3 @c Copyright 1991-2013 Free Software Foundation, Inc.
6 @include configdoc.texi
7 @c (configdoc.texi is generated by the Makefile)
13 @macro gcctabopt{body}
19 @c Configure for the generation of man pages
43 @dircategory Software development
45 * Ld: (ld). The GNU linker.
50 This file documents the @sc{gnu} linker LD
51 @ifset VERSION_PACKAGE
52 @value{VERSION_PACKAGE}
54 version @value{VERSION}.
56 Copyright @copyright{} 1991-2013 Free Software Foundation, Inc.
58 Permission is granted to copy, distribute and/or modify this document
59 under the terms of the GNU Free Documentation License, Version 1.3
60 or any later version published by the Free Software Foundation;
61 with no Invariant Sections, with no Front-Cover Texts, and with no
62 Back-Cover Texts. A copy of the license is included in the
63 section entitled ``GNU Free Documentation License''.
67 @setchapternewpage odd
68 @settitle The GNU linker
73 @ifset VERSION_PACKAGE
74 @subtitle @value{VERSION_PACKAGE}
76 @subtitle Version @value{VERSION}
77 @author Steve Chamberlain
78 @author Ian Lance Taylor
83 \hfill Red Hat Inc\par
84 \hfill nickc\@credhat.com, doc\@redhat.com\par
85 \hfill {\it The GNU linker}\par
86 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
88 \global\parindent=0pt % Steve likes it this way.
91 @vskip 0pt plus 1filll
92 @c man begin COPYRIGHT
93 Copyright @copyright{} 1991-2013 Free Software Foundation, Inc.
95 Permission is granted to copy, distribute and/or modify this document
96 under the terms of the GNU Free Documentation License, Version 1.3
97 or any later version published by the Free Software Foundation;
98 with no Invariant Sections, with no Front-Cover Texts, and with no
99 Back-Cover Texts. A copy of the license is included in the
100 section entitled ``GNU Free Documentation License''.
106 @c FIXME: Talk about importance of *order* of args, cmds to linker!
111 This file documents the @sc{gnu} linker ld
112 @ifset VERSION_PACKAGE
113 @value{VERSION_PACKAGE}
115 version @value{VERSION}.
117 This document is distributed under the terms of the GNU Free
118 Documentation License version 1.3. A copy of the license is included
119 in the section entitled ``GNU Free Documentation License''.
122 * Overview:: Overview
123 * Invocation:: Invocation
124 * Scripts:: Linker Scripts
126 * Machine Dependent:: Machine Dependent Features
130 * H8/300:: ld and the H8/300
133 * Renesas:: ld and other Renesas micros
136 * i960:: ld and the Intel 960 family
139 * ARM:: ld and the ARM family
142 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
145 * HPPA ELF32:: ld and HPPA 32-bit ELF
148 * M68K:: ld and Motorola 68K family
151 * MIPS:: ld and MIPS 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.
367 @kindex --audit @var{AUDITLIB}
368 @item --audit @var{AUDITLIB}
369 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
370 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
371 specified in the library. If specified multiple times @code{DT_AUDIT}
372 will contain a colon separated list of audit interfaces to use. If the linker
373 finds an object with an audit entry while searching for shared libraries,
374 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
375 This option is only meaningful on ELF platforms supporting the rtld-audit
379 @cindex architectures
380 @kindex -A @var{arch}
381 @item -A @var{architecture}
382 @kindex --architecture=@var{arch}
383 @itemx --architecture=@var{architecture}
384 In the current release of @command{ld}, this option is useful only for the
385 Intel 960 family of architectures. In that @command{ld} configuration, the
386 @var{architecture} argument identifies the particular architecture in
387 the 960 family, enabling some safeguards and modifying the
388 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
389 family}, for details.
391 Future releases of @command{ld} may support similar functionality for
392 other architecture families.
395 @ifclear SingleFormat
396 @cindex binary input format
397 @kindex -b @var{format}
398 @kindex --format=@var{format}
401 @item -b @var{input-format}
402 @itemx --format=@var{input-format}
403 @command{ld} may be configured to support more than one kind of object
404 file. If your @command{ld} is configured this way, you can use the
405 @samp{-b} option to specify the binary format for input object files
406 that follow this option on the command line. Even when @command{ld} is
407 configured to support alternative object formats, you don't usually need
408 to specify this, as @command{ld} should be configured to expect as a
409 default input format the most usual format on each machine.
410 @var{input-format} is a text string, the name of a particular format
411 supported by the BFD libraries. (You can list the available binary
412 formats with @samp{objdump -i}.)
415 You may want to use this option if you are linking files with an unusual
416 binary format. You can also use @samp{-b} to switch formats explicitly (when
417 linking object files of different formats), by including
418 @samp{-b @var{input-format}} before each group of object files in a
421 The default format is taken from the environment variable
426 You can also define the input format from a script, using the command
429 see @ref{Format Commands}.
433 @kindex -c @var{MRI-cmdfile}
434 @kindex --mri-script=@var{MRI-cmdfile}
435 @cindex compatibility, MRI
436 @item -c @var{MRI-commandfile}
437 @itemx --mri-script=@var{MRI-commandfile}
438 For compatibility with linkers produced by MRI, @command{ld} accepts script
439 files written in an alternate, restricted command language, described in
441 @ref{MRI,,MRI Compatible Script Files}.
444 the MRI Compatible Script Files section of GNU ld documentation.
446 Introduce MRI script files with
447 the option @samp{-c}; use the @samp{-T} option to run linker
448 scripts written in the general-purpose @command{ld} scripting language.
449 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
450 specified by any @samp{-L} options.
452 @cindex common allocation
459 These three options are equivalent; multiple forms are supported for
460 compatibility with other linkers. They assign space to common symbols
461 even if a relocatable output file is specified (with @samp{-r}). The
462 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
463 @xref{Miscellaneous Commands}.
465 @kindex --depaudit @var{AUDITLIB}
466 @kindex -P @var{AUDITLIB}
467 @item --depaudit @var{AUDITLIB}
468 @itemx -P @var{AUDITLIB}
469 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
470 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
471 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
472 will contain a colon separated list of audit interfaces to use. This
473 option is only meaningful on ELF platforms supporting the rtld-audit interface.
474 The -P option is provided for Solaris compatibility.
476 @cindex entry point, from command line
477 @kindex -e @var{entry}
478 @kindex --entry=@var{entry}
480 @itemx --entry=@var{entry}
481 Use @var{entry} as the explicit symbol for beginning execution of your
482 program, rather than the default entry point. If there is no symbol
483 named @var{entry}, the linker will try to parse @var{entry} as a number,
484 and use that as the entry address (the number will be interpreted in
485 base 10; you may use a leading @samp{0x} for base 16, or a leading
486 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
487 and other ways of specifying the entry point.
489 @kindex --exclude-libs
490 @item --exclude-libs @var{lib},@var{lib},...
491 Specifies a list of archive libraries from which symbols should not be automatically
492 exported. The library names may be delimited by commas or colons. Specifying
493 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
494 automatic export. This option is available only for the i386 PE targeted
495 port of the linker and for ELF targeted ports. For i386 PE, symbols
496 explicitly listed in a .def file are still exported, regardless of this
497 option. For ELF targeted ports, symbols affected by this option will
498 be treated as hidden.
500 @kindex --exclude-modules-for-implib
501 @item --exclude-modules-for-implib @var{module},@var{module},...
502 Specifies a list of object files or archive members, from which symbols
503 should not be automatically exported, but which should be copied wholesale
504 into the import library being generated during the link. The module names
505 may be delimited by commas or colons, and must match exactly the filenames
506 used by @command{ld} to open the files; for archive members, this is simply
507 the member name, but for object files the name listed must include and
508 match precisely any path used to specify the input file on the linker's
509 command-line. This option is available only for the i386 PE targeted port
510 of the linker. Symbols explicitly listed in a .def file are still exported,
511 regardless of this option.
513 @cindex dynamic symbol table
515 @kindex --export-dynamic
516 @kindex --no-export-dynamic
518 @itemx --export-dynamic
519 @itemx --no-export-dynamic
520 When creating a dynamically linked executable, using the @option{-E}
521 option or the @option{--export-dynamic} option causes the linker to add
522 all symbols to the dynamic symbol table. The dynamic symbol table is the
523 set of symbols which are visible from dynamic objects at run time.
525 If you do not use either of these options (or use the
526 @option{--no-export-dynamic} option to restore the default behavior), the
527 dynamic symbol table will normally contain only those symbols which are
528 referenced by some dynamic object mentioned in the link.
530 If you use @code{dlopen} to load a dynamic object which needs to refer
531 back to the symbols defined by the program, rather than some other
532 dynamic object, then you will probably need to use this option when
533 linking the program itself.
535 You can also use the dynamic list to control what symbols should
536 be added to the dynamic symbol table if the output format supports it.
537 See the description of @samp{--dynamic-list}.
539 Note that this option is specific to ELF targeted ports. PE targets
540 support a similar function to export all symbols from a DLL or EXE; see
541 the description of @samp{--export-all-symbols} below.
543 @ifclear SingleFormat
544 @cindex big-endian objects
548 Link big-endian objects. This affects the default output format.
550 @cindex little-endian objects
553 Link little-endian objects. This affects the default output format.
556 @kindex -f @var{name}
557 @kindex --auxiliary=@var{name}
559 @itemx --auxiliary=@var{name}
560 When creating an ELF shared object, set the internal DT_AUXILIARY field
561 to the specified name. This tells the dynamic linker that the symbol
562 table of the shared object should be used as an auxiliary filter on the
563 symbol table of the shared object @var{name}.
565 If you later link a program against this filter object, then, when you
566 run the program, the dynamic linker will see the DT_AUXILIARY field. If
567 the dynamic linker resolves any symbols from the filter object, it will
568 first check whether there is a definition in the shared object
569 @var{name}. If there is one, it will be used instead of the definition
570 in the filter object. The shared object @var{name} need not exist.
571 Thus the shared object @var{name} may be used to provide an alternative
572 implementation of certain functions, perhaps for debugging or for
573 machine specific performance.
575 This option may be specified more than once. The DT_AUXILIARY entries
576 will be created in the order in which they appear on the command line.
578 @kindex -F @var{name}
579 @kindex --filter=@var{name}
581 @itemx --filter=@var{name}
582 When creating an ELF shared object, set the internal DT_FILTER field to
583 the specified name. This tells the dynamic linker that the symbol table
584 of the shared object which is being created should be used as a filter
585 on the symbol table of the shared object @var{name}.
587 If you later link a program against this filter object, then, when you
588 run the program, the dynamic linker will see the DT_FILTER field. The
589 dynamic linker will resolve symbols according to the symbol table of the
590 filter object as usual, but it will actually link to the definitions
591 found in the shared object @var{name}. Thus the filter object can be
592 used to select a subset of the symbols provided by the object
595 Some older linkers used the @option{-F} option throughout a compilation
596 toolchain for specifying object-file format for both input and output
598 @ifclear SingleFormat
599 The @sc{gnu} linker uses other mechanisms for this purpose: the
600 @option{-b}, @option{--format}, @option{--oformat} options, the
601 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
602 environment variable.
604 The @sc{gnu} linker will ignore the @option{-F} option when not
605 creating an ELF shared object.
607 @cindex finalization function
608 @kindex -fini=@var{name}
609 @item -fini=@var{name}
610 When creating an ELF executable or shared object, call NAME when the
611 executable or shared object is unloaded, by setting DT_FINI to the
612 address of the function. By default, the linker uses @code{_fini} as
613 the function to call.
617 Ignored. Provided for compatibility with other tools.
619 @kindex -G @var{value}
620 @kindex --gpsize=@var{value}
623 @itemx --gpsize=@var{value}
624 Set the maximum size of objects to be optimized using the GP register to
625 @var{size}. This is only meaningful for object file formats such as
626 MIPS ELF that support putting large and small objects into different
627 sections. This is ignored for other object file formats.
629 @cindex runtime library name
630 @kindex -h @var{name}
631 @kindex -soname=@var{name}
633 @itemx -soname=@var{name}
634 When creating an ELF shared object, set the internal DT_SONAME field to
635 the specified name. When an executable is linked with a shared object
636 which has a DT_SONAME field, then when the executable is run the dynamic
637 linker will attempt to load the shared object specified by the DT_SONAME
638 field rather than the using the file name given to the linker.
641 @cindex incremental link
643 Perform an incremental link (same as option @samp{-r}).
645 @cindex initialization function
646 @kindex -init=@var{name}
647 @item -init=@var{name}
648 When creating an ELF executable or shared object, call NAME when the
649 executable or shared object is loaded, by setting DT_INIT to the address
650 of the function. By default, the linker uses @code{_init} as the
653 @cindex archive files, from cmd line
654 @kindex -l @var{namespec}
655 @kindex --library=@var{namespec}
656 @item -l @var{namespec}
657 @itemx --library=@var{namespec}
658 Add the archive or object file specified by @var{namespec} to the
659 list of files to link. This option may be used any number of times.
660 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
661 will search the library path for a file called @var{filename}, otherwise it
662 will search the library path for a file called @file{lib@var{namespec}.a}.
664 On systems which support shared libraries, @command{ld} may also search for
665 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
666 and SunOS systems, @command{ld} will search a directory for a library
667 called @file{lib@var{namespec}.so} before searching for one called
668 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
669 indicates a shared library.) Note that this behavior does not apply
670 to @file{:@var{filename}}, which always specifies a file called
673 The linker will search an archive only once, at the location where it is
674 specified on the command line. If the archive defines a symbol which
675 was undefined in some object which appeared before the archive on the
676 command line, the linker will include the appropriate file(s) from the
677 archive. However, an undefined symbol in an object appearing later on
678 the command line will not cause the linker to search the archive again.
680 See the @option{-(} option for a way to force the linker to search
681 archives multiple times.
683 You may list the same archive multiple times on the command line.
686 This type of archive searching is standard for Unix linkers. However,
687 if you are using @command{ld} on AIX, note that it is different from the
688 behaviour of the AIX linker.
691 @cindex search directory, from cmd line
693 @kindex --library-path=@var{dir}
694 @item -L @var{searchdir}
695 @itemx --library-path=@var{searchdir}
696 Add path @var{searchdir} to the list of paths that @command{ld} will search
697 for archive libraries and @command{ld} control scripts. You may use this
698 option any number of times. The directories are searched in the order
699 in which they are specified on the command line. Directories specified
700 on the command line are searched before the default directories. All
701 @option{-L} options apply to all @option{-l} options, regardless of the
702 order in which the options appear. @option{-L} options do not affect
703 how @command{ld} searches for a linker script unless @option{-T}
706 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
707 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
710 The default set of paths searched (without being specified with
711 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
712 some cases also on how it was configured. @xref{Environment}.
715 The paths can also be specified in a link script with the
716 @code{SEARCH_DIR} command. Directories specified this way are searched
717 at the point in which the linker script appears in the command line.
720 @kindex -m @var{emulation}
721 @item -m @var{emulation}
722 Emulate the @var{emulation} linker. You can list the available
723 emulations with the @samp{--verbose} or @samp{-V} options.
725 If the @samp{-m} option is not used, the emulation is taken from the
726 @code{LDEMULATION} environment variable, if that is defined.
728 Otherwise, the default emulation depends upon how the linker was
736 Print a link map to the standard output. A link map provides
737 information about the link, including the following:
741 Where object files are mapped into memory.
743 How common symbols are allocated.
745 All archive members included in the link, with a mention of the symbol
746 which caused the archive member to be brought in.
748 The values assigned to symbols.
750 Note - symbols whose values are computed by an expression which
751 involves a reference to a previous value of the same symbol may not
752 have correct result displayed in the link map. This is because the
753 linker discards intermediate results and only retains the final value
754 of an expression. Under such circumstances the linker will display
755 the final value enclosed by square brackets. Thus for example a
756 linker script containing:
764 will produce the following output in the link map if the @option{-M}
769 [0x0000000c] foo = (foo * 0x4)
770 [0x0000000c] foo = (foo + 0x8)
773 See @ref{Expressions} for more information about expressions in linker
778 @cindex read-only text
783 Turn off page alignment of sections, and disable linking against shared
784 libraries. If the output format supports Unix style magic numbers,
785 mark the output as @code{NMAGIC}.
789 @cindex read/write from cmd line
793 Set the text and data sections to be readable and writable. Also, do
794 not page-align the data segment, and disable linking against shared
795 libraries. If the output format supports Unix style magic numbers,
796 mark the output as @code{OMAGIC}. Note: Although a writable text section
797 is allowed for PE-COFF targets, it does not conform to the format
798 specification published by Microsoft.
803 This option negates most of the effects of the @option{-N} option. It
804 sets the text section to be read-only, and forces the data segment to
805 be page-aligned. Note - this option does not enable linking against
806 shared libraries. Use @option{-Bdynamic} for this.
808 @kindex -o @var{output}
809 @kindex --output=@var{output}
810 @cindex naming the output file
811 @item -o @var{output}
812 @itemx --output=@var{output}
813 Use @var{output} as the name for the program produced by @command{ld}; if this
814 option is not specified, the name @file{a.out} is used by default. The
815 script command @code{OUTPUT} can also specify the output file name.
817 @kindex -O @var{level}
818 @cindex generating optimized output
820 If @var{level} is a numeric values greater than zero @command{ld} optimizes
821 the output. This might take significantly longer and therefore probably
822 should only be enabled for the final binary. At the moment this
823 option only affects ELF shared library generation. Future releases of
824 the linker may make more use of this option. Also currently there is
825 no difference in the linker's behaviour for different non-zero values
826 of this option. Again this may change with future releases.
829 @kindex --emit-relocs
830 @cindex retain relocations in final executable
833 Leave relocation sections and contents in fully linked executables.
834 Post link analysis and optimization tools may need this information in
835 order to perform correct modifications of executables. This results
836 in larger executables.
838 This option is currently only supported on ELF platforms.
840 @kindex --force-dynamic
841 @cindex forcing the creation of dynamic sections
842 @item --force-dynamic
843 Force the output file to have dynamic sections. This option is specific
847 @cindex relocatable output
849 @kindex --relocatable
852 Generate relocatable output---i.e., generate an output file that can in
853 turn serve as input to @command{ld}. This is often called @dfn{partial
854 linking}. As a side effect, in environments that support standard Unix
855 magic numbers, this option also sets the output file's magic number to
857 @c ; see @option{-N}.
858 If this option is not specified, an absolute file is produced. When
859 linking C++ programs, this option @emph{will not} resolve references to
860 constructors; to do that, use @samp{-Ur}.
862 When an input file does not have the same format as the output file,
863 partial linking is only supported if that input file does not contain any
864 relocations. Different output formats can have further restrictions; for
865 example some @code{a.out}-based formats do not support partial linking
866 with input files in other formats at all.
868 This option does the same thing as @samp{-i}.
870 @kindex -R @var{file}
871 @kindex --just-symbols=@var{file}
872 @cindex symbol-only input
873 @item -R @var{filename}
874 @itemx --just-symbols=@var{filename}
875 Read symbol names and their addresses from @var{filename}, but do not
876 relocate it or include it in the output. This allows your output file
877 to refer symbolically to absolute locations of memory defined in other
878 programs. You may use this option more than once.
880 For compatibility with other ELF linkers, if the @option{-R} option is
881 followed by a directory name, rather than a file name, it is treated as
882 the @option{-rpath} option.
886 @cindex strip all symbols
889 Omit all symbol information from the output file.
892 @kindex --strip-debug
893 @cindex strip debugger symbols
896 Omit debugger symbol information (but not all symbols) from the output file.
900 @cindex input files, displaying
903 Print the names of the input files as @command{ld} processes them.
905 @kindex -T @var{script}
906 @kindex --script=@var{script}
908 @item -T @var{scriptfile}
909 @itemx --script=@var{scriptfile}
910 Use @var{scriptfile} as the linker script. This script replaces
911 @command{ld}'s default linker script (rather than adding to it), so
912 @var{commandfile} must specify everything necessary to describe the
913 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
914 the current directory, @code{ld} looks for it in the directories
915 specified by any preceding @samp{-L} options. Multiple @samp{-T}
918 @kindex -dT @var{script}
919 @kindex --default-script=@var{script}
921 @item -dT @var{scriptfile}
922 @itemx --default-script=@var{scriptfile}
923 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
925 This option is similar to the @option{--script} option except that
926 processing of the script is delayed until after the rest of the
927 command line has been processed. This allows options placed after the
928 @option{--default-script} option on the command line to affect the
929 behaviour of the linker script, which can be important when the linker
930 command line cannot be directly controlled by the user. (eg because
931 the command line is being constructed by another tool, such as
934 @kindex -u @var{symbol}
935 @kindex --undefined=@var{symbol}
936 @cindex undefined symbol
937 @item -u @var{symbol}
938 @itemx --undefined=@var{symbol}
939 Force @var{symbol} to be entered in the output file as an undefined
940 symbol. Doing this may, for example, trigger linking of additional
941 modules from standard libraries. @samp{-u} may be repeated with
942 different option arguments to enter additional undefined symbols. This
943 option is equivalent to the @code{EXTERN} linker script command.
948 For anything other than C++ programs, this option is equivalent to
949 @samp{-r}: it generates relocatable output---i.e., an output file that can in
950 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
951 @emph{does} resolve references to constructors, unlike @samp{-r}.
952 It does not work to use @samp{-Ur} on files that were themselves linked
953 with @samp{-Ur}; once the constructor table has been built, it cannot
954 be added to. Use @samp{-Ur} only for the last partial link, and
955 @samp{-r} for the others.
957 @kindex --unique[=@var{SECTION}]
958 @item --unique[=@var{SECTION}]
959 Creates a separate output section for every input section matching
960 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
961 missing, for every orphan input section. An orphan section is one not
962 specifically mentioned in a linker script. You may use this option
963 multiple times on the command line; It prevents the normal merging of
964 input sections with the same name, overriding output section assignments
974 Display the version number for @command{ld}. The @option{-V} option also
975 lists the supported emulations.
978 @kindex --discard-all
979 @cindex deleting local symbols
982 Delete all local symbols.
985 @kindex --discard-locals
986 @cindex local symbols, deleting
988 @itemx --discard-locals
989 Delete all temporary local symbols. (These symbols start with
990 system-specific local label prefixes, typically @samp{.L} for ELF systems
991 or @samp{L} for traditional a.out systems.)
993 @kindex -y @var{symbol}
994 @kindex --trace-symbol=@var{symbol}
995 @cindex symbol tracing
996 @item -y @var{symbol}
997 @itemx --trace-symbol=@var{symbol}
998 Print the name of each linked file in which @var{symbol} appears. This
999 option may be given any number of times. On many systems it is necessary
1000 to prepend an underscore.
1002 This option is useful when you have an undefined symbol in your link but
1003 don't know where the reference is coming from.
1005 @kindex -Y @var{path}
1007 Add @var{path} to the default library search path. This option exists
1008 for Solaris compatibility.
1010 @kindex -z @var{keyword}
1011 @item -z @var{keyword}
1012 The recognized keywords are:
1016 Combines multiple reloc sections and sorts them to make dynamic symbol
1017 lookup caching possible.
1020 Disallows undefined symbols in object files. Undefined symbols in
1021 shared libraries are still allowed.
1024 Marks the object as requiring executable stack.
1027 This option is only meaningful when building a shared object. It makes
1028 the symbols defined by this shared object available for symbol resolution
1029 of subsequently loaded libraries.
1032 This option is only meaningful when building a shared object.
1033 It marks the object so that its runtime initialization will occur
1034 before the runtime initialization of any other objects brought into
1035 the process at the same time. Similarly the runtime finalization of
1036 the object will occur after the runtime finalization of any other
1040 Marks the object that its symbol table interposes before all symbols
1041 but the primary executable.
1044 When generating an executable or shared library, mark it to tell the
1045 dynamic linker to defer function call resolution to the point when
1046 the function is called (lazy binding), rather than at load time.
1047 Lazy binding is the default.
1050 Marks the object that its filters be processed immediately at
1054 Allows multiple definitions.
1057 Disables multiple reloc sections combining.
1060 Disables production of copy relocs.
1063 Marks the object that the search for dependencies of this object will
1064 ignore any default library search paths.
1067 Marks the object shouldn't be unloaded at runtime.
1070 Marks the object not available to @code{dlopen}.
1073 Marks the object can not be dumped by @code{dldump}.
1076 Marks the object as not requiring executable stack.
1079 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1082 When generating an executable or shared library, mark it to tell the
1083 dynamic linker to resolve all symbols when the program is started, or
1084 when the shared library is linked to using dlopen, instead of
1085 deferring function call resolution to the point when the function is
1089 Marks the object may contain $ORIGIN.
1092 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1094 @item max-page-size=@var{value}
1095 Set the emulation maximum page size to @var{value}.
1097 @item common-page-size=@var{value}
1098 Set the emulation common page size to @var{value}.
1100 @item stack-size=@var{value}
1101 Specify a stack size for in an ELF @code{PT_GNU_STACK} segment.
1102 Specifying zero will override any default non-zero sized
1103 @code{PT_GNU_STACK} segment creation.
1107 Other keywords are ignored for Solaris compatibility.
1110 @cindex groups of archives
1111 @item -( @var{archives} -)
1112 @itemx --start-group @var{archives} --end-group
1113 The @var{archives} should be a list of archive files. They may be
1114 either explicit file names, or @samp{-l} options.
1116 The specified archives are searched repeatedly until no new undefined
1117 references are created. Normally, an archive is searched only once in
1118 the order that it is specified on the command line. If a symbol in that
1119 archive is needed to resolve an undefined symbol referred to by an
1120 object in an archive that appears later on the command line, the linker
1121 would not be able to resolve that reference. By grouping the archives,
1122 they all be searched repeatedly until all possible references are
1125 Using this option has a significant performance cost. It is best to use
1126 it only when there are unavoidable circular references between two or
1129 @kindex --accept-unknown-input-arch
1130 @kindex --no-accept-unknown-input-arch
1131 @item --accept-unknown-input-arch
1132 @itemx --no-accept-unknown-input-arch
1133 Tells the linker to accept input files whose architecture cannot be
1134 recognised. The assumption is that the user knows what they are doing
1135 and deliberately wants to link in these unknown input files. This was
1136 the default behaviour of the linker, before release 2.14. The default
1137 behaviour from release 2.14 onwards is to reject such input files, and
1138 so the @samp{--accept-unknown-input-arch} option has been added to
1139 restore the old behaviour.
1142 @kindex --no-as-needed
1144 @itemx --no-as-needed
1145 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1146 on the command line after the @option{--as-needed} option. Normally
1147 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1148 on the command line, regardless of whether the library is actually
1149 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1150 emitted for a library that @emph{at that point in the link} satisfies a
1151 non-weak undefined symbol reference from a regular object file or, if
1152 the library is not found in the DT_NEEDED lists of other libraries, a
1153 non-weak undefined symbol reference from another dynamic library.
1154 Object files or libraries appearing on the command line @emph{after}
1155 the library in question do not affect whether the library is seen as
1156 needed. This is similar to the rules for extraction of object files
1157 from archives. @option{--no-as-needed} restores the default behaviour.
1159 @kindex --add-needed
1160 @kindex --no-add-needed
1162 @itemx --no-add-needed
1163 These two options have been deprecated because of the similarity of
1164 their names to the @option{--as-needed} and @option{--no-as-needed}
1165 options. They have been replaced by @option{--copy-dt-needed-entries}
1166 and @option{--no-copy-dt-needed-entries}.
1168 @kindex -assert @var{keyword}
1169 @item -assert @var{keyword}
1170 This option is ignored for SunOS compatibility.
1174 @kindex -call_shared
1178 Link against dynamic libraries. This is only meaningful on platforms
1179 for which shared libraries are supported. This option is normally the
1180 default on such platforms. The different variants of this option are
1181 for compatibility with various systems. You may use this option
1182 multiple times on the command line: it affects library searching for
1183 @option{-l} options which follow it.
1187 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1188 section. This causes the runtime linker to handle lookups in this
1189 object and its dependencies to be performed only inside the group.
1190 @option{--unresolved-symbols=report-all} is implied. This option is
1191 only meaningful on ELF platforms which support shared libraries.
1201 Do not link against shared libraries. This is only meaningful on
1202 platforms for which shared libraries are supported. The different
1203 variants of this option are for compatibility with various systems. You
1204 may use this option multiple times on the command line: it affects
1205 library searching for @option{-l} options which follow it. This
1206 option also implies @option{--unresolved-symbols=report-all}. This
1207 option can be used with @option{-shared}. Doing so means that a
1208 shared library is being created but that all of the library's external
1209 references must be resolved by pulling in entries from static
1214 When creating a shared library, bind references to global symbols to the
1215 definition within the shared library, if any. Normally, it is possible
1216 for a program linked against a shared library to override the definition
1217 within the shared library. This option is only meaningful on ELF
1218 platforms which support shared libraries.
1220 @kindex -Bsymbolic-functions
1221 @item -Bsymbolic-functions
1222 When creating a shared library, bind references to global function
1223 symbols to the definition within the shared library, if any.
