3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
7 @include configdoc.texi
8 @c (configdoc.texi is generated by the Makefile)
14 @macro gcctabopt{body}
20 @c Configure for the generation of man pages
58 * Ld: (ld). The GNU linker.
64 This file documents the @sc{gnu} linker LD
65 @ifset VERSION_PACKAGE
66 @value{VERSION_PACKAGE}
68 version @value{VERSION}.
70 Copyright (C) 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
71 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
75 Permission is granted to copy, distribute and/or modify this document
76 under the terms of the GNU Free Documentation License, Version 1.1
77 or any later version published by the Free Software Foundation;
78 with no Invariant Sections, with no Front-Cover Texts, and with no
79 Back-Cover Texts. A copy of the license is included in the
80 section entitled ``GNU Free Documentation License''.
82 Permission is granted to process this file through Tex and print the
83 results, provided the printed document carries copying permission
84 notice identical to this one except for the removal of this paragraph
85 (this paragraph not being relevant to the printed manual).
91 @setchapternewpage odd
92 @settitle The GNU linker
97 @ifset VERSION_PACKAGE
98 @subtitle @value{VERSION_PACKAGE}
100 @subtitle Version @value{VERSION}
101 @author Steve Chamberlain
102 @author Ian Lance Taylor
107 \hfill Red Hat Inc\par
108 \hfill nickc\@credhat.com, doc\@redhat.com\par
109 \hfill {\it The GNU linker}\par
110 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
112 \global\parindent=0pt % Steve likes it this way.
115 @vskip 0pt plus 1filll
116 @c man begin COPYRIGHT
117 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
118 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
120 Permission is granted to copy, distribute and/or modify this document
121 under the terms of the GNU Free Documentation License, Version 1.1
122 or any later version published by the Free Software Foundation;
123 with no Invariant Sections, with no Front-Cover Texts, and with no
124 Back-Cover Texts. A copy of the license is included in the
125 section entitled ``GNU Free Documentation License''.
130 @c FIXME: Talk about importance of *order* of args, cmds to linker!
135 This file documents the @sc{gnu} linker ld
136 @ifset VERSION_PACKAGE
137 @value{VERSION_PACKAGE}
139 version @value{VERSION}.
141 This document is distributed under the terms of the GNU Free
142 Documentation License. A copy of the license is included in the
143 section entitled ``GNU Free Documentation License''.
146 * Overview:: Overview
147 * Invocation:: Invocation
148 * Scripts:: Linker Scripts
150 * Machine Dependent:: Machine Dependent Features
154 * H8/300:: ld and the H8/300
157 * Renesas:: ld and other Renesas micros
160 * i960:: ld and the Intel 960 family
163 * ARM:: ld and the ARM family
166 * HPPA ELF32:: ld and HPPA 32-bit ELF
169 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
172 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
175 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
178 * TI COFF:: ld and the TI COFF
181 * Win32:: ld and WIN32 (cygwin/mingw)
184 * Xtensa:: ld and Xtensa Processors
187 @ifclear SingleFormat
190 @c Following blank line required for remaining bug in makeinfo conds/menus
192 * Reporting Bugs:: Reporting Bugs
193 * MRI:: MRI Compatible Script Files
194 * GNU Free Documentation License:: GNU Free Documentation License
195 * LD Index:: LD Index
202 @cindex @sc{gnu} linker
203 @cindex what is this?
206 @c man begin SYNOPSIS
207 ld [@b{options}] @var{objfile} @dots{}
211 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
212 the Info entries for @file{binutils} and
217 @c man begin DESCRIPTION
219 @command{ld} combines a number of object and archive files, relocates
220 their data and ties up symbol references. Usually the last step in
221 compiling a program is to run @command{ld}.
223 @command{ld} accepts Linker Command Language files written in
224 a superset of AT&T's Link Editor Command Language syntax,
225 to provide explicit and total control over the linking process.
229 This man page does not describe the command language; see the
230 @command{ld} entry in @code{info} for full details on the command
231 language and on other aspects of the GNU linker.
234 @ifclear SingleFormat
235 This version of @command{ld} uses the general purpose BFD libraries
236 to operate on object files. This allows @command{ld} to read, combine, and
237 write object files in many different formats---for example, COFF or
238 @code{a.out}. Different formats may be linked together to produce any
239 available kind of object file. @xref{BFD}, for more information.
242 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
243 linkers in providing diagnostic information. Many linkers abandon
244 execution immediately upon encountering an error; whenever possible,
245 @command{ld} continues executing, allowing you to identify other errors
246 (or, in some cases, to get an output file in spite of the error).
253 @c man begin DESCRIPTION
255 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
256 and to be as compatible as possible with other linkers. As a result,
257 you have many choices to control its behavior.
263 * Options:: Command Line Options
264 * Environment:: Environment Variables
268 @section Command Line Options
276 The linker supports a plethora of command-line options, but in actual
277 practice few of them are used in any particular context.
278 @cindex standard Unix system
279 For instance, a frequent use of @command{ld} is to link standard Unix
280 object files on a standard, supported Unix system. On such a system, to
281 link a file @code{hello.o}:
284 ld -o @var{output} /lib/crt0.o hello.o -lc
287 This tells @command{ld} to produce a file called @var{output} as the
288 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
289 the library @code{libc.a}, which will come from the standard search
290 directories. (See the discussion of the @samp{-l} option below.)
292 Some of the command-line options to @command{ld} may be specified at any
293 point in the command line. However, options which refer to files, such
294 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
295 which the option appears in the command line, relative to the object
296 files and other file options. Repeating non-file options with a
297 different argument will either have no further effect, or override prior
298 occurrences (those further to the left on the command line) of that
299 option. Options which may be meaningfully specified more than once are
300 noted in the descriptions below.
303 Non-option arguments are object files or archives which are to be linked
304 together. They may follow, precede, or be mixed in with command-line
305 options, except that an object file argument may not be placed between
306 an option and its argument.
308 Usually the linker is invoked with at least one object file, but you can
309 specify other forms of binary input files using @samp{-l}, @samp{-R},
310 and the script command language. If @emph{no} binary input files at all
311 are specified, the linker does not produce any output, and issues the
312 message @samp{No input files}.
314 If the linker cannot recognize the format of an object file, it will
315 assume that it is a linker script. A script specified in this way
316 augments the main linker script used for the link (either the default
317 linker script or the one specified by using @samp{-T}). This feature
318 permits the linker to link against a file which appears to be an object
319 or an archive, but actually merely defines some symbol values, or uses
320 @code{INPUT} or @code{GROUP} to load other objects. Note that
321 specifying a script in this way merely augments the main linker script;
322 use the @samp{-T} option to replace the default linker script entirely.
325 For options whose names are a single letter,
326 option arguments must either follow the option letter without intervening
327 whitespace, or be given as separate arguments immediately following the
328 option that requires them.
330 For options whose names are multiple letters, either one dash or two can
331 precede the option name; for example, @samp{-trace-symbol} and
332 @samp{--trace-symbol} are equivalent. Note---there is one exception to
333 this rule. Multiple letter options that start with a lower case 'o' can
334 only be preceded by two dashes. This is to reduce confusion with the
335 @samp{-o} option. So for example @samp{-omagic} sets the output file
336 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
339 Arguments to multiple-letter options must either be separated from the
340 option name by an equals sign, or be given as separate arguments
341 immediately following the option that requires them. For example,
342 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
343 Unique abbreviations of the names of multiple-letter options are
346 Note---if the linker is being invoked indirectly, via a compiler driver
347 (e.g. @samp{gcc}) then all the linker command line options should be
348 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
349 compiler driver) like this:
352 gcc -Wl,--startgroup foo.o bar.o -Wl,--endgroup
355 This is important, because otherwise the compiler driver program may
356 silently drop the linker options, resulting in a bad link.
358 Here is a table of the generic command line switches accepted by the GNU
362 @include at-file.texi
364 @kindex -a@var{keyword}
365 @item -a@var{keyword}
366 This option is supported for HP/UX compatibility. The @var{keyword}
367 argument must be one of the strings @samp{archive}, @samp{shared}, or
368 @samp{default}. @samp{-aarchive} is functionally equivalent to
369 @samp{-Bstatic}, and the other two keywords are functionally equivalent
370 to @samp{-Bdynamic}. This option may be used any number of times.
373 @cindex architectures
375 @item -A@var{architecture}
376 @kindex --architecture=@var{arch}
377 @itemx --architecture=@var{architecture}
378 In the current release of @command{ld}, this option is useful only for the
379 Intel 960 family of architectures. In that @command{ld} configuration, the
380 @var{architecture} argument identifies the particular architecture in
381 the 960 family, enabling some safeguards and modifying the
382 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
383 family}, for details.
385 Future releases of @command{ld} may support similar functionality for
386 other architecture families.
389 @ifclear SingleFormat
390 @cindex binary input format
391 @kindex -b @var{format}
392 @kindex --format=@var{format}
395 @item -b @var{input-format}
396 @itemx --format=@var{input-format}
397 @command{ld} may be configured to support more than one kind of object
398 file. If your @command{ld} is configured this way, you can use the
399 @samp{-b} option to specify the binary format for input object files
400 that follow this option on the command line. Even when @command{ld} is
401 configured to support alternative object formats, you don't usually need
402 to specify this, as @command{ld} should be configured to expect as a
403 default input format the most usual format on each machine.
404 @var{input-format} is a text string, the name of a particular format
405 supported by the BFD libraries. (You can list the available binary
406 formats with @samp{objdump -i}.)
409 You may want to use this option if you are linking files with an unusual
410 binary format. You can also use @samp{-b} to switch formats explicitly (when
411 linking object files of different formats), by including
412 @samp{-b @var{input-format}} before each group of object files in a
415 The default format is taken from the environment variable
420 You can also define the input format from a script, using the command
423 see @ref{Format Commands}.
427 @kindex -c @var{MRI-cmdfile}
428 @kindex --mri-script=@var{MRI-cmdfile}
429 @cindex compatibility, MRI
430 @item -c @var{MRI-commandfile}
431 @itemx --mri-script=@var{MRI-commandfile}
432 For compatibility with linkers produced by MRI, @command{ld} accepts script
433 files written in an alternate, restricted command language, described in
435 @ref{MRI,,MRI Compatible Script Files}.
438 the MRI Compatible Script Files section of GNU ld documentation.
440 Introduce MRI script files with
441 the option @samp{-c}; use the @samp{-T} option to run linker
442 scripts written in the general-purpose @command{ld} scripting language.
443 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
444 specified by any @samp{-L} options.
446 @cindex common allocation
453 These three options are equivalent; multiple forms are supported for
454 compatibility with other linkers. They assign space to common symbols
455 even if a relocatable output file is specified (with @samp{-r}). The
456 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
457 @xref{Miscellaneous Commands}.
459 @cindex entry point, from command line
460 @kindex -e @var{entry}
461 @kindex --entry=@var{entry}
463 @itemx --entry=@var{entry}
464 Use @var{entry} as the explicit symbol for beginning execution of your
465 program, rather than the default entry point. If there is no symbol
466 named @var{entry}, the linker will try to parse @var{entry} as a number,
467 and use that as the entry address (the number will be interpreted in
468 base 10; you may use a leading @samp{0x} for base 16, or a leading
469 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
470 and other ways of specifying the entry point.
472 @kindex --exclude-libs
473 @item --exclude-libs @var{lib},@var{lib},...
474 Specifies a list of archive libraries from which symbols should not be automatically
475 exported. The library names may be delimited by commas or colons. Specifying
476 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
477 automatic export. This option is available only for the i386 PE targeted
478 port of the linker and for ELF targeted ports. For i386 PE, symbols
479 explicitly listed in a .def file are still exported, regardless of this
480 option. For ELF targeted ports, symbols affected by this option will
481 be treated as hidden.
483 @cindex dynamic symbol table
485 @kindex --export-dynamic
487 @itemx --export-dynamic
488 When creating a dynamically linked executable, add all symbols to the
489 dynamic symbol table. The dynamic symbol table is the set of symbols
490 which are visible from dynamic objects at run time.
492 If you do not use this option, the dynamic symbol table will normally
493 contain only those symbols which are referenced by some dynamic object
494 mentioned in the link.
496 If you use @code{dlopen} to load a dynamic object which needs to refer
497 back to the symbols defined by the program, rather than some other
498 dynamic object, then you will probably need to use this option when
499 linking the program itself.
501 You can also use the dynamic list to control what symbols should
502 be added to the dynamic symbol table if the output format supports it.
503 See the description of @samp{--dynamic-list}.
505 @ifclear SingleFormat
506 @cindex big-endian objects
510 Link big-endian objects. This affects the default output format.
512 @cindex little-endian objects
515 Link little-endian objects. This affects the default output format.
521 @itemx --auxiliary @var{name}
522 When creating an ELF shared object, set the internal DT_AUXILIARY field
523 to the specified name. This tells the dynamic linker that the symbol
524 table of the shared object should be used as an auxiliary filter on the
525 symbol table of the shared object @var{name}.
527 If you later link a program against this filter object, then, when you
528 run the program, the dynamic linker will see the DT_AUXILIARY field. If
529 the dynamic linker resolves any symbols from the filter object, it will
530 first check whether there is a definition in the shared object
531 @var{name}. If there is one, it will be used instead of the definition
532 in the filter object. The shared object @var{name} need not exist.
533 Thus the shared object @var{name} may be used to provide an alternative
534 implementation of certain functions, perhaps for debugging or for
535 machine specific performance.
537 This option may be specified more than once. The DT_AUXILIARY entries
538 will be created in the order in which they appear on the command line.
543 @itemx --filter @var{name}
544 When creating an ELF shared object, set the internal DT_FILTER field to
545 the specified name. This tells the dynamic linker that the symbol table
546 of the shared object which is being created should be used as a filter
547 on the symbol table of the shared object @var{name}.
549 If you later link a program against this filter object, then, when you
550 run the program, the dynamic linker will see the DT_FILTER field. The
551 dynamic linker will resolve symbols according to the symbol table of the
552 filter object as usual, but it will actually link to the definitions
553 found in the shared object @var{name}. Thus the filter object can be
554 used to select a subset of the symbols provided by the object
557 Some older linkers used the @option{-F} option throughout a compilation
558 toolchain for specifying object-file format for both input and output
560 @ifclear SingleFormat
561 The @sc{gnu} linker uses other mechanisms for this purpose: the
562 @option{-b}, @option{--format}, @option{--oformat} options, the
563 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
564 environment variable.
566 The @sc{gnu} linker will ignore the @option{-F} option when not
567 creating an ELF shared object.
569 @cindex finalization function
571 @item -fini @var{name}
572 When creating an ELF executable or shared object, call NAME when the
573 executable or shared object is unloaded, by setting DT_FINI to the
574 address of the function. By default, the linker uses @code{_fini} as
575 the function to call.
579 Ignored. Provided for compatibility with other tools.
585 @itemx --gpsize=@var{value}
586 Set the maximum size of objects to be optimized using the GP register to
587 @var{size}. This is only meaningful for object file formats such as
588 MIPS ECOFF which supports putting large and small objects into different
589 sections. This is ignored for other object file formats.
591 @cindex runtime library name
593 @kindex -soname=@var{name}
595 @itemx -soname=@var{name}
596 When creating an ELF shared object, set the internal DT_SONAME field to
597 the specified name. When an executable is linked with a shared object
598 which has a DT_SONAME field, then when the executable is run the dynamic
599 linker will attempt to load the shared object specified by the DT_SONAME
600 field rather than the using the file name given to the linker.
603 @cindex incremental link
605 Perform an incremental link (same as option @samp{-r}).
607 @cindex initialization function
609 @item -init @var{name}
610 When creating an ELF executable or shared object, call NAME when the
611 executable or shared object is loaded, by setting DT_INIT to the address
612 of the function. By default, the linker uses @code{_init} as the
615 @cindex archive files, from cmd line
616 @kindex -l@var{archive}
617 @kindex --library=@var{archive}
618 @item -l@var{archive}
619 @itemx --library=@var{archive}
620 Add archive file @var{archive} to the list of files to link. This
621 option may be used any number of times. @command{ld} will search its
622 path-list for occurrences of @code{lib@var{archive}.a} for every
623 @var{archive} specified.
625 On systems which support shared libraries, @command{ld} may also search for
626 libraries with extensions other than @code{.a}. Specifically, on ELF
627 and SunOS systems, @command{ld} will search a directory for a library with
628 an extension of @code{.so} before searching for one with an extension of
629 @code{.a}. By convention, a @code{.so} extension indicates a shared
632 The linker will search an archive only once, at the location where it is
633 specified on the command line. If the archive defines a symbol which
634 was undefined in some object which appeared before the archive on the
635 command line, the linker will include the appropriate file(s) from the
636 archive. However, an undefined symbol in an object appearing later on
637 the command line will not cause the linker to search the archive again.
639 See the @option{-(} option for a way to force the linker to search
640 archives multiple times.
642 You may list the same archive multiple times on the command line.
645 This type of archive searching is standard for Unix linkers. However,
646 if you are using @command{ld} on AIX, note that it is different from the
647 behaviour of the AIX linker.
650 @cindex search directory, from cmd line
652 @kindex --library-path=@var{dir}
653 @item -L@var{searchdir}
654 @itemx --library-path=@var{searchdir}
655 Add path @var{searchdir} to the list of paths that @command{ld} will search
656 for archive libraries and @command{ld} control scripts. You may use this
657 option any number of times. The directories are searched in the order
658 in which they are specified on the command line. Directories specified
659 on the command line are searched before the default directories. All
660 @option{-L} options apply to all @option{-l} options, regardless of the
661 order in which the options appear.
663 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
664 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
667 The default set of paths searched (without being specified with
668 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
669 some cases also on how it was configured. @xref{Environment}.
672 The paths can also be specified in a link script with the
673 @code{SEARCH_DIR} command. Directories specified this way are searched
674 at the point in which the linker script appears in the command line.
677 @kindex -m @var{emulation}
678 @item -m@var{emulation}
679 Emulate the @var{emulation} linker. You can list the available
680 emulations with the @samp{--verbose} or @samp{-V} options.
682 If the @samp{-m} option is not used, the emulation is taken from the
683 @code{LDEMULATION} environment variable, if that is defined.
685 Otherwise, the default emulation depends upon how the linker was
693 Print a link map to the standard output. A link map provides
694 information about the link, including the following:
698 Where object files are mapped into memory.
700 How common symbols are allocated.
702 All archive members included in the link, with a mention of the symbol
703 which caused the archive member to be brought in.
705 The values assigned to symbols.
707 Note - symbols whose values are computed by an expression which
708 involves a reference to a previous value of the same symbol may not
709 have correct result displayed in the link map. This is because the
710 linker discards intermediate results and only retains the final value
711 of an expression. Under such circumstances the linker will display
712 the final value enclosed by square brackets. Thus for example a
713 linker script containing:
721 will produce the following output in the link map if the @option{-M}
726 [0x0000000c] foo = (foo * 0x4)
727 [0x0000000c] foo = (foo + 0x8)
730 See @ref{Expressions} for more information about expressions in linker
735 @cindex read-only text
740 Turn off page alignment of sections, and mark the output as
741 @code{NMAGIC} if possible.
745 @cindex read/write from cmd line
749 Set the text and data sections to be readable and writable. Also, do
750 not page-align the data segment, and disable linking against shared
751 libraries. If the output format supports Unix style magic numbers,
752 mark the output as @code{OMAGIC}. Note: Although a writable text section
753 is allowed for PE-COFF targets, it does not conform to the format
754 specification published by Microsoft.
759 This option negates most of the effects of the @option{-N} option. It
760 sets the text section to be read-only, and forces the data segment to
761 be page-aligned. Note - this option does not enable linking against
762 shared libraries. Use @option{-Bdynamic} for this.
764 @kindex -o @var{output}
765 @kindex --output=@var{output}
766 @cindex naming the output file
767 @item -o @var{output}
768 @itemx --output=@var{output}
769 Use @var{output} as the name for the program produced by @command{ld}; if this
770 option is not specified, the name @file{a.out} is used by default. The
771 script command @code{OUTPUT} can also specify the output file name.
773 @kindex -O @var{level}
774 @cindex generating optimized output
776 If @var{level} is a numeric values greater than zero @command{ld} optimizes
777 the output. This might take significantly longer and therefore probably
778 should only be enabled for the final binary.
781 @kindex --emit-relocs
782 @cindex retain relocations in final executable
785 Leave relocation sections and contents in fully linked executables.
786 Post link analysis and optimization tools may need this information in
787 order to perform correct modifications of executables. This results
788 in larger executables.
790 This option is currently only supported on ELF platforms.
792 @kindex --force-dynamic
793 @cindex forcing the creation of dynamic sections
794 @item --force-dynamic
795 Force the output file to have dynamic sections. This option is specific
799 @cindex relocatable output
801 @kindex --relocatable
804 Generate relocatable output---i.e., generate an output file that can in
805 turn serve as input to @command{ld}. This is often called @dfn{partial
806 linking}. As a side effect, in environments that support standard Unix
807 magic numbers, this option also sets the output file's magic number to
809 @c ; see @option{-N}.
810 If this option is not specified, an absolute file is produced. When
811 linking C++ programs, this option @emph{will not} resolve references to
812 constructors; to do that, use @samp{-Ur}.
814 When an input file does not have the same format as the output file,
815 partial linking is only supported if that input file does not contain any
816 relocations. Different output formats can have further restrictions; for
817 example some @code{a.out}-based formats do not support partial linking
818 with input files in other formats at all.
820 This option does the same thing as @samp{-i}.
822 @kindex -R @var{file}
823 @kindex --just-symbols=@var{file}
824 @cindex symbol-only input
825 @item -R @var{filename}
826 @itemx --just-symbols=@var{filename}
827 Read symbol names and their addresses from @var{filename}, but do not
828 relocate it or include it in the output. This allows your output file
829 to refer symbolically to absolute locations of memory defined in other
830 programs. You may use this option more than once.
832 For compatibility with other ELF linkers, if the @option{-R} option is
833 followed by a directory name, rather than a file name, it is treated as
834 the @option{-rpath} option.
838 @cindex strip all symbols
841 Omit all symbol information from the output file.
844 @kindex --strip-debug
845 @cindex strip debugger symbols
848 Omit debugger symbol information (but not all symbols) from the output file.
852 @cindex input files, displaying
855 Print the names of the input files as @command{ld} processes them.
857 @kindex -T @var{script}
858 @kindex --script=@var{script}
860 @item -T @var{scriptfile}
861 @itemx --script=@var{scriptfile}
862 Use @var{scriptfile} as the linker script. This script replaces
863 @command{ld}'s default linker script (rather than adding to it), so
864 @var{commandfile} must specify everything necessary to describe the
865 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
866 the current directory, @code{ld} looks for it in the directories
867 specified by any preceding @samp{-L} options. Multiple @samp{-T}
870 @kindex -dT @var{script}
871 @kindex --default-script=@var{script}
873 @item -dT @var{scriptfile}
874 @itemx --default-script=@var{scriptfile}
875 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
877 This option is similar to the @option{--script} option except that
878 processing of the script is delayed until after the rest of the
879 command line has been processed. This allows options placed after the
880 @option{--default-script} option on the command line to affect the
881 behaviour of the linker script, which can be important when the linker
882 command line cannot be directly controlled by the user. (eg because
883 the command line is being constructed by another tool, such as
886 @kindex -u @var{symbol}
887 @kindex --undefined=@var{symbol}
888 @cindex undefined symbol
889 @item -u @var{symbol}
890 @itemx --undefined=@var{symbol}
891 Force @var{symbol} to be entered in the output file as an undefined
892 symbol. Doing this may, for example, trigger linking of additional
893 modules from standard libraries. @samp{-u} may be repeated with
894 different option arguments to enter additional undefined symbols. This
895 option is equivalent to the @code{EXTERN} linker script command.
900 For anything other than C++ programs, this option is equivalent to
901 @samp{-r}: it generates relocatable output---i.e., an output file that can in
902 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
903 @emph{does} resolve references to constructors, unlike @samp{-r}.
904 It does not work to use @samp{-Ur} on files that were themselves linked
905 with @samp{-Ur}; once the constructor table has been built, it cannot
906 be added to. Use @samp{-Ur} only for the last partial link, and
907 @samp{-r} for the others.
909 @kindex --unique[=@var{SECTION}]
910 @item --unique[=@var{SECTION}]
911 Creates a separate output section for every input section matching
912 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
913 missing, for every orphan input section. An orphan section is one not
914 specifically mentioned in a linker script. You may use this option
915 multiple times on the command line; It prevents the normal merging of
916 input sections with the same name, overriding output section assignments
926 Display the version number for @command{ld}. The @option{-V} option also
927 lists the supported emulations.
930 @kindex --discard-all
931 @cindex deleting local symbols
934 Delete all local symbols.
937 @kindex --discard-locals
938 @cindex local symbols, deleting
940 @itemx --discard-locals
941 Delete all temporary local symbols. (These symbols start with
942 system-specific local label prefixes, typically @samp{.L} for ELF systems
943 or @samp{L} for traditional a.out systems.)
945 @kindex -y @var{symbol}
946 @kindex --trace-symbol=@var{symbol}
947 @cindex symbol tracing
948 @item -y @var{symbol}
949 @itemx --trace-symbol=@var{symbol}
950 Print the name of each linked file in which @var{symbol} appears. This
951 option may be given any number of times. On many systems it is necessary
952 to prepend an underscore.
954 This option is useful when you have an undefined symbol in your link but
955 don't know where the reference is coming from.
