3 @c Copyright (C) 1991-2022 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
47 @dircategory Software development
49 * Ld: (ld). The GNU linker.
54 This file documents the @sc{gnu} linker LD
55 @ifset VERSION_PACKAGE
56 @value{VERSION_PACKAGE}
58 version @value{VERSION}.
60 Copyright @copyright{} 1991-2022 Free Software Foundation, Inc.
62 Permission is granted to copy, distribute and/or modify this document
63 under the terms of the GNU Free Documentation License, Version 1.3
64 or any later version published by the Free Software Foundation;
65 with no Invariant Sections, with no Front-Cover Texts, and with no
66 Back-Cover Texts. A copy of the license is included in the
67 section entitled ``GNU Free Documentation License''.
71 @setchapternewpage odd
72 @settitle The GNU linker
77 @ifset VERSION_PACKAGE
78 @subtitle @value{VERSION_PACKAGE}
80 @subtitle Version @value{VERSION}
81 @author Steve Chamberlain
82 @author Ian Lance Taylor
87 \hfill Red Hat Inc\par
88 \hfill nickc\@credhat.com, doc\@redhat.com\par
89 \hfill {\it The GNU linker}\par
90 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
92 \global\parindent=0pt % Steve likes it this way.
95 @vskip 0pt plus 1filll
96 @c man begin COPYRIGHT
97 Copyright @copyright{} 1991-2022 Free Software Foundation, Inc.
99 Permission is granted to copy, distribute and/or modify this document
100 under the terms of the GNU Free Documentation License, Version 1.3
101 or any later version published by the Free Software Foundation;
102 with no Invariant Sections, with no Front-Cover Texts, and with no
103 Back-Cover Texts. A copy of the license is included in the
104 section entitled ``GNU Free Documentation License''.
110 @c FIXME: Talk about importance of *order* of args, cmds to linker!
115 This file documents the @sc{gnu} linker ld
116 @ifset VERSION_PACKAGE
117 @value{VERSION_PACKAGE}
119 version @value{VERSION}.
121 This document is distributed under the terms of the GNU Free
122 Documentation License version 1.3. A copy of the license is included
123 in the section entitled ``GNU Free Documentation License''.
126 * Overview:: Overview
127 * Invocation:: Invocation
128 * Scripts:: Linker Scripts
129 * Plugins:: Linker Plugins
131 * Machine Dependent:: Machine Dependent Features
135 * H8/300:: ld and the H8/300
138 * Renesas:: ld and other Renesas micros
141 * ARM:: ld and the ARM family
144 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
147 * HPPA ELF32:: ld and HPPA 32-bit ELF
150 * M68K:: ld and Motorola 68K family
153 * MIPS:: ld and MIPS family
156 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
159 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
162 * S/390 ELF:: ld and S/390 ELF Support
165 * SPU ELF:: ld and SPU ELF Support
168 * TI COFF:: ld and the TI COFF
171 * Win32:: ld and WIN32 (cygwin/mingw)
174 * Xtensa:: ld and Xtensa Processors
177 @ifclear SingleFormat
180 @c Following blank line required for remaining bug in makeinfo conds/menus
182 * Reporting Bugs:: Reporting Bugs
183 * MRI:: MRI Compatible Script Files
184 * GNU Free Documentation License:: GNU Free Documentation License
185 * LD Index:: LD Index
192 @cindex @sc{gnu} linker
193 @cindex what is this?
196 @c man begin SYNOPSIS
197 ld [@b{options}] @var{objfile} @dots{}
201 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
202 the Info entries for @file{binutils} and
207 @c man begin DESCRIPTION
209 @command{ld} combines a number of object and archive files, relocates
210 their data and ties up symbol references. Usually the last step in
211 compiling a program is to run @command{ld}.
213 @command{ld} accepts Linker Command Language files written in
214 a superset of AT&T's Link Editor Command Language syntax,
215 to provide explicit and total control over the linking process.
219 This man page does not describe the command language; see the
220 @command{ld} entry in @code{info} for full details on the command
221 language and on other aspects of the GNU linker.
224 @ifclear SingleFormat
225 This version of @command{ld} uses the general purpose BFD libraries
226 to operate on object files. This allows @command{ld} to read, combine, and
227 write object files in many different formats---for example, COFF or
228 @code{a.out}. Different formats may be linked together to produce any
229 available kind of object file. @xref{BFD}, for more information.
232 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
233 linkers in providing diagnostic information. Many linkers abandon
234 execution immediately upon encountering an error; whenever possible,
235 @command{ld} continues executing, allowing you to identify other errors
236 (or, in some cases, to get an output file in spite of the error).
243 @c man begin DESCRIPTION
245 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
246 and to be as compatible as possible with other linkers. As a result,
247 you have many choices to control its behavior.
253 * Options:: Command-line Options
254 * Environment:: Environment Variables
258 @section Command-line Options
266 The linker supports a plethora of command-line options, but in actual
267 practice few of them are used in any particular context.
268 @cindex standard Unix system
269 For instance, a frequent use of @command{ld} is to link standard Unix
270 object files on a standard, supported Unix system. On such a system, to
271 link a file @code{hello.o}:
274 ld -o @var{output} /lib/crt0.o hello.o -lc
277 This tells @command{ld} to produce a file called @var{output} as the
278 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
279 the library @code{libc.a}, which will come from the standard search
280 directories. (See the discussion of the @samp{-l} option below.)
282 Some of the command-line options to @command{ld} may be specified at any
283 point in the command line. However, options which refer to files, such
284 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
285 which the option appears in the command line, relative to the object
286 files and other file options. Repeating non-file options with a
287 different argument will either have no further effect, or override prior
288 occurrences (those further to the left on the command line) of that
289 option. Options which may be meaningfully specified more than once are
290 noted in the descriptions below.
293 Non-option arguments are object files or archives which are to be linked
294 together. They may follow, precede, or be mixed in with command-line
295 options, except that an object file argument may not be placed between
296 an option and its argument.
298 Usually the linker is invoked with at least one object file, but you can
299 specify other forms of binary input files using @samp{-l}, @samp{-R},
300 and the script command language. If @emph{no} binary input files at all
301 are specified, the linker does not produce any output, and issues the
302 message @samp{No input files}.
304 If the linker cannot recognize the format of an object file, it will
305 assume that it is a linker script. A script specified in this way
306 augments the main linker script used for the link (either the default
307 linker script or the one specified by using @samp{-T}). This feature
308 permits the linker to link against a file which appears to be an object
309 or an archive, but actually merely defines some symbol values, or uses
310 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
311 script in this way merely augments the main linker script, with the
312 extra commands placed after the main script; use the @samp{-T} option
313 to replace the default linker script entirely, but note the effect of
314 the @code{INSERT} command. @xref{Scripts}.
316 For options whose names are a single letter,
317 option arguments must either follow the option letter without intervening
318 whitespace, or be given as separate arguments immediately following the
319 option that requires them.
321 For options whose names are multiple letters, either one dash or two can
322 precede the option name; for example, @samp{-trace-symbol} and
323 @samp{--trace-symbol} are equivalent. Note---there is one exception to
324 this rule. Multiple letter options that start with a lower case 'o' can
325 only be preceded by two dashes. This is to reduce confusion with the
326 @samp{-o} option. So for example @samp{-omagic} sets the output file
327 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
330 Arguments to multiple-letter options must either be separated from the
331 option name by an equals sign, or be given as separate arguments
332 immediately following the option that requires them. For example,
333 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
334 Unique abbreviations of the names of multiple-letter options are
337 Note---if the linker is being invoked indirectly, via a compiler driver
338 (e.g. @samp{gcc}) then all the linker command-line options should be
339 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
340 compiler driver) like this:
343 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
346 This is important, because otherwise the compiler driver program may
347 silently drop the linker options, resulting in a bad link. Confusion
348 may also arise when passing options that require values through a
349 driver, as the use of a space between option and argument acts as
350 a separator, and causes the driver to pass only the option to the linker
351 and the argument to the compiler. In this case, it is simplest to use
352 the joined forms of both single- and multiple-letter options, such as:
355 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
358 Here is a table of the generic command-line switches accepted by the GNU
362 @include at-file.texi
364 @kindex -a @var{keyword}
365 @item -a @var{keyword}
366 This option is supported for HP/UX compatibility. The @var{keyword}
367 argument must be one of the strings @samp{archive}, @samp{shared}, or
368 @samp{default}. @samp{-aarchive} is functionally equivalent to
369 @samp{-Bstatic}, and the other two keywords are functionally equivalent
370 to @samp{-Bdynamic}. This option may be used any number of times.
372 @kindex --audit @var{AUDITLIB}
373 @item --audit @var{AUDITLIB}
374 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
375 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
376 specified in the library. If specified multiple times @code{DT_AUDIT}
377 will contain a colon separated list of audit interfaces to use. If the linker
378 finds an object with an audit entry while searching for shared libraries,
379 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
380 This option is only meaningful on ELF platforms supporting the rtld-audit
383 @ifclear SingleFormat
384 @cindex binary input format
385 @kindex -b @var{format}
386 @kindex --format=@var{format}
389 @item -b @var{input-format}
390 @itemx --format=@var{input-format}
391 @command{ld} may be configured to support more than one kind of object
392 file. If your @command{ld} is configured this way, you can use the
393 @samp{-b} option to specify the binary format for input object files
394 that follow this option on the command line. Even when @command{ld} is
395 configured to support alternative object formats, you don't usually need
396 to specify this, as @command{ld} should be configured to expect as a
397 default input format the most usual format on each machine.
398 @var{input-format} is a text string, the name of a particular format
399 supported by the BFD libraries. (You can list the available binary
400 formats with @samp{objdump -i}.)
403 You may want to use this option if you are linking files with an unusual
404 binary format. You can also use @samp{-b} to switch formats explicitly (when
405 linking object files of different formats), by including
406 @samp{-b @var{input-format}} before each group of object files in a
409 The default format is taken from the environment variable
414 You can also define the input format from a script, using the command
417 see @ref{Format Commands}.
421 @kindex -c @var{MRI-cmdfile}
422 @kindex --mri-script=@var{MRI-cmdfile}
423 @cindex compatibility, MRI
424 @item -c @var{MRI-commandfile}
425 @itemx --mri-script=@var{MRI-commandfile}
426 For compatibility with linkers produced by MRI, @command{ld} accepts script
427 files written in an alternate, restricted command language, described in
429 @ref{MRI,,MRI Compatible Script Files}.
432 the MRI Compatible Script Files section of GNU ld documentation.
434 Introduce MRI script files with
435 the option @samp{-c}; use the @samp{-T} option to run linker
436 scripts written in the general-purpose @command{ld} scripting language.
437 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
438 specified by any @samp{-L} options.
440 @cindex common allocation
447 These three options are equivalent; multiple forms are supported for
448 compatibility with other linkers. They assign space to common symbols
449 even if a relocatable output file is specified (with @samp{-r}). The
450 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
451 @xref{Miscellaneous Commands}.
453 @kindex --depaudit @var{AUDITLIB}
454 @kindex -P @var{AUDITLIB}
455 @item --depaudit @var{AUDITLIB}
456 @itemx -P @var{AUDITLIB}
457 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
458 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
459 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
460 will contain a colon separated list of audit interfaces to use. This
461 option is only meaningful on ELF platforms supporting the rtld-audit interface.
462 The -P option is provided for Solaris compatibility.
464 @kindex --enable-non-contiguous-regions
465 @item --enable-non-contiguous-regions
466 This option avoids generating an error if an input section does not
467 fit a matching output section. The linker tries to allocate the input
468 section to subseque nt matching output sections, and generates an
469 error only if no output section is large enough. This is useful when
470 several non-contiguous memory regions are available and the input
471 section does not require a particular one. The order in which input
472 sections are evaluated does not change, for instance:
476 MEM1 (rwx) : ORIGIN : 0x1000, LENGTH = 0x14
477 MEM2 (rwx) : ORIGIN : 0x1000, LENGTH = 0x40
478 MEM3 (rwx) : ORIGIN : 0x2000, LENGTH = 0x40
481 mem1 : @{ *(.data.*); @} > MEM1
482 mem2 : @{ *(.data.*); @} > MEM2
483 mem3 : @{ *(.data.*); @} > MEM2
491 results in .data.1 affected to mem1, and .data.2 and .data.3
492 affected to mem2, even though .data.3 would fit in mem3.
495 This option is incompatible with INSERT statements because it changes
496 the way input sections are mapped to output sections.
498 @kindex --enable-non-contiguous-regions-warnings
499 @item --enable-non-contiguous-regions-warnings
500 This option enables warnings when
501 @code{--enable-non-contiguous-regions} allows possibly unexpected
502 matches in sections mapping, potentially leading to silently
503 discarding a section instead of failing because it does not fit any
506 @cindex entry point, from command line
507 @kindex -e @var{entry}
508 @kindex --entry=@var{entry}
510 @itemx --entry=@var{entry}
511 Use @var{entry} as the explicit symbol for beginning execution of your
512 program, rather than the default entry point. If there is no symbol
513 named @var{entry}, the linker will try to parse @var{entry} as a number,
514 and use that as the entry address (the number will be interpreted in
515 base 10; you may use a leading @samp{0x} for base 16, or a leading
516 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
517 and other ways of specifying the entry point.
519 @kindex --exclude-libs
520 @item --exclude-libs @var{lib},@var{lib},...
521 Specifies a list of archive libraries from which symbols should not be automatically
522 exported. The library names may be delimited by commas or colons. Specifying
523 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
524 automatic export. This option is available only for the i386 PE targeted
525 port of the linker and for ELF targeted ports. For i386 PE, symbols
526 explicitly listed in a .def file are still exported, regardless of this
527 option. For ELF targeted ports, symbols affected by this option will
528 be treated as hidden.
530 @kindex --exclude-modules-for-implib
531 @item --exclude-modules-for-implib @var{module},@var{module},...
532 Specifies a list of object files or archive members, from which symbols
533 should not be automatically exported, but which should be copied wholesale
534 into the import library being generated during the link. The module names
535 may be delimited by commas or colons, and must match exactly the filenames
536 used by @command{ld} to open the files; for archive members, this is simply
537 the member name, but for object files the name listed must include and
538 match precisely any path used to specify the input file on the linker's
539 command-line. This option is available only for the i386 PE targeted port
540 of the linker. Symbols explicitly listed in a .def file are still exported,
541 regardless of this option.
543 @cindex dynamic symbol table
545 @kindex --export-dynamic
546 @kindex --no-export-dynamic
548 @itemx --export-dynamic
549 @itemx --no-export-dynamic
550 When creating a dynamically linked executable, using the @option{-E}
551 option or the @option{--export-dynamic} option causes the linker to add
552 all symbols to the dynamic symbol table. The dynamic symbol table is the
553 set of symbols which are visible from dynamic objects at run time.
555 If you do not use either of these options (or use the
556 @option{--no-export-dynamic} option to restore the default behavior), the
557 dynamic symbol table will normally contain only those symbols which are
558 referenced by some dynamic object mentioned in the link.
560 If you use @code{dlopen} to load a dynamic object which needs to refer
561 back to the symbols defined by the program, rather than some other
562 dynamic object, then you will probably need to use this option when
563 linking the program itself.
565 You can also use the dynamic list to control what symbols should
566 be added to the dynamic symbol table if the output format supports it.
567 See the description of @samp{--dynamic-list}.
569 Note that this option is specific to ELF targeted ports. PE targets
570 support a similar function to export all symbols from a DLL or EXE; see
571 the description of @samp{--export-all-symbols} below.
573 @kindex --export-dynamic-symbol=@var{glob}
574 @cindex export dynamic symbol
575 @item --export-dynamic-symbol=@var{glob}
576 When creating a dynamically linked executable, symbols matching
577 @var{glob} will be added to the dynamic symbol table. When creating a
578 shared library, references to symbols matching @var{glob} will not be
579 bound to the definitions within the shared library. This option is a
580 no-op when creating a shared library and @samp{-Bsymbolic} or
581 @samp{--dynamic-list} are not specified. This option is only meaningful
582 on ELF platforms which support shared libraries.
584 @kindex --export-dynamic-symbol-list=@var{file}
585 @cindex export dynamic symbol list
586 @item --export-dynamic-symbol-list=@var{file}
587 Specify a @samp{--export-dynamic-symbol} for each pattern in the file.
588 The format of the file is the same as the version node without
589 scope and node name. See @ref{VERSION} for more information.
591 @ifclear SingleFormat
592 @cindex big-endian objects
596 Link big-endian objects. This affects the default output format.
598 @cindex little-endian objects
601 Link little-endian objects. This affects the default output format.
604 @kindex -f @var{name}
605 @kindex --auxiliary=@var{name}
607 @itemx --auxiliary=@var{name}
608 When creating an ELF shared object, set the internal DT_AUXILIARY field
609 to the specified name. This tells the dynamic linker that the symbol
610 table of the shared object should be used as an auxiliary filter on the
611 symbol table of the shared object @var{name}.
613 If you later link a program against this filter object, then, when you
614 run the program, the dynamic linker will see the DT_AUXILIARY field. If
615 the dynamic linker resolves any symbols from the filter object, it will
616 first check whether there is a definition in the shared object
617 @var{name}. If there is one, it will be used instead of the definition
618 in the filter object. The shared object @var{name} need not exist.
619 Thus the shared object @var{name} may be used to provide an alternative
620 implementation of certain functions, perhaps for debugging or for
621 machine-specific performance.
623 This option may be specified more than once. The DT_AUXILIARY entries
624 will be created in the order in which they appear on the command line.
626 @kindex -F @var{name}
627 @kindex --filter=@var{name}
629 @itemx --filter=@var{name}
630 When creating an ELF shared object, set the internal DT_FILTER field to
631 the specified name. This tells the dynamic linker that the symbol table
632 of the shared object which is being created should be used as a filter
633 on the symbol table of the shared object @var{name}.
635 If you later link a program against this filter object, then, when you
636 run the program, the dynamic linker will see the DT_FILTER field. The
637 dynamic linker will resolve symbols according to the symbol table of the
638 filter object as usual, but it will actually link to the definitions
639 found in the shared object @var{name}. Thus the filter object can be
640 used to select a subset of the symbols provided by the object
643 Some older linkers used the @option{-F} option throughout a compilation
644 toolchain for specifying object-file format for both input and output
646 @ifclear SingleFormat
647 The @sc{gnu} linker uses other mechanisms for this purpose: the
648 @option{-b}, @option{--format}, @option{--oformat} options, the
649 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
650 environment variable.
652 The @sc{gnu} linker will ignore the @option{-F} option when not
653 creating an ELF shared object.
655 @cindex finalization function
656 @kindex -fini=@var{name}
657 @item -fini=@var{name}
658 When creating an ELF executable or shared object, call NAME when the
659 executable or shared object is unloaded, by setting DT_FINI to the
660 address of the function. By default, the linker uses @code{_fini} as
661 the function to call.
665 Ignored. Provided for compatibility with other tools.
667 @kindex -G @var{value}
668 @kindex --gpsize=@var{value}
671 @itemx --gpsize=@var{value}
672 Set the maximum size of objects to be optimized using the GP register to
673 @var{size}. This is only meaningful for object file formats such as
674 MIPS ELF that support putting large and small objects into different
675 sections. This is ignored for other object file formats.
677 @cindex runtime library name
678 @kindex -h @var{name}
679 @kindex -soname=@var{name}
681 @itemx -soname=@var{name}
682 When creating an ELF shared object, set the internal DT_SONAME field to
683 the specified name. When an executable is linked with a shared object
684 which has a DT_SONAME field, then when the executable is run the dynamic
685 linker will attempt to load the shared object specified by the DT_SONAME
686 field rather than using the file name given to the linker.
689 @cindex incremental link
691 Perform an incremental link (same as option @samp{-r}).
693 @cindex initialization function
694 @kindex -init=@var{name}
695 @item -init=@var{name}
696 When creating an ELF executable or shared object, call NAME when the
697 executable or shared object is loaded, by setting DT_INIT to the address
698 of the function. By default, the linker uses @code{_init} as the
701 @cindex archive files, from cmd line
702 @kindex -l @var{namespec}
703 @kindex --library=@var{namespec}
704 @item -l @var{namespec}
705 @itemx --library=@var{namespec}
706 Add the archive or object file specified by @var{namespec} to the
707 list of files to link. This option may be used any number of times.
708 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
709 will search the library path for a file called @var{filename}, otherwise it
710 will search the library path for a file called @file{lib@var{namespec}.a}.
712 On systems which support shared libraries, @command{ld} may also search for
713 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
714 and SunOS systems, @command{ld} will search a directory for a library
715 called @file{lib@var{namespec}.so} before searching for one called
716 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
717 indicates a shared library.) Note that this behavior does not apply
718 to @file{:@var{filename}}, which always specifies a file called
721 The linker will search an archive only once, at the location where it is
722 specified on the command line. If the archive defines a symbol which
723 was undefined in some object which appeared before the archive on the
724 command line, the linker will include the appropriate file(s) from the
725 archive. However, an undefined symbol in an object appearing later on
726 the command line will not cause the linker to search the archive again.
728 See the @option{-(} option for a way to force the linker to search
729 archives multiple times.
731 You may list the same archive multiple times on the command line.
734 This type of archive searching is standard for Unix linkers. However,
735 if you are using @command{ld} on AIX, note that it is different from the
736 behaviour of the AIX linker.
739 @cindex search directory, from cmd line
741 @kindex --library-path=@var{dir}
742 @item -L @var{searchdir}
743 @itemx --library-path=@var{searchdir}
744 Add path @var{searchdir} to the list of paths that @command{ld} will search
745 for archive libraries and @command{ld} control scripts. You may use this
746 option any number of times. The directories are searched in the order
747 in which they are specified on the command line. Directories specified
748 on the command line are searched before the default directories. All
749 @option{-L} options apply to all @option{-l} options, regardless of the
750 order in which the options appear. @option{-L} options do not affect
751 how @command{ld} searches for a linker script unless @option{-T}
754 If @var{searchdir} begins with @code{=} or @code{$SYSROOT}, then this
755 prefix will be replaced by the @dfn{sysroot prefix}, controlled by the
756 @samp{--sysroot} option, or specified when the linker is configured.
759 The default set of paths searched (without being specified with
760 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
761 some cases also on how it was configured. @xref{Environment}.
764 The paths can also be specified in a link script with the
765 @code{SEARCH_DIR} command. Directories specified this way are searched
766 at the point in which the linker script appears in the command line.
769 @kindex -m @var{emulation}
770 @item -m @var{emulation}
771 Emulate the @var{emulation} linker. You can list the available
772 emulations with the @samp{--verbose} or @samp{-V} options.
774 If the @samp{-m} option is not used, the emulation is taken from the
775 @code{LDEMULATION} environment variable, if that is defined.
777 Otherwise, the default emulation depends upon how the linker was
785 Print a link map to the standard output. A link map provides
786 information about the link, including the following:
790 Where object files are mapped into memory.
792 How common symbols are allocated.
794 All archive members included in the link, with a mention of the symbol
795 which caused the archive member to be brought in.
797 The values assigned to symbols.
799 Note - symbols whose values are computed by an expression which
800 involves a reference to a previous value of the same symbol may not
801 have correct result displayed in the link map. This is because the
802 linker discards intermediate results and only retains the final value
803 of an expression. Under such circumstances the linker will display
804 the final value enclosed by square brackets. Thus for example a
805 linker script containing:
813 will produce the following output in the link map if the @option{-M}
818 [0x0000000c] foo = (foo * 0x4)
819 [0x0000000c] foo = (foo + 0x8)
822 See @ref{Expressions} for more information about expressions in linker
826 How GNU properties are merged.
828 When the linker merges input .note.gnu.property sections into one output
829 .note.gnu.property section, some properties are removed or updated.
830 These actions are reported in the link map. For example:
833 Removed property 0xc0000002 to merge foo.o (0x1) and bar.o (not found)
836 This indicates that property 0xc0000002 is removed from output when
837 merging properties in @file{foo.o}, whose property 0xc0000002 value
838 is 0x1, and @file{bar.o}, which doesn't have property 0xc0000002.
841 Updated property 0xc0010001 (0x1) to merge foo.o (0x1) and bar.o (0x1)
844 This indicates that property 0xc0010001 value is updated to 0x1 in output
845 when merging properties in @file{foo.o}, whose 0xc0010001 property value
846 is 0x1, and @file{bar.o}, whose 0xc0010001 property value is 0x1.
849 @cindex link map discarded
850 @kindex --print-map-discarded
851 @kindex --no-print-map-discarded
852 @item --print-map-discarded
853 @itemx --no-print-map-discarded
854 Print (or do not print) the list of discarded and garbage collected sections
855 in the link map. Enabled by default.
858 @cindex read-only text
863 Turn off page alignment of sections, and disable linking against shared
864 libraries. If the output format supports Unix style magic numbers,
865 mark the output as @code{NMAGIC}.
869 @cindex read/write from cmd line
873 Set the text and data sections to be readable and writable. Also, do
874 not page-align the data segment, and disable linking against shared
875 libraries. If the output format supports Unix style magic numbers,
876 mark the output as @code{OMAGIC}. Note: Although a writable text section
877 is allowed for PE-COFF targets, it does not conform to the format
878 specification published by Microsoft.
883 This option negates most of the effects of the @option{-N} option. It
884 sets the text section to be read-only, and forces the data segment to
885 be page-aligned. Note - this option does not enable linking against
886 shared libraries. Use @option{-Bdynamic} for this.
888 @kindex -o @var{output}
889 @kindex --output=@var{output}
890 @cindex naming the output file
891 @item -o @var{output}
892 @itemx --output=@var{output}
893 Use @var{output} as the name for the program produced by @command{ld}; if this
894 option is not specified, the name @file{a.out} is used by default. The
895 script command @code{OUTPUT} can also specify the output file name.
897 @kindex --dependency-file=@var{depfile}
898 @cindex dependency file
899 @item --dependency-file=@var{depfile}
900 Write a @dfn{dependency file} to @var{depfile}. This file contains a rule
901 suitable for @code{make} describing the output file and all the input files
902 that were read to produce it. The output is similar to the compiler's
903 output with @samp{-M -MP} (@pxref{Preprocessor Options,, Options
904 Controlling the Preprocessor, gcc.info, Using the GNU Compiler
905 Collection}). Note that there is no option like the compiler's @samp{-MM},
906 to exclude ``system files'' (which is not a well-specified concept in the
907 linker, unlike ``system headers'' in the compiler). So the output from
908 @samp{--dependency-file} is always specific to the exact state of the
909 installation where it was produced, and should not be copied into
910 distributed makefiles without careful editing.
912 @kindex -O @var{level}
913 @cindex generating optimized output
915 If @var{level} is a numeric values greater than zero @command{ld} optimizes
916 the output. This might take significantly longer and therefore probably
917 should only be enabled for the final binary. At the moment this
918 option only affects ELF shared library generation. Future releases of
919 the linker may make more use of this option. Also currently there is
920 no difference in the linker's behaviour for different non-zero values
921 of this option. Again this may change with future releases.
923 @kindex -plugin @var{name}
924 @item -plugin @var{name}
925 Involve a plugin in the linking process. The @var{name} parameter is
926 the absolute filename of the plugin. Usually this parameter is
927 automatically added by the complier, when using link time
928 optimization, but users can also add their own plugins if they so
931 Note that the location of the compiler originated plugins is different
932 from the place where the @command{ar}, @command{nm} and
933 @command{ranlib} programs search for their plugins. In order for
934 those commands to make use of a compiler based plugin it must first be
935 copied into the @file{$@{libdir@}/bfd-plugins} directory. All gcc
936 based linker plugins are backward compatible, so it is sufficient to
937 just copy in the newest one.
940 @cindex push state governing input file handling
942 The @option{--push-state} allows one to preserve the current state of the
943 flags which govern the input file handling so that they can all be
944 restored with one corresponding @option{--pop-state} option.
946 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
947 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
948 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
949 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
950 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
951 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
953 One target for this option are specifications for @file{pkg-config}. When
954 used with the @option{--libs} option all possibly needed libraries are
955 listed and then possibly linked with all the time. It is better to return
956 something as follows:
959 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
963 @cindex pop state governing input file handling
965 Undoes the effect of --push-state, restores the previous values of the
966 flags governing input file handling.
969 @kindex --emit-relocs
970 @cindex retain relocations in final executable
973 Leave relocation sections and contents in fully linked executables.
974 Post link analysis and optimization tools may need this information in
975 order to perform correct modifications of executables. This results
976 in larger executables.
978 This option is currently only supported on ELF platforms.
980 @kindex --force-dynamic
981 @cindex forcing the creation of dynamic sections
982 @item --force-dynamic
983 Force the output file to have dynamic sections. This option is specific
987 @cindex relocatable output
989 @kindex --relocatable
992 Generate relocatable output---i.e., generate an output file that can in
993 turn serve as input to @command{ld}. This is often called @dfn{partial
994 linking}. As a side effect, in environments that support standard Unix
995 magic numbers, this option also sets the output file's magic number to
997 @c ; see @option{-N}.
998 If this option is not specified, an absolute file is produced. When
999 linking C++ programs, this option @emph{will not} resolve references to
1000 constructors; to do that, use @samp{-Ur}.
1002 When an input file does not have the same format as the output file,
1003 partial linking is only supported if that input file does not contain any
1004 relocations. Different output formats can have further restrictions; for
1005 example some @code{a.out}-based formats do not support partial linking
1006 with input files in other formats at all.
1008 This option does the same thing as @samp{-i}.
1010 @kindex -R @var{file}
1011 @kindex --just-symbols=@var{file}
1012 @cindex symbol-only input
1013 @item -R @var{filename}
1014 @itemx --just-symbols=@var{filename}
1015 Read symbol names and their addresses from @var{filename}, but do not
1016 relocate it or include it in the output. This allows your output file
1017 to refer symbolically to absolute locations of memory defined in other
1018 programs. You may use this option more than once.
1020 For compatibility with other ELF linkers, if the @option{-R} option is
1021 followed by a directory name, rather than a file name, it is treated as
1022 the @option{-rpath} option.
1026 @cindex strip all symbols
1029 Omit all symbol information from the output file.
1032 @kindex --strip-debug
1033 @cindex strip debugger symbols
1035 @itemx --strip-debug
1036 Omit debugger symbol information (but not all symbols) from the output file.
1038 @kindex --strip-discarded
1039 @kindex --no-strip-discarded
1040 @item --strip-discarded
1041 @itemx --no-strip-discarded
1042 Omit (or do not omit) global symbols defined in discarded sections.
1047 @cindex input files, displaying
1050 Print the names of the input files as @command{ld} processes them. If
1051 @samp{-t} is given twice then members within archives are also printed.
1052 @samp{-t} output is useful to generate a list of all the object files
1053 and scripts involved in linking, for example, when packaging files for
1054 a linker bug report.
1056 @kindex -T @var{script}
1057 @kindex --script=@var{script}
1058 @cindex script files
1059 @item -T @var{scriptfile}
1060 @itemx --script=@var{scriptfile}
1061 Use @var{scriptfile} as the linker script. This script replaces
1062 @command{ld}'s default linker script (rather than adding to it), so
1063 @var{commandfile} must specify everything necessary to describe the
1064 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
1065 the current directory, @code{ld} looks for it in the directories
1066 specified by any preceding @samp{-L} options. Multiple @samp{-T}
1069 @kindex -dT @var{script}
1070 @kindex --default-script=@var{script}
1071 @cindex script files
1072 @item -dT @var{scriptfile}
1073 @itemx --default-script=@var{scriptfile}
1074 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
1076 This option is similar to the @option{--script} option except that
1077 processing of the script is delayed until after the rest of the
1078 command line has been processed. This allows options placed after the
1079 @option{--default-script} option on the command line to affect the
1080 behaviour of the linker script, which can be important when the linker
1081 command line cannot be directly controlled by the user. (eg because
1082 the command line is being constructed by another tool, such as
1085 @kindex -u @var{symbol}
1086 @kindex --undefined=@var{symbol}
1087 @cindex undefined symbol
1088 @item -u @var{symbol}
1089 @itemx --undefined=@var{symbol}
1090 Force @var{symbol} to be entered in the output file as an undefined
1091 symbol. Doing this may, for example, trigger linking of additional
1092 modules from standard libraries. @samp{-u} may be repeated with
1093 different option arguments to enter additional undefined symbols. This
1094 option is equivalent to the @code{EXTERN} linker script command.
1096 If this option is being used to force additional modules to be pulled
1097 into the link, and if it is an error for the symbol to remain
1098 undefined, then the option @option{--require-defined} should be used
1101 @kindex --require-defined=@var{symbol}
1102 @cindex symbols, require defined
1103 @cindex defined symbol
1104 @item --require-defined=@var{symbol}
1105 Require that @var{symbol} is defined in the output file. This option
1106 is the same as option @option{--undefined} except that if @var{symbol}
1107 is not defined in the output file then the linker will issue an error
1108 and exit. The same effect can be achieved in a linker script by using
1109 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
1110 can be used multiple times to require additional symbols.
1113 @cindex constructors
1115 For anything other than C++ programs, this option is equivalent to
1116 @samp{-r}: it generates relocatable output---i.e., an output file that can in
1117 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
1118 @emph{does} resolve references to constructors, unlike @samp{-r}.
1119 It does not work to use @samp{-Ur} on files that were themselves linked
1120 with @samp{-Ur}; once the constructor table has been built, it cannot
1121 be added to. Use @samp{-Ur} only for the last partial link, and
1122 @samp{-r} for the others.
1124 @kindex --orphan-handling=@var{MODE}
1125 @cindex orphan sections
1126 @cindex sections, orphan
1127 @item --orphan-handling=@var{MODE}
1128 Control how orphan sections are handled. An orphan section is one not
1129 specifically mentioned in a linker script. @xref{Orphan Sections}.
1131 @var{MODE} can have any of the following values:
1135 Orphan sections are placed into a suitable output section following
1136 the strategy described in @ref{Orphan Sections}. The option
1137 @samp{--unique} also affects how sections are placed.
1140 All orphan sections are discarded, by placing them in the
1141 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1144 The linker will place the orphan section as for @code{place} and also
1148 The linker will exit with an error if any orphan section is found.
1151 The default if @samp{--orphan-handling} is not given is @code{place}.
1153 @kindex --unique[=@var{SECTION}]
1154 @item --unique[=@var{SECTION}]
1155 Creates a separate output section for every input section matching
1156 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1157 missing, for every orphan input section. An orphan section is one not
1158 specifically mentioned in a linker script. You may use this option
1159 multiple times on the command line; It prevents the normal merging of
1160 input sections with the same name, overriding output section assignments
1170 Display the version number for @command{ld}. The @option{-V} option also
1171 lists the supported emulations.
1174 @kindex --discard-all
1175 @cindex deleting local symbols
1177 @itemx --discard-all
1178 Delete all local symbols.
1181 @kindex --discard-locals
1182 @cindex local symbols, deleting
1184 @itemx --discard-locals
1185 Delete all temporary local symbols. (These symbols start with
1186 system-specific local label prefixes, typically @samp{.L} for ELF systems
1187 or @samp{L} for traditional a.out systems.)
1189 @kindex -y @var{symbol}
1190 @kindex --trace-symbol=@var{symbol}
1191 @cindex symbol tracing
1192 @item -y @var{symbol}
1193 @itemx --trace-symbol=@var{symbol}
1194 Print the name of each linked file in which @var{symbol} appears. This
1195 option may be given any number of times. On many systems it is necessary
1196 to prepend an underscore.
1198 This option is useful when you have an undefined symbol in your link but
1199 don't know where the reference is coming from.
1201 @kindex -Y @var{path}
1203 Add @var{path} to the default library search path. This option exists
1204 for Solaris compatibility.
1206 @kindex -z @var{keyword}
1207 @item -z @var{keyword}
1208 The recognized keywords are:
1212 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1214 @item call-nop=prefix-addr
1215 @itemx call-nop=suffix-nop
1216 @itemx call-nop=prefix-@var{byte}
1217 @itemx call-nop=suffix-@var{byte}
1218 Specify the 1-byte @code{NOP} padding when transforming indirect call
1219 to a locally defined function, foo, via its GOT slot.
1220 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1221 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1222 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1223 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1224 Supported for i386 and x86_64.
1226 @item cet-report=none
1227 @itemx cet-report=warning
1228 @itemx cet-report=error
1229 Specify how to report the missing GNU_PROPERTY_X86_FEATURE_1_IBT and
1230 GNU_PROPERTY_X86_FEATURE_1_SHSTK properties in input .note.gnu.property
1231 section. @option{cet-report=none}, which is the default, will make the
1232 linker not report missing properties in input files.
1233 @option{cet-report=warning} will make the linker issue a warning for
1234 missing properties in input files. @option{cet-report=error} will make
1235 the linker issue an error for missing properties in input files.
1236 Note that @option{ibt} will turn off the missing
1237 GNU_PROPERTY_X86_FEATURE_1_IBT property report and @option{shstk} will
1238 turn off the missing GNU_PROPERTY_X86_FEATURE_1_SHSTK property report.
