3 @c Copyright (C) 1991-2020 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{} 1991-2020 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{} 1991-2020 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
130 * Machine Dependent:: Machine Dependent Features
134 * H8/300:: ld and the H8/300
137 * Renesas:: ld and other Renesas micros
140 * ARM:: ld and the ARM family
143 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
146 * HPPA ELF32:: ld and HPPA 32-bit ELF
149 * M68K:: ld and Motorola 68K family
152 * MIPS:: ld and MIPS family
155 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
158 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
161 * S/390 ELF:: ld and S/390 ELF Support
164 * SPU ELF:: ld and SPU ELF Support
167 * TI COFF:: ld and the TI COFF
170 * Win32:: ld and WIN32 (cygwin/mingw)
173 * Xtensa:: ld and Xtensa Processors
176 @ifclear SingleFormat
179 @c Following blank line required for remaining bug in makeinfo conds/menus
181 * Reporting Bugs:: Reporting Bugs
182 * MRI:: MRI Compatible Script Files
183 * GNU Free Documentation License:: GNU Free Documentation License
184 * LD Index:: LD Index
191 @cindex @sc{gnu} linker
192 @cindex what is this?
195 @c man begin SYNOPSIS
196 ld [@b{options}] @var{objfile} @dots{}
200 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
201 the Info entries for @file{binutils} and
206 @c man begin DESCRIPTION
208 @command{ld} combines a number of object and archive files, relocates
209 their data and ties up symbol references. Usually the last step in
210 compiling a program is to run @command{ld}.
212 @command{ld} accepts Linker Command Language files written in
213 a superset of AT&T's Link Editor Command Language syntax,
214 to provide explicit and total control over the linking process.
218 This man page does not describe the command language; see the
219 @command{ld} entry in @code{info} for full details on the command
220 language and on other aspects of the GNU linker.
223 @ifclear SingleFormat
224 This version of @command{ld} uses the general purpose BFD libraries
225 to operate on object files. This allows @command{ld} to read, combine, and
226 write object files in many different formats---for example, COFF or
227 @code{a.out}. Different formats may be linked together to produce any
228 available kind of object file. @xref{BFD}, for more information.
231 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
232 linkers in providing diagnostic information. Many linkers abandon
233 execution immediately upon encountering an error; whenever possible,
234 @command{ld} continues executing, allowing you to identify other errors
235 (or, in some cases, to get an output file in spite of the error).
242 @c man begin DESCRIPTION
244 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
245 and to be as compatible as possible with other linkers. As a result,
246 you have many choices to control its behavior.
252 * Options:: Command-line Options
253 * Environment:: Environment Variables
257 @section Command-line Options
265 The linker supports a plethora of command-line options, but in actual
266 practice few of them are used in any particular context.
267 @cindex standard Unix system
268 For instance, a frequent use of @command{ld} is to link standard Unix
269 object files on a standard, supported Unix system. On such a system, to
270 link a file @code{hello.o}:
273 ld -o @var{output} /lib/crt0.o hello.o -lc
276 This tells @command{ld} to produce a file called @var{output} as the
277 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
278 the library @code{libc.a}, which will come from the standard search
279 directories. (See the discussion of the @samp{-l} option below.)
281 Some of the command-line options to @command{ld} may be specified at any
282 point in the command line. However, options which refer to files, such
283 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
284 which the option appears in the command line, relative to the object
285 files and other file options. Repeating non-file options with a
286 different argument will either have no further effect, or override prior
287 occurrences (those further to the left on the command line) of that
288 option. Options which may be meaningfully specified more than once are
289 noted in the descriptions below.
292 Non-option arguments are object files or archives which are to be linked
293 together. They may follow, precede, or be mixed in with command-line
294 options, except that an object file argument may not be placed between
295 an option and its argument.
297 Usually the linker is invoked with at least one object file, but you can
298 specify other forms of binary input files using @samp{-l}, @samp{-R},
299 and the script command language. If @emph{no} binary input files at all
300 are specified, the linker does not produce any output, and issues the
301 message @samp{No input files}.
303 If the linker cannot recognize the format of an object file, it will
304 assume that it is a linker script. A script specified in this way
305 augments the main linker script used for the link (either the default
306 linker script or the one specified by using @samp{-T}). This feature
307 permits the linker to link against a file which appears to be an object
308 or an archive, but actually merely defines some symbol values, or uses
309 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
310 script in this way merely augments the main linker script, with the
311 extra commands placed after the main script; use the @samp{-T} option
312 to replace the default linker script entirely, but note the effect of
313 the @code{INSERT} command. @xref{Scripts}.
315 For options whose names are a single letter,
316 option arguments must either follow the option letter without intervening
317 whitespace, or be given as separate arguments immediately following the
318 option that requires them.
320 For options whose names are multiple letters, either one dash or two can
321 precede the option name; for example, @samp{-trace-symbol} and
322 @samp{--trace-symbol} are equivalent. Note---there is one exception to
323 this rule. Multiple letter options that start with a lower case 'o' can
324 only be preceded by two dashes. This is to reduce confusion with the
325 @samp{-o} option. So for example @samp{-omagic} sets the output file
326 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
329 Arguments to multiple-letter options must either be separated from the
330 option name by an equals sign, or be given as separate arguments
331 immediately following the option that requires them. For example,
332 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
333 Unique abbreviations of the names of multiple-letter options are
336 Note---if the linker is being invoked indirectly, via a compiler driver
337 (e.g. @samp{gcc}) then all the linker command-line options should be
338 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
339 compiler driver) like this:
342 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
345 This is important, because otherwise the compiler driver program may
346 silently drop the linker options, resulting in a bad link. Confusion
347 may also arise when passing options that require values through a
348 driver, as the use of a space between option and argument acts as
349 a separator, and causes the driver to pass only the option to the linker
350 and the argument to the compiler. In this case, it is simplest to use
351 the joined forms of both single- and multiple-letter options, such as:
354 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
357 Here is a table of the generic command-line switches accepted by the GNU
361 @include at-file.texi
363 @kindex -a @var{keyword}
364 @item -a @var{keyword}
365 This option is supported for HP/UX compatibility. The @var{keyword}
366 argument must be one of the strings @samp{archive}, @samp{shared}, or
367 @samp{default}. @samp{-aarchive} is functionally equivalent to
368 @samp{-Bstatic}, and the other two keywords are functionally equivalent
369 to @samp{-Bdynamic}. This option may be used any number of times.
371 @kindex --audit @var{AUDITLIB}
372 @item --audit @var{AUDITLIB}
373 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
374 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
375 specified in the library. If specified multiple times @code{DT_AUDIT}
376 will contain a colon separated list of audit interfaces to use. If the linker
377 finds an object with an audit entry while searching for shared libraries,
378 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
379 This option is only meaningful on ELF platforms supporting the rtld-audit
382 @ifclear SingleFormat
383 @cindex binary input format
384 @kindex -b @var{format}
385 @kindex --format=@var{format}
388 @item -b @var{input-format}
389 @itemx --format=@var{input-format}
390 @command{ld} may be configured to support more than one kind of object
391 file. If your @command{ld} is configured this way, you can use the
392 @samp{-b} option to specify the binary format for input object files
393 that follow this option on the command line. Even when @command{ld} is
394 configured to support alternative object formats, you don't usually need
395 to specify this, as @command{ld} should be configured to expect as a
396 default input format the most usual format on each machine.
397 @var{input-format} is a text string, the name of a particular format
398 supported by the BFD libraries. (You can list the available binary
399 formats with @samp{objdump -i}.)
402 You may want to use this option if you are linking files with an unusual
403 binary format. You can also use @samp{-b} to switch formats explicitly (when
404 linking object files of different formats), by including
405 @samp{-b @var{input-format}} before each group of object files in a
408 The default format is taken from the environment variable
413 You can also define the input format from a script, using the command
416 see @ref{Format Commands}.
420 @kindex -c @var{MRI-cmdfile}
421 @kindex --mri-script=@var{MRI-cmdfile}
422 @cindex compatibility, MRI
423 @item -c @var{MRI-commandfile}
424 @itemx --mri-script=@var{MRI-commandfile}
425 For compatibility with linkers produced by MRI, @command{ld} accepts script
426 files written in an alternate, restricted command language, described in
428 @ref{MRI,,MRI Compatible Script Files}.
431 the MRI Compatible Script Files section of GNU ld documentation.
433 Introduce MRI script files with
434 the option @samp{-c}; use the @samp{-T} option to run linker
435 scripts written in the general-purpose @command{ld} scripting language.
436 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
437 specified by any @samp{-L} options.
439 @cindex common allocation
446 These three options are equivalent; multiple forms are supported for
447 compatibility with other linkers. They assign space to common symbols
448 even if a relocatable output file is specified (with @samp{-r}). The
449 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
450 @xref{Miscellaneous Commands}.
452 @kindex --depaudit @var{AUDITLIB}
453 @kindex -P @var{AUDITLIB}
454 @item --depaudit @var{AUDITLIB}
455 @itemx -P @var{AUDITLIB}
456 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
457 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
458 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
459 will contain a colon separated list of audit interfaces to use. This
460 option is only meaningful on ELF platforms supporting the rtld-audit interface.