957 @kindex -Y @var{path}
959 Add @var{path} to the default library search path. This option exists
960 for Solaris compatibility.
962 @kindex -z @var{keyword}
963 @item -z @var{keyword}
964 The recognized keywords are:
968 Combines multiple reloc sections and sorts them to make dynamic symbol
969 lookup caching possible.
972 Disallows undefined symbols in object files. Undefined symbols in
973 shared libraries are still allowed.
976 Marks the object as requiring executable stack.
979 This option is only meaningful when building a shared object.
980 It marks the object so that its runtime initialization will occur
981 before the runtime initialization of any other objects brought into
982 the process at the same time. Similarly the runtime finalization of
983 the object will occur after the runtime finalization of any other
987 Marks the object that its symbol table interposes before all symbols
988 but the primary executable.
991 When generating an executable or shared library, mark it to tell the
992 dynamic linker to defer function call resolution to the point when
993 the function is called (lazy binding), rather than at load time.
994 Lazy binding is the default.
997 Marks the object that its filters be processed immediately at
1001 Allows multiple definitions.
1004 Disables multiple reloc sections combining.
1007 Disables production of copy relocs.
1010 Marks the object that the search for dependencies of this object will
1011 ignore any default library search paths.
1014 Marks the object shouldn't be unloaded at runtime.
1017 Marks the object not available to @code{dlopen}.
1020 Marks the object can not be dumped by @code{dldump}.
1023 Marks the object as not requiring executable stack.
1026 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1029 When generating an executable or shared library, mark it to tell the
1030 dynamic linker to resolve all symbols when the program is started, or
1031 when the shared library is linked to using dlopen, instead of
1032 deferring function call resolution to the point when the function is
1036 Marks the object may contain $ORIGIN.
1039 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1041 @item max-page-size=@var{value}
1042 Set the emulation maximum page size to @var{value}.
1044 @item common-page-size=@var{value}
1045 Set the emulation common page size to @var{value}.
1049 Other keywords are ignored for Solaris compatibility.
1052 @cindex groups of archives
1053 @item -( @var{archives} -)
1054 @itemx --start-group @var{archives} --end-group
1055 The @var{archives} should be a list of archive files. They may be
1056 either explicit file names, or @samp{-l} options.
1058 The specified archives are searched repeatedly until no new undefined
1059 references are created. Normally, an archive is searched only once in
1060 the order that it is specified on the command line. If a symbol in that
1061 archive is needed to resolve an undefined symbol referred to by an
1062 object in an archive that appears later on the command line, the linker
1063 would not be able to resolve that reference. By grouping the archives,
1064 they all be searched repeatedly until all possible references are
1067 Using this option has a significant performance cost. It is best to use
1068 it only when there are unavoidable circular references between two or
1071 @kindex --accept-unknown-input-arch
1072 @kindex --no-accept-unknown-input-arch
1073 @item --accept-unknown-input-arch
1074 @itemx --no-accept-unknown-input-arch
1075 Tells the linker to accept input files whose architecture cannot be
1076 recognised. The assumption is that the user knows what they are doing
1077 and deliberately wants to link in these unknown input files. This was
1078 the default behaviour of the linker, before release 2.14. The default
1079 behaviour from release 2.14 onwards is to reject such input files, and
1080 so the @samp{--accept-unknown-input-arch} option has been added to
1081 restore the old behaviour.
1084 @kindex --no-as-needed
1086 @itemx --no-as-needed
1087 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1088 on the command line after the @option{--as-needed} option. Normally,
1089 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1090 on the command line, regardless of whether the library is actually
1091 needed. @option{--as-needed} causes DT_NEEDED tags to only be emitted
1092 for libraries that satisfy some symbol reference from regular objects
1093 which is undefined at the point that the library was linked.
1094 @option{--no-as-needed} restores the default behaviour.
1096 @kindex --add-needed
1097 @kindex --no-add-needed
1099 @itemx --no-add-needed
1100 This option affects the treatment of dynamic libraries from ELF
1101 DT_NEEDED tags in dynamic libraries mentioned on the command line after
1102 the @option{--no-add-needed} option. Normally, the linker will add
1103 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1104 @option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1105 for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1106 the default behaviour.
1108 @kindex -assert @var{keyword}
1109 @item -assert @var{keyword}
1110 This option is ignored for SunOS compatibility.
1114 @kindex -call_shared
1118 Link against dynamic libraries. This is only meaningful on platforms
1119 for which shared libraries are supported. This option is normally the
1120 default on such platforms. The different variants of this option are
1121 for compatibility with various systems. You may use this option
1122 multiple times on the command line: it affects library searching for
1123 @option{-l} options which follow it.
1127 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1128 section. This causes the runtime linker to handle lookups in this
1129 object and its dependencies to be performed only inside the group.
1130 @option{--unresolved-symbols=report-all} is implied. This option is
1131 only meaningful on ELF platforms which support shared libraries.
1141 Do not link against shared libraries. This is only meaningful on
1142 platforms for which shared libraries are supported. The different
1143 variants of this option are for compatibility with various systems. You
1144 may use this option multiple times on the command line: it affects
1145 library searching for @option{-l} options which follow it. This
1146 option also implies @option{--unresolved-symbols=report-all}. This
1147 option can be used with @option{-shared}. Doing so means that a
1148 shared library is being created but that all of the library's external
1149 references must be resolved by pulling in entries from static
1154 When creating a shared library, bind references to global symbols to the
1155 definition within the shared library, if any. Normally, it is possible
1156 for a program linked against a shared library to override the definition
1157 within the shared library. This option is only meaningful on ELF
1158 platforms which support shared libraries.
1160 @kindex -Bsymbolic-functions
1161 @item -Bsymbolic-functions
1162 When creating a shared library, bind references to global function
1163 symbols to the definition within the shared library, if any.
1164 This option is only meaningful on ELF platforms which support shared
1167 @kindex --dynamic-list=@var{dynamic-list-file}
1168 @item --dynamic-list=@var{dynamic-list-file}
1169 Specify the name of a dynamic list file to the linker. This is
1170 typically used when creating shared libraries to specify a list of
1171 global symbols whose references shouldn't be bound to the definition
1172 within the shared library, or creating dynamically linked executables
1173 to specify a list of symbols which should be added to the symbol table
1174 in the executable. This option is only meaningful on ELF platforms
1175 which support shared libraries.
1177 The format of the dynamic list is the same as the version node without
1178 scope and node name. See @ref{VERSION} for more information.
1180 @kindex --dynamic-list-data
1181 @item --dynamic-list-data
1182 Include all global data symbols to the dynamic list.
1184 @kindex --dynamic-list-cpp-new
1185 @item --dynamic-list-cpp-new
1186 Provide the builtin dynamic list for C++ operator new and delete. It
1187 is mainly useful for building shared libstdc++.
1189 @kindex --dynamic-list-cpp-typeinfo
1190 @item --dynamic-list-cpp-typeinfo
1191 Provide the builtin dynamic list for C++ runtime type identification.
1193 @kindex --check-sections
1194 @kindex --no-check-sections
1195 @item --check-sections
1196 @itemx --no-check-sections
1197 Asks the linker @emph{not} to check section addresses after they have
1198 been assigned to see if there are any overlaps. Normally the linker will
1199 perform this check, and if it finds any overlaps it will produce
1200 suitable error messages. The linker does know about, and does make
1201 allowances for sections in overlays. The default behaviour can be
1202 restored by using the command line switch @option{--check-sections}.
1204 @cindex cross reference table
1207 Output a cross reference table. If a linker map file is being
1208 generated, the cross reference table is printed to the map file.
1209 Otherwise, it is printed on the standard output.
1211 The format of the table is intentionally simple, so that it may be
1212 easily processed by a script if necessary. The symbols are printed out,
1213 sorted by name. For each symbol, a list of file names is given. If the
1214 symbol is defined, the first file listed is the location of the
1215 definition. The remaining files contain references to the symbol.
1217 @cindex common allocation
1218 @kindex --no-define-common
1219 @item --no-define-common
1220 This option inhibits the assignment of addresses to common symbols.
1221 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1222 @xref{Miscellaneous Commands}.
1224 The @samp{--no-define-common} option allows decoupling
1225 the decision to assign addresses to Common symbols from the choice
1226 of the output file type; otherwise a non-Relocatable output type
1227 forces assigning addresses to Common symbols.
1228 Using @samp{--no-define-common} allows Common symbols that are referenced
1229 from a shared library to be assigned addresses only in the main program.
1230 This eliminates the unused duplicate space in the shared library,
1231 and also prevents any possible confusion over resolving to the wrong
1232 duplicate when there are many dynamic modules with specialized search
1233 paths for runtime symbol resolution.
1235 @cindex symbols, from command line
1236 @kindex --defsym @var{symbol}=@var{exp}
1237 @item --defsym @var{symbol}=@var{expression}
1238 Create a global symbol in the output file, containing the absolute
1239 address given by @var{expression}. You may use this option as many
1240 times as necessary to define multiple symbols in the command line. A
1241 limited form of arithmetic is supported for the @var{expression} in this
1242 context: you may give a hexadecimal constant or the name of an existing
1243 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1244 constants or symbols. If you need more elaborate expressions, consider
1245 using the linker command language from a script (@pxref{Assignments,,
1246 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1247 space between @var{symbol}, the equals sign (``@key{=}''), and
1250 @cindex demangling, from command line
1251 @kindex --demangle[=@var{style}]
1252 @kindex --no-demangle
1253 @item --demangle[=@var{style}]
1254 @itemx --no-demangle
1255 These options control whether to demangle symbol names in error messages
1256 and other output. When the linker is told to demangle, it tries to
1257 present symbol names in a readable fashion: it strips leading
1258 underscores if they are used by the object file format, and converts C++
1259 mangled symbol names into user readable names. Different compilers have
1260 different mangling styles. The optional demangling style argument can be used
1261 to choose an appropriate demangling style for your compiler. The linker will
1262 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1263 is set. These options may be used to override the default.
1265 @cindex dynamic linker, from command line
1266 @kindex -I@var{file}
1267 @kindex --dynamic-linker @var{file}
1268 @item --dynamic-linker @var{file}
1269 Set the name of the dynamic linker. This is only meaningful when
1270 generating dynamically linked ELF executables. The default dynamic
1271 linker is normally correct; don't use this unless you know what you are
1275 @kindex --fatal-warnings
1276 @item --fatal-warnings
1277 Treat all warnings as errors.
1279 @kindex --force-exe-suffix
1280 @item --force-exe-suffix
1281 Make sure that an output file has a .exe suffix.
1283 If a successfully built fully linked output file does not have a
1284 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1285 the output file to one of the same name with a @code{.exe} suffix. This
1286 option is useful when using unmodified Unix makefiles on a Microsoft
1287 Windows host, since some versions of Windows won't run an image unless
1288 it ends in a @code{.exe} suffix.
1290 @kindex --gc-sections
1291 @kindex --no-gc-sections
1292 @cindex garbage collection
1294 @itemx --no-gc-sections
1295 Enable garbage collection of unused input sections. It is ignored on
1296 targets that do not support this option. This option is not compatible
1297 with @samp{-r} or @samp{--emit-relocs}. The default behaviour (of not
1298 performing this garbage collection) can be restored by specifying
1299 @samp{--no-gc-sections} on the command line.
1301 @kindex --print-gc-sections
1302 @kindex --no-print-gc-sections
1303 @cindex garbage collection
1304 @item --print-gc-sections
1305 @itemx --no-print-gc-sections
1306 List all sections removed by garbage collection. The listing is
1307 printed on stderr. This option is only effective if garbage
1308 collection has been enabled via the @samp{--gc-sections}) option. The
1309 default behaviour (of not listing the sections that are removed) can
1310 be restored by specifying @samp{--no-print-gc-sections} on the command
1317 Print a summary of the command-line options on the standard output and exit.
1319 @kindex --target-help
1321 Print a summary of all target specific options on the standard output and exit.
1324 @item -Map @var{mapfile}
1325 Print a link map to the file @var{mapfile}. See the description of the
1326 @option{-M} option, above.
1328 @cindex memory usage
1329 @kindex --no-keep-memory
1330 @item --no-keep-memory
1331 @command{ld} normally optimizes for speed over memory usage by caching the
1332 symbol tables of input files in memory. This option tells @command{ld} to
1333 instead optimize for memory usage, by rereading the symbol tables as
1334 necessary. This may be required if @command{ld} runs out of memory space
1335 while linking a large executable.
1337 @kindex --no-undefined
1339 @item --no-undefined
1341 Report unresolved symbol references from regular object files. This
1342 is done even if the linker is creating a non-symbolic shared library.
1343 The switch @option{--[no-]allow-shlib-undefined} controls the
1344 behaviour for reporting unresolved references found in shared
1345 libraries being linked in.
1347 @kindex --allow-multiple-definition
1349 @item --allow-multiple-definition
1351 Normally when a symbol is defined multiple times, the linker will
1352 report a fatal error. These options allow multiple definitions and the
1353 first definition will be used.
1355 @kindex --allow-shlib-undefined
1356 @kindex --no-allow-shlib-undefined
1357 @item --allow-shlib-undefined
1358 @itemx --no-allow-shlib-undefined
1359 Allows (the default) or disallows undefined symbols in shared libraries.
1360 This switch is similar to @option{--no-undefined} except that it
1361 determines the behaviour when the undefined symbols are in a
1362 shared library rather than a regular object file. It does not affect
1363 how undefined symbols in regular object files are handled.
1365 The reason that @option{--allow-shlib-undefined} is the default is that
1366 the shared library being specified at link time may not be the same as
1367 the one that is available at load time, so the symbols might actually be
1368 resolvable at load time. Plus there are some systems, (eg BeOS) where
1369 undefined symbols in shared libraries is normal. (The kernel patches
1370 them at load time to select which function is most appropriate
1371 for the current architecture. This is used for example to dynamically
1372 select an appropriate memset function). Apparently it is also normal
1373 for HPPA shared libraries to have undefined symbols.
1375 @kindex --no-undefined-version
1376 @item --no-undefined-version
1377 Normally when a symbol has an undefined version, the linker will ignore
1378 it. This option disallows symbols with undefined version and a fatal error
1379 will be issued instead.
1381 @kindex --default-symver
1382 @item --default-symver
1383 Create and use a default symbol version (the soname) for unversioned
1386 @kindex --default-imported-symver
1387 @item --default-imported-symver
1388 Create and use a default symbol version (the soname) for unversioned
1391 @kindex --no-warn-mismatch
1392 @item --no-warn-mismatch
1393 Normally @command{ld} will give an error if you try to link together input
1394 files that are mismatched for some reason, perhaps because they have
1395 been compiled for different processors or for different endiannesses.
1396 This option tells @command{ld} that it should silently permit such possible
1397 errors. This option should only be used with care, in cases when you
1398 have taken some special action that ensures that the linker errors are
1401 @kindex --no-whole-archive
1402 @item --no-whole-archive
1403 Turn off the effect of the @option{--whole-archive} option for subsequent
1406 @cindex output file after errors
1407 @kindex --noinhibit-exec
1408 @item --noinhibit-exec
1409 Retain the executable output file whenever it is still usable.
1410 Normally, the linker will not produce an output file if it encounters
1411 errors during the link process; it exits without writing an output file
1412 when it issues any error whatsoever.
1416 Only search library directories explicitly specified on the
1417 command line. Library directories specified in linker scripts
1418 (including linker scripts specified on the command line) are ignored.
1420 @ifclear SingleFormat
1422 @item --oformat @var{output-format}
1423 @command{ld} may be configured to support more than one kind of object
1424 file. If your @command{ld} is configured this way, you can use the
1425 @samp{--oformat} option to specify the binary format for the output
1426 object file. Even when @command{ld} is configured to support alternative
1427 object formats, you don't usually need to specify this, as @command{ld}
1428 should be configured to produce as a default output format the most
1429 usual format on each machine. @var{output-format} is a text string, the
1430 name of a particular format supported by the BFD libraries. (You can
1431 list the available binary formats with @samp{objdump -i}.) The script
1432 command @code{OUTPUT_FORMAT} can also specify the output format, but
1433 this option overrides it. @xref{BFD}.
1437 @kindex --pic-executable
1439 @itemx --pic-executable
1440 @cindex position independent executables
1441 Create a position independent executable. This is currently only supported on
1442 ELF platforms. Position independent executables are similar to shared
1443 libraries in that they are relocated by the dynamic linker to the virtual
1444 address the OS chooses for them (which can vary between invocations). Like
1445 normal dynamically linked executables they can be executed and symbols
1446 defined in the executable cannot be overridden by shared libraries.
1450 This option is ignored for Linux compatibility.
1454 This option is ignored for SVR4 compatibility.
1457 @cindex synthesizing linker
1458 @cindex relaxing addressing modes
1460 An option with machine dependent effects.
1462 This option is only supported on a few targets.
1465 @xref{H8/300,,@command{ld} and the H8/300}.
1468 @xref{i960,, @command{ld} and the Intel 960 family}.
1471 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1474 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1477 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1480 On some platforms, the @samp{--relax} option performs global
1481 optimizations that become possible when the linker resolves addressing
1482 in the program, such as relaxing address modes and synthesizing new
1483 instructions in the output object file.
1485 On some platforms these link time global optimizations may make symbolic
1486 debugging of the resulting executable impossible.
1489 the case for the Matsushita MN10200 and MN10300 family of processors.
1493 On platforms where this is not supported, @samp{--relax} is accepted,
1497 @cindex retaining specified symbols
1498 @cindex stripping all but some symbols
1499 @cindex symbols, retaining selectively
1500 @item --retain-symbols-file @var{filename}
1501 Retain @emph{only} the symbols listed in the file @var{filename},
1502 discarding all others. @var{filename} is simply a flat file, with one
1503 symbol name per line. This option is especially useful in environments
1507 where a large global symbol table is accumulated gradually, to conserve
1510 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1511 or symbols needed for relocations.
1513 You may only specify @samp{--retain-symbols-file} once in the command
1514 line. It overrides @samp{-s} and @samp{-S}.
1517 @item -rpath @var{dir}
1518 @cindex runtime library search path
1520 Add a directory to the runtime library search path. This is used when
1521 linking an ELF executable with shared objects. All @option{-rpath}
1522 arguments are concatenated and passed to the runtime linker, which uses
1523 them to locate shared objects at runtime. The @option{-rpath} option is
1524 also used when locating shared objects which are needed by shared
1525 objects explicitly included in the link; see the description of the
1526 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1527 ELF executable, the contents of the environment variable
1528 @code{LD_RUN_PATH} will be used if it is defined.
1530 The @option{-rpath} option may also be used on SunOS. By default, on
1531 SunOS, the linker will form a runtime search patch out of all the
1532 @option{-L} options it is given. If a @option{-rpath} option is used, the
1533 runtime search path will be formed exclusively using the @option{-rpath}
1534 options, ignoring the @option{-L} options. This can be useful when using
1535 gcc, which adds many @option{-L} options which may be on NFS mounted
1538 For compatibility with other ELF linkers, if the @option{-R} option is
1539 followed by a directory name, rather than a file name, it is treated as
1540 the @option{-rpath} option.
1544 @cindex link-time runtime library search path
1546 @item -rpath-link @var{DIR}
1547 When using ELF or SunOS, one shared library may require another. This
1548 happens when an @code{ld -shared} link includes a shared library as one
1551 When the linker encounters such a dependency when doing a non-shared,
1552 non-relocatable link, it will automatically try to locate the required
1553 shared library and include it in the link, if it is not included
1554 explicitly. In such a case, the @option{-rpath-link} option
1555 specifies the first set of directories to search. The
1556 @option{-rpath-link} option may specify a sequence of directory names
1557 either by specifying a list of names separated by colons, or by
1558 appearing multiple times.
1560 This option should be used with caution as it overrides the search path
1561 that may have been hard compiled into a shared library. In such a case it
1562 is possible to use unintentionally a different search path than the
1563 runtime linker would do.
1565 The linker uses the following search paths to locate required shared
1569 Any directories specified by @option{-rpath-link} options.
1571 Any directories specified by @option{-rpath} options. The difference
1572 between @option{-rpath} and @option{-rpath-link} is that directories
1573 specified by @option{-rpath} options are included in the executable and
1574 used at runtime, whereas the @option{-rpath-link} option is only effective
1575 at link time. Searching @option{-rpath} in this way is only supported
1576 by native linkers and cross linkers which have been configured with
1577 the @option{--with-sysroot} option.
1579 On an ELF system, if the @option{-rpath} and @code{rpath-link} options
1580 were not used, search the contents of the environment variable
1581 @code{LD_RUN_PATH}. It is for the native linker only.
1583 On SunOS, if the @option{-rpath} option was not used, search any
1584 directories specified using @option{-L} options.
1586 For a native linker, the contents of the environment variable
1587 @code{LD_LIBRARY_PATH}.
1589 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1590 @code{DT_RPATH} of a shared library are searched for shared
1591 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1592 @code{DT_RUNPATH} entries exist.
1594 The default directories, normally @file{/lib} and @file{/usr/lib}.
1596 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1597 exists, the list of directories found in that file.
1600 If the required shared library is not found, the linker will issue a
1601 warning and continue with the link.
1608 @cindex shared libraries
1609 Create a shared library. This is currently only supported on ELF, XCOFF
1610 and SunOS platforms. On SunOS, the linker will automatically create a
1611 shared library if the @option{-e} option is not used and there are
1612 undefined symbols in the link.
1615 @kindex --sort-common
1616 This option tells @command{ld} to sort the common symbols by size when it
1617 places them in the appropriate output sections. First come all the one
1618 byte symbols, then all the two byte, then all the four byte, and then
1619 everything else. This is to prevent gaps between symbols due to
1620 alignment constraints.
1622 @kindex --sort-section name
1623 @item --sort-section name
1624 This option will apply @code{SORT_BY_NAME} to all wildcard section
1625 patterns in the linker script.
1627 @kindex --sort-section alignment
1628 @item --sort-section alignment
1629 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1630 patterns in the linker script.
1632 @kindex --split-by-file
1633 @item --split-by-file [@var{size}]
1634 Similar to @option{--split-by-reloc} but creates a new output section for
1635 each input file when @var{size} is reached. @var{size} defaults to a
1636 size of 1 if not given.
1638 @kindex --split-by-reloc
1639 @item --split-by-reloc [@var{count}]
1640 Tries to creates extra sections in the output file so that no single
1641 output section in the file contains more than @var{count} relocations.
1642 This is useful when generating huge relocatable files for downloading into
1643 certain real time kernels with the COFF object file format; since COFF
1644 cannot represent more than 65535 relocations in a single section. Note
1645 that this will fail to work with object file formats which do not
1646 support arbitrary sections. The linker will not split up individual
1647 input sections for redistribution, so if a single input section contains
1648 more than @var{count} relocations one output section will contain that
1649 many relocations. @var{count} defaults to a value of 32768.
1653 Compute and display statistics about the operation of the linker, such
1654 as execution time and memory usage.
1657 @item --sysroot=@var{directory}
1658 Use @var{directory} as the location of the sysroot, overriding the
1659 configure-time default. This option is only supported by linkers
1660 that were configured using @option{--with-sysroot}.
1662 @kindex --traditional-format
1663 @cindex traditional format
1664 @item --traditional-format
1665 For some targets, the output of @command{ld} is different in some ways from
1666 the output of some existing linker. This switch requests @command{ld} to
1667 use the traditional format instead.
1670 For example, on SunOS, @command{ld} combines duplicate entries in the
1671 symbol string table. This can reduce the size of an output file with
1672 full debugging information by over 30 percent. Unfortunately, the SunOS
1673 @code{dbx} program can not read the resulting program (@code{gdb} has no
1674 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1675 combine duplicate entries.
1677 @kindex --section-start @var{sectionname}=@var{org}
1678 @item --section-start @var{sectionname}=@var{org}
1679 Locate a section in the output file at the absolute
1680 address given by @var{org}. You may use this option as many
1681 times as necessary to locate multiple sections in the command
1683 @var{org} must be a single hexadecimal integer;
1684 for compatibility with other linkers, you may omit the leading
1685 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1686 should be no white space between @var{sectionname}, the equals
1687 sign (``@key{=}''), and @var{org}.
1689 @kindex -Tbss @var{org}
1690 @kindex -Tdata @var{org}
1691 @kindex -Ttext @var{org}
1692 @cindex segment origins, cmd line
1693 @item -Tbss @var{org}
1694 @itemx -Tdata @var{org}
1695 @itemx -Ttext @var{org}
1696 Same as --section-start, with @code{.bss}, @code{.data} or
1697 @code{.text} as the @var{sectionname}.
1699 @kindex --unresolved-symbols
1700 @item --unresolved-symbols=@var{method}
1701 Determine how to handle unresolved symbols. There are four possible
1702 values for @samp{method}:
1706 Do not report any unresolved symbols.
1709 Report all unresolved symbols. This is the default.
1711 @item ignore-in-object-files
1712 Report unresolved symbols that are contained in shared libraries, but
1713 ignore them if they come from regular object files.
1715 @item ignore-in-shared-libs
1716 Report unresolved symbols that come from regular object files, but
1717 ignore them if they come from shared libraries. This can be useful
1718 when creating a dynamic binary and it is known that all the shared
1719 libraries that it should be referencing are included on the linker's
1723 The behaviour for shared libraries on their own can also be controlled
1724 by the @option{--[no-]allow-shlib-undefined} option.
1726 Normally the linker will generate an error message for each reported
1727 unresolved symbol but the option @option{--warn-unresolved-symbols}
1728 can change this to a warning.
1734 Display the version number for @command{ld} and list the linker emulations
1735 supported. Display which input files can and cannot be opened. Display
1736 the linker script being used by the linker.
