3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
5 @c Free Software Foundation, Inc.
8 @include configdoc.texi
9 @c (configdoc.texi is generated by the Makefile)
15 @macro gcctabopt{body}
21 @c Configure for the generation of man pages
45 * Ld: (ld). The GNU linker.
51 This file documents the @sc{gnu} linker LD
52 @ifset VERSION_PACKAGE
53 @value{VERSION_PACKAGE}
55 version @value{VERSION}.
57 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
58 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
60 Permission is granted to copy, distribute and/or modify this document
61 under the terms of the GNU Free Documentation License, Version 1.3
62 or any later version published by the Free Software Foundation;
63 with no Invariant Sections, with no Front-Cover Texts, and with no
64 Back-Cover Texts. A copy of the license is included in the
65 section entitled ``GNU Free Documentation License''.
69 @setchapternewpage odd
70 @settitle The GNU linker
75 @ifset VERSION_PACKAGE
76 @subtitle @value{VERSION_PACKAGE}
78 @subtitle Version @value{VERSION}
79 @author Steve Chamberlain
80 @author Ian Lance Taylor
85 \hfill Red Hat Inc\par
86 \hfill nickc\@credhat.com, doc\@redhat.com\par
87 \hfill {\it The GNU linker}\par
88 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
90 \global\parindent=0pt % Steve likes it this way.
93 @vskip 0pt plus 1filll
94 @c man begin COPYRIGHT
95 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
96 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
98 Permission is granted to copy, distribute and/or modify this document
99 under the terms of the GNU Free Documentation License, Version 1.3
100 or any later version published by the Free Software Foundation;
101 with no Invariant Sections, with no Front-Cover Texts, and with no
102 Back-Cover Texts. A copy of the license is included in the
103 section entitled ``GNU Free Documentation License''.
109 @c FIXME: Talk about importance of *order* of args, cmds to linker!
114 This file documents the @sc{gnu} linker ld
115 @ifset VERSION_PACKAGE
116 @value{VERSION_PACKAGE}
118 version @value{VERSION}.
120 This document is distributed under the terms of the GNU Free
121 Documentation License version 1.3. A copy of the license is included
122 in the section entitled ``GNU Free Documentation License''.
125 * Overview:: Overview
126 * Invocation:: Invocation
127 * Scripts:: Linker Scripts
129 * Machine Dependent:: Machine Dependent Features
133 * H8/300:: ld and the H8/300
136 * Renesas:: ld and other Renesas micros
139 * i960:: ld and the Intel 960 family
142 * ARM:: ld and the ARM family
145 * HPPA ELF32:: ld and HPPA 32-bit ELF
148 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
151 * M68K:: ld and Motorola 68K family
154 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
157 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
160 * SPU ELF:: ld and SPU ELF Support
163 * TI COFF:: ld and the TI COFF
166 * Win32:: ld and WIN32 (cygwin/mingw)
169 * Xtensa:: ld and Xtensa Processors
172 @ifclear SingleFormat
175 @c Following blank line required for remaining bug in makeinfo conds/menus
177 * Reporting Bugs:: Reporting Bugs
178 * MRI:: MRI Compatible Script Files
179 * GNU Free Documentation License:: GNU Free Documentation License
180 * LD Index:: LD Index
187 @cindex @sc{gnu} linker
188 @cindex what is this?
191 @c man begin SYNOPSIS
192 ld [@b{options}] @var{objfile} @dots{}
196 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
197 the Info entries for @file{binutils} and
202 @c man begin DESCRIPTION
204 @command{ld} combines a number of object and archive files, relocates
205 their data and ties up symbol references. Usually the last step in
206 compiling a program is to run @command{ld}.
208 @command{ld} accepts Linker Command Language files written in
209 a superset of AT&T's Link Editor Command Language syntax,
210 to provide explicit and total control over the linking process.
214 This man page does not describe the command language; see the
215 @command{ld} entry in @code{info} for full details on the command
216 language and on other aspects of the GNU linker.
219 @ifclear SingleFormat
220 This version of @command{ld} uses the general purpose BFD libraries
221 to operate on object files. This allows @command{ld} to read, combine, and
222 write object files in many different formats---for example, COFF or
223 @code{a.out}. Different formats may be linked together to produce any
224 available kind of object file. @xref{BFD}, for more information.
227 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
228 linkers in providing diagnostic information. Many linkers abandon
229 execution immediately upon encountering an error; whenever possible,
230 @command{ld} continues executing, allowing you to identify other errors
231 (or, in some cases, to get an output file in spite of the error).
238 @c man begin DESCRIPTION
240 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
241 and to be as compatible as possible with other linkers. As a result,
242 you have many choices to control its behavior.
248 * Options:: Command Line Options
249 * Environment:: Environment Variables
253 @section Command Line Options
261 The linker supports a plethora of command-line options, but in actual
262 practice few of them are used in any particular context.
263 @cindex standard Unix system
264 For instance, a frequent use of @command{ld} is to link standard Unix
265 object files on a standard, supported Unix system. On such a system, to
266 link a file @code{hello.o}:
269 ld -o @var{output} /lib/crt0.o hello.o -lc
272 This tells @command{ld} to produce a file called @var{output} as the
273 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
274 the library @code{libc.a}, which will come from the standard search
275 directories. (See the discussion of the @samp{-l} option below.)
277 Some of the command-line options to @command{ld} may be specified at any
278 point in the command line. However, options which refer to files, such
279 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
280 which the option appears in the command line, relative to the object
281 files and other file options. Repeating non-file options with a
282 different argument will either have no further effect, or override prior
283 occurrences (those further to the left on the command line) of that
284 option. Options which may be meaningfully specified more than once are
285 noted in the descriptions below.
288 Non-option arguments are object files or archives which are to be linked
289 together. They may follow, precede, or be mixed in with command-line
290 options, except that an object file argument may not be placed between
291 an option and its argument.
293 Usually the linker is invoked with at least one object file, but you can
294 specify other forms of binary input files using @samp{-l}, @samp{-R},
295 and the script command language. If @emph{no} binary input files at all
296 are specified, the linker does not produce any output, and issues the
297 message @samp{No input files}.
299 If the linker cannot recognize the format of an object file, it will
300 assume that it is a linker script. A script specified in this way
301 augments the main linker script used for the link (either the default
302 linker script or the one specified by using @samp{-T}). This feature
303 permits the linker to link against a file which appears to be an object
304 or an archive, but actually merely defines some symbol values, or uses
305 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
306 script in this way merely augments the main linker script, with the
307 extra commands placed after the main script; use the @samp{-T} option
308 to replace the default linker script entirely, but note the effect of
309 the @code{INSERT} command. @xref{Scripts}.
311 For options whose names are a single letter,
312 option arguments must either follow the option letter without intervening
313 whitespace, or be given as separate arguments immediately following the
314 option that requires them.
316 For options whose names are multiple letters, either one dash or two can
317 precede the option name; for example, @samp{-trace-symbol} and
318 @samp{--trace-symbol} are equivalent. Note---there is one exception to
319 this rule. Multiple letter options that start with a lower case 'o' can
320 only be preceded by two dashes. This is to reduce confusion with the
321 @samp{-o} option. So for example @samp{-omagic} sets the output file
322 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
325 Arguments to multiple-letter options must either be separated from the
326 option name by an equals sign, or be given as separate arguments
327 immediately following the option that requires them. For example,
328 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
329 Unique abbreviations of the names of multiple-letter options are
332 Note---if the linker is being invoked indirectly, via a compiler driver
333 (e.g. @samp{gcc}) then all the linker command line options should be
334 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
335 compiler driver) like this:
338 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
341 This is important, because otherwise the compiler driver program may
342 silently drop the linker options, resulting in a bad link. Confusion
343 may also arise when passing options that require values through a
344 driver, as the use of a space between option and argument acts as
345 a separator, and causes the driver to pass only the option to the linker
346 and the argument to the compiler. In this case, it is simplest to use
347 the joined forms of both single- and multiple-letter options, such as:
350 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
353 Here is a table of the generic command line switches accepted by the GNU
357 @include at-file.texi
359 @kindex -a @var{keyword}
360 @item -a @var{keyword}
361 This option is supported for HP/UX compatibility. The @var{keyword}
362 argument must be one of the strings @samp{archive}, @samp{shared}, or
363 @samp{default}. @samp{-aarchive} is functionally equivalent to
364 @samp{-Bstatic}, and the other two keywords are functionally equivalent
365 to @samp{-Bdynamic}. This option may be used any number of times.
368 @cindex architectures
369 @kindex -A @var{arch}
370 @item -A @var{architecture}
371 @kindex --architecture=@var{arch}
372 @itemx --architecture=@var{architecture}
373 In the current release of @command{ld}, this option is useful only for the
374 Intel 960 family of architectures. In that @command{ld} configuration, the
375 @var{architecture} argument identifies the particular architecture in
376 the 960 family, enabling some safeguards and modifying the
377 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
378 family}, for details.
380 Future releases of @command{ld} may support similar functionality for
381 other architecture families.
384 @ifclear SingleFormat
385 @cindex binary input format
386 @kindex -b @var{format}
387 @kindex --format=@var{format}
390 @item -b @var{input-format}
391 @itemx --format=@var{input-format}
392 @command{ld} may be configured to support more than one kind of object
393 file. If your @command{ld} is configured this way, you can use the
394 @samp{-b} option to specify the binary format for input object files
395 that follow this option on the command line. Even when @command{ld} is
396 configured to support alternative object formats, you don't usually need
397 to specify this, as @command{ld} should be configured to expect as a
398 default input format the most usual format on each machine.
399 @var{input-format} is a text string, the name of a particular format
400 supported by the BFD libraries. (You can list the available binary
401 formats with @samp{objdump -i}.)
404 You may want to use this option if you are linking files with an unusual
405 binary format. You can also use @samp{-b} to switch formats explicitly (when
406 linking object files of different formats), by including
407 @samp{-b @var{input-format}} before each group of object files in a
410 The default format is taken from the environment variable
415 You can also define the input format from a script, using the command
418 see @ref{Format Commands}.
422 @kindex -c @var{MRI-cmdfile}
423 @kindex --mri-script=@var{MRI-cmdfile}
424 @cindex compatibility, MRI
425 @item -c @var{MRI-commandfile}
426 @itemx --mri-script=@var{MRI-commandfile}
427 For compatibility with linkers produced by MRI, @command{ld} accepts script
428 files written in an alternate, restricted command language, described in
430 @ref{MRI,,MRI Compatible Script Files}.
433 the MRI Compatible Script Files section of GNU ld documentation.
435 Introduce MRI script files with
436 the option @samp{-c}; use the @samp{-T} option to run linker
437 scripts written in the general-purpose @command{ld} scripting language.
438 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
439 specified by any @samp{-L} options.
441 @cindex common allocation
448 These three options are equivalent; multiple forms are supported for
449 compatibility with other linkers. They assign space to common symbols
450 even if a relocatable output file is specified (with @samp{-r}). The
451 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
452 @xref{Miscellaneous Commands}.
454 @cindex entry point, from command line
455 @kindex -e @var{entry}
456 @kindex --entry=@var{entry}
458 @itemx --entry=@var{entry}
459 Use @var{entry} as the explicit symbol for beginning execution of your
460 program, rather than the default entry point. If there is no symbol
461 named @var{entry}, the linker will try to parse @var{entry} as a number,
462 and use that as the entry address (the number will be interpreted in
463 base 10; you may use a leading @samp{0x} for base 16, or a leading
464 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
465 and other ways of specifying the entry point.
467 @kindex --exclude-libs
468 @item --exclude-libs @var{lib},@var{lib},...
469 Specifies a list of archive libraries from which symbols should not be automatically
470 exported. The library names may be delimited by commas or colons. Specifying
471 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
472 automatic export. This option is available only for the i386 PE targeted
473 port of the linker and for ELF targeted ports. For i386 PE, symbols
474 explicitly listed in a .def file are still exported, regardless of this
475 option. For ELF targeted ports, symbols affected by this option will
476 be treated as hidden.
478 @kindex --exclude-modules-for-implib
479 @item --exclude-modules-for-implib @var{module},@var{module},...
480 Specifies a list of object files or archive members, from which symbols
481 should not be automatically exported, but which should be copied wholesale
482 into the import library being generated during the link. The module names
483 may be delimited by commas or colons, and must match exactly the filenames
484 used by @command{ld} to open the files; for archive members, this is simply
485 the member name, but for object files the name listed must include and
486 match precisely any path used to specify the input file on the linker's
487 command-line. This option is available only for the i386 PE targeted port
488 of the linker. Symbols explicitly listed in a .def file are still exported,
489 regardless of this option.
491 @cindex dynamic symbol table
493 @kindex --export-dynamic
495 @itemx --export-dynamic
496 When creating a dynamically linked executable, add all symbols to the
497 dynamic symbol table. The dynamic symbol table is the set of symbols
498 which are visible from dynamic objects at run time.
500 If you do not use this option, the dynamic symbol table will normally
501 contain only those symbols which are referenced by some dynamic object
502 mentioned in the link.
504 If you use @code{dlopen} to load a dynamic object which needs to refer
505 back to the symbols defined by the program, rather than some other
506 dynamic object, then you will probably need to use this option when
507 linking the program itself.
509 You can also use the dynamic list to control what symbols should
510 be added to the dynamic symbol table if the output format supports it.
511 See the description of @samp{--dynamic-list}.
513 Note that this option is specific to ELF targeted ports. PE targets
514 support a similar function to export all symbols from a DLL or EXE; see
515 the description of @samp{--export-all-symbols} below.
517 @ifclear SingleFormat
518 @cindex big-endian objects
522 Link big-endian objects. This affects the default output format.
524 @cindex little-endian objects
527 Link little-endian objects. This affects the default output format.
530 @kindex -f @var{name}
531 @kindex --auxiliary=@var{name}
533 @itemx --auxiliary=@var{name}
534 When creating an ELF shared object, set the internal DT_AUXILIARY field
535 to the specified name. This tells the dynamic linker that the symbol
536 table of the shared object should be used as an auxiliary filter on the
537 symbol table of the shared object @var{name}.
539 If you later link a program against this filter object, then, when you
540 run the program, the dynamic linker will see the DT_AUXILIARY field. If
541 the dynamic linker resolves any symbols from the filter object, it will
542 first check whether there is a definition in the shared object
543 @var{name}. If there is one, it will be used instead of the definition
544 in the filter object. The shared object @var{name} need not exist.
545 Thus the shared object @var{name} may be used to provide an alternative
546 implementation of certain functions, perhaps for debugging or for
547 machine specific performance.
549 This option may be specified more than once. The DT_AUXILIARY entries
550 will be created in the order in which they appear on the command line.
552 @kindex -F @var{name}
553 @kindex --filter=@var{name}
555 @itemx --filter=@var{name}
556 When creating an ELF shared object, set the internal DT_FILTER field to
557 the specified name. This tells the dynamic linker that the symbol table
558 of the shared object which is being created should be used as a filter
559 on the symbol table of the shared object @var{name}.
561 If you later link a program against this filter object, then, when you
562 run the program, the dynamic linker will see the DT_FILTER field. The
563 dynamic linker will resolve symbols according to the symbol table of the
564 filter object as usual, but it will actually link to the definitions
565 found in the shared object @var{name}. Thus the filter object can be
566 used to select a subset of the symbols provided by the object
569 Some older linkers used the @option{-F} option throughout a compilation
570 toolchain for specifying object-file format for both input and output
572 @ifclear SingleFormat
573 The @sc{gnu} linker uses other mechanisms for this purpose: the
574 @option{-b}, @option{--format}, @option{--oformat} options, the
575 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
576 environment variable.
578 The @sc{gnu} linker will ignore the @option{-F} option when not
579 creating an ELF shared object.
581 @cindex finalization function
582 @kindex -fini=@var{name}
583 @item -fini=@var{name}
584 When creating an ELF executable or shared object, call NAME when the
585 executable or shared object is unloaded, by setting DT_FINI to the
586 address of the function. By default, the linker uses @code{_fini} as
587 the function to call.
591 Ignored. Provided for compatibility with other tools.
593 @kindex -G @var{value}
594 @kindex --gpsize=@var{value}
597 @itemx --gpsize=@var{value}
598 Set the maximum size of objects to be optimized using the GP register to
599 @var{size}. This is only meaningful for object file formats such as
600 MIPS ECOFF which supports putting large and small objects into different
601 sections. This is ignored for other object file formats.
603 @cindex runtime library name
604 @kindex -h @var{name}
605 @kindex -soname=@var{name}
607 @itemx -soname=@var{name}
608 When creating an ELF shared object, set the internal DT_SONAME field to
609 the specified name. When an executable is linked with a shared object
610 which has a DT_SONAME field, then when the executable is run the dynamic
611 linker will attempt to load the shared object specified by the DT_SONAME
612 field rather than the using the file name given to the linker.
615 @cindex incremental link
617 Perform an incremental link (same as option @samp{-r}).
619 @cindex initialization function
620 @kindex -init=@var{name}
621 @item -init=@var{name}
622 When creating an ELF executable or shared object, call NAME when the
623 executable or shared object is loaded, by setting DT_INIT to the address
624 of the function. By default, the linker uses @code{_init} as the
627 @cindex archive files, from cmd line
628 @kindex -l @var{namespec}
629 @kindex --library=@var{namespec}
630 @item -l @var{namespec}
631 @itemx --library=@var{namespec}
632 Add the archive or object file specified by @var{namespec} to the
633 list of files to link. This option may be used any number of times.
634 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
635 will search the library path for a file called @var{filename}, otherise it
636 will search the library path for a file called @file{lib@var{namespec}.a}.
638 On systems which support shared libraries, @command{ld} may also search for
639 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
640 and SunOS systems, @command{ld} will search a directory for a library
641 called @file{lib@var{namespec}.so} before searching for one called
642 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
643 indicates a shared library.) Note that this behavior does not apply
644 to @file{:@var{filename}}, which always specifies a file called
647 The linker will search an archive only once, at the location where it is
648 specified on the command line. If the archive defines a symbol which
649 was undefined in some object which appeared before the archive on the
650 command line, the linker will include the appropriate file(s) from the
651 archive. However, an undefined symbol in an object appearing later on
652 the command line will not cause the linker to search the archive again.
654 See the @option{-(} option for a way to force the linker to search
655 archives multiple times.
657 You may list the same archive multiple times on the command line.
660 This type of archive searching is standard for Unix linkers. However,
661 if you are using @command{ld} on AIX, note that it is different from the
662 behaviour of the AIX linker.
665 @cindex search directory, from cmd line
667 @kindex --library-path=@var{dir}
668 @item -L @var{searchdir}
669 @itemx --library-path=@var{searchdir}
670 Add path @var{searchdir} to the list of paths that @command{ld} will search
671 for archive libraries and @command{ld} control scripts. You may use this
672 option any number of times. The directories are searched in the order
673 in which they are specified on the command line. Directories specified
674 on the command line are searched before the default directories. All
675 @option{-L} options apply to all @option{-l} options, regardless of the
676 order in which the options appear. @option{-L} options do not affect
677 how @command{ld} searches for a linker script unless @option{-T}
680 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
681 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
684 The default set of paths searched (without being specified with
685 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
686 some cases also on how it was configured. @xref{Environment}.
689 The paths can also be specified in a link script with the
690 @code{SEARCH_DIR} command. Directories specified this way are searched
691 at the point in which the linker script appears in the command line.
694 @kindex -m @var{emulation}
695 @item -m @var{emulation}
696 Emulate the @var{emulation} linker. You can list the available
697 emulations with the @samp{--verbose} or @samp{-V} options.
699 If the @samp{-m} option is not used, the emulation is taken from the
700 @code{LDEMULATION} environment variable, if that is defined.
702 Otherwise, the default emulation depends upon how the linker was
710 Print a link map to the standard output. A link map provides
711 information about the link, including the following:
715 Where object files are mapped into memory.
717 How common symbols are allocated.
719 All archive members included in the link, with a mention of the symbol
720 which caused the archive member to be brought in.
722 The values assigned to symbols.
724 Note - symbols whose values are computed by an expression which
725 involves a reference to a previous value of the same symbol may not
726 have correct result displayed in the link map. This is because the
727 linker discards intermediate results and only retains the final value
728 of an expression. Under such circumstances the linker will display
729 the final value enclosed by square brackets. Thus for example a
730 linker script containing:
738 will produce the following output in the link map if the @option{-M}
743 [0x0000000c] foo = (foo * 0x4)
744 [0x0000000c] foo = (foo + 0x8)
747 See @ref{Expressions} for more information about expressions in linker
752 @cindex read-only text
757 Turn off page alignment of sections, and mark the output as
758 @code{NMAGIC} if possible.
762 @cindex read/write from cmd line
766 Set the text and data sections to be readable and writable. Also, do
767 not page-align the data segment, and disable linking against shared
768 libraries. If the output format supports Unix style magic numbers,
769 mark the output as @code{OMAGIC}. Note: Although a writable text section
770 is allowed for PE-COFF targets, it does not conform to the format
771 specification published by Microsoft.
776 This option negates most of the effects of the @option{-N} option. It
777 sets the text section to be read-only, and forces the data segment to
778 be page-aligned. Note - this option does not enable linking against
779 shared libraries. Use @option{-Bdynamic} for this.
781 @kindex -o @var{output}
782 @kindex --output=@var{output}
783 @cindex naming the output file
784 @item -o @var{output}
785 @itemx --output=@var{output}
786 Use @var{output} as the name for the program produced by @command{ld}; if this
787 option is not specified, the name @file{a.out} is used by default. The
788 script command @code{OUTPUT} can also specify the output file name.
790 @kindex -O @var{level}
791 @cindex generating optimized output
793 If @var{level} is a numeric values greater than zero @command{ld} optimizes
794 the output. This might take significantly longer and therefore probably
795 should only be enabled for the final binary. At the moment this
796 option only affects ELF shared library generation. Future releases of
797 the linker may make more use of this option. Also currently there is
798 no difference in the linker's behaviour for different non-zero values
799 of this option. Again this may change with future releases.
802 @kindex --emit-relocs
803 @cindex retain relocations in final executable
806 Leave relocation sections and contents in fully linked executables.
807 Post link analysis and optimization tools may need this information in
808 order to perform correct modifications of executables. This results
809 in larger executables.
811 This option is currently only supported on ELF platforms.
813 @kindex --force-dynamic
814 @cindex forcing the creation of dynamic sections
815 @item --force-dynamic
816 Force the output file to have dynamic sections. This option is specific
820 @cindex relocatable output
822 @kindex --relocatable
825 Generate relocatable output---i.e., generate an output file that can in
826 turn serve as input to @command{ld}. This is often called @dfn{partial
827 linking}. As a side effect, in environments that support standard Unix
828 magic numbers, this option also sets the output file's magic number to
830 @c ; see @option{-N}.
831 If this option is not specified, an absolute file is produced. When
832 linking C++ programs, this option @emph{will not} resolve references to
833 constructors; to do that, use @samp{-Ur}.
835 When an input file does not have the same format as the output file,
836 partial linking is only supported if that input file does not contain any
837 relocations. Different output formats can have further restrictions; for
838 example some @code{a.out}-based formats do not support partial linking
839 with input files in other formats at all.
841 This option does the same thing as @samp{-i}.
843 @kindex -R @var{file}
844 @kindex --just-symbols=@var{file}
845 @cindex symbol-only input
846 @item -R @var{filename}
847 @itemx --just-symbols=@var{filename}
848 Read symbol names and their addresses from @var{filename}, but do not
849 relocate it or include it in the output. This allows your output file
850 to refer symbolically to absolute locations of memory defined in other
851 programs. You may use this option more than once.
853 For compatibility with other ELF linkers, if the @option{-R} option is
854 followed by a directory name, rather than a file name, it is treated as
855 the @option{-rpath} option.
859 @cindex strip all symbols
862 Omit all symbol information from the output file.
865 @kindex --strip-debug
866 @cindex strip debugger symbols
869 Omit debugger symbol information (but not all symbols) from the output file.
873 @cindex input files, displaying
876 Print the names of the input files as @command{ld} processes them.
878 @kindex -T @var{script}
879 @kindex --script=@var{script}
881 @item -T @var{scriptfile}
882 @itemx --script=@var{scriptfile}
883 Use @var{scriptfile} as the linker script. This script replaces
884 @command{ld}'s default linker script (rather than adding to it), so
885 @var{commandfile} must specify everything necessary to describe the
886 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
887 the current directory, @code{ld} looks for it in the directories
888 specified by any preceding @samp{-L} options. Multiple @samp{-T}
891 @kindex -dT @var{script}
892 @kindex --default-script=@var{script}
894 @item -dT @var{scriptfile}
895 @itemx --default-script=@var{scriptfile}
896 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
898 This option is similar to the @option{--script} option except that
899 processing of the script is delayed until after the rest of the
900 command line has been processed. This allows options placed after the
901 @option{--default-script} option on the command line to affect the
902 behaviour of the linker script, which can be important when the linker
903 command line cannot be directly controlled by the user. (eg because
904 the command line is being constructed by another tool, such as
907 @kindex -u @var{symbol}
908 @kindex --undefined=@var{symbol}
909 @cindex undefined symbol
910 @item -u @var{symbol}
911 @itemx --undefined=@var{symbol}
912 Force @var{symbol} to be entered in the output file as an undefined
913 symbol. Doing this may, for example, trigger linking of additional
914 modules from standard libraries. @samp{-u} may be repeated with
915 different option arguments to enter additional undefined symbols. This
916 option is equivalent to the @code{EXTERN} linker script command.
921 For anything other than C++ programs, this option is equivalent to
922 @samp{-r}: it generates relocatable output---i.e., an output file that can in
923 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
924 @emph{does} resolve references to constructors, unlike @samp{-r}.
925 It does not work to use @samp{-Ur} on files that were themselves linked
926 with @samp{-Ur}; once the constructor table has been built, it cannot
927 be added to. Use @samp{-Ur} only for the last partial link, and
928 @samp{-r} for the others.
930 @kindex --unique[=@var{SECTION}]
931 @item --unique[=@var{SECTION}]
932 Creates a separate output section for every input section matching
933 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
934 missing, for every orphan input section. An orphan section is one not
935 specifically mentioned in a linker script. You may use this option
936 multiple times on the command line; It prevents the normal merging of
937 input sections with the same name, overriding output section assignments
947 Display the version number for @command{ld}. The @option{-V} option also
948 lists the supported emulations.
951 @kindex --discard-all
952 @cindex deleting local symbols
955 Delete all local symbols.
958 @kindex --discard-locals
959 @cindex local symbols, deleting
961 @itemx --discard-locals
962 Delete all temporary local symbols. (These symbols start with
963 system-specific local label prefixes, typically @samp{.L} for ELF systems
964 or @samp{L} for traditional a.out systems.)
966 @kindex -y @var{symbol}
967 @kindex --trace-symbol=@var{symbol}
968 @cindex symbol tracing
969 @item -y @var{symbol}
970 @itemx --trace-symbol=@var{symbol}
971 Print the name of each linked file in which @var{symbol} appears. This
972 option may be given any number of times. On many systems it is necessary
973 to prepend an underscore.
975 This option is useful when you have an undefined symbol in your link but
976 don't know where the reference is coming from.
978 @kindex -Y @var{path}
980 Add @var{path} to the default library search path. This option exists
981 for Solaris compatibility.
983 @kindex -z @var{keyword}
984 @item -z @var{keyword}
985 The recognized keywords are:
989 Combines multiple reloc sections and sorts them to make dynamic symbol
990 lookup caching possible.
993 Disallows undefined symbols in object files. Undefined symbols in
994 shared libraries are still allowed.
997 Marks the object as requiring executable stack.
1000 This option is only meaningful when building a shared object.
1001 It marks the object so that its runtime initialization will occur
1002 before the runtime initialization of any other objects brought into
1003 the process at the same time. Similarly the runtime finalization of
1004 the object will occur after the runtime finalization of any other
1008 Marks the object that its symbol table interposes before all symbols
1009 but the primary executable.
