3 @c Copyright (C) 1991-2018 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
45 @dircategory Software development
47 * Ld: (ld). The GNU linker.
52 This file documents the @sc{gnu} linker LD
53 @ifset VERSION_PACKAGE
54 @value{VERSION_PACKAGE}
56 version @value{VERSION}.
58 Copyright @copyright{} 1991-2018 Free Software Foundation, Inc.
60 Permission is granted to copy, distribute and/or modify this document
61 under the terms of the GNU Free Documentation License, Version 1.3
62 or any later version published by the Free Software Foundation;
63 with no Invariant Sections, with no Front-Cover Texts, and with no
64 Back-Cover Texts. A copy of the license is included in the
65 section entitled ``GNU Free Documentation License''.
69 @setchapternewpage odd
70 @settitle The GNU linker
75 @ifset VERSION_PACKAGE
76 @subtitle @value{VERSION_PACKAGE}
78 @subtitle Version @value{VERSION}
79 @author Steve Chamberlain
80 @author Ian Lance Taylor
85 \hfill Red Hat Inc\par
86 \hfill nickc\@credhat.com, doc\@redhat.com\par
87 \hfill {\it The GNU linker}\par
88 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
90 \global\parindent=0pt % Steve likes it this way.
93 @vskip 0pt plus 1filll
94 @c man begin COPYRIGHT
95 Copyright @copyright{} 1991-2018 Free Software Foundation, Inc.
97 Permission is granted to copy, distribute and/or modify this document
98 under the terms of the GNU Free Documentation License, Version 1.3
99 or any later version published by the Free Software Foundation;
100 with no Invariant Sections, with no Front-Cover Texts, and with no
101 Back-Cover Texts. A copy of the license is included in the
102 section entitled ``GNU Free Documentation License''.
108 @c FIXME: Talk about importance of *order* of args, cmds to linker!
113 This file documents the @sc{gnu} linker ld
114 @ifset VERSION_PACKAGE
115 @value{VERSION_PACKAGE}
117 version @value{VERSION}.
119 This document is distributed under the terms of the GNU Free
120 Documentation License version 1.3. A copy of the license is included
121 in the section entitled ``GNU Free Documentation License''.
124 * Overview:: Overview
125 * Invocation:: Invocation
126 * Scripts:: Linker Scripts
128 * Machine Dependent:: Machine Dependent Features
132 * H8/300:: ld and the H8/300
135 * Renesas:: ld and other Renesas micros
138 * ARM:: ld and the ARM family
141 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
144 * HPPA ELF32:: ld and HPPA 32-bit ELF
147 * M68K:: ld and Motorola 68K family
150 * MIPS:: ld and MIPS family
153 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
156 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
159 * S/390 ELF:: ld and S/390 ELF Support
162 * SPU ELF:: ld and SPU ELF Support
165 * TI COFF:: ld and the TI COFF
168 * Win32:: ld and WIN32 (cygwin/mingw)
171 * Xtensa:: ld and Xtensa Processors
174 @ifclear SingleFormat
177 @c Following blank line required for remaining bug in makeinfo conds/menus
179 * Reporting Bugs:: Reporting Bugs
180 * MRI:: MRI Compatible Script Files
181 * GNU Free Documentation License:: GNU Free Documentation License
182 * LD Index:: LD Index
189 @cindex @sc{gnu} linker
190 @cindex what is this?
193 @c man begin SYNOPSIS
194 ld [@b{options}] @var{objfile} @dots{}
198 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
199 the Info entries for @file{binutils} and
204 @c man begin DESCRIPTION
206 @command{ld} combines a number of object and archive files, relocates
207 their data and ties up symbol references. Usually the last step in
208 compiling a program is to run @command{ld}.
210 @command{ld} accepts Linker Command Language files written in
211 a superset of AT&T's Link Editor Command Language syntax,
212 to provide explicit and total control over the linking process.
216 This man page does not describe the command language; see the
217 @command{ld} entry in @code{info} for full details on the command
218 language and on other aspects of the GNU linker.
221 @ifclear SingleFormat
222 This version of @command{ld} uses the general purpose BFD libraries
223 to operate on object files. This allows @command{ld} to read, combine, and
224 write object files in many different formats---for example, COFF or
225 @code{a.out}. Different formats may be linked together to produce any
226 available kind of object file. @xref{BFD}, for more information.
229 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
230 linkers in providing diagnostic information. Many linkers abandon
231 execution immediately upon encountering an error; whenever possible,
232 @command{ld} continues executing, allowing you to identify other errors
233 (or, in some cases, to get an output file in spite of the error).
240 @c man begin DESCRIPTION
242 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
243 and to be as compatible as possible with other linkers. As a result,
244 you have many choices to control its behavior.
250 * Options:: Command Line Options
251 * Environment:: Environment Variables
255 @section Command Line Options
263 The linker supports a plethora of command-line options, but in actual
264 practice few of them are used in any particular context.
265 @cindex standard Unix system
266 For instance, a frequent use of @command{ld} is to link standard Unix
267 object files on a standard, supported Unix system. On such a system, to
268 link a file @code{hello.o}:
271 ld -o @var{output} /lib/crt0.o hello.o -lc
274 This tells @command{ld} to produce a file called @var{output} as the
275 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
276 the library @code{libc.a}, which will come from the standard search
277 directories. (See the discussion of the @samp{-l} option below.)
279 Some of the command-line options to @command{ld} may be specified at any
280 point in the command line. However, options which refer to files, such
281 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
282 which the option appears in the command line, relative to the object
283 files and other file options. Repeating non-file options with a
284 different argument will either have no further effect, or override prior
285 occurrences (those further to the left on the command line) of that
286 option. Options which may be meaningfully specified more than once are
287 noted in the descriptions below.
290 Non-option arguments are object files or archives which are to be linked
291 together. They may follow, precede, or be mixed in with command-line
292 options, except that an object file argument may not be placed between
293 an option and its argument.
295 Usually the linker is invoked with at least one object file, but you can
296 specify other forms of binary input files using @samp{-l}, @samp{-R},
297 and the script command language. If @emph{no} binary input files at all
298 are specified, the linker does not produce any output, and issues the
299 message @samp{No input files}.
301 If the linker cannot recognize the format of an object file, it will
302 assume that it is a linker script. A script specified in this way
303 augments the main linker script used for the link (either the default
304 linker script or the one specified by using @samp{-T}). This feature
305 permits the linker to link against a file which appears to be an object
306 or an archive, but actually merely defines some symbol values, or uses
307 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
308 script in this way merely augments the main linker script, with the
309 extra commands placed after the main script; use the @samp{-T} option
310 to replace the default linker script entirely, but note the effect of
311 the @code{INSERT} command. @xref{Scripts}.
313 For options whose names are a single letter,
314 option arguments must either follow the option letter without intervening
315 whitespace, or be given as separate arguments immediately following the
316 option that requires them.
318 For options whose names are multiple letters, either one dash or two can
319 precede the option name; for example, @samp{-trace-symbol} and
320 @samp{--trace-symbol} are equivalent. Note---there is one exception to
321 this rule. Multiple letter options that start with a lower case 'o' can
322 only be preceded by two dashes. This is to reduce confusion with the
323 @samp{-o} option. So for example @samp{-omagic} sets the output file
324 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
327 Arguments to multiple-letter options must either be separated from the
328 option name by an equals sign, or be given as separate arguments
329 immediately following the option that requires them. For example,
330 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
331 Unique abbreviations of the names of multiple-letter options are
334 Note---if the linker is being invoked indirectly, via a compiler driver
335 (e.g. @samp{gcc}) then all the linker command line options should be
336 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
337 compiler driver) like this:
340 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
343 This is important, because otherwise the compiler driver program may
344 silently drop the linker options, resulting in a bad link. Confusion
345 may also arise when passing options that require values through a
346 driver, as the use of a space between option and argument acts as
347 a separator, and causes the driver to pass only the option to the linker
348 and the argument to the compiler. In this case, it is simplest to use
349 the joined forms of both single- and multiple-letter options, such as:
352 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
355 Here is a table of the generic command line switches accepted by the GNU
359 @include at-file.texi
361 @kindex -a @var{keyword}
362 @item -a @var{keyword}
363 This option is supported for HP/UX compatibility. The @var{keyword}
364 argument must be one of the strings @samp{archive}, @samp{shared}, or
365 @samp{default}. @samp{-aarchive} is functionally equivalent to
366 @samp{-Bstatic}, and the other two keywords are functionally equivalent
367 to @samp{-Bdynamic}. This option may be used any number of times.
369 @kindex --audit @var{AUDITLIB}
370 @item --audit @var{AUDITLIB}
371 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
372 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
373 specified in the library. If specified multiple times @code{DT_AUDIT}
374 will contain a colon separated list of audit interfaces to use. If the linker
375 finds an object with an audit entry while searching for shared libraries,
376 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
377 This option is only meaningful on ELF platforms supporting the rtld-audit
380 @ifclear SingleFormat
381 @cindex binary input format
382 @kindex -b @var{format}
383 @kindex --format=@var{format}
386 @item -b @var{input-format}
387 @itemx --format=@var{input-format}
388 @command{ld} may be configured to support more than one kind of object
389 file. If your @command{ld} is configured this way, you can use the
390 @samp{-b} option to specify the binary format for input object files
391 that follow this option on the command line. Even when @command{ld} is
392 configured to support alternative object formats, you don't usually need
393 to specify this, as @command{ld} should be configured to expect as a
394 default input format the most usual format on each machine.
395 @var{input-format} is a text string, the name of a particular format
396 supported by the BFD libraries. (You can list the available binary
397 formats with @samp{objdump -i}.)
400 You may want to use this option if you are linking files with an unusual
401 binary format. You can also use @samp{-b} to switch formats explicitly (when
402 linking object files of different formats), by including
403 @samp{-b @var{input-format}} before each group of object files in a
406 The default format is taken from the environment variable
411 You can also define the input format from a script, using the command
414 see @ref{Format Commands}.
418 @kindex -c @var{MRI-cmdfile}
419 @kindex --mri-script=@var{MRI-cmdfile}
420 @cindex compatibility, MRI
421 @item -c @var{MRI-commandfile}
422 @itemx --mri-script=@var{MRI-commandfile}
423 For compatibility with linkers produced by MRI, @command{ld} accepts script
424 files written in an alternate, restricted command language, described in
426 @ref{MRI,,MRI Compatible Script Files}.
429 the MRI Compatible Script Files section of GNU ld documentation.
431 Introduce MRI script files with
432 the option @samp{-c}; use the @samp{-T} option to run linker
433 scripts written in the general-purpose @command{ld} scripting language.
434 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
435 specified by any @samp{-L} options.
437 @cindex common allocation
444 These three options are equivalent; multiple forms are supported for
445 compatibility with other linkers. They assign space to common symbols
446 even if a relocatable output file is specified (with @samp{-r}). The
447 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
448 @xref{Miscellaneous Commands}.
450 @kindex --depaudit @var{AUDITLIB}
451 @kindex -P @var{AUDITLIB}
452 @item --depaudit @var{AUDITLIB}
453 @itemx -P @var{AUDITLIB}
454 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
455 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
456 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
457 will contain a colon separated list of audit interfaces to use. This
458 option is only meaningful on ELF platforms supporting the rtld-audit interface.
459 The -P option is provided for Solaris compatibility.
461 @cindex entry point, from command line
462 @kindex -e @var{entry}
463 @kindex --entry=@var{entry}
465 @itemx --entry=@var{entry}
466 Use @var{entry} as the explicit symbol for beginning execution of your
467 program, rather than the default entry point. If there is no symbol
468 named @var{entry}, the linker will try to parse @var{entry} as a number,
469 and use that as the entry address (the number will be interpreted in
470 base 10; you may use a leading @samp{0x} for base 16, or a leading
471 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
472 and other ways of specifying the entry point.
474 @kindex --exclude-libs
475 @item --exclude-libs @var{lib},@var{lib},...
476 Specifies a list of archive libraries from which symbols should not be automatically
477 exported. The library names may be delimited by commas or colons. Specifying
478 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
479 automatic export. This option is available only for the i386 PE targeted
480 port of the linker and for ELF targeted ports. For i386 PE, symbols
481 explicitly listed in a .def file are still exported, regardless of this
482 option. For ELF targeted ports, symbols affected by this option will
483 be treated as hidden.
485 @kindex --exclude-modules-for-implib
486 @item --exclude-modules-for-implib @var{module},@var{module},...
487 Specifies a list of object files or archive members, from which symbols
488 should not be automatically exported, but which should be copied wholesale
489 into the import library being generated during the link. The module names
490 may be delimited by commas or colons, and must match exactly the filenames
491 used by @command{ld} to open the files; for archive members, this is simply
492 the member name, but for object files the name listed must include and
493 match precisely any path used to specify the input file on the linker's
494 command-line. This option is available only for the i386 PE targeted port
495 of the linker. Symbols explicitly listed in a .def file are still exported,
496 regardless of this option.
498 @cindex dynamic symbol table
500 @kindex --export-dynamic
501 @kindex --no-export-dynamic
503 @itemx --export-dynamic
504 @itemx --no-export-dynamic
505 When creating a dynamically linked executable, using the @option{-E}
506 option or the @option{--export-dynamic} option causes the linker to add
507 all symbols to the dynamic symbol table. The dynamic symbol table is the
508 set of symbols which are visible from dynamic objects at run time.
510 If you do not use either of these options (or use the
511 @option{--no-export-dynamic} option to restore the default behavior), the
512 dynamic symbol table will normally contain only those symbols which are
513 referenced by some dynamic object mentioned in the link.
515 If you use @code{dlopen} to load a dynamic object which needs to refer
516 back to the symbols defined by the program, rather than some other
517 dynamic object, then you will probably need to use this option when
518 linking the program itself.
520 You can also use the dynamic list to control what symbols should
521 be added to the dynamic symbol table if the output format supports it.
522 See the description of @samp{--dynamic-list}.
524 Note that this option is specific to ELF targeted ports. PE targets
525 support a similar function to export all symbols from a DLL or EXE; see
526 the description of @samp{--export-all-symbols} below.
528 @ifclear SingleFormat
529 @cindex big-endian objects
533 Link big-endian objects. This affects the default output format.
535 @cindex little-endian objects
538 Link little-endian objects. This affects the default output format.
541 @kindex -f @var{name}
542 @kindex --auxiliary=@var{name}
544 @itemx --auxiliary=@var{name}
545 When creating an ELF shared object, set the internal DT_AUXILIARY field
546 to the specified name. This tells the dynamic linker that the symbol
547 table of the shared object should be used as an auxiliary filter on the
548 symbol table of the shared object @var{name}.
550 If you later link a program against this filter object, then, when you
551 run the program, the dynamic linker will see the DT_AUXILIARY field. If
552 the dynamic linker resolves any symbols from the filter object, it will
553 first check whether there is a definition in the shared object
554 @var{name}. If there is one, it will be used instead of the definition
555 in the filter object. The shared object @var{name} need not exist.
556 Thus the shared object @var{name} may be used to provide an alternative
557 implementation of certain functions, perhaps for debugging or for
558 machine specific performance.
560 This option may be specified more than once. The DT_AUXILIARY entries
561 will be created in the order in which they appear on the command line.
563 @kindex -F @var{name}
564 @kindex --filter=@var{name}
566 @itemx --filter=@var{name}
567 When creating an ELF shared object, set the internal DT_FILTER field to
568 the specified name. This tells the dynamic linker that the symbol table
569 of the shared object which is being created should be used as a filter
570 on the symbol table of the shared object @var{name}.
572 If you later link a program against this filter object, then, when you
573 run the program, the dynamic linker will see the DT_FILTER field. The
574 dynamic linker will resolve symbols according to the symbol table of the
575 filter object as usual, but it will actually link to the definitions
576 found in the shared object @var{name}. Thus the filter object can be
577 used to select a subset of the symbols provided by the object
580 Some older linkers used the @option{-F} option throughout a compilation
581 toolchain for specifying object-file format for both input and output
583 @ifclear SingleFormat
584 The @sc{gnu} linker uses other mechanisms for this purpose: the
585 @option{-b}, @option{--format}, @option{--oformat} options, the
586 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
587 environment variable.
589 The @sc{gnu} linker will ignore the @option{-F} option when not
590 creating an ELF shared object.
592 @cindex finalization function
593 @kindex -fini=@var{name}
594 @item -fini=@var{name}
595 When creating an ELF executable or shared object, call NAME when the
596 executable or shared object is unloaded, by setting DT_FINI to the
597 address of the function. By default, the linker uses @code{_fini} as
598 the function to call.
602 Ignored. Provided for compatibility with other tools.
604 @kindex -G @var{value}
605 @kindex --gpsize=@var{value}
608 @itemx --gpsize=@var{value}
609 Set the maximum size of objects to be optimized using the GP register to
610 @var{size}. This is only meaningful for object file formats such as
611 MIPS ELF that support putting large and small objects into different
612 sections. This is ignored for other object file formats.
614 @cindex runtime library name
615 @kindex -h @var{name}
616 @kindex -soname=@var{name}
618 @itemx -soname=@var{name}
619 When creating an ELF shared object, set the internal DT_SONAME field to
620 the specified name. When an executable is linked with a shared object
621 which has a DT_SONAME field, then when the executable is run the dynamic
622 linker will attempt to load the shared object specified by the DT_SONAME
623 field rather than the using the file name given to the linker.
626 @cindex incremental link
628 Perform an incremental link (same as option @samp{-r}).
630 @cindex initialization function
631 @kindex -init=@var{name}
632 @item -init=@var{name}
633 When creating an ELF executable or shared object, call NAME when the
634 executable or shared object is loaded, by setting DT_INIT to the address
635 of the function. By default, the linker uses @code{_init} as the
638 @cindex archive files, from cmd line
639 @kindex -l @var{namespec}
640 @kindex --library=@var{namespec}
641 @item -l @var{namespec}
642 @itemx --library=@var{namespec}
643 Add the archive or object file specified by @var{namespec} to the
644 list of files to link. This option may be used any number of times.
645 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
646 will search the library path for a file called @var{filename}, otherwise it
647 will search the library path for a file called @file{lib@var{namespec}.a}.
649 On systems which support shared libraries, @command{ld} may also search for
650 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
651 and SunOS systems, @command{ld} will search a directory for a library
652 called @file{lib@var{namespec}.so} before searching for one called
653 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
654 indicates a shared library.) Note that this behavior does not apply
655 to @file{:@var{filename}}, which always specifies a file called
658 The linker will search an archive only once, at the location where it is
659 specified on the command line. If the archive defines a symbol which
660 was undefined in some object which appeared before the archive on the
661 command line, the linker will include the appropriate file(s) from the
662 archive. However, an undefined symbol in an object appearing later on
663 the command line will not cause the linker to search the archive again.
665 See the @option{-(} option for a way to force the linker to search
666 archives multiple times.
668 You may list the same archive multiple times on the command line.
671 This type of archive searching is standard for Unix linkers. However,
672 if you are using @command{ld} on AIX, note that it is different from the
673 behaviour of the AIX linker.
676 @cindex search directory, from cmd line
678 @kindex --library-path=@var{dir}
679 @item -L @var{searchdir}
680 @itemx --library-path=@var{searchdir}
681 Add path @var{searchdir} to the list of paths that @command{ld} will search
682 for archive libraries and @command{ld} control scripts. You may use this
683 option any number of times. The directories are searched in the order
684 in which they are specified on the command line. Directories specified
685 on the command line are searched before the default directories. All
686 @option{-L} options apply to all @option{-l} options, regardless of the
687 order in which the options appear. @option{-L} options do not affect
688 how @command{ld} searches for a linker script unless @option{-T}
691 If @var{searchdir} begins with @code{=} or @code{$SYSROOT}, then this
692 prefix will be replaced by the @dfn{sysroot prefix}, controlled by the
693 @samp{--sysroot} option, or specified when the linker is configured.
696 The default set of paths searched (without being specified with
697 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
698 some cases also on how it was configured. @xref{Environment}.
701 The paths can also be specified in a link script with the
702 @code{SEARCH_DIR} command. Directories specified this way are searched
703 at the point in which the linker script appears in the command line.
706 @kindex -m @var{emulation}
707 @item -m @var{emulation}
708 Emulate the @var{emulation} linker. You can list the available
709 emulations with the @samp{--verbose} or @samp{-V} options.
711 If the @samp{-m} option is not used, the emulation is taken from the
712 @code{LDEMULATION} environment variable, if that is defined.
714 Otherwise, the default emulation depends upon how the linker was
722 Print a link map to the standard output. A link map provides
723 information about the link, including the following:
727 Where object files are mapped into memory.
729 How common symbols are allocated.
731 All archive members included in the link, with a mention of the symbol
732 which caused the archive member to be brought in.
734 The values assigned to symbols.
736 Note - symbols whose values are computed by an expression which
737 involves a reference to a previous value of the same symbol may not
738 have correct result displayed in the link map. This is because the
739 linker discards intermediate results and only retains the final value
740 of an expression. Under such circumstances the linker will display
741 the final value enclosed by square brackets. Thus for example a
742 linker script containing:
750 will produce the following output in the link map if the @option{-M}
755 [0x0000000c] foo = (foo * 0x4)
756 [0x0000000c] foo = (foo + 0x8)
759 See @ref{Expressions} for more information about expressions in linker
764 @cindex read-only text
769 Turn off page alignment of sections, and disable linking against shared
770 libraries. If the output format supports Unix style magic numbers,
771 mark the output as @code{NMAGIC}.
775 @cindex read/write from cmd line
779 Set the text and data sections to be readable and writable. Also, do
780 not page-align the data segment, and disable linking against shared
781 libraries. If the output format supports Unix style magic numbers,
782 mark the output as @code{OMAGIC}. Note: Although a writable text section
783 is allowed for PE-COFF targets, it does not conform to the format
784 specification published by Microsoft.
789 This option negates most of the effects of the @option{-N} option. It
790 sets the text section to be read-only, and forces the data segment to
791 be page-aligned. Note - this option does not enable linking against
792 shared libraries. Use @option{-Bdynamic} for this.
794 @kindex -o @var{output}
795 @kindex --output=@var{output}
796 @cindex naming the output file
797 @item -o @var{output}
798 @itemx --output=@var{output}
799 Use @var{output} as the name for the program produced by @command{ld}; if this
800 option is not specified, the name @file{a.out} is used by default. The
801 script command @code{OUTPUT} can also specify the output file name.
803 @kindex -O @var{level}
804 @cindex generating optimized output
806 If @var{level} is a numeric values greater than zero @command{ld} optimizes
807 the output. This might take significantly longer and therefore probably
808 should only be enabled for the final binary. At the moment this
809 option only affects ELF shared library generation. Future releases of
810 the linker may make more use of this option. Also currently there is
811 no difference in the linker's behaviour for different non-zero values
812 of this option. Again this may change with future releases.
814 @kindex -plugin @var{name}
815 @item -plugin @var{name}
816 Involve a plugin in the linking process. The @var{name} parameter is
817 the absolute filename of the plugin. Usually this parameter is
818 automatically added by the complier, when using link time
819 optimization, but users can also add their own plugins if they so
822 Note that the location of the compiler originated plugins is different
823 from the place where the @command{ar}, @command{nm} and
824 @command{ranlib} programs search for their plugins. In order for
825 those commands to make use of a compiler based plugin it must first be
826 copied into the @file{$@{libdir@}/bfd-plugins} directory. All gcc
827 based linker plugins are backward compatible, so it is sufficient to
828 just copy in the newest one.
831 @cindex push state governing input file handling
833 The @option{--push-state} allows to preserve the current state of the
834 flags which govern the input file handling so that they can all be
835 restored with one corresponding @option{--pop-state} option.
837 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
838 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
839 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
840 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
841 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
842 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
844 One target for this option are specifications for @file{pkg-config}. When
845 used with the @option{--libs} option all possibly needed libraries are
846 listed and then possibly linked with all the time. It is better to return
847 something as follows:
850 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
854 @cindex pop state governing input file handling
856 Undoes the effect of --push-state, restores the previous values of the
857 flags governing input file handling.
860 @kindex --emit-relocs
861 @cindex retain relocations in final executable
864 Leave relocation sections and contents in fully linked executables.
865 Post link analysis and optimization tools may need this information in
866 order to perform correct modifications of executables. This results
867 in larger executables.
869 This option is currently only supported on ELF platforms.
871 @kindex --force-dynamic
872 @cindex forcing the creation of dynamic sections
873 @item --force-dynamic
874 Force the output file to have dynamic sections. This option is specific
878 @cindex relocatable output
880 @kindex --relocatable
883 Generate relocatable output---i.e., generate an output file that can in
884 turn serve as input to @command{ld}. This is often called @dfn{partial
885 linking}. As a side effect, in environments that support standard Unix
886 magic numbers, this option also sets the output file's magic number to
888 @c ; see @option{-N}.
889 If this option is not specified, an absolute file is produced. When
890 linking C++ programs, this option @emph{will not} resolve references to
891 constructors; to do that, use @samp{-Ur}.
893 When an input file does not have the same format as the output file,
894 partial linking is only supported if that input file does not contain any
895 relocations. Different output formats can have further restrictions; for
896 example some @code{a.out}-based formats do not support partial linking
897 with input files in other formats at all.
899 This option does the same thing as @samp{-i}.
901 @kindex -R @var{file}
902 @kindex --just-symbols=@var{file}
903 @cindex symbol-only input
904 @item -R @var{filename}
905 @itemx --just-symbols=@var{filename}
906 Read symbol names and their addresses from @var{filename}, but do not
907 relocate it or include it in the output. This allows your output file
908 to refer symbolically to absolute locations of memory defined in other
909 programs. You may use this option more than once.
911 For compatibility with other ELF linkers, if the @option{-R} option is
912 followed by a directory name, rather than a file name, it is treated as
913 the @option{-rpath} option.
917 @cindex strip all symbols
920 Omit all symbol information from the output file.
923 @kindex --strip-debug
924 @cindex strip debugger symbols
927 Omit debugger symbol information (but not all symbols) from the output file.
929 @kindex --strip-discarded
930 @kindex --no-strip-discarded
931 @item --strip-discarded
932 @itemx --no-strip-discarded
933 Omit (or do not omit) global symbols defined in discarded sections.
938 @cindex input files, displaying
941 Print the names of the input files as @command{ld} processes them.
943 @kindex -T @var{script}
944 @kindex --script=@var{script}
946 @item -T @var{scriptfile}
947 @itemx --script=@var{scriptfile}
948 Use @var{scriptfile} as the linker script. This script replaces
949 @command{ld}'s default linker script (rather than adding to it), so
950 @var{commandfile} must specify everything necessary to describe the
951 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
952 the current directory, @code{ld} looks for it in the directories
953 specified by any preceding @samp{-L} options. Multiple @samp{-T}
956 @kindex -dT @var{script}
957 @kindex --default-script=@var{script}
959 @item -dT @var{scriptfile}
960 @itemx --default-script=@var{scriptfile}
961 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
963 This option is similar to the @option{--script} option except that
964 processing of the script is delayed until after the rest of the
965 command line has been processed. This allows options placed after the
966 @option{--default-script} option on the command line to affect the
967 behaviour of the linker script, which can be important when the linker
968 command line cannot be directly controlled by the user. (eg because
969 the command line is being constructed by another tool, such as
972 @kindex -u @var{symbol}
973 @kindex --undefined=@var{symbol}
974 @cindex undefined symbol
975 @item -u @var{symbol}
976 @itemx --undefined=@var{symbol}
977 Force @var{symbol} to be entered in the output file as an undefined
978 symbol. Doing this may, for example, trigger linking of additional
979 modules from standard libraries. @samp{-u} may be repeated with
980 different option arguments to enter additional undefined symbols. This
981 option is equivalent to the @code{EXTERN} linker script command.
983 If this option is being used to force additional modules to be pulled
984 into the link, and if it is an error for the symbol to remain
985 undefined, then the option @option{--require-defined} should be used
988 @kindex --require-defined=@var{symbol}
989 @cindex symbols, require defined
990 @cindex defined symbol
991 @item --require-defined=@var{symbol}
992 Require that @var{symbol} is defined in the output file. This option
993 is the same as option @option{--undefined} except that if @var{symbol}
994 is not defined in the output file then the linker will issue an error
995 and exit. The same effect can be achieved in a linker script by using
996 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
997 can be used multiple times to require additional symbols.
1000 @cindex constructors
1002 For anything other than C++ programs, this option is equivalent to
1003 @samp{-r}: it generates relocatable output---i.e., an output file that can in
1004 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
1005 @emph{does} resolve references to constructors, unlike @samp{-r}.
1006 It does not work to use @samp{-Ur} on files that were themselves linked
1007 with @samp{-Ur}; once the constructor table has been built, it cannot
1008 be added to. Use @samp{-Ur} only for the last partial link, and
1009 @samp{-r} for the others.
1011 @kindex --orphan-handling=@var{MODE}
1012 @cindex orphan sections
1013 @cindex sections, orphan
1014 @item --orphan-handling=@var{MODE}
1015 Control how orphan sections are handled. An orphan section is one not
1016 specifically mentioned in a linker script. @xref{Orphan Sections}.
1018 @var{MODE} can have any of the following values:
1022 Orphan sections are placed into a suitable output section following
1023 the strategy described in @ref{Orphan Sections}. The option
1024 @samp{--unique} also affects how sections are placed.
1027 All orphan sections are discarded, by placing them in the
1028 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1031 The linker will place the orphan section as for @code{place} and also
1035 The linker will exit with an error if any orphan section is found.
1038 The default if @samp{--orphan-handling} is not given is @code{place}.
1040 @kindex --unique[=@var{SECTION}]
1041 @item --unique[=@var{SECTION}]
1042 Creates a separate output section for every input section matching
1043 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1044 missing, for every orphan input section. An orphan section is one not
1045 specifically mentioned in a linker script. You may use this option
1046 multiple times on the command line; It prevents the normal merging of
1047 input sections with the same name, overriding output section assignments
1057 Display the version number for @command{ld}. The @option{-V} option also
1058 lists the supported emulations.
1061 @kindex --discard-all
1062 @cindex deleting local symbols
1064 @itemx --discard-all
1065 Delete all local symbols.
1068 @kindex --discard-locals
1069 @cindex local symbols, deleting
1071 @itemx --discard-locals
1072 Delete all temporary local symbols. (These symbols start with
1073 system-specific local label prefixes, typically @samp{.L} for ELF systems
1074 or @samp{L} for traditional a.out systems.)
1076 @kindex -y @var{symbol}
1077 @kindex --trace-symbol=@var{symbol}
1078 @cindex symbol tracing
1079 @item -y @var{symbol}
1080 @itemx --trace-symbol=@var{symbol}
1081 Print the name of each linked file in which @var{symbol} appears. This
1082 option may be given any number of times. On many systems it is necessary
1083 to prepend an underscore.
1085 This option is useful when you have an undefined symbol in your link but
1086 don't know where the reference is coming from.
1088 @kindex -Y @var{path}
1090 Add @var{path} to the default library search path. This option exists
1091 for Solaris compatibility.
1093 @kindex -z @var{keyword}
1094 @item -z @var{keyword}
1095 The recognized keywords are:
1099 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1101 @item call-nop=prefix-addr
1102 @itemx call-nop=suffix-nop
1103 @itemx call-nop=prefix-@var{byte}
1104 @itemx call-nop=suffix-@var{byte}
1105 Specify the 1-byte @code{NOP} padding when transforming indirect call
1106 to a locally defined function, foo, via its GOT slot.
1107 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1108 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1109 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1110 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1111 Supported for i386 and x86_64.
1115 Combine multiple dynamic relocation sections and sort to improve
1116 dynamic symbol lookup caching. Do not do this if @samp{nocombreloc}.
1120 Generate common symbols with STT_COMMON type during a relocatable
1121 link. Use STT_OBJECT type if @samp{nocommon}.
1123 @item common-page-size=@var{value}
1124 Set the page size most commonly used to @var{value}. Memory image
1125 layout will be optimized to minimize memory pages if the system is
1126 using pages of this size.
1129 Report unresolved symbol references from regular object files. This
1130 is done even if the linker is creating a non-symbolic shared library.
1131 This option is the inverse of @samp{-z undefs}.
1133 @item dynamic-undefined-weak
1134 @itemx nodynamic-undefined-weak
1135 Make undefined weak symbols dynamic when building a dynamic object,
1136 if they are referenced from a regular object file and not forced local
1137 by symbol visibility or versioning. Do not make them dynamic if
1138 @samp{nodynamic-undefined-weak}. If neither option is given, a target
1139 may default to either option being in force, or make some other
1140 selection of undefined weak symbols dynamic. Not all targets support
1144 Marks the object as requiring executable stack.
1147 This option is only meaningful when building a shared object. It makes
1148 the symbols defined by this shared object available for symbol resolution
1149 of subsequently loaded libraries.
1152 This option is only meaningful when building a dynamic executable.
1153 This option marks the executable as requiring global auditing by
1154 setting the @code{DF_1_GLOBAUDIT} bit in the @code{DT_FLAGS_1} dynamic
1155 tag. Global auditing requires that any auditing library defined via
1156 the @option{--depaudit} or @option{-P} command line options be run for
1157 all dynamic objects loaded by the application.