1738 @kindex --version-script=@var{version-scriptfile}
1739 @cindex version script, symbol versions
1740 @itemx --version-script=@var{version-scriptfile}
1741 Specify the name of a version script to the linker. This is typically
1742 used when creating shared libraries to specify additional information
1743 about the version hierarchy for the library being created. This option
1744 is only meaningful on ELF platforms which support shared libraries.
1747 @kindex --warn-common
1748 @cindex warnings, on combining symbols
1749 @cindex combining symbols, warnings on
1751 Warn when a common symbol is combined with another common symbol or with
1752 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1753 but linkers on some other operating systems do not. This option allows
1754 you to find potential problems from combining global symbols.
1755 Unfortunately, some C libraries use this practise, so you may get some
1756 warnings about symbols in the libraries as well as in your programs.
1758 There are three kinds of global symbols, illustrated here by C examples:
1762 A definition, which goes in the initialized data section of the output
1766 An undefined reference, which does not allocate space.
1767 There must be either a definition or a common symbol for the
1771 A common symbol. If there are only (one or more) common symbols for a
1772 variable, it goes in the uninitialized data area of the output file.
1773 The linker merges multiple common symbols for the same variable into a
1774 single symbol. If they are of different sizes, it picks the largest
1775 size. The linker turns a common symbol into a declaration, if there is
1776 a definition of the same variable.
1779 The @samp{--warn-common} option can produce five kinds of warnings.
1780 Each warning consists of a pair of lines: the first describes the symbol
1781 just encountered, and the second describes the previous symbol
1782 encountered with the same name. One or both of the two symbols will be
1787 Turning a common symbol into a reference, because there is already a
1788 definition for the symbol.
1790 @var{file}(@var{section}): warning: common of `@var{symbol}'
1791 overridden by definition
1792 @var{file}(@var{section}): warning: defined here
1796 Turning a common symbol into a reference, because a later definition for
1797 the symbol is encountered. This is the same as the previous case,
1798 except that the symbols are encountered in a different order.
1800 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1802 @var{file}(@var{section}): warning: common is here
1806 Merging a common symbol with a previous same-sized common symbol.
1808 @var{file}(@var{section}): warning: multiple common
1810 @var{file}(@var{section}): warning: previous common is here
1814 Merging a common symbol with a previous larger common symbol.
1816 @var{file}(@var{section}): warning: common of `@var{symbol}'
1817 overridden by larger common
1818 @var{file}(@var{section}): warning: larger common is here
1822 Merging a common symbol with a previous smaller common symbol. This is
1823 the same as the previous case, except that the symbols are
1824 encountered in a different order.
1826 @var{file}(@var{section}): warning: common of `@var{symbol}'
1827 overriding smaller common
1828 @var{file}(@var{section}): warning: smaller common is here
1832 @kindex --warn-constructors
1833 @item --warn-constructors
1834 Warn if any global constructors are used. This is only useful for a few
1835 object file formats. For formats like COFF or ELF, the linker can not
1836 detect the use of global constructors.
1838 @kindex --warn-multiple-gp
1839 @item --warn-multiple-gp
1840 Warn if multiple global pointer values are required in the output file.
1841 This is only meaningful for certain processors, such as the Alpha.
1842 Specifically, some processors put large-valued constants in a special
1843 section. A special register (the global pointer) points into the middle
1844 of this section, so that constants can be loaded efficiently via a
1845 base-register relative addressing mode. Since the offset in
1846 base-register relative mode is fixed and relatively small (e.g., 16
1847 bits), this limits the maximum size of the constant pool. Thus, in
1848 large programs, it is often necessary to use multiple global pointer
1849 values in order to be able to address all possible constants. This
1850 option causes a warning to be issued whenever this case occurs.
1853 @cindex warnings, on undefined symbols
1854 @cindex undefined symbols, warnings on
1856 Only warn once for each undefined symbol, rather than once per module
1859 @kindex --warn-section-align
1860 @cindex warnings, on section alignment
1861 @cindex section alignment, warnings on
1862 @item --warn-section-align
1863 Warn if the address of an output section is changed because of
1864 alignment. Typically, the alignment will be set by an input section.
1865 The address will only be changed if it not explicitly specified; that
1866 is, if the @code{SECTIONS} command does not specify a start address for
1867 the section (@pxref{SECTIONS}).
1869 @kindex --warn-shared-textrel
1870 @item --warn-shared-textrel
1871 Warn if the linker adds a DT_TEXTREL to a shared object.
1873 @kindex --warn-unresolved-symbols
1874 @item --warn-unresolved-symbols
1875 If the linker is going to report an unresolved symbol (see the option
1876 @option{--unresolved-symbols}) it will normally generate an error.
1877 This option makes it generate a warning instead.
1879 @kindex --error-unresolved-symbols
1880 @item --error-unresolved-symbols
1881 This restores the linker's default behaviour of generating errors when
1882 it is reporting unresolved symbols.
1884 @kindex --whole-archive
1885 @cindex including an entire archive
1886 @item --whole-archive
1887 For each archive mentioned on the command line after the
1888 @option{--whole-archive} option, include every object file in the archive
1889 in the link, rather than searching the archive for the required object
1890 files. This is normally used to turn an archive file into a shared
1891 library, forcing every object to be included in the resulting shared
1892 library. This option may be used more than once.
1894 Two notes when using this option from gcc: First, gcc doesn't know
1895 about this option, so you have to use @option{-Wl,-whole-archive}.
1896 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1897 list of archives, because gcc will add its own list of archives to
1898 your link and you may not want this flag to affect those as well.
1901 @item --wrap @var{symbol}
1902 Use a wrapper function for @var{symbol}. Any undefined reference to
1903 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1904 undefined reference to @code{__real_@var{symbol}} will be resolved to
1907 This can be used to provide a wrapper for a system function. The
1908 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1909 wishes to call the system function, it should call
1910 @code{__real_@var{symbol}}.
1912 Here is a trivial example:
1916 __wrap_malloc (size_t c)
1918 printf ("malloc called with %zu\n", c);
1919 return __real_malloc (c);
1923 If you link other code with this file using @option{--wrap malloc}, then
1924 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1925 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1926 call the real @code{malloc} function.
1928 You may wish to provide a @code{__real_malloc} function as well, so that
1929 links without the @option{--wrap} option will succeed. If you do this,
1930 you should not put the definition of @code{__real_malloc} in the same
1931 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
1932 call before the linker has a chance to wrap it to @code{malloc}.
1934 @kindex --eh-frame-hdr
1935 @item --eh-frame-hdr
1936 Request creation of @code{.eh_frame_hdr} section and ELF
1937 @code{PT_GNU_EH_FRAME} segment header.
1939 @kindex --enable-new-dtags
1940 @kindex --disable-new-dtags
1941 @item --enable-new-dtags
1942 @itemx --disable-new-dtags
1943 This linker can create the new dynamic tags in ELF. But the older ELF
1944 systems may not understand them. If you specify
1945 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
1946 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
1947 created. By default, the new dynamic tags are not created. Note that
1948 those options are only available for ELF systems.
1950 @kindex --hash-size=@var{number}
1951 @item --hash-size=@var{number}
1952 Set the default size of the linker's hash tables to a prime number
1953 close to @var{number}. Increasing this value can reduce the length of
1954 time it takes the linker to perform its tasks, at the expense of
1955 increasing the linker's memory requirements. Similarly reducing this
1956 value can reduce the memory requirements at the expense of speed.
1958 @kindex --hash-style=@var{style}
1959 @item --hash-style=@var{style}
1960 Set the type of linker's hash table(s). @var{style} can be either
1961 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
1962 new style GNU @code{.gnu.hash} section or @code{both} for both
1963 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
1964 hash tables. The default is @code{sysv}.
1966 @kindex --reduce-memory-overheads
1967 @item --reduce-memory-overheads
1968 This option reduces memory requirements at ld runtime, at the expense of
1969 linking speed. This was introduced to select the old O(n^2) algorithm
1970 for link map file generation, rather than the new O(n) algorithm which uses
1971 about 40% more memory for symbol storage.
1973 Another effect of the switch is to set the default hash table size to
1974 1021, which again saves memory at the cost of lengthening the linker's
1975 run time. This is not done however if the @option{--hash-size} switch
1978 The @option{--reduce-memory-overheads} switch may be also be used to
1979 enable other tradeoffs in future versions of the linker.
1985 @subsection Options Specific to i386 PE Targets
1987 @c man begin OPTIONS
1989 The i386 PE linker supports the @option{-shared} option, which causes
1990 the output to be a dynamically linked library (DLL) instead of a
1991 normal executable. You should name the output @code{*.dll} when you
1992 use this option. In addition, the linker fully supports the standard
1993 @code{*.def} files, which may be specified on the linker command line
1994 like an object file (in fact, it should precede archives it exports
1995 symbols from, to ensure that they get linked in, just like a normal
1998 In addition to the options common to all targets, the i386 PE linker
1999 support additional command line options that are specific to the i386
2000 PE target. Options that take values may be separated from their
2001 values by either a space or an equals sign.
2005 @kindex --add-stdcall-alias
2006 @item --add-stdcall-alias
2007 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2008 as-is and also with the suffix stripped.
2009 [This option is specific to the i386 PE targeted port of the linker]
2012 @item --base-file @var{file}
2013 Use @var{file} as the name of a file in which to save the base
2014 addresses of all the relocations needed for generating DLLs with
2016 [This is an i386 PE specific option]
2020 Create a DLL instead of a regular executable. You may also use
2021 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2023 [This option is specific to the i386 PE targeted port of the linker]
2025 @kindex --enable-stdcall-fixup
2026 @kindex --disable-stdcall-fixup
2027 @item --enable-stdcall-fixup
2028 @itemx --disable-stdcall-fixup
2029 If the link finds a symbol that it cannot resolve, it will attempt to
2030 do ``fuzzy linking'' by looking for another defined symbol that differs
2031 only in the format of the symbol name (cdecl vs stdcall) and will
2032 resolve that symbol by linking to the match. For example, the
2033 undefined symbol @code{_foo} might be linked to the function
2034 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2035 to the function @code{_bar}. When the linker does this, it prints a
2036 warning, since it normally should have failed to link, but sometimes
2037 import libraries generated from third-party dlls may need this feature
2038 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2039 feature is fully enabled and warnings are not printed. If you specify
2040 @option{--disable-stdcall-fixup}, this feature is disabled and such
2041 mismatches are considered to be errors.
2042 [This option is specific to the i386 PE targeted port of the linker]
2044 @cindex DLLs, creating
2045 @kindex --export-all-symbols
2046 @item --export-all-symbols
2047 If given, all global symbols in the objects used to build a DLL will
2048 be exported by the DLL. Note that this is the default if there
2049 otherwise wouldn't be any exported symbols. When symbols are
2050 explicitly exported via DEF files or implicitly exported via function
2051 attributes, the default is to not export anything else unless this
2052 option is given. Note that the symbols @code{DllMain@@12},
2053 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2054 @code{impure_ptr} will not be automatically
2055 exported. Also, symbols imported from other DLLs will not be
2056 re-exported, nor will symbols specifying the DLL's internal layout
2057 such as those beginning with @code{_head_} or ending with
2058 @code{_iname}. In addition, no symbols from @code{libgcc},
2059 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2060 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2061 not be exported, to help with C++ DLLs. Finally, there is an
2062 extensive list of cygwin-private symbols that are not exported
2063 (obviously, this applies on when building DLLs for cygwin targets).
2064 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2065 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2066 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2067 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2068 @code{cygwin_premain3}, and @code{environ}.
2069 [This option is specific to the i386 PE targeted port of the linker]
2071 @kindex --exclude-symbols
2072 @item --exclude-symbols @var{symbol},@var{symbol},...
2073 Specifies a list of symbols which should not be automatically
2074 exported. The symbol names may be delimited by commas or colons.
2075 [This option is specific to the i386 PE targeted port of the linker]
2077 @kindex --file-alignment
2078 @item --file-alignment
2079 Specify the file alignment. Sections in the file will always begin at
2080 file offsets which are multiples of this number. This defaults to
2082 [This option is specific to the i386 PE targeted port of the linker]
2086 @item --heap @var{reserve}
2087 @itemx --heap @var{reserve},@var{commit}
2088 Specify the amount of memory to reserve (and optionally commit) to be
2089 used as heap for this program. The default is 1Mb reserved, 4K
2091 [This option is specific to the i386 PE targeted port of the linker]
2094 @kindex --image-base
2095 @item --image-base @var{value}
2096 Use @var{value} as the base address of your program or dll. This is
2097 the lowest memory location that will be used when your program or dll
2098 is loaded. To reduce the need to relocate and improve performance of
2099 your dlls, each should have a unique base address and not overlap any
2100 other dlls. The default is 0x400000 for executables, and 0x10000000
2102 [This option is specific to the i386 PE targeted port of the linker]
2106 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2107 symbols before they are exported.
2108 [This option is specific to the i386 PE targeted port of the linker]
2110 @kindex --large-address-aware
2111 @item --large-address-aware
2112 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2113 header is set to indicate that this executable supports virtual addresses
2114 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2115 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2116 section of the BOOT.INI. Otherwise, this bit has no effect.
2117 [This option is specific to PE targeted ports of the linker]
2119 @kindex --major-image-version
2120 @item --major-image-version @var{value}
2121 Sets the major number of the ``image version''. Defaults to 1.
2122 [This option is specific to the i386 PE targeted port of the linker]
2124 @kindex --major-os-version
2125 @item --major-os-version @var{value}
2126 Sets the major number of the ``os version''. Defaults to 4.
2127 [This option is specific to the i386 PE targeted port of the linker]
2129 @kindex --major-subsystem-version
2130 @item --major-subsystem-version @var{value}
2131 Sets the major number of the ``subsystem version''. Defaults to 4.
2132 [This option is specific to the i386 PE targeted port of the linker]
2134 @kindex --minor-image-version
2135 @item --minor-image-version @var{value}
2136 Sets the minor number of the ``image version''. Defaults to 0.
2137 [This option is specific to the i386 PE targeted port of the linker]
2139 @kindex --minor-os-version
2140 @item --minor-os-version @var{value}
2141 Sets the minor number of the ``os version''. Defaults to 0.
2142 [This option is specific to the i386 PE targeted port of the linker]
2144 @kindex --minor-subsystem-version
2145 @item --minor-subsystem-version @var{value}
2146 Sets the minor number of the ``subsystem version''. Defaults to 0.
2147 [This option is specific to the i386 PE targeted port of the linker]
2149 @cindex DEF files, creating
2150 @cindex DLLs, creating
2151 @kindex --output-def
2152 @item --output-def @var{file}
2153 The linker will create the file @var{file} which will contain a DEF
2154 file corresponding to the DLL the linker is generating. This DEF file
2155 (which should be called @code{*.def}) may be used to create an import
2156 library with @code{dlltool} or may be used as a reference to
2157 automatically or implicitly exported symbols.
2158 [This option is specific to the i386 PE targeted port of the linker]
2160 @cindex DLLs, creating
2161 @kindex --out-implib
2162 @item --out-implib @var{file}
2163 The linker will create the file @var{file} which will contain an
2164 import lib corresponding to the DLL the linker is generating. This
2165 import lib (which should be called @code{*.dll.a} or @code{*.a}
2166 may be used to link clients against the generated DLL; this behaviour
2167 makes it possible to skip a separate @code{dlltool} import library
2169 [This option is specific to the i386 PE targeted port of the linker]
2171 @kindex --enable-auto-image-base
2172 @item --enable-auto-image-base
2173 Automatically choose the image base for DLLs, unless one is specified
2174 using the @code{--image-base} argument. By using a hash generated
2175 from the dllname to create unique image bases for each DLL, in-memory
2176 collisions and relocations which can delay program execution are
2178 [This option is specific to the i386 PE targeted port of the linker]
2180 @kindex --disable-auto-image-base
2181 @item --disable-auto-image-base
2182 Do not automatically generate a unique image base. If there is no
2183 user-specified image base (@code{--image-base}) then use the platform
2185 [This option is specific to the i386 PE targeted port of the linker]
2187 @cindex DLLs, linking to
2188 @kindex --dll-search-prefix
2189 @item --dll-search-prefix @var{string}
2190 When linking dynamically to a dll without an import library,
2191 search for @code{<string><basename>.dll} in preference to
2192 @code{lib<basename>.dll}. This behaviour allows easy distinction
2193 between DLLs built for the various "subplatforms": native, cygwin,
2194 uwin, pw, etc. For instance, cygwin DLLs typically use
2195 @code{--dll-search-prefix=cyg}.
2196 [This option is specific to the i386 PE targeted port of the linker]
2198 @kindex --enable-auto-import
2199 @item --enable-auto-import
2200 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2201 DATA imports from DLLs, and create the necessary thunking symbols when
2202 building the import libraries with those DATA exports. Note: Use of the
2203 'auto-import' extension will cause the text section of the image file
2204 to be made writable. This does not conform to the PE-COFF format
2205 specification published by Microsoft.
2207 Using 'auto-import' generally will 'just work' -- but sometimes you may
2210 "variable '<var>' can't be auto-imported. Please read the
2211 documentation for ld's @code{--enable-auto-import} for details."
2213 This message occurs when some (sub)expression accesses an address
2214 ultimately given by the sum of two constants (Win32 import tables only
2215 allow one). Instances where this may occur include accesses to member
2216 fields of struct variables imported from a DLL, as well as using a
2217 constant index into an array variable imported from a DLL. Any
2218 multiword variable (arrays, structs, long long, etc) may trigger
2219 this error condition. However, regardless of the exact data type
2220 of the offending exported variable, ld will always detect it, issue
2221 the warning, and exit.
2223 There are several ways to address this difficulty, regardless of the
2224 data type of the exported variable:
2226 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2227 of adjusting references in your client code for runtime environment, so
2228 this method works only when runtime environment supports this feature.
2230 A second solution is to force one of the 'constants' to be a variable --
2231 that is, unknown and un-optimizable at compile time. For arrays,
2232 there are two possibilities: a) make the indexee (the array's address)
2233 a variable, or b) make the 'constant' index a variable. Thus:
2236 extern type extern_array[];
2238 @{ volatile type *t=extern_array; t[1] @}
2244 extern type extern_array[];
2246 @{ volatile int t=1; extern_array[t] @}
2249 For structs (and most other multiword data types) the only option
2250 is to make the struct itself (or the long long, or the ...) variable:
2253 extern struct s extern_struct;
2254 extern_struct.field -->
2255 @{ volatile struct s *t=&extern_struct; t->field @}
2261 extern long long extern_ll;
2263 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2266 A third method of dealing with this difficulty is to abandon
2267 'auto-import' for the offending symbol and mark it with
2268 @code{__declspec(dllimport)}. However, in practise that
2269 requires using compile-time #defines to indicate whether you are
2270 building a DLL, building client code that will link to the DLL, or
2271 merely building/linking to a static library. In making the choice
2272 between the various methods of resolving the 'direct address with
2273 constant offset' problem, you should consider typical real-world usage:
2281 void main(int argc, char **argv)@{
2282 printf("%d\n",arr[1]);
2292 void main(int argc, char **argv)@{
2293 /* This workaround is for win32 and cygwin; do not "optimize" */
2294 volatile int *parr = arr;
2295 printf("%d\n",parr[1]);
2302 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2303 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2304 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2305 #define FOO_IMPORT __declspec(dllimport)
2309 extern FOO_IMPORT int arr[];
2312 void main(int argc, char **argv)@{
2313 printf("%d\n",arr[1]);
2317 A fourth way to avoid this problem is to re-code your
2318 library to use a functional interface rather than a data interface
2319 for the offending variables (e.g. set_foo() and get_foo() accessor
2321 [This option is specific to the i386 PE targeted port of the linker]
2323 @kindex --disable-auto-import
2324 @item --disable-auto-import
2325 Do not attempt to do sophisticated linking of @code{_symbol} to
2326 @code{__imp__symbol} for DATA imports from DLLs.
2327 [This option is specific to the i386 PE targeted port of the linker]
2329 @kindex --enable-runtime-pseudo-reloc
2330 @item --enable-runtime-pseudo-reloc
2331 If your code contains expressions described in --enable-auto-import section,
2332 that is, DATA imports from DLL with non-zero offset, this switch will create
2333 a vector of 'runtime pseudo relocations' which can be used by runtime
2334 environment to adjust references to such data in your client code.
2335 [This option is specific to the i386 PE targeted port of the linker]
2337 @kindex --disable-runtime-pseudo-reloc
2338 @item --disable-runtime-pseudo-reloc
2339 Do not create pseudo relocations for non-zero offset DATA imports from
2340 DLLs. This is the default.
2341 [This option is specific to the i386 PE targeted port of the linker]
2343 @kindex --enable-extra-pe-debug
2344 @item --enable-extra-pe-debug
2345 Show additional debug info related to auto-import symbol thunking.
2346 [This option is specific to the i386 PE targeted port of the linker]
2348 @kindex --section-alignment
2349 @item --section-alignment
2350 Sets the section alignment. Sections in memory will always begin at
2351 addresses which are a multiple of this number. Defaults to 0x1000.
2352 [This option is specific to the i386 PE targeted port of the linker]
2356 @item --stack @var{reserve}
2357 @itemx --stack @var{reserve},@var{commit}
2358 Specify the amount of memory to reserve (and optionally commit) to be
2359 used as stack for this program. The default is 2Mb reserved, 4K
2361 [This option is specific to the i386 PE targeted port of the linker]
2364 @item --subsystem @var{which}
2365 @itemx --subsystem @var{which}:@var{major}
2366 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2367 Specifies the subsystem under which your program will execute. The
2368 legal values for @var{which} are @code{native}, @code{windows},
2369 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2370 the subsystem version also. Numeric values are also accepted for
2372 [This option is specific to the i386 PE targeted port of the linker]
2379 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2381 @c man begin OPTIONS
2383 The 68HC11 and 68HC12 linkers support specific options to control the
2384 memory bank switching mapping and trampoline code generation.
2388 @kindex --no-trampoline
2389 @item --no-trampoline
2390 This option disables the generation of trampoline. By default a trampoline
2391 is generated for each far function which is called using a @code{jsr}
2392 instruction (this happens when a pointer to a far function is taken).
2394 @kindex --bank-window
2395 @item --bank-window @var{name}
2396 This option indicates to the linker the name of the memory region in
2397 the @samp{MEMORY} specification that describes the memory bank window.
2398 The definition of such region is then used by the linker to compute
2399 paging and addresses within the memory window.
2408 @section Environment Variables
2410 @c man begin ENVIRONMENT
2412 You can change the behaviour of @command{ld} with the environment variables
2413 @ifclear SingleFormat
2416 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2418 @ifclear SingleFormat
2420 @cindex default input format
2421 @code{GNUTARGET} determines the input-file object format if you don't
2422 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2423 of the BFD names for an input format (@pxref{BFD}). If there is no
2424 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2425 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2426 attempts to discover the input format by examining binary input files;
2427 this method often succeeds, but there are potential ambiguities, since
2428 there is no method of ensuring that the magic number used to specify
2429 object-file formats is unique. However, the configuration procedure for
2430 BFD on each system places the conventional format for that system first
2431 in the search-list, so ambiguities are resolved in favor of convention.
2435 @cindex default emulation
2436 @cindex emulation, default
2437 @code{LDEMULATION} determines the default emulation if you don't use the
2438 @samp{-m} option. The emulation can affect various aspects of linker
2439 behaviour, particularly the default linker script. You can list the
2440 available emulations with the @samp{--verbose} or @samp{-V} options. If
2441 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2442 variable is not defined, the default emulation depends upon how the
2443 linker was configured.
2445 @kindex COLLECT_NO_DEMANGLE
2446 @cindex demangling, default
2447 Normally, the linker will default to demangling symbols. However, if
2448 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2449 default to not demangling symbols. This environment variable is used in
2450 a similar fashion by the @code{gcc} linker wrapper program. The default
2451 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2458 @chapter Linker Scripts
2461 @cindex linker scripts
2462 @cindex command files
2463 Every link is controlled by a @dfn{linker script}. This script is
2464 written in the linker command language.
2466 The main purpose of the linker script is to describe how the sections in
2467 the input files should be mapped into the output file, and to control
2468 the memory layout of the output file. Most linker scripts do nothing
2469 more than this. However, when necessary, the linker script can also
2470 direct the linker to perform many other operations, using the commands
2473 The linker always uses a linker script. If you do not supply one
2474 yourself, the linker will use a default script that is compiled into the
2475 linker executable. You can use the @samp{--verbose} command line option
2476 to display the default linker script. Certain command line options,
2477 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2479 You may supply your own linker script by using the @samp{-T} command
2480 line option. When you do this, your linker script will replace the
2481 default linker script.
2483 You may also use linker scripts implicitly by naming them as input files
2484 to the linker, as though they were files to be linked. @xref{Implicit
2488 * Basic Script Concepts:: Basic Linker Script Concepts
2489 * Script Format:: Linker Script Format
2490 * Simple Example:: Simple Linker Script Example
2491 * Simple Commands:: Simple Linker Script Commands
2492 * Assignments:: Assigning Values to Symbols
2493 * SECTIONS:: SECTIONS Command
2494 * MEMORY:: MEMORY Command
2495 * PHDRS:: PHDRS Command
2496 * VERSION:: VERSION Command
2497 * Expressions:: Expressions in Linker Scripts
2498 * Implicit Linker Scripts:: Implicit Linker Scripts
2501 @node Basic Script Concepts
2502 @section Basic Linker Script Concepts
2503 @cindex linker script concepts
2504 We need to define some basic concepts and vocabulary in order to
2505 describe the linker script language.