1012 When generating an executable or shared library, mark it to tell the
1013 dynamic linker to defer function call resolution to the point when
1014 the function is called (lazy binding), rather than at load time.
1015 Lazy binding is the default.
1018 Marks the object that its filters be processed immediately at
1022 Allows multiple definitions.
1025 Disables multiple reloc sections combining.
1028 Disables production of copy relocs.
1031 Marks the object that the search for dependencies of this object will
1032 ignore any default library search paths.
1035 Marks the object shouldn't be unloaded at runtime.
1038 Marks the object not available to @code{dlopen}.
1041 Marks the object can not be dumped by @code{dldump}.
1044 Marks the object as not requiring executable stack.
1047 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1050 When generating an executable or shared library, mark it to tell the
1051 dynamic linker to resolve all symbols when the program is started, or
1052 when the shared library is linked to using dlopen, instead of
1053 deferring function call resolution to the point when the function is
1057 Marks the object may contain $ORIGIN.
1060 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1062 @item max-page-size=@var{value}
1063 Set the emulation maximum page size to @var{value}.
1065 @item common-page-size=@var{value}
1066 Set the emulation common page size to @var{value}.
1070 Other keywords are ignored for Solaris compatibility.
1073 @cindex groups of archives
1074 @item -( @var{archives} -)
1075 @itemx --start-group @var{archives} --end-group
1076 The @var{archives} should be a list of archive files. They may be
1077 either explicit file names, or @samp{-l} options.
1079 The specified archives are searched repeatedly until no new undefined
1080 references are created. Normally, an archive is searched only once in
1081 the order that it is specified on the command line. If a symbol in that
1082 archive is needed to resolve an undefined symbol referred to by an
1083 object in an archive that appears later on the command line, the linker
1084 would not be able to resolve that reference. By grouping the archives,
1085 they all be searched repeatedly until all possible references are
1088 Using this option has a significant performance cost. It is best to use
1089 it only when there are unavoidable circular references between two or
1092 @kindex --accept-unknown-input-arch
1093 @kindex --no-accept-unknown-input-arch
1094 @item --accept-unknown-input-arch
1095 @itemx --no-accept-unknown-input-arch
1096 Tells the linker to accept input files whose architecture cannot be
1097 recognised. The assumption is that the user knows what they are doing
1098 and deliberately wants to link in these unknown input files. This was
1099 the default behaviour of the linker, before release 2.14. The default
1100 behaviour from release 2.14 onwards is to reject such input files, and
1101 so the @samp{--accept-unknown-input-arch} option has been added to
1102 restore the old behaviour.
1105 @kindex --no-as-needed
1107 @itemx --no-as-needed
1108 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1109 on the command line after the @option{--as-needed} option. Normally,
1110 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1111 on the command line, regardless of whether the library is actually
1112 needed. @option{--as-needed} causes a DT_NEEDED tag to only be emitted
1113 for a library that satisfies a symbol reference from regular objects
1114 which is undefined at the point that the library was linked, or, if
1115 the library is not found in the DT_NEEDED lists of other libraries
1116 linked up to that point, a reference from another dynamic library.
1117 @option{--no-as-needed} restores the default behaviour.
1119 @kindex --add-needed
1120 @kindex --no-add-needed
1122 @itemx --no-add-needed
1123 This option affects the treatment of dynamic libraries from ELF
1124 DT_NEEDED tags in dynamic libraries mentioned on the command line after
1125 the @option{--no-add-needed} option. Normally, the linker will add
1126 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1127 @option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1128 for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1129 the default behaviour.
1131 @kindex -assert @var{keyword}
1132 @item -assert @var{keyword}
1133 This option is ignored for SunOS compatibility.
1137 @kindex -call_shared
1141 Link against dynamic libraries. This is only meaningful on platforms
1142 for which shared libraries are supported. This option is normally the
1143 default on such platforms. The different variants of this option are
1144 for compatibility with various systems. You may use this option
1145 multiple times on the command line: it affects library searching for
1146 @option{-l} options which follow it.
1150 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1151 section. This causes the runtime linker to handle lookups in this
1152 object and its dependencies to be performed only inside the group.
1153 @option{--unresolved-symbols=report-all} is implied. This option is
1154 only meaningful on ELF platforms which support shared libraries.
1164 Do not link against shared libraries. This is only meaningful on
1165 platforms for which shared libraries are supported. The different
1166 variants of this option are for compatibility with various systems. You
1167 may use this option multiple times on the command line: it affects
1168 library searching for @option{-l} options which follow it. This
1169 option also implies @option{--unresolved-symbols=report-all}. This
1170 option can be used with @option{-shared}. Doing so means that a
1171 shared library is being created but that all of the library's external
1172 references must be resolved by pulling in entries from static
1177 When creating a shared library, bind references to global symbols to the
1178 definition within the shared library, if any. Normally, it is possible
1179 for a program linked against a shared library to override the definition
1180 within the shared library. This option is only meaningful on ELF
1181 platforms which support shared libraries.
1183 @kindex -Bsymbolic-functions
1184 @item -Bsymbolic-functions
1185 When creating a shared library, bind references to global function
1186 symbols to the definition within the shared library, if any.
1187 This option is only meaningful on ELF platforms which support shared
1190 @kindex --dynamic-list=@var{dynamic-list-file}
1191 @item --dynamic-list=@var{dynamic-list-file}
1192 Specify the name of a dynamic list file to the linker. This is
1193 typically used when creating shared libraries to specify a list of
1194 global symbols whose references shouldn't be bound to the definition
1195 within the shared library, or creating dynamically linked executables
1196 to specify a list of symbols which should be added to the symbol table
1197 in the executable. This option is only meaningful on ELF platforms
1198 which support shared libraries.
1200 The format of the dynamic list is the same as the version node without
1201 scope and node name. See @ref{VERSION} for more information.
1203 @kindex --dynamic-list-data
1204 @item --dynamic-list-data
1205 Include all global data symbols to the dynamic list.
1207 @kindex --dynamic-list-cpp-new
1208 @item --dynamic-list-cpp-new
1209 Provide the builtin dynamic list for C++ operator new and delete. It
1210 is mainly useful for building shared libstdc++.
1212 @kindex --dynamic-list-cpp-typeinfo
1213 @item --dynamic-list-cpp-typeinfo
1214 Provide the builtin dynamic list for C++ runtime type identification.
1216 @kindex --check-sections
1217 @kindex --no-check-sections
1218 @item --check-sections
1219 @itemx --no-check-sections
1220 Asks the linker @emph{not} to check section addresses after they have
1221 been assigned to see if there are any overlaps. Normally the linker will
1222 perform this check, and if it finds any overlaps it will produce
1223 suitable error messages. The linker does know about, and does make
1224 allowances for sections in overlays. The default behaviour can be
1225 restored by using the command line switch @option{--check-sections}.
1226 Section overlap is not usually checked for relocatable links. You can
1227 force checking in that case by using the @option{--check-sections}
1230 @cindex cross reference table
1233 Output a cross reference table. If a linker map file is being
1234 generated, the cross reference table is printed to the map file.
1235 Otherwise, it is printed on the standard output.
1237 The format of the table is intentionally simple, so that it may be
1238 easily processed by a script if necessary. The symbols are printed out,
1239 sorted by name. For each symbol, a list of file names is given. If the
1240 symbol is defined, the first file listed is the location of the
1241 definition. The remaining files contain references to the symbol.
1243 @cindex common allocation
1244 @kindex --no-define-common
1245 @item --no-define-common
1246 This option inhibits the assignment of addresses to common symbols.
1247 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1248 @xref{Miscellaneous Commands}.
1250 The @samp{--no-define-common} option allows decoupling
1251 the decision to assign addresses to Common symbols from the choice
1252 of the output file type; otherwise a non-Relocatable output type
1253 forces assigning addresses to Common symbols.
1254 Using @samp{--no-define-common} allows Common symbols that are referenced
1255 from a shared library to be assigned addresses only in the main program.
1256 This eliminates the unused duplicate space in the shared library,
1257 and also prevents any possible confusion over resolving to the wrong
1258 duplicate when there are many dynamic modules with specialized search
1259 paths for runtime symbol resolution.
1261 @cindex symbols, from command line
1262 @kindex --defsym=@var{symbol}=@var{exp}
1263 @item --defsym=@var{symbol}=@var{expression}
1264 Create a global symbol in the output file, containing the absolute
1265 address given by @var{expression}. You may use this option as many
1266 times as necessary to define multiple symbols in the command line. A
1267 limited form of arithmetic is supported for the @var{expression} in this
1268 context: you may give a hexadecimal constant or the name of an existing
1269 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1270 constants or symbols. If you need more elaborate expressions, consider
1271 using the linker command language from a script (@pxref{Assignments,,
1272 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1273 space between @var{symbol}, the equals sign (``@key{=}''), and
1276 @cindex demangling, from command line
1277 @kindex --demangle[=@var{style}]
1278 @kindex --no-demangle
1279 @item --demangle[=@var{style}]
1280 @itemx --no-demangle
1281 These options control whether to demangle symbol names in error messages
1282 and other output. When the linker is told to demangle, it tries to
1283 present symbol names in a readable fashion: it strips leading
1284 underscores if they are used by the object file format, and converts C++
1285 mangled symbol names into user readable names. Different compilers have
1286 different mangling styles. The optional demangling style argument can be used
1287 to choose an appropriate demangling style for your compiler. The linker will
1288 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1289 is set. These options may be used to override the default.
1291 @cindex dynamic linker, from command line
1292 @kindex -I@var{file}
1293 @kindex --dynamic-linker=@var{file}
1295 @itemx --dynamic-linker=@var{file}
1296 Set the name of the dynamic linker. This is only meaningful when
1297 generating dynamically linked ELF executables. The default dynamic
1298 linker is normally correct; don't use this unless you know what you are
1301 @kindex --fatal-warnings
1302 @kindex --no-fatal-warnings
1303 @item --fatal-warnings
1304 @itemx --no-fatal-warnings
1305 Treat all warnings as errors. The default behaviour can be restored
1306 with the option @option{--no-fatal-warnings}.
1308 @kindex --force-exe-suffix
1309 @item --force-exe-suffix
1310 Make sure that an output file has a .exe suffix.
1312 If a successfully built fully linked output file does not have a
1313 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1314 the output file to one of the same name with a @code{.exe} suffix. This
1315 option is useful when using unmodified Unix makefiles on a Microsoft
1316 Windows host, since some versions of Windows won't run an image unless
1317 it ends in a @code{.exe} suffix.
1319 @kindex --gc-sections
1320 @kindex --no-gc-sections
1321 @cindex garbage collection
1323 @itemx --no-gc-sections
1324 Enable garbage collection of unused input sections. It is ignored on
1325 targets that do not support this option. The default behaviour (of not
1326 performing this garbage collection) can be restored by specifying
1327 @samp{--no-gc-sections} on the command line.
1329 @samp{--gc-sections} decides which input sections are used by
1330 examining symbols and relocations. The section containing the entry
1331 symbol and all sections containing symbols undefined on the
1332 command-line will be kept, as will sections containing symbols
1333 referenced by dynamic objects. Note that when building shared
1334 libraries, the linker must assume that any visible symbol is
1335 referenced. Once this initial set of sections has been determined,
1336 the linker recursively marks as used any section referenced by their
1337 relocations. See @samp{--entry} and @samp{--undefined}.
1339 This option can be set when doing a partial link (enabled with option
1340 @samp{-r}). In this case the root of symbols kept must be explicitely
1341 specified either by an @samp{--entry} or @samp{--undefined} option or by
1342 a @code{ENTRY} command in the linker script.
1344 @kindex --print-gc-sections
1345 @kindex --no-print-gc-sections
1346 @cindex garbage collection
1347 @item --print-gc-sections
1348 @itemx --no-print-gc-sections
1349 List all sections removed by garbage collection. The listing is
1350 printed on stderr. This option is only effective if garbage
1351 collection has been enabled via the @samp{--gc-sections}) option. The
1352 default behaviour (of not listing the sections that are removed) can
1353 be restored by specifying @samp{--no-print-gc-sections} on the command
1360 Print a summary of the command-line options on the standard output and exit.
1362 @kindex --target-help
1364 Print a summary of all target specific options on the standard output and exit.
1366 @kindex -Map=@var{mapfile}
1367 @item -Map=@var{mapfile}
1368 Print a link map to the file @var{mapfile}. See the description of the
1369 @option{-M} option, above.
1371 @cindex memory usage
1372 @kindex --no-keep-memory
1373 @item --no-keep-memory
1374 @command{ld} normally optimizes for speed over memory usage by caching the
1375 symbol tables of input files in memory. This option tells @command{ld} to
1376 instead optimize for memory usage, by rereading the symbol tables as
1377 necessary. This may be required if @command{ld} runs out of memory space
1378 while linking a large executable.
1380 @kindex --no-undefined
1382 @item --no-undefined
1384 Report unresolved symbol references from regular object files. This
1385 is done even if the linker is creating a non-symbolic shared library.
1386 The switch @option{--[no-]allow-shlib-undefined} controls the
1387 behaviour for reporting unresolved references found in shared
1388 libraries being linked in.
1390 @kindex --allow-multiple-definition
1392 @item --allow-multiple-definition
1394 Normally when a symbol is defined multiple times, the linker will
1395 report a fatal error. These options allow multiple definitions and the
1396 first definition will be used.
1398 @kindex --allow-shlib-undefined
1399 @kindex --no-allow-shlib-undefined
1400 @item --allow-shlib-undefined
1401 @itemx --no-allow-shlib-undefined
1402 Allows or disallows undefined symbols in shared libraries.
1403 This switch is similar to @option{--no-undefined} except that it
1404 determines the behaviour when the undefined symbols are in a
1405 shared library rather than a regular object file. It does not affect
1406 how undefined symbols in regular object files are handled.
1408 The default behaviour is to report errors for any undefined symbols
1409 referenced in shared libraries if the linker is being used to create
1410 an executable, but to allow them if the linker is being used to create
1413 The reasons for allowing undefined symbol references in shared
1414 libraries specified at link time are that:
1418 A shared library specified at link time may not be the same as the one
1419 that is available at load time, so the symbol might actually be
1420 resolvable at load time.
1422 There are some operating systems, eg BeOS and HPPA, where undefined
1423 symbols in shared libraries are normal.
1425 The BeOS kernel for example patches shared libraries at load time to
1426 select whichever function is most appropriate for the current
1427 architecture. This is used, for example, to dynamically select an
1428 appropriate memset function.
1431 @kindex --no-undefined-version
1432 @item --no-undefined-version
1433 Normally when a symbol has an undefined version, the linker will ignore
1434 it. This option disallows symbols with undefined version and a fatal error
1435 will be issued instead.
1437 @kindex --default-symver
1438 @item --default-symver
1439 Create and use a default symbol version (the soname) for unversioned
1442 @kindex --default-imported-symver
1443 @item --default-imported-symver
1444 Create and use a default symbol version (the soname) for unversioned
1447 @kindex --no-warn-mismatch
1448 @item --no-warn-mismatch
1449 Normally @command{ld} will give an error if you try to link together input
1450 files that are mismatched for some reason, perhaps because they have
1451 been compiled for different processors or for different endiannesses.
1452 This option tells @command{ld} that it should silently permit such possible
1453 errors. This option should only be used with care, in cases when you
1454 have taken some special action that ensures that the linker errors are
1457 @kindex --no-warn-search-mismatch
1458 @item --no-warn-search-mismatch
1459 Normally @command{ld} will give a warning if it finds an incompatible
1460 library during a library search. This option silences the warning.
1462 @kindex --no-whole-archive
1463 @item --no-whole-archive
1464 Turn off the effect of the @option{--whole-archive} option for subsequent
1467 @cindex output file after errors
1468 @kindex --noinhibit-exec
1469 @item --noinhibit-exec
1470 Retain the executable output file whenever it is still usable.
1471 Normally, the linker will not produce an output file if it encounters
1472 errors during the link process; it exits without writing an output file
1473 when it issues any error whatsoever.
1477 Only search library directories explicitly specified on the
1478 command line. Library directories specified in linker scripts
1479 (including linker scripts specified on the command line) are ignored.
1481 @ifclear SingleFormat
1482 @kindex --oformat=@var{output-format}
1483 @item --oformat=@var{output-format}
1484 @command{ld} may be configured to support more than one kind of object
1485 file. If your @command{ld} is configured this way, you can use the
1486 @samp{--oformat} option to specify the binary format for the output
1487 object file. Even when @command{ld} is configured to support alternative
1488 object formats, you don't usually need to specify this, as @command{ld}
1489 should be configured to produce as a default output format the most
1490 usual format on each machine. @var{output-format} is a text string, the
1491 name of a particular format supported by the BFD libraries. (You can
1492 list the available binary formats with @samp{objdump -i}.) The script
1493 command @code{OUTPUT_FORMAT} can also specify the output format, but
1494 this option overrides it. @xref{BFD}.
1498 @kindex --pic-executable
1500 @itemx --pic-executable
1501 @cindex position independent executables
1502 Create a position independent executable. This is currently only supported on
1503 ELF platforms. Position independent executables are similar to shared
1504 libraries in that they are relocated by the dynamic linker to the virtual
1505 address the OS chooses for them (which can vary between invocations). Like
1506 normal dynamically linked executables they can be executed and symbols
1507 defined in the executable cannot be overridden by shared libraries.
1511 This option is ignored for Linux compatibility.
1515 This option is ignored for SVR4 compatibility.
1518 @cindex synthesizing linker
1519 @cindex relaxing addressing modes
1521 An option with machine dependent effects.
1523 This option is only supported on a few targets.
1526 @xref{H8/300,,@command{ld} and the H8/300}.
1529 @xref{i960,, @command{ld} and the Intel 960 family}.
1532 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1535 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1538 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1541 On some platforms, the @samp{--relax} option performs global
1542 optimizations that become possible when the linker resolves addressing
1543 in the program, such as relaxing address modes and synthesizing new
1544 instructions in the output object file.
1546 On some platforms these link time global optimizations may make symbolic
1547 debugging of the resulting executable impossible.
1550 the case for the Matsushita MN10200 and MN10300 family of processors.
1554 On platforms where this is not supported, @samp{--relax} is accepted,
1558 @cindex retaining specified symbols
1559 @cindex stripping all but some symbols
1560 @cindex symbols, retaining selectively
1561 @kindex --retain-symbols-file=@var{filename}
1562 @item --retain-symbols-file=@var{filename}
1563 Retain @emph{only} the symbols listed in the file @var{filename},
1564 discarding all others. @var{filename} is simply a flat file, with one
1565 symbol name per line. This option is especially useful in environments
1569 where a large global symbol table is accumulated gradually, to conserve
1572 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1573 or symbols needed for relocations.
1575 You may only specify @samp{--retain-symbols-file} once in the command
1576 line. It overrides @samp{-s} and @samp{-S}.
1579 @item -rpath=@var{dir}
1580 @cindex runtime library search path
1581 @kindex -rpath=@var{dir}
1582 Add a directory to the runtime library search path. This is used when
1583 linking an ELF executable with shared objects. All @option{-rpath}
1584 arguments are concatenated and passed to the runtime linker, which uses
1585 them to locate shared objects at runtime. The @option{-rpath} option is
1586 also used when locating shared objects which are needed by shared
1587 objects explicitly included in the link; see the description of the
1588 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1589 ELF executable, the contents of the environment variable
1590 @code{LD_RUN_PATH} will be used if it is defined.
1592 The @option{-rpath} option may also be used on SunOS. By default, on
1593 SunOS, the linker will form a runtime search patch out of all the
1594 @option{-L} options it is given. If a @option{-rpath} option is used, the
1595 runtime search path will be formed exclusively using the @option{-rpath}
1596 options, ignoring the @option{-L} options. This can be useful when using
1597 gcc, which adds many @option{-L} options which may be on NFS mounted
1600 For compatibility with other ELF linkers, if the @option{-R} option is
1601 followed by a directory name, rather than a file name, it is treated as
1602 the @option{-rpath} option.
1606 @cindex link-time runtime library search path
1607 @kindex -rpath-link=@var{dir}
1608 @item -rpath-link=@var{dir}
1609 When using ELF or SunOS, one shared library may require another. This
1610 happens when an @code{ld -shared} link includes a shared library as one
1613 When the linker encounters such a dependency when doing a non-shared,
1614 non-relocatable link, it will automatically try to locate the required
1615 shared library and include it in the link, if it is not included
1616 explicitly. In such a case, the @option{-rpath-link} option
1617 specifies the first set of directories to search. The
1618 @option{-rpath-link} option may specify a sequence of directory names
1619 either by specifying a list of names separated by colons, or by
1620 appearing multiple times.
1622 This option should be used with caution as it overrides the search path
1623 that may have been hard compiled into a shared library. In such a case it
1624 is possible to use unintentionally a different search path than the
1625 runtime linker would do.
1627 The linker uses the following search paths to locate required shared
1631 Any directories specified by @option{-rpath-link} options.
1633 Any directories specified by @option{-rpath} options. The difference
1634 between @option{-rpath} and @option{-rpath-link} is that directories
1635 specified by @option{-rpath} options are included in the executable and
1636 used at runtime, whereas the @option{-rpath-link} option is only effective
1637 at link time. Searching @option{-rpath} in this way is only supported
1638 by native linkers and cross linkers which have been configured with
1639 the @option{--with-sysroot} option.
1641 On an ELF system, for native linkers, if the @option{-rpath} and
1642 @option{-rpath-link} options were not used, search the contents of the
1643 environment variable @code{LD_RUN_PATH}.
1645 On SunOS, if the @option{-rpath} option was not used, search any
1646 directories specified using @option{-L} options.
1648 For a native linker, the search the contents of the environment
1649 variable @code{LD_LIBRARY_PATH}.
1651 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1652 @code{DT_RPATH} of a shared library are searched for shared
1653 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1654 @code{DT_RUNPATH} entries exist.
1656 The default directories, normally @file{/lib} and @file{/usr/lib}.
1658 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1659 exists, the list of directories found in that file.
1662 If the required shared library is not found, the linker will issue a
1663 warning and continue with the link.
1670 @cindex shared libraries
1671 Create a shared library. This is currently only supported on ELF, XCOFF
1672 and SunOS platforms. On SunOS, the linker will automatically create a
1673 shared library if the @option{-e} option is not used and there are
1674 undefined symbols in the link.
1676 @kindex --sort-common
1678 @itemx --sort-common=ascending
1679 @itemx --sort-common=descending
1680 This option tells @command{ld} to sort the common symbols by alignment in
1681 ascending or descending order when it places them in the appropriate output
1682 sections. The symbol alignments considered are sixteen-byte or larger,
1683 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1684 between symbols due to alignment constraints. If no sorting order is
1685 specified, then descending order is assumed.
1687 @kindex --sort-section=name
1688 @item --sort-section=name
1689 This option will apply @code{SORT_BY_NAME} to all wildcard section
1690 patterns in the linker script.
1692 @kindex --sort-section=alignment
1693 @item --sort-section=alignment
1694 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1695 patterns in the linker script.
1697 @kindex --split-by-file
1698 @item --split-by-file[=@var{size}]
1699 Similar to @option{--split-by-reloc} but creates a new output section for
1700 each input file when @var{size} is reached. @var{size} defaults to a
1701 size of 1 if not given.
1703 @kindex --split-by-reloc
1704 @item --split-by-reloc[=@var{count}]
1705 Tries to creates extra sections in the output file so that no single
1706 output section in the file contains more than @var{count} relocations.
1707 This is useful when generating huge relocatable files for downloading into
1708 certain real time kernels with the COFF object file format; since COFF
1709 cannot represent more than 65535 relocations in a single section. Note
1710 that this will fail to work with object file formats which do not
1711 support arbitrary sections. The linker will not split up individual
1712 input sections for redistribution, so if a single input section contains
1713 more than @var{count} relocations one output section will contain that
1714 many relocations. @var{count} defaults to a value of 32768.
1718 Compute and display statistics about the operation of the linker, such
1719 as execution time and memory usage.
1721 @kindex --sysroot=@var{directory}
1722 @item --sysroot=@var{directory}
1723 Use @var{directory} as the location of the sysroot, overriding the
1724 configure-time default. This option is only supported by linkers
1725 that were configured using @option{--with-sysroot}.
1727 @kindex --traditional-format
1728 @cindex traditional format
1729 @item --traditional-format
1730 For some targets, the output of @command{ld} is different in some ways from
1731 the output of some existing linker. This switch requests @command{ld} to
1732 use the traditional format instead.
1735 For example, on SunOS, @command{ld} combines duplicate entries in the
1736 symbol string table. This can reduce the size of an output file with
1737 full debugging information by over 30 percent. Unfortunately, the SunOS
1738 @code{dbx} program can not read the resulting program (@code{gdb} has no
1739 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1740 combine duplicate entries.
1742 @kindex --section-start=@var{sectionname}=@var{org}
1743 @item --section-start=@var{sectionname}=@var{org}
1744 Locate a section in the output file at the absolute
1745 address given by @var{org}. You may use this option as many
1746 times as necessary to locate multiple sections in the command
1748 @var{org} must be a single hexadecimal integer;
1749 for compatibility with other linkers, you may omit the leading
1750 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1751 should be no white space between @var{sectionname}, the equals
1752 sign (``@key{=}''), and @var{org}.
1754 @kindex -Tbss=@var{org}
1755 @kindex -Tdata=@var{org}
1756 @kindex -Ttext=@var{org}
1757 @cindex segment origins, cmd line
1758 @item -Tbss=@var{org}
1759 @itemx -Tdata=@var{org}
1760 @itemx -Ttext=@var{org}
1761 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
1762 @code{.text} as the @var{sectionname}.
1764 @kindex -Ttext-segment=@var{org}
1765 @item -Ttext-segment=@var{org}
1766 @cindex text segment origin, cmd line
1767 When creating an ELF executable or shared object, it will set the address
1768 of the first byte of the text segment.
1770 @kindex --unresolved-symbols
1771 @item --unresolved-symbols=@var{method}
1772 Determine how to handle unresolved symbols. There are four possible
1773 values for @samp{method}:
1777 Do not report any unresolved symbols.
1780 Report all unresolved symbols. This is the default.
1782 @item ignore-in-object-files
1783 Report unresolved symbols that are contained in shared libraries, but
1784 ignore them if they come from regular object files.
1786 @item ignore-in-shared-libs
1787 Report unresolved symbols that come from regular object files, but
1788 ignore them if they come from shared libraries. This can be useful
1789 when creating a dynamic binary and it is known that all the shared
1790 libraries that it should be referencing are included on the linker's
1794 The behaviour for shared libraries on their own can also be controlled
1795 by the @option{--[no-]allow-shlib-undefined} option.
1797 Normally the linker will generate an error message for each reported
1798 unresolved symbol but the option @option{--warn-unresolved-symbols}
1799 can change this to a warning.
1805 Display the version number for @command{ld} and list the linker emulations
1806 supported. Display which input files can and cannot be opened. Display
1807 the linker script being used by the linker.
1809 @kindex --version-script=@var{version-scriptfile}
1810 @cindex version script, symbol versions
1811 @item --version-script=@var{version-scriptfile}
1812 Specify the name of a version script to the linker. This is typically
1813 used when creating shared libraries to specify additional information
1814 about the version hierarchy for the library being created. This option
1815 is only meaningful on ELF platforms which support shared libraries.
1818 @kindex --warn-common
1819 @cindex warnings, on combining symbols
1820 @cindex combining symbols, warnings on
1822 Warn when a common symbol is combined with another common symbol or with
1823 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1824 but linkers on some other operating systems do not. This option allows
1825 you to find potential problems from combining global symbols.
1826 Unfortunately, some C libraries use this practise, so you may get some
1827 warnings about symbols in the libraries as well as in your programs.
1829 There are three kinds of global symbols, illustrated here by C examples:
1833 A definition, which goes in the initialized data section of the output
1837 An undefined reference, which does not allocate space.
1838 There must be either a definition or a common symbol for the
1842 A common symbol. If there are only (one or more) common symbols for a
1843 variable, it goes in the uninitialized data area of the output file.