1160 Generate Intel Indirect Branch Tracking (IBT) enabled PLT entries.
1161 Supported for Linux/i386 and Linux/x86_64.
1164 Generate GNU_PROPERTY_X86_FEATURE_1_IBT in .note.gnu.property section
1165 to indicate compatibility with IBT. This also implies @option{ibtplt}.
1166 Supported for Linux/i386 and Linux/x86_64.
1169 This option is only meaningful when building a shared object.
1170 It marks the object so that its runtime initialization will occur
1171 before the runtime initialization of any other objects brought into
1172 the process at the same time. Similarly the runtime finalization of
1173 the object will occur after the runtime finalization of any other
1177 Specify that the dynamic loader should modify its symbol search order
1178 so that symbols in this shared library interpose all other shared
1179 libraries not so marked.
1182 When generating an executable or shared library, mark it to tell the
1183 dynamic linker to defer function call resolution to the point when
1184 the function is called (lazy binding), rather than at load time.
1185 Lazy binding is the default.
1188 Specify that the object's filters be processed immediately at runtime.
1190 @item max-page-size=@var{value}
1191 Set the maximum memory page size supported to @var{value}.
1194 Allow multiple definitions.
1197 Disable linker generated .dynbss variables used in place of variables
1198 defined in shared libraries. May result in dynamic text relocations.
1201 Specify that the dynamic loader search for dependencies of this object
1202 should ignore any default library search paths.
1205 Specify that the object shouldn't be unloaded at runtime.
1208 Specify that the object is not available to @code{dlopen}.
1211 Specify that the object can not be dumped by @code{dldump}.
1214 Marks the object as not requiring executable stack.
1216 @item noextern-protected-data
1217 Don't treat protected data symbols as external when building a shared
1218 library. This option overrides the linker backend default. It can be
1219 used to work around incorrect relocations against protected data symbols
1220 generated by compiler. Updates on protected data symbols by another
1221 module aren't visible to the resulting shared library. Supported for
1224 @item noreloc-overflow
1225 Disable relocation overflow check. This can be used to disable
1226 relocation overflow check if there will be no dynamic relocation
1227 overflow at run-time. Supported for x86_64.
1230 When generating an executable or shared library, mark it to tell the
1231 dynamic linker to resolve all symbols when the program is started, or
1232 when the shared library is loaded by dlopen, instead of deferring
1233 function call resolution to the point when the function is first
1237 Specify that the object requires @samp{$ORIGIN} handling in paths.
1241 Create an ELF @code{PT_GNU_RELRO} segment header in the object. This
1242 specifies a memory segment that should be made read-only after
1243 relocation, if supported. Specifying @samp{common-page-size} smaller
1244 than the system page size will render this protection ineffective.
1245 Don't create an ELF @code{PT_GNU_RELRO} segment if @samp{norelro}.
1248 @itemx noseparate-code
1249 Create separate code @code{PT_LOAD} segment header in the object. This
1250 specifies a memory segment that should contain only instructions and must
1251 be in wholly disjoint pages from any other data. Don't create separate
1252 code @code{PT_LOAD} segment if @samp{noseparate-code} is used.
1255 Generate GNU_PROPERTY_X86_FEATURE_1_SHSTK in .note.gnu.property section
1256 to indicate compatibility with Intel Shadow Stack. Supported for
1257 Linux/i386 and Linux/x86_64.
1259 @item stack-size=@var{value}
1260 Specify a stack size for an ELF @code{PT_GNU_STACK} segment.
1261 Specifying zero will override any default non-zero sized
1262 @code{PT_GNU_STACK} segment creation.
1267 Report an error if DT_TEXTREL is set, i.e., if the binary has dynamic
1268 relocations in read-only sections. Don't report an error if
1269 @samp{notext} or @samp{textoff}.
1272 Do not report unresolved symbol references from regular object files,
1273 either when creating an executable, or when creating a shared library.
1274 This option is the inverse of @samp{-z defs}.
1278 Other keywords are ignored for Solaris compatibility.
1281 @cindex groups of archives
1282 @item -( @var{archives} -)
1283 @itemx --start-group @var{archives} --end-group
1284 The @var{archives} should be a list of archive files. They may be
1285 either explicit file names, or @samp{-l} options.
1287 The specified archives are searched repeatedly until no new undefined
1288 references are created. Normally, an archive is searched only once in
1289 the order that it is specified on the command line. If a symbol in that
1290 archive is needed to resolve an undefined symbol referred to by an
1291 object in an archive that appears later on the command line, the linker
1292 would not be able to resolve that reference. By grouping the archives,
1293 they all be searched repeatedly until all possible references are
1296 Using this option has a significant performance cost. It is best to use
1297 it only when there are unavoidable circular references between two or
1300 @kindex --accept-unknown-input-arch
1301 @kindex --no-accept-unknown-input-arch
1302 @item --accept-unknown-input-arch
1303 @itemx --no-accept-unknown-input-arch
1304 Tells the linker to accept input files whose architecture cannot be
1305 recognised. The assumption is that the user knows what they are doing
1306 and deliberately wants to link in these unknown input files. This was
1307 the default behaviour of the linker, before release 2.14. The default
1308 behaviour from release 2.14 onwards is to reject such input files, and
1309 so the @samp{--accept-unknown-input-arch} option has been added to
1310 restore the old behaviour.
1313 @kindex --no-as-needed
1315 @itemx --no-as-needed
1316 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1317 on the command line after the @option{--as-needed} option. Normally
1318 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1319 on the command line, regardless of whether the library is actually
1320 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1321 emitted for a library that @emph{at that point in the link} satisfies a
1322 non-weak undefined symbol reference from a regular object file or, if
1323 the library is not found in the DT_NEEDED lists of other needed libraries, a
1324 non-weak undefined symbol reference from another needed dynamic library.
1325 Object files or libraries appearing on the command line @emph{after}
1326 the library in question do not affect whether the library is seen as
1327 needed. This is similar to the rules for extraction of object files
1328 from archives. @option{--no-as-needed} restores the default behaviour.
1330 @kindex --add-needed
1331 @kindex --no-add-needed
1333 @itemx --no-add-needed
1334 These two options have been deprecated because of the similarity of
1335 their names to the @option{--as-needed} and @option{--no-as-needed}
1336 options. They have been replaced by @option{--copy-dt-needed-entries}
1337 and @option{--no-copy-dt-needed-entries}.
1339 @kindex -assert @var{keyword}
1340 @item -assert @var{keyword}
1341 This option is ignored for SunOS compatibility.
1345 @kindex -call_shared
1349 Link against dynamic libraries. This is only meaningful on platforms
1350 for which shared libraries are supported. This option is normally the
1351 default on such platforms. The different variants of this option are
1352 for compatibility with various systems. You may use this option
1353 multiple times on the command line: it affects library searching for
1354 @option{-l} options which follow it.
1358 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1359 section. This causes the runtime linker to handle lookups in this
1360 object and its dependencies to be performed only inside the group.
1361 @option{--unresolved-symbols=report-all} is implied. This option is
1362 only meaningful on ELF platforms which support shared libraries.
1372 Do not link against shared libraries. This is only meaningful on
1373 platforms for which shared libraries are supported. The different
1374 variants of this option are for compatibility with various systems. You
1375 may use this option multiple times on the command line: it affects
1376 library searching for @option{-l} options which follow it. This
1377 option also implies @option{--unresolved-symbols=report-all}. This
1378 option can be used with @option{-shared}. Doing so means that a
1379 shared library is being created but that all of the library's external
1380 references must be resolved by pulling in entries from static
1385 When creating a shared library, bind references to global symbols to the
1386 definition within the shared library, if any. Normally, it is possible
1387 for a program linked against a shared library to override the definition
1388 within the shared library. This option can also be used with the
1389 @option{--export-dynamic} option, when creating a position independent
1390 executable, to bind references to global symbols to the definition within
1391 the executable. This option is only meaningful on ELF platforms which
1392 support shared libraries and position independent executables.
1394 @kindex -Bsymbolic-functions
1395 @item -Bsymbolic-functions
1396 When creating a shared library, bind references to global function
1397 symbols to the definition within the shared library, if any.
1398 This option can also be used with the @option{--export-dynamic} option,
1399 when creating a position independent executable, to bind references
1400 to global function symbols to the definition within the executable.
1401 This option is only meaningful on ELF platforms which support shared
1402 libraries and position independent executables.
1404 @kindex --dynamic-list=@var{dynamic-list-file}
1405 @item --dynamic-list=@var{dynamic-list-file}
1406 Specify the name of a dynamic list file to the linker. This is
1407 typically used when creating shared libraries to specify a list of
1408 global symbols whose references shouldn't be bound to the definition
1409 within the shared library, or creating dynamically linked executables
1410 to specify a list of symbols which should be added to the symbol table
1411 in the executable. This option is only meaningful on ELF platforms
1412 which support shared libraries.
1414 The format of the dynamic list is the same as the version node without
1415 scope and node name. See @ref{VERSION} for more information.
1417 @kindex --dynamic-list-data
1418 @item --dynamic-list-data
1419 Include all global data symbols to the dynamic list.
1421 @kindex --dynamic-list-cpp-new
1422 @item --dynamic-list-cpp-new
1423 Provide the builtin dynamic list for C++ operator new and delete. It
1424 is mainly useful for building shared libstdc++.
1426 @kindex --dynamic-list-cpp-typeinfo
1427 @item --dynamic-list-cpp-typeinfo
1428 Provide the builtin dynamic list for C++ runtime type identification.
1430 @kindex --check-sections
1431 @kindex --no-check-sections
1432 @item --check-sections
1433 @itemx --no-check-sections
1434 Asks the linker @emph{not} to check section addresses after they have
1435 been assigned to see if there are any overlaps. Normally the linker will
1436 perform this check, and if it finds any overlaps it will produce
1437 suitable error messages. The linker does know about, and does make
1438 allowances for sections in overlays. The default behaviour can be
1439 restored by using the command line switch @option{--check-sections}.
1440 Section overlap is not usually checked for relocatable links. You can
1441 force checking in that case by using the @option{--check-sections}
1444 @kindex --copy-dt-needed-entries
1445 @kindex --no-copy-dt-needed-entries
1446 @item --copy-dt-needed-entries
1447 @itemx --no-copy-dt-needed-entries
1448 This option affects the treatment of dynamic libraries referred to
1449 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1450 command line. Normally the linker won't add a DT_NEEDED tag to the
1451 output binary for each library mentioned in a DT_NEEDED tag in an
1452 input dynamic library. With @option{--copy-dt-needed-entries}
1453 specified on the command line however any dynamic libraries that
1454 follow it will have their DT_NEEDED entries added. The default
1455 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1457 This option also has an effect on the resolution of symbols in dynamic
1458 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1459 mentioned on the command line will be recursively searched, following
1460 their DT_NEEDED tags to other libraries, in order to resolve symbols
1461 required by the output binary. With the default setting however
1462 the searching of dynamic libraries that follow it will stop with the
1463 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1466 @cindex cross reference table
1469 Output a cross reference table. If a linker map file is being
1470 generated, the cross reference table is printed to the map file.
1471 Otherwise, it is printed on the standard output.
1473 The format of the table is intentionally simple, so that it may be
1474 easily processed by a script if necessary. The symbols are printed out,
1475 sorted by name. For each symbol, a list of file names is given. If the
1476 symbol is defined, the first file listed is the location of the
1477 definition. If the symbol is defined as a common value then any files
1478 where this happens appear next. Finally any files that reference the
1481 @cindex common allocation
1482 @kindex --no-define-common
1483 @item --no-define-common
1484 This option inhibits the assignment of addresses to common symbols.
1485 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1486 @xref{Miscellaneous Commands}.
1488 The @samp{--no-define-common} option allows decoupling
1489 the decision to assign addresses to Common symbols from the choice
1490 of the output file type; otherwise a non-Relocatable output type
1491 forces assigning addresses to Common symbols.
1492 Using @samp{--no-define-common} allows Common symbols that are referenced
1493 from a shared library to be assigned addresses only in the main program.
1494 This eliminates the unused duplicate space in the shared library,
1495 and also prevents any possible confusion over resolving to the wrong
1496 duplicate when there are many dynamic modules with specialized search
1497 paths for runtime symbol resolution.
1499 @cindex group allocation in linker script
1500 @cindex section groups
1502 @kindex --force-group-allocation
1503 @item --force-group-allocation
1504 This option causes the linker to place section group members like
1505 normal input sections, and to delete the section groups. This is the
1506 default behaviour for a final link but this option can be used to
1507 change the behaviour of a relocatable link (@samp{-r}). The script
1508 command @code{FORCE_GROUP_ALLOCATION} has the same
1509 effect. @xref{Miscellaneous Commands}.
1511 @cindex symbols, from command line
1512 @kindex --defsym=@var{symbol}=@var{exp}
1513 @item --defsym=@var{symbol}=@var{expression}
1514 Create a global symbol in the output file, containing the absolute
1515 address given by @var{expression}. You may use this option as many
1516 times as necessary to define multiple symbols in the command line. A
1517 limited form of arithmetic is supported for the @var{expression} in this
1518 context: you may give a hexadecimal constant or the name of an existing
1519 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1520 constants or symbols. If you need more elaborate expressions, consider
1521 using the linker command language from a script (@pxref{Assignments}).
1522 @emph{Note:} there should be no white space between @var{symbol}, the
1523 equals sign (``@key{=}''), and @var{expression}.
1525 @cindex demangling, from command line
1526 @kindex --demangle[=@var{style}]
1527 @kindex --no-demangle
1528 @item --demangle[=@var{style}]
1529 @itemx --no-demangle
1530 These options control whether to demangle symbol names in error messages
1531 and other output. When the linker is told to demangle, it tries to
1532 present symbol names in a readable fashion: it strips leading
1533 underscores if they are used by the object file format, and converts C++
1534 mangled symbol names into user readable names. Different compilers have
1535 different mangling styles. The optional demangling style argument can be used
1536 to choose an appropriate demangling style for your compiler. The linker will
1537 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1538 is set. These options may be used to override the default.
1540 @cindex dynamic linker, from command line
1541 @kindex -I@var{file}
1542 @kindex --dynamic-linker=@var{file}
1544 @itemx --dynamic-linker=@var{file}
1545 Set the name of the dynamic linker. This is only meaningful when
1546 generating dynamically linked ELF executables. The default dynamic
1547 linker is normally correct; don't use this unless you know what you are
1550 @kindex --no-dynamic-linker
1551 @item --no-dynamic-linker
1552 When producing an executable file, omit the request for a dynamic
1553 linker to be used at load-time. This is only meaningful for ELF
1554 executables that contain dynamic relocations, and usually requires
1555 entry point code that is capable of processing these relocations.
1557 @kindex --embedded-relocs
1558 @item --embedded-relocs
1559 This option is similar to the @option{--emit-relocs} option except
1560 that the relocs are stored in a target specific section. This option
1561 is only supported by the @samp{BFIN}, @samp{CR16} and @emph{M68K}
1564 @kindex --fatal-warnings
1565 @kindex --no-fatal-warnings
1566 @item --fatal-warnings
1567 @itemx --no-fatal-warnings
1568 Treat all warnings as errors. The default behaviour can be restored
1569 with the option @option{--no-fatal-warnings}.
1571 @kindex --force-exe-suffix
1572 @item --force-exe-suffix
1573 Make sure that an output file has a .exe suffix.
1575 If a successfully built fully linked output file does not have a
1576 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1577 the output file to one of the same name with a @code{.exe} suffix. This
1578 option is useful when using unmodified Unix makefiles on a Microsoft
1579 Windows host, since some versions of Windows won't run an image unless
1580 it ends in a @code{.exe} suffix.
1582 @kindex --gc-sections
1583 @kindex --no-gc-sections
1584 @cindex garbage collection
1586 @itemx --no-gc-sections
1587 Enable garbage collection of unused input sections. It is ignored on
1588 targets that do not support this option. The default behaviour (of not
1589 performing this garbage collection) can be restored by specifying
1590 @samp{--no-gc-sections} on the command line. Note that garbage
1591 collection for COFF and PE format targets is supported, but the
1592 implementation is currently considered to be experimental.
1594 @samp{--gc-sections} decides which input sections are used by
1595 examining symbols and relocations. The section containing the entry
1596 symbol and all sections containing symbols undefined on the
1597 command-line will be kept, as will sections containing symbols
1598 referenced by dynamic objects. Note that when building shared
1599 libraries, the linker must assume that any visible symbol is
1600 referenced. Once this initial set of sections has been determined,
1601 the linker recursively marks as used any section referenced by their
1602 relocations. See @samp{--entry} and @samp{--undefined}.
1604 This option can be set when doing a partial link (enabled with option
1605 @samp{-r}). In this case the root of symbols kept must be explicitly
1606 specified either by an @samp{--entry} or @samp{--undefined} option or by
1607 a @code{ENTRY} command in the linker script.
1609 @kindex --print-gc-sections
1610 @kindex --no-print-gc-sections
1611 @cindex garbage collection
1612 @item --print-gc-sections
1613 @itemx --no-print-gc-sections
1614 List all sections removed by garbage collection. The listing is
1615 printed on stderr. This option is only effective if garbage
1616 collection has been enabled via the @samp{--gc-sections}) option. The
1617 default behaviour (of not listing the sections that are removed) can
1618 be restored by specifying @samp{--no-print-gc-sections} on the command
1621 @kindex --gc-keep-exported
1622 @cindex garbage collection
1623 @item --gc-keep-exported
1624 When @samp{--gc-sections} is enabled, this option prevents garbage
1625 collection of unused input sections that contain global symbols having
1626 default or protected visibility. This option is intended to be used for
1627 executables where unreferenced sections would otherwise be garbage
1628 collected regardless of the external visibility of contained symbols.
1629 Note that this option has no effect when linking shared objects since
1630 it is already the default behaviour. This option is only supported for
1633 @kindex --print-output-format
1634 @cindex output format
1635 @item --print-output-format
1636 Print the name of the default output format (perhaps influenced by
1637 other command-line options). This is the string that would appear
1638 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1640 @kindex --print-memory-usage
1641 @cindex memory usage
1642 @item --print-memory-usage
1643 Print used size, total size and used size of memory regions created with
1644 the @ref{MEMORY} command. This is useful on embedded targets to have a
1645 quick view of amount of free memory. The format of the output has one
1646 headline and one line per region. It is both human readable and easily
1647 parsable by tools. Here is an example of an output:
1650 Memory region Used Size Region Size %age Used
1651 ROM: 256 KB 1 MB 25.00%
1652 RAM: 32 B 2 GB 0.00%
1659 Print a summary of the command-line options on the standard output and exit.
1661 @kindex --target-help
1663 Print a summary of all target specific options on the standard output and exit.
1665 @kindex -Map=@var{mapfile}
1666 @item -Map=@var{mapfile}
1667 Print a link map to the file @var{mapfile}. See the description of the
1668 @option{-M} option, above.
1670 @cindex memory usage
1671 @kindex --no-keep-memory
1672 @item --no-keep-memory
1673 @command{ld} normally optimizes for speed over memory usage by caching the
1674 symbol tables of input files in memory. This option tells @command{ld} to
1675 instead optimize for memory usage, by rereading the symbol tables as
1676 necessary. This may be required if @command{ld} runs out of memory space
1677 while linking a large executable.
1679 @kindex --no-undefined
1682 @item --no-undefined
1684 Report unresolved symbol references from regular object files. This
1685 is done even if the linker is creating a non-symbolic shared library.
1686 The switch @option{--[no-]allow-shlib-undefined} controls the
1687 behaviour for reporting unresolved references found in shared
1688 libraries being linked in.
1690 The effects of this option can be reverted by using @code{-z undefs}.
1692 @kindex --allow-multiple-definition
1694 @item --allow-multiple-definition
1696 Normally when a symbol is defined multiple times, the linker will
1697 report a fatal error. These options allow multiple definitions and the
1698 first definition will be used.
1700 @kindex --allow-shlib-undefined
1701 @kindex --no-allow-shlib-undefined
1702 @item --allow-shlib-undefined
1703 @itemx --no-allow-shlib-undefined
1704 Allows or disallows undefined symbols in shared libraries.
1705 This switch is similar to @option{--no-undefined} except that it
1706 determines the behaviour when the undefined symbols are in a
1707 shared library rather than a regular object file. It does not affect
1708 how undefined symbols in regular object files are handled.
1710 The default behaviour is to report errors for any undefined symbols
1711 referenced in shared libraries if the linker is being used to create
1712 an executable, but to allow them if the linker is being used to create
1715 The reasons for allowing undefined symbol references in shared
1716 libraries specified at link time are that:
1720 A shared library specified at link time may not be the same as the one
1721 that is available at load time, so the symbol might actually be
1722 resolvable at load time.
1724 There are some operating systems, eg BeOS and HPPA, where undefined
1725 symbols in shared libraries are normal.
1727 The BeOS kernel for example patches shared libraries at load time to
1728 select whichever function is most appropriate for the current
1729 architecture. This is used, for example, to dynamically select an
1730 appropriate memset function.
1733 @kindex --no-undefined-version
1734 @item --no-undefined-version
1735 Normally when a symbol has an undefined version, the linker will ignore
1736 it. This option disallows symbols with undefined version and a fatal error
1737 will be issued instead.
1739 @kindex --default-symver
1740 @item --default-symver
1741 Create and use a default symbol version (the soname) for unversioned
1744 @kindex --default-imported-symver
1745 @item --default-imported-symver
1746 Create and use a default symbol version (the soname) for unversioned
1749 @kindex --no-warn-mismatch
1750 @item --no-warn-mismatch
1751 Normally @command{ld} will give an error if you try to link together input
1752 files that are mismatched for some reason, perhaps because they have
1753 been compiled for different processors or for different endiannesses.
1754 This option tells @command{ld} that it should silently permit such possible
1755 errors. This option should only be used with care, in cases when you
1756 have taken some special action that ensures that the linker errors are
1759 @kindex --no-warn-search-mismatch
1760 @item --no-warn-search-mismatch
1761 Normally @command{ld} will give a warning if it finds an incompatible
1762 library during a library search. This option silences the warning.
1764 @kindex --no-whole-archive
1765 @item --no-whole-archive
1766 Turn off the effect of the @option{--whole-archive} option for subsequent
1769 @cindex output file after errors
1770 @kindex --noinhibit-exec
1771 @item --noinhibit-exec
1772 Retain the executable output file whenever it is still usable.
1773 Normally, the linker will not produce an output file if it encounters
1774 errors during the link process; it exits without writing an output file
1775 when it issues any error whatsoever.
1779 Only search library directories explicitly specified on the
1780 command line. Library directories specified in linker scripts
1781 (including linker scripts specified on the command line) are ignored.
1783 @ifclear SingleFormat
1784 @kindex --oformat=@var{output-format}
1785 @item --oformat=@var{output-format}
1786 @command{ld} may be configured to support more than one kind of object
1787 file. If your @command{ld} is configured this way, you can use the
1788 @samp{--oformat} option to specify the binary format for the output
1789 object file. Even when @command{ld} is configured to support alternative
1790 object formats, you don't usually need to specify this, as @command{ld}
1791 should be configured to produce as a default output format the most
1792 usual format on each machine. @var{output-format} is a text string, the
1793 name of a particular format supported by the BFD libraries. (You can
1794 list the available binary formats with @samp{objdump -i}.) The script
1795 command @code{OUTPUT_FORMAT} can also specify the output format, but
1796 this option overrides it. @xref{BFD}.
1799 @kindex --out-implib
1800 @item --out-implib @var{file}
1801 Create an import library in @var{file} corresponding to the executable
1802 the linker is generating (eg. a DLL or ELF program). This import
1803 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
1804 may be used to link clients against the generated executable; this
1805 behaviour makes it possible to skip a separate import library creation
1806 step (eg. @code{dlltool} for DLLs). This option is only available for
1807 the i386 PE and ELF targetted ports of the linker.
1810 @kindex --pic-executable
1812 @itemx --pic-executable
1813 @cindex position independent executables
1814 Create a position independent executable. This is currently only supported on
1815 ELF platforms. Position independent executables are similar to shared
1816 libraries in that they are relocated by the dynamic linker to the virtual
1817 address the OS chooses for them (which can vary between invocations). Like
1818 normal dynamically linked executables they can be executed and symbols
1819 defined in the executable cannot be overridden by shared libraries.
1823 This option is ignored for Linux compatibility.
1827 This option is ignored for SVR4 compatibility.
1830 @cindex synthesizing linker
1831 @cindex relaxing addressing modes
1835 An option with machine dependent effects.
1837 This option is only supported on a few targets.
1840 @xref{H8/300,,@command{ld} and the H8/300}.
1843 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1846 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1849 @xref{Nios II,,@command{ld} and the Altera Nios II}.
1852 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1855 On some platforms the @samp{--relax} option performs target specific,
1856 global optimizations that become possible when the linker resolves
1857 addressing in the program, such as relaxing address modes,
1858 synthesizing new instructions, selecting shorter version of current
1859 instructions, and combining constant values.
1861 On some platforms these link time global optimizations may make symbolic
1862 debugging of the resulting executable impossible.
1864 This is known to be the case for the Matsushita MN10200 and MN10300
1865 family of processors.
1869 On platforms where this is not supported, @samp{--relax} is accepted,
1873 On platforms where @samp{--relax} is accepted the option
1874 @samp{--no-relax} can be used to disable the feature.
1876 @cindex retaining specified symbols
1877 @cindex stripping all but some symbols
1878 @cindex symbols, retaining selectively
1879 @kindex --retain-symbols-file=@var{filename}
1880 @item --retain-symbols-file=@var{filename}
1881 Retain @emph{only} the symbols listed in the file @var{filename},
1882 discarding all others. @var{filename} is simply a flat file, with one
1883 symbol name per line. This option is especially useful in environments
1887 where a large global symbol table is accumulated gradually, to conserve
1890 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1891 or symbols needed for relocations.
1893 You may only specify @samp{--retain-symbols-file} once in the command
1894 line. It overrides @samp{-s} and @samp{-S}.
1897 @item -rpath=@var{dir}
1898 @cindex runtime library search path
1899 @kindex -rpath=@var{dir}
1900 Add a directory to the runtime library search path. This is used when
1901 linking an ELF executable with shared objects. All @option{-rpath}
1902 arguments are concatenated and passed to the runtime linker, which uses
1903 them to locate shared objects at runtime. The @option{-rpath} option is
1904 also used when locating shared objects which are needed by shared
1905 objects explicitly included in the link; see the description of the
1906 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1907 ELF executable, the contents of the environment variable
1908 @code{LD_RUN_PATH} will be used if it is defined.
1910 The @option{-rpath} option may also be used on SunOS. By default, on
1911 SunOS, the linker will form a runtime search path out of all the
1912 @option{-L} options it is given. If a @option{-rpath} option is used, the
1913 runtime search path will be formed exclusively using the @option{-rpath}
1914 options, ignoring the @option{-L} options. This can be useful when using
1915 gcc, which adds many @option{-L} options which may be on NFS mounted
1918 For compatibility with other ELF linkers, if the @option{-R} option is
1919 followed by a directory name, rather than a file name, it is treated as
1920 the @option{-rpath} option.
1924 @cindex link-time runtime library search path
1925 @kindex -rpath-link=@var{dir}
1926 @item -rpath-link=@var{dir}
1927 When using ELF or SunOS, one shared library may require another. This
1928 happens when an @code{ld -shared} link includes a shared library as one
1931 When the linker encounters such a dependency when doing a non-shared,
1932 non-relocatable link, it will automatically try to locate the required
1933 shared library and include it in the link, if it is not included
1934 explicitly. In such a case, the @option{-rpath-link} option
1935 specifies the first set of directories to search. The
1936 @option{-rpath-link} option may specify a sequence of directory names
1937 either by specifying a list of names separated by colons, or by
1938 appearing multiple times.
1940 The tokens @var{$ORIGIN} and @var{$LIB} can appear in these search
1941 directories. They will be replaced by the full path to the directory
1942 containing the program or shared object in the case of @var{$ORIGIN}
1943 and either @samp{lib} - for 32-bit binaries - or @samp{lib64} - for
1944 64-bit binaries - in the case of @var{$LIB}.
1946 The alternative form of these tokens - @var{$@{ORIGIN@}} and
1947 @var{$@{LIB@}} can also be used. The token @var{$PLATFORM} is not
1950 This option should be used with caution as it overrides the search path
1951 that may have been hard compiled into a shared library. In such a case it
1952 is possible to use unintentionally a different search path than the
1953 runtime linker would do.
1955 The linker uses the following search paths to locate required shared
1959 Any directories specified by @option{-rpath-link} options.
1961 Any directories specified by @option{-rpath} options. The difference
1962 between @option{-rpath} and @option{-rpath-link} is that directories
1963 specified by @option{-rpath} options are included in the executable and
1964 used at runtime, whereas the @option{-rpath-link} option is only effective
1965 at link time. Searching @option{-rpath} in this way is only supported
1966 by native linkers and cross linkers which have been configured with
1967 the @option{--with-sysroot} option.
1969 On an ELF system, for native linkers, if the @option{-rpath} and
1970 @option{-rpath-link} options were not used, search the contents of the
1971 environment variable @code{LD_RUN_PATH}.
1973 On SunOS, if the @option{-rpath} option was not used, search any
1974 directories specified using @option{-L} options.
1976 For a native linker, search the contents of the environment
1977 variable @code{LD_LIBRARY_PATH}.
1979 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1980 @code{DT_RPATH} of a shared library are searched for shared
1981 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1982 @code{DT_RUNPATH} entries exist.
1984 The default directories, normally @file{/lib} and @file{/usr/lib}.
1986 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1987 exists, the list of directories found in that file.
1990 If the required shared library is not found, the linker will issue a
1991 warning and continue with the link.
1998 @cindex shared libraries
1999 Create a shared library. This is currently only supported on ELF, XCOFF
2000 and SunOS platforms. On SunOS, the linker will automatically create a
2001 shared library if the @option{-e} option is not used and there are
2002 undefined symbols in the link.
2004 @kindex --sort-common
2006 @itemx --sort-common=ascending
2007 @itemx --sort-common=descending
2008 This option tells @command{ld} to sort the common symbols by alignment in
2009 ascending or descending order when it places them in the appropriate output
2010 sections. The symbol alignments considered are sixteen-byte or larger,
2011 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
2012 between symbols due to alignment constraints. If no sorting order is
2013 specified, then descending order is assumed.
2015 @kindex --sort-section=name
2016 @item --sort-section=name
2017 This option will apply @code{SORT_BY_NAME} to all wildcard section
2018 patterns in the linker script.
2020 @kindex --sort-section=alignment
2021 @item --sort-section=alignment
2022 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
2023 patterns in the linker script.
2025 @kindex --spare-dynamic-tags
2026 @item --spare-dynamic-tags=@var{count}
2027 This option specifies the number of empty slots to leave in the
2028 .dynamic section of ELF shared objects. Empty slots may be needed by
2029 post processing tools, such as the prelinker. The default is 5.
2031 @kindex --split-by-file
2032 @item --split-by-file[=@var{size}]
2033 Similar to @option{--split-by-reloc} but creates a new output section for
2034 each input file when @var{size} is reached. @var{size} defaults to a
2035 size of 1 if not given.
2037 @kindex --split-by-reloc
2038 @item --split-by-reloc[=@var{count}]
2039 Tries to creates extra sections in the output file so that no single
2040 output section in the file contains more than @var{count} relocations.
2041 This is useful when generating huge relocatable files for downloading into
2042 certain real time kernels with the COFF object file format; since COFF
2043 cannot represent more than 65535 relocations in a single section. Note
2044 that this will fail to work with object file formats which do not
2045 support arbitrary sections. The linker will not split up individual
2046 input sections for redistribution, so if a single input section contains
2047 more than @var{count} relocations one output section will contain that
2048 many relocations. @var{count} defaults to a value of 32768.
2052 Compute and display statistics about the operation of the linker, such
2053 as execution time and memory usage.
2055 @kindex --sysroot=@var{directory}
2056 @item --sysroot=@var{directory}
2057 Use @var{directory} as the location of the sysroot, overriding the
2058 configure-time default. This option is only supported by linkers
2059 that were configured using @option{--with-sysroot}.
2063 This is used by COFF/PE based targets to create a task-linked object
2064 file where all of the global symbols have been converted to statics.
2066 @kindex --traditional-format
2067 @cindex traditional format
2068 @item --traditional-format
2069 For some targets, the output of @command{ld} is different in some ways from
2070 the output of some existing linker. This switch requests @command{ld} to
2071 use the traditional format instead.
2074 For example, on SunOS, @command{ld} combines duplicate entries in the
2075 symbol string table. This can reduce the size of an output file with
2076 full debugging information by over 30 percent. Unfortunately, the SunOS
2077 @code{dbx} program can not read the resulting program (@code{gdb} has no
2078 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
2079 combine duplicate entries.
2081 @kindex --section-start=@var{sectionname}=@var{org}
2082 @item --section-start=@var{sectionname}=@var{org}
2083 Locate a section in the output file at the absolute
2084 address given by @var{org}. You may use this option as many
2085 times as necessary to locate multiple sections in the command
2087 @var{org} must be a single hexadecimal integer;
2088 for compatibility with other linkers, you may omit the leading
2089 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
2090 should be no white space between @var{sectionname}, the equals
2091 sign (``@key{=}''), and @var{org}.