2507 The linker combines input files into a single output file. The output
2508 file and each input file are in a special data format known as an
2509 @dfn{object file format}. Each file is called an @dfn{object file}.
2510 The output file is often called an @dfn{executable}, but for our
2511 purposes we will also call it an object file. Each object file has,
2512 among other things, a list of @dfn{sections}. We sometimes refer to a
2513 section in an input file as an @dfn{input section}; similarly, a section
2514 in the output file is an @dfn{output section}.
2516 Each section in an object file has a name and a size. Most sections
2517 also have an associated block of data, known as the @dfn{section
2518 contents}. A section may be marked as @dfn{loadable}, which mean that
2519 the contents should be loaded into memory when the output file is run.
2520 A section with no contents may be @dfn{allocatable}, which means that an
2521 area in memory should be set aside, but nothing in particular should be
2522 loaded there (in some cases this memory must be zeroed out). A section
2523 which is neither loadable nor allocatable typically contains some sort
2524 of debugging information.
2526 Every loadable or allocatable output section has two addresses. The
2527 first is the @dfn{VMA}, or virtual memory address. This is the address
2528 the section will have when the output file is run. The second is the
2529 @dfn{LMA}, or load memory address. This is the address at which the
2530 section will be loaded. In most cases the two addresses will be the
2531 same. An example of when they might be different is when a data section
2532 is loaded into ROM, and then copied into RAM when the program starts up
2533 (this technique is often used to initialize global variables in a ROM
2534 based system). In this case the ROM address would be the LMA, and the
2535 RAM address would be the VMA.
2537 You can see the sections in an object file by using the @code{objdump}
2538 program with the @samp{-h} option.
2540 Every object file also has a list of @dfn{symbols}, known as the
2541 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2542 has a name, and each defined symbol has an address, among other
2543 information. If you compile a C or C++ program into an object file, you
2544 will get a defined symbol for every defined function and global or
2545 static variable. Every undefined function or global variable which is
2546 referenced in the input file will become an undefined symbol.
2548 You can see the symbols in an object file by using the @code{nm}
2549 program, or by using the @code{objdump} program with the @samp{-t}
2553 @section Linker Script Format
2554 @cindex linker script format
2555 Linker scripts are text files.
2557 You write a linker script as a series of commands. Each command is
2558 either a keyword, possibly followed by arguments, or an assignment to a
2559 symbol. You may separate commands using semicolons. Whitespace is
2562 Strings such as file or format names can normally be entered directly.
2563 If the file name contains a character such as a comma which would
2564 otherwise serve to separate file names, you may put the file name in
2565 double quotes. There is no way to use a double quote character in a
2568 You may include comments in linker scripts just as in C, delimited by
2569 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2572 @node Simple Example
2573 @section Simple Linker Script Example
2574 @cindex linker script example
2575 @cindex example of linker script
2576 Many linker scripts are fairly simple.
2578 The simplest possible linker script has just one command:
2579 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2580 memory layout of the output file.
2582 The @samp{SECTIONS} command is a powerful command. Here we will
2583 describe a simple use of it. Let's assume your program consists only of
2584 code, initialized data, and uninitialized data. These will be in the
2585 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2586 Let's assume further that these are the only sections which appear in
2589 For this example, let's say that the code should be loaded at address
2590 0x10000, and that the data should start at address 0x8000000. Here is a
2591 linker script which will do that:
2596 .text : @{ *(.text) @}
2598 .data : @{ *(.data) @}
2599 .bss : @{ *(.bss) @}
2603 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2604 followed by a series of symbol assignments and output section
2605 descriptions enclosed in curly braces.
2607 The first line inside the @samp{SECTIONS} command of the above example
2608 sets the value of the special symbol @samp{.}, which is the location
2609 counter. If you do not specify the address of an output section in some
2610 other way (other ways are described later), the address is set from the
2611 current value of the location counter. The location counter is then
2612 incremented by the size of the output section. At the start of the
2613 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2615 The second line defines an output section, @samp{.text}. The colon is
2616 required syntax which may be ignored for now. Within the curly braces
2617 after the output section name, you list the names of the input sections
2618 which should be placed into this output section. The @samp{*} is a
2619 wildcard which matches any file name. The expression @samp{*(.text)}
2620 means all @samp{.text} input sections in all input files.
2622 Since the location counter is @samp{0x10000} when the output section
2623 @samp{.text} is defined, the linker will set the address of the
2624 @samp{.text} section in the output file to be @samp{0x10000}.
2626 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2627 the output file. The linker will place the @samp{.data} output section
2628 at address @samp{0x8000000}. After the linker places the @samp{.data}
2629 output section, the value of the location counter will be
2630 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2631 effect is that the linker will place the @samp{.bss} output section
2632 immediately after the @samp{.data} output section in memory.
2634 The linker will ensure that each output section has the required
2635 alignment, by increasing the location counter if necessary. In this
2636 example, the specified addresses for the @samp{.text} and @samp{.data}
2637 sections will probably satisfy any alignment constraints, but the linker
2638 may have to create a small gap between the @samp{.data} and @samp{.bss}
2641 That's it! That's a simple and complete linker script.
2643 @node Simple Commands
2644 @section Simple Linker Script Commands
2645 @cindex linker script simple commands
2646 In this section we describe the simple linker script commands.
2649 * Entry Point:: Setting the entry point
2650 * File Commands:: Commands dealing with files
2651 @ifclear SingleFormat
2652 * Format Commands:: Commands dealing with object file formats
2655 * Miscellaneous Commands:: Other linker script commands
2659 @subsection Setting the Entry Point
2660 @kindex ENTRY(@var{symbol})
2661 @cindex start of execution
2662 @cindex first instruction
2664 The first instruction to execute in a program is called the @dfn{entry
2665 point}. You can use the @code{ENTRY} linker script command to set the
2666 entry point. The argument is a symbol name:
2671 There are several ways to set the entry point. The linker will set the
2672 entry point by trying each of the following methods in order, and
2673 stopping when one of them succeeds:
2676 the @samp{-e} @var{entry} command-line option;
2678 the @code{ENTRY(@var{symbol})} command in a linker script;
2680 the value of the symbol @code{start}, if defined;
2682 the address of the first byte of the @samp{.text} section, if present;
2684 The address @code{0}.
2688 @subsection Commands Dealing with Files
2689 @cindex linker script file commands
2690 Several linker script commands deal with files.
2693 @item INCLUDE @var{filename}
2694 @kindex INCLUDE @var{filename}
2695 @cindex including a linker script
2696 Include the linker script @var{filename} at this point. The file will
2697 be searched for in the current directory, and in any directory specified
2698 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2701 @item INPUT(@var{file}, @var{file}, @dots{})
2702 @itemx INPUT(@var{file} @var{file} @dots{})
2703 @kindex INPUT(@var{files})
2704 @cindex input files in linker scripts
2705 @cindex input object files in linker scripts
2706 @cindex linker script input object files
2707 The @code{INPUT} command directs the linker to include the named files
2708 in the link, as though they were named on the command line.
2710 For example, if you always want to include @file{subr.o} any time you do
2711 a link, but you can't be bothered to put it on every link command line,
2712 then you can put @samp{INPUT (subr.o)} in your linker script.
2714 In fact, if you like, you can list all of your input files in the linker
2715 script, and then invoke the linker with nothing but a @samp{-T} option.
2717 In case a @dfn{sysroot prefix} is configured, and the filename starts
2718 with the @samp{/} character, and the script being processed was
2719 located inside the @dfn{sysroot prefix}, the filename will be looked
2720 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2721 open the file in the current directory. If it is not found, the
2722 linker will search through the archive library search path. See the
2723 description of @samp{-L} in @ref{Options,,Command Line Options}.
2725 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2726 name to @code{lib@var{file}.a}, as with the command line argument
2729 When you use the @code{INPUT} command in an implicit linker script, the
2730 files will be included in the link at the point at which the linker
2731 script file is included. This can affect archive searching.
2733 @item GROUP(@var{file}, @var{file}, @dots{})
2734 @itemx GROUP(@var{file} @var{file} @dots{})
2735 @kindex GROUP(@var{files})
2736 @cindex grouping input files
2737 The @code{GROUP} command is like @code{INPUT}, except that the named
2738 files should all be archives, and they are searched repeatedly until no
2739 new undefined references are created. See the description of @samp{-(}
2740 in @ref{Options,,Command Line Options}.
2742 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
2743 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
2744 @kindex AS_NEEDED(@var{files})
2745 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2746 commands, among other filenames. The files listed will be handled
2747 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2748 with the exception of ELF shared libraries, that will be added only
2749 when they are actually needed. This construct essentially enables
2750 @option{--as-needed} option for all the files listed inside of it
2751 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2754 @item OUTPUT(@var{filename})
2755 @kindex OUTPUT(@var{filename})
2756 @cindex output file name in linker script
2757 The @code{OUTPUT} command names the output file. Using
2758 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2759 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2760 Line Options}). If both are used, the command line option takes
2763 You can use the @code{OUTPUT} command to define a default name for the
2764 output file other than the usual default of @file{a.out}.
2766 @item SEARCH_DIR(@var{path})
2767 @kindex SEARCH_DIR(@var{path})
2768 @cindex library search path in linker script
2769 @cindex archive search path in linker script
2770 @cindex search path in linker script
2771 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2772 @command{ld} looks for archive libraries. Using
2773 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2774 on the command line (@pxref{Options,,Command Line Options}). If both
2775 are used, then the linker will search both paths. Paths specified using
2776 the command line option are searched first.
2778 @item STARTUP(@var{filename})
2779 @kindex STARTUP(@var{filename})
2780 @cindex first input file
2781 The @code{STARTUP} command is just like the @code{INPUT} command, except
2782 that @var{filename} will become the first input file to be linked, as
2783 though it were specified first on the command line. This may be useful
2784 when using a system in which the entry point is always the start of the
2788 @ifclear SingleFormat
2789 @node Format Commands
2790 @subsection Commands Dealing with Object File Formats
2791 A couple of linker script commands deal with object file formats.
2794 @item OUTPUT_FORMAT(@var{bfdname})
2795 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2796 @kindex OUTPUT_FORMAT(@var{bfdname})
2797 @cindex output file format in linker script
2798 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2799 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2800 exactly like using @samp{--oformat @var{bfdname}} on the command line
2801 (@pxref{Options,,Command Line Options}). If both are used, the command
2802 line option takes precedence.
2804 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2805 formats based on the @samp{-EB} and @samp{-EL} command line options.
2806 This permits the linker script to set the output format based on the
2809 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2810 will be the first argument, @var{default}. If @samp{-EB} is used, the
2811 output format will be the second argument, @var{big}. If @samp{-EL} is
2812 used, the output format will be the third argument, @var{little}.
2814 For example, the default linker script for the MIPS ELF target uses this
2817 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2819 This says that the default format for the output file is
2820 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
2821 option, the output file will be created in the @samp{elf32-littlemips}
2824 @item TARGET(@var{bfdname})
2825 @kindex TARGET(@var{bfdname})
2826 @cindex input file format in linker script
2827 The @code{TARGET} command names the BFD format to use when reading input
2828 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
2829 This command is like using @samp{-b @var{bfdname}} on the command line
2830 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
2831 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
2832 command is also used to set the format for the output file. @xref{BFD}.
2836 @node Miscellaneous Commands
2837 @subsection Other Linker Script Commands
2838 There are a few other linker scripts commands.
2841 @item ASSERT(@var{exp}, @var{message})
2843 @cindex assertion in linker script
2844 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
2845 with an error code, and print @var{message}.
2847 @item EXTERN(@var{symbol} @var{symbol} @dots{})
2849 @cindex undefined symbol in linker script
2850 Force @var{symbol} to be entered in the output file as an undefined
2851 symbol. Doing this may, for example, trigger linking of additional
2852 modules from standard libraries. You may list several @var{symbol}s for
2853 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
2854 command has the same effect as the @samp{-u} command-line option.
2856 @item FORCE_COMMON_ALLOCATION
2857 @kindex FORCE_COMMON_ALLOCATION
2858 @cindex common allocation in linker script
2859 This command has the same effect as the @samp{-d} command-line option:
2860 to make @command{ld} assign space to common symbols even if a relocatable
2861 output file is specified (@samp{-r}).
2863 @item INHIBIT_COMMON_ALLOCATION
2864 @kindex INHIBIT_COMMON_ALLOCATION
2865 @cindex common allocation in linker script
2866 This command has the same effect as the @samp{--no-define-common}
2867 command-line option: to make @code{ld} omit the assignment of addresses
2868 to common symbols even for a non-relocatable output file.
2870 @item NOCROSSREFS(@var{section} @var{section} @dots{})
2871 @kindex NOCROSSREFS(@var{sections})
2872 @cindex cross references
2873 This command may be used to tell @command{ld} to issue an error about any
2874 references among certain output sections.
2876 In certain types of programs, particularly on embedded systems when
2877 using overlays, when one section is loaded into memory, another section
2878 will not be. Any direct references between the two sections would be
2879 errors. For example, it would be an error if code in one section called
2880 a function defined in the other section.
2882 The @code{NOCROSSREFS} command takes a list of output section names. If
2883 @command{ld} detects any cross references between the sections, it reports
2884 an error and returns a non-zero exit status. Note that the
2885 @code{NOCROSSREFS} command uses output section names, not input section
2888 @ifclear SingleFormat
2889 @item OUTPUT_ARCH(@var{bfdarch})
2890 @kindex OUTPUT_ARCH(@var{bfdarch})
2891 @cindex machine architecture
2892 @cindex architecture
2893 Specify a particular output machine architecture. The argument is one
2894 of the names used by the BFD library (@pxref{BFD}). You can see the
2895 architecture of an object file by using the @code{objdump} program with
2896 the @samp{-f} option.
2901 @section Assigning Values to Symbols
2902 @cindex assignment in scripts
2903 @cindex symbol definition, scripts
2904 @cindex variables, defining
2905 You may assign a value to a symbol in a linker script. This will define
2906 the symbol and place it into the symbol table with a global scope.
2909 * Simple Assignments:: Simple Assignments
2911 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2912 * Source Code Reference:: How to use a linker script defined symbol in source code
2915 @node Simple Assignments
2916 @subsection Simple Assignments
2918 You may assign to a symbol using any of the C assignment operators:
2921 @item @var{symbol} = @var{expression} ;
2922 @itemx @var{symbol} += @var{expression} ;
2923 @itemx @var{symbol} -= @var{expression} ;
2924 @itemx @var{symbol} *= @var{expression} ;
2925 @itemx @var{symbol} /= @var{expression} ;
2926 @itemx @var{symbol} <<= @var{expression} ;
2927 @itemx @var{symbol} >>= @var{expression} ;
2928 @itemx @var{symbol} &= @var{expression} ;
2929 @itemx @var{symbol} |= @var{expression} ;
2932 The first case will define @var{symbol} to the value of
2933 @var{expression}. In the other cases, @var{symbol} must already be
2934 defined, and the value will be adjusted accordingly.
2936 The special symbol name @samp{.} indicates the location counter. You
2937 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
2939 The semicolon after @var{expression} is required.
2941 Expressions are defined below; see @ref{Expressions}.
2943 You may write symbol assignments as commands in their own right, or as
2944 statements within a @code{SECTIONS} command, or as part of an output
2945 section description in a @code{SECTIONS} command.
2947 The section of the symbol will be set from the section of the
2948 expression; for more information, see @ref{Expression Section}.
2950 Here is an example showing the three different places that symbol
2951 assignments may be used:
2962 _bdata = (. + 3) & ~ 3;
2963 .data : @{ *(.data) @}
2967 In this example, the symbol @samp{floating_point} will be defined as
2968 zero. The symbol @samp{_etext} will be defined as the address following
2969 the last @samp{.text} input section. The symbol @samp{_bdata} will be
2970 defined as the address following the @samp{.text} output section aligned
2971 upward to a 4 byte boundary.
2976 In some cases, it is desirable for a linker script to define a symbol
2977 only if it is referenced and is not defined by any object included in
2978 the link. For example, traditional linkers defined the symbol
2979 @samp{etext}. However, ANSI C requires that the user be able to use
2980 @samp{etext} as a function name without encountering an error. The
2981 @code{PROVIDE} keyword may be used to define a symbol, such as
2982 @samp{etext}, only if it is referenced but not defined. The syntax is
2983 @code{PROVIDE(@var{symbol} = @var{expression})}.
2985 Here is an example of using @code{PROVIDE} to define @samp{etext}:
2998 In this example, if the program defines @samp{_etext} (with a leading
2999 underscore), the linker will give a multiple definition error. If, on
3000 the other hand, the program defines @samp{etext} (with no leading
3001 underscore), the linker will silently use the definition in the program.
3002 If the program references @samp{etext} but does not define it, the
3003 linker will use the definition in the linker script.
3005 @node PROVIDE_HIDDEN
3006 @subsection PROVIDE_HIDDEN
3007 @cindex PROVIDE_HIDDEN
3008 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3009 hidden and won't be exported.
3011 @node Source Code Reference
3012 @subsection Source Code Reference
3014 Accessing a linker script defined variable from source code is not
3015 intuitive. In particular a linker script symbol is not equivalent to
3016 a variable declaration in a high level language, it is instead a
3017 symbol that does not have a value.
3019 Before going further, it is important to note that compilers often
3020 transform names in the source code into different names when they are
3021 stored in the symbol table. For example, Fortran compilers commonly
3022 prepend or append an underscore, and C++ performs extensive @samp{name
3023 mangling}. Therefore there might be a discrepancy between the name
3024 of a variable as it is used in source code and the name of the same
3025 variable as it is defined in a linker script. For example in C a
3026 linker script variable might be referred to as:
3032 But in the linker script it might be defined as:
3038 In the remaining examples however it is assumed that no name
3039 transformation has taken place.
3041 When a symbol is declared in a high level language such as C, two
3042 things happen. The first is that the compiler reserves enough space
3043 in the program's memory to hold the @emph{value} of the symbol. The
3044 second is that the compiler creates an entry in the program's symbol
3045 table which holds the symbol's @emph{address}. ie the symbol table
3046 contains the address of the block of memory holding the symbol's
3047 value. So for example the following C declaration, at file scope:
3053 creates a entry called @samp{foo} in the symbol table. This entry
3054 holds the address of an @samp{int} sized block of memory where the
3055 number 1000 is initially stored.
3057 When a program references a symbol the compiler generates code that
3058 first accesses the symbol table to find the address of the symbol's
3059 memory block and then code to read the value from that memory block.
3066 looks up the symbol @samp{foo} in the symbol table, gets the address
3067 associated with this symbol and then writes the value 1 into that
3074 looks up the symbol @samp{foo} in the symbol table, gets it address
3075 and then copies this address into the block of memory associated with
3076 the variable @samp{a}.
3078 Linker scripts symbol declarations, by contrast, create an entry in
3079 the symbol table but do not assign any memory to them. Thus they are
3080 an address without a value. So for example the linker script definition:
3086 creates an entry in the symbol table called @samp{foo} which holds
3087 the address of memory location 1000, but nothing special is stored at
3088 address 1000. This means that you cannot access the @emph{value} of a
3089 linker script defined symbol - it has no value - all you can do is
3090 access the @emph{address} of a linker script defined symbol.
3092 Hence when you are using a linker script defined symbol in source code
3093 you should always take the address of the symbol, and never attempt to
3094 use its value. For example suppose you want to copy the contents of a
3095 section of memory called .ROM into a section called .FLASH and the
3096 linker script contains these declarations:
3100 start_of_ROM = .ROM;
3101 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3102 start_of_FLASH = .FLASH;
3106 Then the C source code to perform the copy would be:
3110 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3112 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3116 Note the use of the @samp{&} operators. These are correct.
3119 @section SECTIONS Command
3121 The @code{SECTIONS} command tells the linker how to map input sections
3122 into output sections, and how to place the output sections in memory.
3124 The format of the @code{SECTIONS} command is:
3128 @var{sections-command}
3129 @var{sections-command}
3134 Each @var{sections-command} may of be one of the following:
3138 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3140 a symbol assignment (@pxref{Assignments})
3142 an output section description
3144 an overlay description
3147 The @code{ENTRY} command and symbol assignments are permitted inside the
3148 @code{SECTIONS} command for convenience in using the location counter in
3149 those commands. This can also make the linker script easier to
3150 understand because you can use those commands at meaningful points in
3151 the layout of the output file.
3153 Output section descriptions and overlay descriptions are described
3156 If you do not use a @code{SECTIONS} command in your linker script, the
3157 linker will place each input section into an identically named output
3158 section in the order that the sections are first encountered in the
3159 input files. If all input sections are present in the first file, for
3160 example, the order of sections in the output file will match the order
3161 in the first input file. The first section will be at address zero.
3164 * Output Section Description:: Output section description
3165 * Output Section Name:: Output section name
3166 * Output Section Address:: Output section address
3167 * Input Section:: Input section description
3168 * Output Section Data:: Output section data
3169 * Output Section Keywords:: Output section keywords
3170 * Output Section Discarding:: Output section discarding
3171 * Output Section Attributes:: Output section attributes
3172 * Overlay Description:: Overlay description
3175 @node Output Section Description
3176 @subsection Output Section Description
3177 The full description of an output section looks like this:
3180 @var{section} [@var{address}] [(@var{type})] :
3181 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3183 @var{output-section-command}
3184 @var{output-section-command}
3186 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3190 Most output sections do not use most of the optional section attributes.
3192 The whitespace around @var{section} is required, so that the section
3193 name is unambiguous. The colon and the curly braces are also required.
3194 The line breaks and other white space are optional.
3196 Each @var{output-section-command} may be one of the following:
3200 a symbol assignment (@pxref{Assignments})
3202 an input section description (@pxref{Input Section})
3204 data values to include directly (@pxref{Output Section Data})
3206 a special output section keyword (@pxref{Output Section Keywords})
3209 @node Output Section Name
3210 @subsection Output Section Name
3211 @cindex name, section
3212 @cindex section name
3213 The name of the output section is @var{section}. @var{section} must
3214 meet the constraints of your output format. In formats which only
3215 support a limited number of sections, such as @code{a.out}, the name
3216 must be one of the names supported by the format (@code{a.out}, for
3217 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3218 output format supports any number of sections, but with numbers and not
3219 names (as is the case for Oasys), the name should be supplied as a
3220 quoted numeric string. A section name may consist of any sequence of
3221 characters, but a name which contains any unusual characters such as
3222 commas must be quoted.
3224 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3227 @node Output Section Address
3228 @subsection Output Section Address
3229 @cindex address, section
3230 @cindex section address
3231 The @var{address} is an expression for the VMA (the virtual memory
3232 address) of the output section. If you do not provide @var{address},
3233 the linker will set it based on @var{region} if present, or otherwise
3234 based on the current value of the location counter.
3236 If you provide @var{address}, the address of the output section will be
3237 set to precisely that. If you provide neither @var{address} nor
3238 @var{region}, then the address of the output section will be set to the
3239 current value of the location counter aligned to the alignment
3240 requirements of the output section. The alignment requirement of the
3241 output section is the strictest alignment of any input section contained
3242 within the output section.
3246 .text . : @{ *(.text) @}
3251 .text : @{ *(.text) @}
3254 are subtly different. The first will set the address of the
3255 @samp{.text} output section to the current value of the location
3256 counter. The second will set it to the current value of the location
3257 counter aligned to the strictest alignment of a @samp{.text} input
3260 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3261 For example, if you want to align the section on a 0x10 byte boundary,
3262 so that the lowest four bits of the section address are zero, you could
3263 do something like this:
3265 .text ALIGN(0x10) : @{ *(.text) @}
3268 This works because @code{ALIGN} returns the current location counter
3269 aligned upward to the specified value.
3271 Specifying @var{address} for a section will change the value of the
3275 @subsection Input Section Description
3276 @cindex input sections
3277 @cindex mapping input sections to output sections
3278 The most common output section command is an input section description.
3280 The input section description is the most basic linker script operation.
3281 You use output sections to tell the linker how to lay out your program
3282 in memory. You use input section descriptions to tell the linker how to
3283 map the input files into your memory layout.
3286 * Input Section Basics:: Input section basics
3287 * Input Section Wildcards:: Input section wildcard patterns
3288 * Input Section Common:: Input section for common symbols
3289 * Input Section Keep:: Input section and garbage collection
3290 * Input Section Example:: Input section example
3293 @node Input Section Basics
3294 @subsubsection Input Section Basics
3295 @cindex input section basics
3296 An input section description consists of a file name optionally followed
3297 by a list of section names in parentheses.
3299 The file name and the section name may be wildcard patterns, which we
3300 describe further below (@pxref{Input Section Wildcards}).
3302 The most common input section description is to include all input
3303 sections with a particular name in the output section. For example, to
3304 include all input @samp{.text} sections, you would write:
3309 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3310 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3311 match all files except the ones specified in the EXCLUDE_FILE list. For
3314 (*(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors))
3316 will cause all .ctors sections from all files except @file{crtend.o} and
3317 @file{otherfile.o} to be included.
3319 There are two ways to include more than one section:
3325 The difference between these is the order in which the @samp{.text} and
3326 @samp{.rdata} input sections will appear in the output section. In the
3327 first example, they will be intermingled, appearing in the same order as
3328 they are found in the linker input. In the second example, all
3329 @samp{.text} input sections will appear first, followed by all
3330 @samp{.rdata} input sections.