1844 The linker merges multiple common symbols for the same variable into a
1845 single symbol. If they are of different sizes, it picks the largest
1846 size. The linker turns a common symbol into a declaration, if there is
1847 a definition of the same variable.
1850 The @samp{--warn-common} option can produce five kinds of warnings.
1851 Each warning consists of a pair of lines: the first describes the symbol
1852 just encountered, and the second describes the previous symbol
1853 encountered with the same name. One or both of the two symbols will be
1858 Turning a common symbol into a reference, because there is already a
1859 definition for the symbol.
1861 @var{file}(@var{section}): warning: common of `@var{symbol}'
1862 overridden by definition
1863 @var{file}(@var{section}): warning: defined here
1867 Turning a common symbol into a reference, because a later definition for
1868 the symbol is encountered. This is the same as the previous case,
1869 except that the symbols are encountered in a different order.
1871 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1873 @var{file}(@var{section}): warning: common is here
1877 Merging a common symbol with a previous same-sized common symbol.
1879 @var{file}(@var{section}): warning: multiple common
1881 @var{file}(@var{section}): warning: previous common is here
1885 Merging a common symbol with a previous larger common symbol.
1887 @var{file}(@var{section}): warning: common of `@var{symbol}'
1888 overridden by larger common
1889 @var{file}(@var{section}): warning: larger common is here
1893 Merging a common symbol with a previous smaller common symbol. This is
1894 the same as the previous case, except that the symbols are
1895 encountered in a different order.
1897 @var{file}(@var{section}): warning: common of `@var{symbol}'
1898 overriding smaller common
1899 @var{file}(@var{section}): warning: smaller common is here
1903 @kindex --warn-constructors
1904 @item --warn-constructors
1905 Warn if any global constructors are used. This is only useful for a few
1906 object file formats. For formats like COFF or ELF, the linker can not
1907 detect the use of global constructors.
1909 @kindex --warn-multiple-gp
1910 @item --warn-multiple-gp
1911 Warn if multiple global pointer values are required in the output file.
1912 This is only meaningful for certain processors, such as the Alpha.
1913 Specifically, some processors put large-valued constants in a special
1914 section. A special register (the global pointer) points into the middle
1915 of this section, so that constants can be loaded efficiently via a
1916 base-register relative addressing mode. Since the offset in
1917 base-register relative mode is fixed and relatively small (e.g., 16
1918 bits), this limits the maximum size of the constant pool. Thus, in
1919 large programs, it is often necessary to use multiple global pointer
1920 values in order to be able to address all possible constants. This
1921 option causes a warning to be issued whenever this case occurs.
1924 @cindex warnings, on undefined symbols
1925 @cindex undefined symbols, warnings on
1927 Only warn once for each undefined symbol, rather than once per module
1930 @kindex --warn-section-align
1931 @cindex warnings, on section alignment
1932 @cindex section alignment, warnings on
1933 @item --warn-section-align
1934 Warn if the address of an output section is changed because of
1935 alignment. Typically, the alignment will be set by an input section.
1936 The address will only be changed if it not explicitly specified; that
1937 is, if the @code{SECTIONS} command does not specify a start address for
1938 the section (@pxref{SECTIONS}).
1940 @kindex --warn-shared-textrel
1941 @item --warn-shared-textrel
1942 Warn if the linker adds a DT_TEXTREL to a shared object.
1944 @kindex --warn-unresolved-symbols
1945 @item --warn-unresolved-symbols
1946 If the linker is going to report an unresolved symbol (see the option
1947 @option{--unresolved-symbols}) it will normally generate an error.
1948 This option makes it generate a warning instead.
1950 @kindex --error-unresolved-symbols
1951 @item --error-unresolved-symbols
1952 This restores the linker's default behaviour of generating errors when
1953 it is reporting unresolved symbols.
1955 @kindex --whole-archive
1956 @cindex including an entire archive
1957 @item --whole-archive
1958 For each archive mentioned on the command line after the
1959 @option{--whole-archive} option, include every object file in the archive
1960 in the link, rather than searching the archive for the required object
1961 files. This is normally used to turn an archive file into a shared
1962 library, forcing every object to be included in the resulting shared
1963 library. This option may be used more than once.
1965 Two notes when using this option from gcc: First, gcc doesn't know
1966 about this option, so you have to use @option{-Wl,-whole-archive}.
1967 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1968 list of archives, because gcc will add its own list of archives to
1969 your link and you may not want this flag to affect those as well.
1971 @kindex --wrap=@var{symbol}
1972 @item --wrap=@var{symbol}
1973 Use a wrapper function for @var{symbol}. Any undefined reference to
1974 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1975 undefined reference to @code{__real_@var{symbol}} will be resolved to
1978 This can be used to provide a wrapper for a system function. The
1979 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1980 wishes to call the system function, it should call
1981 @code{__real_@var{symbol}}.
1983 Here is a trivial example:
1987 __wrap_malloc (size_t c)
1989 printf ("malloc called with %zu\n", c);
1990 return __real_malloc (c);
1994 If you link other code with this file using @option{--wrap malloc}, then
1995 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1996 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1997 call the real @code{malloc} function.
1999 You may wish to provide a @code{__real_malloc} function as well, so that
2000 links without the @option{--wrap} option will succeed. If you do this,
2001 you should not put the definition of @code{__real_malloc} in the same
2002 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2003 call before the linker has a chance to wrap it to @code{malloc}.
2005 @kindex --eh-frame-hdr
2006 @item --eh-frame-hdr
2007 Request creation of @code{.eh_frame_hdr} section and ELF
2008 @code{PT_GNU_EH_FRAME} segment header.
2010 @kindex --enable-new-dtags
2011 @kindex --disable-new-dtags
2012 @item --enable-new-dtags
2013 @itemx --disable-new-dtags
2014 This linker can create the new dynamic tags in ELF. But the older ELF
2015 systems may not understand them. If you specify
2016 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
2017 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2018 created. By default, the new dynamic tags are not created. Note that
2019 those options are only available for ELF systems.
2021 @kindex --hash-size=@var{number}
2022 @item --hash-size=@var{number}
2023 Set the default size of the linker's hash tables to a prime number
2024 close to @var{number}. Increasing this value can reduce the length of
2025 time it takes the linker to perform its tasks, at the expense of
2026 increasing the linker's memory requirements. Similarly reducing this
2027 value can reduce the memory requirements at the expense of speed.
2029 @kindex --hash-style=@var{style}
2030 @item --hash-style=@var{style}
2031 Set the type of linker's hash table(s). @var{style} can be either
2032 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2033 new style GNU @code{.gnu.hash} section or @code{both} for both
2034 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2035 hash tables. The default is @code{sysv}.
2037 @kindex --reduce-memory-overheads
2038 @item --reduce-memory-overheads
2039 This option reduces memory requirements at ld runtime, at the expense of
2040 linking speed. This was introduced to select the old O(n^2) algorithm
2041 for link map file generation, rather than the new O(n) algorithm which uses
2042 about 40% more memory for symbol storage.
2044 Another effect of the switch is to set the default hash table size to
2045 1021, which again saves memory at the cost of lengthening the linker's
2046 run time. This is not done however if the @option{--hash-size} switch
2049 The @option{--reduce-memory-overheads} switch may be also be used to
2050 enable other tradeoffs in future versions of the linker.
2053 @kindex --build-id=@var{style}
2055 @itemx --build-id=@var{style}
2056 Request creation of @code{.note.gnu.build-id} ELF note section.
2057 The contents of the note are unique bits identifying this linked
2058 file. @var{style} can be @code{uuid} to use 128 random bits,
2059 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
2060 parts of the output contents, @code{md5} to use a 128-bit
2061 @sc{MD5} hash on the normative parts of the output contents, or
2062 @code{0x@var{hexstring}} to use a chosen bit string specified as
2063 an even number of hexadecimal digits (@code{-} and @code{:}
2064 characters between digit pairs are ignored). If @var{style} is
2065 omitted, @code{sha1} is used.
2067 The @code{md5} and @code{sha1} styles produces an identifier
2068 that is always the same in an identical output file, but will be
2069 unique among all nonidentical output files. It is not intended
2070 to be compared as a checksum for the file's contents. A linked
2071 file may be changed later by other tools, but the build ID bit
2072 string identifying the original linked file does not change.
2074 Passing @code{none} for @var{style} disables the setting from any
2075 @code{--build-id} options earlier on the command line.
2080 @subsection Options Specific to i386 PE Targets
2082 @c man begin OPTIONS
2084 The i386 PE linker supports the @option{-shared} option, which causes
2085 the output to be a dynamically linked library (DLL) instead of a
2086 normal executable. You should name the output @code{*.dll} when you
2087 use this option. In addition, the linker fully supports the standard
2088 @code{*.def} files, which may be specified on the linker command line
2089 like an object file (in fact, it should precede archives it exports
2090 symbols from, to ensure that they get linked in, just like a normal
2093 In addition to the options common to all targets, the i386 PE linker
2094 support additional command line options that are specific to the i386
2095 PE target. Options that take values may be separated from their
2096 values by either a space or an equals sign.
2100 @kindex --add-stdcall-alias
2101 @item --add-stdcall-alias
2102 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2103 as-is and also with the suffix stripped.
2104 [This option is specific to the i386 PE targeted port of the linker]
2107 @item --base-file @var{file}
2108 Use @var{file} as the name of a file in which to save the base
2109 addresses of all the relocations needed for generating DLLs with
2111 [This is an i386 PE specific option]
2115 Create a DLL instead of a regular executable. You may also use
2116 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2118 [This option is specific to the i386 PE targeted port of the linker]
2120 @kindex --enable-long-section-names
2121 @kindex --disable-long-section-names
2122 @item --enable-long-section-names
2123 @itemx --disable-long-section-names
2124 The PE variants of the Coff object format add an extension that permits
2125 the use of section names longer than eight characters, the normal limit
2126 for Coff. By default, these names are only allowed in object files, as
2127 fully-linked executable images do not carry the Coff string table required
2128 to support the longer names. As a GNU extension, it is possible to
2129 allow their use in executable images as well, or to (probably pointlessly!)
2130 disallow it in object files, by using these two options. Executable images
2131 generated with these long section names are slightly non-standard, carrying
2132 as they do a string table, and may generate confusing output when examined
2133 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2134 GDB relies on the use of PE long section names to find Dwarf-2 debug
2135 information sections in an executable image at runtime, and so if neither
2136 option is specified on the command-line, @command{ld} will enable long
2137 section names, overriding the default and technically correct behaviour,
2138 when it finds the presence of debug information while linking an executable
2139 image and not stripping symbols.
2140 [This option is valid for all PE targeted ports of the linker]
2142 @kindex --enable-stdcall-fixup
2143 @kindex --disable-stdcall-fixup
2144 @item --enable-stdcall-fixup
2145 @itemx --disable-stdcall-fixup
2146 If the link finds a symbol that it cannot resolve, it will attempt to
2147 do ``fuzzy linking'' by looking for another defined symbol that differs
2148 only in the format of the symbol name (cdecl vs stdcall) and will
2149 resolve that symbol by linking to the match. For example, the
2150 undefined symbol @code{_foo} might be linked to the function
2151 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2152 to the function @code{_bar}. When the linker does this, it prints a
2153 warning, since it normally should have failed to link, but sometimes
2154 import libraries generated from third-party dlls may need this feature
2155 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2156 feature is fully enabled and warnings are not printed. If you specify
2157 @option{--disable-stdcall-fixup}, this feature is disabled and such
2158 mismatches are considered to be errors.
2159 [This option is specific to the i386 PE targeted port of the linker]
2161 @cindex DLLs, creating
2162 @kindex --export-all-symbols
2163 @item --export-all-symbols
2164 If given, all global symbols in the objects used to build a DLL will
2165 be exported by the DLL. Note that this is the default if there
2166 otherwise wouldn't be any exported symbols. When symbols are
2167 explicitly exported via DEF files or implicitly exported via function
2168 attributes, the default is to not export anything else unless this
2169 option is given. Note that the symbols @code{DllMain@@12},
2170 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2171 @code{impure_ptr} will not be automatically
2172 exported. Also, symbols imported from other DLLs will not be
2173 re-exported, nor will symbols specifying the DLL's internal layout
2174 such as those beginning with @code{_head_} or ending with
2175 @code{_iname}. In addition, no symbols from @code{libgcc},
2176 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2177 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2178 not be exported, to help with C++ DLLs. Finally, there is an
2179 extensive list of cygwin-private symbols that are not exported
2180 (obviously, this applies on when building DLLs for cygwin targets).
2181 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2182 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2183 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2184 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2185 @code{cygwin_premain3}, and @code{environ}.
2186 [This option is specific to the i386 PE targeted port of the linker]
2188 @kindex --exclude-symbols
2189 @item --exclude-symbols @var{symbol},@var{symbol},...
2190 Specifies a list of symbols which should not be automatically
2191 exported. The symbol names may be delimited by commas or colons.
2192 [This option is specific to the i386 PE targeted port of the linker]
2194 @kindex --file-alignment
2195 @item --file-alignment
2196 Specify the file alignment. Sections in the file will always begin at
2197 file offsets which are multiples of this number. This defaults to
2199 [This option is specific to the i386 PE targeted port of the linker]
2203 @item --heap @var{reserve}
2204 @itemx --heap @var{reserve},@var{commit}
2205 Specify the number of bytes of memory to reserve (and optionally commit)
2206 to be used as heap for this program. The default is 1Mb reserved, 4K
2208 [This option is specific to the i386 PE targeted port of the linker]
2211 @kindex --image-base
2212 @item --image-base @var{value}
2213 Use @var{value} as the base address of your program or dll. This is
2214 the lowest memory location that will be used when your program or dll
2215 is loaded. To reduce the need to relocate and improve performance of
2216 your dlls, each should have a unique base address and not overlap any
2217 other dlls. The default is 0x400000 for executables, and 0x10000000
2219 [This option is specific to the i386 PE targeted port of the linker]
2223 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2224 symbols before they are exported.
2225 [This option is specific to the i386 PE targeted port of the linker]
2227 @kindex --large-address-aware
2228 @item --large-address-aware
2229 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2230 header is set to indicate that this executable supports virtual addresses
2231 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2232 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2233 section of the BOOT.INI. Otherwise, this bit has no effect.
2234 [This option is specific to PE targeted ports of the linker]
2236 @kindex --major-image-version
2237 @item --major-image-version @var{value}
2238 Sets the major number of the ``image version''. Defaults to 1.
2239 [This option is specific to the i386 PE targeted port of the linker]
2241 @kindex --major-os-version
2242 @item --major-os-version @var{value}
2243 Sets the major number of the ``os version''. Defaults to 4.
2244 [This option is specific to the i386 PE targeted port of the linker]
2246 @kindex --major-subsystem-version
2247 @item --major-subsystem-version @var{value}
2248 Sets the major number of the ``subsystem version''. Defaults to 4.
2249 [This option is specific to the i386 PE targeted port of the linker]
2251 @kindex --minor-image-version
2252 @item --minor-image-version @var{value}
2253 Sets the minor number of the ``image version''. Defaults to 0.
2254 [This option is specific to the i386 PE targeted port of the linker]
2256 @kindex --minor-os-version
2257 @item --minor-os-version @var{value}
2258 Sets the minor number of the ``os version''. Defaults to 0.
2259 [This option is specific to the i386 PE targeted port of the linker]
2261 @kindex --minor-subsystem-version
2262 @item --minor-subsystem-version @var{value}
2263 Sets the minor number of the ``subsystem version''. Defaults to 0.
2264 [This option is specific to the i386 PE targeted port of the linker]
2266 @cindex DEF files, creating
2267 @cindex DLLs, creating
2268 @kindex --output-def
2269 @item --output-def @var{file}
2270 The linker will create the file @var{file} which will contain a DEF
2271 file corresponding to the DLL the linker is generating. This DEF file
2272 (which should be called @code{*.def}) may be used to create an import
2273 library with @code{dlltool} or may be used as a reference to
2274 automatically or implicitly exported symbols.
2275 [This option is specific to the i386 PE targeted port of the linker]
2277 @cindex DLLs, creating
2278 @kindex --out-implib
2279 @item --out-implib @var{file}
2280 The linker will create the file @var{file} which will contain an
2281 import lib corresponding to the DLL the linker is generating. This
2282 import lib (which should be called @code{*.dll.a} or @code{*.a}
2283 may be used to link clients against the generated DLL; this behaviour
2284 makes it possible to skip a separate @code{dlltool} import library
2286 [This option is specific to the i386 PE targeted port of the linker]
2288 @kindex --enable-auto-image-base
2289 @item --enable-auto-image-base
2290 Automatically choose the image base for DLLs, unless one is specified
2291 using the @code{--image-base} argument. By using a hash generated
2292 from the dllname to create unique image bases for each DLL, in-memory
2293 collisions and relocations which can delay program execution are
2295 [This option is specific to the i386 PE targeted port of the linker]
2297 @kindex --disable-auto-image-base
2298 @item --disable-auto-image-base
2299 Do not automatically generate a unique image base. If there is no
2300 user-specified image base (@code{--image-base}) then use the platform
2302 [This option is specific to the i386 PE targeted port of the linker]
2304 @cindex DLLs, linking to
2305 @kindex --dll-search-prefix
2306 @item --dll-search-prefix @var{string}
2307 When linking dynamically to a dll without an import library,
2308 search for @code{<string><basename>.dll} in preference to
2309 @code{lib<basename>.dll}. This behaviour allows easy distinction
2310 between DLLs built for the various "subplatforms": native, cygwin,
2311 uwin, pw, etc. For instance, cygwin DLLs typically use
2312 @code{--dll-search-prefix=cyg}.
2313 [This option is specific to the i386 PE targeted port of the linker]
2315 @kindex --enable-auto-import
2316 @item --enable-auto-import
2317 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2318 DATA imports from DLLs, and create the necessary thunking symbols when
2319 building the import libraries with those DATA exports. Note: Use of the
2320 'auto-import' extension will cause the text section of the image file
2321 to be made writable. This does not conform to the PE-COFF format
2322 specification published by Microsoft.
2324 Note - use of the 'auto-import' extension will also cause read only
2325 data which would normally be placed into the .rdata section to be
2326 placed into the .data section instead. This is in order to work
2327 around a problem with consts that is described here:
2328 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2330 Using 'auto-import' generally will 'just work' -- but sometimes you may
2333 "variable '<var>' can't be auto-imported. Please read the
2334 documentation for ld's @code{--enable-auto-import} for details."
2336 This message occurs when some (sub)expression accesses an address
2337 ultimately given by the sum of two constants (Win32 import tables only
2338 allow one). Instances where this may occur include accesses to member
2339 fields of struct variables imported from a DLL, as well as using a
2340 constant index into an array variable imported from a DLL. Any
2341 multiword variable (arrays, structs, long long, etc) may trigger
2342 this error condition. However, regardless of the exact data type
2343 of the offending exported variable, ld will always detect it, issue
2344 the warning, and exit.
2346 There are several ways to address this difficulty, regardless of the
2347 data type of the exported variable:
2349 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2350 of adjusting references in your client code for runtime environment, so
2351 this method works only when runtime environment supports this feature.
2353 A second solution is to force one of the 'constants' to be a variable --
2354 that is, unknown and un-optimizable at compile time. For arrays,
2355 there are two possibilities: a) make the indexee (the array's address)
2356 a variable, or b) make the 'constant' index a variable. Thus:
2359 extern type extern_array[];
2361 @{ volatile type *t=extern_array; t[1] @}
2367 extern type extern_array[];
2369 @{ volatile int t=1; extern_array[t] @}
2372 For structs (and most other multiword data types) the only option
2373 is to make the struct itself (or the long long, or the ...) variable:
2376 extern struct s extern_struct;
2377 extern_struct.field -->
2378 @{ volatile struct s *t=&extern_struct; t->field @}
2384 extern long long extern_ll;
2386 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2389 A third method of dealing with this difficulty is to abandon
2390 'auto-import' for the offending symbol and mark it with
2391 @code{__declspec(dllimport)}. However, in practise that
2392 requires using compile-time #defines to indicate whether you are
2393 building a DLL, building client code that will link to the DLL, or
2394 merely building/linking to a static library. In making the choice
2395 between the various methods of resolving the 'direct address with
2396 constant offset' problem, you should consider typical real-world usage:
2404 void main(int argc, char **argv)@{
2405 printf("%d\n",arr[1]);
2415 void main(int argc, char **argv)@{
2416 /* This workaround is for win32 and cygwin; do not "optimize" */
2417 volatile int *parr = arr;
2418 printf("%d\n",parr[1]);
2425 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2426 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2427 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2428 #define FOO_IMPORT __declspec(dllimport)
2432 extern FOO_IMPORT int arr[];
2435 void main(int argc, char **argv)@{
2436 printf("%d\n",arr[1]);
2440 A fourth way to avoid this problem is to re-code your
2441 library to use a functional interface rather than a data interface
2442 for the offending variables (e.g. set_foo() and get_foo() accessor
2444 [This option is specific to the i386 PE targeted port of the linker]
2446 @kindex --disable-auto-import
2447 @item --disable-auto-import
2448 Do not attempt to do sophisticated linking of @code{_symbol} to
2449 @code{__imp__symbol} for DATA imports from DLLs.
2450 [This option is specific to the i386 PE targeted port of the linker]
2452 @kindex --enable-runtime-pseudo-reloc
2453 @item --enable-runtime-pseudo-reloc
2454 If your code contains expressions described in --enable-auto-import section,
2455 that is, DATA imports from DLL with non-zero offset, this switch will create
2456 a vector of 'runtime pseudo relocations' which can be used by runtime
2457 environment to adjust references to such data in your client code.
2458 [This option is specific to the i386 PE targeted port of the linker]
2460 @kindex --disable-runtime-pseudo-reloc
2461 @item --disable-runtime-pseudo-reloc
2462 Do not create pseudo relocations for non-zero offset DATA imports from
2463 DLLs. This is the default.
2464 [This option is specific to the i386 PE targeted port of the linker]
2466 @kindex --enable-extra-pe-debug
2467 @item --enable-extra-pe-debug
2468 Show additional debug info related to auto-import symbol thunking.
2469 [This option is specific to the i386 PE targeted port of the linker]
2471 @kindex --section-alignment
2472 @item --section-alignment
2473 Sets the section alignment. Sections in memory will always begin at
2474 addresses which are a multiple of this number. Defaults to 0x1000.
2475 [This option is specific to the i386 PE targeted port of the linker]
2479 @item --stack @var{reserve}
2480 @itemx --stack @var{reserve},@var{commit}
2481 Specify the number of bytes of memory to reserve (and optionally commit)
2482 to be used as stack for this program. The default is 2Mb reserved, 4K
2484 [This option is specific to the i386 PE targeted port of the linker]
2487 @item --subsystem @var{which}
2488 @itemx --subsystem @var{which}:@var{major}
2489 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2490 Specifies the subsystem under which your program will execute. The
2491 legal values for @var{which} are @code{native}, @code{windows},
2492 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2493 the subsystem version also. Numeric values are also accepted for
2495 [This option is specific to the i386 PE targeted port of the linker]
2497 The following options set flags in the @code{DllCharacteristics} field
2498 of the PE file header:
2499 [These options are specific to PE targeted ports of the linker]
2501 @kindex --dynamicbase
2503 The image base address may be relocated using address space layout
2504 randomization (ASLR). This feature was introduced with MS Windows
2505 Vista for i386 PE targets.
2507 @kindex --forceinteg
2509 Code integrity checks are enforced.
2513 The image is compatible with the Data Execution Prevention.
2514 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2516 @kindex --no-isolation
2517 @item --no-isolation
2518 Although the image understands isolation, do not isolate the image.
2522 The image does not use SEH. No SE handler may be called from
2527 Do not bind this image.
2531 The driver uses the MS Windows Driver Model.
2535 The image is Terminal Server aware.
2542 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2544 @c man begin OPTIONS
2546 The 68HC11 and 68HC12 linkers support specific options to control the
2547 memory bank switching mapping and trampoline code generation.
2551 @kindex --no-trampoline
2552 @item --no-trampoline
2553 This option disables the generation of trampoline. By default a trampoline
2554 is generated for each far function which is called using a @code{jsr}
2555 instruction (this happens when a pointer to a far function is taken).
2557 @kindex --bank-window
2558 @item --bank-window @var{name}
2559 This option indicates to the linker the name of the memory region in
2560 the @samp{MEMORY} specification that describes the memory bank window.
2561 The definition of such region is then used by the linker to compute
2562 paging and addresses within the memory window.
2570 @subsection Options specific to Motorola 68K target
2572 @c man begin OPTIONS
2574 The following options are supported to control handling of GOT generation
2575 when linking for 68K targets.
2580 @item --got=@var{type}
2581 This option tells the linker which GOT generation scheme to use.
2582 @var{type} should be one of @samp{single}, @samp{negative},
2583 @samp{multigot} or @samp{target}. For more information refer to the
2584 Info entry for @file{ld}.
2593 @section Environment Variables
2595 @c man begin ENVIRONMENT
2597 You can change the behaviour of @command{ld} with the environment variables
2598 @ifclear SingleFormat
2601 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2603 @ifclear SingleFormat
2605 @cindex default input format
2606 @code{GNUTARGET} determines the input-file object format if you don't
2607 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2608 of the BFD names for an input format (@pxref{BFD}). If there is no
2609 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2610 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2611 attempts to discover the input format by examining binary input files;
2612 this method often succeeds, but there are potential ambiguities, since
2613 there is no method of ensuring that the magic number used to specify
2614 object-file formats is unique. However, the configuration procedure for
2615 BFD on each system places the conventional format for that system first
2616 in the search-list, so ambiguities are resolved in favor of convention.
2620 @cindex default emulation
2621 @cindex emulation, default
2622 @code{LDEMULATION} determines the default emulation if you don't use the
2623 @samp{-m} option. The emulation can affect various aspects of linker
2624 behaviour, particularly the default linker script. You can list the
2625 available emulations with the @samp{--verbose} or @samp{-V} options. If
2626 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2627 variable is not defined, the default emulation depends upon how the
2628 linker was configured.
2630 @kindex COLLECT_NO_DEMANGLE
2631 @cindex demangling, default
2632 Normally, the linker will default to demangling symbols. However, if
2633 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2634 default to not demangling symbols. This environment variable is used in
2635 a similar fashion by the @code{gcc} linker wrapper program. The default
2636 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2643 @chapter Linker Scripts
2646 @cindex linker scripts
2647 @cindex command files
2648 Every link is controlled by a @dfn{linker script}. This script is
2649 written in the linker command language.
2651 The main purpose of the linker script is to describe how the sections in
2652 the input files should be mapped into the output file, and to control
2653 the memory layout of the output file. Most linker scripts do nothing
2654 more than this. However, when necessary, the linker script can also
2655 direct the linker to perform many other operations, using the commands
2658 The linker always uses a linker script. If you do not supply one
2659 yourself, the linker will use a default script that is compiled into the
2660 linker executable. You can use the @samp{--verbose} command line option
2661 to display the default linker script. Certain command line options,
2662 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2664 You may supply your own linker script by using the @samp{-T} command
2665 line option. When you do this, your linker script will replace the
2666 default linker script.
2668 You may also use linker scripts implicitly by naming them as input files
2669 to the linker, as though they were files to be linked. @xref{Implicit
2673 * Basic Script Concepts:: Basic Linker Script Concepts
2674 * Script Format:: Linker Script Format
2675 * Simple Example:: Simple Linker Script Example
2676 * Simple Commands:: Simple Linker Script Commands
2677 * Assignments:: Assigning Values to Symbols
2678 * SECTIONS:: SECTIONS Command
2679 * MEMORY:: MEMORY Command
2680 * PHDRS:: PHDRS Command
2681 * VERSION:: VERSION Command
2682 * Expressions:: Expressions in Linker Scripts
2683 * Implicit Linker Scripts:: Implicit Linker Scripts
2686 @node Basic Script Concepts
2687 @section Basic Linker Script Concepts
2688 @cindex linker script concepts
2689 We need to define some basic concepts and vocabulary in order to
2690 describe the linker script language.