1224 This option is only meaningful on ELF platforms which support shared
1227 @kindex --dynamic-list=@var{dynamic-list-file}
1228 @item --dynamic-list=@var{dynamic-list-file}
1229 Specify the name of a dynamic list file to the linker. This is
1230 typically used when creating shared libraries to specify a list of
1231 global symbols whose references shouldn't be bound to the definition
1232 within the shared library, or creating dynamically linked executables
1233 to specify a list of symbols which should be added to the symbol table
1234 in the executable. This option is only meaningful on ELF platforms
1235 which support shared libraries.
1237 The format of the dynamic list is the same as the version node without
1238 scope and node name. See @ref{VERSION} for more information.
1240 @kindex --dynamic-list-data
1241 @item --dynamic-list-data
1242 Include all global data symbols to the dynamic list.
1244 @kindex --dynamic-list-cpp-new
1245 @item --dynamic-list-cpp-new
1246 Provide the builtin dynamic list for C++ operator new and delete. It
1247 is mainly useful for building shared libstdc++.
1249 @kindex --dynamic-list-cpp-typeinfo
1250 @item --dynamic-list-cpp-typeinfo
1251 Provide the builtin dynamic list for C++ runtime type identification.
1253 @kindex --check-sections
1254 @kindex --no-check-sections
1255 @item --check-sections
1256 @itemx --no-check-sections
1257 Asks the linker @emph{not} to check section addresses after they have
1258 been assigned to see if there are any overlaps. Normally the linker will
1259 perform this check, and if it finds any overlaps it will produce
1260 suitable error messages. The linker does know about, and does make
1261 allowances for sections in overlays. The default behaviour can be
1262 restored by using the command line switch @option{--check-sections}.
1263 Section overlap is not usually checked for relocatable links. You can
1264 force checking in that case by using the @option{--check-sections}
1267 @kindex --copy-dt-needed-entries
1268 @kindex --no-copy-dt-needed-entries
1269 @item --copy-dt-needed-entries
1270 @itemx --no-copy-dt-needed-entries
1271 This option affects the treatment of dynamic libraries referred to
1272 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1273 command line. Normally the linker won't add a DT_NEEDED tag to the
1274 output binary for each library mentioned in a DT_NEEDED tag in an
1275 input dynamic library. With @option{--copy-dt-needed-entries}
1276 specified on the command line however any dynamic libraries that
1277 follow it will have their DT_NEEDED entries added. The default
1278 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1280 This option also has an effect on the resolution of symbols in dynamic
1281 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1282 mentioned on the command line will be recursively searched, following
1283 their DT_NEEDED tags to other libraries, in order to resolve symbols
1284 required by the output binary. With the default setting however
1285 the searching of dynamic libraries that follow it will stop with the
1286 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1289 @cindex cross reference table
1292 Output a cross reference table. If a linker map file is being
1293 generated, the cross reference table is printed to the map file.
1294 Otherwise, it is printed on the standard output.
1296 The format of the table is intentionally simple, so that it may be
1297 easily processed by a script if necessary. The symbols are printed out,
1298 sorted by name. For each symbol, a list of file names is given. If the
1299 symbol is defined, the first file listed is the location of the
1300 definition. If the symbol is defined as a common value then any files
1301 where this happens appear next. Finally any files that reference the
1304 @cindex common allocation
1305 @kindex --no-define-common
1306 @item --no-define-common
1307 This option inhibits the assignment of addresses to common symbols.
1308 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1309 @xref{Miscellaneous Commands}.
1311 The @samp{--no-define-common} option allows decoupling
1312 the decision to assign addresses to Common symbols from the choice
1313 of the output file type; otherwise a non-Relocatable output type
1314 forces assigning addresses to Common symbols.
1315 Using @samp{--no-define-common} allows Common symbols that are referenced
1316 from a shared library to be assigned addresses only in the main program.
1317 This eliminates the unused duplicate space in the shared library,
1318 and also prevents any possible confusion over resolving to the wrong
1319 duplicate when there are many dynamic modules with specialized search
1320 paths for runtime symbol resolution.
1322 @cindex symbols, from command line
1323 @kindex --defsym=@var{symbol}=@var{exp}
1324 @item --defsym=@var{symbol}=@var{expression}
1325 Create a global symbol in the output file, containing the absolute
1326 address given by @var{expression}. You may use this option as many
1327 times as necessary to define multiple symbols in the command line. A
1328 limited form of arithmetic is supported for the @var{expression} in this
1329 context: you may give a hexadecimal constant or the name of an existing
1330 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1331 constants or symbols. If you need more elaborate expressions, consider
1332 using the linker command language from a script (@pxref{Assignments,,
1333 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1334 space between @var{symbol}, the equals sign (``@key{=}''), and
1337 @cindex demangling, from command line
1338 @kindex --demangle[=@var{style}]
1339 @kindex --no-demangle
1340 @item --demangle[=@var{style}]
1341 @itemx --no-demangle
1342 These options control whether to demangle symbol names in error messages
1343 and other output. When the linker is told to demangle, it tries to
1344 present symbol names in a readable fashion: it strips leading
1345 underscores if they are used by the object file format, and converts C++
1346 mangled symbol names into user readable names. Different compilers have
1347 different mangling styles. The optional demangling style argument can be used
1348 to choose an appropriate demangling style for your compiler. The linker will
1349 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1350 is set. These options may be used to override the default.
1352 @cindex dynamic linker, from command line
1353 @kindex -I@var{file}
1354 @kindex --dynamic-linker=@var{file}
1356 @itemx --dynamic-linker=@var{file}
1357 Set the name of the dynamic linker. This is only meaningful when
1358 generating dynamically linked ELF executables. The default dynamic
1359 linker is normally correct; don't use this unless you know what you are
1362 @kindex --fatal-warnings
1363 @kindex --no-fatal-warnings
1364 @item --fatal-warnings
1365 @itemx --no-fatal-warnings
1366 Treat all warnings as errors. The default behaviour can be restored
1367 with the option @option{--no-fatal-warnings}.
1369 @kindex --force-exe-suffix
1370 @item --force-exe-suffix
1371 Make sure that an output file has a .exe suffix.
1373 If a successfully built fully linked output file does not have a
1374 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1375 the output file to one of the same name with a @code{.exe} suffix. This
1376 option is useful when using unmodified Unix makefiles on a Microsoft
1377 Windows host, since some versions of Windows won't run an image unless
1378 it ends in a @code{.exe} suffix.
1380 @kindex --gc-sections
1381 @kindex --no-gc-sections
1382 @cindex garbage collection
1384 @itemx --no-gc-sections
1385 Enable garbage collection of unused input sections. It is ignored on
1386 targets that do not support this option. The default behaviour (of not
1387 performing this garbage collection) can be restored by specifying
1388 @samp{--no-gc-sections} on the command line.
1390 @samp{--gc-sections} decides which input sections are used by
1391 examining symbols and relocations. The section containing the entry
1392 symbol and all sections containing symbols undefined on the
1393 command-line will be kept, as will sections containing symbols
1394 referenced by dynamic objects. Note that when building shared
1395 libraries, the linker must assume that any visible symbol is
1396 referenced. Once this initial set of sections has been determined,
1397 the linker recursively marks as used any section referenced by their
1398 relocations. See @samp{--entry} and @samp{--undefined}.
1400 This option can be set when doing a partial link (enabled with option
1401 @samp{-r}). In this case the root of symbols kept must be explicitly
1402 specified either by an @samp{--entry} or @samp{--undefined} option or by
1403 a @code{ENTRY} command in the linker script.
1405 @kindex --print-gc-sections
1406 @kindex --no-print-gc-sections
1407 @cindex garbage collection
1408 @item --print-gc-sections
1409 @itemx --no-print-gc-sections
1410 List all sections removed by garbage collection. The listing is
1411 printed on stderr. This option is only effective if garbage
1412 collection has been enabled via the @samp{--gc-sections}) option. The
1413 default behaviour (of not listing the sections that are removed) can
1414 be restored by specifying @samp{--no-print-gc-sections} on the command
1417 @kindex --print-output-format
1418 @cindex output format
1419 @item --print-output-format
1420 Print the name of the default output format (perhaps influenced by
1421 other command-line options). This is the string that would appear
1422 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1428 Print a summary of the command-line options on the standard output and exit.
1430 @kindex --target-help
1432 Print a summary of all target specific options on the standard output and exit.
1434 @kindex -Map=@var{mapfile}
1435 @item -Map=@var{mapfile}
1436 Print a link map to the file @var{mapfile}. See the description of the
1437 @option{-M} option, above.
1439 @cindex memory usage
1440 @kindex --no-keep-memory
1441 @item --no-keep-memory
1442 @command{ld} normally optimizes for speed over memory usage by caching the
1443 symbol tables of input files in memory. This option tells @command{ld} to
1444 instead optimize for memory usage, by rereading the symbol tables as
1445 necessary. This may be required if @command{ld} runs out of memory space
1446 while linking a large executable.
1448 @kindex --no-undefined
1450 @item --no-undefined
1452 Report unresolved symbol references from regular object files. This
1453 is done even if the linker is creating a non-symbolic shared library.
1454 The switch @option{--[no-]allow-shlib-undefined} controls the
1455 behaviour for reporting unresolved references found in shared
1456 libraries being linked in.
1458 @kindex --allow-multiple-definition
1460 @item --allow-multiple-definition
1462 Normally when a symbol is defined multiple times, the linker will
1463 report a fatal error. These options allow multiple definitions and the
1464 first definition will be used.
1466 @kindex --allow-shlib-undefined
1467 @kindex --no-allow-shlib-undefined
1468 @item --allow-shlib-undefined
1469 @itemx --no-allow-shlib-undefined
1470 Allows or disallows undefined symbols in shared libraries.
1471 This switch is similar to @option{--no-undefined} except that it
1472 determines the behaviour when the undefined symbols are in a
1473 shared library rather than a regular object file. It does not affect
1474 how undefined symbols in regular object files are handled.
1476 The default behaviour is to report errors for any undefined symbols
1477 referenced in shared libraries if the linker is being used to create
1478 an executable, but to allow them if the linker is being used to create
1481 The reasons for allowing undefined symbol references in shared
1482 libraries specified at link time are that:
1486 A shared library specified at link time may not be the same as the one
1487 that is available at load time, so the symbol might actually be
1488 resolvable at load time.
1490 There are some operating systems, eg BeOS and HPPA, where undefined
1491 symbols in shared libraries are normal.
1493 The BeOS kernel for example patches shared libraries at load time to
1494 select whichever function is most appropriate for the current
1495 architecture. This is used, for example, to dynamically select an
1496 appropriate memset function.
1499 @kindex --no-undefined-version
1500 @item --no-undefined-version
1501 Normally when a symbol has an undefined version, the linker will ignore
1502 it. This option disallows symbols with undefined version and a fatal error
1503 will be issued instead.
1505 @kindex --default-symver
1506 @item --default-symver
1507 Create and use a default symbol version (the soname) for unversioned
1510 @kindex --default-imported-symver
1511 @item --default-imported-symver
1512 Create and use a default symbol version (the soname) for unversioned
1515 @kindex --no-warn-mismatch
1516 @item --no-warn-mismatch
1517 Normally @command{ld} will give an error if you try to link together input
1518 files that are mismatched for some reason, perhaps because they have
1519 been compiled for different processors or for different endiannesses.
1520 This option tells @command{ld} that it should silently permit such possible
1521 errors. This option should only be used with care, in cases when you
1522 have taken some special action that ensures that the linker errors are
1525 @kindex --no-warn-search-mismatch
1526 @item --no-warn-search-mismatch
1527 Normally @command{ld} will give a warning if it finds an incompatible
1528 library during a library search. This option silences the warning.
1530 @kindex --no-whole-archive
1531 @item --no-whole-archive
1532 Turn off the effect of the @option{--whole-archive} option for subsequent
1535 @cindex output file after errors
1536 @kindex --noinhibit-exec
1537 @item --noinhibit-exec
1538 Retain the executable output file whenever it is still usable.
1539 Normally, the linker will not produce an output file if it encounters
1540 errors during the link process; it exits without writing an output file
1541 when it issues any error whatsoever.
1545 Only search library directories explicitly specified on the
1546 command line. Library directories specified in linker scripts
1547 (including linker scripts specified on the command line) are ignored.
1549 @ifclear SingleFormat
1550 @kindex --oformat=@var{output-format}
1551 @item --oformat=@var{output-format}
1552 @command{ld} may be configured to support more than one kind of object
1553 file. If your @command{ld} is configured this way, you can use the
1554 @samp{--oformat} option to specify the binary format for the output
1555 object file. Even when @command{ld} is configured to support alternative
1556 object formats, you don't usually need to specify this, as @command{ld}
1557 should be configured to produce as a default output format the most
1558 usual format on each machine. @var{output-format} is a text string, the
1559 name of a particular format supported by the BFD libraries. (You can
1560 list the available binary formats with @samp{objdump -i}.) The script
1561 command @code{OUTPUT_FORMAT} can also specify the output format, but
1562 this option overrides it. @xref{BFD}.
1566 @kindex --pic-executable
1568 @itemx --pic-executable
1569 @cindex position independent executables
1570 Create a position independent executable. This is currently only supported on
1571 ELF platforms. Position independent executables are similar to shared
1572 libraries in that they are relocated by the dynamic linker to the virtual
1573 address the OS chooses for them (which can vary between invocations). Like
1574 normal dynamically linked executables they can be executed and symbols
1575 defined in the executable cannot be overridden by shared libraries.
1579 This option is ignored for Linux compatibility.
1583 This option is ignored for SVR4 compatibility.
1586 @cindex synthesizing linker
1587 @cindex relaxing addressing modes
1591 An option with machine dependent effects.
1593 This option is only supported on a few targets.
1596 @xref{H8/300,,@command{ld} and the H8/300}.
1599 @xref{i960,, @command{ld} and the Intel 960 family}.
1602 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1605 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1608 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1611 On some platforms the @samp{--relax} option performs target specific,
1612 global optimizations that become possible when the linker resolves
1613 addressing in the program, such as relaxing address modes,
1614 synthesizing new instructions, selecting shorter version of current
1615 instructions, and combining constant values.
1617 On some platforms these link time global optimizations may make symbolic
1618 debugging of the resulting executable impossible.
1620 This is known to be the case for the Matsushita MN10200 and MN10300
1621 family of processors.
1625 On platforms where this is not supported, @samp{--relax} is accepted,
1629 On platforms where @samp{--relax} is accepted the option
1630 @samp{--no-relax} can be used to disable the feature.
1632 @cindex retaining specified symbols
1633 @cindex stripping all but some symbols
1634 @cindex symbols, retaining selectively
1635 @kindex --retain-symbols-file=@var{filename}
1636 @item --retain-symbols-file=@var{filename}
1637 Retain @emph{only} the symbols listed in the file @var{filename},
1638 discarding all others. @var{filename} is simply a flat file, with one
1639 symbol name per line. This option is especially useful in environments
1643 where a large global symbol table is accumulated gradually, to conserve
1646 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1647 or symbols needed for relocations.
1649 You may only specify @samp{--retain-symbols-file} once in the command
1650 line. It overrides @samp{-s} and @samp{-S}.
1653 @item -rpath=@var{dir}
1654 @cindex runtime library search path
1655 @kindex -rpath=@var{dir}
1656 Add a directory to the runtime library search path. This is used when
1657 linking an ELF executable with shared objects. All @option{-rpath}
1658 arguments are concatenated and passed to the runtime linker, which uses
1659 them to locate shared objects at runtime. The @option{-rpath} option is
1660 also used when locating shared objects which are needed by shared
1661 objects explicitly included in the link; see the description of the
1662 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1663 ELF executable, the contents of the environment variable
1664 @code{LD_RUN_PATH} will be used if it is defined.
1666 The @option{-rpath} option may also be used on SunOS. By default, on
1667 SunOS, the linker will form a runtime search patch out of all the
1668 @option{-L} options it is given. If a @option{-rpath} option is used, the
1669 runtime search path will be formed exclusively using the @option{-rpath}
1670 options, ignoring the @option{-L} options. This can be useful when using
1671 gcc, which adds many @option{-L} options which may be on NFS mounted
1674 For compatibility with other ELF linkers, if the @option{-R} option is
1675 followed by a directory name, rather than a file name, it is treated as
1676 the @option{-rpath} option.
1680 @cindex link-time runtime library search path
1681 @kindex -rpath-link=@var{dir}
1682 @item -rpath-link=@var{dir}
1683 When using ELF or SunOS, one shared library may require another. This
1684 happens when an @code{ld -shared} link includes a shared library as one
1687 When the linker encounters such a dependency when doing a non-shared,
1688 non-relocatable link, it will automatically try to locate the required
1689 shared library and include it in the link, if it is not included
1690 explicitly. In such a case, the @option{-rpath-link} option
1691 specifies the first set of directories to search. The
1692 @option{-rpath-link} option may specify a sequence of directory names
1693 either by specifying a list of names separated by colons, or by
1694 appearing multiple times.
1696 This option should be used with caution as it overrides the search path
1697 that may have been hard compiled into a shared library. In such a case it
1698 is possible to use unintentionally a different search path than the
1699 runtime linker would do.
1701 The linker uses the following search paths to locate required shared
1705 Any directories specified by @option{-rpath-link} options.
1707 Any directories specified by @option{-rpath} options. The difference
1708 between @option{-rpath} and @option{-rpath-link} is that directories
1709 specified by @option{-rpath} options are included in the executable and
1710 used at runtime, whereas the @option{-rpath-link} option is only effective
1711 at link time. Searching @option{-rpath} in this way is only supported
1712 by native linkers and cross linkers which have been configured with
1713 the @option{--with-sysroot} option.
1715 On an ELF system, for native linkers, if the @option{-rpath} and
1716 @option{-rpath-link} options were not used, search the contents of the
1717 environment variable @code{LD_RUN_PATH}.
1719 On SunOS, if the @option{-rpath} option was not used, search any
1720 directories specified using @option{-L} options.
1722 For a native linker, search the contents of the environment
1723 variable @code{LD_LIBRARY_PATH}.
1725 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1726 @code{DT_RPATH} of a shared library are searched for shared
1727 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1728 @code{DT_RUNPATH} entries exist.
1730 The default directories, normally @file{/lib} and @file{/usr/lib}.
1732 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1733 exists, the list of directories found in that file.
1736 If the required shared library is not found, the linker will issue a
1737 warning and continue with the link.
1744 @cindex shared libraries
1745 Create a shared library. This is currently only supported on ELF, XCOFF
1746 and SunOS platforms. On SunOS, the linker will automatically create a
1747 shared library if the @option{-e} option is not used and there are
1748 undefined symbols in the link.
1750 @kindex --sort-common
1752 @itemx --sort-common=ascending
1753 @itemx --sort-common=descending
1754 This option tells @command{ld} to sort the common symbols by alignment in
1755 ascending or descending order when it places them in the appropriate output
1756 sections. The symbol alignments considered are sixteen-byte or larger,
1757 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1758 between symbols due to alignment constraints. If no sorting order is
1759 specified, then descending order is assumed.
1761 @kindex --sort-section=name
1762 @item --sort-section=name
1763 This option will apply @code{SORT_BY_NAME} to all wildcard section
1764 patterns in the linker script.
1766 @kindex --sort-section=alignment
1767 @item --sort-section=alignment
1768 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1769 patterns in the linker script.
1771 @kindex --split-by-file
1772 @item --split-by-file[=@var{size}]
1773 Similar to @option{--split-by-reloc} but creates a new output section for
1774 each input file when @var{size} is reached. @var{size} defaults to a
1775 size of 1 if not given.
1777 @kindex --split-by-reloc
1778 @item --split-by-reloc[=@var{count}]
1779 Tries to creates extra sections in the output file so that no single
1780 output section in the file contains more than @var{count} relocations.
1781 This is useful when generating huge relocatable files for downloading into
1782 certain real time kernels with the COFF object file format; since COFF
1783 cannot represent more than 65535 relocations in a single section. Note
1784 that this will fail to work with object file formats which do not
1785 support arbitrary sections. The linker will not split up individual
1786 input sections for redistribution, so if a single input section contains
1787 more than @var{count} relocations one output section will contain that
1788 many relocations. @var{count} defaults to a value of 32768.
1792 Compute and display statistics about the operation of the linker, such
1793 as execution time and memory usage.
1795 @kindex --sysroot=@var{directory}
1796 @item --sysroot=@var{directory}
1797 Use @var{directory} as the location of the sysroot, overriding the
1798 configure-time default. This option is only supported by linkers
1799 that were configured using @option{--with-sysroot}.
1801 @kindex --traditional-format
1802 @cindex traditional format
1803 @item --traditional-format
1804 For some targets, the output of @command{ld} is different in some ways from
1805 the output of some existing linker. This switch requests @command{ld} to
1806 use the traditional format instead.
1809 For example, on SunOS, @command{ld} combines duplicate entries in the
1810 symbol string table. This can reduce the size of an output file with
1811 full debugging information by over 30 percent. Unfortunately, the SunOS
1812 @code{dbx} program can not read the resulting program (@code{gdb} has no
1813 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1814 combine duplicate entries.
1816 @kindex --section-start=@var{sectionname}=@var{org}
1817 @item --section-start=@var{sectionname}=@var{org}
1818 Locate a section in the output file at the absolute
1819 address given by @var{org}. You may use this option as many
1820 times as necessary to locate multiple sections in the command
1822 @var{org} must be a single hexadecimal integer;
1823 for compatibility with other linkers, you may omit the leading
1824 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1825 should be no white space between @var{sectionname}, the equals
1826 sign (``@key{=}''), and @var{org}.
1828 @kindex -Tbss=@var{org}
1829 @kindex -Tdata=@var{org}
1830 @kindex -Ttext=@var{org}
1831 @cindex segment origins, cmd line
1832 @item -Tbss=@var{org}
1833 @itemx -Tdata=@var{org}
1834 @itemx -Ttext=@var{org}
1835 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
1836 @code{.text} as the @var{sectionname}.
1838 @kindex -Ttext-segment=@var{org}
1839 @item -Ttext-segment=@var{org}
1840 @cindex text segment origin, cmd line
1841 When creating an ELF executable or shared object, it will set the address
1842 of the first byte of the text segment.
1844 @kindex -Trodata-segment=@var{org}
1845 @item -Trodata-segment=@var{org}
1846 @cindex rodata segment origin, cmd line
1847 When creating an ELF executable or shared object for a target where
1848 the read-only data is in its own segment separate from the executable
1849 text, it will set the address of the first byte of the read-only data segment.
1851 @kindex -Tldata-segment=@var{org}
1852 @item -Tldata-segment=@var{org}
1853 @cindex ldata segment origin, cmd line
1854 When creating an ELF executable or shared object for x86-64 medium memory
1855 model, it will set the address of the first byte of the ldata segment.
1857 @kindex --unresolved-symbols
1858 @item --unresolved-symbols=@var{method}
1859 Determine how to handle unresolved symbols. There are four possible
1860 values for @samp{method}:
1864 Do not report any unresolved symbols.
1867 Report all unresolved symbols. This is the default.
1869 @item ignore-in-object-files
1870 Report unresolved symbols that are contained in shared libraries, but
1871 ignore them if they come from regular object files.
1873 @item ignore-in-shared-libs
1874 Report unresolved symbols that come from regular object files, but
1875 ignore them if they come from shared libraries. This can be useful
1876 when creating a dynamic binary and it is known that all the shared
1877 libraries that it should be referencing are included on the linker's
1881 The behaviour for shared libraries on their own can also be controlled
1882 by the @option{--[no-]allow-shlib-undefined} option.
1884 Normally the linker will generate an error message for each reported
1885 unresolved symbol but the option @option{--warn-unresolved-symbols}
1886 can change this to a warning.
1888 @kindex --verbose[=@var{NUMBER}]
1889 @cindex verbose[=@var{NUMBER}]
1891 @itemx --verbose[=@var{NUMBER}]
1892 Display the version number for @command{ld} and list the linker emulations
1893 supported. Display which input files can and cannot be opened. Display
1894 the linker script being used by the linker. If the optional @var{NUMBER}
1895 argument > 1, plugin symbol status will also be displayed.
1897 @kindex --version-script=@var{version-scriptfile}
1898 @cindex version script, symbol versions
1899 @item --version-script=@var{version-scriptfile}
1900 Specify the name of a version script to the linker. This is typically
1901 used when creating shared libraries to specify additional information
1902 about the version hierarchy for the library being created. This option
1903 is only fully supported on ELF platforms which support shared libraries;
1904 see @ref{VERSION}. It is partially supported on PE platforms, which can
1905 use version scripts to filter symbol visibility in auto-export mode: any
1906 symbols marked @samp{local} in the version script will not be exported.
1909 @kindex --warn-common
1910 @cindex warnings, on combining symbols
1911 @cindex combining symbols, warnings on
1913 Warn when a common symbol is combined with another common symbol or with
1914 a symbol definition. Unix linkers allow this somewhat sloppy practice,
1915 but linkers on some other operating systems do not. This option allows
1916 you to find potential problems from combining global symbols.
1917 Unfortunately, some C libraries use this practice, so you may get some
1918 warnings about symbols in the libraries as well as in your programs.
1920 There are three kinds of global symbols, illustrated here by C examples:
1924 A definition, which goes in the initialized data section of the output
1928 An undefined reference, which does not allocate space.
1929 There must be either a definition or a common symbol for the
1933 A common symbol. If there are only (one or more) common symbols for a
1934 variable, it goes in the uninitialized data area of the output file.
1935 The linker merges multiple common symbols for the same variable into a
1936 single symbol. If they are of different sizes, it picks the largest
1937 size. The linker turns a common symbol into a declaration, if there is
1938 a definition of the same variable.
1941 The @samp{--warn-common} option can produce five kinds of warnings.
1942 Each warning consists of a pair of lines: the first describes the symbol
1943 just encountered, and the second describes the previous symbol
1944 encountered with the same name. One or both of the two symbols will be
1949 Turning a common symbol into a reference, because there is already a
1950 definition for the symbol.
1952 @var{file}(@var{section}): warning: common of `@var{symbol}'
1953 overridden by definition
1954 @var{file}(@var{section}): warning: defined here
1958 Turning a common symbol into a reference, because a later definition for
1959 the symbol is encountered. This is the same as the previous case,
1960 except that the symbols are encountered in a different order.
1962 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1964 @var{file}(@var{section}): warning: common is here
1968 Merging a common symbol with a previous same-sized common symbol.
1970 @var{file}(@var{section}): warning: multiple common
1972 @var{file}(@var{section}): warning: previous common is here
1976 Merging a common symbol with a previous larger common symbol.
1978 @var{file}(@var{section}): warning: common of `@var{symbol}'
1979 overridden by larger common
1980 @var{file}(@var{section}): warning: larger common is here
1984 Merging a common symbol with a previous smaller common symbol. This is
1985 the same as the previous case, except that the symbols are
1986 encountered in a different order.
1988 @var{file}(@var{section}): warning: common of `@var{symbol}'
1989 overriding smaller common
1990 @var{file}(@var{section}): warning: smaller common is here
1994 @kindex --warn-constructors
1995 @item --warn-constructors
1996 Warn if any global constructors are used. This is only useful for a few
1997 object file formats. For formats like COFF or ELF, the linker can not
1998 detect the use of global constructors.
2000 @kindex --warn-multiple-gp
2001 @item --warn-multiple-gp
2002 Warn if multiple global pointer values are required in the output file.
2003 This is only meaningful for certain processors, such as the Alpha.
2004 Specifically, some processors put large-valued constants in a special
2005 section. A special register (the global pointer) points into the middle
2006 of this section, so that constants can be loaded efficiently via a
2007 base-register relative addressing mode. Since the offset in
2008 base-register relative mode is fixed and relatively small (e.g., 16
2009 bits), this limits the maximum size of the constant pool. Thus, in
2010 large programs, it is often necessary to use multiple global pointer
2011 values in order to be able to address all possible constants. This
2012 option causes a warning to be issued whenever this case occurs.
2015 @cindex warnings, on undefined symbols
2016 @cindex undefined symbols, warnings on
2018 Only warn once for each undefined symbol, rather than once per module
2021 @kindex --warn-section-align
2022 @cindex warnings, on section alignment
2023 @cindex section alignment, warnings on
2024 @item --warn-section-align
2025 Warn if the address of an output section is changed because of
2026 alignment. Typically, the alignment will be set by an input section.
2027 The address will only be changed if it not explicitly specified; that
2028 is, if the @code{SECTIONS} command does not specify a start address for
2029 the section (@pxref{SECTIONS}).
2031 @kindex --warn-shared-textrel
2032 @item --warn-shared-textrel
2033 Warn if the linker adds a DT_TEXTREL to a shared object.
2035 @kindex --warn-alternate-em
2036 @item --warn-alternate-em
2037 Warn if an object has alternate ELF machine code.
2039 @kindex --warn-unresolved-symbols
2040 @item --warn-unresolved-symbols
2041 If the linker is going to report an unresolved symbol (see the option
2042 @option{--unresolved-symbols}) it will normally generate an error.
2043 This option makes it generate a warning instead.
2045 @kindex --error-unresolved-symbols
2046 @item --error-unresolved-symbols
2047 This restores the linker's default behaviour of generating errors when
2048 it is reporting unresolved symbols.
2050 @kindex --whole-archive
2051 @cindex including an entire archive
2052 @item --whole-archive
2053 For each archive mentioned on the command line after the
2054 @option{--whole-archive} option, include every object file in the archive
2055 in the link, rather than searching the archive for the required object
2056 files. This is normally used to turn an archive file into a shared
2057 library, forcing every object to be included in the resulting shared
2058 library. This option may be used more than once.