461 The -P option is provided for Solaris compatibility.
463 @kindex --enable-non-contiguous-regions
464 @item --enable-non-contiguous-regions
465 This option avoids generating an error if an input section does not
466 fit a matching output section. The linker tries to allocate the input
467 section to subseque nt matching output sections, and generates an
468 error only if no output section is large enough. This is useful when
469 several non-contiguous memory regions are available and the input
470 section does not require a particular one. The order in which input
471 sections are evaluated does not change, for instance:
475 MEM1 (rwx) : ORIGIN : 0x1000, LENGTH = 0x14
476 MEM2 (rwx) : ORIGIN : 0x1000, LENGTH = 0x40
477 MEM3 (rwx) : ORIGIN : 0x2000, LENGTH = 0x40
480 mem1 : @{ *(.data.*); @} > MEM1
481 mem2 : @{ *(.data.*); @} > MEM2
482 mem3 : @{ *(.data.*); @} > MEM2
490 results in .data.1 affected to mem1, and .data.2 and .data.3
491 affected to mem2, even though .data.3 would fit in mem3.
494 This option is incompatible with INSERT statements because it changes
495 the way input sections are mapped to output sections.
497 @kindex --enable-non-contiguous-regions-warnings
498 @item --enable-non-contiguous-regions-warnings
499 This option enables warnings when
500 @code{--enable-non-contiguous-regions} allows possibly unexpected
501 matches in sections mapping, potentially leading to silently
502 discarding a section instead of failing because it does not fit any
505 @cindex entry point, from command line
506 @kindex -e @var{entry}
507 @kindex --entry=@var{entry}
509 @itemx --entry=@var{entry}
510 Use @var{entry} as the explicit symbol for beginning execution of your
511 program, rather than the default entry point. If there is no symbol
512 named @var{entry}, the linker will try to parse @var{entry} as a number,
513 and use that as the entry address (the number will be interpreted in
514 base 10; you may use a leading @samp{0x} for base 16, or a leading
515 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
516 and other ways of specifying the entry point.
518 @kindex --exclude-libs
519 @item --exclude-libs @var{lib},@var{lib},...
520 Specifies a list of archive libraries from which symbols should not be automatically
521 exported. The library names may be delimited by commas or colons. Specifying
522 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
523 automatic export. This option is available only for the i386 PE targeted
524 port of the linker and for ELF targeted ports. For i386 PE, symbols
525 explicitly listed in a .def file are still exported, regardless of this
526 option. For ELF targeted ports, symbols affected by this option will
527 be treated as hidden.
529 @kindex --exclude-modules-for-implib
530 @item --exclude-modules-for-implib @var{module},@var{module},...
531 Specifies a list of object files or archive members, from which symbols
532 should not be automatically exported, but which should be copied wholesale
533 into the import library being generated during the link. The module names
534 may be delimited by commas or colons, and must match exactly the filenames
535 used by @command{ld} to open the files; for archive members, this is simply
536 the member name, but for object files the name listed must include and
537 match precisely any path used to specify the input file on the linker's
538 command-line. This option is available only for the i386 PE targeted port
539 of the linker. Symbols explicitly listed in a .def file are still exported,
540 regardless of this option.
542 @cindex dynamic symbol table
544 @kindex --export-dynamic
545 @kindex --no-export-dynamic
547 @itemx --export-dynamic
548 @itemx --no-export-dynamic
549 When creating a dynamically linked executable, using the @option{-E}
550 option or the @option{--export-dynamic} option causes the linker to add
551 all symbols to the dynamic symbol table. The dynamic symbol table is the
552 set of symbols which are visible from dynamic objects at run time.
554 If you do not use either of these options (or use the
555 @option{--no-export-dynamic} option to restore the default behavior), the
556 dynamic symbol table will normally contain only those symbols which are
557 referenced by some dynamic object mentioned in the link.
559 If you use @code{dlopen} to load a dynamic object which needs to refer
560 back to the symbols defined by the program, rather than some other
561 dynamic object, then you will probably need to use this option when
562 linking the program itself.
564 You can also use the dynamic list to control what symbols should
565 be added to the dynamic symbol table if the output format supports it.
566 See the description of @samp{--dynamic-list}.
568 Note that this option is specific to ELF targeted ports. PE targets
569 support a similar function to export all symbols from a DLL or EXE; see
570 the description of @samp{--export-all-symbols} below.
572 @kindex --export-dynamic-symbol=@var{glob}
573 @cindex export dynamic symbol
574 @item --export-dynamic-symbol=@var{glob}
575 When creating a dynamically linked executable, symbols matching
576 @var{glob} will be added to the dynamic symbol table. When creating a
577 shared library, references to symbols matching @var{glob} will not be
578 bound to the definitions within the shared library. This option is a
579 no-op when creating a shared library and @samp{-Bsymbolic} or
580 @samp{--dynamic-list} are not specified. This option is only meaningful
581 on ELF platforms which support shared libraries.
583 @kindex --export-dynamic-symbol-list=@var{file}
584 @cindex export dynamic symbol list
585 @item --export-dynamic-symbol-list=@var{file}
586 Specify a @samp{--export-dynamic-symbol} for each pattern in the file.
587 The format of the file is the same as the version node without
588 scope and node name. See @ref{VERSION} for more information.
590 @ifclear SingleFormat
591 @cindex big-endian objects
595 Link big-endian objects. This affects the default output format.
597 @cindex little-endian objects
600 Link little-endian objects. This affects the default output format.
603 @kindex -f @var{name}
604 @kindex --auxiliary=@var{name}
606 @itemx --auxiliary=@var{name}
607 When creating an ELF shared object, set the internal DT_AUXILIARY field
608 to the specified name. This tells the dynamic linker that the symbol
609 table of the shared object should be used as an auxiliary filter on the
610 symbol table of the shared object @var{name}.
612 If you later link a program against this filter object, then, when you
613 run the program, the dynamic linker will see the DT_AUXILIARY field. If
614 the dynamic linker resolves any symbols from the filter object, it will
615 first check whether there is a definition in the shared object
616 @var{name}. If there is one, it will be used instead of the definition
617 in the filter object. The shared object @var{name} need not exist.
618 Thus the shared object @var{name} may be used to provide an alternative
619 implementation of certain functions, perhaps for debugging or for
620 machine-specific performance.
622 This option may be specified more than once. The DT_AUXILIARY entries
623 will be created in the order in which they appear on the command line.
625 @kindex -F @var{name}
626 @kindex --filter=@var{name}
628 @itemx --filter=@var{name}
629 When creating an ELF shared object, set the internal DT_FILTER field to
630 the specified name. This tells the dynamic linker that the symbol table
631 of the shared object which is being created should be used as a filter
632 on the symbol table of the shared object @var{name}.
634 If you later link a program against this filter object, then, when you
635 run the program, the dynamic linker will see the DT_FILTER field. The
636 dynamic linker will resolve symbols according to the symbol table of the
637 filter object as usual, but it will actually link to the definitions
638 found in the shared object @var{name}. Thus the filter object can be
639 used to select a subset of the symbols provided by the object
642 Some older linkers used the @option{-F} option throughout a compilation
643 toolchain for specifying object-file format for both input and output
645 @ifclear SingleFormat
646 The @sc{gnu} linker uses other mechanisms for this purpose: the
647 @option{-b}, @option{--format}, @option{--oformat} options, the
648 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
649 environment variable.
651 The @sc{gnu} linker will ignore the @option{-F} option when not
652 creating an ELF shared object.
654 @cindex finalization function
655 @kindex -fini=@var{name}
656 @item -fini=@var{name}
657 When creating an ELF executable or shared object, call NAME when the
658 executable or shared object is unloaded, by setting DT_FINI to the
659 address of the function. By default, the linker uses @code{_fini} as
660 the function to call.
664 Ignored. Provided for compatibility with other tools.
666 @kindex -G @var{value}
667 @kindex --gpsize=@var{value}
670 @itemx --gpsize=@var{value}
671 Set the maximum size of objects to be optimized using the GP register to
672 @var{size}. This is only meaningful for object file formats such as
673 MIPS ELF that support putting large and small objects into different
674 sections. This is ignored for other object file formats.
676 @cindex runtime library name
677 @kindex -h @var{name}
678 @kindex -soname=@var{name}
680 @itemx -soname=@var{name}
681 When creating an ELF shared object, set the internal DT_SONAME field to
682 the specified name. When an executable is linked with a shared object
683 which has a DT_SONAME field, then when the executable is run the dynamic
684 linker will attempt to load the shared object specified by the DT_SONAME
685 field rather than the using the file name given to the linker.
688 @cindex incremental link
690 Perform an incremental link (same as option @samp{-r}).
692 @cindex initialization function
693 @kindex -init=@var{name}
694 @item -init=@var{name}
695 When creating an ELF executable or shared object, call NAME when the
696 executable or shared object is loaded, by setting DT_INIT to the address
697 of the function. By default, the linker uses @code{_init} as the
700 @cindex archive files, from cmd line
701 @kindex -l @var{namespec}
702 @kindex --library=@var{namespec}
703 @item -l @var{namespec}
704 @itemx --library=@var{namespec}
705 Add the archive or object file specified by @var{namespec} to the
706 list of files to link. This option may be used any number of times.
707 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
708 will search the library path for a file called @var{filename}, otherwise it
709 will search the library path for a file called @file{lib@var{namespec}.a}.
711 On systems which support shared libraries, @command{ld} may also search for
712 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
713 and SunOS systems, @command{ld} will search a directory for a library
714 called @file{lib@var{namespec}.so} before searching for one called
715 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
716 indicates a shared library.) Note that this behavior does not apply
717 to @file{:@var{filename}}, which always specifies a file called
720 The linker will search an archive only once, at the location where it is
721 specified on the command line. If the archive defines a symbol which
722 was undefined in some object which appeared before the archive on the
723 command line, the linker will include the appropriate file(s) from the
724 archive. However, an undefined symbol in an object appearing later on
725 the command line will not cause the linker to search the archive again.