1239 Supported for Linux/i386 and Linux/x86_64.
1243 Combine multiple dynamic relocation sections and sort to improve
1244 dynamic symbol lookup caching. Do not do this if @samp{nocombreloc}.
1248 Generate common symbols with STT_COMMON type during a relocatable
1249 link. Use STT_OBJECT type if @samp{nocommon}.
1251 @item common-page-size=@var{value}
1252 Set the page size most commonly used to @var{value}. Memory image
1253 layout will be optimized to minimize memory pages if the system is
1254 using pages of this size.
1257 Report unresolved symbol references from regular object files. This
1258 is done even if the linker is creating a non-symbolic shared library.
1259 This option is the inverse of @samp{-z undefs}.
1261 @item dynamic-undefined-weak
1262 @itemx nodynamic-undefined-weak
1263 Make undefined weak symbols dynamic when building a dynamic object,
1264 if they are referenced from a regular object file and not forced local
1265 by symbol visibility or versioning. Do not make them dynamic if
1266 @samp{nodynamic-undefined-weak}. If neither option is given, a target
1267 may default to either option being in force, or make some other
1268 selection of undefined weak symbols dynamic. Not all targets support
1272 Marks the object as requiring executable stack.
1275 This option is only meaningful when building a shared object. It makes
1276 the symbols defined by this shared object available for symbol resolution
1277 of subsequently loaded libraries.
1280 This option is only meaningful when building a dynamic executable.
1281 This option marks the executable as requiring global auditing by
1282 setting the @code{DF_1_GLOBAUDIT} bit in the @code{DT_FLAGS_1} dynamic
1283 tag. Global auditing requires that any auditing library defined via
1284 the @option{--depaudit} or @option{-P} command-line options be run for
1285 all dynamic objects loaded by the application.
1288 Generate Intel Indirect Branch Tracking (IBT) enabled PLT entries.
1289 Supported for Linux/i386 and Linux/x86_64.
1292 Generate GNU_PROPERTY_X86_FEATURE_1_IBT in .note.gnu.property section
1293 to indicate compatibility with IBT. This also implies @option{ibtplt}.
1294 Supported for Linux/i386 and Linux/x86_64.
1296 @item indirect-extern-access
1297 @itemx noindirect-extern-access
1298 Generate GNU_PROPERTY_1_NEEDED_INDIRECT_EXTERN_ACCESS in
1299 .note.gnu.property section to indicate that object file requires
1300 canonical function pointers and cannot be used with copy relocation.
1301 This option also implies @option{noextern-protected-data} and
1302 @option{nocopyreloc}. Supported for i386 and x86-64.
1304 @option{noindirect-extern-access} removes
1305 GNU_PROPERTY_1_NEEDED_INDIRECT_EXTERN_ACCESS from .note.gnu.property
1309 This option is only meaningful when building a shared object.
1310 It marks the object so that its runtime initialization will occur
1311 before the runtime initialization of any other objects brought into
1312 the process at the same time. Similarly the runtime finalization of
1313 the object will occur after the runtime finalization of any other
1317 Specify that the dynamic loader should modify its symbol search order
1318 so that symbols in this shared library interpose all other shared
1319 libraries not so marked.
1323 When generating a shared library or other dynamically loadable ELF
1324 object mark it as one that should (by default) only ever be loaded once,
1325 and only in the main namespace (when using @code{dlmopen}). This is
1326 primarily used to mark fundamental libraries such as libc, libpthread et
1327 al which do not usually function correctly unless they are the sole instances
1328 of themselves. This behaviour can be overridden by the @code{dlmopen} caller
1329 and does not apply to certain loading mechanisms (such as audit libraries).
1332 Generate GNU_PROPERTY_X86_FEATURE_1_LAM_U48 in .note.gnu.property section
1333 to indicate compatibility with Intel LAM_U48. Supported for Linux/x86_64.
1336 Generate GNU_PROPERTY_X86_FEATURE_1_LAM_U57 in .note.gnu.property section
1337 to indicate compatibility with Intel LAM_U57. Supported for Linux/x86_64.
1339 @item lam-u48-report=none
1340 @itemx lam-u48-report=warning
1341 @itemx lam-u48-report=error
1342 Specify how to report the missing GNU_PROPERTY_X86_FEATURE_1_LAM_U48
1343 property in input .note.gnu.property section.
1344 @option{lam-u48-report=none}, which is the default, will make the
1345 linker not report missing properties in input files.
1346 @option{lam-u48-report=warning} will make the linker issue a warning for
1347 missing properties in input files. @option{lam-u48-report=error} will
1348 make the linker issue an error for missing properties in input files.
1349 Supported for Linux/x86_64.
1351 @item lam-u57-report=none
1352 @itemx lam-u57-report=warning
1353 @itemx lam-u57-report=error
1354 Specify how to report the missing GNU_PROPERTY_X86_FEATURE_1_LAM_U57
1355 property in input .note.gnu.property section.
1356 @option{lam-u57-report=none}, which is the default, will make the
1357 linker not report missing properties in input files.
1358 @option{lam-u57-report=warning} will make the linker issue a warning for
1359 missing properties in input files. @option{lam-u57-report=error} will
1360 make the linker issue an error for missing properties in input files.
1361 Supported for Linux/x86_64.
1363 @item lam-report=none
1364 @itemx lam-report=warning
1365 @itemx lam-report=error
1366 Specify how to report the missing GNU_PROPERTY_X86_FEATURE_1_LAM_U48 and
1367 GNU_PROPERTY_X86_FEATURE_1_LAM_U57 properties in input .note.gnu.property
1368 section. @option{lam-report=none}, which is the default, will make the
1369 linker not report missing properties in input files.
1370 @option{lam-report=warning} will make the linker issue a warning for
1371 missing properties in input files. @option{lam-report=error} will make
1372 the linker issue an error for missing properties in input files.
1373 Supported for Linux/x86_64.
1376 When generating an executable or shared library, mark it to tell the
1377 dynamic linker to defer function call resolution to the point when
1378 the function is called (lazy binding), rather than at load time.
1379 Lazy binding is the default.
1382 Specify that the object's filters be processed immediately at runtime.
1384 @item max-page-size=@var{value}
1385 Set the maximum memory page size supported to @var{value}.
1388 Allow multiple definitions.
1391 Disable linker generated .dynbss variables used in place of variables
1392 defined in shared libraries. May result in dynamic text relocations.
1395 Specify that the dynamic loader search for dependencies of this object
1396 should ignore any default library search paths.
1399 Specify that the object shouldn't be unloaded at runtime.
1402 Specify that the object is not available to @code{dlopen}.
1405 Specify that the object can not be dumped by @code{dldump}.
1408 Marks the object as not requiring executable stack.
1410 @item noextern-protected-data
1411 Don't treat protected data symbols as external when building a shared
1412 library. This option overrides the linker backend default. It can be
1413 used to work around incorrect relocations against protected data symbols
1414 generated by compiler. Updates on protected data symbols by another
1415 module aren't visible to the resulting shared library. Supported for
1418 @item noreloc-overflow
1419 Disable relocation overflow check. This can be used to disable
1420 relocation overflow check if there will be no dynamic relocation
1421 overflow at run-time. Supported for x86_64.
1424 When generating an executable or shared library, mark it to tell the
1425 dynamic linker to resolve all symbols when the program is started, or
1426 when the shared library is loaded by dlopen, instead of deferring
1427 function call resolution to the point when the function is first
1431 Specify that the object requires @samp{$ORIGIN} handling in paths.
1433 @item pack-relative-relocs
1434 @itemx nopack-relative-relocs
1435 Generate compact relative relocation in position-independent executable
1436 and shared library. It adds @code{DT_RELR}, @code{DT_RELRSZ} and
1437 @code{DT_RELRENT} entries to the dynamic section. It is ignored when
1438 building position-dependent executable and relocatable output.
1439 @option{nopack-relative-relocs} is the default, which disables compact
1440 relative relocation. When linked against the GNU C Library, a
1441 GLIBC_ABI_DT_RELR symbol version dependency on the shared C Library is
1442 added to the output. Supported for i386 and x86-64.
1446 Create an ELF @code{PT_GNU_RELRO} segment header in the object. This
1447 specifies a memory segment that should be made read-only after
1448 relocation, if supported. Specifying @samp{common-page-size} smaller
1449 than the system page size will render this protection ineffective.
1450 Don't create an ELF @code{PT_GNU_RELRO} segment if @samp{norelro}.
1452 @item report-relative-reloc
1453 Report dynamic relative relocations generated by linker. Supported for
1454 Linux/i386 and Linux/x86_64.
1457 @itemx noseparate-code
1458 Create separate code @code{PT_LOAD} segment header in the object. This
1459 specifies a memory segment that should contain only instructions and must
1460 be in wholly disjoint pages from any other data. Don't create separate
1461 code @code{PT_LOAD} segment if @samp{noseparate-code} is used.
1464 Generate GNU_PROPERTY_X86_FEATURE_1_SHSTK in .note.gnu.property section
1465 to indicate compatibility with Intel Shadow Stack. Supported for
1466 Linux/i386 and Linux/x86_64.
1468 @item stack-size=@var{value}
1469 Specify a stack size for an ELF @code{PT_GNU_STACK} segment.
1470 Specifying zero will override any default non-zero sized
1471 @code{PT_GNU_STACK} segment creation.
1474 @itemx nostart-stop-gc
1475 @cindex start-stop-gc
1476 When @samp{--gc-sections} is in effect, a reference from a retained
1477 section to @code{__start_SECNAME} or @code{__stop_SECNAME} causes all
1478 input sections named @code{SECNAME} to also be retained, if
1479 @code{SECNAME} is representable as a C identifier and either
1480 @code{__start_SECNAME} or @code{__stop_SECNAME} is synthesized by the
1481 linker. @samp{-z start-stop-gc} disables this effect, allowing
1482 sections to be garbage collected as if the special synthesized symbols
1483 were not defined. @samp{-z start-stop-gc} has no effect on a
1484 definition of @code{__start_SECNAME} or @code{__stop_SECNAME} in an
1485 object file or linker script. Such a definition will prevent the
1486 linker providing a synthesized @code{__start_SECNAME} or
1487 @code{__stop_SECNAME} respectively, and therefore the special
1488 treatment by garbage collection for those references.
1490 @item start-stop-visibility=@var{value}
1492 @cindex ELF symbol visibility
1493 Specify the ELF symbol visibility for synthesized
1494 @code{__start_SECNAME} and @code{__stop_SECNAME} symbols (@pxref{Input
1495 Section Example}). @var{value} must be exactly @samp{default},
1496 @samp{internal}, @samp{hidden}, or @samp{protected}. If no @samp{-z
1497 start-stop-visibility} option is given, @samp{protected} is used for
1498 compatibility with historical practice. However, it's highly
1499 recommended to use @samp{-z start-stop-visibility=hidden} in new
1500 programs and shared libraries so that these symbols are not exported
1501 between shared objects, which is not usually what's intended.
1506 Report an error if DT_TEXTREL is set, i.e., if the position-independent
1507 or shared object has dynamic relocations in read-only sections. Don't
1508 report an error if @samp{notext} or @samp{textoff}.
1511 Do not report unresolved symbol references from regular object files,
1512 either when creating an executable, or when creating a shared library.
1513 This option is the inverse of @samp{-z defs}.
1516 @itemx nounique-symbol
1517 Avoid duplicated local symbol names in the symbol string table. Append
1518 ".@code{number}" to duplicated local symbol names if @samp{unique-symbol}
1519 is used. @option{nounique-symbol} is the default.
1521 @item x86-64-baseline
1525 Specify the x86-64 ISA level needed in .note.gnu.property section.
1526 @option{x86-64-baseline} generates @code{GNU_PROPERTY_X86_ISA_1_BASELINE}.
1527 @option{x86-64-v2} generates @code{GNU_PROPERTY_X86_ISA_1_V2}.
1528 @option{x86-64-v3} generates @code{GNU_PROPERTY_X86_ISA_1_V3}.
1529 @option{x86-64-v4} generates @code{GNU_PROPERTY_X86_ISA_1_V4}.
1530 Supported for Linux/i386 and Linux/x86_64.
1534 Other keywords are ignored for Solaris compatibility.
1537 @cindex groups of archives
1538 @item -( @var{archives} -)
1539 @itemx --start-group @var{archives} --end-group
1540 The @var{archives} should be a list of archive files. They may be
1541 either explicit file names, or @samp{-l} options.
1543 The specified archives are searched repeatedly until no new undefined
1544 references are created. Normally, an archive is searched only once in
1545 the order that it is specified on the command line. If a symbol in that
1546 archive is needed to resolve an undefined symbol referred to by an
1547 object in an archive that appears later on the command line, the linker
1548 would not be able to resolve that reference. By grouping the archives,
1549 they will all be searched repeatedly until all possible references are
1552 Using this option has a significant performance cost. It is best to use
1553 it only when there are unavoidable circular references between two or
1556 @kindex --accept-unknown-input-arch
1557 @kindex --no-accept-unknown-input-arch
1558 @item --accept-unknown-input-arch
1559 @itemx --no-accept-unknown-input-arch
1560 Tells the linker to accept input files whose architecture cannot be
1561 recognised. The assumption is that the user knows what they are doing
1562 and deliberately wants to link in these unknown input files. This was
1563 the default behaviour of the linker, before release 2.14. The default
1564 behaviour from release 2.14 onwards is to reject such input files, and
1565 so the @samp{--accept-unknown-input-arch} option has been added to
1566 restore the old behaviour.
1569 @kindex --no-as-needed
1571 @itemx --no-as-needed
1572 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1573 on the command line after the @option{--as-needed} option. Normally
1574 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1575 on the command line, regardless of whether the library is actually
1576 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1577 emitted for a library that @emph{at that point in the link} satisfies a
1578 non-weak undefined symbol reference from a regular object file or, if
1579 the library is not found in the DT_NEEDED lists of other needed libraries, a
1580 non-weak undefined symbol reference from another needed dynamic library.
1581 Object files or libraries appearing on the command line @emph{after}
1582 the library in question do not affect whether the library is seen as
1583 needed. This is similar to the rules for extraction of object files
1584 from archives. @option{--no-as-needed} restores the default behaviour.
1586 Note: On Linux based systems the @option{--as-needed} option also has
1587 an affect on the behaviour of the @option{--rpath} and
1588 @option{--rpath-link} options. See the description of
1589 @option{--rpath-link} for more details.
1591 @kindex --add-needed
1592 @kindex --no-add-needed
1594 @itemx --no-add-needed
1595 These two options have been deprecated because of the similarity of
1596 their names to the @option{--as-needed} and @option{--no-as-needed}
1597 options. They have been replaced by @option{--copy-dt-needed-entries}
1598 and @option{--no-copy-dt-needed-entries}.
1600 @kindex -assert @var{keyword}
1601 @item -assert @var{keyword}
1602 This option is ignored for SunOS compatibility.
1606 @kindex -call_shared
1610 Link against dynamic libraries. This is only meaningful on platforms
1611 for which shared libraries are supported. This option is normally the
1612 default on such platforms. The different variants of this option are
1613 for compatibility with various systems. You may use this option
1614 multiple times on the command line: it affects library searching for
1615 @option{-l} options which follow it.
1619 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1620 section. This causes the runtime linker to handle lookups in this
1621 object and its dependencies to be performed only inside the group.
1622 @option{--unresolved-symbols=report-all} is implied. This option is
1623 only meaningful on ELF platforms which support shared libraries.
1633 Do not link against shared libraries. This is only meaningful on
1634 platforms for which shared libraries are supported. The different
1635 variants of this option are for compatibility with various systems. You
1636 may use this option multiple times on the command line: it affects
1637 library searching for @option{-l} options which follow it. This
1638 option also implies @option{--unresolved-symbols=report-all}. This
1639 option can be used with @option{-shared}. Doing so means that a
1640 shared library is being created but that all of the library's external
1641 references must be resolved by pulling in entries from static
1646 When creating a shared library, bind references to global symbols to the
1647 definition within the shared library, if any. Normally, it is possible
1648 for a program linked against a shared library to override the definition
1649 within the shared library. This option is only meaningful on ELF
1650 platforms which support shared libraries.
1652 @kindex -Bsymbolic-functions
1653 @item -Bsymbolic-functions
1654 When creating a shared library, bind references to global function
1655 symbols to the definition within the shared library, if any.
1656 This option is only meaningful on ELF platforms which support shared
1659 @kindex -Bno-symbolic
1661 This option can cancel previously specified @samp{-Bsymbolic} and
1662 @samp{-Bsymbolic-functions}.
1664 @kindex --dynamic-list=@var{dynamic-list-file}
1665 @item --dynamic-list=@var{dynamic-list-file}
1666 Specify the name of a dynamic list file to the linker. This is
1667 typically used when creating shared libraries to specify a list of
1668 global symbols whose references shouldn't be bound to the definition
1669 within the shared library, or creating dynamically linked executables
1670 to specify a list of symbols which should be added to the symbol table
1671 in the executable. This option is only meaningful on ELF platforms
1672 which support shared libraries.
1674 The format of the dynamic list is the same as the version node without
1675 scope and node name. See @ref{VERSION} for more information.
1677 @kindex --dynamic-list-data
1678 @item --dynamic-list-data
1679 Include all global data symbols to the dynamic list.
1681 @kindex --dynamic-list-cpp-new
1682 @item --dynamic-list-cpp-new
1683 Provide the builtin dynamic list for C++ operator new and delete. It
1684 is mainly useful for building shared libstdc++.
1686 @kindex --dynamic-list-cpp-typeinfo
1687 @item --dynamic-list-cpp-typeinfo
1688 Provide the builtin dynamic list for C++ runtime type identification.
1690 @kindex --check-sections
1691 @kindex --no-check-sections
1692 @item --check-sections
1693 @itemx --no-check-sections
1694 Asks the linker @emph{not} to check section addresses after they have
1695 been assigned to see if there are any overlaps. Normally the linker will
1696 perform this check, and if it finds any overlaps it will produce
1697 suitable error messages. The linker does know about, and does make
1698 allowances for sections in overlays. The default behaviour can be
1699 restored by using the command-line switch @option{--check-sections}.
1700 Section overlap is not usually checked for relocatable links. You can
1701 force checking in that case by using the @option{--check-sections}
1704 @kindex --copy-dt-needed-entries
1705 @kindex --no-copy-dt-needed-entries
1706 @item --copy-dt-needed-entries
1707 @itemx --no-copy-dt-needed-entries
1708 This option affects the treatment of dynamic libraries referred to
1709 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1710 command line. Normally the linker won't add a DT_NEEDED tag to the
1711 output binary for each library mentioned in a DT_NEEDED tag in an
1712 input dynamic library. With @option{--copy-dt-needed-entries}
1713 specified on the command line however any dynamic libraries that
1714 follow it will have their DT_NEEDED entries added. The default
1715 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1717 This option also has an effect on the resolution of symbols in dynamic
1718 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1719 mentioned on the command line will be recursively searched, following
1720 their DT_NEEDED tags to other libraries, in order to resolve symbols
1721 required by the output binary. With the default setting however
1722 the searching of dynamic libraries that follow it will stop with the
1723 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1726 @cindex cross reference table
1729 Output a cross reference table. If a linker map file is being
1730 generated, the cross reference table is printed to the map file.
1731 Otherwise, it is printed on the standard output.
1733 The format of the table is intentionally simple, so that it may be
1734 easily processed by a script if necessary. The symbols are printed out,
1735 sorted by name. For each symbol, a list of file names is given. If the
1736 symbol is defined, the first file listed is the location of the
1737 definition. If the symbol is defined as a common value then any files
1738 where this happens appear next. Finally any files that reference the
1741 @cindex ctf variables
1742 @kindex --ctf-variables
1743 @kindex --no-ctf-variables
1744 @item --ctf-variables
1745 @item --no-ctf-variables
1746 The CTF debuginfo format supports a section which encodes the names and
1747 types of variables found in the program which do not appear in any symbol
1748 table. These variables clearly cannot be looked up by address by
1749 conventional debuggers, so the space used for their types and names is
1750 usually wasted: the types are usually small but the names are often not.
1751 @option{--ctf-variables} causes the generation of such a section.
1752 The default behaviour can be restored with @option{--no-ctf-variables}.
1754 @cindex ctf type sharing
1755 @kindex --ctf-share-types
1756 @item --ctf-share-types=@var{method}
1757 Adjust the method used to share types between translation units in CTF.
1760 @item share-unconflicted
1761 Put all types that do not have ambiguous definitions into the shared dictionary,
1762 where debuggers can easily access them, even if they only occur in one
1763 translation unit. This is the default.
1765 @item share-duplicated
1766 Put only types that occur in multiple translation units into the shared
1767 dictionary: types with only one definition go into per-translation-unit
1768 dictionaries. Types with ambiguous definitions in multiple translation units
1769 always go into per-translation-unit dictionaries. This tends to make the CTF
1770 larger, but may reduce the amount of CTF in the shared dictionary. For very
1771 large projects this may speed up opening the CTF and save memory in the CTF
1772 consumer at runtime.
1775 @cindex common allocation
1776 @kindex --no-define-common
1777 @item --no-define-common
1778 This option inhibits the assignment of addresses to common symbols.
1779 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1780 @xref{Miscellaneous Commands}.
1782 The @samp{--no-define-common} option allows decoupling
1783 the decision to assign addresses to Common symbols from the choice
1784 of the output file type; otherwise a non-Relocatable output type
1785 forces assigning addresses to Common symbols.
1786 Using @samp{--no-define-common} allows Common symbols that are referenced
1787 from a shared library to be assigned addresses only in the main program.
1788 This eliminates the unused duplicate space in the shared library,
1789 and also prevents any possible confusion over resolving to the wrong
1790 duplicate when there are many dynamic modules with specialized search
1791 paths for runtime symbol resolution.
1793 @cindex group allocation in linker script
1794 @cindex section groups
1796 @kindex --force-group-allocation
1797 @item --force-group-allocation
1798 This option causes the linker to place section group members like
1799 normal input sections, and to delete the section groups. This is the
1800 default behaviour for a final link but this option can be used to
1801 change the behaviour of a relocatable link (@samp{-r}). The script
1802 command @code{FORCE_GROUP_ALLOCATION} has the same
1803 effect. @xref{Miscellaneous Commands}.
1805 @cindex symbols, from command line
1806 @kindex --defsym=@var{symbol}=@var{exp}
1807 @item --defsym=@var{symbol}=@var{expression}
1808 Create a global symbol in the output file, containing the absolute
1809 address given by @var{expression}. You may use this option as many
1810 times as necessary to define multiple symbols in the command line. A
1811 limited form of arithmetic is supported for the @var{expression} in this
1812 context: you may give a hexadecimal constant or the name of an existing
1813 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1814 constants or symbols. If you need more elaborate expressions, consider
1815 using the linker command language from a script (@pxref{Assignments}).
1816 @emph{Note:} there should be no white space between @var{symbol}, the
1817 equals sign (``@key{=}''), and @var{expression}.
1819 The linker processes @samp{--defsym} arguments and @samp{-T} arguments
1820 in order, placing @samp{--defsym} before @samp{-T} will define the
1821 symbol before the linker script from @samp{-T} is processed, while
1822 placing @samp{--defsym} after @samp{-T} will define the symbol after
1823 the linker script has been processed. This difference has
1824 consequences for expressions within the linker script that use the
1825 @samp{--defsym} symbols, which order is correct will depend on what
1826 you are trying to achieve.
1828 @cindex demangling, from command line
1829 @kindex --demangle[=@var{style}]
1830 @kindex --no-demangle
1831 @item --demangle[=@var{style}]
1832 @itemx --no-demangle
1833 These options control whether to demangle symbol names in error messages
1834 and other output. When the linker is told to demangle, it tries to
1835 present symbol names in a readable fashion: it strips leading
1836 underscores if they are used by the object file format, and converts C++
1837 mangled symbol names into user readable names. Different compilers have
1838 different mangling styles. The optional demangling style argument can be used
1839 to choose an appropriate demangling style for your compiler. The linker will
1840 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1841 is set. These options may be used to override the default.
1843 @cindex dynamic linker, from command line
1844 @kindex -I@var{file}
1845 @kindex --dynamic-linker=@var{file}
1847 @itemx --dynamic-linker=@var{file}
1848 Set the name of the dynamic linker. This is only meaningful when
1849 generating dynamically linked ELF executables. The default dynamic
1850 linker is normally correct; don't use this unless you know what you are
1853 @kindex --no-dynamic-linker
1854 @item --no-dynamic-linker
1855 When producing an executable file, omit the request for a dynamic
1856 linker to be used at load-time. This is only meaningful for ELF
1857 executables that contain dynamic relocations, and usually requires
1858 entry point code that is capable of processing these relocations.
1860 @kindex --embedded-relocs
1861 @item --embedded-relocs
1862 This option is similar to the @option{--emit-relocs} option except
1863 that the relocs are stored in a target-specific section. This option
1864 is only supported by the @samp{BFIN}, @samp{CR16} and @emph{M68K}
1867 @kindex --disable-multiple-abs-defs
1868 @item --disable-multiple-abs-defs
1869 Do not allow multiple definitions with symbols included
1870 in filename invoked by -R or --just-symbols
1872 @kindex --fatal-warnings
1873 @kindex --no-fatal-warnings
1874 @item --fatal-warnings
1875 @itemx --no-fatal-warnings
1876 Treat all warnings as errors. The default behaviour can be restored
1877 with the option @option{--no-fatal-warnings}.
1879 @kindex --force-exe-suffix
1880 @item --force-exe-suffix
1881 Make sure that an output file has a .exe suffix.
1883 If a successfully built fully linked output file does not have a
1884 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1885 the output file to one of the same name with a @code{.exe} suffix. This
1886 option is useful when using unmodified Unix makefiles on a Microsoft
1887 Windows host, since some versions of Windows won't run an image unless
1888 it ends in a @code{.exe} suffix.
1890 @kindex --gc-sections
1891 @kindex --no-gc-sections
1892 @cindex garbage collection
1894 @itemx --no-gc-sections
1895 Enable garbage collection of unused input sections. It is ignored on
1896 targets that do not support this option. The default behaviour (of not
1897 performing this garbage collection) can be restored by specifying
1898 @samp{--no-gc-sections} on the command line. Note that garbage
1899 collection for COFF and PE format targets is supported, but the
1900 implementation is currently considered to be experimental.
1902 @samp{--gc-sections} decides which input sections are used by
1903 examining symbols and relocations. The section containing the entry
1904 symbol and all sections containing symbols undefined on the
1905 command-line will be kept, as will sections containing symbols
1906 referenced by dynamic objects. Note that when building shared
1907 libraries, the linker must assume that any visible symbol is
1908 referenced. Once this initial set of sections has been determined,
1909 the linker recursively marks as used any section referenced by their
1910 relocations. See @samp{--entry}, @samp{--undefined}, and
1911 @samp{--gc-keep-exported}.
1913 This option can be set when doing a partial link (enabled with option
1914 @samp{-r}). In this case the root of symbols kept must be explicitly
1915 specified either by one of the options @samp{--entry},
1916 @samp{--undefined}, or @samp{--gc-keep-exported} or by a @code{ENTRY}
1917 command in the linker script.
1919 As a GNU extension, ELF input sections marked with the
1920 @code{SHF_GNU_RETAIN} flag will not be garbage collected.
1922 @kindex --print-gc-sections
1923 @kindex --no-print-gc-sections
1924 @cindex garbage collection
1925 @item --print-gc-sections
1926 @itemx --no-print-gc-sections
1927 List all sections removed by garbage collection. The listing is
1928 printed on stderr. This option is only effective if garbage
1929 collection has been enabled via the @samp{--gc-sections}) option. The
1930 default behaviour (of not listing the sections that are removed) can
1931 be restored by specifying @samp{--no-print-gc-sections} on the command
1934 @kindex --gc-keep-exported
1935 @cindex garbage collection
1936 @item --gc-keep-exported
1937 When @samp{--gc-sections} is enabled, this option prevents garbage
1938 collection of unused input sections that contain global symbols having
1939 default or protected visibility. This option is intended to be used for
1940 executables where unreferenced sections would otherwise be garbage
1941 collected regardless of the external visibility of contained symbols.
1942 Note that this option has no effect when linking shared objects since
1943 it is already the default behaviour. This option is only supported for
1946 @kindex --print-output-format
1947 @cindex output format
1948 @item --print-output-format
1949 Print the name of the default output format (perhaps influenced by
1950 other command-line options). This is the string that would appear
1951 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1953 @kindex --print-memory-usage
1954 @cindex memory usage
1955 @item --print-memory-usage
1956 Print used size, total size and used size of memory regions created with
1957 the @ref{MEMORY} command. This is useful on embedded targets to have a
1958 quick view of amount of free memory. The format of the output has one
1959 headline and one line per region. It is both human readable and easily
1960 parsable by tools. Here is an example of an output:
1963 Memory region Used Size Region Size %age Used
1964 ROM: 256 KB 1 MB 25.00%
1965 RAM: 32 B 2 GB 0.00%
1972 Print a summary of the command-line options on the standard output and exit.
1974 @kindex --target-help
1976 Print a summary of all target-specific options on the standard output and exit.
1978 @kindex -Map=@var{mapfile}
1979 @item -Map=@var{mapfile}
1980 Print a link map to the file @var{mapfile}. See the description of the
1981 @option{-M} option, above. If @var{mapfile} is just the character
1982 @code{-} then the map will be written to stdout.
1984 Specifying a directory as @var{mapfile} causes the linker map to be
1985 written as a file inside the directory. Normally name of the file
1986 inside the directory is computed as the basename of the @var{output}
1987 file with @code{.map} appended. If however the special character
1988 @code{%} is used then this will be replaced by the full path of the
1989 output file. Additionally if there are any characters after the
1990 @var{%} symbol then @code{.map} will no longer be appended.
1993 -o foo.exe -Map=bar [Creates ./bar]
1994 -o ../dir/foo.exe -Map=bar [Creates ./bar]
1995 -o foo.exe -Map=../dir [Creates ../dir/foo.exe.map]
1996 -o ../dir2/foo.exe -Map=../dir [Creates ../dir/foo.exe.map]
1997 -o foo.exe -Map=% [Creates ./foo.exe.map]
1998 -o ../dir/foo.exe -Map=% [Creates ../dir/foo.exe.map]
1999 -o foo.exe -Map=%.bar [Creates ./foo.exe.bar]
2000 -o ../dir/foo.exe -Map=%.bar [Creates ../dir/foo.exe.bar]
2001 -o ../dir2/foo.exe -Map=../dir/% [Creates ../dir/../dir2/foo.exe.map]
2002 -o ../dir2/foo.exe -Map=../dir/%.bar [Creates ../dir/../dir2/foo.exe.bar]
2005 It is an error to specify more than one @code{%} character.
2007 If the map file already exists then it will be overwritten by this
2010 @cindex memory usage
2011 @kindex --no-keep-memory
2012 @item --no-keep-memory
2013 @command{ld} normally optimizes for speed over memory usage by caching the
2014 symbol tables of input files in memory. This option tells @command{ld} to
2015 instead optimize for memory usage, by rereading the symbol tables as
2016 necessary. This may be required if @command{ld} runs out of memory space
2017 while linking a large executable.
2019 @kindex --no-undefined
2022 @item --no-undefined
2024 Report unresolved symbol references from regular object files. This
2025 is done even if the linker is creating a non-symbolic shared library.
2026 The switch @option{--[no-]allow-shlib-undefined} controls the
2027 behaviour for reporting unresolved references found in shared
2028 libraries being linked in.
2030 The effects of this option can be reverted by using @code{-z undefs}.
2032 @kindex --allow-multiple-definition
2034 @item --allow-multiple-definition
2036 Normally when a symbol is defined multiple times, the linker will
2037 report a fatal error. These options allow multiple definitions and the
2038 first definition will be used.
2040 @kindex --allow-shlib-undefined
2041 @kindex --no-allow-shlib-undefined
2042 @item --allow-shlib-undefined
2043 @itemx --no-allow-shlib-undefined
2044 Allows or disallows undefined symbols in shared libraries.
2045 This switch is similar to @option{--no-undefined} except that it
2046 determines the behaviour when the undefined symbols are in a
2047 shared library rather than a regular object file. It does not affect
2048 how undefined symbols in regular object files are handled.
2050 The default behaviour is to report errors for any undefined symbols
2051 referenced in shared libraries if the linker is being used to create
2052 an executable, but to allow them if the linker is being used to create
2055 The reasons for allowing undefined symbol references in shared
2056 libraries specified at link time are that:
2060 A shared library specified at link time may not be the same as the one
2061 that is available at load time, so the symbol might actually be
2062 resolvable at load time.
2064 There are some operating systems, eg BeOS and HPPA, where undefined
2065 symbols in shared libraries are normal.
2067 The BeOS kernel for example patches shared libraries at load time to
2068 select whichever function is most appropriate for the current
2069 architecture. This is used, for example, to dynamically select an
2070 appropriate memset function.
2073 @kindex --error-handling-script=@var{scriptname}
2074 @item --error-handling-script=@var{scriptname}
2075 If this option is provided then the linker will invoke
2076 @var{scriptname} whenever an error is encountered. Currently however
2077 only two kinds of error are supported: missing symbols and missing
2078 libraries. Two arguments will be passed to script: the keyword
2079 ``undefined-symbol'' or `missing-lib'' and the @var{name} of the
2080 undefined symbol or missing library. The intention is that the script
2081 will provide suggestions to the user as to where the symbol or library
2082 might be found. After the script has finished then the normal linker
2083 error message will be displayed.
2085 The availability of this option is controlled by a configure time
2086 switch, so it may not be present in specific implementations.
2088 @kindex --no-undefined-version
2089 @item --no-undefined-version
2090 Normally when a symbol has an undefined version, the linker will ignore
2091 it. This option disallows symbols with undefined version and a fatal error
2092 will be issued instead.
2094 @kindex --default-symver
2095 @item --default-symver
2096 Create and use a default symbol version (the soname) for unversioned
2099 @kindex --default-imported-symver
2100 @item --default-imported-symver
2101 Create and use a default symbol version (the soname) for unversioned
2104 @kindex --no-warn-mismatch
2105 @item --no-warn-mismatch
2106 Normally @command{ld} will give an error if you try to link together input
2107 files that are mismatched for some reason, perhaps because they have
2108 been compiled for different processors or for different endiannesses.
2109 This option tells @command{ld} that it should silently permit such possible
2110 errors. This option should only be used with care, in cases when you
2111 have taken some special action that ensures that the linker errors are
2114 @kindex --no-warn-search-mismatch
2115 @item --no-warn-search-mismatch
2116 Normally @command{ld} will give a warning if it finds an incompatible
2117 library during a library search. This option silences the warning.
2119 @kindex --no-whole-archive
2120 @item --no-whole-archive
2121 Turn off the effect of the @option{--whole-archive} option for subsequent
2124 @cindex output file after errors
2125 @kindex --noinhibit-exec
2126 @item --noinhibit-exec
2127 Retain the executable output file whenever it is still usable.
2128 Normally, the linker will not produce an output file if it encounters
2129 errors during the link process; it exits without writing an output file
2130 when it issues any error whatsoever.
2134 Only search library directories explicitly specified on the
2135 command line. Library directories specified in linker scripts
2136 (including linker scripts specified on the command line) are ignored.
2138 @ifclear SingleFormat
2139 @kindex --oformat=@var{output-format}
2140 @item --oformat=@var{output-format}
2141 @command{ld} may be configured to support more than one kind of object
2142 file. If your @command{ld} is configured this way, you can use the
2143 @samp{--oformat} option to specify the binary format for the output
2144 object file. Even when @command{ld} is configured to support alternative
2145 object formats, you don't usually need to specify this, as @command{ld}
2146 should be configured to produce as a default output format the most
2147 usual format on each machine. @var{output-format} is a text string, the
2148 name of a particular format supported by the BFD libraries. (You can
2149 list the available binary formats with @samp{objdump -i}.) The script
2150 command @code{OUTPUT_FORMAT} can also specify the output format, but
2151 this option overrides it. @xref{BFD}.
2154 @kindex --out-implib
2155 @item --out-implib @var{file}
2156 Create an import library in @var{file} corresponding to the executable
2157 the linker is generating (eg. a DLL or ELF program). This import
2158 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
2159 may be used to link clients against the generated executable; this
2160 behaviour makes it possible to skip a separate import library creation
2161 step (eg. @code{dlltool} for DLLs). This option is only available for
2162 the i386 PE and ELF targetted ports of the linker.
2165 @kindex --pic-executable
2167 @itemx --pic-executable
2168 @cindex position independent executables
2169 Create a position independent executable. This is currently only supported on
2170 ELF platforms. Position independent executables are similar to shared
2171 libraries in that they are relocated by the dynamic linker to the virtual
2172 address the OS chooses for them (which can vary between invocations). Like
2173 normal dynamically linked executables they can be executed and symbols
2174 defined in the executable cannot be overridden by shared libraries.
2178 @cindex position dependent executables
2179 Create a position dependent executable. This is the default.
2183 This option is ignored for Linux compatibility.
2187 This option is ignored for SVR4 compatibility.
2190 @cindex synthesizing linker
2191 @cindex relaxing addressing modes
2195 An option with machine dependent effects.
2197 This option is only supported on a few targets.
2200 @xref{H8/300,,@command{ld} and the H8/300}.