3332 You can specify a file name to include sections from a particular file.
3333 You would do this if one or more of your files contain special data that
3334 needs to be at a particular location in memory. For example:
3339 If you use a file name without a list of sections, then all sections in
3340 the input file will be included in the output section. This is not
3341 commonly done, but it may by useful on occasion. For example:
3346 When you use a file name which does not contain any wild card
3347 characters, the linker will first see if you also specified the file
3348 name on the linker command line or in an @code{INPUT} command. If you
3349 did not, the linker will attempt to open the file as an input file, as
3350 though it appeared on the command line. Note that this differs from an
3351 @code{INPUT} command, because the linker will not search for the file in
3352 the archive search path.
3354 @node Input Section Wildcards
3355 @subsubsection Input Section Wildcard Patterns
3356 @cindex input section wildcards
3357 @cindex wildcard file name patterns
3358 @cindex file name wildcard patterns
3359 @cindex section name wildcard patterns
3360 In an input section description, either the file name or the section
3361 name or both may be wildcard patterns.
3363 The file name of @samp{*} seen in many examples is a simple wildcard
3364 pattern for the file name.
3366 The wildcard patterns are like those used by the Unix shell.
3370 matches any number of characters
3372 matches any single character
3374 matches a single instance of any of the @var{chars}; the @samp{-}
3375 character may be used to specify a range of characters, as in
3376 @samp{[a-z]} to match any lower case letter
3378 quotes the following character
3381 When a file name is matched with a wildcard, the wildcard characters
3382 will not match a @samp{/} character (used to separate directory names on
3383 Unix). A pattern consisting of a single @samp{*} character is an
3384 exception; it will always match any file name, whether it contains a
3385 @samp{/} or not. In a section name, the wildcard characters will match
3386 a @samp{/} character.
3388 File name wildcard patterns only match files which are explicitly
3389 specified on the command line or in an @code{INPUT} command. The linker
3390 does not search directories to expand wildcards.
3392 If a file name matches more than one wildcard pattern, or if a file name
3393 appears explicitly and is also matched by a wildcard pattern, the linker
3394 will use the first match in the linker script. For example, this
3395 sequence of input section descriptions is probably in error, because the
3396 @file{data.o} rule will not be used:
3398 .data : @{ *(.data) @}
3399 .data1 : @{ data.o(.data) @}
3402 @cindex SORT_BY_NAME
3403 Normally, the linker will place files and sections matched by wildcards
3404 in the order in which they are seen during the link. You can change
3405 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3406 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
3407 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3408 into ascending order by name before placing them in the output file.
3410 @cindex SORT_BY_ALIGNMENT
3411 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3412 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3413 ascending order by alignment before placing them in the output file.
3416 @code{SORT} is an alias for @code{SORT_BY_NAME}.
3418 When there are nested section sorting commands in linker script, there
3419 can be at most 1 level of nesting for section sorting commands.
3423 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3424 It will sort the input sections by name first, then by alignment if 2
3425 sections have the same name.
3427 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3428 It will sort the input sections by alignment first, then by name if 2
3429 sections have the same alignment.
3431 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3432 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3434 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3435 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3437 All other nested section sorting commands are invalid.
3440 When both command line section sorting option and linker script
3441 section sorting command are used, section sorting command always
3442 takes precedence over the command line option.
3444 If the section sorting command in linker script isn't nested, the
3445 command line option will make the section sorting command to be
3446 treated as nested sorting command.
3450 @code{SORT_BY_NAME} (wildcard section pattern ) with
3451 @option{--sort-sections alignment} is equivalent to
3452 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3454 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3455 @option{--sort-section name} is equivalent to
3456 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3459 If the section sorting command in linker script is nested, the
3460 command line option will be ignored.
3462 If you ever get confused about where input sections are going, use the
3463 @samp{-M} linker option to generate a map file. The map file shows
3464 precisely how input sections are mapped to output sections.
3466 This example shows how wildcard patterns might be used to partition
3467 files. This linker script directs the linker to place all @samp{.text}
3468 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3469 The linker will place the @samp{.data} section from all files beginning
3470 with an upper case character in @samp{.DATA}; for all other files, the
3471 linker will place the @samp{.data} section in @samp{.data}.
3475 .text : @{ *(.text) @}
3476 .DATA : @{ [A-Z]*(.data) @}
3477 .data : @{ *(.data) @}
3478 .bss : @{ *(.bss) @}
3483 @node Input Section Common
3484 @subsubsection Input Section for Common Symbols
3485 @cindex common symbol placement
3486 @cindex uninitialized data placement
3487 A special notation is needed for common symbols, because in many object
3488 file formats common symbols do not have a particular input section. The
3489 linker treats common symbols as though they are in an input section
3490 named @samp{COMMON}.
3492 You may use file names with the @samp{COMMON} section just as with any
3493 other input sections. You can use this to place common symbols from a
3494 particular input file in one section while common symbols from other
3495 input files are placed in another section.
3497 In most cases, common symbols in input files will be placed in the
3498 @samp{.bss} section in the output file. For example:
3500 .bss @{ *(.bss) *(COMMON) @}
3503 @cindex scommon section
3504 @cindex small common symbols
3505 Some object file formats have more than one type of common symbol. For
3506 example, the MIPS ELF object file format distinguishes standard common
3507 symbols and small common symbols. In this case, the linker will use a
3508 different special section name for other types of common symbols. In
3509 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3510 symbols and @samp{.scommon} for small common symbols. This permits you
3511 to map the different types of common symbols into memory at different
3515 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3516 notation is now considered obsolete. It is equivalent to
3519 @node Input Section Keep
3520 @subsubsection Input Section and Garbage Collection
3522 @cindex garbage collection
3523 When link-time garbage collection is in use (@samp{--gc-sections}),
3524 it is often useful to mark sections that should not be eliminated.
3525 This is accomplished by surrounding an input section's wildcard entry
3526 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3527 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3529 @node Input Section Example
3530 @subsubsection Input Section Example
3531 The following example is a complete linker script. It tells the linker
3532 to read all of the sections from file @file{all.o} and place them at the
3533 start of output section @samp{outputa} which starts at location
3534 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3535 follows immediately, in the same output section. All of section
3536 @samp{.input2} from @file{foo.o} goes into output section
3537 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3538 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3539 files are written to output section @samp{outputc}.
3567 @node Output Section Data
3568 @subsection Output Section Data
3570 @cindex section data
3571 @cindex output section data
3572 @kindex BYTE(@var{expression})
3573 @kindex SHORT(@var{expression})
3574 @kindex LONG(@var{expression})
3575 @kindex QUAD(@var{expression})
3576 @kindex SQUAD(@var{expression})
3577 You can include explicit bytes of data in an output section by using
3578 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3579 an output section command. Each keyword is followed by an expression in
3580 parentheses providing the value to store (@pxref{Expressions}). The
3581 value of the expression is stored at the current value of the location
3584 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3585 store one, two, four, and eight bytes (respectively). After storing the
3586 bytes, the location counter is incremented by the number of bytes
3589 For example, this will store the byte 1 followed by the four byte value
3590 of the symbol @samp{addr}:
3596 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3597 same; they both store an 8 byte, or 64 bit, value. When both host and
3598 target are 32 bits, an expression is computed as 32 bits. In this case
3599 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3600 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
3602 If the object file format of the output file has an explicit endianness,
3603 which is the normal case, the value will be stored in that endianness.
3604 When the object file format does not have an explicit endianness, as is
3605 true of, for example, S-records, the value will be stored in the
3606 endianness of the first input object file.
3608 Note---these commands only work inside a section description and not
3609 between them, so the following will produce an error from the linker:
3611 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
3613 whereas this will work:
3615 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
3618 @kindex FILL(@var{expression})
3619 @cindex holes, filling
3620 @cindex unspecified memory
3621 You may use the @code{FILL} command to set the fill pattern for the
3622 current section. It is followed by an expression in parentheses. Any
3623 otherwise unspecified regions of memory within the section (for example,
3624 gaps left due to the required alignment of input sections) are filled
3625 with the value of the expression, repeated as
3626 necessary. A @code{FILL} statement covers memory locations after the
3627 point at which it occurs in the section definition; by including more
3628 than one @code{FILL} statement, you can have different fill patterns in
3629 different parts of an output section.
3631 This example shows how to fill unspecified regions of memory with the
3637 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
3638 section attribute, but it only affects the
3639 part of the section following the @code{FILL} command, rather than the
3640 entire section. If both are used, the @code{FILL} command takes
3641 precedence. @xref{Output Section Fill}, for details on the fill
3644 @node Output Section Keywords
3645 @subsection Output Section Keywords
3646 There are a couple of keywords which can appear as output section
3650 @kindex CREATE_OBJECT_SYMBOLS
3651 @cindex input filename symbols
3652 @cindex filename symbols
3653 @item CREATE_OBJECT_SYMBOLS
3654 The command tells the linker to create a symbol for each input file.
3655 The name of each symbol will be the name of the corresponding input
3656 file. The section of each symbol will be the output section in which
3657 the @code{CREATE_OBJECT_SYMBOLS} command appears.
3659 This is conventional for the a.out object file format. It is not
3660 normally used for any other object file format.
3662 @kindex CONSTRUCTORS
3663 @cindex C++ constructors, arranging in link
3664 @cindex constructors, arranging in link
3666 When linking using the a.out object file format, the linker uses an
3667 unusual set construct to support C++ global constructors and
3668 destructors. When linking object file formats which do not support
3669 arbitrary sections, such as ECOFF and XCOFF, the linker will
3670 automatically recognize C++ global constructors and destructors by name.
3671 For these object file formats, the @code{CONSTRUCTORS} command tells the
3672 linker to place constructor information in the output section where the
3673 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
3674 ignored for other object file formats.
3676 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
3677 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
3678 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
3679 the start and end of the global destructors. The
3680 first word in the list is the number of entries, followed by the address
3681 of each constructor or destructor, followed by a zero word. The
3682 compiler must arrange to actually run the code. For these object file
3683 formats @sc{gnu} C++ normally calls constructors from a subroutine
3684 @code{__main}; a call to @code{__main} is automatically inserted into
3685 the startup code for @code{main}. @sc{gnu} C++ normally runs
3686 destructors either by using @code{atexit}, or directly from the function
3689 For object file formats such as @code{COFF} or @code{ELF} which support
3690 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
3691 addresses of global constructors and destructors into the @code{.ctors}
3692 and @code{.dtors} sections. Placing the following sequence into your
3693 linker script will build the sort of table which the @sc{gnu} C++
3694 runtime code expects to see.
3698 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3703 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3709 If you are using the @sc{gnu} C++ support for initialization priority,
3710 which provides some control over the order in which global constructors
3711 are run, you must sort the constructors at link time to ensure that they
3712 are executed in the correct order. When using the @code{CONSTRUCTORS}
3713 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
3714 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
3715 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
3718 Normally the compiler and linker will handle these issues automatically,
3719 and you will not need to concern yourself with them. However, you may
3720 need to consider this if you are using C++ and writing your own linker
3725 @node Output Section Discarding
3726 @subsection Output Section Discarding
3727 @cindex discarding sections
3728 @cindex sections, discarding
3729 @cindex removing sections
3730 The linker will not create output sections with no contents. This is
3731 for convenience when referring to input sections that may or may not
3732 be present in any of the input files. For example:
3734 .foo : @{ *(.foo) @}
3737 will only create a @samp{.foo} section in the output file if there is a
3738 @samp{.foo} section in at least one input file, and if the input
3739 sections are not all empty. Other link script directives that allocate
3740 space in an output section will also create the output section.
3742 The linker will ignore address assignments (@pxref{Output Section Address})
3743 on discarded output sections, except when the linker script defines
3744 symbols in the output section. In that case the linker will obey
3745 the address assignments, possibly advancing dot even though the
3746 section is discarded.
3749 The special output section name @samp{/DISCARD/} may be used to discard
3750 input sections. Any input sections which are assigned to an output
3751 section named @samp{/DISCARD/} are not included in the output file.
3753 @node Output Section Attributes
3754 @subsection Output Section Attributes
3755 @cindex output section attributes
3756 We showed above that the full description of an output section looked
3760 @var{section} [@var{address}] [(@var{type})] :
3761 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3763 @var{output-section-command}
3764 @var{output-section-command}
3766 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3769 We've already described @var{section}, @var{address}, and
3770 @var{output-section-command}. In this section we will describe the
3771 remaining section attributes.
3774 * Output Section Type:: Output section type
3775 * Output Section LMA:: Output section LMA
3776 * Forced Output Alignment:: Forced Output Alignment
3777 * Forced Input Alignment:: Forced Input Alignment
3778 * Output Section Region:: Output section region
3779 * Output Section Phdr:: Output section phdr
3780 * Output Section Fill:: Output section fill
3783 @node Output Section Type
3784 @subsubsection Output Section Type
3785 Each output section may have a type. The type is a keyword in
3786 parentheses. The following types are defined:
3790 The section should be marked as not loadable, so that it will not be
3791 loaded into memory when the program is run.
3796 These type names are supported for backward compatibility, and are
3797 rarely used. They all have the same effect: the section should be
3798 marked as not allocatable, so that no memory is allocated for the
3799 section when the program is run.
3803 @cindex prevent unnecessary loading
3804 @cindex loading, preventing
3805 The linker normally sets the attributes of an output section based on
3806 the input sections which map into it. You can override this by using
3807 the section type. For example, in the script sample below, the
3808 @samp{ROM} section is addressed at memory location @samp{0} and does not
3809 need to be loaded when the program is run. The contents of the
3810 @samp{ROM} section will appear in the linker output file as usual.
3814 ROM 0 (NOLOAD) : @{ @dots{} @}
3820 @node Output Section LMA
3821 @subsubsection Output Section LMA
3822 @kindex AT>@var{lma_region}
3823 @kindex AT(@var{lma})
3824 @cindex load address
3825 @cindex section load address
3826 Every section has a virtual address (VMA) and a load address (LMA); see
3827 @ref{Basic Script Concepts}. The address expression which may appear in
3828 an output section description sets the VMA (@pxref{Output Section
3831 The expression @var{lma} that follows the @code{AT} keyword specifies
3832 the load address of the section.
3834 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
3835 specify a memory region for the section's load address. @xref{MEMORY}.
3836 Note that if the section has not had a VMA assigned to it then the
3837 linker will use the @var{lma_region} as the VMA region as well.
3839 If neither @code{AT} nor @code{AT>} is specified for an allocatable
3840 section, the linker will set the LMA such that the difference between
3841 VMA and LMA for the section is the same as the preceding output
3842 section in the same region. If there is no preceding output section
3843 or the section is not allocatable, the linker will set the LMA equal
3845 @xref{Output Section Region}.
3847 @cindex ROM initialized data
3848 @cindex initialized data in ROM
3849 This feature is designed to make it easy to build a ROM image. For
3850 example, the following linker script creates three output sections: one
3851 called @samp{.text}, which starts at @code{0x1000}, one called
3852 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
3853 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
3854 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
3855 defined with the value @code{0x2000}, which shows that the location
3856 counter holds the VMA value, not the LMA value.
3862 .text 0x1000 : @{ *(.text) _etext = . ; @}
3864 AT ( ADDR (.text) + SIZEOF (.text) )
3865 @{ _data = . ; *(.data); _edata = . ; @}
3867 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
3872 The run-time initialization code for use with a program generated with
3873 this linker script would include something like the following, to copy
3874 the initialized data from the ROM image to its runtime address. Notice
3875 how this code takes advantage of the symbols defined by the linker
3880 extern char _etext, _data, _edata, _bstart, _bend;
3881 char *src = &_etext;
3884 /* ROM has data at end of text; copy it. */
3885 while (dst < &_edata) @{
3890 for (dst = &_bstart; dst< &_bend; dst++)
3895 @node Forced Output Alignment
3896 @subsubsection Forced Output Alignment
3897 @kindex ALIGN(@var{section_align})
3898 @cindex forcing output section alignment
3899 @cindex output section alignment
3900 You can increase an output section's alignment by using ALIGN.
3902 @node Forced Input Alignment
3903 @subsubsection Forced Input Alignment
3904 @kindex SUBALIGN(@var{subsection_align})
3905 @cindex forcing input section alignment
3906 @cindex input section alignment
3907 You can force input section alignment within an output section by using
3908 SUBALIGN. The value specified overrides any alignment given by input
3909 sections, whether larger or smaller.
3911 @node Output Section Region
3912 @subsubsection Output Section Region
3913 @kindex >@var{region}
3914 @cindex section, assigning to memory region
3915 @cindex memory regions and sections
3916 You can assign a section to a previously defined region of memory by
3917 using @samp{>@var{region}}. @xref{MEMORY}.
3919 Here is a simple example:
3922 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
3923 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
3927 @node Output Section Phdr
3928 @subsubsection Output Section Phdr
3930 @cindex section, assigning to program header
3931 @cindex program headers and sections
3932 You can assign a section to a previously defined program segment by
3933 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
3934 one or more segments, then all subsequent allocated sections will be
3935 assigned to those segments as well, unless they use an explicitly
3936 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
3937 linker to not put the section in any segment at all.
3939 Here is a simple example:
3942 PHDRS @{ text PT_LOAD ; @}
3943 SECTIONS @{ .text : @{ *(.text) @} :text @}
3947 @node Output Section Fill
3948 @subsubsection Output Section Fill
3949 @kindex =@var{fillexp}
3950 @cindex section fill pattern
3951 @cindex fill pattern, entire section
3952 You can set the fill pattern for an entire section by using
3953 @samp{=@var{fillexp}}. @var{fillexp} is an expression
3954 (@pxref{Expressions}). Any otherwise unspecified regions of memory
3955 within the output section (for example, gaps left due to the required
3956 alignment of input sections) will be filled with the value, repeated as
3957 necessary. If the fill expression is a simple hex number, ie. a string
3958 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
3959 an arbitrarily long sequence of hex digits can be used to specify the
3960 fill pattern; Leading zeros become part of the pattern too. For all
3961 other cases, including extra parentheses or a unary @code{+}, the fill
3962 pattern is the four least significant bytes of the value of the
3963 expression. In all cases, the number is big-endian.
3965 You can also change the fill value with a @code{FILL} command in the
3966 output section commands; (@pxref{Output Section Data}).
3968 Here is a simple example:
3971 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
3975 @node Overlay Description
3976 @subsection Overlay Description
3979 An overlay description provides an easy way to describe sections which
3980 are to be loaded as part of a single memory image but are to be run at
3981 the same memory address. At run time, some sort of overlay manager will
3982 copy the overlaid sections in and out of the runtime memory address as
3983 required, perhaps by simply manipulating addressing bits. This approach
3984 can be useful, for example, when a certain region of memory is faster
3987 Overlays are described using the @code{OVERLAY} command. The
3988 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
3989 output section description. The full syntax of the @code{OVERLAY}
3990 command is as follows:
3993 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
3997 @var{output-section-command}
3998 @var{output-section-command}
4000 @} [:@var{phdr}@dots{}] [=@var{fill}]
4003 @var{output-section-command}
4004 @var{output-section-command}
4006 @} [:@var{phdr}@dots{}] [=@var{fill}]
4008 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4012 Everything is optional except @code{OVERLAY} (a keyword), and each
4013 section must have a name (@var{secname1} and @var{secname2} above). The
4014 section definitions within the @code{OVERLAY} construct are identical to
4015 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
4016 except that no addresses and no memory regions may be defined for
4017 sections within an @code{OVERLAY}.
4019 The sections are all defined with the same starting address. The load
4020 addresses of the sections are arranged such that they are consecutive in
4021 memory starting at the load address used for the @code{OVERLAY} as a
4022 whole (as with normal section definitions, the load address is optional,
4023 and defaults to the start address; the start address is also optional,
4024 and defaults to the current value of the location counter).
4026 If the @code{NOCROSSREFS} keyword is used, and there any references
4027 among the sections, the linker will report an error. Since the sections
4028 all run at the same address, it normally does not make sense for one
4029 section to refer directly to another. @xref{Miscellaneous Commands,
4032 For each section within the @code{OVERLAY}, the linker automatically
4033 defines two symbols. The symbol @code{__load_start_@var{secname}} is
4034 defined as the starting load address of the section. The symbol
4035 @code{__load_stop_@var{secname}} is defined as the final load address of
4036 the section. Any characters within @var{secname} which are not legal
4037 within C identifiers are removed. C (or assembler) code may use these
4038 symbols to move the overlaid sections around as necessary.
4040 At the end of the overlay, the value of the location counter is set to
4041 the start address of the overlay plus the size of the largest section.
4043 Here is an example. Remember that this would appear inside a
4044 @code{SECTIONS} construct.
4047 OVERLAY 0x1000 : AT (0x4000)
4049 .text0 @{ o1/*.o(.text) @}
4050 .text1 @{ o2/*.o(.text) @}
4055 This will define both @samp{.text0} and @samp{.text1} to start at
4056 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4057 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4058 following symbols will be defined: @code{__load_start_text0},
4059 @code{__load_stop_text0}, @code{__load_start_text1},
4060 @code{__load_stop_text1}.
4062 C code to copy overlay @code{.text1} into the overlay area might look
4067 extern char __load_start_text1, __load_stop_text1;
4068 memcpy ((char *) 0x1000, &__load_start_text1,
4069 &__load_stop_text1 - &__load_start_text1);
4073 Note that the @code{OVERLAY} command is just syntactic sugar, since
4074 everything it does can be done using the more basic commands. The above
4075 example could have been written identically as follows.
4079 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4080 __load_start_text0 = LOADADDR (.text0);
4081 __load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0);
4082 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4083 __load_start_text1 = LOADADDR (.text1);
4084 __load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1);
4085 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4090 @section MEMORY Command
4092 @cindex memory regions
4093 @cindex regions of memory
4094 @cindex allocating memory
4095 @cindex discontinuous memory
4096 The linker's default configuration permits allocation of all available
4097 memory. You can override this by using the @code{MEMORY} command.
4099 The @code{MEMORY} command describes the location and size of blocks of
4100 memory in the target. You can use it to describe which memory regions
4101 may be used by the linker, and which memory regions it must avoid. You
4102 can then assign sections to particular memory regions. The linker will
4103 set section addresses based on the memory regions, and will warn about
4104 regions that become too full. The linker will not shuffle sections
4105 around to fit into the available regions.
4107 A linker script may contain at most one use of the @code{MEMORY}
4108 command. However, you can define as many blocks of memory within it as
4109 you wish. The syntax is:
4114 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4120 The @var{name} is a name used in the linker script to refer to the
4121 region. The region name has no meaning outside of the linker script.
4122 Region names are stored in a separate name space, and will not conflict
4123 with symbol names, file names, or section names. Each memory region
4124 must have a distinct name.
4126 @cindex memory region attributes
4127 The @var{attr} string is an optional list of attributes that specify
4128 whether to use a particular memory region for an input section which is
4129 not explicitly mapped in the linker script. As described in
4130 @ref{SECTIONS}, if you do not specify an output section for some input
4131 section, the linker will create an output section with the same name as
4132 the input section. If you define region attributes, the linker will use
4133 them to select the memory region for the output section that it creates.
4135 The @var{attr} string must consist only of the following characters:
4150 Invert the sense of any of the preceding attributes
4153 If a unmapped section matches any of the listed attributes other than
4154 @samp{!}, it will be placed in the memory region. The @samp{!}
4155 attribute reverses this test, so that an unmapped section will be placed
4156 in the memory region only if it does not match any of the listed
4162 The @var{origin} is an numerical expression for the start address of
4163 the memory region. The expression must evaluate to a constant and it
4164 cannot involve any symbols. The keyword @code{ORIGIN} may be
4165 abbreviated to @code{org} or @code{o} (but not, for example,
4171 The @var{len} is an expression for the size in bytes of the memory
4172 region. As with the @var{origin} expression, the expression must
4173 be numerical only and must evaluate to a constant. The keyword
4174 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4176 In the following example, we specify that there are two memory regions
4177 available for allocation: one starting at @samp{0} for 256 kilobytes,
4178 and the other starting at @samp{0x40000000} for four megabytes. The
4179 linker will place into the @samp{rom} memory region every section which
4180 is not explicitly mapped into a memory region, and is either read-only
4181 or executable. The linker will place other sections which are not
4182 explicitly mapped into a memory region into the @samp{ram} memory
4189 rom (rx) : ORIGIN = 0, LENGTH = 256K
4190 ram (!rx) : org = 0x40000000, l = 4M
4195 Once you define a memory region, you can direct the linker to place
4196 specific output sections into that memory region by using the
4197 @samp{>@var{region}} output section attribute. For example, if you have
4198 a memory region named @samp{mem}, you would use @samp{>mem} in the
4199 output section definition. @xref{Output Section Region}. If no address
4200 was specified for the output section, the linker will set the address to
4201 the next available address within the memory region. If the combined
4202 output sections directed to a memory region are too large for the
4203 region, the linker will issue an error message.
4205 It is possible to access the origin and length of a memory in an
4206 expression via the @code{ORIGIN(@var{memory})} and
4207 @code{LENGTH(@var{memory})} functions:
4211 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4216 @section PHDRS Command
4218 @cindex program headers
4219 @cindex ELF program headers
4220 @cindex program segments
4221 @cindex segments, ELF
4222 The ELF object file format uses @dfn{program headers}, also knows as
4223 @dfn{segments}. The program headers describe how the program should be
4224 loaded into memory. You can print them out by using the @code{objdump}
4225 program with the @samp{-p} option.
4227 When you run an ELF program on a native ELF system, the system loader
4228 reads the program headers in order to figure out how to load the
4229 program. This will only work if the program headers are set correctly.
4230 This manual does not describe the details of how the system loader
4231 interprets program headers; for more information, see the ELF ABI.