727 See the @option{-(} option for a way to force the linker to search
728 archives multiple times.
730 You may list the same archive multiple times on the command line.
733 This type of archive searching is standard for Unix linkers. However,
734 if you are using @command{ld} on AIX, note that it is different from the
735 behaviour of the AIX linker.
738 @cindex search directory, from cmd line
740 @kindex --library-path=@var{dir}
741 @item -L @var{searchdir}
742 @itemx --library-path=@var{searchdir}
743 Add path @var{searchdir} to the list of paths that @command{ld} will search
744 for archive libraries and @command{ld} control scripts. You may use this
745 option any number of times. The directories are searched in the order
746 in which they are specified on the command line. Directories specified
747 on the command line are searched before the default directories. All
748 @option{-L} options apply to all @option{-l} options, regardless of the
749 order in which the options appear. @option{-L} options do not affect
750 how @command{ld} searches for a linker script unless @option{-T}
753 If @var{searchdir} begins with @code{=} or @code{$SYSROOT}, then this
754 prefix will be replaced by the @dfn{sysroot prefix}, controlled by the
755 @samp{--sysroot} option, or specified when the linker is configured.
758 The default set of paths searched (without being specified with
759 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
760 some cases also on how it was configured. @xref{Environment}.
763 The paths can also be specified in a link script with the
764 @code{SEARCH_DIR} command. Directories specified this way are searched
765 at the point in which the linker script appears in the command line.
768 @kindex -m @var{emulation}
769 @item -m @var{emulation}
770 Emulate the @var{emulation} linker. You can list the available
771 emulations with the @samp{--verbose} or @samp{-V} options.
773 If the @samp{-m} option is not used, the emulation is taken from the
774 @code{LDEMULATION} environment variable, if that is defined.
776 Otherwise, the default emulation depends upon how the linker was
784 Print a link map to the standard output. A link map provides
785 information about the link, including the following:
789 Where object files are mapped into memory.
791 How common symbols are allocated.
793 All archive members included in the link, with a mention of the symbol
794 which caused the archive member to be brought in.
796 The values assigned to symbols.
798 Note - symbols whose values are computed by an expression which
799 involves a reference to a previous value of the same symbol may not
800 have correct result displayed in the link map. This is because the
801 linker discards intermediate results and only retains the final value
802 of an expression. Under such circumstances the linker will display
803 the final value enclosed by square brackets. Thus for example a
804 linker script containing:
812 will produce the following output in the link map if the @option{-M}
817 [0x0000000c] foo = (foo * 0x4)
818 [0x0000000c] foo = (foo + 0x8)
821 See @ref{Expressions} for more information about expressions in linker
825 How GNU properties are merged.
827 When the linker merges input .note.gnu.property sections into one output
828 .note.gnu.property section, some properties are removed or updated.
829 These actions are reported in the link map. For example:
832 Removed property 0xc0000002 to merge foo.o (0x1) and bar.o (not found)
835 This indicates that property 0xc0000002 is removed from output when
836 merging properties in @file{foo.o}, whose property 0xc0000002 value
837 is 0x1, and @file{bar.o}, which doesn't have property 0xc0000002.
840 Updated property 0xc0010001 (0x1) to merge foo.o (0x1) and bar.o (0x1)
843 This indicates that property 0xc0010001 value is updated to 0x1 in output
844 when merging properties in @file{foo.o}, whose 0xc0010001 property value
845 is 0x1, and @file{bar.o}, whose 0xc0010001 property value is 0x1.
848 @cindex link map discarded
849 @kindex --print-map-discarded
850 @kindex --no-print-map-discarded
851 @item --print-map-discarded
852 @itemx --no-print-map-discarded
853 Print (or do not print) the list of discarded and garbage collected sections
854 in the link map. Enabled by default.
857 @cindex read-only text
862 Turn off page alignment of sections, and disable linking against shared
863 libraries. If the output format supports Unix style magic numbers,
864 mark the output as @code{NMAGIC}.
868 @cindex read/write from cmd line
872 Set the text and data sections to be readable and writable. Also, do
873 not page-align the data segment, and disable linking against shared
874 libraries. If the output format supports Unix style magic numbers,
875 mark the output as @code{OMAGIC}. Note: Although a writable text section
876 is allowed for PE-COFF targets, it does not conform to the format
877 specification published by Microsoft.
882 This option negates most of the effects of the @option{-N} option. It
883 sets the text section to be read-only, and forces the data segment to
884 be page-aligned. Note - this option does not enable linking against
885 shared libraries. Use @option{-Bdynamic} for this.
887 @kindex -o @var{output}
888 @kindex --output=@var{output}
889 @cindex naming the output file
890 @item -o @var{output}
891 @itemx --output=@var{output}
892 Use @var{output} as the name for the program produced by @command{ld}; if this
893 option is not specified, the name @file{a.out} is used by default. The
894 script command @code{OUTPUT} can also specify the output file name.
896 @kindex --dependency-file=@var{depfile}
897 @cindex dependency file
898 @item --dependency-file=@var{depfile}
899 Write a @dfn{dependency file} to @var{depfile}. This file contains a rule
900 suitable for @code{make} describing the output file and all the input files
901 that were read to produce it. The output is similar to the compiler's
902 output with @samp{-M -MP} (@pxref{Preprocessor Options,, Options
903 Controlling the Preprocessor, gcc.info, Using the GNU Compiler
904 Collection}). Note that there is no option like the compiler's @samp{-MM},
905 to exclude ``system files'' (which is not a well-specified concept in the
906 linker, unlike ``system headers'' in the compiler). So the output from
907 @samp{--dependency-file} is always specific to the exact state of the
908 installation where it was produced, and should not be copied into
909 distributed makefiles without careful editing.
911 @kindex -O @var{level}
912 @cindex generating optimized output
914 If @var{level} is a numeric values greater than zero @command{ld} optimizes
915 the output. This might take significantly longer and therefore probably
916 should only be enabled for the final binary. At the moment this
917 option only affects ELF shared library generation. Future releases of
918 the linker may make more use of this option. Also currently there is
919 no difference in the linker's behaviour for different non-zero values
920 of this option. Again this may change with future releases.
922 @kindex -plugin @var{name}
923 @item -plugin @var{name}
924 Involve a plugin in the linking process. The @var{name} parameter is
925 the absolute filename of the plugin. Usually this parameter is
926 automatically added by the complier, when using link time
927 optimization, but users can also add their own plugins if they so
930 Note that the location of the compiler originated plugins is different
931 from the place where the @command{ar}, @command{nm} and
932 @command{ranlib} programs search for their plugins. In order for
933 those commands to make use of a compiler based plugin it must first be
934 copied into the @file{$@{libdir@}/bfd-plugins} directory. All gcc
935 based linker plugins are backward compatible, so it is sufficient to
936 just copy in the newest one.
939 @cindex push state governing input file handling
941 The @option{--push-state} allows to preserve the current state of the
942 flags which govern the input file handling so that they can all be
943 restored with one corresponding @option{--pop-state} option.
945 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
946 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
947 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
948 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
949 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
950 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
952 One target for this option are specifications for @file{pkg-config}. When
953 used with the @option{--libs} option all possibly needed libraries are
954 listed and then possibly linked with all the time. It is better to return
955 something as follows:
958 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
962 @cindex pop state governing input file handling
964 Undoes the effect of --push-state, restores the previous values of the
965 flags governing input file handling.
968 @kindex --emit-relocs
969 @cindex retain relocations in final executable
972 Leave relocation sections and contents in fully linked executables.
973 Post link analysis and optimization tools may need this information in
974 order to perform correct modifications of executables. This results
975 in larger executables.
977 This option is currently only supported on ELF platforms.
979 @kindex --force-dynamic
980 @cindex forcing the creation of dynamic sections
981 @item --force-dynamic
982 Force the output file to have dynamic sections. This option is specific
986 @cindex relocatable output
988 @kindex --relocatable
991 Generate relocatable output---i.e., generate an output file that can in
992 turn serve as input to @command{ld}. This is often called @dfn{partial
993 linking}. As a side effect, in environments that support standard Unix
994 magic numbers, this option also sets the output file's magic number to
996 @c ; see @option{-N}.
997 If this option is not specified, an absolute file is produced. When
998 linking C++ programs, this option @emph{will not} resolve references to
999 constructors; to do that, use @samp{-Ur}.
1001 When an input file does not have the same format as the output file,
1002 partial linking is only supported if that input file does not contain any
1003 relocations. Different output formats can have further restrictions; for
1004 example some @code{a.out}-based formats do not support partial linking
1005 with input files in other formats at all.
1007 This option does the same thing as @samp{-i}.
1009 @kindex -R @var{file}
1010 @kindex --just-symbols=@var{file}
1011 @cindex symbol-only input
1012 @item -R @var{filename}
1013 @itemx --just-symbols=@var{filename}
1014 Read symbol names and their addresses from @var{filename}, but do not
1015 relocate it or include it in the output. This allows your output file
1016 to refer symbolically to absolute locations of memory defined in other
1017 programs. You may use this option more than once.
1019 For compatibility with other ELF linkers, if the @option{-R} option is
1020 followed by a directory name, rather than a file name, it is treated as
1021 the @option{-rpath} option.