2093 @kindex -Tbss=@var{org}
2094 @kindex -Tdata=@var{org}
2095 @kindex -Ttext=@var{org}
2096 @cindex segment origins, cmd line
2097 @item -Tbss=@var{org}
2098 @itemx -Tdata=@var{org}
2099 @itemx -Ttext=@var{org}
2100 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2101 @code{.text} as the @var{sectionname}.
2103 @kindex -Ttext-segment=@var{org}
2104 @item -Ttext-segment=@var{org}
2105 @cindex text segment origin, cmd line
2106 When creating an ELF executable, it will set the address of the first
2107 byte of the text segment.
2109 @kindex -Trodata-segment=@var{org}
2110 @item -Trodata-segment=@var{org}
2111 @cindex rodata segment origin, cmd line
2112 When creating an ELF executable or shared object for a target where
2113 the read-only data is in its own segment separate from the executable
2114 text, it will set the address of the first byte of the read-only data segment.
2116 @kindex -Tldata-segment=@var{org}
2117 @item -Tldata-segment=@var{org}
2118 @cindex ldata segment origin, cmd line
2119 When creating an ELF executable or shared object for x86-64 medium memory
2120 model, it will set the address of the first byte of the ldata segment.
2122 @kindex --unresolved-symbols
2123 @item --unresolved-symbols=@var{method}
2124 Determine how to handle unresolved symbols. There are four possible
2125 values for @samp{method}:
2129 Do not report any unresolved symbols.
2132 Report all unresolved symbols. This is the default.
2134 @item ignore-in-object-files
2135 Report unresolved symbols that are contained in shared libraries, but
2136 ignore them if they come from regular object files.
2138 @item ignore-in-shared-libs
2139 Report unresolved symbols that come from regular object files, but
2140 ignore them if they come from shared libraries. This can be useful
2141 when creating a dynamic binary and it is known that all the shared
2142 libraries that it should be referencing are included on the linker's
2146 The behaviour for shared libraries on their own can also be controlled
2147 by the @option{--[no-]allow-shlib-undefined} option.
2149 Normally the linker will generate an error message for each reported
2150 unresolved symbol but the option @option{--warn-unresolved-symbols}
2151 can change this to a warning.
2153 @kindex --verbose[=@var{NUMBER}]
2154 @cindex verbose[=@var{NUMBER}]
2156 @itemx --verbose[=@var{NUMBER}]
2157 Display the version number for @command{ld} and list the linker emulations
2158 supported. Display which input files can and cannot be opened. Display
2159 the linker script being used by the linker. If the optional @var{NUMBER}
2160 argument > 1, plugin symbol status will also be displayed.
2162 @kindex --version-script=@var{version-scriptfile}
2163 @cindex version script, symbol versions
2164 @item --version-script=@var{version-scriptfile}
2165 Specify the name of a version script to the linker. This is typically
2166 used when creating shared libraries to specify additional information
2167 about the version hierarchy for the library being created. This option
2168 is only fully supported on ELF platforms which support shared libraries;
2169 see @ref{VERSION}. It is partially supported on PE platforms, which can
2170 use version scripts to filter symbol visibility in auto-export mode: any
2171 symbols marked @samp{local} in the version script will not be exported.
2174 @kindex --warn-common
2175 @cindex warnings, on combining symbols
2176 @cindex combining symbols, warnings on
2178 Warn when a common symbol is combined with another common symbol or with
2179 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2180 but linkers on some other operating systems do not. This option allows
2181 you to find potential problems from combining global symbols.
2182 Unfortunately, some C libraries use this practice, so you may get some
2183 warnings about symbols in the libraries as well as in your programs.
2185 There are three kinds of global symbols, illustrated here by C examples:
2189 A definition, which goes in the initialized data section of the output
2193 An undefined reference, which does not allocate space.
2194 There must be either a definition or a common symbol for the
2198 A common symbol. If there are only (one or more) common symbols for a
2199 variable, it goes in the uninitialized data area of the output file.
2200 The linker merges multiple common symbols for the same variable into a
2201 single symbol. If they are of different sizes, it picks the largest
2202 size. The linker turns a common symbol into a declaration, if there is
2203 a definition of the same variable.
2206 The @samp{--warn-common} option can produce five kinds of warnings.
2207 Each warning consists of a pair of lines: the first describes the symbol
2208 just encountered, and the second describes the previous symbol
2209 encountered with the same name. One or both of the two symbols will be
2214 Turning a common symbol into a reference, because there is already a
2215 definition for the symbol.
2217 @var{file}(@var{section}): warning: common of `@var{symbol}'
2218 overridden by definition
2219 @var{file}(@var{section}): warning: defined here
2223 Turning a common symbol into a reference, because a later definition for
2224 the symbol is encountered. This is the same as the previous case,
2225 except that the symbols are encountered in a different order.
2227 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2229 @var{file}(@var{section}): warning: common is here
2233 Merging a common symbol with a previous same-sized common symbol.
2235 @var{file}(@var{section}): warning: multiple common
2237 @var{file}(@var{section}): warning: previous common is here
2241 Merging a common symbol with a previous larger common symbol.
2243 @var{file}(@var{section}): warning: common of `@var{symbol}'
2244 overridden by larger common
2245 @var{file}(@var{section}): warning: larger common is here
2249 Merging a common symbol with a previous smaller common symbol. This is
2250 the same as the previous case, except that the symbols are
2251 encountered in a different order.
2253 @var{file}(@var{section}): warning: common of `@var{symbol}'
2254 overriding smaller common
2255 @var{file}(@var{section}): warning: smaller common is here
2259 @kindex --warn-constructors
2260 @item --warn-constructors
2261 Warn if any global constructors are used. This is only useful for a few
2262 object file formats. For formats like COFF or ELF, the linker can not
2263 detect the use of global constructors.
2265 @kindex --warn-multiple-gp
2266 @item --warn-multiple-gp
2267 Warn if multiple global pointer values are required in the output file.
2268 This is only meaningful for certain processors, such as the Alpha.
2269 Specifically, some processors put large-valued constants in a special
2270 section. A special register (the global pointer) points into the middle
2271 of this section, so that constants can be loaded efficiently via a
2272 base-register relative addressing mode. Since the offset in
2273 base-register relative mode is fixed and relatively small (e.g., 16
2274 bits), this limits the maximum size of the constant pool. Thus, in
2275 large programs, it is often necessary to use multiple global pointer
2276 values in order to be able to address all possible constants. This
2277 option causes a warning to be issued whenever this case occurs.
2280 @cindex warnings, on undefined symbols
2281 @cindex undefined symbols, warnings on
2283 Only warn once for each undefined symbol, rather than once per module
2286 @kindex --warn-section-align
2287 @cindex warnings, on section alignment
2288 @cindex section alignment, warnings on
2289 @item --warn-section-align
2290 Warn if the address of an output section is changed because of
2291 alignment. Typically, the alignment will be set by an input section.
2292 The address will only be changed if it not explicitly specified; that
2293 is, if the @code{SECTIONS} command does not specify a start address for
2294 the section (@pxref{SECTIONS}).
2296 @kindex --warn-shared-textrel
2297 @item --warn-shared-textrel
2298 Warn if the linker adds a DT_TEXTREL to a shared object.
2300 @kindex --warn-alternate-em
2301 @item --warn-alternate-em
2302 Warn if an object has alternate ELF machine code.
2304 @kindex --warn-unresolved-symbols
2305 @item --warn-unresolved-symbols
2306 If the linker is going to report an unresolved symbol (see the option
2307 @option{--unresolved-symbols}) it will normally generate an error.
2308 This option makes it generate a warning instead.
2310 @kindex --error-unresolved-symbols
2311 @item --error-unresolved-symbols
2312 This restores the linker's default behaviour of generating errors when
2313 it is reporting unresolved symbols.
2315 @kindex --whole-archive
2316 @cindex including an entire archive
2317 @item --whole-archive
2318 For each archive mentioned on the command line after the
2319 @option{--whole-archive} option, include every object file in the archive
2320 in the link, rather than searching the archive for the required object
2321 files. This is normally used to turn an archive file into a shared
2322 library, forcing every object to be included in the resulting shared
2323 library. This option may be used more than once.
2325 Two notes when using this option from gcc: First, gcc doesn't know
2326 about this option, so you have to use @option{-Wl,-whole-archive}.
2327 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2328 list of archives, because gcc will add its own list of archives to
2329 your link and you may not want this flag to affect those as well.
2331 @kindex --wrap=@var{symbol}
2332 @item --wrap=@var{symbol}
2333 Use a wrapper function for @var{symbol}. Any undefined reference to
2334 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2335 undefined reference to @code{__real_@var{symbol}} will be resolved to
2338 This can be used to provide a wrapper for a system function. The
2339 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2340 wishes to call the system function, it should call
2341 @code{__real_@var{symbol}}.
2343 Here is a trivial example:
2347 __wrap_malloc (size_t c)
2349 printf ("malloc called with %zu\n", c);
2350 return __real_malloc (c);
2354 If you link other code with this file using @option{--wrap malloc}, then
2355 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2356 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2357 call the real @code{malloc} function.
2359 You may wish to provide a @code{__real_malloc} function as well, so that
2360 links without the @option{--wrap} option will succeed. If you do this,
2361 you should not put the definition of @code{__real_malloc} in the same
2362 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2363 call before the linker has a chance to wrap it to @code{malloc}.
2365 @kindex --eh-frame-hdr
2366 @kindex --no-eh-frame-hdr
2367 @item --eh-frame-hdr
2368 @itemx --no-eh-frame-hdr
2369 Request (@option{--eh-frame-hdr}) or suppress
2370 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2371 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2373 @kindex --ld-generated-unwind-info
2374 @item --no-ld-generated-unwind-info
2375 Request creation of @code{.eh_frame} unwind info for linker
2376 generated code sections like PLT. This option is on by default
2377 if linker generated unwind info is supported.
2379 @kindex --enable-new-dtags
2380 @kindex --disable-new-dtags
2381 @item --enable-new-dtags
2382 @itemx --disable-new-dtags
2383 This linker can create the new dynamic tags in ELF. But the older ELF
2384 systems may not understand them. If you specify
2385 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2386 and older dynamic tags will be omitted.
2387 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2388 created. By default, the new dynamic tags are not created. Note that
2389 those options are only available for ELF systems.
2391 @kindex --hash-size=@var{number}
2392 @item --hash-size=@var{number}
2393 Set the default size of the linker's hash tables to a prime number
2394 close to @var{number}. Increasing this value can reduce the length of
2395 time it takes the linker to perform its tasks, at the expense of
2396 increasing the linker's memory requirements. Similarly reducing this
2397 value can reduce the memory requirements at the expense of speed.
2399 @kindex --hash-style=@var{style}
2400 @item --hash-style=@var{style}
2401 Set the type of linker's hash table(s). @var{style} can be either
2402 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2403 new style GNU @code{.gnu.hash} section or @code{both} for both
2404 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2405 hash tables. The default is @code{sysv}.
2407 @kindex --compress-debug-sections=none
2408 @kindex --compress-debug-sections=zlib
2409 @kindex --compress-debug-sections=zlib-gnu
2410 @kindex --compress-debug-sections=zlib-gabi
2411 @item --compress-debug-sections=none
2412 @itemx --compress-debug-sections=zlib
2413 @itemx --compress-debug-sections=zlib-gnu
2414 @itemx --compress-debug-sections=zlib-gabi
2415 On ELF platforms, these options control how DWARF debug sections are
2416 compressed using zlib.
2418 @option{--compress-debug-sections=none} doesn't compress DWARF debug
2419 sections. @option{--compress-debug-sections=zlib-gnu} compresses
2420 DWARF debug sections and renames them to begin with @samp{.zdebug}
2421 instead of @samp{.debug}. @option{--compress-debug-sections=zlib-gabi}
2422 also compresses DWARF debug sections, but rather than renaming them it
2423 sets the SHF_COMPRESSED flag in the sections' headers.
2425 The @option{--compress-debug-sections=zlib} option is an alias for
2426 @option{--compress-debug-sections=zlib-gabi}.
2428 Note that this option overrides any compression in input debug
2429 sections, so if a binary is linked with @option{--compress-debug-sections=none}
2430 for example, then any compressed debug sections in input files will be
2431 uncompressed before they are copied into the output binary.
2433 The default compression behaviour varies depending upon the target
2434 involved and the configure options used to build the toolchain. The
2435 default can be determined by examining the output from the linker's
2436 @option{--help} option.
2438 @kindex --reduce-memory-overheads
2439 @item --reduce-memory-overheads
2440 This option reduces memory requirements at ld runtime, at the expense of
2441 linking speed. This was introduced to select the old O(n^2) algorithm
2442 for link map file generation, rather than the new O(n) algorithm which uses
2443 about 40% more memory for symbol storage.
2445 Another effect of the switch is to set the default hash table size to
2446 1021, which again saves memory at the cost of lengthening the linker's
2447 run time. This is not done however if the @option{--hash-size} switch
2450 The @option{--reduce-memory-overheads} switch may be also be used to
2451 enable other tradeoffs in future versions of the linker.
2454 @kindex --build-id=@var{style}
2456 @itemx --build-id=@var{style}
2457 Request the creation of a @code{.note.gnu.build-id} ELF note section
2458 or a @code{.buildid} COFF section. The contents of the note are
2459 unique bits identifying this linked file. @var{style} can be
2460 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2461 @sc{SHA1} hash on the normative parts of the output contents,
2462 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2463 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2464 string specified as an even number of hexadecimal digits (@code{-} and
2465 @code{:} characters between digit pairs are ignored). If @var{style}
2466 is omitted, @code{sha1} is used.
2468 The @code{md5} and @code{sha1} styles produces an identifier
2469 that is always the same in an identical output file, but will be
2470 unique among all nonidentical output files. It is not intended
2471 to be compared as a checksum for the file's contents. A linked
2472 file may be changed later by other tools, but the build ID bit
2473 string identifying the original linked file does not change.
2475 Passing @code{none} for @var{style} disables the setting from any
2476 @code{--build-id} options earlier on the command line.
2481 @subsection Options Specific to i386 PE Targets
2483 @c man begin OPTIONS
2485 The i386 PE linker supports the @option{-shared} option, which causes
2486 the output to be a dynamically linked library (DLL) instead of a
2487 normal executable. You should name the output @code{*.dll} when you
2488 use this option. In addition, the linker fully supports the standard
2489 @code{*.def} files, which may be specified on the linker command line
2490 like an object file (in fact, it should precede archives it exports
2491 symbols from, to ensure that they get linked in, just like a normal
2494 In addition to the options common to all targets, the i386 PE linker
2495 support additional command line options that are specific to the i386
2496 PE target. Options that take values may be separated from their
2497 values by either a space or an equals sign.
2501 @kindex --add-stdcall-alias
2502 @item --add-stdcall-alias
2503 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2504 as-is and also with the suffix stripped.
2505 [This option is specific to the i386 PE targeted port of the linker]
2508 @item --base-file @var{file}
2509 Use @var{file} as the name of a file in which to save the base
2510 addresses of all the relocations needed for generating DLLs with
2512 [This is an i386 PE specific option]
2516 Create a DLL instead of a regular executable. You may also use
2517 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2519 [This option is specific to the i386 PE targeted port of the linker]
2521 @kindex --enable-long-section-names
2522 @kindex --disable-long-section-names
2523 @item --enable-long-section-names
2524 @itemx --disable-long-section-names
2525 The PE variants of the COFF object format add an extension that permits
2526 the use of section names longer than eight characters, the normal limit
2527 for COFF. By default, these names are only allowed in object files, as
2528 fully-linked executable images do not carry the COFF string table required
2529 to support the longer names. As a GNU extension, it is possible to
2530 allow their use in executable images as well, or to (probably pointlessly!)
2531 disallow it in object files, by using these two options. Executable images
2532 generated with these long section names are slightly non-standard, carrying
2533 as they do a string table, and may generate confusing output when examined
2534 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2535 GDB relies on the use of PE long section names to find Dwarf-2 debug
2536 information sections in an executable image at runtime, and so if neither
2537 option is specified on the command-line, @command{ld} will enable long
2538 section names, overriding the default and technically correct behaviour,
2539 when it finds the presence of debug information while linking an executable
2540 image and not stripping symbols.
2541 [This option is valid for all PE targeted ports of the linker]
2543 @kindex --enable-stdcall-fixup
2544 @kindex --disable-stdcall-fixup
2545 @item --enable-stdcall-fixup
2546 @itemx --disable-stdcall-fixup
2547 If the link finds a symbol that it cannot resolve, it will attempt to
2548 do ``fuzzy linking'' by looking for another defined symbol that differs
2549 only in the format of the symbol name (cdecl vs stdcall) and will
2550 resolve that symbol by linking to the match. For example, the
2551 undefined symbol @code{_foo} might be linked to the function
2552 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2553 to the function @code{_bar}. When the linker does this, it prints a
2554 warning, since it normally should have failed to link, but sometimes
2555 import libraries generated from third-party dlls may need this feature
2556 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2557 feature is fully enabled and warnings are not printed. If you specify
2558 @option{--disable-stdcall-fixup}, this feature is disabled and such
2559 mismatches are considered to be errors.
2560 [This option is specific to the i386 PE targeted port of the linker]
2562 @kindex --leading-underscore
2563 @kindex --no-leading-underscore
2564 @item --leading-underscore
2565 @itemx --no-leading-underscore
2566 For most targets default symbol-prefix is an underscore and is defined
2567 in target's description. By this option it is possible to
2568 disable/enable the default underscore symbol-prefix.
2570 @cindex DLLs, creating
2571 @kindex --export-all-symbols
2572 @item --export-all-symbols
2573 If given, all global symbols in the objects used to build a DLL will
2574 be exported by the DLL. Note that this is the default if there
2575 otherwise wouldn't be any exported symbols. When symbols are
2576 explicitly exported via DEF files or implicitly exported via function
2577 attributes, the default is to not export anything else unless this
2578 option is given. Note that the symbols @code{DllMain@@12},
2579 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2580 @code{impure_ptr} will not be automatically
2581 exported. Also, symbols imported from other DLLs will not be
2582 re-exported, nor will symbols specifying the DLL's internal layout
2583 such as those beginning with @code{_head_} or ending with
2584 @code{_iname}. In addition, no symbols from @code{libgcc},
2585 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2586 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2587 not be exported, to help with C++ DLLs. Finally, there is an
2588 extensive list of cygwin-private symbols that are not exported
2589 (obviously, this applies on when building DLLs for cygwin targets).
2590 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2591 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2592 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2593 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2594 @code{cygwin_premain3}, and @code{environ}.
2595 [This option is specific to the i386 PE targeted port of the linker]
2597 @kindex --exclude-symbols
2598 @item --exclude-symbols @var{symbol},@var{symbol},...
2599 Specifies a list of symbols which should not be automatically
2600 exported. The symbol names may be delimited by commas or colons.
2601 [This option is specific to the i386 PE targeted port of the linker]
2603 @kindex --exclude-all-symbols
2604 @item --exclude-all-symbols
2605 Specifies no symbols should be automatically exported.
2606 [This option is specific to the i386 PE targeted port of the linker]
2608 @kindex --file-alignment
2609 @item --file-alignment
2610 Specify the file alignment. Sections in the file will always begin at
2611 file offsets which are multiples of this number. This defaults to
2613 [This option is specific to the i386 PE targeted port of the linker]
2617 @item --heap @var{reserve}
2618 @itemx --heap @var{reserve},@var{commit}
2619 Specify the number of bytes of memory to reserve (and optionally commit)
2620 to be used as heap for this program. The default is 1MB reserved, 4K
2622 [This option is specific to the i386 PE targeted port of the linker]
2625 @kindex --image-base
2626 @item --image-base @var{value}
2627 Use @var{value} as the base address of your program or dll. This is
2628 the lowest memory location that will be used when your program or dll
2629 is loaded. To reduce the need to relocate and improve performance of
2630 your dlls, each should have a unique base address and not overlap any
2631 other dlls. The default is 0x400000 for executables, and 0x10000000
2633 [This option is specific to the i386 PE targeted port of the linker]
2637 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2638 symbols before they are exported.
2639 [This option is specific to the i386 PE targeted port of the linker]
2641 @kindex --large-address-aware
2642 @item --large-address-aware
2643 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2644 header is set to indicate that this executable supports virtual addresses
2645 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2646 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2647 section of the BOOT.INI. Otherwise, this bit has no effect.
2648 [This option is specific to PE targeted ports of the linker]
2650 @kindex --disable-large-address-aware
2651 @item --disable-large-address-aware
2652 Reverts the effect of a previous @samp{--large-address-aware} option.
2653 This is useful if @samp{--large-address-aware} is always set by the compiler
2654 driver (e.g. Cygwin gcc) and the executable does not support virtual
2655 addresses greater than 2 gigabytes.
2656 [This option is specific to PE targeted ports of the linker]
2658 @kindex --major-image-version
2659 @item --major-image-version @var{value}
2660 Sets the major number of the ``image version''. Defaults to 1.
2661 [This option is specific to the i386 PE targeted port of the linker]
2663 @kindex --major-os-version
2664 @item --major-os-version @var{value}
2665 Sets the major number of the ``os version''. Defaults to 4.
2666 [This option is specific to the i386 PE targeted port of the linker]
2668 @kindex --major-subsystem-version
2669 @item --major-subsystem-version @var{value}
2670 Sets the major number of the ``subsystem version''. Defaults to 4.
2671 [This option is specific to the i386 PE targeted port of the linker]
2673 @kindex --minor-image-version
2674 @item --minor-image-version @var{value}
2675 Sets the minor number of the ``image version''. Defaults to 0.
2676 [This option is specific to the i386 PE targeted port of the linker]
2678 @kindex --minor-os-version
2679 @item --minor-os-version @var{value}
2680 Sets the minor number of the ``os version''. Defaults to 0.
2681 [This option is specific to the i386 PE targeted port of the linker]
2683 @kindex --minor-subsystem-version
2684 @item --minor-subsystem-version @var{value}
2685 Sets the minor number of the ``subsystem version''. Defaults to 0.
2686 [This option is specific to the i386 PE targeted port of the linker]
2688 @cindex DEF files, creating
2689 @cindex DLLs, creating
2690 @kindex --output-def
2691 @item --output-def @var{file}
2692 The linker will create the file @var{file} which will contain a DEF
2693 file corresponding to the DLL the linker is generating. This DEF file
2694 (which should be called @code{*.def}) may be used to create an import
2695 library with @code{dlltool} or may be used as a reference to
2696 automatically or implicitly exported symbols.
2697 [This option is specific to the i386 PE targeted port of the linker]
2699 @cindex DLLs, creating
2700 @kindex --enable-auto-image-base
2701 @item --enable-auto-image-base
2702 @itemx --enable-auto-image-base=@var{value}
2703 Automatically choose the image base for DLLs, optionally starting with base
2704 @var{value}, unless one is specified using the @code{--image-base} argument.
2705 By using a hash generated from the dllname to create unique image bases
2706 for each DLL, in-memory collisions and relocations which can delay program
2707 execution are avoided.
2708 [This option is specific to the i386 PE targeted port of the linker]
2710 @kindex --disable-auto-image-base
2711 @item --disable-auto-image-base
2712 Do not automatically generate a unique image base. If there is no
2713 user-specified image base (@code{--image-base}) then use the platform
2715 [This option is specific to the i386 PE targeted port of the linker]
2717 @cindex DLLs, linking to
2718 @kindex --dll-search-prefix
2719 @item --dll-search-prefix @var{string}
2720 When linking dynamically to a dll without an import library,
2721 search for @code{<string><basename>.dll} in preference to
2722 @code{lib<basename>.dll}. This behaviour allows easy distinction
2723 between DLLs built for the various "subplatforms": native, cygwin,
2724 uwin, pw, etc. For instance, cygwin DLLs typically use
2725 @code{--dll-search-prefix=cyg}.
2726 [This option is specific to the i386 PE targeted port of the linker]
2728 @kindex --enable-auto-import
2729 @item --enable-auto-import
2730 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2731 DATA imports from DLLs, thus making it possible to bypass the dllimport
2732 mechanism on the user side and to reference unmangled symbol names.
2733 [This option is specific to the i386 PE targeted port of the linker]
2735 The following remarks pertain to the original implementation of the
2736 feature and are obsolete nowadays for Cygwin and MinGW targets.
2738 Note: Use of the 'auto-import' extension will cause the text section
2739 of the image file to be made writable. This does not conform to the
2740 PE-COFF format specification published by Microsoft.
2742 Note - use of the 'auto-import' extension will also cause read only
2743 data which would normally be placed into the .rdata section to be
2744 placed into the .data section instead. This is in order to work
2745 around a problem with consts that is described here:
2746 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2748 Using 'auto-import' generally will 'just work' -- but sometimes you may
2751 "variable '<var>' can't be auto-imported. Please read the
2752 documentation for ld's @code{--enable-auto-import} for details."
2754 This message occurs when some (sub)expression accesses an address
2755 ultimately given by the sum of two constants (Win32 import tables only
2756 allow one). Instances where this may occur include accesses to member
2757 fields of struct variables imported from a DLL, as well as using a
2758 constant index into an array variable imported from a DLL. Any
2759 multiword variable (arrays, structs, long long, etc) may trigger
2760 this error condition. However, regardless of the exact data type
2761 of the offending exported variable, ld will always detect it, issue
2762 the warning, and exit.
2764 There are several ways to address this difficulty, regardless of the
2765 data type of the exported variable:
2767 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2768 of adjusting references in your client code for runtime environment, so
2769 this method works only when runtime environment supports this feature.
2771 A second solution is to force one of the 'constants' to be a variable --
2772 that is, unknown and un-optimizable at compile time. For arrays,
2773 there are two possibilities: a) make the indexee (the array's address)
2774 a variable, or b) make the 'constant' index a variable. Thus:
2777 extern type extern_array[];
2779 @{ volatile type *t=extern_array; t[1] @}
2785 extern type extern_array[];
2787 @{ volatile int t=1; extern_array[t] @}
2790 For structs (and most other multiword data types) the only option
2791 is to make the struct itself (or the long long, or the ...) variable:
2794 extern struct s extern_struct;
2795 extern_struct.field -->
2796 @{ volatile struct s *t=&extern_struct; t->field @}
2802 extern long long extern_ll;
2804 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2807 A third method of dealing with this difficulty is to abandon
2808 'auto-import' for the offending symbol and mark it with
2809 @code{__declspec(dllimport)}. However, in practice that
2810 requires using compile-time #defines to indicate whether you are
2811 building a DLL, building client code that will link to the DLL, or
2812 merely building/linking to a static library. In making the choice
2813 between the various methods of resolving the 'direct address with
2814 constant offset' problem, you should consider typical real-world usage:
2822 void main(int argc, char **argv)@{
2823 printf("%d\n",arr[1]);
2833 void main(int argc, char **argv)@{
2834 /* This workaround is for win32 and cygwin; do not "optimize" */
2835 volatile int *parr = arr;
2836 printf("%d\n",parr[1]);
2843 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2844 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2845 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2846 #define FOO_IMPORT __declspec(dllimport)
2850 extern FOO_IMPORT int arr[];
2853 void main(int argc, char **argv)@{
2854 printf("%d\n",arr[1]);
2858 A fourth way to avoid this problem is to re-code your
2859 library to use a functional interface rather than a data interface
2860 for the offending variables (e.g. set_foo() and get_foo() accessor
2863 @kindex --disable-auto-import
2864 @item --disable-auto-import
2865 Do not attempt to do sophisticated linking of @code{_symbol} to
2866 @code{__imp__symbol} for DATA imports from DLLs.
2867 [This option is specific to the i386 PE targeted port of the linker]
2869 @kindex --enable-runtime-pseudo-reloc
2870 @item --enable-runtime-pseudo-reloc
2871 If your code contains expressions described in --enable-auto-import section,
2872 that is, DATA imports from DLL with non-zero offset, this switch will create
2873 a vector of 'runtime pseudo relocations' which can be used by runtime
2874 environment to adjust references to such data in your client code.
2875 [This option is specific to the i386 PE targeted port of the linker]
2877 @kindex --disable-runtime-pseudo-reloc
2878 @item --disable-runtime-pseudo-reloc
2879 Do not create pseudo relocations for non-zero offset DATA imports from DLLs.
2880 [This option is specific to the i386 PE targeted port of the linker]
2882 @kindex --enable-extra-pe-debug
2883 @item --enable-extra-pe-debug
2884 Show additional debug info related to auto-import symbol thunking.
2885 [This option is specific to the i386 PE targeted port of the linker]
2887 @kindex --section-alignment
2888 @item --section-alignment
2889 Sets the section alignment. Sections in memory will always begin at
2890 addresses which are a multiple of this number. Defaults to 0x1000.
2891 [This option is specific to the i386 PE targeted port of the linker]
2895 @item --stack @var{reserve}
2896 @itemx --stack @var{reserve},@var{commit}
2897 Specify the number of bytes of memory to reserve (and optionally commit)
2898 to be used as stack for this program. The default is 2MB reserved, 4K
2900 [This option is specific to the i386 PE targeted port of the linker]
2903 @item --subsystem @var{which}
2904 @itemx --subsystem @var{which}:@var{major}
2905 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2906 Specifies the subsystem under which your program will execute. The
2907 legal values for @var{which} are @code{native}, @code{windows},
2908 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2909 the subsystem version also. Numeric values are also accepted for
2911 [This option is specific to the i386 PE targeted port of the linker]
2913 The following options set flags in the @code{DllCharacteristics} field
2914 of the PE file header:
2915 [These options are specific to PE targeted ports of the linker]
2917 @kindex --high-entropy-va
2918 @item --high-entropy-va
2919 Image is compatible with 64-bit address space layout randomization
2922 @kindex --dynamicbase
2924 The image base address may be relocated using address space layout
2925 randomization (ASLR). This feature was introduced with MS Windows
2926 Vista for i386 PE targets.
2928 @kindex --forceinteg
2930 Code integrity checks are enforced.
2934 The image is compatible with the Data Execution Prevention.
2935 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2937 @kindex --no-isolation
2938 @item --no-isolation
2939 Although the image understands isolation, do not isolate the image.
2943 The image does not use SEH. No SE handler may be called from
2948 Do not bind this image.
2952 The driver uses the MS Windows Driver Model.
2956 The image is Terminal Server aware.
2958 @kindex --insert-timestamp
2959 @item --insert-timestamp
2960 @itemx --no-insert-timestamp
2961 Insert a real timestamp into the image. This is the default behaviour
2962 as it matches legacy code and it means that the image will work with
2963 other, proprietary tools. The problem with this default is that it
2964 will result in slightly different images being produced each time the
2965 same sources are linked. The option @option{--no-insert-timestamp}
2966 can be used to insert a zero value for the timestamp, this ensuring
2967 that binaries produced from identical sources will compare
2974 @subsection Options specific to C6X uClinux targets
2976 @c man begin OPTIONS
2978 The C6X uClinux target uses a binary format called DSBT to support shared
2979 libraries. Each shared library in the system needs to have a unique index;
2980 all executables use an index of 0.
2985 @item --dsbt-size @var{size}
2986 This option sets the number of entries in the DSBT of the current executable
2987 or shared library to @var{size}. The default is to create a table with 64
2990 @kindex --dsbt-index
2991 @item --dsbt-index @var{index}
2992 This option sets the DSBT index of the current executable or shared library
2993 to @var{index}. The default is 0, which is appropriate for generating
2994 executables. If a shared library is generated with a DSBT index of 0, the
2995 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
2997 @kindex --no-merge-exidx-entries
2998 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
2999 exidx entries in frame unwind info.
3007 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
3009 @c man begin OPTIONS
3011 The 68HC11 and 68HC12 linkers support specific options to control the
3012 memory bank switching mapping and trampoline code generation.
3016 @kindex --no-trampoline
3017 @item --no-trampoline
3018 This option disables the generation of trampoline. By default a trampoline
3019 is generated for each far function which is called using a @code{jsr}
3020 instruction (this happens when a pointer to a far function is taken).
3022 @kindex --bank-window
3023 @item --bank-window @var{name}
3024 This option indicates to the linker the name of the memory region in
3025 the @samp{MEMORY} specification that describes the memory bank window.
3026 The definition of such region is then used by the linker to compute
3027 paging and addresses within the memory window.
3035 @subsection Options specific to Motorola 68K target
3037 @c man begin OPTIONS
3039 The following options are supported to control handling of GOT generation
3040 when linking for 68K targets.
3045 @item --got=@var{type}
3046 This option tells the linker which GOT generation scheme to use.
3047 @var{type} should be one of @samp{single}, @samp{negative},
3048 @samp{multigot} or @samp{target}. For more information refer to the
3049 Info entry for @file{ld}.
3057 @subsection Options specific to MIPS targets
3059 @c man begin OPTIONS
3061 The following options are supported to control microMIPS instruction
3062 generation and branch relocation checks for ISA mode transitions when
3063 linking for MIPS targets.