2203 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
2206 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
2209 @xref{Nios II,,@command{ld} and the Altera Nios II}.
2212 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
2215 On some platforms the @option{--relax} option performs target specific,
2216 global optimizations that become possible when the linker resolves
2217 addressing in the program, such as relaxing address modes,
2218 synthesizing new instructions, selecting shorter version of current
2219 instructions, and combining constant values.
2221 On some platforms these link time global optimizations may make symbolic
2222 debugging of the resulting executable impossible.
2224 This is known to be the case for the Matsushita MN10200 and MN10300
2225 family of processors.
2228 On platforms where the feature is supported, the option
2229 @option{--no-relax} will disable it.
2231 On platforms where the feature is not supported, both @option{--relax}
2232 and @option{--no-relax} are accepted, but ignored.
2234 @cindex retaining specified symbols
2235 @cindex stripping all but some symbols
2236 @cindex symbols, retaining selectively
2237 @kindex --retain-symbols-file=@var{filename}
2238 @item --retain-symbols-file=@var{filename}
2239 Retain @emph{only} the symbols listed in the file @var{filename},
2240 discarding all others. @var{filename} is simply a flat file, with one
2241 symbol name per line. This option is especially useful in environments
2245 where a large global symbol table is accumulated gradually, to conserve
2248 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
2249 or symbols needed for relocations.
2251 You may only specify @samp{--retain-symbols-file} once in the command
2252 line. It overrides @samp{-s} and @samp{-S}.
2255 @item -rpath=@var{dir}
2256 @cindex runtime library search path
2257 @kindex -rpath=@var{dir}
2258 Add a directory to the runtime library search path. This is used when
2259 linking an ELF executable with shared objects. All @option{-rpath}
2260 arguments are concatenated and passed to the runtime linker, which uses
2261 them to locate shared objects at runtime.
2263 The @option{-rpath} option is also used when locating shared objects which
2264 are needed by shared objects explicitly included in the link; see the
2265 description of the @option{-rpath-link} option. Searching @option{-rpath}
2266 in this way is only supported by native linkers and cross linkers which
2267 have been configured with the @option{--with-sysroot} option.
2269 If @option{-rpath} is not used when linking an ELF executable, the
2270 contents of the environment variable @code{LD_RUN_PATH} will be used if it
2273 The @option{-rpath} option may also be used on SunOS. By default, on
2274 SunOS, the linker will form a runtime search path out of all the
2275 @option{-L} options it is given. If a @option{-rpath} option is used, the
2276 runtime search path will be formed exclusively using the @option{-rpath}
2277 options, ignoring the @option{-L} options. This can be useful when using
2278 gcc, which adds many @option{-L} options which may be on NFS mounted
2281 For compatibility with other ELF linkers, if the @option{-R} option is
2282 followed by a directory name, rather than a file name, it is treated as
2283 the @option{-rpath} option.
2287 @cindex link-time runtime library search path
2288 @kindex -rpath-link=@var{dir}
2289 @item -rpath-link=@var{dir}
2290 When using ELF or SunOS, one shared library may require another. This
2291 happens when an @code{ld -shared} link includes a shared library as one
2294 When the linker encounters such a dependency when doing a non-shared,
2295 non-relocatable link, it will automatically try to locate the required
2296 shared library and include it in the link, if it is not included
2297 explicitly. In such a case, the @option{-rpath-link} option
2298 specifies the first set of directories to search. The
2299 @option{-rpath-link} option may specify a sequence of directory names
2300 either by specifying a list of names separated by colons, or by
2301 appearing multiple times.
2303 The tokens @var{$ORIGIN} and @var{$LIB} can appear in these search
2304 directories. They will be replaced by the full path to the directory
2305 containing the program or shared object in the case of @var{$ORIGIN}
2306 and either @samp{lib} - for 32-bit binaries - or @samp{lib64} - for
2307 64-bit binaries - in the case of @var{$LIB}.
2309 The alternative form of these tokens - @var{$@{ORIGIN@}} and
2310 @var{$@{LIB@}} can also be used. The token @var{$PLATFORM} is not
2313 This option should be used with caution as it overrides the search path
2314 that may have been hard compiled into a shared library. In such a case it
2315 is possible to use unintentionally a different search path than the
2316 runtime linker would do.
2318 The linker uses the following search paths to locate required shared
2323 Any directories specified by @option{-rpath-link} options.
2325 Any directories specified by @option{-rpath} options. The difference
2326 between @option{-rpath} and @option{-rpath-link} is that directories
2327 specified by @option{-rpath} options are included in the executable and
2328 used at runtime, whereas the @option{-rpath-link} option is only effective
2329 at link time. Searching @option{-rpath} in this way is only supported
2330 by native linkers and cross linkers which have been configured with
2331 the @option{--with-sysroot} option.
2333 On an ELF system, for native linkers, if the @option{-rpath} and
2334 @option{-rpath-link} options were not used, search the contents of the
2335 environment variable @code{LD_RUN_PATH}.
2337 On SunOS, if the @option{-rpath} option was not used, search any
2338 directories specified using @option{-L} options.
2340 For a native linker, search the contents of the environment
2341 variable @code{LD_LIBRARY_PATH}.
2343 For a native ELF linker, the directories in @code{DT_RUNPATH} or
2344 @code{DT_RPATH} of a shared library are searched for shared
2345 libraries needed by it. The @code{DT_RPATH} entries are ignored if
2346 @code{DT_RUNPATH} entries exist.
2348 For a linker for a Linux system, if the file @file{/etc/ld.so.conf}
2349 exists, the list of directories found in that file. Note: the path
2350 to this file is prefixed with the @code{sysroot} value, if that is
2351 defined, and then any @code{prefix} string if the linker was
2352 configured with the @command{--prefix=<path>} option.
2354 For a native linker on a FreeBSD system, any directories specified by
2355 the @code{_PATH_ELF_HINTS} macro defined in the @file{elf-hints.h}
2358 Any directories specified by a @code{SEARCH_DIR} command in a
2359 linker script given on the command line, including scripts specified
2360 by @option{-T} (but not @option{-dT}).
2362 The default directories, normally @file{/lib} and @file{/usr/lib}.
2364 Any directories specified by a plugin LDPT_SET_EXTRA_LIBRARY_PATH.
2366 Any directories specified by a @code{SEARCH_DIR} command in a default
2370 Note however on Linux based systems there is an additional caveat: If
2371 the @option{--as-needed} option is active @emph{and} a shared library
2372 is located which would normally satisfy the search @emph{and} this
2373 library does not have DT_NEEDED tag for @file{libc.so}
2374 @emph{and} there is a shared library later on in the set of search
2375 directories which also satisfies the search @emph{and}
2376 this second shared library does have a DT_NEEDED tag for
2377 @file{libc.so} @emph{then} the second library will be selected instead
2380 If the required shared library is not found, the linker will issue a
2381 warning and continue with the link.
2389 @cindex shared libraries
2390 Create a shared library. This is currently only supported on ELF, XCOFF
2391 and SunOS platforms. On SunOS, the linker will automatically create a
2392 shared library if the @option{-e} option is not used and there are
2393 undefined symbols in the link.
2395 @kindex --sort-common
2397 @itemx --sort-common=ascending
2398 @itemx --sort-common=descending
2399 This option tells @command{ld} to sort the common symbols by alignment in
2400 ascending or descending order when it places them in the appropriate output
2401 sections. The symbol alignments considered are sixteen-byte or larger,
2402 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
2403 between symbols due to alignment constraints. If no sorting order is
2404 specified, then descending order is assumed.
2406 @kindex --sort-section=name
2407 @item --sort-section=name
2408 This option will apply @code{SORT_BY_NAME} to all wildcard section
2409 patterns in the linker script.
2411 @kindex --sort-section=alignment
2412 @item --sort-section=alignment
2413 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
2414 patterns in the linker script.
2416 @kindex --spare-dynamic-tags
2417 @item --spare-dynamic-tags=@var{count}
2418 This option specifies the number of empty slots to leave in the
2419 .dynamic section of ELF shared objects. Empty slots may be needed by
2420 post processing tools, such as the prelinker. The default is 5.
2422 @kindex --split-by-file
2423 @item --split-by-file[=@var{size}]
2424 Similar to @option{--split-by-reloc} but creates a new output section for
2425 each input file when @var{size} is reached. @var{size} defaults to a
2426 size of 1 if not given.
2428 @kindex --split-by-reloc
2429 @item --split-by-reloc[=@var{count}]
2430 Tries to creates extra sections in the output file so that no single
2431 output section in the file contains more than @var{count} relocations.
2432 This is useful when generating huge relocatable files for downloading into
2433 certain real time kernels with the COFF object file format; since COFF
2434 cannot represent more than 65535 relocations in a single section. Note
2435 that this will fail to work with object file formats which do not
2436 support arbitrary sections. The linker will not split up individual
2437 input sections for redistribution, so if a single input section contains
2438 more than @var{count} relocations one output section will contain that
2439 many relocations. @var{count} defaults to a value of 32768.
2443 Compute and display statistics about the operation of the linker, such
2444 as execution time and memory usage.
2446 @kindex --sysroot=@var{directory}
2447 @item --sysroot=@var{directory}
2448 Use @var{directory} as the location of the sysroot, overriding the
2449 configure-time default. This option is only supported by linkers
2450 that were configured using @option{--with-sysroot}.
2454 This is used by COFF/PE based targets to create a task-linked object
2455 file where all of the global symbols have been converted to statics.
2457 @kindex --traditional-format
2458 @cindex traditional format
2459 @item --traditional-format
2460 For some targets, the output of @command{ld} is different in some ways from
2461 the output of some existing linker. This switch requests @command{ld} to
2462 use the traditional format instead.
2465 For example, on SunOS, @command{ld} combines duplicate entries in the
2466 symbol string table. This can reduce the size of an output file with
2467 full debugging information by over 30 percent. Unfortunately, the SunOS
2468 @code{dbx} program can not read the resulting program (@code{gdb} has no
2469 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
2470 combine duplicate entries.
2472 @kindex --section-start=@var{sectionname}=@var{org}
2473 @item --section-start=@var{sectionname}=@var{org}
2474 Locate a section in the output file at the absolute
2475 address given by @var{org}. You may use this option as many
2476 times as necessary to locate multiple sections in the command
2478 @var{org} must be a single hexadecimal integer;
2479 for compatibility with other linkers, you may omit the leading
2480 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
2481 should be no white space between @var{sectionname}, the equals
2482 sign (``@key{=}''), and @var{org}.
2484 @kindex -Tbss=@var{org}
2485 @kindex -Tdata=@var{org}
2486 @kindex -Ttext=@var{org}
2487 @cindex segment origins, cmd line
2488 @item -Tbss=@var{org}
2489 @itemx -Tdata=@var{org}
2490 @itemx -Ttext=@var{org}
2491 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2492 @code{.text} as the @var{sectionname}.
2494 @kindex -Ttext-segment=@var{org}
2495 @item -Ttext-segment=@var{org}
2496 @cindex text segment origin, cmd line
2497 When creating an ELF executable, it will set the address of the first
2498 byte of the text segment.
2500 @kindex -Trodata-segment=@var{org}
2501 @item -Trodata-segment=@var{org}
2502 @cindex rodata segment origin, cmd line
2503 When creating an ELF executable or shared object for a target where
2504 the read-only data is in its own segment separate from the executable
2505 text, it will set the address of the first byte of the read-only data segment.
2507 @kindex -Tldata-segment=@var{org}
2508 @item -Tldata-segment=@var{org}
2509 @cindex ldata segment origin, cmd line
2510 When creating an ELF executable or shared object for x86-64 medium memory
2511 model, it will set the address of the first byte of the ldata segment.
2513 @kindex --unresolved-symbols
2514 @item --unresolved-symbols=@var{method}
2515 Determine how to handle unresolved symbols. There are four possible
2516 values for @samp{method}:
2520 Do not report any unresolved symbols.
2523 Report all unresolved symbols. This is the default.
2525 @item ignore-in-object-files
2526 Report unresolved symbols that are contained in shared libraries, but
2527 ignore them if they come from regular object files.
2529 @item ignore-in-shared-libs
2530 Report unresolved symbols that come from regular object files, but
2531 ignore them if they come from shared libraries. This can be useful
2532 when creating a dynamic binary and it is known that all the shared
2533 libraries that it should be referencing are included on the linker's
2537 The behaviour for shared libraries on their own can also be controlled
2538 by the @option{--[no-]allow-shlib-undefined} option.
2540 Normally the linker will generate an error message for each reported
2541 unresolved symbol but the option @option{--warn-unresolved-symbols}
2542 can change this to a warning.
2544 @kindex --verbose[=@var{NUMBER}]
2545 @cindex verbose[=@var{NUMBER}]
2547 @itemx --verbose[=@var{NUMBER}]
2548 Display the version number for @command{ld} and list the linker emulations
2549 supported. Display which input files can and cannot be opened. Display
2550 the linker script being used by the linker. If the optional @var{NUMBER}
2551 argument > 1, plugin symbol status will also be displayed.
2553 @kindex --version-script=@var{version-scriptfile}
2554 @cindex version script, symbol versions
2555 @item --version-script=@var{version-scriptfile}
2556 Specify the name of a version script to the linker. This is typically
2557 used when creating shared libraries to specify additional information
2558 about the version hierarchy for the library being created. This option
2559 is only fully supported on ELF platforms which support shared libraries;
2560 see @ref{VERSION}. It is partially supported on PE platforms, which can
2561 use version scripts to filter symbol visibility in auto-export mode: any
2562 symbols marked @samp{local} in the version script will not be exported.
2565 @kindex --warn-common
2566 @cindex warnings, on combining symbols
2567 @cindex combining symbols, warnings on
2569 Warn when a common symbol is combined with another common symbol or with
2570 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2571 but linkers on some other operating systems do not. This option allows
2572 you to find potential problems from combining global symbols.
2573 Unfortunately, some C libraries use this practice, so you may get some
2574 warnings about symbols in the libraries as well as in your programs.
2576 There are three kinds of global symbols, illustrated here by C examples:
2580 A definition, which goes in the initialized data section of the output
2584 An undefined reference, which does not allocate space.
2585 There must be either a definition or a common symbol for the
2589 A common symbol. If there are only (one or more) common symbols for a
2590 variable, it goes in the uninitialized data area of the output file.
2591 The linker merges multiple common symbols for the same variable into a
2592 single symbol. If they are of different sizes, it picks the largest
2593 size. The linker turns a common symbol into a declaration, if there is
2594 a definition of the same variable.
2597 The @samp{--warn-common} option can produce five kinds of warnings.
2598 Each warning consists of a pair of lines: the first describes the symbol
2599 just encountered, and the second describes the previous symbol
2600 encountered with the same name. One or both of the two symbols will be
2605 Turning a common symbol into a reference, because there is already a
2606 definition for the symbol.
2608 @var{file}(@var{section}): warning: common of `@var{symbol}'
2609 overridden by definition
2610 @var{file}(@var{section}): warning: defined here
2614 Turning a common symbol into a reference, because a later definition for
2615 the symbol is encountered. This is the same as the previous case,
2616 except that the symbols are encountered in a different order.
2618 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2620 @var{file}(@var{section}): warning: common is here
2624 Merging a common symbol with a previous same-sized common symbol.
2626 @var{file}(@var{section}): warning: multiple common
2628 @var{file}(@var{section}): warning: previous common is here
2632 Merging a common symbol with a previous larger common symbol.
2634 @var{file}(@var{section}): warning: common of `@var{symbol}'
2635 overridden by larger common
2636 @var{file}(@var{section}): warning: larger common is here
2640 Merging a common symbol with a previous smaller common symbol. This is
2641 the same as the previous case, except that the symbols are
2642 encountered in a different order.
2644 @var{file}(@var{section}): warning: common of `@var{symbol}'
2645 overriding smaller common
2646 @var{file}(@var{section}): warning: smaller common is here
2650 @kindex --warn-constructors
2651 @item --warn-constructors
2652 Warn if any global constructors are used. This is only useful for a few
2653 object file formats. For formats like COFF or ELF, the linker can not
2654 detect the use of global constructors.
2656 @kindex --warn-multiple-gp
2657 @item --warn-multiple-gp
2658 Warn if multiple global pointer values are required in the output file.
2659 This is only meaningful for certain processors, such as the Alpha.
2660 Specifically, some processors put large-valued constants in a special
2661 section. A special register (the global pointer) points into the middle
2662 of this section, so that constants can be loaded efficiently via a
2663 base-register relative addressing mode. Since the offset in
2664 base-register relative mode is fixed and relatively small (e.g., 16
2665 bits), this limits the maximum size of the constant pool. Thus, in
2666 large programs, it is often necessary to use multiple global pointer
2667 values in order to be able to address all possible constants. This
2668 option causes a warning to be issued whenever this case occurs.
2671 @cindex warnings, on undefined symbols
2672 @cindex undefined symbols, warnings on
2674 Only warn once for each undefined symbol, rather than once per module
2677 @kindex --warn-section-align
2678 @cindex warnings, on section alignment
2679 @cindex section alignment, warnings on
2680 @item --warn-section-align
2681 Warn if the address of an output section is changed because of
2682 alignment. Typically, the alignment will be set by an input section.
2683 The address will only be changed if it not explicitly specified; that
2684 is, if the @code{SECTIONS} command does not specify a start address for
2685 the section (@pxref{SECTIONS}).
2687 @kindex --warn-textrel
2688 @item --warn-textrel
2689 Warn if the linker adds DT_TEXTREL to a position-independent executable
2692 @kindex --warn-alternate-em
2693 @item --warn-alternate-em
2694 Warn if an object has alternate ELF machine code.
2696 @kindex --warn-unresolved-symbols
2697 @item --warn-unresolved-symbols
2698 If the linker is going to report an unresolved symbol (see the option
2699 @option{--unresolved-symbols}) it will normally generate an error.
2700 This option makes it generate a warning instead.
2702 @kindex --error-unresolved-symbols
2703 @item --error-unresolved-symbols
2704 This restores the linker's default behaviour of generating errors when
2705 it is reporting unresolved symbols.
2707 @kindex --whole-archive
2708 @cindex including an entire archive
2709 @item --whole-archive
2710 For each archive mentioned on the command line after the
2711 @option{--whole-archive} option, include every object file in the archive
2712 in the link, rather than searching the archive for the required object
2713 files. This is normally used to turn an archive file into a shared
2714 library, forcing every object to be included in the resulting shared
2715 library. This option may be used more than once.
2717 Two notes when using this option from gcc: First, gcc doesn't know
2718 about this option, so you have to use @option{-Wl,-whole-archive}.
2719 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2720 list of archives, because gcc will add its own list of archives to
2721 your link and you may not want this flag to affect those as well.
2723 @kindex --wrap=@var{symbol}
2724 @item --wrap=@var{symbol}
2725 Use a wrapper function for @var{symbol}. Any undefined reference to
2726 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2727 undefined reference to @code{__real_@var{symbol}} will be resolved to
2730 This can be used to provide a wrapper for a system function. The
2731 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2732 wishes to call the system function, it should call
2733 @code{__real_@var{symbol}}.
2735 Here is a trivial example:
2739 __wrap_malloc (size_t c)
2741 printf ("malloc called with %zu\n", c);
2742 return __real_malloc (c);
2746 If you link other code with this file using @option{--wrap malloc}, then
2747 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2748 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2749 call the real @code{malloc} function.
2751 You may wish to provide a @code{__real_malloc} function as well, so that
2752 links without the @option{--wrap} option will succeed. If you do this,
2753 you should not put the definition of @code{__real_malloc} in the same
2754 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2755 call before the linker has a chance to wrap it to @code{malloc}.
2757 Only undefined references are replaced by the linker. So, translation unit
2758 internal references to @var{symbol} are not resolved to
2759 @code{__wrap_@var{symbol}}. In the next example, the call to @code{f} in
2760 @code{g} is not resolved to @code{__wrap_f}.
2776 @kindex --eh-frame-hdr
2777 @kindex --no-eh-frame-hdr
2778 @item --eh-frame-hdr
2779 @itemx --no-eh-frame-hdr
2780 Request (@option{--eh-frame-hdr}) or suppress
2781 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2782 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2784 @kindex --ld-generated-unwind-info
2785 @item --no-ld-generated-unwind-info
2786 Request creation of @code{.eh_frame} unwind info for linker
2787 generated code sections like PLT. This option is on by default
2788 if linker generated unwind info is supported.
2790 @kindex --enable-new-dtags
2791 @kindex --disable-new-dtags
2792 @item --enable-new-dtags
2793 @itemx --disable-new-dtags
2794 This linker can create the new dynamic tags in ELF. But the older ELF
2795 systems may not understand them. If you specify
2796 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2797 and older dynamic tags will be omitted.
2798 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2799 created. By default, the new dynamic tags are not created. Note that
2800 those options are only available for ELF systems.
2802 @kindex --hash-size=@var{number}
2803 @item --hash-size=@var{number}
2804 Set the default size of the linker's hash tables to a prime number
2805 close to @var{number}. Increasing this value can reduce the length of
2806 time it takes the linker to perform its tasks, at the expense of
2807 increasing the linker's memory requirements. Similarly reducing this
2808 value can reduce the memory requirements at the expense of speed.
2810 @kindex --hash-style=@var{style}
2811 @item --hash-style=@var{style}
2812 Set the type of linker's hash table(s). @var{style} can be either
2813 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2814 new style GNU @code{.gnu.hash} section or @code{both} for both
2815 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2816 hash tables. The default depends upon how the linker was configured,
2817 but for most Linux based systems it will be @code{both}.
2819 @kindex --compress-debug-sections=none
2820 @kindex --compress-debug-sections=zlib
2821 @kindex --compress-debug-sections=zlib-gnu
2822 @kindex --compress-debug-sections=zlib-gabi
2823 @item --compress-debug-sections=none
2824 @itemx --compress-debug-sections=zlib
2825 @itemx --compress-debug-sections=zlib-gnu
2826 @itemx --compress-debug-sections=zlib-gabi
2827 On ELF platforms, these options control how DWARF debug sections are
2828 compressed using zlib.
2830 @option{--compress-debug-sections=none} doesn't compress DWARF debug
2831 sections. @option{--compress-debug-sections=zlib-gnu} compresses
2832 DWARF debug sections and renames them to begin with @samp{.zdebug}
2833 instead of @samp{.debug}. @option{--compress-debug-sections=zlib-gabi}
2834 also compresses DWARF debug sections, but rather than renaming them it
2835 sets the SHF_COMPRESSED flag in the sections' headers.
2837 The @option{--compress-debug-sections=zlib} option is an alias for
2838 @option{--compress-debug-sections=zlib-gabi}.
2840 Note that this option overrides any compression in input debug
2841 sections, so if a binary is linked with @option{--compress-debug-sections=none}
2842 for example, then any compressed debug sections in input files will be
2843 uncompressed before they are copied into the output binary.
2845 The default compression behaviour varies depending upon the target
2846 involved and the configure options used to build the toolchain. The
2847 default can be determined by examining the output from the linker's
2848 @option{--help} option.
2850 @kindex --reduce-memory-overheads
2851 @item --reduce-memory-overheads
2852 This option reduces memory requirements at ld runtime, at the expense of
2853 linking speed. This was introduced to select the old O(n^2) algorithm
2854 for link map file generation, rather than the new O(n) algorithm which uses
2855 about 40% more memory for symbol storage.
2857 Another effect of the switch is to set the default hash table size to
2858 1021, which again saves memory at the cost of lengthening the linker's
2859 run time. This is not done however if the @option{--hash-size} switch
2862 The @option{--reduce-memory-overheads} switch may be also be used to
2863 enable other tradeoffs in future versions of the linker.
2865 @kindex --max-cache-size=@var{size}
2866 @item --max-cache-size=@var{size}
2867 @command{ld} normally caches the relocation information and symbol tables
2868 of input files in memory with the unlimited size. This option sets the
2869 maximum cache size to @var{size}.
2872 @kindex --build-id=@var{style}
2874 @itemx --build-id=@var{style}
2875 Request the creation of a @code{.note.gnu.build-id} ELF note section
2876 or a @code{.buildid} COFF section. The contents of the note are
2877 unique bits identifying this linked file. @var{style} can be
2878 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2879 @sc{SHA1} hash on the normative parts of the output contents,
2880 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2881 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2882 string specified as an even number of hexadecimal digits (@code{-} and
2883 @code{:} characters between digit pairs are ignored). If @var{style}
2884 is omitted, @code{sha1} is used.
2886 The @code{md5} and @code{sha1} styles produces an identifier
2887 that is always the same in an identical output file, but will be
2888 unique among all nonidentical output files. It is not intended
2889 to be compared as a checksum for the file's contents. A linked
2890 file may be changed later by other tools, but the build ID bit
2891 string identifying the original linked file does not change.
2893 Passing @code{none} for @var{style} disables the setting from any
2894 @code{--build-id} options earlier on the command line.
2899 @subsection Options Specific to i386 PE Targets
2901 @c man begin OPTIONS
2903 The i386 PE linker supports the @option{-shared} option, which causes
2904 the output to be a dynamically linked library (DLL) instead of a
2905 normal executable. You should name the output @code{*.dll} when you
2906 use this option. In addition, the linker fully supports the standard
2907 @code{*.def} files, which may be specified on the linker command line
2908 like an object file (in fact, it should precede archives it exports
2909 symbols from, to ensure that they get linked in, just like a normal
2912 In addition to the options common to all targets, the i386 PE linker
2913 support additional command-line options that are specific to the i386
2914 PE target. Options that take values may be separated from their
2915 values by either a space or an equals sign.
2919 @kindex --add-stdcall-alias
2920 @item --add-stdcall-alias
2921 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2922 as-is and also with the suffix stripped.
2923 [This option is specific to the i386 PE targeted port of the linker]
2926 @item --base-file @var{file}
2927 Use @var{file} as the name of a file in which to save the base
2928 addresses of all the relocations needed for generating DLLs with
2930 [This is an i386 PE specific option]
2934 Create a DLL instead of a regular executable. You may also use
2935 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2937 [This option is specific to the i386 PE targeted port of the linker]
2939 @kindex --enable-long-section-names
2940 @kindex --disable-long-section-names
2941 @item --enable-long-section-names
2942 @itemx --disable-long-section-names
2943 The PE variants of the COFF object format add an extension that permits
2944 the use of section names longer than eight characters, the normal limit
2945 for COFF. By default, these names are only allowed in object files, as
2946 fully-linked executable images do not carry the COFF string table required
2947 to support the longer names. As a GNU extension, it is possible to
2948 allow their use in executable images as well, or to (probably pointlessly!)
2949 disallow it in object files, by using these two options. Executable images
2950 generated with these long section names are slightly non-standard, carrying
2951 as they do a string table, and may generate confusing output when examined
2952 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2953 GDB relies on the use of PE long section names to find Dwarf-2 debug
2954 information sections in an executable image at runtime, and so if neither
2955 option is specified on the command-line, @command{ld} will enable long
2956 section names, overriding the default and technically correct behaviour,
2957 when it finds the presence of debug information while linking an executable
2958 image and not stripping symbols.
2959 [This option is valid for all PE targeted ports of the linker]
2961 @kindex --enable-stdcall-fixup
2962 @kindex --disable-stdcall-fixup
2963 @item --enable-stdcall-fixup
2964 @itemx --disable-stdcall-fixup
2965 If the link finds a symbol that it cannot resolve, it will attempt to
2966 do ``fuzzy linking'' by looking for another defined symbol that differs
2967 only in the format of the symbol name (cdecl vs stdcall) and will
2968 resolve that symbol by linking to the match. For example, the
2969 undefined symbol @code{_foo} might be linked to the function
2970 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2971 to the function @code{_bar}. When the linker does this, it prints a
2972 warning, since it normally should have failed to link, but sometimes
2973 import libraries generated from third-party dlls may need this feature
2974 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2975 feature is fully enabled and warnings are not printed. If you specify
2976 @option{--disable-stdcall-fixup}, this feature is disabled and such
2977 mismatches are considered to be errors.
2978 [This option is specific to the i386 PE targeted port of the linker]
2980 @kindex --leading-underscore
2981 @kindex --no-leading-underscore
2982 @item --leading-underscore
2983 @itemx --no-leading-underscore
2984 For most targets default symbol-prefix is an underscore and is defined
2985 in target's description. By this option it is possible to
2986 disable/enable the default underscore symbol-prefix.
2988 @cindex DLLs, creating
2989 @kindex --export-all-symbols
2990 @item --export-all-symbols
2991 If given, all global symbols in the objects used to build a DLL will
2992 be exported by the DLL. Note that this is the default if there
2993 otherwise wouldn't be any exported symbols. When symbols are
2994 explicitly exported via DEF files or implicitly exported via function
2995 attributes, the default is to not export anything else unless this
2996 option is given. Note that the symbols @code{DllMain@@12},
2997 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2998 @code{impure_ptr} will not be automatically
2999 exported. Also, symbols imported from other DLLs will not be
3000 re-exported, nor will symbols specifying the DLL's internal layout
3001 such as those beginning with @code{_head_} or ending with
3002 @code{_iname}. In addition, no symbols from @code{libgcc},
3003 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
3004 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
3005 not be exported, to help with C++ DLLs. Finally, there is an
3006 extensive list of cygwin-private symbols that are not exported
3007 (obviously, this applies on when building DLLs for cygwin targets).
3008 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
3009 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
3010 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
3011 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
3012 @code{cygwin_premain3}, and @code{environ}.
3013 [This option is specific to the i386 PE targeted port of the linker]
3015 @kindex --exclude-symbols
3016 @item --exclude-symbols @var{symbol},@var{symbol},...
3017 Specifies a list of symbols which should not be automatically
3018 exported. The symbol names may be delimited by commas or colons.
3019 [This option is specific to the i386 PE targeted port of the linker]
3021 @kindex --exclude-all-symbols
3022 @item --exclude-all-symbols
3023 Specifies no symbols should be automatically exported.
3024 [This option is specific to the i386 PE targeted port of the linker]
3026 @kindex --file-alignment
3027 @item --file-alignment
3028 Specify the file alignment. Sections in the file will always begin at
3029 file offsets which are multiples of this number. This defaults to
3031 [This option is specific to the i386 PE targeted port of the linker]
3035 @item --heap @var{reserve}
3036 @itemx --heap @var{reserve},@var{commit}
3037 Specify the number of bytes of memory to reserve (and optionally commit)
3038 to be used as heap for this program. The default is 1MB reserved, 4K
3040 [This option is specific to the i386 PE targeted port of the linker]
3043 @kindex --image-base
3044 @item --image-base @var{value}
3045 Use @var{value} as the base address of your program or dll. This is
3046 the lowest memory location that will be used when your program or dll
3047 is loaded. To reduce the need to relocate and improve performance of
3048 your dlls, each should have a unique base address and not overlap any
3049 other dlls. The default is 0x400000 for executables, and 0x10000000
3051 [This option is specific to the i386 PE targeted port of the linker]
3055 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
3056 symbols before they are exported.
3057 [This option is specific to the i386 PE targeted port of the linker]
3059 @kindex --large-address-aware
3060 @item --large-address-aware
3061 If given, the appropriate bit in the ``Characteristics'' field of the COFF
3062 header is set to indicate that this executable supports virtual addresses
3063 greater than 2 gigabytes. This should be used in conjunction with the /3GB
3064 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
3065 section of the BOOT.INI. Otherwise, this bit has no effect.
3066 [This option is specific to PE targeted ports of the linker]
3068 @kindex --disable-large-address-aware
3069 @item --disable-large-address-aware
3070 Reverts the effect of a previous @samp{--large-address-aware} option.
3071 This is useful if @samp{--large-address-aware} is always set by the compiler
3072 driver (e.g. Cygwin gcc) and the executable does not support virtual
3073 addresses greater than 2 gigabytes.
3074 [This option is specific to PE targeted ports of the linker]
3076 @kindex --major-image-version
3077 @item --major-image-version @var{value}
3078 Sets the major number of the ``image version''. Defaults to 1.
3079 [This option is specific to the i386 PE targeted port of the linker]
3081 @kindex --major-os-version
3082 @item --major-os-version @var{value}
3083 Sets the major number of the ``os version''. Defaults to 4.
3084 [This option is specific to the i386 PE targeted port of the linker]
3086 @kindex --major-subsystem-version
3087 @item --major-subsystem-version @var{value}
3088 Sets the major number of the ``subsystem version''. Defaults to 4.
3089 [This option is specific to the i386 PE targeted port of the linker]
3091 @kindex --minor-image-version
3092 @item --minor-image-version @var{value}
3093 Sets the minor number of the ``image version''. Defaults to 0.
3094 [This option is specific to the i386 PE targeted port of the linker]
3096 @kindex --minor-os-version
3097 @item --minor-os-version @var{value}
3098 Sets the minor number of the ``os version''. Defaults to 0.
3099 [This option is specific to the i386 PE targeted port of the linker]
3101 @kindex --minor-subsystem-version
3102 @item --minor-subsystem-version @var{value}
3103 Sets the minor number of the ``subsystem version''. Defaults to 0.
3104 [This option is specific to the i386 PE targeted port of the linker]
3106 @cindex DEF files, creating
3107 @cindex DLLs, creating
3108 @kindex --output-def
3109 @item --output-def @var{file}
3110 The linker will create the file @var{file} which will contain a DEF
3111 file corresponding to the DLL the linker is generating. This DEF file
3112 (which should be called @code{*.def}) may be used to create an import
3113 library with @code{dlltool} or may be used as a reference to
3114 automatically or implicitly exported symbols.
3115 [This option is specific to the i386 PE targeted port of the linker]
3117 @cindex DLLs, creating
3118 @kindex --enable-auto-image-base
3119 @item --enable-auto-image-base
3120 @itemx --enable-auto-image-base=@var{value}
3121 Automatically choose the image base for DLLs, optionally starting with base
3122 @var{value}, unless one is specified using the @code{--image-base} argument.
3123 By using a hash generated from the dllname to create unique image bases
3124 for each DLL, in-memory collisions and relocations which can delay program
3125 execution are avoided.
3126 [This option is specific to the i386 PE targeted port of the linker]
3128 @kindex --disable-auto-image-base
3129 @item --disable-auto-image-base
3130 Do not automatically generate a unique image base. If there is no
3131 user-specified image base (@code{--image-base}) then use the platform
3133 [This option is specific to the i386 PE targeted port of the linker]
3135 @cindex DLLs, linking to
3136 @kindex --dll-search-prefix
3137 @item --dll-search-prefix @var{string}
3138 When linking dynamically to a dll without an import library,
3139 search for @code{<string><basename>.dll} in preference to
3140 @code{lib<basename>.dll}. This behaviour allows easy distinction
3141 between DLLs built for the various "subplatforms": native, cygwin,
3142 uwin, pw, etc. For instance, cygwin DLLs typically use
3143 @code{--dll-search-prefix=cyg}.
3144 [This option is specific to the i386 PE targeted port of the linker]
3146 @kindex --enable-auto-import
3147 @item --enable-auto-import
3148 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
3149 DATA imports from DLLs, thus making it possible to bypass the dllimport
3150 mechanism on the user side and to reference unmangled symbol names.
3151 [This option is specific to the i386 PE targeted port of the linker]
3153 The following remarks pertain to the original implementation of the
3154 feature and are obsolete nowadays for Cygwin and MinGW targets.
3156 Note: Use of the 'auto-import' extension will cause the text section
3157 of the image file to be made writable. This does not conform to the
3158 PE-COFF format specification published by Microsoft.
3160 Note - use of the 'auto-import' extension will also cause read only
3161 data which would normally be placed into the .rdata section to be
3162 placed into the .data section instead. This is in order to work
3163 around a problem with consts that is described here:
3164 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
3166 Using 'auto-import' generally will 'just work' -- but sometimes you may
3169 "variable '<var>' can't be auto-imported. Please read the
3170 documentation for ld's @code{--enable-auto-import} for details."
3172 This message occurs when some (sub)expression accesses an address
3173 ultimately given by the sum of two constants (Win32 import tables only
3174 allow one). Instances where this may occur include accesses to member
3175 fields of struct variables imported from a DLL, as well as using a
3176 constant index into an array variable imported from a DLL. Any
3177 multiword variable (arrays, structs, long long, etc) may trigger
3178 this error condition. However, regardless of the exact data type
3179 of the offending exported variable, ld will always detect it, issue
3180 the warning, and exit.
3182 There are several ways to address this difficulty, regardless of the
3183 data type of the exported variable:
3185 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
3186 of adjusting references in your client code for runtime environment, so
3187 this method works only when runtime environment supports this feature.