4233 The linker will create reasonable program headers by default. However,
4234 in some cases, you may need to specify the program headers more
4235 precisely. You may use the @code{PHDRS} command for this purpose. When
4236 the linker sees the @code{PHDRS} command in the linker script, it will
4237 not create any program headers other than the ones specified.
4239 The linker only pays attention to the @code{PHDRS} command when
4240 generating an ELF output file. In other cases, the linker will simply
4241 ignore @code{PHDRS}.
4243 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4244 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4250 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4251 [ FLAGS ( @var{flags} ) ] ;
4256 The @var{name} is used only for reference in the @code{SECTIONS} command
4257 of the linker script. It is not put into the output file. Program
4258 header names are stored in a separate name space, and will not conflict
4259 with symbol names, file names, or section names. Each program header
4260 must have a distinct name.
4262 Certain program header types describe segments of memory which the
4263 system loader will load from the file. In the linker script, you
4264 specify the contents of these segments by placing allocatable output
4265 sections in the segments. You use the @samp{:@var{phdr}} output section
4266 attribute to place a section in a particular segment. @xref{Output
4269 It is normal to put certain sections in more than one segment. This
4270 merely implies that one segment of memory contains another. You may
4271 repeat @samp{:@var{phdr}}, using it once for each segment which should
4272 contain the section.
4274 If you place a section in one or more segments using @samp{:@var{phdr}},
4275 then the linker will place all subsequent allocatable sections which do
4276 not specify @samp{:@var{phdr}} in the same segments. This is for
4277 convenience, since generally a whole set of contiguous sections will be
4278 placed in a single segment. You can use @code{:NONE} to override the
4279 default segment and tell the linker to not put the section in any
4284 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4285 the program header type to further describe the contents of the segment.
4286 The @code{FILEHDR} keyword means that the segment should include the ELF
4287 file header. The @code{PHDRS} keyword means that the segment should
4288 include the ELF program headers themselves.
4290 The @var{type} may be one of the following. The numbers indicate the
4291 value of the keyword.
4294 @item @code{PT_NULL} (0)
4295 Indicates an unused program header.
4297 @item @code{PT_LOAD} (1)
4298 Indicates that this program header describes a segment to be loaded from
4301 @item @code{PT_DYNAMIC} (2)
4302 Indicates a segment where dynamic linking information can be found.
4304 @item @code{PT_INTERP} (3)
4305 Indicates a segment where the name of the program interpreter may be
4308 @item @code{PT_NOTE} (4)
4309 Indicates a segment holding note information.
4311 @item @code{PT_SHLIB} (5)
4312 A reserved program header type, defined but not specified by the ELF
4315 @item @code{PT_PHDR} (6)
4316 Indicates a segment where the program headers may be found.
4318 @item @var{expression}
4319 An expression giving the numeric type of the program header. This may
4320 be used for types not defined above.
4323 You can specify that a segment should be loaded at a particular address
4324 in memory by using an @code{AT} expression. This is identical to the
4325 @code{AT} command used as an output section attribute (@pxref{Output
4326 Section LMA}). The @code{AT} command for a program header overrides the
4327 output section attribute.
4329 The linker will normally set the segment flags based on the sections
4330 which comprise the segment. You may use the @code{FLAGS} keyword to
4331 explicitly specify the segment flags. The value of @var{flags} must be
4332 an integer. It is used to set the @code{p_flags} field of the program
4335 Here is an example of @code{PHDRS}. This shows a typical set of program
4336 headers used on a native ELF system.
4342 headers PT_PHDR PHDRS ;
4344 text PT_LOAD FILEHDR PHDRS ;
4346 dynamic PT_DYNAMIC ;
4352 .interp : @{ *(.interp) @} :text :interp
4353 .text : @{ *(.text) @} :text
4354 .rodata : @{ *(.rodata) @} /* defaults to :text */
4356 . = . + 0x1000; /* move to a new page in memory */
4357 .data : @{ *(.data) @} :data
4358 .dynamic : @{ *(.dynamic) @} :data :dynamic
4365 @section VERSION Command
4366 @kindex VERSION @{script text@}
4367 @cindex symbol versions
4368 @cindex version script
4369 @cindex versions of symbols
4370 The linker supports symbol versions when using ELF. Symbol versions are
4371 only useful when using shared libraries. The dynamic linker can use
4372 symbol versions to select a specific version of a function when it runs
4373 a program that may have been linked against an earlier version of the
4376 You can include a version script directly in the main linker script, or
4377 you can supply the version script as an implicit linker script. You can
4378 also use the @samp{--version-script} linker option.
4380 The syntax of the @code{VERSION} command is simply
4382 VERSION @{ version-script-commands @}
4385 The format of the version script commands is identical to that used by
4386 Sun's linker in Solaris 2.5. The version script defines a tree of
4387 version nodes. You specify the node names and interdependencies in the
4388 version script. You can specify which symbols are bound to which
4389 version nodes, and you can reduce a specified set of symbols to local
4390 scope so that they are not globally visible outside of the shared
4393 The easiest way to demonstrate the version script language is with a few
4414 "int f(int, double)";
4419 This example version script defines three version nodes. The first
4420 version node defined is @samp{VERS_1.1}; it has no other dependencies.
4421 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
4422 a number of symbols to local scope so that they are not visible outside
4423 of the shared library; this is done using wildcard patterns, so that any
4424 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4425 is matched. The wildcard patterns available are the same as those used
4426 in the shell when matching filenames (also known as ``globbing'').
4427 However, if you specify the symbol name inside double quotes, then the
4428 name is treated as literal, rather than as a glob pattern.
4430 Next, the version script defines node @samp{VERS_1.2}. This node
4431 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
4432 to the version node @samp{VERS_1.2}.
4434 Finally, the version script defines node @samp{VERS_2.0}. This node
4435 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
4436 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4438 When the linker finds a symbol defined in a library which is not
4439 specifically bound to a version node, it will effectively bind it to an
4440 unspecified base version of the library. You can bind all otherwise
4441 unspecified symbols to a given version node by using @samp{global: *;}
4442 somewhere in the version script.
4444 The names of the version nodes have no specific meaning other than what
4445 they might suggest to the person reading them. The @samp{2.0} version
4446 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4447 However, this would be a confusing way to write a version script.
4449 Node name can be omitted, provided it is the only version node
4450 in the version script. Such version script doesn't assign any versions to
4451 symbols, only selects which symbols will be globally visible out and which
4455 @{ global: foo; bar; local: *; @};
4458 When you link an application against a shared library that has versioned
4459 symbols, the application itself knows which version of each symbol it
4460 requires, and it also knows which version nodes it needs from each
4461 shared library it is linked against. Thus at runtime, the dynamic
4462 loader can make a quick check to make sure that the libraries you have
4463 linked against do in fact supply all of the version nodes that the
4464 application will need to resolve all of the dynamic symbols. In this
4465 way it is possible for the dynamic linker to know with certainty that
4466 all external symbols that it needs will be resolvable without having to
4467 search for each symbol reference.
4469 The symbol versioning is in effect a much more sophisticated way of
4470 doing minor version checking that SunOS does. The fundamental problem
4471 that is being addressed here is that typically references to external
4472 functions are bound on an as-needed basis, and are not all bound when
4473 the application starts up. If a shared library is out of date, a
4474 required interface may be missing; when the application tries to use
4475 that interface, it may suddenly and unexpectedly fail. With symbol
4476 versioning, the user will get a warning when they start their program if
4477 the libraries being used with the application are too old.
4479 There are several GNU extensions to Sun's versioning approach. The
4480 first of these is the ability to bind a symbol to a version node in the
4481 source file where the symbol is defined instead of in the versioning
4482 script. This was done mainly to reduce the burden on the library
4483 maintainer. You can do this by putting something like:
4485 __asm__(".symver original_foo,foo@@VERS_1.1");
4488 in the C source file. This renames the function @samp{original_foo} to
4489 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4490 The @samp{local:} directive can be used to prevent the symbol
4491 @samp{original_foo} from being exported. A @samp{.symver} directive
4492 takes precedence over a version script.
4494 The second GNU extension is to allow multiple versions of the same
4495 function to appear in a given shared library. In this way you can make
4496 an incompatible change to an interface without increasing the major
4497 version number of the shared library, while still allowing applications
4498 linked against the old interface to continue to function.
4500 To do this, you must use multiple @samp{.symver} directives in the
4501 source file. Here is an example:
4504 __asm__(".symver original_foo,foo@@");
4505 __asm__(".symver old_foo,foo@@VERS_1.1");
4506 __asm__(".symver old_foo1,foo@@VERS_1.2");
4507 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4510 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4511 unspecified base version of the symbol. The source file that contains this
4512 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4513 @samp{old_foo1}, and @samp{new_foo}.
4515 When you have multiple definitions of a given symbol, there needs to be
4516 some way to specify a default version to which external references to
4517 this symbol will be bound. You can do this with the
4518 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4519 declare one version of a symbol as the default in this manner; otherwise
4520 you would effectively have multiple definitions of the same symbol.
4522 If you wish to bind a reference to a specific version of the symbol
4523 within the shared library, you can use the aliases of convenience
4524 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4525 specifically bind to an external version of the function in question.
4527 You can also specify the language in the version script:
4530 VERSION extern "lang" @{ version-script-commands @}
4533 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4534 The linker will iterate over the list of symbols at the link time and
4535 demangle them according to @samp{lang} before matching them to the
4536 patterns specified in @samp{version-script-commands}.
4538 Demangled names may contains spaces and other special characters. As
4539 described above, you can use a glob pattern to match demangled names,
4540 or you can use a double-quoted string to match the string exactly. In
4541 the latter case, be aware that minor differences (such as differing
4542 whitespace) between the version script and the demangler output will
4543 cause a mismatch. As the exact string generated by the demangler
4544 might change in the future, even if the mangled name does not, you
4545 should check that all of your version directives are behaving as you
4546 expect when you upgrade.
4549 @section Expressions in Linker Scripts
4552 The syntax for expressions in the linker script language is identical to
4553 that of C expressions. All expressions are evaluated as integers. All
4554 expressions are evaluated in the same size, which is 32 bits if both the
4555 host and target are 32 bits, and is otherwise 64 bits.
4557 You can use and set symbol values in expressions.
4559 The linker defines several special purpose builtin functions for use in
4563 * Constants:: Constants
4564 * Symbols:: Symbol Names
4565 * Orphan Sections:: Orphan Sections
4566 * Location Counter:: The Location Counter
4567 * Operators:: Operators
4568 * Evaluation:: Evaluation
4569 * Expression Section:: The Section of an Expression
4570 * Builtin Functions:: Builtin Functions
4574 @subsection Constants
4575 @cindex integer notation
4576 @cindex constants in linker scripts
4577 All constants are integers.
4579 As in C, the linker considers an integer beginning with @samp{0} to be
4580 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4581 hexadecimal. The linker considers other integers to be decimal.
4583 @cindex scaled integers
4584 @cindex K and M integer suffixes
4585 @cindex M and K integer suffixes
4586 @cindex suffixes for integers
4587 @cindex integer suffixes
4588 In addition, you can use the suffixes @code{K} and @code{M} to scale a
4592 @c END TEXI2ROFF-KILL
4593 @code{1024} or @code{1024*1024}
4597 ${\rm 1024}$ or ${\rm 1024}^2$
4599 @c END TEXI2ROFF-KILL
4600 respectively. For example, the following all refer to the same quantity:
4608 @subsection Symbol Names
4609 @cindex symbol names
4611 @cindex quoted symbol names
4613 Unless quoted, symbol names start with a letter, underscore, or period
4614 and may include letters, digits, underscores, periods, and hyphens.
4615 Unquoted symbol names must not conflict with any keywords. You can
4616 specify a symbol which contains odd characters or has the same name as a
4617 keyword by surrounding the symbol name in double quotes:
4620 "with a space" = "also with a space" + 10;
4623 Since symbols can contain many non-alphabetic characters, it is safest
4624 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
4625 whereas @samp{A - B} is an expression involving subtraction.
4627 @node Orphan Sections
4628 @subsection Orphan Sections
4630 Orphan sections are sections present in the input files which
4631 are not explicitly placed into the output file by the linker
4632 script. The linker will still copy these sections into the
4633 output file, but it has to guess as to where they should be
4634 placed. The linker uses a simple heuristic to do this. It
4635 attempts to place orphan sections after non-orphan sections of the
4636 same attribute, such as code vs data, loadable vs non-loadable, etc.
4637 If there is not enough room to do this then it places
4638 at the end of the file.
4640 For ELF targets, the attribute of the section includes section type as
4641 well as section flag.
4643 @node Location Counter
4644 @subsection The Location Counter
4647 @cindex location counter
4648 @cindex current output location
4649 The special linker variable @dfn{dot} @samp{.} always contains the
4650 current output location counter. Since the @code{.} always refers to a
4651 location in an output section, it may only appear in an expression
4652 within a @code{SECTIONS} command. The @code{.} symbol may appear
4653 anywhere that an ordinary symbol is allowed in an expression.
4656 Assigning a value to @code{.} will cause the location counter to be
4657 moved. This may be used to create holes in the output section. The
4658 location counter may not be moved backwards inside an output section,
4659 and may not be moved backwards outside of an output section if so
4660 doing creates areas with overlapping LMAs.
4676 In the previous example, the @samp{.text} section from @file{file1} is
4677 located at the beginning of the output section @samp{output}. It is
4678 followed by a 1000 byte gap. Then the @samp{.text} section from
4679 @file{file2} appears, also with a 1000 byte gap following before the
4680 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
4681 specifies what data to write in the gaps (@pxref{Output Section Fill}).
4683 @cindex dot inside sections
4684 Note: @code{.} actually refers to the byte offset from the start of the
4685 current containing object. Normally this is the @code{SECTIONS}
4686 statement, whose start address is 0, hence @code{.} can be used as an
4687 absolute address. If @code{.} is used inside a section description
4688 however, it refers to the byte offset from the start of that section,
4689 not an absolute address. Thus in a script like this:
4707 The @samp{.text} section will be assigned a starting address of 0x100
4708 and a size of exactly 0x200 bytes, even if there is not enough data in
4709 the @samp{.text} input sections to fill this area. (If there is too
4710 much data, an error will be produced because this would be an attempt to
4711 move @code{.} backwards). The @samp{.data} section will start at 0x500
4712 and it will have an extra 0x600 bytes worth of space after the end of
4713 the values from the @samp{.data} input sections and before the end of
4714 the @samp{.data} output section itself.
4716 @cindex dot outside sections
4717 Setting symbols to the value of the location counter outside of an
4718 output section statement can result in unexpected values if the linker
4719 needs to place orphan sections. For example, given the following:
4725 .text: @{ *(.text) @}
4729 .data: @{ *(.data) @}
4734 If the linker needs to place some input section, e.g. @code{.rodata},
4735 not mentioned in the script, it might choose to place that section
4736 between @code{.text} and @code{.data}. You might think the linker
4737 should place @code{.rodata} on the blank line in the above script, but
4738 blank lines are of no particular significance to the linker. As well,
4739 the linker doesn't associate the above symbol names with their
4740 sections. Instead, it assumes that all assignments or other
4741 statements belong to the previous output section, except for the
4742 special case of an assignment to @code{.}. I.e., the linker will
4743 place the orphan @code{.rodata} section as if the script was written
4750 .text: @{ *(.text) @}
4754 .rodata: @{ *(.rodata) @}
4755 .data: @{ *(.data) @}
4760 This may or may not be the script author's intention for the value of
4761 @code{start_of_data}. One way to influence the orphan section
4762 placement is to assign the location counter to itself, as the linker
4763 assumes that an assignment to @code{.} is setting the start address of
4764 a following output section and thus should be grouped with that
4765 section. So you could write:
4771 .text: @{ *(.text) @}
4776 .data: @{ *(.data) @}
4781 Now, the orphan @code{.rodata} section will be placed between
4782 @code{end_of_text} and @code{start_of_data}.
4786 @subsection Operators
4787 @cindex operators for arithmetic
4788 @cindex arithmetic operators
4789 @cindex precedence in expressions
4790 The linker recognizes the standard C set of arithmetic operators, with
4791 the standard bindings and precedence levels:
4794 @c END TEXI2ROFF-KILL
4796 precedence associativity Operators Notes
4802 5 left == != > < <= >=
4808 11 right &= += -= *= /= (2)
4812 (1) Prefix operators
4813 (2) @xref{Assignments}.
4817 \vskip \baselineskip
4818 %"lispnarrowing" is the extra indent used generally for smallexample
4819 \hskip\lispnarrowing\vbox{\offinterlineskip
4822 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
4823 height2pt&\omit&&\omit&&\omit&\cr
4824 &Precedence&& Associativity &&{\rm Operators}&\cr
4825 height2pt&\omit&&\omit&&\omit&\cr
4827 height2pt&\omit&&\omit&&\omit&\cr
4829 % '176 is tilde, '~' in tt font
4830 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
4831 &2&&left&&* / \%&\cr
4834 &5&&left&&== != > < <= >=&\cr
4837 &8&&left&&{\&\&}&\cr
4840 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
4842 height2pt&\omit&&\omit&&\omit&\cr}
4847 @obeylines@parskip=0pt@parindent=0pt
4848 @dag@quad Prefix operators.
4849 @ddag@quad @xref{Assignments}.
4852 @c END TEXI2ROFF-KILL
4855 @subsection Evaluation
4856 @cindex lazy evaluation
4857 @cindex expression evaluation order
4858 The linker evaluates expressions lazily. It only computes the value of
4859 an expression when absolutely necessary.
4861 The linker needs some information, such as the value of the start
4862 address of the first section, and the origins and lengths of memory
4863 regions, in order to do any linking at all. These values are computed
4864 as soon as possible when the linker reads in the linker script.
4866 However, other values (such as symbol values) are not known or needed
4867 until after storage allocation. Such values are evaluated later, when
4868 other information (such as the sizes of output sections) is available
4869 for use in the symbol assignment expression.
4871 The sizes of sections cannot be known until after allocation, so
4872 assignments dependent upon these are not performed until after
4875 Some expressions, such as those depending upon the location counter
4876 @samp{.}, must be evaluated during section allocation.
4878 If the result of an expression is required, but the value is not
4879 available, then an error results. For example, a script like the
4885 .text 9+this_isnt_constant :
4891 will cause the error message @samp{non constant expression for initial
4894 @node Expression Section
4895 @subsection The Section of an Expression
4896 @cindex expression sections
4897 @cindex absolute expressions
4898 @cindex relative expressions
4899 @cindex absolute and relocatable symbols
4900 @cindex relocatable and absolute symbols
4901 @cindex symbols, relocatable and absolute
4902 When the linker evaluates an expression, the result is either absolute
4903 or relative to some section. A relative expression is expressed as a
4904 fixed offset from the base of a section.
4906 The position of the expression within the linker script determines
4907 whether it is absolute or relative. An expression which appears within
4908 an output section definition is relative to the base of the output
4909 section. An expression which appears elsewhere will be absolute.
4911 A symbol set to a relative expression will be relocatable if you request
4912 relocatable output using the @samp{-r} option. That means that a
4913 further link operation may change the value of the symbol. The symbol's
4914 section will be the section of the relative expression.
4916 A symbol set to an absolute expression will retain the same value
4917 through any further link operation. The symbol will be absolute, and
4918 will not have any particular associated section.
4920 You can use the builtin function @code{ABSOLUTE} to force an expression
4921 to be absolute when it would otherwise be relative. For example, to
4922 create an absolute symbol set to the address of the end of the output
4923 section @samp{.data}:
4927 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
4931 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
4932 @samp{.data} section.
4934 @node Builtin Functions
4935 @subsection Builtin Functions
4936 @cindex functions in expressions
4937 The linker script language includes a number of builtin functions for
4938 use in linker script expressions.
4941 @item ABSOLUTE(@var{exp})
4942 @kindex ABSOLUTE(@var{exp})
4943 @cindex expression, absolute
4944 Return the absolute (non-relocatable, as opposed to non-negative) value
4945 of the expression @var{exp}. Primarily useful to assign an absolute
4946 value to a symbol within a section definition, where symbol values are
4947 normally section relative. @xref{Expression Section}.
4949 @item ADDR(@var{section})
4950 @kindex ADDR(@var{section})
4951 @cindex section address in expression
4952 Return the absolute address (the VMA) of the named @var{section}. Your
4953 script must previously have defined the location of that section. In
4954 the following example, @code{symbol_1} and @code{symbol_2} are assigned
4961 start_of_output_1 = ABSOLUTE(.);
4966 symbol_1 = ADDR(.output1);
4967 symbol_2 = start_of_output_1;
4973 @item ALIGN(@var{align})
4974 @itemx ALIGN(@var{exp},@var{align})
4975 @kindex ALIGN(@var{align})
4976 @kindex ALIGN(@var{exp},@var{align})
4977 @cindex round up location counter
4978 @cindex align location counter
4979 @cindex round up expression
4980 @cindex align expression
4981 Return the location counter (@code{.}) or arbitrary expression aligned
4982 to the next @var{align} boundary. The single operand @code{ALIGN}
4983 doesn't change the value of the location counter---it just does
4984 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
4985 expression to be aligned upwards (@code{ALIGN(@var{align})} is
4986 equivalent to @code{ALIGN(., @var{align})}).
4988 Here is an example which aligns the output @code{.data} section to the
4989 next @code{0x2000} byte boundary after the preceding section and sets a
4990 variable within the section to the next @code{0x8000} boundary after the
4995 .data ALIGN(0x2000): @{
4997 variable = ALIGN(0x8000);
5003 The first use of @code{ALIGN} in this example specifies the location of
5004 a section because it is used as the optional @var{address} attribute of
5005 a section definition (@pxref{Output Section Address}). The second use
5006 of @code{ALIGN} is used to defines the value of a symbol.
5008 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5010 @item BLOCK(@var{exp})
5011 @kindex BLOCK(@var{exp})
5012 This is a synonym for @code{ALIGN}, for compatibility with older linker
5013 scripts. It is most often seen when setting the address of an output
5016 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5017 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5018 This is equivalent to either
5020 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5024 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5027 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5028 for the data segment (area between the result of this expression and
5029 @code{DATA_SEGMENT_END}) than the former or not.
5030 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5031 memory will be saved at the expense of up to @var{commonpagesize} wasted
5032 bytes in the on-disk file.
5034 This expression can only be used directly in @code{SECTIONS} commands, not in
5035 any output section descriptions and only once in the linker script.
5036 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5037 be the system page size the object wants to be optimized for (while still
5038 working on system page sizes up to @var{maxpagesize}).
5043 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5046 @item DATA_SEGMENT_END(@var{exp})
5047 @kindex DATA_SEGMENT_END(@var{exp})
5048 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5049 evaluation purposes.
5052 . = DATA_SEGMENT_END(.);
5055 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5056 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5057 This defines the end of the @code{PT_GNU_RELRO} segment when
5058 @samp{-z relro} option is used. Second argument is returned.
5059 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5060 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5061 @var{exp} + @var{offset} is aligned to the most commonly used page
5062 boundary for particular target. If present in the linker script,
5063 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5064 @code{DATA_SEGMENT_END}.
5067 . = DATA_SEGMENT_RELRO_END(24, .);
5070 @item DEFINED(@var{symbol})
5071 @kindex DEFINED(@var{symbol})
5072 @cindex symbol defaults
5073 Return 1 if @var{symbol} is in the linker global symbol table and is
5074 defined before the statement using DEFINED in the script, otherwise
5075 return 0. You can use this function to provide
5076 default values for symbols. For example, the following script fragment
5077 shows how to set a global symbol @samp{begin} to the first location in
5078 the @samp{.text} section---but if a symbol called @samp{begin} already
5079 existed, its value is preserved:
5085 begin = DEFINED(begin) ? begin : . ;
5093 @item LENGTH(@var{memory})
5094 @kindex LENGTH(@var{memory})
5095 Return the length of the memory region named @var{memory}.
5097 @item LOADADDR(@var{section})
5098 @kindex LOADADDR(@var{section})
5099 @cindex section load address in expression
5100 Return the absolute LMA of the named @var{section}. This is normally
5101 the same as @code{ADDR}, but it may be different if the @code{AT}
5102 attribute is used in the output section definition (@pxref{Output
5106 @item MAX(@var{exp1}, @var{exp2})
5107 Returns the maximum of @var{exp1} and @var{exp2}.
5110 @item MIN(@var{exp1}, @var{exp2})
5111 Returns the minimum of @var{exp1} and @var{exp2}.
5113 @item NEXT(@var{exp})
5114 @kindex NEXT(@var{exp})
5115 @cindex unallocated address, next
5116 Return the next unallocated address that is a multiple of @var{exp}.
5117 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5118 use the @code{MEMORY} command to define discontinuous memory for the
5119 output file, the two functions are equivalent.
5121 @item ORIGIN(@var{memory})
5122 @kindex ORIGIN(@var{memory})
5123 Return the origin of the memory region named @var{memory}.
5125 @item SEGMENT_START(@var{segment}, @var{default})
5126 @kindex SEGMENT_START(@var{segment}, @var{default})
5127 Return the base address of the named @var{segment}. If an explicit
5128 value has been given for this segment (with a command-line @samp{-T}
5129 option) that value will be returned; otherwise the value will be
5130 @var{default}. At present, the @samp{-T} command-line option can only
5131 be used to set the base address for the ``text'', ``data'', and
5132 ``bss'' sections, but you use @code{SEGMENT_START} with any segment
5135 @item SIZEOF(@var{section})
5136 @kindex SIZEOF(@var{section})
5137 @cindex section size
5138 Return the size in bytes of the named @var{section}, if that section has
5139 been allocated. If the section has not been allocated when this is
5140 evaluated, the linker will report an error. In the following example,
5141 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5150 symbol_1 = .end - .start ;
5151 symbol_2 = SIZEOF(.output);
5156 @item SIZEOF_HEADERS
5157 @itemx sizeof_headers
5158 @kindex SIZEOF_HEADERS
5160 Return the size in bytes of the output file's headers. This is
5161 information which appears at the start of the output file. You can use
5162 this number when setting the start address of the first section, if you
5163 choose, to facilitate paging.