2692 The linker combines input files into a single output file. The output
2693 file and each input file are in a special data format known as an
2694 @dfn{object file format}. Each file is called an @dfn{object file}.
2695 The output file is often called an @dfn{executable}, but for our
2696 purposes we will also call it an object file. Each object file has,
2697 among other things, a list of @dfn{sections}. We sometimes refer to a
2698 section in an input file as an @dfn{input section}; similarly, a section
2699 in the output file is an @dfn{output section}.
2701 Each section in an object file has a name and a size. Most sections
2702 also have an associated block of data, known as the @dfn{section
2703 contents}. A section may be marked as @dfn{loadable}, which mean that
2704 the contents should be loaded into memory when the output file is run.
2705 A section with no contents may be @dfn{allocatable}, which means that an
2706 area in memory should be set aside, but nothing in particular should be
2707 loaded there (in some cases this memory must be zeroed out). A section
2708 which is neither loadable nor allocatable typically contains some sort
2709 of debugging information.
2711 Every loadable or allocatable output section has two addresses. The
2712 first is the @dfn{VMA}, or virtual memory address. This is the address
2713 the section will have when the output file is run. The second is the
2714 @dfn{LMA}, or load memory address. This is the address at which the
2715 section will be loaded. In most cases the two addresses will be the
2716 same. An example of when they might be different is when a data section
2717 is loaded into ROM, and then copied into RAM when the program starts up
2718 (this technique is often used to initialize global variables in a ROM
2719 based system). In this case the ROM address would be the LMA, and the
2720 RAM address would be the VMA.
2722 You can see the sections in an object file by using the @code{objdump}
2723 program with the @samp{-h} option.
2725 Every object file also has a list of @dfn{symbols}, known as the
2726 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2727 has a name, and each defined symbol has an address, among other
2728 information. If you compile a C or C++ program into an object file, you
2729 will get a defined symbol for every defined function and global or
2730 static variable. Every undefined function or global variable which is
2731 referenced in the input file will become an undefined symbol.
2733 You can see the symbols in an object file by using the @code{nm}
2734 program, or by using the @code{objdump} program with the @samp{-t}
2738 @section Linker Script Format
2739 @cindex linker script format
2740 Linker scripts are text files.
2742 You write a linker script as a series of commands. Each command is
2743 either a keyword, possibly followed by arguments, or an assignment to a
2744 symbol. You may separate commands using semicolons. Whitespace is
2747 Strings such as file or format names can normally be entered directly.
2748 If the file name contains a character such as a comma which would
2749 otherwise serve to separate file names, you may put the file name in
2750 double quotes. There is no way to use a double quote character in a
2753 You may include comments in linker scripts just as in C, delimited by
2754 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2757 @node Simple Example
2758 @section Simple Linker Script Example
2759 @cindex linker script example
2760 @cindex example of linker script
2761 Many linker scripts are fairly simple.
2763 The simplest possible linker script has just one command:
2764 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2765 memory layout of the output file.
2767 The @samp{SECTIONS} command is a powerful command. Here we will
2768 describe a simple use of it. Let's assume your program consists only of
2769 code, initialized data, and uninitialized data. These will be in the
2770 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2771 Let's assume further that these are the only sections which appear in
2774 For this example, let's say that the code should be loaded at address
2775 0x10000, and that the data should start at address 0x8000000. Here is a
2776 linker script which will do that:
2781 .text : @{ *(.text) @}
2783 .data : @{ *(.data) @}
2784 .bss : @{ *(.bss) @}
2788 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2789 followed by a series of symbol assignments and output section
2790 descriptions enclosed in curly braces.
2792 The first line inside the @samp{SECTIONS} command of the above example
2793 sets the value of the special symbol @samp{.}, which is the location
2794 counter. If you do not specify the address of an output section in some
2795 other way (other ways are described later), the address is set from the
2796 current value of the location counter. The location counter is then
2797 incremented by the size of the output section. At the start of the
2798 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2800 The second line defines an output section, @samp{.text}. The colon is
2801 required syntax which may be ignored for now. Within the curly braces
2802 after the output section name, you list the names of the input sections
2803 which should be placed into this output section. The @samp{*} is a
2804 wildcard which matches any file name. The expression @samp{*(.text)}
2805 means all @samp{.text} input sections in all input files.
2807 Since the location counter is @samp{0x10000} when the output section
2808 @samp{.text} is defined, the linker will set the address of the
2809 @samp{.text} section in the output file to be @samp{0x10000}.
2811 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2812 the output file. The linker will place the @samp{.data} output section
2813 at address @samp{0x8000000}. After the linker places the @samp{.data}
2814 output section, the value of the location counter will be
2815 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2816 effect is that the linker will place the @samp{.bss} output section
2817 immediately after the @samp{.data} output section in memory.
2819 The linker will ensure that each output section has the required
2820 alignment, by increasing the location counter if necessary. In this
2821 example, the specified addresses for the @samp{.text} and @samp{.data}
2822 sections will probably satisfy any alignment constraints, but the linker
2823 may have to create a small gap between the @samp{.data} and @samp{.bss}
2826 That's it! That's a simple and complete linker script.
2828 @node Simple Commands
2829 @section Simple Linker Script Commands
2830 @cindex linker script simple commands
2831 In this section we describe the simple linker script commands.
2834 * Entry Point:: Setting the entry point
2835 * File Commands:: Commands dealing with files
2836 @ifclear SingleFormat
2837 * Format Commands:: Commands dealing with object file formats
2840 * REGION_ALIAS:: Assign alias names to memory regions
2841 * Miscellaneous Commands:: Other linker script commands
2845 @subsection Setting the Entry Point
2846 @kindex ENTRY(@var{symbol})
2847 @cindex start of execution
2848 @cindex first instruction
2850 The first instruction to execute in a program is called the @dfn{entry
2851 point}. You can use the @code{ENTRY} linker script command to set the
2852 entry point. The argument is a symbol name:
2857 There are several ways to set the entry point. The linker will set the
2858 entry point by trying each of the following methods in order, and
2859 stopping when one of them succeeds:
2862 the @samp{-e} @var{entry} command-line option;
2864 the @code{ENTRY(@var{symbol})} command in a linker script;
2866 the value of the symbol @code{start}, if defined;
2868 the address of the first byte of the @samp{.text} section, if present;
2870 The address @code{0}.
2874 @subsection Commands Dealing with Files
2875 @cindex linker script file commands
2876 Several linker script commands deal with files.
2879 @item INCLUDE @var{filename}
2880 @kindex INCLUDE @var{filename}
2881 @cindex including a linker script
2882 Include the linker script @var{filename} at this point. The file will
2883 be searched for in the current directory, and in any directory specified
2884 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2887 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
2888 @code{SECTIONS} commands, or in output section descriptions.
2890 @item INPUT(@var{file}, @var{file}, @dots{})
2891 @itemx INPUT(@var{file} @var{file} @dots{})
2892 @kindex INPUT(@var{files})
2893 @cindex input files in linker scripts
2894 @cindex input object files in linker scripts
2895 @cindex linker script input object files
2896 The @code{INPUT} command directs the linker to include the named files
2897 in the link, as though they were named on the command line.
2899 For example, if you always want to include @file{subr.o} any time you do
2900 a link, but you can't be bothered to put it on every link command line,
2901 then you can put @samp{INPUT (subr.o)} in your linker script.
2903 In fact, if you like, you can list all of your input files in the linker
2904 script, and then invoke the linker with nothing but a @samp{-T} option.
2906 In case a @dfn{sysroot prefix} is configured, and the filename starts
2907 with the @samp{/} character, and the script being processed was
2908 located inside the @dfn{sysroot prefix}, the filename will be looked
2909 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2910 open the file in the current directory. If it is not found, the
2911 linker will search through the archive library search path. See the
2912 description of @samp{-L} in @ref{Options,,Command Line Options}.
2914 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2915 name to @code{lib@var{file}.a}, as with the command line argument
2918 When you use the @code{INPUT} command in an implicit linker script, the
2919 files will be included in the link at the point at which the linker
2920 script file is included. This can affect archive searching.
2922 @item GROUP(@var{file}, @var{file}, @dots{})
2923 @itemx GROUP(@var{file} @var{file} @dots{})
2924 @kindex GROUP(@var{files})
2925 @cindex grouping input files
2926 The @code{GROUP} command is like @code{INPUT}, except that the named
2927 files should all be archives, and they are searched repeatedly until no
2928 new undefined references are created. See the description of @samp{-(}
2929 in @ref{Options,,Command Line Options}.
2931 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
2932 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
2933 @kindex AS_NEEDED(@var{files})
2934 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2935 commands, among other filenames. The files listed will be handled
2936 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2937 with the exception of ELF shared libraries, that will be added only
2938 when they are actually needed. This construct essentially enables
2939 @option{--as-needed} option for all the files listed inside of it
2940 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2943 @item OUTPUT(@var{filename})
2944 @kindex OUTPUT(@var{filename})
2945 @cindex output file name in linker script
2946 The @code{OUTPUT} command names the output file. Using
2947 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2948 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2949 Line Options}). If both are used, the command line option takes
2952 You can use the @code{OUTPUT} command to define a default name for the
2953 output file other than the usual default of @file{a.out}.
2955 @item SEARCH_DIR(@var{path})
2956 @kindex SEARCH_DIR(@var{path})
2957 @cindex library search path in linker script
2958 @cindex archive search path in linker script
2959 @cindex search path in linker script
2960 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2961 @command{ld} looks for archive libraries. Using
2962 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2963 on the command line (@pxref{Options,,Command Line Options}). If both
2964 are used, then the linker will search both paths. Paths specified using
2965 the command line option are searched first.
2967 @item STARTUP(@var{filename})
2968 @kindex STARTUP(@var{filename})
2969 @cindex first input file
2970 The @code{STARTUP} command is just like the @code{INPUT} command, except
2971 that @var{filename} will become the first input file to be linked, as
2972 though it were specified first on the command line. This may be useful
2973 when using a system in which the entry point is always the start of the
2977 @ifclear SingleFormat
2978 @node Format Commands
2979 @subsection Commands Dealing with Object File Formats
2980 A couple of linker script commands deal with object file formats.
2983 @item OUTPUT_FORMAT(@var{bfdname})
2984 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2985 @kindex OUTPUT_FORMAT(@var{bfdname})
2986 @cindex output file format in linker script
2987 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2988 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2989 exactly like using @samp{--oformat @var{bfdname}} on the command line
2990 (@pxref{Options,,Command Line Options}). If both are used, the command
2991 line option takes precedence.
2993 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2994 formats based on the @samp{-EB} and @samp{-EL} command line options.
2995 This permits the linker script to set the output format based on the
2998 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2999 will be the first argument, @var{default}. If @samp{-EB} is used, the
3000 output format will be the second argument, @var{big}. If @samp{-EL} is
3001 used, the output format will be the third argument, @var{little}.
3003 For example, the default linker script for the MIPS ELF target uses this
3006 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3008 This says that the default format for the output file is
3009 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3010 option, the output file will be created in the @samp{elf32-littlemips}
3013 @item TARGET(@var{bfdname})
3014 @kindex TARGET(@var{bfdname})
3015 @cindex input file format in linker script
3016 The @code{TARGET} command names the BFD format to use when reading input
3017 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3018 This command is like using @samp{-b @var{bfdname}} on the command line
3019 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3020 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3021 command is also used to set the format for the output file. @xref{BFD}.
3026 @subsection Assign alias names to memory regions
3027 @kindex REGION_ALIAS(@var{alias}, @var{region})
3028 @cindex region alias
3029 @cindex region names
3031 Alias names can be added to existing memory regions created with the
3032 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3035 REGION_ALIAS(@var{alias}, @var{region})
3038 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3039 memory region @var{region}. This allows a flexible mapping of output sections
3040 to memory regions. An example follows.
3042 Suppose we have an application for embedded systems which come with various
3043 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3044 that allows code execution or data storage. Some may have a read-only,
3045 non-volatile memory @code{ROM} that allows code execution and read-only data
3046 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3047 read-only data access and no code execution capability. We have four output
3052 @code{.text} program code;
3054 @code{.rodata} read-only data;
3056 @code{.data} read-write initialized data;
3058 @code{.bss} read-write zero initialized data.
3061 The goal is to provide a linker command file that contains a system independent
3062 part defining the output sections and a system dependent part mapping the
3063 output sections to the memory regions available on the system. Our embedded
3064 systems come with three different memory setups @code{A}, @code{B} and
3066 @multitable @columnfractions .25 .25 .25 .25
3067 @item Section @tab Variant A @tab Variant B @tab Variant C
3068 @item .text @tab RAM @tab ROM @tab ROM
3069 @item .rodata @tab RAM @tab ROM @tab ROM2
3070 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3071 @item .bss @tab RAM @tab RAM @tab RAM
3073 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3074 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3075 the load address of the @code{.data} section starts in all three variants at
3076 the end of the @code{.rodata} section.
3078 The base linker script that deals with the output sections follows. It
3079 includes the system dependent @code{linkcmds.memory} file that describes the
3082 INCLUDE linkcmds.memory
3095 .data : AT (rodata_end)
3100 data_size = SIZEOF(.data);
3101 data_load_start = LOADADDR(.data);
3109 Now we need three different @code{linkcmds.memory} files to define memory
3110 regions and alias names. The content of @code{linkcmds.memory} for the three
3111 variants @code{A}, @code{B} and @code{C}:
3114 Here everything goes into the @code{RAM}.
3118 RAM : ORIGIN = 0, LENGTH = 4M
3121 REGION_ALIAS("REGION_TEXT", RAM);
3122 REGION_ALIAS("REGION_RODATA", RAM);
3123 REGION_ALIAS("REGION_DATA", RAM);
3124 REGION_ALIAS("REGION_BSS", RAM);
3127 Program code and read-only data go into the @code{ROM}. Read-write data goes
3128 into the @code{RAM}. An image of the initialized data is loaded into the
3129 @code{ROM} and will be copied during system start into the @code{RAM}.
3133 ROM : ORIGIN = 0, LENGTH = 3M
3134 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3137 REGION_ALIAS("REGION_TEXT", ROM);
3138 REGION_ALIAS("REGION_RODATA", ROM);
3139 REGION_ALIAS("REGION_DATA", RAM);
3140 REGION_ALIAS("REGION_BSS", RAM);
3143 Program code goes into the @code{ROM}. Read-only data goes into the
3144 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3145 initialized data is loaded into the @code{ROM2} and will be copied during
3146 system start into the @code{RAM}.
3150 ROM : ORIGIN = 0, LENGTH = 2M
3151 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3152 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3155 REGION_ALIAS("REGION_TEXT", ROM);
3156 REGION_ALIAS("REGION_RODATA", ROM2);
3157 REGION_ALIAS("REGION_DATA", RAM);
3158 REGION_ALIAS("REGION_BSS", RAM);
3162 It is possible to write a common system initialization routine to copy the
3163 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3168 extern char data_start [];
3169 extern char data_size [];
3170 extern char data_load_start [];
3172 void copy_data(void)
3174 if (data_start != data_load_start)
3176 memcpy(data_start, data_load_start, (size_t) data_size);
3181 @node Miscellaneous Commands
3182 @subsection Other Linker Script Commands
3183 There are a few other linker scripts commands.
3186 @item ASSERT(@var{exp}, @var{message})
3188 @cindex assertion in linker script
3189 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3190 with an error code, and print @var{message}.
3192 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3194 @cindex undefined symbol in linker script
3195 Force @var{symbol} to be entered in the output file as an undefined
3196 symbol. Doing this may, for example, trigger linking of additional
3197 modules from standard libraries. You may list several @var{symbol}s for
3198 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3199 command has the same effect as the @samp{-u} command-line option.
3201 @item FORCE_COMMON_ALLOCATION
3202 @kindex FORCE_COMMON_ALLOCATION
3203 @cindex common allocation in linker script
3204 This command has the same effect as the @samp{-d} command-line option:
3205 to make @command{ld} assign space to common symbols even if a relocatable
3206 output file is specified (@samp{-r}).
3208 @item INHIBIT_COMMON_ALLOCATION
3209 @kindex INHIBIT_COMMON_ALLOCATION
3210 @cindex common allocation in linker script
3211 This command has the same effect as the @samp{--no-define-common}
3212 command-line option: to make @code{ld} omit the assignment of addresses
3213 to common symbols even for a non-relocatable output file.
3215 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3217 @cindex insert user script into default script
3218 This command is typically used in a script specified by @samp{-T} to
3219 augment the default @code{SECTIONS} with, for example, overlays. It
3220 inserts all prior linker script statements after (or before)
3221 @var{output_section}, and also causes @samp{-T} to not override the
3222 default linker script. The exact insertion point is as for orphan
3223 sections. @xref{Location Counter}. The insertion happens after the
3224 linker has mapped input sections to output sections. Prior to the
3225 insertion, since @samp{-T} scripts are parsed before the default
3226 linker script, statements in the @samp{-T} script occur before the
3227 default linker script statements in the internal linker representation
3228 of the script. In particular, input section assignments will be made
3229 to @samp{-T} output sections before those in the default script. Here
3230 is an example of how a @samp{-T} script using @code{INSERT} might look:
3237 .ov1 @{ ov1*(.text) @}
3238 .ov2 @{ ov2*(.text) @}
3244 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3245 @kindex NOCROSSREFS(@var{sections})
3246 @cindex cross references
3247 This command may be used to tell @command{ld} to issue an error about any
3248 references among certain output sections.
3250 In certain types of programs, particularly on embedded systems when
3251 using overlays, when one section is loaded into memory, another section
3252 will not be. Any direct references between the two sections would be
3253 errors. For example, it would be an error if code in one section called
3254 a function defined in the other section.
3256 The @code{NOCROSSREFS} command takes a list of output section names. If
3257 @command{ld} detects any cross references between the sections, it reports
3258 an error and returns a non-zero exit status. Note that the
3259 @code{NOCROSSREFS} command uses output section names, not input section
3262 @ifclear SingleFormat
3263 @item OUTPUT_ARCH(@var{bfdarch})
3264 @kindex OUTPUT_ARCH(@var{bfdarch})
3265 @cindex machine architecture
3266 @cindex architecture
3267 Specify a particular output machine architecture. The argument is one
3268 of the names used by the BFD library (@pxref{BFD}). You can see the
3269 architecture of an object file by using the @code{objdump} program with
3270 the @samp{-f} option.
3275 @section Assigning Values to Symbols
3276 @cindex assignment in scripts
3277 @cindex symbol definition, scripts
3278 @cindex variables, defining
3279 You may assign a value to a symbol in a linker script. This will define
3280 the symbol and place it into the symbol table with a global scope.
3283 * Simple Assignments:: Simple Assignments
3285 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3286 * Source Code Reference:: How to use a linker script defined symbol in source code
3289 @node Simple Assignments
3290 @subsection Simple Assignments
3292 You may assign to a symbol using any of the C assignment operators:
3295 @item @var{symbol} = @var{expression} ;
3296 @itemx @var{symbol} += @var{expression} ;
3297 @itemx @var{symbol} -= @var{expression} ;
3298 @itemx @var{symbol} *= @var{expression} ;
3299 @itemx @var{symbol} /= @var{expression} ;
3300 @itemx @var{symbol} <<= @var{expression} ;
3301 @itemx @var{symbol} >>= @var{expression} ;
3302 @itemx @var{symbol} &= @var{expression} ;
3303 @itemx @var{symbol} |= @var{expression} ;
3306 The first case will define @var{symbol} to the value of
3307 @var{expression}. In the other cases, @var{symbol} must already be
3308 defined, and the value will be adjusted accordingly.
3310 The special symbol name @samp{.} indicates the location counter. You
3311 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3313 The semicolon after @var{expression} is required.
3315 Expressions are defined below; see @ref{Expressions}.
3317 You may write symbol assignments as commands in their own right, or as
3318 statements within a @code{SECTIONS} command, or as part of an output
3319 section description in a @code{SECTIONS} command.
3321 The section of the symbol will be set from the section of the
3322 expression; for more information, see @ref{Expression Section}.
3324 Here is an example showing the three different places that symbol
3325 assignments may be used:
3336 _bdata = (. + 3) & ~ 3;
3337 .data : @{ *(.data) @}
3341 In this example, the symbol @samp{floating_point} will be defined as
3342 zero. The symbol @samp{_etext} will be defined as the address following
3343 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3344 defined as the address following the @samp{.text} output section aligned
3345 upward to a 4 byte boundary.
3350 In some cases, it is desirable for a linker script to define a symbol
3351 only if it is referenced and is not defined by any object included in
3352 the link. For example, traditional linkers defined the symbol
3353 @samp{etext}. However, ANSI C requires that the user be able to use
3354 @samp{etext} as a function name without encountering an error. The
3355 @code{PROVIDE} keyword may be used to define a symbol, such as
3356 @samp{etext}, only if it is referenced but not defined. The syntax is
3357 @code{PROVIDE(@var{symbol} = @var{expression})}.
3359 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3372 In this example, if the program defines @samp{_etext} (with a leading
3373 underscore), the linker will give a multiple definition error. If, on
3374 the other hand, the program defines @samp{etext} (with no leading
3375 underscore), the linker will silently use the definition in the program.
3376 If the program references @samp{etext} but does not define it, the
3377 linker will use the definition in the linker script.
3379 @node PROVIDE_HIDDEN
3380 @subsection PROVIDE_HIDDEN
3381 @cindex PROVIDE_HIDDEN
3382 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3383 hidden and won't be exported.
3385 @node Source Code Reference
3386 @subsection Source Code Reference
3388 Accessing a linker script defined variable from source code is not
3389 intuitive. In particular a linker script symbol is not equivalent to
3390 a variable declaration in a high level language, it is instead a
3391 symbol that does not have a value.
3393 Before going further, it is important to note that compilers often
3394 transform names in the source code into different names when they are
3395 stored in the symbol table. For example, Fortran compilers commonly
3396 prepend or append an underscore, and C++ performs extensive @samp{name
3397 mangling}. Therefore there might be a discrepancy between the name
3398 of a variable as it is used in source code and the name of the same
3399 variable as it is defined in a linker script. For example in C a
3400 linker script variable might be referred to as:
3406 But in the linker script it might be defined as:
3412 In the remaining examples however it is assumed that no name
3413 transformation has taken place.
3415 When a symbol is declared in a high level language such as C, two
3416 things happen. The first is that the compiler reserves enough space
3417 in the program's memory to hold the @emph{value} of the symbol. The
3418 second is that the compiler creates an entry in the program's symbol
3419 table which holds the symbol's @emph{address}. ie the symbol table
3420 contains the address of the block of memory holding the symbol's
3421 value. So for example the following C declaration, at file scope:
3427 creates a entry called @samp{foo} in the symbol table. This entry
3428 holds the address of an @samp{int} sized block of memory where the
3429 number 1000 is initially stored.
3431 When a program references a symbol the compiler generates code that
3432 first accesses the symbol table to find the address of the symbol's
3433 memory block and then code to read the value from that memory block.
3440 looks up the symbol @samp{foo} in the symbol table, gets the address
3441 associated with this symbol and then writes the value 1 into that
3448 looks up the symbol @samp{foo} in the symbol table, gets it address
3449 and then copies this address into the block of memory associated with
3450 the variable @samp{a}.
3452 Linker scripts symbol declarations, by contrast, create an entry in
3453 the symbol table but do not assign any memory to them. Thus they are
3454 an address without a value. So for example the linker script definition:
3460 creates an entry in the symbol table called @samp{foo} which holds
3461 the address of memory location 1000, but nothing special is stored at
3462 address 1000. This means that you cannot access the @emph{value} of a
3463 linker script defined symbol - it has no value - all you can do is
3464 access the @emph{address} of a linker script defined symbol.
3466 Hence when you are using a linker script defined symbol in source code
3467 you should always take the address of the symbol, and never attempt to
3468 use its value. For example suppose you want to copy the contents of a
3469 section of memory called .ROM into a section called .FLASH and the
3470 linker script contains these declarations:
3474 start_of_ROM = .ROM;
3475 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3476 start_of_FLASH = .FLASH;
3480 Then the C source code to perform the copy would be:
3484 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3486 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3490 Note the use of the @samp{&} operators. These are correct.
3493 @section SECTIONS Command
3495 The @code{SECTIONS} command tells the linker how to map input sections
3496 into output sections, and how to place the output sections in memory.
3498 The format of the @code{SECTIONS} command is:
3502 @var{sections-command}
3503 @var{sections-command}
3508 Each @var{sections-command} may of be one of the following:
3512 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3514 a symbol assignment (@pxref{Assignments})
3516 an output section description
3518 an overlay description
3521 The @code{ENTRY} command and symbol assignments are permitted inside the
3522 @code{SECTIONS} command for convenience in using the location counter in
3523 those commands. This can also make the linker script easier to
3524 understand because you can use those commands at meaningful points in
3525 the layout of the output file.
3527 Output section descriptions and overlay descriptions are described
3530 If you do not use a @code{SECTIONS} command in your linker script, the
3531 linker will place each input section into an identically named output
3532 section in the order that the sections are first encountered in the
3533 input files. If all input sections are present in the first file, for
3534 example, the order of sections in the output file will match the order
3535 in the first input file. The first section will be at address zero.
3538 * Output Section Description:: Output section description
3539 * Output Section Name:: Output section name
3540 * Output Section Address:: Output section address
3541 * Input Section:: Input section description
3542 * Output Section Data:: Output section data
3543 * Output Section Keywords:: Output section keywords
3544 * Output Section Discarding:: Output section discarding
3545 * Output Section Attributes:: Output section attributes
3546 * Overlay Description:: Overlay description
3549 @node Output Section Description
3550 @subsection Output Section Description
3551 The full description of an output section looks like this:
3554 @var{section} [@var{address}] [(@var{type})] :
3555 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3557 @var{output-section-command}
3558 @var{output-section-command}
3560 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3564 Most output sections do not use most of the optional section attributes.
3566 The whitespace around @var{section} is required, so that the section
3567 name is unambiguous. The colon and the curly braces are also required.
3568 The line breaks and other white space are optional.
3570 Each @var{output-section-command} may be one of the following:
3574 a symbol assignment (@pxref{Assignments})
3576 an input section description (@pxref{Input Section})
3578 data values to include directly (@pxref{Output Section Data})
3580 a special output section keyword (@pxref{Output Section Keywords})
3583 @node Output Section Name
3584 @subsection Output Section Name
3585 @cindex name, section
3586 @cindex section name
3587 The name of the output section is @var{section}. @var{section} must
3588 meet the constraints of your output format. In formats which only
3589 support a limited number of sections, such as @code{a.out}, the name
3590 must be one of the names supported by the format (@code{a.out}, for
3591 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3592 output format supports any number of sections, but with numbers and not
3593 names (as is the case for Oasys), the name should be supplied as a
3594 quoted numeric string. A section name may consist of any sequence of
3595 characters, but a name which contains any unusual characters such as
3596 commas must be quoted.
3598 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3601 @node Output Section Address
3602 @subsection Output Section Address
3603 @cindex address, section
3604 @cindex section address
3605 The @var{address} is an expression for the VMA (the virtual memory
3606 address) of the output section. If you do not provide @var{address},
3607 the linker will set it based on @var{region} if present, or otherwise
3608 based on the current value of the location counter.
3610 If you provide @var{address}, the address of the output section will be
3611 set to precisely that. If you provide neither @var{address} nor
3612 @var{region}, then the address of the output section will be set to the
3613 current value of the location counter aligned to the alignment
3614 requirements of the output section. The alignment requirement of the
3615 output section is the strictest alignment of any input section contained
3616 within the output section.