3189 A second solution is to force one of the 'constants' to be a variable --
3190 that is, unknown and un-optimizable at compile time. For arrays,
3191 there are two possibilities: a) make the indexee (the array's address)
3192 a variable, or b) make the 'constant' index a variable. Thus:
3195 extern type extern_array[];
3197 @{ volatile type *t=extern_array; t[1] @}
3203 extern type extern_array[];
3205 @{ volatile int t=1; extern_array[t] @}
3208 For structs (and most other multiword data types) the only option
3209 is to make the struct itself (or the long long, or the ...) variable:
3212 extern struct s extern_struct;
3213 extern_struct.field -->
3214 @{ volatile struct s *t=&extern_struct; t->field @}
3220 extern long long extern_ll;
3222 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
3225 A third method of dealing with this difficulty is to abandon
3226 'auto-import' for the offending symbol and mark it with
3227 @code{__declspec(dllimport)}. However, in practice that
3228 requires using compile-time #defines to indicate whether you are
3229 building a DLL, building client code that will link to the DLL, or
3230 merely building/linking to a static library. In making the choice
3231 between the various methods of resolving the 'direct address with
3232 constant offset' problem, you should consider typical real-world usage:
3240 void main(int argc, char **argv)@{
3241 printf("%d\n",arr[1]);
3251 void main(int argc, char **argv)@{
3252 /* This workaround is for win32 and cygwin; do not "optimize" */
3253 volatile int *parr = arr;
3254 printf("%d\n",parr[1]);
3261 /* Note: auto-export is assumed (no __declspec(dllexport)) */
3262 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
3263 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
3264 #define FOO_IMPORT __declspec(dllimport)
3268 extern FOO_IMPORT int arr[];
3271 void main(int argc, char **argv)@{
3272 printf("%d\n",arr[1]);
3276 A fourth way to avoid this problem is to re-code your
3277 library to use a functional interface rather than a data interface
3278 for the offending variables (e.g. set_foo() and get_foo() accessor
3281 @kindex --disable-auto-import
3282 @item --disable-auto-import
3283 Do not attempt to do sophisticated linking of @code{_symbol} to
3284 @code{__imp__symbol} for DATA imports from DLLs.
3285 [This option is specific to the i386 PE targeted port of the linker]
3287 @kindex --enable-runtime-pseudo-reloc
3288 @item --enable-runtime-pseudo-reloc
3289 If your code contains expressions described in --enable-auto-import section,
3290 that is, DATA imports from DLL with non-zero offset, this switch will create
3291 a vector of 'runtime pseudo relocations' which can be used by runtime
3292 environment to adjust references to such data in your client code.
3293 [This option is specific to the i386 PE targeted port of the linker]
3295 @kindex --disable-runtime-pseudo-reloc
3296 @item --disable-runtime-pseudo-reloc
3297 Do not create pseudo relocations for non-zero offset DATA imports from DLLs.
3298 [This option is specific to the i386 PE targeted port of the linker]
3300 @kindex --enable-extra-pe-debug
3301 @item --enable-extra-pe-debug
3302 Show additional debug info related to auto-import symbol thunking.
3303 [This option is specific to the i386 PE targeted port of the linker]
3305 @kindex --section-alignment
3306 @item --section-alignment
3307 Sets the section alignment. Sections in memory will always begin at
3308 addresses which are a multiple of this number. Defaults to 0x1000.
3309 [This option is specific to the i386 PE targeted port of the linker]
3313 @item --stack @var{reserve}
3314 @itemx --stack @var{reserve},@var{commit}
3315 Specify the number of bytes of memory to reserve (and optionally commit)
3316 to be used as stack for this program. The default is 2MB reserved, 4K
3318 [This option is specific to the i386 PE targeted port of the linker]
3321 @item --subsystem @var{which}
3322 @itemx --subsystem @var{which}:@var{major}
3323 @itemx --subsystem @var{which}:@var{major}.@var{minor}
3324 Specifies the subsystem under which your program will execute. The
3325 legal values for @var{which} are @code{native}, @code{windows},
3326 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
3327 the subsystem version also. Numeric values are also accepted for
3329 [This option is specific to the i386 PE targeted port of the linker]
3331 The following options set flags in the @code{DllCharacteristics} field
3332 of the PE file header:
3333 [These options are specific to PE targeted ports of the linker]
3335 @kindex --high-entropy-va
3336 @item --high-entropy-va
3337 @itemx --disable-high-entropy-va
3338 Image is compatible with 64-bit address space layout randomization
3339 (ASLR). This option is enabled by default for 64-bit PE images.
3341 This option also implies @option{--dynamicbase} and
3342 @option{--enable-reloc-section}.
3344 @kindex --dynamicbase
3346 @itemx --disable-dynamicbase
3347 The image base address may be relocated using address space layout
3348 randomization (ASLR). This feature was introduced with MS Windows
3349 Vista for i386 PE targets. This option is enabled by default but
3350 can be disabled via the @option{--disable-dynamicbase} option.
3351 This option also implies @option{--enable-reloc-section}.
3353 @kindex --forceinteg
3355 @itemx --disable-forceinteg
3356 Code integrity checks are enforced. This option is disabled by
3361 @item --disable-nxcompat
3362 The image is compatible with the Data Execution Prevention.
3363 This feature was introduced with MS Windows XP SP2 for i386 PE
3364 targets. The option is enabled by default.
3366 @kindex --no-isolation
3367 @item --no-isolation
3368 @itemx --disable-no-isolation
3369 Although the image understands isolation, do not isolate the image.
3370 This option is disabled by default.
3374 @itemx --disable-no-seh
3375 The image does not use SEH. No SE handler may be called from
3376 this image. This option is disabled by default.
3380 @itemx --disable-no-bind
3381 Do not bind this image. This option is disabled by default.
3385 @itemx --disable-wdmdriver
3386 The driver uses the MS Windows Driver Model. This option is disabled
3391 @itemx --disable-tsaware
3392 The image is Terminal Server aware. This option is disabled by
3395 @kindex --insert-timestamp
3396 @item --insert-timestamp
3397 @itemx --no-insert-timestamp
3398 Insert a real timestamp into the image. This is the default behaviour
3399 as it matches legacy code and it means that the image will work with
3400 other, proprietary tools. The problem with this default is that it
3401 will result in slightly different images being produced each time the
3402 same sources are linked. The option @option{--no-insert-timestamp}
3403 can be used to insert a zero value for the timestamp, this ensuring
3404 that binaries produced from identical sources will compare
3407 @kindex --enable-reloc-section
3408 @item --enable-reloc-section
3409 @itemx --disable-reloc-section
3410 Create the base relocation table, which is necessary if the image
3411 is loaded at a different image base than specified in the PE header.
3412 This option is enabled by default.
3418 @subsection Options specific to C6X uClinux targets
3420 @c man begin OPTIONS
3422 The C6X uClinux target uses a binary format called DSBT to support shared
3423 libraries. Each shared library in the system needs to have a unique index;
3424 all executables use an index of 0.
3429 @item --dsbt-size @var{size}
3430 This option sets the number of entries in the DSBT of the current executable
3431 or shared library to @var{size}. The default is to create a table with 64
3434 @kindex --dsbt-index
3435 @item --dsbt-index @var{index}
3436 This option sets the DSBT index of the current executable or shared library
3437 to @var{index}. The default is 0, which is appropriate for generating
3438 executables. If a shared library is generated with a DSBT index of 0, the
3439 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
3441 @kindex --no-merge-exidx-entries
3442 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
3443 exidx entries in frame unwind info.
3451 @subsection Options specific to C-SKY targets
3453 @c man begin OPTIONS
3457 @kindex --branch-stub on C-SKY
3459 This option enables linker branch relaxation by inserting branch stub
3460 sections when needed to extend the range of branches. This option is
3461 usually not required since C-SKY supports branch and call instructions that
3462 can access the full memory range and branch relaxation is normally handled by
3463 the compiler or assembler.
3465 @kindex --stub-group-size on C-SKY
3466 @item --stub-group-size=@var{N}
3467 This option allows finer control of linker branch stub creation.
3468 It sets the maximum size of a group of input sections that can
3469 be handled by one stub section. A negative value of @var{N} locates
3470 stub sections after their branches, while a positive value allows stub
3471 sections to appear either before or after the branches. Values of
3472 @samp{1} or @samp{-1} indicate that the
3473 linker should choose suitable defaults.
3481 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
3483 @c man begin OPTIONS
3485 The 68HC11 and 68HC12 linkers support specific options to control the
3486 memory bank switching mapping and trampoline code generation.
3490 @kindex --no-trampoline
3491 @item --no-trampoline
3492 This option disables the generation of trampoline. By default a trampoline
3493 is generated for each far function which is called using a @code{jsr}
3494 instruction (this happens when a pointer to a far function is taken).
3496 @kindex --bank-window
3497 @item --bank-window @var{name}
3498 This option indicates to the linker the name of the memory region in
3499 the @samp{MEMORY} specification that describes the memory bank window.
3500 The definition of such region is then used by the linker to compute
3501 paging and addresses within the memory window.
3509 @subsection Options specific to Motorola 68K target
3511 @c man begin OPTIONS
3513 The following options are supported to control handling of GOT generation
3514 when linking for 68K targets.
3519 @item --got=@var{type}
3520 This option tells the linker which GOT generation scheme to use.
3521 @var{type} should be one of @samp{single}, @samp{negative},
3522 @samp{multigot} or @samp{target}. For more information refer to the
3523 Info entry for @file{ld}.
3531 @subsection Options specific to MIPS targets
3533 @c man begin OPTIONS
3535 The following options are supported to control microMIPS instruction
3536 generation and branch relocation checks for ISA mode transitions when
3537 linking for MIPS targets.
3545 These options control the choice of microMIPS instructions used in code
3546 generated by the linker, such as that in the PLT or lazy binding stubs,
3547 or in relaxation. If @samp{--insn32} is used, then the linker only uses
3548 32-bit instruction encodings. By default or if @samp{--no-insn32} is
3549 used, all instruction encodings are used, including 16-bit ones where
3552 @kindex --ignore-branch-isa
3553 @item --ignore-branch-isa
3554 @kindex --no-ignore-branch-isa
3555 @itemx --no-ignore-branch-isa
3556 These options control branch relocation checks for invalid ISA mode
3557 transitions. If @samp{--ignore-branch-isa} is used, then the linker
3558 accepts any branch relocations and any ISA mode transition required
3559 is lost in relocation calculation, except for some cases of @code{BAL}
3560 instructions which meet relaxation conditions and are converted to
3561 equivalent @code{JALX} instructions as the associated relocation is
3562 calculated. By default or if @samp{--no-ignore-branch-isa} is used
3563 a check is made causing the loss of an ISA mode transition to produce
3566 @kindex --compact-branches
3567 @item --compact-branches
3568 @kindex --no-compact-branches
3569 @itemx --no-compact-branches
3570 These options control the generation of compact instructions by the linker
3571 in the PLT entries for MIPS R6.
3580 @subsection Options specific to PDP11 targets
3582 @c man begin OPTIONS
3584 For the pdp11-aout target, three variants of the output format can be
3585 produced as selected by the following options. The default variant
3586 for pdp11-aout is the @samp{--omagic} option, whereas for other
3587 targets @samp{--nmagic} is the default. The @samp{--imagic} option is
3588 defined only for the pdp11-aout target, while the others are described
3589 here as they apply to the pdp11-aout target.
3598 Mark the output as @code{OMAGIC} (0407) in the @file{a.out} header to
3599 indicate that the text segment is not to be write-protected and
3600 shared. Since the text and data sections are both readable and
3601 writable, the data section is allocated immediately contiguous after
3602 the text segment. This is the oldest format for PDP11 executable
3603 programs and is the default for @command{ld} on PDP11 Unix systems
3604 from the beginning through 2.11BSD.
3611 Mark the output as @code{NMAGIC} (0410) in the @file{a.out} header to
3612 indicate that when the output file is executed, the text portion will
3613 be read-only and shareable among all processes executing the same
3614 file. This involves moving the data areas up to the first possible 8K
3615 byte page boundary following the end of the text. This option creates
3616 a @emph{pure executable} format.
3623 Mark the output as @code{IMAGIC} (0411) in the @file{a.out} header to
3624 indicate that when the output file is executed, the program text and
3625 data areas will be loaded into separate address spaces using the split
3626 instruction and data space feature of the memory management unit in
3627 larger models of the PDP11. This doubles the address space available
3628 to the program. The text segment is again pure, write-protected, and
3629 shareable. The only difference in the output format between this
3630 option and the others, besides the magic number, is that both the text
3631 and data sections start at location 0. The @samp{-z} option selected
3632 this format in 2.11BSD. This option creates a @emph{separate
3638 Equivalent to @samp{--nmagic} for pdp11-aout.
3647 @section Environment Variables
3649 @c man begin ENVIRONMENT
3651 You can change the behaviour of @command{ld} with the environment variables
3652 @ifclear SingleFormat
3655 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
3657 @ifclear SingleFormat
3659 @cindex default input format
3660 @code{GNUTARGET} determines the input-file object format if you don't
3661 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
3662 of the BFD names for an input format (@pxref{BFD}). If there is no
3663 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
3664 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
3665 attempts to discover the input format by examining binary input files;
3666 this method often succeeds, but there are potential ambiguities, since
3667 there is no method of ensuring that the magic number used to specify
3668 object-file formats is unique. However, the configuration procedure for
3669 BFD on each system places the conventional format for that system first
3670 in the search-list, so ambiguities are resolved in favor of convention.
3674 @cindex default emulation
3675 @cindex emulation, default
3676 @code{LDEMULATION} determines the default emulation if you don't use the
3677 @samp{-m} option. The emulation can affect various aspects of linker
3678 behaviour, particularly the default linker script. You can list the
3679 available emulations with the @samp{--verbose} or @samp{-V} options. If
3680 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3681 variable is not defined, the default emulation depends upon how the
3682 linker was configured.
3684 @kindex COLLECT_NO_DEMANGLE
3685 @cindex demangling, default
3686 Normally, the linker will default to demangling symbols. However, if
3687 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3688 default to not demangling symbols. This environment variable is used in
3689 a similar fashion by the @code{gcc} linker wrapper program. The default
3690 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3697 @chapter Linker Scripts
3700 @cindex linker scripts
3701 @cindex command files
3702 Every link is controlled by a @dfn{linker script}. This script is
3703 written in the linker command language.
3705 The main purpose of the linker script is to describe how the sections in
3706 the input files should be mapped into the output file, and to control
3707 the memory layout of the output file. Most linker scripts do nothing
3708 more than this. However, when necessary, the linker script can also
3709 direct the linker to perform many other operations, using the commands
3712 The linker always uses a linker script. If you do not supply one
3713 yourself, the linker will use a default script that is compiled into the
3714 linker executable. You can use the @samp{--verbose} command-line option
3715 to display the default linker script. Certain command-line options,
3716 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3718 You may supply your own linker script by using the @samp{-T} command
3719 line option. When you do this, your linker script will replace the
3720 default linker script.
3722 You may also use linker scripts implicitly by naming them as input files
3723 to the linker, as though they were files to be linked. @xref{Implicit
3727 * Basic Script Concepts:: Basic Linker Script Concepts
3728 * Script Format:: Linker Script Format
3729 * Simple Example:: Simple Linker Script Example
3730 * Simple Commands:: Simple Linker Script Commands
3731 * Assignments:: Assigning Values to Symbols
3732 * SECTIONS:: SECTIONS Command
3733 * MEMORY:: MEMORY Command
3734 * PHDRS:: PHDRS Command
3735 * VERSION:: VERSION Command
3736 * Expressions:: Expressions in Linker Scripts
3737 * Implicit Linker Scripts:: Implicit Linker Scripts
3740 @node Basic Script Concepts
3741 @section Basic Linker Script Concepts
3742 @cindex linker script concepts
3743 We need to define some basic concepts and vocabulary in order to
3744 describe the linker script language.
3746 The linker combines input files into a single output file. The output
3747 file and each input file are in a special data format known as an
3748 @dfn{object file format}. Each file is called an @dfn{object file}.
3749 The output file is often called an @dfn{executable}, but for our
3750 purposes we will also call it an object file. Each object file has,
3751 among other things, a list of @dfn{sections}. We sometimes refer to a
3752 section in an input file as an @dfn{input section}; similarly, a section
3753 in the output file is an @dfn{output section}.
3755 Each section in an object file has a name and a size. Most sections
3756 also have an associated block of data, known as the @dfn{section
3757 contents}. A section may be marked as @dfn{loadable}, which means that
3758 the contents should be loaded into memory when the output file is run.
3759 A section with no contents may be @dfn{allocatable}, which means that an
3760 area in memory should be set aside, but nothing in particular should be
3761 loaded there (in some cases this memory must be zeroed out). A section
3762 which is neither loadable nor allocatable typically contains some sort
3763 of debugging information.
3765 Every loadable or allocatable output section has two addresses. The
3766 first is the @dfn{VMA}, or virtual memory address. This is the address
3767 the section will have when the output file is run. The second is the
3768 @dfn{LMA}, or load memory address. This is the address at which the
3769 section will be loaded. In most cases the two addresses will be the
3770 same. An example of when they might be different is when a data section
3771 is loaded into ROM, and then copied into RAM when the program starts up
3772 (this technique is often used to initialize global variables in a ROM
3773 based system). In this case the ROM address would be the LMA, and the
3774 RAM address would be the VMA.
3776 You can see the sections in an object file by using the @code{objdump}
3777 program with the @samp{-h} option.
3779 Every object file also has a list of @dfn{symbols}, known as the
3780 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3781 has a name, and each defined symbol has an address, among other
3782 information. If you compile a C or C++ program into an object file, you
3783 will get a defined symbol for every defined function and global or
3784 static variable. Every undefined function or global variable which is
3785 referenced in the input file will become an undefined symbol.
3787 You can see the symbols in an object file by using the @code{nm}
3788 program, or by using the @code{objdump} program with the @samp{-t}
3792 @section Linker Script Format
3793 @cindex linker script format
3794 Linker scripts are text files.
3796 You write a linker script as a series of commands. Each command is
3797 either a keyword, possibly followed by arguments, or an assignment to a
3798 symbol. You may separate commands using semicolons. Whitespace is
3801 Strings such as file or format names can normally be entered directly.
3802 If the file name contains a character such as a comma which would
3803 otherwise serve to separate file names, you may put the file name in
3804 double quotes. There is no way to use a double quote character in a
3807 You may include comments in linker scripts just as in C, delimited by
3808 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3811 @node Simple Example
3812 @section Simple Linker Script Example
3813 @cindex linker script example
3814 @cindex example of linker script
3815 Many linker scripts are fairly simple.
3817 The simplest possible linker script has just one command:
3818 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3819 memory layout of the output file.
3821 The @samp{SECTIONS} command is a powerful command. Here we will
3822 describe a simple use of it. Let's assume your program consists only of
3823 code, initialized data, and uninitialized data. These will be in the
3824 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3825 Let's assume further that these are the only sections which appear in
3828 For this example, let's say that the code should be loaded at address
3829 0x10000, and that the data should start at address 0x8000000. Here is a
3830 linker script which will do that:
3835 .text : @{ *(.text) @}
3837 .data : @{ *(.data) @}
3838 .bss : @{ *(.bss) @}
3842 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3843 followed by a series of symbol assignments and output section
3844 descriptions enclosed in curly braces.
3846 The first line inside the @samp{SECTIONS} command of the above example
3847 sets the value of the special symbol @samp{.}, which is the location
3848 counter. If you do not specify the address of an output section in some
3849 other way (other ways are described later), the address is set from the
3850 current value of the location counter. The location counter is then
3851 incremented by the size of the output section. At the start of the
3852 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3854 The second line defines an output section, @samp{.text}. The colon is
3855 required syntax which may be ignored for now. Within the curly braces
3856 after the output section name, you list the names of the input sections
3857 which should be placed into this output section. The @samp{*} is a
3858 wildcard which matches any file name. The expression @samp{*(.text)}
3859 means all @samp{.text} input sections in all input files.
3861 Since the location counter is @samp{0x10000} when the output section
3862 @samp{.text} is defined, the linker will set the address of the
3863 @samp{.text} section in the output file to be @samp{0x10000}.
3865 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3866 the output file. The linker will place the @samp{.data} output section
3867 at address @samp{0x8000000}. After the linker places the @samp{.data}
3868 output section, the value of the location counter will be
3869 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3870 effect is that the linker will place the @samp{.bss} output section
3871 immediately after the @samp{.data} output section in memory.
3873 The linker will ensure that each output section has the required
3874 alignment, by increasing the location counter if necessary. In this
3875 example, the specified addresses for the @samp{.text} and @samp{.data}
3876 sections will probably satisfy any alignment constraints, but the linker
3877 may have to create a small gap between the @samp{.data} and @samp{.bss}
3880 That's it! That's a simple and complete linker script.
3882 @node Simple Commands
3883 @section Simple Linker Script Commands
3884 @cindex linker script simple commands
3885 In this section we describe the simple linker script commands.
3888 * Entry Point:: Setting the entry point
3889 * File Commands:: Commands dealing with files
3890 @ifclear SingleFormat
3891 * Format Commands:: Commands dealing with object file formats
3894 * REGION_ALIAS:: Assign alias names to memory regions
3895 * Miscellaneous Commands:: Other linker script commands
3899 @subsection Setting the Entry Point
3900 @kindex ENTRY(@var{symbol})
3901 @cindex start of execution
3902 @cindex first instruction
3904 The first instruction to execute in a program is called the @dfn{entry
3905 point}. You can use the @code{ENTRY} linker script command to set the
3906 entry point. The argument is a symbol name:
3911 There are several ways to set the entry point. The linker will set the
3912 entry point by trying each of the following methods in order, and
3913 stopping when one of them succeeds:
3916 the @samp{-e} @var{entry} command-line option;
3918 the @code{ENTRY(@var{symbol})} command in a linker script;
3920 the value of a target-specific symbol, if it is defined; For many
3921 targets this is @code{start}, but PE- and BeOS-based systems for example
3922 check a list of possible entry symbols, matching the first one found.
3924 the address of the first byte of the code section, if present and an
3925 executable is being created - the code section is usually
3926 @samp{.text}, but can be something else;
3928 The address @code{0}.
3932 @subsection Commands Dealing with Files
3933 @cindex linker script file commands
3934 Several linker script commands deal with files.
3937 @item INCLUDE @var{filename}
3938 @kindex INCLUDE @var{filename}
3939 @cindex including a linker script
3940 Include the linker script @var{filename} at this point. The file will
3941 be searched for in the current directory, and in any directory specified
3942 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3945 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3946 @code{SECTIONS} commands, or in output section descriptions.
3948 @item INPUT(@var{file}, @var{file}, @dots{})
3949 @itemx INPUT(@var{file} @var{file} @dots{})
3950 @kindex INPUT(@var{files})
3951 @cindex input files in linker scripts
3952 @cindex input object files in linker scripts
3953 @cindex linker script input object files
3954 The @code{INPUT} command directs the linker to include the named files
3955 in the link, as though they were named on the command line.
3957 For example, if you always want to include @file{subr.o} any time you do
3958 a link, but you can't be bothered to put it on every link command line,
3959 then you can put @samp{INPUT (subr.o)} in your linker script.
3961 In fact, if you like, you can list all of your input files in the linker
3962 script, and then invoke the linker with nothing but a @samp{-T} option.
3964 In case a @dfn{sysroot prefix} is configured, and the filename starts
3965 with the @samp{/} character, and the script being processed was
3966 located inside the @dfn{sysroot prefix}, the filename will be looked
3967 for in the @dfn{sysroot prefix}. The @dfn{sysroot prefix} can also be forced by specifying
3968 @code{=} as the first character in the filename path, or prefixing the
3969 filename path with @code{$SYSROOT}. See also the description of
3970 @samp{-L} in @ref{Options,,Command-line Options}.
3972 If a @dfn{sysroot prefix} is not used then the linker will try to open
3973 the file in the directory containing the linker script. If it is not
3974 found the linker will then search the current directory. If it is still
3975 not found the linker will search through the archive library search
3978 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3979 name to @code{lib@var{file}.a}, as with the command-line argument
3982 When you use the @code{INPUT} command in an implicit linker script, the
3983 files will be included in the link at the point at which the linker
3984 script file is included. This can affect archive searching.
3986 @item GROUP(@var{file}, @var{file}, @dots{})
3987 @itemx GROUP(@var{file} @var{file} @dots{})
3988 @kindex GROUP(@var{files})
3989 @cindex grouping input files
3990 The @code{GROUP} command is like @code{INPUT}, except that the named
3991 files should all be archives, and they are searched repeatedly until no
3992 new undefined references are created. See the description of @samp{-(}
3993 in @ref{Options,,Command-line Options}.
3995 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3996 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3997 @kindex AS_NEEDED(@var{files})
3998 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3999 commands, among other filenames. The files listed will be handled
4000 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
4001 with the exception of ELF shared libraries, that will be added only
4002 when they are actually needed. This construct essentially enables
4003 @option{--as-needed} option for all the files listed inside of it
4004 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
4007 @item OUTPUT(@var{filename})
4008 @kindex OUTPUT(@var{filename})
4009 @cindex output file name in linker script
4010 The @code{OUTPUT} command names the output file. Using
4011 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
4012 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
4013 Line Options}). If both are used, the command-line option takes
4016 You can use the @code{OUTPUT} command to define a default name for the
4017 output file other than the usual default of @file{a.out}.
4019 @item SEARCH_DIR(@var{path})
4020 @kindex SEARCH_DIR(@var{path})
4021 @cindex library search path in linker script
4022 @cindex archive search path in linker script
4023 @cindex search path in linker script
4024 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
4025 @command{ld} looks for archive libraries. Using
4026 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
4027 on the command line (@pxref{Options,,Command-line Options}). If both
4028 are used, then the linker will search both paths. Paths specified using
4029 the command-line option are searched first.
4031 @item STARTUP(@var{filename})
4032 @kindex STARTUP(@var{filename})
4033 @cindex first input file
4034 The @code{STARTUP} command is just like the @code{INPUT} command, except
4035 that @var{filename} will become the first input file to be linked, as
4036 though it were specified first on the command line. This may be useful
4037 when using a system in which the entry point is always the start of the
4041 @ifclear SingleFormat
4042 @node Format Commands
4043 @subsection Commands Dealing with Object File Formats
4044 A couple of linker script commands deal with object file formats.
4047 @item OUTPUT_FORMAT(@var{bfdname})
4048 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
4049 @kindex OUTPUT_FORMAT(@var{bfdname})
4050 @cindex output file format in linker script
4051 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
4052 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
4053 exactly like using @samp{--oformat @var{bfdname}} on the command line
4054 (@pxref{Options,,Command-line Options}). If both are used, the command
4055 line option takes precedence.
4057 You can use @code{OUTPUT_FORMAT} with three arguments to use different
4058 formats based on the @samp{-EB} and @samp{-EL} command-line options.
4059 This permits the linker script to set the output format based on the
4062 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
4063 will be the first argument, @var{default}. If @samp{-EB} is used, the
4064 output format will be the second argument, @var{big}. If @samp{-EL} is
4065 used, the output format will be the third argument, @var{little}.
4067 For example, the default linker script for the MIPS ELF target uses this
4070 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
4072 This says that the default format for the output file is
4073 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command-line
4074 option, the output file will be created in the @samp{elf32-littlemips}
4077 @item TARGET(@var{bfdname})
4078 @kindex TARGET(@var{bfdname})
4079 @cindex input file format in linker script
4080 The @code{TARGET} command names the BFD format to use when reading input
4081 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
4082 This command is like using @samp{-b @var{bfdname}} on the command line
4083 (@pxref{Options,,Command-line Options}). If the @code{TARGET} command
4084 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
4085 command is also used to set the format for the output file. @xref{BFD}.
4090 @subsection Assign alias names to memory regions
4091 @kindex REGION_ALIAS(@var{alias}, @var{region})
4092 @cindex region alias
4093 @cindex region names
4095 Alias names can be added to existing memory regions created with the
4096 @ref{MEMORY} command. Each name corresponds to at most one memory region.
4099 REGION_ALIAS(@var{alias}, @var{region})
4102 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
4103 memory region @var{region}. This allows a flexible mapping of output sections
4104 to memory regions. An example follows.
4106 Suppose we have an application for embedded systems which come with various
4107 memory storage devices. All have a general purpose, volatile memory @code{RAM}
4108 that allows code execution or data storage. Some may have a read-only,
4109 non-volatile memory @code{ROM} that allows code execution and read-only data
4110 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
4111 read-only data access and no code execution capability. We have four output
4116 @code{.text} program code;
4118 @code{.rodata} read-only data;
4120 @code{.data} read-write initialized data;
4122 @code{.bss} read-write zero initialized data.
4125 The goal is to provide a linker command file that contains a system independent
4126 part defining the output sections and a system dependent part mapping the
4127 output sections to the memory regions available on the system. Our embedded
4128 systems come with three different memory setups @code{A}, @code{B} and
4130 @multitable @columnfractions .25 .25 .25 .25
4131 @item Section @tab Variant A @tab Variant B @tab Variant C
4132 @item .text @tab RAM @tab ROM @tab ROM
4133 @item .rodata @tab RAM @tab ROM @tab ROM2
4134 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
4135 @item .bss @tab RAM @tab RAM @tab RAM
4137 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
4138 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
4139 the load address of the @code{.data} section starts in all three variants at
4140 the end of the @code{.rodata} section.
4142 The base linker script that deals with the output sections follows. It
4143 includes the system dependent @code{linkcmds.memory} file that describes the
4146 INCLUDE linkcmds.memory
4159 .data : AT (rodata_end)
4164 data_size = SIZEOF(.data);
4165 data_load_start = LOADADDR(.data);
4173 Now we need three different @code{linkcmds.memory} files to define memory
4174 regions and alias names. The content of @code{linkcmds.memory} for the three
4175 variants @code{A}, @code{B} and @code{C}:
4178 Here everything goes into the @code{RAM}.
4182 RAM : ORIGIN = 0, LENGTH = 4M
4185 REGION_ALIAS("REGION_TEXT", RAM);
4186 REGION_ALIAS("REGION_RODATA", RAM);
4187 REGION_ALIAS("REGION_DATA", RAM);
4188 REGION_ALIAS("REGION_BSS", RAM);
4191 Program code and read-only data go into the @code{ROM}. Read-write data goes
4192 into the @code{RAM}. An image of the initialized data is loaded into the
4193 @code{ROM} and will be copied during system start into the @code{RAM}.
4197 ROM : ORIGIN = 0, LENGTH = 3M
4198 RAM : ORIGIN = 0x10000000, LENGTH = 1M
4201 REGION_ALIAS("REGION_TEXT", ROM);
4202 REGION_ALIAS("REGION_RODATA", ROM);
4203 REGION_ALIAS("REGION_DATA", RAM);
4204 REGION_ALIAS("REGION_BSS", RAM);
4207 Program code goes into the @code{ROM}. Read-only data goes into the
4208 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
4209 initialized data is loaded into the @code{ROM2} and will be copied during
4210 system start into the @code{RAM}.
4214 ROM : ORIGIN = 0, LENGTH = 2M
4215 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
4216 RAM : ORIGIN = 0x20000000, LENGTH = 1M
4219 REGION_ALIAS("REGION_TEXT", ROM);
4220 REGION_ALIAS("REGION_RODATA", ROM2);
4221 REGION_ALIAS("REGION_DATA", RAM);
4222 REGION_ALIAS("REGION_BSS", RAM);
4226 It is possible to write a common system initialization routine to copy the
4227 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
4232 extern char data_start [];
4233 extern char data_size [];
4234 extern char data_load_start [];
4236 void copy_data(void)
4238 if (data_start != data_load_start)
4240 memcpy(data_start, data_load_start, (size_t) data_size);
4245 @node Miscellaneous Commands
4246 @subsection Other Linker Script Commands
4247 There are a few other linker scripts commands.
4250 @item ASSERT(@var{exp}, @var{message})
4252 @cindex assertion in linker script
4253 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
4254 with an error code, and print @var{message}.
4256 Note that assertions are checked before the final stages of linking
4257 take place. This means that expressions involving symbols PROVIDEd
4258 inside section definitions will fail if the user has not set values
4259 for those symbols. The only exception to this rule is PROVIDEd
4260 symbols that just reference dot. Thus an assertion like this:
4265 PROVIDE (__stack = .);
4266 PROVIDE (__stack_size = 0x100);
4267 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
4271 will fail if @code{__stack_size} is not defined elsewhere. Symbols
4272 PROVIDEd outside of section definitions are evaluated earlier, so they
4273 can be used inside ASSERTions. Thus:
4276 PROVIDE (__stack_size = 0x100);
4279 PROVIDE (__stack = .);
4280 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
4286 @item EXTERN(@var{symbol} @var{symbol} @dots{})
4288 @cindex undefined symbol in linker script
4289 Force @var{symbol} to be entered in the output file as an undefined
4290 symbol. Doing this may, for example, trigger linking of additional
4291 modules from standard libraries. You may list several @var{symbol}s for
4292 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
4293 command has the same effect as the @samp{-u} command-line option.
4295 @item FORCE_COMMON_ALLOCATION
4296 @kindex FORCE_COMMON_ALLOCATION
4297 @cindex common allocation in linker script
4298 This command has the same effect as the @samp{-d} command-line option:
4299 to make @command{ld} assign space to common symbols even if a relocatable
4300 output file is specified (@samp{-r}).
4302 @item INHIBIT_COMMON_ALLOCATION
4303 @kindex INHIBIT_COMMON_ALLOCATION
4304 @cindex common allocation in linker script
4305 This command has the same effect as the @samp{--no-define-common}
4306 command-line option: to make @code{ld} omit the assignment of addresses
4307 to common symbols even for a non-relocatable output file.
4309 @item FORCE_GROUP_ALLOCATION
4310 @kindex FORCE_GROUP_ALLOCATION
4311 @cindex group allocation in linker script
4312 @cindex section groups
4314 This command has the same effect as the
4315 @samp{--force-group-allocation} command-line option: to make
4316 @command{ld} place section group members like normal input sections,
4317 and to delete the section groups even if a relocatable output file is
4318 specified (@samp{-r}).
4320 @item INSERT [ AFTER | BEFORE ] @var{output_section}
4322 @cindex insert user script into default script
4323 This command is typically used in a script specified by @samp{-T} to
4324 augment the default @code{SECTIONS} with, for example, overlays. It
4325 inserts all prior linker script statements after (or before)
4326 @var{output_section}, and also causes @samp{-T} to not override the
4327 default linker script. The exact insertion point is as for orphan
4328 sections. @xref{Location Counter}. The insertion happens after the
4329 linker has mapped input sections to output sections. Prior to the
4330 insertion, since @samp{-T} scripts are parsed before the default
4331 linker script, statements in the @samp{-T} script occur before the
4332 default linker script statements in the internal linker representation
4333 of the script. In particular, input section assignments will be made
4334 to @samp{-T} output sections before those in the default script. Here
4335 is an example of how a @samp{-T} script using @code{INSERT} might look:
4342 .ov1 @{ ov1*(.text) @}
4343 .ov2 @{ ov2*(.text) @}
4349 @item NOCROSSREFS(@var{section} @var{section} @dots{})
4350 @kindex NOCROSSREFS(@var{sections})
4351 @cindex cross references
4352 This command may be used to tell @command{ld} to issue an error about any
4353 references among certain output sections.
4355 In certain types of programs, particularly on embedded systems when
4356 using overlays, when one section is loaded into memory, another section
4357 will not be. Any direct references between the two sections would be
4358 errors. For example, it would be an error if code in one section called
4359 a function defined in the other section.
4361 The @code{NOCROSSREFS} command takes a list of output section names. If
4362 @command{ld} detects any cross references between the sections, it reports
4363 an error and returns a non-zero exit status. Note that the
4364 @code{NOCROSSREFS} command uses output section names, not input section
4367 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
4368 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
4369 @cindex cross references
4370 This command may be used to tell @command{ld} to issue an error about any
4371 references to one section from a list of other sections.
4373 The @code{NOCROSSREFS} command is useful when ensuring that two or more
4374 output sections are entirely independent but there are situations where
4375 a one-way dependency is needed. For example, in a multi-core application
4376 there may be shared code that can be called from each core but for safety
4377 must never call back.
4379 The @code{NOCROSSREFS_TO} command takes a list of output section names.
4380 The first section can not be referenced from any of the other sections.
4381 If @command{ld} detects any references to the first section from any of
4382 the other sections, it reports an error and returns a non-zero exit
4383 status. Note that the @code{NOCROSSREFS_TO} command uses output section
4384 names, not input section names.
4386 @ifclear SingleFormat
4387 @item OUTPUT_ARCH(@var{bfdarch})
4388 @kindex OUTPUT_ARCH(@var{bfdarch})
4389 @cindex machine architecture
4390 @cindex architecture
4391 Specify a particular output machine architecture. The argument is one
4392 of the names used by the BFD library (@pxref{BFD}). You can see the
4393 architecture of an object file by using the @code{objdump} program with
4394 the @samp{-f} option.
4397 @item LD_FEATURE(@var{string})
4398 @kindex LD_FEATURE(@var{string})
4399 This command may be used to modify @command{ld} behavior. If
4400 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
4401 in a script are simply treated as numbers everywhere.
4402 @xref{Expression Section}.
4406 @section Assigning Values to Symbols
4407 @cindex assignment in scripts
4408 @cindex symbol definition, scripts
4409 @cindex variables, defining
4410 You may assign a value to a symbol in a linker script. This will define
4411 the symbol and place it into the symbol table with a global scope.