1025 @cindex strip all symbols
1028 Omit all symbol information from the output file.
1031 @kindex --strip-debug
1032 @cindex strip debugger symbols
1034 @itemx --strip-debug
1035 Omit debugger symbol information (but not all symbols) from the output file.
1037 @kindex --strip-discarded
1038 @kindex --no-strip-discarded
1039 @item --strip-discarded
1040 @itemx --no-strip-discarded
1041 Omit (or do not omit) global symbols defined in discarded sections.
1046 @cindex input files, displaying
1049 Print the names of the input files as @command{ld} processes them. If
1050 @samp{-t} is given twice then members within archives are also printed.
1051 @samp{-t} output is useful to generate a list of all the object files
1052 and scripts involved in linking, for example, when packaging files for
1053 a linker bug report.
1055 @kindex -T @var{script}
1056 @kindex --script=@var{script}
1057 @cindex script files
1058 @item -T @var{scriptfile}
1059 @itemx --script=@var{scriptfile}
1060 Use @var{scriptfile} as the linker script. This script replaces
1061 @command{ld}'s default linker script (rather than adding to it), so
1062 @var{commandfile} must specify everything necessary to describe the
1063 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
1064 the current directory, @code{ld} looks for it in the directories
1065 specified by any preceding @samp{-L} options. Multiple @samp{-T}
1068 @kindex -dT @var{script}
1069 @kindex --default-script=@var{script}
1070 @cindex script files
1071 @item -dT @var{scriptfile}
1072 @itemx --default-script=@var{scriptfile}
1073 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
1075 This option is similar to the @option{--script} option except that
1076 processing of the script is delayed until after the rest of the
1077 command line has been processed. This allows options placed after the
1078 @option{--default-script} option on the command line to affect the
1079 behaviour of the linker script, which can be important when the linker
1080 command line cannot be directly controlled by the user. (eg because
1081 the command line is being constructed by another tool, such as
1084 @kindex -u @var{symbol}
1085 @kindex --undefined=@var{symbol}
1086 @cindex undefined symbol
1087 @item -u @var{symbol}
1088 @itemx --undefined=@var{symbol}
1089 Force @var{symbol} to be entered in the output file as an undefined
1090 symbol. Doing this may, for example, trigger linking of additional
1091 modules from standard libraries. @samp{-u} may be repeated with
1092 different option arguments to enter additional undefined symbols. This
1093 option is equivalent to the @code{EXTERN} linker script command.
1095 If this option is being used to force additional modules to be pulled
1096 into the link, and if it is an error for the symbol to remain
1097 undefined, then the option @option{--require-defined} should be used
1100 @kindex --require-defined=@var{symbol}
1101 @cindex symbols, require defined
1102 @cindex defined symbol
1103 @item --require-defined=@var{symbol}
1104 Require that @var{symbol} is defined in the output file. This option
1105 is the same as option @option{--undefined} except that if @var{symbol}
1106 is not defined in the output file then the linker will issue an error
1107 and exit. The same effect can be achieved in a linker script by using
1108 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
1109 can be used multiple times to require additional symbols.
1112 @cindex constructors
1114 For anything other than C++ programs, this option is equivalent to
1115 @samp{-r}: it generates relocatable output---i.e., an output file that can in
1116 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
1117 @emph{does} resolve references to constructors, unlike @samp{-r}.
1118 It does not work to use @samp{-Ur} on files that were themselves linked
1119 with @samp{-Ur}; once the constructor table has been built, it cannot
1120 be added to. Use @samp{-Ur} only for the last partial link, and
1121 @samp{-r} for the others.
1123 @kindex --orphan-handling=@var{MODE}
1124 @cindex orphan sections
1125 @cindex sections, orphan
1126 @item --orphan-handling=@var{MODE}
1127 Control how orphan sections are handled. An orphan section is one not
1128 specifically mentioned in a linker script. @xref{Orphan Sections}.
1130 @var{MODE} can have any of the following values:
1134 Orphan sections are placed into a suitable output section following
1135 the strategy described in @ref{Orphan Sections}. The option
1136 @samp{--unique} also affects how sections are placed.
1139 All orphan sections are discarded, by placing them in the
1140 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1143 The linker will place the orphan section as for @code{place} and also
1147 The linker will exit with an error if any orphan section is found.
1150 The default if @samp{--orphan-handling} is not given is @code{place}.
1152 @kindex --unique[=@var{SECTION}]
1153 @item --unique[=@var{SECTION}]
1154 Creates a separate output section for every input section matching
1155 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1156 missing, for every orphan input section. An orphan section is one not
1157 specifically mentioned in a linker script. You may use this option
1158 multiple times on the command line; It prevents the normal merging of
1159 input sections with the same name, overriding output section assignments
1169 Display the version number for @command{ld}. The @option{-V} option also
1170 lists the supported emulations.
1173 @kindex --discard-all
1174 @cindex deleting local symbols
1176 @itemx --discard-all
1177 Delete all local symbols.
1180 @kindex --discard-locals
1181 @cindex local symbols, deleting
1183 @itemx --discard-locals
1184 Delete all temporary local symbols. (These symbols start with
1185 system-specific local label prefixes, typically @samp{.L} for ELF systems
1186 or @samp{L} for traditional a.out systems.)
1188 @kindex -y @var{symbol}
1189 @kindex --trace-symbol=@var{symbol}
1190 @cindex symbol tracing
1191 @item -y @var{symbol}
1192 @itemx --trace-symbol=@var{symbol}
1193 Print the name of each linked file in which @var{symbol} appears. This
1194 option may be given any number of times. On many systems it is necessary
1195 to prepend an underscore.
1197 This option is useful when you have an undefined symbol in your link but
1198 don't know where the reference is coming from.
1200 @kindex -Y @var{path}
1202 Add @var{path} to the default library search path. This option exists
1203 for Solaris compatibility.
1205 @kindex -z @var{keyword}
1206 @item -z @var{keyword}
1207 The recognized keywords are:
1211 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1213 @item call-nop=prefix-addr
1214 @itemx call-nop=suffix-nop
1215 @itemx call-nop=prefix-@var{byte}
1216 @itemx call-nop=suffix-@var{byte}
1217 Specify the 1-byte @code{NOP} padding when transforming indirect call
1218 to a locally defined function, foo, via its GOT slot.
1219 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1220 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1221 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1222 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1223 Supported for i386 and x86_64.
1225 @item cet-report=none
1226 @itemx cet-report=warning
1227 @itemx cet-report=error
1228 Specify how to report the missing GNU_PROPERTY_X86_FEATURE_1_IBT and
1229 GNU_PROPERTY_X86_FEATURE_1_SHSTK properties in input .note.gnu.property
1230 section. @option{cet-report=none}, which is the default, will make the
1231 linker not report missing properties in input files.
1232 @option{cet-report=warning} will make the linker issue a warning for
1233 missing properties in input files. @option{cet-report=error} will make
1234 the linker issue an error for missing properties in input files.
1235 Note that @option{ibt} will turn off the missing
1236 GNU_PROPERTY_X86_FEATURE_1_IBT property report and @option{shstk} will
1237 turn off the missing GNU_PROPERTY_X86_FEATURE_1_SHSTK property report.
1238 Supported for Linux/i386 and Linux/x86_64.
1242 Combine multiple dynamic relocation sections and sort to improve
1243 dynamic symbol lookup caching. Do not do this if @samp{nocombreloc}.
1247 Generate common symbols with STT_COMMON type during a relocatable
1248 link. Use STT_OBJECT type if @samp{nocommon}.
1250 @item common-page-size=@var{value}
1251 Set the page size most commonly used to @var{value}. Memory image
1252 layout will be optimized to minimize memory pages if the system is
1253 using pages of this size.
1256 Report unresolved symbol references from regular object files. This
1257 is done even if the linker is creating a non-symbolic shared library.
1258 This option is the inverse of @samp{-z undefs}.
1260 @item dynamic-undefined-weak
1261 @itemx nodynamic-undefined-weak
1262 Make undefined weak symbols dynamic when building a dynamic object,
1263 if they are referenced from a regular object file and not forced local
1264 by symbol visibility or versioning. Do not make them dynamic if
1265 @samp{nodynamic-undefined-weak}. If neither option is given, a target
1266 may default to either option being in force, or make some other
1267 selection of undefined weak symbols dynamic. Not all targets support
1271 Marks the object as requiring executable stack.
1274 This option is only meaningful when building a shared object. It makes
1275 the symbols defined by this shared object available for symbol resolution
1276 of subsequently loaded libraries.
1279 This option is only meaningful when building a dynamic executable.
1280 This option marks the executable as requiring global auditing by
1281 setting the @code{DF_1_GLOBAUDIT} bit in the @code{DT_FLAGS_1} dynamic
1282 tag. Global auditing requires that any auditing library defined via
1283 the @option{--depaudit} or @option{-P} command-line options be run for
1284 all dynamic objects loaded by the application.
1287 Generate Intel Indirect Branch Tracking (IBT) enabled PLT entries.
1288 Supported for Linux/i386 and Linux/x86_64.
1291 Generate GNU_PROPERTY_X86_FEATURE_1_IBT in .note.gnu.property section
1292 to indicate compatibility with IBT. This also implies @option{ibtplt}.