3071 These options control the choice of microMIPS instructions used in code
3072 generated by the linker, such as that in the PLT or lazy binding stubs,
3073 or in relaxation. If @samp{--insn32} is used, then the linker only uses
3074 32-bit instruction encodings. By default or if @samp{--no-insn32} is
3075 used, all instruction encodings are used, including 16-bit ones where
3078 @kindex --ignore-branch-isa
3079 @item --ignore-branch-isa
3080 @kindex --no-ignore-branch-isa
3081 @itemx --no-ignore-branch-isa
3082 These options control branch relocation checks for invalid ISA mode
3083 transitions. If @samp{--ignore-branch-isa} is used, then the linker
3084 accepts any branch relocations and any ISA mode transition required
3085 is lost in relocation calculation, except for some cases of @code{BAL}
3086 instructions which meet relaxation conditions and are converted to
3087 equivalent @code{JALX} instructions as the associated relocation is
3088 calculated. By default or if @samp{--no-ignore-branch-isa} is used
3089 a check is made causing the loss of an ISA mode transition to produce
3099 @section Environment Variables
3101 @c man begin ENVIRONMENT
3103 You can change the behaviour of @command{ld} with the environment variables
3104 @ifclear SingleFormat
3107 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
3109 @ifclear SingleFormat
3111 @cindex default input format
3112 @code{GNUTARGET} determines the input-file object format if you don't
3113 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
3114 of the BFD names for an input format (@pxref{BFD}). If there is no
3115 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
3116 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
3117 attempts to discover the input format by examining binary input files;
3118 this method often succeeds, but there are potential ambiguities, since
3119 there is no method of ensuring that the magic number used to specify
3120 object-file formats is unique. However, the configuration procedure for
3121 BFD on each system places the conventional format for that system first
3122 in the search-list, so ambiguities are resolved in favor of convention.
3126 @cindex default emulation
3127 @cindex emulation, default
3128 @code{LDEMULATION} determines the default emulation if you don't use the
3129 @samp{-m} option. The emulation can affect various aspects of linker
3130 behaviour, particularly the default linker script. You can list the
3131 available emulations with the @samp{--verbose} or @samp{-V} options. If
3132 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3133 variable is not defined, the default emulation depends upon how the
3134 linker was configured.
3136 @kindex COLLECT_NO_DEMANGLE
3137 @cindex demangling, default
3138 Normally, the linker will default to demangling symbols. However, if
3139 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3140 default to not demangling symbols. This environment variable is used in
3141 a similar fashion by the @code{gcc} linker wrapper program. The default
3142 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3149 @chapter Linker Scripts
3152 @cindex linker scripts
3153 @cindex command files
3154 Every link is controlled by a @dfn{linker script}. This script is
3155 written in the linker command language.
3157 The main purpose of the linker script is to describe how the sections in
3158 the input files should be mapped into the output file, and to control
3159 the memory layout of the output file. Most linker scripts do nothing
3160 more than this. However, when necessary, the linker script can also
3161 direct the linker to perform many other operations, using the commands
3164 The linker always uses a linker script. If you do not supply one
3165 yourself, the linker will use a default script that is compiled into the
3166 linker executable. You can use the @samp{--verbose} command line option
3167 to display the default linker script. Certain command line options,
3168 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3170 You may supply your own linker script by using the @samp{-T} command
3171 line option. When you do this, your linker script will replace the
3172 default linker script.
3174 You may also use linker scripts implicitly by naming them as input files
3175 to the linker, as though they were files to be linked. @xref{Implicit
3179 * Basic Script Concepts:: Basic Linker Script Concepts
3180 * Script Format:: Linker Script Format
3181 * Simple Example:: Simple Linker Script Example
3182 * Simple Commands:: Simple Linker Script Commands
3183 * Assignments:: Assigning Values to Symbols
3184 * SECTIONS:: SECTIONS Command
3185 * MEMORY:: MEMORY Command
3186 * PHDRS:: PHDRS Command
3187 * VERSION:: VERSION Command
3188 * Expressions:: Expressions in Linker Scripts
3189 * Implicit Linker Scripts:: Implicit Linker Scripts
3192 @node Basic Script Concepts
3193 @section Basic Linker Script Concepts
3194 @cindex linker script concepts
3195 We need to define some basic concepts and vocabulary in order to
3196 describe the linker script language.
3198 The linker combines input files into a single output file. The output
3199 file and each input file are in a special data format known as an
3200 @dfn{object file format}. Each file is called an @dfn{object file}.
3201 The output file is often called an @dfn{executable}, but for our
3202 purposes we will also call it an object file. Each object file has,
3203 among other things, a list of @dfn{sections}. We sometimes refer to a
3204 section in an input file as an @dfn{input section}; similarly, a section
3205 in the output file is an @dfn{output section}.
3207 Each section in an object file has a name and a size. Most sections
3208 also have an associated block of data, known as the @dfn{section
3209 contents}. A section may be marked as @dfn{loadable}, which means that
3210 the contents should be loaded into memory when the output file is run.
3211 A section with no contents may be @dfn{allocatable}, which means that an
3212 area in memory should be set aside, but nothing in particular should be
3213 loaded there (in some cases this memory must be zeroed out). A section
3214 which is neither loadable nor allocatable typically contains some sort
3215 of debugging information.
3217 Every loadable or allocatable output section has two addresses. The
3218 first is the @dfn{VMA}, or virtual memory address. This is the address
3219 the section will have when the output file is run. The second is the
3220 @dfn{LMA}, or load memory address. This is the address at which the
3221 section will be loaded. In most cases the two addresses will be the
3222 same. An example of when they might be different is when a data section
3223 is loaded into ROM, and then copied into RAM when the program starts up
3224 (this technique is often used to initialize global variables in a ROM
3225 based system). In this case the ROM address would be the LMA, and the
3226 RAM address would be the VMA.
3228 You can see the sections in an object file by using the @code{objdump}
3229 program with the @samp{-h} option.
3231 Every object file also has a list of @dfn{symbols}, known as the
3232 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3233 has a name, and each defined symbol has an address, among other
3234 information. If you compile a C or C++ program into an object file, you
3235 will get a defined symbol for every defined function and global or
3236 static variable. Every undefined function or global variable which is
3237 referenced in the input file will become an undefined symbol.
3239 You can see the symbols in an object file by using the @code{nm}
3240 program, or by using the @code{objdump} program with the @samp{-t}
3244 @section Linker Script Format
3245 @cindex linker script format
3246 Linker scripts are text files.
3248 You write a linker script as a series of commands. Each command is
3249 either a keyword, possibly followed by arguments, or an assignment to a
3250 symbol. You may separate commands using semicolons. Whitespace is
3253 Strings such as file or format names can normally be entered directly.
3254 If the file name contains a character such as a comma which would
3255 otherwise serve to separate file names, you may put the file name in
3256 double quotes. There is no way to use a double quote character in a
3259 You may include comments in linker scripts just as in C, delimited by
3260 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3263 @node Simple Example
3264 @section Simple Linker Script Example
3265 @cindex linker script example
3266 @cindex example of linker script
3267 Many linker scripts are fairly simple.
3269 The simplest possible linker script has just one command:
3270 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3271 memory layout of the output file.
3273 The @samp{SECTIONS} command is a powerful command. Here we will
3274 describe a simple use of it. Let's assume your program consists only of
3275 code, initialized data, and uninitialized data. These will be in the
3276 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3277 Let's assume further that these are the only sections which appear in
3280 For this example, let's say that the code should be loaded at address
3281 0x10000, and that the data should start at address 0x8000000. Here is a
3282 linker script which will do that:
3287 .text : @{ *(.text) @}
3289 .data : @{ *(.data) @}
3290 .bss : @{ *(.bss) @}
3294 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3295 followed by a series of symbol assignments and output section
3296 descriptions enclosed in curly braces.
3298 The first line inside the @samp{SECTIONS} command of the above example
3299 sets the value of the special symbol @samp{.}, which is the location
3300 counter. If you do not specify the address of an output section in some
3301 other way (other ways are described later), the address is set from the
3302 current value of the location counter. The location counter is then
3303 incremented by the size of the output section. At the start of the
3304 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3306 The second line defines an output section, @samp{.text}. The colon is
3307 required syntax which may be ignored for now. Within the curly braces
3308 after the output section name, you list the names of the input sections
3309 which should be placed into this output section. The @samp{*} is a
3310 wildcard which matches any file name. The expression @samp{*(.text)}
3311 means all @samp{.text} input sections in all input files.
3313 Since the location counter is @samp{0x10000} when the output section
3314 @samp{.text} is defined, the linker will set the address of the
3315 @samp{.text} section in the output file to be @samp{0x10000}.
3317 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3318 the output file. The linker will place the @samp{.data} output section
3319 at address @samp{0x8000000}. After the linker places the @samp{.data}
3320 output section, the value of the location counter will be
3321 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3322 effect is that the linker will place the @samp{.bss} output section
3323 immediately after the @samp{.data} output section in memory.
3325 The linker will ensure that each output section has the required
3326 alignment, by increasing the location counter if necessary. In this
3327 example, the specified addresses for the @samp{.text} and @samp{.data}
3328 sections will probably satisfy any alignment constraints, but the linker
3329 may have to create a small gap between the @samp{.data} and @samp{.bss}
3332 That's it! That's a simple and complete linker script.
3334 @node Simple Commands
3335 @section Simple Linker Script Commands
3336 @cindex linker script simple commands
3337 In this section we describe the simple linker script commands.
3340 * Entry Point:: Setting the entry point
3341 * File Commands:: Commands dealing with files
3342 @ifclear SingleFormat
3343 * Format Commands:: Commands dealing with object file formats
3346 * REGION_ALIAS:: Assign alias names to memory regions
3347 * Miscellaneous Commands:: Other linker script commands
3351 @subsection Setting the Entry Point
3352 @kindex ENTRY(@var{symbol})
3353 @cindex start of execution
3354 @cindex first instruction
3356 The first instruction to execute in a program is called the @dfn{entry
3357 point}. You can use the @code{ENTRY} linker script command to set the
3358 entry point. The argument is a symbol name:
3363 There are several ways to set the entry point. The linker will set the
3364 entry point by trying each of the following methods in order, and
3365 stopping when one of them succeeds:
3368 the @samp{-e} @var{entry} command-line option;
3370 the @code{ENTRY(@var{symbol})} command in a linker script;
3372 the value of a target specific symbol, if it is defined; For many
3373 targets this is @code{start}, but PE and BeOS based systems for example
3374 check a list of possible entry symbols, matching the first one found.
3376 the address of the first byte of the @samp{.text} section, if present;
3378 The address @code{0}.
3382 @subsection Commands Dealing with Files
3383 @cindex linker script file commands
3384 Several linker script commands deal with files.
3387 @item INCLUDE @var{filename}
3388 @kindex INCLUDE @var{filename}
3389 @cindex including a linker script
3390 Include the linker script @var{filename} at this point. The file will
3391 be searched for in the current directory, and in any directory specified
3392 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3395 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3396 @code{SECTIONS} commands, or in output section descriptions.
3398 @item INPUT(@var{file}, @var{file}, @dots{})
3399 @itemx INPUT(@var{file} @var{file} @dots{})
3400 @kindex INPUT(@var{files})
3401 @cindex input files in linker scripts
3402 @cindex input object files in linker scripts
3403 @cindex linker script input object files
3404 The @code{INPUT} command directs the linker to include the named files
3405 in the link, as though they were named on the command line.
3407 For example, if you always want to include @file{subr.o} any time you do
3408 a link, but you can't be bothered to put it on every link command line,
3409 then you can put @samp{INPUT (subr.o)} in your linker script.
3411 In fact, if you like, you can list all of your input files in the linker
3412 script, and then invoke the linker with nothing but a @samp{-T} option.
3414 In case a @dfn{sysroot prefix} is configured, and the filename starts
3415 with the @samp{/} character, and the script being processed was
3416 located inside the @dfn{sysroot prefix}, the filename will be looked
3417 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3418 open the file in the current directory. If it is not found, the
3419 linker will search through the archive library search path.
3420 The @dfn{sysroot prefix} can also be forced by specifying @code{=}
3421 as the first character in the filename path, or prefixing the filename
3422 path with @code{$SYSROOT}. See also the description of @samp{-L} in
3423 @ref{Options,,Command Line Options}.
3425 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3426 name to @code{lib@var{file}.a}, as with the command line argument
3429 When you use the @code{INPUT} command in an implicit linker script, the
3430 files will be included in the link at the point at which the linker
3431 script file is included. This can affect archive searching.
3433 @item GROUP(@var{file}, @var{file}, @dots{})
3434 @itemx GROUP(@var{file} @var{file} @dots{})
3435 @kindex GROUP(@var{files})
3436 @cindex grouping input files
3437 The @code{GROUP} command is like @code{INPUT}, except that the named
3438 files should all be archives, and they are searched repeatedly until no
3439 new undefined references are created. See the description of @samp{-(}
3440 in @ref{Options,,Command Line Options}.
3442 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3443 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3444 @kindex AS_NEEDED(@var{files})
3445 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3446 commands, among other filenames. The files listed will be handled
3447 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3448 with the exception of ELF shared libraries, that will be added only
3449 when they are actually needed. This construct essentially enables
3450 @option{--as-needed} option for all the files listed inside of it
3451 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3454 @item OUTPUT(@var{filename})
3455 @kindex OUTPUT(@var{filename})
3456 @cindex output file name in linker script
3457 The @code{OUTPUT} command names the output file. Using
3458 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3459 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3460 Line Options}). If both are used, the command line option takes
3463 You can use the @code{OUTPUT} command to define a default name for the
3464 output file other than the usual default of @file{a.out}.
3466 @item SEARCH_DIR(@var{path})
3467 @kindex SEARCH_DIR(@var{path})
3468 @cindex library search path in linker script
3469 @cindex archive search path in linker script
3470 @cindex search path in linker script
3471 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3472 @command{ld} looks for archive libraries. Using
3473 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3474 on the command line (@pxref{Options,,Command Line Options}). If both
3475 are used, then the linker will search both paths. Paths specified using
3476 the command line option are searched first.
3478 @item STARTUP(@var{filename})
3479 @kindex STARTUP(@var{filename})
3480 @cindex first input file
3481 The @code{STARTUP} command is just like the @code{INPUT} command, except
3482 that @var{filename} will become the first input file to be linked, as
3483 though it were specified first on the command line. This may be useful
3484 when using a system in which the entry point is always the start of the
3488 @ifclear SingleFormat
3489 @node Format Commands
3490 @subsection Commands Dealing with Object File Formats
3491 A couple of linker script commands deal with object file formats.
3494 @item OUTPUT_FORMAT(@var{bfdname})
3495 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3496 @kindex OUTPUT_FORMAT(@var{bfdname})
3497 @cindex output file format in linker script
3498 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3499 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3500 exactly like using @samp{--oformat @var{bfdname}} on the command line
3501 (@pxref{Options,,Command Line Options}). If both are used, the command
3502 line option takes precedence.
3504 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3505 formats based on the @samp{-EB} and @samp{-EL} command line options.
3506 This permits the linker script to set the output format based on the
3509 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3510 will be the first argument, @var{default}. If @samp{-EB} is used, the
3511 output format will be the second argument, @var{big}. If @samp{-EL} is
3512 used, the output format will be the third argument, @var{little}.
3514 For example, the default linker script for the MIPS ELF target uses this
3517 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3519 This says that the default format for the output file is
3520 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3521 option, the output file will be created in the @samp{elf32-littlemips}
3524 @item TARGET(@var{bfdname})
3525 @kindex TARGET(@var{bfdname})
3526 @cindex input file format in linker script
3527 The @code{TARGET} command names the BFD format to use when reading input
3528 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3529 This command is like using @samp{-b @var{bfdname}} on the command line
3530 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3531 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3532 command is also used to set the format for the output file. @xref{BFD}.
3537 @subsection Assign alias names to memory regions
3538 @kindex REGION_ALIAS(@var{alias}, @var{region})
3539 @cindex region alias
3540 @cindex region names
3542 Alias names can be added to existing memory regions created with the
3543 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3546 REGION_ALIAS(@var{alias}, @var{region})
3549 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3550 memory region @var{region}. This allows a flexible mapping of output sections
3551 to memory regions. An example follows.
3553 Suppose we have an application for embedded systems which come with various
3554 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3555 that allows code execution or data storage. Some may have a read-only,
3556 non-volatile memory @code{ROM} that allows code execution and read-only data
3557 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3558 read-only data access and no code execution capability. We have four output
3563 @code{.text} program code;
3565 @code{.rodata} read-only data;
3567 @code{.data} read-write initialized data;
3569 @code{.bss} read-write zero initialized data.
3572 The goal is to provide a linker command file that contains a system independent
3573 part defining the output sections and a system dependent part mapping the
3574 output sections to the memory regions available on the system. Our embedded
3575 systems come with three different memory setups @code{A}, @code{B} and
3577 @multitable @columnfractions .25 .25 .25 .25
3578 @item Section @tab Variant A @tab Variant B @tab Variant C
3579 @item .text @tab RAM @tab ROM @tab ROM
3580 @item .rodata @tab RAM @tab ROM @tab ROM2
3581 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3582 @item .bss @tab RAM @tab RAM @tab RAM
3584 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3585 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3586 the load address of the @code{.data} section starts in all three variants at
3587 the end of the @code{.rodata} section.
3589 The base linker script that deals with the output sections follows. It
3590 includes the system dependent @code{linkcmds.memory} file that describes the
3593 INCLUDE linkcmds.memory
3606 .data : AT (rodata_end)
3611 data_size = SIZEOF(.data);
3612 data_load_start = LOADADDR(.data);
3620 Now we need three different @code{linkcmds.memory} files to define memory
3621 regions and alias names. The content of @code{linkcmds.memory} for the three
3622 variants @code{A}, @code{B} and @code{C}:
3625 Here everything goes into the @code{RAM}.
3629 RAM : ORIGIN = 0, LENGTH = 4M
3632 REGION_ALIAS("REGION_TEXT", RAM);
3633 REGION_ALIAS("REGION_RODATA", RAM);
3634 REGION_ALIAS("REGION_DATA", RAM);
3635 REGION_ALIAS("REGION_BSS", RAM);
3638 Program code and read-only data go into the @code{ROM}. Read-write data goes
3639 into the @code{RAM}. An image of the initialized data is loaded into the
3640 @code{ROM} and will be copied during system start into the @code{RAM}.
3644 ROM : ORIGIN = 0, LENGTH = 3M
3645 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3648 REGION_ALIAS("REGION_TEXT", ROM);
3649 REGION_ALIAS("REGION_RODATA", ROM);
3650 REGION_ALIAS("REGION_DATA", RAM);
3651 REGION_ALIAS("REGION_BSS", RAM);
3654 Program code goes into the @code{ROM}. Read-only data goes into the
3655 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3656 initialized data is loaded into the @code{ROM2} and will be copied during
3657 system start into the @code{RAM}.
3661 ROM : ORIGIN = 0, LENGTH = 2M
3662 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3663 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3666 REGION_ALIAS("REGION_TEXT", ROM);
3667 REGION_ALIAS("REGION_RODATA", ROM2);
3668 REGION_ALIAS("REGION_DATA", RAM);
3669 REGION_ALIAS("REGION_BSS", RAM);
3673 It is possible to write a common system initialization routine to copy the
3674 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3679 extern char data_start [];
3680 extern char data_size [];
3681 extern char data_load_start [];
3683 void copy_data(void)
3685 if (data_start != data_load_start)
3687 memcpy(data_start, data_load_start, (size_t) data_size);
3692 @node Miscellaneous Commands
3693 @subsection Other Linker Script Commands
3694 There are a few other linker scripts commands.
3697 @item ASSERT(@var{exp}, @var{message})
3699 @cindex assertion in linker script
3700 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3701 with an error code, and print @var{message}.
3703 Note that assertions are checked before the final stages of linking
3704 take place. This means that expressions involving symbols PROVIDEd
3705 inside section definitions will fail if the user has not set values
3706 for those symbols. The only exception to this rule is PROVIDEd
3707 symbols that just reference dot. Thus an assertion like this:
3712 PROVIDE (__stack = .);
3713 PROVIDE (__stack_size = 0x100);
3714 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3718 will fail if @code{__stack_size} is not defined elsewhere. Symbols
3719 PROVIDEd outside of section definitions are evaluated earlier, so they
3720 can be used inside ASSERTions. Thus:
3723 PROVIDE (__stack_size = 0x100);
3726 PROVIDE (__stack = .);
3727 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3733 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3735 @cindex undefined symbol in linker script
3736 Force @var{symbol} to be entered in the output file as an undefined
3737 symbol. Doing this may, for example, trigger linking of additional
3738 modules from standard libraries. You may list several @var{symbol}s for
3739 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3740 command has the same effect as the @samp{-u} command-line option.
3742 @item FORCE_COMMON_ALLOCATION
3743 @kindex FORCE_COMMON_ALLOCATION
3744 @cindex common allocation in linker script
3745 This command has the same effect as the @samp{-d} command-line option:
3746 to make @command{ld} assign space to common symbols even if a relocatable
3747 output file is specified (@samp{-r}).
3749 @item INHIBIT_COMMON_ALLOCATION
3750 @kindex INHIBIT_COMMON_ALLOCATION
3751 @cindex common allocation in linker script
3752 This command has the same effect as the @samp{--no-define-common}
3753 command-line option: to make @code{ld} omit the assignment of addresses
3754 to common symbols even for a non-relocatable output file.
3756 @item FORCE_GROUP_ALLOCATION
3757 @kindex FORCE_GROUP_ALLOCATION
3758 @cindex group allocation in linker script
3759 @cindex section groups
3761 This command has the same effect as the
3762 @samp{--force-group-allocation} command-line option: to make
3763 @command{ld} place section group members like normal input sections,
3764 and to delete the section groups even if a relocatable output file is
3765 specified (@samp{-r}).
3767 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3769 @cindex insert user script into default script
3770 This command is typically used in a script specified by @samp{-T} to
3771 augment the default @code{SECTIONS} with, for example, overlays. It
3772 inserts all prior linker script statements after (or before)
3773 @var{output_section}, and also causes @samp{-T} to not override the
3774 default linker script. The exact insertion point is as for orphan
3775 sections. @xref{Location Counter}. The insertion happens after the
3776 linker has mapped input sections to output sections. Prior to the
3777 insertion, since @samp{-T} scripts are parsed before the default
3778 linker script, statements in the @samp{-T} script occur before the
3779 default linker script statements in the internal linker representation
3780 of the script. In particular, input section assignments will be made
3781 to @samp{-T} output sections before those in the default script. Here
3782 is an example of how a @samp{-T} script using @code{INSERT} might look:
3789 .ov1 @{ ov1*(.text) @}
3790 .ov2 @{ ov2*(.text) @}
3796 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3797 @kindex NOCROSSREFS(@var{sections})
3798 @cindex cross references
3799 This command may be used to tell @command{ld} to issue an error about any
3800 references among certain output sections.
3802 In certain types of programs, particularly on embedded systems when
3803 using overlays, when one section is loaded into memory, another section
3804 will not be. Any direct references between the two sections would be
3805 errors. For example, it would be an error if code in one section called
3806 a function defined in the other section.
3808 The @code{NOCROSSREFS} command takes a list of output section names. If
3809 @command{ld} detects any cross references between the sections, it reports
3810 an error and returns a non-zero exit status. Note that the
3811 @code{NOCROSSREFS} command uses output section names, not input section
3814 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
3815 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
3816 @cindex cross references
3817 This command may be used to tell @command{ld} to issue an error about any
3818 references to one section from a list of other sections.
3820 The @code{NOCROSSREFS} command is useful when ensuring that two or more
3821 output sections are entirely independent but there are situations where
3822 a one-way dependency is needed. For example, in a multi-core application
3823 there may be shared code that can be called from each core but for safety
3824 must never call back.
3826 The @code{NOCROSSREFS_TO} command takes a list of output section names.
3827 The first section can not be referenced from any of the other sections.
3828 If @command{ld} detects any references to the first section from any of
3829 the other sections, it reports an error and returns a non-zero exit
3830 status. Note that the @code{NOCROSSREFS_TO} command uses output section
3831 names, not input section names.
3833 @ifclear SingleFormat
3834 @item OUTPUT_ARCH(@var{bfdarch})
3835 @kindex OUTPUT_ARCH(@var{bfdarch})
3836 @cindex machine architecture
3837 @cindex architecture
3838 Specify a particular output machine architecture. The argument is one
3839 of the names used by the BFD library (@pxref{BFD}). You can see the
3840 architecture of an object file by using the @code{objdump} program with
3841 the @samp{-f} option.
3844 @item LD_FEATURE(@var{string})
3845 @kindex LD_FEATURE(@var{string})
3846 This command may be used to modify @command{ld} behavior. If
3847 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3848 in a script are simply treated as numbers everywhere.
3849 @xref{Expression Section}.
3853 @section Assigning Values to Symbols
3854 @cindex assignment in scripts
3855 @cindex symbol definition, scripts
3856 @cindex variables, defining
3857 You may assign a value to a symbol in a linker script. This will define
3858 the symbol and place it into the symbol table with a global scope.
3861 * Simple Assignments:: Simple Assignments
3864 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3865 * Source Code Reference:: How to use a linker script defined symbol in source code
3868 @node Simple Assignments
3869 @subsection Simple Assignments
3871 You may assign to a symbol using any of the C assignment operators:
3874 @item @var{symbol} = @var{expression} ;
3875 @itemx @var{symbol} += @var{expression} ;
3876 @itemx @var{symbol} -= @var{expression} ;
3877 @itemx @var{symbol} *= @var{expression} ;
3878 @itemx @var{symbol} /= @var{expression} ;
3879 @itemx @var{symbol} <<= @var{expression} ;
3880 @itemx @var{symbol} >>= @var{expression} ;
3881 @itemx @var{symbol} &= @var{expression} ;
3882 @itemx @var{symbol} |= @var{expression} ;
3885 The first case will define @var{symbol} to the value of
3886 @var{expression}. In the other cases, @var{symbol} must already be
3887 defined, and the value will be adjusted accordingly.
3889 The special symbol name @samp{.} indicates the location counter. You
3890 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3892 The semicolon after @var{expression} is required.
3894 Expressions are defined below; see @ref{Expressions}.
3896 You may write symbol assignments as commands in their own right, or as
3897 statements within a @code{SECTIONS} command, or as part of an output
3898 section description in a @code{SECTIONS} command.
3900 The section of the symbol will be set from the section of the
3901 expression; for more information, see @ref{Expression Section}.
3903 Here is an example showing the three different places that symbol
3904 assignments may be used:
3915 _bdata = (. + 3) & ~ 3;
3916 .data : @{ *(.data) @}
3920 In this example, the symbol @samp{floating_point} will be defined as
3921 zero. The symbol @samp{_etext} will be defined as the address following
3922 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3923 defined as the address following the @samp{.text} output section aligned
3924 upward to a 4 byte boundary.
3929 For ELF targeted ports, define a symbol that will be hidden and won't be
3930 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
3932 Here is the example from @ref{Simple Assignments}, rewritten to use
3936 HIDDEN(floating_point = 0);
3944 HIDDEN(_bdata = (. + 3) & ~ 3);
3945 .data : @{ *(.data) @}
3949 In this case none of the three symbols will be visible outside this module.
3954 In some cases, it is desirable for a linker script to define a symbol
3955 only if it is referenced and is not defined by any object included in
3956 the link. For example, traditional linkers defined the symbol
3957 @samp{etext}. However, ANSI C requires that the user be able to use
3958 @samp{etext} as a function name without encountering an error. The
3959 @code{PROVIDE} keyword may be used to define a symbol, such as
3960 @samp{etext}, only if it is referenced but not defined. The syntax is
3961 @code{PROVIDE(@var{symbol} = @var{expression})}.
3963 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3976 In this example, if the program defines @samp{_etext} (with a leading
3977 underscore), the linker will give a multiple definition error. If, on
3978 the other hand, the program defines @samp{etext} (with no leading
3979 underscore), the linker will silently use the definition in the program.
3980 If the program references @samp{etext} but does not define it, the
3981 linker will use the definition in the linker script.
3983 Note - the @code{PROVIDE} directive considers a common symbol to be
3984 defined, even though such a symbol could be combined with the symbol
3985 that the @code{PROVIDE} would create. This is particularly important
3986 when considering constructor and destructor list symbols such as
3987 @samp{__CTOR_LIST__} as these are often defined as common symbols.
3989 @node PROVIDE_HIDDEN
3990 @subsection PROVIDE_HIDDEN
3991 @cindex PROVIDE_HIDDEN
3992 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3993 hidden and won't be exported.
3995 @node Source Code Reference
3996 @subsection Source Code Reference
3998 Accessing a linker script defined variable from source code is not
3999 intuitive. In particular a linker script symbol is not equivalent to
4000 a variable declaration in a high level language, it is instead a
4001 symbol that does not have a value.
4003 Before going further, it is important to note that compilers often
4004 transform names in the source code into different names when they are
4005 stored in the symbol table. For example, Fortran compilers commonly
4006 prepend or append an underscore, and C++ performs extensive @samp{name
4007 mangling}. Therefore there might be a discrepancy between the name
4008 of a variable as it is used in source code and the name of the same
4009 variable as it is defined in a linker script. For example in C a
4010 linker script variable might be referred to as:
4016 But in the linker script it might be defined as:
4022 In the remaining examples however it is assumed that no name
4023 transformation has taken place.
4025 When a symbol is declared in a high level language such as C, two
4026 things happen. The first is that the compiler reserves enough space
4027 in the program's memory to hold the @emph{value} of the symbol. The
4028 second is that the compiler creates an entry in the program's symbol
4029 table which holds the symbol's @emph{address}. ie the symbol table
4030 contains the address of the block of memory holding the symbol's
4031 value. So for example the following C declaration, at file scope:
4037 creates an entry called @samp{foo} in the symbol table. This entry
4038 holds the address of an @samp{int} sized block of memory where the
4039 number 1000 is initially stored.
4041 When a program references a symbol the compiler generates code that
4042 first accesses the symbol table to find the address of the symbol's
4043 memory block and then code to read the value from that memory block.
4050 looks up the symbol @samp{foo} in the symbol table, gets the address
4051 associated with this symbol and then writes the value 1 into that
4058 looks up the symbol @samp{foo} in the symbol table, gets its address
4059 and then copies this address into the block of memory associated with
4060 the variable @samp{a}.
4062 Linker scripts symbol declarations, by contrast, create an entry in
4063 the symbol table but do not assign any memory to them. Thus they are
4064 an address without a value. So for example the linker script definition:
4070 creates an entry in the symbol table called @samp{foo} which holds
4071 the address of memory location 1000, but nothing special is stored at
4072 address 1000. This means that you cannot access the @emph{value} of a
4073 linker script defined symbol - it has no value - all you can do is
4074 access the @emph{address} of a linker script defined symbol.
4076 Hence when you are using a linker script defined symbol in source code
4077 you should always take the address of the symbol, and never attempt to
4078 use its value. For example suppose you want to copy the contents of a
4079 section of memory called .ROM into a section called .FLASH and the
4080 linker script contains these declarations:
4084 start_of_ROM = .ROM;
4085 end_of_ROM = .ROM + sizeof (.ROM);
4086 start_of_FLASH = .FLASH;
4090 Then the C source code to perform the copy would be:
4094 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
4096 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
4100 Note the use of the @samp{&} operators. These are correct.
4101 Alternatively the symbols can be treated as the names of vectors or
4102 arrays and then the code will again work as expected:
4106 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
4108 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
4112 Note how using this method does not require the use of @samp{&}
4116 @section SECTIONS Command
4118 The @code{SECTIONS} command tells the linker how to map input sections
4119 into output sections, and how to place the output sections in memory.
4121 The format of the @code{SECTIONS} command is:
4125 @var{sections-command}
4126 @var{sections-command}
4131 Each @var{sections-command} may of be one of the following:
4135 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
4137 a symbol assignment (@pxref{Assignments})
4139 an output section description
4141 an overlay description
4144 The @code{ENTRY} command and symbol assignments are permitted inside the
4145 @code{SECTIONS} command for convenience in using the location counter in
4146 those commands. This can also make the linker script easier to
4147 understand because you can use those commands at meaningful points in
4148 the layout of the output file.
4150 Output section descriptions and overlay descriptions are described
4153 If you do not use a @code{SECTIONS} command in your linker script, the
4154 linker will place each input section into an identically named output
4155 section in the order that the sections are first encountered in the
4156 input files. If all input sections are present in the first file, for
4157 example, the order of sections in the output file will match the order
4158 in the first input file. The first section will be at address zero.
4161 * Output Section Description:: Output section description
4162 * Output Section Name:: Output section name
4163 * Output Section Address:: Output section address
4164 * Input Section:: Input section description
4165 * Output Section Data:: Output section data
4166 * Output Section Keywords:: Output section keywords
4167 * Output Section Discarding:: Output section discarding
4168 * Output Section Attributes:: Output section attributes
4169 * Overlay Description:: Overlay description
4172 @node Output Section Description
4173 @subsection Output Section Description
4174 The full description of an output section looks like this:
4177 @var{section} [@var{address}] [(@var{type})] :
4179 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4180 [SUBALIGN(@var{subsection_align})]
4183 @var{output-section-command}
4184 @var{output-section-command}
4186 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4190 Most output sections do not use most of the optional section attributes.
4192 The whitespace around @var{section} is required, so that the section
4193 name is unambiguous. The colon and the curly braces are also required.
4194 The comma at the end may be required if a @var{fillexp} is used and
4195 the next @var{sections-command} looks like a continuation of the expression.