3620 .text . : @{ *(.text) @}
3625 .text : @{ *(.text) @}
3628 are subtly different. The first will set the address of the
3629 @samp{.text} output section to the current value of the location
3630 counter. The second will set it to the current value of the location
3631 counter aligned to the strictest alignment of a @samp{.text} input
3634 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3635 For example, if you want to align the section on a 0x10 byte boundary,
3636 so that the lowest four bits of the section address are zero, you could
3637 do something like this:
3639 .text ALIGN(0x10) : @{ *(.text) @}
3642 This works because @code{ALIGN} returns the current location counter
3643 aligned upward to the specified value.
3645 Specifying @var{address} for a section will change the value of the
3646 location counter, provided that the section is non-empty. (Empty
3647 sections are ignored).
3650 @subsection Input Section Description
3651 @cindex input sections
3652 @cindex mapping input sections to output sections
3653 The most common output section command is an input section description.
3655 The input section description is the most basic linker script operation.
3656 You use output sections to tell the linker how to lay out your program
3657 in memory. You use input section descriptions to tell the linker how to
3658 map the input files into your memory layout.
3661 * Input Section Basics:: Input section basics
3662 * Input Section Wildcards:: Input section wildcard patterns
3663 * Input Section Common:: Input section for common symbols
3664 * Input Section Keep:: Input section and garbage collection
3665 * Input Section Example:: Input section example
3668 @node Input Section Basics
3669 @subsubsection Input Section Basics
3670 @cindex input section basics
3671 An input section description consists of a file name optionally followed
3672 by a list of section names in parentheses.
3674 The file name and the section name may be wildcard patterns, which we
3675 describe further below (@pxref{Input Section Wildcards}).
3677 The most common input section description is to include all input
3678 sections with a particular name in the output section. For example, to
3679 include all input @samp{.text} sections, you would write:
3684 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3685 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3686 match all files except the ones specified in the EXCLUDE_FILE list. For
3689 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3691 will cause all .ctors sections from all files except @file{crtend.o} and
3692 @file{otherfile.o} to be included.
3694 There are two ways to include more than one section:
3700 The difference between these is the order in which the @samp{.text} and
3701 @samp{.rdata} input sections will appear in the output section. In the
3702 first example, they will be intermingled, appearing in the same order as
3703 they are found in the linker input. In the second example, all
3704 @samp{.text} input sections will appear first, followed by all
3705 @samp{.rdata} input sections.
3707 You can specify a file name to include sections from a particular file.
3708 You would do this if one or more of your files contain special data that
3709 needs to be at a particular location in memory. For example:
3714 You can also specify files within archives by writing a pattern
3715 matching the archive, a colon, then the pattern matching the file,
3716 with no whitespace around the colon.
3720 matches file within archive
3722 matches the whole archive
3724 matches file but not one in an archive
3727 Either one or both of @samp{archive} and @samp{file} can contain shell
3728 wildcards. On DOS based file systems, the linker will assume that a
3729 single letter followed by a colon is a drive specifier, so
3730 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
3731 within an archive called @samp{c}. @samp{archive:file} filespecs may
3732 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
3733 other linker script contexts. For instance, you cannot extract a file
3734 from an archive by using @samp{archive:file} in an @code{INPUT}
3737 If you use a file name without a list of sections, then all sections in
3738 the input file will be included in the output section. This is not
3739 commonly done, but it may by useful on occasion. For example:
3744 When you use a file name which is not an @samp{archive:file} specifier
3745 and does not contain any wild card
3746 characters, the linker will first see if you also specified the file
3747 name on the linker command line or in an @code{INPUT} command. If you
3748 did not, the linker will attempt to open the file as an input file, as
3749 though it appeared on the command line. Note that this differs from an
3750 @code{INPUT} command, because the linker will not search for the file in
3751 the archive search path.
3753 @node Input Section Wildcards
3754 @subsubsection Input Section Wildcard Patterns
3755 @cindex input section wildcards
3756 @cindex wildcard file name patterns
3757 @cindex file name wildcard patterns
3758 @cindex section name wildcard patterns
3759 In an input section description, either the file name or the section
3760 name or both may be wildcard patterns.
3762 The file name of @samp{*} seen in many examples is a simple wildcard
3763 pattern for the file name.
3765 The wildcard patterns are like those used by the Unix shell.
3769 matches any number of characters
3771 matches any single character
3773 matches a single instance of any of the @var{chars}; the @samp{-}
3774 character may be used to specify a range of characters, as in
3775 @samp{[a-z]} to match any lower case letter
3777 quotes the following character
3780 When a file name is matched with a wildcard, the wildcard characters
3781 will not match a @samp{/} character (used to separate directory names on
3782 Unix). A pattern consisting of a single @samp{*} character is an
3783 exception; it will always match any file name, whether it contains a
3784 @samp{/} or not. In a section name, the wildcard characters will match
3785 a @samp{/} character.
3787 File name wildcard patterns only match files which are explicitly
3788 specified on the command line or in an @code{INPUT} command. The linker
3789 does not search directories to expand wildcards.
3791 If a file name matches more than one wildcard pattern, or if a file name
3792 appears explicitly and is also matched by a wildcard pattern, the linker
3793 will use the first match in the linker script. For example, this
3794 sequence of input section descriptions is probably in error, because the
3795 @file{data.o} rule will not be used:
3797 .data : @{ *(.data) @}
3798 .data1 : @{ data.o(.data) @}
3801 @cindex SORT_BY_NAME
3802 Normally, the linker will place files and sections matched by wildcards
3803 in the order in which they are seen during the link. You can change
3804 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3805 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
3806 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3807 into ascending order by name before placing them in the output file.
3809 @cindex SORT_BY_ALIGNMENT
3810 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3811 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3812 ascending order by alignment before placing them in the output file.
3815 @code{SORT} is an alias for @code{SORT_BY_NAME}.
3817 When there are nested section sorting commands in linker script, there
3818 can be at most 1 level of nesting for section sorting commands.
3822 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3823 It will sort the input sections by name first, then by alignment if 2
3824 sections have the same name.
3826 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3827 It will sort the input sections by alignment first, then by name if 2
3828 sections have the same alignment.
3830 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3831 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3833 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3834 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3836 All other nested section sorting commands are invalid.
3839 When both command line section sorting option and linker script
3840 section sorting command are used, section sorting command always
3841 takes precedence over the command line option.
3843 If the section sorting command in linker script isn't nested, the
3844 command line option will make the section sorting command to be
3845 treated as nested sorting command.
3849 @code{SORT_BY_NAME} (wildcard section pattern ) with
3850 @option{--sort-sections alignment} is equivalent to
3851 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3853 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3854 @option{--sort-section name} is equivalent to
3855 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3858 If the section sorting command in linker script is nested, the
3859 command line option will be ignored.
3861 If you ever get confused about where input sections are going, use the
3862 @samp{-M} linker option to generate a map file. The map file shows
3863 precisely how input sections are mapped to output sections.
3865 This example shows how wildcard patterns might be used to partition
3866 files. This linker script directs the linker to place all @samp{.text}
3867 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3868 The linker will place the @samp{.data} section from all files beginning
3869 with an upper case character in @samp{.DATA}; for all other files, the
3870 linker will place the @samp{.data} section in @samp{.data}.
3874 .text : @{ *(.text) @}
3875 .DATA : @{ [A-Z]*(.data) @}
3876 .data : @{ *(.data) @}
3877 .bss : @{ *(.bss) @}
3882 @node Input Section Common
3883 @subsubsection Input Section for Common Symbols
3884 @cindex common symbol placement
3885 @cindex uninitialized data placement
3886 A special notation is needed for common symbols, because in many object
3887 file formats common symbols do not have a particular input section. The
3888 linker treats common symbols as though they are in an input section
3889 named @samp{COMMON}.
3891 You may use file names with the @samp{COMMON} section just as with any
3892 other input sections. You can use this to place common symbols from a
3893 particular input file in one section while common symbols from other
3894 input files are placed in another section.
3896 In most cases, common symbols in input files will be placed in the
3897 @samp{.bss} section in the output file. For example:
3899 .bss @{ *(.bss) *(COMMON) @}
3902 @cindex scommon section
3903 @cindex small common symbols
3904 Some object file formats have more than one type of common symbol. For
3905 example, the MIPS ELF object file format distinguishes standard common
3906 symbols and small common symbols. In this case, the linker will use a
3907 different special section name for other types of common symbols. In
3908 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3909 symbols and @samp{.scommon} for small common symbols. This permits you
3910 to map the different types of common symbols into memory at different
3914 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3915 notation is now considered obsolete. It is equivalent to
3918 @node Input Section Keep
3919 @subsubsection Input Section and Garbage Collection
3921 @cindex garbage collection
3922 When link-time garbage collection is in use (@samp{--gc-sections}),
3923 it is often useful to mark sections that should not be eliminated.
3924 This is accomplished by surrounding an input section's wildcard entry
3925 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3926 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3928 @node Input Section Example
3929 @subsubsection Input Section Example
3930 The following example is a complete linker script. It tells the linker
3931 to read all of the sections from file @file{all.o} and place them at the
3932 start of output section @samp{outputa} which starts at location
3933 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3934 follows immediately, in the same output section. All of section
3935 @samp{.input2} from @file{foo.o} goes into output section
3936 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3937 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3938 files are written to output section @samp{outputc}.
3966 @node Output Section Data
3967 @subsection Output Section Data
3969 @cindex section data
3970 @cindex output section data
3971 @kindex BYTE(@var{expression})
3972 @kindex SHORT(@var{expression})
3973 @kindex LONG(@var{expression})
3974 @kindex QUAD(@var{expression})
3975 @kindex SQUAD(@var{expression})
3976 You can include explicit bytes of data in an output section by using
3977 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3978 an output section command. Each keyword is followed by an expression in
3979 parentheses providing the value to store (@pxref{Expressions}). The
3980 value of the expression is stored at the current value of the location
3983 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3984 store one, two, four, and eight bytes (respectively). After storing the
3985 bytes, the location counter is incremented by the number of bytes
3988 For example, this will store the byte 1 followed by the four byte value
3989 of the symbol @samp{addr}:
3995 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3996 same; they both store an 8 byte, or 64 bit, value. When both host and
3997 target are 32 bits, an expression is computed as 32 bits. In this case
3998 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3999 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4001 If the object file format of the output file has an explicit endianness,
4002 which is the normal case, the value will be stored in that endianness.
4003 When the object file format does not have an explicit endianness, as is
4004 true of, for example, S-records, the value will be stored in the
4005 endianness of the first input object file.
4007 Note---these commands only work inside a section description and not
4008 between them, so the following will produce an error from the linker:
4010 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4012 whereas this will work:
4014 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4017 @kindex FILL(@var{expression})
4018 @cindex holes, filling
4019 @cindex unspecified memory
4020 You may use the @code{FILL} command to set the fill pattern for the
4021 current section. It is followed by an expression in parentheses. Any
4022 otherwise unspecified regions of memory within the section (for example,
4023 gaps left due to the required alignment of input sections) are filled
4024 with the value of the expression, repeated as
4025 necessary. A @code{FILL} statement covers memory locations after the
4026 point at which it occurs in the section definition; by including more
4027 than one @code{FILL} statement, you can have different fill patterns in
4028 different parts of an output section.
4030 This example shows how to fill unspecified regions of memory with the
4036 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4037 section attribute, but it only affects the
4038 part of the section following the @code{FILL} command, rather than the
4039 entire section. If both are used, the @code{FILL} command takes
4040 precedence. @xref{Output Section Fill}, for details on the fill
4043 @node Output Section Keywords
4044 @subsection Output Section Keywords
4045 There are a couple of keywords which can appear as output section
4049 @kindex CREATE_OBJECT_SYMBOLS
4050 @cindex input filename symbols
4051 @cindex filename symbols
4052 @item CREATE_OBJECT_SYMBOLS
4053 The command tells the linker to create a symbol for each input file.
4054 The name of each symbol will be the name of the corresponding input
4055 file. The section of each symbol will be the output section in which
4056 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4058 This is conventional for the a.out object file format. It is not
4059 normally used for any other object file format.
4061 @kindex CONSTRUCTORS
4062 @cindex C++ constructors, arranging in link
4063 @cindex constructors, arranging in link
4065 When linking using the a.out object file format, the linker uses an
4066 unusual set construct to support C++ global constructors and
4067 destructors. When linking object file formats which do not support
4068 arbitrary sections, such as ECOFF and XCOFF, the linker will
4069 automatically recognize C++ global constructors and destructors by name.
4070 For these object file formats, the @code{CONSTRUCTORS} command tells the
4071 linker to place constructor information in the output section where the
4072 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4073 ignored for other object file formats.
4075 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4076 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4077 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4078 the start and end of the global destructors. The
4079 first word in the list is the number of entries, followed by the address
4080 of each constructor or destructor, followed by a zero word. The
4081 compiler must arrange to actually run the code. For these object file
4082 formats @sc{gnu} C++ normally calls constructors from a subroutine
4083 @code{__main}; a call to @code{__main} is automatically inserted into
4084 the startup code for @code{main}. @sc{gnu} C++ normally runs
4085 destructors either by using @code{atexit}, or directly from the function
4088 For object file formats such as @code{COFF} or @code{ELF} which support
4089 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4090 addresses of global constructors and destructors into the @code{.ctors}
4091 and @code{.dtors} sections. Placing the following sequence into your
4092 linker script will build the sort of table which the @sc{gnu} C++
4093 runtime code expects to see.
4097 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4102 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4108 If you are using the @sc{gnu} C++ support for initialization priority,
4109 which provides some control over the order in which global constructors
4110 are run, you must sort the constructors at link time to ensure that they
4111 are executed in the correct order. When using the @code{CONSTRUCTORS}
4112 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4113 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4114 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4117 Normally the compiler and linker will handle these issues automatically,
4118 and you will not need to concern yourself with them. However, you may
4119 need to consider this if you are using C++ and writing your own linker
4124 @node Output Section Discarding
4125 @subsection Output Section Discarding
4126 @cindex discarding sections
4127 @cindex sections, discarding
4128 @cindex removing sections
4129 The linker will not create output sections with no contents. This is
4130 for convenience when referring to input sections that may or may not
4131 be present in any of the input files. For example:
4133 .foo : @{ *(.foo) @}
4136 will only create a @samp{.foo} section in the output file if there is a
4137 @samp{.foo} section in at least one input file, and if the input
4138 sections are not all empty. Other link script directives that allocate
4139 space in an output section will also create the output section.
4141 The linker will ignore address assignments (@pxref{Output Section Address})
4142 on discarded output sections, except when the linker script defines
4143 symbols in the output section. In that case the linker will obey
4144 the address assignments, possibly advancing dot even though the
4145 section is discarded.
4148 The special output section name @samp{/DISCARD/} may be used to discard
4149 input sections. Any input sections which are assigned to an output
4150 section named @samp{/DISCARD/} are not included in the output file.
4152 @node Output Section Attributes
4153 @subsection Output Section Attributes
4154 @cindex output section attributes
4155 We showed above that the full description of an output section looked
4159 @var{section} [@var{address}] [(@var{type})] :
4160 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
4162 @var{output-section-command}
4163 @var{output-section-command}
4165 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4168 We've already described @var{section}, @var{address}, and
4169 @var{output-section-command}. In this section we will describe the
4170 remaining section attributes.
4173 * Output Section Type:: Output section type
4174 * Output Section LMA:: Output section LMA
4175 * Forced Output Alignment:: Forced Output Alignment
4176 * Forced Input Alignment:: Forced Input Alignment
4177 * Output Section Region:: Output section region
4178 * Output Section Phdr:: Output section phdr
4179 * Output Section Fill:: Output section fill
4182 @node Output Section Type
4183 @subsubsection Output Section Type
4184 Each output section may have a type. The type is a keyword in
4185 parentheses. The following types are defined:
4189 The section should be marked as not loadable, so that it will not be
4190 loaded into memory when the program is run.
4195 These type names are supported for backward compatibility, and are
4196 rarely used. They all have the same effect: the section should be
4197 marked as not allocatable, so that no memory is allocated for the
4198 section when the program is run.
4202 @cindex prevent unnecessary loading
4203 @cindex loading, preventing
4204 The linker normally sets the attributes of an output section based on
4205 the input sections which map into it. You can override this by using
4206 the section type. For example, in the script sample below, the
4207 @samp{ROM} section is addressed at memory location @samp{0} and does not
4208 need to be loaded when the program is run. The contents of the
4209 @samp{ROM} section will appear in the linker output file as usual.
4213 ROM 0 (NOLOAD) : @{ @dots{} @}
4219 @node Output Section LMA
4220 @subsubsection Output Section LMA
4221 @kindex AT>@var{lma_region}
4222 @kindex AT(@var{lma})
4223 @cindex load address
4224 @cindex section load address
4225 Every section has a virtual address (VMA) and a load address (LMA); see
4226 @ref{Basic Script Concepts}. The address expression which may appear in
4227 an output section description sets the VMA (@pxref{Output Section
4230 The expression @var{lma} that follows the @code{AT} keyword specifies
4231 the load address of the section.
4233 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
4234 specify a memory region for the section's load address. @xref{MEMORY}.
4235 Note that if the section has not had a VMA assigned to it then the
4236 linker will use the @var{lma_region} as the VMA region as well.
4238 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4239 section, the linker will set the LMA such that the difference between
4240 VMA and LMA for the section is the same as the preceding output
4241 section in the same region. If there is no preceding output section
4242 or the section is not allocatable, the linker will set the LMA equal
4244 @xref{Output Section Region}.
4246 @cindex ROM initialized data
4247 @cindex initialized data in ROM
4248 This feature is designed to make it easy to build a ROM image. For
4249 example, the following linker script creates three output sections: one
4250 called @samp{.text}, which starts at @code{0x1000}, one called
4251 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
4252 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
4253 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
4254 defined with the value @code{0x2000}, which shows that the location
4255 counter holds the VMA value, not the LMA value.
4261 .text 0x1000 : @{ *(.text) _etext = . ; @}
4263 AT ( ADDR (.text) + SIZEOF (.text) )
4264 @{ _data = . ; *(.data); _edata = . ; @}
4266 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
4271 The run-time initialization code for use with a program generated with
4272 this linker script would include something like the following, to copy
4273 the initialized data from the ROM image to its runtime address. Notice
4274 how this code takes advantage of the symbols defined by the linker
4279 extern char _etext, _data, _edata, _bstart, _bend;
4280 char *src = &_etext;
4283 /* ROM has data at end of text; copy it. */
4284 while (dst < &_edata) @{
4289 for (dst = &_bstart; dst< &_bend; dst++)
4294 @node Forced Output Alignment
4295 @subsubsection Forced Output Alignment
4296 @kindex ALIGN(@var{section_align})
4297 @cindex forcing output section alignment
4298 @cindex output section alignment
4299 You can increase an output section's alignment by using ALIGN.
4301 @node Forced Input Alignment
4302 @subsubsection Forced Input Alignment
4303 @kindex SUBALIGN(@var{subsection_align})
4304 @cindex forcing input section alignment
4305 @cindex input section alignment
4306 You can force input section alignment within an output section by using
4307 SUBALIGN. The value specified overrides any alignment given by input
4308 sections, whether larger or smaller.
4310 @node Output Section Region
4311 @subsubsection Output Section Region
4312 @kindex >@var{region}
4313 @cindex section, assigning to memory region
4314 @cindex memory regions and sections
4315 You can assign a section to a previously defined region of memory by
4316 using @samp{>@var{region}}. @xref{MEMORY}.
4318 Here is a simple example:
4321 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4322 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4326 @node Output Section Phdr
4327 @subsubsection Output Section Phdr
4329 @cindex section, assigning to program header
4330 @cindex program headers and sections
4331 You can assign a section to a previously defined program segment by
4332 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4333 one or more segments, then all subsequent allocated sections will be
4334 assigned to those segments as well, unless they use an explicitly
4335 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4336 linker to not put the section in any segment at all.
4338 Here is a simple example:
4341 PHDRS @{ text PT_LOAD ; @}
4342 SECTIONS @{ .text : @{ *(.text) @} :text @}
4346 @node Output Section Fill
4347 @subsubsection Output Section Fill
4348 @kindex =@var{fillexp}
4349 @cindex section fill pattern
4350 @cindex fill pattern, entire section
4351 You can set the fill pattern for an entire section by using
4352 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4353 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4354 within the output section (for example, gaps left due to the required
4355 alignment of input sections) will be filled with the value, repeated as
4356 necessary. If the fill expression is a simple hex number, ie. a string
4357 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4358 an arbitrarily long sequence of hex digits can be used to specify the
4359 fill pattern; Leading zeros become part of the pattern too. For all
4360 other cases, including extra parentheses or a unary @code{+}, the fill
4361 pattern is the four least significant bytes of the value of the
4362 expression. In all cases, the number is big-endian.
4364 You can also change the fill value with a @code{FILL} command in the
4365 output section commands; (@pxref{Output Section Data}).
4367 Here is a simple example:
4370 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4374 @node Overlay Description
4375 @subsection Overlay Description
4378 An overlay description provides an easy way to describe sections which
4379 are to be loaded as part of a single memory image but are to be run at
4380 the same memory address. At run time, some sort of overlay manager will
4381 copy the overlaid sections in and out of the runtime memory address as
4382 required, perhaps by simply manipulating addressing bits. This approach
4383 can be useful, for example, when a certain region of memory is faster
4386 Overlays are described using the @code{OVERLAY} command. The
4387 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4388 output section description. The full syntax of the @code{OVERLAY}
4389 command is as follows:
4392 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4396 @var{output-section-command}
4397 @var{output-section-command}
4399 @} [:@var{phdr}@dots{}] [=@var{fill}]
4402 @var{output-section-command}
4403 @var{output-section-command}
4405 @} [:@var{phdr}@dots{}] [=@var{fill}]
4407 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4411 Everything is optional except @code{OVERLAY} (a keyword), and each
4412 section must have a name (@var{secname1} and @var{secname2} above). The
4413 section definitions within the @code{OVERLAY} construct are identical to
4414 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
4415 except that no addresses and no memory regions may be defined for
4416 sections within an @code{OVERLAY}.
4418 The sections are all defined with the same starting address. The load
4419 addresses of the sections are arranged such that they are consecutive in
4420 memory starting at the load address used for the @code{OVERLAY} as a
4421 whole (as with normal section definitions, the load address is optional,
4422 and defaults to the start address; the start address is also optional,
4423 and defaults to the current value of the location counter).
4425 If the @code{NOCROSSREFS} keyword is used, and there any references
4426 among the sections, the linker will report an error. Since the sections
4427 all run at the same address, it normally does not make sense for one
4428 section to refer directly to another. @xref{Miscellaneous Commands,
4431 For each section within the @code{OVERLAY}, the linker automatically
4432 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4433 defined as the starting load address of the section. The symbol
4434 @code{__load_stop_@var{secname}} is defined as the final load address of
4435 the section. Any characters within @var{secname} which are not legal
4436 within C identifiers are removed. C (or assembler) code may use these
4437 symbols to move the overlaid sections around as necessary.
4439 At the end of the overlay, the value of the location counter is set to
4440 the start address of the overlay plus the size of the largest section.
4442 Here is an example. Remember that this would appear inside a
4443 @code{SECTIONS} construct.
4446 OVERLAY 0x1000 : AT (0x4000)
4448 .text0 @{ o1/*.o(.text) @}
4449 .text1 @{ o2/*.o(.text) @}
4454 This will define both @samp{.text0} and @samp{.text1} to start at
4455 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4456 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4457 following symbols will be defined if referenced: @code{__load_start_text0},
4458 @code{__load_stop_text0}, @code{__load_start_text1},
4459 @code{__load_stop_text1}.
4461 C code to copy overlay @code{.text1} into the overlay area might look
4466 extern char __load_start_text1, __load_stop_text1;
4467 memcpy ((char *) 0x1000, &__load_start_text1,
4468 &__load_stop_text1 - &__load_start_text1);
4472 Note that the @code{OVERLAY} command is just syntactic sugar, since
4473 everything it does can be done using the more basic commands. The above
4474 example could have been written identically as follows.
4478 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4479 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4480 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4481 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4482 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4483 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4484 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4489 @section MEMORY Command
4491 @cindex memory regions
4492 @cindex regions of memory
4493 @cindex allocating memory
4494 @cindex discontinuous memory
4495 The linker's default configuration permits allocation of all available
4496 memory. You can override this by using the @code{MEMORY} command.
4498 The @code{MEMORY} command describes the location and size of blocks of
4499 memory in the target. You can use it to describe which memory regions
4500 may be used by the linker, and which memory regions it must avoid. You
4501 can then assign sections to particular memory regions. The linker will
4502 set section addresses based on the memory regions, and will warn about
4503 regions that become too full. The linker will not shuffle sections
4504 around to fit into the available regions.
4506 A linker script may contain at most one use of the @code{MEMORY}
4507 command. However, you can define as many blocks of memory within it as
4508 you wish. The syntax is:
4513 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4519 The @var{name} is a name used in the linker script to refer to the
4520 region. The region name has no meaning outside of the linker script.
4521 Region names are stored in a separate name space, and will not conflict
4522 with symbol names, file names, or section names. Each memory region
4523 must have a distinct name within the @code{MEMORY} command. However you can
4524 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
4527 @cindex memory region attributes
4528 The @var{attr} string is an optional list of attributes that specify
4529 whether to use a particular memory region for an input section which is
4530 not explicitly mapped in the linker script. As described in
4531 @ref{SECTIONS}, if you do not specify an output section for some input
4532 section, the linker will create an output section with the same name as
4533 the input section. If you define region attributes, the linker will use
4534 them to select the memory region for the output section that it creates.
4536 The @var{attr} string must consist only of the following characters:
4551 Invert the sense of any of the preceding attributes
4554 If a unmapped section matches any of the listed attributes other than
4555 @samp{!}, it will be placed in the memory region. The @samp{!}
4556 attribute reverses this test, so that an unmapped section will be placed
4557 in the memory region only if it does not match any of the listed
4563 The @var{origin} is an numerical expression for the start address of
4564 the memory region. The expression must evaluate to a constant and it
4565 cannot involve any symbols. The keyword @code{ORIGIN} may be
4566 abbreviated to @code{org} or @code{o} (but not, for example,
4572 The @var{len} is an expression for the size in bytes of the memory
4573 region. As with the @var{origin} expression, the expression must
4574 be numerical only and must evaluate to a constant. The keyword
4575 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4577 In the following example, we specify that there are two memory regions
4578 available for allocation: one starting at @samp{0} for 256 kilobytes,
4579 and the other starting at @samp{0x40000000} for four megabytes. The
4580 linker will place into the @samp{rom} memory region every section which
4581 is not explicitly mapped into a memory region, and is either read-only
4582 or executable. The linker will place other sections which are not
4583 explicitly mapped into a memory region into the @samp{ram} memory
4590 rom (rx) : ORIGIN = 0, LENGTH = 256K
4591 ram (!rx) : org = 0x40000000, l = 4M
4596 Once you define a memory region, you can direct the linker to place
4597 specific output sections into that memory region by using the
4598 @samp{>@var{region}} output section attribute. For example, if you have
4599 a memory region named @samp{mem}, you would use @samp{>mem} in the
4600 output section definition. @xref{Output Section Region}. If no address
4601 was specified for the output section, the linker will set the address to
4602 the next available address within the memory region. If the combined
4603 output sections directed to a memory region are too large for the
4604 region, the linker will issue an error message.
4606 It is possible to access the origin and length of a memory in an
4607 expression via the @code{ORIGIN(@var{memory})} and
4608 @code{LENGTH(@var{memory})} functions:
4612 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4617 @section PHDRS Command
4619 @cindex program headers
4620 @cindex ELF program headers
4621 @cindex program segments
4622 @cindex segments, ELF
4623 The ELF object file format uses @dfn{program headers}, also knows as
4624 @dfn{segments}. The program headers describe how the program should be
4625 loaded into memory. You can print them out by using the @code{objdump}
4626 program with the @samp{-p} option.
4628 When you run an ELF program on a native ELF system, the system loader
4629 reads the program headers in order to figure out how to load the
4630 program. This will only work if the program headers are set correctly.