1293 Supported for Linux/i386 and Linux/x86_64.
1296 This option is only meaningful when building a shared object.
1297 It marks the object so that its runtime initialization will occur
1298 before the runtime initialization of any other objects brought into
1299 the process at the same time. Similarly the runtime finalization of
1300 the object will occur after the runtime finalization of any other
1304 Specify that the dynamic loader should modify its symbol search order
1305 so that symbols in this shared library interpose all other shared
1306 libraries not so marked.
1309 When generating an executable or shared library, mark it to tell the
1310 dynamic linker to defer function call resolution to the point when
1311 the function is called (lazy binding), rather than at load time.
1312 Lazy binding is the default.
1315 Specify that the object's filters be processed immediately at runtime.
1317 @item max-page-size=@var{value}
1318 Set the maximum memory page size supported to @var{value}.
1321 Allow multiple definitions.
1324 Disable linker generated .dynbss variables used in place of variables
1325 defined in shared libraries. May result in dynamic text relocations.
1328 Specify that the dynamic loader search for dependencies of this object
1329 should ignore any default library search paths.
1332 Specify that the object shouldn't be unloaded at runtime.
1335 Specify that the object is not available to @code{dlopen}.
1338 Specify that the object can not be dumped by @code{dldump}.
1341 Marks the object as not requiring executable stack.
1343 @item noextern-protected-data
1344 Don't treat protected data symbols as external when building a shared
1345 library. This option overrides the linker backend default. It can be
1346 used to work around incorrect relocations against protected data symbols
1347 generated by compiler. Updates on protected data symbols by another
1348 module aren't visible to the resulting shared library. Supported for
1351 @item noreloc-overflow
1352 Disable relocation overflow check. This can be used to disable
1353 relocation overflow check if there will be no dynamic relocation
1354 overflow at run-time. Supported for x86_64.
1357 When generating an executable or shared library, mark it to tell the
1358 dynamic linker to resolve all symbols when the program is started, or
1359 when the shared library is loaded by dlopen, instead of deferring
1360 function call resolution to the point when the function is first
1364 Specify that the object requires @samp{$ORIGIN} handling in paths.
1368 Create an ELF @code{PT_GNU_RELRO} segment header in the object. This
1369 specifies a memory segment that should be made read-only after
1370 relocation, if supported. Specifying @samp{common-page-size} smaller
1371 than the system page size will render this protection ineffective.
1372 Don't create an ELF @code{PT_GNU_RELRO} segment if @samp{norelro}.
1375 @itemx noseparate-code
1376 Create separate code @code{PT_LOAD} segment header in the object. This
1377 specifies a memory segment that should contain only instructions and must
1378 be in wholly disjoint pages from any other data. Don't create separate
1379 code @code{PT_LOAD} segment if @samp{noseparate-code} is used.
1382 Generate GNU_PROPERTY_X86_FEATURE_1_SHSTK in .note.gnu.property section
1383 to indicate compatibility with Intel Shadow Stack. Supported for
1384 Linux/i386 and Linux/x86_64.
1386 @item stack-size=@var{value}
1387 Specify a stack size for an ELF @code{PT_GNU_STACK} segment.
1388 Specifying zero will override any default non-zero sized
1389 @code{PT_GNU_STACK} segment creation.
1391 @item start-stop-visibility=@var{value}
1393 @cindex ELF symbol visibility
1394 Specify the ELF symbol visibility for synthesized
1395 @code{__start_SECNAME} and @code{__stop_SECNAME} symbols (@pxref{Input
1396 Section Example}). @var{value} must be exactly @samp{default},
1397 @samp{internal}, @samp{hidden}, or @samp{protected}. If no @samp{-z
1398 start-stop-visibility} option is given, @samp{protected} is used for
1399 compatibility with historical practice. However, it's highly
1400 recommended to use @samp{-z start-stop-visibility=hidden} in new
1401 programs and shared libraries so that these symbols are not exported
1402 between shared objects, which is not usually what's intended.
1407 Report an error if DT_TEXTREL is set, i.e., if the position-independent
1408 or shared object has dynamic relocations in read-only sections. Don't
1409 report an error if @samp{notext} or @samp{textoff}.
1412 Do not report unresolved symbol references from regular object files,
1413 either when creating an executable, or when creating a shared library.
1414 This option is the inverse of @samp{-z defs}.
1418 Other keywords are ignored for Solaris compatibility.
1421 @cindex groups of archives
1422 @item -( @var{archives} -)
1423 @itemx --start-group @var{archives} --end-group
1424 The @var{archives} should be a list of archive files. They may be
1425 either explicit file names, or @samp{-l} options.
1427 The specified archives are searched repeatedly until no new undefined
1428 references are created. Normally, an archive is searched only once in
1429 the order that it is specified on the command line. If a symbol in that
1430 archive is needed to resolve an undefined symbol referred to by an
1431 object in an archive that appears later on the command line, the linker
1432 would not be able to resolve that reference. By grouping the archives,
1433 they will all be searched repeatedly until all possible references are
1436 Using this option has a significant performance cost. It is best to use
1437 it only when there are unavoidable circular references between two or
1440 @kindex --accept-unknown-input-arch
1441 @kindex --no-accept-unknown-input-arch
1442 @item --accept-unknown-input-arch
1443 @itemx --no-accept-unknown-input-arch
1444 Tells the linker to accept input files whose architecture cannot be
1445 recognised. The assumption is that the user knows what they are doing
1446 and deliberately wants to link in these unknown input files. This was
1447 the default behaviour of the linker, before release 2.14. The default
1448 behaviour from release 2.14 onwards is to reject such input files, and
1449 so the @samp{--accept-unknown-input-arch} option has been added to
1450 restore the old behaviour.
1453 @kindex --no-as-needed
1455 @itemx --no-as-needed
1456 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1457 on the command line after the @option{--as-needed} option. Normally
1458 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1459 on the command line, regardless of whether the library is actually
1460 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1461 emitted for a library that @emph{at that point in the link} satisfies a
1462 non-weak undefined symbol reference from a regular object file or, if
1463 the library is not found in the DT_NEEDED lists of other needed libraries, a
1464 non-weak undefined symbol reference from another needed dynamic library.
1465 Object files or libraries appearing on the command line @emph{after}
1466 the library in question do not affect whether the library is seen as
1467 needed. This is similar to the rules for extraction of object files
1468 from archives. @option{--no-as-needed} restores the default behaviour.
1470 @kindex --add-needed
1471 @kindex --no-add-needed
1473 @itemx --no-add-needed
1474 These two options have been deprecated because of the similarity of
1475 their names to the @option{--as-needed} and @option{--no-as-needed}
1476 options. They have been replaced by @option{--copy-dt-needed-entries}
1477 and @option{--no-copy-dt-needed-entries}.
1479 @kindex -assert @var{keyword}
1480 @item -assert @var{keyword}
1481 This option is ignored for SunOS compatibility.
1485 @kindex -call_shared
1489 Link against dynamic libraries. This is only meaningful on platforms
1490 for which shared libraries are supported. This option is normally the
1491 default on such platforms. The different variants of this option are
1492 for compatibility with various systems. You may use this option
1493 multiple times on the command line: it affects library searching for
1494 @option{-l} options which follow it.
1498 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1499 section. This causes the runtime linker to handle lookups in this
1500 object and its dependencies to be performed only inside the group.
1501 @option{--unresolved-symbols=report-all} is implied. This option is
1502 only meaningful on ELF platforms which support shared libraries.
1512 Do not link against shared libraries. This is only meaningful on
1513 platforms for which shared libraries are supported. The different
1514 variants of this option are for compatibility with various systems. You
1515 may use this option multiple times on the command line: it affects
1516 library searching for @option{-l} options which follow it. This
1517 option also implies @option{--unresolved-symbols=report-all}. This
1518 option can be used with @option{-shared}. Doing so means that a
1519 shared library is being created but that all of the library's external
1520 references must be resolved by pulling in entries from static
1525 When creating a shared library, bind references to global symbols to the
1526 definition within the shared library, if any. Normally, it is possible
1527 for a program linked against a shared library to override the definition
1528 within the shared library. This option is only meaningful on ELF
1529 platforms which support shared libraries.
1531 @kindex -Bsymbolic-functions
1532 @item -Bsymbolic-functions
1533 When creating a shared library, bind references to global function
1534 symbols to the definition within the shared library, if any.
1535 This option is only meaningful on ELF platforms which support shared
1538 @kindex --dynamic-list=@var{dynamic-list-file}
1539 @item --dynamic-list=@var{dynamic-list-file}
1540 Specify the name of a dynamic list file to the linker. This is
1541 typically used when creating shared libraries to specify a list of
1542 global symbols whose references shouldn't be bound to the definition
1543 within the shared library, or creating dynamically linked executables
1544 to specify a list of symbols which should be added to the symbol table
1545 in the executable. This option is only meaningful on ELF platforms
1546 which support shared libraries.
1548 The format of the dynamic list is the same as the version node without
1549 scope and node name. See @ref{VERSION} for more information.
1551 @kindex --dynamic-list-data
1552 @item --dynamic-list-data
1553 Include all global data symbols to the dynamic list.
1555 @kindex --dynamic-list-cpp-new
1556 @item --dynamic-list-cpp-new
1557 Provide the builtin dynamic list for C++ operator new and delete. It
1558 is mainly useful for building shared libstdc++.