4196 The line breaks and other white space are optional.
4198 Each @var{output-section-command} may be one of the following:
4202 a symbol assignment (@pxref{Assignments})
4204 an input section description (@pxref{Input Section})
4206 data values to include directly (@pxref{Output Section Data})
4208 a special output section keyword (@pxref{Output Section Keywords})
4211 @node Output Section Name
4212 @subsection Output Section Name
4213 @cindex name, section
4214 @cindex section name
4215 The name of the output section is @var{section}. @var{section} must
4216 meet the constraints of your output format. In formats which only
4217 support a limited number of sections, such as @code{a.out}, the name
4218 must be one of the names supported by the format (@code{a.out}, for
4219 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4220 output format supports any number of sections, but with numbers and not
4221 names (as is the case for Oasys), the name should be supplied as a
4222 quoted numeric string. A section name may consist of any sequence of
4223 characters, but a name which contains any unusual characters such as
4224 commas must be quoted.
4226 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4229 @node Output Section Address
4230 @subsection Output Section Address
4231 @cindex address, section
4232 @cindex section address
4233 The @var{address} is an expression for the VMA (the virtual memory
4234 address) of the output section. This address is optional, but if it
4235 is provided then the output address will be set exactly as specified.
4237 If the output address is not specified then one will be chosen for the
4238 section, based on the heuristic below. This address will be adjusted
4239 to fit the alignment requirement of the output section. The
4240 alignment requirement is the strictest alignment of any input section
4241 contained within the output section.
4243 The output section address heuristic is as follows:
4247 If an output memory @var{region} is set for the section then it
4248 is added to this region and its address will be the next free address
4252 If the MEMORY command has been used to create a list of memory
4253 regions then the first region which has attributes compatible with the
4254 section is selected to contain it. The section's output address will
4255 be the next free address in that region; @ref{MEMORY}.
4258 If no memory regions were specified, or none match the section then
4259 the output address will be based on the current value of the location
4267 .text . : @{ *(.text) @}
4274 .text : @{ *(.text) @}
4278 are subtly different. The first will set the address of the
4279 @samp{.text} output section to the current value of the location
4280 counter. The second will set it to the current value of the location
4281 counter aligned to the strictest alignment of any of the @samp{.text}
4284 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4285 For example, if you want to align the section on a 0x10 byte boundary,
4286 so that the lowest four bits of the section address are zero, you could
4287 do something like this:
4289 .text ALIGN(0x10) : @{ *(.text) @}
4292 This works because @code{ALIGN} returns the current location counter
4293 aligned upward to the specified value.
4295 Specifying @var{address} for a section will change the value of the
4296 location counter, provided that the section is non-empty. (Empty
4297 sections are ignored).
4300 @subsection Input Section Description
4301 @cindex input sections
4302 @cindex mapping input sections to output sections
4303 The most common output section command is an input section description.
4305 The input section description is the most basic linker script operation.
4306 You use output sections to tell the linker how to lay out your program
4307 in memory. You use input section descriptions to tell the linker how to
4308 map the input files into your memory layout.
4311 * Input Section Basics:: Input section basics
4312 * Input Section Wildcards:: Input section wildcard patterns
4313 * Input Section Common:: Input section for common symbols
4314 * Input Section Keep:: Input section and garbage collection
4315 * Input Section Example:: Input section example
4318 @node Input Section Basics
4319 @subsubsection Input Section Basics
4320 @cindex input section basics
4321 An input section description consists of a file name optionally followed
4322 by a list of section names in parentheses.
4324 The file name and the section name may be wildcard patterns, which we
4325 describe further below (@pxref{Input Section Wildcards}).
4327 The most common input section description is to include all input
4328 sections with a particular name in the output section. For example, to
4329 include all input @samp{.text} sections, you would write:
4334 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4335 @cindex EXCLUDE_FILE
4336 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4337 match all files except the ones specified in the EXCLUDE_FILE list. For
4340 EXCLUDE_FILE (*crtend.o *otherfile.o) *(.ctors)
4343 will cause all .ctors sections from all files except @file{crtend.o}
4344 and @file{otherfile.o} to be included. The EXCLUDE_FILE can also be
4345 placed inside the section list, for example:
4347 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4350 The result of this is identically to the previous example. Supporting
4351 two syntaxes for EXCLUDE_FILE is useful if the section list contains
4352 more than one section, as described below.
4354 There are two ways to include more than one section:
4360 The difference between these is the order in which the @samp{.text} and
4361 @samp{.rdata} input sections will appear in the output section. In the
4362 first example, they will be intermingled, appearing in the same order as
4363 they are found in the linker input. In the second example, all
4364 @samp{.text} input sections will appear first, followed by all
4365 @samp{.rdata} input sections.
4367 When using EXCLUDE_FILE with more than one section, if the exclusion
4368 is within the section list then the exclusion only applies to the
4369 immediately following section, for example:
4371 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4374 will cause all @samp{.text} sections from all files except
4375 @file{somefile.o} to be included, while all @samp{.rdata} sections
4376 from all files, including @file{somefile.o}, will be included. To
4377 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4378 could be modified to:
4380 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4383 Alternatively, placing the EXCLUDE_FILE outside of the section list,
4384 before the input file selection, will cause the exclusion to apply for
4385 all sections. Thus the previous example can be rewritten as:
4387 EXCLUDE_FILE (*somefile.o) *(.text .rdata)
4390 You can specify a file name to include sections from a particular file.
4391 You would do this if one or more of your files contain special data that
4392 needs to be at a particular location in memory. For example:
4397 To refine the sections that are included based on the section flags
4398 of an input section, INPUT_SECTION_FLAGS may be used.
4400 Here is a simple example for using Section header flags for ELF sections:
4405 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4406 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4411 In this example, the output section @samp{.text} will be comprised of any
4412 input section matching the name *(.text) whose section header flags
4413 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4414 @samp{.text2} will be comprised of any input section matching the name *(.text)
4415 whose section header flag @code{SHF_WRITE} is clear.
4417 You can also specify files within archives by writing a pattern
4418 matching the archive, a colon, then the pattern matching the file,
4419 with no whitespace around the colon.
4423 matches file within archive
4425 matches the whole archive
4427 matches file but not one in an archive
4430 Either one or both of @samp{archive} and @samp{file} can contain shell
4431 wildcards. On DOS based file systems, the linker will assume that a
4432 single letter followed by a colon is a drive specifier, so
4433 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4434 within an archive called @samp{c}. @samp{archive:file} filespecs may
4435 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4436 other linker script contexts. For instance, you cannot extract a file
4437 from an archive by using @samp{archive:file} in an @code{INPUT}
4440 If you use a file name without a list of sections, then all sections in
4441 the input file will be included in the output section. This is not
4442 commonly done, but it may by useful on occasion. For example:
4447 When you use a file name which is not an @samp{archive:file} specifier
4448 and does not contain any wild card
4449 characters, the linker will first see if you also specified the file
4450 name on the linker command line or in an @code{INPUT} command. If you
4451 did not, the linker will attempt to open the file as an input file, as
4452 though it appeared on the command line. Note that this differs from an
4453 @code{INPUT} command, because the linker will not search for the file in
4454 the archive search path.
4456 @node Input Section Wildcards
4457 @subsubsection Input Section Wildcard Patterns
4458 @cindex input section wildcards
4459 @cindex wildcard file name patterns
4460 @cindex file name wildcard patterns
4461 @cindex section name wildcard patterns
4462 In an input section description, either the file name or the section
4463 name or both may be wildcard patterns.
4465 The file name of @samp{*} seen in many examples is a simple wildcard
4466 pattern for the file name.
4468 The wildcard patterns are like those used by the Unix shell.
4472 matches any number of characters
4474 matches any single character
4476 matches a single instance of any of the @var{chars}; the @samp{-}
4477 character may be used to specify a range of characters, as in
4478 @samp{[a-z]} to match any lower case letter
4480 quotes the following character
4483 When a file name is matched with a wildcard, the wildcard characters
4484 will not match a @samp{/} character (used to separate directory names on
4485 Unix). A pattern consisting of a single @samp{*} character is an
4486 exception; it will always match any file name, whether it contains a
4487 @samp{/} or not. In a section name, the wildcard characters will match
4488 a @samp{/} character.
4490 File name wildcard patterns only match files which are explicitly
4491 specified on the command line or in an @code{INPUT} command. The linker
4492 does not search directories to expand wildcards.
4494 If a file name matches more than one wildcard pattern, or if a file name
4495 appears explicitly and is also matched by a wildcard pattern, the linker
4496 will use the first match in the linker script. For example, this
4497 sequence of input section descriptions is probably in error, because the
4498 @file{data.o} rule will not be used:
4500 .data : @{ *(.data) @}
4501 .data1 : @{ data.o(.data) @}
4504 @cindex SORT_BY_NAME
4505 Normally, the linker will place files and sections matched by wildcards
4506 in the order in which they are seen during the link. You can change
4507 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4508 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4509 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4510 into ascending order by name before placing them in the output file.
4512 @cindex SORT_BY_ALIGNMENT
4513 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4514 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4515 descending order by alignment before placing them in the output file.
4516 Larger alignments are placed before smaller alignments in order to
4517 reduce the amount of padding necessary.
4519 @cindex SORT_BY_INIT_PRIORITY
4520 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4521 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4522 ascending order by numerical value of the GCC init_priority attribute
4523 encoded in the section name before placing them in the output file.
4526 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4528 When there are nested section sorting commands in linker script, there
4529 can be at most 1 level of nesting for section sorting commands.
4533 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4534 It will sort the input sections by name first, then by alignment if two
4535 sections have the same name.
4537 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4538 It will sort the input sections by alignment first, then by name if two
4539 sections have the same alignment.
4541 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4542 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4544 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4545 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4547 All other nested section sorting commands are invalid.
4550 When both command line section sorting option and linker script
4551 section sorting command are used, section sorting command always
4552 takes precedence over the command line option.
4554 If the section sorting command in linker script isn't nested, the
4555 command line option will make the section sorting command to be
4556 treated as nested sorting command.
4560 @code{SORT_BY_NAME} (wildcard section pattern ) with
4561 @option{--sort-sections alignment} is equivalent to
4562 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4564 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4565 @option{--sort-section name} is equivalent to
4566 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4569 If the section sorting command in linker script is nested, the
4570 command line option will be ignored.
4573 @code{SORT_NONE} disables section sorting by ignoring the command line
4574 section sorting option.
4576 If you ever get confused about where input sections are going, use the
4577 @samp{-M} linker option to generate a map file. The map file shows
4578 precisely how input sections are mapped to output sections.
4580 This example shows how wildcard patterns might be used to partition
4581 files. This linker script directs the linker to place all @samp{.text}
4582 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4583 The linker will place the @samp{.data} section from all files beginning
4584 with an upper case character in @samp{.DATA}; for all other files, the
4585 linker will place the @samp{.data} section in @samp{.data}.
4589 .text : @{ *(.text) @}
4590 .DATA : @{ [A-Z]*(.data) @}
4591 .data : @{ *(.data) @}
4592 .bss : @{ *(.bss) @}
4597 @node Input Section Common
4598 @subsubsection Input Section for Common Symbols
4599 @cindex common symbol placement
4600 @cindex uninitialized data placement
4601 A special notation is needed for common symbols, because in many object
4602 file formats common symbols do not have a particular input section. The
4603 linker treats common symbols as though they are in an input section
4604 named @samp{COMMON}.
4606 You may use file names with the @samp{COMMON} section just as with any
4607 other input sections. You can use this to place common symbols from a
4608 particular input file in one section while common symbols from other
4609 input files are placed in another section.
4611 In most cases, common symbols in input files will be placed in the
4612 @samp{.bss} section in the output file. For example:
4614 .bss @{ *(.bss) *(COMMON) @}
4617 @cindex scommon section
4618 @cindex small common symbols
4619 Some object file formats have more than one type of common symbol. For
4620 example, the MIPS ELF object file format distinguishes standard common
4621 symbols and small common symbols. In this case, the linker will use a
4622 different special section name for other types of common symbols. In
4623 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4624 symbols and @samp{.scommon} for small common symbols. This permits you
4625 to map the different types of common symbols into memory at different
4629 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4630 notation is now considered obsolete. It is equivalent to
4633 @node Input Section Keep
4634 @subsubsection Input Section and Garbage Collection
4636 @cindex garbage collection
4637 When link-time garbage collection is in use (@samp{--gc-sections}),
4638 it is often useful to mark sections that should not be eliminated.
4639 This is accomplished by surrounding an input section's wildcard entry
4640 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4641 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4643 @node Input Section Example
4644 @subsubsection Input Section Example
4645 The following example is a complete linker script. It tells the linker
4646 to read all of the sections from file @file{all.o} and place them at the
4647 start of output section @samp{outputa} which starts at location
4648 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4649 follows immediately, in the same output section. All of section
4650 @samp{.input2} from @file{foo.o} goes into output section
4651 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4652 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4653 files are written to output section @samp{outputc}.
4681 If an output section's name is the same as the input section's name
4682 and is representable as a C identifier, then the linker will
4683 automatically @pxref{PROVIDE} two symbols: __start_SECNAME and
4684 __stop_SECNAME, where SECNAME is the name of the section. These
4685 indicate the start address and end address of the output section
4686 respectively. Note: most section names are not representable as
4687 C identifiers because they contain a @samp{.} character.
4689 @node Output Section Data
4690 @subsection Output Section Data
4692 @cindex section data
4693 @cindex output section data
4694 @kindex BYTE(@var{expression})
4695 @kindex SHORT(@var{expression})
4696 @kindex LONG(@var{expression})
4697 @kindex QUAD(@var{expression})
4698 @kindex SQUAD(@var{expression})
4699 You can include explicit bytes of data in an output section by using
4700 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4701 an output section command. Each keyword is followed by an expression in
4702 parentheses providing the value to store (@pxref{Expressions}). The
4703 value of the expression is stored at the current value of the location
4706 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4707 store one, two, four, and eight bytes (respectively). After storing the
4708 bytes, the location counter is incremented by the number of bytes
4711 For example, this will store the byte 1 followed by the four byte value
4712 of the symbol @samp{addr}:
4718 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4719 same; they both store an 8 byte, or 64 bit, value. When both host and
4720 target are 32 bits, an expression is computed as 32 bits. In this case
4721 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4722 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4724 If the object file format of the output file has an explicit endianness,
4725 which is the normal case, the value will be stored in that endianness.
4726 When the object file format does not have an explicit endianness, as is
4727 true of, for example, S-records, the value will be stored in the
4728 endianness of the first input object file.
4730 Note---these commands only work inside a section description and not
4731 between them, so the following will produce an error from the linker:
4733 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4735 whereas this will work:
4737 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4740 @kindex FILL(@var{expression})
4741 @cindex holes, filling
4742 @cindex unspecified memory
4743 You may use the @code{FILL} command to set the fill pattern for the
4744 current section. It is followed by an expression in parentheses. Any
4745 otherwise unspecified regions of memory within the section (for example,
4746 gaps left due to the required alignment of input sections) are filled
4747 with the value of the expression, repeated as
4748 necessary. A @code{FILL} statement covers memory locations after the
4749 point at which it occurs in the section definition; by including more
4750 than one @code{FILL} statement, you can have different fill patterns in
4751 different parts of an output section.
4753 This example shows how to fill unspecified regions of memory with the
4759 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4760 section attribute, but it only affects the
4761 part of the section following the @code{FILL} command, rather than the
4762 entire section. If both are used, the @code{FILL} command takes
4763 precedence. @xref{Output Section Fill}, for details on the fill
4766 @node Output Section Keywords
4767 @subsection Output Section Keywords
4768 There are a couple of keywords which can appear as output section
4772 @kindex CREATE_OBJECT_SYMBOLS
4773 @cindex input filename symbols
4774 @cindex filename symbols
4775 @item CREATE_OBJECT_SYMBOLS
4776 The command tells the linker to create a symbol for each input file.
4777 The name of each symbol will be the name of the corresponding input
4778 file. The section of each symbol will be the output section in which
4779 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4781 This is conventional for the a.out object file format. It is not
4782 normally used for any other object file format.
4784 @kindex CONSTRUCTORS
4785 @cindex C++ constructors, arranging in link
4786 @cindex constructors, arranging in link
4788 When linking using the a.out object file format, the linker uses an
4789 unusual set construct to support C++ global constructors and
4790 destructors. When linking object file formats which do not support
4791 arbitrary sections, such as ECOFF and XCOFF, the linker will
4792 automatically recognize C++ global constructors and destructors by name.
4793 For these object file formats, the @code{CONSTRUCTORS} command tells the
4794 linker to place constructor information in the output section where the
4795 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4796 ignored for other object file formats.
4798 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4799 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4800 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4801 the start and end of the global destructors. The
4802 first word in the list is the number of entries, followed by the address
4803 of each constructor or destructor, followed by a zero word. The
4804 compiler must arrange to actually run the code. For these object file
4805 formats @sc{gnu} C++ normally calls constructors from a subroutine
4806 @code{__main}; a call to @code{__main} is automatically inserted into
4807 the startup code for @code{main}. @sc{gnu} C++ normally runs
4808 destructors either by using @code{atexit}, or directly from the function
4811 For object file formats such as @code{COFF} or @code{ELF} which support
4812 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4813 addresses of global constructors and destructors into the @code{.ctors}
4814 and @code{.dtors} sections. Placing the following sequence into your
4815 linker script will build the sort of table which the @sc{gnu} C++
4816 runtime code expects to see.
4820 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4825 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4831 If you are using the @sc{gnu} C++ support for initialization priority,
4832 which provides some control over the order in which global constructors
4833 are run, you must sort the constructors at link time to ensure that they
4834 are executed in the correct order. When using the @code{CONSTRUCTORS}
4835 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4836 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4837 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4840 Normally the compiler and linker will handle these issues automatically,
4841 and you will not need to concern yourself with them. However, you may
4842 need to consider this if you are using C++ and writing your own linker
4847 @node Output Section Discarding
4848 @subsection Output Section Discarding
4849 @cindex discarding sections
4850 @cindex sections, discarding
4851 @cindex removing sections
4852 The linker will not normally create output sections with no contents.
4853 This is for convenience when referring to input sections that may or
4854 may not be present in any of the input files. For example:
4856 .foo : @{ *(.foo) @}
4859 will only create a @samp{.foo} section in the output file if there is a
4860 @samp{.foo} section in at least one input file, and if the input
4861 sections are not all empty. Other link script directives that allocate
4862 space in an output section will also create the output section. So
4863 too will assignments to dot even if the assignment does not create
4864 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
4865 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
4866 @samp{sym} is an absolute symbol of value 0 defined in the script.
4867 This allows you to force output of an empty section with @samp{. = .}.
4869 The linker will ignore address assignments (@pxref{Output Section Address})
4870 on discarded output sections, except when the linker script defines
4871 symbols in the output section. In that case the linker will obey
4872 the address assignments, possibly advancing dot even though the
4873 section is discarded.
4876 The special output section name @samp{/DISCARD/} may be used to discard
4877 input sections. Any input sections which are assigned to an output
4878 section named @samp{/DISCARD/} are not included in the output file.
4880 @node Output Section Attributes
4881 @subsection Output Section Attributes
4882 @cindex output section attributes
4883 We showed above that the full description of an output section looked
4888 @var{section} [@var{address}] [(@var{type})] :
4890 [ALIGN(@var{section_align})]
4891 [SUBALIGN(@var{subsection_align})]
4894 @var{output-section-command}
4895 @var{output-section-command}
4897 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4901 We've already described @var{section}, @var{address}, and
4902 @var{output-section-command}. In this section we will describe the
4903 remaining section attributes.
4906 * Output Section Type:: Output section type
4907 * Output Section LMA:: Output section LMA
4908 * Forced Output Alignment:: Forced Output Alignment
4909 * Forced Input Alignment:: Forced Input Alignment
4910 * Output Section Constraint:: Output section constraint
4911 * Output Section Region:: Output section region
4912 * Output Section Phdr:: Output section phdr
4913 * Output Section Fill:: Output section fill
4916 @node Output Section Type
4917 @subsubsection Output Section Type
4918 Each output section may have a type. The type is a keyword in
4919 parentheses. The following types are defined:
4923 The section should be marked as not loadable, so that it will not be
4924 loaded into memory when the program is run.
4929 These type names are supported for backward compatibility, and are
4930 rarely used. They all have the same effect: the section should be
4931 marked as not allocatable, so that no memory is allocated for the
4932 section when the program is run.
4936 @cindex prevent unnecessary loading
4937 @cindex loading, preventing
4938 The linker normally sets the attributes of an output section based on
4939 the input sections which map into it. You can override this by using
4940 the section type. For example, in the script sample below, the
4941 @samp{ROM} section is addressed at memory location @samp{0} and does not
4942 need to be loaded when the program is run.
4946 ROM 0 (NOLOAD) : @{ @dots{} @}
4952 @node Output Section LMA
4953 @subsubsection Output Section LMA
4954 @kindex AT>@var{lma_region}
4955 @kindex AT(@var{lma})
4956 @cindex load address
4957 @cindex section load address
4958 Every section has a virtual address (VMA) and a load address (LMA); see
4959 @ref{Basic Script Concepts}. The virtual address is specified by the
4960 @pxref{Output Section Address} described earlier. The load address is
4961 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
4962 address is optional.
4964 The @code{AT} keyword takes an expression as an argument. This
4965 specifies the exact load address of the section. The @code{AT>} keyword
4966 takes the name of a memory region as an argument. @xref{MEMORY}. The
4967 load address of the section is set to the next free address in the
4968 region, aligned to the section's alignment requirements.
4970 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4971 section, the linker will use the following heuristic to determine the
4976 If the section has a specific VMA address, then this is used as
4977 the LMA address as well.
4980 If the section is not allocatable then its LMA is set to its VMA.
4983 Otherwise if a memory region can be found that is compatible
4984 with the current section, and this region contains at least one
4985 section, then the LMA is set so the difference between the
4986 VMA and LMA is the same as the difference between the VMA and LMA of
4987 the last section in the located region.
4990 If no memory regions have been declared then a default region
4991 that covers the entire address space is used in the previous step.
4994 If no suitable region could be found, or there was no previous
4995 section then the LMA is set equal to the VMA.
4998 @cindex ROM initialized data
4999 @cindex initialized data in ROM
5000 This feature is designed to make it easy to build a ROM image. For
5001 example, the following linker script creates three output sections: one
5002 called @samp{.text}, which starts at @code{0x1000}, one called
5003 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
5004 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
5005 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
5006 defined with the value @code{0x2000}, which shows that the location
5007 counter holds the VMA value, not the LMA value.
5013 .text 0x1000 : @{ *(.text) _etext = . ; @}
5015 AT ( ADDR (.text) + SIZEOF (.text) )
5016 @{ _data = . ; *(.data); _edata = . ; @}
5018 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
5023 The run-time initialization code for use with a program generated with
5024 this linker script would include something like the following, to copy
5025 the initialized data from the ROM image to its runtime address. Notice
5026 how this code takes advantage of the symbols defined by the linker
5031 extern char _etext, _data, _edata, _bstart, _bend;
5032 char *src = &_etext;
5035 /* ROM has data at end of text; copy it. */
5036 while (dst < &_edata)
5040 for (dst = &_bstart; dst< &_bend; dst++)
5045 @node Forced Output Alignment
5046 @subsubsection Forced Output Alignment
5047 @kindex ALIGN(@var{section_align})
5048 @cindex forcing output section alignment
5049 @cindex output section alignment
5050 You can increase an output section's alignment by using ALIGN. As an
5051 alternative you can enforce that the difference between the VMA and LMA remains
5052 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
5054 @node Forced Input Alignment
5055 @subsubsection Forced Input Alignment
5056 @kindex SUBALIGN(@var{subsection_align})
5057 @cindex forcing input section alignment
5058 @cindex input section alignment
5059 You can force input section alignment within an output section by using
5060 SUBALIGN. The value specified overrides any alignment given by input
5061 sections, whether larger or smaller.
5063 @node Output Section Constraint
5064 @subsubsection Output Section Constraint
5067 @cindex constraints on output sections
5068 You can specify that an output section should only be created if all
5069 of its input sections are read-only or all of its input sections are
5070 read-write by using the keyword @code{ONLY_IF_RO} and
5071 @code{ONLY_IF_RW} respectively.
5073 @node Output Section Region
5074 @subsubsection Output Section Region
5075 @kindex >@var{region}
5076 @cindex section, assigning to memory region
5077 @cindex memory regions and sections
5078 You can assign a section to a previously defined region of memory by
5079 using @samp{>@var{region}}. @xref{MEMORY}.
5081 Here is a simple example:
5084 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
5085 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
5089 @node Output Section Phdr
5090 @subsubsection Output Section Phdr
5092 @cindex section, assigning to program header
5093 @cindex program headers and sections
5094 You can assign a section to a previously defined program segment by
5095 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
5096 one or more segments, then all subsequent allocated sections will be
5097 assigned to those segments as well, unless they use an explicitly
5098 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
5099 linker to not put the section in any segment at all.
5101 Here is a simple example:
5104 PHDRS @{ text PT_LOAD ; @}
5105 SECTIONS @{ .text : @{ *(.text) @} :text @}
5109 @node Output Section Fill
5110 @subsubsection Output Section Fill
5111 @kindex =@var{fillexp}
5112 @cindex section fill pattern
5113 @cindex fill pattern, entire section
5114 You can set the fill pattern for an entire section by using
5115 @samp{=@var{fillexp}}. @var{fillexp} is an expression
5116 (@pxref{Expressions}). Any otherwise unspecified regions of memory
5117 within the output section (for example, gaps left due to the required
5118 alignment of input sections) will be filled with the value, repeated as
5119 necessary. If the fill expression is a simple hex number, ie. a string
5120 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
5121 an arbitrarily long sequence of hex digits can be used to specify the
5122 fill pattern; Leading zeros become part of the pattern too. For all
5123 other cases, including extra parentheses or a unary @code{+}, the fill
5124 pattern is the four least significant bytes of the value of the
5125 expression. In all cases, the number is big-endian.
5127 You can also change the fill value with a @code{FILL} command in the
5128 output section commands; (@pxref{Output Section Data}).
5130 Here is a simple example:
5133 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
5137 @node Overlay Description
5138 @subsection Overlay Description
5141 An overlay description provides an easy way to describe sections which
5142 are to be loaded as part of a single memory image but are to be run at
5143 the same memory address. At run time, some sort of overlay manager will
5144 copy the overlaid sections in and out of the runtime memory address as
5145 required, perhaps by simply manipulating addressing bits. This approach
5146 can be useful, for example, when a certain region of memory is faster
5149 Overlays are described using the @code{OVERLAY} command. The
5150 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
5151 output section description. The full syntax of the @code{OVERLAY}
5152 command is as follows:
5155 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
5159 @var{output-section-command}
5160 @var{output-section-command}
5162 @} [:@var{phdr}@dots{}] [=@var{fill}]
5165 @var{output-section-command}
5166 @var{output-section-command}
5168 @} [:@var{phdr}@dots{}] [=@var{fill}]
5170 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5174 Everything is optional except @code{OVERLAY} (a keyword), and each
5175 section must have a name (@var{secname1} and @var{secname2} above). The
5176 section definitions within the @code{OVERLAY} construct are identical to
5177 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5178 except that no addresses and no memory regions may be defined for
5179 sections within an @code{OVERLAY}.
5181 The comma at the end may be required if a @var{fill} is used and
5182 the next @var{sections-command} looks like a continuation of the expression.
5184 The sections are all defined with the same starting address. The load
5185 addresses of the sections are arranged such that they are consecutive in
5186 memory starting at the load address used for the @code{OVERLAY} as a
5187 whole (as with normal section definitions, the load address is optional,
5188 and defaults to the start address; the start address is also optional,
5189 and defaults to the current value of the location counter).
5191 If the @code{NOCROSSREFS} keyword is used, and there are any
5192 references among the sections, the linker will report an error. Since
5193 the sections all run at the same address, it normally does not make
5194 sense for one section to refer directly to another.
5195 @xref{Miscellaneous Commands, NOCROSSREFS}.
5197 For each section within the @code{OVERLAY}, the linker automatically
5198 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5199 defined as the starting load address of the section. The symbol
5200 @code{__load_stop_@var{secname}} is defined as the final load address of
5201 the section. Any characters within @var{secname} which are not legal
5202 within C identifiers are removed. C (or assembler) code may use these
5203 symbols to move the overlaid sections around as necessary.
5205 At the end of the overlay, the value of the location counter is set to
5206 the start address of the overlay plus the size of the largest section.
5208 Here is an example. Remember that this would appear inside a
5209 @code{SECTIONS} construct.
5212 OVERLAY 0x1000 : AT (0x4000)
5214 .text0 @{ o1/*.o(.text) @}
5215 .text1 @{ o2/*.o(.text) @}
5220 This will define both @samp{.text0} and @samp{.text1} to start at
5221 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5222 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5223 following symbols will be defined if referenced: @code{__load_start_text0},
5224 @code{__load_stop_text0}, @code{__load_start_text1},
5225 @code{__load_stop_text1}.
5227 C code to copy overlay @code{.text1} into the overlay area might look
5232 extern char __load_start_text1, __load_stop_text1;
5233 memcpy ((char *) 0x1000, &__load_start_text1,
5234 &__load_stop_text1 - &__load_start_text1);
5238 Note that the @code{OVERLAY} command is just syntactic sugar, since
5239 everything it does can be done using the more basic commands. The above
5240 example could have been written identically as follows.
5244 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5245 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5246 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5247 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5248 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5249 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5250 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5255 @section MEMORY Command
5257 @cindex memory regions
5258 @cindex regions of memory
5259 @cindex allocating memory
5260 @cindex discontinuous memory
5261 The linker's default configuration permits allocation of all available
5262 memory. You can override this by using the @code{MEMORY} command.
5264 The @code{MEMORY} command describes the location and size of blocks of
5265 memory in the target. You can use it to describe which memory regions
5266 may be used by the linker, and which memory regions it must avoid. You
5267 can then assign sections to particular memory regions. The linker will
5268 set section addresses based on the memory regions, and will warn about
5269 regions that become too full. The linker will not shuffle sections
5270 around to fit into the available regions.
5272 A linker script may contain many uses of the @code{MEMORY} command,
5273 however, all memory blocks defined are treated as if they were
5274 specified inside a single @code{MEMORY} command. The syntax for
5280 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5286 The @var{name} is a name used in the linker script to refer to the
5287 region. The region name has no meaning outside of the linker script.
5288 Region names are stored in a separate name space, and will not conflict
5289 with symbol names, file names, or section names. Each memory region
5290 must have a distinct name within the @code{MEMORY} command. However you can
5291 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5294 @cindex memory region attributes
5295 The @var{attr} string is an optional list of attributes that specify
5296 whether to use a particular memory region for an input section which is
5297 not explicitly mapped in the linker script. As described in
5298 @ref{SECTIONS}, if you do not specify an output section for some input
5299 section, the linker will create an output section with the same name as
5300 the input section. If you define region attributes, the linker will use
5301 them to select the memory region for the output section that it creates.
5303 The @var{attr} string must consist only of the following characters:
5318 Invert the sense of any of the attributes that follow
5321 If a unmapped section matches any of the listed attributes other than
5322 @samp{!}, it will be placed in the memory region. The @samp{!}
5323 attribute reverses this test, so that an unmapped section will be placed
5324 in the memory region only if it does not match any of the listed
5330 The @var{origin} is an numerical expression for the start address of
5331 the memory region. The expression must evaluate to a constant and it
5332 cannot involve any symbols. The keyword @code{ORIGIN} may be
5333 abbreviated to @code{org} or @code{o} (but not, for example,
5339 The @var{len} is an expression for the size in bytes of the memory
5340 region. As with the @var{origin} expression, the expression must
5341 be numerical only and must evaluate to a constant. The keyword
5342 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5344 In the following example, we specify that there are two memory regions
5345 available for allocation: one starting at @samp{0} for 256 kilobytes,
5346 and the other starting at @samp{0x40000000} for four megabytes. The
5347 linker will place into the @samp{rom} memory region every section which
5348 is not explicitly mapped into a memory region, and is either read-only
5349 or executable. The linker will place other sections which are not
5350 explicitly mapped into a memory region into the @samp{ram} memory
5357 rom (rx) : ORIGIN = 0, LENGTH = 256K
5358 ram (!rx) : org = 0x40000000, l = 4M
5363 Once you define a memory region, you can direct the linker to place
5364 specific output sections into that memory region by using the
5365 @samp{>@var{region}} output section attribute. For example, if you have
5366 a memory region named @samp{mem}, you would use @samp{>mem} in the
5367 output section definition. @xref{Output Section Region}. If no address
5368 was specified for the output section, the linker will set the address to
5369 the next available address within the memory region. If the combined
5370 output sections directed to a memory region are too large for the
5371 region, the linker will issue an error message.
5373 It is possible to access the origin and length of a memory in an
5374 expression via the @code{ORIGIN(@var{memory})} and
5375 @code{LENGTH(@var{memory})} functions:
5379 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5384 @section PHDRS Command
5386 @cindex program headers
5387 @cindex ELF program headers
5388 @cindex program segments
5389 @cindex segments, ELF
5390 The ELF object file format uses @dfn{program headers}, also knows as
5391 @dfn{segments}. The program headers describe how the program should be
5392 loaded into memory. You can print them out by using the @code{objdump}
5393 program with the @samp{-p} option.