4414 * Simple Assignments:: Simple Assignments
4417 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
4418 * Source Code Reference:: How to use a linker script defined symbol in source code
4421 @node Simple Assignments
4422 @subsection Simple Assignments
4424 You may assign to a symbol using any of the C assignment operators:
4427 @item @var{symbol} = @var{expression} ;
4428 @itemx @var{symbol} += @var{expression} ;
4429 @itemx @var{symbol} -= @var{expression} ;
4430 @itemx @var{symbol} *= @var{expression} ;
4431 @itemx @var{symbol} /= @var{expression} ;
4432 @itemx @var{symbol} <<= @var{expression} ;
4433 @itemx @var{symbol} >>= @var{expression} ;
4434 @itemx @var{symbol} &= @var{expression} ;
4435 @itemx @var{symbol} |= @var{expression} ;
4438 The first case will define @var{symbol} to the value of
4439 @var{expression}. In the other cases, @var{symbol} must already be
4440 defined, and the value will be adjusted accordingly.
4442 The special symbol name @samp{.} indicates the location counter. You
4443 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
4445 The semicolon after @var{expression} is required.
4447 Expressions are defined below; see @ref{Expressions}.
4449 You may write symbol assignments as commands in their own right, or as
4450 statements within a @code{SECTIONS} command, or as part of an output
4451 section description in a @code{SECTIONS} command.
4453 The section of the symbol will be set from the section of the
4454 expression; for more information, see @ref{Expression Section}.
4456 Here is an example showing the three different places that symbol
4457 assignments may be used:
4468 _bdata = (. + 3) & ~ 3;
4469 .data : @{ *(.data) @}
4473 In this example, the symbol @samp{floating_point} will be defined as
4474 zero. The symbol @samp{_etext} will be defined as the address following
4475 the last @samp{.text} input section. The symbol @samp{_bdata} will be
4476 defined as the address following the @samp{.text} output section aligned
4477 upward to a 4 byte boundary.
4482 For ELF targeted ports, define a symbol that will be hidden and won't be
4483 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
4485 Here is the example from @ref{Simple Assignments}, rewritten to use
4489 HIDDEN(floating_point = 0);
4497 HIDDEN(_bdata = (. + 3) & ~ 3);
4498 .data : @{ *(.data) @}
4502 In this case none of the three symbols will be visible outside this module.
4507 In some cases, it is desirable for a linker script to define a symbol
4508 only if it is referenced and is not defined by any object included in
4509 the link. For example, traditional linkers defined the symbol
4510 @samp{etext}. However, ANSI C requires that the user be able to use
4511 @samp{etext} as a function name without encountering an error. The
4512 @code{PROVIDE} keyword may be used to define a symbol, such as
4513 @samp{etext}, only if it is referenced but not defined. The syntax is
4514 @code{PROVIDE(@var{symbol} = @var{expression})}.
4516 Here is an example of using @code{PROVIDE} to define @samp{etext}:
4529 In this example, if the program defines @samp{_etext} (with a leading
4530 underscore), the linker will give a multiple definition diagnostic. If,
4531 on the other hand, the program defines @samp{etext} (with no leading
4532 underscore), the linker will silently use the definition in the program.
4533 If the program references @samp{etext} but does not define it, the
4534 linker will use the definition in the linker script.
4536 Note - the @code{PROVIDE} directive considers a common symbol to be
4537 defined, even though such a symbol could be combined with the symbol
4538 that the @code{PROVIDE} would create. This is particularly important
4539 when considering constructor and destructor list symbols such as
4540 @samp{__CTOR_LIST__} as these are often defined as common symbols.
4542 @node PROVIDE_HIDDEN
4543 @subsection PROVIDE_HIDDEN
4544 @cindex PROVIDE_HIDDEN
4545 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
4546 hidden and won't be exported.
4548 @node Source Code Reference
4549 @subsection Source Code Reference
4551 Accessing a linker script defined variable from source code is not
4552 intuitive. In particular a linker script symbol is not equivalent to
4553 a variable declaration in a high level language, it is instead a
4554 symbol that does not have a value.
4556 Before going further, it is important to note that compilers often
4557 transform names in the source code into different names when they are
4558 stored in the symbol table. For example, Fortran compilers commonly
4559 prepend or append an underscore, and C++ performs extensive @samp{name
4560 mangling}. Therefore there might be a discrepancy between the name
4561 of a variable as it is used in source code and the name of the same
4562 variable as it is defined in a linker script. For example in C a
4563 linker script variable might be referred to as:
4569 But in the linker script it might be defined as:
4575 In the remaining examples however it is assumed that no name
4576 transformation has taken place.
4578 When a symbol is declared in a high level language such as C, two
4579 things happen. The first is that the compiler reserves enough space
4580 in the program's memory to hold the @emph{value} of the symbol. The
4581 second is that the compiler creates an entry in the program's symbol
4582 table which holds the symbol's @emph{address}. ie the symbol table
4583 contains the address of the block of memory holding the symbol's
4584 value. So for example the following C declaration, at file scope:
4590 creates an entry called @samp{foo} in the symbol table. This entry
4591 holds the address of an @samp{int} sized block of memory where the
4592 number 1000 is initially stored.
4594 When a program references a symbol the compiler generates code that
4595 first accesses the symbol table to find the address of the symbol's
4596 memory block and then code to read the value from that memory block.
4603 looks up the symbol @samp{foo} in the symbol table, gets the address
4604 associated with this symbol and then writes the value 1 into that
4611 looks up the symbol @samp{foo} in the symbol table, gets its address
4612 and then copies this address into the block of memory associated with
4613 the variable @samp{a}.
4615 Linker scripts symbol declarations, by contrast, create an entry in
4616 the symbol table but do not assign any memory to them. Thus they are
4617 an address without a value. So for example the linker script definition:
4623 creates an entry in the symbol table called @samp{foo} which holds
4624 the address of memory location 1000, but nothing special is stored at
4625 address 1000. This means that you cannot access the @emph{value} of a
4626 linker script defined symbol - it has no value - all you can do is
4627 access the @emph{address} of a linker script defined symbol.
4629 Hence when you are using a linker script defined symbol in source code
4630 you should always take the address of the symbol, and never attempt to
4631 use its value. For example suppose you want to copy the contents of a
4632 section of memory called .ROM into a section called .FLASH and the
4633 linker script contains these declarations:
4637 start_of_ROM = .ROM;
4638 end_of_ROM = .ROM + sizeof (.ROM);
4639 start_of_FLASH = .FLASH;
4643 Then the C source code to perform the copy would be:
4647 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
4649 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
4653 Note the use of the @samp{&} operators. These are correct.
4654 Alternatively the symbols can be treated as the names of vectors or
4655 arrays and then the code will again work as expected:
4659 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
4661 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
4665 Note how using this method does not require the use of @samp{&}
4669 @section SECTIONS Command
4671 The @code{SECTIONS} command tells the linker how to map input sections
4672 into output sections, and how to place the output sections in memory.
4674 The format of the @code{SECTIONS} command is:
4678 @var{sections-command}
4679 @var{sections-command}
4684 Each @var{sections-command} may of be one of the following:
4688 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
4690 a symbol assignment (@pxref{Assignments})
4692 an output section description
4694 an overlay description
4697 The @code{ENTRY} command and symbol assignments are permitted inside the
4698 @code{SECTIONS} command for convenience in using the location counter in
4699 those commands. This can also make the linker script easier to
4700 understand because you can use those commands at meaningful points in
4701 the layout of the output file.
4703 Output section descriptions and overlay descriptions are described
4706 If you do not use a @code{SECTIONS} command in your linker script, the
4707 linker will place each input section into an identically named output
4708 section in the order that the sections are first encountered in the
4709 input files. If all input sections are present in the first file, for
4710 example, the order of sections in the output file will match the order
4711 in the first input file. The first section will be at address zero.
4714 * Output Section Description:: Output section description
4715 * Output Section Name:: Output section name
4716 * Output Section Address:: Output section address
4717 * Input Section:: Input section description
4718 * Output Section Data:: Output section data
4719 * Output Section Keywords:: Output section keywords
4720 * Output Section Discarding:: Output section discarding
4721 * Output Section Attributes:: Output section attributes
4722 * Overlay Description:: Overlay description
4725 @node Output Section Description
4726 @subsection Output Section Description
4727 The full description of an output section looks like this:
4730 @var{section} [@var{address}] [(@var{type})] :
4732 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4733 [SUBALIGN(@var{subsection_align})]
4736 @var{output-section-command}
4737 @var{output-section-command}
4739 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4743 Most output sections do not use most of the optional section attributes.
4745 The whitespace around @var{section} is required, so that the section
4746 name is unambiguous. The colon and the curly braces are also required.
4747 The comma at the end may be required if a @var{fillexp} is used and
4748 the next @var{sections-command} looks like a continuation of the expression.
4749 The line breaks and other white space are optional.
4751 Each @var{output-section-command} may be one of the following:
4755 a symbol assignment (@pxref{Assignments})
4757 an input section description (@pxref{Input Section})
4759 data values to include directly (@pxref{Output Section Data})
4761 a special output section keyword (@pxref{Output Section Keywords})
4764 @node Output Section Name
4765 @subsection Output Section Name
4766 @cindex name, section
4767 @cindex section name
4768 The name of the output section is @var{section}. @var{section} must
4769 meet the constraints of your output format. In formats which only
4770 support a limited number of sections, such as @code{a.out}, the name
4771 must be one of the names supported by the format (@code{a.out}, for
4772 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4773 output format supports any number of sections, but with numbers and not
4774 names (as is the case for Oasys), the name should be supplied as a
4775 quoted numeric string. A section name may consist of any sequence of
4776 characters, but a name which contains any unusual characters such as
4777 commas must be quoted.
4779 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4782 @node Output Section Address
4783 @subsection Output Section Address
4784 @cindex address, section
4785 @cindex section address
4786 The @var{address} is an expression for the VMA (the virtual memory
4787 address) of the output section. This address is optional, but if it
4788 is provided then the output address will be set exactly as specified.
4790 If the output address is not specified then one will be chosen for the
4791 section, based on the heuristic below. This address will be adjusted
4792 to fit the alignment requirement of the output section. The
4793 alignment requirement is the strictest alignment of any input section
4794 contained within the output section.
4796 The output section address heuristic is as follows:
4800 If an output memory @var{region} is set for the section then it
4801 is added to this region and its address will be the next free address
4805 If the MEMORY command has been used to create a list of memory
4806 regions then the first region which has attributes compatible with the
4807 section is selected to contain it. The section's output address will
4808 be the next free address in that region; @ref{MEMORY}.
4811 If no memory regions were specified, or none match the section then
4812 the output address will be based on the current value of the location
4820 .text . : @{ *(.text) @}
4827 .text : @{ *(.text) @}
4831 are subtly different. The first will set the address of the
4832 @samp{.text} output section to the current value of the location
4833 counter. The second will set it to the current value of the location
4834 counter aligned to the strictest alignment of any of the @samp{.text}
4837 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4838 For example, if you want to align the section on a 0x10 byte boundary,
4839 so that the lowest four bits of the section address are zero, you could
4840 do something like this:
4842 .text ALIGN(0x10) : @{ *(.text) @}
4845 This works because @code{ALIGN} returns the current location counter
4846 aligned upward to the specified value.
4848 Specifying @var{address} for a section will change the value of the
4849 location counter, provided that the section is non-empty. (Empty
4850 sections are ignored).
4853 @subsection Input Section Description
4854 @cindex input sections
4855 @cindex mapping input sections to output sections
4856 The most common output section command is an input section description.
4858 The input section description is the most basic linker script operation.
4859 You use output sections to tell the linker how to lay out your program
4860 in memory. You use input section descriptions to tell the linker how to
4861 map the input files into your memory layout.
4864 * Input Section Basics:: Input section basics
4865 * Input Section Wildcards:: Input section wildcard patterns
4866 * Input Section Common:: Input section for common symbols
4867 * Input Section Keep:: Input section and garbage collection
4868 * Input Section Example:: Input section example
4871 @node Input Section Basics
4872 @subsubsection Input Section Basics
4873 @cindex input section basics
4874 An input section description consists of a file name optionally followed
4875 by a list of section names in parentheses.
4877 The file name and the section name may be wildcard patterns, which we
4878 describe further below (@pxref{Input Section Wildcards}).
4880 The most common input section description is to include all input
4881 sections with a particular name in the output section. For example, to
4882 include all input @samp{.text} sections, you would write:
4887 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4888 @cindex EXCLUDE_FILE
4889 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4890 match all files except the ones specified in the EXCLUDE_FILE list. For
4893 EXCLUDE_FILE (*crtend.o *otherfile.o) *(.ctors)
4896 will cause all .ctors sections from all files except @file{crtend.o}
4897 and @file{otherfile.o} to be included. The EXCLUDE_FILE can also be
4898 placed inside the section list, for example:
4900 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4903 The result of this is identically to the previous example. Supporting
4904 two syntaxes for EXCLUDE_FILE is useful if the section list contains
4905 more than one section, as described below.
4907 There are two ways to include more than one section:
4913 The difference between these is the order in which the @samp{.text} and
4914 @samp{.rdata} input sections will appear in the output section. In the
4915 first example, they will be intermingled, appearing in the same order as
4916 they are found in the linker input. In the second example, all
4917 @samp{.text} input sections will appear first, followed by all
4918 @samp{.rdata} input sections.
4920 When using EXCLUDE_FILE with more than one section, if the exclusion
4921 is within the section list then the exclusion only applies to the
4922 immediately following section, for example:
4924 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4927 will cause all @samp{.text} sections from all files except
4928 @file{somefile.o} to be included, while all @samp{.rdata} sections
4929 from all files, including @file{somefile.o}, will be included. To
4930 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4931 could be modified to:
4933 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4936 Alternatively, placing the EXCLUDE_FILE outside of the section list,
4937 before the input file selection, will cause the exclusion to apply for
4938 all sections. Thus the previous example can be rewritten as:
4940 EXCLUDE_FILE (*somefile.o) *(.text .rdata)
4943 You can specify a file name to include sections from a particular file.
4944 You would do this if one or more of your files contain special data that
4945 needs to be at a particular location in memory. For example:
4950 To refine the sections that are included based on the section flags
4951 of an input section, INPUT_SECTION_FLAGS may be used.
4953 Here is a simple example for using Section header flags for ELF sections:
4958 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4959 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4964 In this example, the output section @samp{.text} will be comprised of any
4965 input section matching the name *(.text) whose section header flags
4966 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4967 @samp{.text2} will be comprised of any input section matching the name *(.text)
4968 whose section header flag @code{SHF_WRITE} is clear.
4970 You can also specify files within archives by writing a pattern
4971 matching the archive, a colon, then the pattern matching the file,
4972 with no whitespace around the colon.
4976 matches file within archive
4978 matches the whole archive
4980 matches file but not one in an archive
4983 Either one or both of @samp{archive} and @samp{file} can contain shell
4984 wildcards. On DOS based file systems, the linker will assume that a
4985 single letter followed by a colon is a drive specifier, so
4986 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4987 within an archive called @samp{c}. @samp{archive:file} filespecs may
4988 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4989 other linker script contexts. For instance, you cannot extract a file
4990 from an archive by using @samp{archive:file} in an @code{INPUT}
4993 If you use a file name without a list of sections, then all sections in
4994 the input file will be included in the output section. This is not
4995 commonly done, but it may by useful on occasion. For example:
5000 When you use a file name which is not an @samp{archive:file} specifier
5001 and does not contain any wild card
5002 characters, the linker will first see if you also specified the file
5003 name on the linker command line or in an @code{INPUT} command. If you
5004 did not, the linker will attempt to open the file as an input file, as
5005 though it appeared on the command line. Note that this differs from an
5006 @code{INPUT} command, because the linker will not search for the file in
5007 the archive search path.
5009 @node Input Section Wildcards
5010 @subsubsection Input Section Wildcard Patterns
5011 @cindex input section wildcards
5012 @cindex wildcard file name patterns
5013 @cindex file name wildcard patterns
5014 @cindex section name wildcard patterns
5015 In an input section description, either the file name or the section
5016 name or both may be wildcard patterns.
5018 The file name of @samp{*} seen in many examples is a simple wildcard
5019 pattern for the file name.
5021 The wildcard patterns are like those used by the Unix shell.
5025 matches any number of characters
5027 matches any single character
5029 matches a single instance of any of the @var{chars}; the @samp{-}
5030 character may be used to specify a range of characters, as in
5031 @samp{[a-z]} to match any lower case letter
5033 quotes the following character
5036 File name wildcard patterns only match files which are explicitly
5037 specified on the command line or in an @code{INPUT} command. The linker
5038 does not search directories to expand wildcards.
5040 If a file name matches more than one wildcard pattern, or if a file name
5041 appears explicitly and is also matched by a wildcard pattern, the linker
5042 will use the first match in the linker script. For example, this
5043 sequence of input section descriptions is probably in error, because the
5044 @file{data.o} rule will not be used:
5046 .data : @{ *(.data) @}
5047 .data1 : @{ data.o(.data) @}
5050 @cindex SORT_BY_NAME
5051 Normally, the linker will place files and sections matched by wildcards
5052 in the order in which they are seen during the link. You can change
5053 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
5054 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
5055 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
5056 into ascending order by name before placing them in the output file.
5058 @cindex SORT_BY_ALIGNMENT
5059 @code{SORT_BY_ALIGNMENT} is similar to @code{SORT_BY_NAME}.
5060 @code{SORT_BY_ALIGNMENT} will sort sections into descending order of
5061 alignment before placing them in the output file. Placing larger
5062 alignments before smaller alignments can reduce the amount of padding
5065 @cindex SORT_BY_INIT_PRIORITY
5066 @code{SORT_BY_INIT_PRIORITY} is also similar to @code{SORT_BY_NAME}.
5067 @code{SORT_BY_INIT_PRIORITY} will sort sections into ascending
5068 numerical order of the GCC init_priority attribute encoded in the
5069 section name before placing them in the output file. In
5070 @code{.init_array.NNNNN} and @code{.fini_array.NNNNN}, @code{NNNNN} is
5071 the init_priority. In @code{.ctors.NNNNN} and @code{.dtors.NNNNN},
5072 @code{NNNNN} is 65535 minus the init_priority.
5075 @code{SORT} is an alias for @code{SORT_BY_NAME}.
5077 When there are nested section sorting commands in linker script, there
5078 can be at most 1 level of nesting for section sorting commands.
5082 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
5083 It will sort the input sections by name first, then by alignment if two
5084 sections have the same name.
5086 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
5087 It will sort the input sections by alignment first, then by name if two
5088 sections have the same alignment.
5090 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
5091 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
5093 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
5094 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
5096 All other nested section sorting commands are invalid.
5099 When both command-line section sorting option and linker script
5100 section sorting command are used, section sorting command always
5101 takes precedence over the command-line option.
5103 If the section sorting command in linker script isn't nested, the
5104 command-line option will make the section sorting command to be
5105 treated as nested sorting command.
5109 @code{SORT_BY_NAME} (wildcard section pattern ) with
5110 @option{--sort-sections alignment} is equivalent to
5111 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
5113 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
5114 @option{--sort-section name} is equivalent to
5115 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
5118 If the section sorting command in linker script is nested, the
5119 command-line option will be ignored.
5122 @code{SORT_NONE} disables section sorting by ignoring the command-line
5123 section sorting option.
5125 If you ever get confused about where input sections are going, use the
5126 @samp{-M} linker option to generate a map file. The map file shows
5127 precisely how input sections are mapped to output sections.
5129 This example shows how wildcard patterns might be used to partition
5130 files. This linker script directs the linker to place all @samp{.text}
5131 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
5132 The linker will place the @samp{.data} section from all files beginning
5133 with an upper case character in @samp{.DATA}; for all other files, the
5134 linker will place the @samp{.data} section in @samp{.data}.
5138 .text : @{ *(.text) @}
5139 .DATA : @{ [A-Z]*(.data) @}
5140 .data : @{ *(.data) @}
5141 .bss : @{ *(.bss) @}
5146 @node Input Section Common
5147 @subsubsection Input Section for Common Symbols
5148 @cindex common symbol placement
5149 @cindex uninitialized data placement
5150 A special notation is needed for common symbols, because in many object
5151 file formats common symbols do not have a particular input section. The
5152 linker treats common symbols as though they are in an input section
5153 named @samp{COMMON}.
5155 You may use file names with the @samp{COMMON} section just as with any
5156 other input sections. You can use this to place common symbols from a
5157 particular input file in one section while common symbols from other
5158 input files are placed in another section.
5160 In most cases, common symbols in input files will be placed in the
5161 @samp{.bss} section in the output file. For example:
5163 .bss @{ *(.bss) *(COMMON) @}
5166 @cindex scommon section
5167 @cindex small common symbols
5168 Some object file formats have more than one type of common symbol. For
5169 example, the MIPS ELF object file format distinguishes standard common
5170 symbols and small common symbols. In this case, the linker will use a
5171 different special section name for other types of common symbols. In
5172 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
5173 symbols and @samp{.scommon} for small common symbols. This permits you
5174 to map the different types of common symbols into memory at different
5178 You will sometimes see @samp{[COMMON]} in old linker scripts. This
5179 notation is now considered obsolete. It is equivalent to
5182 @node Input Section Keep
5183 @subsubsection Input Section and Garbage Collection
5185 @cindex garbage collection
5186 When link-time garbage collection is in use (@samp{--gc-sections}),
5187 it is often useful to mark sections that should not be eliminated.
5188 This is accomplished by surrounding an input section's wildcard entry
5189 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
5190 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
5192 @node Input Section Example
5193 @subsubsection Input Section Example
5194 The following example is a complete linker script. It tells the linker
5195 to read all of the sections from file @file{all.o} and place them at the
5196 start of output section @samp{outputa} which starts at location
5197 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
5198 follows immediately, in the same output section. All of section
5199 @samp{.input2} from @file{foo.o} goes into output section
5200 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
5201 All of the remaining @samp{.input1} and @samp{.input2} sections from any
5202 files are written to output section @samp{outputc}.
5230 If an output section's name is the same as the input section's name
5231 and is representable as a C identifier, then the linker will
5232 automatically @pxref{PROVIDE} two symbols: __start_SECNAME and
5233 __stop_SECNAME, where SECNAME is the name of the section. These
5234 indicate the start address and end address of the output section
5235 respectively. Note: most section names are not representable as
5236 C identifiers because they contain a @samp{.} character.
5238 @node Output Section Data
5239 @subsection Output Section Data
5241 @cindex section data
5242 @cindex output section data
5243 @kindex BYTE(@var{expression})
5244 @kindex SHORT(@var{expression})
5245 @kindex LONG(@var{expression})
5246 @kindex QUAD(@var{expression})
5247 @kindex SQUAD(@var{expression})
5248 You can include explicit bytes of data in an output section by using
5249 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
5250 an output section command. Each keyword is followed by an expression in
5251 parentheses providing the value to store (@pxref{Expressions}). The
5252 value of the expression is stored at the current value of the location
5255 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
5256 store one, two, four, and eight bytes (respectively). After storing the
5257 bytes, the location counter is incremented by the number of bytes
5260 For example, this will store the byte 1 followed by the four byte value
5261 of the symbol @samp{addr}:
5267 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
5268 same; they both store an 8 byte, or 64 bit, value. When both host and
5269 target are 32 bits, an expression is computed as 32 bits. In this case
5270 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
5271 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
5273 If the object file format of the output file has an explicit endianness,
5274 which is the normal case, the value will be stored in that endianness.
5275 When the object file format does not have an explicit endianness, as is
5276 true of, for example, S-records, the value will be stored in the
5277 endianness of the first input object file.
5279 Note---these commands only work inside a section description and not
5280 between them, so the following will produce an error from the linker:
5282 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
5284 whereas this will work:
5286 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
5289 @kindex FILL(@var{expression})
5290 @cindex holes, filling
5291 @cindex unspecified memory
5292 You may use the @code{FILL} command to set the fill pattern for the
5293 current section. It is followed by an expression in parentheses. Any
5294 otherwise unspecified regions of memory within the section (for example,
5295 gaps left due to the required alignment of input sections) are filled
5296 with the value of the expression, repeated as
5297 necessary. A @code{FILL} statement covers memory locations after the
5298 point at which it occurs in the section definition; by including more
5299 than one @code{FILL} statement, you can have different fill patterns in
5300 different parts of an output section.
5302 This example shows how to fill unspecified regions of memory with the
5308 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
5309 section attribute, but it only affects the
5310 part of the section following the @code{FILL} command, rather than the
5311 entire section. If both are used, the @code{FILL} command takes
5312 precedence. @xref{Output Section Fill}, for details on the fill
5315 @node Output Section Keywords
5316 @subsection Output Section Keywords
5317 There are a couple of keywords which can appear as output section
5321 @kindex CREATE_OBJECT_SYMBOLS
5322 @cindex input filename symbols
5323 @cindex filename symbols
5324 @item CREATE_OBJECT_SYMBOLS
5325 The command tells the linker to create a symbol for each input file.
5326 The name of each symbol will be the name of the corresponding input
5327 file. The section of each symbol will be the output section in which
5328 the @code{CREATE_OBJECT_SYMBOLS} command appears.
5330 This is conventional for the a.out object file format. It is not
5331 normally used for any other object file format.
5333 @kindex CONSTRUCTORS
5334 @cindex C++ constructors, arranging in link
5335 @cindex constructors, arranging in link
5337 When linking using the a.out object file format, the linker uses an
5338 unusual set construct to support C++ global constructors and
5339 destructors. When linking object file formats which do not support
5340 arbitrary sections, such as ECOFF and XCOFF, the linker will
5341 automatically recognize C++ global constructors and destructors by name.
5342 For these object file formats, the @code{CONSTRUCTORS} command tells the
5343 linker to place constructor information in the output section where the
5344 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
5345 ignored for other object file formats.
5347 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
5348 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
5349 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
5350 the start and end of the global destructors. The
5351 first word in the list is the number of entries, followed by the address
5352 of each constructor or destructor, followed by a zero word. The
5353 compiler must arrange to actually run the code. For these object file
5354 formats @sc{gnu} C++ normally calls constructors from a subroutine
5355 @code{__main}; a call to @code{__main} is automatically inserted into
5356 the startup code for @code{main}. @sc{gnu} C++ normally runs
5357 destructors either by using @code{atexit}, or directly from the function
5360 For object file formats such as @code{COFF} or @code{ELF} which support
5361 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
5362 addresses of global constructors and destructors into the @code{.ctors}
5363 and @code{.dtors} sections. Placing the following sequence into your
5364 linker script will build the sort of table which the @sc{gnu} C++
5365 runtime code expects to see.
5369 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
5374 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
5380 If you are using the @sc{gnu} C++ support for initialization priority,
5381 which provides some control over the order in which global constructors
5382 are run, you must sort the constructors at link time to ensure that they
5383 are executed in the correct order. When using the @code{CONSTRUCTORS}
5384 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
5385 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
5386 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
5389 Normally the compiler and linker will handle these issues automatically,
5390 and you will not need to concern yourself with them. However, you may
5391 need to consider this if you are using C++ and writing your own linker
5396 @node Output Section Discarding
5397 @subsection Output Section Discarding
5398 @cindex discarding sections
5399 @cindex sections, discarding
5400 @cindex removing sections
5401 The linker will not normally create output sections with no contents.
5402 This is for convenience when referring to input sections that may or
5403 may not be present in any of the input files. For example:
5405 .foo : @{ *(.foo) @}
5408 will only create a @samp{.foo} section in the output file if there is a
5409 @samp{.foo} section in at least one input file, and if the input
5410 sections are not all empty. Other link script directives that allocate
5411 space in an output section will also create the output section. So
5412 too will assignments to dot even if the assignment does not create
5413 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
5414 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
5415 @samp{sym} is an absolute symbol of value 0 defined in the script.
5416 This allows you to force output of an empty section with @samp{. = .}.
5418 The linker will ignore address assignments (@pxref{Output Section Address})
5419 on discarded output sections, except when the linker script defines
5420 symbols in the output section. In that case the linker will obey
5421 the address assignments, possibly advancing dot even though the
5422 section is discarded.
5425 The special output section name @samp{/DISCARD/} may be used to discard
5426 input sections. Any input sections which are assigned to an output
5427 section named @samp{/DISCARD/} are not included in the output file.
5429 This can be used to discard input sections marked with the ELF flag
5430 @code{SHF_GNU_RETAIN}, which would otherwise have been saved from linker
5433 Note, sections that match the @samp{/DISCARD/} output section will be
5434 discarded even if they are in an ELF section group which has other
5435 members which are not being discarded. This is deliberate.
5436 Discarding takes precedence over grouping.
5438 @node Output Section Attributes
5439 @subsection Output Section Attributes
5440 @cindex output section attributes
5441 We showed above that the full description of an output section looked
5446 @var{section} [@var{address}] [(@var{type})] :
5448 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
5449 [SUBALIGN(@var{subsection_align})]
5452 @var{output-section-command}
5453 @var{output-section-command}
5455 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
5459 We've already described @var{section}, @var{address}, and
5460 @var{output-section-command}. In this section we will describe the
5461 remaining section attributes.
5464 * Output Section Type:: Output section type
5465 * Output Section LMA:: Output section LMA
5466 * Forced Output Alignment:: Forced Output Alignment
5467 * Forced Input Alignment:: Forced Input Alignment
5468 * Output Section Constraint:: Output section constraint
5469 * Output Section Region:: Output section region
5470 * Output Section Phdr:: Output section phdr
5471 * Output Section Fill:: Output section fill
5474 @node Output Section Type
5475 @subsubsection Output Section Type
5476 Each output section may have a type. The type is a keyword in
5477 parentheses. The following types are defined:
5481 The section should be marked as not loadable, so that it will not be
5482 loaded into memory when the program is run.
5484 The section should be marked as read-only.
5489 These type names are supported for backward compatibility, and are
5490 rarely used. They all have the same effect: the section should be
5491 marked as not allocatable, so that no memory is allocated for the
5492 section when the program is run.
5493 @item TYPE = @var{type}
5494 Set the section type to the integer @var{type}. When generating an ELF
5495 output file, type names @code{SHT_PROGBITS}, @code{SHT_STRTAB},
5496 @code{SHT_NOTE}, @code{SHT_NOBITS}, @code{SHT_INIT_ARRAY},
5497 @code{SHT_FINI_ARRAY}, and @code{SHT_PREINIT_ARRAY} are also allowed
5498 for @var{type}. It is the user's responsibility to ensure that any
5499 special requirements of the section type are met.
5500 @item READONLY ( TYPE = @var{type} )
5501 This form of the syntax combines the @var{READONLY} type with the
5502 type specified by @var{type}.
5506 @cindex prevent unnecessary loading
5507 @cindex loading, preventing
5508 The linker normally sets the attributes of an output section based on
5509 the input sections which map into it. You can override this by using
5510 the section type. For example, in the script sample below, the
5511 @samp{ROM} section is addressed at memory location @samp{0} and does not
5512 need to be loaded when the program is run.
5516 ROM 0 (NOLOAD) : @{ @dots{} @}
5522 @node Output Section LMA
5523 @subsubsection Output Section LMA
5524 @kindex AT>@var{lma_region}
5525 @kindex AT(@var{lma})
5526 @cindex load address
5527 @cindex section load address
5528 Every section has a virtual address (VMA) and a load address (LMA); see
5529 @ref{Basic Script Concepts}. The virtual address is specified by the
5530 @pxref{Output Section Address} described earlier. The load address is
5531 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
5532 address is optional.
5534 The @code{AT} keyword takes an expression as an argument. This
5535 specifies the exact load address of the section. The @code{AT>} keyword
5536 takes the name of a memory region as an argument. @xref{MEMORY}. The
5537 load address of the section is set to the next free address in the
5538 region, aligned to the section's alignment requirements.
5540 If neither @code{AT} nor @code{AT>} is specified for an allocatable
5541 section, the linker will use the following heuristic to determine the
5546 If the section has a specific VMA address, then this is used as
5547 the LMA address as well.
5550 If the section is not allocatable then its LMA is set to its VMA.
5553 Otherwise if a memory region can be found that is compatible
5554 with the current section, and this region contains at least one
5555 section, then the LMA is set so the difference between the
5556 VMA and LMA is the same as the difference between the VMA and LMA of
5557 the last section in the located region.
5560 If no memory regions have been declared then a default region
5561 that covers the entire address space is used in the previous step.
5564 If no suitable region could be found, or there was no previous
5565 section then the LMA is set equal to the VMA.
5568 @cindex ROM initialized data
5569 @cindex initialized data in ROM
5570 This feature is designed to make it easy to build a ROM image. For
5571 example, the following linker script creates three output sections: one
5572 called @samp{.text}, which starts at @code{0x1000}, one called
5573 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
5574 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
5575 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
5576 defined with the value @code{0x2000}, which shows that the location
5577 counter holds the VMA value, not the LMA value.
5583 .text 0x1000 : @{ *(.text) _etext = . ; @}
5585 AT ( ADDR (.text) + SIZEOF (.text) )
5586 @{ _data = . ; *(.data); _edata = . ; @}
5588 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
5593 The run-time initialization code for use with a program generated with
5594 this linker script would include something like the following, to copy
5595 the initialized data from the ROM image to its runtime address. Notice
5596 how this code takes advantage of the symbols defined by the linker
5601 extern char _etext, _data, _edata, _bstart, _bend;
5602 char *src = &_etext;
5605 /* ROM has data at end of text; copy it. */
5606 while (dst < &_edata)
5610 for (dst = &_bstart; dst< &_bend; dst++)
5615 @node Forced Output Alignment
5616 @subsubsection Forced Output Alignment
5617 @kindex ALIGN(@var{section_align})
5618 @cindex forcing output section alignment
5619 @cindex output section alignment
5620 You can increase an output section's alignment by using ALIGN. As an
5621 alternative you can enforce that the difference between the VMA and LMA remains
5622 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
5624 @node Forced Input Alignment
5625 @subsubsection Forced Input Alignment
5626 @kindex SUBALIGN(@var{subsection_align})
5627 @cindex forcing input section alignment
5628 @cindex input section alignment
5629 You can force input section alignment within an output section by using
5630 SUBALIGN. The value specified overrides any alignment given by input
5631 sections, whether larger or smaller.
5633 @node Output Section Constraint
5634 @subsubsection Output Section Constraint
5637 @cindex constraints on output sections
5638 You can specify that an output section should only be created if all
5639 of its input sections are read-only or all of its input sections are
5640 read-write by using the keyword @code{ONLY_IF_RO} and
5641 @code{ONLY_IF_RW} respectively.
5643 @node Output Section Region
5644 @subsubsection Output Section Region
5645 @kindex >@var{region}
5646 @cindex section, assigning to memory region
5647 @cindex memory regions and sections
5648 You can assign a section to a previously defined region of memory by
5649 using @samp{>@var{region}}. @xref{MEMORY}.
5651 Here is a simple example:
5654 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
5655 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
5659 @node Output Section Phdr
5660 @subsubsection Output Section Phdr
5662 @cindex section, assigning to program header
5663 @cindex program headers and sections
5664 You can assign a section to a previously defined program segment by
5665 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
5666 one or more segments, then all subsequent allocated sections will be
5667 assigned to those segments as well, unless they use an explicitly
5668 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
5669 linker to not put the section in any segment at all.
5671 Here is a simple example:
5674 PHDRS @{ text PT_LOAD ; @}
5675 SECTIONS @{ .text : @{ *(.text) @} :text @}
5679 @node Output Section Fill
5680 @subsubsection Output Section Fill
5681 @kindex =@var{fillexp}
5682 @cindex section fill pattern
5683 @cindex fill pattern, entire section
5684 You can set the fill pattern for an entire section by using
5685 @samp{=@var{fillexp}}. @var{fillexp} is an expression
5686 (@pxref{Expressions}). Any otherwise unspecified regions of memory
5687 within the output section (for example, gaps left due to the required
5688 alignment of input sections) will be filled with the value, repeated as
5689 necessary. If the fill expression is a simple hex number, ie. a string
5690 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
5691 an arbitrarily long sequence of hex digits can be used to specify the
5692 fill pattern; Leading zeros become part of the pattern too. For all
5693 other cases, including extra parentheses or a unary @code{+}, the fill
5694 pattern is the four least significant bytes of the value of the
5695 expression. In all cases, the number is big-endian.
5697 You can also change the fill value with a @code{FILL} command in the
5698 output section commands; (@pxref{Output Section Data}).