4631 This manual does not describe the details of how the system loader
4632 interprets program headers; for more information, see the ELF ABI.
4634 The linker will create reasonable program headers by default. However,
4635 in some cases, you may need to specify the program headers more
4636 precisely. You may use the @code{PHDRS} command for this purpose. When
4637 the linker sees the @code{PHDRS} command in the linker script, it will
4638 not create any program headers other than the ones specified.
4640 The linker only pays attention to the @code{PHDRS} command when
4641 generating an ELF output file. In other cases, the linker will simply
4642 ignore @code{PHDRS}.
4644 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4645 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4651 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4652 [ FLAGS ( @var{flags} ) ] ;
4657 The @var{name} is used only for reference in the @code{SECTIONS} command
4658 of the linker script. It is not put into the output file. Program
4659 header names are stored in a separate name space, and will not conflict
4660 with symbol names, file names, or section names. Each program header
4661 must have a distinct name.
4663 Certain program header types describe segments of memory which the
4664 system loader will load from the file. In the linker script, you
4665 specify the contents of these segments by placing allocatable output
4666 sections in the segments. You use the @samp{:@var{phdr}} output section
4667 attribute to place a section in a particular segment. @xref{Output
4670 It is normal to put certain sections in more than one segment. This
4671 merely implies that one segment of memory contains another. You may
4672 repeat @samp{:@var{phdr}}, using it once for each segment which should
4673 contain the section.
4675 If you place a section in one or more segments using @samp{:@var{phdr}},
4676 then the linker will place all subsequent allocatable sections which do
4677 not specify @samp{:@var{phdr}} in the same segments. This is for
4678 convenience, since generally a whole set of contiguous sections will be
4679 placed in a single segment. You can use @code{:NONE} to override the
4680 default segment and tell the linker to not put the section in any
4685 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4686 the program header type to further describe the contents of the segment.
4687 The @code{FILEHDR} keyword means that the segment should include the ELF
4688 file header. The @code{PHDRS} keyword means that the segment should
4689 include the ELF program headers themselves.
4691 The @var{type} may be one of the following. The numbers indicate the
4692 value of the keyword.
4695 @item @code{PT_NULL} (0)
4696 Indicates an unused program header.
4698 @item @code{PT_LOAD} (1)
4699 Indicates that this program header describes a segment to be loaded from
4702 @item @code{PT_DYNAMIC} (2)
4703 Indicates a segment where dynamic linking information can be found.
4705 @item @code{PT_INTERP} (3)
4706 Indicates a segment where the name of the program interpreter may be
4709 @item @code{PT_NOTE} (4)
4710 Indicates a segment holding note information.
4712 @item @code{PT_SHLIB} (5)
4713 A reserved program header type, defined but not specified by the ELF
4716 @item @code{PT_PHDR} (6)
4717 Indicates a segment where the program headers may be found.
4719 @item @var{expression}
4720 An expression giving the numeric type of the program header. This may
4721 be used for types not defined above.
4724 You can specify that a segment should be loaded at a particular address
4725 in memory by using an @code{AT} expression. This is identical to the
4726 @code{AT} command used as an output section attribute (@pxref{Output
4727 Section LMA}). The @code{AT} command for a program header overrides the
4728 output section attribute.
4730 The linker will normally set the segment flags based on the sections
4731 which comprise the segment. You may use the @code{FLAGS} keyword to
4732 explicitly specify the segment flags. The value of @var{flags} must be
4733 an integer. It is used to set the @code{p_flags} field of the program
4736 Here is an example of @code{PHDRS}. This shows a typical set of program
4737 headers used on a native ELF system.
4743 headers PT_PHDR PHDRS ;
4745 text PT_LOAD FILEHDR PHDRS ;
4747 dynamic PT_DYNAMIC ;
4753 .interp : @{ *(.interp) @} :text :interp
4754 .text : @{ *(.text) @} :text
4755 .rodata : @{ *(.rodata) @} /* defaults to :text */
4757 . = . + 0x1000; /* move to a new page in memory */
4758 .data : @{ *(.data) @} :data
4759 .dynamic : @{ *(.dynamic) @} :data :dynamic
4766 @section VERSION Command
4767 @kindex VERSION @{script text@}
4768 @cindex symbol versions
4769 @cindex version script
4770 @cindex versions of symbols
4771 The linker supports symbol versions when using ELF. Symbol versions are
4772 only useful when using shared libraries. The dynamic linker can use
4773 symbol versions to select a specific version of a function when it runs
4774 a program that may have been linked against an earlier version of the
4777 You can include a version script directly in the main linker script, or
4778 you can supply the version script as an implicit linker script. You can
4779 also use the @samp{--version-script} linker option.
4781 The syntax of the @code{VERSION} command is simply
4783 VERSION @{ version-script-commands @}
4786 The format of the version script commands is identical to that used by
4787 Sun's linker in Solaris 2.5. The version script defines a tree of
4788 version nodes. You specify the node names and interdependencies in the
4789 version script. You can specify which symbols are bound to which
4790 version nodes, and you can reduce a specified set of symbols to local
4791 scope so that they are not globally visible outside of the shared
4794 The easiest way to demonstrate the version script language is with a few
4815 "int f(int, double)";
4820 This example version script defines three version nodes. The first
4821 version node defined is @samp{VERS_1.1}; it has no other dependencies.
4822 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
4823 a number of symbols to local scope so that they are not visible outside
4824 of the shared library; this is done using wildcard patterns, so that any
4825 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4826 is matched. The wildcard patterns available are the same as those used
4827 in the shell when matching filenames (also known as ``globbing'').
4828 However, if you specify the symbol name inside double quotes, then the
4829 name is treated as literal, rather than as a glob pattern.
4831 Next, the version script defines node @samp{VERS_1.2}. This node
4832 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
4833 to the version node @samp{VERS_1.2}.
4835 Finally, the version script defines node @samp{VERS_2.0}. This node
4836 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
4837 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4839 When the linker finds a symbol defined in a library which is not
4840 specifically bound to a version node, it will effectively bind it to an
4841 unspecified base version of the library. You can bind all otherwise
4842 unspecified symbols to a given version node by using @samp{global: *;}
4843 somewhere in the version script. Note that it's slightly crazy to use
4844 wildcards in a global spec except on the last version node. Global
4845 wildcards elsewhere run the risk of accidentally adding symbols to the
4846 set exported for an old version. That's wrong since older versions
4847 ought to have a fixed set of symbols.
4849 The names of the version nodes have no specific meaning other than what
4850 they might suggest to the person reading them. The @samp{2.0} version
4851 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4852 However, this would be a confusing way to write a version script.
4854 Node name can be omitted, provided it is the only version node
4855 in the version script. Such version script doesn't assign any versions to
4856 symbols, only selects which symbols will be globally visible out and which
4860 @{ global: foo; bar; local: *; @};
4863 When you link an application against a shared library that has versioned
4864 symbols, the application itself knows which version of each symbol it
4865 requires, and it also knows which version nodes it needs from each
4866 shared library it is linked against. Thus at runtime, the dynamic
4867 loader can make a quick check to make sure that the libraries you have
4868 linked against do in fact supply all of the version nodes that the
4869 application will need to resolve all of the dynamic symbols. In this
4870 way it is possible for the dynamic linker to know with certainty that
4871 all external symbols that it needs will be resolvable without having to
4872 search for each symbol reference.
4874 The symbol versioning is in effect a much more sophisticated way of
4875 doing minor version checking that SunOS does. The fundamental problem
4876 that is being addressed here is that typically references to external
4877 functions are bound on an as-needed basis, and are not all bound when
4878 the application starts up. If a shared library is out of date, a
4879 required interface may be missing; when the application tries to use
4880 that interface, it may suddenly and unexpectedly fail. With symbol
4881 versioning, the user will get a warning when they start their program if
4882 the libraries being used with the application are too old.
4884 There are several GNU extensions to Sun's versioning approach. The
4885 first of these is the ability to bind a symbol to a version node in the
4886 source file where the symbol is defined instead of in the versioning
4887 script. This was done mainly to reduce the burden on the library
4888 maintainer. You can do this by putting something like:
4890 __asm__(".symver original_foo,foo@@VERS_1.1");
4893 in the C source file. This renames the function @samp{original_foo} to
4894 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4895 The @samp{local:} directive can be used to prevent the symbol
4896 @samp{original_foo} from being exported. A @samp{.symver} directive
4897 takes precedence over a version script.
4899 The second GNU extension is to allow multiple versions of the same
4900 function to appear in a given shared library. In this way you can make
4901 an incompatible change to an interface without increasing the major
4902 version number of the shared library, while still allowing applications
4903 linked against the old interface to continue to function.
4905 To do this, you must use multiple @samp{.symver} directives in the
4906 source file. Here is an example:
4909 __asm__(".symver original_foo,foo@@");
4910 __asm__(".symver old_foo,foo@@VERS_1.1");
4911 __asm__(".symver old_foo1,foo@@VERS_1.2");
4912 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4915 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4916 unspecified base version of the symbol. The source file that contains this
4917 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4918 @samp{old_foo1}, and @samp{new_foo}.
4920 When you have multiple definitions of a given symbol, there needs to be
4921 some way to specify a default version to which external references to
4922 this symbol will be bound. You can do this with the
4923 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4924 declare one version of a symbol as the default in this manner; otherwise
4925 you would effectively have multiple definitions of the same symbol.
4927 If you wish to bind a reference to a specific version of the symbol
4928 within the shared library, you can use the aliases of convenience
4929 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4930 specifically bind to an external version of the function in question.
4932 You can also specify the language in the version script:
4935 VERSION extern "lang" @{ version-script-commands @}
4938 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4939 The linker will iterate over the list of symbols at the link time and
4940 demangle them according to @samp{lang} before matching them to the
4941 patterns specified in @samp{version-script-commands}.
4943 Demangled names may contains spaces and other special characters. As
4944 described above, you can use a glob pattern to match demangled names,
4945 or you can use a double-quoted string to match the string exactly. In
4946 the latter case, be aware that minor differences (such as differing
4947 whitespace) between the version script and the demangler output will
4948 cause a mismatch. As the exact string generated by the demangler
4949 might change in the future, even if the mangled name does not, you
4950 should check that all of your version directives are behaving as you
4951 expect when you upgrade.
4954 @section Expressions in Linker Scripts
4957 The syntax for expressions in the linker script language is identical to
4958 that of C expressions. All expressions are evaluated as integers. All
4959 expressions are evaluated in the same size, which is 32 bits if both the
4960 host and target are 32 bits, and is otherwise 64 bits.
4962 You can use and set symbol values in expressions.
4964 The linker defines several special purpose builtin functions for use in
4968 * Constants:: Constants
4969 * Symbols:: Symbol Names
4970 * Orphan Sections:: Orphan Sections
4971 * Location Counter:: The Location Counter
4972 * Operators:: Operators
4973 * Evaluation:: Evaluation
4974 * Expression Section:: The Section of an Expression
4975 * Builtin Functions:: Builtin Functions
4979 @subsection Constants
4980 @cindex integer notation
4981 @cindex constants in linker scripts
4982 All constants are integers.
4984 As in C, the linker considers an integer beginning with @samp{0} to be
4985 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4986 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
4987 @samp{H} for hexadeciaml, @samp{o} or @samp{O} for octal, @samp{b} or
4988 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
4989 value without a prefix or a suffix is considered to be decimal.
4991 @cindex scaled integers
4992 @cindex K and M integer suffixes
4993 @cindex M and K integer suffixes
4994 @cindex suffixes for integers
4995 @cindex integer suffixes
4996 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5000 @c END TEXI2ROFF-KILL
5001 @code{1024} or @code{1024*1024}
5005 ${\rm 1024}$ or ${\rm 1024}^2$
5007 @c END TEXI2ROFF-KILL
5008 respectively. For example, the following
5009 all refer to the same quantity:
5018 Note - the @code{K} and @code{M} suffixes cannot be used in
5019 conjunction with the base suffixes mentioned above.
5022 @subsection Symbol Names
5023 @cindex symbol names
5025 @cindex quoted symbol names
5027 Unless quoted, symbol names start with a letter, underscore, or period
5028 and may include letters, digits, underscores, periods, and hyphens.
5029 Unquoted symbol names must not conflict with any keywords. You can
5030 specify a symbol which contains odd characters or has the same name as a
5031 keyword by surrounding the symbol name in double quotes:
5034 "with a space" = "also with a space" + 10;
5037 Since symbols can contain many non-alphabetic characters, it is safest
5038 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5039 whereas @samp{A - B} is an expression involving subtraction.
5041 @node Orphan Sections
5042 @subsection Orphan Sections
5044 Orphan sections are sections present in the input files which
5045 are not explicitly placed into the output file by the linker
5046 script. The linker will still copy these sections into the
5047 output file, but it has to guess as to where they should be
5048 placed. The linker uses a simple heuristic to do this. It
5049 attempts to place orphan sections after non-orphan sections of the
5050 same attribute, such as code vs data, loadable vs non-loadable, etc.
5051 If there is not enough room to do this then it places
5052 at the end of the file.
5054 For ELF targets, the attribute of the section includes section type as
5055 well as section flag.
5057 If an orphaned section's name is representable as a C identifier then
5058 the linker will automatically @pxref{PROVIDE} two symbols:
5059 __start_SECNAME and __end_SECNAME, where SECNAME is the name of the
5060 section. These indicate the start address and end address of the
5061 orphaned section respectively. Note: most section names are not
5062 representable as C identifiers because they contain a @samp{.}
5065 @node Location Counter
5066 @subsection The Location Counter
5069 @cindex location counter
5070 @cindex current output location
5071 The special linker variable @dfn{dot} @samp{.} always contains the
5072 current output location counter. Since the @code{.} always refers to a
5073 location in an output section, it may only appear in an expression
5074 within a @code{SECTIONS} command. The @code{.} symbol may appear
5075 anywhere that an ordinary symbol is allowed in an expression.
5078 Assigning a value to @code{.} will cause the location counter to be
5079 moved. This may be used to create holes in the output section. The
5080 location counter may not be moved backwards inside an output section,
5081 and may not be moved backwards outside of an output section if so
5082 doing creates areas with overlapping LMAs.
5098 In the previous example, the @samp{.text} section from @file{file1} is
5099 located at the beginning of the output section @samp{output}. It is
5100 followed by a 1000 byte gap. Then the @samp{.text} section from
5101 @file{file2} appears, also with a 1000 byte gap following before the
5102 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5103 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5105 @cindex dot inside sections
5106 Note: @code{.} actually refers to the byte offset from the start of the
5107 current containing object. Normally this is the @code{SECTIONS}
5108 statement, whose start address is 0, hence @code{.} can be used as an
5109 absolute address. If @code{.} is used inside a section description
5110 however, it refers to the byte offset from the start of that section,
5111 not an absolute address. Thus in a script like this:
5129 The @samp{.text} section will be assigned a starting address of 0x100
5130 and a size of exactly 0x200 bytes, even if there is not enough data in
5131 the @samp{.text} input sections to fill this area. (If there is too
5132 much data, an error will be produced because this would be an attempt to
5133 move @code{.} backwards). The @samp{.data} section will start at 0x500
5134 and it will have an extra 0x600 bytes worth of space after the end of
5135 the values from the @samp{.data} input sections and before the end of
5136 the @samp{.data} output section itself.
5138 @cindex dot outside sections
5139 Setting symbols to the value of the location counter outside of an
5140 output section statement can result in unexpected values if the linker
5141 needs to place orphan sections. For example, given the following:
5147 .text: @{ *(.text) @}
5151 .data: @{ *(.data) @}
5156 If the linker needs to place some input section, e.g. @code{.rodata},
5157 not mentioned in the script, it might choose to place that section
5158 between @code{.text} and @code{.data}. You might think the linker
5159 should place @code{.rodata} on the blank line in the above script, but
5160 blank lines are of no particular significance to the linker. As well,
5161 the linker doesn't associate the above symbol names with their
5162 sections. Instead, it assumes that all assignments or other
5163 statements belong to the previous output section, except for the
5164 special case of an assignment to @code{.}. I.e., the linker will
5165 place the orphan @code{.rodata} section as if the script was written
5172 .text: @{ *(.text) @}
5176 .rodata: @{ *(.rodata) @}
5177 .data: @{ *(.data) @}
5182 This may or may not be the script author's intention for the value of
5183 @code{start_of_data}. One way to influence the orphan section
5184 placement is to assign the location counter to itself, as the linker
5185 assumes that an assignment to @code{.} is setting the start address of
5186 a following output section and thus should be grouped with that
5187 section. So you could write:
5193 .text: @{ *(.text) @}
5198 .data: @{ *(.data) @}
5203 Now, the orphan @code{.rodata} section will be placed between
5204 @code{end_of_text} and @code{start_of_data}.
5208 @subsection Operators
5209 @cindex operators for arithmetic
5210 @cindex arithmetic operators
5211 @cindex precedence in expressions
5212 The linker recognizes the standard C set of arithmetic operators, with
5213 the standard bindings and precedence levels:
5216 @c END TEXI2ROFF-KILL
5218 precedence associativity Operators Notes
5224 5 left == != > < <= >=
5230 11 right &= += -= *= /= (2)
5234 (1) Prefix operators
5235 (2) @xref{Assignments}.
5239 \vskip \baselineskip
5240 %"lispnarrowing" is the extra indent used generally for smallexample
5241 \hskip\lispnarrowing\vbox{\offinterlineskip
5244 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
5245 height2pt&\omit&&\omit&&\omit&\cr
5246 &Precedence&& Associativity &&{\rm Operators}&\cr
5247 height2pt&\omit&&\omit&&\omit&\cr
5249 height2pt&\omit&&\omit&&\omit&\cr
5251 % '176 is tilde, '~' in tt font
5252 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
5253 &2&&left&&* / \%&\cr
5256 &5&&left&&== != > < <= >=&\cr
5259 &8&&left&&{\&\&}&\cr
5262 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
5264 height2pt&\omit&&\omit&&\omit&\cr}
5269 @obeylines@parskip=0pt@parindent=0pt
5270 @dag@quad Prefix operators.
5271 @ddag@quad @xref{Assignments}.
5274 @c END TEXI2ROFF-KILL
5277 @subsection Evaluation
5278 @cindex lazy evaluation
5279 @cindex expression evaluation order
5280 The linker evaluates expressions lazily. It only computes the value of
5281 an expression when absolutely necessary.
5283 The linker needs some information, such as the value of the start
5284 address of the first section, and the origins and lengths of memory
5285 regions, in order to do any linking at all. These values are computed
5286 as soon as possible when the linker reads in the linker script.
5288 However, other values (such as symbol values) are not known or needed
5289 until after storage allocation. Such values are evaluated later, when
5290 other information (such as the sizes of output sections) is available
5291 for use in the symbol assignment expression.
5293 The sizes of sections cannot be known until after allocation, so
5294 assignments dependent upon these are not performed until after
5297 Some expressions, such as those depending upon the location counter
5298 @samp{.}, must be evaluated during section allocation.
5300 If the result of an expression is required, but the value is not
5301 available, then an error results. For example, a script like the
5307 .text 9+this_isnt_constant :
5313 will cause the error message @samp{non constant expression for initial
5316 @node Expression Section
5317 @subsection The Section of an Expression
5318 @cindex expression sections
5319 @cindex absolute expressions
5320 @cindex relative expressions
5321 @cindex absolute and relocatable symbols
5322 @cindex relocatable and absolute symbols
5323 @cindex symbols, relocatable and absolute
5324 When the linker evaluates an expression, the result is either absolute
5325 or relative to some section. A relative expression is expressed as a
5326 fixed offset from the base of a section.
5328 The position of the expression within the linker script determines
5329 whether it is absolute or relative. An expression which appears within
5330 an output section definition is relative to the base of the output
5331 section. An expression which appears elsewhere will be absolute.
5333 A symbol set to a relative expression will be relocatable if you request
5334 relocatable output using the @samp{-r} option. That means that a
5335 further link operation may change the value of the symbol. The symbol's
5336 section will be the section of the relative expression.
5338 A symbol set to an absolute expression will retain the same value
5339 through any further link operation. The symbol will be absolute, and
5340 will not have any particular associated section.
5342 You can use the builtin function @code{ABSOLUTE} to force an expression
5343 to be absolute when it would otherwise be relative. For example, to
5344 create an absolute symbol set to the address of the end of the output
5345 section @samp{.data}:
5349 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
5353 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
5354 @samp{.data} section.
5356 @node Builtin Functions
5357 @subsection Builtin Functions
5358 @cindex functions in expressions
5359 The linker script language includes a number of builtin functions for
5360 use in linker script expressions.
5363 @item ABSOLUTE(@var{exp})
5364 @kindex ABSOLUTE(@var{exp})
5365 @cindex expression, absolute
5366 Return the absolute (non-relocatable, as opposed to non-negative) value
5367 of the expression @var{exp}. Primarily useful to assign an absolute
5368 value to a symbol within a section definition, where symbol values are
5369 normally section relative. @xref{Expression Section}.
5371 @item ADDR(@var{section})
5372 @kindex ADDR(@var{section})
5373 @cindex section address in expression
5374 Return the absolute address (the VMA) of the named @var{section}. Your
5375 script must previously have defined the location of that section. In
5376 the following example, @code{symbol_1} and @code{symbol_2} are assigned
5383 start_of_output_1 = ABSOLUTE(.);
5388 symbol_1 = ADDR(.output1);
5389 symbol_2 = start_of_output_1;
5395 @item ALIGN(@var{align})
5396 @itemx ALIGN(@var{exp},@var{align})
5397 @kindex ALIGN(@var{align})
5398 @kindex ALIGN(@var{exp},@var{align})
5399 @cindex round up location counter
5400 @cindex align location counter
5401 @cindex round up expression
5402 @cindex align expression
5403 Return the location counter (@code{.}) or arbitrary expression aligned
5404 to the next @var{align} boundary. The single operand @code{ALIGN}
5405 doesn't change the value of the location counter---it just does
5406 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5407 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5408 equivalent to @code{ALIGN(., @var{align})}).
5410 Here is an example which aligns the output @code{.data} section to the
5411 next @code{0x2000} byte boundary after the preceding section and sets a
5412 variable within the section to the next @code{0x8000} boundary after the
5417 .data ALIGN(0x2000): @{
5419 variable = ALIGN(0x8000);
5425 The first use of @code{ALIGN} in this example specifies the location of
5426 a section because it is used as the optional @var{address} attribute of
5427 a section definition (@pxref{Output Section Address}). The second use
5428 of @code{ALIGN} is used to defines the value of a symbol.
5430 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5432 @item ALIGNOF(@var{section})
5433 @kindex ALIGNOF(@var{section})
5434 @cindex section alignment
5435 Return the alignment in bytes of the named @var{section}, if that section has
5436 been allocated. If the section has not been allocated when this is
5437 evaluated, the linker will report an error. In the following example,
5438 the alignment of the @code{.output} section is stored as the first
5439 value in that section.
5444 LONG (ALIGNOF (.output))
5451 @item BLOCK(@var{exp})
5452 @kindex BLOCK(@var{exp})
5453 This is a synonym for @code{ALIGN}, for compatibility with older linker
5454 scripts. It is most often seen when setting the address of an output
5457 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5458 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5459 This is equivalent to either
5461 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5465 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5468 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5469 for the data segment (area between the result of this expression and
5470 @code{DATA_SEGMENT_END}) than the former or not.
5471 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5472 memory will be saved at the expense of up to @var{commonpagesize} wasted
5473 bytes in the on-disk file.
5475 This expression can only be used directly in @code{SECTIONS} commands, not in
5476 any output section descriptions and only once in the linker script.
5477 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5478 be the system page size the object wants to be optimized for (while still
5479 working on system page sizes up to @var{maxpagesize}).
5484 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5487 @item DATA_SEGMENT_END(@var{exp})
5488 @kindex DATA_SEGMENT_END(@var{exp})
5489 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5490 evaluation purposes.
5493 . = DATA_SEGMENT_END(.);
5496 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5497 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5498 This defines the end of the @code{PT_GNU_RELRO} segment when
5499 @samp{-z relro} option is used. Second argument is returned.
5500 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5501 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5502 @var{exp} + @var{offset} is aligned to the most commonly used page
5503 boundary for particular target. If present in the linker script,
5504 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5505 @code{DATA_SEGMENT_END}.
5508 . = DATA_SEGMENT_RELRO_END(24, .);
5511 @item DEFINED(@var{symbol})
5512 @kindex DEFINED(@var{symbol})
5513 @cindex symbol defaults
5514 Return 1 if @var{symbol} is in the linker global symbol table and is
5515 defined before the statement using DEFINED in the script, otherwise
5516 return 0. You can use this function to provide
5517 default values for symbols. For example, the following script fragment
5518 shows how to set a global symbol @samp{begin} to the first location in
5519 the @samp{.text} section---but if a symbol called @samp{begin} already
5520 existed, its value is preserved:
5526 begin = DEFINED(begin) ? begin : . ;
5534 @item LENGTH(@var{memory})
5535 @kindex LENGTH(@var{memory})
5536 Return the length of the memory region named @var{memory}.
5538 @item LOADADDR(@var{section})
5539 @kindex LOADADDR(@var{section})
5540 @cindex section load address in expression
5541 Return the absolute LMA of the named @var{section}. This is normally
5542 the same as @code{ADDR}, but it may be different if the @code{AT}
5543 attribute is used in the output section definition (@pxref{Output
5547 @item MAX(@var{exp1}, @var{exp2})
5548 Returns the maximum of @var{exp1} and @var{exp2}.
5551 @item MIN(@var{exp1}, @var{exp2})
5552 Returns the minimum of @var{exp1} and @var{exp2}.
5554 @item NEXT(@var{exp})
5555 @kindex NEXT(@var{exp})
5556 @cindex unallocated address, next
5557 Return the next unallocated address that is a multiple of @var{exp}.
5558 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5559 use the @code{MEMORY} command to define discontinuous memory for the
5560 output file, the two functions are equivalent.
5562 @item ORIGIN(@var{memory})
5563 @kindex ORIGIN(@var{memory})
5564 Return the origin of the memory region named @var{memory}.
5566 @item SEGMENT_START(@var{segment}, @var{default})
5567 @kindex SEGMENT_START(@var{segment}, @var{default})
5568 Return the base address of the named @var{segment}. If an explicit
5569 value has been given for this segment (with a command-line @samp{-T}
5570 option) that value will be returned; otherwise the value will be
5571 @var{default}. At present, the @samp{-T} command-line option can only
5572 be used to set the base address for the ``text'', ``data'', and
5573 ``bss'' sections, but you use @code{SEGMENT_START} with any segment
5576 @item SIZEOF(@var{section})
5577 @kindex SIZEOF(@var{section})
5578 @cindex section size
5579 Return the size in bytes of the named @var{section}, if that section has
5580 been allocated. If the section has not been allocated when this is
5581 evaluated, the linker will report an error. In the following example,
5582 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5591 symbol_1 = .end - .start ;
5592 symbol_2 = SIZEOF(.output);
5597 @item SIZEOF_HEADERS
5598 @itemx sizeof_headers
5599 @kindex SIZEOF_HEADERS
5601 Return the size in bytes of the output file's headers. This is
5602 information which appears at the start of the output file. You can use
5603 this number when setting the start address of the first section, if you
5604 choose, to facilitate paging.
5606 @cindex not enough room for program headers
5607 @cindex program headers, not enough room
5608 When producing an ELF output file, if the linker script uses the
5609 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5610 number of program headers before it has determined all the section
5611 addresses and sizes. If the linker later discovers that it needs
5612 additional program headers, it will report an error @samp{not enough
5613 room for program headers}. To avoid this error, you must avoid using
5614 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5615 script to avoid forcing the linker to use additional program headers, or
5616 you must define the program headers yourself using the @code{PHDRS}
5617 command (@pxref{PHDRS}).
5620 @node Implicit Linker Scripts
5621 @section Implicit Linker Scripts
5622 @cindex implicit linker scripts
5623 If you specify a linker input file which the linker can not recognize as
5624 an object file or an archive file, it will try to read the file as a
5625 linker script. If the file can not be parsed as a linker script, the
5626 linker will report an error.