5395 When you run an ELF program on a native ELF system, the system loader
5396 reads the program headers in order to figure out how to load the
5397 program. This will only work if the program headers are set correctly.
5398 This manual does not describe the details of how the system loader
5399 interprets program headers; for more information, see the ELF ABI.
5401 The linker will create reasonable program headers by default. However,
5402 in some cases, you may need to specify the program headers more
5403 precisely. You may use the @code{PHDRS} command for this purpose. When
5404 the linker sees the @code{PHDRS} command in the linker script, it will
5405 not create any program headers other than the ones specified.
5407 The linker only pays attention to the @code{PHDRS} command when
5408 generating an ELF output file. In other cases, the linker will simply
5409 ignore @code{PHDRS}.
5411 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5412 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5418 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5419 [ FLAGS ( @var{flags} ) ] ;
5424 The @var{name} is used only for reference in the @code{SECTIONS} command
5425 of the linker script. It is not put into the output file. Program
5426 header names are stored in a separate name space, and will not conflict
5427 with symbol names, file names, or section names. Each program header
5428 must have a distinct name. The headers are processed in order and it
5429 is usual for them to map to sections in ascending load address order.
5431 Certain program header types describe segments of memory which the
5432 system loader will load from the file. In the linker script, you
5433 specify the contents of these segments by placing allocatable output
5434 sections in the segments. You use the @samp{:@var{phdr}} output section
5435 attribute to place a section in a particular segment. @xref{Output
5438 It is normal to put certain sections in more than one segment. This
5439 merely implies that one segment of memory contains another. You may
5440 repeat @samp{:@var{phdr}}, using it once for each segment which should
5441 contain the section.
5443 If you place a section in one or more segments using @samp{:@var{phdr}},
5444 then the linker will place all subsequent allocatable sections which do
5445 not specify @samp{:@var{phdr}} in the same segments. This is for
5446 convenience, since generally a whole set of contiguous sections will be
5447 placed in a single segment. You can use @code{:NONE} to override the
5448 default segment and tell the linker to not put the section in any
5453 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
5454 the program header type to further describe the contents of the segment.
5455 The @code{FILEHDR} keyword means that the segment should include the ELF
5456 file header. The @code{PHDRS} keyword means that the segment should
5457 include the ELF program headers themselves. If applied to a loadable
5458 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5461 The @var{type} may be one of the following. The numbers indicate the
5462 value of the keyword.
5465 @item @code{PT_NULL} (0)
5466 Indicates an unused program header.
5468 @item @code{PT_LOAD} (1)
5469 Indicates that this program header describes a segment to be loaded from
5472 @item @code{PT_DYNAMIC} (2)
5473 Indicates a segment where dynamic linking information can be found.
5475 @item @code{PT_INTERP} (3)
5476 Indicates a segment where the name of the program interpreter may be
5479 @item @code{PT_NOTE} (4)
5480 Indicates a segment holding note information.
5482 @item @code{PT_SHLIB} (5)
5483 A reserved program header type, defined but not specified by the ELF
5486 @item @code{PT_PHDR} (6)
5487 Indicates a segment where the program headers may be found.
5489 @item @code{PT_TLS} (7)
5490 Indicates a segment containing thread local storage.
5492 @item @var{expression}
5493 An expression giving the numeric type of the program header. This may
5494 be used for types not defined above.
5497 You can specify that a segment should be loaded at a particular address
5498 in memory by using an @code{AT} expression. This is identical to the
5499 @code{AT} command used as an output section attribute (@pxref{Output
5500 Section LMA}). The @code{AT} command for a program header overrides the
5501 output section attribute.
5503 The linker will normally set the segment flags based on the sections
5504 which comprise the segment. You may use the @code{FLAGS} keyword to
5505 explicitly specify the segment flags. The value of @var{flags} must be
5506 an integer. It is used to set the @code{p_flags} field of the program
5509 Here is an example of @code{PHDRS}. This shows a typical set of program
5510 headers used on a native ELF system.
5516 headers PT_PHDR PHDRS ;
5518 text PT_LOAD FILEHDR PHDRS ;
5520 dynamic PT_DYNAMIC ;
5526 .interp : @{ *(.interp) @} :text :interp
5527 .text : @{ *(.text) @} :text
5528 .rodata : @{ *(.rodata) @} /* defaults to :text */
5530 . = . + 0x1000; /* move to a new page in memory */
5531 .data : @{ *(.data) @} :data
5532 .dynamic : @{ *(.dynamic) @} :data :dynamic
5539 @section VERSION Command
5540 @kindex VERSION @{script text@}
5541 @cindex symbol versions
5542 @cindex version script
5543 @cindex versions of symbols
5544 The linker supports symbol versions when using ELF. Symbol versions are
5545 only useful when using shared libraries. The dynamic linker can use
5546 symbol versions to select a specific version of a function when it runs
5547 a program that may have been linked against an earlier version of the
5550 You can include a version script directly in the main linker script, or
5551 you can supply the version script as an implicit linker script. You can
5552 also use the @samp{--version-script} linker option.
5554 The syntax of the @code{VERSION} command is simply
5556 VERSION @{ version-script-commands @}
5559 The format of the version script commands is identical to that used by
5560 Sun's linker in Solaris 2.5. The version script defines a tree of
5561 version nodes. You specify the node names and interdependencies in the
5562 version script. You can specify which symbols are bound to which
5563 version nodes, and you can reduce a specified set of symbols to local
5564 scope so that they are not globally visible outside of the shared
5567 The easiest way to demonstrate the version script language is with a few
5593 This example version script defines three version nodes. The first
5594 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5595 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5596 a number of symbols to local scope so that they are not visible outside
5597 of the shared library; this is done using wildcard patterns, so that any
5598 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5599 is matched. The wildcard patterns available are the same as those used
5600 in the shell when matching filenames (also known as ``globbing'').
5601 However, if you specify the symbol name inside double quotes, then the
5602 name is treated as literal, rather than as a glob pattern.
5604 Next, the version script defines node @samp{VERS_1.2}. This node
5605 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5606 to the version node @samp{VERS_1.2}.
5608 Finally, the version script defines node @samp{VERS_2.0}. This node
5609 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5610 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5612 When the linker finds a symbol defined in a library which is not
5613 specifically bound to a version node, it will effectively bind it to an
5614 unspecified base version of the library. You can bind all otherwise
5615 unspecified symbols to a given version node by using @samp{global: *;}
5616 somewhere in the version script. Note that it's slightly crazy to use
5617 wildcards in a global spec except on the last version node. Global
5618 wildcards elsewhere run the risk of accidentally adding symbols to the
5619 set exported for an old version. That's wrong since older versions
5620 ought to have a fixed set of symbols.
5622 The names of the version nodes have no specific meaning other than what
5623 they might suggest to the person reading them. The @samp{2.0} version
5624 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5625 However, this would be a confusing way to write a version script.
5627 Node name can be omitted, provided it is the only version node
5628 in the version script. Such version script doesn't assign any versions to
5629 symbols, only selects which symbols will be globally visible out and which
5633 @{ global: foo; bar; local: *; @};
5636 When you link an application against a shared library that has versioned
5637 symbols, the application itself knows which version of each symbol it
5638 requires, and it also knows which version nodes it needs from each
5639 shared library it is linked against. Thus at runtime, the dynamic
5640 loader can make a quick check to make sure that the libraries you have
5641 linked against do in fact supply all of the version nodes that the
5642 application will need to resolve all of the dynamic symbols. In this
5643 way it is possible for the dynamic linker to know with certainty that
5644 all external symbols that it needs will be resolvable without having to
5645 search for each symbol reference.
5647 The symbol versioning is in effect a much more sophisticated way of
5648 doing minor version checking that SunOS does. The fundamental problem
5649 that is being addressed here is that typically references to external
5650 functions are bound on an as-needed basis, and are not all bound when
5651 the application starts up. If a shared library is out of date, a
5652 required interface may be missing; when the application tries to use
5653 that interface, it may suddenly and unexpectedly fail. With symbol
5654 versioning, the user will get a warning when they start their program if
5655 the libraries being used with the application are too old.
5657 There are several GNU extensions to Sun's versioning approach. The
5658 first of these is the ability to bind a symbol to a version node in the
5659 source file where the symbol is defined instead of in the versioning
5660 script. This was done mainly to reduce the burden on the library
5661 maintainer. You can do this by putting something like:
5663 __asm__(".symver original_foo,foo@@VERS_1.1");
5666 in the C source file. This renames the function @samp{original_foo} to
5667 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5668 The @samp{local:} directive can be used to prevent the symbol
5669 @samp{original_foo} from being exported. A @samp{.symver} directive
5670 takes precedence over a version script.
5672 The second GNU extension is to allow multiple versions of the same
5673 function to appear in a given shared library. In this way you can make
5674 an incompatible change to an interface without increasing the major
5675 version number of the shared library, while still allowing applications
5676 linked against the old interface to continue to function.
5678 To do this, you must use multiple @samp{.symver} directives in the
5679 source file. Here is an example:
5682 __asm__(".symver original_foo,foo@@");
5683 __asm__(".symver old_foo,foo@@VERS_1.1");
5684 __asm__(".symver old_foo1,foo@@VERS_1.2");
5685 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5688 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5689 unspecified base version of the symbol. The source file that contains this
5690 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5691 @samp{old_foo1}, and @samp{new_foo}.
5693 When you have multiple definitions of a given symbol, there needs to be
5694 some way to specify a default version to which external references to
5695 this symbol will be bound. You can do this with the
5696 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5697 declare one version of a symbol as the default in this manner; otherwise
5698 you would effectively have multiple definitions of the same symbol.
5700 If you wish to bind a reference to a specific version of the symbol
5701 within the shared library, you can use the aliases of convenience
5702 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5703 specifically bind to an external version of the function in question.
5705 You can also specify the language in the version script:
5708 VERSION extern "lang" @{ version-script-commands @}
5711 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5712 The linker will iterate over the list of symbols at the link time and
5713 demangle them according to @samp{lang} before matching them to the
5714 patterns specified in @samp{version-script-commands}. The default
5715 @samp{lang} is @samp{C}.
5717 Demangled names may contains spaces and other special characters. As
5718 described above, you can use a glob pattern to match demangled names,
5719 or you can use a double-quoted string to match the string exactly. In
5720 the latter case, be aware that minor differences (such as differing
5721 whitespace) between the version script and the demangler output will
5722 cause a mismatch. As the exact string generated by the demangler
5723 might change in the future, even if the mangled name does not, you
5724 should check that all of your version directives are behaving as you
5725 expect when you upgrade.
5728 @section Expressions in Linker Scripts
5731 The syntax for expressions in the linker script language is identical to
5732 that of C expressions. All expressions are evaluated as integers. All
5733 expressions are evaluated in the same size, which is 32 bits if both the
5734 host and target are 32 bits, and is otherwise 64 bits.
5736 You can use and set symbol values in expressions.
5738 The linker defines several special purpose builtin functions for use in
5742 * Constants:: Constants
5743 * Symbolic Constants:: Symbolic constants
5744 * Symbols:: Symbol Names
5745 * Orphan Sections:: Orphan Sections
5746 * Location Counter:: The Location Counter
5747 * Operators:: Operators
5748 * Evaluation:: Evaluation
5749 * Expression Section:: The Section of an Expression
5750 * Builtin Functions:: Builtin Functions
5754 @subsection Constants
5755 @cindex integer notation
5756 @cindex constants in linker scripts
5757 All constants are integers.
5759 As in C, the linker considers an integer beginning with @samp{0} to be
5760 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5761 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5762 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5763 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5764 value without a prefix or a suffix is considered to be decimal.
5766 @cindex scaled integers
5767 @cindex K and M integer suffixes
5768 @cindex M and K integer suffixes
5769 @cindex suffixes for integers
5770 @cindex integer suffixes
5771 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5775 @c END TEXI2ROFF-KILL
5776 @code{1024} or @code{1024*1024}
5780 ${\rm 1024}$ or ${\rm 1024}^2$
5782 @c END TEXI2ROFF-KILL
5783 respectively. For example, the following
5784 all refer to the same quantity:
5793 Note - the @code{K} and @code{M} suffixes cannot be used in
5794 conjunction with the base suffixes mentioned above.
5796 @node Symbolic Constants
5797 @subsection Symbolic Constants
5798 @cindex symbolic constants
5800 It is possible to refer to target specific constants via the use of
5801 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5806 The target's maximum page size.
5808 @item COMMONPAGESIZE
5809 @kindex COMMONPAGESIZE
5810 The target's default page size.
5816 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5819 will create a text section aligned to the largest page boundary
5820 supported by the target.
5823 @subsection Symbol Names
5824 @cindex symbol names
5826 @cindex quoted symbol names
5828 Unless quoted, symbol names start with a letter, underscore, or period
5829 and may include letters, digits, underscores, periods, and hyphens.
5830 Unquoted symbol names must not conflict with any keywords. You can
5831 specify a symbol which contains odd characters or has the same name as a
5832 keyword by surrounding the symbol name in double quotes:
5835 "with a space" = "also with a space" + 10;
5838 Since symbols can contain many non-alphabetic characters, it is safest
5839 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5840 whereas @samp{A - B} is an expression involving subtraction.
5842 @node Orphan Sections
5843 @subsection Orphan Sections
5845 Orphan sections are sections present in the input files which
5846 are not explicitly placed into the output file by the linker
5847 script. The linker will still copy these sections into the
5848 output file by either finding, or creating a suitable output section
5849 in which to place the orphaned input section.
5851 If the name of an orphaned input section exactly matches the name of
5852 an existing output section, then the orphaned input section will be
5853 placed at the end of that output section.
5855 If there is no output section with a matching name then new output
5856 sections will be created. Each new output section will have the same
5857 name as the orphan section placed within it. If there are multiple
5858 orphan sections with the same name, these will all be combined into
5859 one new output section.
5861 If new output sections are created to hold orphaned input sections,
5862 then the linker must decide where to place these new output sections
5863 in relation to existing output sections. On most modern targets, the
5864 linker attempts to place orphan sections after sections of the same
5865 attribute, such as code vs data, loadable vs non-loadable, etc. If no
5866 sections with matching attributes are found, or your target lacks this
5867 support, the orphan section is placed at the end of the file.
5869 The command line options @samp{--orphan-handling} and @samp{--unique}
5870 (@pxref{Options,,Command Line Options}) can be used to control which
5871 output sections an orphan is placed in.
5873 @node Location Counter
5874 @subsection The Location Counter
5877 @cindex location counter
5878 @cindex current output location
5879 The special linker variable @dfn{dot} @samp{.} always contains the
5880 current output location counter. Since the @code{.} always refers to a
5881 location in an output section, it may only appear in an expression
5882 within a @code{SECTIONS} command. The @code{.} symbol may appear
5883 anywhere that an ordinary symbol is allowed in an expression.
5886 Assigning a value to @code{.} will cause the location counter to be
5887 moved. This may be used to create holes in the output section. The
5888 location counter may not be moved backwards inside an output section,
5889 and may not be moved backwards outside of an output section if so
5890 doing creates areas with overlapping LMAs.
5906 In the previous example, the @samp{.text} section from @file{file1} is
5907 located at the beginning of the output section @samp{output}. It is
5908 followed by a 1000 byte gap. Then the @samp{.text} section from
5909 @file{file2} appears, also with a 1000 byte gap following before the
5910 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5911 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5913 @cindex dot inside sections
5914 Note: @code{.} actually refers to the byte offset from the start of the
5915 current containing object. Normally this is the @code{SECTIONS}
5916 statement, whose start address is 0, hence @code{.} can be used as an
5917 absolute address. If @code{.} is used inside a section description
5918 however, it refers to the byte offset from the start of that section,
5919 not an absolute address. Thus in a script like this:
5937 The @samp{.text} section will be assigned a starting address of 0x100
5938 and a size of exactly 0x200 bytes, even if there is not enough data in
5939 the @samp{.text} input sections to fill this area. (If there is too
5940 much data, an error will be produced because this would be an attempt to
5941 move @code{.} backwards). The @samp{.data} section will start at 0x500
5942 and it will have an extra 0x600 bytes worth of space after the end of
5943 the values from the @samp{.data} input sections and before the end of
5944 the @samp{.data} output section itself.
5946 @cindex dot outside sections
5947 Setting symbols to the value of the location counter outside of an
5948 output section statement can result in unexpected values if the linker
5949 needs to place orphan sections. For example, given the following:
5955 .text: @{ *(.text) @}
5959 .data: @{ *(.data) @}
5964 If the linker needs to place some input section, e.g. @code{.rodata},
5965 not mentioned in the script, it might choose to place that section
5966 between @code{.text} and @code{.data}. You might think the linker
5967 should place @code{.rodata} on the blank line in the above script, but
5968 blank lines are of no particular significance to the linker. As well,
5969 the linker doesn't associate the above symbol names with their
5970 sections. Instead, it assumes that all assignments or other
5971 statements belong to the previous output section, except for the
5972 special case of an assignment to @code{.}. I.e., the linker will
5973 place the orphan @code{.rodata} section as if the script was written
5980 .text: @{ *(.text) @}
5984 .rodata: @{ *(.rodata) @}
5985 .data: @{ *(.data) @}
5990 This may or may not be the script author's intention for the value of
5991 @code{start_of_data}. One way to influence the orphan section
5992 placement is to assign the location counter to itself, as the linker
5993 assumes that an assignment to @code{.} is setting the start address of
5994 a following output section and thus should be grouped with that
5995 section. So you could write:
6001 .text: @{ *(.text) @}
6006 .data: @{ *(.data) @}
6011 Now, the orphan @code{.rodata} section will be placed between
6012 @code{end_of_text} and @code{start_of_data}.
6016 @subsection Operators
6017 @cindex operators for arithmetic
6018 @cindex arithmetic operators
6019 @cindex precedence in expressions
6020 The linker recognizes the standard C set of arithmetic operators, with
6021 the standard bindings and precedence levels:
6024 @c END TEXI2ROFF-KILL
6026 precedence associativity Operators Notes
6032 5 left == != > < <= >=
6038 11 right &= += -= *= /= (2)
6042 (1) Prefix operators
6043 (2) @xref{Assignments}.
6047 \vskip \baselineskip
6048 %"lispnarrowing" is the extra indent used generally for smallexample
6049 \hskip\lispnarrowing\vbox{\offinterlineskip
6052 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
6053 height2pt&\omit&&\omit&&\omit&\cr
6054 &Precedence&& Associativity &&{\rm Operators}&\cr
6055 height2pt&\omit&&\omit&&\omit&\cr
6057 height2pt&\omit&&\omit&&\omit&\cr
6059 % '176 is tilde, '~' in tt font
6060 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
6061 &2&&left&&* / \%&\cr
6064 &5&&left&&== != > < <= >=&\cr
6067 &8&&left&&{\&\&}&\cr
6070 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
6072 height2pt&\omit&&\omit&&\omit&\cr}
6077 @obeylines@parskip=0pt@parindent=0pt
6078 @dag@quad Prefix operators.
6079 @ddag@quad @xref{Assignments}.
6082 @c END TEXI2ROFF-KILL
6085 @subsection Evaluation
6086 @cindex lazy evaluation
6087 @cindex expression evaluation order
6088 The linker evaluates expressions lazily. It only computes the value of
6089 an expression when absolutely necessary.
6091 The linker needs some information, such as the value of the start
6092 address of the first section, and the origins and lengths of memory
6093 regions, in order to do any linking at all. These values are computed
6094 as soon as possible when the linker reads in the linker script.
6096 However, other values (such as symbol values) are not known or needed
6097 until after storage allocation. Such values are evaluated later, when
6098 other information (such as the sizes of output sections) is available
6099 for use in the symbol assignment expression.
6101 The sizes of sections cannot be known until after allocation, so
6102 assignments dependent upon these are not performed until after
6105 Some expressions, such as those depending upon the location counter
6106 @samp{.}, must be evaluated during section allocation.
6108 If the result of an expression is required, but the value is not
6109 available, then an error results. For example, a script like the
6115 .text 9+this_isnt_constant :
6121 will cause the error message @samp{non constant expression for initial
6124 @node Expression Section
6125 @subsection The Section of an Expression
6126 @cindex expression sections
6127 @cindex absolute expressions
6128 @cindex relative expressions
6129 @cindex absolute and relocatable symbols
6130 @cindex relocatable and absolute symbols
6131 @cindex symbols, relocatable and absolute
6132 Addresses and symbols may be section relative, or absolute. A section
6133 relative symbol is relocatable. If you request relocatable output
6134 using the @samp{-r} option, a further link operation may change the
6135 value of a section relative symbol. On the other hand, an absolute
6136 symbol will retain the same value throughout any further link
6139 Some terms in linker expressions are addresses. This is true of
6140 section relative symbols and for builtin functions that return an
6141 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
6142 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
6143 functions that return a non-address value, such as @code{LENGTH}.
6144 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
6145 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
6146 differently depending on their location, for compatibility with older
6147 versions of @code{ld}. Expressions appearing outside an output
6148 section definition treat all numbers as absolute addresses.
6149 Expressions appearing inside an output section definition treat
6150 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
6151 given, then absolute symbols and numbers are simply treated as numbers
6154 In the following simple example,
6161 __executable_start = 0x100;
6165 __data_start = 0x10;
6173 both @code{.} and @code{__executable_start} are set to the absolute
6174 address 0x100 in the first two assignments, then both @code{.} and
6175 @code{__data_start} are set to 0x10 relative to the @code{.data}
6176 section in the second two assignments.
6178 For expressions involving numbers, relative addresses and absolute
6179 addresses, ld follows these rules to evaluate terms:
6183 Unary operations on an absolute address or number, and binary
6184 operations on two absolute addresses or two numbers, or between one
6185 absolute address and a number, apply the operator to the value(s).
6187 Unary operations on a relative address, and binary operations on two
6188 relative addresses in the same section or between one relative address
6189 and a number, apply the operator to the offset part of the address(es).
6191 Other binary operations, that is, between two relative addresses not
6192 in the same section, or between a relative address and an absolute
6193 address, first convert any non-absolute term to an absolute address
6194 before applying the operator.
6197 The result section of each sub-expression is as follows:
6201 An operation involving only numbers results in a number.
6203 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6205 The result of other binary arithmetic and logical operations on two
6206 relative addresses in the same section or two absolute addresses
6207 (after above conversions) is also a number when
6208 @code{LD_FEATURE ("SANE_EXPR")} or inside an output section definition
6209 but an absolute address otherwise.
6211 The result of other operations on relative addresses or one
6212 relative address and a number, is a relative address in the same
6213 section as the relative operand(s).
6215 The result of other operations on absolute addresses (after above
6216 conversions) is an absolute address.
6219 You can use the builtin function @code{ABSOLUTE} to force an expression
6220 to be absolute when it would otherwise be relative. For example, to
6221 create an absolute symbol set to the address of the end of the output
6222 section @samp{.data}:
6226 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6230 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6231 @samp{.data} section.
6233 Using @code{LOADADDR} also forces an expression absolute, since this
6234 particular builtin function returns an absolute address.
6236 @node Builtin Functions
6237 @subsection Builtin Functions
6238 @cindex functions in expressions
6239 The linker script language includes a number of builtin functions for
6240 use in linker script expressions.
6243 @item ABSOLUTE(@var{exp})
6244 @kindex ABSOLUTE(@var{exp})
6245 @cindex expression, absolute
6246 Return the absolute (non-relocatable, as opposed to non-negative) value
6247 of the expression @var{exp}. Primarily useful to assign an absolute
6248 value to a symbol within a section definition, where symbol values are
6249 normally section relative. @xref{Expression Section}.
6251 @item ADDR(@var{section})
6252 @kindex ADDR(@var{section})
6253 @cindex section address in expression
6254 Return the address (VMA) of the named @var{section}. Your
6255 script must previously have defined the location of that section. In
6256 the following example, @code{start_of_output_1}, @code{symbol_1} and
6257 @code{symbol_2} are assigned equivalent values, except that
6258 @code{symbol_1} will be relative to the @code{.output1} section while
6259 the other two will be absolute:
6265 start_of_output_1 = ABSOLUTE(.);
6270 symbol_1 = ADDR(.output1);
6271 symbol_2 = start_of_output_1;
6277 @item ALIGN(@var{align})
6278 @itemx ALIGN(@var{exp},@var{align})
6279 @kindex ALIGN(@var{align})
6280 @kindex ALIGN(@var{exp},@var{align})
6281 @cindex round up location counter
6282 @cindex align location counter
6283 @cindex round up expression
6284 @cindex align expression
6285 Return the location counter (@code{.}) or arbitrary expression aligned
6286 to the next @var{align} boundary. The single operand @code{ALIGN}
6287 doesn't change the value of the location counter---it just does
6288 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6289 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6290 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6292 Here is an example which aligns the output @code{.data} section to the
6293 next @code{0x2000} byte boundary after the preceding section and sets a
6294 variable within the section to the next @code{0x8000} boundary after the
6299 .data ALIGN(0x2000): @{
6301 variable = ALIGN(0x8000);
6307 The first use of @code{ALIGN} in this example specifies the location of
6308 a section because it is used as the optional @var{address} attribute of
6309 a section definition (@pxref{Output Section Address}). The second use
6310 of @code{ALIGN} is used to defines the value of a symbol.
6312 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6314 @item ALIGNOF(@var{section})
6315 @kindex ALIGNOF(@var{section})
6316 @cindex section alignment
6317 Return the alignment in bytes of the named @var{section}, if that section has
6318 been allocated. If the section has not been allocated when this is
6319 evaluated, the linker will report an error. In the following example,
6320 the alignment of the @code{.output} section is stored as the first
6321 value in that section.
6326 LONG (ALIGNOF (.output))
6333 @item BLOCK(@var{exp})
6334 @kindex BLOCK(@var{exp})
6335 This is a synonym for @code{ALIGN}, for compatibility with older linker
6336 scripts. It is most often seen when setting the address of an output
6339 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6340 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6341 This is equivalent to either
6343 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6347 (ALIGN(@var{maxpagesize})
6348 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6351 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6352 for the data segment (area between the result of this expression and
6353 @code{DATA_SEGMENT_END}) than the former or not.
6354 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6355 memory will be saved at the expense of up to @var{commonpagesize} wasted
6356 bytes in the on-disk file.
6358 This expression can only be used directly in @code{SECTIONS} commands, not in
6359 any output section descriptions and only once in the linker script.
6360 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6361 be the system page size the object wants to be optimized for while still
6362 running on system page sizes up to @var{maxpagesize}. Note however
6363 that @samp{-z relro} protection will not be effective if the system
6364 page size is larger than @var{commonpagesize}.
6369 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6372 @item DATA_SEGMENT_END(@var{exp})
6373 @kindex DATA_SEGMENT_END(@var{exp})
6374 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6375 evaluation purposes.
6378 . = DATA_SEGMENT_END(.);
6381 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6382 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6383 This defines the end of the @code{PT_GNU_RELRO} segment when
6384 @samp{-z relro} option is used.
6385 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6386 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6387 @var{exp} + @var{offset} is aligned to the @var{commonpagesize}
6388 argument given to @code{DATA_SEGMENT_ALIGN}. If present in the linker
6389 script, it must be placed between @code{DATA_SEGMENT_ALIGN} and
6390 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6391 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6395 . = DATA_SEGMENT_RELRO_END(24, .);
6398 @item DEFINED(@var{symbol})
6399 @kindex DEFINED(@var{symbol})
6400 @cindex symbol defaults
6401 Return 1 if @var{symbol} is in the linker global symbol table and is
6402 defined before the statement using DEFINED in the script, otherwise
6403 return 0. You can use this function to provide
6404 default values for symbols. For example, the following script fragment
6405 shows how to set a global symbol @samp{begin} to the first location in
6406 the @samp{.text} section---but if a symbol called @samp{begin} already
6407 existed, its value is preserved:
6413 begin = DEFINED(begin) ? begin : . ;
6421 @item LENGTH(@var{memory})
6422 @kindex LENGTH(@var{memory})
6423 Return the length of the memory region named @var{memory}.
6425 @item LOADADDR(@var{section})
6426 @kindex LOADADDR(@var{section})
6427 @cindex section load address in expression
6428 Return the absolute LMA of the named @var{section}. (@pxref{Output
6431 @item LOG2CEIL(@var{exp})
6432 @kindex LOG2CEIL(@var{exp})
6433 Return the binary logarithm of @var{exp} rounded towards infinity.
6434 @code{LOG2CEIL(0)} returns 0.
6437 @item MAX(@var{exp1}, @var{exp2})
6438 Returns the maximum of @var{exp1} and @var{exp2}.
6441 @item MIN(@var{exp1}, @var{exp2})
6442 Returns the minimum of @var{exp1} and @var{exp2}.
6444 @item NEXT(@var{exp})
6445 @kindex NEXT(@var{exp})
6446 @cindex unallocated address, next
6447 Return the next unallocated address that is a multiple of @var{exp}.
6448 This function is closely related to @code{ALIGN(@var{exp})}; unless you
6449 use the @code{MEMORY} command to define discontinuous memory for the
6450 output file, the two functions are equivalent.
6452 @item ORIGIN(@var{memory})
6453 @kindex ORIGIN(@var{memory})
6454 Return the origin of the memory region named @var{memory}.
6456 @item SEGMENT_START(@var{segment}, @var{default})
6457 @kindex SEGMENT_START(@var{segment}, @var{default})
6458 Return the base address of the named @var{segment}. If an explicit
6459 value has already been given for this segment (with a command-line
6460 @samp{-T} option) then that value will be returned otherwise the value
6461 will be @var{default}. At present, the @samp{-T} command-line option
6462 can only be used to set the base address for the ``text'', ``data'', and
6463 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
6466 @item SIZEOF(@var{section})
6467 @kindex SIZEOF(@var{section})
6468 @cindex section size
6469 Return the size in bytes of the named @var{section}, if that section has
6470 been allocated. If the section has not been allocated when this is
6471 evaluated, the linker will report an error. In the following example,
6472 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6481 symbol_1 = .end - .start ;
6482 symbol_2 = SIZEOF(.output);
6487 @item SIZEOF_HEADERS
6488 @itemx sizeof_headers
6489 @kindex SIZEOF_HEADERS
6491 Return the size in bytes of the output file's headers. This is
6492 information which appears at the start of the output file. You can use
6493 this number when setting the start address of the first section, if you
6494 choose, to facilitate paging.
6496 @cindex not enough room for program headers
6497 @cindex program headers, not enough room
6498 When producing an ELF output file, if the linker script uses the
6499 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6500 number of program headers before it has determined all the section
6501 addresses and sizes. If the linker later discovers that it needs
6502 additional program headers, it will report an error @samp{not enough
6503 room for program headers}. To avoid this error, you must avoid using
6504 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6505 script to avoid forcing the linker to use additional program headers, or
6506 you must define the program headers yourself using the @code{PHDRS}
6507 command (@pxref{PHDRS}).
6510 @node Implicit Linker Scripts
6511 @section Implicit Linker Scripts
6512 @cindex implicit linker scripts
6513 If you specify a linker input file which the linker can not recognize as
6514 an object file or an archive file, it will try to read the file as a
6515 linker script. If the file can not be parsed as a linker script, the
6516 linker will report an error.
6518 An implicit linker script will not replace the default linker script.
6520 Typically an implicit linker script would contain only symbol
6521 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6524 Any input files read because of an implicit linker script will be read
6525 at the position in the command line where the implicit linker script was
6526 read. This can affect archive searching.
6529 @node Machine Dependent
6530 @chapter Machine Dependent Features
6532 @cindex machine dependencies
6533 @command{ld} has additional features on some platforms; the following
6534 sections describe them. Machines where @command{ld} has no additional
6535 functionality are not listed.
6539 * H8/300:: @command{ld} and the H8/300
6542 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6545 * ARM:: @command{ld} and the ARM family
6548 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6551 * M68K:: @command{ld} and the Motorola 68K family
6554 * MIPS:: @command{ld} and the MIPS family
6557 * MMIX:: @command{ld} and MMIX
6560 * MSP430:: @command{ld} and MSP430
6563 * NDS32:: @command{ld} and NDS32
6566 * Nios II:: @command{ld} and the Altera Nios II
6569 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6572 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6575 * S/390 ELF:: @command{ld} and S/390 ELF Support
6578 * SPU ELF:: @command{ld} and SPU ELF Support
6581 * TI COFF:: @command{ld} and TI COFF
6584 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6587 * Xtensa:: @command{ld} and Xtensa Processors
6598 @section @command{ld} and the H8/300
6600 @cindex H8/300 support
6601 For the H8/300, @command{ld} can perform these global optimizations when
6602 you specify the @samp{--relax} command-line option.
6605 @cindex relaxing on H8/300
6606 @item relaxing address modes
6607 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6608 targets are within eight bits, and turns them into eight-bit
6609 program-counter relative @code{bsr} and @code{bra} instructions,
6612 @cindex synthesizing on H8/300
6613 @item synthesizing instructions
6614 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6615 @command{ld} finds all @code{mov.b} instructions which use the
6616 sixteen-bit absolute address form, but refer to the top
6617 page of memory, and changes them to use the eight-bit address form.
6618 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6619 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6620 top page of memory).
6622 @command{ld} finds all @code{mov} instructions which use the register
6623 indirect with 32-bit displacement addressing mode, but use a small
6624 displacement inside 16-bit displacement range, and changes them to use
6625 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6626 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6627 whenever the displacement @var{d} is in the 16 bit signed integer
6628 range. Only implemented in ELF-format ld).
