samp{--verbose} or @samp{-V} options.
648 If the @samp{-m} option is not used, the emulation is taken from the
649 @code{LDEMULATION} environment variable, if that is defined.
651 Otherwise, the default emulation depends upon how the linker was
659 Print a link map to the standard output. A link map provides
660 information about the link, including the following:
664 Where object files and symbols are mapped into memory.
666 How common symbols are allocated.
668 All archive members included in the link, with a mention of the symbol
669 which caused the archive member to be brought in.
673 @cindex read-only text
678 Turn off page alignment of sections, and mark the output as
679 @code{NMAGIC} if possible.
683 @cindex read/write from cmd line
687 Set the text and data sections to be readable and writable. Also, do
688 not page-align the data segment, and disable linking against shared
689 libraries. If the output format supports Unix style magic numbers,
690 mark the output as @code{OMAGIC}. Note: Although a writable text section
691 is allowed for PE-COFF targets, it does not conform to the format
692 specification published by Microsoft.
697 This option negates most of the effects of the @option{-N} option. It
698 sets the text section to be read-only, and forces the data segment to
699 be page-aligned. Note - this option does not enable linking against
700 shared libraries. Use @option{-Bdynamic} for this.
702 @kindex -o @var{output}
703 @kindex --output=@var{output}
704 @cindex naming the output file
705 @item -o @var{output}
706 @itemx --output=@var{output}
707 Use @var{output} as the name for the program produced by @command{ld}; if this
708 option is not specified, the name @file{a.out} is used by default. The
709 script command @code{OUTPUT} can also specify the output file name.
711 @kindex -O @var{level}
712 @cindex generating optimized output
714 If @var{level} is a numeric values greater than zero @command{ld} optimizes
715 the output. This might take significantly longer and therefore probably
716 should only be enabled for the final binary.
719 @kindex --emit-relocs
720 @cindex retain relocations in final executable
723 Leave relocation sections and contents in fully linked exececutables.
724 Post link analysis and optimization tools may need this information in
725 order to perform correct modifications of executables. This results
726 in larger executables.
728 This option is currently only supported on ELF platforms.
731 @cindex relocatable output
733 @kindex --relocatable
736 Generate relocatable output---i.e., generate an output file that can in
737 turn serve as input to @command{ld}. This is often called @dfn{partial
738 linking}. As a side effect, in environments that support standard Unix
739 magic numbers, this option also sets the output file's magic number to
741 @c ; see @option{-N}.
742 If this option is not specified, an absolute file is produced. When
743 linking C++ programs, this option @emph{will not} resolve references to
744 constructors; to do that, use @samp{-Ur}.
746 When an input file does not have the same format as the output file,
747 partial linking is only supported if that input file does not contain any
748 relocations. Different output formats can have further restrictions; for
749 example some @code{a.out}-based formats do not support partial linking
750 with input files in other formats at all.
752 This option does the same thing as @samp{-i}.
754 @kindex -R @var{file}
755 @kindex --just-symbols=@var{file}
756 @cindex symbol-only input
757 @item -R @var{filename}
758 @itemx --just-symbols=@var{filename}
759 Read symbol names and their addresses from @var{filename}, but do not
760 relocate it or include it in the output. This allows your output file
761 to refer symbolically to absolute locations of memory defined in other
762 programs. You may use this option more than once.
764 For compatibility with other ELF linkers, if the @option{-R} option is
765 followed by a directory name, rather than a file name, it is treated as
766 the @option{-rpath} option.
770 @cindex strip all symbols
773 Omit all symbol information from the output file.
776 @kindex --strip-debug
777 @cindex strip debugger symbols
780 Omit debugger symbol information (but not all symbols) from the output file.
784 @cindex input files, displaying
787 Print the names of the input files as @command{ld} processes them.
789 @kindex -T @var{script}
790 @kindex --script=@var{script}
792 @item -T @var{scriptfile}
793 @itemx --script=@var{scriptfile}
794 Use @var{scriptfile} as the linker script. This script replaces
795 @command{ld}'s default linker script (rather than adding to it), so
796 @var{commandfile} must specify everything necessary to describe the
797 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
798 the current directory, @code{ld} looks for it in the directories
799 specified by any preceding @samp{-L} options. Multiple @samp{-T}
802 @kindex -u @var{symbol}
803 @kindex --undefined=@var{symbol}
804 @cindex undefined symbol
805 @item -u @var{symbol}
806 @itemx --undefined=@var{symbol}
807 Force @var{symbol} to be entered in the output file as an undefined
808 symbol. Doing this may, for example, trigger linking of additional
809 modules from standard libraries. @samp{-u} may be repeated with
810 different option arguments to enter additional undefined symbols. This
811 option is equivalent to the @code{EXTERN} linker script command.
816 For anything other than C++ programs, this option is equivalent to
817 @samp{-r}: it generates relocatable output---i.e., an output file that can in
818 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
819 @emph{does} resolve references to constructors, unlike @samp{-r}.
820 It does not work to use @samp{-Ur} on files that were themselves linked
821 with @samp{-Ur}; once the constructor table has been built, it cannot
822 be added to. Use @samp{-Ur} only for the last partial link, and
823 @samp{-r} for the others.
825 @kindex --unique[=@var{SECTION}]
826 @item --unique[=@var{SECTION}]
827 Creates a separate output section for every input section matching
828 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
829 missing, for every orphan input section. An orphan section is one not
830 specifically mentioned in a linker script. You may use this option
831 multiple times on the command line; It prevents the normal merging of
832 input sections with the same name, overriding output section assignments
842 Display the version number for @command{ld}. The @option{-V} option also
843 lists the supported emulations.
846 @kindex --discard-all
847 @cindex deleting local symbols
850 Delete all local symbols.
853 @kindex --discard-locals
854 @cindex local symbols, deleting
855 @cindex L, deleting symbols beginning
857 @itemx --discard-locals
858 Delete all temporary local symbols. For most targets, this is all local
859 symbols whose names begin with @samp{L}.
861 @kindex -y @var{symbol}
862 @kindex --trace-symbol=@var{symbol}
863 @cindex symbol tracing
864 @item -y @var{symbol}
865 @itemx --trace-symbol=@var{symbol}
866 Print the name of each linked file in which @var{symbol} appears. This
867 option may be given any number of times. On many systems it is necessary
868 to prepend an underscore.
870 This option is useful when you have an undefined symbol in your link but
871 don't know where the reference is coming from.
873 @kindex -Y @var{path}
875 Add @var{path} to the default library search path. This option exists
876 for Solaris compatibility.
878 @kindex -z @var{keyword}
879 @item -z @var{keyword}
880 The recognized keywords are:
884 Combines multiple reloc sections and sorts them to make dynamic symbol
885 lookup caching possible.
888 Disallows undefined symbols in object files. Undefined symbols in
889 shared libraries are still allowed.
892 This option is only meaningful when building a shared object.
893 It marks the object so that its runtime initialization will occur
894 before the runtime initialization of any other objects brought into
895 the process at the same time. Similarly the runtime finalization of
896 the object will occur after the runtime finalization of any other
900 Marks the object that its symbol table interposes before all symbols
901 but the primary executable.
904 Marks the object that its filters be processed immediately at
908 Allows multiple definitions.
911 Disables multiple reloc sections combining.
914 Disables production of copy relocs.
917 Marks the object that the search for dependencies of this object will
918 ignore any default library search paths.
921 Marks the object shouldn't be unloaded at runtime.
924 Marks the object not available to @code{dlopen}.
927 Marks the object can not be dumped by @code{dldump}.
930 When generating an executable or shared library, mark it to tell the
931 dynamic linker to resolve all symbols when the program is started, or
932 when the shared library is linked to using dlopen, instead of
933 deferring function call resolution to the point when the function is
937 Marks the object may contain $ORIGIN.
941 Other keywords are ignored for Solaris compatibility.
944 @cindex groups of archives
945 @item -( @var{archives} -)
946 @itemx --start-group @var{archives} --end-group
947 The @var{archives} should be a list of archive files. They may be
948 either explicit file names, or @samp{-l} options.
950 The specified archives are searched repeatedly until no new undefined
951 references are created. Normally, an archive is searched only once in
952 the order that it is specified on the command line. If a symbol in that
953 archive is needed to resolve an undefined symbol referred to by an
954 object in an archive that appears later on the command line, the linker
955 would not be able to resolve that reference. By grouping the archives,
956 they all be searched repeatedly until all possible references are
959 Using this option has a significant performance cost. It is best to use
960 it only when there are unavoidable circular references between two or
963 @kindex --accept-unknown-input-arch
964 @kindex --no-accept-unknown-input-arch
965 @item --accept-unknown-input-arch
966 @itemx --no-accept-unknown-input-arch
967 Tells the linker to accept input files whose architecture cannot be
968 recognised. The assumption is that the user knows what they are doing
969 and deliberately wants to link in these unknown input files. This was
970 the default behaviour of the linker, before release 2.14. The default
971 behaviour from release 2.14 onwards is to reject such input files, and
972 so the @samp{--accept-unknown-input-arch} option has been added to
973 restore the old behaviour.
976 @kindex --no-as-needed
978 @itemx --no-as-needed
979 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
980 on the command line after the @option{--as-needed} option. Normally,
981 the linker will add a DT_NEEDED tag for each dynamic library mentioned
982 on the command line, regardless of whether the library is actually
983 needed. @option{--as-needed} causes DT_NEEDED tags to only be emitted
984 for libraries that satisfy some reference from regular objects.
985 @option{--no-as-needed} restores the default behaviour.
987 @kindex -assert @var{keyword}
988 @item -assert @var{keyword}
989 This option is ignored for SunOS compatibility.
997 Link against dynamic libraries. This is only meaningful on platforms
998 for which shared libraries are supported. This option is normally the
999 default on such platforms. The different variants of this option are
1000 for compatibility with various systems. You may use this option
1001 multiple times on the command line: it affects library searching for
1002 @option{-l} options which follow it.
1006 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1007 section. This causes the runtime linker to handle lookups in this
1008 object and its dependencies to be performed only inside the group.
1009 @option{--unresolved-symbols=report-all} is implied. This option is
1010 only meaningful on ELF platforms which support shared libraries.
1020 Do not link against shared libraries. This is only meaningful on
1021 platforms for which shared libraries are supported. The different
1022 variants of this option are for compatibility with various systems. You
1023 may use this option multiple times on the command line: it affects
1024 library searching for @option{-l} options which follow it. This
1025 option also implies @option{--unresolved-symbols=report-all}.
1029 When creating a shared library, bind references to global symbols to the
1030 definition within the shared library, if any. Normally, it is possible
1031 for a program linked against a shared library to override the definition
1032 within the shared library. This option is only meaningful on ELF
1033 platforms which support shared libraries.
1035 @kindex --check-sections
1036 @kindex --no-check-sections
1037 @item --check-sections
1038 @itemx --no-check-sections
1039 Asks the linker @emph{not} to check section addresses after they have
1040 been assigned to see if there any overlaps. Normally the linker will
1041 perform this check, and if it finds any overlaps it will produce
1042 suitable error messages. The linker does know about, and does make
1043 allowances for sections in overlays. The default behaviour can be
1044 restored by using the command line switch @option{--check-sections}.
1046 @cindex cross reference table
1049 Output a cross reference table. If a linker map file is being
1050 generated, the cross reference table is printed to the map file.
1051 Otherwise, it is printed on the standard output.
1053 The format of the table is intentionally simple, so that it may be
1054 easily processed by a script if necessary. The symbols are printed out,
1055 sorted by name. For each symbol, a list of file names is given. If the
1056 symbol is defined, the first file listed is the location of the
1057 definition. The remaining files contain references to the symbol.
1059 @cindex common allocation
1060 @kindex --no-define-common
1061 @item --no-define-common
1062 This option inhibits the assignment of addresses to common symbols.
1063 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1064 @xref{Miscellaneous Commands}.
1066 The @samp{--no-define-common} option allows decoupling
1067 the decision to assign addresses to Common symbols from the choice
1068 of the output file type; otherwise a non-Relocatable output type
1069 forces assigning addresses to Common symbols.
1070 Using @samp{--no-define-common} allows Common symbols that are referenced
1071 from a shared library to be assigned addresses only in the main program.
1072 This eliminates the unused duplicate space in the shared library,
1073 and also prevents any possible confusion over resolving to the wrong
1074 duplicate when there are many dynamic modules with specialized search
1075 paths for runtime symbol resolution.
1077 @cindex symbols, from command line
1078 @kindex --defsym @var{symbol}=@var{exp}
1079 @item --defsym @var{symbol}=@var{expression}
1080 Create a global symbol in the output file, containing the absolute
1081 address given by @var{expression}. You may use this option as many
1082 times as necessary to define multiple symbols in the command line. A
1083 limited form of arithmetic is supported for the @var{expression} in this
1084 context: you may give a hexadecimal constant or the name of an existing
1085 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1086 constants or symbols. If you need more elaborate expressions, consider
1087 using the linker command language from a script (@pxref{Assignments,,
1088 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1089 space between @var{symbol}, the equals sign (``@key{=}''), and
1092 @cindex demangling, from command line
1093 @kindex --demangle[=@var{style}]
1094 @kindex --no-demangle
1095 @item --demangle[=@var{style}]
1096 @itemx --no-demangle
1097 These options control whether to demangle symbol names in error messages
1098 and other output. When the linker is told to demangle, it tries to
1099 present symbol names in a readable fashion: it strips leading
1100 underscores if they are used by the object file format, and converts C++
1101 mangled symbol names into user readable names. Different compilers have
1102 different mangling styles. The optional demangling style argument can be used
1103 to choose an appropriate demangling style for your compiler. The linker will
1104 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1105 is set. These options may be used to override the default.
1107 @cindex dynamic linker, from command line
1108 @kindex -I@var{file}
1109 @kindex --dynamic-linker @var{file}
1110 @item --dynamic-linker @var{file}
1111 Set the name of the dynamic linker. This is only meaningful when
1112 generating dynamically linked ELF executables. The default dynamic
1113 linker is normally correct; don't use this unless you know what you are
1116 @cindex MIPS embedded PIC code
1117 @kindex --embedded-relocs
1118 @item --embedded-relocs
1119 This option is only meaningful when linking MIPS embedded PIC code,
1120 generated by the -membedded-pic option to the @sc{gnu} compiler and
1121 assembler. It causes the linker to create a table which may be used at
1122 runtime to relocate any data which was statically initialized to pointer
1123 values. See the code in testsuite/ld-empic for details.
1126 @kindex --fatal-warnings
1127 @item --fatal-warnings
1128 Treat all warnings as errors.
1130 @kindex --force-exe-suffix
1131 @item --force-exe-suffix
1132 Make sure that an output file has a .exe suffix.
1134 If a successfully built fully linked output file does not have a
1135 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1136 the output file to one of the same name with a @code{.exe} suffix. This
1137 option is useful when using unmodified Unix makefiles on a Microsoft
1138 Windows host, since some versions of Windows won't run an image unless
1139 it ends in a @code{.exe} suffix.
1141 @kindex --gc-sections
1142 @kindex --no-gc-sections
1143 @cindex garbage collection
1144 @item --no-gc-sections
1145 @itemx --gc-sections
1146 Enable garbage collection of unused input sections. It is ignored on
1147 targets that do not support this option. This option is not compatible
1148 with @samp{-r}, nor should it be used with dynamic linking. The default
1149 behaviour (of not performing this garbage collection) can be restored by
1150 specifying @samp{--no-gc-sections} on the command line.
1156 Print a summary of the command-line options on the standard output and exit.
1158 @kindex --target-help
1160 Print a summary of all target specific options on the standard output and exit.
1163 @item -Map @var{mapfile}
1164 Print a link map to the file @var{mapfile}. See the description of the
1165 @option{-M} option, above.
1167 @cindex memory usage
1168 @kindex --no-keep-memory
1169 @item --no-keep-memory
1170 @command{ld} normally optimizes for speed over memory usage by caching the
1171 symbol tables of input files in memory. This option tells @command{ld} to
1172 instead optimize for memory usage, by rereading the symbol tables as
1173 necessary. This may be required if @command{ld} runs out of memory space
1174 while linking a large executable.
1176 @kindex --no-undefined
1178 @item --no-undefined
1180 Report unresolved symbol references from regular object files. This
1181 is done even if the linker is creating a non-symbolic shared library.
1182 The switch @option{--[no-]allow-shlib-undefined} controls the
1183 behaviour for reporting unresolved references found in shared
1184 libraries being linked in.
1186 @kindex --allow-multiple-definition
1188 @item --allow-multiple-definition
1190 Normally when a symbol is defined multiple times, the linker will
1191 report a fatal error. These options allow multiple definitions and the
1192 first definition will be used.
1194 @kindex --allow-shlib-undefined
1195 @kindex --no-allow-shlib-undefined
1196 @item --allow-shlib-undefined
1197 @itemx --no-allow-shlib-undefined
1198 Allows (the default) or disallows undefined symbols in shared libraries.
1199 This switch is similar to @option{--no-undefined} except that it
1200 determines the behaviour when the undefined symbols are in a
1201 shared library rather than a regular object file. It does not affect
1202 how undefined symbols in regular object files are handled.
1204 The reason that @option{--allow-shlib-undefined} is the default is that
1205 the shared library being specified at link time may not be the same as
1206 the one that is available at load time, so the symbols might actually be
1207 resolvable at load time. Plus there are some systems, (eg BeOS) where
1208 undefined symbols in shared libraries is normal. (The kernel patches
1209 them at load time to select which function is most appropriate
1210 for the current architecture. This is used for example to dynamically
1211 select an appropriate memset function). Apparently it is also normal
1212 for HPPA shared libraries to have undefined symbols.
1214 @kindex --no-undefined-version
1215 @item --no-undefined-version
1216 Normally when a symbol has an undefined version, the linker will ignore
1217 it. This option disallows symbols with undefined version and a fatal error
1218 will be issued instead.
1220 @kindex --no-warn-mismatch
1221 @item --no-warn-mismatch
1222 Normally @command{ld} will give an error if you try to link together input
1223 files that are mismatched for some reason, perhaps because they have
1224 been compiled for different processors or for different endiannesses.
1225 This option tells @command{ld} that it should silently permit such possible
1226 errors. This option should only be used with care, in cases when you
1227 have taken some special action that ensures that the linker errors are
1230 @kindex --no-whole-archive
1231 @item --no-whole-archive
1232 Turn off the effect of the @option{--whole-archive} option for subsequent
1235 @cindex output file after errors
1236 @kindex --noinhibit-exec
1237 @item --noinhibit-exec
1238 Retain the executable output file whenever it is still usable.
1239 Normally, the linker will not produce an output file if it encounters
1240 errors during the link process; it exits without writing an output file
1241 when it issues any error whatsoever.
1245 Only search library directories explicitly specified on the
1246 command line. Library directories specified in linker scripts
1247 (including linker scripts specified on the command line) are ignored.
1249 @ifclear SingleFormat
1251 @item --oformat @var{output-format}
1252 @command{ld} may be configured to support more than one kind of object
1253 file. If your @command{ld} is configured this way, you can use the
1254 @samp{--oformat} option to specify the binary format for the output
1255 object file. Even when @command{ld} is configured to support alternative
1256 object formats, you don't usually need to specify this, as @command{ld}
1257 should be configured to produce as a default output format the most
1258 usual format on each machine. @var{output-format} is a text string, the
1259 name of a particular format supported by the BFD libraries. (You can
1260 list the available binary formats with @samp{objdump -i}.) The script
1261 command @code{OUTPUT_FORMAT} can also specify the output format, but
1262 this option overrides it. @xref{BFD}.