5165 @cindex not enough room for program headers
5166 @cindex program headers, not enough room
5167 When producing an ELF output file, if the linker script uses the
5168 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5169 number of program headers before it has determined all the section
5170 addresses and sizes. If the linker later discovers that it needs
5171 additional program headers, it will report an error @samp{not enough
5172 room for program headers}. To avoid this error, you must avoid using
5173 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5174 script to avoid forcing the linker to use additional program headers, or
5175 you must define the program headers yourself using the @code{PHDRS}
5176 command (@pxref{PHDRS}).
5179 @node Implicit Linker Scripts
5180 @section Implicit Linker Scripts
5181 @cindex implicit linker scripts
5182 If you specify a linker input file which the linker can not recognize as
5183 an object file or an archive file, it will try to read the file as a
5184 linker script. If the file can not be parsed as a linker script, the
5185 linker will report an error.
5187 An implicit linker script will not replace the default linker script.
5189 Typically an implicit linker script would contain only symbol
5190 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5193 Any input files read because of an implicit linker script will be read
5194 at the position in the command line where the implicit linker script was
5195 read. This can affect archive searching.
5198 @node Machine Dependent
5199 @chapter Machine Dependent Features
5201 @cindex machine dependencies
5202 @command{ld} has additional features on some platforms; the following
5203 sections describe them. Machines where @command{ld} has no additional
5204 functionality are not listed.
5208 * H8/300:: @command{ld} and the H8/300
5211 * i960:: @command{ld} and the Intel 960 family
5214 * ARM:: @command{ld} and the ARM family
5217 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
5220 * MMIX:: @command{ld} and MMIX
5223 * MSP430:: @command{ld} and MSP430
5226 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
5229 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
5232 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
5235 * TI COFF:: @command{ld} and TI COFF
5238 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
5241 * Xtensa:: @command{ld} and Xtensa Processors
5252 @section @command{ld} and the H8/300
5254 @cindex H8/300 support
5255 For the H8/300, @command{ld} can perform these global optimizations when
5256 you specify the @samp{--relax} command-line option.
5259 @cindex relaxing on H8/300
5260 @item relaxing address modes
5261 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5262 targets are within eight bits, and turns them into eight-bit
5263 program-counter relative @code{bsr} and @code{bra} instructions,
5266 @cindex synthesizing on H8/300
5267 @item synthesizing instructions
5268 @c FIXME: specifically mov.b, or any mov instructions really?
5269 @command{ld} finds all @code{mov.b} instructions which use the
5270 sixteen-bit absolute address form, but refer to the top
5271 page of memory, and changes them to use the eight-bit address form.
5272 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5273 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5274 top page of memory).
5276 @item bit manipulation instructions
5277 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
5278 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5279 which use 32 bit and 16 bit absolute address form, but refer to the top
5280 page of memory, and changes them to use the 8 bit address form.
5281 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5282 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5283 the top page of memory).
5285 @item system control instructions
5286 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
5287 32 bit absolute address form, but refer to the top page of memory, and
5288 changes them to use 16 bit address form.
5289 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5290 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5291 the top page of memory).
5301 @c This stuff is pointless to say unless you're especially concerned
5302 @c with Renesas chips; don't enable it for generic case, please.
5304 @chapter @command{ld} and Other Renesas Chips
5306 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5307 H8/500, and SH chips. No special features, commands, or command-line
5308 options are required for these chips.
5318 @section @command{ld} and the Intel 960 Family
5320 @cindex i960 support
5322 You can use the @samp{-A@var{architecture}} command line option to
5323 specify one of the two-letter names identifying members of the 960
5324 family; the option specifies the desired output target, and warns of any
5325 incompatible instructions in the input files. It also modifies the
5326 linker's search strategy for archive libraries, to support the use of
5327 libraries specific to each particular architecture, by including in the
5328 search loop names suffixed with the string identifying the architecture.
5330 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5331 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5332 paths, and in any paths you specify with @samp{-L}) for a library with
5345 The first two possibilities would be considered in any event; the last
5346 two are due to the use of @w{@samp{-ACA}}.
5348 You can meaningfully use @samp{-A} more than once on a command line, since
5349 the 960 architecture family allows combination of target architectures; each
5350 use will add another pair of name variants to search for when @w{@samp{-l}}
5351 specifies a library.
5353 @cindex @option{--relax} on i960
5354 @cindex relaxing on i960
5355 @command{ld} supports the @samp{--relax} option for the i960 family. If
5356 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5357 @code{calx} instructions whose targets are within 24 bits, and turns
5358 them into 24-bit program-counter relative @code{bal} and @code{cal}
5359 instructions, respectively. @command{ld} also turns @code{cal}
5360 instructions into @code{bal} instructions when it determines that the
5361 target subroutine is a leaf routine (that is, the target subroutine does
5362 not itself call any subroutines).
5379 @node M68HC11/68HC12
5380 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
5382 @cindex M68HC11 and 68HC12 support
5384 @subsection Linker Relaxation
5386 For the Motorola 68HC11, @command{ld} can perform these global
5387 optimizations when you specify the @samp{--relax} command-line option.
5390 @cindex relaxing on M68HC11
5391 @item relaxing address modes
5392 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5393 targets are within eight bits, and turns them into eight-bit
5394 program-counter relative @code{bsr} and @code{bra} instructions,
5397 @command{ld} also looks at all 16-bit extended addressing modes and
5398 transforms them in a direct addressing mode when the address is in
5399 page 0 (between 0 and 0x0ff).
5401 @item relaxing gcc instruction group
5402 When @command{gcc} is called with @option{-mrelax}, it can emit group
5403 of instructions that the linker can optimize to use a 68HC11 direct
5404 addressing mode. These instructions consists of @code{bclr} or
5405 @code{bset} instructions.
5409 @subsection Trampoline Generation
5411 @cindex trampoline generation on M68HC11
5412 @cindex trampoline generation on M68HC12
5413 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5414 call a far function using a normal @code{jsr} instruction. The linker
5415 will also change the relocation to some far function to use the
5416 trampoline address instead of the function address. This is typically the
5417 case when a pointer to a function is taken. The pointer will in fact
5418 point to the function trampoline.
5421 @kindex --pic-veneer
5422 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
5423 ARM/Thumb interworking veneers, even if the rest of the binary
5424 is not PIC. This avoids problems on uClinux targets where
5425 @samp{--emit-relocs} is used to generate relocatable binaries.
5433 @section @command{ld} and the ARM family
5435 @cindex ARM interworking support
5436 @kindex --support-old-code
5437 For the ARM, @command{ld} will generate code stubs to allow functions calls
5438 between ARM and Thumb code. These stubs only work with code that has
5439 been compiled and assembled with the @samp{-mthumb-interwork} command
5440 line option. If it is necessary to link with old ARM object files or
5441 libraries, which have not been compiled with the -mthumb-interwork
5442 option then the @samp{--support-old-code} command line switch should be
5443 given to the linker. This will make it generate larger stub functions
5444 which will work with non-interworking aware ARM code. Note, however,
5445 the linker does not support generating stubs for function calls to
5446 non-interworking aware Thumb code.
5448 @cindex thumb entry point
5449 @cindex entry point, thumb
5450 @kindex --thumb-entry=@var{entry}
5451 The @samp{--thumb-entry} switch is a duplicate of the generic
5452 @samp{--entry} switch, in that it sets the program's starting address.
5453 But it also sets the bottom bit of the address, so that it can be
5454 branched to using a BX instruction, and the program will start
5455 executing in Thumb mode straight away.
5459 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5460 executables. This option is only valid when linking big-endian objects.
5461 The resulting image will contain big-endian data and little-endian code.
5464 @kindex --target1-rel
5465 @kindex --target1-abs
5466 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5467 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
5468 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
5469 and @samp{--target1-abs} switches override the default.
5472 @kindex --target2=@var{type}
5473 The @samp{--target2=type} switch overrides the default definition of the
5474 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
5475 meanings, and target defaults are as follows:
5478 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5480 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
5482 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5487 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5488 specification) enables objects compiled for the ARMv4 architecture to be
5489 interworking-safe when linked with other objects compiled for ARMv4t, but
5490 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
5492 In the latter case, the switch @option{--fix-v4bx} must be passed to the
5493 linker, which causes v4t @code{BX rM} instructions to be rewritten as
5494 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
5496 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
5497 relocations are ignored.
5501 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
5502 BLX instructions (available on ARMv5t and above) in various
5503 situations. Currently it is used to perform calls via the PLT from Thumb
5504 code using BLX rather than using BX and a mode-switching stub before
5505 each PLT entry. This should lead to such calls executing slightly faster.
5507 This option is enabled implicitly for SymbianOS, so there is no need to
5508 specify it if you are using that target.
5510 @cindex VFP11_DENORM_FIX
5511 @kindex --vfp11-denorm-fix
5512 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
5513 bug in certain VFP11 coprocessor hardware, which sometimes allows
5514 instructions with denorm operands (which must be handled by support code)
5515 to have those operands overwritten by subsequent instructions before
5516 the support code can read the intended values.
5518 The bug may be avoided in scalar mode if you allow at least one
5519 intervening instruction between a VFP11 instruction which uses a register
5520 and another instruction which writes to the same register, or at least two
5521 intervening instructions if vector mode is in use. The bug only affects
5522 full-compliance floating-point mode: you do not need this workaround if
5523 you are using "runfast" mode. Please contact ARM for further details.
5525 If you know you are using buggy VFP11 hardware, you can
5526 enable this workaround by specifying the linker option
5527 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
5528 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
5529 vector mode (the latter also works for scalar code). The default is
5530 @samp{--vfp-denorm-fix=none}.
5532 If the workaround is enabled, instructions are scanned for
5533 potentially-troublesome sequences, and a veneer is created for each
5534 such sequence which may trigger the erratum. The veneer consists of the
5535 first instruction of the sequence and a branch back to the subsequent
5536 instruction. The original instruction is then replaced with a branch to
5537 the veneer. The extra cycles required to call and return from the veneer
5538 are sufficient to avoid the erratum in both the scalar and vector cases.
5540 @cindex NO_ENUM_SIZE_WARNING
5541 @kindex --no-enum-size-warning
5542 The @samp{--no-enum-size-warning} switch prevents the linker from
5543 warning when linking object files that specify incompatible EABI
5544 enumeration size attributes. For example, with this switch enabled,
5545 linking of an object file using 32-bit enumeration values with another
5546 using enumeration values fitted into the smallest possible space will
5560 @section @command{ld} and HPPA 32-bit ELF Support
5561 @cindex HPPA multiple sub-space stubs
5562 @kindex --multi-subspace
5563 When generating a shared library, @command{ld} will by default generate
5564 import stubs suitable for use with a single sub-space application.
5565 The @samp{--multi-subspace} switch causes @command{ld} to generate export
5566 stubs, and different (larger) import stubs suitable for use with
5567 multiple sub-spaces.
5569 @cindex HPPA stub grouping
5570 @kindex --stub-group-size=@var{N}
5571 Long branch stubs and import/export stubs are placed by @command{ld} in
5572 stub sections located between groups of input sections.
5573 @samp{--stub-group-size} specifies the maximum size of a group of input
5574 sections handled by one stub section. Since branch offsets are signed,
5575 a stub section may serve two groups of input sections, one group before
5576 the stub section, and one group after it. However, when using
5577 conditional branches that require stubs, it may be better (for branch
5578 prediction) that stub sections only serve one group of input sections.
5579 A negative value for @samp{N} chooses this scheme, ensuring that
5580 branches to stubs always use a negative offset. Two special values of
5581 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5582 @command{ld} to automatically size input section groups for the branch types
5583 detected, with the same behaviour regarding stub placement as other
5584 positive or negative values of @samp{N} respectively.
5586 Note that @samp{--stub-group-size} does not split input sections. A
5587 single input section larger than the group size specified will of course
5588 create a larger group (of one section). If input sections are too
5589 large, it may not be possible for a branch to reach its stub.
5602 @section @code{ld} and MMIX
5603 For MMIX, there is a choice of generating @code{ELF} object files or
5604 @code{mmo} object files when linking. The simulator @code{mmix}
5605 understands the @code{mmo} format. The binutils @code{objcopy} utility
5606 can translate between the two formats.
5608 There is one special section, the @samp{.MMIX.reg_contents} section.
5609 Contents in this section is assumed to correspond to that of global
5610 registers, and symbols referring to it are translated to special symbols,
5611 equal to registers. In a final link, the start address of the
5612 @samp{.MMIX.reg_contents} section corresponds to the first allocated
5613 global register multiplied by 8. Register @code{$255} is not included in
5614 this section; it is always set to the program entry, which is at the
5615 symbol @code{Main} for @code{mmo} files.
5617 Symbols with the prefix @code{__.MMIX.start.}, for example
5618 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special;
5619 there must be only one each, even if they are local. The default linker
5620 script uses these to set the default start address of a section.
5622 Initial and trailing multiples of zero-valued 32-bit words in a section,
5623 are left out from an mmo file.
5636 @section @code{ld} and MSP430
5637 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
5638 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
5639 just pass @samp{-m help} option to the linker).
5641 @cindex MSP430 extra sections
5642 The linker will recognize some extra sections which are MSP430 specific:
5645 @item @samp{.vectors}
5646 Defines a portion of ROM where interrupt vectors located.
5648 @item @samp{.bootloader}
5649 Defines the bootloader portion of the ROM (if applicable). Any code
5650 in this section will be uploaded to the MPU.
5652 @item @samp{.infomem}
5653 Defines an information memory section (if applicable). Any code in
5654 this section will be uploaded to the MPU.
5656 @item @samp{.infomemnobits}
5657 This is the same as the @samp{.infomem} section except that any code
5658 in this section will not be uploaded to the MPU.
5660 @item @samp{.noinit}
5661 Denotes a portion of RAM located above @samp{.bss} section.
5663 The last two sections are used by gcc.
5677 @section @command{ld} and PowerPC 32-bit ELF Support
5678 @cindex PowerPC long branches
5679 @kindex --relax on PowerPC
5680 Branches on PowerPC processors are limited to a signed 26-bit
5681 displacement, which may result in @command{ld} giving
5682 @samp{relocation truncated to fit} errors with very large programs.
5683 @samp{--relax} enables the generation of trampolines that can access
5684 the entire 32-bit address space. These trampolines are inserted at
5685 section boundaries, so may not themselves be reachable if an input
5686 section exceeds 33M in size.
5688 @cindex PowerPC ELF32 options
5693 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
5694 generates code capable of using a newer PLT and GOT layout that has
5695 the security advantage of no executable section ever needing to be
5696 writable and no writable section ever being executable. PowerPC
5697 @command{ld} will generate this layout, including stubs to access the
5698 PLT, if all input files (including startup and static libraries) were
5699 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
5700 BSS PLT (and GOT layout) which can give slightly better performance.
5705 The new secure PLT and GOT are placed differently relative to other
5706 sections compared to older BSS PLT and GOT placement. The location of
5707 @code{.plt} must change because the new secure PLT is an initialized
5708 section while the old PLT is uninitialized. The reason for the
5709 @code{.got} change is more subtle: The new placement allows
5710 @code{.got} to be read-only in applications linked with
5711 @samp{-z relro -z now}. However, this placement means that
5712 @code{.sdata} cannot always be used in shared libraries, because the
5713 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
5714 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
5715 GCC doesn't use @code{.sdata} in shared libraries, so this option is
5716 really only useful for other compilers that may do so.
5718 @cindex PowerPC stub symbols
5719 @kindex --emit-stub-syms
5720 @item --emit-stub-syms
5721 This option causes @command{ld} to label linker stubs with a local
5722 symbol that encodes the stub type and destination.
5724 @cindex PowerPC TLS optimization
5725 @kindex --no-tls-optimize
5726 @item --no-tls-optimize
5727 PowerPC @command{ld} normally performs some optimization of code
5728 sequences used to access Thread-Local Storage. Use this option to
5729 disable the optimization.
5742 @node PowerPC64 ELF64
5743 @section @command{ld} and PowerPC64 64-bit ELF Support
5745 @cindex PowerPC64 ELF64 options
5747 @cindex PowerPC64 stub grouping
5748 @kindex --stub-group-size
5749 @item --stub-group-size
5750 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
5751 by @command{ld} in stub sections located between groups of input sections.
5752 @samp{--stub-group-size} specifies the maximum size of a group of input
5753 sections handled by one stub section. Since branch offsets are signed,
5754 a stub section may serve two groups of input sections, one group before
5755 the stub section, and one group after it. However, when using
5756 conditional branches that require stubs, it may be better (for branch
5757 prediction) that stub sections only serve one group of input sections.
5758 A negative value for @samp{N} chooses this scheme, ensuring that
5759 branches to stubs always use a negative offset. Two special values of
5760 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5761 @command{ld} to automatically size input section groups for the branch types
5762 detected, with the same behaviour regarding stub placement as other
5763 positive or negative values of @samp{N} respectively.
5765 Note that @samp{--stub-group-size} does not split input sections. A
5766 single input section larger than the group size specified will of course
5767 create a larger group (of one section). If input sections are too
5768 large, it may not be possible for a branch to reach its stub.
5770 @cindex PowerPC64 stub symbols
5771 @kindex --emit-stub-syms
5772 @item --emit-stub-syms
5773 This option causes @command{ld} to label linker stubs with a local
5774 symbol that encodes the stub type and destination.
5776 @cindex PowerPC64 dot symbols
5778 @kindex --no-dotsyms
5779 @item --dotsyms, --no-dotsyms
5780 These two options control how @command{ld} interprets version patterns
5781 in a version script. Older PowerPC64 compilers emitted both a
5782 function descriptor symbol with the same name as the function, and a
5783 code entry symbol with the name prefixed by a dot (@samp{.}). To
5784 properly version a function @samp{foo}, the version script thus needs
5785 to control both @samp{foo} and @samp{.foo}. The option
5786 @samp{--dotsyms}, on by default, automatically adds the required
5787 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
5790 @cindex PowerPC64 TLS optimization
5791 @kindex --no-tls-optimize
5792 @item --no-tls-optimize
5793 PowerPC64 @command{ld} normally performs some optimization of code
5794 sequences used to access Thread-Local Storage. Use this option to
5795 disable the optimization.
5797 @cindex PowerPC64 OPD optimization
5798 @kindex --no-opd-optimize
5799 @item --no-opd-optimize
5800 PowerPC64 @command{ld} normally removes @code{.opd} section entries
5801 corresponding to deleted link-once functions, or functions removed by
5802 the action of @samp{--gc-sections} or linker scrip @code{/DISCARD/}.
5803 Use this option to disable @code{.opd} optimization.
5805 @cindex PowerPC64 OPD spacing
5806 @kindex --non-overlapping-opd
5807 @item --non-overlapping-opd
5808 Some PowerPC64 compilers have an option to generate compressed
5809 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
5810 the static chain pointer (unused in C) with the first word of the next
5811 entry. This option expands such entries to the full 24 bytes.
5813 @cindex PowerPC64 TOC optimization
5814 @kindex --no-toc-optimize
5815 @item --no-toc-optimize
5816 PowerPC64 @command{ld} normally removes unused @code{.toc} section
5817 entries. Such entries are detected by examining relocations that
5818 reference the TOC in code sections. A reloc in a deleted code section
5819 marks a TOC word as unneeded, while a reloc in a kept code section
5820 marks a TOC word as needed. Since the TOC may reference itself, TOC
5821 relocs are also examined. TOC words marked as both needed and
5822 unneeded will of course be kept. TOC words without any referencing
5823 reloc are assumed to be part of a multi-word entry, and are kept or
5824 discarded as per the nearest marked preceding word. This works
5825 reliably for compiler generated code, but may be incorrect if assembly
5826 code is used to insert TOC entries. Use this option to disable the
5829 @cindex PowerPC64 multi-TOC
5830 @kindex --no-multi-toc
5831 @item --no-multi-toc
5832 By default, PowerPC64 GCC generates code for a TOC model where TOC
5833 entries are accessed with a 16-bit offset from r2. This limits the
5834 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
5835 grouping code sections such that each group uses less than 64K for its
5836 TOC entries, then inserts r2 adjusting stubs between inter-group
5837 calls. @command{ld} does not split apart input sections, so cannot
5838 help if a single input file has a @code{.toc} section that exceeds
5839 64K, most likely from linking multiple files with @command{ld -r}.
5840 Use this option to turn off this feature.
5854 @section @command{ld}'s Support for Various TI COFF Versions
5855 @cindex TI COFF versions
5856 @kindex --format=@var{version}
5857 The @samp{--format} switch allows selection of one of the various
5858 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
5859 also supported. The TI COFF versions also vary in header byte-order
5860 format; @command{ld} will read any version or byte order, but the output
5861 header format depends on the default specified by the specific target.
5874 @section @command{ld} and WIN32 (cygwin/mingw)
5876 This section describes some of the win32 specific @command{ld} issues.
5877 See @ref{Options,,Command Line Options} for detailed description of the
5878 command line options mentioned here.
5881 @cindex import libraries
5882 @item import libraries
5883 The standard Windows linker creates and uses so-called import
5884 libraries, which contains information for linking to dll's. They are
5885 regular static archives and are handled as any other static
5886 archive. The cygwin and mingw ports of @command{ld} have specific
5887 support for creating such libraries provided with the
5888 @samp{--out-implib} command line option.
5890 @item exporting DLL symbols
5891 @cindex exporting DLL symbols
5892 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
5895 @item using auto-export functionality
5896 @cindex using auto-export functionality
5897 By default @command{ld} exports symbols with the auto-export functionality,
5898 which is controlled by the following command line options:
5901 @item --export-all-symbols [This is the default]
5902 @item --exclude-symbols
5903 @item --exclude-libs
5906 If, however, @samp{--export-all-symbols} is not given explicitly on the
5907 command line, then the default auto-export behavior will be @emph{disabled}
5908 if either of the following are true:
5911 @item A DEF file is used.
5912 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
5915 @item using a DEF file
5916 @cindex using a DEF file
5917 Another way of exporting symbols is using a DEF file. A DEF file is
5918 an ASCII file containing definitions of symbols which should be
5919 exported when a dll is created. Usually it is named @samp{<dll
5920 name>.def} and is added as any other object file to the linker's
5921 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
5924 gcc -o <output> <objectfiles> <dll name>.def
5927 Using a DEF file turns off the normal auto-export behavior, unless the
5928 @samp{--export-all-symbols} option is also used.
5930 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
5933 LIBRARY "xyz.dll" BASE=0x20000000
5939 another_foo = abc.dll.afoo
5943 This example defines a DLL with a non-default base address and five
5944 symbols in the export table. The third exported symbol @code{_bar} is an
5945 alias for the second. The fourth symbol, @code{another_foo} is resolved
5946 by "forwarding" to another module and treating it as an alias for
5947 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
5948 @code{var1} is declared to be a data object.
5950 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
5951 name of the output DLL. If @samp{<name>} does not include a suffix,
5952 the default library suffix, @samp{.DLL} is appended.
5954 When the .DEF file is used to build an application, rather than a
5955 library, the @code{NAME <name>} command should be used instead of
5956 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
5957 executable suffix, @samp{.EXE} is appended.
5959 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
5960 specification @code{BASE = <number>} may be used to specify a
5961 non-default base address for the image.
5963 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
5964 or they specify an empty string, the internal name is the same as the
5965 filename specified on the command line.
5967 The complete specification of an export symbol is:
5971 ( ( ( <name1> [ = <name2> ] )
5972 | ( <name1> = <module-name> . <external-name>))
5973 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
5976 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
5977 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
5978 @samp{<name1>} as a "forward" alias for the symbol
5979 @samp{<external-name>} in the DLL @samp{<module-name>}.
5980 Optionally, the symbol may be exported by the specified ordinal
5981 @samp{<integer>} alias.
5983 The optional keywords that follow the declaration indicate:
5985 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
5986 will still be exported by its ordinal alias (either the value specified
5987 by the .def specification or, otherwise, the value assigned by the
5988 linker). The symbol name, however, does remain visible in the import
5989 library (if any), unless @code{PRIVATE} is also specified.
5991 @code{DATA}: The symbol is a variable or object, rather than a function.
5992 The import lib will export only an indirect reference to @code{foo} as
5993 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
5996 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
5997 well as @code{_imp__foo} into the import library. Both refer to the
5998 read-only import address table's pointer to the variable, not to the
5999 variable itself. This can be dangerous. If the user code fails to add
6000 the @code{dllimport} attribute and also fails to explicitly add the
6001 extra indirection that the use of the attribute enforces, the
6002 application will behave unexpectedly.
6004 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
6005 it into the static import library used to resolve imports at link time. The
6006 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
6007 API at runtime or by by using the GNU ld extension of linking directly to
6008 the DLL without an import library.
6010 See ld/deffilep.y in the binutils sources for the full specification of
6011 other DEF file statements
6013 @cindex creating a DEF file
6014 While linking a shared dll, @command{ld} is able to create a DEF file
6015 with the @samp{--output-def <file>} command line option.