6630 @item bit manipulation instructions
6631 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6632 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6633 which use 32 bit and 16 bit absolute address form, but refer to the top
6634 page of memory, and changes them to use the 8 bit address form.
6635 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6636 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6637 the top page of memory).
6639 @item system control instructions
6640 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6641 32 bit absolute address form, but refer to the top page of memory, and
6642 changes them to use 16 bit address form.
6643 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6644 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6645 the top page of memory).
6655 @c This stuff is pointless to say unless you're especially concerned
6656 @c with Renesas chips; don't enable it for generic case, please.
6658 @chapter @command{ld} and Other Renesas Chips
6660 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6661 H8/500, and SH chips. No special features, commands, or command-line
6662 options are required for these chips.
6676 @node M68HC11/68HC12
6677 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6679 @cindex M68HC11 and 68HC12 support
6681 @subsection Linker Relaxation
6683 For the Motorola 68HC11, @command{ld} can perform these global
6684 optimizations when you specify the @samp{--relax} command-line option.
6687 @cindex relaxing on M68HC11
6688 @item relaxing address modes
6689 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6690 targets are within eight bits, and turns them into eight-bit
6691 program-counter relative @code{bsr} and @code{bra} instructions,
6694 @command{ld} also looks at all 16-bit extended addressing modes and
6695 transforms them in a direct addressing mode when the address is in
6696 page 0 (between 0 and 0x0ff).
6698 @item relaxing gcc instruction group
6699 When @command{gcc} is called with @option{-mrelax}, it can emit group
6700 of instructions that the linker can optimize to use a 68HC11 direct
6701 addressing mode. These instructions consists of @code{bclr} or
6702 @code{bset} instructions.
6706 @subsection Trampoline Generation
6708 @cindex trampoline generation on M68HC11
6709 @cindex trampoline generation on M68HC12
6710 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6711 call a far function using a normal @code{jsr} instruction. The linker
6712 will also change the relocation to some far function to use the
6713 trampoline address instead of the function address. This is typically the
6714 case when a pointer to a function is taken. The pointer will in fact
6715 point to the function trampoline.
6723 @section @command{ld} and the ARM family
6725 @cindex ARM interworking support
6726 @kindex --support-old-code
6727 For the ARM, @command{ld} will generate code stubs to allow functions calls
6728 between ARM and Thumb code. These stubs only work with code that has
6729 been compiled and assembled with the @samp{-mthumb-interwork} command
6730 line option. If it is necessary to link with old ARM object files or
6731 libraries, which have not been compiled with the -mthumb-interwork
6732 option then the @samp{--support-old-code} command line switch should be
6733 given to the linker. This will make it generate larger stub functions
6734 which will work with non-interworking aware ARM code. Note, however,
6735 the linker does not support generating stubs for function calls to
6736 non-interworking aware Thumb code.
6738 @cindex thumb entry point
6739 @cindex entry point, thumb
6740 @kindex --thumb-entry=@var{entry}
6741 The @samp{--thumb-entry} switch is a duplicate of the generic
6742 @samp{--entry} switch, in that it sets the program's starting address.
6743 But it also sets the bottom bit of the address, so that it can be
6744 branched to using a BX instruction, and the program will start
6745 executing in Thumb mode straight away.
6747 @cindex PE import table prefixing
6748 @kindex --use-nul-prefixed-import-tables
6749 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6750 the import tables idata4 and idata5 have to be generated with a zero
6751 element prefix for import libraries. This is the old style to generate
6752 import tables. By default this option is turned off.
6756 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6757 executables. This option is only valid when linking big-endian
6758 objects - ie ones which have been assembled with the @option{-EB}
6759 option. The resulting image will contain big-endian data and
6763 @kindex --target1-rel
6764 @kindex --target1-abs
6765 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6766 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6767 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6768 and @samp{--target1-abs} switches override the default.
6771 @kindex --target2=@var{type}
6772 The @samp{--target2=type} switch overrides the default definition of the
6773 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6774 meanings, and target defaults are as follows:
6777 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6779 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6781 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6786 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6787 specification) enables objects compiled for the ARMv4 architecture to be
6788 interworking-safe when linked with other objects compiled for ARMv4t, but
6789 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6791 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6792 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6793 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6795 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6796 relocations are ignored.
6798 @cindex FIX_V4BX_INTERWORKING
6799 @kindex --fix-v4bx-interworking
6800 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6801 relocations with a branch to the following veneer:
6809 This allows generation of libraries/applications that work on ARMv4 cores
6810 and are still interworking safe. Note that the above veneer clobbers the
6811 condition flags, so may cause incorrect program behavior in rare cases.
6815 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6816 BLX instructions (available on ARMv5t and above) in various
6817 situations. Currently it is used to perform calls via the PLT from Thumb
6818 code using BLX rather than using BX and a mode-switching stub before
6819 each PLT entry. This should lead to such calls executing slightly faster.
6821 This option is enabled implicitly for SymbianOS, so there is no need to
6822 specify it if you are using that target.
6824 @cindex VFP11_DENORM_FIX
6825 @kindex --vfp11-denorm-fix
6826 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6827 bug in certain VFP11 coprocessor hardware, which sometimes allows
6828 instructions with denorm operands (which must be handled by support code)
6829 to have those operands overwritten by subsequent instructions before
6830 the support code can read the intended values.
6832 The bug may be avoided in scalar mode if you allow at least one
6833 intervening instruction between a VFP11 instruction which uses a register
6834 and another instruction which writes to the same register, or at least two
6835 intervening instructions if vector mode is in use. The bug only affects
6836 full-compliance floating-point mode: you do not need this workaround if
6837 you are using "runfast" mode. Please contact ARM for further details.
6839 If you know you are using buggy VFP11 hardware, you can
6840 enable this workaround by specifying the linker option
6841 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6842 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6843 vector mode (the latter also works for scalar code). The default is
6844 @samp{--vfp-denorm-fix=none}.
6846 If the workaround is enabled, instructions are scanned for
6847 potentially-troublesome sequences, and a veneer is created for each
6848 such sequence which may trigger the erratum. The veneer consists of the
6849 first instruction of the sequence and a branch back to the subsequent
6850 instruction. The original instruction is then replaced with a branch to
6851 the veneer. The extra cycles required to call and return from the veneer
6852 are sufficient to avoid the erratum in both the scalar and vector cases.
6854 @cindex ARM1176 erratum workaround
6855 @kindex --fix-arm1176
6856 @kindex --no-fix-arm1176
6857 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6858 in certain ARM1176 processors. The workaround is enabled by default if you
6859 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6860 unconditionally by specifying @samp{--no-fix-arm1176}.
6862 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6863 Programmer Advice Notice'' available on the ARM documentation website at:
6864 http://infocenter.arm.com/.
6866 @cindex STM32L4xx erratum workaround
6867 @kindex --fix-stm32l4xx-629360
6869 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
6870 workaround for a bug in the bus matrix / memory controller for some of
6871 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
6872 off-chip memory via the affected bus for bus reads of 9 words or more,
6873 the bus can generate corrupt data and/or abort. These are only
6874 core-initiated accesses (not DMA), and might affect any access:
6875 integer loads such as LDM, POP and floating-point loads such as VLDM,
6876 VPOP. Stores are not affected.
6878 The bug can be avoided by splitting memory accesses into the
6879 necessary chunks to keep bus reads below 8 words.
6881 The workaround is not enabled by default, this is equivalent to use
6882 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
6883 STM32L4xx hardware, you can enable the workaround by specifying the
6884 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
6885 @samp{--fix-stm32l4xx-629360=default}.
6887 If the workaround is enabled, instructions are scanned for
6888 potentially-troublesome sequences, and a veneer is created for each
6889 such sequence which may trigger the erratum. The veneer consists in a
6890 replacement sequence emulating the behaviour of the original one and a
6891 branch back to the subsequent instruction. The original instruction is
6892 then replaced with a branch to the veneer.
6894 The workaround does not always preserve the memory access order for
6895 the LDMDB instruction, when the instruction loads the PC.
6897 The workaround is not able to handle problematic instructions when
6898 they are in the middle of an IT block, since a branch is not allowed
6899 there. In that case, the linker reports a warning and no replacement
6902 The workaround is not able to replace problematic instructions with a
6903 PC-relative branch instruction if the @samp{.text} section is too
6904 large. In that case, when the branch that replaces the original code
6905 cannot be encoded, the linker reports a warning and no replacement
6908 @cindex NO_ENUM_SIZE_WARNING
6909 @kindex --no-enum-size-warning
6910 The @option{--no-enum-size-warning} switch prevents the linker from
6911 warning when linking object files that specify incompatible EABI
6912 enumeration size attributes. For example, with this switch enabled,
6913 linking of an object file using 32-bit enumeration values with another
6914 using enumeration values fitted into the smallest possible space will
6917 @cindex NO_WCHAR_SIZE_WARNING
6918 @kindex --no-wchar-size-warning
6919 The @option{--no-wchar-size-warning} switch prevents the linker from
6920 warning when linking object files that specify incompatible EABI
6921 @code{wchar_t} size attributes. For example, with this switch enabled,
6922 linking of an object file using 32-bit @code{wchar_t} values with another
6923 using 16-bit @code{wchar_t} values will not be diagnosed.
6926 @kindex --pic-veneer
6927 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6928 ARM/Thumb interworking veneers, even if the rest of the binary
6929 is not PIC. This avoids problems on uClinux targets where
6930 @samp{--emit-relocs} is used to generate relocatable binaries.
6932 @cindex STUB_GROUP_SIZE
6933 @kindex --stub-group-size=@var{N}
6934 The linker will automatically generate and insert small sequences of
6935 code into a linked ARM ELF executable whenever an attempt is made to
6936 perform a function call to a symbol that is too far away. The
6937 placement of these sequences of instructions - called stubs - is
6938 controlled by the command line option @option{--stub-group-size=N}.
6939 The placement is important because a poor choice can create a need for
6940 duplicate stubs, increasing the code size. The linker will try to
6941 group stubs together in order to reduce interruptions to the flow of
6942 code, but it needs guidance as to how big these groups should be and
6943 where they should be placed.
6945 The value of @samp{N}, the parameter to the
6946 @option{--stub-group-size=} option controls where the stub groups are
6947 placed. If it is negative then all stubs are placed after the first
6948 branch that needs them. If it is positive then the stubs can be
6949 placed either before or after the branches that need them. If the
6950 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6951 exactly where to place groups of stubs, using its built in heuristics.
6952 A value of @samp{N} greater than 1 (or smaller than -1) tells the
6953 linker that a single group of stubs can service at most @samp{N} bytes
6954 from the input sections.
6956 The default, if @option{--stub-group-size=} is not specified, is
6959 Farcalls stubs insertion is fully supported for the ARM-EABI target
6960 only, because it relies on object files properties not present
6963 @cindex Cortex-A8 erratum workaround
6964 @kindex --fix-cortex-a8
6965 @kindex --no-fix-cortex-a8
6966 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}.
6968 The erratum only affects Thumb-2 code. Please contact ARM for further details.
6970 @cindex Cortex-A53 erratum 835769 workaround
6971 @kindex --fix-cortex-a53-835769
6972 @kindex --no-fix-cortex-a53-835769
6973 The @samp{--fix-cortex-a53-835769} switch enables a link-time workaround for erratum 835769 present on certain early revisions of Cortex-A53 processors. The workaround is disabled by default. It can be enabled by specifying @samp{--fix-cortex-a53-835769}, or disabled unconditionally by specifying @samp{--no-fix-cortex-a53-835769}.
6975 Please contact ARM for further details.
6977 @kindex --merge-exidx-entries
6978 @kindex --no-merge-exidx-entries
6979 @cindex Merging exidx entries
6980 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
6983 @cindex 32-bit PLT entries
6984 The @samp{--long-plt} option enables the use of 16 byte PLT entries
6985 which support up to 4Gb of code. The default is to use 12 byte PLT
6986 entries which only support 512Mb of code.
6988 @kindex --no-apply-dynamic-relocs
6989 @cindex AArch64 rela addend
6990 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
6991 link-time values for dynamic relocations.
6993 @cindex Placement of SG veneers
6994 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
6995 Its start address must be set, either with the command line option
6996 @samp{--section-start} or in a linker script, to indicate where to place these
6999 @kindex --cmse-implib
7000 @cindex Secure gateway import library
7001 The @samp{--cmse-implib} option requests that the import libraries
7002 specified by the @samp{--out-implib} and @samp{--in-implib} options are
7003 secure gateway import libraries, suitable for linking a non-secure
7004 executable against secure code as per ARMv8-M Security Extensions.
7006 @kindex --in-implib=@var{file}
7007 @cindex Input import library
7008 The @samp{--in-implib=file} specifies an input import library whose symbols
7009 must keep the same address in the executable being produced. A warning is
7010 given if no @samp{--out-implib} is given but new symbols have been introduced
7011 in the executable that should be listed in its import library. Otherwise, if
7012 @samp{--out-implib} is specified, the symbols are added to the output import
7013 library. A warning is also given if some symbols present in the input import
7014 library have disappeared from the executable. This option is only effective
7015 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
7029 @section @command{ld} and HPPA 32-bit ELF Support
7030 @cindex HPPA multiple sub-space stubs
7031 @kindex --multi-subspace
7032 When generating a shared library, @command{ld} will by default generate
7033 import stubs suitable for use with a single sub-space application.
7034 The @samp{--multi-subspace} switch causes @command{ld} to generate export
7035 stubs, and different (larger) import stubs suitable for use with
7036 multiple sub-spaces.
7038 @cindex HPPA stub grouping
7039 @kindex --stub-group-size=@var{N}
7040 Long branch stubs and import/export stubs are placed by @command{ld} in
7041 stub sections located between groups of input sections.
7042 @samp{--stub-group-size} specifies the maximum size of a group of input
7043 sections handled by one stub section. Since branch offsets are signed,
7044 a stub section may serve two groups of input sections, one group before
7045 the stub section, and one group after it. However, when using
7046 conditional branches that require stubs, it may be better (for branch
7047 prediction) that stub sections only serve one group of input sections.
7048 A negative value for @samp{N} chooses this scheme, ensuring that
7049 branches to stubs always use a negative offset. Two special values of
7050 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7051 @command{ld} to automatically size input section groups for the branch types
7052 detected, with the same behaviour regarding stub placement as other
7053 positive or negative values of @samp{N} respectively.
7055 Note that @samp{--stub-group-size} does not split input sections. A
7056 single input section larger than the group size specified will of course
7057 create a larger group (of one section). If input sections are too
7058 large, it may not be possible for a branch to reach its stub.
7071 @section @command{ld} and the Motorola 68K family
7073 @cindex Motorola 68K GOT generation
7074 @kindex --got=@var{type}
7075 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
7076 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
7077 @samp{target}. When @samp{target} is selected the linker chooses
7078 the default GOT generation scheme for the current target.
7079 @samp{single} tells the linker to generate a single GOT with
7080 entries only at non-negative offsets.
7081 @samp{negative} instructs the linker to generate a single GOT with
7082 entries at both negative and positive offsets. Not all environments
7084 @samp{multigot} allows the linker to generate several GOTs in the
7085 output file. All GOT references from a single input object
7086 file access the same GOT, but references from different input object
7087 files might access different GOTs. Not all environments support such GOTs.
7100 @section @command{ld} and the MIPS family
7102 @cindex MIPS microMIPS instruction choice selection
7105 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
7106 microMIPS instructions used in code generated by the linker, such as that
7107 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
7108 used, then the linker only uses 32-bit instruction encodings. By default
7109 or if @samp{--no-insn32} is used, all instruction encodings are used,
7110 including 16-bit ones where possible.
7112 @cindex MIPS branch relocation check control
7113 @kindex --ignore-branch-isa
7114 @kindex --no-ignore-branch-isa
7115 The @samp{--ignore-branch-isa} and @samp{--no-ignore-branch-isa} options
7116 control branch relocation checks for invalid ISA mode transitions. If
7117 @samp{--ignore-branch-isa} is used, then the linker accepts any branch
7118 relocations and any ISA mode transition required is lost in relocation
7119 calculation, except for some cases of @code{BAL} instructions which meet
7120 relaxation conditions and are converted to equivalent @code{JALX}
7121 instructions as the associated relocation is calculated. By default
7122 or if @samp{--no-ignore-branch-isa} is used a check is made causing
7123 the loss of an ISA mode transition to produce an error.
7136 @section @code{ld} and MMIX
7137 For MMIX, there is a choice of generating @code{ELF} object files or
7138 @code{mmo} object files when linking. The simulator @code{mmix}
7139 understands the @code{mmo} format. The binutils @code{objcopy} utility
7140 can translate between the two formats.
7142 There is one special section, the @samp{.MMIX.reg_contents} section.
7143 Contents in this section is assumed to correspond to that of global
7144 registers, and symbols referring to it are translated to special symbols,
7145 equal to registers. In a final link, the start address of the
7146 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7147 global register multiplied by 8. Register @code{$255} is not included in
7148 this section; it is always set to the program entry, which is at the
7149 symbol @code{Main} for @code{mmo} files.
7151 Global symbols with the prefix @code{__.MMIX.start.}, for example
7152 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7153 The default linker script uses these to set the default start address
7156 Initial and trailing multiples of zero-valued 32-bit words in a section,
7157 are left out from an mmo file.
7170 @section @code{ld} and MSP430
7171 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7172 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7173 just pass @samp{-m help} option to the linker).
7175 @cindex MSP430 extra sections
7176 The linker will recognize some extra sections which are MSP430 specific:
7179 @item @samp{.vectors}
7180 Defines a portion of ROM where interrupt vectors located.
7182 @item @samp{.bootloader}
7183 Defines the bootloader portion of the ROM (if applicable). Any code
7184 in this section will be uploaded to the MPU.
7186 @item @samp{.infomem}
7187 Defines an information memory section (if applicable). Any code in
7188 this section will be uploaded to the MPU.
7190 @item @samp{.infomemnobits}
7191 This is the same as the @samp{.infomem} section except that any code
7192 in this section will not be uploaded to the MPU.
7194 @item @samp{.noinit}
7195 Denotes a portion of RAM located above @samp{.bss} section.
7197 The last two sections are used by gcc.
7201 @cindex MSP430 Options
7202 @kindex --code-region
7203 @item --code-region=[either,lower,upper,none]
7204 This will transform .text* sections to [either,lower,upper].text* sections. The
7205 argument passed to GCC for -mcode-region is propagated to the linker
7208 @kindex --data-region
7209 @item --data-region=[either,lower,upper,none]
7210 This will transform .data*, .bss* and .rodata* sections to
7211 [either,lower,upper].[data,bss,rodata]* sections. The argument passed to GCC
7212 for -mdata-region is propagated to the linker using this option.
7214 @kindex --disable-sec-transformation
7215 @item --disable-sec-transformation
7216 Prevent the transformation of sections as specified by the @code{--code-region}
7217 and @code{--data-region} options.
7218 This is useful if you are compiling and linking using a single call to the GCC
7219 wrapper, and want to compile the source files using -m[code,data]-region but
7220 not transform the sections for prebuilt libraries and objects.
7234 @section @code{ld} and NDS32
7235 @kindex relaxing on NDS32
7236 For NDS32, there are some options to select relaxation behavior. The linker
7237 relaxes objects according to these options.
7240 @item @samp{--m[no-]fp-as-gp}
7241 Disable/enable fp-as-gp relaxation.
7243 @item @samp{--mexport-symbols=FILE}
7244 Exporting symbols and their address into FILE as linker script.
7246 @item @samp{--m[no-]ex9}
7247 Disable/enable link-time EX9 relaxation.
7249 @item @samp{--mexport-ex9=FILE}
7250 Export the EX9 table after linking.
7252 @item @samp{--mimport-ex9=FILE}
7253 Import the Ex9 table for EX9 relaxation.
7255 @item @samp{--mupdate-ex9}
7256 Update the existing EX9 table.
7258 @item @samp{--mex9-limit=NUM}
7259 Maximum number of entries in the ex9 table.
7261 @item @samp{--mex9-loop-aware}
7262 Avoid generating the EX9 instruction inside the loop.
7264 @item @samp{--m[no-]ifc}
7265 Disable/enable the link-time IFC optimization.
7267 @item @samp{--mifc-loop-aware}
7268 Avoid generating the IFC instruction inside the loop.
7282 @section @command{ld} and the Altera Nios II
7283 @cindex Nios II call relaxation
7284 @kindex --relax on Nios II
7286 Call and immediate jump instructions on Nios II processors are limited to
7287 transferring control to addresses in the same 256MB memory segment,
7288 which may result in @command{ld} giving
7289 @samp{relocation truncated to fit} errors with very large programs.
7290 The command-line option @option{--relax} enables the generation of
7291 trampolines that can access the entire 32-bit address space for calls
7292 outside the normal @code{call} and @code{jmpi} address range. These
7293 trampolines are inserted at section boundaries, so may not themselves
7294 be reachable if an input section and its associated call trampolines are
7297 The @option{--relax} option is enabled by default unless @option{-r}
7298 is also specified. You can disable trampoline generation by using the
7299 @option{--no-relax} linker option. You can also disable this optimization
7300 locally by using the @samp{set .noat} directive in assembly-language
7301 source files, as the linker-inserted trampolines use the @code{at}
7302 register as a temporary.
7304 Note that the linker @option{--relax} option is independent of assembler
7305 relaxation options, and that using the GNU assembler's @option{-relax-all}
7306 option interferes with the linker's more selective call instruction relaxation.
7319 @section @command{ld} and PowerPC 32-bit ELF Support
7320 @cindex PowerPC long branches
7321 @kindex --relax on PowerPC
7322 Branches on PowerPC processors are limited to a signed 26-bit
7323 displacement, which may result in @command{ld} giving
7324 @samp{relocation truncated to fit} errors with very large programs.
7325 @samp{--relax} enables the generation of trampolines that can access
7326 the entire 32-bit address space. These trampolines are inserted at
7327 section boundaries, so may not themselves be reachable if an input
7328 section exceeds 33M in size. You may combine @samp{-r} and
7329 @samp{--relax} to add trampolines in a partial link. In that case
7330 both branches to undefined symbols and inter-section branches are also
7331 considered potentially out of range, and trampolines inserted.
7333 @cindex PowerPC ELF32 options
7338 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7339 generates code capable of using a newer PLT and GOT layout that has
7340 the security advantage of no executable section ever needing to be
7341 writable and no writable section ever being executable. PowerPC
7342 @command{ld} will generate this layout, including stubs to access the
7343 PLT, if all input files (including startup and static libraries) were
7344 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7345 BSS PLT (and GOT layout) which can give slightly better performance.
7347 @kindex --secure-plt
7349 @command{ld} will use the new PLT and GOT layout if it is linking new
7350 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7351 when linking non-PIC code. This option requests the new PLT and GOT
7352 layout. A warning will be given if some object file requires the old
7358 The new secure PLT and GOT are placed differently relative to other
7359 sections compared to older BSS PLT and GOT placement. The location of
7360 @code{.plt} must change because the new secure PLT is an initialized
7361 section while the old PLT is uninitialized. The reason for the
7362 @code{.got} change is more subtle: The new placement allows
7363 @code{.got} to be read-only in applications linked with
7364 @samp{-z relro -z now}. However, this placement means that
7365 @code{.sdata} cannot always be used in shared libraries, because the
7366 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7367 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7368 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7369 really only useful for other compilers that may do so.
7371 @cindex PowerPC stub symbols
7372 @kindex --emit-stub-syms
7373 @item --emit-stub-syms
7374 This option causes @command{ld} to label linker stubs with a local
7375 symbol that encodes the stub type and destination.
7377 @cindex PowerPC TLS optimization
7378 @kindex --no-tls-optimize
7379 @item --no-tls-optimize
7380 PowerPC @command{ld} normally performs some optimization of code
7381 sequences used to access Thread-Local Storage. Use this option to
7382 disable the optimization.
7395 @node PowerPC64 ELF64
7396 @section @command{ld} and PowerPC64 64-bit ELF Support
7398 @cindex PowerPC64 ELF64 options
7400 @cindex PowerPC64 stub grouping
7401 @kindex --stub-group-size
7402 @item --stub-group-size
7403 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
7404 by @command{ld} in stub sections located between groups of input sections.
7405 @samp{--stub-group-size} specifies the maximum size of a group of input
7406 sections handled by one stub section. Since branch offsets are signed,
7407 a stub section may serve two groups of input sections, one group before
7408 the stub section, and one group after it. However, when using
7409 conditional branches that require stubs, it may be better (for branch
7410 prediction) that stub sections only serve one group of input sections.
7411 A negative value for @samp{N} chooses this scheme, ensuring that
7412 branches to stubs always use a negative offset. Two special values of
7413 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7414 @command{ld} to automatically size input section groups for the branch types
7415 detected, with the same behaviour regarding stub placement as other
7416 positive or negative values of @samp{N} respectively.
7418 Note that @samp{--stub-group-size} does not split input sections. A
7419 single input section larger than the group size specified will of course
7420 create a larger group (of one section). If input sections are too
7421 large, it may not be possible for a branch to reach its stub.
7423 @cindex PowerPC64 stub symbols
7424 @kindex --emit-stub-syms
7425 @item --emit-stub-syms
7426 This option causes @command{ld} to label linker stubs with a local
7427 symbol that encodes the stub type and destination.
7429 @cindex PowerPC64 dot symbols
7431 @kindex --no-dotsyms
7434 These two options control how @command{ld} interprets version patterns
7435 in a version script. Older PowerPC64 compilers emitted both a
7436 function descriptor symbol with the same name as the function, and a
7437 code entry symbol with the name prefixed by a dot (@samp{.}). To
7438 properly version a function @samp{foo}, the version script thus needs
7439 to control both @samp{foo} and @samp{.foo}. The option
7440 @samp{--dotsyms}, on by default, automatically adds the required
7441 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
7444 @cindex PowerPC64 register save/restore functions
7445 @kindex --save-restore-funcs
7446 @kindex --no-save-restore-funcs
7447 @item --save-restore-funcs
7448 @itemx --no-save-restore-funcs
7449 These two options control whether PowerPC64 @command{ld} automatically
7450 provides out-of-line register save and restore functions used by
7451 @samp{-Os} code. The default is to provide any such referenced
7452 function for a normal final link, and to not do so for a relocatable
7455 @cindex PowerPC64 TLS optimization
7456 @kindex --no-tls-optimize
7457 @item --no-tls-optimize
7458 PowerPC64 @command{ld} normally performs some optimization of code
7459 sequences used to access Thread-Local Storage. Use this option to
7460 disable the optimization.
7462 @cindex PowerPC64 __tls_get_addr optimization
7463 @kindex --tls-get-addr-optimize
7464 @kindex --no-tls-get-addr-optimize
7465 @item --tls-get-addr-optimize
7466 @itemx --no-tls-get-addr-optimize
7467 These options control whether PowerPC64 @command{ld} uses a special
7468 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
7469 an optimization that allows the second and subsequent calls to
7470 @code{__tls_get_addr} for a given symbol to be resolved by the special
7471 stub without calling in to glibc. By default the linker enables this
7472 option when glibc advertises the availability of __tls_get_addr_opt.
7473 Forcing this option on when using an older glibc won't do much besides
7474 slow down your applications, but may be useful if linking an
7475 application against an older glibc with the expectation that it will
7476 normally be used on systems having a newer glibc.
7478 @cindex PowerPC64 OPD optimization
7479 @kindex --no-opd-optimize
7480 @item --no-opd-optimize
7481 PowerPC64 @command{ld} normally removes @code{.opd} section entries
7482 corresponding to deleted link-once functions, or functions removed by
7483 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
7484 Use this option to disable @code{.opd} optimization.
7486 @cindex PowerPC64 OPD spacing
7487 @kindex --non-overlapping-opd
7488 @item --non-overlapping-opd
7489 Some PowerPC64 compilers have an option to generate compressed
7490 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
7491 the static chain pointer (unused in C) with the first word of the next
7492 entry. This option expands such entries to the full 24 bytes.
7494 @cindex PowerPC64 TOC optimization
7495 @kindex --no-toc-optimize
7496 @item --no-toc-optimize
7497 PowerPC64 @command{ld} normally removes unused @code{.toc} section
7498 entries. Such entries are detected by examining relocations that
7499 reference the TOC in code sections. A reloc in a deleted code section
7500 marks a TOC word as unneeded, while a reloc in a kept code section
7501 marks a TOC word as needed. Since the TOC may reference itself, TOC
7502 relocs are also examined. TOC words marked as both needed and
7503 unneeded will of course be kept. TOC words without any referencing
7504 reloc are assumed to be part of a multi-word entry, and are kept or
7505 discarded as per the nearest marked preceding word. This works
7506 reliably for compiler generated code, but may be incorrect if assembly
7507 code is used to insert TOC entries. Use this option to disable the
7510 @cindex PowerPC64 multi-TOC
7511 @kindex --no-multi-toc
7512 @item --no-multi-toc
7513 If given any toc option besides @code{-mcmodel=medium} or
7514 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
7516 entries are accessed with a 16-bit offset from r2. This limits the
7517 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
7518 grouping code sections such that each group uses less than 64K for its
7519 TOC entries, then inserts r2 adjusting stubs between inter-group
7520 calls. @command{ld} does not split apart input sections, so cannot
7521 help if a single input file has a @code{.toc} section that exceeds
7522 64K, most likely from linking multiple files with @command{ld -r}.
7523 Use this option to turn off this feature.
7525 @cindex PowerPC64 TOC sorting
7526 @kindex --no-toc-sort
7528 By default, @command{ld} sorts TOC sections so that those whose file
7529 happens to have a section called @code{.init} or @code{.fini} are
7530 placed first, followed by TOC sections referenced by code generated
7531 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
7532 referenced only by code generated with PowerPC64 gcc's
7533 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
7534 results in better TOC grouping for multi-TOC. Use this option to turn
7537 @cindex PowerPC64 PLT stub alignment
7539 @kindex --no-plt-align
7541 @itemx --no-plt-align
7542 Use these options to control whether individual PLT call stubs are
7543 aligned to a 32-byte boundary, or to the specified power of two
7544 boundary when using @code{--plt-align=}. A negative value may be
7545 specified to pad PLT call stubs so that they do not cross the
7546 specified power of two boundary (or the minimum number of boundaries
7547 if a PLT stub is so large that it must cross a boundary). By default
7548 PLT call stubs are aligned to 32-byte boundaries.
7550 @cindex PowerPC64 PLT call stub static chain
7551 @kindex --plt-static-chain
7552 @kindex --no-plt-static-chain
7553 @item --plt-static-chain
7554 @itemx --no-plt-static-chain
7555 Use these options to control whether PLT call stubs load the static
7556 chain pointer (r11). @code{ld} defaults to not loading the static
7557 chain since there is never any need to do so on a PLT call.
7559 @cindex PowerPC64 PLT call stub thread safety
7560 @kindex --plt-thread-safe
7561 @kindex --no-plt-thread-safe
7562 @item --plt-thread-safe
7563 @itemx --no-plt-thread-safe
7564 With power7's weakly ordered memory model, it is possible when using
7565 lazy binding for ld.so to update a plt entry in one thread and have
7566 another thread see the individual plt entry words update in the wrong
7567 order, despite ld.so carefully writing in the correct order and using
7568 memory write barriers. To avoid this we need some sort of read
7569 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
7570 looks for calls to commonly used functions that create threads, and if
7571 seen, adds the necessary barriers. Use these options to change the
7574 @cindex PowerPC64 ELFv2 PLT localentry optimization
7575 @kindex --plt-localentry
7576 @kindex --no-plt-localentry
7577 @item --plt-localentry
7578 @itemx --no-localentry
7579 ELFv2 functions with localentry:0 are those with a single entry point,
7580 ie. global entry == local entry, and that have no requirement on r2
7581 (the TOC/GOT pointer) or r12, and guarantee r2 is unchanged on return.
7582 Such an external function can be called via the PLT without saving r2
7583 or restoring it on return, avoiding a common load-hit-store for small
7584 functions. The optimization is attractive, with up to 40% reduction
7585 in execution time for a small function, but can result in symbol
7586 interposition failures. Also, minor changes in a shared library,
7587 including system libraries, can cause a function that was localentry:0
7588 to become localentry:8. This will result in a dynamic loader
7589 complaint and failure to run. The option is experimental, use with
7590 care. @option{--no-plt-localentry} is the default.
7604 @section @command{ld} and S/390 ELF Support
7606 @cindex S/390 ELF options
7610 @kindex --s390-pgste
7612 This option marks the result file with a @code{PT_S390_PGSTE}
7613 segment. The Linux kernel is supposed to allocate 4k page tables for
7614 binaries marked that way.
7628 @section @command{ld} and SPU ELF Support
7630 @cindex SPU ELF options
7636 This option marks an executable as a PIC plugin module.
7638 @cindex SPU overlays
7639 @kindex --no-overlays
7641 Normally, @command{ld} recognizes calls to functions within overlay
7642 regions, and redirects such calls to an overlay manager via a stub.
7643 @command{ld} also provides a built-in overlay manager. This option
7644 turns off all this special overlay handling.