2060 Two notes when using this option from gcc: First, gcc doesn't know
2061 about this option, so you have to use @option{-Wl,-whole-archive}.
2062 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2063 list of archives, because gcc will add its own list of archives to
2064 your link and you may not want this flag to affect those as well.
2066 @kindex --wrap=@var{symbol}
2067 @item --wrap=@var{symbol}
2068 Use a wrapper function for @var{symbol}. Any undefined reference to
2069 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2070 undefined reference to @code{__real_@var{symbol}} will be resolved to
2073 This can be used to provide a wrapper for a system function. The
2074 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2075 wishes to call the system function, it should call
2076 @code{__real_@var{symbol}}.
2078 Here is a trivial example:
2082 __wrap_malloc (size_t c)
2084 printf ("malloc called with %zu\n", c);
2085 return __real_malloc (c);
2089 If you link other code with this file using @option{--wrap malloc}, then
2090 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2091 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2092 call the real @code{malloc} function.
2094 You may wish to provide a @code{__real_malloc} function as well, so that
2095 links without the @option{--wrap} option will succeed. If you do this,
2096 you should not put the definition of @code{__real_malloc} in the same
2097 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2098 call before the linker has a chance to wrap it to @code{malloc}.
2100 @kindex --eh-frame-hdr
2101 @item --eh-frame-hdr
2102 Request creation of @code{.eh_frame_hdr} section and ELF
2103 @code{PT_GNU_EH_FRAME} segment header.
2105 @kindex --ld-generated-unwind-info
2106 @item --no-ld-generated-unwind-info
2107 Request creation of @code{.eh_frame} unwind info for linker
2108 generated code sections like PLT. This option is on by default
2109 if linker generated unwind info is supported.
2111 @kindex --enable-new-dtags
2112 @kindex --disable-new-dtags
2113 @item --enable-new-dtags
2114 @itemx --disable-new-dtags
2115 This linker can create the new dynamic tags in ELF. But the older ELF
2116 systems may not understand them. If you specify
2117 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2118 and older dynamic tags will be omitted.
2119 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2120 created. By default, the new dynamic tags are not created. Note that
2121 those options are only available for ELF systems.
2123 @kindex --hash-size=@var{number}
2124 @item --hash-size=@var{number}
2125 Set the default size of the linker's hash tables to a prime number
2126 close to @var{number}. Increasing this value can reduce the length of
2127 time it takes the linker to perform its tasks, at the expense of
2128 increasing the linker's memory requirements. Similarly reducing this
2129 value can reduce the memory requirements at the expense of speed.
2131 @kindex --hash-style=@var{style}
2132 @item --hash-style=@var{style}
2133 Set the type of linker's hash table(s). @var{style} can be either
2134 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2135 new style GNU @code{.gnu.hash} section or @code{both} for both
2136 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2137 hash tables. The default is @code{sysv}.
2139 @kindex --reduce-memory-overheads
2140 @item --reduce-memory-overheads
2141 This option reduces memory requirements at ld runtime, at the expense of
2142 linking speed. This was introduced to select the old O(n^2) algorithm
2143 for link map file generation, rather than the new O(n) algorithm which uses
2144 about 40% more memory for symbol storage.
2146 Another effect of the switch is to set the default hash table size to
2147 1021, which again saves memory at the cost of lengthening the linker's
2148 run time. This is not done however if the @option{--hash-size} switch
2151 The @option{--reduce-memory-overheads} switch may be also be used to
2152 enable other tradeoffs in future versions of the linker.
2155 @kindex --build-id=@var{style}
2157 @itemx --build-id=@var{style}
2158 Request creation of @code{.note.gnu.build-id} ELF note section.
2159 The contents of the note are unique bits identifying this linked
2160 file. @var{style} can be @code{uuid} to use 128 random bits,
2161 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
2162 parts of the output contents, @code{md5} to use a 128-bit
2163 @sc{MD5} hash on the normative parts of the output contents, or
2164 @code{0x@var{hexstring}} to use a chosen bit string specified as
2165 an even number of hexadecimal digits (@code{-} and @code{:}
2166 characters between digit pairs are ignored). If @var{style} is
2167 omitted, @code{sha1} is used.
2169 The @code{md5} and @code{sha1} styles produces an identifier
2170 that is always the same in an identical output file, but will be
2171 unique among all nonidentical output files. It is not intended
2172 to be compared as a checksum for the file's contents. A linked
2173 file may be changed later by other tools, but the build ID bit
2174 string identifying the original linked file does not change.
2176 Passing @code{none} for @var{style} disables the setting from any
2177 @code{--build-id} options earlier on the command line.
2182 @subsection Options Specific to i386 PE Targets
2184 @c man begin OPTIONS
2186 The i386 PE linker supports the @option{-shared} option, which causes
2187 the output to be a dynamically linked library (DLL) instead of a
2188 normal executable. You should name the output @code{*.dll} when you
2189 use this option. In addition, the linker fully supports the standard
2190 @code{*.def} files, which may be specified on the linker command line
2191 like an object file (in fact, it should precede archives it exports
2192 symbols from, to ensure that they get linked in, just like a normal
2195 In addition to the options common to all targets, the i386 PE linker
2196 support additional command line options that are specific to the i386
2197 PE target. Options that take values may be separated from their
2198 values by either a space or an equals sign.
2202 @kindex --add-stdcall-alias
2203 @item --add-stdcall-alias
2204 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2205 as-is and also with the suffix stripped.
2206 [This option is specific to the i386 PE targeted port of the linker]
2209 @item --base-file @var{file}
2210 Use @var{file} as the name of a file in which to save the base
2211 addresses of all the relocations needed for generating DLLs with
2213 [This is an i386 PE specific option]
2217 Create a DLL instead of a regular executable. You may also use
2218 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2220 [This option is specific to the i386 PE targeted port of the linker]
2222 @kindex --enable-long-section-names
2223 @kindex --disable-long-section-names
2224 @item --enable-long-section-names
2225 @itemx --disable-long-section-names
2226 The PE variants of the Coff object format add an extension that permits
2227 the use of section names longer than eight characters, the normal limit
2228 for Coff. By default, these names are only allowed in object files, as
2229 fully-linked executable images do not carry the Coff string table required
2230 to support the longer names. As a GNU extension, it is possible to
2231 allow their use in executable images as well, or to (probably pointlessly!)
2232 disallow it in object files, by using these two options. Executable images
2233 generated with these long section names are slightly non-standard, carrying
2234 as they do a string table, and may generate confusing output when examined
2235 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2236 GDB relies on the use of PE long section names to find Dwarf-2 debug
2237 information sections in an executable image at runtime, and so if neither
2238 option is specified on the command-line, @command{ld} will enable long
2239 section names, overriding the default and technically correct behaviour,
2240 when it finds the presence of debug information while linking an executable
2241 image and not stripping symbols.
2242 [This option is valid for all PE targeted ports of the linker]
2244 @kindex --enable-stdcall-fixup
2245 @kindex --disable-stdcall-fixup
2246 @item --enable-stdcall-fixup
2247 @itemx --disable-stdcall-fixup
2248 If the link finds a symbol that it cannot resolve, it will attempt to
2249 do ``fuzzy linking'' by looking for another defined symbol that differs
2250 only in the format of the symbol name (cdecl vs stdcall) and will
2251 resolve that symbol by linking to the match. For example, the
2252 undefined symbol @code{_foo} might be linked to the function
2253 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2254 to the function @code{_bar}. When the linker does this, it prints a
2255 warning, since it normally should have failed to link, but sometimes
2256 import libraries generated from third-party dlls may need this feature
2257 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2258 feature is fully enabled and warnings are not printed. If you specify
2259 @option{--disable-stdcall-fixup}, this feature is disabled and such
2260 mismatches are considered to be errors.
2261 [This option is specific to the i386 PE targeted port of the linker]
2263 @kindex --leading-underscore
2264 @kindex --no-leading-underscore
2265 @item --leading-underscore
2266 @itemx --no-leading-underscore
2267 For most targets default symbol-prefix is an underscore and is defined
2268 in target's description. By this option it is possible to
2269 disable/enable the default underscore symbol-prefix.
2271 @cindex DLLs, creating
2272 @kindex --export-all-symbols
2273 @item --export-all-symbols
2274 If given, all global symbols in the objects used to build a DLL will
2275 be exported by the DLL. Note that this is the default if there
2276 otherwise wouldn't be any exported symbols. When symbols are
2277 explicitly exported via DEF files or implicitly exported via function
2278 attributes, the default is to not export anything else unless this
2279 option is given. Note that the symbols @code{DllMain@@12},
2280 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2281 @code{impure_ptr} will not be automatically
2282 exported. Also, symbols imported from other DLLs will not be
2283 re-exported, nor will symbols specifying the DLL's internal layout
2284 such as those beginning with @code{_head_} or ending with
2285 @code{_iname}. In addition, no symbols from @code{libgcc},
2286 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2287 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2288 not be exported, to help with C++ DLLs. Finally, there is an
2289 extensive list of cygwin-private symbols that are not exported
2290 (obviously, this applies on when building DLLs for cygwin targets).
2291 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2292 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2293 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2294 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2295 @code{cygwin_premain3}, and @code{environ}.
2296 [This option is specific to the i386 PE targeted port of the linker]
2298 @kindex --exclude-symbols
2299 @item --exclude-symbols @var{symbol},@var{symbol},...
2300 Specifies a list of symbols which should not be automatically
2301 exported. The symbol names may be delimited by commas or colons.
2302 [This option is specific to the i386 PE targeted port of the linker]
2304 @kindex --exclude-all-symbols
2305 @item --exclude-all-symbols
2306 Specifies no symbols should be automatically exported.
2307 [This option is specific to the i386 PE targeted port of the linker]
2309 @kindex --file-alignment
2310 @item --file-alignment
2311 Specify the file alignment. Sections in the file will always begin at
2312 file offsets which are multiples of this number. This defaults to
2314 [This option is specific to the i386 PE targeted port of the linker]
2318 @item --heap @var{reserve}
2319 @itemx --heap @var{reserve},@var{commit}
2320 Specify the number of bytes of memory to reserve (and optionally commit)
2321 to be used as heap for this program. The default is 1Mb reserved, 4K
2323 [This option is specific to the i386 PE targeted port of the linker]
2326 @kindex --image-base
2327 @item --image-base @var{value}
2328 Use @var{value} as the base address of your program or dll. This is
2329 the lowest memory location that will be used when your program or dll
2330 is loaded. To reduce the need to relocate and improve performance of
2331 your dlls, each should have a unique base address and not overlap any
2332 other dlls. The default is 0x400000 for executables, and 0x10000000
2334 [This option is specific to the i386 PE targeted port of the linker]
2338 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2339 symbols before they are exported.
2340 [This option is specific to the i386 PE targeted port of the linker]
2342 @kindex --large-address-aware
2343 @item --large-address-aware
2344 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2345 header is set to indicate that this executable supports virtual addresses
2346 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2347 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2348 section of the BOOT.INI. Otherwise, this bit has no effect.
2349 [This option is specific to PE targeted ports of the linker]
2351 @kindex --disable-large-address-aware
2352 @item --disable-large-address-aware
2353 Reverts the effect of a previous @samp{--large-address-aware} option.
2354 This is useful if @samp{--large-address-aware} is always set by the compiler
2355 driver (e.g. Cygwin gcc) and the executable does not support virtual
2356 addresses greater than 2 gigabytes.
2357 [This option is specific to PE targeted ports of the linker]
2359 @kindex --major-image-version
2360 @item --major-image-version @var{value}
2361 Sets the major number of the ``image version''. Defaults to 1.
2362 [This option is specific to the i386 PE targeted port of the linker]
2364 @kindex --major-os-version
2365 @item --major-os-version @var{value}
2366 Sets the major number of the ``os version''. Defaults to 4.
2367 [This option is specific to the i386 PE targeted port of the linker]
2369 @kindex --major-subsystem-version
2370 @item --major-subsystem-version @var{value}
2371 Sets the major number of the ``subsystem version''. Defaults to 4.
2372 [This option is specific to the i386 PE targeted port of the linker]
2374 @kindex --minor-image-version
2375 @item --minor-image-version @var{value}
2376 Sets the minor number of the ``image version''. Defaults to 0.
2377 [This option is specific to the i386 PE targeted port of the linker]
2379 @kindex --minor-os-version
2380 @item --minor-os-version @var{value}
2381 Sets the minor number of the ``os version''. Defaults to 0.
2382 [This option is specific to the i386 PE targeted port of the linker]
2384 @kindex --minor-subsystem-version
2385 @item --minor-subsystem-version @var{value}
2386 Sets the minor number of the ``subsystem version''. Defaults to 0.
2387 [This option is specific to the i386 PE targeted port of the linker]
2389 @cindex DEF files, creating
2390 @cindex DLLs, creating
2391 @kindex --output-def
2392 @item --output-def @var{file}
2393 The linker will create the file @var{file} which will contain a DEF
2394 file corresponding to the DLL the linker is generating. This DEF file
2395 (which should be called @code{*.def}) may be used to create an import
2396 library with @code{dlltool} or may be used as a reference to
2397 automatically or implicitly exported symbols.
2398 [This option is specific to the i386 PE targeted port of the linker]
2400 @cindex DLLs, creating
2401 @kindex --out-implib
2402 @item --out-implib @var{file}
2403 The linker will create the file @var{file} which will contain an
2404 import lib corresponding to the DLL the linker is generating. This
2405 import lib (which should be called @code{*.dll.a} or @code{*.a}
2406 may be used to link clients against the generated DLL; this behaviour
2407 makes it possible to skip a separate @code{dlltool} import library
2409 [This option is specific to the i386 PE targeted port of the linker]
2411 @kindex --enable-auto-image-base
2412 @item --enable-auto-image-base
2413 Automatically choose the image base for DLLs, unless one is specified
2414 using the @code{--image-base} argument. By using a hash generated
2415 from the dllname to create unique image bases for each DLL, in-memory
2416 collisions and relocations which can delay program execution are
2418 [This option is specific to the i386 PE targeted port of the linker]
2420 @kindex --disable-auto-image-base
2421 @item --disable-auto-image-base
2422 Do not automatically generate a unique image base. If there is no
2423 user-specified image base (@code{--image-base}) then use the platform
2425 [This option is specific to the i386 PE targeted port of the linker]
2427 @cindex DLLs, linking to
2428 @kindex --dll-search-prefix
2429 @item --dll-search-prefix @var{string}
2430 When linking dynamically to a dll without an import library,
2431 search for @code{<string><basename>.dll} in preference to
2432 @code{lib<basename>.dll}. This behaviour allows easy distinction
2433 between DLLs built for the various "subplatforms": native, cygwin,
2434 uwin, pw, etc. For instance, cygwin DLLs typically use
2435 @code{--dll-search-prefix=cyg}.
2436 [This option is specific to the i386 PE targeted port of the linker]
2438 @kindex --enable-auto-import
2439 @item --enable-auto-import
2440 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2441 DATA imports from DLLs, and create the necessary thunking symbols when
2442 building the import libraries with those DATA exports. Note: Use of the
2443 'auto-import' extension will cause the text section of the image file
2444 to be made writable. This does not conform to the PE-COFF format
2445 specification published by Microsoft.
2447 Note - use of the 'auto-import' extension will also cause read only
2448 data which would normally be placed into the .rdata section to be
2449 placed into the .data section instead. This is in order to work
2450 around a problem with consts that is described here:
2451 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2453 Using 'auto-import' generally will 'just work' -- but sometimes you may
2456 "variable '<var>' can't be auto-imported. Please read the
2457 documentation for ld's @code{--enable-auto-import} for details."
2459 This message occurs when some (sub)expression accesses an address
2460 ultimately given by the sum of two constants (Win32 import tables only
2461 allow one). Instances where this may occur include accesses to member
2462 fields of struct variables imported from a DLL, as well as using a
2463 constant index into an array variable imported from a DLL. Any
2464 multiword variable (arrays, structs, long long, etc) may trigger
2465 this error condition. However, regardless of the exact data type
2466 of the offending exported variable, ld will always detect it, issue
2467 the warning, and exit.
2469 There are several ways to address this difficulty, regardless of the
2470 data type of the exported variable:
2472 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2473 of adjusting references in your client code for runtime environment, so
2474 this method works only when runtime environment supports this feature.
2476 A second solution is to force one of the 'constants' to be a variable --
2477 that is, unknown and un-optimizable at compile time. For arrays,
2478 there are two possibilities: a) make the indexee (the array's address)
2479 a variable, or b) make the 'constant' index a variable. Thus:
2482 extern type extern_array[];
2484 @{ volatile type *t=extern_array; t[1] @}
2490 extern type extern_array[];
2492 @{ volatile int t=1; extern_array[t] @}
2495 For structs (and most other multiword data types) the only option
2496 is to make the struct itself (or the long long, or the ...) variable:
2499 extern struct s extern_struct;
2500 extern_struct.field -->
2501 @{ volatile struct s *t=&extern_struct; t->field @}
2507 extern long long extern_ll;
2509 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2512 A third method of dealing with this difficulty is to abandon
2513 'auto-import' for the offending symbol and mark it with
2514 @code{__declspec(dllimport)}. However, in practice that
2515 requires using compile-time #defines to indicate whether you are
2516 building a DLL, building client code that will link to the DLL, or
2517 merely building/linking to a static library. In making the choice
2518 between the various methods of resolving the 'direct address with
2519 constant offset' problem, you should consider typical real-world usage:
2527 void main(int argc, char **argv)@{
2528 printf("%d\n",arr[1]);
2538 void main(int argc, char **argv)@{
2539 /* This workaround is for win32 and cygwin; do not "optimize" */
2540 volatile int *parr = arr;
2541 printf("%d\n",parr[1]);
2548 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2549 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2550 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2551 #define FOO_IMPORT __declspec(dllimport)
2555 extern FOO_IMPORT int arr[];
2558 void main(int argc, char **argv)@{
2559 printf("%d\n",arr[1]);
2563 A fourth way to avoid this problem is to re-code your
2564 library to use a functional interface rather than a data interface
2565 for the offending variables (e.g. set_foo() and get_foo() accessor
2567 [This option is specific to the i386 PE targeted port of the linker]
2569 @kindex --disable-auto-import
2570 @item --disable-auto-import
2571 Do not attempt to do sophisticated linking of @code{_symbol} to
2572 @code{__imp__symbol} for DATA imports from DLLs.
2573 [This option is specific to the i386 PE targeted port of the linker]
2575 @kindex --enable-runtime-pseudo-reloc
2576 @item --enable-runtime-pseudo-reloc
2577 If your code contains expressions described in --enable-auto-import section,
2578 that is, DATA imports from DLL with non-zero offset, this switch will create
2579 a vector of 'runtime pseudo relocations' which can be used by runtime
2580 environment to adjust references to such data in your client code.
2581 [This option is specific to the i386 PE targeted port of the linker]
2583 @kindex --disable-runtime-pseudo-reloc
2584 @item --disable-runtime-pseudo-reloc
2585 Do not create pseudo relocations for non-zero offset DATA imports from
2587 [This option is specific to the i386 PE targeted port of the linker]
2589 @kindex --enable-extra-pe-debug
2590 @item --enable-extra-pe-debug
2591 Show additional debug info related to auto-import symbol thunking.
2592 [This option is specific to the i386 PE targeted port of the linker]
2594 @kindex --section-alignment
2595 @item --section-alignment
2596 Sets the section alignment. Sections in memory will always begin at
2597 addresses which are a multiple of this number. Defaults to 0x1000.
2598 [This option is specific to the i386 PE targeted port of the linker]
2602 @item --stack @var{reserve}
2603 @itemx --stack @var{reserve},@var{commit}
2604 Specify the number of bytes of memory to reserve (and optionally commit)
2605 to be used as stack for this program. The default is 2Mb reserved, 4K
2607 [This option is specific to the i386 PE targeted port of the linker]
2610 @item --subsystem @var{which}
2611 @itemx --subsystem @var{which}:@var{major}
2612 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2613 Specifies the subsystem under which your program will execute. The
2614 legal values for @var{which} are @code{native}, @code{windows},
2615 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2616 the subsystem version also. Numeric values are also accepted for
2618 [This option is specific to the i386 PE targeted port of the linker]
2620 The following options set flags in the @code{DllCharacteristics} field
2621 of the PE file header:
2622 [These options are specific to PE targeted ports of the linker]
2624 @kindex --dynamicbase
2626 The image base address may be relocated using address space layout
2627 randomization (ASLR). This feature was introduced with MS Windows
2628 Vista for i386 PE targets.
2630 @kindex --forceinteg
2632 Code integrity checks are enforced.
2636 The image is compatible with the Data Execution Prevention.
2637 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2639 @kindex --no-isolation
2640 @item --no-isolation
2641 Although the image understands isolation, do not isolate the image.
2645 The image does not use SEH. No SE handler may be called from
2650 Do not bind this image.
2654 The driver uses the MS Windows Driver Model.
2658 The image is Terminal Server aware.
2665 @subsection Options specific to C6X uClinux targets
2667 @c man begin OPTIONS
2669 The C6X uClinux target uses a binary format called DSBT to support shared
2670 libraries. Each shared library in the system needs to have a unique index;
2671 all executables use an index of 0.
2676 @item --dsbt-size @var{size}
2677 This option sets the number of entires in the DSBT of the current executable
2678 or shared library to @var{size}. The default is to create a table with 64
2681 @kindex --dsbt-index
2682 @item --dsbt-index @var{index}
2683 This option sets the DSBT index of the current executable or shared library
2684 to @var{index}. The default is 0, which is appropriate for generating
2685 executables. If a shared library is generated with a DSBT index of 0, the
2686 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
2688 @kindex --no-merge-exidx-entries
2689 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
2690 exidx entries in frame unwind info.
2698 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2700 @c man begin OPTIONS
2702 The 68HC11 and 68HC12 linkers support specific options to control the
2703 memory bank switching mapping and trampoline code generation.
2707 @kindex --no-trampoline
2708 @item --no-trampoline
2709 This option disables the generation of trampoline. By default a trampoline
2710 is generated for each far function which is called using a @code{jsr}
2711 instruction (this happens when a pointer to a far function is taken).
2713 @kindex --bank-window
2714 @item --bank-window @var{name}
2715 This option indicates to the linker the name of the memory region in
2716 the @samp{MEMORY} specification that describes the memory bank window.
2717 The definition of such region is then used by the linker to compute
2718 paging and addresses within the memory window.
2726 @subsection Options specific to Motorola 68K target
2728 @c man begin OPTIONS
2730 The following options are supported to control handling of GOT generation
2731 when linking for 68K targets.
2736 @item --got=@var{type}
2737 This option tells the linker which GOT generation scheme to use.
2738 @var{type} should be one of @samp{single}, @samp{negative},
2739 @samp{multigot} or @samp{target}. For more information refer to the
2740 Info entry for @file{ld}.
2748 @subsection Options specific to MIPS targets
2750 @c man begin OPTIONS
2752 The following options are supported to control microMIPS instruction
2753 generation when linking for MIPS targets.
2761 These options control the choice of microMIPS instructions used in code
2762 generated by the linker, such as that in the PLT or lazy binding stubs,
2763 or in relaxation. If @samp{--insn32} is used, then the linker only uses
2764 32-bit instruction encodings. By default or if @samp{--no-insn32} is
2765 used, all instruction encodings are used, including 16-bit ones where
2775 @section Environment Variables
2777 @c man begin ENVIRONMENT
2779 You can change the behaviour of @command{ld} with the environment variables
2780 @ifclear SingleFormat
2783 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2785 @ifclear SingleFormat
2787 @cindex default input format
2788 @code{GNUTARGET} determines the input-file object format if you don't
2789 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2790 of the BFD names for an input format (@pxref{BFD}). If there is no
2791 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2792 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2793 attempts to discover the input format by examining binary input files;
2794 this method often succeeds, but there are potential ambiguities, since
2795 there is no method of ensuring that the magic number used to specify
2796 object-file formats is unique. However, the configuration procedure for
2797 BFD on each system places the conventional format for that system first
2798 in the search-list, so ambiguities are resolved in favor of convention.
2802 @cindex default emulation
2803 @cindex emulation, default
2804 @code{LDEMULATION} determines the default emulation if you don't use the
2805 @samp{-m} option. The emulation can affect various aspects of linker
2806 behaviour, particularly the default linker script. You can list the
2807 available emulations with the @samp{--verbose} or @samp{-V} options. If
2808 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2809 variable is not defined, the default emulation depends upon how the
2810 linker was configured.
2812 @kindex COLLECT_NO_DEMANGLE
2813 @cindex demangling, default
2814 Normally, the linker will default to demangling symbols. However, if
2815 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2816 default to not demangling symbols. This environment variable is used in
2817 a similar fashion by the @code{gcc} linker wrapper program. The default
2818 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2825 @chapter Linker Scripts
2828 @cindex linker scripts
2829 @cindex command files
2830 Every link is controlled by a @dfn{linker script}. This script is
2831 written in the linker command language.
2833 The main purpose of the linker script is to describe how the sections in
2834 the input files should be mapped into the output file, and to control
2835 the memory layout of the output file. Most linker scripts do nothing
2836 more than this. However, when necessary, the linker script can also
2837 direct the linker to perform many other operations, using the commands
2840 The linker always uses a linker script. If you do not supply one
2841 yourself, the linker will use a default script that is compiled into the
2842 linker executable. You can use the @samp{--verbose} command line option
2843 to display the default linker script. Certain command line options,
2844 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2846 You may supply your own linker script by using the @samp{-T} command
2847 line option. When you do this, your linker script will replace the
2848 default linker script.
2850 You may also use linker scripts implicitly by naming them as input files
2851 to the linker, as though they were files to be linked. @xref{Implicit
2855 * Basic Script Concepts:: Basic Linker Script Concepts
2856 * Script Format:: Linker Script Format
2857 * Simple Example:: Simple Linker Script Example
2858 * Simple Commands:: Simple Linker Script Commands
2859 * Assignments:: Assigning Values to Symbols
2860 * SECTIONS:: SECTIONS Command
2861 * MEMORY:: MEMORY Command
2862 * PHDRS:: PHDRS Command
2863 * VERSION:: VERSION Command
2864 * Expressions:: Expressions in Linker Scripts
2865 * Implicit Linker Scripts:: Implicit Linker Scripts
2868 @node Basic Script Concepts
2869 @section Basic Linker Script Concepts
2870 @cindex linker script concepts
2871 We need to define some basic concepts and vocabulary in order to
2872 describe the linker script language.
2874 The linker combines input files into a single output file. The output
2875 file and each input file are in a special data format known as an
2876 @dfn{object file format}. Each file is called an @dfn{object file}.
2877 The output file is often called an @dfn{executable}, but for our
2878 purposes we will also call it an object file. Each object file has,
2879 among other things, a list of @dfn{sections}. We sometimes refer to a
2880 section in an input file as an @dfn{input section}; similarly, a section
2881 in the output file is an @dfn{output section}.
2883 Each section in an object file has a name and a size. Most sections
2884 also have an associated block of data, known as the @dfn{section
2885 contents}. A section may be marked as @dfn{loadable}, which means that
2886 the contents should be loaded into memory when the output file is run.
2887 A section with no contents may be @dfn{allocatable}, which means that an
2888 area in memory should be set aside, but nothing in particular should be
2889 loaded there (in some cases this memory must be zeroed out). A section
2890 which is neither loadable nor allocatable typically contains some sort
2891 of debugging information.
2893 Every loadable or allocatable output section has two addresses. The
2894 first is the @dfn{VMA}, or virtual memory address. This is the address
2895 the section will have when the output file is run. The second is the
2896 @dfn{LMA}, or load memory address. This is the address at which the
2897 section will be loaded. In most cases the two addresses will be the
2898 same. An example of when they might be different is when a data section
2899 is loaded into ROM, and then copied into RAM when the program starts up
2900 (this technique is often used to initialize global variables in a ROM
2901 based system). In this case the ROM address would be the LMA, and the
2902 RAM address would be the VMA.
2904 You can see the sections in an object file by using the @code{objdump}
2905 program with the @samp{-h} option.
2907 Every object file also has a list of @dfn{symbols}, known as the
2908 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2909 has a name, and each defined symbol has an address, among other
2910 information. If you compile a C or C++ program into an object file, you
2911 will get a defined symbol for every defined function and global or
2912 static variable. Every undefined function or global variable which is
2913 referenced in the input file will become an undefined symbol.
2915 You can see the symbols in an object file by using the @code{nm}
2916 program, or by using the @code{objdump} program with the @samp{-t}
2920 @section Linker Script Format
2921 @cindex linker script format
2922 Linker scripts are text files.
2924 You write a linker script as a series of commands. Each command is
2925 either a keyword, possibly followed by arguments, or an assignment to a
2926 symbol. You may separate commands using semicolons. Whitespace is
2929 Strings such as file or format names can normally be entered directly.
2930 If the file name contains a character such as a comma which would
2931 otherwise serve to separate file names, you may put the file name in
2932 double quotes. There is no way to use a double quote character in a
2935 You may include comments in linker scripts just as in C, delimited by
2936 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2939 @node Simple Example
2940 @section Simple Linker Script Example
2941 @cindex linker script example
2942 @cindex example of linker script
2943 Many linker scripts are fairly simple.
2945 The simplest possible linker script has just one command:
2946 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2947 memory layout of the output file.