5628 An implicit linker script will not replace the default linker script.
5630 Typically an implicit linker script would contain only symbol
5631 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5634 Any input files read because of an implicit linker script will be read
5635 at the position in the command line where the implicit linker script was
5636 read. This can affect archive searching.
5639 @node Machine Dependent
5640 @chapter Machine Dependent Features
5642 @cindex machine dependencies
5643 @command{ld} has additional features on some platforms; the following
5644 sections describe them. Machines where @command{ld} has no additional
5645 functionality are not listed.
5649 * H8/300:: @command{ld} and the H8/300
5652 * i960:: @command{ld} and the Intel 960 family
5655 * ARM:: @command{ld} and the ARM family
5658 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
5661 * M68K:: @command{ld} and the Motorola 68K family
5664 * MMIX:: @command{ld} and MMIX
5667 * MSP430:: @command{ld} and MSP430
5670 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
5673 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
5676 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
5679 * SPU ELF:: @command{ld} and SPU ELF Support
5682 * TI COFF:: @command{ld} and TI COFF
5685 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
5688 * Xtensa:: @command{ld} and Xtensa Processors
5699 @section @command{ld} and the H8/300
5701 @cindex H8/300 support
5702 For the H8/300, @command{ld} can perform these global optimizations when
5703 you specify the @samp{--relax} command-line option.
5706 @cindex relaxing on H8/300
5707 @item relaxing address modes
5708 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5709 targets are within eight bits, and turns them into eight-bit
5710 program-counter relative @code{bsr} and @code{bra} instructions,
5713 @cindex synthesizing on H8/300
5714 @item synthesizing instructions
5715 @c FIXME: specifically mov.b, or any mov instructions really?
5716 @command{ld} finds all @code{mov.b} instructions which use the
5717 sixteen-bit absolute address form, but refer to the top
5718 page of memory, and changes them to use the eight-bit address form.
5719 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5720 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5721 top page of memory).
5723 @item bit manipulation instructions
5724 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
5725 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5726 which use 32 bit and 16 bit absolute address form, but refer to the top
5727 page of memory, and changes them to use the 8 bit address form.
5728 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5729 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5730 the top page of memory).
5732 @item system control instructions
5733 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
5734 32 bit absolute address form, but refer to the top page of memory, and
5735 changes them to use 16 bit address form.
5736 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5737 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5738 the top page of memory).
5748 @c This stuff is pointless to say unless you're especially concerned
5749 @c with Renesas chips; don't enable it for generic case, please.
5751 @chapter @command{ld} and Other Renesas Chips
5753 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5754 H8/500, and SH chips. No special features, commands, or command-line
5755 options are required for these chips.
5765 @section @command{ld} and the Intel 960 Family
5767 @cindex i960 support
5769 You can use the @samp{-A@var{architecture}} command line option to
5770 specify one of the two-letter names identifying members of the 960
5771 family; the option specifies the desired output target, and warns of any
5772 incompatible instructions in the input files. It also modifies the
5773 linker's search strategy for archive libraries, to support the use of
5774 libraries specific to each particular architecture, by including in the
5775 search loop names suffixed with the string identifying the architecture.
5777 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5778 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5779 paths, and in any paths you specify with @samp{-L}) for a library with
5792 The first two possibilities would be considered in any event; the last
5793 two are due to the use of @w{@samp{-ACA}}.
5795 You can meaningfully use @samp{-A} more than once on a command line, since
5796 the 960 architecture family allows combination of target architectures; each
5797 use will add another pair of name variants to search for when @w{@samp{-l}}
5798 specifies a library.
5800 @cindex @option{--relax} on i960
5801 @cindex relaxing on i960
5802 @command{ld} supports the @samp{--relax} option for the i960 family. If
5803 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5804 @code{calx} instructions whose targets are within 24 bits, and turns
5805 them into 24-bit program-counter relative @code{bal} and @code{cal}
5806 instructions, respectively. @command{ld} also turns @code{cal}
5807 instructions into @code{bal} instructions when it determines that the
5808 target subroutine is a leaf routine (that is, the target subroutine does
5809 not itself call any subroutines).
5826 @node M68HC11/68HC12
5827 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
5829 @cindex M68HC11 and 68HC12 support
5831 @subsection Linker Relaxation
5833 For the Motorola 68HC11, @command{ld} can perform these global
5834 optimizations when you specify the @samp{--relax} command-line option.
5837 @cindex relaxing on M68HC11
5838 @item relaxing address modes
5839 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5840 targets are within eight bits, and turns them into eight-bit
5841 program-counter relative @code{bsr} and @code{bra} instructions,
5844 @command{ld} also looks at all 16-bit extended addressing modes and
5845 transforms them in a direct addressing mode when the address is in
5846 page 0 (between 0 and 0x0ff).
5848 @item relaxing gcc instruction group
5849 When @command{gcc} is called with @option{-mrelax}, it can emit group
5850 of instructions that the linker can optimize to use a 68HC11 direct
5851 addressing mode. These instructions consists of @code{bclr} or
5852 @code{bset} instructions.
5856 @subsection Trampoline Generation
5858 @cindex trampoline generation on M68HC11
5859 @cindex trampoline generation on M68HC12
5860 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5861 call a far function using a normal @code{jsr} instruction. The linker
5862 will also change the relocation to some far function to use the
5863 trampoline address instead of the function address. This is typically the
5864 case when a pointer to a function is taken. The pointer will in fact
5865 point to the function trampoline.
5873 @section @command{ld} and the ARM family
5875 @cindex ARM interworking support
5876 @kindex --support-old-code
5877 For the ARM, @command{ld} will generate code stubs to allow functions calls
5878 between ARM and Thumb code. These stubs only work with code that has
5879 been compiled and assembled with the @samp{-mthumb-interwork} command
5880 line option. If it is necessary to link with old ARM object files or
5881 libraries, which have not been compiled with the -mthumb-interwork
5882 option then the @samp{--support-old-code} command line switch should be
5883 given to the linker. This will make it generate larger stub functions
5884 which will work with non-interworking aware ARM code. Note, however,
5885 the linker does not support generating stubs for function calls to
5886 non-interworking aware Thumb code.
5888 @cindex thumb entry point
5889 @cindex entry point, thumb
5890 @kindex --thumb-entry=@var{entry}
5891 The @samp{--thumb-entry} switch is a duplicate of the generic
5892 @samp{--entry} switch, in that it sets the program's starting address.
5893 But it also sets the bottom bit of the address, so that it can be
5894 branched to using a BX instruction, and the program will start
5895 executing in Thumb mode straight away.
5897 @cindex PE import table prefixing
5898 @kindex --use-nul-prefixed-import-tables
5899 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
5900 the import tables idata4 and idata5 have to be generated with a zero
5901 elememt prefix for import libraries. This is the old style to generate
5902 import tables. By default this option is turned off.
5906 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5907 executables. This option is only valid when linking big-endian objects.
5908 The resulting image will contain big-endian data and little-endian code.
5911 @kindex --target1-rel
5912 @kindex --target1-abs
5913 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5914 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
5915 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
5916 and @samp{--target1-abs} switches override the default.
5919 @kindex --target2=@var{type}
5920 The @samp{--target2=type} switch overrides the default definition of the
5921 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
5922 meanings, and target defaults are as follows:
5925 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5927 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
5929 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5934 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5935 specification) enables objects compiled for the ARMv4 architecture to be
5936 interworking-safe when linked with other objects compiled for ARMv4t, but
5937 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
5939 In the latter case, the switch @option{--fix-v4bx} must be passed to the
5940 linker, which causes v4t @code{BX rM} instructions to be rewritten as
5941 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
5943 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
5944 relocations are ignored.
5946 @cindex FIX_V4BX_INTERWORKING
5947 @kindex --fix-v4bx-interworking
5948 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
5949 relocations with a branch to the following veneer:
5957 This allows generation of libraries/applications that work on ARMv4 cores
5958 and are still interworking safe. Note that the above veneer clobbers the
5959 condition flags, so may cause incorrect progrm behavior in rare cases.
5963 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
5964 BLX instructions (available on ARMv5t and above) in various
5965 situations. Currently it is used to perform calls via the PLT from Thumb
5966 code using BLX rather than using BX and a mode-switching stub before
5967 each PLT entry. This should lead to such calls executing slightly faster.
5969 This option is enabled implicitly for SymbianOS, so there is no need to
5970 specify it if you are using that target.
5972 @cindex VFP11_DENORM_FIX
5973 @kindex --vfp11-denorm-fix
5974 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
5975 bug in certain VFP11 coprocessor hardware, which sometimes allows
5976 instructions with denorm operands (which must be handled by support code)
5977 to have those operands overwritten by subsequent instructions before
5978 the support code can read the intended values.
5980 The bug may be avoided in scalar mode if you allow at least one
5981 intervening instruction between a VFP11 instruction which uses a register
5982 and another instruction which writes to the same register, or at least two
5983 intervening instructions if vector mode is in use. The bug only affects
5984 full-compliance floating-point mode: you do not need this workaround if
5985 you are using "runfast" mode. Please contact ARM for further details.
5987 If you know you are using buggy VFP11 hardware, you can
5988 enable this workaround by specifying the linker option
5989 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
5990 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
5991 vector mode (the latter also works for scalar code). The default is
5992 @samp{--vfp-denorm-fix=none}.
5994 If the workaround is enabled, instructions are scanned for
5995 potentially-troublesome sequences, and a veneer is created for each
5996 such sequence which may trigger the erratum. The veneer consists of the
5997 first instruction of the sequence and a branch back to the subsequent
5998 instruction. The original instruction is then replaced with a branch to
5999 the veneer. The extra cycles required to call and return from the veneer
6000 are sufficient to avoid the erratum in both the scalar and vector cases.
6002 @cindex NO_ENUM_SIZE_WARNING
6003 @kindex --no-enum-size-warning
6004 The @option{--no-enum-size-warning} switch prevents the linker from
6005 warning when linking object files that specify incompatible EABI
6006 enumeration size attributes. For example, with this switch enabled,
6007 linking of an object file using 32-bit enumeration values with another
6008 using enumeration values fitted into the smallest possible space will
6011 @cindex NO_WCHAR_SIZE_WARNING
6012 @kindex --no-wchar-size-warning
6013 The @option{--no-wchar-size-warning} switch prevents the linker from
6014 warning when linking object files that specify incompatible EABI
6015 @code{wchar_t} size attributes. For example, with this switch enabled,
6016 linking of an object file using 32-bit @code{wchar_t} values with another
6017 using 16-bit @code{wchar_t} values will not be diagnosed.
6020 @kindex --pic-veneer
6021 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6022 ARM/Thumb interworking veneers, even if the rest of the binary
6023 is not PIC. This avoids problems on uClinux targets where
6024 @samp{--emit-relocs} is used to generate relocatable binaries.
6026 @cindex STUB_GROUP_SIZE
6027 @kindex --stub-group-size=@var{N}
6028 The linker will automatically generate and insert small sequences of
6029 code into a linked ARM ELF executable whenever an attempt is made to
6030 perform a function call to a symbol that is too far away. The
6031 placement of these sequences of instructions - called stubs - is
6032 controlled by the command line option @option{--stub-group-size=N}.
6033 The placement is important because a poor choice can create a need for
6034 duplicate stubs, increasing the code sizw. The linker will try to
6035 group stubs together in order to reduce interruptions to the flow of
6036 code, but it needs guidance as to how big these groups should be and
6037 where they should be placed.
6039 The value of @samp{N}, the parameter to the
6040 @option{--stub-group-size=} option controls where the stub groups are
6041 placed. If it is negative then all stubs are placed after the first
6042 branch that needs them. If it is positive then the stubs can be
6043 placed either before or after the branches that need them. If the
6044 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6045 exactly where to place groups of stubs, using its built in heuristics.
6046 A value of @samp{N} greater than 1 (or smaller than -1) tells the
6047 linker that a single group of stubs can service at most @samp{N} bytes
6048 from the input sections.
6050 The default, if @option{--stub-group-size=} is not specified, is
6053 Farcalls stubs insertion is fully supported for the ARM-EABI target
6054 only, because it relies on object files properties not present
6068 @section @command{ld} and HPPA 32-bit ELF Support
6069 @cindex HPPA multiple sub-space stubs
6070 @kindex --multi-subspace
6071 When generating a shared library, @command{ld} will by default generate
6072 import stubs suitable for use with a single sub-space application.
6073 The @samp{--multi-subspace} switch causes @command{ld} to generate export
6074 stubs, and different (larger) import stubs suitable for use with
6075 multiple sub-spaces.
6077 @cindex HPPA stub grouping
6078 @kindex --stub-group-size=@var{N}
6079 Long branch stubs and import/export stubs are placed by @command{ld} in
6080 stub sections located between groups of input sections.
6081 @samp{--stub-group-size} specifies the maximum size of a group of input
6082 sections handled by one stub section. Since branch offsets are signed,
6083 a stub section may serve two groups of input sections, one group before
6084 the stub section, and one group after it. However, when using
6085 conditional branches that require stubs, it may be better (for branch
6086 prediction) that stub sections only serve one group of input sections.
6087 A negative value for @samp{N} chooses this scheme, ensuring that
6088 branches to stubs always use a negative offset. Two special values of
6089 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6090 @command{ld} to automatically size input section groups for the branch types
6091 detected, with the same behaviour regarding stub placement as other
6092 positive or negative values of @samp{N} respectively.
6094 Note that @samp{--stub-group-size} does not split input sections. A
6095 single input section larger than the group size specified will of course
6096 create a larger group (of one section). If input sections are too
6097 large, it may not be possible for a branch to reach its stub.
6110 @section @command{ld} and the Motorola 68K family
6112 @cindex Motorola 68K GOT generation
6113 @kindex --got=@var{type}
6114 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
6115 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
6116 @samp{target}. When @samp{target} is selected the linker chooses
6117 the default GOT generation scheme for the current target.
6118 @samp{single} tells the linker to generate a single GOT with
6119 entries only at non-negative offsets.
6120 @samp{negative} instructs the linker to generate a single GOT with
6121 entries at both negative and positive offsets. Not all environments
6123 @samp{multigot} allows the linker to generate several GOTs in the
6124 output file. All GOT references from a single input object
6125 file access the same GOT, but references from different input object
6126 files might access different GOTs. Not all environments support such GOTs.
6139 @section @code{ld} and MMIX
6140 For MMIX, there is a choice of generating @code{ELF} object files or
6141 @code{mmo} object files when linking. The simulator @code{mmix}
6142 understands the @code{mmo} format. The binutils @code{objcopy} utility
6143 can translate between the two formats.
6145 There is one special section, the @samp{.MMIX.reg_contents} section.
6146 Contents in this section is assumed to correspond to that of global
6147 registers, and symbols referring to it are translated to special symbols,
6148 equal to registers. In a final link, the start address of the
6149 @samp{.MMIX.reg_contents} section corresponds to the first allocated
6150 global register multiplied by 8. Register @code{$255} is not included in
6151 this section; it is always set to the program entry, which is at the
6152 symbol @code{Main} for @code{mmo} files.
6154 Global symbols with the prefix @code{__.MMIX.start.}, for example
6155 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
6156 The default linker script uses these to set the default start address
6159 Initial and trailing multiples of zero-valued 32-bit words in a section,
6160 are left out from an mmo file.
6173 @section @code{ld} and MSP430
6174 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
6175 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
6176 just pass @samp{-m help} option to the linker).
6178 @cindex MSP430 extra sections
6179 The linker will recognize some extra sections which are MSP430 specific:
6182 @item @samp{.vectors}
6183 Defines a portion of ROM where interrupt vectors located.
6185 @item @samp{.bootloader}
6186 Defines the bootloader portion of the ROM (if applicable). Any code
6187 in this section will be uploaded to the MPU.
6189 @item @samp{.infomem}
6190 Defines an information memory section (if applicable). Any code in
6191 this section will be uploaded to the MPU.
6193 @item @samp{.infomemnobits}
6194 This is the same as the @samp{.infomem} section except that any code
6195 in this section will not be uploaded to the MPU.
6197 @item @samp{.noinit}
6198 Denotes a portion of RAM located above @samp{.bss} section.
6200 The last two sections are used by gcc.
6214 @section @command{ld} and PowerPC 32-bit ELF Support
6215 @cindex PowerPC long branches
6216 @kindex --relax on PowerPC
6217 Branches on PowerPC processors are limited to a signed 26-bit
6218 displacement, which may result in @command{ld} giving
6219 @samp{relocation truncated to fit} errors with very large programs.
6220 @samp{--relax} enables the generation of trampolines that can access
6221 the entire 32-bit address space. These trampolines are inserted at
6222 section boundaries, so may not themselves be reachable if an input
6223 section exceeds 33M in size.
6225 @cindex PowerPC ELF32 options
6230 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
6231 generates code capable of using a newer PLT and GOT layout that has
6232 the security advantage of no executable section ever needing to be
6233 writable and no writable section ever being executable. PowerPC
6234 @command{ld} will generate this layout, including stubs to access the
6235 PLT, if all input files (including startup and static libraries) were
6236 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
6237 BSS PLT (and GOT layout) which can give slightly better performance.
6239 @kindex --secure-plt
6241 @command{ld} will use the new PLT and GOT layout if it is linking new
6242 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
6243 when linking non-PIC code. This option requests the new PLT and GOT
6244 layout. A warning will be given if some object file requires the old
6250 The new secure PLT and GOT are placed differently relative to other
6251 sections compared to older BSS PLT and GOT placement. The location of
6252 @code{.plt} must change because the new secure PLT is an initialized
6253 section while the old PLT is uninitialized. The reason for the
6254 @code{.got} change is more subtle: The new placement allows
6255 @code{.got} to be read-only in applications linked with
6256 @samp{-z relro -z now}. However, this placement means that
6257 @code{.sdata} cannot always be used in shared libraries, because the
6258 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
6259 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
6260 GCC doesn't use @code{.sdata} in shared libraries, so this option is
6261 really only useful for other compilers that may do so.
6263 @cindex PowerPC stub symbols
6264 @kindex --emit-stub-syms
6265 @item --emit-stub-syms
6266 This option causes @command{ld} to label linker stubs with a local
6267 symbol that encodes the stub type and destination.
6269 @cindex PowerPC TLS optimization
6270 @kindex --no-tls-optimize
6271 @item --no-tls-optimize
6272 PowerPC @command{ld} normally performs some optimization of code
6273 sequences used to access Thread-Local Storage. Use this option to
6274 disable the optimization.
6287 @node PowerPC64 ELF64
6288 @section @command{ld} and PowerPC64 64-bit ELF Support
6290 @cindex PowerPC64 ELF64 options
6292 @cindex PowerPC64 stub grouping
6293 @kindex --stub-group-size
6294 @item --stub-group-size
6295 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
6296 by @command{ld} in stub sections located between groups of input sections.
6297 @samp{--stub-group-size} specifies the maximum size of a group of input
6298 sections handled by one stub section. Since branch offsets are signed,
6299 a stub section may serve two groups of input sections, one group before
6300 the stub section, and one group after it. However, when using
6301 conditional branches that require stubs, it may be better (for branch
6302 prediction) that stub sections only serve one group of input sections.
6303 A negative value for @samp{N} chooses this scheme, ensuring that
6304 branches to stubs always use a negative offset. Two special values of
6305 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6306 @command{ld} to automatically size input section groups for the branch types
6307 detected, with the same behaviour regarding stub placement as other
6308 positive or negative values of @samp{N} respectively.
6310 Note that @samp{--stub-group-size} does not split input sections. A
6311 single input section larger than the group size specified will of course
6312 create a larger group (of one section). If input sections are too
6313 large, it may not be possible for a branch to reach its stub.
6315 @cindex PowerPC64 stub symbols
6316 @kindex --emit-stub-syms
6317 @item --emit-stub-syms
6318 This option causes @command{ld} to label linker stubs with a local
6319 symbol that encodes the stub type and destination.
6321 @cindex PowerPC64 dot symbols
6323 @kindex --no-dotsyms
6324 @item --dotsyms, --no-dotsyms
6325 These two options control how @command{ld} interprets version patterns
6326 in a version script. Older PowerPC64 compilers emitted both a
6327 function descriptor symbol with the same name as the function, and a
6328 code entry symbol with the name prefixed by a dot (@samp{.}). To
6329 properly version a function @samp{foo}, the version script thus needs
6330 to control both @samp{foo} and @samp{.foo}. The option
6331 @samp{--dotsyms}, on by default, automatically adds the required
6332 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
6335 @cindex PowerPC64 TLS optimization
6336 @kindex --no-tls-optimize
6337 @item --no-tls-optimize
6338 PowerPC64 @command{ld} normally performs some optimization of code
6339 sequences used to access Thread-Local Storage. Use this option to
6340 disable the optimization.
6342 @cindex PowerPC64 OPD optimization
6343 @kindex --no-opd-optimize
6344 @item --no-opd-optimize
6345 PowerPC64 @command{ld} normally removes @code{.opd} section entries
6346 corresponding to deleted link-once functions, or functions removed by
6347 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
6348 Use this option to disable @code{.opd} optimization.
6350 @cindex PowerPC64 OPD spacing
6351 @kindex --non-overlapping-opd
6352 @item --non-overlapping-opd
6353 Some PowerPC64 compilers have an option to generate compressed
6354 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
6355 the static chain pointer (unused in C) with the first word of the next
6356 entry. This option expands such entries to the full 24 bytes.
6358 @cindex PowerPC64 TOC optimization
6359 @kindex --no-toc-optimize
6360 @item --no-toc-optimize
6361 PowerPC64 @command{ld} normally removes unused @code{.toc} section
6362 entries. Such entries are detected by examining relocations that
6363 reference the TOC in code sections. A reloc in a deleted code section
6364 marks a TOC word as unneeded, while a reloc in a kept code section
6365 marks a TOC word as needed. Since the TOC may reference itself, TOC
6366 relocs are also examined. TOC words marked as both needed and
6367 unneeded will of course be kept. TOC words without any referencing
6368 reloc are assumed to be part of a multi-word entry, and are kept or
6369 discarded as per the nearest marked preceding word. This works
6370 reliably for compiler generated code, but may be incorrect if assembly
6371 code is used to insert TOC entries. Use this option to disable the
6374 @cindex PowerPC64 multi-TOC
6375 @kindex --no-multi-toc
6376 @item --no-multi-toc
6377 By default, PowerPC64 GCC generates code for a TOC model where TOC
6378 entries are accessed with a 16-bit offset from r2. This limits the
6379 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
6380 grouping code sections such that each group uses less than 64K for its
6381 TOC entries, then inserts r2 adjusting stubs between inter-group
6382 calls. @command{ld} does not split apart input sections, so cannot
6383 help if a single input file has a @code{.toc} section that exceeds
6384 64K, most likely from linking multiple files with @command{ld -r}.
6385 Use this option to turn off this feature.
6399 @section @command{ld} and SPU ELF Support
6401 @cindex SPU ELF options
6407 This option marks an executable as a PIC plugin module.
6409 @cindex SPU overlays
6410 @kindex --no-overlays
6412 Normally, @command{ld} recognizes calls to functions within overlay
6413 regions, and redirects such calls to an overlay manager via a stub.
6414 @command{ld} also provides a built-in overlay manager. This option
6415 turns off all this special overlay handling.
6417 @cindex SPU overlay stub symbols
6418 @kindex --emit-stub-syms
6419 @item --emit-stub-syms
6420 This option causes @command{ld} to label overlay stubs with a local
6421 symbol that encodes the stub type and destination.
6423 @cindex SPU extra overlay stubs
6424 @kindex --extra-overlay-stubs
6425 @item --extra-overlay-stubs
6426 This option causes @command{ld} to add overlay call stubs on all
6427 function calls out of overlay regions. Normally stubs are not added
6428 on calls to non-overlay regions.
6430 @cindex SPU local store size
6431 @kindex --local-store=lo:hi
6432 @item --local-store=lo:hi
6433 @command{ld} usually checks that a final executable for SPU fits in
6434 the address range 0 to 256k. This option may be used to change the
6435 range. Disable the check entirely with @option{--local-store=0:0}.
6438 @kindex --stack-analysis
6439 @item --stack-analysis
6440 SPU local store space is limited. Over-allocation of stack space
6441 unnecessarily limits space available for code and data, while
6442 under-allocation results in runtime failures. If given this option,
6443 @command{ld} will provide an estimate of maximum stack usage.
6444 @command{ld} does this by examining symbols in code sections to
6445 determine the extents of functions, and looking at function prologues
6446 for stack adjusting instructions. A call-graph is created by looking
6447 for relocations on branch instructions. The graph is then searched
6448 for the maximum stack usage path. Note that this analysis does not
6449 find calls made via function pointers, and does not handle recursion
6450 and other cycles in the call graph. Stack usage may be
6451 under-estimated if your code makes such calls. Also, stack usage for
6452 dynamic allocation, e.g. alloca, will not be detected. If a link map
6453 is requested, detailed information about each function's stack usage
6454 and calls will be given.
6457 @kindex --emit-stack-syms
6458 @item --emit-stack-syms
6459 This option, if given along with @option{--stack-analysis} will result
6460 in @command{ld} emitting stack sizing symbols for each function.
6461 These take the form @code{__stack_<function_name>} for global
6462 functions, and @code{__stack_<number>_<function_name>} for static
6463 functions. @code{<number>} is the section id in hex. The value of
6464 such symbols is the stack requirement for the corresponding function.
6465 The symbol size will be zero, type @code{STT_NOTYPE}, binding
6466 @code{STB_LOCAL}, and section @code{SHN_ABS}.
6480 @section @command{ld}'s Support for Various TI COFF Versions
6481 @cindex TI COFF versions
6482 @kindex --format=@var{version}
6483 The @samp{--format} switch allows selection of one of the various
6484 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
6485 also supported. The TI COFF versions also vary in header byte-order
6486 format; @command{ld} will read any version or byte order, but the output
6487 header format depends on the default specified by the specific target.
6500 @section @command{ld} and WIN32 (cygwin/mingw)
6502 This section describes some of the win32 specific @command{ld} issues.
6503 See @ref{Options,,Command Line Options} for detailed description of the
6504 command line options mentioned here.
6507 @cindex import libraries
6508 @item import libraries
6509 The standard Windows linker creates and uses so-called import
6510 libraries, which contains information for linking to dll's. They are
6511 regular static archives and are handled as any other static
6512 archive. The cygwin and mingw ports of @command{ld} have specific
6513 support for creating such libraries provided with the
6514 @samp{--out-implib} command line option.
6516 @item exporting DLL symbols
6517 @cindex exporting DLL symbols
6518 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6521 @item using auto-export functionality
6522 @cindex using auto-export functionality
6523 By default @command{ld} exports symbols with the auto-export functionality,
6524 which is controlled by the following command line options:
6527 @item --export-all-symbols [This is the default]
6528 @item --exclude-symbols
6529 @item --exclude-libs
6530 @item --exclude-modules-for-implib
6533 If, however, @samp{--export-all-symbols} is not given explicitly on the
6534 command line, then the default auto-export behavior will be @emph{disabled}
6535 if either of the following are true:
6538 @item A DEF file is used.
6539 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
6542 @item using a DEF file
6543 @cindex using a DEF file
6544 Another way of exporting symbols is using a DEF file. A DEF file is
6545 an ASCII file containing definitions of symbols which should be
6546 exported when a dll is created. Usually it is named @samp{<dll
6547 name>.def} and is added as any other object file to the linker's
6548 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
6551 gcc -o <output> <objectfiles> <dll name>.def
6554 Using a DEF file turns off the normal auto-export behavior, unless the
6555 @samp{--export-all-symbols} option is also used.
6557 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
6560 LIBRARY "xyz.dll" BASE=0x20000000
6566 another_foo = abc.dll.afoo
6570 This example defines a DLL with a non-default base address and five
6571 symbols in the export table. The third exported symbol @code{_bar} is an
6572 alias for the second. The fourth symbol, @code{another_foo} is resolved
6573 by "forwarding" to another module and treating it as an alias for
6574 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
6575 @code{var1} is declared to be a data object.