1560 @kindex --dynamic-list-cpp-typeinfo
1561 @item --dynamic-list-cpp-typeinfo
1562 Provide the builtin dynamic list for C++ runtime type identification.
1564 @kindex --check-sections
1565 @kindex --no-check-sections
1566 @item --check-sections
1567 @itemx --no-check-sections
1568 Asks the linker @emph{not} to check section addresses after they have
1569 been assigned to see if there are any overlaps. Normally the linker will
1570 perform this check, and if it finds any overlaps it will produce
1571 suitable error messages. The linker does know about, and does make
1572 allowances for sections in overlays. The default behaviour can be
1573 restored by using the command-line switch @option{--check-sections}.
1574 Section overlap is not usually checked for relocatable links. You can
1575 force checking in that case by using the @option{--check-sections}
1578 @kindex --copy-dt-needed-entries
1579 @kindex --no-copy-dt-needed-entries
1580 @item --copy-dt-needed-entries
1581 @itemx --no-copy-dt-needed-entries
1582 This option affects the treatment of dynamic libraries referred to
1583 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1584 command line. Normally the linker won't add a DT_NEEDED tag to the
1585 output binary for each library mentioned in a DT_NEEDED tag in an
1586 input dynamic library. With @option{--copy-dt-needed-entries}
1587 specified on the command line however any dynamic libraries that
1588 follow it will have their DT_NEEDED entries added. The default
1589 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1591 This option also has an effect on the resolution of symbols in dynamic
1592 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1593 mentioned on the command line will be recursively searched, following
1594 their DT_NEEDED tags to other libraries, in order to resolve symbols
1595 required by the output binary. With the default setting however
1596 the searching of dynamic libraries that follow it will stop with the
1597 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1600 @cindex cross reference table
1603 Output a cross reference table. If a linker map file is being
1604 generated, the cross reference table is printed to the map file.
1605 Otherwise, it is printed on the standard output.
1607 The format of the table is intentionally simple, so that it may be
1608 easily processed by a script if necessary. The symbols are printed out,
1609 sorted by name. For each symbol, a list of file names is given. If the
1610 symbol is defined, the first file listed is the location of the
1611 definition. If the symbol is defined as a common value then any files
1612 where this happens appear next. Finally any files that reference the
1615 @cindex common allocation
1616 @kindex --no-define-common
1617 @item --no-define-common
1618 This option inhibits the assignment of addresses to common symbols.
1619 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1620 @xref{Miscellaneous Commands}.
1622 The @samp{--no-define-common} option allows decoupling
1623 the decision to assign addresses to Common symbols from the choice
1624 of the output file type; otherwise a non-Relocatable output type
1625 forces assigning addresses to Common symbols.
1626 Using @samp{--no-define-common} allows Common symbols that are referenced
1627 from a shared library to be assigned addresses only in the main program.
1628 This eliminates the unused duplicate space in the shared library,
1629 and also prevents any possible confusion over resolving to the wrong
1630 duplicate when there are many dynamic modules with specialized search
1631 paths for runtime symbol resolution.
1633 @cindex group allocation in linker script
1634 @cindex section groups
1636 @kindex --force-group-allocation
1637 @item --force-group-allocation
1638 This option causes the linker to place section group members like
1639 normal input sections, and to delete the section groups. This is the
1640 default behaviour for a final link but this option can be used to
1641 change the behaviour of a relocatable link (@samp{-r}). The script
1642 command @code{FORCE_GROUP_ALLOCATION} has the same
1643 effect. @xref{Miscellaneous Commands}.
1645 @cindex symbols, from command line
1646 @kindex --defsym=@var{symbol}=@var{exp}
1647 @item --defsym=@var{symbol}=@var{expression}
1648 Create a global symbol in the output file, containing the absolute
1649 address given by @var{expression}. You may use this option as many
1650 times as necessary to define multiple symbols in the command line. A
1651 limited form of arithmetic is supported for the @var{expression} in this
1652 context: you may give a hexadecimal constant or the name of an existing
1653 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1654 constants or symbols. If you need more elaborate expressions, consider
1655 using the linker command language from a script (@pxref{Assignments}).
1656 @emph{Note:} there should be no white space between @var{symbol}, the
1657 equals sign (``@key{=}''), and @var{expression}.
1659 @cindex demangling, from command line
1660 @kindex --demangle[=@var{style}]
1661 @kindex --no-demangle
1662 @item --demangle[=@var{style}]
1663 @itemx --no-demangle
1664 These options control whether to demangle symbol names in error messages
1665 and other output. When the linker is told to demangle, it tries to
1666 present symbol names in a readable fashion: it strips leading
1667 underscores if they are used by the object file format, and converts C++
1668 mangled symbol names into user readable names. Different compilers have
1669 different mangling styles. The optional demangling style argument can be used
1670 to choose an appropriate demangling style for your compiler. The linker will
1671 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1672 is set. These options may be used to override the default.
1674 @cindex dynamic linker, from command line
1675 @kindex -I@var{file}
1676 @kindex --dynamic-linker=@var{file}
1678 @itemx --dynamic-linker=@var{file}
1679 Set the name of the dynamic linker. This is only meaningful when
1680 generating dynamically linked ELF executables. The default dynamic
1681 linker is normally correct; don't use this unless you know what you are
1684 @kindex --no-dynamic-linker
1685 @item --no-dynamic-linker
1686 When producing an executable file, omit the request for a dynamic
1687 linker to be used at load-time. This is only meaningful for ELF
1688 executables that contain dynamic relocations, and usually requires
1689 entry point code that is capable of processing these relocations.
1691 @kindex --embedded-relocs
1692 @item --embedded-relocs
1693 This option is similar to the @option{--emit-relocs} option except
1694 that the relocs are stored in a target-specific section. This option
1695 is only supported by the @samp{BFIN}, @samp{CR16} and @emph{M68K}
1698 @kindex --disable-multiple-abs-defs
1699 @item --disable-multiple-abs-defs
1700 Do not allow multiple definitions with symbols included
1701 in filename invoked by -R or --just-symbols
1703 @kindex --fatal-warnings
1704 @kindex --no-fatal-warnings
1705 @item --fatal-warnings
1706 @itemx --no-fatal-warnings
1707 Treat all warnings as errors. The default behaviour can be restored
1708 with the option @option{--no-fatal-warnings}.
1710 @kindex --force-exe-suffix
1711 @item --force-exe-suffix
1712 Make sure that an output file has a .exe suffix.
1714 If a successfully built fully linked output file does not have a
1715 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1716 the output file to one of the same name with a @code{.exe} suffix. This
1717 option is useful when using unmodified Unix makefiles on a Microsoft
1718 Windows host, since some versions of Windows won't run an image unless
1719 it ends in a @code{.exe} suffix.
1721 @kindex --gc-sections
1722 @kindex --no-gc-sections
1723 @cindex garbage collection
1725 @itemx --no-gc-sections
1726 Enable garbage collection of unused input sections. It is ignored on
1727 targets that do not support this option. The default behaviour (of not
1728 performing this garbage collection) can be restored by specifying
1729 @samp{--no-gc-sections} on the command line. Note that garbage
1730 collection for COFF and PE format targets is supported, but the
1731 implementation is currently considered to be experimental.
1733 @samp{--gc-sections} decides which input sections are used by
1734 examining symbols and relocations. The section containing the entry
1735 symbol and all sections containing symbols undefined on the
1736 command-line will be kept, as will sections containing symbols
1737 referenced by dynamic objects. Note that when building shared
1738 libraries, the linker must assume that any visible symbol is
1739 referenced. Once this initial set of sections has been determined,
1740 the linker recursively marks as used any section referenced by their
1741 relocations. See @samp{--entry}, @samp{--undefined}, and
1742 @samp{--gc-keep-exported}.
1744 This option can be set when doing a partial link (enabled with option
1745 @samp{-r}). In this case the root of symbols kept must be explicitly
1746 specified either by one of the options @samp{--entry},
1747 @samp{--undefined}, or @samp{--gc-keep-exported} or by a @code{ENTRY}
1748 command in the linker script.
1750 @kindex --print-gc-sections
1751 @kindex --no-print-gc-sections
1752 @cindex garbage collection
1753 @item --print-gc-sections
1754 @itemx --no-print-gc-sections
1755 List all sections removed by garbage collection. The listing is
1756 printed on stderr. This option is only effective if garbage
1757 collection has been enabled via the @samp{--gc-sections}) option. The
1758 default behaviour (of not listing the sections that are removed) can
1759 be restored by specifying @samp{--no-print-gc-sections} on the command
1762 @kindex --gc-keep-exported
1763 @cindex garbage collection
1764 @item --gc-keep-exported
1765 When @samp{--gc-sections} is enabled, this option prevents garbage
1766 collection of unused input sections that contain global symbols having
1767 default or protected visibility. This option is intended to be used for
1768 executables where unreferenced sections would otherwise be garbage
1769 collected regardless of the external visibility of contained symbols.
1770 Note that this option has no effect when linking shared objects since
1771 it is already the default behaviour. This option is only supported for
1774 @kindex --print-output-format
1775 @cindex output format
1776 @item --print-output-format
1777 Print the name of the default output format (perhaps influenced by
1778 other command-line options). This is the string that would appear
1779 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1781 @kindex --print-memory-usage
1782 @cindex memory usage
1783 @item --print-memory-usage
1784 Print used size, total size and used size of memory regions created with
1785 the @ref{MEMORY} command. This is useful on embedded targets to have a
1786 quick view of amount of free memory. The format of the output has one
1787 headline and one line per region. It is both human readable and easily
1788 parsable by tools. Here is an example of an output:
1791 Memory region Used Size Region Size %age Used
1792 ROM: 256 KB 1 MB 25.00%
1793 RAM: 32 B 2 GB 0.00%
1800 Print a summary of the command-line options on the standard output and exit.
1802 @kindex --target-help
1804 Print a summary of all target-specific options on the standard output and exit.
1806 @kindex -Map=@var{mapfile}
1807 @item -Map=@var{mapfile}
1808 Print a link map to the file @var{mapfile}. See the description of the
1809 @option{-M} option, above. Specifying a directory as @var{mapfile}
1810 causes the linker map to be written into a file inside the directory.