6017 @item Using decorations
6018 @cindex Using decorations
6019 Another way of marking symbols for export is to modify the source code
6020 itself, so that when building the DLL each symbol to be exported is
6024 __declspec(dllexport) int a_variable
6025 __declspec(dllexport) void a_function(int with_args)
6028 All such symbols will be exported from the DLL. If, however,
6029 any of the object files in the DLL contain symbols decorated in
6030 this way, then the normal auto-export behavior is disabled, unless
6031 the @samp{--export-all-symbols} option is also used.
6033 Note that object files that wish to access these symbols must @emph{not}
6034 decorate them with dllexport. Instead, they should use dllimport,
6038 __declspec(dllimport) int a_variable
6039 __declspec(dllimport) void a_function(int with_args)
6042 This complicates the structure of library header files, because
6043 when included by the library itself the header must declare the
6044 variables and functions as dllexport, but when included by client
6045 code the header must declare them as dllimport. There are a number
6046 of idioms that are typically used to do this; often client code can
6047 omit the __declspec() declaration completely. See
6048 @samp{--enable-auto-import} and @samp{automatic data imports} for more
6052 @cindex automatic data imports
6053 @item automatic data imports
6054 The standard Windows dll format supports data imports from dlls only
6055 by adding special decorations (dllimport/dllexport), which let the
6056 compiler produce specific assembler instructions to deal with this
6057 issue. This increases the effort necessary to port existing Un*x
6058 code to these platforms, especially for large
6059 c++ libraries and applications. The auto-import feature, which was
6060 initially provided by Paul Sokolovsky, allows one to omit the
6061 decorations to achieve a behavior that conforms to that on POSIX/Un*x
6062 platforms. This feature is enabled with the @samp{--enable-auto-import}
6063 command-line option, although it is enabled by default on cygwin/mingw.
6064 The @samp{--enable-auto-import} option itself now serves mainly to
6065 suppress any warnings that are ordinarily emitted when linked objects
6066 trigger the feature's use.
6068 auto-import of variables does not always work flawlessly without
6069 additional assistance. Sometimes, you will see this message
6071 "variable '<var>' can't be auto-imported. Please read the
6072 documentation for ld's @code{--enable-auto-import} for details."
6074 The @samp{--enable-auto-import} documentation explains why this error
6075 occurs, and several methods that can be used to overcome this difficulty.
6076 One of these methods is the @emph{runtime pseudo-relocs} feature, described
6079 @cindex runtime pseudo-relocation
6080 For complex variables imported from DLLs (such as structs or classes),
6081 object files typically contain a base address for the variable and an
6082 offset (@emph{addend}) within the variable--to specify a particular
6083 field or public member, for instance. Unfortunately, the runtime loader used
6084 in win32 environments is incapable of fixing these references at runtime
6085 without the additional information supplied by dllimport/dllexport decorations.
6086 The standard auto-import feature described above is unable to resolve these
6089 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
6090 be resolved without error, while leaving the task of adjusting the references
6091 themselves (with their non-zero addends) to specialized code provided by the
6092 runtime environment. Recent versions of the cygwin and mingw environments and
6093 compilers provide this runtime support; older versions do not. However, the
6094 support is only necessary on the developer's platform; the compiled result will
6095 run without error on an older system.
6097 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
6100 @cindex direct linking to a dll
6101 @item direct linking to a dll
6102 The cygwin/mingw ports of @command{ld} support the direct linking,
6103 including data symbols, to a dll without the usage of any import
6104 libraries. This is much faster and uses much less memory than does the
6105 traditional import library method, especially when linking large
6106 libraries or applications. When @command{ld} creates an import lib, each
6107 function or variable exported from the dll is stored in its own bfd, even
6108 though a single bfd could contain many exports. The overhead involved in
6109 storing, loading, and processing so many bfd's is quite large, and explains the
6110 tremendous time, memory, and storage needed to link against particularly
6111 large or complex libraries when using import libs.
6113 Linking directly to a dll uses no extra command-line switches other than
6114 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
6115 of names to match each library. All that is needed from the developer's
6116 perspective is an understanding of this search, in order to force ld to
6117 select the dll instead of an import library.
6120 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
6121 to find, in the first directory of its search path,
6133 before moving on to the next directory in the search path.
6135 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
6136 where @samp{<prefix>} is set by the @command{ld} option
6137 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6138 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6141 Other win32-based unix environments, such as mingw or pw32, may use other
6142 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
6143 was originally intended to help avoid name conflicts among dll's built for the
6144 various win32/un*x environments, so that (for example) two versions of a zlib dll
6145 could coexist on the same machine.
6147 The generic cygwin/mingw path layout uses a @samp{bin} directory for
6148 applications and dll's and a @samp{lib} directory for the import
6149 libraries (using cygwin nomenclature):
6155 libxxx.dll.a (in case of dll's)
6156 libxxx.a (in case of static archive)
6159 Linking directly to a dll without using the import library can be
6162 1. Use the dll directly by adding the @samp{bin} path to the link line
6164 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6167 However, as the dll's often have version numbers appended to their names
6168 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6169 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
6170 not versioned, and do not have this difficulty.
6172 2. Create a symbolic link from the dll to a file in the @samp{lib}
6173 directory according to the above mentioned search pattern. This
6174 should be used to avoid unwanted changes in the tools needed for
6178 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6181 Then you can link without any make environment changes.
6184 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6187 This technique also avoids the version number problems, because the following is
6194 libxxx.dll.a -> ../bin/cygxxx-5.dll
6197 Linking directly to a dll without using an import lib will work
6198 even when auto-import features are exercised, and even when
6199 @samp{--enable-runtime-pseudo-relocs} is used.
6201 Given the improvements in speed and memory usage, one might justifiably
6202 wonder why import libraries are used at all. There are three reasons:
6204 1. Until recently, the link-directly-to-dll functionality did @emph{not}
6205 work with auto-imported data.
6207 2. Sometimes it is necessary to include pure static objects within the
6208 import library (which otherwise contains only bfd's for indirection
6209 symbols that point to the exports of a dll). Again, the import lib
6210 for the cygwin kernel makes use of this ability, and it is not
6211 possible to do this without an import lib.
6213 3. Symbol aliases can only be resolved using an import lib. This is
6214 critical when linking against OS-supplied dll's (eg, the win32 API)
6215 in which symbols are usually exported as undecorated aliases of their
6216 stdcall-decorated assembly names.
6218 So, import libs are not going away. But the ability to replace
6219 true import libs with a simple symbolic link to (or a copy of)
6220 a dll, in many cases, is a useful addition to the suite of tools
6221 binutils makes available to the win32 developer. Given the
6222 massive improvements in memory requirements during linking, storage
6223 requirements, and linking speed, we expect that many developers
6224 will soon begin to use this feature whenever possible.
6226 @item symbol aliasing
6228 @item adding additional names
6229 Sometimes, it is useful to export symbols with additional names.
6230 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6231 exported as @samp{_foo} by using special directives in the DEF file
6232 when creating the dll. This will affect also the optional created
6233 import library. Consider the following DEF file:
6236 LIBRARY "xyz.dll" BASE=0x61000000
6243 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6245 Another method for creating a symbol alias is to create it in the
6246 source code using the "weak" attribute:
6249 void foo () @{ /* Do something. */; @}
6250 void _foo () __attribute__ ((weak, alias ("foo")));
6253 See the gcc manual for more information about attributes and weak
6256 @item renaming symbols
6257 Sometimes it is useful to rename exports. For instance, the cygwin
6258 kernel does this regularly. A symbol @samp{_foo} can be exported as
6259 @samp{foo} but not as @samp{_foo} by using special directives in the
6260 DEF file. (This will also affect the import library, if it is
6261 created). In the following example:
6264 LIBRARY "xyz.dll" BASE=0x61000000
6270 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6274 Note: using a DEF file disables the default auto-export behavior,
6275 unless the @samp{--export-all-symbols} command line option is used.
6276 If, however, you are trying to rename symbols, then you should list
6277 @emph{all} desired exports in the DEF file, including the symbols
6278 that are not being renamed, and do @emph{not} use the
6279 @samp{--export-all-symbols} option. If you list only the
6280 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6281 to handle the other symbols, then the both the new names @emph{and}
6282 the original names for the renamed symbols will be exported.
6283 In effect, you'd be aliasing those symbols, not renaming them,
6284 which is probably not what you wanted.
6286 @cindex weak externals
6287 @item weak externals
6288 The Windows object format, PE, specifies a form of weak symbols called
6289 weak externals. When a weak symbol is linked and the symbol is not
6290 defined, the weak symbol becomes an alias for some other symbol. There
6291 are three variants of weak externals:
6293 @item Definition is searched for in objects and libraries, historically
6294 called lazy externals.
6295 @item Definition is searched for only in other objects, not in libraries.
6296 This form is not presently implemented.
6297 @item No search; the symbol is an alias. This form is not presently
6300 As a GNU extension, weak symbols that do not specify an alternate symbol
6301 are supported. If the symbol is undefined when linking, the symbol
6302 uses a default value.
6316 @section @code{ld} and Xtensa Processors
6318 @cindex Xtensa processors
6319 The default @command{ld} behavior for Xtensa processors is to interpret
6320 @code{SECTIONS} commands so that lists of explicitly named sections in a
6321 specification with a wildcard file will be interleaved when necessary to
6322 keep literal pools within the range of PC-relative load offsets. For
6323 example, with the command:
6335 @command{ld} may interleave some of the @code{.literal}
6336 and @code{.text} sections from different object files to ensure that the
6337 literal pools are within the range of PC-relative load offsets. A valid
6338 interleaving might place the @code{.literal} sections from an initial
6339 group of files followed by the @code{.text} sections of that group of
6340 files. Then, the @code{.literal} sections from the rest of the files
6341 and the @code{.text} sections from the rest of the files would follow.
6343 @cindex @option{--relax} on Xtensa
6344 @cindex relaxing on Xtensa
6345 Relaxation is enabled by default for the Xtensa version of @command{ld} and
6346 provides two important link-time optimizations. The first optimization
6347 is to combine identical literal values to reduce code size. A redundant
6348 literal will be removed and all the @code{L32R} instructions that use it
6349 will be changed to reference an identical literal, as long as the
6350 location of the replacement literal is within the offset range of all
6351 the @code{L32R} instructions. The second optimization is to remove
6352 unnecessary overhead from assembler-generated ``longcall'' sequences of
6353 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
6354 range of direct @code{CALL@var{n}} instructions.
6356 For each of these cases where an indirect call sequence can be optimized
6357 to a direct call, the linker will change the @code{CALLX@var{n}}
6358 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
6359 instruction, and remove the literal referenced by the @code{L32R}
6360 instruction if it is not used for anything else. Removing the
6361 @code{L32R} instruction always reduces code size but can potentially
6362 hurt performance by changing the alignment of subsequent branch targets.
6363 By default, the linker will always preserve alignments, either by
6364 switching some instructions between 24-bit encodings and the equivalent
6365 density instructions or by inserting a no-op in place of the @code{L32R}
6366 instruction that was removed. If code size is more important than
6367 performance, the @option{--size-opt} option can be used to prevent the
6368 linker from widening density instructions or inserting no-ops, except in
6369 a few cases where no-ops are required for correctness.
6371 The following Xtensa-specific command-line options can be used to
6374 @cindex Xtensa options
6378 Since the Xtensa version of @code{ld} enables the @option{--relax} option
6379 by default, the @option{--no-relax} option is provided to disable
6383 When optimizing indirect calls to direct calls, optimize for code size
6384 more than performance. With this option, the linker will not insert
6385 no-ops or widen density instructions to preserve branch target
6386 alignment. There may still be some cases where no-ops are required to
6387 preserve the correctness of the code.
6395 @ifclear SingleFormat
6400 @cindex object file management
6401 @cindex object formats available
6403 The linker accesses object and archive files using the BFD libraries.
6404 These libraries allow the linker to use the same routines to operate on
6405 object files whatever the object file format. A different object file
6406 format can be supported simply by creating a new BFD back end and adding
6407 it to the library. To conserve runtime memory, however, the linker and
6408 associated tools are usually configured to support only a subset of the
6409 object file formats available. You can use @code{objdump -i}
6410 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
6411 list all the formats available for your configuration.
6413 @cindex BFD requirements
6414 @cindex requirements for BFD
6415 As with most implementations, BFD is a compromise between
6416 several conflicting requirements. The major factor influencing
6417 BFD design was efficiency: any time used converting between
6418 formats is time which would not have been spent had BFD not
6419 been involved. This is partly offset by abstraction payback; since
6420 BFD simplifies applications and back ends, more time and care
6421 may be spent optimizing algorithms for a greater speed.
6423 One minor artifact of the BFD solution which you should bear in
6424 mind is the potential for information loss. There are two places where
6425 useful information can be lost using the BFD mechanism: during
6426 conversion and during output. @xref{BFD information loss}.
6429 * BFD outline:: How it works: an outline of BFD
6433 @section How It Works: An Outline of BFD
6434 @cindex opening object files
6435 @include bfdsumm.texi
6438 @node Reporting Bugs
6439 @chapter Reporting Bugs
6440 @cindex bugs in @command{ld}
6441 @cindex reporting bugs in @command{ld}
6443 Your bug reports play an essential role in making @command{ld} reliable.
6445 Reporting a bug may help you by bringing a solution to your problem, or
6446 it may not. But in any case the principal function of a bug report is
6447 to help the entire community by making the next version of @command{ld}
6448 work better. Bug reports are your contribution to the maintenance of
6451 In order for a bug report to serve its purpose, you must include the
6452 information that enables us to fix the bug.
6455 * Bug Criteria:: Have you found a bug?
6456 * Bug Reporting:: How to report bugs
6460 @section Have You Found a Bug?
6461 @cindex bug criteria
6463 If you are not sure whether you have found a bug, here are some guidelines:
6466 @cindex fatal signal
6467 @cindex linker crash
6468 @cindex crash of linker
6470 If the linker gets a fatal signal, for any input whatever, that is a
6471 @command{ld} bug. Reliable linkers never crash.
6473 @cindex error on valid input
6475 If @command{ld} produces an error message for valid input, that is a bug.
6477 @cindex invalid input
6479 If @command{ld} does not produce an error message for invalid input, that
6480 may be a bug. In the general case, the linker can not verify that
6481 object files are correct.
6484 If you are an experienced user of linkers, your suggestions for
6485 improvement of @command{ld} are welcome in any case.
6489 @section How to Report Bugs
6491 @cindex @command{ld} bugs, reporting
6493 A number of companies and individuals offer support for @sc{gnu}
6494 products. If you obtained @command{ld} from a support organization, we
6495 recommend you contact that organization first.
6497 You can find contact information for many support companies and
6498 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
6502 Otherwise, send bug reports for @command{ld} to
6506 The fundamental principle of reporting bugs usefully is this:
6507 @strong{report all the facts}. If you are not sure whether to state a
6508 fact or leave it out, state it!
6510 Often people omit facts because they think they know what causes the
6511 problem and assume that some details do not matter. Thus, you might
6512 assume that the name of a symbol you use in an example does not
6513 matter. Well, probably it does not, but one cannot be sure. Perhaps
6514 the bug is a stray memory reference which happens to fetch from the
6515 location where that name is stored in memory; perhaps, if the name
6516 were different, the contents of that location would fool the linker
6517 into doing the right thing despite the bug. Play it safe and give a
6518 specific, complete example. That is the easiest thing for you to do,
6519 and the most helpful.
6521 Keep in mind that the purpose of a bug report is to enable us to fix
6522 the bug if it is new to us. Therefore, always write your bug reports
6523 on the assumption that the bug has not been reported previously.
6525 Sometimes people give a few sketchy facts and ask, ``Does this ring a
6526 bell?'' This cannot help us fix a bug, so it is basically useless. We
6527 respond by asking for enough details to enable us to investigate.
6528 You might as well expedite matters by sending them to begin with.
6530 To enable us to fix the bug, you should include all these things:
6534 The version of @command{ld}. @command{ld} announces it if you start it with
6535 the @samp{--version} argument.
6537 Without this, we will not know whether there is any point in looking for
6538 the bug in the current version of @command{ld}.
6541 Any patches you may have applied to the @command{ld} source, including any
6542 patches made to the @code{BFD} library.
6545 The type of machine you are using, and the operating system name and
6549 What compiler (and its version) was used to compile @command{ld}---e.g.
6553 The command arguments you gave the linker to link your example and
6554 observe the bug. To guarantee you will not omit something important,
6555 list them all. A copy of the Makefile (or the output from make) is
6558 If we were to try to guess the arguments, we would probably guess wrong
6559 and then we might not encounter the bug.
6562 A complete input file, or set of input files, that will reproduce the
6563 bug. It is generally most helpful to send the actual object files
6564 provided that they are reasonably small. Say no more than 10K. For
6565 bigger files you can either make them available by FTP or HTTP or else
6566 state that you are willing to send the object file(s) to whomever
6567 requests them. (Note - your email will be going to a mailing list, so
6568 we do not want to clog it up with large attachments). But small
6569 attachments are best.
6571 If the source files were assembled using @code{gas} or compiled using
6572 @code{gcc}, then it may be OK to send the source files rather than the
6573 object files. In this case, be sure to say exactly what version of
6574 @code{gas} or @code{gcc} was used to produce the object files. Also say
6575 how @code{gas} or @code{gcc} were configured.
6578 A description of what behavior you observe that you believe is
6579 incorrect. For example, ``It gets a fatal signal.''
6581 Of course, if the bug is that @command{ld} gets a fatal signal, then we
6582 will certainly notice it. But if the bug is incorrect output, we might
6583 not notice unless it is glaringly wrong. You might as well not give us
6584 a chance to make a mistake.
6586 Even if the problem you experience is a fatal signal, you should still
6587 say so explicitly. Suppose something strange is going on, such as, your
6588 copy of @command{ld} is out of sync, or you have encountered a bug in the
6589 C library on your system. (This has happened!) Your copy might crash
6590 and ours would not. If you told us to expect a crash, then when ours
6591 fails to crash, we would know that the bug was not happening for us. If
6592 you had not told us to expect a crash, then we would not be able to draw
6593 any conclusion from our observations.
6596 If you wish to suggest changes to the @command{ld} source, send us context
6597 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
6598 @samp{-p} option. Always send diffs from the old file to the new file.
6599 If you even discuss something in the @command{ld} source, refer to it by
6600 context, not by line number.
6602 The line numbers in our development sources will not match those in your
6603 sources. Your line numbers would convey no useful information to us.
6606 Here are some things that are not necessary:
6610 A description of the envelope of the bug.
6612 Often people who encounter a bug spend a lot of time investigating
6613 which changes to the input file will make the bug go away and which
6614 changes will not affect it.
6616 This is often time consuming and not very useful, because the way we
6617 will find the bug is by running a single example under the debugger
6618 with breakpoints, not by pure deduction from a series of examples.
6619 We recommend that you save your time for something else.
6621 Of course, if you can find a simpler example to report @emph{instead}
6622 of the original one, that is a convenience for us. Errors in the
6623 output will be easier to spot, running under the debugger will take
6624 less time, and so on.
6626 However, simplification is not vital; if you do not want to do this,
6627 report the bug anyway and send us the entire test case you used.
6630 A patch for the bug.
6632 A patch for the bug does help us if it is a good one. But do not omit
6633 the necessary information, such as the test case, on the assumption that
6634 a patch is all we need. We might see problems with your patch and decide
6635 to fix the problem another way, or we might not understand it at all.
6637 Sometimes with a program as complicated as @command{ld} it is very hard to
6638 construct an example that will make the program follow a certain path
6639 through the code. If you do not send us the example, we will not be
6640 able to construct one, so we will not be able to verify that the bug is
6643 And if we cannot understand what bug you are trying to fix, or why your
6644 patch should be an improvement, we will not install it. A test case will
6645 help us to understand.
6648 A guess about what the bug is or what it depends on.
6650 Such guesses are usually wrong. Even we cannot guess right about such
6651 things without first using the debugger to find the facts.
6655 @appendix MRI Compatible Script Files
6656 @cindex MRI compatibility
6657 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
6658 linker, @command{ld} can use MRI compatible linker scripts as an
6659 alternative to the more general-purpose linker scripting language
6660 described in @ref{Scripts}. MRI compatible linker scripts have a much
6661 simpler command set than the scripting language otherwise used with
6662 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
6663 linker commands; these commands are described here.
6665 In general, MRI scripts aren't of much use with the @code{a.out} object
6666 file format, since it only has three sections and MRI scripts lack some
6667 features to make use of them.
6669 You can specify a file containing an MRI-compatible script using the
6670 @samp{-c} command-line option.
6672 Each command in an MRI-compatible script occupies its own line; each
6673 command line starts with the keyword that identifies the command (though
6674 blank lines are also allowed for punctuation). If a line of an
6675 MRI-compatible script begins with an unrecognized keyword, @command{ld}
6676 issues a warning message, but continues processing the script.
6678 Lines beginning with @samp{*} are comments.
6680 You can write these commands using all upper-case letters, or all
6681 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
6682 The following list shows only the upper-case form of each command.
6685 @cindex @code{ABSOLUTE} (MRI)
6686 @item ABSOLUTE @var{secname}
6687 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
6688 Normally, @command{ld} includes in the output file all sections from all
6689 the input files. However, in an MRI-compatible script, you can use the
6690 @code{ABSOLUTE} command to restrict the sections that will be present in
6691 your output program. If the @code{ABSOLUTE} command is used at all in a
6692 script, then only the sections named explicitly in @code{ABSOLUTE}
6693 commands will appear in the linker output. You can still use other
6694 input sections (whatever you select on the command line, or using
6695 @code{LOAD}) to resolve addresses in the output file.
6697 @cindex @code{ALIAS} (MRI)
6698 @item ALIAS @var{out-secname}, @var{in-secname}
6699 Use this command to place the data from input section @var{in-secname}
6700 in a section called @var{out-secname} in the linker output file.
6702 @var{in-secname} may be an integer.
6704 @cindex @code{ALIGN} (MRI)
6705 @item ALIGN @var{secname} = @var{expression}
6706 Align the section called @var{secname} to @var{expression}. The
6707 @var{expression} should be a power of two.
6709 @cindex @code{BASE} (MRI)
6710 @item BASE @var{expression}
6711 Use the value of @var{expression} as the lowest address (other than
6712 absolute addresses) in the output file.
6714 @cindex @code{CHIP} (MRI)
6715 @item CHIP @var{expression}
6716 @itemx CHIP @var{expression}, @var{expression}
6717 This command does nothing; it is accepted only for compatibility.
6719 @cindex @code{END} (MRI)
6721 This command does nothing whatever; it's only accepted for compatibility.
6723 @cindex @code{FORMAT} (MRI)
6724 @item FORMAT @var{output-format}
6725 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
6726 language, but restricted to one of these output formats:
6730 S-records, if @var{output-format} is @samp{S}
6733 IEEE, if @var{output-format} is @samp{IEEE}
6736 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
6740 @cindex @code{LIST} (MRI)
6741 @item LIST @var{anything}@dots{}
6742 Print (to the standard output file) a link map, as produced by the
6743 @command{ld} command-line option @samp{-M}.
6745 The keyword @code{LIST} may be followed by anything on the
6746 same line, with no change in its effect.
6748 @cindex @code{LOAD} (MRI)
6749 @item LOAD @var{filename}
6750 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
6751 Include one or more object file @var{filename} in the link; this has the
6752 same effect as specifying @var{filename} directly on the @command{ld}
6755 @cindex @code{NAME} (MRI)
6756 @item NAME @var{output-name}
6757 @var{output-name} is the name for the program produced by @command{ld}; the
6758 MRI-compatible command @code{NAME} is equivalent to the command-line
6759 option @samp{-o} or the general script language command @code{OUTPUT}.
6761 @cindex @code{ORDER} (MRI)
6762 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
6763 @itemx ORDER @var{secname} @var{secname} @var{secname}
6764 Normally, @command{ld} orders the sections in its output file in the
6765 order in which they first appear in the input files. In an MRI-compatible
6766 script, you can override this ordering with the @code{ORDER} command. The
6767 sections you list with @code{ORDER} will appear first in your output
6768 file, in the order specified.
6770 @cindex @code{PUBLIC} (MRI)
6771 @item PUBLIC @var{name}=@var{expression}
6772 @itemx PUBLIC @var{name},@var{expression}
6773 @itemx PUBLIC @var{name} @var{expression}
6774 Supply a value (@var{expression}) for external symbol
6775 @var{name} used in the linker input files.
6777 @cindex @code{SECT} (MRI)
6778 @item SECT @var{secname}, @var{expression}
6779 @itemx SECT @var{secname}=@var{expression}
6780 @itemx SECT @var{secname} @var{expression}
6781 You can use any of these three forms of the @code{SECT} command to
6782 specify the start address (@var{expression}) for section @var{secname}.
6783 If you have more than one @code{SECT} statement for the same
6784 @var{secname}, only the @emph{first} sets the start address.
6790 @unnumbered LD Index
6795 % I think something like @colophon should be in texinfo. In the
6797 \long\def\colophon{\hbox to0pt{}\vfill
6798 \centerline{The body of this manual is set in}
6799 \centerline{\fontname\tenrm,}
6800 \centerline{with headings in {\bf\fontname\tenbf}}
6801 \centerline{and examples in {\tt\fontname\tentt}.}
6802 \centerline{{\it\fontname\tenit\/} and}
6803 \centerline{{\sl\fontname\tensl\/}}
6804 \centerline{are used for emphasis.}\vfill}
6806 % Blame: doc@cygnus.com, 28mar91.