1266 @kindex --pic-executable
1268 @itemx --pic-executable
1269 @cindex position independent executables
1270 Create a position independent executable. This is currently only supported on
1271 ELF platforms. Position independent executables are similar to shared
1272 libraries in that they are relocated by the dynamic linker to the virtual
1273 address the OS chooses for them (which can vary between invocations). Like
1274 normal dynamically linked executables they can be executed and symbols
1275 defined in the executable cannot be overridden by shared libraries.
1279 This option is ignored for Linux compatibility.
1283 This option is ignored for SVR4 compatibility.
1286 @cindex synthesizing linker
1287 @cindex relaxing addressing modes
1289 An option with machine dependent effects.
1291 This option is only supported on a few targets.
1294 @xref{H8/300,,@command{ld} and the H8/300}.
1297 @xref{i960,, @command{ld} and the Intel 960 family}.
1300 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1303 On some platforms, the @samp{--relax} option performs global
1304 optimizations that become possible when the linker resolves addressing
1305 in the program, such as relaxing address modes and synthesizing new
1306 instructions in the output object file.
1308 On some platforms these link time global optimizations may make symbolic
1309 debugging of the resulting executable impossible.
1312 the case for the Matsushita MN10200 and MN10300 family of processors.
1316 On platforms where this is not supported, @samp{--relax} is accepted,
1320 @cindex retaining specified symbols
1321 @cindex stripping all but some symbols
1322 @cindex symbols, retaining selectively
1323 @item --retain-symbols-file @var{filename}
1324 Retain @emph{only} the symbols listed in the file @var{filename},
1325 discarding all others. @var{filename} is simply a flat file, with one
1326 symbol name per line. This option is especially useful in environments
1330 where a large global symbol table is accumulated gradually, to conserve
1333 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1334 or symbols needed for relocations.
1336 You may only specify @samp{--retain-symbols-file} once in the command
1337 line. It overrides @samp{-s} and @samp{-S}.
1340 @item -rpath @var{dir}
1341 @cindex runtime library search path
1343 Add a directory to the runtime library search path. This is used when
1344 linking an ELF executable with shared objects. All @option{-rpath}
1345 arguments are concatenated and passed to the runtime linker, which uses
1346 them to locate shared objects at runtime. The @option{-rpath} option is
1347 also used when locating shared objects which are needed by shared
1348 objects explicitly included in the link; see the description of the
1349 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1350 ELF executable, the contents of the environment variable
1351 @code{LD_RUN_PATH} will be used if it is defined.
1353 The @option{-rpath} option may also be used on SunOS. By default, on
1354 SunOS, the linker will form a runtime search patch out of all the
1355 @option{-L} options it is given. If a @option{-rpath} option is used, the
1356 runtime search path will be formed exclusively using the @option{-rpath}
1357 options, ignoring the @option{-L} options. This can be useful when using
1358 gcc, which adds many @option{-L} options which may be on NFS mounted
1361 For compatibility with other ELF linkers, if the @option{-R} option is
1362 followed by a directory name, rather than a file name, it is treated as
1363 the @option{-rpath} option.
1367 @cindex link-time runtime library search path
1369 @item -rpath-link @var{DIR}
1370 When using ELF or SunOS, one shared library may require another. This
1371 happens when an @code{ld -shared} link includes a shared library as one
1374 When the linker encounters such a dependency when doing a non-shared,
1375 non-relocatable link, it will automatically try to locate the required
1376 shared library and include it in the link, if it is not included
1377 explicitly. In such a case, the @option{-rpath-link} option
1378 specifies the first set of directories to search. The
1379 @option{-rpath-link} option may specify a sequence of directory names
1380 either by specifying a list of names separated by colons, or by
1381 appearing multiple times.
1383 This option should be used with caution as it overrides the search path
1384 that may have been hard compiled into a shared library. In such a case it
1385 is possible to use unintentionally a different search path than the
1386 runtime linker would do.
1388 The linker uses the following search paths to locate required shared
1392 Any directories specified by @option{-rpath-link} options.
1394 Any directories specified by @option{-rpath} options. The difference
1395 between @option{-rpath} and @option{-rpath-link} is that directories
1396 specified by @option{-rpath} options are included in the executable and
1397 used at runtime, whereas the @option{-rpath-link} option is only effective
1398 at link time. It is for the native linker only.
1400 On an ELF system, if the @option{-rpath} and @code{rpath-link} options
1401 were not used, search the contents of the environment variable
1402 @code{LD_RUN_PATH}. It is for the native linker only.
1404 On SunOS, if the @option{-rpath} option was not used, search any
1405 directories specified using @option{-L} options.
1407 For a native linker, the contents of the environment variable
1408 @code{LD_LIBRARY_PATH}.
1410 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1411 @code{DT_RPATH} of a shared library are searched for shared
1412 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1413 @code{DT_RUNPATH} entries exist.
1415 The default directories, normally @file{/lib} and @file{/usr/lib}.
1417 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1418 exists, the list of directories found in that file.
1421 If the required shared library is not found, the linker will issue a
1422 warning and continue with the link.
1429 @cindex shared libraries
1430 Create a shared library. This is currently only supported on ELF, XCOFF
1431 and SunOS platforms. On SunOS, the linker will automatically create a
1432 shared library if the @option{-e} option is not used and there are
1433 undefined symbols in the link.
1436 @kindex --sort-common
1437 This option tells @command{ld} to sort the common symbols by size when it
1438 places them in the appropriate output sections. First come all the one
1439 byte symbols, then all the two byte, then all the four byte, and then
1440 everything else. This is to prevent gaps between symbols due to
1441 alignment constraints.
1443 @kindex --split-by-file
1444 @item --split-by-file [@var{size}]
1445 Similar to @option{--split-by-reloc} but creates a new output section for
1446 each input file when @var{size} is reached. @var{size} defaults to a
1447 size of 1 if not given.
1449 @kindex --split-by-reloc
1450 @item --split-by-reloc [@var{count}]
1451 Tries to creates extra sections in the output file so that no single
1452 output section in the file contains more than @var{count} relocations.
1453 This is useful when generating huge relocatable files for downloading into
1454 certain real time kernels with the COFF object file format; since COFF
1455 cannot represent more than 65535 relocations in a single section. Note
1456 that this will fail to work with object file formats which do not
1457 support arbitrary sections. The linker will not split up individual
1458 input sections for redistribution, so if a single input section contains
1459 more than @var{count} relocations one output section will contain that
1460 many relocations. @var{count} defaults to a value of 32768.
1464 Compute and display statistics about the operation of the linker, such
1465 as execution time and memory usage.
1467 @kindex --traditional-format
1468 @cindex traditional format
1469 @item --traditional-format
1470 For some targets, the output of @command{ld} is different in some ways from
1471 the output of some existing linker. This switch requests @command{ld} to
1472 use the traditional format instead.
1475 For example, on SunOS, @command{ld} combines duplicate entries in the
1476 symbol string table. This can reduce the size of an output file with
1477 full debugging information by over 30 percent. Unfortunately, the SunOS
1478 @code{dbx} program can not read the resulting program (@code{gdb} has no
1479 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1480 combine duplicate entries.
1482 @kindex --section-start @var{sectionname}=@var{org}
1483 @item --section-start @var{sectionname}=@var{org}
1484 Locate a section in the output file at the absolute
1485 address given by @var{org}. You may use this option as many
1486 times as necessary to locate multiple sections in the command
1488 @var{org} must be a single hexadecimal integer;
1489 for compatibility with other linkers, you may omit the leading
1490 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1491 should be no white space between @var{sectionname}, the equals
1492 sign (``@key{=}''), and @var{org}.
1494 @kindex -Tbss @var{org}
1495 @kindex -Tdata @var{org}
1496 @kindex -Ttext @var{org}
1497 @cindex segment origins, cmd line
1498 @item -Tbss @var{org}
1499 @itemx -Tdata @var{org}
1500 @itemx -Ttext @var{org}
1501 Same as --section-start, with @code{.bss}, @code{.data} or
1502 @code{.text} as the @var{sectionname}.
1504 @kindex --unresolved-symbols
1505 @item --unresolved-symbols=@var{method}
1506 Determine how to handle unresolved symbols. There are four possible
1507 values for @samp{method}:
1511 Do not report any unresolved symbols.
1514 Report all unresolved symbols. This is the default.
1516 @item ignore-in-object-files
1517 Report unresolved symbols that are contained in shared libraries, but
1518 ignore them if they come from regular object files.
1520 @item ignore-in-shared-libs
1521 Report unresolved symbols that come from regular object files, but
1522 ignore them if they come from shared libraries. This can be useful
1523 when creating a dynamic binary and it is known that all the shared
1524 libraries that it should be referencing are included on the linker's
1528 The behaviour for shared libraries on their own can also be controlled
1529 by the @option{--[no-]allow-shlib-undefined} option.
1531 Normally the linker will generate an error message for each reported
1532 unresolved symbol but the option @option{--warn-unresolved-symbols}
1533 can change this to a warning.
1539 Display the version number for @command{ld} and list the linker emulations
1540 supported. Display which input files can and cannot be opened. Display
1541 the linker script being used by the linker.
1543 @kindex --version-script=@var{version-scriptfile}
1544 @cindex version script, symbol versions
1545 @itemx --version-script=@var{version-scriptfile}
1546 Specify the name of a version script to the linker. This is typically
1547 used when creating shared libraries to specify additional information
1548 about the version hierarchy for the library being created. This option
1549 is only meaningful on ELF platforms which support shared libraries.
1552 @kindex --warn-common
1553 @cindex warnings, on combining symbols
1554 @cindex combining symbols, warnings on
1556 Warn when a common symbol is combined with another common symbol or with
1557 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1558 but linkers on some other operating systems do not. This option allows
1559 you to find potential problems from combining global symbols.
1560 Unfortunately, some C libraries use this practise, so you may get some
1561 warnings about symbols in the libraries as well as in your programs.
1563 There are three kinds of global symbols, illustrated here by C examples:
1567 A definition, which goes in the initialized data section of the output
1571 An undefined reference, which does not allocate space.
1572 There must be either a definition or a common symbol for the
1576 A common symbol. If there are only (one or more) common symbols for a
1577 variable, it goes in the uninitialized data area of the output file.
1578 The linker merges multiple common symbols for the same variable into a
1579 single symbol. If they are of different sizes, it picks the largest
1580 size. The linker turns a common symbol into a declaration, if there is
1581 a definition of the same variable.
1584 The @samp{--warn-common} option can produce five kinds of warnings.
1585 Each warning consists of a pair of lines: the first describes the symbol
1586 just encountered, and the second describes the previous symbol
1587 encountered with the same name. One or both of the two symbols will be
1592 Turning a common symbol into a reference, because there is already a
1593 definition for the symbol.
1595 @var{file}(@var{section}): warning: common of `@var{symbol}'
1596 overridden by definition
1597 @var{file}(@var{section}): warning: defined here
1601 Turning a common symbol into a reference, because a later definition for
1602 the symbol is encountered. This is the same as the previous case,
1603 except that the symbols are encountered in a different order.
1605 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1607 @var{file}(@var{section}): warning: common is here
1611 Merging a common symbol with a previous same-sized common symbol.
1613 @var{file}(@var{section}): warning: multiple common
1615 @var{file}(@var{section}): warning: previous common is here
1619 Merging a common symbol with a previous larger common symbol.
1621 @var{file}(@var{section}): warning: common of `@var{symbol}'
1622 overridden by larger common
1623 @var{file}(@var{section}): warning: larger common is here
1627 Merging a common symbol with a previous smaller common symbol. This is
1628 the same as the previous case, except that the symbols are
1629 encountered in a different order.
1631 @var{file}(@var{section}): warning: common of `@var{symbol}'
1632 overriding smaller common
1633 @var{file}(@var{section}): warning: smaller common is here
1637 @kindex --warn-constructors
1638 @item --warn-constructors
1639 Warn if any global constructors are used. This is only useful for a few
1640 object file formats. For formats like COFF or ELF, the linker can not
1641 detect the use of global constructors.
1643 @kindex --warn-multiple-gp
1644 @item --warn-multiple-gp
1645 Warn if multiple global pointer values are required in the output file.
1646 This is only meaningful for certain processors, such as the Alpha.
1647 Specifically, some processors put large-valued constants in a special
1648 section. A special register (the global pointer) points into the middle
1649 of this section, so that constants can be loaded efficiently via a
1650 base-register relative addressing mode. Since the offset in
1651 base-register relative mode is fixed and relatively small (e.g., 16
1652 bits), this limits the maximum size of the constant pool. Thus, in
1653 large programs, it is often necessary to use multiple global pointer
1654 values in order to be able to address all possible constants. This
1655 option causes a warning to be issued whenever this case occurs.
1658 @cindex warnings, on undefined symbols
1659 @cindex undefined symbols, warnings on
1661 Only warn once for each undefined symbol, rather than once per module
1664 @kindex --warn-section-align
1665 @cindex warnings, on section alignment
1666 @cindex section alignment, warnings on
1667 @item --warn-section-align
1668 Warn if the address of an output section is changed because of
1669 alignment. Typically, the alignment will be set by an input section.
1670 The address will only be changed if it not explicitly specified; that
1671 is, if the @code{SECTIONS} command does not specify a start address for
1672 the section (@pxref{SECTIONS}).
1674 @kindex --warn-unresolved-symbols
1675 @item --warn-unresolved-symbols
1676 If the linker is going to report an unresolved symbol (see the option
1677 @option{--unresolved-symbols}) it will normally generate an error.
1678 This option makes it generate a warning instead.
1680 @kindex --error-unresolved-symbols
1681 @item --error-unresolved-symbols
1682 This restores the linker's default behaviour of generating errors when
1683 it is reporting unresolved symbols.
1685 @kindex --whole-archive
1686 @cindex including an entire archive
1687 @item --whole-archive
1688 For each archive mentioned on the command line after the
1689 @option{--whole-archive} option, include every object file in the archive
1690 in the link, rather than searching the archive for the required object
1691 files. This is normally used to turn an archive file into a shared
1692 library, forcing every object to be included in the resulting shared
1693 library. This option may be used more than once.
1695 Two notes when using this option from gcc: First, gcc doesn't know
1696 about this option, so you have to use @option{-Wl,-whole-archive}.
1697 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1698 list of archives, because gcc will add its own list of archives to
1699 your link and you may not want this flag to affect those as well.
1702 @item --wrap @var{symbol}
1703 Use a wrapper function for @var{symbol}. Any undefined reference to
1704 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1705 undefined reference to @code{__real_@var{symbol}} will be resolved to
1708 This can be used to provide a wrapper for a system function. The
1709 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1710 wishes to call the system function, it should call
1711 @code{__real_@var{symbol}}.
1713 Here is a trivial example:
1717 __wrap_malloc (size_t c)
1719 printf ("malloc called with %zu\n", c);
1720 return __real_malloc (c);
1724 If you link other code with this file using @option{--wrap malloc}, then
1725 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1726 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1727 call the real @code{malloc} function.
1729 You may wish to provide a @code{__real_malloc} function as well, so that
1730 links without the @option{--wrap} option will succeed. If you do this,
1731 you should not put the definition of @code{__real_malloc} in the same
1732 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
1733 call before the linker has a chance to wrap it to @code{malloc}.
1735 @kindex --enable-new-dtags
1736 @kindex --disable-new-dtags
1737 @item --enable-new-dtags
1738 @itemx --disable-new-dtags
1739 This linker can create the new dynamic tags in ELF. But the older ELF
1740 systems may not understand them. If you specify
1741 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
1742 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
1743 created. By default, the new dynamic tags are not created. Note that
1744 those options are only available for ELF systems.
1750 @subsection Options Specific to i386 PE Targets
1752 @c man begin OPTIONS
1754 The i386 PE linker supports the @option{-shared} option, which causes
1755 the output to be a dynamically linked library (DLL) instead of a
1756 normal executable. You should name the output @code{*.dll} when you
1757 use this option. In addition, the linker fully supports the standard
1758 @code{*.def} files, which may be specified on the linker command line
1759 like an object file (in fact, it should precede archives it exports
1760 symbols from, to ensure that they get linked in, just like a normal
1763 In addition to the options common to all targets, the i386 PE linker
1764 support additional command line options that are specific to the i386
1765 PE target. Options that take values may be separated from their
1766 values by either a space or an equals sign.
1770 @kindex --add-stdcall-alias
1771 @item --add-stdcall-alias
1772 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
1773 as-is and also with the suffix stripped.
1774 [This option is specific to the i386 PE targeted port of the linker]
1777 @item --base-file @var{file}
1778 Use @var{file} as the name of a file in which to save the base
1779 addresses of all the relocations needed for generating DLLs with
1781 [This is an i386 PE specific option]
1785 Create a DLL instead of a regular executable. You may also use
1786 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
1788 [This option is specific to the i386 PE targeted port of the linker]
1790 @kindex --enable-stdcall-fixup
1791 @kindex --disable-stdcall-fixup
1792 @item --enable-stdcall-fixup
1793 @itemx --disable-stdcall-fixup
1794 If the link finds a symbol that it cannot resolve, it will attempt to
1795 do ``fuzzy linking'' by looking for another defined symbol that differs
1796 only in the format of the symbol name (cdecl vs stdcall) and will
1797 resolve that symbol by linking to the match. For example, the
1798 undefined symbol @code{_foo} might be linked to the function
1799 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
1800 to the function @code{_bar}. When the linker does this, it prints a
1801 warning, since it normally should have failed to link, but sometimes
1802 import libraries generated from third-party dlls may need this feature
1803 to be usable. If you specify @option{--enable-stdcall-fixup}, this
1804 feature is fully enabled and warnings are not printed. If you specify
1805 @option{--disable-stdcall-fixup}, this feature is disabled and such
1806 mismatches are considered to be errors.
1807 [This option is specific to the i386 PE targeted port of the linker]
1809 @cindex DLLs, creating
1810 @kindex --export-all-symbols
1811 @item --export-all-symbols
1812 If given, all global symbols in the objects used to build a DLL will
1813 be exported by the DLL. Note that this is the default if there
1814 otherwise wouldn't be any exported symbols. When symbols are
1815 explicitly exported via DEF files or implicitly exported via function
1816 attributes, the default is to not export anything else unless this
1817 option is given. Note that the symbols @code{DllMain@@12},
1818 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
1819 @code{impure_ptr} will not be automatically
1820 exported. Also, symbols imported from other DLLs will not be
1821 re-exported, nor will symbols specifying the DLL's internal layout
1822 such as those beginning with @code{_head_} or ending with
1823 @code{_iname}. In addition, no symbols from @code{libgcc},
1824 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
1825 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
1826 not be exported, to help with C++ DLLs. Finally, there is an
1827 extensive list of cygwin-private symbols that are not exported
1828 (obviously, this applies on when building DLLs for cygwin targets).
1829 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
1830 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
1831 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
1832 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
1833 @code{cygwin_premain3}, and @code{environ}.
1834 [This option is specific to the i386 PE targeted port of the linker]
1836 @kindex --exclude-symbols
1837 @item --exclude-symbols @var{symbol},@var{symbol},...
1838 Specifies a list of symbols which should not be automatically
1839 exported. The symbol names may be delimited by commas or colons.
1840 [This option is specific to the i386 PE targeted port of the linker]
1842 @kindex --exclude-libs
1843 @item --exclude-libs @var{lib},@var{lib},...
1844 Specifies a list of archive libraries from which symbols should not be automatically
1845 exported. The library names may be delimited by commas or colons. Specifying
1846 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
1847 automatic export. Symbols explicitly listed in a .def file are still exported,
1848 regardless of this option.