5700 Here is a simple example:
5703 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
5707 @node Overlay Description
5708 @subsection Overlay Description
5711 An overlay description provides an easy way to describe sections which
5712 are to be loaded as part of a single memory image but are to be run at
5713 the same memory address. At run time, some sort of overlay manager will
5714 copy the overlaid sections in and out of the runtime memory address as
5715 required, perhaps by simply manipulating addressing bits. This approach
5716 can be useful, for example, when a certain region of memory is faster
5719 Overlays are described using the @code{OVERLAY} command. The
5720 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
5721 output section description. The full syntax of the @code{OVERLAY}
5722 command is as follows:
5725 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
5729 @var{output-section-command}
5730 @var{output-section-command}
5732 @} [:@var{phdr}@dots{}] [=@var{fill}]
5735 @var{output-section-command}
5736 @var{output-section-command}
5738 @} [:@var{phdr}@dots{}] [=@var{fill}]
5740 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5744 Everything is optional except @code{OVERLAY} (a keyword), and each
5745 section must have a name (@var{secname1} and @var{secname2} above). The
5746 section definitions within the @code{OVERLAY} construct are identical to
5747 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5748 except that no addresses and no memory regions may be defined for
5749 sections within an @code{OVERLAY}.
5751 The comma at the end may be required if a @var{fill} is used and
5752 the next @var{sections-command} looks like a continuation of the expression.
5754 The sections are all defined with the same starting address. The load
5755 addresses of the sections are arranged such that they are consecutive in
5756 memory starting at the load address used for the @code{OVERLAY} as a
5757 whole (as with normal section definitions, the load address is optional,
5758 and defaults to the start address; the start address is also optional,
5759 and defaults to the current value of the location counter).
5761 If the @code{NOCROSSREFS} keyword is used, and there are any
5762 references among the sections, the linker will report an error. Since
5763 the sections all run at the same address, it normally does not make
5764 sense for one section to refer directly to another.
5765 @xref{Miscellaneous Commands, NOCROSSREFS}.
5767 For each section within the @code{OVERLAY}, the linker automatically
5768 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5769 defined as the starting load address of the section. The symbol
5770 @code{__load_stop_@var{secname}} is defined as the final load address of
5771 the section. Any characters within @var{secname} which are not legal
5772 within C identifiers are removed. C (or assembler) code may use these
5773 symbols to move the overlaid sections around as necessary.
5775 At the end of the overlay, the value of the location counter is set to
5776 the start address of the overlay plus the size of the largest section.
5778 Here is an example. Remember that this would appear inside a
5779 @code{SECTIONS} construct.
5782 OVERLAY 0x1000 : AT (0x4000)
5784 .text0 @{ o1/*.o(.text) @}
5785 .text1 @{ o2/*.o(.text) @}
5790 This will define both @samp{.text0} and @samp{.text1} to start at
5791 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5792 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5793 following symbols will be defined if referenced: @code{__load_start_text0},
5794 @code{__load_stop_text0}, @code{__load_start_text1},
5795 @code{__load_stop_text1}.
5797 C code to copy overlay @code{.text1} into the overlay area might look
5802 extern char __load_start_text1, __load_stop_text1;
5803 memcpy ((char *) 0x1000, &__load_start_text1,
5804 &__load_stop_text1 - &__load_start_text1);
5808 Note that the @code{OVERLAY} command is just syntactic sugar, since
5809 everything it does can be done using the more basic commands. The above
5810 example could have been written identically as follows.
5814 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5815 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5816 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5817 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5818 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5819 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5820 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5825 @section MEMORY Command
5827 @cindex memory regions
5828 @cindex regions of memory
5829 @cindex allocating memory
5830 @cindex discontinuous memory
5831 The linker's default configuration permits allocation of all available
5832 memory. You can override this by using the @code{MEMORY} command.
5834 The @code{MEMORY} command describes the location and size of blocks of
5835 memory in the target. You can use it to describe which memory regions
5836 may be used by the linker, and which memory regions it must avoid. You
5837 can then assign sections to particular memory regions. The linker will
5838 set section addresses based on the memory regions, and will warn about
5839 regions that become too full. The linker will not shuffle sections
5840 around to fit into the available regions.
5842 A linker script may contain many uses of the @code{MEMORY} command,
5843 however, all memory blocks defined are treated as if they were
5844 specified inside a single @code{MEMORY} command. The syntax for
5850 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5856 The @var{name} is a name used in the linker script to refer to the
5857 region. The region name has no meaning outside of the linker script.
5858 Region names are stored in a separate name space, and will not conflict
5859 with symbol names, file names, or section names. Each memory region
5860 must have a distinct name within the @code{MEMORY} command. However you can
5861 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5864 @cindex memory region attributes
5865 The @var{attr} string is an optional list of attributes that specify
5866 whether to use a particular memory region for an input section which is
5867 not explicitly mapped in the linker script. As described in
5868 @ref{SECTIONS}, if you do not specify an output section for some input
5869 section, the linker will create an output section with the same name as
5870 the input section. If you define region attributes, the linker will use
5871 them to select the memory region for the output section that it creates.
5873 The @var{attr} string must consist only of the following characters:
5888 Invert the sense of any of the attributes that follow
5891 If an unmapped section matches any of the listed attributes other than
5892 @samp{!}, it will be placed in the memory region. The @samp{!}
5893 attribute reverses the test for the characters that follow, so that an
5894 unmapped section will be placed in the memory region only if it does
5895 not match any of the attributes listed afterwards. Thus an attribute
5896 string of @samp{RW!X} will match any unmapped section that has either
5897 or both of the @samp{R} and @samp{W} attributes, but only as long as
5898 the section does not also have the @samp{X} attribute.
5903 The @var{origin} is an numerical expression for the start address of
5904 the memory region. The expression must evaluate to a constant and it
5905 cannot involve any symbols. The keyword @code{ORIGIN} may be
5906 abbreviated to @code{org} or @code{o} (but not, for example,
5912 The @var{len} is an expression for the size in bytes of the memory
5913 region. As with the @var{origin} expression, the expression must
5914 be numerical only and must evaluate to a constant. The keyword
5915 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5917 In the following example, we specify that there are two memory regions
5918 available for allocation: one starting at @samp{0} for 256 kilobytes,
5919 and the other starting at @samp{0x40000000} for four megabytes. The
5920 linker will place into the @samp{rom} memory region every section which
5921 is not explicitly mapped into a memory region, and is either read-only
5922 or executable. The linker will place other sections which are not
5923 explicitly mapped into a memory region into the @samp{ram} memory
5930 rom (rx) : ORIGIN = 0, LENGTH = 256K
5931 ram (!rx) : org = 0x40000000, l = 4M
5936 Once you define a memory region, you can direct the linker to place
5937 specific output sections into that memory region by using the
5938 @samp{>@var{region}} output section attribute. For example, if you have
5939 a memory region named @samp{mem}, you would use @samp{>mem} in the
5940 output section definition. @xref{Output Section Region}. If no address
5941 was specified for the output section, the linker will set the address to
5942 the next available address within the memory region. If the combined
5943 output sections directed to a memory region are too large for the
5944 region, the linker will issue an error message.
5946 It is possible to access the origin and length of a memory in an
5947 expression via the @code{ORIGIN(@var{memory})} and
5948 @code{LENGTH(@var{memory})} functions:
5952 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5957 @section PHDRS Command
5959 @cindex program headers
5960 @cindex ELF program headers
5961 @cindex program segments
5962 @cindex segments, ELF
5963 The ELF object file format uses @dfn{program headers}, also knows as
5964 @dfn{segments}. The program headers describe how the program should be
5965 loaded into memory. You can print them out by using the @code{objdump}
5966 program with the @samp{-p} option.
5968 When you run an ELF program on a native ELF system, the system loader
5969 reads the program headers in order to figure out how to load the
5970 program. This will only work if the program headers are set correctly.
5971 This manual does not describe the details of how the system loader
5972 interprets program headers; for more information, see the ELF ABI.
5974 The linker will create reasonable program headers by default. However,
5975 in some cases, you may need to specify the program headers more
5976 precisely. You may use the @code{PHDRS} command for this purpose. When
5977 the linker sees the @code{PHDRS} command in the linker script, it will
5978 not create any program headers other than the ones specified.
5980 The linker only pays attention to the @code{PHDRS} command when
5981 generating an ELF output file. In other cases, the linker will simply
5982 ignore @code{PHDRS}.
5984 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5985 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5991 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5992 [ FLAGS ( @var{flags} ) ] ;
5997 The @var{name} is used only for reference in the @code{SECTIONS} command
5998 of the linker script. It is not put into the output file. Program
5999 header names are stored in a separate name space, and will not conflict
6000 with symbol names, file names, or section names. Each program header
6001 must have a distinct name. The headers are processed in order and it
6002 is usual for them to map to sections in ascending load address order.
6004 Certain program header types describe segments of memory which the
6005 system loader will load from the file. In the linker script, you
6006 specify the contents of these segments by placing allocatable output
6007 sections in the segments. You use the @samp{:@var{phdr}} output section
6008 attribute to place a section in a particular segment. @xref{Output
6011 It is normal to put certain sections in more than one segment. This
6012 merely implies that one segment of memory contains another. You may
6013 repeat @samp{:@var{phdr}}, using it once for each segment which should
6014 contain the section.
6016 If you place a section in one or more segments using @samp{:@var{phdr}},
6017 then the linker will place all subsequent allocatable sections which do
6018 not specify @samp{:@var{phdr}} in the same segments. This is for
6019 convenience, since generally a whole set of contiguous sections will be
6020 placed in a single segment. You can use @code{:NONE} to override the
6021 default segment and tell the linker to not put the section in any
6026 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
6027 the program header type to further describe the contents of the segment.
6028 The @code{FILEHDR} keyword means that the segment should include the ELF
6029 file header. The @code{PHDRS} keyword means that the segment should
6030 include the ELF program headers themselves. If applied to a loadable
6031 segment (@code{PT_LOAD}), all prior loadable segments must have one of
6034 The @var{type} may be one of the following. The numbers indicate the
6035 value of the keyword.
6038 @item @code{PT_NULL} (0)
6039 Indicates an unused program header.
6041 @item @code{PT_LOAD} (1)
6042 Indicates that this program header describes a segment to be loaded from
6045 @item @code{PT_DYNAMIC} (2)
6046 Indicates a segment where dynamic linking information can be found.
6048 @item @code{PT_INTERP} (3)
6049 Indicates a segment where the name of the program interpreter may be
6052 @item @code{PT_NOTE} (4)
6053 Indicates a segment holding note information.
6055 @item @code{PT_SHLIB} (5)
6056 A reserved program header type, defined but not specified by the ELF
6059 @item @code{PT_PHDR} (6)
6060 Indicates a segment where the program headers may be found.
6062 @item @code{PT_TLS} (7)
6063 Indicates a segment containing thread local storage.
6065 @item @var{expression}
6066 An expression giving the numeric type of the program header. This may
6067 be used for types not defined above.
6070 You can specify that a segment should be loaded at a particular address
6071 in memory by using an @code{AT} expression. This is identical to the
6072 @code{AT} command used as an output section attribute (@pxref{Output
6073 Section LMA}). The @code{AT} command for a program header overrides the
6074 output section attribute.
6076 The linker will normally set the segment flags based on the sections
6077 which comprise the segment. You may use the @code{FLAGS} keyword to
6078 explicitly specify the segment flags. The value of @var{flags} must be
6079 an integer. It is used to set the @code{p_flags} field of the program
6082 Here is an example of @code{PHDRS}. This shows a typical set of program
6083 headers used on a native ELF system.
6089 headers PT_PHDR PHDRS ;
6091 text PT_LOAD FILEHDR PHDRS ;
6093 dynamic PT_DYNAMIC ;
6099 .interp : @{ *(.interp) @} :text :interp
6100 .text : @{ *(.text) @} :text
6101 .rodata : @{ *(.rodata) @} /* defaults to :text */
6103 . = . + 0x1000; /* move to a new page in memory */
6104 .data : @{ *(.data) @} :data
6105 .dynamic : @{ *(.dynamic) @} :data :dynamic
6112 @section VERSION Command
6113 @kindex VERSION @{script text@}
6114 @cindex symbol versions
6115 @cindex version script
6116 @cindex versions of symbols
6117 The linker supports symbol versions when using ELF. Symbol versions are
6118 only useful when using shared libraries. The dynamic linker can use
6119 symbol versions to select a specific version of a function when it runs
6120 a program that may have been linked against an earlier version of the
6123 You can include a version script directly in the main linker script, or
6124 you can supply the version script as an implicit linker script. You can
6125 also use the @samp{--version-script} linker option.
6127 The syntax of the @code{VERSION} command is simply
6129 VERSION @{ version-script-commands @}
6132 The format of the version script commands is identical to that used by
6133 Sun's linker in Solaris 2.5. The version script defines a tree of
6134 version nodes. You specify the node names and interdependencies in the
6135 version script. You can specify which symbols are bound to which
6136 version nodes, and you can reduce a specified set of symbols to local
6137 scope so that they are not globally visible outside of the shared
6140 The easiest way to demonstrate the version script language is with a few
6166 This example version script defines three version nodes. The first
6167 version node defined is @samp{VERS_1.1}; it has no other dependencies.
6168 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
6169 a number of symbols to local scope so that they are not visible outside
6170 of the shared library; this is done using wildcard patterns, so that any
6171 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
6172 is matched. The wildcard patterns available are the same as those used
6173 in the shell when matching filenames (also known as ``globbing'').
6174 However, if you specify the symbol name inside double quotes, then the
6175 name is treated as literal, rather than as a glob pattern.
6177 Next, the version script defines node @samp{VERS_1.2}. This node
6178 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
6179 to the version node @samp{VERS_1.2}.
6181 Finally, the version script defines node @samp{VERS_2.0}. This node
6182 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
6183 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
6185 When the linker finds a symbol defined in a library which is not
6186 specifically bound to a version node, it will effectively bind it to an
6187 unspecified base version of the library. You can bind all otherwise
6188 unspecified symbols to a given version node by using @samp{global: *;}
6189 somewhere in the version script. Note that it's slightly crazy to use
6190 wildcards in a global spec except on the last version node. Global
6191 wildcards elsewhere run the risk of accidentally adding symbols to the
6192 set exported for an old version. That's wrong since older versions
6193 ought to have a fixed set of symbols.
6195 The names of the version nodes have no specific meaning other than what
6196 they might suggest to the person reading them. The @samp{2.0} version
6197 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
6198 However, this would be a confusing way to write a version script.
6200 Node name can be omitted, provided it is the only version node
6201 in the version script. Such version script doesn't assign any versions to
6202 symbols, only selects which symbols will be globally visible out and which
6206 @{ global: foo; bar; local: *; @};
6209 When you link an application against a shared library that has versioned
6210 symbols, the application itself knows which version of each symbol it
6211 requires, and it also knows which version nodes it needs from each
6212 shared library it is linked against. Thus at runtime, the dynamic
6213 loader can make a quick check to make sure that the libraries you have
6214 linked against do in fact supply all of the version nodes that the
6215 application will need to resolve all of the dynamic symbols. In this
6216 way it is possible for the dynamic linker to know with certainty that
6217 all external symbols that it needs will be resolvable without having to
6218 search for each symbol reference.
6220 The symbol versioning is in effect a much more sophisticated way of
6221 doing minor version checking that SunOS does. The fundamental problem
6222 that is being addressed here is that typically references to external
6223 functions are bound on an as-needed basis, and are not all bound when
6224 the application starts up. If a shared library is out of date, a
6225 required interface may be missing; when the application tries to use
6226 that interface, it may suddenly and unexpectedly fail. With symbol
6227 versioning, the user will get a warning when they start their program if
6228 the libraries being used with the application are too old.
6230 There are several GNU extensions to Sun's versioning approach. The
6231 first of these is the ability to bind a symbol to a version node in the
6232 source file where the symbol is defined instead of in the versioning
6233 script. This was done mainly to reduce the burden on the library
6234 maintainer. You can do this by putting something like:
6236 __asm__(".symver original_foo,foo@@VERS_1.1");
6239 in the C source file. This renames the function @samp{original_foo} to
6240 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
6241 The @samp{local:} directive can be used to prevent the symbol
6242 @samp{original_foo} from being exported. A @samp{.symver} directive
6243 takes precedence over a version script.
6245 The second GNU extension is to allow multiple versions of the same
6246 function to appear in a given shared library. In this way you can make
6247 an incompatible change to an interface without increasing the major
6248 version number of the shared library, while still allowing applications
6249 linked against the old interface to continue to function.
6251 To do this, you must use multiple @samp{.symver} directives in the
6252 source file. Here is an example:
6255 __asm__(".symver original_foo,foo@@");
6256 __asm__(".symver old_foo,foo@@VERS_1.1");
6257 __asm__(".symver old_foo1,foo@@VERS_1.2");
6258 __asm__(".symver new_foo,foo@@@@VERS_2.0");
6261 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
6262 unspecified base version of the symbol. The source file that contains this
6263 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
6264 @samp{old_foo1}, and @samp{new_foo}.
6266 When you have multiple definitions of a given symbol, there needs to be
6267 some way to specify a default version to which external references to
6268 this symbol will be bound. You can do this with the
6269 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
6270 declare one version of a symbol as the default in this manner; otherwise
6271 you would effectively have multiple definitions of the same symbol.
6273 If you wish to bind a reference to a specific version of the symbol
6274 within the shared library, you can use the aliases of convenience
6275 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
6276 specifically bind to an external version of the function in question.
6278 You can also specify the language in the version script:
6281 VERSION extern "lang" @{ version-script-commands @}
6284 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
6285 The linker will iterate over the list of symbols at the link time and
6286 demangle them according to @samp{lang} before matching them to the
6287 patterns specified in @samp{version-script-commands}. The default
6288 @samp{lang} is @samp{C}.
6290 Demangled names may contains spaces and other special characters. As
6291 described above, you can use a glob pattern to match demangled names,
6292 or you can use a double-quoted string to match the string exactly. In
6293 the latter case, be aware that minor differences (such as differing
6294 whitespace) between the version script and the demangler output will
6295 cause a mismatch. As the exact string generated by the demangler
6296 might change in the future, even if the mangled name does not, you
6297 should check that all of your version directives are behaving as you
6298 expect when you upgrade.
6301 @section Expressions in Linker Scripts
6304 The syntax for expressions in the linker script language is identical to
6305 that of C expressions, except that whitespace is required in some
6306 places to resolve syntactic ambiguities. All expressions are
6307 evaluated as integers. All expressions are evaluated in the same
6308 size, which is 32 bits if both the host and target are 32 bits, and is
6311 You can use and set symbol values in expressions.
6313 The linker defines several special purpose builtin functions for use in
6317 * Constants:: Constants
6318 * Symbolic Constants:: Symbolic constants
6319 * Symbols:: Symbol Names
6320 * Orphan Sections:: Orphan Sections
6321 * Location Counter:: The Location Counter
6322 * Operators:: Operators
6323 * Evaluation:: Evaluation
6324 * Expression Section:: The Section of an Expression
6325 * Builtin Functions:: Builtin Functions
6329 @subsection Constants
6330 @cindex integer notation
6331 @cindex constants in linker scripts
6332 All constants are integers.
6334 As in C, the linker considers an integer beginning with @samp{0} to be
6335 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
6336 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
6337 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
6338 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
6339 value without a prefix or a suffix is considered to be decimal.
6341 @cindex scaled integers
6342 @cindex K and M integer suffixes
6343 @cindex M and K integer suffixes
6344 @cindex suffixes for integers
6345 @cindex integer suffixes
6346 In addition, you can use the suffixes @code{K} and @code{M} to scale a
6350 @c END TEXI2ROFF-KILL
6351 @code{1024} or @code{1024*1024}
6355 ${\rm 1024}$ or ${\rm 1024}^2$
6357 @c END TEXI2ROFF-KILL
6358 respectively. For example, the following
6359 all refer to the same quantity:
6368 Note - the @code{K} and @code{M} suffixes cannot be used in
6369 conjunction with the base suffixes mentioned above.
6371 @node Symbolic Constants
6372 @subsection Symbolic Constants
6373 @cindex symbolic constants
6375 It is possible to refer to target-specific constants via the use of
6376 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
6381 The target's maximum page size.
6383 @item COMMONPAGESIZE
6384 @kindex COMMONPAGESIZE
6385 The target's default page size.
6391 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
6394 will create a text section aligned to the largest page boundary
6395 supported by the target.
6398 @subsection Symbol Names
6399 @cindex symbol names
6401 @cindex quoted symbol names
6403 Unless quoted, symbol names start with a letter, underscore, or period
6404 and may include letters, digits, underscores, periods, and hyphens.
6405 Unquoted symbol names must not conflict with any keywords. You can
6406 specify a symbol which contains odd characters or has the same name as a
6407 keyword by surrounding the symbol name in double quotes:
6410 "with a space" = "also with a space" + 10;
6413 Since symbols can contain many non-alphabetic characters, it is safest
6414 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
6415 whereas @samp{A - B} is an expression involving subtraction.
6417 @node Orphan Sections
6418 @subsection Orphan Sections
6420 Orphan sections are sections present in the input files which
6421 are not explicitly placed into the output file by the linker
6422 script. The linker will still copy these sections into the
6423 output file by either finding, or creating a suitable output section
6424 in which to place the orphaned input section.
6426 If the name of an orphaned input section exactly matches the name of
6427 an existing output section, then the orphaned input section will be
6428 placed at the end of that output section.
6430 If there is no output section with a matching name then new output
6431 sections will be created. Each new output section will have the same
6432 name as the orphan section placed within it. If there are multiple
6433 orphan sections with the same name, these will all be combined into
6434 one new output section.
6436 If new output sections are created to hold orphaned input sections,
6437 then the linker must decide where to place these new output sections
6438 in relation to existing output sections. On most modern targets, the
6439 linker attempts to place orphan sections after sections of the same
6440 attribute, such as code vs data, loadable vs non-loadable, etc. If no
6441 sections with matching attributes are found, or your target lacks this
6442 support, the orphan section is placed at the end of the file.
6444 The command-line options @samp{--orphan-handling} and @samp{--unique}
6445 (@pxref{Options,,Command-line Options}) can be used to control which
6446 output sections an orphan is placed in.
6448 @node Location Counter
6449 @subsection The Location Counter
6452 @cindex location counter
6453 @cindex current output location
6454 The special linker variable @dfn{dot} @samp{.} always contains the
6455 current output location counter. Since the @code{.} always refers to a
6456 location in an output section, it may only appear in an expression
6457 within a @code{SECTIONS} command. The @code{.} symbol may appear
6458 anywhere that an ordinary symbol is allowed in an expression.
6461 Assigning a value to @code{.} will cause the location counter to be
6462 moved. This may be used to create holes in the output section. The
6463 location counter may not be moved backwards inside an output section,
6464 and may not be moved backwards outside of an output section if so
6465 doing creates areas with overlapping LMAs.
6481 In the previous example, the @samp{.text} section from @file{file1} is
6482 located at the beginning of the output section @samp{output}. It is
6483 followed by a 1000 byte gap. Then the @samp{.text} section from
6484 @file{file2} appears, also with a 1000 byte gap following before the
6485 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
6486 specifies what data to write in the gaps (@pxref{Output Section Fill}).
6488 @cindex dot inside sections
6489 Note: @code{.} actually refers to the byte offset from the start of the
6490 current containing object. Normally this is the @code{SECTIONS}
6491 statement, whose start address is 0, hence @code{.} can be used as an
6492 absolute address. If @code{.} is used inside a section description
6493 however, it refers to the byte offset from the start of that section,
6494 not an absolute address. Thus in a script like this:
6512 The @samp{.text} section will be assigned a starting address of 0x100
6513 and a size of exactly 0x200 bytes, even if there is not enough data in
6514 the @samp{.text} input sections to fill this area. (If there is too
6515 much data, an error will be produced because this would be an attempt to
6516 move @code{.} backwards). The @samp{.data} section will start at 0x500
6517 and it will have an extra 0x600 bytes worth of space after the end of
6518 the values from the @samp{.data} input sections and before the end of
6519 the @samp{.data} output section itself.
6521 @cindex dot outside sections
6522 Setting symbols to the value of the location counter outside of an
6523 output section statement can result in unexpected values if the linker
6524 needs to place orphan sections. For example, given the following:
6530 .text: @{ *(.text) @}
6534 .data: @{ *(.data) @}
6539 If the linker needs to place some input section, e.g. @code{.rodata},
6540 not mentioned in the script, it might choose to place that section
6541 between @code{.text} and @code{.data}. You might think the linker
6542 should place @code{.rodata} on the blank line in the above script, but
6543 blank lines are of no particular significance to the linker. As well,
6544 the linker doesn't associate the above symbol names with their
6545 sections. Instead, it assumes that all assignments or other
6546 statements belong to the previous output section, except for the
6547 special case of an assignment to @code{.}. I.e., the linker will
6548 place the orphan @code{.rodata} section as if the script was written
6555 .text: @{ *(.text) @}
6559 .rodata: @{ *(.rodata) @}
6560 .data: @{ *(.data) @}
6565 This may or may not be the script author's intention for the value of
6566 @code{start_of_data}. One way to influence the orphan section
6567 placement is to assign the location counter to itself, as the linker
6568 assumes that an assignment to @code{.} is setting the start address of
6569 a following output section and thus should be grouped with that
6570 section. So you could write:
6576 .text: @{ *(.text) @}
6581 .data: @{ *(.data) @}
6586 Now, the orphan @code{.rodata} section will be placed between
6587 @code{end_of_text} and @code{start_of_data}.
6591 @subsection Operators
6592 @cindex operators for arithmetic
6593 @cindex arithmetic operators
6594 @cindex precedence in expressions
6595 The linker recognizes the standard C set of arithmetic operators, with
6596 the standard bindings and precedence levels:
6599 @c END TEXI2ROFF-KILL
6601 precedence associativity Operators Notes
6607 5 left == != > < <= >=
6613 11 right &= += -= *= /= (2)
6617 (1) Prefix operators
6618 (2) @xref{Assignments}.
6622 \vskip \baselineskip
6623 %"lispnarrowing" is the extra indent used generally for smallexample
6624 \hskip\lispnarrowing\vbox{\offinterlineskip
6627 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
6628 height2pt&\omit&&\omit&&\omit&\cr
6629 &Precedence&& Associativity &&{\rm Operators}&\cr
6630 height2pt&\omit&&\omit&&\omit&\cr
6632 height2pt&\omit&&\omit&&\omit&\cr
6634 % '176 is tilde, '~' in tt font
6635 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
6636 &2&&left&&* / \%&\cr
6639 &5&&left&&== != > < <= >=&\cr
6642 &8&&left&&{\&\&}&\cr
6645 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
6647 height2pt&\omit&&\omit&&\omit&\cr}
6652 @obeylines@parskip=0pt@parindent=0pt
6653 @dag@quad Prefix operators.
6654 @ddag@quad @xref{Assignments}.
6657 @c END TEXI2ROFF-KILL
6660 @subsection Evaluation
6661 @cindex lazy evaluation
6662 @cindex expression evaluation order
6663 The linker evaluates expressions lazily. It only computes the value of
6664 an expression when absolutely necessary.
6666 The linker needs some information, such as the value of the start
6667 address of the first section, and the origins and lengths of memory
6668 regions, in order to do any linking at all. These values are computed
6669 as soon as possible when the linker reads in the linker script.
6671 However, other values (such as symbol values) are not known or needed
6672 until after storage allocation. Such values are evaluated later, when
6673 other information (such as the sizes of output sections) is available
6674 for use in the symbol assignment expression.
6676 The sizes of sections cannot be known until after allocation, so
6677 assignments dependent upon these are not performed until after
6680 Some expressions, such as those depending upon the location counter
6681 @samp{.}, must be evaluated during section allocation.
6683 If the result of an expression is required, but the value is not
6684 available, then an error results. For example, a script like the
6690 .text 9+this_isnt_constant :
6696 will cause the error message @samp{non constant expression for initial
6699 @node Expression Section
6700 @subsection The Section of an Expression
6701 @cindex expression sections
6702 @cindex absolute expressions
6703 @cindex relative expressions
6704 @cindex absolute and relocatable symbols
6705 @cindex relocatable and absolute symbols
6706 @cindex symbols, relocatable and absolute
6707 Addresses and symbols may be section relative, or absolute. A section
6708 relative symbol is relocatable. If you request relocatable output
6709 using the @samp{-r} option, a further link operation may change the
6710 value of a section relative symbol. On the other hand, an absolute
6711 symbol will retain the same value throughout any further link
6714 Some terms in linker expressions are addresses. This is true of
6715 section relative symbols and for builtin functions that return an
6716 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
6717 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
6718 functions that return a non-address value, such as @code{LENGTH}.
6719 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
6720 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
6721 differently depending on their location, for compatibility with older
6722 versions of @code{ld}. Expressions appearing outside an output
6723 section definition treat all numbers as absolute addresses.
6724 Expressions appearing inside an output section definition treat
6725 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
6726 given, then absolute symbols and numbers are simply treated as numbers
6729 In the following simple example,
6736 __executable_start = 0x100;
6740 __data_start = 0x10;
6748 both @code{.} and @code{__executable_start} are set to the absolute
6749 address 0x100 in the first two assignments, then both @code{.} and
6750 @code{__data_start} are set to 0x10 relative to the @code{.data}
6751 section in the second two assignments.
6753 For expressions involving numbers, relative addresses and absolute
6754 addresses, ld follows these rules to evaluate terms:
6758 Unary operations on an absolute address or number, and binary
6759 operations on two absolute addresses or two numbers, or between one
6760 absolute address and a number, apply the operator to the value(s).
6762 Unary operations on a relative address, and binary operations on two
6763 relative addresses in the same section or between one relative address
6764 and a number, apply the operator to the offset part of the address(es).
6766 Other binary operations, that is, between two relative addresses not
6767 in the same section, or between a relative address and an absolute
6768 address, first convert any non-absolute term to an absolute address
6769 before applying the operator.
6772 The result section of each sub-expression is as follows:
6776 An operation involving only numbers results in a number.
6778 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6780 The result of other binary arithmetic and logical operations on two
6781 relative addresses in the same section or two absolute addresses
6782 (after above conversions) is also a number when
6783 @code{LD_FEATURE ("SANE_EXPR")} or inside an output section definition
6784 but an absolute address otherwise.
6786 The result of other operations on relative addresses or one
6787 relative address and a number, is a relative address in the same
6788 section as the relative operand(s).
6790 The result of other operations on absolute addresses (after above
6791 conversions) is an absolute address.
6794 You can use the builtin function @code{ABSOLUTE} to force an expression
6795 to be absolute when it would otherwise be relative. For example, to
6796 create an absolute symbol set to the address of the end of the output
6797 section @samp{.data}:
6801 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6805 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6806 @samp{.data} section.
6808 Using @code{LOADADDR} also forces an expression absolute, since this
6809 particular builtin function returns an absolute address.
6811 @node Builtin Functions
6812 @subsection Builtin Functions
6813 @cindex functions in expressions
6814 The linker script language includes a number of builtin functions for
6815 use in linker script expressions.
6818 @item ABSOLUTE(@var{exp})
6819 @kindex ABSOLUTE(@var{exp})
6820 @cindex expression, absolute
6821 Return the absolute (non-relocatable, as opposed to non-negative) value
6822 of the expression @var{exp}. Primarily useful to assign an absolute
6823 value to a symbol within a section definition, where symbol values are
6824 normally section relative. @xref{Expression Section}.
6826 @item ADDR(@var{section})
6827 @kindex ADDR(@var{section})
6828 @cindex section address in expression
6829 Return the address (VMA) of the named @var{section}. Your
6830 script must previously have defined the location of that section. In
6831 the following example, @code{start_of_output_1}, @code{symbol_1} and
6832 @code{symbol_2} are assigned equivalent values, except that
6833 @code{symbol_1} will be relative to the @code{.output1} section while
6834 the other two will be absolute:
6840 start_of_output_1 = ABSOLUTE(.);
6845 symbol_1 = ADDR(.output1);
6846 symbol_2 = start_of_output_1;
6852 @item ALIGN(@var{align})
6853 @itemx ALIGN(@var{exp},@var{align})
6854 @kindex ALIGN(@var{align})
6855 @kindex ALIGN(@var{exp},@var{align})
6856 @cindex round up location counter
6857 @cindex align location counter
6858 @cindex round up expression
6859 @cindex align expression
6860 Return the location counter (@code{.}) or arbitrary expression aligned
6861 to the next @var{align} boundary. The single operand @code{ALIGN}
6862 doesn't change the value of the location counter---it just does
6863 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6864 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6865 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6867 Here is an example which aligns the output @code{.data} section to the
6868 next @code{0x2000} byte boundary after the preceding section and sets a
6869 variable within the section to the next @code{0x8000} boundary after the
6874 .data ALIGN(0x2000): @{
6876 variable = ALIGN(0x8000);
6882 The first use of @code{ALIGN} in this example specifies the location of
6883 a section because it is used as the optional @var{address} attribute of
6884 a section definition (@pxref{Output Section Address}). The second use
6885 of @code{ALIGN} is used to defines the value of a symbol.
6887 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6889 @item ALIGNOF(@var{section})
6890 @kindex ALIGNOF(@var{section})
6891 @cindex section alignment
6892 Return the alignment in bytes of the named @var{section}, if that section has
6893 been allocated. If the section has not been allocated when this is
6894 evaluated, the linker will report an error. In the following example,
6895 the alignment of the @code{.output} section is stored as the first
6896 value in that section.
6901 LONG (ALIGNOF (.output))
6908 @item BLOCK(@var{exp})
6909 @kindex BLOCK(@var{exp})
6910 This is a synonym for @code{ALIGN}, for compatibility with older linker
6911 scripts. It is most often seen when setting the address of an output
6914 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6915 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6916 This is equivalent to either
6918 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6922 (ALIGN(@var{maxpagesize})
6923 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6926 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6927 for the data segment (area between the result of this expression and
6928 @code{DATA_SEGMENT_END}) than the former or not.
6929 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6930 memory will be saved at the expense of up to @var{commonpagesize} wasted
6931 bytes in the on-disk file.
6933 This expression can only be used directly in @code{SECTIONS} commands, not in
6934 any output section descriptions and only once in the linker script.
6935 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6936 be the system page size the object wants to be optimized for while still
6937 running on system page sizes up to @var{maxpagesize}. Note however
6938 that @samp{-z relro} protection will not be effective if the system
6939 page size is larger than @var{commonpagesize}.
6944 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6947 @item DATA_SEGMENT_END(@var{exp})
6948 @kindex DATA_SEGMENT_END(@var{exp})
6949 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6950 evaluation purposes.
6953 . = DATA_SEGMENT_END(.);
6956 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6957 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6958 This defines the end of the @code{PT_GNU_RELRO} segment when
6959 @samp{-z relro} option is used.
6960 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6961 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6962 @var{exp} + @var{offset} is aligned to the @var{commonpagesize}
6963 argument given to @code{DATA_SEGMENT_ALIGN}. If present in the linker
6964 script, it must be placed between @code{DATA_SEGMENT_ALIGN} and
6965 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6966 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6970 . = DATA_SEGMENT_RELRO_END(24, .);
6973 @item DEFINED(@var{symbol})
6974 @kindex DEFINED(@var{symbol})
6975 @cindex symbol defaults
6976 Return 1 if @var{symbol} is in the linker global symbol table and is
6977 defined before the statement using DEFINED in the script, otherwise
6978 return 0. You can use this function to provide
6979 default values for symbols. For example, the following script fragment
6980 shows how to set a global symbol @samp{begin} to the first location in
6981 the @samp{.text} section---but if a symbol called @samp{begin} already
6982 existed, its value is preserved:
6988 begin = DEFINED(begin) ? begin : . ;
6996 @item LENGTH(@var{memory})
6997 @kindex LENGTH(@var{memory})
6998 Return the length of the memory region named @var{memory}.
7000 @item LOADADDR(@var{section})
7001 @kindex LOADADDR(@var{section})
7002 @cindex section load address in expression
7003 Return the absolute LMA of the named @var{section}. (@pxref{Output
7006 @item LOG2CEIL(@var{exp})
7007 @kindex LOG2CEIL(@var{exp})
7008 Return the binary logarithm of @var{exp} rounded towards infinity.
7009 @code{LOG2CEIL(0)} returns 0.
7012 @item MAX(@var{exp1}, @var{exp2})
7013 Returns the maximum of @var{exp1} and @var{exp2}.
7016 @item MIN(@var{exp1}, @var{exp2})
7017 Returns the minimum of @var{exp1} and @var{exp2}.
7019 @item NEXT(@var{exp})
7020 @kindex NEXT(@var{exp})
7021 @cindex unallocated address, next
7022 Return the next unallocated address that is a multiple of @var{exp}.
7023 This function is closely related to @code{ALIGN(@var{exp})}; unless you
7024 use the @code{MEMORY} command to define discontinuous memory for the
7025 output file, the two functions are equivalent.
7027 @item ORIGIN(@var{memory})
7028 @kindex ORIGIN(@var{memory})
7029 Return the origin of the memory region named @var{memory}.
7031 @item SEGMENT_START(@var{segment}, @var{default})
7032 @kindex SEGMENT_START(@var{segment}, @var{default})
7033 Return the base address of the named @var{segment}. If an explicit
7034 value has already been given for this segment (with a command-line
7035 @samp{-T} option) then that value will be returned otherwise the value
7036 will be @var{default}. At present, the @samp{-T} command-line option
7037 can only be used to set the base address for the ``text'', ``data'', and
7038 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
7041 @item SIZEOF(@var{section})
7042 @kindex SIZEOF(@var{section})
7043 @cindex section size
7044 Return the size in bytes of the named @var{section}, if that section has
7045 been allocated. If the section has not been allocated when this is
7046 evaluated, the linker will report an error. In the following example,
7047 @code{symbol_1} and @code{symbol_2} are assigned identical values:
7056 symbol_1 = .end - .start ;
7057 symbol_2 = SIZEOF(.output);
7062 @item SIZEOF_HEADERS
7063 @kindex SIZEOF_HEADERS
7065 Return the size in bytes of the output file's headers. This is
7066 information which appears at the start of the output file. You can use
7067 this number when setting the start address of the first section, if you
7068 choose, to facilitate paging.