2949 The @samp{SECTIONS} command is a powerful command. Here we will
2950 describe a simple use of it. Let's assume your program consists only of
2951 code, initialized data, and uninitialized data. These will be in the
2952 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2953 Let's assume further that these are the only sections which appear in
2956 For this example, let's say that the code should be loaded at address
2957 0x10000, and that the data should start at address 0x8000000. Here is a
2958 linker script which will do that:
2963 .text : @{ *(.text) @}
2965 .data : @{ *(.data) @}
2966 .bss : @{ *(.bss) @}
2970 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2971 followed by a series of symbol assignments and output section
2972 descriptions enclosed in curly braces.
2974 The first line inside the @samp{SECTIONS} command of the above example
2975 sets the value of the special symbol @samp{.}, which is the location
2976 counter. If you do not specify the address of an output section in some
2977 other way (other ways are described later), the address is set from the
2978 current value of the location counter. The location counter is then
2979 incremented by the size of the output section. At the start of the
2980 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2982 The second line defines an output section, @samp{.text}. The colon is
2983 required syntax which may be ignored for now. Within the curly braces
2984 after the output section name, you list the names of the input sections
2985 which should be placed into this output section. The @samp{*} is a
2986 wildcard which matches any file name. The expression @samp{*(.text)}
2987 means all @samp{.text} input sections in all input files.
2989 Since the location counter is @samp{0x10000} when the output section
2990 @samp{.text} is defined, the linker will set the address of the
2991 @samp{.text} section in the output file to be @samp{0x10000}.
2993 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2994 the output file. The linker will place the @samp{.data} output section
2995 at address @samp{0x8000000}. After the linker places the @samp{.data}
2996 output section, the value of the location counter will be
2997 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2998 effect is that the linker will place the @samp{.bss} output section
2999 immediately after the @samp{.data} output section in memory.
3001 The linker will ensure that each output section has the required
3002 alignment, by increasing the location counter if necessary. In this
3003 example, the specified addresses for the @samp{.text} and @samp{.data}
3004 sections will probably satisfy any alignment constraints, but the linker
3005 may have to create a small gap between the @samp{.data} and @samp{.bss}
3008 That's it! That's a simple and complete linker script.
3010 @node Simple Commands
3011 @section Simple Linker Script Commands
3012 @cindex linker script simple commands
3013 In this section we describe the simple linker script commands.
3016 * Entry Point:: Setting the entry point
3017 * File Commands:: Commands dealing with files
3018 @ifclear SingleFormat
3019 * Format Commands:: Commands dealing with object file formats
3022 * REGION_ALIAS:: Assign alias names to memory regions
3023 * Miscellaneous Commands:: Other linker script commands
3027 @subsection Setting the Entry Point
3028 @kindex ENTRY(@var{symbol})
3029 @cindex start of execution
3030 @cindex first instruction
3032 The first instruction to execute in a program is called the @dfn{entry
3033 point}. You can use the @code{ENTRY} linker script command to set the
3034 entry point. The argument is a symbol name:
3039 There are several ways to set the entry point. The linker will set the
3040 entry point by trying each of the following methods in order, and
3041 stopping when one of them succeeds:
3044 the @samp{-e} @var{entry} command-line option;
3046 the @code{ENTRY(@var{symbol})} command in a linker script;
3048 the value of a target specific symbol, if it is defined; For many
3049 targets this is @code{start}, but PE and BeOS based systems for example
3050 check a list of possible entry symbols, matching the first one found.
3052 the address of the first byte of the @samp{.text} section, if present;
3054 The address @code{0}.
3058 @subsection Commands Dealing with Files
3059 @cindex linker script file commands
3060 Several linker script commands deal with files.
3063 @item INCLUDE @var{filename}
3064 @kindex INCLUDE @var{filename}
3065 @cindex including a linker script
3066 Include the linker script @var{filename} at this point. The file will
3067 be searched for in the current directory, and in any directory specified
3068 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3071 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3072 @code{SECTIONS} commands, or in output section descriptions.
3074 @item INPUT(@var{file}, @var{file}, @dots{})
3075 @itemx INPUT(@var{file} @var{file} @dots{})
3076 @kindex INPUT(@var{files})
3077 @cindex input files in linker scripts
3078 @cindex input object files in linker scripts
3079 @cindex linker script input object files
3080 The @code{INPUT} command directs the linker to include the named files
3081 in the link, as though they were named on the command line.
3083 For example, if you always want to include @file{subr.o} any time you do
3084 a link, but you can't be bothered to put it on every link command line,
3085 then you can put @samp{INPUT (subr.o)} in your linker script.
3087 In fact, if you like, you can list all of your input files in the linker
3088 script, and then invoke the linker with nothing but a @samp{-T} option.
3090 In case a @dfn{sysroot prefix} is configured, and the filename starts
3091 with the @samp{/} character, and the script being processed was
3092 located inside the @dfn{sysroot prefix}, the filename will be looked
3093 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3094 open the file in the current directory. If it is not found, the
3095 linker will search through the archive library search path. See the
3096 description of @samp{-L} in @ref{Options,,Command Line Options}.
3098 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3099 name to @code{lib@var{file}.a}, as with the command line argument
3102 When you use the @code{INPUT} command in an implicit linker script, the
3103 files will be included in the link at the point at which the linker
3104 script file is included. This can affect archive searching.
3106 @item GROUP(@var{file}, @var{file}, @dots{})
3107 @itemx GROUP(@var{file} @var{file} @dots{})
3108 @kindex GROUP(@var{files})
3109 @cindex grouping input files
3110 The @code{GROUP} command is like @code{INPUT}, except that the named
3111 files should all be archives, and they are searched repeatedly until no
3112 new undefined references are created. See the description of @samp{-(}
3113 in @ref{Options,,Command Line Options}.
3115 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3116 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3117 @kindex AS_NEEDED(@var{files})
3118 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3119 commands, among other filenames. The files listed will be handled
3120 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3121 with the exception of ELF shared libraries, that will be added only
3122 when they are actually needed. This construct essentially enables
3123 @option{--as-needed} option for all the files listed inside of it
3124 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3127 @item OUTPUT(@var{filename})
3128 @kindex OUTPUT(@var{filename})
3129 @cindex output file name in linker script
3130 The @code{OUTPUT} command names the output file. Using
3131 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3132 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3133 Line Options}). If both are used, the command line option takes
3136 You can use the @code{OUTPUT} command to define a default name for the
3137 output file other than the usual default of @file{a.out}.
3139 @item SEARCH_DIR(@var{path})
3140 @kindex SEARCH_DIR(@var{path})
3141 @cindex library search path in linker script
3142 @cindex archive search path in linker script
3143 @cindex search path in linker script
3144 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3145 @command{ld} looks for archive libraries. Using
3146 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3147 on the command line (@pxref{Options,,Command Line Options}). If both
3148 are used, then the linker will search both paths. Paths specified using
3149 the command line option are searched first.
3151 @item STARTUP(@var{filename})
3152 @kindex STARTUP(@var{filename})
3153 @cindex first input file
3154 The @code{STARTUP} command is just like the @code{INPUT} command, except
3155 that @var{filename} will become the first input file to be linked, as
3156 though it were specified first on the command line. This may be useful
3157 when using a system in which the entry point is always the start of the
3161 @ifclear SingleFormat
3162 @node Format Commands
3163 @subsection Commands Dealing with Object File Formats
3164 A couple of linker script commands deal with object file formats.
3167 @item OUTPUT_FORMAT(@var{bfdname})
3168 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3169 @kindex OUTPUT_FORMAT(@var{bfdname})
3170 @cindex output file format in linker script
3171 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3172 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3173 exactly like using @samp{--oformat @var{bfdname}} on the command line
3174 (@pxref{Options,,Command Line Options}). If both are used, the command
3175 line option takes precedence.
3177 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3178 formats based on the @samp{-EB} and @samp{-EL} command line options.
3179 This permits the linker script to set the output format based on the
3182 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3183 will be the first argument, @var{default}. If @samp{-EB} is used, the
3184 output format will be the second argument, @var{big}. If @samp{-EL} is
3185 used, the output format will be the third argument, @var{little}.
3187 For example, the default linker script for the MIPS ELF target uses this
3190 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3192 This says that the default format for the output file is
3193 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3194 option, the output file will be created in the @samp{elf32-littlemips}
3197 @item TARGET(@var{bfdname})
3198 @kindex TARGET(@var{bfdname})
3199 @cindex input file format in linker script
3200 The @code{TARGET} command names the BFD format to use when reading input
3201 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3202 This command is like using @samp{-b @var{bfdname}} on the command line
3203 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3204 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3205 command is also used to set the format for the output file. @xref{BFD}.
3210 @subsection Assign alias names to memory regions
3211 @kindex REGION_ALIAS(@var{alias}, @var{region})
3212 @cindex region alias
3213 @cindex region names
3215 Alias names can be added to existing memory regions created with the
3216 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3219 REGION_ALIAS(@var{alias}, @var{region})
3222 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3223 memory region @var{region}. This allows a flexible mapping of output sections
3224 to memory regions. An example follows.
3226 Suppose we have an application for embedded systems which come with various
3227 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3228 that allows code execution or data storage. Some may have a read-only,
3229 non-volatile memory @code{ROM} that allows code execution and read-only data
3230 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3231 read-only data access and no code execution capability. We have four output
3236 @code{.text} program code;
3238 @code{.rodata} read-only data;
3240 @code{.data} read-write initialized data;
3242 @code{.bss} read-write zero initialized data.
3245 The goal is to provide a linker command file that contains a system independent
3246 part defining the output sections and a system dependent part mapping the
3247 output sections to the memory regions available on the system. Our embedded
3248 systems come with three different memory setups @code{A}, @code{B} and
3250 @multitable @columnfractions .25 .25 .25 .25
3251 @item Section @tab Variant A @tab Variant B @tab Variant C
3252 @item .text @tab RAM @tab ROM @tab ROM
3253 @item .rodata @tab RAM @tab ROM @tab ROM2
3254 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3255 @item .bss @tab RAM @tab RAM @tab RAM
3257 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3258 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3259 the load address of the @code{.data} section starts in all three variants at
3260 the end of the @code{.rodata} section.
3262 The base linker script that deals with the output sections follows. It
3263 includes the system dependent @code{linkcmds.memory} file that describes the
3266 INCLUDE linkcmds.memory
3279 .data : AT (rodata_end)
3284 data_size = SIZEOF(.data);
3285 data_load_start = LOADADDR(.data);
3293 Now we need three different @code{linkcmds.memory} files to define memory
3294 regions and alias names. The content of @code{linkcmds.memory} for the three
3295 variants @code{A}, @code{B} and @code{C}:
3298 Here everything goes into the @code{RAM}.
3302 RAM : ORIGIN = 0, LENGTH = 4M
3305 REGION_ALIAS("REGION_TEXT", RAM);
3306 REGION_ALIAS("REGION_RODATA", RAM);
3307 REGION_ALIAS("REGION_DATA", RAM);
3308 REGION_ALIAS("REGION_BSS", RAM);
3311 Program code and read-only data go into the @code{ROM}. Read-write data goes
3312 into the @code{RAM}. An image of the initialized data is loaded into the
3313 @code{ROM} and will be copied during system start into the @code{RAM}.
3317 ROM : ORIGIN = 0, LENGTH = 3M
3318 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3321 REGION_ALIAS("REGION_TEXT", ROM);
3322 REGION_ALIAS("REGION_RODATA", ROM);
3323 REGION_ALIAS("REGION_DATA", RAM);
3324 REGION_ALIAS("REGION_BSS", RAM);
3327 Program code goes into the @code{ROM}. Read-only data goes into the
3328 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3329 initialized data is loaded into the @code{ROM2} and will be copied during
3330 system start into the @code{RAM}.
3334 ROM : ORIGIN = 0, LENGTH = 2M
3335 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3336 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3339 REGION_ALIAS("REGION_TEXT", ROM);
3340 REGION_ALIAS("REGION_RODATA", ROM2);
3341 REGION_ALIAS("REGION_DATA", RAM);
3342 REGION_ALIAS("REGION_BSS", RAM);
3346 It is possible to write a common system initialization routine to copy the
3347 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3352 extern char data_start [];
3353 extern char data_size [];
3354 extern char data_load_start [];
3356 void copy_data(void)
3358 if (data_start != data_load_start)
3360 memcpy(data_start, data_load_start, (size_t) data_size);
3365 @node Miscellaneous Commands
3366 @subsection Other Linker Script Commands
3367 There are a few other linker scripts commands.
3370 @item ASSERT(@var{exp}, @var{message})
3372 @cindex assertion in linker script
3373 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3374 with an error code, and print @var{message}.
3376 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3378 @cindex undefined symbol in linker script
3379 Force @var{symbol} to be entered in the output file as an undefined
3380 symbol. Doing this may, for example, trigger linking of additional
3381 modules from standard libraries. You may list several @var{symbol}s for
3382 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3383 command has the same effect as the @samp{-u} command-line option.
3385 @item FORCE_COMMON_ALLOCATION
3386 @kindex FORCE_COMMON_ALLOCATION
3387 @cindex common allocation in linker script
3388 This command has the same effect as the @samp{-d} command-line option:
3389 to make @command{ld} assign space to common symbols even if a relocatable
3390 output file is specified (@samp{-r}).
3392 @item INHIBIT_COMMON_ALLOCATION
3393 @kindex INHIBIT_COMMON_ALLOCATION
3394 @cindex common allocation in linker script
3395 This command has the same effect as the @samp{--no-define-common}
3396 command-line option: to make @code{ld} omit the assignment of addresses
3397 to common symbols even for a non-relocatable output file.
3399 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3401 @cindex insert user script into default script
3402 This command is typically used in a script specified by @samp{-T} to
3403 augment the default @code{SECTIONS} with, for example, overlays. It
3404 inserts all prior linker script statements after (or before)
3405 @var{output_section}, and also causes @samp{-T} to not override the
3406 default linker script. The exact insertion point is as for orphan
3407 sections. @xref{Location Counter}. The insertion happens after the
3408 linker has mapped input sections to output sections. Prior to the
3409 insertion, since @samp{-T} scripts are parsed before the default
3410 linker script, statements in the @samp{-T} script occur before the
3411 default linker script statements in the internal linker representation
3412 of the script. In particular, input section assignments will be made
3413 to @samp{-T} output sections before those in the default script. Here
3414 is an example of how a @samp{-T} script using @code{INSERT} might look:
3421 .ov1 @{ ov1*(.text) @}
3422 .ov2 @{ ov2*(.text) @}
3428 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3429 @kindex NOCROSSREFS(@var{sections})
3430 @cindex cross references
3431 This command may be used to tell @command{ld} to issue an error about any
3432 references among certain output sections.
3434 In certain types of programs, particularly on embedded systems when
3435 using overlays, when one section is loaded into memory, another section
3436 will not be. Any direct references between the two sections would be
3437 errors. For example, it would be an error if code in one section called
3438 a function defined in the other section.
3440 The @code{NOCROSSREFS} command takes a list of output section names. If
3441 @command{ld} detects any cross references between the sections, it reports
3442 an error and returns a non-zero exit status. Note that the
3443 @code{NOCROSSREFS} command uses output section names, not input section
3446 @ifclear SingleFormat
3447 @item OUTPUT_ARCH(@var{bfdarch})
3448 @kindex OUTPUT_ARCH(@var{bfdarch})
3449 @cindex machine architecture
3450 @cindex architecture
3451 Specify a particular output machine architecture. The argument is one
3452 of the names used by the BFD library (@pxref{BFD}). You can see the
3453 architecture of an object file by using the @code{objdump} program with
3454 the @samp{-f} option.
3457 @item LD_FEATURE(@var{string})
3458 @kindex LD_FEATURE(@var{string})
3459 This command may be used to modify @command{ld} behavior. If
3460 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3461 in a script are simply treated as numbers everywhere.
3462 @xref{Expression Section}.
3466 @section Assigning Values to Symbols
3467 @cindex assignment in scripts
3468 @cindex symbol definition, scripts
3469 @cindex variables, defining
3470 You may assign a value to a symbol in a linker script. This will define
3471 the symbol and place it into the symbol table with a global scope.
3474 * Simple Assignments:: Simple Assignments
3477 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3478 * Source Code Reference:: How to use a linker script defined symbol in source code
3481 @node Simple Assignments
3482 @subsection Simple Assignments
3484 You may assign to a symbol using any of the C assignment operators:
3487 @item @var{symbol} = @var{expression} ;
3488 @itemx @var{symbol} += @var{expression} ;
3489 @itemx @var{symbol} -= @var{expression} ;
3490 @itemx @var{symbol} *= @var{expression} ;
3491 @itemx @var{symbol} /= @var{expression} ;
3492 @itemx @var{symbol} <<= @var{expression} ;
3493 @itemx @var{symbol} >>= @var{expression} ;
3494 @itemx @var{symbol} &= @var{expression} ;
3495 @itemx @var{symbol} |= @var{expression} ;
3498 The first case will define @var{symbol} to the value of
3499 @var{expression}. In the other cases, @var{symbol} must already be
3500 defined, and the value will be adjusted accordingly.
3502 The special symbol name @samp{.} indicates the location counter. You
3503 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3505 The semicolon after @var{expression} is required.
3507 Expressions are defined below; see @ref{Expressions}.
3509 You may write symbol assignments as commands in their own right, or as
3510 statements within a @code{SECTIONS} command, or as part of an output
3511 section description in a @code{SECTIONS} command.
3513 The section of the symbol will be set from the section of the
3514 expression; for more information, see @ref{Expression Section}.
3516 Here is an example showing the three different places that symbol
3517 assignments may be used:
3528 _bdata = (. + 3) & ~ 3;
3529 .data : @{ *(.data) @}
3533 In this example, the symbol @samp{floating_point} will be defined as
3534 zero. The symbol @samp{_etext} will be defined as the address following
3535 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3536 defined as the address following the @samp{.text} output section aligned
3537 upward to a 4 byte boundary.
3542 For ELF targeted ports, define a symbol that will be hidden and won't be
3543 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
3545 Here is the example from @ref{Simple Assignments}, rewritten to use
3549 HIDDEN(floating_point = 0);
3557 HIDDEN(_bdata = (. + 3) & ~ 3);
3558 .data : @{ *(.data) @}
3562 In this case none of the three symbols will be visible outside this module.
3567 In some cases, it is desirable for a linker script to define a symbol
3568 only if it is referenced and is not defined by any object included in
3569 the link. For example, traditional linkers defined the symbol
3570 @samp{etext}. However, ANSI C requires that the user be able to use
3571 @samp{etext} as a function name without encountering an error. The
3572 @code{PROVIDE} keyword may be used to define a symbol, such as
3573 @samp{etext}, only if it is referenced but not defined. The syntax is
3574 @code{PROVIDE(@var{symbol} = @var{expression})}.
3576 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3589 In this example, if the program defines @samp{_etext} (with a leading
3590 underscore), the linker will give a multiple definition error. If, on
3591 the other hand, the program defines @samp{etext} (with no leading
3592 underscore), the linker will silently use the definition in the program.
3593 If the program references @samp{etext} but does not define it, the
3594 linker will use the definition in the linker script.
3596 @node PROVIDE_HIDDEN
3597 @subsection PROVIDE_HIDDEN
3598 @cindex PROVIDE_HIDDEN
3599 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3600 hidden and won't be exported.
3602 @node Source Code Reference
3603 @subsection Source Code Reference
3605 Accessing a linker script defined variable from source code is not
3606 intuitive. In particular a linker script symbol is not equivalent to
3607 a variable declaration in a high level language, it is instead a
3608 symbol that does not have a value.
3610 Before going further, it is important to note that compilers often
3611 transform names in the source code into different names when they are
3612 stored in the symbol table. For example, Fortran compilers commonly
3613 prepend or append an underscore, and C++ performs extensive @samp{name
3614 mangling}. Therefore there might be a discrepancy between the name
3615 of a variable as it is used in source code and the name of the same
3616 variable as it is defined in a linker script. For example in C a
3617 linker script variable might be referred to as:
3623 But in the linker script it might be defined as:
3629 In the remaining examples however it is assumed that no name
3630 transformation has taken place.
3632 When a symbol is declared in a high level language such as C, two
3633 things happen. The first is that the compiler reserves enough space
3634 in the program's memory to hold the @emph{value} of the symbol. The
3635 second is that the compiler creates an entry in the program's symbol
3636 table which holds the symbol's @emph{address}. ie the symbol table
3637 contains the address of the block of memory holding the symbol's
3638 value. So for example the following C declaration, at file scope:
3644 creates an entry called @samp{foo} in the symbol table. This entry
3645 holds the address of an @samp{int} sized block of memory where the
3646 number 1000 is initially stored.
3648 When a program references a symbol the compiler generates code that
3649 first accesses the symbol table to find the address of the symbol's
3650 memory block and then code to read the value from that memory block.
3657 looks up the symbol @samp{foo} in the symbol table, gets the address
3658 associated with this symbol and then writes the value 1 into that
3665 looks up the symbol @samp{foo} in the symbol table, gets its address
3666 and then copies this address into the block of memory associated with
3667 the variable @samp{a}.
3669 Linker scripts symbol declarations, by contrast, create an entry in
3670 the symbol table but do not assign any memory to them. Thus they are
3671 an address without a value. So for example the linker script definition:
3677 creates an entry in the symbol table called @samp{foo} which holds
3678 the address of memory location 1000, but nothing special is stored at
3679 address 1000. This means that you cannot access the @emph{value} of a
3680 linker script defined symbol - it has no value - all you can do is
3681 access the @emph{address} of a linker script defined symbol.
3683 Hence when you are using a linker script defined symbol in source code
3684 you should always take the address of the symbol, and never attempt to
3685 use its value. For example suppose you want to copy the contents of a
3686 section of memory called .ROM into a section called .FLASH and the
3687 linker script contains these declarations:
3691 start_of_ROM = .ROM;
3692 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3693 start_of_FLASH = .FLASH;
3697 Then the C source code to perform the copy would be:
3701 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3703 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3707 Note the use of the @samp{&} operators. These are correct.
3710 @section SECTIONS Command
3712 The @code{SECTIONS} command tells the linker how to map input sections
3713 into output sections, and how to place the output sections in memory.
3715 The format of the @code{SECTIONS} command is:
3719 @var{sections-command}
3720 @var{sections-command}
3725 Each @var{sections-command} may of be one of the following:
3729 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3731 a symbol assignment (@pxref{Assignments})
3733 an output section description
3735 an overlay description
3738 The @code{ENTRY} command and symbol assignments are permitted inside the
3739 @code{SECTIONS} command for convenience in using the location counter in
3740 those commands. This can also make the linker script easier to
3741 understand because you can use those commands at meaningful points in
3742 the layout of the output file.
3744 Output section descriptions and overlay descriptions are described
3747 If you do not use a @code{SECTIONS} command in your linker script, the
3748 linker will place each input section into an identically named output
3749 section in the order that the sections are first encountered in the
3750 input files. If all input sections are present in the first file, for
3751 example, the order of sections in the output file will match the order
3752 in the first input file. The first section will be at address zero.
3755 * Output Section Description:: Output section description
3756 * Output Section Name:: Output section name
3757 * Output Section Address:: Output section address
3758 * Input Section:: Input section description
3759 * Output Section Data:: Output section data
3760 * Output Section Keywords:: Output section keywords
3761 * Output Section Discarding:: Output section discarding
3762 * Output Section Attributes:: Output section attributes
3763 * Overlay Description:: Overlay description
3766 @node Output Section Description
3767 @subsection Output Section Description
3768 The full description of an output section looks like this:
3771 @var{section} [@var{address}] [(@var{type})] :
3773 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
3774 [SUBALIGN(@var{subsection_align})]
3777 @var{output-section-command}
3778 @var{output-section-command}
3780 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3784 Most output sections do not use most of the optional section attributes.
3786 The whitespace around @var{section} is required, so that the section
3787 name is unambiguous. The colon and the curly braces are also required.
3788 The line breaks and other white space are optional.
3790 Each @var{output-section-command} may be one of the following:
3794 a symbol assignment (@pxref{Assignments})
3796 an input section description (@pxref{Input Section})
3798 data values to include directly (@pxref{Output Section Data})
3800 a special output section keyword (@pxref{Output Section Keywords})
3803 @node Output Section Name
3804 @subsection Output Section Name
3805 @cindex name, section
3806 @cindex section name
3807 The name of the output section is @var{section}. @var{section} must
3808 meet the constraints of your output format. In formats which only
3809 support a limited number of sections, such as @code{a.out}, the name
3810 must be one of the names supported by the format (@code{a.out}, for
3811 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3812 output format supports any number of sections, but with numbers and not
3813 names (as is the case for Oasys), the name should be supplied as a
3814 quoted numeric string. A section name may consist of any sequence of
3815 characters, but a name which contains any unusual characters such as
3816 commas must be quoted.
3818 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3821 @node Output Section Address
3822 @subsection Output Section Address
3823 @cindex address, section
3824 @cindex section address
3825 The @var{address} is an expression for the VMA (the virtual memory
3826 address) of the output section. This address is optional, but if it
3827 is provided then the output address will be set exactly as specified.
3829 If the output address is not specified then one will be chosen for the
3830 section, based on the heuristic below. This address will be adjusted
3831 to fit the alignment requirement of the output section. The
3832 alignment requirement is the strictest alignment of any input section
3833 contained within the output section.
3835 The output section address heuristic is as follows:
3839 If an output memory @var{region} is set for the section then it
3840 is added to this region and its address will be the next free address
3844 If the MEMORY command has been used to create a list of memory
3845 regions then the first region which has attributes compatible with the
3846 section is selected to contain it. The section's output address will
3847 be the next free address in that region; @ref{MEMORY}.
3850 If no memory regions were specified, or none match the section then
3851 the output address will be based on the current value of the location
3859 .text . : @{ *(.text) @}
3866 .text : @{ *(.text) @}
3870 are subtly different. The first will set the address of the
3871 @samp{.text} output section to the current value of the location
3872 counter. The second will set it to the current value of the location
3873 counter aligned to the strictest alignment of any of the @samp{.text}
3876 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3877 For example, if you want to align the section on a 0x10 byte boundary,
3878 so that the lowest four bits of the section address are zero, you could
3879 do something like this:
3881 .text ALIGN(0x10) : @{ *(.text) @}
3884 This works because @code{ALIGN} returns the current location counter
3885 aligned upward to the specified value.
3887 Specifying @var{address} for a section will change the value of the
3888 location counter, provided that the section is non-empty. (Empty
3889 sections are ignored).
3892 @subsection Input Section Description
3893 @cindex input sections
3894 @cindex mapping input sections to output sections
3895 The most common output section command is an input section description.
3897 The input section description is the most basic linker script operation.
3898 You use output sections to tell the linker how to lay out your program
3899 in memory. You use input section descriptions to tell the linker how to
3900 map the input files into your memory layout.
3903 * Input Section Basics:: Input section basics
3904 * Input Section Wildcards:: Input section wildcard patterns
3905 * Input Section Common:: Input section for common symbols
3906 * Input Section Keep:: Input section and garbage collection
3907 * Input Section Example:: Input section example
3910 @node Input Section Basics
3911 @subsubsection Input Section Basics
3912 @cindex input section basics
3913 An input section description consists of a file name optionally followed
3914 by a list of section names in parentheses.
3916 The file name and the section name may be wildcard patterns, which we
3917 describe further below (@pxref{Input Section Wildcards}).
3919 The most common input section description is to include all input
3920 sections with a particular name in the output section. For example, to
3921 include all input @samp{.text} sections, you would write:
3926 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3927 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3928 match all files except the ones specified in the EXCLUDE_FILE list. For
3931 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3933 will cause all .ctors sections from all files except @file{crtend.o} and
3934 @file{otherfile.o} to be included.
3936 There are two ways to include more than one section:
3942 The difference between these is the order in which the @samp{.text} and
3943 @samp{.rdata} input sections will appear in the output section. In the
3944 first example, they will be intermingled, appearing in the same order as
3945 they are found in the linker input. In the second example, all
3946 @samp{.text} input sections will appear first, followed by all
3947 @samp{.rdata} input sections.
3949 You can specify a file name to include sections from a particular file.
3950 You would do this if one or more of your files contain special data that
3951 needs to be at a particular location in memory. For example:
3956 To refine the sections that are included based on the section flags
3957 of an input section, INPUT_SECTION_FLAGS may be used.
3959 Here is a simple example for using Section header flags for ELF sections:
3964 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
3965 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
3970 In this example, the output section @samp{.text} will be comprised of any
3971 input section matching the name *(.text) whose section header flags
3972 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
3973 @samp{.text2} will be comprised of any input section matching the name *(.text)
3974 whose section header flag @code{SHF_WRITE} is clear.
3976 You can also specify files within archives by writing a pattern
3977 matching the archive, a colon, then the pattern matching the file,
3978 with no whitespace around the colon.