1849 [This option is specific to the i386 PE targeted port of the linker]
1851 @kindex --file-alignment
1852 @item --file-alignment
1853 Specify the file alignment. Sections in the file will always begin at
1854 file offsets which are multiples of this number. This defaults to
1856 [This option is specific to the i386 PE targeted port of the linker]
1860 @item --heap @var{reserve}
1861 @itemx --heap @var{reserve},@var{commit}
1862 Specify the amount of memory to reserve (and optionally commit) to be
1863 used as heap for this program. The default is 1Mb reserved, 4K
1865 [This option is specific to the i386 PE targeted port of the linker]
1868 @kindex --image-base
1869 @item --image-base @var{value}
1870 Use @var{value} as the base address of your program or dll. This is
1871 the lowest memory location that will be used when your program or dll
1872 is loaded. To reduce the need to relocate and improve performance of
1873 your dlls, each should have a unique base address and not overlap any
1874 other dlls. The default is 0x400000 for executables, and 0x10000000
1876 [This option is specific to the i386 PE targeted port of the linker]
1880 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
1881 symbols before they are exported.
1882 [This option is specific to the i386 PE targeted port of the linker]
1884 @kindex --major-image-version
1885 @item --major-image-version @var{value}
1886 Sets the major number of the ``image version''. Defaults to 1.
1887 [This option is specific to the i386 PE targeted port of the linker]
1889 @kindex --major-os-version
1890 @item --major-os-version @var{value}
1891 Sets the major number of the ``os version''. Defaults to 4.
1892 [This option is specific to the i386 PE targeted port of the linker]
1894 @kindex --major-subsystem-version
1895 @item --major-subsystem-version @var{value}
1896 Sets the major number of the ``subsystem version''. Defaults to 4.
1897 [This option is specific to the i386 PE targeted port of the linker]
1899 @kindex --minor-image-version
1900 @item --minor-image-version @var{value}
1901 Sets the minor number of the ``image version''. Defaults to 0.
1902 [This option is specific to the i386 PE targeted port of the linker]
1904 @kindex --minor-os-version
1905 @item --minor-os-version @var{value}
1906 Sets the minor number of the ``os version''. Defaults to 0.
1907 [This option is specific to the i386 PE targeted port of the linker]
1909 @kindex --minor-subsystem-version
1910 @item --minor-subsystem-version @var{value}
1911 Sets the minor number of the ``subsystem version''. Defaults to 0.
1912 [This option is specific to the i386 PE targeted port of the linker]
1914 @cindex DEF files, creating
1915 @cindex DLLs, creating
1916 @kindex --output-def
1917 @item --output-def @var{file}
1918 The linker will create the file @var{file} which will contain a DEF
1919 file corresponding to the DLL the linker is generating. This DEF file
1920 (which should be called @code{*.def}) may be used to create an import
1921 library with @code{dlltool} or may be used as a reference to
1922 automatically or implicitly exported symbols.
1923 [This option is specific to the i386 PE targeted port of the linker]
1925 @cindex DLLs, creating
1926 @kindex --out-implib
1927 @item --out-implib @var{file}
1928 The linker will create the file @var{file} which will contain an
1929 import lib corresponding to the DLL the linker is generating. This
1930 import lib (which should be called @code{*.dll.a} or @code{*.a}
1931 may be used to link clients against the generated DLL; this behaviour
1932 makes it possible to skip a separate @code{dlltool} import library
1934 [This option is specific to the i386 PE targeted port of the linker]
1936 @kindex --enable-auto-image-base
1937 @item --enable-auto-image-base
1938 Automatically choose the image base for DLLs, unless one is specified
1939 using the @code{--image-base} argument. By using a hash generated
1940 from the dllname to create unique image bases for each DLL, in-memory
1941 collisions and relocations which can delay program execution are
1943 [This option is specific to the i386 PE targeted port of the linker]
1945 @kindex --disable-auto-image-base
1946 @item --disable-auto-image-base
1947 Do not automatically generate a unique image base. If there is no
1948 user-specified image base (@code{--image-base}) then use the platform
1950 [This option is specific to the i386 PE targeted port of the linker]
1952 @cindex DLLs, linking to
1953 @kindex --dll-search-prefix
1954 @item --dll-search-prefix @var{string}
1955 When linking dynamically to a dll without an import library,
1956 search for @code{<string><basename>.dll} in preference to
1957 @code{lib<basename>.dll}. This behaviour allows easy distinction
1958 between DLLs built for the various "subplatforms": native, cygwin,
1959 uwin, pw, etc. For instance, cygwin DLLs typically use
1960 @code{--dll-search-prefix=cyg}.
1961 [This option is specific to the i386 PE targeted port of the linker]
1963 @kindex --enable-auto-import
1964 @item --enable-auto-import
1965 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
1966 DATA imports from DLLs, and create the necessary thunking symbols when
1967 building the import libraries with those DATA exports. Note: Use of the
1968 'auto-import' extension will cause the text section of the image file
1969 to be made writable. This does not conform to the PE-COFF format
1970 specification published by Microsoft.
1972 Using 'auto-import' generally will 'just work' -- but sometimes you may
1975 "variable '<var>' can't be auto-imported. Please read the
1976 documentation for ld's @code{--enable-auto-import} for details."
1978 This message occurs when some (sub)expression accesses an address
1979 ultimately given by the sum of two constants (Win32 import tables only
1980 allow one). Instances where this may occur include accesses to member
1981 fields of struct variables imported from a DLL, as well as using a
1982 constant index into an array variable imported from a DLL. Any
1983 multiword variable (arrays, structs, long long, etc) may trigger
1984 this error condition. However, regardless of the exact data type
1985 of the offending exported variable, ld will always detect it, issue
1986 the warning, and exit.
1988 There are several ways to address this difficulty, regardless of the
1989 data type of the exported variable:
1991 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
1992 of adjusting references in your client code for runtime environment, so
1993 this method works only when runtime environment supports this feature.
1995 A second solution is to force one of the 'constants' to be a variable --
1996 that is, unknown and un-optimizable at compile time. For arrays,
1997 there are two possibilities: a) make the indexee (the array's address)
1998 a variable, or b) make the 'constant' index a variable. Thus:
2001 extern type extern_array[];
2003 @{ volatile type *t=extern_array; t[1] @}
2009 extern type extern_array[];
2011 @{ volatile int t=1; extern_array[t] @}
2014 For structs (and most other multiword data types) the only option
2015 is to make the struct itself (or the long long, or the ...) variable:
2018 extern struct s extern_struct;
2019 extern_struct.field -->
2020 @{ volatile struct s *t=&extern_struct; t->field @}
2026 extern long long extern_ll;
2028 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2031 A third method of dealing with this difficulty is to abandon
2032 'auto-import' for the offending symbol and mark it with
2033 @code{__declspec(dllimport)}. However, in practise that
2034 requires using compile-time #defines to indicate whether you are
2035 building a DLL, building client code that will link to the DLL, or
2036 merely building/linking to a static library. In making the choice
2037 between the various methods of resolving the 'direct address with
2038 constant offset' problem, you should consider typical real-world usage:
2046 void main(int argc, char **argv)@{
2047 printf("%d\n",arr[1]);
2057 void main(int argc, char **argv)@{
2058 /* This workaround is for win32 and cygwin; do not "optimize" */
2059 volatile int *parr = arr;
2060 printf("%d\n",parr[1]);
2067 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2068 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2069 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2070 #define FOO_IMPORT __declspec(dllimport)
2074 extern FOO_IMPORT int arr[];
2077 void main(int argc, char **argv)@{
2078 printf("%d\n",arr[1]);
2082 A fourth way to avoid this problem is to re-code your
2083 library to use a functional interface rather than a data interface
2084 for the offending variables (e.g. set_foo() and get_foo() accessor
2086 [This option is specific to the i386 PE targeted port of the linker]
2088 @kindex --disable-auto-import
2089 @item --disable-auto-import
2090 Do not attempt to do sophisticated linking of @code{_symbol} to
2091 @code{__imp__symbol} for DATA imports from DLLs.
2092 [This option is specific to the i386 PE targeted port of the linker]
2094 @kindex --enable-runtime-pseudo-reloc
2095 @item --enable-runtime-pseudo-reloc
2096 If your code contains expressions described in --enable-auto-import section,
2097 that is, DATA imports from DLL with non-zero offset, this switch will create
2098 a vector of 'runtime pseudo relocations' which can be used by runtime
2099 environment to adjust references to such data in your client code.
2100 [This option is specific to the i386 PE targeted port of the linker]
2102 @kindex --disable-runtime-pseudo-reloc
2103 @item --disable-runtime-pseudo-reloc
2104 Do not create pseudo relocations for non-zero offset DATA imports from
2105 DLLs. This is the default.
2106 [This option is specific to the i386 PE targeted port of the linker]
2108 @kindex --enable-extra-pe-debug
2109 @item --enable-extra-pe-debug
2110 Show additional debug info related to auto-import symbol thunking.
2111 [This option is specific to the i386 PE targeted port of the linker]
2113 @kindex --section-alignment
2114 @item --section-alignment
2115 Sets the section alignment. Sections in memory will always begin at
2116 addresses which are a multiple of this number. Defaults to 0x1000.
2117 [This option is specific to the i386 PE targeted port of the linker]
2121 @item --stack @var{reserve}
2122 @itemx --stack @var{reserve},@var{commit}
2123 Specify the amount of memory to reserve (and optionally commit) to be
2124 used as stack for this program. The default is 2Mb reserved, 4K
2126 [This option is specific to the i386 PE targeted port of the linker]
2129 @item --subsystem @var{which}
2130 @itemx --subsystem @var{which}:@var{major}
2131 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2132 Specifies the subsystem under which your program will execute. The
2133 legal values for @var{which} are @code{native}, @code{windows},
2134 @code{console}, and @code{posix}. You may optionally set the
2135 subsystem version also.
2136 [This option is specific to the i386 PE targeted port of the linker]
2144 @section Environment Variables
2146 @c man begin ENVIRONMENT
2148 You can change the behaviour of @command{ld} with the environment variables
2149 @ifclear SingleFormat
2152 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2154 @ifclear SingleFormat
2156 @cindex default input format
2157 @code{GNUTARGET} determines the input-file object format if you don't
2158 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2159 of the BFD names for an input format (@pxref{BFD}). If there is no
2160 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2161 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2162 attempts to discover the input format by examining binary input files;
2163 this method often succeeds, but there are potential ambiguities, since
2164 there is no method of ensuring that the magic number used to specify
2165 object-file formats is unique. However, the configuration procedure for
2166 BFD on each system places the conventional format for that system first
2167 in the search-list, so ambiguities are resolved in favor of convention.
2171 @cindex default emulation
2172 @cindex emulation, default
2173 @code{LDEMULATION} determines the default emulation if you don't use the
2174 @samp{-m} option. The emulation can affect various aspects of linker
2175 behaviour, particularly the default linker script. You can list the
2176 available emulations with the @samp{--verbose} or @samp{-V} options. If
2177 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2178 variable is not defined, the default emulation depends upon how the
2179 linker was configured.
2181 @kindex COLLECT_NO_DEMANGLE
2182 @cindex demangling, default
2183 Normally, the linker will default to demangling symbols. However, if
2184 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2185 default to not demangling symbols. This environment variable is used in
2186 a similar fashion by the @code{gcc} linker wrapper program. The default
2187 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2194 @chapter Linker Scripts
2197 @cindex linker scripts
2198 @cindex command files
2199 Every link is controlled by a @dfn{linker script}. This script is
2200 written in the linker command language.
2202 The main purpose of the linker script is to describe how the sections in
2203 the input files should be mapped into the output file, and to control
2204 the memory layout of the output file. Most linker scripts do nothing
2205 more than this. However, when necessary, the linker script can also
2206 direct the linker to perform many other operations, using the commands
2209 The linker always uses a linker script. If you do not supply one
2210 yourself, the linker will use a default script that is compiled into the
2211 linker executable. You can use the @samp{--verbose} command line option
2212 to display the default linker script. Certain command line options,
2213 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2215 You may supply your own linker script by using the @samp{-T} command
2216 line option. When you do this, your linker script will replace the
2217 default linker script.
2219 You may also use linker scripts implicitly by naming them as input files
2220 to the linker, as though they were files to be linked. @xref{Implicit
2224 * Basic Script Concepts:: Basic Linker Script Concepts
2225 * Script Format:: Linker Script Format
2226 * Simple Example:: Simple Linker Script Example
2227 * Simple Commands:: Simple Linker Script Commands
2228 * Assignments:: Assigning Values to Symbols
2229 * SECTIONS:: SECTIONS Command
2230 * MEMORY:: MEMORY Command
2231 * PHDRS:: PHDRS Command
2232 * VERSION:: VERSION Command
2233 * Expressions:: Expressions in Linker Scripts
2234 * Implicit Linker Scripts:: Implicit Linker Scripts
2237 @node Basic Script Concepts
2238 @section Basic Linker Script Concepts
2239 @cindex linker script concepts
2240 We need to define some basic concepts and vocabulary in order to
2241 describe the linker script language.
2243 The linker combines input files into a single output file. The output
2244 file and each input file are in a special data format known as an
2245 @dfn{object file format}. Each file is called an @dfn{object file}.
2246 The output file is often called an @dfn{executable}, but for our
2247 purposes we will also call it an object file. Each object file has,
2248 among other things, a list of @dfn{sections}. We sometimes refer to a
2249 section in an input file as an @dfn{input section}; similarly, a section
2250 in the output file is an @dfn{output section}.
2252 Each section in an object file has a name and a size. Most sections
2253 also have an associated block of data, known as the @dfn{section
2254 contents}. A section may be marked as @dfn{loadable}, which mean that
2255 the contents should be loaded into memory when the output file is run.
2256 A section with no contents may be @dfn{allocatable}, which means that an
2257 area in memory should be set aside, but nothing in particular should be
2258 loaded there (in some cases this memory must be zeroed out). A section
2259 which is neither loadable nor allocatable typically contains some sort
2260 of debugging information.
2262 Every loadable or allocatable output section has two addresses. The
2263 first is the @dfn{VMA}, or virtual memory address. This is the address
2264 the section will have when the output file is run. The second is the
2265 @dfn{LMA}, or load memory address. This is the address at which the
2266 section will be loaded. In most cases the two addresses will be the
2267 same. An example of when they might be different is when a data section
2268 is loaded into ROM, and then copied into RAM when the program starts up
2269 (this technique is often used to initialize global variables in a ROM
2270 based system). In this case the ROM address would be the LMA, and the
2271 RAM address would be the VMA.
2273 You can see the sections in an object file by using the @code{objdump}
2274 program with the @samp{-h} option.
2276 Every object file also has a list of @dfn{symbols}, known as the
2277 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2278 has a name, and each defined symbol has an address, among other
2279 information. If you compile a C or C++ program into an object file, you
2280 will get a defined symbol for every defined function and global or
2281 static variable. Every undefined function or global variable which is
2282 referenced in the input file will become an undefined symbol.
2284 You can see the symbols in an object file by using the @code{nm}
2285 program, or by using the @code{objdump} program with the @samp{-t}
2289 @section Linker Script Format
2290 @cindex linker script format
2291 Linker scripts are text files.
2293 You write a linker script as a series of commands. Each command is
2294 either a keyword, possibly followed by arguments, or an assignment to a
2295 symbol. You may separate commands using semicolons. Whitespace is
2298 Strings such as file or format names can normally be entered directly.
2299 If the file name contains a character such as a comma which would
2300 otherwise serve to separate file names, you may put the file name in
2301 double quotes. There is no way to use a double quote character in a
2304 You may include comments in linker scripts just as in C, delimited by
2305 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2308 @node Simple Example
2309 @section Simple Linker Script Example
2310 @cindex linker script example
2311 @cindex example of linker script
2312 Many linker scripts are fairly simple.
2314 The simplest possible linker script has just one command:
2315 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2316 memory layout of the output file.
2318 The @samp{SECTIONS} command is a powerful command. Here we will
2319 describe a simple use of it. Let's assume your program consists only of
2320 code, initialized data, and uninitialized data. These will be in the
2321 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2322 Let's assume further that these are the only sections which appear in
2325 For this example, let's say that the code should be loaded at address
2326 0x10000, and that the data should start at address 0x8000000. Here is a
2327 linker script which will do that:
2332 .text : @{ *(.text) @}
2334 .data : @{ *(.data) @}
2335 .bss : @{ *(.bss) @}
2339 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2340 followed by a series of symbol assignments and output section
2341 descriptions enclosed in curly braces.
2343 The first line inside the @samp{SECTIONS} command of the above example
2344 sets the value of the special symbol @samp{.}, which is the location
2345 counter. If you do not specify the address of an output section in some
2346 other way (other ways are described later), the address is set from the
2347 current value of the location counter. The location counter is then
2348 incremented by the size of the output section. At the start of the
2349 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2351 The second line defines an output section, @samp{.text}. The colon is
2352 required syntax which may be ignored for now. Within the curly braces
2353 after the output section name, you list the names of the input sections
2354 which should be placed into this output section. The @samp{*} is a
2355 wildcard which matches any file name. The expression @samp{*(.text)}
2356 means all @samp{.text} input sections in all input files.
2358 Since the location counter is @samp{0x10000} when the output section
2359 @samp{.text} is defined, the linker will set the address of the
2360 @samp{.text} section in the output file to be @samp{0x10000}.
2362 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2363 the output file. The linker will place the @samp{.data} output section
2364 at address @samp{0x8000000}. After the linker places the @samp{.data}
2365 output section, the value of the location counter will be
2366 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2367 effect is that the linker will place the @samp{.bss} output section
2368 immediately after the @samp{.data} output section in memory.
2370 The linker will ensure that each output section has the required
2371 alignment, by increasing the location counter if necessary. In this
2372 example, the specified addresses for the @samp{.text} and @samp{.data}
2373 sections will probably satisfy any alignment constraints, but the linker
2374 may have to create a small gap between the @samp{.data} and @samp{.bss}
2377 That's it! That's a simple and complete linker script.
2379 @node Simple Commands
2380 @section Simple Linker Script Commands
2381 @cindex linker script simple commands
2382 In this section we describe the simple linker script commands.
2385 * Entry Point:: Setting the entry point
2386 * File Commands:: Commands dealing with files
2387 @ifclear SingleFormat
2388 * Format Commands:: Commands dealing with object file formats
2391 * Miscellaneous Commands:: Other linker script commands
2395 @subsection Setting the Entry Point
2396 @kindex ENTRY(@var{symbol})
2397 @cindex start of execution
2398 @cindex first instruction
2400 The first instruction to execute in a program is called the @dfn{entry
2401 point}. You can use the @code{ENTRY} linker script command to set the
2402 entry point. The argument is a symbol name:
2407 There are several ways to set the entry point. The linker will set the
2408 entry point by trying each of the following methods in order, and
2409 stopping when one of them succeeds:
2412 the @samp{-e} @var{entry} command-line option;
2414 the @code{ENTRY(@var{symbol})} command in a linker script;
2416 the value of the symbol @code{start}, if defined;
2418 the address of the first byte of the @samp{.text} section, if present;
2420 The address @code{0}.
2424 @subsection Commands Dealing with Files
2425 @cindex linker script file commands
2426 Several linker script commands deal with files.
2429 @item INCLUDE @var{filename}
2430 @kindex INCLUDE @var{filename}
2431 @cindex including a linker script
2432 Include the linker script @var{filename} at this point. The file will
2433 be searched for in the current directory, and in any directory specified
2434 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2437 @item INPUT(@var{file}, @var{file}, @dots{})
2438 @itemx INPUT(@var{file} @var{file} @dots{})
2439 @kindex INPUT(@var{files})
2440 @cindex input files in linker scripts
2441 @cindex input object files in linker scripts
2442 @cindex linker script input object files
2443 The @code{INPUT} command directs the linker to include the named files
2444 in the link, as though they were named on the command line.