1811 The name of the file is based upon the @var{output} filename with
1812 @code{.map} appended.
1814 @cindex memory usage
1815 @kindex --no-keep-memory
1816 @item --no-keep-memory
1817 @command{ld} normally optimizes for speed over memory usage by caching the
1818 symbol tables of input files in memory. This option tells @command{ld} to
1819 instead optimize for memory usage, by rereading the symbol tables as
1820 necessary. This may be required if @command{ld} runs out of memory space
1821 while linking a large executable.
1823 @kindex --no-undefined
1826 @item --no-undefined
1828 Report unresolved symbol references from regular object files. This
1829 is done even if the linker is creating a non-symbolic shared library.
1830 The switch @option{--[no-]allow-shlib-undefined} controls the
1831 behaviour for reporting unresolved references found in shared
1832 libraries being linked in.
1834 The effects of this option can be reverted by using @code{-z undefs}.
1836 @kindex --allow-multiple-definition
1838 @item --allow-multiple-definition
1840 Normally when a symbol is defined multiple times, the linker will
1841 report a fatal error. These options allow multiple definitions and the
1842 first definition will be used.
1844 @kindex --allow-shlib-undefined
1845 @kindex --no-allow-shlib-undefined
1846 @item --allow-shlib-undefined
1847 @itemx --no-allow-shlib-undefined
1848 Allows or disallows undefined symbols in shared libraries.
1849 This switch is similar to @option{--no-undefined} except that it
1850 determines the behaviour when the undefined symbols are in a
1851 shared library rather than a regular object file. It does not affect
1852 how undefined symbols in regular object files are handled.
1854 The default behaviour is to report errors for any undefined symbols
1855 referenced in shared libraries if the linker is being used to create
1856 an executable, but to allow them if the linker is being used to create
1859 The reasons for allowing undefined symbol references in shared
1860 libraries specified at link time are that:
1864 A shared library specified at link time may not be the same as the one
1865 that is available at load time, so the symbol might actually be
1866 resolvable at load time.
1868 There are some operating systems, eg BeOS and HPPA, where undefined
1869 symbols in shared libraries are normal.
1871 The BeOS kernel for example patches shared libraries at load time to
1872 select whichever function is most appropriate for the current
1873 architecture. This is used, for example, to dynamically select an
1874 appropriate memset function.
1877 @kindex --no-undefined-version
1878 @item --no-undefined-version
1879 Normally when a symbol has an undefined version, the linker will ignore
1880 it. This option disallows symbols with undefined version and a fatal error
1881 will be issued instead.
1883 @kindex --default-symver
1884 @item --default-symver
1885 Create and use a default symbol version (the soname) for unversioned
1888 @kindex --default-imported-symver
1889 @item --default-imported-symver
1890 Create and use a default symbol version (the soname) for unversioned
1893 @kindex --no-warn-mismatch
1894 @item --no-warn-mismatch
1895 Normally @command{ld} will give an error if you try to link together input
1896 files that are mismatched for some reason, perhaps because they have
1897 been compiled for different processors or for different endiannesses.
1898 This option tells @command{ld} that it should silently permit such possible
1899 errors. This option should only be used with care, in cases when you
1900 have taken some special action that ensures that the linker errors are
1903 @kindex --no-warn-search-mismatch
1904 @item --no-warn-search-mismatch
1905 Normally @command{ld} will give a warning if it finds an incompatible
1906 library during a library search. This option silences the warning.
1908 @kindex --no-whole-archive
1909 @item --no-whole-archive
1910 Turn off the effect of the @option{--whole-archive} option for subsequent
1913 @cindex output file after errors
1914 @kindex --noinhibit-exec
1915 @item --noinhibit-exec
1916 Retain the executable output file whenever it is still usable.
1917 Normally, the linker will not produce an output file if it encounters
1918 errors during the link process; it exits without writing an output file
1919 when it issues any error whatsoever.
1923 Only search library directories explicitly specified on the
1924 command line. Library directories specified in linker scripts
1925 (including linker scripts specified on the command line) are ignored.
1927 @ifclear SingleFormat
1928 @kindex --oformat=@var{output-format}
1929 @item --oformat=@var{output-format}
1930 @command{ld} may be configured to support more than one kind of object
1931 file. If your @command{ld} is configured this way, you can use the
1932 @samp{--oformat} option to specify the binary format for the output
1933 object file. Even when @command{ld} is configured to support alternative
1934 object formats, you don't usually need to specify this, as @command{ld}
1935 should be configured to produce as a default output format the most
1936 usual format on each machine. @var{output-format} is a text string, the
1937 name of a particular format supported by the BFD libraries. (You can
1938 list the available binary formats with @samp{objdump -i}.) The script
1939 command @code{OUTPUT_FORMAT} can also specify the output format, but
1940 this option overrides it. @xref{BFD}.
1943 @kindex --out-implib
1944 @item --out-implib @var{file}
1945 Create an import library in @var{file} corresponding to the executable
1946 the linker is generating (eg. a DLL or ELF program). This import
1947 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
1948 may be used to link clients against the generated executable; this
1949 behaviour makes it possible to skip a separate import library creation
1950 step (eg. @code{dlltool} for DLLs). This option is only available for
1951 the i386 PE and ELF targetted ports of the linker.
1954 @kindex --pic-executable
1956 @itemx --pic-executable
1957 @cindex position independent executables
1958 Create a position independent executable. This is currently only supported on
1959 ELF platforms. Position independent executables are similar to shared
1960 libraries in that they are relocated by the dynamic linker to the virtual
1961 address the OS chooses for them (which can vary between invocations). Like
1962 normal dynamically linked executables they can be executed and symbols
1963 defined in the executable cannot be overridden by shared libraries.
1967 This option is ignored for Linux compatibility.
1971 This option is ignored for SVR4 compatibility.
1974 @cindex synthesizing linker
1975 @cindex relaxing addressing modes
1979 An option with machine dependent effects.
1981 This option is only supported on a few targets.
1984 @xref{H8/300,,@command{ld} and the H8/300}.
1987 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1990 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1993 @xref{Nios II,,@command{ld} and the Altera Nios II}.
1996 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1999 On some platforms the @samp{--relax} option performs target-specific,
2000 global optimizations that become possible when the linker resolves
2001 addressing in the program, such as relaxing address modes,
2002 synthesizing new instructions, selecting shorter version of current
2003 instructions, and combining constant values.
2005 On some platforms these link time global optimizations may make symbolic
2006 debugging of the resulting executable impossible.
2008 This is known to be the case for the Matsushita MN10200 and MN10300
2009 family of processors.
2013 On platforms where this is not supported, @samp{--relax} is accepted,
2017 On platforms where @samp{--relax} is accepted the option
2018 @samp{--no-relax} can be used to disable the feature.
2020 @cindex retaining specified symbols
2021 @cindex stripping all but some symbols
2022 @cindex symbols, retaining selectively
2023 @kindex --retain-symbols-file=@var{filename}
2024 @item --retain-symbols-file=@var{filename}
2025 Retain @emph{only} the symbols listed in the file @var{filename},
2026 discarding all others. @var{filename} is simply a flat file, with one
2027 symbol name per line. This option is especially useful in environments
2031 where a large global symbol table is accumulated gradually, to conserve
2034 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
2035 or symbols needed for relocations.
2037 You may only specify @samp{--retain-symbols-file} once in the command
2038 line. It overrides @samp{-s} and @samp{-S}.
2041 @item -rpath=@var{dir}
2042 @cindex runtime library search path
2043 @kindex -rpath=@var{dir}
2044 Add a directory to the runtime library search path. This is used when
2045 linking an ELF executable with shared objects. All @option{-rpath}
2046 arguments are concatenated and passed to the runtime linker, which uses
2047 them to locate shared objects at runtime.
2049 The @option{-rpath} option is also used when locating shared objects which
2050 are needed by shared objects explicitly included in the link; see the
2051 description of the @option{-rpath-link} option. Searching @option{-rpath}
2052 in this way is only supported by native linkers and cross linkers which
2053 have been configured with the @option{--with-sysroot} option.
2055 If @option{-rpath} is not used when linking an ELF executable, the
2056 contents of the environment variable @code{LD_RUN_PATH} will be used if it
2059 The @option{-rpath} option may also be used on SunOS. By default, on
2060 SunOS, the linker will form a runtime search path out of all the
2061 @option{-L} options it is given. If a @option{-rpath} option is used, the
2062 runtime search path will be formed exclusively using the @option{-rpath}
2063 options, ignoring the @option{-L} options. This can be useful when using
2064 gcc, which adds many @option{-L} options which may be on NFS mounted
2067 For compatibility with other ELF linkers, if the @option{-R} option is
2068 followed by a directory name, rather than a file name, it is treated as
2069 the @option{-rpath} option.