3982 matches file within archive
3984 matches the whole archive
3986 matches file but not one in an archive
3989 Either one or both of @samp{archive} and @samp{file} can contain shell
3990 wildcards. On DOS based file systems, the linker will assume that a
3991 single letter followed by a colon is a drive specifier, so
3992 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
3993 within an archive called @samp{c}. @samp{archive:file} filespecs may
3994 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
3995 other linker script contexts. For instance, you cannot extract a file
3996 from an archive by using @samp{archive:file} in an @code{INPUT}
3999 If you use a file name without a list of sections, then all sections in
4000 the input file will be included in the output section. This is not
4001 commonly done, but it may by useful on occasion. For example:
4006 When you use a file name which is not an @samp{archive:file} specifier
4007 and does not contain any wild card
4008 characters, the linker will first see if you also specified the file
4009 name on the linker command line or in an @code{INPUT} command. If you
4010 did not, the linker will attempt to open the file as an input file, as
4011 though it appeared on the command line. Note that this differs from an
4012 @code{INPUT} command, because the linker will not search for the file in
4013 the archive search path.
4015 @node Input Section Wildcards
4016 @subsubsection Input Section Wildcard Patterns
4017 @cindex input section wildcards
4018 @cindex wildcard file name patterns
4019 @cindex file name wildcard patterns
4020 @cindex section name wildcard patterns
4021 In an input section description, either the file name or the section
4022 name or both may be wildcard patterns.
4024 The file name of @samp{*} seen in many examples is a simple wildcard
4025 pattern for the file name.
4027 The wildcard patterns are like those used by the Unix shell.
4031 matches any number of characters
4033 matches any single character
4035 matches a single instance of any of the @var{chars}; the @samp{-}
4036 character may be used to specify a range of characters, as in
4037 @samp{[a-z]} to match any lower case letter
4039 quotes the following character
4042 When a file name is matched with a wildcard, the wildcard characters
4043 will not match a @samp{/} character (used to separate directory names on
4044 Unix). A pattern consisting of a single @samp{*} character is an
4045 exception; it will always match any file name, whether it contains a
4046 @samp{/} or not. In a section name, the wildcard characters will match
4047 a @samp{/} character.
4049 File name wildcard patterns only match files which are explicitly
4050 specified on the command line or in an @code{INPUT} command. The linker
4051 does not search directories to expand wildcards.
4053 If a file name matches more than one wildcard pattern, or if a file name
4054 appears explicitly and is also matched by a wildcard pattern, the linker
4055 will use the first match in the linker script. For example, this
4056 sequence of input section descriptions is probably in error, because the
4057 @file{data.o} rule will not be used:
4059 .data : @{ *(.data) @}
4060 .data1 : @{ data.o(.data) @}
4063 @cindex SORT_BY_NAME
4064 Normally, the linker will place files and sections matched by wildcards
4065 in the order in which they are seen during the link. You can change
4066 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4067 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4068 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4069 into ascending order by name before placing them in the output file.
4071 @cindex SORT_BY_ALIGNMENT
4072 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4073 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4074 descending order by alignment before placing them in the output file.
4075 Larger alignments are placed before smaller alignments in order to
4076 reduce the amount of padding necessary.
4078 @cindex SORT_BY_INIT_PRIORITY
4079 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4080 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4081 ascending order by numerical value of the GCC init_priority attribute
4082 encoded in the section name before placing them in the output file.
4085 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4087 When there are nested section sorting commands in linker script, there
4088 can be at most 1 level of nesting for section sorting commands.
4092 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4093 It will sort the input sections by name first, then by alignment if two
4094 sections have the same name.
4096 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4097 It will sort the input sections by alignment first, then by name if two
4098 sections have the same alignment.
4100 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4101 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4103 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4104 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4106 All other nested section sorting commands are invalid.
4109 When both command line section sorting option and linker script
4110 section sorting command are used, section sorting command always
4111 takes precedence over the command line option.
4113 If the section sorting command in linker script isn't nested, the
4114 command line option will make the section sorting command to be
4115 treated as nested sorting command.
4119 @code{SORT_BY_NAME} (wildcard section pattern ) with
4120 @option{--sort-sections alignment} is equivalent to
4121 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4123 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4124 @option{--sort-section name} is equivalent to
4125 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4128 If the section sorting command in linker script is nested, the
4129 command line option will be ignored.
4132 @code{SORT_NONE} disables section sorting by ignoring the command line
4133 section sorting option.
4135 If you ever get confused about where input sections are going, use the
4136 @samp{-M} linker option to generate a map file. The map file shows
4137 precisely how input sections are mapped to output sections.
4139 This example shows how wildcard patterns might be used to partition
4140 files. This linker script directs the linker to place all @samp{.text}
4141 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4142 The linker will place the @samp{.data} section from all files beginning
4143 with an upper case character in @samp{.DATA}; for all other files, the
4144 linker will place the @samp{.data} section in @samp{.data}.
4148 .text : @{ *(.text) @}
4149 .DATA : @{ [A-Z]*(.data) @}
4150 .data : @{ *(.data) @}
4151 .bss : @{ *(.bss) @}
4156 @node Input Section Common
4157 @subsubsection Input Section for Common Symbols
4158 @cindex common symbol placement
4159 @cindex uninitialized data placement
4160 A special notation is needed for common symbols, because in many object
4161 file formats common symbols do not have a particular input section. The
4162 linker treats common symbols as though they are in an input section
4163 named @samp{COMMON}.
4165 You may use file names with the @samp{COMMON} section just as with any
4166 other input sections. You can use this to place common symbols from a
4167 particular input file in one section while common symbols from other
4168 input files are placed in another section.
4170 In most cases, common symbols in input files will be placed in the
4171 @samp{.bss} section in the output file. For example:
4173 .bss @{ *(.bss) *(COMMON) @}
4176 @cindex scommon section
4177 @cindex small common symbols
4178 Some object file formats have more than one type of common symbol. For
4179 example, the MIPS ELF object file format distinguishes standard common
4180 symbols and small common symbols. In this case, the linker will use a
4181 different special section name for other types of common symbols. In
4182 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4183 symbols and @samp{.scommon} for small common symbols. This permits you
4184 to map the different types of common symbols into memory at different
4188 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4189 notation is now considered obsolete. It is equivalent to
4192 @node Input Section Keep
4193 @subsubsection Input Section and Garbage Collection
4195 @cindex garbage collection
4196 When link-time garbage collection is in use (@samp{--gc-sections}),
4197 it is often useful to mark sections that should not be eliminated.
4198 This is accomplished by surrounding an input section's wildcard entry
4199 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4200 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4202 @node Input Section Example
4203 @subsubsection Input Section Example
4204 The following example is a complete linker script. It tells the linker
4205 to read all of the sections from file @file{all.o} and place them at the
4206 start of output section @samp{outputa} which starts at location
4207 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4208 follows immediately, in the same output section. All of section
4209 @samp{.input2} from @file{foo.o} goes into output section
4210 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4211 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4212 files are written to output section @samp{outputc}.
4240 @node Output Section Data
4241 @subsection Output Section Data
4243 @cindex section data
4244 @cindex output section data
4245 @kindex BYTE(@var{expression})
4246 @kindex SHORT(@var{expression})
4247 @kindex LONG(@var{expression})
4248 @kindex QUAD(@var{expression})
4249 @kindex SQUAD(@var{expression})
4250 You can include explicit bytes of data in an output section by using
4251 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4252 an output section command. Each keyword is followed by an expression in
4253 parentheses providing the value to store (@pxref{Expressions}). The
4254 value of the expression is stored at the current value of the location
4257 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4258 store one, two, four, and eight bytes (respectively). After storing the
4259 bytes, the location counter is incremented by the number of bytes
4262 For example, this will store the byte 1 followed by the four byte value
4263 of the symbol @samp{addr}:
4269 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4270 same; they both store an 8 byte, or 64 bit, value. When both host and
4271 target are 32 bits, an expression is computed as 32 bits. In this case
4272 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4273 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4275 If the object file format of the output file has an explicit endianness,
4276 which is the normal case, the value will be stored in that endianness.
4277 When the object file format does not have an explicit endianness, as is
4278 true of, for example, S-records, the value will be stored in the
4279 endianness of the first input object file.
4281 Note---these commands only work inside a section description and not
4282 between them, so the following will produce an error from the linker:
4284 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4286 whereas this will work:
4288 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4291 @kindex FILL(@var{expression})
4292 @cindex holes, filling
4293 @cindex unspecified memory
4294 You may use the @code{FILL} command to set the fill pattern for the
4295 current section. It is followed by an expression in parentheses. Any
4296 otherwise unspecified regions of memory within the section (for example,
4297 gaps left due to the required alignment of input sections) are filled
4298 with the value of the expression, repeated as
4299 necessary. A @code{FILL} statement covers memory locations after the
4300 point at which it occurs in the section definition; by including more
4301 than one @code{FILL} statement, you can have different fill patterns in
4302 different parts of an output section.
4304 This example shows how to fill unspecified regions of memory with the
4310 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4311 section attribute, but it only affects the
4312 part of the section following the @code{FILL} command, rather than the
4313 entire section. If both are used, the @code{FILL} command takes
4314 precedence. @xref{Output Section Fill}, for details on the fill
4317 @node Output Section Keywords
4318 @subsection Output Section Keywords
4319 There are a couple of keywords which can appear as output section
4323 @kindex CREATE_OBJECT_SYMBOLS
4324 @cindex input filename symbols
4325 @cindex filename symbols
4326 @item CREATE_OBJECT_SYMBOLS
4327 The command tells the linker to create a symbol for each input file.
4328 The name of each symbol will be the name of the corresponding input
4329 file. The section of each symbol will be the output section in which
4330 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4332 This is conventional for the a.out object file format. It is not
4333 normally used for any other object file format.
4335 @kindex CONSTRUCTORS
4336 @cindex C++ constructors, arranging in link
4337 @cindex constructors, arranging in link
4339 When linking using the a.out object file format, the linker uses an
4340 unusual set construct to support C++ global constructors and
4341 destructors. When linking object file formats which do not support
4342 arbitrary sections, such as ECOFF and XCOFF, the linker will
4343 automatically recognize C++ global constructors and destructors by name.
4344 For these object file formats, the @code{CONSTRUCTORS} command tells the
4345 linker to place constructor information in the output section where the
4346 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4347 ignored for other object file formats.
4349 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4350 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4351 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4352 the start and end of the global destructors. The
4353 first word in the list is the number of entries, followed by the address
4354 of each constructor or destructor, followed by a zero word. The
4355 compiler must arrange to actually run the code. For these object file
4356 formats @sc{gnu} C++ normally calls constructors from a subroutine
4357 @code{__main}; a call to @code{__main} is automatically inserted into
4358 the startup code for @code{main}. @sc{gnu} C++ normally runs
4359 destructors either by using @code{atexit}, or directly from the function
4362 For object file formats such as @code{COFF} or @code{ELF} which support
4363 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4364 addresses of global constructors and destructors into the @code{.ctors}
4365 and @code{.dtors} sections. Placing the following sequence into your
4366 linker script will build the sort of table which the @sc{gnu} C++
4367 runtime code expects to see.
4371 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4376 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4382 If you are using the @sc{gnu} C++ support for initialization priority,
4383 which provides some control over the order in which global constructors
4384 are run, you must sort the constructors at link time to ensure that they
4385 are executed in the correct order. When using the @code{CONSTRUCTORS}
4386 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4387 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4388 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4391 Normally the compiler and linker will handle these issues automatically,
4392 and you will not need to concern yourself with them. However, you may
4393 need to consider this if you are using C++ and writing your own linker
4398 @node Output Section Discarding
4399 @subsection Output Section Discarding
4400 @cindex discarding sections
4401 @cindex sections, discarding
4402 @cindex removing sections
4403 The linker will not create output sections with no contents. This is
4404 for convenience when referring to input sections that may or may not
4405 be present in any of the input files. For example:
4407 .foo : @{ *(.foo) @}
4410 will only create a @samp{.foo} section in the output file if there is a
4411 @samp{.foo} section in at least one input file, and if the input
4412 sections are not all empty. Other link script directives that allocate
4413 space in an output section will also create the output section.
4415 The linker will ignore address assignments (@pxref{Output Section Address})
4416 on discarded output sections, except when the linker script defines
4417 symbols in the output section. In that case the linker will obey
4418 the address assignments, possibly advancing dot even though the
4419 section is discarded.
4422 The special output section name @samp{/DISCARD/} may be used to discard
4423 input sections. Any input sections which are assigned to an output
4424 section named @samp{/DISCARD/} are not included in the output file.
4426 @node Output Section Attributes
4427 @subsection Output Section Attributes
4428 @cindex output section attributes
4429 We showed above that the full description of an output section looked
4434 @var{section} [@var{address}] [(@var{type})] :
4436 [ALIGN(@var{section_align})]
4437 [SUBALIGN(@var{subsection_align})]
4440 @var{output-section-command}
4441 @var{output-section-command}
4443 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4447 We've already described @var{section}, @var{address}, and
4448 @var{output-section-command}. In this section we will describe the
4449 remaining section attributes.
4452 * Output Section Type:: Output section type
4453 * Output Section LMA:: Output section LMA
4454 * Forced Output Alignment:: Forced Output Alignment
4455 * Forced Input Alignment:: Forced Input Alignment
4456 * Output Section Constraint:: Output section constraint
4457 * Output Section Region:: Output section region
4458 * Output Section Phdr:: Output section phdr
4459 * Output Section Fill:: Output section fill
4462 @node Output Section Type
4463 @subsubsection Output Section Type
4464 Each output section may have a type. The type is a keyword in
4465 parentheses. The following types are defined:
4469 The section should be marked as not loadable, so that it will not be
4470 loaded into memory when the program is run.
4475 These type names are supported for backward compatibility, and are
4476 rarely used. They all have the same effect: the section should be
4477 marked as not allocatable, so that no memory is allocated for the
4478 section when the program is run.
4482 @cindex prevent unnecessary loading
4483 @cindex loading, preventing
4484 The linker normally sets the attributes of an output section based on
4485 the input sections which map into it. You can override this by using
4486 the section type. For example, in the script sample below, the
4487 @samp{ROM} section is addressed at memory location @samp{0} and does not
4488 need to be loaded when the program is run.
4492 ROM 0 (NOLOAD) : @{ @dots{} @}
4498 @node Output Section LMA
4499 @subsubsection Output Section LMA
4500 @kindex AT>@var{lma_region}
4501 @kindex AT(@var{lma})
4502 @cindex load address
4503 @cindex section load address
4504 Every section has a virtual address (VMA) and a load address (LMA); see
4505 @ref{Basic Script Concepts}. The virtual address is specified by the
4506 @pxref{Output Section Address} described earlier. The load address is
4507 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
4508 address is optional.
4510 The @code{AT} keyword takes an expression as an argument. This
4511 specifies the exact load address of the section. The @code{AT>} keyword
4512 takes the name of a memory region as an argument. @xref{MEMORY}. The
4513 load address of the section is set to the next free address in the
4514 region, aligned to the section's alignment requirements.
4516 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4517 section, the linker will use the following heuristic to determine the
4522 If the section has a specific VMA address, then this is used as
4523 the LMA address as well.
4526 If the section is not allocatable then its LMA is set to its VMA.
4529 Otherwise if a memory region can be found that is compatible
4530 with the current section, and this region contains at least one
4531 section, then the LMA is set so the difference between the
4532 VMA and LMA is the same as the difference between the VMA and LMA of
4533 the last section in the located region.
4536 If no memory regions have been declared then a default region
4537 that covers the entire address space is used in the previous step.
4540 If no suitable region could be found, or there was no previous
4541 section then the LMA is set equal to the VMA.
4544 @cindex ROM initialized data
4545 @cindex initialized data in ROM
4546 This feature is designed to make it easy to build a ROM image. For
4547 example, the following linker script creates three output sections: one
4548 called @samp{.text}, which starts at @code{0x1000}, one called
4549 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
4550 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
4551 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
4552 defined with the value @code{0x2000}, which shows that the location
4553 counter holds the VMA value, not the LMA value.
4559 .text 0x1000 : @{ *(.text) _etext = . ; @}
4561 AT ( ADDR (.text) + SIZEOF (.text) )
4562 @{ _data = . ; *(.data); _edata = . ; @}
4564 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
4569 The run-time initialization code for use with a program generated with
4570 this linker script would include something like the following, to copy
4571 the initialized data from the ROM image to its runtime address. Notice
4572 how this code takes advantage of the symbols defined by the linker
4577 extern char _etext, _data, _edata, _bstart, _bend;
4578 char *src = &_etext;
4581 /* ROM has data at end of text; copy it. */
4582 while (dst < &_edata)
4586 for (dst = &_bstart; dst< &_bend; dst++)
4591 @node Forced Output Alignment
4592 @subsubsection Forced Output Alignment
4593 @kindex ALIGN(@var{section_align})
4594 @cindex forcing output section alignment
4595 @cindex output section alignment
4596 You can increase an output section's alignment by using ALIGN. As an
4597 alternative you can force the output section alignment to the maximum alignment
4598 of all its input sections with ALIGN_WITH_INPUT. The alignment forced by
4599 ALIGN_WITH_INPUT is used even in case the load and virtual memory regions are
4602 @node Forced Input Alignment
4603 @subsubsection Forced Input Alignment
4604 @kindex SUBALIGN(@var{subsection_align})
4605 @cindex forcing input section alignment
4606 @cindex input section alignment
4607 You can force input section alignment within an output section by using
4608 SUBALIGN. The value specified overrides any alignment given by input
4609 sections, whether larger or smaller.
4611 @node Output Section Constraint
4612 @subsubsection Output Section Constraint
4615 @cindex constraints on output sections
4616 You can specify that an output section should only be created if all
4617 of its input sections are read-only or all of its input sections are
4618 read-write by using the keyword @code{ONLY_IF_RO} and
4619 @code{ONLY_IF_RW} respectively.
4621 @node Output Section Region
4622 @subsubsection Output Section Region
4623 @kindex >@var{region}
4624 @cindex section, assigning to memory region
4625 @cindex memory regions and sections
4626 You can assign a section to a previously defined region of memory by
4627 using @samp{>@var{region}}. @xref{MEMORY}.
4629 Here is a simple example:
4632 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4633 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4637 @node Output Section Phdr
4638 @subsubsection Output Section Phdr
4640 @cindex section, assigning to program header
4641 @cindex program headers and sections
4642 You can assign a section to a previously defined program segment by
4643 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4644 one or more segments, then all subsequent allocated sections will be
4645 assigned to those segments as well, unless they use an explicitly
4646 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4647 linker to not put the section in any segment at all.
4649 Here is a simple example:
4652 PHDRS @{ text PT_LOAD ; @}
4653 SECTIONS @{ .text : @{ *(.text) @} :text @}
4657 @node Output Section Fill
4658 @subsubsection Output Section Fill
4659 @kindex =@var{fillexp}
4660 @cindex section fill pattern
4661 @cindex fill pattern, entire section
4662 You can set the fill pattern for an entire section by using
4663 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4664 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4665 within the output section (for example, gaps left due to the required
4666 alignment of input sections) will be filled with the value, repeated as
4667 necessary. If the fill expression is a simple hex number, ie. a string
4668 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4669 an arbitrarily long sequence of hex digits can be used to specify the
4670 fill pattern; Leading zeros become part of the pattern too. For all
4671 other cases, including extra parentheses or a unary @code{+}, the fill
4672 pattern is the four least significant bytes of the value of the
4673 expression. In all cases, the number is big-endian.
4675 You can also change the fill value with a @code{FILL} command in the
4676 output section commands; (@pxref{Output Section Data}).
4678 Here is a simple example:
4681 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4685 @node Overlay Description
4686 @subsection Overlay Description
4689 An overlay description provides an easy way to describe sections which
4690 are to be loaded as part of a single memory image but are to be run at
4691 the same memory address. At run time, some sort of overlay manager will
4692 copy the overlaid sections in and out of the runtime memory address as
4693 required, perhaps by simply manipulating addressing bits. This approach
4694 can be useful, for example, when a certain region of memory is faster
4697 Overlays are described using the @code{OVERLAY} command. The
4698 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4699 output section description. The full syntax of the @code{OVERLAY}
4700 command is as follows:
4703 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4707 @var{output-section-command}
4708 @var{output-section-command}
4710 @} [:@var{phdr}@dots{}] [=@var{fill}]
4713 @var{output-section-command}
4714 @var{output-section-command}
4716 @} [:@var{phdr}@dots{}] [=@var{fill}]
4718 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4722 Everything is optional except @code{OVERLAY} (a keyword), and each
4723 section must have a name (@var{secname1} and @var{secname2} above). The
4724 section definitions within the @code{OVERLAY} construct are identical to
4725 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
4726 except that no addresses and no memory regions may be defined for
4727 sections within an @code{OVERLAY}.
4729 The sections are all defined with the same starting address. The load
4730 addresses of the sections are arranged such that they are consecutive in
4731 memory starting at the load address used for the @code{OVERLAY} as a
4732 whole (as with normal section definitions, the load address is optional,
4733 and defaults to the start address; the start address is also optional,
4734 and defaults to the current value of the location counter).
4736 If the @code{NOCROSSREFS} keyword is used, and there are any
4737 references among the sections, the linker will report an error. Since
4738 the sections all run at the same address, it normally does not make
4739 sense for one section to refer directly to another.
4740 @xref{Miscellaneous Commands, NOCROSSREFS}.
4742 For each section within the @code{OVERLAY}, the linker automatically
4743 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4744 defined as the starting load address of the section. The symbol
4745 @code{__load_stop_@var{secname}} is defined as the final load address of
4746 the section. Any characters within @var{secname} which are not legal
4747 within C identifiers are removed. C (or assembler) code may use these
4748 symbols to move the overlaid sections around as necessary.
4750 At the end of the overlay, the value of the location counter is set to
4751 the start address of the overlay plus the size of the largest section.
4753 Here is an example. Remember that this would appear inside a
4754 @code{SECTIONS} construct.
4757 OVERLAY 0x1000 : AT (0x4000)
4759 .text0 @{ o1/*.o(.text) @}
4760 .text1 @{ o2/*.o(.text) @}
4765 This will define both @samp{.text0} and @samp{.text1} to start at
4766 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4767 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4768 following symbols will be defined if referenced: @code{__load_start_text0},
4769 @code{__load_stop_text0}, @code{__load_start_text1},
4770 @code{__load_stop_text1}.
4772 C code to copy overlay @code{.text1} into the overlay area might look
4777 extern char __load_start_text1, __load_stop_text1;
4778 memcpy ((char *) 0x1000, &__load_start_text1,
4779 &__load_stop_text1 - &__load_start_text1);
4783 Note that the @code{OVERLAY} command is just syntactic sugar, since
4784 everything it does can be done using the more basic commands. The above
4785 example could have been written identically as follows.
4789 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4790 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4791 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4792 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4793 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4794 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4795 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4800 @section MEMORY Command
4802 @cindex memory regions
4803 @cindex regions of memory
4804 @cindex allocating memory
4805 @cindex discontinuous memory
4806 The linker's default configuration permits allocation of all available
4807 memory. You can override this by using the @code{MEMORY} command.
4809 The @code{MEMORY} command describes the location and size of blocks of
4810 memory in the target. You can use it to describe which memory regions
4811 may be used by the linker, and which memory regions it must avoid. You
4812 can then assign sections to particular memory regions. The linker will
4813 set section addresses based on the memory regions, and will warn about
4814 regions that become too full. The linker will not shuffle sections
4815 around to fit into the available regions.
4817 A linker script may contain at most one use of the @code{MEMORY}
4818 command. However, you can define as many blocks of memory within it as
4819 you wish. The syntax is:
4824 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4830 The @var{name} is a name used in the linker script to refer to the
4831 region. The region name has no meaning outside of the linker script.
4832 Region names are stored in a separate name space, and will not conflict
4833 with symbol names, file names, or section names. Each memory region
4834 must have a distinct name within the @code{MEMORY} command. However you can
4835 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
4838 @cindex memory region attributes
4839 The @var{attr} string is an optional list of attributes that specify
4840 whether to use a particular memory region for an input section which is
4841 not explicitly mapped in the linker script. As described in
4842 @ref{SECTIONS}, if you do not specify an output section for some input
4843 section, the linker will create an output section with the same name as
4844 the input section. If you define region attributes, the linker will use
4845 them to select the memory region for the output section that it creates.
4847 The @var{attr} string must consist only of the following characters:
4862 Invert the sense of any of the attributes that follow
4865 If a unmapped section matches any of the listed attributes other than
4866 @samp{!}, it will be placed in the memory region. The @samp{!}
4867 attribute reverses this test, so that an unmapped section will be placed
4868 in the memory region only if it does not match any of the listed
4874 The @var{origin} is an numerical expression for the start address of
4875 the memory region. The expression must evaluate to a constant and it
4876 cannot involve any symbols. The keyword @code{ORIGIN} may be
4877 abbreviated to @code{org} or @code{o} (but not, for example,
4883 The @var{len} is an expression for the size in bytes of the memory
4884 region. As with the @var{origin} expression, the expression must
4885 be numerical only and must evaluate to a constant. The keyword
4886 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4888 In the following example, we specify that there are two memory regions
4889 available for allocation: one starting at @samp{0} for 256 kilobytes,
4890 and the other starting at @samp{0x40000000} for four megabytes. The
4891 linker will place into the @samp{rom} memory region every section which
4892 is not explicitly mapped into a memory region, and is either read-only
4893 or executable. The linker will place other sections which are not
4894 explicitly mapped into a memory region into the @samp{ram} memory
4901 rom (rx) : ORIGIN = 0, LENGTH = 256K
4902 ram (!rx) : org = 0x40000000, l = 4M
4907 Once you define a memory region, you can direct the linker to place
4908 specific output sections into that memory region by using the
4909 @samp{>@var{region}} output section attribute. For example, if you have
4910 a memory region named @samp{mem}, you would use @samp{>mem} in the
4911 output section definition. @xref{Output Section Region}. If no address
4912 was specified for the output section, the linker will set the address to
4913 the next available address within the memory region. If the combined
4914 output sections directed to a memory region are too large for the
4915 region, the linker will issue an error message.
4917 It is possible to access the origin and length of a memory in an
4918 expression via the @code{ORIGIN(@var{memory})} and
4919 @code{LENGTH(@var{memory})} functions:
4923 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4928 @section PHDRS Command
4930 @cindex program headers
4931 @cindex ELF program headers
4932 @cindex program segments
4933 @cindex segments, ELF
4934 The ELF object file format uses @dfn{program headers}, also knows as
4935 @dfn{segments}. The program headers describe how the program should be
4936 loaded into memory. You can print them out by using the @code{objdump}
4937 program with the @samp{-p} option.
4939 When you run an ELF program on a native ELF system, the system loader
4940 reads the program headers in order to figure out how to load the
4941 program. This will only work if the program headers are set correctly.
4942 This manual does not describe the details of how the system loader
4943 interprets program headers; for more information, see the ELF ABI.
4945 The linker will create reasonable program headers by default. However,
4946 in some cases, you may need to specify the program headers more
4947 precisely. You may use the @code{PHDRS} command for this purpose. When
4948 the linker sees the @code{PHDRS} command in the linker script, it will
4949 not create any program headers other than the ones specified.
4951 The linker only pays attention to the @code{PHDRS} command when
4952 generating an ELF output file. In other cases, the linker will simply
4953 ignore @code{PHDRS}.
4955 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4956 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4962 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4963 [ FLAGS ( @var{flags} ) ] ;
4968 The @var{name} is used only for reference in the @code{SECTIONS} command
4969 of the linker script. It is not put into the output file. Program
4970 header names are stored in a separate name space, and will not conflict
4971 with symbol names, file names, or section names. Each program header
4972 must have a distinct name. The headers are processed in order and it
4973 is usual for them to map to sections in ascending load address order.
4975 Certain program header types describe segments of memory which the
4976 system loader will load from the file. In the linker script, you
4977 specify the contents of these segments by placing allocatable output
4978 sections in the segments. You use the @samp{:@var{phdr}} output section
4979 attribute to place a section in a particular segment. @xref{Output
4982 It is normal to put certain sections in more than one segment. This
4983 merely implies that one segment of memory contains another. You may
4984 repeat @samp{:@var{phdr}}, using it once for each segment which should
4985 contain the section.
4987 If you place a section in one or more segments using @samp{:@var{phdr}},
4988 then the linker will place all subsequent allocatable sections which do
4989 not specify @samp{:@var{phdr}} in the same segments. This is for
4990 convenience, since generally a whole set of contiguous sections will be
4991 placed in a single segment. You can use @code{:NONE} to override the
4992 default segment and tell the linker to not put the section in any
4997 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
4998 the program header type to further describe the contents of the segment.
4999 The @code{FILEHDR} keyword means that the segment should include the ELF
5000 file header. The @code{PHDRS} keyword means that the segment should
5001 include the ELF program headers themselves. If applied to a loadable
5002 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5005 The @var{type} may be one of the following. The numbers indicate the
5006 value of the keyword.
5009 @item @code{PT_NULL} (0)
5010 Indicates an unused program header.
5012 @item @code{PT_LOAD} (1)
5013 Indicates that this program header describes a segment to be loaded from
5016 @item @code{PT_DYNAMIC} (2)
5017 Indicates a segment where dynamic linking information can be found.
5019 @item @code{PT_INTERP} (3)
5020 Indicates a segment where the name of the program interpreter may be
5023 @item @code{PT_NOTE} (4)
5024 Indicates a segment holding note information.
5026 @item @code{PT_SHLIB} (5)
5027 A reserved program header type, defined but not specified by the ELF
5030 @item @code{PT_PHDR} (6)
5031 Indicates a segment where the program headers may be found.
5033 @item @var{expression}
5034 An expression giving the numeric type of the program header. This may
5035 be used for types not defined above.