2446 For example, if you always want to include @file{subr.o} any time you do
2447 a link, but you can't be bothered to put it on every link command line,
2448 then you can put @samp{INPUT (subr.o)} in your linker script.
2450 In fact, if you like, you can list all of your input files in the linker
2451 script, and then invoke the linker with nothing but a @samp{-T} option.
2453 In case a @dfn{sysroot prefix} is configured, and the filename starts
2454 with the @samp{/} character, and the script being processed was
2455 located inside the @dfn{sysroot prefix}, the filename will be looked
2456 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2457 open the file in the current directory. If it is not found, the
2458 linker will search through the archive library search path. See the
2459 description of @samp{-L} in @ref{Options,,Command Line Options}.
2461 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2462 name to @code{lib@var{file}.a}, as with the command line argument
2465 When you use the @code{INPUT} command in an implicit linker script, the
2466 files will be included in the link at the point at which the linker
2467 script file is included. This can affect archive searching.
2469 @item GROUP(@var{file}, @var{file}, @dots{})
2470 @itemx GROUP(@var{file} @var{file} @dots{})
2471 @kindex GROUP(@var{files})
2472 @cindex grouping input files
2473 The @code{GROUP} command is like @code{INPUT}, except that the named
2474 files should all be archives, and they are searched repeatedly until no
2475 new undefined references are created. See the description of @samp{-(}
2476 in @ref{Options,,Command Line Options}.
2478 @item OUTPUT(@var{filename})
2479 @kindex OUTPUT(@var{filename})
2480 @cindex output file name in linker scripot
2481 The @code{OUTPUT} command names the output file. Using
2482 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2483 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2484 Line Options}). If both are used, the command line option takes
2487 You can use the @code{OUTPUT} command to define a default name for the
2488 output file other than the usual default of @file{a.out}.
2490 @item SEARCH_DIR(@var{path})
2491 @kindex SEARCH_DIR(@var{path})
2492 @cindex library search path in linker script
2493 @cindex archive search path in linker script
2494 @cindex search path in linker script
2495 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2496 @command{ld} looks for archive libraries. Using
2497 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2498 on the command line (@pxref{Options,,Command Line Options}). If both
2499 are used, then the linker will search both paths. Paths specified using
2500 the command line option are searched first.
2502 @item STARTUP(@var{filename})
2503 @kindex STARTUP(@var{filename})
2504 @cindex first input file
2505 The @code{STARTUP} command is just like the @code{INPUT} command, except
2506 that @var{filename} will become the first input file to be linked, as
2507 though it were specified first on the command line. This may be useful
2508 when using a system in which the entry point is always the start of the
2512 @ifclear SingleFormat
2513 @node Format Commands
2514 @subsection Commands Dealing with Object File Formats
2515 A couple of linker script commands deal with object file formats.
2518 @item OUTPUT_FORMAT(@var{bfdname})
2519 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2520 @kindex OUTPUT_FORMAT(@var{bfdname})
2521 @cindex output file format in linker script
2522 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2523 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2524 exactly like using @samp{--oformat @var{bfdname}} on the command line
2525 (@pxref{Options,,Command Line Options}). If both are used, the command
2526 line option takes precedence.
2528 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2529 formats based on the @samp{-EB} and @samp{-EL} command line options.
2530 This permits the linker script to set the output format based on the
2533 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2534 will be the first argument, @var{default}. If @samp{-EB} is used, the
2535 output format will be the second argument, @var{big}. If @samp{-EL} is
2536 used, the output format will be the third argument, @var{little}.
2538 For example, the default linker script for the MIPS ELF target uses this
2541 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2543 This says that the default format for the output file is
2544 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
2545 option, the output file will be created in the @samp{elf32-littlemips}
2548 @item TARGET(@var{bfdname})
2549 @kindex TARGET(@var{bfdname})
2550 @cindex input file format in linker script
2551 The @code{TARGET} command names the BFD format to use when reading input
2552 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
2553 This command is like using @samp{-b @var{bfdname}} on the command line
2554 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
2555 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
2556 command is also used to set the format for the output file. @xref{BFD}.
2560 @node Miscellaneous Commands
2561 @subsection Other Linker Script Commands
2562 There are a few other linker scripts commands.
2565 @item ASSERT(@var{exp}, @var{message})
2567 @cindex assertion in linker script
2568 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
2569 with an error code, and print @var{message}.
2571 @item EXTERN(@var{symbol} @var{symbol} @dots{})
2573 @cindex undefined symbol in linker script
2574 Force @var{symbol} to be entered in the output file as an undefined
2575 symbol. Doing this may, for example, trigger linking of additional
2576 modules from standard libraries. You may list several @var{symbol}s for
2577 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
2578 command has the same effect as the @samp{-u} command-line option.
2580 @item FORCE_COMMON_ALLOCATION
2581 @kindex FORCE_COMMON_ALLOCATION
2582 @cindex common allocation in linker script
2583 This command has the same effect as the @samp{-d} command-line option:
2584 to make @command{ld} assign space to common symbols even if a relocatable
2585 output file is specified (@samp{-r}).
2587 @item INHIBIT_COMMON_ALLOCATION
2588 @kindex INHIBIT_COMMON_ALLOCATION
2589 @cindex common allocation in linker script
2590 This command has the same effect as the @samp{--no-define-common}
2591 command-line option: to make @code{ld} omit the assignment of addresses
2592 to common symbols even for a non-relocatable output file.
2594 @item NOCROSSREFS(@var{section} @var{section} @dots{})
2595 @kindex NOCROSSREFS(@var{sections})
2596 @cindex cross references
2597 This command may be used to tell @command{ld} to issue an error about any
2598 references among certain output sections.
2600 In certain types of programs, particularly on embedded systems when
2601 using overlays, when one section is loaded into memory, another section
2602 will not be. Any direct references between the two sections would be
2603 errors. For example, it would be an error if code in one section called
2604 a function defined in the other section.
2606 The @code{NOCROSSREFS} command takes a list of output section names. If
2607 @command{ld} detects any cross references between the sections, it reports
2608 an error and returns a non-zero exit status. Note that the
2609 @code{NOCROSSREFS} command uses output section names, not input section
2612 @ifclear SingleFormat
2613 @item OUTPUT_ARCH(@var{bfdarch})
2614 @kindex OUTPUT_ARCH(@var{bfdarch})
2615 @cindex machine architecture
2616 @cindex architecture
2617 Specify a particular output machine architecture. The argument is one
2618 of the names used by the BFD library (@pxref{BFD}). You can see the
2619 architecture of an object file by using the @code{objdump} program with
2620 the @samp{-f} option.
2625 @section Assigning Values to Symbols
2626 @cindex assignment in scripts
2627 @cindex symbol definition, scripts
2628 @cindex variables, defining
2629 You may assign a value to a symbol in a linker script. This will define
2630 the symbol as a global symbol.
2633 * Simple Assignments:: Simple Assignments
2637 @node Simple Assignments
2638 @subsection Simple Assignments
2640 You may assign to a symbol using any of the C assignment operators:
2643 @item @var{symbol} = @var{expression} ;
2644 @itemx @var{symbol} += @var{expression} ;
2645 @itemx @var{symbol} -= @var{expression} ;
2646 @itemx @var{symbol} *= @var{expression} ;
2647 @itemx @var{symbol} /= @var{expression} ;
2648 @itemx @var{symbol} <<= @var{expression} ;
2649 @itemx @var{symbol} >>= @var{expression} ;
2650 @itemx @var{symbol} &= @var{expression} ;
2651 @itemx @var{symbol} |= @var{expression} ;
2654 The first case will define @var{symbol} to the value of
2655 @var{expression}. In the other cases, @var{symbol} must already be
2656 defined, and the value will be adjusted accordingly.
2658 The special symbol name @samp{.} indicates the location counter. You
2659 may only use this within a @code{SECTIONS} command.
2661 The semicolon after @var{expression} is required.
2663 Expressions are defined below; see @ref{Expressions}.
2665 You may write symbol assignments as commands in their own right, or as
2666 statements within a @code{SECTIONS} command, or as part of an output
2667 section description in a @code{SECTIONS} command.
2669 The section of the symbol will be set from the section of the
2670 expression; for more information, see @ref{Expression Section}.
2672 Here is an example showing the three different places that symbol
2673 assignments may be used:
2684 _bdata = (. + 3) & ~ 3;
2685 .data : @{ *(.data) @}
2689 In this example, the symbol @samp{floating_point} will be defined as
2690 zero. The symbol @samp{_etext} will be defined as the address following
2691 the last @samp{.text} input section. The symbol @samp{_bdata} will be
2692 defined as the address following the @samp{.text} output section aligned
2693 upward to a 4 byte boundary.
2698 In some cases, it is desirable for a linker script to define a symbol
2699 only if it is referenced and is not defined by any object included in
2700 the link. For example, traditional linkers defined the symbol
2701 @samp{etext}. However, ANSI C requires that the user be able to use
2702 @samp{etext} as a function name without encountering an error. The
2703 @code{PROVIDE} keyword may be used to define a symbol, such as
2704 @samp{etext}, only if it is referenced but not defined. The syntax is
2705 @code{PROVIDE(@var{symbol} = @var{expression})}.
2707 Here is an example of using @code{PROVIDE} to define @samp{etext}:
2720 In this example, if the program defines @samp{_etext} (with a leading
2721 underscore), the linker will give a multiple definition error. If, on
2722 the other hand, the program defines @samp{etext} (with no leading
2723 underscore), the linker will silently use the definition in the program.
2724 If the program references @samp{etext} but does not define it, the
2725 linker will use the definition in the linker script.
2728 @section SECTIONS Command
2730 The @code{SECTIONS} command tells the linker how to map input sections
2731 into output sections, and how to place the output sections in memory.
2733 The format of the @code{SECTIONS} command is:
2737 @var{sections-command}
2738 @var{sections-command}
2743 Each @var{sections-command} may of be one of the following:
2747 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
2749 a symbol assignment (@pxref{Assignments})
2751 an output section description
2753 an overlay description
2756 The @code{ENTRY} command and symbol assignments are permitted inside the
2757 @code{SECTIONS} command for convenience in using the location counter in
2758 those commands. This can also make the linker script easier to
2759 understand because you can use those commands at meaningful points in
2760 the layout of the output file.
2762 Output section descriptions and overlay descriptions are described
2765 If you do not use a @code{SECTIONS} command in your linker script, the
2766 linker will place each input section into an identically named output
2767 section in the order that the sections are first encountered in the
2768 input files. If all input sections are present in the first file, for
2769 example, the order of sections in the output file will match the order
2770 in the first input file. The first section will be at address zero.
2773 * Output Section Description:: Output section description
2774 * Output Section Name:: Output section name
2775 * Output Section Address:: Output section address
2776 * Input Section:: Input section description
2777 * Output Section Data:: Output section data
2778 * Output Section Keywords:: Output section keywords
2779 * Output Section Discarding:: Output section discarding
2780 * Output Section Attributes:: Output section attributes
2781 * Overlay Description:: Overlay description
2784 @node Output Section Description
2785 @subsection Output Section Description
2786 The full description of an output section looks like this:
2789 @var{section} [@var{address}] [(@var{type})] :
2790 [AT(@var{lma})] [SUBALIGN(@var{subsection_align})]
2792 @var{output-section-command}
2793 @var{output-section-command}
2795 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
2799 Most output sections do not use most of the optional section attributes.
2801 The whitespace around @var{section} is required, so that the section
2802 name is unambiguous. The colon and the curly braces are also required.
2803 The line breaks and other white space are optional.
2805 Each @var{output-section-command} may be one of the following:
2809 a symbol assignment (@pxref{Assignments})
2811 an input section description (@pxref{Input Section})
2813 data values to include directly (@pxref{Output Section Data})
2815 a special output section keyword (@pxref{Output Section Keywords})
2818 @node Output Section Name
2819 @subsection Output Section Name
2820 @cindex name, section
2821 @cindex section name
2822 The name of the output section is @var{section}. @var{section} must
2823 meet the constraints of your output format. In formats which only
2824 support a limited number of sections, such as @code{a.out}, the name
2825 must be one of the names supported by the format (@code{a.out}, for
2826 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
2827 output format supports any number of sections, but with numbers and not
2828 names (as is the case for Oasys), the name should be supplied as a
2829 quoted numeric string. A section name may consist of any sequence of
2830 characters, but a name which contains any unusual characters such as
2831 commas must be quoted.
2833 The output section name @samp{/DISCARD/} is special; @ref{Output Section
2836 @node Output Section Address
2837 @subsection Output Section Description
2838 @cindex address, section
2839 @cindex section address
2840 The @var{address} is an expression for the VMA (the virtual memory
2841 address) of the output section. If you do not provide @var{address},
2842 the linker will set it based on @var{region} if present, or otherwise
2843 based on the current value of the location counter.
2845 If you provide @var{address}, the address of the output section will be
2846 set to precisely that. If you provide neither @var{address} nor
2847 @var{region}, then the address of the output section will be set to the
2848 current value of the location counter aligned to the alignment
2849 requirements of the output section. The alignment requirement of the
2850 output section is the strictest alignment of any input section contained
2851 within the output section.
2855 .text . : @{ *(.text) @}
2860 .text : @{ *(.text) @}
2863 are subtly different. The first will set the address of the
2864 @samp{.text} output section to the current value of the location
2865 counter. The second will set it to the current value of the location
2866 counter aligned to the strictest alignment of a @samp{.text} input
2869 The @var{address} may be an arbitrary expression; @ref{Expressions}.
2870 For example, if you want to align the section on a 0x10 byte boundary,
2871 so that the lowest four bits of the section address are zero, you could
2872 do something like this:
2874 .text ALIGN(0x10) : @{ *(.text) @}
2877 This works because @code{ALIGN} returns the current location counter
2878 aligned upward to the specified value.
2880 Specifying @var{address} for a section will change the value of the
2884 @subsection Input Section Description
2885 @cindex input sections
2886 @cindex mapping input sections to output sections
2887 The most common output section command is an input section description.
2889 The input section description is the most basic linker script operation.
2890 You use output sections to tell the linker how to lay out your program
2891 in memory. You use input section descriptions to tell the linker how to
2892 map the input files into your memory layout.
2895 * Input Section Basics:: Input section basics
2896 * Input Section Wildcards:: Input section wildcard patterns
2897 * Input Section Common:: Input section for common symbols
2898 * Input Section Keep:: Input section and garbage collection
2899 * Input Section Example:: Input section example
2902 @node Input Section Basics
2903 @subsubsection Input Section Basics
2904 @cindex input section basics
2905 An input section description consists of a file name optionally followed
2906 by a list of section names in parentheses.
2908 The file name and the section name may be wildcard patterns, which we
2909 describe further below (@pxref{Input Section Wildcards}).
2911 The most common input section description is to include all input
2912 sections with a particular name in the output section. For example, to
2913 include all input @samp{.text} sections, you would write:
2918 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
2919 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
2920 match all files except the ones specified in the EXCLUDE_FILE list. For
2923 (*(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors))
2925 will cause all .ctors sections from all files except @file{crtend.o} and
2926 @file{otherfile.o} to be included.
2928 There are two ways to include more than one section:
2934 The difference between these is the order in which the @samp{.text} and
2935 @samp{.rdata} input sections will appear in the output section. In the
2936 first example, they will be intermingled, appearing in the same order as
2937 they are found in the linker input. In the second example, all
2938 @samp{.text} input sections will appear first, followed by all
2939 @samp{.rdata} input sections.
2941 You can specify a file name to include sections from a particular file.
2942 You would do this if one or more of your files contain special data that
2943 needs to be at a particular location in memory. For example:
2948 If you use a file name without a list of sections, then all sections in
2949 the input file will be included in the output section. This is not
2950 commonly done, but it may by useful on occasion. For example:
2955 When you use a file name which does not contain any wild card
2956 characters, the linker will first see if you also specified the file
2957 name on the linker command line or in an @code{INPUT} command. If you
2958 did not, the linker will attempt to open the file as an input file, as
2959 though it appeared on the command line. Note that this differs from an
2960 @code{INPUT} command, because the linker will not search for the file in
2961 the archive search path.
2963 @node Input Section Wildcards
2964 @subsubsection Input Section Wildcard Patterns
2965 @cindex input section wildcards
2966 @cindex wildcard file name patterns
2967 @cindex file name wildcard patterns
2968 @cindex section name wildcard patterns
2969 In an input section description, either the file name or the section
2970 name or both may be wildcard patterns.
2972 The file name of @samp{*} seen in many examples is a simple wildcard
2973 pattern for the file name.
2975 The wildcard patterns are like those used by the Unix shell.
2979 matches any number of characters
2981 matches any single character
2983 matches a single instance of any of the @var{chars}; the @samp{-}
2984 character may be used to specify a range of characters, as in
2985 @samp{[a-z]} to match any lower case letter
2987 quotes the following character
2990 When a file name is matched with a wildcard, the wildcard characters
2991 will not match a @samp{/} character (used to separate directory names on
2992 Unix). A pattern consisting of a single @samp{*} character is an
2993 exception; it will always match any file name, whether it contains a
2994 @samp{/} or not. In a section name, the wildcard characters will match
2995 a @samp{/} character.
2997 File name wildcard patterns only match files which are explicitly
2998 specified on the command line or in an @code{INPUT} command. The linker
2999 does not search directories to expand wildcards.
3001 If a file name matches more than one wildcard pattern, or if a file name
3002 appears explicitly and is also matched by a wildcard pattern, the linker
3003 will use the first match in the linker script. For example, this
3004 sequence of input section descriptions is probably in error, because the
3005 @file{data.o} rule will not be used:
3007 .data : @{ *(.data) @}
3008 .data1 : @{ data.o(.data) @}
3012 Normally, the linker will place files and sections matched by wildcards
3013 in the order in which they are seen during the link. You can change
3014 this by using the @code{SORT} keyword, which appears before a wildcard
3015 pattern in parentheses (e.g., @code{SORT(.text*)}). When the
3016 @code{SORT} keyword is used, the linker will sort the files or sections
3017 into ascending order by name before placing them in the output file.
3019 If you ever get confused about where input sections are going, use the
3020 @samp{-M} linker option to generate a map file. The map file shows
3021 precisely how input sections are mapped to output sections.
3023 This example shows how wildcard patterns might be used to partition
3024 files. This linker script directs the linker to place all @samp{.text}
3025 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3026 The linker will place the @samp{.data} section from all files beginning
3027 with an upper case character in @samp{.DATA}; for all other files, the
3028 linker will place the @samp{.data} section in @samp{.data}.
3032 .text : @{ *(.text) @}
3033 .DATA : @{ [A-Z]*(.data) @}
3034 .data : @{ *(.data) @}
3035 .bss : @{ *(.bss) @}
3040 @node Input Section Common
3041 @subsubsection Input Section for Common Symbols
3042 @cindex common symbol placement
3043 @cindex uninitialized data placement
3044 A special notation is needed for common symbols, because in many object
3045 file formats common symbols do not have a particular input section. The
3046 linker treats common symbols as though they are in an input section
3047 named @samp{COMMON}.
3049 You may use file names with the @samp{COMMON} section just as with any
3050 other input sections. You can use this to place common symbols from a
3051 particular input file in one section while common symbols from other
3052 input files are placed in another section.