7070 @cindex not enough room for program headers
7071 @cindex program headers, not enough room
7072 When producing an ELF output file, if the linker script uses the
7073 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
7074 number of program headers before it has determined all the section
7075 addresses and sizes. If the linker later discovers that it needs
7076 additional program headers, it will report an error @samp{not enough
7077 room for program headers}. To avoid this error, you must avoid using
7078 the @code{SIZEOF_HEADERS} function, or you must rework your linker
7079 script to avoid forcing the linker to use additional program headers, or
7080 you must define the program headers yourself using the @code{PHDRS}
7081 command (@pxref{PHDRS}).
7084 @node Implicit Linker Scripts
7085 @section Implicit Linker Scripts
7086 @cindex implicit linker scripts
7087 If you specify a linker input file which the linker can not recognize as
7088 an object file or an archive file, it will try to read the file as a
7089 linker script. If the file can not be parsed as a linker script, the
7090 linker will report an error.
7092 An implicit linker script will not replace the default linker script.
7094 Typically an implicit linker script would contain only symbol
7095 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
7098 Any input files read because of an implicit linker script will be read
7099 at the position in the command line where the implicit linker script was
7100 read. This can affect archive searching.
7103 @chapter Linker Plugins
7106 @cindex linker plugins
7107 The linker can use dynamically loaded plugins to modify its behavior.
7108 For example, the link-time optimization feature that some compilers
7109 support is implemented with a linker plugin.
7111 Currently there is only one plugin shipped by default, but more may
7112 be added here later.
7115 * libdep Plugin:: Static Library Dependencies Plugin
7119 @section Static Library Dependencies Plugin
7120 @cindex static library dependencies
7121 Originally, static libraries were contained in an archive file consisting
7122 just of a collection of relocatable object files. Later they evolved to
7123 optionally include a symbol table, to assist in finding the needed objects
7124 within a library. There their evolution ended, and dynamic libraries
7127 One useful feature of dynamic libraries was that, more than just collecting
7128 multiple objects into a single file, they also included a list of their
7129 dependencies, such that one could specify just the name of a single dynamic
7130 library at link time, and all of its dependencies would be implicitly
7131 referenced as well. But static libraries lacked this feature, so if a
7132 link invocation was switched from using dynamic libraries to static
7133 libraries, the link command would usually fail unless it was rewritten to
7134 explicitly list the dependencies of the static library.
7136 The GNU @command{ar} utility now supports a @option{--record-libdeps} option
7137 to embed dependency lists into static libraries as well, and the @file{libdep}
7138 plugin may be used to read this dependency information at link time. The
7139 dependency information is stored as a single string, carrying @option{-l}
7140 and @option{-L} arguments as they would normally appear in a linker
7141 command line. As such, the information can be written with any text
7142 utility and stored into any archive, even if GNU @command{ar} is not
7143 being used to create the archive. The information is stored in an
7144 archive member named @samp{__.LIBDEP}.
7146 For example, given a library @file{libssl.a} that depends on another
7147 library @file{libcrypto.a} which may be found in @file{/usr/local/lib},
7148 the @samp{__.LIBDEP} member of @file{libssl.a} would contain
7151 -L/usr/local/lib -lcrypto
7155 @node Machine Dependent
7156 @chapter Machine Dependent Features
7158 @cindex machine dependencies
7159 @command{ld} has additional features on some platforms; the following
7160 sections describe them. Machines where @command{ld} has no additional
7161 functionality are not listed.
7165 * H8/300:: @command{ld} and the H8/300
7168 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
7171 * ARM:: @command{ld} and the ARM family
7174 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
7177 * M68K:: @command{ld} and the Motorola 68K family
7180 * MIPS:: @command{ld} and the MIPS family
7183 * MMIX:: @command{ld} and MMIX
7186 * MSP430:: @command{ld} and MSP430
7189 * NDS32:: @command{ld} and NDS32
7192 * Nios II:: @command{ld} and the Altera Nios II
7195 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
7198 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
7201 * S/390 ELF:: @command{ld} and S/390 ELF Support
7204 * SPU ELF:: @command{ld} and SPU ELF Support
7207 * TI COFF:: @command{ld} and TI COFF
7210 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
7213 * Xtensa:: @command{ld} and Xtensa Processors
7224 @section @command{ld} and the H8/300
7226 @cindex H8/300 support
7227 For the H8/300, @command{ld} can perform these global optimizations when
7228 you specify the @samp{--relax} command-line option.
7231 @cindex relaxing on H8/300
7232 @item relaxing address modes
7233 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
7234 targets are within eight bits, and turns them into eight-bit
7235 program-counter relative @code{bsr} and @code{bra} instructions,
7238 @cindex synthesizing on H8/300
7239 @item synthesizing instructions
7240 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
7241 @command{ld} finds all @code{mov.b} instructions which use the
7242 sixteen-bit absolute address form, but refer to the top
7243 page of memory, and changes them to use the eight-bit address form.
7244 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
7245 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
7246 top page of memory).
7248 @command{ld} finds all @code{mov} instructions which use the register
7249 indirect with 32-bit displacement addressing mode, but use a small
7250 displacement inside 16-bit displacement range, and changes them to use
7251 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
7252 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
7253 whenever the displacement @var{d} is in the 16 bit signed integer
7254 range. Only implemented in ELF-format ld).
7256 @item bit manipulation instructions
7257 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
7258 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
7259 which use 32 bit and 16 bit absolute address form, but refer to the top
7260 page of memory, and changes them to use the 8 bit address form.
7261 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
7262 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
7263 the top page of memory).
7265 @item system control instructions
7266 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
7267 32 bit absolute address form, but refer to the top page of memory, and
7268 changes them to use 16 bit address form.
7269 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
7270 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
7271 the top page of memory).
7281 @c This stuff is pointless to say unless you're especially concerned
7282 @c with Renesas chips; don't enable it for generic case, please.
7284 @chapter @command{ld} and Other Renesas Chips
7286 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
7287 H8/500, and SH chips. No special features, commands, or command-line
7288 options are required for these chips.
7302 @node M68HC11/68HC12
7303 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
7305 @cindex M68HC11 and 68HC12 support
7307 @subsection Linker Relaxation
7309 For the Motorola 68HC11, @command{ld} can perform these global
7310 optimizations when you specify the @samp{--relax} command-line option.
7313 @cindex relaxing on M68HC11
7314 @item relaxing address modes
7315 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
7316 targets are within eight bits, and turns them into eight-bit
7317 program-counter relative @code{bsr} and @code{bra} instructions,
7320 @command{ld} also looks at all 16-bit extended addressing modes and
7321 transforms them in a direct addressing mode when the address is in
7322 page 0 (between 0 and 0x0ff).
7324 @item relaxing gcc instruction group
7325 When @command{gcc} is called with @option{-mrelax}, it can emit group
7326 of instructions that the linker can optimize to use a 68HC11 direct
7327 addressing mode. These instructions consists of @code{bclr} or
7328 @code{bset} instructions.
7332 @subsection Trampoline Generation
7334 @cindex trampoline generation on M68HC11
7335 @cindex trampoline generation on M68HC12
7336 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
7337 call a far function using a normal @code{jsr} instruction. The linker
7338 will also change the relocation to some far function to use the
7339 trampoline address instead of the function address. This is typically the
7340 case when a pointer to a function is taken. The pointer will in fact
7341 point to the function trampoline.
7349 @section @command{ld} and the ARM family
7351 @cindex ARM interworking support
7352 @kindex --support-old-code
7353 For the ARM, @command{ld} will generate code stubs to allow functions calls
7354 between ARM and Thumb code. These stubs only work with code that has
7355 been compiled and assembled with the @samp{-mthumb-interwork} command
7356 line option. If it is necessary to link with old ARM object files or
7357 libraries, which have not been compiled with the -mthumb-interwork
7358 option then the @samp{--support-old-code} command-line switch should be
7359 given to the linker. This will make it generate larger stub functions
7360 which will work with non-interworking aware ARM code. Note, however,
7361 the linker does not support generating stubs for function calls to
7362 non-interworking aware Thumb code.
7364 @cindex thumb entry point
7365 @cindex entry point, thumb
7366 @kindex --thumb-entry=@var{entry}
7367 The @samp{--thumb-entry} switch is a duplicate of the generic
7368 @samp{--entry} switch, in that it sets the program's starting address.
7369 But it also sets the bottom bit of the address, so that it can be
7370 branched to using a BX instruction, and the program will start
7371 executing in Thumb mode straight away.
7373 @cindex PE import table prefixing
7374 @kindex --use-nul-prefixed-import-tables
7375 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
7376 the import tables idata4 and idata5 have to be generated with a zero
7377 element prefix for import libraries. This is the old style to generate
7378 import tables. By default this option is turned off.
7382 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
7383 executables. This option is only valid when linking big-endian
7384 objects - ie ones which have been assembled with the @option{-EB}
7385 option. The resulting image will contain big-endian data and
7389 @kindex --target1-rel
7390 @kindex --target1-abs
7391 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
7392 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
7393 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
7394 and @samp{--target1-abs} switches override the default.
7397 @kindex --target2=@var{type}
7398 The @samp{--target2=type} switch overrides the default definition of the
7399 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
7400 meanings, and target defaults are as follows:
7403 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
7407 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
7412 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
7413 specification) enables objects compiled for the ARMv4 architecture to be
7414 interworking-safe when linked with other objects compiled for ARMv4t, but
7415 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
7417 In the latter case, the switch @option{--fix-v4bx} must be passed to the
7418 linker, which causes v4t @code{BX rM} instructions to be rewritten as
7419 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
7421 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
7422 relocations are ignored.
7424 @cindex FIX_V4BX_INTERWORKING
7425 @kindex --fix-v4bx-interworking
7426 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
7427 relocations with a branch to the following veneer:
7435 This allows generation of libraries/applications that work on ARMv4 cores
7436 and are still interworking safe. Note that the above veneer clobbers the
7437 condition flags, so may cause incorrect program behavior in rare cases.
7441 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
7442 BLX instructions (available on ARMv5t and above) in various
7443 situations. Currently it is used to perform calls via the PLT from Thumb
7444 code using BLX rather than using BX and a mode-switching stub before
7445 each PLT entry. This should lead to such calls executing slightly faster.
7447 @cindex VFP11_DENORM_FIX
7448 @kindex --vfp11-denorm-fix
7449 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
7450 bug in certain VFP11 coprocessor hardware, which sometimes allows
7451 instructions with denorm operands (which must be handled by support code)
7452 to have those operands overwritten by subsequent instructions before
7453 the support code can read the intended values.
7455 The bug may be avoided in scalar mode if you allow at least one
7456 intervening instruction between a VFP11 instruction which uses a register
7457 and another instruction which writes to the same register, or at least two
7458 intervening instructions if vector mode is in use. The bug only affects
7459 full-compliance floating-point mode: you do not need this workaround if
7460 you are using "runfast" mode. Please contact ARM for further details.
7462 If you know you are using buggy VFP11 hardware, you can
7463 enable this workaround by specifying the linker option
7464 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
7465 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
7466 vector mode (the latter also works for scalar code). The default is
7467 @samp{--vfp-denorm-fix=none}.
7469 If the workaround is enabled, instructions are scanned for
7470 potentially-troublesome sequences, and a veneer is created for each
7471 such sequence which may trigger the erratum. The veneer consists of the
7472 first instruction of the sequence and a branch back to the subsequent
7473 instruction. The original instruction is then replaced with a branch to
7474 the veneer. The extra cycles required to call and return from the veneer
7475 are sufficient to avoid the erratum in both the scalar and vector cases.
7477 @cindex ARM1176 erratum workaround
7478 @kindex --fix-arm1176
7479 @kindex --no-fix-arm1176
7480 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
7481 in certain ARM1176 processors. The workaround is enabled by default if you
7482 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
7483 unconditionally by specifying @samp{--no-fix-arm1176}.
7485 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
7486 Programmer Advice Notice'' available on the ARM documentation website at:
7487 http://infocenter.arm.com/.
7489 @cindex STM32L4xx erratum workaround
7490 @kindex --fix-stm32l4xx-629360
7492 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
7493 workaround for a bug in the bus matrix / memory controller for some of
7494 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
7495 off-chip memory via the affected bus for bus reads of 9 words or more,
7496 the bus can generate corrupt data and/or abort. These are only
7497 core-initiated accesses (not DMA), and might affect any access:
7498 integer loads such as LDM, POP and floating-point loads such as VLDM,
7499 VPOP. Stores are not affected.
7501 The bug can be avoided by splitting memory accesses into the
7502 necessary chunks to keep bus reads below 8 words.
7504 The workaround is not enabled by default, this is equivalent to use
7505 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
7506 STM32L4xx hardware, you can enable the workaround by specifying the
7507 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
7508 @samp{--fix-stm32l4xx-629360=default}.
7510 If the workaround is enabled, instructions are scanned for
7511 potentially-troublesome sequences, and a veneer is created for each
7512 such sequence which may trigger the erratum. The veneer consists in a
7513 replacement sequence emulating the behaviour of the original one and a
7514 branch back to the subsequent instruction. The original instruction is
7515 then replaced with a branch to the veneer.
7517 The workaround does not always preserve the memory access order for
7518 the LDMDB instruction, when the instruction loads the PC.
7520 The workaround is not able to handle problematic instructions when
7521 they are in the middle of an IT block, since a branch is not allowed
7522 there. In that case, the linker reports a warning and no replacement
7525 The workaround is not able to replace problematic instructions with a
7526 PC-relative branch instruction if the @samp{.text} section is too
7527 large. In that case, when the branch that replaces the original code
7528 cannot be encoded, the linker reports a warning and no replacement
7531 @cindex NO_ENUM_SIZE_WARNING
7532 @kindex --no-enum-size-warning
7533 The @option{--no-enum-size-warning} switch prevents the linker from
7534 warning when linking object files that specify incompatible EABI
7535 enumeration size attributes. For example, with this switch enabled,
7536 linking of an object file using 32-bit enumeration values with another
7537 using enumeration values fitted into the smallest possible space will
7540 @cindex NO_WCHAR_SIZE_WARNING
7541 @kindex --no-wchar-size-warning
7542 The @option{--no-wchar-size-warning} switch prevents the linker from
7543 warning when linking object files that specify incompatible EABI
7544 @code{wchar_t} size attributes. For example, with this switch enabled,
7545 linking of an object file using 32-bit @code{wchar_t} values with another
7546 using 16-bit @code{wchar_t} values will not be diagnosed.
7549 @kindex --pic-veneer
7550 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
7551 ARM/Thumb interworking veneers, even if the rest of the binary
7552 is not PIC. This avoids problems on uClinux targets where
7553 @samp{--emit-relocs} is used to generate relocatable binaries.
7555 @cindex STUB_GROUP_SIZE
7556 @kindex --stub-group-size=@var{N}
7557 The linker will automatically generate and insert small sequences of
7558 code into a linked ARM ELF executable whenever an attempt is made to
7559 perform a function call to a symbol that is too far away. The
7560 placement of these sequences of instructions - called stubs - is
7561 controlled by the command-line option @option{--stub-group-size=N}.
7562 The placement is important because a poor choice can create a need for
7563 duplicate stubs, increasing the code size. The linker will try to
7564 group stubs together in order to reduce interruptions to the flow of
7565 code, but it needs guidance as to how big these groups should be and
7566 where they should be placed.
7568 The value of @samp{N}, the parameter to the
7569 @option{--stub-group-size=} option controls where the stub groups are
7570 placed. If it is negative then all stubs are placed after the first
7571 branch that needs them. If it is positive then the stubs can be
7572 placed either before or after the branches that need them. If the
7573 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
7574 exactly where to place groups of stubs, using its built in heuristics.
7575 A value of @samp{N} greater than 1 (or smaller than -1) tells the
7576 linker that a single group of stubs can service at most @samp{N} bytes
7577 from the input sections.
7579 The default, if @option{--stub-group-size=} is not specified, is
7582 Farcalls stubs insertion is fully supported for the ARM-EABI target
7583 only, because it relies on object files properties not present
7586 @cindex Cortex-A8 erratum workaround
7587 @kindex --fix-cortex-a8
7588 @kindex --no-fix-cortex-a8
7589 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}.
7591 The erratum only affects Thumb-2 code. Please contact ARM for further details.
7593 @cindex Cortex-A53 erratum 835769 workaround
7594 @kindex --fix-cortex-a53-835769
7595 @kindex --no-fix-cortex-a53-835769
7596 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}.
7598 Please contact ARM for further details.
7600 @kindex --merge-exidx-entries
7601 @kindex --no-merge-exidx-entries
7602 @cindex Merging exidx entries
7603 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
7606 @cindex 32-bit PLT entries
7607 The @samp{--long-plt} option enables the use of 16 byte PLT entries
7608 which support up to 4Gb of code. The default is to use 12 byte PLT
7609 entries which only support 512Mb of code.
7611 @kindex --no-apply-dynamic-relocs
7612 @cindex AArch64 rela addend
7613 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
7614 link-time values for dynamic relocations.
7616 @cindex Placement of SG veneers
7617 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
7618 Its start address must be set, either with the command-line option
7619 @samp{--section-start} or in a linker script, to indicate where to place these
7622 @kindex --cmse-implib
7623 @cindex Secure gateway import library
7624 The @samp{--cmse-implib} option requests that the import libraries
7625 specified by the @samp{--out-implib} and @samp{--in-implib} options are
7626 secure gateway import libraries, suitable for linking a non-secure
7627 executable against secure code as per ARMv8-M Security Extensions.
7629 @kindex --in-implib=@var{file}
7630 @cindex Input import library
7631 The @samp{--in-implib=file} specifies an input import library whose symbols
7632 must keep the same address in the executable being produced. A warning is
7633 given if no @samp{--out-implib} is given but new symbols have been introduced
7634 in the executable that should be listed in its import library. Otherwise, if
7635 @samp{--out-implib} is specified, the symbols are added to the output import
7636 library. A warning is also given if some symbols present in the input import
7637 library have disappeared from the executable. This option is only effective
7638 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
7652 @section @command{ld} and HPPA 32-bit ELF Support
7653 @cindex HPPA multiple sub-space stubs
7654 @kindex --multi-subspace
7655 When generating a shared library, @command{ld} will by default generate
7656 import stubs suitable for use with a single sub-space application.
7657 The @samp{--multi-subspace} switch causes @command{ld} to generate export
7658 stubs, and different (larger) import stubs suitable for use with
7659 multiple sub-spaces.
7661 @cindex HPPA stub grouping
7662 @kindex --stub-group-size=@var{N}
7663 Long branch stubs and import/export stubs are placed by @command{ld} in
7664 stub sections located between groups of input sections.
7665 @samp{--stub-group-size} specifies the maximum size of a group of input
7666 sections handled by one stub section. Since branch offsets are signed,
7667 a stub section may serve two groups of input sections, one group before
7668 the stub section, and one group after it. However, when using
7669 conditional branches that require stubs, it may be better (for branch
7670 prediction) that stub sections only serve one group of input sections.
7671 A negative value for @samp{N} chooses this scheme, ensuring that
7672 branches to stubs always use a negative offset. Two special values of
7673 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7674 @command{ld} to automatically size input section groups for the branch types
7675 detected, with the same behaviour regarding stub placement as other
7676 positive or negative values of @samp{N} respectively.
7678 Note that @samp{--stub-group-size} does not split input sections. A
7679 single input section larger than the group size specified will of course
7680 create a larger group (of one section). If input sections are too
7681 large, it may not be possible for a branch to reach its stub.
7694 @section @command{ld} and the Motorola 68K family
7696 @cindex Motorola 68K GOT generation
7697 @kindex --got=@var{type}
7698 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
7699 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
7700 @samp{target}. When @samp{target} is selected the linker chooses
7701 the default GOT generation scheme for the current target.
7702 @samp{single} tells the linker to generate a single GOT with
7703 entries only at non-negative offsets.
7704 @samp{negative} instructs the linker to generate a single GOT with
7705 entries at both negative and positive offsets. Not all environments
7707 @samp{multigot} allows the linker to generate several GOTs in the
7708 output file. All GOT references from a single input object
7709 file access the same GOT, but references from different input object
7710 files might access different GOTs. Not all environments support such GOTs.
7723 @section @command{ld} and the MIPS family
7725 @cindex MIPS microMIPS instruction choice selection
7728 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
7729 microMIPS instructions used in code generated by the linker, such as that
7730 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
7731 used, then the linker only uses 32-bit instruction encodings. By default
7732 or if @samp{--no-insn32} is used, all instruction encodings are used,
7733 including 16-bit ones where possible.
7735 @cindex MIPS branch relocation check control
7736 @kindex --ignore-branch-isa
7737 @kindex --no-ignore-branch-isa
7738 The @samp{--ignore-branch-isa} and @samp{--no-ignore-branch-isa} options
7739 control branch relocation checks for invalid ISA mode transitions. If
7740 @samp{--ignore-branch-isa} is used, then the linker accepts any branch
7741 relocations and any ISA mode transition required is lost in relocation
7742 calculation, except for some cases of @code{BAL} instructions which meet
7743 relaxation conditions and are converted to equivalent @code{JALX}
7744 instructions as the associated relocation is calculated. By default
7745 or if @samp{--no-ignore-branch-isa} is used a check is made causing
7746 the loss of an ISA mode transition to produce an error.
7759 @section @code{ld} and MMIX
7760 For MMIX, there is a choice of generating @code{ELF} object files or
7761 @code{mmo} object files when linking. The simulator @code{mmix}
7762 understands the @code{mmo} format. The binutils @code{objcopy} utility
7763 can translate between the two formats.
7765 There is one special section, the @samp{.MMIX.reg_contents} section.
7766 Contents in this section is assumed to correspond to that of global
7767 registers, and symbols referring to it are translated to special symbols,
7768 equal to registers. In a final link, the start address of the
7769 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7770 global register multiplied by 8. Register @code{$255} is not included in
7771 this section; it is always set to the program entry, which is at the
7772 symbol @code{Main} for @code{mmo} files.
7774 Global symbols with the prefix @code{__.MMIX.start.}, for example
7775 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7776 The default linker script uses these to set the default start address
7779 Initial and trailing multiples of zero-valued 32-bit words in a section,
7780 are left out from an mmo file.
7793 @section @code{ld} and MSP430
7794 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7795 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7796 just pass @samp{-m help} option to the linker).
7798 @cindex MSP430 extra sections
7799 The linker will recognize some extra sections which are MSP430 specific:
7802 @item @samp{.vectors}
7803 Defines a portion of ROM where interrupt vectors located.
7805 @item @samp{.bootloader}
7806 Defines the bootloader portion of the ROM (if applicable). Any code
7807 in this section will be uploaded to the MPU.
7809 @item @samp{.infomem}
7810 Defines an information memory section (if applicable). Any code in
7811 this section will be uploaded to the MPU.
7813 @item @samp{.infomemnobits}
7814 This is the same as the @samp{.infomem} section except that any code
7815 in this section will not be uploaded to the MPU.
7817 @item @samp{.noinit}
7818 Denotes a portion of RAM located above @samp{.bss} section.
7820 The last two sections are used by gcc.
7824 @cindex MSP430 Options
7825 @kindex --code-region
7826 @item --code-region=[either,lower,upper,none]
7827 This will transform .text* sections to [either,lower,upper].text* sections. The
7828 argument passed to GCC for -mcode-region is propagated to the linker
7831 @kindex --data-region
7832 @item --data-region=[either,lower,upper,none]
7833 This will transform .data*, .bss* and .rodata* sections to
7834 [either,lower,upper].[data,bss,rodata]* sections. The argument passed to GCC
7835 for -mdata-region is propagated to the linker using this option.
7837 @kindex --disable-sec-transformation
7838 @item --disable-sec-transformation
7839 Prevent the transformation of sections as specified by the @code{--code-region}
7840 and @code{--data-region} options.
7841 This is useful if you are compiling and linking using a single call to the GCC
7842 wrapper, and want to compile the source files using -m[code,data]-region but
7843 not transform the sections for prebuilt libraries and objects.
7857 @section @code{ld} and NDS32
7858 @kindex relaxing on NDS32
7859 For NDS32, there are some options to select relaxation behavior. The linker
7860 relaxes objects according to these options.
7863 @item @samp{--m[no-]fp-as-gp}
7864 Disable/enable fp-as-gp relaxation.
7866 @item @samp{--mexport-symbols=FILE}
7867 Exporting symbols and their address into FILE as linker script.
7869 @item @samp{--m[no-]ex9}
7870 Disable/enable link-time EX9 relaxation.
7872 @item @samp{--mexport-ex9=FILE}
7873 Export the EX9 table after linking.
7875 @item @samp{--mimport-ex9=FILE}
7876 Import the Ex9 table for EX9 relaxation.
7878 @item @samp{--mupdate-ex9}
7879 Update the existing EX9 table.
7881 @item @samp{--mex9-limit=NUM}
7882 Maximum number of entries in the ex9 table.
7884 @item @samp{--mex9-loop-aware}
7885 Avoid generating the EX9 instruction inside the loop.
7887 @item @samp{--m[no-]ifc}
7888 Disable/enable the link-time IFC optimization.
7890 @item @samp{--mifc-loop-aware}
7891 Avoid generating the IFC instruction inside the loop.
7905 @section @command{ld} and the Altera Nios II
7906 @cindex Nios II call relaxation
7907 @kindex --relax on Nios II
7909 Call and immediate jump instructions on Nios II processors are limited to
7910 transferring control to addresses in the same 256MB memory segment,
7911 which may result in @command{ld} giving
7912 @samp{relocation truncated to fit} errors with very large programs.
7913 The command-line option @option{--relax} enables the generation of
7914 trampolines that can access the entire 32-bit address space for calls
7915 outside the normal @code{call} and @code{jmpi} address range. These
7916 trampolines are inserted at section boundaries, so may not themselves
7917 be reachable if an input section and its associated call trampolines are
7920 The @option{--relax} option is enabled by default unless @option{-r}
7921 is also specified. You can disable trampoline generation by using the
7922 @option{--no-relax} linker option. You can also disable this optimization
7923 locally by using the @samp{set .noat} directive in assembly-language
7924 source files, as the linker-inserted trampolines use the @code{at}
7925 register as a temporary.
7927 Note that the linker @option{--relax} option is independent of assembler
7928 relaxation options, and that using the GNU assembler's @option{-relax-all}
7929 option interferes with the linker's more selective call instruction relaxation.
7942 @section @command{ld} and PowerPC 32-bit ELF Support
7943 @cindex PowerPC long branches
7944 @kindex --relax on PowerPC
7945 Branches on PowerPC processors are limited to a signed 26-bit
7946 displacement, which may result in @command{ld} giving
7947 @samp{relocation truncated to fit} errors with very large programs.
7948 @samp{--relax} enables the generation of trampolines that can access
7949 the entire 32-bit address space. These trampolines are inserted at
7950 section boundaries, so may not themselves be reachable if an input
7951 section exceeds 33M in size. You may combine @samp{-r} and
7952 @samp{--relax} to add trampolines in a partial link. In that case
7953 both branches to undefined symbols and inter-section branches are also
7954 considered potentially out of range, and trampolines inserted.
7956 @cindex PowerPC ELF32 options
7961 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7962 generates code capable of using a newer PLT and GOT layout that has
7963 the security advantage of no executable section ever needing to be
7964 writable and no writable section ever being executable. PowerPC
7965 @command{ld} will generate this layout, including stubs to access the
7966 PLT, if all input files (including startup and static libraries) were
7967 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7968 BSS PLT (and GOT layout) which can give slightly better performance.
7970 @kindex --secure-plt
7972 @command{ld} will use the new PLT and GOT layout if it is linking new
7973 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7974 when linking non-PIC code. This option requests the new PLT and GOT
7975 layout. A warning will be given if some object file requires the old
7981 The new secure PLT and GOT are placed differently relative to other
7982 sections compared to older BSS PLT and GOT placement. The location of
7983 @code{.plt} must change because the new secure PLT is an initialized
7984 section while the old PLT is uninitialized. The reason for the
7985 @code{.got} change is more subtle: The new placement allows
7986 @code{.got} to be read-only in applications linked with
7987 @samp{-z relro -z now}. However, this placement means that
7988 @code{.sdata} cannot always be used in shared libraries, because the
7989 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7990 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7991 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7992 really only useful for other compilers that may do so.
7994 @cindex PowerPC stub symbols
7995 @kindex --emit-stub-syms
7996 @item --emit-stub-syms
7997 This option causes @command{ld} to label linker stubs with a local
7998 symbol that encodes the stub type and destination.
8000 @cindex PowerPC TLS optimization
8001 @kindex --no-tls-optimize
8002 @item --no-tls-optimize
8003 PowerPC @command{ld} normally performs some optimization of code
8004 sequences used to access Thread-Local Storage. Use this option to
8005 disable the optimization.
8018 @node PowerPC64 ELF64
8019 @section @command{ld} and PowerPC64 64-bit ELF Support
8021 @cindex PowerPC64 ELF64 options
8023 @cindex PowerPC64 stub grouping
8024 @kindex --stub-group-size
8025 @item --stub-group-size
8026 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
8027 by @command{ld} in stub sections located between groups of input sections.
8028 @samp{--stub-group-size} specifies the maximum size of a group of input
8029 sections handled by one stub section. Since branch offsets are signed,
8030 a stub section may serve two groups of input sections, one group before
8031 the stub section, and one group after it. However, when using
8032 conditional branches that require stubs, it may be better (for branch
8033 prediction) that stub sections only serve one group of input sections.
8034 A negative value for @samp{N} chooses this scheme, ensuring that
8035 branches to stubs always use a negative offset. Two special values of
8036 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
8037 @command{ld} to automatically size input section groups for the branch types
8038 detected, with the same behaviour regarding stub placement as other
8039 positive or negative values of @samp{N} respectively.
8041 Note that @samp{--stub-group-size} does not split input sections. A
8042 single input section larger than the group size specified will of course
8043 create a larger group (of one section). If input sections are too
8044 large, it may not be possible for a branch to reach its stub.
8046 @cindex PowerPC64 stub symbols
8047 @kindex --emit-stub-syms
8048 @item --emit-stub-syms
8049 This option causes @command{ld} to label linker stubs with a local
8050 symbol that encodes the stub type and destination.
8052 @cindex PowerPC64 dot symbols
8054 @kindex --no-dotsyms
8057 These two options control how @command{ld} interprets version patterns
8058 in a version script. Older PowerPC64 compilers emitted both a
8059 function descriptor symbol with the same name as the function, and a
8060 code entry symbol with the name prefixed by a dot (@samp{.}). To
8061 properly version a function @samp{foo}, the version script thus needs
8062 to control both @samp{foo} and @samp{.foo}. The option
8063 @samp{--dotsyms}, on by default, automatically adds the required
8064 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
8067 @cindex PowerPC64 register save/restore functions
8068 @kindex --save-restore-funcs
8069 @kindex --no-save-restore-funcs
8070 @item --save-restore-funcs
8071 @itemx --no-save-restore-funcs
8072 These two options control whether PowerPC64 @command{ld} automatically
8073 provides out-of-line register save and restore functions used by
8074 @samp{-Os} code. The default is to provide any such referenced
8075 function for a normal final link, and to not do so for a relocatable
8078 @cindex PowerPC64 TLS optimization
8079 @kindex --no-tls-optimize
8080 @item --no-tls-optimize
8081 PowerPC64 @command{ld} normally performs some optimization of code
8082 sequences used to access Thread-Local Storage. Use this option to
8083 disable the optimization.
8085 @cindex PowerPC64 __tls_get_addr optimization
8086 @kindex --tls-get-addr-optimize
8087 @kindex --no-tls-get-addr-optimize
8088 @kindex --tls-get-addr-regsave
8089 @kindex --no-tls-get-addr-regsave
8090 @item --tls-get-addr-optimize
8091 @itemx --no-tls-get-addr-optimize
8092 These options control how PowerPC64 @command{ld} uses a special
8093 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
8094 an optimization that allows the second and subsequent calls to
8095 @code{__tls_get_addr} for a given symbol to be resolved by the special
8096 stub without calling in to glibc. By default the linker enables
8097 generation of the stub when glibc advertises the availability of
8099 Using @option{--tls-get-addr-optimize} with an older glibc won't do
8100 much besides slow down your applications, but may be useful if linking
8101 an application against an older glibc with the expectation that it
8102 will normally be used on systems having a newer glibc.
8103 @option{--tls-get-addr-regsave} forces generation of a stub that saves
8104 and restores volatile registers around the call into glibc. Normally,
8105 this is done when the linker detects a call to __tls_get_addr_desc.
8106 Such calls then go via the register saving stub to __tls_get_addr_opt.
8107 @option{--no-tls-get-addr-regsave} disables generation of the
8110 @cindex PowerPC64 OPD optimization
8111 @kindex --no-opd-optimize
8112 @item --no-opd-optimize
8113 PowerPC64 @command{ld} normally removes @code{.opd} section entries
8114 corresponding to deleted link-once functions, or functions removed by
8115 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
8116 Use this option to disable @code{.opd} optimization.
8118 @cindex PowerPC64 OPD spacing
8119 @kindex --non-overlapping-opd
8120 @item --non-overlapping-opd
8121 Some PowerPC64 compilers have an option to generate compressed
8122 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
8123 the static chain pointer (unused in C) with the first word of the next
8124 entry. This option expands such entries to the full 24 bytes.
8126 @cindex PowerPC64 TOC optimization
8127 @kindex --no-toc-optimize
8128 @item --no-toc-optimize
8129 PowerPC64 @command{ld} normally removes unused @code{.toc} section
8130 entries. Such entries are detected by examining relocations that
8131 reference the TOC in code sections. A reloc in a deleted code section
8132 marks a TOC word as unneeded, while a reloc in a kept code section
8133 marks a TOC word as needed. Since the TOC may reference itself, TOC
8134 relocs are also examined. TOC words marked as both needed and
8135 unneeded will of course be kept. TOC words without any referencing
8136 reloc are assumed to be part of a multi-word entry, and are kept or
8137 discarded as per the nearest marked preceding word. This works
8138 reliably for compiler generated code, but may be incorrect if assembly
8139 code is used to insert TOC entries. Use this option to disable the
8142 @cindex PowerPC64 inline PLT call optimization
8143 @kindex --no-inline-optimize
8144 @item --no-inline-optimize
8145 PowerPC64 @command{ld} normally replaces inline PLT call sequences
8146 marked with @code{R_PPC64_PLTSEQ}, @code{R_PPC64_PLTCALL},
8147 @code{R_PPC64_PLT16_HA} and @code{R_PPC64_PLT16_LO_DS} relocations by
8148 a number of @code{nop}s and a direct call when the function is defined
8149 locally and can't be overridden by some other definition. This option
8150 disables that optimization.
8152 @cindex PowerPC64 multi-TOC
8153 @kindex --no-multi-toc
8154 @item --no-multi-toc
8155 If given any toc option besides @code{-mcmodel=medium} or
8156 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
8158 entries are accessed with a 16-bit offset from r2. This limits the
8159 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
8160 grouping code sections such that each group uses less than 64K for its
8161 TOC entries, then inserts r2 adjusting stubs between inter-group
8162 calls. @command{ld} does not split apart input sections, so cannot
8163 help if a single input file has a @code{.toc} section that exceeds
8164 64K, most likely from linking multiple files with @command{ld -r}.
8165 Use this option to turn off this feature.
8167 @cindex PowerPC64 TOC sorting
8168 @kindex --no-toc-sort
8170 By default, @command{ld} sorts TOC sections so that those whose file
8171 happens to have a section called @code{.init} or @code{.fini} are
8172 placed first, followed by TOC sections referenced by code generated
8173 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
8174 referenced only by code generated with PowerPC64 gcc's
8175 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
8176 results in better TOC grouping for multi-TOC. Use this option to turn
8179 @cindex PowerPC64 PLT stub alignment
8181 @kindex --no-plt-align
8183 @itemx --no-plt-align
8184 Use these options to control whether individual PLT call stubs are
8185 aligned to a 32-byte boundary, or to the specified power of two
8186 boundary when using @code{--plt-align=}. A negative value may be
8187 specified to pad PLT call stubs so that they do not cross the
8188 specified power of two boundary (or the minimum number of boundaries
8189 if a PLT stub is so large that it must cross a boundary). By default
8190 PLT call stubs are aligned to 32-byte boundaries.
8192 @cindex PowerPC64 PLT call stub static chain
8193 @kindex --plt-static-chain
8194 @kindex --no-plt-static-chain
8195 @item --plt-static-chain
8196 @itemx --no-plt-static-chain
8197 Use these options to control whether PLT call stubs load the static
8198 chain pointer (r11). @code{ld} defaults to not loading the static
8199 chain since there is never any need to do so on a PLT call.