5038 You can specify that a segment should be loaded at a particular address
5039 in memory by using an @code{AT} expression. This is identical to the
5040 @code{AT} command used as an output section attribute (@pxref{Output
5041 Section LMA}). The @code{AT} command for a program header overrides the
5042 output section attribute.
5044 The linker will normally set the segment flags based on the sections
5045 which comprise the segment. You may use the @code{FLAGS} keyword to
5046 explicitly specify the segment flags. The value of @var{flags} must be
5047 an integer. It is used to set the @code{p_flags} field of the program
5050 Here is an example of @code{PHDRS}. This shows a typical set of program
5051 headers used on a native ELF system.
5057 headers PT_PHDR PHDRS ;
5059 text PT_LOAD FILEHDR PHDRS ;
5061 dynamic PT_DYNAMIC ;
5067 .interp : @{ *(.interp) @} :text :interp
5068 .text : @{ *(.text) @} :text
5069 .rodata : @{ *(.rodata) @} /* defaults to :text */
5071 . = . + 0x1000; /* move to a new page in memory */
5072 .data : @{ *(.data) @} :data
5073 .dynamic : @{ *(.dynamic) @} :data :dynamic
5080 @section VERSION Command
5081 @kindex VERSION @{script text@}
5082 @cindex symbol versions
5083 @cindex version script
5084 @cindex versions of symbols
5085 The linker supports symbol versions when using ELF. Symbol versions are
5086 only useful when using shared libraries. The dynamic linker can use
5087 symbol versions to select a specific version of a function when it runs
5088 a program that may have been linked against an earlier version of the
5091 You can include a version script directly in the main linker script, or
5092 you can supply the version script as an implicit linker script. You can
5093 also use the @samp{--version-script} linker option.
5095 The syntax of the @code{VERSION} command is simply
5097 VERSION @{ version-script-commands @}
5100 The format of the version script commands is identical to that used by
5101 Sun's linker in Solaris 2.5. The version script defines a tree of
5102 version nodes. You specify the node names and interdependencies in the
5103 version script. You can specify which symbols are bound to which
5104 version nodes, and you can reduce a specified set of symbols to local
5105 scope so that they are not globally visible outside of the shared
5108 The easiest way to demonstrate the version script language is with a few
5134 This example version script defines three version nodes. The first
5135 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5136 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5137 a number of symbols to local scope so that they are not visible outside
5138 of the shared library; this is done using wildcard patterns, so that any
5139 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5140 is matched. The wildcard patterns available are the same as those used
5141 in the shell when matching filenames (also known as ``globbing'').
5142 However, if you specify the symbol name inside double quotes, then the
5143 name is treated as literal, rather than as a glob pattern.
5145 Next, the version script defines node @samp{VERS_1.2}. This node
5146 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5147 to the version node @samp{VERS_1.2}.
5149 Finally, the version script defines node @samp{VERS_2.0}. This node
5150 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5151 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5153 When the linker finds a symbol defined in a library which is not
5154 specifically bound to a version node, it will effectively bind it to an
5155 unspecified base version of the library. You can bind all otherwise
5156 unspecified symbols to a given version node by using @samp{global: *;}
5157 somewhere in the version script. Note that it's slightly crazy to use
5158 wildcards in a global spec except on the last version node. Global
5159 wildcards elsewhere run the risk of accidentally adding symbols to the
5160 set exported for an old version. That's wrong since older versions
5161 ought to have a fixed set of symbols.
5163 The names of the version nodes have no specific meaning other than what
5164 they might suggest to the person reading them. The @samp{2.0} version
5165 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5166 However, this would be a confusing way to write a version script.
5168 Node name can be omitted, provided it is the only version node
5169 in the version script. Such version script doesn't assign any versions to
5170 symbols, only selects which symbols will be globally visible out and which
5174 @{ global: foo; bar; local: *; @};
5177 When you link an application against a shared library that has versioned
5178 symbols, the application itself knows which version of each symbol it
5179 requires, and it also knows which version nodes it needs from each
5180 shared library it is linked against. Thus at runtime, the dynamic
5181 loader can make a quick check to make sure that the libraries you have
5182 linked against do in fact supply all of the version nodes that the
5183 application will need to resolve all of the dynamic symbols. In this
5184 way it is possible for the dynamic linker to know with certainty that
5185 all external symbols that it needs will be resolvable without having to
5186 search for each symbol reference.
5188 The symbol versioning is in effect a much more sophisticated way of
5189 doing minor version checking that SunOS does. The fundamental problem
5190 that is being addressed here is that typically references to external
5191 functions are bound on an as-needed basis, and are not all bound when
5192 the application starts up. If a shared library is out of date, a
5193 required interface may be missing; when the application tries to use
5194 that interface, it may suddenly and unexpectedly fail. With symbol
5195 versioning, the user will get a warning when they start their program if
5196 the libraries being used with the application are too old.
5198 There are several GNU extensions to Sun's versioning approach. The
5199 first of these is the ability to bind a symbol to a version node in the
5200 source file where the symbol is defined instead of in the versioning
5201 script. This was done mainly to reduce the burden on the library
5202 maintainer. You can do this by putting something like:
5204 __asm__(".symver original_foo,foo@@VERS_1.1");
5207 in the C source file. This renames the function @samp{original_foo} to
5208 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5209 The @samp{local:} directive can be used to prevent the symbol
5210 @samp{original_foo} from being exported. A @samp{.symver} directive
5211 takes precedence over a version script.
5213 The second GNU extension is to allow multiple versions of the same
5214 function to appear in a given shared library. In this way you can make
5215 an incompatible change to an interface without increasing the major
5216 version number of the shared library, while still allowing applications
5217 linked against the old interface to continue to function.
5219 To do this, you must use multiple @samp{.symver} directives in the
5220 source file. Here is an example:
5223 __asm__(".symver original_foo,foo@@");
5224 __asm__(".symver old_foo,foo@@VERS_1.1");
5225 __asm__(".symver old_foo1,foo@@VERS_1.2");
5226 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5229 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5230 unspecified base version of the symbol. The source file that contains this
5231 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5232 @samp{old_foo1}, and @samp{new_foo}.
5234 When you have multiple definitions of a given symbol, there needs to be
5235 some way to specify a default version to which external references to
5236 this symbol will be bound. You can do this with the
5237 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5238 declare one version of a symbol as the default in this manner; otherwise
5239 you would effectively have multiple definitions of the same symbol.
5241 If you wish to bind a reference to a specific version of the symbol
5242 within the shared library, you can use the aliases of convenience
5243 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5244 specifically bind to an external version of the function in question.
5246 You can also specify the language in the version script:
5249 VERSION extern "lang" @{ version-script-commands @}
5252 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5253 The linker will iterate over the list of symbols at the link time and
5254 demangle them according to @samp{lang} before matching them to the
5255 patterns specified in @samp{version-script-commands}. The default
5256 @samp{lang} is @samp{C}.
5258 Demangled names may contains spaces and other special characters. As
5259 described above, you can use a glob pattern to match demangled names,
5260 or you can use a double-quoted string to match the string exactly. In
5261 the latter case, be aware that minor differences (such as differing
5262 whitespace) between the version script and the demangler output will
5263 cause a mismatch. As the exact string generated by the demangler
5264 might change in the future, even if the mangled name does not, you
5265 should check that all of your version directives are behaving as you
5266 expect when you upgrade.
5269 @section Expressions in Linker Scripts
5272 The syntax for expressions in the linker script language is identical to
5273 that of C expressions. All expressions are evaluated as integers. All
5274 expressions are evaluated in the same size, which is 32 bits if both the
5275 host and target are 32 bits, and is otherwise 64 bits.
5277 You can use and set symbol values in expressions.
5279 The linker defines several special purpose builtin functions for use in
5283 * Constants:: Constants
5284 * Symbolic Constants:: Symbolic constants
5285 * Symbols:: Symbol Names
5286 * Orphan Sections:: Orphan Sections
5287 * Location Counter:: The Location Counter
5288 * Operators:: Operators
5289 * Evaluation:: Evaluation
5290 * Expression Section:: The Section of an Expression
5291 * Builtin Functions:: Builtin Functions
5295 @subsection Constants
5296 @cindex integer notation
5297 @cindex constants in linker scripts
5298 All constants are integers.
5300 As in C, the linker considers an integer beginning with @samp{0} to be
5301 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5302 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5303 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5304 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5305 value without a prefix or a suffix is considered to be decimal.
5307 @cindex scaled integers
5308 @cindex K and M integer suffixes
5309 @cindex M and K integer suffixes
5310 @cindex suffixes for integers
5311 @cindex integer suffixes
5312 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5316 @c END TEXI2ROFF-KILL
5317 @code{1024} or @code{1024*1024}
5321 ${\rm 1024}$ or ${\rm 1024}^2$
5323 @c END TEXI2ROFF-KILL
5324 respectively. For example, the following
5325 all refer to the same quantity:
5334 Note - the @code{K} and @code{M} suffixes cannot be used in
5335 conjunction with the base suffixes mentioned above.
5337 @node Symbolic Constants
5338 @subsection Symbolic Constants
5339 @cindex symbolic constants
5341 It is possible to refer to target specific constants via the use of
5342 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5347 The target's maximum page size.
5349 @item COMMONPAGESIZE
5350 @kindex COMMONPAGESIZE
5351 The target's default page size.
5357 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5360 will create a text section aligned to the largest page boundary
5361 supported by the target.
5364 @subsection Symbol Names
5365 @cindex symbol names
5367 @cindex quoted symbol names
5369 Unless quoted, symbol names start with a letter, underscore, or period
5370 and may include letters, digits, underscores, periods, and hyphens.
5371 Unquoted symbol names must not conflict with any keywords. You can
5372 specify a symbol which contains odd characters or has the same name as a
5373 keyword by surrounding the symbol name in double quotes:
5376 "with a space" = "also with a space" + 10;
5379 Since symbols can contain many non-alphabetic characters, it is safest
5380 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5381 whereas @samp{A - B} is an expression involving subtraction.
5383 @node Orphan Sections
5384 @subsection Orphan Sections
5386 Orphan sections are sections present in the input files which
5387 are not explicitly placed into the output file by the linker
5388 script. The linker will still copy these sections into the
5389 output file, but it has to guess as to where they should be
5390 placed. The linker uses a simple heuristic to do this. It
5391 attempts to place orphan sections after non-orphan sections of the
5392 same attribute, such as code vs data, loadable vs non-loadable, etc.
5393 If there is not enough room to do this then it places
5394 at the end of the file.
5396 For ELF targets, the attribute of the section includes section type as
5397 well as section flag.
5399 If an orphaned section's name is representable as a C identifier then
5400 the linker will automatically @pxref{PROVIDE} two symbols:
5401 __start_SECNAME and __stop_SECNAME, where SECNAME is the name of the
5402 section. These indicate the start address and end address of the
5403 orphaned section respectively. Note: most section names are not
5404 representable as C identifiers because they contain a @samp{.}
5407 @node Location Counter
5408 @subsection The Location Counter
5411 @cindex location counter
5412 @cindex current output location
5413 The special linker variable @dfn{dot} @samp{.} always contains the
5414 current output location counter. Since the @code{.} always refers to a
5415 location in an output section, it may only appear in an expression
5416 within a @code{SECTIONS} command. The @code{.} symbol may appear
5417 anywhere that an ordinary symbol is allowed in an expression.
5420 Assigning a value to @code{.} will cause the location counter to be
5421 moved. This may be used to create holes in the output section. The
5422 location counter may not be moved backwards inside an output section,
5423 and may not be moved backwards outside of an output section if so
5424 doing creates areas with overlapping LMAs.
5440 In the previous example, the @samp{.text} section from @file{file1} is
5441 located at the beginning of the output section @samp{output}. It is
5442 followed by a 1000 byte gap. Then the @samp{.text} section from
5443 @file{file2} appears, also with a 1000 byte gap following before the
5444 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5445 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5447 @cindex dot inside sections
5448 Note: @code{.} actually refers to the byte offset from the start of the
5449 current containing object. Normally this is the @code{SECTIONS}
5450 statement, whose start address is 0, hence @code{.} can be used as an
5451 absolute address. If @code{.} is used inside a section description
5452 however, it refers to the byte offset from the start of that section,
5453 not an absolute address. Thus in a script like this:
5471 The @samp{.text} section will be assigned a starting address of 0x100
5472 and a size of exactly 0x200 bytes, even if there is not enough data in
5473 the @samp{.text} input sections to fill this area. (If there is too
5474 much data, an error will be produced because this would be an attempt to
5475 move @code{.} backwards). The @samp{.data} section will start at 0x500
5476 and it will have an extra 0x600 bytes worth of space after the end of
5477 the values from the @samp{.data} input sections and before the end of
5478 the @samp{.data} output section itself.
5480 @cindex dot outside sections
5481 Setting symbols to the value of the location counter outside of an
5482 output section statement can result in unexpected values if the linker
5483 needs to place orphan sections. For example, given the following:
5489 .text: @{ *(.text) @}
5493 .data: @{ *(.data) @}
5498 If the linker needs to place some input section, e.g. @code{.rodata},
5499 not mentioned in the script, it might choose to place that section
5500 between @code{.text} and @code{.data}. You might think the linker
5501 should place @code{.rodata} on the blank line in the above script, but
5502 blank lines are of no particular significance to the linker. As well,
5503 the linker doesn't associate the above symbol names with their
5504 sections. Instead, it assumes that all assignments or other
5505 statements belong to the previous output section, except for the
5506 special case of an assignment to @code{.}. I.e., the linker will
5507 place the orphan @code{.rodata} section as if the script was written
5514 .text: @{ *(.text) @}
5518 .rodata: @{ *(.rodata) @}
5519 .data: @{ *(.data) @}
5524 This may or may not be the script author's intention for the value of
5525 @code{start_of_data}. One way to influence the orphan section
5526 placement is to assign the location counter to itself, as the linker
5527 assumes that an assignment to @code{.} is setting the start address of
5528 a following output section and thus should be grouped with that
5529 section. So you could write:
5535 .text: @{ *(.text) @}
5540 .data: @{ *(.data) @}
5545 Now, the orphan @code{.rodata} section will be placed between
5546 @code{end_of_text} and @code{start_of_data}.
5550 @subsection Operators
5551 @cindex operators for arithmetic
5552 @cindex arithmetic operators
5553 @cindex precedence in expressions
5554 The linker recognizes the standard C set of arithmetic operators, with
5555 the standard bindings and precedence levels:
5558 @c END TEXI2ROFF-KILL
5560 precedence associativity Operators Notes
5566 5 left == != > < <= >=
5572 11 right &= += -= *= /= (2)
5576 (1) Prefix operators
5577 (2) @xref{Assignments}.
5581 \vskip \baselineskip
5582 %"lispnarrowing" is the extra indent used generally for smallexample
5583 \hskip\lispnarrowing\vbox{\offinterlineskip
5586 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
5587 height2pt&\omit&&\omit&&\omit&\cr
5588 &Precedence&& Associativity &&{\rm Operators}&\cr
5589 height2pt&\omit&&\omit&&\omit&\cr
5591 height2pt&\omit&&\omit&&\omit&\cr
5593 % '176 is tilde, '~' in tt font
5594 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
5595 &2&&left&&* / \%&\cr
5598 &5&&left&&== != > < <= >=&\cr
5601 &8&&left&&{\&\&}&\cr
5604 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
5606 height2pt&\omit&&\omit&&\omit&\cr}
5611 @obeylines@parskip=0pt@parindent=0pt
5612 @dag@quad Prefix operators.
5613 @ddag@quad @xref{Assignments}.
5616 @c END TEXI2ROFF-KILL
5619 @subsection Evaluation
5620 @cindex lazy evaluation
5621 @cindex expression evaluation order
5622 The linker evaluates expressions lazily. It only computes the value of
5623 an expression when absolutely necessary.
5625 The linker needs some information, such as the value of the start
5626 address of the first section, and the origins and lengths of memory
5627 regions, in order to do any linking at all. These values are computed
5628 as soon as possible when the linker reads in the linker script.
5630 However, other values (such as symbol values) are not known or needed
5631 until after storage allocation. Such values are evaluated later, when
5632 other information (such as the sizes of output sections) is available
5633 for use in the symbol assignment expression.
5635 The sizes of sections cannot be known until after allocation, so
5636 assignments dependent upon these are not performed until after
5639 Some expressions, such as those depending upon the location counter
5640 @samp{.}, must be evaluated during section allocation.
5642 If the result of an expression is required, but the value is not
5643 available, then an error results. For example, a script like the
5649 .text 9+this_isnt_constant :
5655 will cause the error message @samp{non constant expression for initial
5658 @node Expression Section
5659 @subsection The Section of an Expression
5660 @cindex expression sections
5661 @cindex absolute expressions
5662 @cindex relative expressions
5663 @cindex absolute and relocatable symbols
5664 @cindex relocatable and absolute symbols
5665 @cindex symbols, relocatable and absolute
5666 Addresses and symbols may be section relative, or absolute. A section
5667 relative symbol is relocatable. If you request relocatable output
5668 using the @samp{-r} option, a further link operation may change the
5669 value of a section relative symbol. On the other hand, an absolute
5670 symbol will retain the same value throughout any further link
5673 Some terms in linker expressions are addresses. This is true of
5674 section relative symbols and for builtin functions that return an
5675 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
5676 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
5677 functions that return a non-address value, such as @code{LENGTH}.
5678 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
5679 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
5680 differently depending on their location, for compatibility with older
5681 versions of @code{ld}. Expressions appearing outside an output
5682 section definition treat all numbers as absolute addresses.
5683 Expressions appearing inside an output section definition treat
5684 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
5685 given, then absolute symbols and numbers are simply treated as numbers
5688 In the following simple example,
5695 __executable_start = 0x100;
5699 __data_start = 0x10;
5707 both @code{.} and @code{__executable_start} are set to the absolute
5708 address 0x100 in the first two assignments, then both @code{.} and
5709 @code{__data_start} are set to 0x10 relative to the @code{.data}
5710 section in the second two assignments.
5712 For expressions involving numbers, relative addresses and absolute
5713 addresses, ld follows these rules to evaluate terms:
5717 Unary operations on an absolute address or number, and binary
5718 operations on two absolute addresses or two numbers, or between one
5719 absolute address and a number, apply the operator to the value(s).
5721 Unary operations on a relative address, and binary operations on two
5722 relative addresses in the same section or between one relative address
5723 and a number, apply the operator to the offset part of the address(es).
5725 Other binary operations, that is, between two relative addresses not
5726 in the same section, or between a relative address and an absolute
5727 address, first convert any non-absolute term to an absolute address
5728 before applying the operator.
5731 The result section of each sub-expression is as follows:
5735 An operation involving only numbers results in a number.
5737 The result of comparisons, @samp{&&} and @samp{||} is also a number.
5739 The result of other binary arithmetic and logical operations on two
5740 relative addresses in the same section or two absolute addresses
5741 (after above conversions) is also a number.
5743 The result of other operations on relative addresses or one
5744 relative address and a number, is a relative address in the same
5745 section as the relative operand(s).
5747 The result of other operations on absolute addresses (after above
5748 conversions) is an absolute address.
5751 You can use the builtin function @code{ABSOLUTE} to force an expression
5752 to be absolute when it would otherwise be relative. For example, to
5753 create an absolute symbol set to the address of the end of the output
5754 section @samp{.data}:
5758 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
5762 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
5763 @samp{.data} section.
5765 Using @code{LOADADDR} also forces an expression absolute, since this
5766 particular builtin function returns an absolute address.
5768 @node Builtin Functions
5769 @subsection Builtin Functions
5770 @cindex functions in expressions
5771 The linker script language includes a number of builtin functions for
5772 use in linker script expressions.
5775 @item ABSOLUTE(@var{exp})
5776 @kindex ABSOLUTE(@var{exp})
5777 @cindex expression, absolute
5778 Return the absolute (non-relocatable, as opposed to non-negative) value
5779 of the expression @var{exp}. Primarily useful to assign an absolute
5780 value to a symbol within a section definition, where symbol values are
5781 normally section relative. @xref{Expression Section}.
5783 @item ADDR(@var{section})
5784 @kindex ADDR(@var{section})
5785 @cindex section address in expression
5786 Return the address (VMA) of the named @var{section}. Your
5787 script must previously have defined the location of that section. In
5788 the following example, @code{start_of_output_1}, @code{symbol_1} and
5789 @code{symbol_2} are assigned equivalent values, except that
5790 @code{symbol_1} will be relative to the @code{.output1} section while
5791 the other two will be absolute:
5797 start_of_output_1 = ABSOLUTE(.);
5802 symbol_1 = ADDR(.output1);
5803 symbol_2 = start_of_output_1;
5809 @item ALIGN(@var{align})
5810 @itemx ALIGN(@var{exp},@var{align})
5811 @kindex ALIGN(@var{align})
5812 @kindex ALIGN(@var{exp},@var{align})
5813 @cindex round up location counter
5814 @cindex align location counter
5815 @cindex round up expression
5816 @cindex align expression
5817 Return the location counter (@code{.}) or arbitrary expression aligned
5818 to the next @var{align} boundary. The single operand @code{ALIGN}
5819 doesn't change the value of the location counter---it just does
5820 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5821 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5822 equivalent to @code{ALIGN(., @var{align})}).
5824 Here is an example which aligns the output @code{.data} section to the
5825 next @code{0x2000} byte boundary after the preceding section and sets a
5826 variable within the section to the next @code{0x8000} boundary after the
5831 .data ALIGN(0x2000): @{
5833 variable = ALIGN(0x8000);
5839 The first use of @code{ALIGN} in this example specifies the location of
5840 a section because it is used as the optional @var{address} attribute of
5841 a section definition (@pxref{Output Section Address}). The second use
5842 of @code{ALIGN} is used to defines the value of a symbol.
5844 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5846 @item ALIGNOF(@var{section})
5847 @kindex ALIGNOF(@var{section})
5848 @cindex section alignment
5849 Return the alignment in bytes of the named @var{section}, if that section has
5850 been allocated. If the section has not been allocated when this is
5851 evaluated, the linker will report an error. In the following example,
5852 the alignment of the @code{.output} section is stored as the first
5853 value in that section.
5858 LONG (ALIGNOF (.output))
5865 @item BLOCK(@var{exp})
5866 @kindex BLOCK(@var{exp})
5867 This is a synonym for @code{ALIGN}, for compatibility with older linker
5868 scripts. It is most often seen when setting the address of an output
5871 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5872 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5873 This is equivalent to either
5875 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5879 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5882 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5883 for the data segment (area between the result of this expression and
5884 @code{DATA_SEGMENT_END}) than the former or not.
5885 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5886 memory will be saved at the expense of up to @var{commonpagesize} wasted
5887 bytes in the on-disk file.
5889 This expression can only be used directly in @code{SECTIONS} commands, not in
5890 any output section descriptions and only once in the linker script.
5891 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5892 be the system page size the object wants to be optimized for (while still
5893 working on system page sizes up to @var{maxpagesize}).
5898 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5901 @item DATA_SEGMENT_END(@var{exp})
5902 @kindex DATA_SEGMENT_END(@var{exp})
5903 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5904 evaluation purposes.
5907 . = DATA_SEGMENT_END(.);
5910 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5911 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5912 This defines the end of the @code{PT_GNU_RELRO} segment when
5913 @samp{-z relro} option is used. Second argument is returned.
5914 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5915 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5916 @var{exp} + @var{offset} is aligned to the most commonly used page
5917 boundary for particular target. If present in the linker script,
5918 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5919 @code{DATA_SEGMENT_END}.
5922 . = DATA_SEGMENT_RELRO_END(24, .);
5925 @item DEFINED(@var{symbol})
5926 @kindex DEFINED(@var{symbol})
5927 @cindex symbol defaults
5928 Return 1 if @var{symbol} is in the linker global symbol table and is
5929 defined before the statement using DEFINED in the script, otherwise
5930 return 0. You can use this function to provide
5931 default values for symbols. For example, the following script fragment
5932 shows how to set a global symbol @samp{begin} to the first location in
5933 the @samp{.text} section---but if a symbol called @samp{begin} already
5934 existed, its value is preserved:
5940 begin = DEFINED(begin) ? begin : . ;
5948 @item LENGTH(@var{memory})
5949 @kindex LENGTH(@var{memory})
5950 Return the length of the memory region named @var{memory}.
5952 @item LOADADDR(@var{section})
5953 @kindex LOADADDR(@var{section})
5954 @cindex section load address in expression
5955 Return the absolute LMA of the named @var{section}. (@pxref{Output
5958 @item LOG2CEIL(@var{exp})
5959 @kindex LOG2CEIL(@var{exp})
5960 Return the binary logarithm of @var{exp} rounded towards infinity.
5961 @code{LOG2CEIL(0)} returns 0.
5964 @item MAX(@var{exp1}, @var{exp2})
5965 Returns the maximum of @var{exp1} and @var{exp2}.
5968 @item MIN(@var{exp1}, @var{exp2})
5969 Returns the minimum of @var{exp1} and @var{exp2}.
5971 @item NEXT(@var{exp})
5972 @kindex NEXT(@var{exp})
5973 @cindex unallocated address, next
5974 Return the next unallocated address that is a multiple of @var{exp}.
5975 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5976 use the @code{MEMORY} command to define discontinuous memory for the
5977 output file, the two functions are equivalent.
5979 @item ORIGIN(@var{memory})
5980 @kindex ORIGIN(@var{memory})
5981 Return the origin of the memory region named @var{memory}.
5983 @item SEGMENT_START(@var{segment}, @var{default})
5984 @kindex SEGMENT_START(@var{segment}, @var{default})
5985 Return the base address of the named @var{segment}. If an explicit
5986 value has already been given for this segment (with a command-line
5987 @samp{-T} option) then that value will be returned otherwise the value
5988 will be @var{default}. At present, the @samp{-T} command-line option
5989 can only be used to set the base address for the ``text'', ``data'', and
5990 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
5993 @item SIZEOF(@var{section})
5994 @kindex SIZEOF(@var{section})
5995 @cindex section size
5996 Return the size in bytes of the named @var{section}, if that section has
5997 been allocated. If the section has not been allocated when this is
5998 evaluated, the linker will report an error. In the following example,
5999 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6008 symbol_1 = .end - .start ;
6009 symbol_2 = SIZEOF(.output);
6014 @item SIZEOF_HEADERS
6015 @itemx sizeof_headers
6016 @kindex SIZEOF_HEADERS
6018 Return the size in bytes of the output file's headers. This is
6019 information which appears at the start of the output file. You can use
6020 this number when setting the start address of the first section, if you
6021 choose, to facilitate paging.
6023 @cindex not enough room for program headers
6024 @cindex program headers, not enough room
6025 When producing an ELF output file, if the linker script uses the
6026 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6027 number of program headers before it has determined all the section
6028 addresses and sizes. If the linker later discovers that it needs
6029 additional program headers, it will report an error @samp{not enough
6030 room for program headers}. To avoid this error, you must avoid using
6031 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6032 script to avoid forcing the linker to use additional program headers, or
6033 you must define the program headers yourself using the @code{PHDRS}
6034 command (@pxref{PHDRS}).
6037 @node Implicit Linker Scripts
6038 @section Implicit Linker Scripts
6039 @cindex implicit linker scripts
6040 If you specify a linker input file which the linker can not recognize as
6041 an object file or an archive file, it will try to read the file as a
6042 linker script. If the file can not be parsed as a linker script, the
6043 linker will report an error.
6045 An implicit linker script will not replace the default linker script.
6047 Typically an implicit linker script would contain only symbol
6048 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6051 Any input files read because of an implicit linker script will be read
6052 at the position in the command line where the implicit linker script was
6053 read. This can affect archive searching.
6056 @node Machine Dependent
6057 @chapter Machine Dependent Features
6059 @cindex machine dependencies
6060 @command{ld} has additional features on some platforms; the following
6061 sections describe them. Machines where @command{ld} has no additional
6062 functionality are not listed.
6066 * H8/300:: @command{ld} and the H8/300
6069 * i960:: @command{ld} and the Intel 960 family
6072 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6075 * ARM:: @command{ld} and the ARM family
6078 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6081 * M68K:: @command{ld} and the Motorola 68K family
6084 * MIPS:: @command{ld} and the MIPS family
6087 * MMIX:: @command{ld} and MMIX
6090 * MSP430:: @command{ld} and MSP430
6093 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6096 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6099 * SPU ELF:: @command{ld} and SPU ELF Support
6102 * TI COFF:: @command{ld} and TI COFF
6105 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6108 * Xtensa:: @command{ld} and Xtensa Processors
6119 @section @command{ld} and the H8/300
6121 @cindex H8/300 support
6122 For the H8/300, @command{ld} can perform these global optimizations when
6123 you specify the @samp{--relax} command-line option.
6126 @cindex relaxing on H8/300
6127 @item relaxing address modes
6128 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6129 targets are within eight bits, and turns them into eight-bit
6130 program-counter relative @code{bsr} and @code{bra} instructions,
6133 @cindex synthesizing on H8/300
6134 @item synthesizing instructions
6135 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6136 @command{ld} finds all @code{mov.b} instructions which use the
6137 sixteen-bit absolute address form, but refer to the top
6138 page of memory, and changes them to use the eight-bit address form.