3054 In most cases, common symbols in input files will be placed in the
3055 @samp{.bss} section in the output file. For example:
3057 .bss @{ *(.bss) *(COMMON) @}
3060 @cindex scommon section
3061 @cindex small common symbols
3062 Some object file formats have more than one type of common symbol. For
3063 example, the MIPS ELF object file format distinguishes standard common
3064 symbols and small common symbols. In this case, the linker will use a
3065 different special section name for other types of common symbols. In
3066 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3067 symbols and @samp{.scommon} for small common symbols. This permits you
3068 to map the different types of common symbols into memory at different
3072 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3073 notation is now considered obsolete. It is equivalent to
3076 @node Input Section Keep
3077 @subsubsection Input Section and Garbage Collection
3079 @cindex garbage collection
3080 When link-time garbage collection is in use (@samp{--gc-sections}),
3081 it is often useful to mark sections that should not be eliminated.
3082 This is accomplished by surrounding an input section's wildcard entry
3083 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3084 @code{KEEP(SORT(*)(.ctors))}.
3086 @node Input Section Example
3087 @subsubsection Input Section Example
3088 The following example is a complete linker script. It tells the linker
3089 to read all of the sections from file @file{all.o} and place them at the
3090 start of output section @samp{outputa} which starts at location
3091 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3092 follows immediately, in the same output section. All of section
3093 @samp{.input2} from @file{foo.o} goes into output section
3094 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3095 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3096 files are written to output section @samp{outputc}.
3124 @node Output Section Data
3125 @subsection Output Section Data
3127 @cindex section data
3128 @cindex output section data
3129 @kindex BYTE(@var{expression})
3130 @kindex SHORT(@var{expression})
3131 @kindex LONG(@var{expression})
3132 @kindex QUAD(@var{expression})
3133 @kindex SQUAD(@var{expression})
3134 You can include explicit bytes of data in an output section by using
3135 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3136 an output section command. Each keyword is followed by an expression in
3137 parentheses providing the value to store (@pxref{Expressions}). The
3138 value of the expression is stored at the current value of the location
3141 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3142 store one, two, four, and eight bytes (respectively). After storing the
3143 bytes, the location counter is incremented by the number of bytes
3146 For example, this will store the byte 1 followed by the four byte value
3147 of the symbol @samp{addr}:
3153 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3154 same; they both store an 8 byte, or 64 bit, value. When both host and
3155 target are 32 bits, an expression is computed as 32 bits. In this case
3156 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3157 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
3159 If the object file format of the output file has an explicit endianness,
3160 which is the normal case, the value will be stored in that endianness.
3161 When the object file format does not have an explicit endianness, as is
3162 true of, for example, S-records, the value will be stored in the
3163 endianness of the first input object file.
3165 Note---these commands only work inside a section description and not
3166 between them, so the following will produce an error from the linker:
3168 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
3170 whereas this will work:
3172 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
3175 @kindex FILL(@var{expression})
3176 @cindex holes, filling
3177 @cindex unspecified memory
3178 You may use the @code{FILL} command to set the fill pattern for the
3179 current section. It is followed by an expression in parentheses. Any
3180 otherwise unspecified regions of memory within the section (for example,
3181 gaps left due to the required alignment of input sections) are filled
3182 with the value of the expression, repeated as
3183 necessary. A @code{FILL} statement covers memory locations after the
3184 point at which it occurs in the section definition; by including more
3185 than one @code{FILL} statement, you can have different fill patterns in
3186 different parts of an output section.
3188 This example shows how to fill unspecified regions of memory with the
3194 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
3195 section attribute, but it only affects the
3196 part of the section following the @code{FILL} command, rather than the
3197 entire section. If both are used, the @code{FILL} command takes
3198 precedence. @xref{Output Section Fill}, for details on the fill
3201 @node Output Section Keywords
3202 @subsection Output Section Keywords
3203 There are a couple of keywords which can appear as output section
3207 @kindex CREATE_OBJECT_SYMBOLS
3208 @cindex input filename symbols
3209 @cindex filename symbols
3210 @item CREATE_OBJECT_SYMBOLS
3211 The command tells the linker to create a symbol for each input file.
3212 The name of each symbol will be the name of the corresponding input
3213 file. The section of each symbol will be the output section in which
3214 the @code{CREATE_OBJECT_SYMBOLS} command appears.
3216 This is conventional for the a.out object file format. It is not
3217 normally used for any other object file format.
3219 @kindex CONSTRUCTORS
3220 @cindex C++ constructors, arranging in link
3221 @cindex constructors, arranging in link
3223 When linking using the a.out object file format, the linker uses an
3224 unusual set construct to support C++ global constructors and
3225 destructors. When linking object file formats which do not support
3226 arbitrary sections, such as ECOFF and XCOFF, the linker will
3227 automatically recognize C++ global constructors and destructors by name.
3228 For these object file formats, the @code{CONSTRUCTORS} command tells the
3229 linker to place constructor information in the output section where the
3230 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
3231 ignored for other object file formats.
3233 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
3234 constructors, and the symbol @w{@code{__DTOR_LIST}} marks the end. The
3235 first word in the list is the number of entries, followed by the address
3236 of each constructor or destructor, followed by a zero word. The
3237 compiler must arrange to actually run the code. For these object file
3238 formats @sc{gnu} C++ normally calls constructors from a subroutine
3239 @code{__main}; a call to @code{__main} is automatically inserted into
3240 the startup code for @code{main}. @sc{gnu} C++ normally runs
3241 destructors either by using @code{atexit}, or directly from the function
3244 For object file formats such as @code{COFF} or @code{ELF} which support
3245 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
3246 addresses of global constructors and destructors into the @code{.ctors}
3247 and @code{.dtors} sections. Placing the following sequence into your
3248 linker script will build the sort of table which the @sc{gnu} C++
3249 runtime code expects to see.
3253 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3258 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3264 If you are using the @sc{gnu} C++ support for initialization priority,
3265 which provides some control over the order in which global constructors
3266 are run, you must sort the constructors at link time to ensure that they
3267 are executed in the correct order. When using the @code{CONSTRUCTORS}
3268 command, use @samp{SORT(CONSTRUCTORS)} instead. When using the
3269 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT(.ctors))} and
3270 @samp{*(SORT(.dtors))} instead of just @samp{*(.ctors)} and
3273 Normally the compiler and linker will handle these issues automatically,
3274 and you will not need to concern yourself with them. However, you may
3275 need to consider this if you are using C++ and writing your own linker
3280 @node Output Section Discarding
3281 @subsection Output Section Discarding
3282 @cindex discarding sections
3283 @cindex sections, discarding
3284 @cindex removing sections
3285 The linker will not create output section which do not have any
3286 contents. This is for convenience when referring to input sections that
3287 may or may not be present in any of the input files. For example:
3292 will only create a @samp{.foo} section in the output file if there is a
3293 @samp{.foo} section in at least one input file.
3295 If you use anything other than an input section description as an output
3296 section command, such as a symbol assignment, then the output section
3297 will always be created, even if there are no matching input sections.
3300 The special output section name @samp{/DISCARD/} may be used to discard
3301 input sections. Any input sections which are assigned to an output
3302 section named @samp{/DISCARD/} are not included in the output file.
3304 @node Output Section Attributes
3305 @subsection Output Section Attributes
3306 @cindex output section attributes
3307 We showed above that the full description of an output section looked
3311 @var{section} [@var{address}] [(@var{type})] :
3312 [AT(@var{lma})] [SUBALIGN(@var{subsection_align})]
3314 @var{output-section-command}
3315 @var{output-section-command}
3317 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3320 We've already described @var{section}, @var{address}, and
3321 @var{output-section-command}. In this section we will describe the
3322 remaining section attributes.
3325 * Output Section Type:: Output section type
3326 * Output Section LMA:: Output section LMA
3327 * Forced Input Alignment:: Forced Input Alignment
3328 * Output Section Region:: Output section region
3329 * Output Section Phdr:: Output section phdr
3330 * Output Section Fill:: Output section fill
3333 @node Output Section Type
3334 @subsubsection Output Section Type
3335 Each output section may have a type. The type is a keyword in
3336 parentheses. The following types are defined:
3340 The section should be marked as not loadable, so that it will not be
3341 loaded into memory when the program is run.
3346 These type names are supported for backward compatibility, and are
3347 rarely used. They all have the same effect: the section should be
3348 marked as not allocatable, so that no memory is allocated for the
3349 section when the program is run.
3353 @cindex prevent unnecessary loading
3354 @cindex loading, preventing
3355 The linker normally sets the attributes of an output section based on
3356 the input sections which map into it. You can override this by using
3357 the section type. For example, in the script sample below, the
3358 @samp{ROM} section is addressed at memory location @samp{0} and does not
3359 need to be loaded when the program is run. The contents of the
3360 @samp{ROM} section will appear in the linker output file as usual.
3364 ROM 0 (NOLOAD) : @{ @dots{} @}
3370 @node Output Section LMA
3371 @subsubsection Output Section LMA
3372 @kindex AT>@var{lma_region}
3373 @kindex AT(@var{lma})
3374 @cindex load address
3375 @cindex section load address
3376 Every section has a virtual address (VMA) and a load address (LMA); see
3377 @ref{Basic Script Concepts}. The address expression which may appear in
3378 an output section description sets the VMA (@pxref{Output Section
3381 The linker will normally set the LMA equal to the VMA. You can change
3382 that by using the @code{AT} keyword. The expression @var{lma} that
3383 follows the @code{AT} keyword specifies the load address of the
3386 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
3387 specify a memory region for the section's load address. @xref{MEMORY}.
3388 Note that if the section has not had a VMA assigned to it then the
3389 linker will use the @var{lma_region} as the VMA region as well.
3390 @xref{Output Section Region}.
3392 @cindex ROM initialized data
3393 @cindex initialized data in ROM
3394 This feature is designed to make it easy to build a ROM image. For
3395 example, the following linker script creates three output sections: one
3396 called @samp{.text}, which starts at @code{0x1000}, one called
3397 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
3398 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
3399 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
3400 defined with the value @code{0x2000}, which shows that the location
3401 counter holds the VMA value, not the LMA value.
3407 .text 0x1000 : @{ *(.text) _etext = . ; @}
3409 AT ( ADDR (.text) + SIZEOF (.text) )
3410 @{ _data = . ; *(.data); _edata = . ; @}
3412 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
3417 The run-time initialization code for use with a program generated with
3418 this linker script would include something like the following, to copy
3419 the initialized data from the ROM image to its runtime address. Notice
3420 how this code takes advantage of the symbols defined by the linker
3425 extern char _etext, _data, _edata, _bstart, _bend;
3426 char *src = &_etext;
3429 /* ROM has data at end of text; copy it. */
3430 while (dst < &_edata) @{
3435 for (dst = &_bstart; dst< &_bend; dst++)
3440 @node Forced Input Alignment
3441 @subsubsection Forced Input Alignment
3442 @kindex SUBALIGN(@var{subsection_align})
3443 @cindex forcing input section alignment
3444 @cindex input section alignment
3445 You can force input section alignment within an output section by using
3446 SUBALIGN. The value specified overrides any alignment given by input
3447 sections, whether larger or smaller.
3449 @node Output Section Region
3450 @subsubsection Output Section Region
3451 @kindex >@var{region}
3452 @cindex section, assigning to memory region
3453 @cindex memory regions and sections
3454 You can assign a section to a previously defined region of memory by
3455 using @samp{>@var{region}}. @xref{MEMORY}.
3457 Here is a simple example:
3460 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
3461 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
3465 @node Output Section Phdr
3466 @subsubsection Output Section Phdr
3468 @cindex section, assigning to program header
3469 @cindex program headers and sections
3470 You can assign a section to a previously defined program segment by
3471 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
3472 one or more segments, then all subsequent allocated sections will be
3473 assigned to those segments as well, unless they use an explicitly
3474 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
3475 linker to not put the section in any segment at all.
3477 Here is a simple example:
3480 PHDRS @{ text PT_LOAD ; @}
3481 SECTIONS @{ .text : @{ *(.text) @} :text @}
3485 @node Output Section Fill
3486 @subsubsection Output Section Fill
3487 @kindex =@var{fillexp}
3488 @cindex section fill pattern
3489 @cindex fill pattern, entire section
3490 You can set the fill pattern for an entire section by using
3491 @samp{=@var{fillexp}}. @var{fillexp} is an expression
3492 (@pxref{Expressions}). Any otherwise unspecified regions of memory
3493 within the output section (for example, gaps left due to the required
3494 alignment of input sections) will be filled with the value, repeated as
3495 necessary. If the fill expression is a simple hex number, ie. a string
3496 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
3497 an arbitrarily long sequence of hex digits can be used to specify the
3498 fill pattern; Leading zeros become part of the pattern too. For all
3499 other cases, including extra parentheses or a unary @code{+}, the fill
3500 pattern is the four least significant bytes of the value of the
3501 expression. In all cases, the number is big-endian.
3503 You can also change the fill value with a @code{FILL} command in the
3504 output section commands; (@pxref{Output Section Data}).
3506 Here is a simple example:
3509 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
3513 @node Overlay Description
3514 @subsection Overlay Description
3517 An overlay description provides an easy way to describe sections which
3518 are to be loaded as part of a single memory image but are to be run at
3519 the same memory address. At run time, some sort of overlay manager will
3520 copy the overlaid sections in and out of the runtime memory address as
3521 required, perhaps by simply manipulating addressing bits. This approach
3522 can be useful, for example, when a certain region of memory is faster
3525 Overlays are described using the @code{OVERLAY} command. The
3526 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
3527 output section description. The full syntax of the @code{OVERLAY}
3528 command is as follows:
3531 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
3535 @var{output-section-command}
3536 @var{output-section-command}
3538 @} [:@var{phdr}@dots{}] [=@var{fill}]
3541 @var{output-section-command}
3542 @var{output-section-command}
3544 @} [:@var{phdr}@dots{}] [=@var{fill}]
3546 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
3550 Everything is optional except @code{OVERLAY} (a keyword), and each
3551 section must have a name (@var{secname1} and @var{secname2} above). The
3552 section definitions within the @code{OVERLAY} construct are identical to
3553 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
3554 except that no addresses and no memory regions may be defined for
3555 sections within an @code{OVERLAY}.
3557 The sections are all defined with the same starting address. The load
3558 addresses of the sections are arranged such that they are consecutive in
3559 memory starting at the load address used for the @code{OVERLAY} as a
3560 whole (as with normal section definitions, the load address is optional,
3561 and defaults to the start address; the start address is also optional,
3562 and defaults to the current value of the location counter).
3564 If the @code{NOCROSSREFS} keyword is used, and there any references
3565 among the sections, the linker will report an error. Since the sections
3566 all run at the same address, it normally does not make sense for one
3567 section to refer directly to another. @xref{Miscellaneous Commands,
3570 For each section within the @code{OVERLAY}, the linker automatically
3571 defines two symbols. The symbol @code{__load_start_@var{secname}} is
3572 defined as the starting load address of the section. The symbol
3573 @code{__load_stop_@var{secname}} is defined as the final load address of
3574 the section. Any characters within @var{secname} which are not legal
3575 within C identifiers are removed. C (or assembler) code may use these
3576 symbols to move the overlaid sections around as necessary.
3578 At the end of the overlay, the value of the location counter is set to
3579 the start address of the overlay plus the size of the largest section.
3581 Here is an example. Remember that this would appear inside a
3582 @code{SECTIONS} construct.
3585 OVERLAY 0x1000 : AT (0x4000)
3587 .text0 @{ o1/*.o(.text) @}
3588 .text1 @{ o2/*.o(.text) @}
3593 This will define both @samp{.text0} and @samp{.text1} to start at
3594 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
3595 @samp{.text1} will be loaded immediately after @samp{.text0}. The
3596 following symbols will be defined: @code{__load_start_text0},
3597 @code{__load_stop_text0}, @code{__load_start_text1},
3598 @code{__load_stop_text1}.
3600 C code to copy overlay @code{.text1} into the overlay area might look
3605 extern char __load_start_text1, __load_stop_text1;
3606 memcpy ((char *) 0x1000, &__load_start_text1,
3607 &__load_stop_text1 - &__load_start_text1);
3611 Note that the @code{OVERLAY} command is just syntactic sugar, since
3612 everything it does can be done using the more basic commands. The above
3613 example could have been written identically as follows.
3617 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
3618 __load_start_text0 = LOADADDR (.text0);
3619 __load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0);
3620 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
3621 __load_start_text1 = LOADADDR (.text1);
3622 __load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1);
3623 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
3628 @section MEMORY Command
3630 @cindex memory regions
3631 @cindex regions of memory
3632 @cindex allocating memory
3633 @cindex discontinuous memory
3634 The linker's default configuration permits allocation of all available
3635 memory. You can override this by using the @code{MEMORY} command.
3637 The @code{MEMORY} command describes the location and size of blocks of
3638 memory in the target. You can use it to describe which memory regions
3639 may be used by the linker, and which memory regions it must avoid. You
3640 can then assign sections to particular memory regions. The linker will
3641 set section addresses based on the memory regions, and will warn about
3642 regions that become too full. The linker will not shuffle sections
3643 around to fit into the available regions.
3645 A linker script may contain at most one use of the @code{MEMORY}
3646 command. However, you can define as many blocks of memory within it as
3647 you wish. The syntax is:
3652 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
3658 The @var{name} is a name used in the linker script to refer to the
3659 region. The region name has no meaning outside of the linker script.
3660 Region names are stored in a separate name space, and will not conflict
3661 with symbol names, file names, or section names. Each memory region
3662 must have a distinct name.
3664 @cindex memory region attributes
3665 The @var{attr} string is an optional list of attributes that specify
3666 whether to use a particular memory region for an input section which is
3667 not explicitly mapped in the linker script. As described in
3668 @ref{SECTIONS}, if you do not specify an output section for some input
3669 section, the linker will create an output section with the same name as
3670 the input section. If you define region attributes, the linker will use
3671 them to select the memory region for the output section that it creates.
3673 The @var{attr} string must consist only of the following characters:
3688 Invert the sense of any of the preceding attributes
3691 If a unmapped section matches any of the listed attributes other than
3692 @samp{!}, it will be placed in the memory region. The @samp{!}
3693 attribute reverses this test, so that an unmapped section will be placed
3694 in the memory region only if it does not match any of the listed
3700 The @var{origin} is an expression for the start address of the memory
3701 region. The expression must evaluate to a constant before memory
3702 allocation is performed, which means that you may not use any section
3703 relative symbols. The keyword @code{ORIGIN} may be abbreviated to
3704 @code{org} or @code{o} (but not, for example, @code{ORG}).
3709 The @var{len} is an expression for the size in bytes of the memory
3710 region. As with the @var{origin} expression, the expression must
3711 evaluate to a constant before memory allocation is performed. The
3712 keyword @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
3714 In the following example, we specify that there are two memory regions
3715 available for allocation: one starting at @samp{0} for 256 kilobytes,
3716 and the other starting at @samp{0x40000000} for four megabytes. The
3717 linker will place into the @samp{rom} memory region every section which
3718 is not explicitly mapped into a memory region, and is either read-only
3719 or executable. The linker will place other sections which are not
3720 explicitly mapped into a memory region into the @samp{ram} memory
3727 rom (rx) : ORIGIN = 0, LENGTH = 256K
3728 ram (!rx) : org = 0x40000000, l = 4M
3733 Once you define a memory region, you can direct the linker to place
3734 specific output sections into that memory region by using the
3735 @samp{>@var{region}} output section attribute. For example, if you have
3736 a memory region named @samp{mem}, you would use @samp{>mem} in the
3737 output section definition. @xref{Output Section Region}. If no address
3738 was specified for the output section, the linker will set the address to
3739 the next available address within the memory region. If the combined
3740 output sections directed to a memory region are too large for the
3741 region, the linker will issue an error message.