8201 @cindex PowerPC64 PLT call stub thread safety
8202 @kindex --plt-thread-safe
8203 @kindex --no-plt-thread-safe
8204 @item --plt-thread-safe
8205 @itemx --no-plt-thread-safe
8206 With power7's weakly ordered memory model, it is possible when using
8207 lazy binding for ld.so to update a plt entry in one thread and have
8208 another thread see the individual plt entry words update in the wrong
8209 order, despite ld.so carefully writing in the correct order and using
8210 memory write barriers. To avoid this we need some sort of read
8211 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
8212 looks for calls to commonly used functions that create threads, and if
8213 seen, adds the necessary barriers. Use these options to change the
8216 @cindex PowerPC64 ELFv2 PLT localentry optimization
8217 @kindex --plt-localentry
8218 @kindex --no-plt-localentry
8219 @item --plt-localentry
8220 @itemx --no-localentry
8221 ELFv2 functions with localentry:0 are those with a single entry point,
8222 ie. global entry == local entry, and that have no requirement on r2
8223 (the TOC/GOT pointer) or r12, and guarantee r2 is unchanged on return.
8224 Such an external function can be called via the PLT without saving r2
8225 or restoring it on return, avoiding a common load-hit-store for small
8226 functions. The optimization is attractive, with up to 40% reduction
8227 in execution time for a small function, but can result in symbol
8228 interposition failures. Also, minor changes in a shared library,
8229 including system libraries, can cause a function that was localentry:0
8230 to become localentry:8. This will result in a dynamic loader
8231 complaint and failure to run. The option is experimental, use with
8232 care. @option{--no-plt-localentry} is the default.
8234 @cindex PowerPC64 Power10 stubs
8235 @kindex --power10-stubs
8236 @kindex --no-power10-stubs
8237 @item --power10-stubs
8238 @itemx --no-power10-stubs
8239 When PowerPC64 @command{ld} links input object files containing
8240 relocations used on power10 prefixed instructions it normally creates
8241 linkage stubs (PLT call and long branch) using power10 instructions
8242 for @code{@@notoc} PLT calls where @code{r2} is not known. The
8243 power10 notoc stubs are smaller and faster, so are preferred for
8244 power10. @option{--power10-stubs} and @option{--no-power10-stubs}
8245 allow you to override the linker's selection of stub instructions.
8246 @option{--power10-stubs=auto} allows the user to select the default
8261 @section @command{ld} and S/390 ELF Support
8263 @cindex S/390 ELF options
8267 @kindex --s390-pgste
8269 This option marks the result file with a @code{PT_S390_PGSTE}
8270 segment. The Linux kernel is supposed to allocate 4k page tables for
8271 binaries marked that way.
8285 @section @command{ld} and SPU ELF Support
8287 @cindex SPU ELF options
8293 This option marks an executable as a PIC plugin module.
8295 @cindex SPU overlays
8296 @kindex --no-overlays
8298 Normally, @command{ld} recognizes calls to functions within overlay
8299 regions, and redirects such calls to an overlay manager via a stub.
8300 @command{ld} also provides a built-in overlay manager. This option
8301 turns off all this special overlay handling.
8303 @cindex SPU overlay stub symbols
8304 @kindex --emit-stub-syms
8305 @item --emit-stub-syms
8306 This option causes @command{ld} to label overlay stubs with a local
8307 symbol that encodes the stub type and destination.
8309 @cindex SPU extra overlay stubs
8310 @kindex --extra-overlay-stubs
8311 @item --extra-overlay-stubs
8312 This option causes @command{ld} to add overlay call stubs on all
8313 function calls out of overlay regions. Normally stubs are not added
8314 on calls to non-overlay regions.
8316 @cindex SPU local store size
8317 @kindex --local-store=lo:hi
8318 @item --local-store=lo:hi
8319 @command{ld} usually checks that a final executable for SPU fits in
8320 the address range 0 to 256k. This option may be used to change the
8321 range. Disable the check entirely with @option{--local-store=0:0}.
8324 @kindex --stack-analysis
8325 @item --stack-analysis
8326 SPU local store space is limited. Over-allocation of stack space
8327 unnecessarily limits space available for code and data, while
8328 under-allocation results in runtime failures. If given this option,
8329 @command{ld} will provide an estimate of maximum stack usage.
8330 @command{ld} does this by examining symbols in code sections to
8331 determine the extents of functions, and looking at function prologues
8332 for stack adjusting instructions. A call-graph is created by looking
8333 for relocations on branch instructions. The graph is then searched
8334 for the maximum stack usage path. Note that this analysis does not
8335 find calls made via function pointers, and does not handle recursion
8336 and other cycles in the call graph. Stack usage may be
8337 under-estimated if your code makes such calls. Also, stack usage for
8338 dynamic allocation, e.g. alloca, will not be detected. If a link map
8339 is requested, detailed information about each function's stack usage
8340 and calls will be given.
8343 @kindex --emit-stack-syms
8344 @item --emit-stack-syms
8345 This option, if given along with @option{--stack-analysis} will result
8346 in @command{ld} emitting stack sizing symbols for each function.
8347 These take the form @code{__stack_<function_name>} for global
8348 functions, and @code{__stack_<number>_<function_name>} for static
8349 functions. @code{<number>} is the section id in hex. The value of
8350 such symbols is the stack requirement for the corresponding function.
8351 The symbol size will be zero, type @code{STT_NOTYPE}, binding
8352 @code{STB_LOCAL}, and section @code{SHN_ABS}.
8366 @section @command{ld}'s Support for Various TI COFF Versions
8367 @cindex TI COFF versions
8368 @kindex --format=@var{version}
8369 The @samp{--format} switch allows selection of one of the various
8370 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
8371 also supported. The TI COFF versions also vary in header byte-order
8372 format; @command{ld} will read any version or byte order, but the output
8373 header format depends on the default specified by the specific target.
8386 @section @command{ld} and WIN32 (cygwin/mingw)
8388 This section describes some of the win32 specific @command{ld} issues.
8389 See @ref{Options,,Command-line Options} for detailed description of the
8390 command-line options mentioned here.
8393 @cindex import libraries
8394 @item import libraries
8395 The standard Windows linker creates and uses so-called import
8396 libraries, which contains information for linking to dll's. They are
8397 regular static archives and are handled as any other static
8398 archive. The cygwin and mingw ports of @command{ld} have specific
8399 support for creating such libraries provided with the
8400 @samp{--out-implib} command-line option.
8402 @item exporting DLL symbols
8403 @cindex exporting DLL symbols
8404 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
8407 @item using auto-export functionality
8408 @cindex using auto-export functionality
8409 By default @command{ld} exports symbols with the auto-export functionality,
8410 which is controlled by the following command-line options:
8413 @item --export-all-symbols [This is the default]
8414 @item --exclude-symbols
8415 @item --exclude-libs
8416 @item --exclude-modules-for-implib
8417 @item --version-script
8420 When auto-export is in operation, @command{ld} will export all the non-local
8421 (global and common) symbols it finds in a DLL, with the exception of a few
8422 symbols known to belong to the system's runtime and libraries. As it will
8423 often not be desirable to export all of a DLL's symbols, which may include
8424 private functions that are not part of any public interface, the command-line
8425 options listed above may be used to filter symbols out from the list for
8426 exporting. The @samp{--output-def} option can be used in order to see the
8427 final list of exported symbols with all exclusions taken into effect.
8429 If @samp{--export-all-symbols} is not given explicitly on the
8430 command line, then the default auto-export behavior will be @emph{disabled}
8431 if either of the following are true:
8434 @item A DEF file is used.
8435 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
8438 @item using a DEF file
8439 @cindex using a DEF file
8440 Another way of exporting symbols is using a DEF file. A DEF file is
8441 an ASCII file containing definitions of symbols which should be
8442 exported when a dll is created. Usually it is named @samp{<dll
8443 name>.def} and is added as any other object file to the linker's
8444 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
8447 gcc -o <output> <objectfiles> <dll name>.def
8450 Using a DEF file turns off the normal auto-export behavior, unless the
8451 @samp{--export-all-symbols} option is also used.
8453 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
8456 LIBRARY "xyz.dll" BASE=0x20000000
8462 another_foo = abc.dll.afoo
8468 This example defines a DLL with a non-default base address and seven
8469 symbols in the export table. The third exported symbol @code{_bar} is an
8470 alias for the second. The fourth symbol, @code{another_foo} is resolved
8471 by "forwarding" to another module and treating it as an alias for
8472 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
8473 @code{var1} is declared to be a data object. The @samp{doo} symbol in
8474 export library is an alias of @samp{foo}, which gets the string name
8475 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
8476 symbol, which gets in export table the name @samp{var1}.
8478 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
8479 name of the output DLL. If @samp{<name>} does not include a suffix,
8480 the default library suffix, @samp{.DLL} is appended.
8482 When the .DEF file is used to build an application, rather than a
8483 library, the @code{NAME <name>} command should be used instead of
8484 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
8485 executable suffix, @samp{.EXE} is appended.
8487 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
8488 specification @code{BASE = <number>} may be used to specify a
8489 non-default base address for the image.
8491 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
8492 or they specify an empty string, the internal name is the same as the
8493 filename specified on the command line.
8495 The complete specification of an export symbol is:
8499 ( ( ( <name1> [ = <name2> ] )
8500 | ( <name1> = <module-name> . <external-name>))
8501 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
8504 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
8505 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
8506 @samp{<name1>} as a "forward" alias for the symbol
8507 @samp{<external-name>} in the DLL @samp{<module-name>}.
8508 Optionally, the symbol may be exported by the specified ordinal
8509 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
8510 string in import/export table for the symbol.
8512 The optional keywords that follow the declaration indicate:
8514 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
8515 will still be exported by its ordinal alias (either the value specified
8516 by the .def specification or, otherwise, the value assigned by the
8517 linker). The symbol name, however, does remain visible in the import
8518 library (if any), unless @code{PRIVATE} is also specified.
8520 @code{DATA}: The symbol is a variable or object, rather than a function.
8521 The import lib will export only an indirect reference to @code{foo} as
8522 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
8525 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
8526 well as @code{_imp__foo} into the import library. Both refer to the
8527 read-only import address table's pointer to the variable, not to the
8528 variable itself. This can be dangerous. If the user code fails to add
8529 the @code{dllimport} attribute and also fails to explicitly add the
8530 extra indirection that the use of the attribute enforces, the
8531 application will behave unexpectedly.
8533 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
8534 it into the static import library used to resolve imports at link time. The
8535 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
8536 API at runtime or by using the GNU ld extension of linking directly to
8537 the DLL without an import library.
8539 See ld/deffilep.y in the binutils sources for the full specification of
8540 other DEF file statements
8542 @cindex creating a DEF file
8543 While linking a shared dll, @command{ld} is able to create a DEF file
8544 with the @samp{--output-def <file>} command-line option.
8546 @item Using decorations
8547 @cindex Using decorations
8548 Another way of marking symbols for export is to modify the source code
8549 itself, so that when building the DLL each symbol to be exported is
8553 __declspec(dllexport) int a_variable
8554 __declspec(dllexport) void a_function(int with_args)
8557 All such symbols will be exported from the DLL. If, however,
8558 any of the object files in the DLL contain symbols decorated in
8559 this way, then the normal auto-export behavior is disabled, unless
8560 the @samp{--export-all-symbols} option is also used.
8562 Note that object files that wish to access these symbols must @emph{not}
8563 decorate them with dllexport. Instead, they should use dllimport,
8567 __declspec(dllimport) int a_variable
8568 __declspec(dllimport) void a_function(int with_args)
8571 This complicates the structure of library header files, because
8572 when included by the library itself the header must declare the
8573 variables and functions as dllexport, but when included by client
8574 code the header must declare them as dllimport. There are a number
8575 of idioms that are typically used to do this; often client code can
8576 omit the __declspec() declaration completely. See
8577 @samp{--enable-auto-import} and @samp{automatic data imports} for more
8581 @cindex automatic data imports
8582 @item automatic data imports
8583 The standard Windows dll format supports data imports from dlls only
8584 by adding special decorations (dllimport/dllexport), which let the
8585 compiler produce specific assembler instructions to deal with this
8586 issue. This increases the effort necessary to port existing Un*x
8587 code to these platforms, especially for large
8588 c++ libraries and applications. The auto-import feature, which was
8589 initially provided by Paul Sokolovsky, allows one to omit the
8590 decorations to achieve a behavior that conforms to that on POSIX/Un*x
8591 platforms. This feature is enabled with the @samp{--enable-auto-import}
8592 command-line option, although it is enabled by default on cygwin/mingw.
8593 The @samp{--enable-auto-import} option itself now serves mainly to
8594 suppress any warnings that are ordinarily emitted when linked objects
8595 trigger the feature's use.
8597 auto-import of variables does not always work flawlessly without
8598 additional assistance. Sometimes, you will see this message
8600 "variable '<var>' can't be auto-imported. Please read the
8601 documentation for ld's @code{--enable-auto-import} for details."
8603 The @samp{--enable-auto-import} documentation explains why this error
8604 occurs, and several methods that can be used to overcome this difficulty.
8605 One of these methods is the @emph{runtime pseudo-relocs} feature, described
8608 @cindex runtime pseudo-relocation
8609 For complex variables imported from DLLs (such as structs or classes),
8610 object files typically contain a base address for the variable and an
8611 offset (@emph{addend}) within the variable--to specify a particular
8612 field or public member, for instance. Unfortunately, the runtime loader used
8613 in win32 environments is incapable of fixing these references at runtime
8614 without the additional information supplied by dllimport/dllexport decorations.
8615 The standard auto-import feature described above is unable to resolve these
8618 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
8619 be resolved without error, while leaving the task of adjusting the references
8620 themselves (with their non-zero addends) to specialized code provided by the
8621 runtime environment. Recent versions of the cygwin and mingw environments and
8622 compilers provide this runtime support; older versions do not. However, the
8623 support is only necessary on the developer's platform; the compiled result will
8624 run without error on an older system.
8626 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
8629 @cindex direct linking to a dll
8630 @item direct linking to a dll
8631 The cygwin/mingw ports of @command{ld} support the direct linking,
8632 including data symbols, to a dll without the usage of any import
8633 libraries. This is much faster and uses much less memory than does the
8634 traditional import library method, especially when linking large
8635 libraries or applications. When @command{ld} creates an import lib, each
8636 function or variable exported from the dll is stored in its own bfd, even
8637 though a single bfd could contain many exports. The overhead involved in
8638 storing, loading, and processing so many bfd's is quite large, and explains the
8639 tremendous time, memory, and storage needed to link against particularly
8640 large or complex libraries when using import libs.
8642 Linking directly to a dll uses no extra command-line switches other than
8643 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
8644 of names to match each library. All that is needed from the developer's
8645 perspective is an understanding of this search, in order to force ld to
8646 select the dll instead of an import library.
8649 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
8650 to find, in the first directory of its search path,
8663 before moving on to the next directory in the search path.
8665 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
8666 where @samp{<prefix>} is set by the @command{ld} option
8667 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
8668 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
8671 Other win32-based unix environments, such as mingw or pw32, may use other
8672 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
8673 was originally intended to help avoid name conflicts among dll's built for the
8674 various win32/un*x environments, so that (for example) two versions of a zlib dll
8675 could coexist on the same machine.
8677 The generic cygwin/mingw path layout uses a @samp{bin} directory for
8678 applications and dll's and a @samp{lib} directory for the import
8679 libraries (using cygwin nomenclature):
8685 libxxx.dll.a (in case of dll's)
8686 libxxx.a (in case of static archive)
8689 Linking directly to a dll without using the import library can be
8692 1. Use the dll directly by adding the @samp{bin} path to the link line
8694 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
8697 However, as the dll's often have version numbers appended to their names
8698 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
8699 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
8700 not versioned, and do not have this difficulty.
8702 2. Create a symbolic link from the dll to a file in the @samp{lib}
8703 directory according to the above mentioned search pattern. This
8704 should be used to avoid unwanted changes in the tools needed for
8708 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
8711 Then you can link without any make environment changes.
8714 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
8717 This technique also avoids the version number problems, because the following is
8724 libxxx.dll.a -> ../bin/cygxxx-5.dll
8727 Linking directly to a dll without using an import lib will work
8728 even when auto-import features are exercised, and even when
8729 @samp{--enable-runtime-pseudo-relocs} is used.
8731 Given the improvements in speed and memory usage, one might justifiably
8732 wonder why import libraries are used at all. There are three reasons:
8734 1. Until recently, the link-directly-to-dll functionality did @emph{not}
8735 work with auto-imported data.
8737 2. Sometimes it is necessary to include pure static objects within the
8738 import library (which otherwise contains only bfd's for indirection
8739 symbols that point to the exports of a dll). Again, the import lib
8740 for the cygwin kernel makes use of this ability, and it is not
8741 possible to do this without an import lib.
8743 3. Symbol aliases can only be resolved using an import lib. This is
8744 critical when linking against OS-supplied dll's (eg, the win32 API)
8745 in which symbols are usually exported as undecorated aliases of their
8746 stdcall-decorated assembly names.
8748 So, import libs are not going away. But the ability to replace
8749 true import libs with a simple symbolic link to (or a copy of)
8750 a dll, in many cases, is a useful addition to the suite of tools
8751 binutils makes available to the win32 developer. Given the
8752 massive improvements in memory requirements during linking, storage
8753 requirements, and linking speed, we expect that many developers
8754 will soon begin to use this feature whenever possible.
8756 @item symbol aliasing
8758 @item adding additional names
8759 Sometimes, it is useful to export symbols with additional names.
8760 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
8761 exported as @samp{_foo} by using special directives in the DEF file
8762 when creating the dll. This will affect also the optional created
8763 import library. Consider the following DEF file:
8766 LIBRARY "xyz.dll" BASE=0x61000000
8773 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
8775 Another method for creating a symbol alias is to create it in the
8776 source code using the "weak" attribute:
8779 void foo () @{ /* Do something. */; @}
8780 void _foo () __attribute__ ((weak, alias ("foo")));
8783 See the gcc manual for more information about attributes and weak
8786 @item renaming symbols
8787 Sometimes it is useful to rename exports. For instance, the cygwin
8788 kernel does this regularly. A symbol @samp{_foo} can be exported as
8789 @samp{foo} but not as @samp{_foo} by using special directives in the
8790 DEF file. (This will also affect the import library, if it is
8791 created). In the following example:
8794 LIBRARY "xyz.dll" BASE=0x61000000
8800 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
8804 Note: using a DEF file disables the default auto-export behavior,
8805 unless the @samp{--export-all-symbols} command-line option is used.
8806 If, however, you are trying to rename symbols, then you should list
8807 @emph{all} desired exports in the DEF file, including the symbols
8808 that are not being renamed, and do @emph{not} use the
8809 @samp{--export-all-symbols} option. If you list only the
8810 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
8811 to handle the other symbols, then the both the new names @emph{and}
8812 the original names for the renamed symbols will be exported.
8813 In effect, you'd be aliasing those symbols, not renaming them,
8814 which is probably not what you wanted.
8816 @cindex weak externals
8817 @item weak externals
8818 The Windows object format, PE, specifies a form of weak symbols called
8819 weak externals. When a weak symbol is linked and the symbol is not
8820 defined, the weak symbol becomes an alias for some other symbol. There
8821 are three variants of weak externals:
8823 @item Definition is searched for in objects and libraries, historically
8824 called lazy externals.
8825 @item Definition is searched for only in other objects, not in libraries.
8826 This form is not presently implemented.
8827 @item No search; the symbol is an alias. This form is not presently
8830 As a GNU extension, weak symbols that do not specify an alternate symbol
8831 are supported. If the symbol is undefined when linking, the symbol
8832 uses a default value.
8834 @cindex aligned common symbols
8835 @item aligned common symbols
8836 As a GNU extension to the PE file format, it is possible to specify the
8837 desired alignment for a common symbol. This information is conveyed from
8838 the assembler or compiler to the linker by means of GNU-specific commands
8839 carried in the object file's @samp{.drectve} section, which are recognized
8840 by @command{ld} and respected when laying out the common symbols. Native
8841 tools will be able to process object files employing this GNU extension,
8842 but will fail to respect the alignment instructions, and may issue noisy
8843 warnings about unknown linker directives.
8858 @section @code{ld} and Xtensa Processors
8860 @cindex Xtensa processors
8861 The default @command{ld} behavior for Xtensa processors is to interpret
8862 @code{SECTIONS} commands so that lists of explicitly named sections in a
8863 specification with a wildcard file will be interleaved when necessary to
8864 keep literal pools within the range of PC-relative load offsets. For
8865 example, with the command:
8877 @command{ld} may interleave some of the @code{.literal}
8878 and @code{.text} sections from different object files to ensure that the
8879 literal pools are within the range of PC-relative load offsets. A valid
8880 interleaving might place the @code{.literal} sections from an initial
8881 group of files followed by the @code{.text} sections of that group of
8882 files. Then, the @code{.literal} sections from the rest of the files
8883 and the @code{.text} sections from the rest of the files would follow.
8885 @cindex @option{--relax} on Xtensa
8886 @cindex relaxing on Xtensa
8887 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8888 provides two important link-time optimizations. The first optimization
8889 is to combine identical literal values to reduce code size. A redundant
8890 literal will be removed and all the @code{L32R} instructions that use it
8891 will be changed to reference an identical literal, as long as the
8892 location of the replacement literal is within the offset range of all
8893 the @code{L32R} instructions. The second optimization is to remove
8894 unnecessary overhead from assembler-generated ``longcall'' sequences of
8895 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8896 range of direct @code{CALL@var{n}} instructions.
8898 For each of these cases where an indirect call sequence can be optimized
8899 to a direct call, the linker will change the @code{CALLX@var{n}}
8900 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8901 instruction, and remove the literal referenced by the @code{L32R}
8902 instruction if it is not used for anything else. Removing the
8903 @code{L32R} instruction always reduces code size but can potentially
8904 hurt performance by changing the alignment of subsequent branch targets.
8905 By default, the linker will always preserve alignments, either by
8906 switching some instructions between 24-bit encodings and the equivalent
8907 density instructions or by inserting a no-op in place of the @code{L32R}
8908 instruction that was removed. If code size is more important than
8909 performance, the @option{--size-opt} option can be used to prevent the
8910 linker from widening density instructions or inserting no-ops, except in
8911 a few cases where no-ops are required for correctness.
8913 The following Xtensa-specific command-line options can be used to
8916 @cindex Xtensa options
8919 When optimizing indirect calls to direct calls, optimize for code size
8920 more than performance. With this option, the linker will not insert
8921 no-ops or widen density instructions to preserve branch target
8922 alignment. There may still be some cases where no-ops are required to
8923 preserve the correctness of the code.
8925 @item --abi-windowed
8927 Choose ABI for the output object and for the generated PLT code.
8928 PLT code inserted by the linker must match ABI of the output object
8929 because windowed and call0 ABI use incompatible function call
8931 Default ABI is chosen by the ABI tag in the @code{.xtensa.info} section
8932 of the first input object.
8933 A warning is issued if ABI tags of input objects do not match each other
8934 or the chosen output object ABI.
8942 @ifclear SingleFormat
8947 @cindex object file management
8948 @cindex object formats available
8950 The linker accesses object and archive files using the BFD libraries.
8951 These libraries allow the linker to use the same routines to operate on
8952 object files whatever the object file format. A different object file
8953 format can be supported simply by creating a new BFD back end and adding
8954 it to the library. To conserve runtime memory, however, the linker and
8955 associated tools are usually configured to support only a subset of the
8956 object file formats available. You can use @code{objdump -i}
8957 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8958 list all the formats available for your configuration.
8960 @cindex BFD requirements
8961 @cindex requirements for BFD
8962 As with most implementations, BFD is a compromise between
8963 several conflicting requirements. The major factor influencing
8964 BFD design was efficiency: any time used converting between
8965 formats is time which would not have been spent had BFD not
8966 been involved. This is partly offset by abstraction payback; since
8967 BFD simplifies applications and back ends, more time and care
8968 may be spent optimizing algorithms for a greater speed.
8970 One minor artifact of the BFD solution which you should bear in
8971 mind is the potential for information loss. There are two places where
8972 useful information can be lost using the BFD mechanism: during
8973 conversion and during output. @xref{BFD information loss}.
8976 * BFD outline:: How it works: an outline of BFD
8980 @section How It Works: An Outline of BFD
8981 @cindex opening object files
8982 @include bfdsumm.texi
8985 @node Reporting Bugs
8986 @chapter Reporting Bugs
8987 @cindex bugs in @command{ld}
8988 @cindex reporting bugs in @command{ld}
8990 Your bug reports play an essential role in making @command{ld} reliable.
8992 Reporting a bug may help you by bringing a solution to your problem, or
8993 it may not. But in any case the principal function of a bug report is
8994 to help the entire community by making the next version of @command{ld}
8995 work better. Bug reports are your contribution to the maintenance of
8998 In order for a bug report to serve its purpose, you must include the
8999 information that enables us to fix the bug.
9002 * Bug Criteria:: Have you found a bug?
9003 * Bug Reporting:: How to report bugs
9007 @section Have You Found a Bug?
9008 @cindex bug criteria
9010 If you are not sure whether you have found a bug, here are some guidelines:
9013 @cindex fatal signal
9014 @cindex linker crash
9015 @cindex crash of linker
9017 If the linker gets a fatal signal, for any input whatever, that is a
9018 @command{ld} bug. Reliable linkers never crash.
9020 @cindex error on valid input
9022 If @command{ld} produces an error message for valid input, that is a bug.
9024 @cindex invalid input
9026 If @command{ld} does not produce an error message for invalid input, that
9027 may be a bug. In the general case, the linker can not verify that
9028 object files are correct.
9031 If you are an experienced user of linkers, your suggestions for
9032 improvement of @command{ld} are welcome in any case.
9036 @section How to Report Bugs
9038 @cindex @command{ld} bugs, reporting
9040 A number of companies and individuals offer support for @sc{gnu}
9041 products. If you obtained @command{ld} from a support organization, we
9042 recommend you contact that organization first.
9044 You can find contact information for many support companies and
9045 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
9049 Otherwise, send bug reports for @command{ld} to
9053 The fundamental principle of reporting bugs usefully is this:
9054 @strong{report all the facts}. If you are not sure whether to state a
9055 fact or leave it out, state it!
9057 Often people omit facts because they think they know what causes the
9058 problem and assume that some details do not matter. Thus, you might
9059 assume that the name of a symbol you use in an example does not
9060 matter. Well, probably it does not, but one cannot be sure. Perhaps
9061 the bug is a stray memory reference which happens to fetch from the
9062 location where that name is stored in memory; perhaps, if the name
9063 were different, the contents of that location would fool the linker
9064 into doing the right thing despite the bug. Play it safe and give a
9065 specific, complete example. That is the easiest thing for you to do,
9066 and the most helpful.
9068 Keep in mind that the purpose of a bug report is to enable us to fix
9069 the bug if it is new to us. Therefore, always write your bug reports
9070 on the assumption that the bug has not been reported previously.
9072 Sometimes people give a few sketchy facts and ask, ``Does this ring a
9073 bell?'' This cannot help us fix a bug, so it is basically useless. We
9074 respond by asking for enough details to enable us to investigate.
9075 You might as well expedite matters by sending them to begin with.
9077 To enable us to fix the bug, you should include all these things:
9081 The version of @command{ld}. @command{ld} announces it if you start it with
9082 the @samp{--version} argument.
9084 Without this, we will not know whether there is any point in looking for
9085 the bug in the current version of @command{ld}.
9088 Any patches you may have applied to the @command{ld} source, including any
9089 patches made to the @code{BFD} library.
9092 The type of machine you are using, and the operating system name and
9096 What compiler (and its version) was used to compile @command{ld}---e.g.
9100 The command arguments you gave the linker to link your example and
9101 observe the bug. To guarantee you will not omit something important,
9102 list them all. A copy of the Makefile (or the output from make) is
9105 If we were to try to guess the arguments, we would probably guess wrong
9106 and then we might not encounter the bug.
9109 A complete input file, or set of input files, that will reproduce the
9110 bug. It is generally most helpful to send the actual object files
9111 provided that they are reasonably small. Say no more than 10K. For
9112 bigger files you can either make them available by FTP or HTTP or else
9113 state that you are willing to send the object file(s) to whomever
9114 requests them. (Note - your email will be going to a mailing list, so
9115 we do not want to clog it up with large attachments). But small
9116 attachments are best.
9118 If the source files were assembled using @code{gas} or compiled using
9119 @code{gcc}, then it may be OK to send the source files rather than the
9120 object files. In this case, be sure to say exactly what version of
9121 @code{gas} or @code{gcc} was used to produce the object files. Also say
9122 how @code{gas} or @code{gcc} were configured.
9125 A description of what behavior you observe that you believe is
9126 incorrect. For example, ``It gets a fatal signal.''
9128 Of course, if the bug is that @command{ld} gets a fatal signal, then we
9129 will certainly notice it. But if the bug is incorrect output, we might
9130 not notice unless it is glaringly wrong. You might as well not give us
9131 a chance to make a mistake.
9133 Even if the problem you experience is a fatal signal, you should still
9134 say so explicitly. Suppose something strange is going on, such as, your
9135 copy of @command{ld} is out of sync, or you have encountered a bug in the
9136 C library on your system. (This has happened!) Your copy might crash
9137 and ours would not. If you told us to expect a crash, then when ours
9138 fails to crash, we would know that the bug was not happening for us. If
9139 you had not told us to expect a crash, then we would not be able to draw
9140 any conclusion from our observations.
9143 If you wish to suggest changes to the @command{ld} source, send us context
9144 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
9145 @samp{-p} option. Always send diffs from the old file to the new file.
9146 If you even discuss something in the @command{ld} source, refer to it by
9147 context, not by line number.
9149 The line numbers in our development sources will not match those in your
9150 sources. Your line numbers would convey no useful information to us.
9153 Here are some things that are not necessary:
9157 A description of the envelope of the bug.
9159 Often people who encounter a bug spend a lot of time investigating
9160 which changes to the input file will make the bug go away and which
9161 changes will not affect it.
9163 This is often time consuming and not very useful, because the way we
9164 will find the bug is by running a single example under the debugger
9165 with breakpoints, not by pure deduction from a series of examples.
9166 We recommend that you save your time for something else.
9168 Of course, if you can find a simpler example to report @emph{instead}
9169 of the original one, that is a convenience for us. Errors in the
9170 output will be easier to spot, running under the debugger will take
9171 less time, and so on.
9173 However, simplification is not vital; if you do not want to do this,
9174 report the bug anyway and send us the entire test case you used.
9177 A patch for the bug.
9179 A patch for the bug does help us if it is a good one. But do not omit
9180 the necessary information, such as the test case, on the assumption that
9181 a patch is all we need. We might see problems with your patch and decide
9182 to fix the problem another way, or we might not understand it at all.
9184 Sometimes with a program as complicated as @command{ld} it is very hard to
9185 construct an example that will make the program follow a certain path
9186 through the code. If you do not send us the example, we will not be
9187 able to construct one, so we will not be able to verify that the bug is
9190 And if we cannot understand what bug you are trying to fix, or why your
9191 patch should be an improvement, we will not install it. A test case will
9192 help us to understand.
9195 A guess about what the bug is or what it depends on.
9197 Such guesses are usually wrong. Even we cannot guess right about such
9198 things without first using the debugger to find the facts.
9202 @appendix MRI Compatible Script Files
9203 @cindex MRI compatibility
9204 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
9205 linker, @command{ld} can use MRI compatible linker scripts as an
9206 alternative to the more general-purpose linker scripting language
9207 described in @ref{Scripts}. MRI compatible linker scripts have a much
9208 simpler command set than the scripting language otherwise used with
9209 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
9210 linker commands; these commands are described here.
9212 In general, MRI scripts aren't of much use with the @code{a.out} object
9213 file format, since it only has three sections and MRI scripts lack some
9214 features to make use of them.
9216 You can specify a file containing an MRI-compatible script using the
9217 @samp{-c} command-line option.
9219 Each command in an MRI-compatible script occupies its own line; each
9220 command line starts with the keyword that identifies the command (though
9221 blank lines are also allowed for punctuation). If a line of an
9222 MRI-compatible script begins with an unrecognized keyword, @command{ld}
9223 issues a warning message, but continues processing the script.
9225 Lines beginning with @samp{*} are comments.
9227 You can write these commands using all upper-case letters, or all
9228 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
9229 The following list shows only the upper-case form of each command.
9232 @cindex @code{ABSOLUTE} (MRI)
9233 @item ABSOLUTE @var{secname}
9234 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
9235 Normally, @command{ld} includes in the output file all sections from all
9236 the input files. However, in an MRI-compatible script, you can use the
9237 @code{ABSOLUTE} command to restrict the sections that will be present in
9238 your output program. If the @code{ABSOLUTE} command is used at all in a
9239 script, then only the sections named explicitly in @code{ABSOLUTE}
9240 commands will appear in the linker output. You can still use other
9241 input sections (whatever you select on the command line, or using
9242 @code{LOAD}) to resolve addresses in the output file.
9244 @cindex @code{ALIAS} (MRI)
9245 @item ALIAS @var{out-secname}, @var{in-secname}
9246 Use this command to place the data from input section @var{in-secname}
9247 in a section called @var{out-secname} in the linker output file.
9249 @var{in-secname} may be an integer.
9251 @cindex @code{ALIGN} (MRI)
9252 @item ALIGN @var{secname} = @var{expression}
9253 Align the section called @var{secname} to @var{expression}. The
9254 @var{expression} should be a power of two.
9256 @cindex @code{BASE} (MRI)
9257 @item BASE @var{expression}
9258 Use the value of @var{expression} as the lowest address (other than
9259 absolute addresses) in the output file.
9261 @cindex @code{CHIP} (MRI)
9262 @item CHIP @var{expression}
9263 @itemx CHIP @var{expression}, @var{expression}
9264 This command does nothing; it is accepted only for compatibility.
9266 @cindex @code{END} (MRI)
9268 This command does nothing whatever; it's only accepted for compatibility.
9270 @cindex @code{FORMAT} (MRI)
9271 @item FORMAT @var{output-format}
9272 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
9273 language, but restricted to S-records, if @var{output-format} is @samp{S}
9275 @cindex @code{LIST} (MRI)
9276 @item LIST @var{anything}@dots{}
9277 Print (to the standard output file) a link map, as produced by the
9278 @command{ld} command-line option @samp{-M}.
9280 The keyword @code{LIST} may be followed by anything on the
9281 same line, with no change in its effect.
9283 @cindex @code{LOAD} (MRI)
9284 @item LOAD @var{filename}
9285 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
9286 Include one or more object file @var{filename} in the link; this has the
9287 same effect as specifying @var{filename} directly on the @command{ld}
9290 @cindex @code{NAME} (MRI)
9291 @item NAME @var{output-name}
9292 @var{output-name} is the name for the program produced by @command{ld}; the
9293 MRI-compatible command @code{NAME} is equivalent to the command-line
9294 option @samp{-o} or the general script language command @code{OUTPUT}.
9296 @cindex @code{ORDER} (MRI)
9297 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
9298 @itemx ORDER @var{secname} @var{secname} @var{secname}
9299 Normally, @command{ld} orders the sections in its output file in the
9300 order in which they first appear in the input files. In an MRI-compatible
9301 script, you can override this ordering with the @code{ORDER} command. The
9302 sections you list with @code{ORDER} will appear first in your output
9303 file, in the order specified.
9305 @cindex @code{PUBLIC} (MRI)
9306 @item PUBLIC @var{name}=@var{expression}
9307 @itemx PUBLIC @var{name},@var{expression}
9308 @itemx PUBLIC @var{name} @var{expression}
9309 Supply a value (@var{expression}) for external symbol
9310 @var{name} used in the linker input files.
9312 @cindex @code{SECT} (MRI)
9313 @item SECT @var{secname}, @var{expression}
9314 @itemx SECT @var{secname}=@var{expression}
9315 @itemx SECT @var{secname} @var{expression}
9316 You can use any of these three forms of the @code{SECT} command to
9317 specify the start address (@var{expression}) for section @var{secname}.
9318 If you have more than one @code{SECT} statement for the same
9319 @var{secname}, only the @emph{first} sets the start address.
9322 @node GNU Free Documentation License
9323 @appendix GNU Free Documentation License
9327 @unnumbered LD Index
9332 % I think something like @@colophon should be in texinfo. In the
9334 \long\def\colophon{\hbox to0pt{}\vfill
9335 \centerline{The body of this manual is set in}
9336 \centerline{\fontname\tenrm,}
9337 \centerline{with headings in {\bf\fontname\tenbf}}
9338 \centerline{and examples in {\tt\fontname\tentt}.}
9339 \centerline{{\it\fontname\tenit\/} and}
9340 \centerline{{\sl\fontname\tensl\/}}
9341 \centerline{are used for emphasis.}\vfill}
9343 % Blame: doc@@cygnus.com, 28mar91.