6139 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6140 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6141 top page of memory).
6143 @command{ld} finds all @code{mov} instructions which use the register
6144 indirect with 32-bit displacement addressing mode, but use a small
6145 displacement inside 16-bit displacement range, and changes them to use
6146 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6147 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6148 whenever the displacement @var{d} is in the 16 bit signed integer
6149 range. Only implemented in ELF-format ld).
6151 @item bit manipulation instructions
6152 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6153 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6154 which use 32 bit and 16 bit absolute address form, but refer to the top
6155 page of memory, and changes them to use the 8 bit address form.
6156 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6157 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6158 the top page of memory).
6160 @item system control instructions
6161 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6162 32 bit absolute address form, but refer to the top page of memory, and
6163 changes them to use 16 bit address form.
6164 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6165 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6166 the top page of memory).
6176 @c This stuff is pointless to say unless you're especially concerned
6177 @c with Renesas chips; don't enable it for generic case, please.
6179 @chapter @command{ld} and Other Renesas Chips
6181 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6182 H8/500, and SH chips. No special features, commands, or command-line
6183 options are required for these chips.
6193 @section @command{ld} and the Intel 960 Family
6195 @cindex i960 support
6197 You can use the @samp{-A@var{architecture}} command line option to
6198 specify one of the two-letter names identifying members of the 960
6199 family; the option specifies the desired output target, and warns of any
6200 incompatible instructions in the input files. It also modifies the
6201 linker's search strategy for archive libraries, to support the use of
6202 libraries specific to each particular architecture, by including in the
6203 search loop names suffixed with the string identifying the architecture.
6205 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
6206 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
6207 paths, and in any paths you specify with @samp{-L}) for a library with
6220 The first two possibilities would be considered in any event; the last
6221 two are due to the use of @w{@samp{-ACA}}.
6223 You can meaningfully use @samp{-A} more than once on a command line, since
6224 the 960 architecture family allows combination of target architectures; each
6225 use will add another pair of name variants to search for when @w{@samp{-l}}
6226 specifies a library.
6228 @cindex @option{--relax} on i960
6229 @cindex relaxing on i960
6230 @command{ld} supports the @samp{--relax} option for the i960 family. If
6231 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
6232 @code{calx} instructions whose targets are within 24 bits, and turns
6233 them into 24-bit program-counter relative @code{bal} and @code{cal}
6234 instructions, respectively. @command{ld} also turns @code{cal}
6235 instructions into @code{bal} instructions when it determines that the
6236 target subroutine is a leaf routine (that is, the target subroutine does
6237 not itself call any subroutines).
6239 @cindex Cortex-A8 erratum workaround
6240 @kindex --fix-cortex-a8
6241 @kindex --no-fix-cortex-a8
6242 The @samp{--fix-cortex-a8} switch enables a link-time workaround for an erratum in certain Cortex-A8 processors. The workaround is enabled by default if you are targeting the ARM v7-A architecture profile. It can be enabled otherwise by specifying @samp{--fix-cortex-a8}, or disabled unconditionally by specifying @samp{--no-fix-cortex-a8}.
6244 The erratum only affects Thumb-2 code. Please contact ARM for further details.
6246 @kindex --merge-exidx-entries
6247 @kindex --no-merge-exidx-entries
6248 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
6265 @node M68HC11/68HC12
6266 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6268 @cindex M68HC11 and 68HC12 support
6270 @subsection Linker Relaxation
6272 For the Motorola 68HC11, @command{ld} can perform these global
6273 optimizations when you specify the @samp{--relax} command-line option.
6276 @cindex relaxing on M68HC11
6277 @item relaxing address modes
6278 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6279 targets are within eight bits, and turns them into eight-bit
6280 program-counter relative @code{bsr} and @code{bra} instructions,
6283 @command{ld} also looks at all 16-bit extended addressing modes and
6284 transforms them in a direct addressing mode when the address is in
6285 page 0 (between 0 and 0x0ff).
6287 @item relaxing gcc instruction group
6288 When @command{gcc} is called with @option{-mrelax}, it can emit group
6289 of instructions that the linker can optimize to use a 68HC11 direct
6290 addressing mode. These instructions consists of @code{bclr} or
6291 @code{bset} instructions.
6295 @subsection Trampoline Generation
6297 @cindex trampoline generation on M68HC11
6298 @cindex trampoline generation on M68HC12
6299 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6300 call a far function using a normal @code{jsr} instruction. The linker
6301 will also change the relocation to some far function to use the
6302 trampoline address instead of the function address. This is typically the
6303 case when a pointer to a function is taken. The pointer will in fact
6304 point to the function trampoline.
6312 @section @command{ld} and the ARM family
6314 @cindex ARM interworking support
6315 @kindex --support-old-code
6316 For the ARM, @command{ld} will generate code stubs to allow functions calls
6317 between ARM and Thumb code. These stubs only work with code that has
6318 been compiled and assembled with the @samp{-mthumb-interwork} command
6319 line option. If it is necessary to link with old ARM object files or
6320 libraries, which have not been compiled with the -mthumb-interwork
6321 option then the @samp{--support-old-code} command line switch should be
6322 given to the linker. This will make it generate larger stub functions
6323 which will work with non-interworking aware ARM code. Note, however,
6324 the linker does not support generating stubs for function calls to
6325 non-interworking aware Thumb code.
6327 @cindex thumb entry point
6328 @cindex entry point, thumb
6329 @kindex --thumb-entry=@var{entry}
6330 The @samp{--thumb-entry} switch is a duplicate of the generic
6331 @samp{--entry} switch, in that it sets the program's starting address.
6332 But it also sets the bottom bit of the address, so that it can be
6333 branched to using a BX instruction, and the program will start
6334 executing in Thumb mode straight away.
6336 @cindex PE import table prefixing
6337 @kindex --use-nul-prefixed-import-tables
6338 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6339 the import tables idata4 and idata5 have to be generated with a zero
6340 element prefix for import libraries. This is the old style to generate
6341 import tables. By default this option is turned off.
6345 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6346 executables. This option is only valid when linking big-endian objects.
6347 The resulting image will contain big-endian data and little-endian code.
6350 @kindex --target1-rel
6351 @kindex --target1-abs
6352 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6353 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6354 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6355 and @samp{--target1-abs} switches override the default.
6358 @kindex --target2=@var{type}
6359 The @samp{--target2=type} switch overrides the default definition of the
6360 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6361 meanings, and target defaults are as follows:
6364 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6366 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6368 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6373 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6374 specification) enables objects compiled for the ARMv4 architecture to be
6375 interworking-safe when linked with other objects compiled for ARMv4t, but
6376 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6378 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6379 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6380 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6382 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6383 relocations are ignored.
6385 @cindex FIX_V4BX_INTERWORKING
6386 @kindex --fix-v4bx-interworking
6387 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6388 relocations with a branch to the following veneer:
6396 This allows generation of libraries/applications that work on ARMv4 cores
6397 and are still interworking safe. Note that the above veneer clobbers the
6398 condition flags, so may cause incorrect program behavior in rare cases.
6402 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6403 BLX instructions (available on ARMv5t and above) in various
6404 situations. Currently it is used to perform calls via the PLT from Thumb
6405 code using BLX rather than using BX and a mode-switching stub before
6406 each PLT entry. This should lead to such calls executing slightly faster.
6408 This option is enabled implicitly for SymbianOS, so there is no need to
6409 specify it if you are using that target.
6411 @cindex VFP11_DENORM_FIX
6412 @kindex --vfp11-denorm-fix
6413 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6414 bug in certain VFP11 coprocessor hardware, which sometimes allows
6415 instructions with denorm operands (which must be handled by support code)
6416 to have those operands overwritten by subsequent instructions before
6417 the support code can read the intended values.
6419 The bug may be avoided in scalar mode if you allow at least one
6420 intervening instruction between a VFP11 instruction which uses a register
6421 and another instruction which writes to the same register, or at least two
6422 intervening instructions if vector mode is in use. The bug only affects
6423 full-compliance floating-point mode: you do not need this workaround if
6424 you are using "runfast" mode. Please contact ARM for further details.
6426 If you know you are using buggy VFP11 hardware, you can
6427 enable this workaround by specifying the linker option
6428 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6429 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6430 vector mode (the latter also works for scalar code). The default is
6431 @samp{--vfp-denorm-fix=none}.
6433 If the workaround is enabled, instructions are scanned for
6434 potentially-troublesome sequences, and a veneer is created for each
6435 such sequence which may trigger the erratum. The veneer consists of the
6436 first instruction of the sequence and a branch back to the subsequent
6437 instruction. The original instruction is then replaced with a branch to
6438 the veneer. The extra cycles required to call and return from the veneer
6439 are sufficient to avoid the erratum in both the scalar and vector cases.
6441 @cindex ARM1176 erratum workaround
6442 @kindex --fix-arm1176
6443 @kindex --no-fix-arm1176
6444 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6445 in certain ARM1176 processors. The workaround is enabled by default if you
6446 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6447 unconditionally by specifying @samp{--no-fix-arm1176}.
6449 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6450 Programmer Advice Notice'' available on the ARM documentation website at:
6451 http://infocenter.arm.com/.
6453 @cindex NO_ENUM_SIZE_WARNING
6454 @kindex --no-enum-size-warning
6455 The @option{--no-enum-size-warning} switch prevents the linker from
6456 warning when linking object files that specify incompatible EABI
6457 enumeration size attributes. For example, with this switch enabled,
6458 linking of an object file using 32-bit enumeration values with another
6459 using enumeration values fitted into the smallest possible space will
6462 @cindex NO_WCHAR_SIZE_WARNING
6463 @kindex --no-wchar-size-warning
6464 The @option{--no-wchar-size-warning} switch prevents the linker from
6465 warning when linking object files that specify incompatible EABI
6466 @code{wchar_t} size attributes. For example, with this switch enabled,
6467 linking of an object file using 32-bit @code{wchar_t} values with another
6468 using 16-bit @code{wchar_t} values will not be diagnosed.
6471 @kindex --pic-veneer
6472 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6473 ARM/Thumb interworking veneers, even if the rest of the binary
6474 is not PIC. This avoids problems on uClinux targets where
6475 @samp{--emit-relocs} is used to generate relocatable binaries.
6477 @cindex STUB_GROUP_SIZE
6478 @kindex --stub-group-size=@var{N}
6479 The linker will automatically generate and insert small sequences of
6480 code into a linked ARM ELF executable whenever an attempt is made to
6481 perform a function call to a symbol that is too far away. The
6482 placement of these sequences of instructions - called stubs - is
6483 controlled by the command line option @option{--stub-group-size=N}.
6484 The placement is important because a poor choice can create a need for
6485 duplicate stubs, increasing the code size. The linker will try to
6486 group stubs together in order to reduce interruptions to the flow of
6487 code, but it needs guidance as to how big these groups should be and
6488 where they should be placed.
6490 The value of @samp{N}, the parameter to the
6491 @option{--stub-group-size=} option controls where the stub groups are
6492 placed. If it is negative then all stubs are placed after the first
6493 branch that needs them. If it is positive then the stubs can be
6494 placed either before or after the branches that need them. If the
6495 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6496 exactly where to place groups of stubs, using its built in heuristics.
6497 A value of @samp{N} greater than 1 (or smaller than -1) tells the
6498 linker that a single group of stubs can service at most @samp{N} bytes
6499 from the input sections.
6501 The default, if @option{--stub-group-size=} is not specified, is
6504 Farcalls stubs insertion is fully supported for the ARM-EABI target
6505 only, because it relies on object files properties not present
6519 @section @command{ld} and HPPA 32-bit ELF Support
6520 @cindex HPPA multiple sub-space stubs
6521 @kindex --multi-subspace
6522 When generating a shared library, @command{ld} will by default generate
6523 import stubs suitable for use with a single sub-space application.
6524 The @samp{--multi-subspace} switch causes @command{ld} to generate export
6525 stubs, and different (larger) import stubs suitable for use with
6526 multiple sub-spaces.
6528 @cindex HPPA stub grouping
6529 @kindex --stub-group-size=@var{N}
6530 Long branch stubs and import/export stubs are placed by @command{ld} in
6531 stub sections located between groups of input sections.
6532 @samp{--stub-group-size} specifies the maximum size of a group of input
6533 sections handled by one stub section. Since branch offsets are signed,
6534 a stub section may serve two groups of input sections, one group before
6535 the stub section, and one group after it. However, when using
6536 conditional branches that require stubs, it may be better (for branch
6537 prediction) that stub sections only serve one group of input sections.
6538 A negative value for @samp{N} chooses this scheme, ensuring that
6539 branches to stubs always use a negative offset. Two special values of
6540 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6541 @command{ld} to automatically size input section groups for the branch types
6542 detected, with the same behaviour regarding stub placement as other
6543 positive or negative values of @samp{N} respectively.
6545 Note that @samp{--stub-group-size} does not split input sections. A
6546 single input section larger than the group size specified will of course
6547 create a larger group (of one section). If input sections are too
6548 large, it may not be possible for a branch to reach its stub.
6561 @section @command{ld} and the Motorola 68K family
6563 @cindex Motorola 68K GOT generation
6564 @kindex --got=@var{type}
6565 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
6566 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
6567 @samp{target}. When @samp{target} is selected the linker chooses
6568 the default GOT generation scheme for the current target.
6569 @samp{single} tells the linker to generate a single GOT with
6570 entries only at non-negative offsets.
6571 @samp{negative} instructs the linker to generate a single GOT with
6572 entries at both negative and positive offsets. Not all environments
6574 @samp{multigot} allows the linker to generate several GOTs in the
6575 output file. All GOT references from a single input object
6576 file access the same GOT, but references from different input object
6577 files might access different GOTs. Not all environments support such GOTs.
6590 @section @command{ld} and the MIPS family
6592 @cindex MIPS microMIPS instruction choice selection
6595 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
6596 microMIPS instructions used in code generated by the linker, such as that
6597 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
6598 used, then the linker only uses 32-bit instruction encodings. By default
6599 or if @samp{--no-insn32} is used, all instruction encodings are used,
6600 including 16-bit ones where possible.
6613 @section @code{ld} and MMIX
6614 For MMIX, there is a choice of generating @code{ELF} object files or
6615 @code{mmo} object files when linking. The simulator @code{mmix}
6616 understands the @code{mmo} format. The binutils @code{objcopy} utility
6617 can translate between the two formats.
6619 There is one special section, the @samp{.MMIX.reg_contents} section.
6620 Contents in this section is assumed to correspond to that of global
6621 registers, and symbols referring to it are translated to special symbols,
6622 equal to registers. In a final link, the start address of the
6623 @samp{.MMIX.reg_contents} section corresponds to the first allocated
6624 global register multiplied by 8. Register @code{$255} is not included in
6625 this section; it is always set to the program entry, which is at the
6626 symbol @code{Main} for @code{mmo} files.
6628 Global symbols with the prefix @code{__.MMIX.start.}, for example
6629 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
6630 The default linker script uses these to set the default start address
6633 Initial and trailing multiples of zero-valued 32-bit words in a section,
6634 are left out from an mmo file.
6647 @section @code{ld} and MSP430
6648 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
6649 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
6650 just pass @samp{-m help} option to the linker).
6652 @cindex MSP430 extra sections
6653 The linker will recognize some extra sections which are MSP430 specific:
6656 @item @samp{.vectors}
6657 Defines a portion of ROM where interrupt vectors located.
6659 @item @samp{.bootloader}
6660 Defines the bootloader portion of the ROM (if applicable). Any code
6661 in this section will be uploaded to the MPU.
6663 @item @samp{.infomem}
6664 Defines an information memory section (if applicable). Any code in
6665 this section will be uploaded to the MPU.
6667 @item @samp{.infomemnobits}
6668 This is the same as the @samp{.infomem} section except that any code
6669 in this section will not be uploaded to the MPU.
6671 @item @samp{.noinit}
6672 Denotes a portion of RAM located above @samp{.bss} section.
6674 The last two sections are used by gcc.
6688 @section @command{ld} and PowerPC 32-bit ELF Support
6689 @cindex PowerPC long branches
6690 @kindex --relax on PowerPC
6691 Branches on PowerPC processors are limited to a signed 26-bit
6692 displacement, which may result in @command{ld} giving
6693 @samp{relocation truncated to fit} errors with very large programs.
6694 @samp{--relax} enables the generation of trampolines that can access
6695 the entire 32-bit address space. These trampolines are inserted at
6696 section boundaries, so may not themselves be reachable if an input
6697 section exceeds 33M in size. You may combine @samp{-r} and
6698 @samp{--relax} to add trampolines in a partial link. In that case
6699 both branches to undefined symbols and inter-section branches are also
6700 considered potentially out of range, and trampolines inserted.
6702 @cindex PowerPC ELF32 options
6707 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
6708 generates code capable of using a newer PLT and GOT layout that has
6709 the security advantage of no executable section ever needing to be
6710 writable and no writable section ever being executable. PowerPC
6711 @command{ld} will generate this layout, including stubs to access the
6712 PLT, if all input files (including startup and static libraries) were
6713 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
6714 BSS PLT (and GOT layout) which can give slightly better performance.
6716 @kindex --secure-plt
6718 @command{ld} will use the new PLT and GOT layout if it is linking new
6719 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
6720 when linking non-PIC code. This option requests the new PLT and GOT
6721 layout. A warning will be given if some object file requires the old
6727 The new secure PLT and GOT are placed differently relative to other
6728 sections compared to older BSS PLT and GOT placement. The location of
6729 @code{.plt} must change because the new secure PLT is an initialized
6730 section while the old PLT is uninitialized. The reason for the
6731 @code{.got} change is more subtle: The new placement allows
6732 @code{.got} to be read-only in applications linked with
6733 @samp{-z relro -z now}. However, this placement means that
6734 @code{.sdata} cannot always be used in shared libraries, because the
6735 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
6736 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
6737 GCC doesn't use @code{.sdata} in shared libraries, so this option is
6738 really only useful for other compilers that may do so.
6740 @cindex PowerPC stub symbols
6741 @kindex --emit-stub-syms
6742 @item --emit-stub-syms
6743 This option causes @command{ld} to label linker stubs with a local
6744 symbol that encodes the stub type and destination.
6746 @cindex PowerPC TLS optimization
6747 @kindex --no-tls-optimize
6748 @item --no-tls-optimize
6749 PowerPC @command{ld} normally performs some optimization of code
6750 sequences used to access Thread-Local Storage. Use this option to
6751 disable the optimization.
6764 @node PowerPC64 ELF64
6765 @section @command{ld} and PowerPC64 64-bit ELF Support
6767 @cindex PowerPC64 ELF64 options
6769 @cindex PowerPC64 stub grouping
6770 @kindex --stub-group-size
6771 @item --stub-group-size
6772 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
6773 by @command{ld} in stub sections located between groups of input sections.
6774 @samp{--stub-group-size} specifies the maximum size of a group of input
6775 sections handled by one stub section. Since branch offsets are signed,
6776 a stub section may serve two groups of input sections, one group before
6777 the stub section, and one group after it. However, when using
6778 conditional branches that require stubs, it may be better (for branch
6779 prediction) that stub sections only serve one group of input sections.
6780 A negative value for @samp{N} chooses this scheme, ensuring that
6781 branches to stubs always use a negative offset. Two special values of
6782 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6783 @command{ld} to automatically size input section groups for the branch types
6784 detected, with the same behaviour regarding stub placement as other
6785 positive or negative values of @samp{N} respectively.
6787 Note that @samp{--stub-group-size} does not split input sections. A
6788 single input section larger than the group size specified will of course
6789 create a larger group (of one section). If input sections are too
6790 large, it may not be possible for a branch to reach its stub.
6792 @cindex PowerPC64 stub symbols
6793 @kindex --emit-stub-syms
6794 @item --emit-stub-syms
6795 This option causes @command{ld} to label linker stubs with a local
6796 symbol that encodes the stub type and destination.
6798 @cindex PowerPC64 dot symbols
6800 @kindex --no-dotsyms
6801 @item --dotsyms, --no-dotsyms
6802 These two options control how @command{ld} interprets version patterns
6803 in a version script. Older PowerPC64 compilers emitted both a
6804 function descriptor symbol with the same name as the function, and a
6805 code entry symbol with the name prefixed by a dot (@samp{.}). To
6806 properly version a function @samp{foo}, the version script thus needs
6807 to control both @samp{foo} and @samp{.foo}. The option
6808 @samp{--dotsyms}, on by default, automatically adds the required
6809 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
6812 @cindex PowerPC64 TLS optimization
6813 @kindex --no-tls-optimize
6814 @item --no-tls-optimize
6815 PowerPC64 @command{ld} normally performs some optimization of code
6816 sequences used to access Thread-Local Storage. Use this option to
6817 disable the optimization.
6819 @cindex PowerPC64 OPD optimization
6820 @kindex --no-opd-optimize
6821 @item --no-opd-optimize
6822 PowerPC64 @command{ld} normally removes @code{.opd} section entries
6823 corresponding to deleted link-once functions, or functions removed by
6824 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
6825 Use this option to disable @code{.opd} optimization.
6827 @cindex PowerPC64 OPD spacing
6828 @kindex --non-overlapping-opd
6829 @item --non-overlapping-opd
6830 Some PowerPC64 compilers have an option to generate compressed
6831 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
6832 the static chain pointer (unused in C) with the first word of the next
6833 entry. This option expands such entries to the full 24 bytes.
6835 @cindex PowerPC64 TOC optimization
6836 @kindex --no-toc-optimize
6837 @item --no-toc-optimize
6838 PowerPC64 @command{ld} normally removes unused @code{.toc} section
6839 entries. Such entries are detected by examining relocations that
6840 reference the TOC in code sections. A reloc in a deleted code section
6841 marks a TOC word as unneeded, while a reloc in a kept code section
6842 marks a TOC word as needed. Since the TOC may reference itself, TOC
6843 relocs are also examined. TOC words marked as both needed and
6844 unneeded will of course be kept. TOC words without any referencing
6845 reloc are assumed to be part of a multi-word entry, and are kept or
6846 discarded as per the nearest marked preceding word. This works
6847 reliably for compiler generated code, but may be incorrect if assembly
6848 code is used to insert TOC entries. Use this option to disable the
6851 @cindex PowerPC64 multi-TOC
6852 @kindex --no-multi-toc
6853 @item --no-multi-toc
6854 If given any toc option besides @code{-mcmodel=medium} or
6855 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
6857 entries are accessed with a 16-bit offset from r2. This limits the
6858 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
6859 grouping code sections such that each group uses less than 64K for its
6860 TOC entries, then inserts r2 adjusting stubs between inter-group
6861 calls. @command{ld} does not split apart input sections, so cannot
6862 help if a single input file has a @code{.toc} section that exceeds
6863 64K, most likely from linking multiple files with @command{ld -r}.
6864 Use this option to turn off this feature.
6866 @cindex PowerPC64 TOC sorting
6867 @kindex --no-toc-sort
6869 By default, @command{ld} sorts TOC sections so that those whose file
6870 happens to have a section called @code{.init} or @code{.fini} are
6871 placed first, followed by TOC sections referenced by code generated
6872 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
6873 referenced only by code generated with PowerPC64 gcc's
6874 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
6875 results in better TOC grouping for multi-TOC. Use this option to turn
6878 @cindex PowerPC64 PLT stub alignment
6880 @kindex --no-plt-align
6882 @itemx --no-plt-align
6883 Use these options to control whether individual PLT call stubs are
6884 aligned to a 32-byte boundary, or to the specified power of two
6885 boundary when using @code{--plt-align=}. By default PLT call stubs
6888 @cindex PowerPC64 PLT call stub static chain
6889 @kindex --plt-static-chain
6890 @kindex --no-plt-static-chain
6891 @item --plt-static-chain
6892 @itemx --no-plt-static-chain
6893 Use these options to control whether PLT call stubs load the static
6894 chain pointer (r11). @code{ld} defaults to not loading the static
6895 chain since there is never any need to do so on a PLT call.
6897 @cindex PowerPC64 PLT call stub thread safety
6898 @kindex --plt-thread-safe
6899 @kindex --no-plt-thread-safe
6900 @item --plt-thread-safe
6901 @itemx --no-thread-safe
6902 With power7's weakly ordered memory model, it is possible when using
6903 lazy binding for ld.so to update a plt entry in one thread and have
6904 another thread see the individual plt entry words update in the wrong
6905 order, despite ld.so carefully writing in the correct order and using
6906 memory write barriers. To avoid this we need some sort of read
6907 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
6908 looks for calls to commonly used functions that create threads, and if
6909 seen, adds the necessary barriers. Use these options to change the
6924 @section @command{ld} and SPU ELF Support
6926 @cindex SPU ELF options
6932 This option marks an executable as a PIC plugin module.
6934 @cindex SPU overlays
6935 @kindex --no-overlays
6937 Normally, @command{ld} recognizes calls to functions within overlay
6938 regions, and redirects such calls to an overlay manager via a stub.
6939 @command{ld} also provides a built-in overlay manager. This option
6940 turns off all this special overlay handling.
6942 @cindex SPU overlay stub symbols
6943 @kindex --emit-stub-syms
6944 @item --emit-stub-syms
6945 This option causes @command{ld} to label overlay stubs with a local
6946 symbol that encodes the stub type and destination.
6948 @cindex SPU extra overlay stubs
6949 @kindex --extra-overlay-stubs
6950 @item --extra-overlay-stubs
6951 This option causes @command{ld} to add overlay call stubs on all
6952 function calls out of overlay regions. Normally stubs are not added
6953 on calls to non-overlay regions.
6955 @cindex SPU local store size
6956 @kindex --local-store=lo:hi
6957 @item --local-store=lo:hi
6958 @command{ld} usually checks that a final executable for SPU fits in
6959 the address range 0 to 256k. This option may be used to change the
6960 range. Disable the check entirely with @option{--local-store=0:0}.
6963 @kindex --stack-analysis
6964 @item --stack-analysis
6965 SPU local store space is limited. Over-allocation of stack space
6966 unnecessarily limits space available for code and data, while
6967 under-allocation results in runtime failures. If given this option,
6968 @command{ld} will provide an estimate of maximum stack usage.
6969 @command{ld} does this by examining symbols in code sections to
6970 determine the extents of functions, and looking at function prologues
6971 for stack adjusting instructions. A call-graph is created by looking
6972 for relocations on branch instructions. The graph is then searched
6973 for the maximum stack usage path. Note that this analysis does not
6974 find calls made via function pointers, and does not handle recursion
6975 and other cycles in the call graph. Stack usage may be
6976 under-estimated if your code makes such calls. Also, stack usage for
6977 dynamic allocation, e.g. alloca, will not be detected. If a link map
6978 is requested, detailed information about each function's stack usage
6979 and calls will be given.
6982 @kindex --emit-stack-syms
6983 @item --emit-stack-syms
6984 This option, if given along with @option{--stack-analysis} will result
6985 in @command{ld} emitting stack sizing symbols for each function.
6986 These take the form @code{__stack_<function_name>} for global
6987 functions, and @code{__stack_<number>_<function_name>} for static
6988 functions. @code{<number>} is the section id in hex. The value of
6989 such symbols is the stack requirement for the corresponding function.
6990 The symbol size will be zero, type @code{STT_NOTYPE}, binding
6991 @code{STB_LOCAL}, and section @code{SHN_ABS}.
7005 @section @command{ld}'s Support for Various TI COFF Versions
7006 @cindex TI COFF versions
7007 @kindex --format=@var{version}
7008 The @samp{--format} switch allows selection of one of the various
7009 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
7010 also supported. The TI COFF versions also vary in header byte-order
7011 format; @command{ld} will read any version or byte order, but the output
7012 header format depends on the default specified by the specific target.
7025 @section @command{ld} and WIN32 (cygwin/mingw)
7027 This section describes some of the win32 specific @command{ld} issues.
7028 See @ref{Options,,Command Line Options} for detailed description of the
7029 command line options mentioned here.
7032 @cindex import libraries
7033 @item import libraries
7034 The standard Windows linker creates and uses so-called import
7035 libraries, which contains information for linking to dll's. They are
7036 regular static archives and are handled as any other static
7037 archive. The cygwin and mingw ports of @command{ld} have specific
7038 support for creating such libraries provided with the
7039 @samp{--out-implib} command line option.
7041 @item exporting DLL symbols
7042 @cindex exporting DLL symbols
7043 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
7046 @item using auto-export functionality
7047 @cindex using auto-export functionality
7048 By default @command{ld} exports symbols with the auto-export functionality,
7049 which is controlled by the following command line options:
7052 @item --export-all-symbols [This is the default]
7053 @item --exclude-symbols
7054 @item --exclude-libs
7055 @item --exclude-modules-for-implib
7056 @item --version-script
7059 When auto-export is in operation, @command{ld} will export all the non-local
7060 (global and common) symbols it finds in a DLL, with the exception of a few
7061 symbols known to belong to the system's runtime and libraries. As it will
7062 often not be desirable to export all of a DLL's symbols, which may include
7063 private functions that are not part of any public interface, the command-line
7064 options listed above may be used to filter symbols out from the list for
7065 exporting. The @samp{--output-def} option can be used in order to see the
7066 final list of exported symbols with all exclusions taken into effect.