3744 @section PHDRS Command
3746 @cindex program headers
3747 @cindex ELF program headers
3748 @cindex program segments
3749 @cindex segments, ELF
3750 The ELF object file format uses @dfn{program headers}, also knows as
3751 @dfn{segments}. The program headers describe how the program should be
3752 loaded into memory. You can print them out by using the @code{objdump}
3753 program with the @samp{-p} option.
3755 When you run an ELF program on a native ELF system, the system loader
3756 reads the program headers in order to figure out how to load the
3757 program. This will only work if the program headers are set correctly.
3758 This manual does not describe the details of how the system loader
3759 interprets program headers; for more information, see the ELF ABI.
3761 The linker will create reasonable program headers by default. However,
3762 in some cases, you may need to specify the program headers more
3763 precisely. You may use the @code{PHDRS} command for this purpose. When
3764 the linker sees the @code{PHDRS} command in the linker script, it will
3765 not create any program headers other than the ones specified.
3767 The linker only pays attention to the @code{PHDRS} command when
3768 generating an ELF output file. In other cases, the linker will simply
3769 ignore @code{PHDRS}.
3771 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
3772 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
3778 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
3779 [ FLAGS ( @var{flags} ) ] ;
3784 The @var{name} is used only for reference in the @code{SECTIONS} command
3785 of the linker script. It is not put into the output file. Program
3786 header names are stored in a separate name space, and will not conflict
3787 with symbol names, file names, or section names. Each program header
3788 must have a distinct name.
3790 Certain program header types describe segments of memory which the
3791 system loader will load from the file. In the linker script, you
3792 specify the contents of these segments by placing allocatable output
3793 sections in the segments. You use the @samp{:@var{phdr}} output section
3794 attribute to place a section in a particular segment. @xref{Output
3797 It is normal to put certain sections in more than one segment. This
3798 merely implies that one segment of memory contains another. You may
3799 repeat @samp{:@var{phdr}}, using it once for each segment which should
3800 contain the section.
3802 If you place a section in one or more segments using @samp{:@var{phdr}},
3803 then the linker will place all subsequent allocatable sections which do
3804 not specify @samp{:@var{phdr}} in the same segments. This is for
3805 convenience, since generally a whole set of contiguous sections will be
3806 placed in a single segment. You can use @code{:NONE} to override the
3807 default segment and tell the linker to not put the section in any
3812 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
3813 the program header type to further describe the contents of the segment.
3814 The @code{FILEHDR} keyword means that the segment should include the ELF
3815 file header. The @code{PHDRS} keyword means that the segment should
3816 include the ELF program headers themselves.
3818 The @var{type} may be one of the following. The numbers indicate the
3819 value of the keyword.
3822 @item @code{PT_NULL} (0)
3823 Indicates an unused program header.
3825 @item @code{PT_LOAD} (1)
3826 Indicates that this program header describes a segment to be loaded from
3829 @item @code{PT_DYNAMIC} (2)
3830 Indicates a segment where dynamic linking information can be found.
3832 @item @code{PT_INTERP} (3)
3833 Indicates a segment where the name of the program interpreter may be
3836 @item @code{PT_NOTE} (4)
3837 Indicates a segment holding note information.
3839 @item @code{PT_SHLIB} (5)
3840 A reserved program header type, defined but not specified by the ELF
3843 @item @code{PT_PHDR} (6)
3844 Indicates a segment where the program headers may be found.
3846 @item @var{expression}
3847 An expression giving the numeric type of the program header. This may
3848 be used for types not defined above.
3851 You can specify that a segment should be loaded at a particular address
3852 in memory by using an @code{AT} expression. This is identical to the
3853 @code{AT} command used as an output section attribute (@pxref{Output
3854 Section LMA}). The @code{AT} command for a program header overrides the
3855 output section attribute.
3857 The linker will normally set the segment flags based on the sections
3858 which comprise the segment. You may use the @code{FLAGS} keyword to
3859 explicitly specify the segment flags. The value of @var{flags} must be
3860 an integer. It is used to set the @code{p_flags} field of the program
3863 Here is an example of @code{PHDRS}. This shows a typical set of program
3864 headers used on a native ELF system.
3870 headers PT_PHDR PHDRS ;
3872 text PT_LOAD FILEHDR PHDRS ;
3874 dynamic PT_DYNAMIC ;
3880 .interp : @{ *(.interp) @} :text :interp
3881 .text : @{ *(.text) @} :text
3882 .rodata : @{ *(.rodata) @} /* defaults to :text */
3884 . = . + 0x1000; /* move to a new page in memory */
3885 .data : @{ *(.data) @} :data
3886 .dynamic : @{ *(.dynamic) @} :data :dynamic
3893 @section VERSION Command
3894 @kindex VERSION @{script text@}
3895 @cindex symbol versions
3896 @cindex version script
3897 @cindex versions of symbols
3898 The linker supports symbol versions when using ELF. Symbol versions are
3899 only useful when using shared libraries. The dynamic linker can use
3900 symbol versions to select a specific version of a function when it runs
3901 a program that may have been linked against an earlier version of the
3904 You can include a version script directly in the main linker script, or
3905 you can supply the version script as an implicit linker script. You can
3906 also use the @samp{--version-script} linker option.
3908 The syntax of the @code{VERSION} command is simply
3910 VERSION @{ version-script-commands @}
3913 The format of the version script commands is identical to that used by
3914 Sun's linker in Solaris 2.5. The version script defines a tree of
3915 version nodes. You specify the node names and interdependencies in the
3916 version script. You can specify which symbols are bound to which
3917 version nodes, and you can reduce a specified set of symbols to local
3918 scope so that they are not globally visible outside of the shared
3921 The easiest way to demonstrate the version script language is with a few
3943 This example version script defines three version nodes. The first
3944 version node defined is @samp{VERS_1.1}; it has no other dependencies.
3945 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
3946 a number of symbols to local scope so that they are not visible outside
3947 of the shared library; this is done using wildcard patterns, so that any
3948 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
3949 is matched. The wildcard patterns available are the same as those used
3950 in the shell when matching filenames (also known as ``globbing'').
3952 Next, the version script defines node @samp{VERS_1.2}. This node
3953 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
3954 to the version node @samp{VERS_1.2}.
3956 Finally, the version script defines node @samp{VERS_2.0}. This node
3957 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
3958 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
3960 When the linker finds a symbol defined in a library which is not
3961 specifically bound to a version node, it will effectively bind it to an
3962 unspecified base version of the library. You can bind all otherwise
3963 unspecified symbols to a given version node by using @samp{global: *;}
3964 somewhere in the version script.
3966 The names of the version nodes have no specific meaning other than what
3967 they might suggest to the person reading them. The @samp{2.0} version
3968 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
3969 However, this would be a confusing way to write a version script.
3971 Node name can be omited, provided it is the only version node
3972 in the version script. Such version script doesn't assign any versions to
3973 symbols, only selects which symbols will be globally visible out and which
3977 @{ global: foo; bar; local: *; @};
3980 When you link an application against a shared library that has versioned
3981 symbols, the application itself knows which version of each symbol it
3982 requires, and it also knows which version nodes it needs from each
3983 shared library it is linked against. Thus at runtime, the dynamic
3984 loader can make a quick check to make sure that the libraries you have
3985 linked against do in fact supply all of the version nodes that the
3986 application will need to resolve all of the dynamic symbols. In this
3987 way it is possible for the dynamic linker to know with certainty that
3988 all external symbols that it needs will be resolvable without having to
3989 search for each symbol reference.
3991 The symbol versioning is in effect a much more sophisticated way of
3992 doing minor version checking that SunOS does. The fundamental problem
3993 that is being addressed here is that typically references to external
3994 functions are bound on an as-needed basis, and are not all bound when
3995 the application starts up. If a shared library is out of date, a
3996 required interface may be missing; when the application tries to use
3997 that interface, it may suddenly and unexpectedly fail. With symbol
3998 versioning, the user will get a warning when they start their program if
3999 the libraries being used with the application are too old.
4001 There are several GNU extensions to Sun's versioning approach. The
4002 first of these is the ability to bind a symbol to a version node in the
4003 source file where the symbol is defined instead of in the versioning
4004 script. This was done mainly to reduce the burden on the library
4005 maintainer. You can do this by putting something like:
4007 __asm__(".symver original_foo,foo@@VERS_1.1");
4010 in the C source file. This renames the function @samp{original_foo} to
4011 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4012 The @samp{local:} directive can be used to prevent the symbol
4013 @samp{original_foo} from being exported. A @samp{.symver} directive
4014 takes precedence over a version script.
4016 The second GNU extension is to allow multiple versions of the same
4017 function to appear in a given shared library. In this way you can make
4018 an incompatible change to an interface without increasing the major
4019 version number of the shared library, while still allowing applications
4020 linked against the old interface to continue to function.
4022 To do this, you must use multiple @samp{.symver} directives in the
4023 source file. Here is an example:
4026 __asm__(".symver original_foo,foo@@");
4027 __asm__(".symver old_foo,foo@@VERS_1.1");
4028 __asm__(".symver old_foo1,foo@@VERS_1.2");
4029 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4032 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4033 unspecified base version of the symbol. The source file that contains this
4034 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4035 @samp{old_foo1}, and @samp{new_foo}.
4037 When you have multiple definitions of a given symbol, there needs to be
4038 some way to specify a default version to which external references to
4039 this symbol will be bound. You can do this with the
4040 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4041 declare one version of a symbol as the default in this manner; otherwise
4042 you would effectively have multiple definitions of the same symbol.
4044 If you wish to bind a reference to a specific version of the symbol
4045 within the shared library, you can use the aliases of convenience
4046 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4047 specifically bind to an external version of the function in question.
4049 You can also specify the language in the version script:
4052 VERSION extern "lang" @{ version-script-commands @}
4055 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4056 The linker will iterate over the list of symbols at the link time and
4057 demangle them according to @samp{lang} before matching them to the
4058 patterns specified in @samp{version-script-commands}.
4061 @section Expressions in Linker Scripts
4064 The syntax for expressions in the linker script language is identical to
4065 that of C expressions. All expressions are evaluated as integers. All
4066 expressions are evaluated in the same size, which is 32 bits if both the
4067 host and target are 32 bits, and is otherwise 64 bits.
4069 You can use and set symbol values in expressions.
4071 The linker defines several special purpose builtin functions for use in
4075 * Constants:: Constants
4076 * Symbols:: Symbol Names
4077 * Location Counter:: The Location Counter
4078 * Operators:: Operators
4079 * Evaluation:: Evaluation
4080 * Expression Section:: The Section of an Expression
4081 * Builtin Functions:: Builtin Functions
4085 @subsection Constants
4086 @cindex integer notation
4087 @cindex constants in linker scripts
4088 All constants are integers.
4090 As in C, the linker considers an integer beginning with @samp{0} to be
4091 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4092 hexadecimal. The linker considers other integers to be decimal.
4094 @cindex scaled integers
4095 @cindex K and M integer suffixes
4096 @cindex M and K integer suffixes
4097 @cindex suffixes for integers
4098 @cindex integer suffixes
4099 In addition, you can use the suffixes @code{K} and @code{M} to scale a
4103 @c END TEXI2ROFF-KILL
4104 @code{1024} or @code{1024*1024}
4108 ${\rm 1024}$ or ${\rm 1024}^2$
4110 @c END TEXI2ROFF-KILL
4111 respectively. For example, the following all refer to the same quantity:
4119 @subsection Symbol Names
4120 @cindex symbol names
4122 @cindex quoted symbol names
4124 Unless quoted, symbol names start with a letter, underscore, or period
4125 and may include letters, digits, underscores, periods, and hyphens.
4126 Unquoted symbol names must not conflict with any keywords. You can
4127 specify a symbol which contains odd characters or has the same name as a
4128 keyword by surrounding the symbol name in double quotes:
4131 "with a space" = "also with a space" + 10;
4134 Since symbols can contain many non-alphabetic characters, it is safest
4135 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
4136 whereas @samp{A - B} is an expression involving subtraction.
4138 @node Location Counter
4139 @subsection The Location Counter
4142 @cindex location counter
4143 @cindex current output location
4144 The special linker variable @dfn{dot} @samp{.} always contains the
4145 current output location counter. Since the @code{.} always refers to a
4146 location in an output section, it may only appear in an expression
4147 within a @code{SECTIONS} command. The @code{.} symbol may appear
4148 anywhere that an ordinary symbol is allowed in an expression.
4151 Assigning a value to @code{.} will cause the location counter to be
4152 moved. This may be used to create holes in the output section. The
4153 location counter may never be moved backwards.
4169 In the previous example, the @samp{.text} section from @file{file1} is
4170 located at the beginning of the output section @samp{output}. It is
4171 followed by a 1000 byte gap. Then the @samp{.text} section from
4172 @file{file2} appears, also with a 1000 byte gap following before the
4173 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
4174 specifies what data to write in the gaps (@pxref{Output Section Fill}).
4176 @cindex dot inside sections
4177 Note: @code{.} actually refers to the byte offset from the start of the
4178 current containing object. Normally this is the @code{SECTIONS}
4179 statement, whose start address is 0, hence @code{.} can be used as an
4180 absolute address. If @code{.} is used inside a section description
4181 however, it refers to the byte offset from the start of that section,
4182 not an absolute address. Thus in a script like this:
4200 The @samp{.text} section will be assigned a starting address of 0x100
4201 and a size of exactly 0x200 bytes, even if there is not enough data in
4202 the @samp{.text} input sections to fill this area. (If there is too
4203 much data, an error will be produced because this would be an attempt to
4204 move @code{.} backwards). The @samp{.data} section will start at 0x500
4205 and it will have an extra 0x600 bytes worth of space after the end of
4206 the values from the @samp{.data} input sections and before the end of
4207 the @samp{.data} output section itself.
4211 @subsection Operators
4212 @cindex operators for arithmetic
4213 @cindex arithmetic operators
4214 @cindex precedence in expressions
4215 The linker recognizes the standard C set of arithmetic operators, with
4216 the standard bindings and precedence levels:
4219 @c END TEXI2ROFF-KILL
4221 precedence associativity Operators Notes
4227 5 left == != > < <= >=
4233 11 right &= += -= *= /= (2)
4237 (1) Prefix operators
4238 (2) @xref{Assignments}.
4242 \vskip \baselineskip
4243 %"lispnarrowing" is the extra indent used generally for smallexample
4244 \hskip\lispnarrowing\vbox{\offinterlineskip
4247 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
4248 height2pt&\omit&&\omit&&\omit&\cr
4249 &Precedence&& Associativity &&{\rm Operators}&\cr
4250 height2pt&\omit&&\omit&&\omit&\cr
4252 height2pt&\omit&&\omit&&\omit&\cr
4254 % '176 is tilde, '~' in tt font
4255 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
4256 &2&&left&&* / \%&\cr
4259 &5&&left&&== != > < <= >=&\cr
4262 &8&&left&&{\&\&}&\cr
4265 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
4267 height2pt&\omit&&\omit&&\omit&\cr}
4272 @obeylines@parskip=0pt@parindent=0pt
4273 @dag@quad Prefix operators.
4274 @ddag@quad @xref{Assignments}.
4277 @c END TEXI2ROFF-KILL
4280 @subsection Evaluation
4281 @cindex lazy evaluation
4282 @cindex expression evaluation order
4283 The linker evaluates expressions lazily. It only computes the value of
4284 an expression when absolutely necessary.
4286 The linker needs some information, such as the value of the start
4287 address of the first section, and the origins and lengths of memory
4288 regions, in order to do any linking at all. These values are computed
4289 as soon as possible when the linker reads in the linker script.
4291 However, other values (such as symbol values) are not known or needed
4292 until after storage allocation. Such values are evaluated later, when
4293 other information (such as the sizes of output sections) is available
4294 for use in the symbol assignment expression.
4296 The sizes of sections cannot be known until after allocation, so
4297 assignments dependent upon these are not performed until after
4300 Some expressions, such as those depending upon the location counter
4301 @samp{.}, must be evaluated during section allocation.
4303 If the result of an expression is required, but the value is not
4304 available, then an error results. For example, a script like the
4310 .text 9+this_isnt_constant :
4316 will cause the error message @samp{non constant expression for initial
4319 @node Expression Section
4320 @subsection The Section of an Expression
4321 @cindex expression sections
4322 @cindex absolute expressions
4323 @cindex relative expressions
4324 @cindex absolute and relocatable symbols
4325 @cindex relocatable and absolute symbols
4326 @cindex symbols, relocatable and absolute
4327 When the linker evaluates an expression, the result is either absolute
4328 or relative to some section. A relative expression is expressed as a
4329 fixed offset from the base of a section.
4331 The position of the expression within the linker script determines
4332 whether it is absolute or relative. An expression which appears within
4333 an output section definition is relative to the base of the output
4334 section. An expression which appears elsewhere will be absolute.
4336 A symbol set to a relative expression will be relocatable if you request
4337 relocatable output using the @samp{-r} option. That means that a
4338 further link operation may change the value of the symbol. The symbol's
4339 section will be the section of the relative expression.
4341 A symbol set to an absolute expression will retain the same value
4342 through any further link operation. The symbol will be absolute, and
4343 will not have any particular associated section.
4345 You can use the builtin function @code{ABSOLUTE} to force an expression
4346 to be absolute when it would otherwise be relative. For example, to
4347 create an absolute symbol set to the address of the end of the output
4348 section @samp{.data}:
4352 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
4356 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
4357 @samp{.data} section.
4359 @node Builtin Functions
4360 @subsection Builtin Functions
4361 @cindex functions in expressions
4362 The linker script language includes a number of builtin functions for
4363 use in linker script expressions.
4366 @item ABSOLUTE(@var{exp})
4367 @kindex ABSOLUTE(@var{exp})
4368 @cindex expression, absolute
4369 Return the absolute (non-relocatable, as opposed to non-negative) value
4370 of the expression @var{exp}. Primarily useful to assign an absolute
4371 value to a symbol within a section definition, where symbol values are
4372 normally section relative. @xref{Expression Section}.
4374 @item ADDR(@var{section})
4375 @kindex ADDR(@var{section})
4376 @cindex section address in expression
4377 Return the absolute address (the VMA) of the named @var{section}. Your
4378 script must previously have defined the location of that section. In
4379 the following example, @code{symbol_1} and @code{symbol_2} are assigned
4386 start_of_output_1 = ABSOLUTE(.);
4391 symbol_1 = ADDR(.output1);
4392 symbol_2 = start_of_output_1;
4398 @item ALIGN(@var{align})
4399 @itemx ALIGN(@var{exp},@var{align})
4400 @kindex ALIGN(@var{align})
4401 @kindex ALIGN(@var{exp},@var{align})
4402 @cindex round up location counter
4403 @cindex align location counter
4404 @cindex round up expression
4405 @cindex align expression
4406 Return the location counter (@code{.}) or arbitrary expression aligned
4407 to the next @var{align} boundary. The single operand @code{ALIGN}
4408 doesn't change the value of the location counter---it just does
4409 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
4410 expression to be aligned upwards (@code{ALIGN(@var{align})} is
4411 equivalent to @code{ALIGN(., @var{align})}).