6577 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
6578 name of the output DLL. If @samp{<name>} does not include a suffix,
6579 the default library suffix, @samp{.DLL} is appended.
6581 When the .DEF file is used to build an application, rather than a
6582 library, the @code{NAME <name>} command should be used instead of
6583 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
6584 executable suffix, @samp{.EXE} is appended.
6586 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
6587 specification @code{BASE = <number>} may be used to specify a
6588 non-default base address for the image.
6590 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
6591 or they specify an empty string, the internal name is the same as the
6592 filename specified on the command line.
6594 The complete specification of an export symbol is:
6598 ( ( ( <name1> [ = <name2> ] )
6599 | ( <name1> = <module-name> . <external-name>))
6600 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
6603 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
6604 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
6605 @samp{<name1>} as a "forward" alias for the symbol
6606 @samp{<external-name>} in the DLL @samp{<module-name>}.
6607 Optionally, the symbol may be exported by the specified ordinal
6608 @samp{<integer>} alias.
6610 The optional keywords that follow the declaration indicate:
6612 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
6613 will still be exported by its ordinal alias (either the value specified
6614 by the .def specification or, otherwise, the value assigned by the
6615 linker). The symbol name, however, does remain visible in the import
6616 library (if any), unless @code{PRIVATE} is also specified.
6618 @code{DATA}: The symbol is a variable or object, rather than a function.
6619 The import lib will export only an indirect reference to @code{foo} as
6620 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
6623 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
6624 well as @code{_imp__foo} into the import library. Both refer to the
6625 read-only import address table's pointer to the variable, not to the
6626 variable itself. This can be dangerous. If the user code fails to add
6627 the @code{dllimport} attribute and also fails to explicitly add the
6628 extra indirection that the use of the attribute enforces, the
6629 application will behave unexpectedly.
6631 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
6632 it into the static import library used to resolve imports at link time. The
6633 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
6634 API at runtime or by by using the GNU ld extension of linking directly to
6635 the DLL without an import library.
6637 See ld/deffilep.y in the binutils sources for the full specification of
6638 other DEF file statements
6640 @cindex creating a DEF file
6641 While linking a shared dll, @command{ld} is able to create a DEF file
6642 with the @samp{--output-def <file>} command line option.
6644 @item Using decorations
6645 @cindex Using decorations
6646 Another way of marking symbols for export is to modify the source code
6647 itself, so that when building the DLL each symbol to be exported is
6651 __declspec(dllexport) int a_variable
6652 __declspec(dllexport) void a_function(int with_args)
6655 All such symbols will be exported from the DLL. If, however,
6656 any of the object files in the DLL contain symbols decorated in
6657 this way, then the normal auto-export behavior is disabled, unless
6658 the @samp{--export-all-symbols} option is also used.
6660 Note that object files that wish to access these symbols must @emph{not}
6661 decorate them with dllexport. Instead, they should use dllimport,
6665 __declspec(dllimport) int a_variable
6666 __declspec(dllimport) void a_function(int with_args)
6669 This complicates the structure of library header files, because
6670 when included by the library itself the header must declare the
6671 variables and functions as dllexport, but when included by client
6672 code the header must declare them as dllimport. There are a number
6673 of idioms that are typically used to do this; often client code can
6674 omit the __declspec() declaration completely. See
6675 @samp{--enable-auto-import} and @samp{automatic data imports} for more
6679 @cindex automatic data imports
6680 @item automatic data imports
6681 The standard Windows dll format supports data imports from dlls only
6682 by adding special decorations (dllimport/dllexport), which let the
6683 compiler produce specific assembler instructions to deal with this
6684 issue. This increases the effort necessary to port existing Un*x
6685 code to these platforms, especially for large
6686 c++ libraries and applications. The auto-import feature, which was
6687 initially provided by Paul Sokolovsky, allows one to omit the
6688 decorations to achieve a behavior that conforms to that on POSIX/Un*x
6689 platforms. This feature is enabled with the @samp{--enable-auto-import}
6690 command-line option, although it is enabled by default on cygwin/mingw.
6691 The @samp{--enable-auto-import} option itself now serves mainly to
6692 suppress any warnings that are ordinarily emitted when linked objects
6693 trigger the feature's use.
6695 auto-import of variables does not always work flawlessly without
6696 additional assistance. Sometimes, you will see this message
6698 "variable '<var>' can't be auto-imported. Please read the
6699 documentation for ld's @code{--enable-auto-import} for details."
6701 The @samp{--enable-auto-import} documentation explains why this error
6702 occurs, and several methods that can be used to overcome this difficulty.
6703 One of these methods is the @emph{runtime pseudo-relocs} feature, described
6706 @cindex runtime pseudo-relocation
6707 For complex variables imported from DLLs (such as structs or classes),
6708 object files typically contain a base address for the variable and an
6709 offset (@emph{addend}) within the variable--to specify a particular
6710 field or public member, for instance. Unfortunately, the runtime loader used
6711 in win32 environments is incapable of fixing these references at runtime
6712 without the additional information supplied by dllimport/dllexport decorations.
6713 The standard auto-import feature described above is unable to resolve these
6716 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
6717 be resolved without error, while leaving the task of adjusting the references
6718 themselves (with their non-zero addends) to specialized code provided by the
6719 runtime environment. Recent versions of the cygwin and mingw environments and
6720 compilers provide this runtime support; older versions do not. However, the
6721 support is only necessary on the developer's platform; the compiled result will
6722 run without error on an older system.
6724 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
6727 @cindex direct linking to a dll
6728 @item direct linking to a dll
6729 The cygwin/mingw ports of @command{ld} support the direct linking,
6730 including data symbols, to a dll without the usage of any import
6731 libraries. This is much faster and uses much less memory than does the
6732 traditional import library method, especially when linking large
6733 libraries or applications. When @command{ld} creates an import lib, each
6734 function or variable exported from the dll is stored in its own bfd, even
6735 though a single bfd could contain many exports. The overhead involved in
6736 storing, loading, and processing so many bfd's is quite large, and explains the
6737 tremendous time, memory, and storage needed to link against particularly
6738 large or complex libraries when using import libs.
6740 Linking directly to a dll uses no extra command-line switches other than
6741 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
6742 of names to match each library. All that is needed from the developer's
6743 perspective is an understanding of this search, in order to force ld to
6744 select the dll instead of an import library.
6747 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
6748 to find, in the first directory of its search path,
6760 before moving on to the next directory in the search path.
6762 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
6763 where @samp{<prefix>} is set by the @command{ld} option
6764 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6765 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6768 Other win32-based unix environments, such as mingw or pw32, may use other
6769 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
6770 was originally intended to help avoid name conflicts among dll's built for the
6771 various win32/un*x environments, so that (for example) two versions of a zlib dll
6772 could coexist on the same machine.
6774 The generic cygwin/mingw path layout uses a @samp{bin} directory for
6775 applications and dll's and a @samp{lib} directory for the import
6776 libraries (using cygwin nomenclature):
6782 libxxx.dll.a (in case of dll's)
6783 libxxx.a (in case of static archive)
6786 Linking directly to a dll without using the import library can be
6789 1. Use the dll directly by adding the @samp{bin} path to the link line
6791 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6794 However, as the dll's often have version numbers appended to their names
6795 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6796 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
6797 not versioned, and do not have this difficulty.
6799 2. Create a symbolic link from the dll to a file in the @samp{lib}
6800 directory according to the above mentioned search pattern. This
6801 should be used to avoid unwanted changes in the tools needed for
6805 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6808 Then you can link without any make environment changes.
6811 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6814 This technique also avoids the version number problems, because the following is
6821 libxxx.dll.a -> ../bin/cygxxx-5.dll
6824 Linking directly to a dll without using an import lib will work
6825 even when auto-import features are exercised, and even when
6826 @samp{--enable-runtime-pseudo-relocs} is used.
6828 Given the improvements in speed and memory usage, one might justifiably
6829 wonder why import libraries are used at all. There are three reasons:
6831 1. Until recently, the link-directly-to-dll functionality did @emph{not}
6832 work with auto-imported data.
6834 2. Sometimes it is necessary to include pure static objects within the
6835 import library (which otherwise contains only bfd's for indirection
6836 symbols that point to the exports of a dll). Again, the import lib
6837 for the cygwin kernel makes use of this ability, and it is not
6838 possible to do this without an import lib.
6840 3. Symbol aliases can only be resolved using an import lib. This is
6841 critical when linking against OS-supplied dll's (eg, the win32 API)
6842 in which symbols are usually exported as undecorated aliases of their
6843 stdcall-decorated assembly names.
6845 So, import libs are not going away. But the ability to replace
6846 true import libs with a simple symbolic link to (or a copy of)
6847 a dll, in many cases, is a useful addition to the suite of tools
6848 binutils makes available to the win32 developer. Given the
6849 massive improvements in memory requirements during linking, storage
6850 requirements, and linking speed, we expect that many developers
6851 will soon begin to use this feature whenever possible.
6853 @item symbol aliasing
6855 @item adding additional names
6856 Sometimes, it is useful to export symbols with additional names.
6857 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6858 exported as @samp{_foo} by using special directives in the DEF file
6859 when creating the dll. This will affect also the optional created
6860 import library. Consider the following DEF file:
6863 LIBRARY "xyz.dll" BASE=0x61000000
6870 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6872 Another method for creating a symbol alias is to create it in the
6873 source code using the "weak" attribute:
6876 void foo () @{ /* Do something. */; @}
6877 void _foo () __attribute__ ((weak, alias ("foo")));
6880 See the gcc manual for more information about attributes and weak
6883 @item renaming symbols
6884 Sometimes it is useful to rename exports. For instance, the cygwin
6885 kernel does this regularly. A symbol @samp{_foo} can be exported as
6886 @samp{foo} but not as @samp{_foo} by using special directives in the
6887 DEF file. (This will also affect the import library, if it is
6888 created). In the following example:
6891 LIBRARY "xyz.dll" BASE=0x61000000
6897 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6901 Note: using a DEF file disables the default auto-export behavior,
6902 unless the @samp{--export-all-symbols} command line option is used.
6903 If, however, you are trying to rename symbols, then you should list
6904 @emph{all} desired exports in the DEF file, including the symbols
6905 that are not being renamed, and do @emph{not} use the
6906 @samp{--export-all-symbols} option. If you list only the
6907 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6908 to handle the other symbols, then the both the new names @emph{and}
6909 the original names for the renamed symbols will be exported.
6910 In effect, you'd be aliasing those symbols, not renaming them,
6911 which is probably not what you wanted.
6913 @cindex weak externals
6914 @item weak externals
6915 The Windows object format, PE, specifies a form of weak symbols called
6916 weak externals. When a weak symbol is linked and the symbol is not
6917 defined, the weak symbol becomes an alias for some other symbol. There
6918 are three variants of weak externals:
6920 @item Definition is searched for in objects and libraries, historically
6921 called lazy externals.
6922 @item Definition is searched for only in other objects, not in libraries.
6923 This form is not presently implemented.
6924 @item No search; the symbol is an alias. This form is not presently
6927 As a GNU extension, weak symbols that do not specify an alternate symbol
6928 are supported. If the symbol is undefined when linking, the symbol
6929 uses a default value.
6943 @section @code{ld} and Xtensa Processors
6945 @cindex Xtensa processors
6946 The default @command{ld} behavior for Xtensa processors is to interpret
6947 @code{SECTIONS} commands so that lists of explicitly named sections in a
6948 specification with a wildcard file will be interleaved when necessary to
6949 keep literal pools within the range of PC-relative load offsets. For
6950 example, with the command:
6962 @command{ld} may interleave some of the @code{.literal}
6963 and @code{.text} sections from different object files to ensure that the
6964 literal pools are within the range of PC-relative load offsets. A valid
6965 interleaving might place the @code{.literal} sections from an initial
6966 group of files followed by the @code{.text} sections of that group of
6967 files. Then, the @code{.literal} sections from the rest of the files
6968 and the @code{.text} sections from the rest of the files would follow.
6970 @cindex @option{--relax} on Xtensa
6971 @cindex relaxing on Xtensa
6972 Relaxation is enabled by default for the Xtensa version of @command{ld} and
6973 provides two important link-time optimizations. The first optimization
6974 is to combine identical literal values to reduce code size. A redundant
6975 literal will be removed and all the @code{L32R} instructions that use it
6976 will be changed to reference an identical literal, as long as the
6977 location of the replacement literal is within the offset range of all
6978 the @code{L32R} instructions. The second optimization is to remove
6979 unnecessary overhead from assembler-generated ``longcall'' sequences of
6980 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
6981 range of direct @code{CALL@var{n}} instructions.
6983 For each of these cases where an indirect call sequence can be optimized
6984 to a direct call, the linker will change the @code{CALLX@var{n}}
6985 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
6986 instruction, and remove the literal referenced by the @code{L32R}
6987 instruction if it is not used for anything else. Removing the
6988 @code{L32R} instruction always reduces code size but can potentially
6989 hurt performance by changing the alignment of subsequent branch targets.
6990 By default, the linker will always preserve alignments, either by
6991 switching some instructions between 24-bit encodings and the equivalent
6992 density instructions or by inserting a no-op in place of the @code{L32R}
6993 instruction that was removed. If code size is more important than
6994 performance, the @option{--size-opt} option can be used to prevent the
6995 linker from widening density instructions or inserting no-ops, except in
6996 a few cases where no-ops are required for correctness.
6998 The following Xtensa-specific command-line options can be used to
7001 @cindex Xtensa options
7005 Since the Xtensa version of @code{ld} enables the @option{--relax} option
7006 by default, the @option{--no-relax} option is provided to disable
7010 When optimizing indirect calls to direct calls, optimize for code size
7011 more than performance. With this option, the linker will not insert
7012 no-ops or widen density instructions to preserve branch target
7013 alignment. There may still be some cases where no-ops are required to
7014 preserve the correctness of the code.
7022 @ifclear SingleFormat
7027 @cindex object file management
7028 @cindex object formats available
7030 The linker accesses object and archive files using the BFD libraries.
7031 These libraries allow the linker to use the same routines to operate on
7032 object files whatever the object file format. A different object file
7033 format can be supported simply by creating a new BFD back end and adding
7034 it to the library. To conserve runtime memory, however, the linker and
7035 associated tools are usually configured to support only a subset of the
7036 object file formats available. You can use @code{objdump -i}
7037 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
7038 list all the formats available for your configuration.
7040 @cindex BFD requirements
7041 @cindex requirements for BFD
7042 As with most implementations, BFD is a compromise between
7043 several conflicting requirements. The major factor influencing
7044 BFD design was efficiency: any time used converting between
7045 formats is time which would not have been spent had BFD not
7046 been involved. This is partly offset by abstraction payback; since
7047 BFD simplifies applications and back ends, more time and care
7048 may be spent optimizing algorithms for a greater speed.
7050 One minor artifact of the BFD solution which you should bear in
7051 mind is the potential for information loss. There are two places where
7052 useful information can be lost using the BFD mechanism: during
7053 conversion and during output. @xref{BFD information loss}.
7056 * BFD outline:: How it works: an outline of BFD
7060 @section How It Works: An Outline of BFD
7061 @cindex opening object files
7062 @include bfdsumm.texi
7065 @node Reporting Bugs
7066 @chapter Reporting Bugs
7067 @cindex bugs in @command{ld}
7068 @cindex reporting bugs in @command{ld}
7070 Your bug reports play an essential role in making @command{ld} reliable.
7072 Reporting a bug may help you by bringing a solution to your problem, or
7073 it may not. But in any case the principal function of a bug report is
7074 to help the entire community by making the next version of @command{ld}
7075 work better. Bug reports are your contribution to the maintenance of
7078 In order for a bug report to serve its purpose, you must include the
7079 information that enables us to fix the bug.
7082 * Bug Criteria:: Have you found a bug?
7083 * Bug Reporting:: How to report bugs
7087 @section Have You Found a Bug?
7088 @cindex bug criteria
7090 If you are not sure whether you have found a bug, here are some guidelines:
7093 @cindex fatal signal
7094 @cindex linker crash
7095 @cindex crash of linker
7097 If the linker gets a fatal signal, for any input whatever, that is a
7098 @command{ld} bug. Reliable linkers never crash.
7100 @cindex error on valid input
7102 If @command{ld} produces an error message for valid input, that is a bug.
7104 @cindex invalid input
7106 If @command{ld} does not produce an error message for invalid input, that
7107 may be a bug. In the general case, the linker can not verify that
7108 object files are correct.
7111 If you are an experienced user of linkers, your suggestions for
7112 improvement of @command{ld} are welcome in any case.
7116 @section How to Report Bugs
7118 @cindex @command{ld} bugs, reporting
7120 A number of companies and individuals offer support for @sc{gnu}
7121 products. If you obtained @command{ld} from a support organization, we
7122 recommend you contact that organization first.
7124 You can find contact information for many support companies and
7125 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
7129 Otherwise, send bug reports for @command{ld} to
7133 The fundamental principle of reporting bugs usefully is this:
7134 @strong{report all the facts}. If you are not sure whether to state a
7135 fact or leave it out, state it!
7137 Often people omit facts because they think they know what causes the
7138 problem and assume that some details do not matter. Thus, you might
7139 assume that the name of a symbol you use in an example does not
7140 matter. Well, probably it does not, but one cannot be sure. Perhaps
7141 the bug is a stray memory reference which happens to fetch from the
7142 location where that name is stored in memory; perhaps, if the name
7143 were different, the contents of that location would fool the linker
7144 into doing the right thing despite the bug. Play it safe and give a
7145 specific, complete example. That is the easiest thing for you to do,
7146 and the most helpful.
7148 Keep in mind that the purpose of a bug report is to enable us to fix
7149 the bug if it is new to us. Therefore, always write your bug reports
7150 on the assumption that the bug has not been reported previously.
7152 Sometimes people give a few sketchy facts and ask, ``Does this ring a
7153 bell?'' This cannot help us fix a bug, so it is basically useless. We
7154 respond by asking for enough details to enable us to investigate.
7155 You might as well expedite matters by sending them to begin with.
7157 To enable us to fix the bug, you should include all these things:
7161 The version of @command{ld}. @command{ld} announces it if you start it with
7162 the @samp{--version} argument.
7164 Without this, we will not know whether there is any point in looking for
7165 the bug in the current version of @command{ld}.
7168 Any patches you may have applied to the @command{ld} source, including any
7169 patches made to the @code{BFD} library.
7172 The type of machine you are using, and the operating system name and
7176 What compiler (and its version) was used to compile @command{ld}---e.g.
7180 The command arguments you gave the linker to link your example and
7181 observe the bug. To guarantee you will not omit something important,
7182 list them all. A copy of the Makefile (or the output from make) is
7185 If we were to try to guess the arguments, we would probably guess wrong
7186 and then we might not encounter the bug.
7189 A complete input file, or set of input files, that will reproduce the
7190 bug. It is generally most helpful to send the actual object files
7191 provided that they are reasonably small. Say no more than 10K. For
7192 bigger files you can either make them available by FTP or HTTP or else
7193 state that you are willing to send the object file(s) to whomever
7194 requests them. (Note - your email will be going to a mailing list, so
7195 we do not want to clog it up with large attachments). But small
7196 attachments are best.
7198 If the source files were assembled using @code{gas} or compiled using
7199 @code{gcc}, then it may be OK to send the source files rather than the
7200 object files. In this case, be sure to say exactly what version of
7201 @code{gas} or @code{gcc} was used to produce the object files. Also say
7202 how @code{gas} or @code{gcc} were configured.
7205 A description of what behavior you observe that you believe is
7206 incorrect. For example, ``It gets a fatal signal.''
7208 Of course, if the bug is that @command{ld} gets a fatal signal, then we
7209 will certainly notice it. But if the bug is incorrect output, we might
7210 not notice unless it is glaringly wrong. You might as well not give us
7211 a chance to make a mistake.
7213 Even if the problem you experience is a fatal signal, you should still
7214 say so explicitly. Suppose something strange is going on, such as, your
7215 copy of @command{ld} is out of sync, or you have encountered a bug in the
7216 C library on your system. (This has happened!) Your copy might crash
7217 and ours would not. If you told us to expect a crash, then when ours
7218 fails to crash, we would know that the bug was not happening for us. If
7219 you had not told us to expect a crash, then we would not be able to draw
7220 any conclusion from our observations.
7223 If you wish to suggest changes to the @command{ld} source, send us context
7224 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
7225 @samp{-p} option. Always send diffs from the old file to the new file.
7226 If you even discuss something in the @command{ld} source, refer to it by
7227 context, not by line number.
7229 The line numbers in our development sources will not match those in your
7230 sources. Your line numbers would convey no useful information to us.
7233 Here are some things that are not necessary:
7237 A description of the envelope of the bug.
7239 Often people who encounter a bug spend a lot of time investigating
7240 which changes to the input file will make the bug go away and which
7241 changes will not affect it.
7243 This is often time consuming and not very useful, because the way we
7244 will find the bug is by running a single example under the debugger
7245 with breakpoints, not by pure deduction from a series of examples.
7246 We recommend that you save your time for something else.
7248 Of course, if you can find a simpler example to report @emph{instead}
7249 of the original one, that is a convenience for us. Errors in the
7250 output will be easier to spot, running under the debugger will take
7251 less time, and so on.
7253 However, simplification is not vital; if you do not want to do this,
7254 report the bug anyway and send us the entire test case you used.
7257 A patch for the bug.
7259 A patch for the bug does help us if it is a good one. But do not omit
7260 the necessary information, such as the test case, on the assumption that
7261 a patch is all we need. We might see problems with your patch and decide
7262 to fix the problem another way, or we might not understand it at all.
7264 Sometimes with a program as complicated as @command{ld} it is very hard to
7265 construct an example that will make the program follow a certain path
7266 through the code. If you do not send us the example, we will not be
7267 able to construct one, so we will not be able to verify that the bug is
7270 And if we cannot understand what bug you are trying to fix, or why your
7271 patch should be an improvement, we will not install it. A test case will
7272 help us to understand.
7275 A guess about what the bug is or what it depends on.
7277 Such guesses are usually wrong. Even we cannot guess right about such
7278 things without first using the debugger to find the facts.
7282 @appendix MRI Compatible Script Files
7283 @cindex MRI compatibility
7284 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
7285 linker, @command{ld} can use MRI compatible linker scripts as an
7286 alternative to the more general-purpose linker scripting language
7287 described in @ref{Scripts}. MRI compatible linker scripts have a much
7288 simpler command set than the scripting language otherwise used with
7289 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
7290 linker commands; these commands are described here.
7292 In general, MRI scripts aren't of much use with the @code{a.out} object
7293 file format, since it only has three sections and MRI scripts lack some
7294 features to make use of them.
7296 You can specify a file containing an MRI-compatible script using the
7297 @samp{-c} command-line option.
7299 Each command in an MRI-compatible script occupies its own line; each
7300 command line starts with the keyword that identifies the command (though
7301 blank lines are also allowed for punctuation). If a line of an
7302 MRI-compatible script begins with an unrecognized keyword, @command{ld}
7303 issues a warning message, but continues processing the script.
7305 Lines beginning with @samp{*} are comments.
7307 You can write these commands using all upper-case letters, or all
7308 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
7309 The following list shows only the upper-case form of each command.
7312 @cindex @code{ABSOLUTE} (MRI)
7313 @item ABSOLUTE @var{secname}
7314 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
7315 Normally, @command{ld} includes in the output file all sections from all
7316 the input files. However, in an MRI-compatible script, you can use the
7317 @code{ABSOLUTE} command to restrict the sections that will be present in
7318 your output program. If the @code{ABSOLUTE} command is used at all in a
7319 script, then only the sections named explicitly in @code{ABSOLUTE}
7320 commands will appear in the linker output. You can still use other
7321 input sections (whatever you select on the command line, or using
7322 @code{LOAD}) to resolve addresses in the output file.
7324 @cindex @code{ALIAS} (MRI)
7325 @item ALIAS @var{out-secname}, @var{in-secname}
7326 Use this command to place the data from input section @var{in-secname}
7327 in a section called @var{out-secname} in the linker output file.
7329 @var{in-secname} may be an integer.
7331 @cindex @code{ALIGN} (MRI)
7332 @item ALIGN @var{secname} = @var{expression}
7333 Align the section called @var{secname} to @var{expression}. The
7334 @var{expression} should be a power of two.
7336 @cindex @code{BASE} (MRI)
7337 @item BASE @var{expression}
7338 Use the value of @var{expression} as the lowest address (other than
7339 absolute addresses) in the output file.
7341 @cindex @code{CHIP} (MRI)
7342 @item CHIP @var{expression}
7343 @itemx CHIP @var{expression}, @var{expression}
7344 This command does nothing; it is accepted only for compatibility.
7346 @cindex @code{END} (MRI)
7348 This command does nothing whatever; it's only accepted for compatibility.
7350 @cindex @code{FORMAT} (MRI)
7351 @item FORMAT @var{output-format}
7352 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
7353 language, but restricted to one of these output formats:
7357 S-records, if @var{output-format} is @samp{S}
7360 IEEE, if @var{output-format} is @samp{IEEE}
7363 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
7367 @cindex @code{LIST} (MRI)
7368 @item LIST @var{anything}@dots{}
7369 Print (to the standard output file) a link map, as produced by the
7370 @command{ld} command-line option @samp{-M}.
7372 The keyword @code{LIST} may be followed by anything on the
7373 same line, with no change in its effect.
7375 @cindex @code{LOAD} (MRI)
7376 @item LOAD @var{filename}
7377 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
7378 Include one or more object file @var{filename} in the link; this has the
7379 same effect as specifying @var{filename} directly on the @command{ld}
7382 @cindex @code{NAME} (MRI)
7383 @item NAME @var{output-name}
7384 @var{output-name} is the name for the program produced by @command{ld}; the
7385 MRI-compatible command @code{NAME} is equivalent to the command-line
7386 option @samp{-o} or the general script language command @code{OUTPUT}.
7388 @cindex @code{ORDER} (MRI)
7389 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
7390 @itemx ORDER @var{secname} @var{secname} @var{secname}
7391 Normally, @command{ld} orders the sections in its output file in the
7392 order in which they first appear in the input files. In an MRI-compatible
7393 script, you can override this ordering with the @code{ORDER} command. The
7394 sections you list with @code{ORDER} will appear first in your output
7395 file, in the order specified.
7397 @cindex @code{PUBLIC} (MRI)
7398 @item PUBLIC @var{name}=@var{expression}
7399 @itemx PUBLIC @var{name},@var{expression}
7400 @itemx PUBLIC @var{name} @var{expression}
7401 Supply a value (@var{expression}) for external symbol
7402 @var{name} used in the linker input files.
7404 @cindex @code{SECT} (MRI)
7405 @item SECT @var{secname}, @var{expression}
7406 @itemx SECT @var{secname}=@var{expression}
7407 @itemx SECT @var{secname} @var{expression}
7408 You can use any of these three forms of the @code{SECT} command to
7409 specify the start address (@var{expression}) for section @var{secname}.
7410 If you have more than one @code{SECT} statement for the same
7411 @var{secname}, only the @emph{first} sets the start address.
7414 @node GNU Free Documentation License
7415 @appendix GNU Free Documentation License
7419 @unnumbered LD Index
7424 % I think something like @colophon should be in texinfo. In the
7426 \long\def\colophon{\hbox to0pt{}\vfill
7427 \centerline{The body of this manual is set in}
7428 \centerline{\fontname\tenrm,}
7429 \centerline{with headings in {\bf\fontname\tenbf}}
7430 \centerline{and examples in {\tt\fontname\tentt}.}
7431 \centerline{{\it\fontname\tenit\/} and}
7432 \centerline{{\sl\fontname\tensl\/}}
7433 \centerline{are used for emphasis.}\vfill}
7435 % Blame: doc@cygnus.com, 28mar91.