7068 If @samp{--export-all-symbols} is not given explicitly on the
7069 command line, then the default auto-export behavior will be @emph{disabled}
7070 if either of the following are true:
7073 @item A DEF file is used.
7074 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
7077 @item using a DEF file
7078 @cindex using a DEF file
7079 Another way of exporting symbols is using a DEF file. A DEF file is
7080 an ASCII file containing definitions of symbols which should be
7081 exported when a dll is created. Usually it is named @samp{<dll
7082 name>.def} and is added as any other object file to the linker's
7083 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7086 gcc -o <output> <objectfiles> <dll name>.def
7089 Using a DEF file turns off the normal auto-export behavior, unless the
7090 @samp{--export-all-symbols} option is also used.
7092 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7095 LIBRARY "xyz.dll" BASE=0x20000000
7101 another_foo = abc.dll.afoo
7107 This example defines a DLL with a non-default base address and seven
7108 symbols in the export table. The third exported symbol @code{_bar} is an
7109 alias for the second. The fourth symbol, @code{another_foo} is resolved
7110 by "forwarding" to another module and treating it as an alias for
7111 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7112 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7113 export library is an alias of @samp{foo}, which gets the string name
7114 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7115 symbol, which gets in export table the name @samp{var1}.
7117 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7118 name of the output DLL. If @samp{<name>} does not include a suffix,
7119 the default library suffix, @samp{.DLL} is appended.
7121 When the .DEF file is used to build an application, rather than a
7122 library, the @code{NAME <name>} command should be used instead of
7123 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7124 executable suffix, @samp{.EXE} is appended.
7126 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7127 specification @code{BASE = <number>} may be used to specify a
7128 non-default base address for the image.
7130 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7131 or they specify an empty string, the internal name is the same as the
7132 filename specified on the command line.
7134 The complete specification of an export symbol is:
7138 ( ( ( <name1> [ = <name2> ] )
7139 | ( <name1> = <module-name> . <external-name>))
7140 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7143 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7144 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7145 @samp{<name1>} as a "forward" alias for the symbol
7146 @samp{<external-name>} in the DLL @samp{<module-name>}.
7147 Optionally, the symbol may be exported by the specified ordinal
7148 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7149 string in import/export table for the symbol.
7151 The optional keywords that follow the declaration indicate:
7153 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7154 will still be exported by its ordinal alias (either the value specified
7155 by the .def specification or, otherwise, the value assigned by the
7156 linker). The symbol name, however, does remain visible in the import
7157 library (if any), unless @code{PRIVATE} is also specified.
7159 @code{DATA}: The symbol is a variable or object, rather than a function.
7160 The import lib will export only an indirect reference to @code{foo} as
7161 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7164 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7165 well as @code{_imp__foo} into the import library. Both refer to the
7166 read-only import address table's pointer to the variable, not to the
7167 variable itself. This can be dangerous. If the user code fails to add
7168 the @code{dllimport} attribute and also fails to explicitly add the
7169 extra indirection that the use of the attribute enforces, the
7170 application will behave unexpectedly.
7172 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7173 it into the static import library used to resolve imports at link time. The
7174 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7175 API at runtime or by by using the GNU ld extension of linking directly to
7176 the DLL without an import library.
7178 See ld/deffilep.y in the binutils sources for the full specification of
7179 other DEF file statements
7181 @cindex creating a DEF file
7182 While linking a shared dll, @command{ld} is able to create a DEF file
7183 with the @samp{--output-def <file>} command line option.
7185 @item Using decorations
7186 @cindex Using decorations
7187 Another way of marking symbols for export is to modify the source code
7188 itself, so that when building the DLL each symbol to be exported is
7192 __declspec(dllexport) int a_variable
7193 __declspec(dllexport) void a_function(int with_args)
7196 All such symbols will be exported from the DLL. If, however,
7197 any of the object files in the DLL contain symbols decorated in
7198 this way, then the normal auto-export behavior is disabled, unless
7199 the @samp{--export-all-symbols} option is also used.
7201 Note that object files that wish to access these symbols must @emph{not}
7202 decorate them with dllexport. Instead, they should use dllimport,
7206 __declspec(dllimport) int a_variable
7207 __declspec(dllimport) void a_function(int with_args)
7210 This complicates the structure of library header files, because
7211 when included by the library itself the header must declare the
7212 variables and functions as dllexport, but when included by client
7213 code the header must declare them as dllimport. There are a number
7214 of idioms that are typically used to do this; often client code can
7215 omit the __declspec() declaration completely. See
7216 @samp{--enable-auto-import} and @samp{automatic data imports} for more
7220 @cindex automatic data imports
7221 @item automatic data imports
7222 The standard Windows dll format supports data imports from dlls only
7223 by adding special decorations (dllimport/dllexport), which let the
7224 compiler produce specific assembler instructions to deal with this
7225 issue. This increases the effort necessary to port existing Un*x
7226 code to these platforms, especially for large
7227 c++ libraries and applications. The auto-import feature, which was
7228 initially provided by Paul Sokolovsky, allows one to omit the
7229 decorations to achieve a behavior that conforms to that on POSIX/Un*x
7230 platforms. This feature is enabled with the @samp{--enable-auto-import}
7231 command-line option, although it is enabled by default on cygwin/mingw.
7232 The @samp{--enable-auto-import} option itself now serves mainly to
7233 suppress any warnings that are ordinarily emitted when linked objects
7234 trigger the feature's use.
7236 auto-import of variables does not always work flawlessly without
7237 additional assistance. Sometimes, you will see this message
7239 "variable '<var>' can't be auto-imported. Please read the
7240 documentation for ld's @code{--enable-auto-import} for details."
7242 The @samp{--enable-auto-import} documentation explains why this error
7243 occurs, and several methods that can be used to overcome this difficulty.
7244 One of these methods is the @emph{runtime pseudo-relocs} feature, described
7247 @cindex runtime pseudo-relocation
7248 For complex variables imported from DLLs (such as structs or classes),
7249 object files typically contain a base address for the variable and an
7250 offset (@emph{addend}) within the variable--to specify a particular
7251 field or public member, for instance. Unfortunately, the runtime loader used
7252 in win32 environments is incapable of fixing these references at runtime
7253 without the additional information supplied by dllimport/dllexport decorations.
7254 The standard auto-import feature described above is unable to resolve these
7257 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
7258 be resolved without error, while leaving the task of adjusting the references
7259 themselves (with their non-zero addends) to specialized code provided by the
7260 runtime environment. Recent versions of the cygwin and mingw environments and
7261 compilers provide this runtime support; older versions do not. However, the
7262 support is only necessary on the developer's platform; the compiled result will
7263 run without error on an older system.
7265 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
7268 @cindex direct linking to a dll
7269 @item direct linking to a dll
7270 The cygwin/mingw ports of @command{ld} support the direct linking,
7271 including data symbols, to a dll without the usage of any import
7272 libraries. This is much faster and uses much less memory than does the
7273 traditional import library method, especially when linking large
7274 libraries or applications. When @command{ld} creates an import lib, each
7275 function or variable exported from the dll is stored in its own bfd, even
7276 though a single bfd could contain many exports. The overhead involved in
7277 storing, loading, and processing so many bfd's is quite large, and explains the
7278 tremendous time, memory, and storage needed to link against particularly
7279 large or complex libraries when using import libs.
7281 Linking directly to a dll uses no extra command-line switches other than
7282 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
7283 of names to match each library. All that is needed from the developer's
7284 perspective is an understanding of this search, in order to force ld to
7285 select the dll instead of an import library.
7288 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
7289 to find, in the first directory of its search path,
7301 before moving on to the next directory in the search path.
7303 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
7304 where @samp{<prefix>} is set by the @command{ld} option
7305 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
7306 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
7309 Other win32-based unix environments, such as mingw or pw32, may use other
7310 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
7311 was originally intended to help avoid name conflicts among dll's built for the
7312 various win32/un*x environments, so that (for example) two versions of a zlib dll
7313 could coexist on the same machine.
7315 The generic cygwin/mingw path layout uses a @samp{bin} directory for
7316 applications and dll's and a @samp{lib} directory for the import
7317 libraries (using cygwin nomenclature):
7323 libxxx.dll.a (in case of dll's)
7324 libxxx.a (in case of static archive)
7327 Linking directly to a dll without using the import library can be
7330 1. Use the dll directly by adding the @samp{bin} path to the link line
7332 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
7335 However, as the dll's often have version numbers appended to their names
7336 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
7337 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
7338 not versioned, and do not have this difficulty.
7340 2. Create a symbolic link from the dll to a file in the @samp{lib}
7341 directory according to the above mentioned search pattern. This
7342 should be used to avoid unwanted changes in the tools needed for
7346 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
7349 Then you can link without any make environment changes.
7352 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
7355 This technique also avoids the version number problems, because the following is
7362 libxxx.dll.a -> ../bin/cygxxx-5.dll
7365 Linking directly to a dll without using an import lib will work
7366 even when auto-import features are exercised, and even when
7367 @samp{--enable-runtime-pseudo-relocs} is used.
7369 Given the improvements in speed and memory usage, one might justifiably
7370 wonder why import libraries are used at all. There are three reasons:
7372 1. Until recently, the link-directly-to-dll functionality did @emph{not}
7373 work with auto-imported data.
7375 2. Sometimes it is necessary to include pure static objects within the
7376 import library (which otherwise contains only bfd's for indirection
7377 symbols that point to the exports of a dll). Again, the import lib
7378 for the cygwin kernel makes use of this ability, and it is not
7379 possible to do this without an import lib.
7381 3. Symbol aliases can only be resolved using an import lib. This is
7382 critical when linking against OS-supplied dll's (eg, the win32 API)
7383 in which symbols are usually exported as undecorated aliases of their
7384 stdcall-decorated assembly names.
7386 So, import libs are not going away. But the ability to replace
7387 true import libs with a simple symbolic link to (or a copy of)
7388 a dll, in many cases, is a useful addition to the suite of tools
7389 binutils makes available to the win32 developer. Given the
7390 massive improvements in memory requirements during linking, storage
7391 requirements, and linking speed, we expect that many developers
7392 will soon begin to use this feature whenever possible.
7394 @item symbol aliasing
7396 @item adding additional names
7397 Sometimes, it is useful to export symbols with additional names.
7398 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
7399 exported as @samp{_foo} by using special directives in the DEF file
7400 when creating the dll. This will affect also the optional created
7401 import library. Consider the following DEF file:
7404 LIBRARY "xyz.dll" BASE=0x61000000
7411 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
7413 Another method for creating a symbol alias is to create it in the
7414 source code using the "weak" attribute:
7417 void foo () @{ /* Do something. */; @}
7418 void _foo () __attribute__ ((weak, alias ("foo")));
7421 See the gcc manual for more information about attributes and weak
7424 @item renaming symbols
7425 Sometimes it is useful to rename exports. For instance, the cygwin
7426 kernel does this regularly. A symbol @samp{_foo} can be exported as
7427 @samp{foo} but not as @samp{_foo} by using special directives in the
7428 DEF file. (This will also affect the import library, if it is
7429 created). In the following example:
7432 LIBRARY "xyz.dll" BASE=0x61000000
7438 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
7442 Note: using a DEF file disables the default auto-export behavior,
7443 unless the @samp{--export-all-symbols} command line option is used.
7444 If, however, you are trying to rename symbols, then you should list
7445 @emph{all} desired exports in the DEF file, including the symbols
7446 that are not being renamed, and do @emph{not} use the
7447 @samp{--export-all-symbols} option. If you list only the
7448 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
7449 to handle the other symbols, then the both the new names @emph{and}
7450 the original names for the renamed symbols will be exported.
7451 In effect, you'd be aliasing those symbols, not renaming them,
7452 which is probably not what you wanted.
7454 @cindex weak externals
7455 @item weak externals
7456 The Windows object format, PE, specifies a form of weak symbols called
7457 weak externals. When a weak symbol is linked and the symbol is not
7458 defined, the weak symbol becomes an alias for some other symbol. There
7459 are three variants of weak externals:
7461 @item Definition is searched for in objects and libraries, historically
7462 called lazy externals.
7463 @item Definition is searched for only in other objects, not in libraries.
7464 This form is not presently implemented.
7465 @item No search; the symbol is an alias. This form is not presently
7468 As a GNU extension, weak symbols that do not specify an alternate symbol
7469 are supported. If the symbol is undefined when linking, the symbol
7470 uses a default value.
7472 @cindex aligned common symbols
7473 @item aligned common symbols
7474 As a GNU extension to the PE file format, it is possible to specify the
7475 desired alignment for a common symbol. This information is conveyed from
7476 the assembler or compiler to the linker by means of GNU-specific commands
7477 carried in the object file's @samp{.drectve} section, which are recognized
7478 by @command{ld} and respected when laying out the common symbols. Native
7479 tools will be able to process object files employing this GNU extension,
7480 but will fail to respect the alignment instructions, and may issue noisy
7481 warnings about unknown linker directives.
7495 @section @code{ld} and Xtensa Processors
7497 @cindex Xtensa processors
7498 The default @command{ld} behavior for Xtensa processors is to interpret
7499 @code{SECTIONS} commands so that lists of explicitly named sections in a
7500 specification with a wildcard file will be interleaved when necessary to
7501 keep literal pools within the range of PC-relative load offsets. For
7502 example, with the command:
7514 @command{ld} may interleave some of the @code{.literal}
7515 and @code{.text} sections from different object files to ensure that the
7516 literal pools are within the range of PC-relative load offsets. A valid
7517 interleaving might place the @code{.literal} sections from an initial
7518 group of files followed by the @code{.text} sections of that group of
7519 files. Then, the @code{.literal} sections from the rest of the files
7520 and the @code{.text} sections from the rest of the files would follow.
7522 @cindex @option{--relax} on Xtensa
7523 @cindex relaxing on Xtensa
7524 Relaxation is enabled by default for the Xtensa version of @command{ld} and
7525 provides two important link-time optimizations. The first optimization
7526 is to combine identical literal values to reduce code size. A redundant
7527 literal will be removed and all the @code{L32R} instructions that use it
7528 will be changed to reference an identical literal, as long as the
7529 location of the replacement literal is within the offset range of all
7530 the @code{L32R} instructions. The second optimization is to remove
7531 unnecessary overhead from assembler-generated ``longcall'' sequences of
7532 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
7533 range of direct @code{CALL@var{n}} instructions.
7535 For each of these cases where an indirect call sequence can be optimized
7536 to a direct call, the linker will change the @code{CALLX@var{n}}
7537 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
7538 instruction, and remove the literal referenced by the @code{L32R}
7539 instruction if it is not used for anything else. Removing the
7540 @code{L32R} instruction always reduces code size but can potentially
7541 hurt performance by changing the alignment of subsequent branch targets.
7542 By default, the linker will always preserve alignments, either by
7543 switching some instructions between 24-bit encodings and the equivalent
7544 density instructions or by inserting a no-op in place of the @code{L32R}
7545 instruction that was removed. If code size is more important than
7546 performance, the @option{--size-opt} option can be used to prevent the
7547 linker from widening density instructions or inserting no-ops, except in
7548 a few cases where no-ops are required for correctness.
7550 The following Xtensa-specific command-line options can be used to
7553 @cindex Xtensa options
7556 When optimizing indirect calls to direct calls, optimize for code size
7557 more than performance. With this option, the linker will not insert
7558 no-ops or widen density instructions to preserve branch target
7559 alignment. There may still be some cases where no-ops are required to
7560 preserve the correctness of the code.
7568 @ifclear SingleFormat
7573 @cindex object file management
7574 @cindex object formats available
7576 The linker accesses object and archive files using the BFD libraries.
7577 These libraries allow the linker to use the same routines to operate on
7578 object files whatever the object file format. A different object file
7579 format can be supported simply by creating a new BFD back end and adding
7580 it to the library. To conserve runtime memory, however, the linker and
7581 associated tools are usually configured to support only a subset of the
7582 object file formats available. You can use @code{objdump -i}
7583 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
7584 list all the formats available for your configuration.
7586 @cindex BFD requirements
7587 @cindex requirements for BFD
7588 As with most implementations, BFD is a compromise between
7589 several conflicting requirements. The major factor influencing
7590 BFD design was efficiency: any time used converting between
7591 formats is time which would not have been spent had BFD not
7592 been involved. This is partly offset by abstraction payback; since
7593 BFD simplifies applications and back ends, more time and care
7594 may be spent optimizing algorithms for a greater speed.
7596 One minor artifact of the BFD solution which you should bear in
7597 mind is the potential for information loss. There are two places where
7598 useful information can be lost using the BFD mechanism: during
7599 conversion and during output. @xref{BFD information loss}.
7602 * BFD outline:: How it works: an outline of BFD
7606 @section How It Works: An Outline of BFD
7607 @cindex opening object files
7608 @include bfdsumm.texi
7611 @node Reporting Bugs
7612 @chapter Reporting Bugs
7613 @cindex bugs in @command{ld}
7614 @cindex reporting bugs in @command{ld}
7616 Your bug reports play an essential role in making @command{ld} reliable.
7618 Reporting a bug may help you by bringing a solution to your problem, or
7619 it may not. But in any case the principal function of a bug report is
7620 to help the entire community by making the next version of @command{ld}
7621 work better. Bug reports are your contribution to the maintenance of
7624 In order for a bug report to serve its purpose, you must include the
7625 information that enables us to fix the bug.
7628 * Bug Criteria:: Have you found a bug?
7629 * Bug Reporting:: How to report bugs
7633 @section Have You Found a Bug?
7634 @cindex bug criteria
7636 If you are not sure whether you have found a bug, here are some guidelines:
7639 @cindex fatal signal
7640 @cindex linker crash
7641 @cindex crash of linker
7643 If the linker gets a fatal signal, for any input whatever, that is a
7644 @command{ld} bug. Reliable linkers never crash.
7646 @cindex error on valid input
7648 If @command{ld} produces an error message for valid input, that is a bug.
7650 @cindex invalid input
7652 If @command{ld} does not produce an error message for invalid input, that
7653 may be a bug. In the general case, the linker can not verify that
7654 object files are correct.
7657 If you are an experienced user of linkers, your suggestions for
7658 improvement of @command{ld} are welcome in any case.
7662 @section How to Report Bugs
7664 @cindex @command{ld} bugs, reporting
7666 A number of companies and individuals offer support for @sc{gnu}
7667 products. If you obtained @command{ld} from a support organization, we
7668 recommend you contact that organization first.
7670 You can find contact information for many support companies and
7671 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
7675 Otherwise, send bug reports for @command{ld} to
7679 The fundamental principle of reporting bugs usefully is this:
7680 @strong{report all the facts}. If you are not sure whether to state a
7681 fact or leave it out, state it!
7683 Often people omit facts because they think they know what causes the
7684 problem and assume that some details do not matter. Thus, you might
7685 assume that the name of a symbol you use in an example does not
7686 matter. Well, probably it does not, but one cannot be sure. Perhaps
7687 the bug is a stray memory reference which happens to fetch from the
7688 location where that name is stored in memory; perhaps, if the name
7689 were different, the contents of that location would fool the linker
7690 into doing the right thing despite the bug. Play it safe and give a
7691 specific, complete example. That is the easiest thing for you to do,
7692 and the most helpful.
7694 Keep in mind that the purpose of a bug report is to enable us to fix
7695 the bug if it is new to us. Therefore, always write your bug reports
7696 on the assumption that the bug has not been reported previously.
7698 Sometimes people give a few sketchy facts and ask, ``Does this ring a
7699 bell?'' This cannot help us fix a bug, so it is basically useless. We
7700 respond by asking for enough details to enable us to investigate.
7701 You might as well expedite matters by sending them to begin with.
7703 To enable us to fix the bug, you should include all these things:
7707 The version of @command{ld}. @command{ld} announces it if you start it with
7708 the @samp{--version} argument.
7710 Without this, we will not know whether there is any point in looking for
7711 the bug in the current version of @command{ld}.
7714 Any patches you may have applied to the @command{ld} source, including any
7715 patches made to the @code{BFD} library.
7718 The type of machine you are using, and the operating system name and
7722 What compiler (and its version) was used to compile @command{ld}---e.g.
7726 The command arguments you gave the linker to link your example and
7727 observe the bug. To guarantee you will not omit something important,
7728 list them all. A copy of the Makefile (or the output from make) is
7731 If we were to try to guess the arguments, we would probably guess wrong
7732 and then we might not encounter the bug.
7735 A complete input file, or set of input files, that will reproduce the
7736 bug. It is generally most helpful to send the actual object files
7737 provided that they are reasonably small. Say no more than 10K. For
7738 bigger files you can either make them available by FTP or HTTP or else
7739 state that you are willing to send the object file(s) to whomever
7740 requests them. (Note - your email will be going to a mailing list, so
7741 we do not want to clog it up with large attachments). But small
7742 attachments are best.
7744 If the source files were assembled using @code{gas} or compiled using
7745 @code{gcc}, then it may be OK to send the source files rather than the
7746 object files. In this case, be sure to say exactly what version of
7747 @code{gas} or @code{gcc} was used to produce the object files. Also say
7748 how @code{gas} or @code{gcc} were configured.
7751 A description of what behavior you observe that you believe is
7752 incorrect. For example, ``It gets a fatal signal.''
7754 Of course, if the bug is that @command{ld} gets a fatal signal, then we
7755 will certainly notice it. But if the bug is incorrect output, we might
7756 not notice unless it is glaringly wrong. You might as well not give us
7757 a chance to make a mistake.
7759 Even if the problem you experience is a fatal signal, you should still
7760 say so explicitly. Suppose something strange is going on, such as, your
7761 copy of @command{ld} is out of sync, or you have encountered a bug in the
7762 C library on your system. (This has happened!) Your copy might crash
7763 and ours would not. If you told us to expect a crash, then when ours
7764 fails to crash, we would know that the bug was not happening for us. If
7765 you had not told us to expect a crash, then we would not be able to draw
7766 any conclusion from our observations.
7769 If you wish to suggest changes to the @command{ld} source, send us context
7770 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
7771 @samp{-p} option. Always send diffs from the old file to the new file.
7772 If you even discuss something in the @command{ld} source, refer to it by
7773 context, not by line number.
7775 The line numbers in our development sources will not match those in your
7776 sources. Your line numbers would convey no useful information to us.
7779 Here are some things that are not necessary:
7783 A description of the envelope of the bug.
7785 Often people who encounter a bug spend a lot of time investigating
7786 which changes to the input file will make the bug go away and which
7787 changes will not affect it.
7789 This is often time consuming and not very useful, because the way we
7790 will find the bug is by running a single example under the debugger
7791 with breakpoints, not by pure deduction from a series of examples.
7792 We recommend that you save your time for something else.
7794 Of course, if you can find a simpler example to report @emph{instead}
7795 of the original one, that is a convenience for us. Errors in the
7796 output will be easier to spot, running under the debugger will take
7797 less time, and so on.
7799 However, simplification is not vital; if you do not want to do this,
7800 report the bug anyway and send us the entire test case you used.
7803 A patch for the bug.
7805 A patch for the bug does help us if it is a good one. But do not omit
7806 the necessary information, such as the test case, on the assumption that
7807 a patch is all we need. We might see problems with your patch and decide
7808 to fix the problem another way, or we might not understand it at all.
7810 Sometimes with a program as complicated as @command{ld} it is very hard to
7811 construct an example that will make the program follow a certain path
7812 through the code. If you do not send us the example, we will not be
7813 able to construct one, so we will not be able to verify that the bug is
7816 And if we cannot understand what bug you are trying to fix, or why your
7817 patch should be an improvement, we will not install it. A test case will
7818 help us to understand.
7821 A guess about what the bug is or what it depends on.
7823 Such guesses are usually wrong. Even we cannot guess right about such
7824 things without first using the debugger to find the facts.
7828 @appendix MRI Compatible Script Files
7829 @cindex MRI compatibility
7830 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
7831 linker, @command{ld} can use MRI compatible linker scripts as an
7832 alternative to the more general-purpose linker scripting language
7833 described in @ref{Scripts}. MRI compatible linker scripts have a much
7834 simpler command set than the scripting language otherwise used with
7835 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
7836 linker commands; these commands are described here.
7838 In general, MRI scripts aren't of much use with the @code{a.out} object
7839 file format, since it only has three sections and MRI scripts lack some
7840 features to make use of them.
7842 You can specify a file containing an MRI-compatible script using the
7843 @samp{-c} command-line option.
7845 Each command in an MRI-compatible script occupies its own line; each
7846 command line starts with the keyword that identifies the command (though
7847 blank lines are also allowed for punctuation). If a line of an
7848 MRI-compatible script begins with an unrecognized keyword, @command{ld}
7849 issues a warning message, but continues processing the script.
7851 Lines beginning with @samp{*} are comments.
7853 You can write these commands using all upper-case letters, or all
7854 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
7855 The following list shows only the upper-case form of each command.
7858 @cindex @code{ABSOLUTE} (MRI)
7859 @item ABSOLUTE @var{secname}
7860 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
7861 Normally, @command{ld} includes in the output file all sections from all
7862 the input files. However, in an MRI-compatible script, you can use the
7863 @code{ABSOLUTE} command to restrict the sections that will be present in
7864 your output program. If the @code{ABSOLUTE} command is used at all in a
7865 script, then only the sections named explicitly in @code{ABSOLUTE}
7866 commands will appear in the linker output. You can still use other
7867 input sections (whatever you select on the command line, or using
7868 @code{LOAD}) to resolve addresses in the output file.
7870 @cindex @code{ALIAS} (MRI)
7871 @item ALIAS @var{out-secname}, @var{in-secname}
7872 Use this command to place the data from input section @var{in-secname}
7873 in a section called @var{out-secname} in the linker output file.
7875 @var{in-secname} may be an integer.
7877 @cindex @code{ALIGN} (MRI)
7878 @item ALIGN @var{secname} = @var{expression}
7879 Align the section called @var{secname} to @var{expression}. The
7880 @var{expression} should be a power of two.
7882 @cindex @code{BASE} (MRI)
7883 @item BASE @var{expression}
7884 Use the value of @var{expression} as the lowest address (other than
7885 absolute addresses) in the output file.
7887 @cindex @code{CHIP} (MRI)
7888 @item CHIP @var{expression}
7889 @itemx CHIP @var{expression}, @var{expression}
7890 This command does nothing; it is accepted only for compatibility.
7892 @cindex @code{END} (MRI)
7894 This command does nothing whatever; it's only accepted for compatibility.
7896 @cindex @code{FORMAT} (MRI)
7897 @item FORMAT @var{output-format}
7898 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
7899 language, but restricted to one of these output formats:
7903 S-records, if @var{output-format} is @samp{S}
7906 IEEE, if @var{output-format} is @samp{IEEE}
7909 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
7913 @cindex @code{LIST} (MRI)
7914 @item LIST @var{anything}@dots{}
7915 Print (to the standard output file) a link map, as produced by the
7916 @command{ld} command-line option @samp{-M}.
7918 The keyword @code{LIST} may be followed by anything on the
7919 same line, with no change in its effect.
7921 @cindex @code{LOAD} (MRI)
7922 @item LOAD @var{filename}
7923 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
7924 Include one or more object file @var{filename} in the link; this has the
7925 same effect as specifying @var{filename} directly on the @command{ld}
7928 @cindex @code{NAME} (MRI)
7929 @item NAME @var{output-name}
7930 @var{output-name} is the name for the program produced by @command{ld}; the
7931 MRI-compatible command @code{NAME} is equivalent to the command-line
7932 option @samp{-o} or the general script language command @code{OUTPUT}.
7934 @cindex @code{ORDER} (MRI)
7935 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
7936 @itemx ORDER @var{secname} @var{secname} @var{secname}
7937 Normally, @command{ld} orders the sections in its output file in the
7938 order in which they first appear in the input files. In an MRI-compatible
7939 script, you can override this ordering with the @code{ORDER} command. The
7940 sections you list with @code{ORDER} will appear first in your output
7941 file, in the order specified.
7943 @cindex @code{PUBLIC} (MRI)
7944 @item PUBLIC @var{name}=@var{expression}
7945 @itemx PUBLIC @var{name},@var{expression}
7946 @itemx PUBLIC @var{name} @var{expression}
7947 Supply a value (@var{expression}) for external symbol
7948 @var{name} used in the linker input files.
7950 @cindex @code{SECT} (MRI)
7951 @item SECT @var{secname}, @var{expression}
7952 @itemx SECT @var{secname}=@var{expression}
7953 @itemx SECT @var{secname} @var{expression}
7954 You can use any of these three forms of the @code{SECT} command to
7955 specify the start address (@var{expression}) for section @var{secname}.
7956 If you have more than one @code{SECT} statement for the same
7957 @var{secname}, only the @emph{first} sets the start address.
7960 @node GNU Free Documentation License
7961 @appendix GNU Free Documentation License
7965 @unnumbered LD Index
7970 % I think something like @@colophon should be in texinfo. In the
7972 \long\def\colophon{\hbox to0pt{}\vfill
7973 \centerline{The body of this manual is set in}
7974 \centerline{\fontname\tenrm,}
7975 \centerline{with headings in {\bf\fontname\tenbf}}
7976 \centerline{and examples in {\tt\fontname\tentt}.}
7977 \centerline{{\it\fontname\tenit\/} and}
7978 \centerline{{\sl\fontname\tensl\/}}
7979 \centerline{are used for emphasis.}\vfill}
7981 % Blame: doc@@cygnus.com, 28mar91.