4413 Here is an example which aligns the output @code{.data} section to the
4414 next @code{0x2000} byte boundary after the preceding section and sets a
4415 variable within the section to the next @code{0x8000} boundary after the
4420 .data ALIGN(0x2000): @{
4422 variable = ALIGN(0x8000);
4428 The first use of @code{ALIGN} in this example specifies the location of
4429 a section because it is used as the optional @var{address} attribute of
4430 a section definition (@pxref{Output Section Address}). The second use
4431 of @code{ALIGN} is used to defines the value of a symbol.
4433 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
4435 @item BLOCK(@var{exp})
4436 @kindex BLOCK(@var{exp})
4437 This is a synonym for @code{ALIGN}, for compatibility with older linker
4438 scripts. It is most often seen when setting the address of an output
4441 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
4442 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
4443 This is equivalent to either
4445 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
4449 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
4452 depending on whether the latter uses fewer @var{commonpagesize} sized pages
4453 for the data segment (area between the result of this expression and
4454 @code{DATA_SEGMENT_END}) than the former or not.
4455 If the latter form is used, it means @var{commonpagesize} bytes of runtime
4456 memory will be saved at the expense of up to @var{commonpagesize} wasted
4457 bytes in the on-disk file.
4459 This expression can only be used directly in @code{SECTIONS} commands, not in
4460 any output section descriptions and only once in the linker script.
4461 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
4462 be the system page size the object wants to be optimized for (while still
4463 working on system page sizes up to @var{maxpagesize}).
4468 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
4471 @item DATA_SEGMENT_END(@var{exp})
4472 @kindex DATA_SEGMENT_END(@var{exp})
4473 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
4474 evaluation purposes.
4477 . = DATA_SEGMENT_END(.);
4480 @item DEFINED(@var{symbol})
4481 @kindex DEFINED(@var{symbol})
4482 @cindex symbol defaults
4483 Return 1 if @var{symbol} is in the linker global symbol table and is
4484 defined before the statement using DEFINED in the script, otherwise
4485 return 0. You can use this function to provide
4486 default values for symbols. For example, the following script fragment
4487 shows how to set a global symbol @samp{begin} to the first location in
4488 the @samp{.text} section---but if a symbol called @samp{begin} already
4489 existed, its value is preserved:
4495 begin = DEFINED(begin) ? begin : . ;
4503 @item LOADADDR(@var{section})
4504 @kindex LOADADDR(@var{section})
4505 @cindex section load address in expression
4506 Return the absolute LMA of the named @var{section}. This is normally
4507 the same as @code{ADDR}, but it may be different if the @code{AT}
4508 attribute is used in the output section definition (@pxref{Output
4512 @item MAX(@var{exp1}, @var{exp2})
4513 Returns the maximum of @var{exp1} and @var{exp2}.
4516 @item MIN(@var{exp1}, @var{exp2})
4517 Returns the minimum of @var{exp1} and @var{exp2}.
4519 @item NEXT(@var{exp})
4520 @kindex NEXT(@var{exp})
4521 @cindex unallocated address, next
4522 Return the next unallocated address that is a multiple of @var{exp}.
4523 This function is closely related to @code{ALIGN(@var{exp})}; unless you
4524 use the @code{MEMORY} command to define discontinuous memory for the
4525 output file, the two functions are equivalent.
4527 @item SIZEOF(@var{section})
4528 @kindex SIZEOF(@var{section})
4529 @cindex section size
4530 Return the size in bytes of the named @var{section}, if that section has
4531 been allocated. If the section has not been allocated when this is
4532 evaluated, the linker will report an error. In the following example,
4533 @code{symbol_1} and @code{symbol_2} are assigned identical values:
4542 symbol_1 = .end - .start ;
4543 symbol_2 = SIZEOF(.output);
4548 @item SIZEOF_HEADERS
4549 @itemx sizeof_headers
4550 @kindex SIZEOF_HEADERS
4552 Return the size in bytes of the output file's headers. This is
4553 information which appears at the start of the output file. You can use
4554 this number when setting the start address of the first section, if you
4555 choose, to facilitate paging.
4557 @cindex not enough room for program headers
4558 @cindex program headers, not enough room
4559 When producing an ELF output file, if the linker script uses the
4560 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
4561 number of program headers before it has determined all the section
4562 addresses and sizes. If the linker later discovers that it needs
4563 additional program headers, it will report an error @samp{not enough
4564 room for program headers}. To avoid this error, you must avoid using
4565 the @code{SIZEOF_HEADERS} function, or you must rework your linker
4566 script to avoid forcing the linker to use additional program headers, or
4567 you must define the program headers yourself using the @code{PHDRS}
4568 command (@pxref{PHDRS}).
4571 @node Implicit Linker Scripts
4572 @section Implicit Linker Scripts
4573 @cindex implicit linker scripts
4574 If you specify a linker input file which the linker can not recognize as
4575 an object file or an archive file, it will try to read the file as a
4576 linker script. If the file can not be parsed as a linker script, the
4577 linker will report an error.
4579 An implicit linker script will not replace the default linker script.
4581 Typically an implicit linker script would contain only symbol
4582 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
4585 Any input files read because of an implicit linker script will be read
4586 at the position in the command line where the implicit linker script was
4587 read. This can affect archive searching.
4590 @node Machine Dependent
4591 @chapter Machine Dependent Features
4593 @cindex machine dependencies
4594 @command{ld} has additional features on some platforms; the following
4595 sections describe them. Machines where @command{ld} has no additional
4596 functionality are not listed.
4600 * H8/300:: @command{ld} and the H8/300
4603 * i960:: @command{ld} and the Intel 960 family
4606 * ARM:: @command{ld} and the ARM family
4609 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
4612 * MMIX:: @command{ld} and MMIX
4615 * MSP430:: @command{ld} and MSP430
4618 * TI COFF:: @command{ld} and TI COFF
4621 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
4624 * Xtensa:: @command{ld} and Xtensa Processors
4635 @section @command{ld} and the H8/300
4637 @cindex H8/300 support
4638 For the H8/300, @command{ld} can perform these global optimizations when
4639 you specify the @samp{--relax} command-line option.
4642 @cindex relaxing on H8/300
4643 @item relaxing address modes
4644 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
4645 targets are within eight bits, and turns them into eight-bit
4646 program-counter relative @code{bsr} and @code{bra} instructions,
4649 @cindex synthesizing on H8/300
4650 @item synthesizing instructions
4651 @c FIXME: specifically mov.b, or any mov instructions really?
4652 @command{ld} finds all @code{mov.b} instructions which use the
4653 sixteen-bit absolute address form, but refer to the top
4654 page of memory, and changes them to use the eight-bit address form.
4655 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
4656 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
4657 top page of memory).
4667 @c This stuff is pointless to say unless you're especially concerned
4668 @c with Renesas chips; don't enable it for generic case, please.
4670 @chapter @command{ld} and Other Renesas Chips
4672 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
4673 H8/500, and SH chips. No special features, commands, or command-line
4674 options are required for these chips.
4684 @section @command{ld} and the Intel 960 Family
4686 @cindex i960 support
4688 You can use the @samp{-A@var{architecture}} command line option to
4689 specify one of the two-letter names identifying members of the 960
4690 family; the option specifies the desired output target, and warns of any
4691 incompatible instructions in the input files. It also modifies the
4692 linker's search strategy for archive libraries, to support the use of
4693 libraries specific to each particular architecture, by including in the
4694 search loop names suffixed with the string identifying the architecture.
4696 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
4697 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
4698 paths, and in any paths you specify with @samp{-L}) for a library with
4711 The first two possibilities would be considered in any event; the last
4712 two are due to the use of @w{@samp{-ACA}}.
4714 You can meaningfully use @samp{-A} more than once on a command line, since
4715 the 960 architecture family allows combination of target architectures; each
4716 use will add another pair of name variants to search for when @w{@samp{-l}}
4717 specifies a library.
4719 @cindex @option{--relax} on i960
4720 @cindex relaxing on i960
4721 @command{ld} supports the @samp{--relax} option for the i960 family. If
4722 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
4723 @code{calx} instructions whose targets are within 24 bits, and turns
4724 them into 24-bit program-counter relative @code{bal} and @code{cal}
4725 instructions, respectively. @command{ld} also turns @code{cal}
4726 instructions into @code{bal} instructions when it determines that the
4727 target subroutine is a leaf routine (that is, the target subroutine does
4728 not itself call any subroutines).
4741 @section @command{ld}'s Support for Interworking Between ARM and Thumb Code
4743 @cindex ARM interworking support
4744 @kindex --support-old-code
4745 For the ARM, @command{ld} will generate code stubs to allow functions calls
4746 betweem ARM and Thumb code. These stubs only work with code that has
4747 been compiled and assembled with the @samp{-mthumb-interwork} command
4748 line option. If it is necessary to link with old ARM object files or
4749 libraries, which have not been compiled with the -mthumb-interwork
4750 option then the @samp{--support-old-code} command line switch should be
4751 given to the linker. This will make it generate larger stub functions
4752 which will work with non-interworking aware ARM code. Note, however,
4753 the linker does not support generating stubs for function calls to
4754 non-interworking aware Thumb code.
4756 @cindex thumb entry point
4757 @cindex entry point, thumb
4758 @kindex --thumb-entry=@var{entry}
4759 The @samp{--thumb-entry} switch is a duplicate of the generic
4760 @samp{--entry} switch, in that it sets the program's starting address.
4761 But it also sets the bottom bit of the address, so that it can be
4762 branched to using a BX instruction, and the program will start
4763 executing in Thumb mode straight away.
4767 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
4768 executables. This option is only valid when linking big-endian objects.
4769 The resulting image will contain big-endian data and little-endian code.
4782 @section @command{ld} and HPPA 32-bit ELF Support
4783 @cindex HPPA multiple sub-space stubs
4784 @kindex --multi-subspace
4785 When generating a shared library, @command{ld} will by default generate
4786 import stubs suitable for use with a single sub-space application.
4787 The @samp{--multi-subspace} switch causes @command{ld} to generate export
4788 stubs, and different (larger) import stubs suitable for use with
4789 multiple sub-spaces.
4791 @cindex HPPA stub grouping
4792 @kindex --stub-group-size=@var{N}
4793 Long branch stubs and import/export stubs are placed by @command{ld} in
4794 stub sections located between groups of input sections.
4795 @samp{--stub-group-size} specifies the maximum size of a group of input
4796 sections handled by one stub section. Since branch offsets are signed,
4797 a stub section may serve two groups of input sections, one group before
4798 the stub section, and one group after it. However, when using
4799 conditional branches that require stubs, it may be better (for branch
4800 prediction) that stub sections only serve one group of input sections.
4801 A negative value for @samp{N} chooses this scheme, ensuring that
4802 branches to stubs always use a negative offset. Two special values of
4803 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
4804 @command{ld} to automatically size input section groups for the branch types
4805 detected, with the same behaviour regarding stub placement as other
4806 positive or negative values of @samp{N} respectively.
4808 Note that @samp{--stub-group-size} does not split input sections. A
4809 single input section larger than the group size specified will of course
4810 create a larger group (of one section). If input sections are too
4811 large, it may not be possible for a branch to reach its stub.
4824 @section @code{ld} and MMIX
4825 For MMIX, there is a choice of generating @code{ELF} object files or
4826 @code{mmo} object files when linking. The simulator @code{mmix}
4827 understands the @code{mmo} format. The binutils @code{objcopy} utility
4828 can translate between the two formats.
4830 There is one special section, the @samp{.MMIX.reg_contents} section.
4831 Contents in this section is assumed to correspond to that of global
4832 registers, and symbols referring to it are translated to special symbols,
4833 equal to registers. In a final link, the start address of the
4834 @samp{.MMIX.reg_contents} section corresponds to the first allocated
4835 global register multiplied by 8. Register @code{$255} is not included in
4836 this section; it is always set to the program entry, which is at the
4837 symbol @code{Main} for @code{mmo} files.
4839 Symbols with the prefix @code{__.MMIX.start.}, for example
4840 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special;
4841 there must be only one each, even if they are local. The default linker
4842 script uses these to set the default start address of a section.
4844 Initial and trailing multiples of zero-valued 32-bit words in a section,
4845 are left out from an mmo file.
4858 @section @code{ld} and MSP430
4859 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
4860 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
4861 just pass @samp{-m help} option to the linker).
4863 @cindex MSP430 extra sections
4864 The linker will recognize some extra sections which are MSP430 specific:
4867 @item @samp{.vectors}
4868 Defines a portion of ROM where interrupt vectors located.
4870 @item @samp{.bootloader}
4871 Defines the bootloader portion of the ROM (if applicable). Any code
4872 in this section will be uploaded to the MPU.
4874 @item @samp{.infomem}
4875 Defines an information memory section (if applicable). Any code in
4876 this section will be uploaded to the MPU.
4878 @item @samp{.infomemnobits}
4879 This is the same as the @samp{.infomem} section except that any code
4880 in this section will not be uploaded to the MPU.
4882 @item @samp{.noinit}
4883 Denotes a portion of RAM located above @samp{.bss} section.
4885 The last two sections are used by gcc.
4899 @section @command{ld}'s Support for Various TI COFF Versions
4900 @cindex TI COFF versions
4901 @kindex --format=@var{version}
4902 The @samp{--format} switch allows selection of one of the various
4903 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
4904 also supported. The TI COFF versions also vary in header byte-order
4905 format; @command{ld} will read any version or byte order, but the output
4906 header format depends on the default specified by the specific target.
4919 @section @command{ld} and WIN32 (cygwin/mingw)
4921 This section describes some of the win32 specific @command{ld} issues.
4922 See @ref{Options,,Command Line Options} for detailed decription of the
4923 command line options mentioned here.
4926 @cindex import libraries
4927 @item import libraries
4928 The standard Windows linker creates and uses so-called import
4929 libraries, which contains information for linking to dll's. They are
4930 regular static archives and are handled as any other static
4931 archive. The cygwin and mingw ports of @command{ld} have specific
4932 support for creating such libraries provided with the
4933 @samp{--out-implib} command line option.
4935 @item exporting DLL symbols
4936 @cindex exporting DLL symbols
4937 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
4940 @item using auto-export functionality
4941 @cindex using auto-export functionality
4942 By default @command{ld} exports symbols with the auto-export functionality,
4943 which is controlled by the following command line options:
4946 @item --export-all-symbols [This is the default]
4947 @item --exclude-symbols
4948 @item --exclude-libs
4951 If, however, @samp{--export-all-symbols} is not given explicitly on the
4952 command line, then the default auto-export behavior will be @emph{disabled}
4953 if either of the following are true:
4956 @item A DEF file is used.
4957 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
4960 @item using a DEF file
4961 @cindex using a DEF file
4962 Another way of exporting symbols is using a DEF file. A DEF file is
4963 an ASCII file containing definitions of symbols which should be
4964 exported when a dll is created. Usually it is named @samp{<dll
4965 name>.def} and is added as any other object file to the linker's
4966 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
4969 gcc -o <output> <objectfiles> <dll name>.def
4972 Using a DEF file turns off the normal auto-export behavior, unless the
4973 @samp{--export-all-symbols} option is also used.
4975 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
4978 LIBRARY "xyz.dll" BASE=0x10000000
4986 This example defines a base address and three symbols. The third
4987 symbol is an alias for the second. For the complete format
4988 specification see ld/deffilep.y in the binutils sources.
4990 @cindex creating a DEF file
4991 While linking a shared dll, @command{ld} is able to create a DEF file
4992 with the @samp{--output-def <file>} command line option.
4994 @item Using decorations
4995 @cindex Using decorations
4996 Another way of marking symbols for export is to modify the source code
4997 itself, so that when building the DLL each symbol to be exported is
5001 __declspec(dllexport) int a_variable
5002 __declspec(dllexport) void a_function(int with_args)
5005 All such symbols will be exported from the DLL. If, however,
5006 any of the object files in the DLL contain symbols decorated in
5007 this way, then the normal auto-export behavior is disabled, unless
5008 the @samp{--export-all-symbols} option is also used.
5010 Note that object files that wish to access these symbols must @emph{not}
5011 decorate them with dllexport. Instead, they should use dllimport,
5015 __declspec(dllimport) int a_variable
5016 __declspec(dllimport) void a_function(int with_args)
5019 This complicates the structure of library header files, because
5020 when included by the library itself the header must declare the
5021 variables and functions as dllexport, but when included by client
5022 code the header must declare them as dllimport. There are a number
5023 of idioms that are typically used to do this; often client code can
5024 omit the __declspec() declaration completely. See
5025 @samp{--enable-auto-import} and @samp{automatic data imports} for more
5029 @cindex automatic data imports
5030 @item automatic data imports
5031 The standard Windows dll format supports data imports from dlls only
5032 by adding special decorations (dllimport/dllexport), which let the
5033 compiler produce specific assembler instructions to deal with this
5034 issue. This increases the effort necessary to port existing Un*x
5035 code to these platforms, especially for large
5036 c++ libraries and applications. The auto-import feature, which was
5037 initially provided by Paul Sokolovsky, allows one to omit the
5038 decorations to archieve a behavior that conforms to that on POSIX/Un*x
5039 platforms. This feature is enabled with the @samp{--enable-auto-import}
5040 command-line option, although it is enabled by default on cygwin/mingw.
5041 The @samp{--enable-auto-import} option itself now serves mainly to
5042 suppress any warnings that are ordinarily emitted when linked objects
5043 trigger the feature's use.
5045 auto-import of variables does not always work flawlessly without
5046 additional assistance. Sometimes, you will see this message
5048 "variable '<var>' can't be auto-imported. Please read the
5049 documentation for ld's @code{--enable-auto-import} for details."
5051 The @samp{--enable-auto-import} documentation explains why this error
5052 occurs, and several methods that can be used to overcome this difficulty.
5053 One of these methods is the @emph{runtime pseudo-relocs} feature, described
5056 @cindex runtime pseudo-relocation
5057 For complex variables imported from DLLs (such as structs or classes),
5058 object files typically contain a base address for the variable and an
5059 offset (@emph{addend}) within the variable--to specify a particular
5060 field or public member, for instance. Unfortunately, the runtime loader used
5061 in win32 environments is incapable of fixing these references at runtime
5062 without the additional information supplied by dllimport/dllexport decorations.
5063 The standard auto-import feature described above is unable to resolve these
5066 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
5067 be resolved without error, while leaving the task of adjusting the references
5068 themselves (with their non-zero addends) to specialized code provided by the
5069 runtime environment. Recent versions of the cygwin and mingw environments and
5070 compilers provide this runtime support; older versions do not. However, the
5071 support is only necessary on the developer's platform; the compiled result will
5072 run without error on an older system.
5074 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
5077 @cindex direct linking to a dll
5078 @item direct linking to a dll
5079 The cygwin/mingw ports of @command{ld} support the direct linking,
5080 including data symbols, to a dll without the usage of any import
5081 libraries. This is much faster and uses much less memory than does the
5082 traditional import library method, expecially when linking large
5083 libraries or applications. When @command{ld} creates an import lib, each
5084 function or variable exported from the dll is stored in its own bfd, even
5085 though a single bfd could contain many exports. The overhead involved in
5086 storing, loading, and processing so many bfd's is quite large, and explains the
5087 tremendous time, memory, and storage needed to link against particularly
5088 large or complex libraries when using import libs.
5090 Linking directly to a dll uses no extra command-line switches other than
5091 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
5092 of names to match each library. All that is needed from the developer's
5093 perspective is an understanding of this search, in order to force ld to
5094 select the dll instead of an import library.
5097 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
5098 to find, in the first directory of its search path,
5109 before moving on to the next directory in the search path.