7646 @cindex SPU overlay stub symbols
7647 @kindex --emit-stub-syms
7648 @item --emit-stub-syms
7649 This option causes @command{ld} to label overlay stubs with a local
7650 symbol that encodes the stub type and destination.
7652 @cindex SPU extra overlay stubs
7653 @kindex --extra-overlay-stubs
7654 @item --extra-overlay-stubs
7655 This option causes @command{ld} to add overlay call stubs on all
7656 function calls out of overlay regions. Normally stubs are not added
7657 on calls to non-overlay regions.
7659 @cindex SPU local store size
7660 @kindex --local-store=lo:hi
7661 @item --local-store=lo:hi
7662 @command{ld} usually checks that a final executable for SPU fits in
7663 the address range 0 to 256k. This option may be used to change the
7664 range. Disable the check entirely with @option{--local-store=0:0}.
7667 @kindex --stack-analysis
7668 @item --stack-analysis
7669 SPU local store space is limited. Over-allocation of stack space
7670 unnecessarily limits space available for code and data, while
7671 under-allocation results in runtime failures. If given this option,
7672 @command{ld} will provide an estimate of maximum stack usage.
7673 @command{ld} does this by examining symbols in code sections to
7674 determine the extents of functions, and looking at function prologues
7675 for stack adjusting instructions. A call-graph is created by looking
7676 for relocations on branch instructions. The graph is then searched
7677 for the maximum stack usage path. Note that this analysis does not
7678 find calls made via function pointers, and does not handle recursion
7679 and other cycles in the call graph. Stack usage may be
7680 under-estimated if your code makes such calls. Also, stack usage for
7681 dynamic allocation, e.g. alloca, will not be detected. If a link map
7682 is requested, detailed information about each function's stack usage
7683 and calls will be given.
7686 @kindex --emit-stack-syms
7687 @item --emit-stack-syms
7688 This option, if given along with @option{--stack-analysis} will result
7689 in @command{ld} emitting stack sizing symbols for each function.
7690 These take the form @code{__stack_<function_name>} for global
7691 functions, and @code{__stack_<number>_<function_name>} for static
7692 functions. @code{<number>} is the section id in hex. The value of
7693 such symbols is the stack requirement for the corresponding function.
7694 The symbol size will be zero, type @code{STT_NOTYPE}, binding
7695 @code{STB_LOCAL}, and section @code{SHN_ABS}.
7709 @section @command{ld}'s Support for Various TI COFF Versions
7710 @cindex TI COFF versions
7711 @kindex --format=@var{version}
7712 The @samp{--format} switch allows selection of one of the various
7713 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
7714 also supported. The TI COFF versions also vary in header byte-order
7715 format; @command{ld} will read any version or byte order, but the output
7716 header format depends on the default specified by the specific target.
7729 @section @command{ld} and WIN32 (cygwin/mingw)
7731 This section describes some of the win32 specific @command{ld} issues.
7732 See @ref{Options,,Command Line Options} for detailed description of the
7733 command line options mentioned here.
7736 @cindex import libraries
7737 @item import libraries
7738 The standard Windows linker creates and uses so-called import
7739 libraries, which contains information for linking to dll's. They are
7740 regular static archives and are handled as any other static
7741 archive. The cygwin and mingw ports of @command{ld} have specific
7742 support for creating such libraries provided with the
7743 @samp{--out-implib} command line option.
7745 @item exporting DLL symbols
7746 @cindex exporting DLL symbols
7747 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
7750 @item using auto-export functionality
7751 @cindex using auto-export functionality
7752 By default @command{ld} exports symbols with the auto-export functionality,
7753 which is controlled by the following command line options:
7756 @item --export-all-symbols [This is the default]
7757 @item --exclude-symbols
7758 @item --exclude-libs
7759 @item --exclude-modules-for-implib
7760 @item --version-script
7763 When auto-export is in operation, @command{ld} will export all the non-local
7764 (global and common) symbols it finds in a DLL, with the exception of a few
7765 symbols known to belong to the system's runtime and libraries. As it will
7766 often not be desirable to export all of a DLL's symbols, which may include
7767 private functions that are not part of any public interface, the command-line
7768 options listed above may be used to filter symbols out from the list for
7769 exporting. The @samp{--output-def} option can be used in order to see the
7770 final list of exported symbols with all exclusions taken into effect.
7772 If @samp{--export-all-symbols} is not given explicitly on the
7773 command line, then the default auto-export behavior will be @emph{disabled}
7774 if either of the following are true:
7777 @item A DEF file is used.
7778 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
7781 @item using a DEF file
7782 @cindex using a DEF file
7783 Another way of exporting symbols is using a DEF file. A DEF file is
7784 an ASCII file containing definitions of symbols which should be
7785 exported when a dll is created. Usually it is named @samp{<dll
7786 name>.def} and is added as any other object file to the linker's
7787 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7790 gcc -o <output> <objectfiles> <dll name>.def
7793 Using a DEF file turns off the normal auto-export behavior, unless the
7794 @samp{--export-all-symbols} option is also used.
7796 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7799 LIBRARY "xyz.dll" BASE=0x20000000
7805 another_foo = abc.dll.afoo
7811 This example defines a DLL with a non-default base address and seven
7812 symbols in the export table. The third exported symbol @code{_bar} is an
7813 alias for the second. The fourth symbol, @code{another_foo} is resolved
7814 by "forwarding" to another module and treating it as an alias for
7815 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7816 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7817 export library is an alias of @samp{foo}, which gets the string name
7818 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7819 symbol, which gets in export table the name @samp{var1}.
7821 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7822 name of the output DLL. If @samp{<name>} does not include a suffix,
7823 the default library suffix, @samp{.DLL} is appended.
7825 When the .DEF file is used to build an application, rather than a
7826 library, the @code{NAME <name>} command should be used instead of
7827 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7828 executable suffix, @samp{.EXE} is appended.
7830 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7831 specification @code{BASE = <number>} may be used to specify a
7832 non-default base address for the image.
7834 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7835 or they specify an empty string, the internal name is the same as the
7836 filename specified on the command line.
7838 The complete specification of an export symbol is:
7842 ( ( ( <name1> [ = <name2> ] )
7843 | ( <name1> = <module-name> . <external-name>))
7844 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7847 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7848 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7849 @samp{<name1>} as a "forward" alias for the symbol
7850 @samp{<external-name>} in the DLL @samp{<module-name>}.
7851 Optionally, the symbol may be exported by the specified ordinal
7852 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7853 string in import/export table for the symbol.
7855 The optional keywords that follow the declaration indicate:
7857 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7858 will still be exported by its ordinal alias (either the value specified
7859 by the .def specification or, otherwise, the value assigned by the
7860 linker). The symbol name, however, does remain visible in the import
7861 library (if any), unless @code{PRIVATE} is also specified.
7863 @code{DATA}: The symbol is a variable or object, rather than a function.
7864 The import lib will export only an indirect reference to @code{foo} as
7865 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7868 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7869 well as @code{_imp__foo} into the import library. Both refer to the
7870 read-only import address table's pointer to the variable, not to the
7871 variable itself. This can be dangerous. If the user code fails to add
7872 the @code{dllimport} attribute and also fails to explicitly add the
7873 extra indirection that the use of the attribute enforces, the
7874 application will behave unexpectedly.
7876 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7877 it into the static import library used to resolve imports at link time. The
7878 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7879 API at runtime or by using the GNU ld extension of linking directly to
7880 the DLL without an import library.
7882 See ld/deffilep.y in the binutils sources for the full specification of
7883 other DEF file statements
7885 @cindex creating a DEF file
7886 While linking a shared dll, @command{ld} is able to create a DEF file
7887 with the @samp{--output-def <file>} command line option.
7889 @item Using decorations
7890 @cindex Using decorations
7891 Another way of marking symbols for export is to modify the source code
7892 itself, so that when building the DLL each symbol to be exported is
7896 __declspec(dllexport) int a_variable
7897 __declspec(dllexport) void a_function(int with_args)
7900 All such symbols will be exported from the DLL. If, however,
7901 any of the object files in the DLL contain symbols decorated in
7902 this way, then the normal auto-export behavior is disabled, unless
7903 the @samp{--export-all-symbols} option is also used.
7905 Note that object files that wish to access these symbols must @emph{not}
7906 decorate them with dllexport. Instead, they should use dllimport,
7910 __declspec(dllimport) int a_variable
7911 __declspec(dllimport) void a_function(int with_args)
7914 This complicates the structure of library header files, because
7915 when included by the library itself the header must declare the
7916 variables and functions as dllexport, but when included by client
7917 code the header must declare them as dllimport. There are a number
7918 of idioms that are typically used to do this; often client code can
7919 omit the __declspec() declaration completely. See
7920 @samp{--enable-auto-import} and @samp{automatic data imports} for more
7924 @cindex automatic data imports
7925 @item automatic data imports
7926 The standard Windows dll format supports data imports from dlls only
7927 by adding special decorations (dllimport/dllexport), which let the
7928 compiler produce specific assembler instructions to deal with this
7929 issue. This increases the effort necessary to port existing Un*x
7930 code to these platforms, especially for large
7931 c++ libraries and applications. The auto-import feature, which was
7932 initially provided by Paul Sokolovsky, allows one to omit the
7933 decorations to achieve a behavior that conforms to that on POSIX/Un*x
7934 platforms. This feature is enabled with the @samp{--enable-auto-import}
7935 command-line option, although it is enabled by default on cygwin/mingw.
7936 The @samp{--enable-auto-import} option itself now serves mainly to
7937 suppress any warnings that are ordinarily emitted when linked objects
7938 trigger the feature's use.
7940 auto-import of variables does not always work flawlessly without
7941 additional assistance. Sometimes, you will see this message
7943 "variable '<var>' can't be auto-imported. Please read the
7944 documentation for ld's @code{--enable-auto-import} for details."
7946 The @samp{--enable-auto-import} documentation explains why this error
7947 occurs, and several methods that can be used to overcome this difficulty.
7948 One of these methods is the @emph{runtime pseudo-relocs} feature, described
7951 @cindex runtime pseudo-relocation
7952 For complex variables imported from DLLs (such as structs or classes),
7953 object files typically contain a base address for the variable and an
7954 offset (@emph{addend}) within the variable--to specify a particular
7955 field or public member, for instance. Unfortunately, the runtime loader used
7956 in win32 environments is incapable of fixing these references at runtime
7957 without the additional information supplied by dllimport/dllexport decorations.
7958 The standard auto-import feature described above is unable to resolve these
7961 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
7962 be resolved without error, while leaving the task of adjusting the references
7963 themselves (with their non-zero addends) to specialized code provided by the
7964 runtime environment. Recent versions of the cygwin and mingw environments and
7965 compilers provide this runtime support; older versions do not. However, the
7966 support is only necessary on the developer's platform; the compiled result will
7967 run without error on an older system.
7969 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
7972 @cindex direct linking to a dll
7973 @item direct linking to a dll
7974 The cygwin/mingw ports of @command{ld} support the direct linking,
7975 including data symbols, to a dll without the usage of any import
7976 libraries. This is much faster and uses much less memory than does the
7977 traditional import library method, especially when linking large
7978 libraries or applications. When @command{ld} creates an import lib, each
7979 function or variable exported from the dll is stored in its own bfd, even
7980 though a single bfd could contain many exports. The overhead involved in
7981 storing, loading, and processing so many bfd's is quite large, and explains the
7982 tremendous time, memory, and storage needed to link against particularly
7983 large or complex libraries when using import libs.
7985 Linking directly to a dll uses no extra command-line switches other than
7986 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
7987 of names to match each library. All that is needed from the developer's
7988 perspective is an understanding of this search, in order to force ld to
7989 select the dll instead of an import library.
7992 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
7993 to find, in the first directory of its search path,
8006 before moving on to the next directory in the search path.
8008 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
8009 where @samp{<prefix>} is set by the @command{ld} option
8010 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
8011 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
8014 Other win32-based unix environments, such as mingw or pw32, may use other
8015 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
8016 was originally intended to help avoid name conflicts among dll's built for the
8017 various win32/un*x environments, so that (for example) two versions of a zlib dll
8018 could coexist on the same machine.
8020 The generic cygwin/mingw path layout uses a @samp{bin} directory for
8021 applications and dll's and a @samp{lib} directory for the import
8022 libraries (using cygwin nomenclature):
8028 libxxx.dll.a (in case of dll's)
8029 libxxx.a (in case of static archive)
8032 Linking directly to a dll without using the import library can be
8035 1. Use the dll directly by adding the @samp{bin} path to the link line
8037 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
8040 However, as the dll's often have version numbers appended to their names
8041 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
8042 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
8043 not versioned, and do not have this difficulty.
8045 2. Create a symbolic link from the dll to a file in the @samp{lib}
8046 directory according to the above mentioned search pattern. This
8047 should be used to avoid unwanted changes in the tools needed for
8051 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
8054 Then you can link without any make environment changes.
8057 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
8060 This technique also avoids the version number problems, because the following is
8067 libxxx.dll.a -> ../bin/cygxxx-5.dll
8070 Linking directly to a dll without using an import lib will work
8071 even when auto-import features are exercised, and even when
8072 @samp{--enable-runtime-pseudo-relocs} is used.
8074 Given the improvements in speed and memory usage, one might justifiably
8075 wonder why import libraries are used at all. There are three reasons:
8077 1. Until recently, the link-directly-to-dll functionality did @emph{not}
8078 work with auto-imported data.
8080 2. Sometimes it is necessary to include pure static objects within the
8081 import library (which otherwise contains only bfd's for indirection
8082 symbols that point to the exports of a dll). Again, the import lib
8083 for the cygwin kernel makes use of this ability, and it is not
8084 possible to do this without an import lib.
8086 3. Symbol aliases can only be resolved using an import lib. This is
8087 critical when linking against OS-supplied dll's (eg, the win32 API)
8088 in which symbols are usually exported as undecorated aliases of their
8089 stdcall-decorated assembly names.
8091 So, import libs are not going away. But the ability to replace
8092 true import libs with a simple symbolic link to (or a copy of)
8093 a dll, in many cases, is a useful addition to the suite of tools
8094 binutils makes available to the win32 developer. Given the
8095 massive improvements in memory requirements during linking, storage
8096 requirements, and linking speed, we expect that many developers
8097 will soon begin to use this feature whenever possible.
8099 @item symbol aliasing
8101 @item adding additional names
8102 Sometimes, it is useful to export symbols with additional names.
8103 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
8104 exported as @samp{_foo} by using special directives in the DEF file
8105 when creating the dll. This will affect also the optional created
8106 import library. Consider the following DEF file:
8109 LIBRARY "xyz.dll" BASE=0x61000000
8116 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
8118 Another method for creating a symbol alias is to create it in the
8119 source code using the "weak" attribute:
8122 void foo () @{ /* Do something. */; @}
8123 void _foo () __attribute__ ((weak, alias ("foo")));
8126 See the gcc manual for more information about attributes and weak
8129 @item renaming symbols
8130 Sometimes it is useful to rename exports. For instance, the cygwin
8131 kernel does this regularly. A symbol @samp{_foo} can be exported as
8132 @samp{foo} but not as @samp{_foo} by using special directives in the
8133 DEF file. (This will also affect the import library, if it is
8134 created). In the following example:
8137 LIBRARY "xyz.dll" BASE=0x61000000
8143 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
8147 Note: using a DEF file disables the default auto-export behavior,
8148 unless the @samp{--export-all-symbols} command line option is used.
8149 If, however, you are trying to rename symbols, then you should list
8150 @emph{all} desired exports in the DEF file, including the symbols
8151 that are not being renamed, and do @emph{not} use the
8152 @samp{--export-all-symbols} option. If you list only the
8153 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
8154 to handle the other symbols, then the both the new names @emph{and}
8155 the original names for the renamed symbols will be exported.
8156 In effect, you'd be aliasing those symbols, not renaming them,
8157 which is probably not what you wanted.
8159 @cindex weak externals
8160 @item weak externals
8161 The Windows object format, PE, specifies a form of weak symbols called
8162 weak externals. When a weak symbol is linked and the symbol is not
8163 defined, the weak symbol becomes an alias for some other symbol. There
8164 are three variants of weak externals:
8166 @item Definition is searched for in objects and libraries, historically
8167 called lazy externals.
8168 @item Definition is searched for only in other objects, not in libraries.
8169 This form is not presently implemented.
8170 @item No search; the symbol is an alias. This form is not presently
8173 As a GNU extension, weak symbols that do not specify an alternate symbol
8174 are supported. If the symbol is undefined when linking, the symbol
8175 uses a default value.
8177 @cindex aligned common symbols
8178 @item aligned common symbols
8179 As a GNU extension to the PE file format, it is possible to specify the
8180 desired alignment for a common symbol. This information is conveyed from
8181 the assembler or compiler to the linker by means of GNU-specific commands
8182 carried in the object file's @samp{.drectve} section, which are recognized
8183 by @command{ld} and respected when laying out the common symbols. Native
8184 tools will be able to process object files employing this GNU extension,
8185 but will fail to respect the alignment instructions, and may issue noisy
8186 warnings about unknown linker directives.
8201 @section @code{ld} and Xtensa Processors
8203 @cindex Xtensa processors
8204 The default @command{ld} behavior for Xtensa processors is to interpret
8205 @code{SECTIONS} commands so that lists of explicitly named sections in a
8206 specification with a wildcard file will be interleaved when necessary to
8207 keep literal pools within the range of PC-relative load offsets. For
8208 example, with the command:
8220 @command{ld} may interleave some of the @code{.literal}
8221 and @code{.text} sections from different object files to ensure that the
8222 literal pools are within the range of PC-relative load offsets. A valid
8223 interleaving might place the @code{.literal} sections from an initial
8224 group of files followed by the @code{.text} sections of that group of
8225 files. Then, the @code{.literal} sections from the rest of the files
8226 and the @code{.text} sections from the rest of the files would follow.
8228 @cindex @option{--relax} on Xtensa
8229 @cindex relaxing on Xtensa
8230 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8231 provides two important link-time optimizations. The first optimization
8232 is to combine identical literal values to reduce code size. A redundant
8233 literal will be removed and all the @code{L32R} instructions that use it
8234 will be changed to reference an identical literal, as long as the
8235 location of the replacement literal is within the offset range of all
8236 the @code{L32R} instructions. The second optimization is to remove
8237 unnecessary overhead from assembler-generated ``longcall'' sequences of
8238 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8239 range of direct @code{CALL@var{n}} instructions.
8241 For each of these cases where an indirect call sequence can be optimized
8242 to a direct call, the linker will change the @code{CALLX@var{n}}
8243 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8244 instruction, and remove the literal referenced by the @code{L32R}
8245 instruction if it is not used for anything else. Removing the
8246 @code{L32R} instruction always reduces code size but can potentially
8247 hurt performance by changing the alignment of subsequent branch targets.
8248 By default, the linker will always preserve alignments, either by
8249 switching some instructions between 24-bit encodings and the equivalent
8250 density instructions or by inserting a no-op in place of the @code{L32R}
8251 instruction that was removed. If code size is more important than
8252 performance, the @option{--size-opt} option can be used to prevent the
8253 linker from widening density instructions or inserting no-ops, except in
8254 a few cases where no-ops are required for correctness.
8256 The following Xtensa-specific command-line options can be used to
8259 @cindex Xtensa options
8262 When optimizing indirect calls to direct calls, optimize for code size
8263 more than performance. With this option, the linker will not insert
8264 no-ops or widen density instructions to preserve branch target
8265 alignment. There may still be some cases where no-ops are required to
8266 preserve the correctness of the code.
8274 @ifclear SingleFormat
8279 @cindex object file management
8280 @cindex object formats available
8282 The linker accesses object and archive files using the BFD libraries.
8283 These libraries allow the linker to use the same routines to operate on
8284 object files whatever the object file format. A different object file
8285 format can be supported simply by creating a new BFD back end and adding
8286 it to the library. To conserve runtime memory, however, the linker and
8287 associated tools are usually configured to support only a subset of the
8288 object file formats available. You can use @code{objdump -i}
8289 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8290 list all the formats available for your configuration.
8292 @cindex BFD requirements
8293 @cindex requirements for BFD
8294 As with most implementations, BFD is a compromise between
8295 several conflicting requirements. The major factor influencing
8296 BFD design was efficiency: any time used converting between
8297 formats is time which would not have been spent had BFD not
8298 been involved. This is partly offset by abstraction payback; since
8299 BFD simplifies applications and back ends, more time and care
8300 may be spent optimizing algorithms for a greater speed.
8302 One minor artifact of the BFD solution which you should bear in
8303 mind is the potential for information loss. There are two places where
8304 useful information can be lost using the BFD mechanism: during
8305 conversion and during output. @xref{BFD information loss}.
8308 * BFD outline:: How it works: an outline of BFD
8312 @section How It Works: An Outline of BFD
8313 @cindex opening object files
8314 @include bfdsumm.texi
8317 @node Reporting Bugs
8318 @chapter Reporting Bugs
8319 @cindex bugs in @command{ld}
8320 @cindex reporting bugs in @command{ld}
8322 Your bug reports play an essential role in making @command{ld} reliable.
8324 Reporting a bug may help you by bringing a solution to your problem, or
8325 it may not. But in any case the principal function of a bug report is
8326 to help the entire community by making the next version of @command{ld}
8327 work better. Bug reports are your contribution to the maintenance of
8330 In order for a bug report to serve its purpose, you must include the
8331 information that enables us to fix the bug.
8334 * Bug Criteria:: Have you found a bug?
8335 * Bug Reporting:: How to report bugs
8339 @section Have You Found a Bug?
8340 @cindex bug criteria
8342 If you are not sure whether you have found a bug, here are some guidelines:
8345 @cindex fatal signal
8346 @cindex linker crash
8347 @cindex crash of linker
8349 If the linker gets a fatal signal, for any input whatever, that is a
8350 @command{ld} bug. Reliable linkers never crash.
8352 @cindex error on valid input
8354 If @command{ld} produces an error message for valid input, that is a bug.
8356 @cindex invalid input
8358 If @command{ld} does not produce an error message for invalid input, that
8359 may be a bug. In the general case, the linker can not verify that
8360 object files are correct.
8363 If you are an experienced user of linkers, your suggestions for
8364 improvement of @command{ld} are welcome in any case.
8368 @section How to Report Bugs
8370 @cindex @command{ld} bugs, reporting
8372 A number of companies and individuals offer support for @sc{gnu}
8373 products. If you obtained @command{ld} from a support organization, we
8374 recommend you contact that organization first.
8376 You can find contact information for many support companies and
8377 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
8381 Otherwise, send bug reports for @command{ld} to
8385 The fundamental principle of reporting bugs usefully is this:
8386 @strong{report all the facts}. If you are not sure whether to state a
8387 fact or leave it out, state it!
8389 Often people omit facts because they think they know what causes the
8390 problem and assume that some details do not matter. Thus, you might
8391 assume that the name of a symbol you use in an example does not
8392 matter. Well, probably it does not, but one cannot be sure. Perhaps
8393 the bug is a stray memory reference which happens to fetch from the
8394 location where that name is stored in memory; perhaps, if the name
8395 were different, the contents of that location would fool the linker
8396 into doing the right thing despite the bug. Play it safe and give a
8397 specific, complete example. That is the easiest thing for you to do,
8398 and the most helpful.
8400 Keep in mind that the purpose of a bug report is to enable us to fix
8401 the bug if it is new to us. Therefore, always write your bug reports
8402 on the assumption that the bug has not been reported previously.
8404 Sometimes people give a few sketchy facts and ask, ``Does this ring a
8405 bell?'' This cannot help us fix a bug, so it is basically useless. We
8406 respond by asking for enough details to enable us to investigate.
8407 You might as well expedite matters by sending them to begin with.
8409 To enable us to fix the bug, you should include all these things:
8413 The version of @command{ld}. @command{ld} announces it if you start it with
8414 the @samp{--version} argument.
8416 Without this, we will not know whether there is any point in looking for
8417 the bug in the current version of @command{ld}.
8420 Any patches you may have applied to the @command{ld} source, including any
8421 patches made to the @code{BFD} library.
8424 The type of machine you are using, and the operating system name and
8428 What compiler (and its version) was used to compile @command{ld}---e.g.
8432 The command arguments you gave the linker to link your example and
8433 observe the bug. To guarantee you will not omit something important,
8434 list them all. A copy of the Makefile (or the output from make) is
8437 If we were to try to guess the arguments, we would probably guess wrong
8438 and then we might not encounter the bug.
8441 A complete input file, or set of input files, that will reproduce the
8442 bug. It is generally most helpful to send the actual object files
8443 provided that they are reasonably small. Say no more than 10K. For
8444 bigger files you can either make them available by FTP or HTTP or else
8445 state that you are willing to send the object file(s) to whomever
8446 requests them. (Note - your email will be going to a mailing list, so
8447 we do not want to clog it up with large attachments). But small
8448 attachments are best.
8450 If the source files were assembled using @code{gas} or compiled using
8451 @code{gcc}, then it may be OK to send the source files rather than the
8452 object files. In this case, be sure to say exactly what version of
8453 @code{gas} or @code{gcc} was used to produce the object files. Also say
8454 how @code{gas} or @code{gcc} were configured.
8457 A description of what behavior you observe that you believe is
8458 incorrect. For example, ``It gets a fatal signal.''
8460 Of course, if the bug is that @command{ld} gets a fatal signal, then we
8461 will certainly notice it. But if the bug is incorrect output, we might
8462 not notice unless it is glaringly wrong. You might as well not give us
8463 a chance to make a mistake.
8465 Even if the problem you experience is a fatal signal, you should still
8466 say so explicitly. Suppose something strange is going on, such as, your
8467 copy of @command{ld} is out of sync, or you have encountered a bug in the
8468 C library on your system. (This has happened!) Your copy might crash
8469 and ours would not. If you told us to expect a crash, then when ours
8470 fails to crash, we would know that the bug was not happening for us. If
8471 you had not told us to expect a crash, then we would not be able to draw
8472 any conclusion from our observations.
8475 If you wish to suggest changes to the @command{ld} source, send us context
8476 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
8477 @samp{-p} option. Always send diffs from the old file to the new file.
8478 If you even discuss something in the @command{ld} source, refer to it by
8479 context, not by line number.
8481 The line numbers in our development sources will not match those in your
8482 sources. Your line numbers would convey no useful information to us.
8485 Here are some things that are not necessary:
8489 A description of the envelope of the bug.
8491 Often people who encounter a bug spend a lot of time investigating
8492 which changes to the input file will make the bug go away and which
8493 changes will not affect it.
8495 This is often time consuming and not very useful, because the way we
8496 will find the bug is by running a single example under the debugger
8497 with breakpoints, not by pure deduction from a series of examples.
8498 We recommend that you save your time for something else.
8500 Of course, if you can find a simpler example to report @emph{instead}
8501 of the original one, that is a convenience for us. Errors in the
8502 output will be easier to spot, running under the debugger will take
8503 less time, and so on.
8505 However, simplification is not vital; if you do not want to do this,
8506 report the bug anyway and send us the entire test case you used.
8509 A patch for the bug.
8511 A patch for the bug does help us if it is a good one. But do not omit
8512 the necessary information, such as the test case, on the assumption that
8513 a patch is all we need. We might see problems with your patch and decide
8514 to fix the problem another way, or we might not understand it at all.
8516 Sometimes with a program as complicated as @command{ld} it is very hard to
8517 construct an example that will make the program follow a certain path
8518 through the code. If you do not send us the example, we will not be
8519 able to construct one, so we will not be able to verify that the bug is
8522 And if we cannot understand what bug you are trying to fix, or why your
8523 patch should be an improvement, we will not install it. A test case will
8524 help us to understand.
8527 A guess about what the bug is or what it depends on.
8529 Such guesses are usually wrong. Even we cannot guess right about such
8530 things without first using the debugger to find the facts.
8534 @appendix MRI Compatible Script Files
8535 @cindex MRI compatibility
8536 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
8537 linker, @command{ld} can use MRI compatible linker scripts as an
8538 alternative to the more general-purpose linker scripting language
8539 described in @ref{Scripts}. MRI compatible linker scripts have a much
8540 simpler command set than the scripting language otherwise used with
8541 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
8542 linker commands; these commands are described here.
8544 In general, MRI scripts aren't of much use with the @code{a.out} object
8545 file format, since it only has three sections and MRI scripts lack some
8546 features to make use of them.
8548 You can specify a file containing an MRI-compatible script using the
8549 @samp{-c} command-line option.
8551 Each command in an MRI-compatible script occupies its own line; each
8552 command line starts with the keyword that identifies the command (though
8553 blank lines are also allowed for punctuation). If a line of an
8554 MRI-compatible script begins with an unrecognized keyword, @command{ld}
8555 issues a warning message, but continues processing the script.
8557 Lines beginning with @samp{*} are comments.
8559 You can write these commands using all upper-case letters, or all
8560 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
8561 The following list shows only the upper-case form of each command.
8564 @cindex @code{ABSOLUTE} (MRI)
8565 @item ABSOLUTE @var{secname}
8566 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
8567 Normally, @command{ld} includes in the output file all sections from all
8568 the input files. However, in an MRI-compatible script, you can use the
8569 @code{ABSOLUTE} command to restrict the sections that will be present in
8570 your output program. If the @code{ABSOLUTE} command is used at all in a
8571 script, then only the sections named explicitly in @code{ABSOLUTE}
8572 commands will appear in the linker output. You can still use other
8573 input sections (whatever you select on the command line, or using
8574 @code{LOAD}) to resolve addresses in the output file.
8576 @cindex @code{ALIAS} (MRI)
8577 @item ALIAS @var{out-secname}, @var{in-secname}
8578 Use this command to place the data from input section @var{in-secname}
8579 in a section called @var{out-secname} in the linker output file.
8581 @var{in-secname} may be an integer.
8583 @cindex @code{ALIGN} (MRI)
8584 @item ALIGN @var{secname} = @var{expression}
8585 Align the section called @var{secname} to @var{expression}. The
8586 @var{expression} should be a power of two.
8588 @cindex @code{BASE} (MRI)
8589 @item BASE @var{expression}
8590 Use the value of @var{expression} as the lowest address (other than
8591 absolute addresses) in the output file.
8593 @cindex @code{CHIP} (MRI)
8594 @item CHIP @var{expression}
8595 @itemx CHIP @var{expression}, @var{expression}
8596 This command does nothing; it is accepted only for compatibility.
8598 @cindex @code{END} (MRI)
8600 This command does nothing whatever; it's only accepted for compatibility.
8602 @cindex @code{FORMAT} (MRI)
8603 @item FORMAT @var{output-format}
8604 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
8605 language, but restricted to S-records, if @var{output-format} is @samp{S}
8607 @cindex @code{LIST} (MRI)
8608 @item LIST @var{anything}@dots{}
8609 Print (to the standard output file) a link map, as produced by the
8610 @command{ld} command-line option @samp{-M}.
8612 The keyword @code{LIST} may be followed by anything on the
8613 same line, with no change in its effect.
8615 @cindex @code{LOAD} (MRI)
8616 @item LOAD @var{filename}
8617 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
8618 Include one or more object file @var{filename} in the link; this has the
8619 same effect as specifying @var{filename} directly on the @command{ld}
8622 @cindex @code{NAME} (MRI)
8623 @item NAME @var{output-name}
8624 @var{output-name} is the name for the program produced by @command{ld}; the
8625 MRI-compatible command @code{NAME} is equivalent to the command-line
8626 option @samp{-o} or the general script language command @code{OUTPUT}.
8628 @cindex @code{ORDER} (MRI)
8629 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
8630 @itemx ORDER @var{secname} @var{secname} @var{secname}
8631 Normally, @command{ld} orders the sections in its output file in the
8632 order in which they first appear in the input files. In an MRI-compatible
8633 script, you can override this ordering with the @code{ORDER} command. The
8634 sections you list with @code{ORDER} will appear first in your output
8635 file, in the order specified.
8637 @cindex @code{PUBLIC} (MRI)
8638 @item PUBLIC @var{name}=@var{expression}
8639 @itemx PUBLIC @var{name},@var{expression}
8640 @itemx PUBLIC @var{name} @var{expression}
8641 Supply a value (@var{expression}) for external symbol
8642 @var{name} used in the linker input files.
8644 @cindex @code{SECT} (MRI)
8645 @item SECT @var{secname}, @var{expression}
8646 @itemx SECT @var{secname}=@var{expression}
8647 @itemx SECT @var{secname} @var{expression}
8648 You can use any of these three forms of the @code{SECT} command to
8649 specify the start address (@var{expression}) for section @var{secname}.
8650 If you have more than one @code{SECT} statement for the same
8651 @var{secname}, only the @emph{first} sets the start address.
8654 @node GNU Free Documentation License
8655 @appendix GNU Free Documentation License
8659 @unnumbered LD Index
8664 % I think something like @@colophon should be in texinfo. In the
8666 \long\def\colophon{\hbox to0pt{}\vfill
8667 \centerline{The body of this manual is set in}
8668 \centerline{\fontname\tenrm,}
8669 \centerline{with headings in {\bf\fontname\tenbf}}
8670 \centerline{and examples in {\tt\fontname\tentt}.}
8671 \centerline{{\it\fontname\tenit\/} and}
8672 \centerline{{\sl\fontname\tensl\/}}
8673 \centerline{are used for emphasis.}\vfill}
8675 % Blame: doc@@cygnus.com, 28mar91.