5111 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
5112 where @samp{<prefix>} is set by the @command{ld} option
5113 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
5114 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
5117 Other win32-based unix environments, such as mingw or pw32, may use other
5118 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
5119 was originally intended to help avoid name conflicts among dll's built for the
5120 various win32/un*x environments, so that (for example) two versions of a zlib dll
5121 could coexist on the same machine.
5123 The generic cygwin/mingw path layout uses a @samp{bin} directory for
5124 applications and dll's and a @samp{lib} directory for the import
5125 libraries (using cygwin nomenclature):
5131 libxxx.dll.a (in case of dll's)
5132 libxxx.a (in case of static archive)
5135 Linking directly to a dll without using the import library can be
5138 1. Use the dll directly by adding the @samp{bin} path to the link line
5140 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
5143 However, as the dll's often have version numbers appended to their names
5144 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
5145 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
5146 not versioned, and do not have this difficulty.
5148 2. Create a symbolic link from the dll to a file in the @samp{lib}
5149 directory according to the above mentioned search pattern. This
5150 should be used to avoid unwanted changes in the tools needed for
5154 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
5157 Then you can link without any make environment changes.
5160 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
5163 This technique also avoids the version number problems, because the following is
5170 libxxx.dll.a -> ../bin/cygxxx-5.dll
5173 Linking directly to a dll without using an import lib will work
5174 even when auto-import features are exercised, and even when
5175 @samp{--enable-runtime-pseudo-relocs} is used.
5177 Given the improvements in speed and memory usage, one might justifiably
5178 wonder why import libraries are used at all. There are two reasons:
5180 1. Until recently, the link-directly-to-dll functionality did @emph{not}
5181 work with auto-imported data.
5183 2. Sometimes it is necessary to include pure static objects within the
5184 import library (which otherwise contains only bfd's for indirection
5185 symbols that point to the exports of a dll). Again, the import lib
5186 for the cygwin kernel makes use of this ability, and it is not
5187 possible to do this without an import lib.
5189 So, import libs are not going away. But the ability to replace
5190 true import libs with a simple symbolic link to (or a copy of)
5191 a dll, in most cases, is a useful addition to the suite of tools
5192 binutils makes available to the win32 developer. Given the
5193 massive improvements in memory requirements during linking, storage
5194 requirements, and linking speed, we expect that many developers
5195 will soon begin to use this feature whenever possible.
5197 @item symbol aliasing
5199 @item adding additional names
5200 Sometimes, it is useful to export symbols with additional names.
5201 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
5202 exported as @samp{_foo} by using special directives in the DEF file
5203 when creating the dll. This will affect also the optional created
5204 import library. Consider the following DEF file:
5207 LIBRARY "xyz.dll" BASE=0x61000000
5214 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
5216 Another method for creating a symbol alias is to create it in the
5217 source code using the "weak" attribute:
5220 void foo () @{ /* Do something. */; @}
5221 void _foo () __attribute__ ((weak, alias ("foo")));
5224 See the gcc manual for more information about attributes and weak
5227 @item renaming symbols
5228 Sometimes it is useful to rename exports. For instance, the cygwin
5229 kernel does this regularly. A symbol @samp{_foo} can be exported as
5230 @samp{foo} but not as @samp{_foo} by using special directives in the
5231 DEF file. (This will also affect the import library, if it is
5232 created). In the following example:
5235 LIBRARY "xyz.dll" BASE=0x61000000
5241 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
5245 Note: using a DEF file disables the default auto-export behavior,
5246 unless the @samp{--export-all-symbols} command line option is used.
5247 If, however, you are trying to rename symbols, then you should list
5248 @emph{all} desired exports in the DEF file, including the symbols
5249 that are not being renamed, and do @emph{not} use the
5250 @samp{--export-all-symbols} option. If you list only the
5251 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
5252 to handle the other symbols, then the both the new names @emph{and}
5253 the original names for the renamed symbols will be exported.
5254 In effect, you'd be aliasing those symbols, not renaming them,
5255 which is probably not what you wanted.
5269 @section @code{ld} and Xtensa Processors
5271 @cindex Xtensa processors
5272 The default @command{ld} behavior for Xtensa processors is to interpret
5273 @code{SECTIONS} commands so that lists of explicitly named sections in a
5274 specification with a wildcard file will be interleaved when necessary to
5275 keep literal pools within the range of PC-relative load offsets. For
5276 example, with the command:
5288 @command{ld} may interleave some of the @code{.literal}
5289 and @code{.text} sections from different object files to ensure that the
5290 literal pools are within the range of PC-relative load offsets. A valid
5291 interleaving might place the @code{.literal} sections from an initial
5292 group of files followed by the @code{.text} sections of that group of
5293 files. Then, the @code{.literal} sections from the rest of the files
5294 and the @code{.text} sections from the rest of the files would follow.
5295 The non-interleaved order can still be specified as:
5301 *(.literal) *(.text)
5306 @cindex @code{--relax} on Xtensa
5307 @cindex relaxing on Xtensa
5309 The Xtensa version of @command{ld} enables the @option{--relax} option by
5310 default to attempt to reduce space in the output image by combining
5311 literals with identical values. It also provides the
5312 @option{--no-relax} option to disable this optimization. When enabled,
5313 the relaxation algorithm ensures that a literal will only be merged with
5314 another literal when the new merged literal location is within the
5315 offset range of all of its uses.
5317 The relaxation mechanism will also attempt to optimize
5318 assembler-generated ``longcall'' sequences of
5319 @code{L32R}/@code{CALLX@var{n}} when the target is known to fit into a
5320 @code{CALL@var{n}} instruction encoding. The current optimization
5321 converts the sequence into @code{NOP}/@code{CALL@var{n}} and removes the
5322 literal referenced by the @code{L32R} instruction.
5329 @ifclear SingleFormat
5334 @cindex object file management
5335 @cindex object formats available
5337 The linker accesses object and archive files using the BFD libraries.
5338 These libraries allow the linker to use the same routines to operate on
5339 object files whatever the object file format. A different object file
5340 format can be supported simply by creating a new BFD back end and adding
5341 it to the library. To conserve runtime memory, however, the linker and
5342 associated tools are usually configured to support only a subset of the
5343 object file formats available. You can use @code{objdump -i}
5344 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
5345 list all the formats available for your configuration.
5347 @cindex BFD requirements
5348 @cindex requirements for BFD
5349 As with most implementations, BFD is a compromise between
5350 several conflicting requirements. The major factor influencing
5351 BFD design was efficiency: any time used converting between
5352 formats is time which would not have been spent had BFD not
5353 been involved. This is partly offset by abstraction payback; since
5354 BFD simplifies applications and back ends, more time and care
5355 may be spent optimizing algorithms for a greater speed.
5357 One minor artifact of the BFD solution which you should bear in
5358 mind is the potential for information loss. There are two places where
5359 useful information can be lost using the BFD mechanism: during
5360 conversion and during output. @xref{BFD information loss}.
5363 * BFD outline:: How it works: an outline of BFD
5367 @section How It Works: An Outline of BFD
5368 @cindex opening object files
5369 @include bfdsumm.texi
5372 @node Reporting Bugs
5373 @chapter Reporting Bugs
5374 @cindex bugs in @command{ld}
5375 @cindex reporting bugs in @command{ld}
5377 Your bug reports play an essential role in making @command{ld} reliable.
5379 Reporting a bug may help you by bringing a solution to your problem, or
5380 it may not. But in any case the principal function of a bug report is
5381 to help the entire community by making the next version of @command{ld}
5382 work better. Bug reports are your contribution to the maintenance of
5385 In order for a bug report to serve its purpose, you must include the
5386 information that enables us to fix the bug.
5389 * Bug Criteria:: Have you found a bug?
5390 * Bug Reporting:: How to report bugs
5394 @section Have You Found a Bug?
5395 @cindex bug criteria
5397 If you are not sure whether you have found a bug, here are some guidelines:
5400 @cindex fatal signal
5401 @cindex linker crash
5402 @cindex crash of linker
5404 If the linker gets a fatal signal, for any input whatever, that is a
5405 @command{ld} bug. Reliable linkers never crash.
5407 @cindex error on valid input
5409 If @command{ld} produces an error message for valid input, that is a bug.
5411 @cindex invalid input
5413 If @command{ld} does not produce an error message for invalid input, that
5414 may be a bug. In the general case, the linker can not verify that
5415 object files are correct.
5418 If you are an experienced user of linkers, your suggestions for
5419 improvement of @command{ld} are welcome in any case.
5423 @section How to Report Bugs
5425 @cindex @command{ld} bugs, reporting
5427 A number of companies and individuals offer support for @sc{gnu}
5428 products. If you obtained @command{ld} from a support organization, we
5429 recommend you contact that organization first.
5431 You can find contact information for many support companies and
5432 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
5435 Otherwise, send bug reports for @command{ld} to
5436 @samp{bug-binutils@@gnu.org}.
5438 The fundamental principle of reporting bugs usefully is this:
5439 @strong{report all the facts}. If you are not sure whether to state a
5440 fact or leave it out, state it!
5442 Often people omit facts because they think they know what causes the
5443 problem and assume that some details do not matter. Thus, you might
5444 assume that the name of a symbol you use in an example does not
5445 matter. Well, probably it does not, but one cannot be sure. Perhaps
5446 the bug is a stray memory reference which happens to fetch from the
5447 location where that name is stored in memory; perhaps, if the name
5448 were different, the contents of that location would fool the linker
5449 into doing the right thing despite the bug. Play it safe and give a
5450 specific, complete example. That is the easiest thing for you to do,
5451 and the most helpful.
5453 Keep in mind that the purpose of a bug report is to enable us to fix
5454 the bug if it is new to us. Therefore, always write your bug reports
5455 on the assumption that the bug has not been reported previously.
5457 Sometimes people give a few sketchy facts and ask, ``Does this ring a
5458 bell?'' This cannot help us fix a bug, so it is basically useless. We
5459 respond by asking for enough details to enable us to investigate.
5460 You might as well expedite matters by sending them to begin with.
5462 To enable us to fix the bug, you should include all these things:
5466 The version of @command{ld}. @command{ld} announces it if you start it with
5467 the @samp{--version} argument.
5469 Without this, we will not know whether there is any point in looking for
5470 the bug in the current version of @command{ld}.
5473 Any patches you may have applied to the @command{ld} source, including any
5474 patches made to the @code{BFD} library.
5477 The type of machine you are using, and the operating system name and
5481 What compiler (and its version) was used to compile @command{ld}---e.g.
5485 The command arguments you gave the linker to link your example and
5486 observe the bug. To guarantee you will not omit something important,
5487 list them all. A copy of the Makefile (or the output from make) is
5490 If we were to try to guess the arguments, we would probably guess wrong
5491 and then we might not encounter the bug.
5494 A complete input file, or set of input files, that will reproduce the
5495 bug. It is generally most helpful to send the actual object files
5496 provided that they are reasonably small. Say no more than 10K. For
5497 bigger files you can either make them available by FTP or HTTP or else
5498 state that you are willing to send the object file(s) to whomever
5499 requests them. (Note - your email will be going to a mailing list, so
5500 we do not want to clog it up with large attachments). But small
5501 attachments are best.
5503 If the source files were assembled using @code{gas} or compiled using
5504 @code{gcc}, then it may be OK to send the source files rather than the
5505 object files. In this case, be sure to say exactly what version of
5506 @code{gas} or @code{gcc} was used to produce the object files. Also say
5507 how @code{gas} or @code{gcc} were configured.
5510 A description of what behavior you observe that you believe is
5511 incorrect. For example, ``It gets a fatal signal.''
5513 Of course, if the bug is that @command{ld} gets a fatal signal, then we
5514 will certainly notice it. But if the bug is incorrect output, we might
5515 not notice unless it is glaringly wrong. You might as well not give us
5516 a chance to make a mistake.
5518 Even if the problem you experience is a fatal signal, you should still
5519 say so explicitly. Suppose something strange is going on, such as, your
5520 copy of @command{ld} is out of synch, or you have encountered a bug in the
5521 C library on your system. (This has happened!) Your copy might crash
5522 and ours would not. If you told us to expect a crash, then when ours
5523 fails to crash, we would know that the bug was not happening for us. If
5524 you had not told us to expect a crash, then we would not be able to draw
5525 any conclusion from our observations.
5528 If you wish to suggest changes to the @command{ld} source, send us context
5529 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
5530 @samp{-p} option. Always send diffs from the old file to the new file.
5531 If you even discuss something in the @command{ld} source, refer to it by
5532 context, not by line number.
5534 The line numbers in our development sources will not match those in your
5535 sources. Your line numbers would convey no useful information to us.
5538 Here are some things that are not necessary:
5542 A description of the envelope of the bug.
5544 Often people who encounter a bug spend a lot of time investigating
5545 which changes to the input file will make the bug go away and which
5546 changes will not affect it.
5548 This is often time consuming and not very useful, because the way we
5549 will find the bug is by running a single example under the debugger
5550 with breakpoints, not by pure deduction from a series of examples.
5551 We recommend that you save your time for something else.
5553 Of course, if you can find a simpler example to report @emph{instead}
5554 of the original one, that is a convenience for us. Errors in the
5555 output will be easier to spot, running under the debugger will take
5556 less time, and so on.
5558 However, simplification is not vital; if you do not want to do this,
5559 report the bug anyway and send us the entire test case you used.
5562 A patch for the bug.
5564 A patch for the bug does help us if it is a good one. But do not omit
5565 the necessary information, such as the test case, on the assumption that
5566 a patch is all we need. We might see problems with your patch and decide
5567 to fix the problem another way, or we might not understand it at all.
5569 Sometimes with a program as complicated as @command{ld} it is very hard to
5570 construct an example that will make the program follow a certain path
5571 through the code. If you do not send us the example, we will not be
5572 able to construct one, so we will not be able to verify that the bug is
5575 And if we cannot understand what bug you are trying to fix, or why your
5576 patch should be an improvement, we will not install it. A test case will
5577 help us to understand.
5580 A guess about what the bug is or what it depends on.
5582 Such guesses are usually wrong. Even we cannot guess right about such
5583 things without first using the debugger to find the facts.
5587 @appendix MRI Compatible Script Files
5588 @cindex MRI compatibility
5589 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
5590 linker, @command{ld} can use MRI compatible linker scripts as an
5591 alternative to the more general-purpose linker scripting language
5592 described in @ref{Scripts}. MRI compatible linker scripts have a much
5593 simpler command set than the scripting language otherwise used with
5594 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
5595 linker commands; these commands are described here.
5597 In general, MRI scripts aren't of much use with the @code{a.out} object
5598 file format, since it only has three sections and MRI scripts lack some
5599 features to make use of them.
5601 You can specify a file containing an MRI-compatible script using the
5602 @samp{-c} command-line option.
5604 Each command in an MRI-compatible script occupies its own line; each
5605 command line starts with the keyword that identifies the command (though
5606 blank lines are also allowed for punctuation). If a line of an
5607 MRI-compatible script begins with an unrecognized keyword, @command{ld}
5608 issues a warning message, but continues processing the script.
5610 Lines beginning with @samp{*} are comments.
5612 You can write these commands using all upper-case letters, or all
5613 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
5614 The following list shows only the upper-case form of each command.
5617 @cindex @code{ABSOLUTE} (MRI)
5618 @item ABSOLUTE @var{secname}
5619 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
5620 Normally, @command{ld} includes in the output file all sections from all
5621 the input files. However, in an MRI-compatible script, you can use the
5622 @code{ABSOLUTE} command to restrict the sections that will be present in
5623 your output program. If the @code{ABSOLUTE} command is used at all in a
5624 script, then only the sections named explicitly in @code{ABSOLUTE}
5625 commands will appear in the linker output. You can still use other
5626 input sections (whatever you select on the command line, or using
5627 @code{LOAD}) to resolve addresses in the output file.
5629 @cindex @code{ALIAS} (MRI)
5630 @item ALIAS @var{out-secname}, @var{in-secname}
5631 Use this command to place the data from input section @var{in-secname}
5632 in a section called @var{out-secname} in the linker output file.
5634 @var{in-secname} may be an integer.
5636 @cindex @code{ALIGN} (MRI)
5637 @item ALIGN @var{secname} = @var{expression}
5638 Align the section called @var{secname} to @var{expression}. The
5639 @var{expression} should be a power of two.
5641 @cindex @code{BASE} (MRI)
5642 @item BASE @var{expression}
5643 Use the value of @var{expression} as the lowest address (other than
5644 absolute addresses) in the output file.
5646 @cindex @code{CHIP} (MRI)
5647 @item CHIP @var{expression}
5648 @itemx CHIP @var{expression}, @var{expression}
5649 This command does nothing; it is accepted only for compatibility.
5651 @cindex @code{END} (MRI)
5653 This command does nothing whatever; it's only accepted for compatibility.
5655 @cindex @code{FORMAT} (MRI)
5656 @item FORMAT @var{output-format}
5657 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
5658 language, but restricted to one of these output formats:
5662 S-records, if @var{output-format} is @samp{S}
5665 IEEE, if @var{output-format} is @samp{IEEE}
5668 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
5672 @cindex @code{LIST} (MRI)
5673 @item LIST @var{anything}@dots{}
5674 Print (to the standard output file) a link map, as produced by the
5675 @command{ld} command-line option @samp{-M}.
5677 The keyword @code{LIST} may be followed by anything on the
5678 same line, with no change in its effect.
5680 @cindex @code{LOAD} (MRI)
5681 @item LOAD @var{filename}
5682 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
5683 Include one or more object file @var{filename} in the link; this has the
5684 same effect as specifying @var{filename} directly on the @command{ld}
5687 @cindex @code{NAME} (MRI)
5688 @item NAME @var{output-name}
5689 @var{output-name} is the name for the program produced by @command{ld}; the
5690 MRI-compatible command @code{NAME} is equivalent to the command-line
5691 option @samp{-o} or the general script language command @code{OUTPUT}.
5693 @cindex @code{ORDER} (MRI)
5694 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
5695 @itemx ORDER @var{secname} @var{secname} @var{secname}
5696 Normally, @command{ld} orders the sections in its output file in the
5697 order in which they first appear in the input files. In an MRI-compatible
5698 script, you can override this ordering with the @code{ORDER} command. The
5699 sections you list with @code{ORDER} will appear first in your output
5700 file, in the order specified.
5702 @cindex @code{PUBLIC} (MRI)
5703 @item PUBLIC @var{name}=@var{expression}
5704 @itemx PUBLIC @var{name},@var{expression}
5705 @itemx PUBLIC @var{name} @var{expression}
5706 Supply a value (@var{expression}) for external symbol
5707 @var{name} used in the linker input files.
5709 @cindex @code{SECT} (MRI)
5710 @item SECT @var{secname}, @var{expression}
5711 @itemx SECT @var{secname}=@var{expression}
5712 @itemx SECT @var{secname} @var{expression}
5713 You can use any of these three forms of the @code{SECT} command to
5714 specify the start address (@var{expression}) for section @var{secname}.
5715 If you have more than one @code{SECT} statement for the same
5716 @var{secname}, only the @emph{first} sets the start address.
5727 % I think something like @colophon should be in texinfo. In the
5729 \long\def\colophon{\hbox to0pt{}\vfill
5730 \centerline{The body of this manual is set in}
5731 \centerline{\fontname\tenrm,}
5732 \centerline{with headings in {\bf\fontname\tenbf}}
5733 \centerline{and examples in {\tt\fontname\tentt}.}
5734 \centerline{{\it\fontname\tenit\/} and}
5735 \centerline{{\sl\fontname\tensl\/}}
5736 \centerline{are used for emphasis.}\vfill}
5738 % Blame: doc@cygnus.com, 28mar91.