2073 @cindex link-time runtime library search path
2074 @kindex -rpath-link=@var{dir}
2075 @item -rpath-link=@var{dir}
2076 When using ELF or SunOS, one shared library may require another. This
2077 happens when an @code{ld -shared} link includes a shared library as one
2080 When the linker encounters such a dependency when doing a non-shared,
2081 non-relocatable link, it will automatically try to locate the required
2082 shared library and include it in the link, if it is not included
2083 explicitly. In such a case, the @option{-rpath-link} option
2084 specifies the first set of directories to search. The
2085 @option{-rpath-link} option may specify a sequence of directory names
2086 either by specifying a list of names separated by colons, or by
2087 appearing multiple times.
2089 The tokens @var{$ORIGIN} and @var{$LIB} can appear in these search
2090 directories. They will be replaced by the full path to the directory
2091 containing the program or shared object in the case of @var{$ORIGIN}
2092 and either @samp{lib} - for 32-bit binaries - or @samp{lib64} - for
2093 64-bit binaries - in the case of @var{$LIB}.
2095 The alternative form of these tokens - @var{$@{ORIGIN@}} and
2096 @var{$@{LIB@}} can also be used. The token @var{$PLATFORM} is not
2099 This option should be used with caution as it overrides the search path
2100 that may have been hard compiled into a shared library. In such a case it
2101 is possible to use unintentionally a different search path than the
2102 runtime linker would do.
2104 The linker uses the following search paths to locate required shared
2109 Any directories specified by @option{-rpath-link} options.
2111 Any directories specified by @option{-rpath} options. The difference
2112 between @option{-rpath} and @option{-rpath-link} is that directories
2113 specified by @option{-rpath} options are included in the executable and
2114 used at runtime, whereas the @option{-rpath-link} option is only effective
2115 at link time. Searching @option{-rpath} in this way is only supported
2116 by native linkers and cross linkers which have been configured with
2117 the @option{--with-sysroot} option.
2119 On an ELF system, for native linkers, if the @option{-rpath} and
2120 @option{-rpath-link} options were not used, search the contents of the
2121 environment variable @code{LD_RUN_PATH}.
2123 On SunOS, if the @option{-rpath} option was not used, search any
2124 directories specified using @option{-L} options.
2126 For a native linker, search the contents of the environment
2127 variable @code{LD_LIBRARY_PATH}.
2129 For a native ELF linker, the directories in @code{DT_RUNPATH} or
2130 @code{DT_RPATH} of a shared library are searched for shared
2131 libraries needed by it. The @code{DT_RPATH} entries are ignored if
2132 @code{DT_RUNPATH} entries exist.
2134 The default directories, normally @file{/lib} and @file{/usr/lib}.
2136 For a linker for a Linux system, if the file @file{/etc/ld.so.conf}
2137 exists, the list of directories found in that file. Note: the path
2138 to this file is prefixed with the @code{sysroot} value, if that is
2139 defined, and then any @code{prefix} string if the linker was
2140 configured with the @command{--prefix=<path>} option.
2142 For a native linker on a FreeBSD system, any directories specified by
2143 the @code{_PATH_ELF_HINTS} macro defined in the @file{elf-hints.h}
2146 Any directories specifed by a @code{SEARCH_DIR} command in the
2147 linker script being used.
2150 If the required shared library is not found, the linker will issue a
2151 warning and continue with the link.
2158 @cindex shared libraries
2159 Create a shared library. This is currently only supported on ELF, XCOFF
2160 and SunOS platforms. On SunOS, the linker will automatically create a
2161 shared library if the @option{-e} option is not used and there are
2162 undefined symbols in the link.
2164 @kindex --sort-common
2166 @itemx --sort-common=ascending
2167 @itemx --sort-common=descending
2168 This option tells @command{ld} to sort the common symbols by alignment in
2169 ascending or descending order when it places them in the appropriate output
2170 sections. The symbol alignments considered are sixteen-byte or larger,
2171 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
2172 between symbols due to alignment constraints. If no sorting order is
2173 specified, then descending order is assumed.
2175 @kindex --sort-section=name
2176 @item --sort-section=name
2177 This option will apply @code{SORT_BY_NAME} to all wildcard section
2178 patterns in the linker script.
2180 @kindex --sort-section=alignment
2181 @item --sort-section=alignment
2182 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
2183 patterns in the linker script.
2185 @kindex --spare-dynamic-tags
2186 @item --spare-dynamic-tags=@var{count}
2187 This option specifies the number of empty slots to leave in the
2188 .dynamic section of ELF shared objects. Empty slots may be needed by
2189 post processing tools, such as the prelinker. The default is 5.
2191 @kindex --split-by-file
2192 @item --split-by-file[=@var{size}]
2193 Similar to @option{--split-by-reloc} but creates a new output section for
2194 each input file when @var{size} is reached. @var{size} defaults to a
2195 size of 1 if not given.
2197 @kindex --split-by-reloc
2198 @item --split-by-reloc[=@var{count}]
2199 Tries to creates extra sections in the output file so that no single
2200 output section in the file contains more than @var{count} relocations.
2201 This is useful when generating huge relocatable files for downloading into
2202 certain real time kernels with the COFF object file format; since COFF
2203 cannot represent more than 65535 relocations in a single section. Note
2204 that this will fail to work with object file formats which do not
2205 support arbitrary sections. The linker will not split up individual
2206 input sections for redistribution, so if a single input section contains
2207 more than @var{count} relocations one output section will contain that
2208 many relocations. @var{count} defaults to a value of 32768.
2212 Compute and display statistics about the operation of the linker, such
2213 as execution time and memory usage.
2215 @kindex --sysroot=@var{directory}
2216 @item --sysroot=@var{directory}
2217 Use @var{directory} as the location of the sysroot, overriding the
2218 configure-time default. This option is only supported by linkers
2219 that were configured using @option{--with-sysroot}.
2223 This is used by COFF/PE based targets to create a task-linked object
2224 file where all of the global symbols have been converted to statics.
2226 @kindex --traditional-format
2227 @cindex traditional format
2228 @item --traditional-format
2229 For some targets, the output of @command{ld} is different in some ways from
2230 the output of some existing linker. This switch requests @command{ld} to
2231 use the traditional format instead.
2234 For example, on SunOS, @command{ld} combines duplicate entries in the
2235 symbol string table. This can reduce the size of an output file with
2236 full debugging information by over 30 percent. Unfortunately, the SunOS
2237 @code{dbx} program can not read the resulting program (@code{gdb} has no
2238 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
2239 combine duplicate entries.
2241 @kindex --section-start=@var{sectionname}=@var{org}
2242 @item --section-start=@var{sectionname}=@var{org}
2243 Locate a section in the output file at the absolute
2244 address given by @var{org}. You may use this option as many
2245 times as necessary to locate multiple sections in the command
2247 @var{org} must be a single hexadecimal integer;
2248 for compatibility with other linkers, you may omit the leading
2249 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
2250 should be no white space between @var{sectionname}, the equals
2251 sign (``@key{=}''), and @var{org}.
2253 @kindex -Tbss=@var{org}
2254 @kindex -Tdata=@var{org}
2255 @kindex -Ttext=@var{org}
2256 @cindex segment origins, cmd line
2257 @item -Tbss=@var{org}
2258 @itemx -Tdata=@var{org}
2259 @itemx -Ttext=@var{org}
2260 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2261 @code{.text} as the @var{sectionname}.
2263 @kindex -Ttext-segment=@var{org}
2264 @item -Ttext-segment=@var{org}
2265 @cindex text segment origin, cmd line
2266 When creating an ELF executable, it will set the address of the first
2267 byte of the text segment.
2269 @kindex -Trodata-segment=@var{org}
2270 @item -Trodata-segment=@var{org}
2271 @cindex rodata segment origin, cmd line
2272 When creating an ELF executable or shared object for a target where
2273 the read-only data is in its own segment separate from the executable
2274 text, it will set the address of the first byte of the read-only data segment.
2276 @kindex -Tldata-segment=@var{org}
2277 @item -Tldata-segment=@var{org}
2278 @cindex ldata segment origin, cmd line
2279 When creating an ELF executable or shared object for x86-64 medium memory
2280 model, it will set the address of the first byte of the ldata segment.
2282 @kindex --unresolved-symbols
2283 @item --unresolved-symbols=@var{method}
2284 Determine how to handle unresolved symbols. There are four possible
2285 values for @samp{method}:
2289 Do not report any unresolved symbols.
2292 Report all unresolved symbols. This is the default.
2294 @item ignore-in-object-files
2295 Report unresolved symbols that are contained in shared libraries, but
2296 ignore them if they come from regular object files.
2298 @item ignore-in-shared-libs
2299 Report unresolved symbols that come from regular object files, but
2300 ignore them if they come from shared libraries. This can be useful
2301 when creating a dynamic binary and it is known that all the shared
2302 libraries that it should be referencing are included on the linker's
2306 The behaviour for shared libraries on their own can also be controlled
2307 by the @option{--[no-]allow-shlib-undefined} option.
2309 Normally the linker will generate an error message for each reported
2310 unresolved symbol but the option @option{--warn-unresolved-symbols}
2311 can change this to a warning.
2313 @kindex --verbose[=@var{NUMBER}]
2314 @cindex verbose[=@var{NUMBER}]
2316 @itemx --verbose[=@var{NUMBER}]
2317 Display the version number for @command{ld} and list the linker emulations
2318 supported. Display which input files can and cannot be opened. Display
2319 the linker script being used by the linker. If the optional @var{NUMBER}
2320 argument > 1, plugin symbol status will also be displayed.
2322 @kindex --version-script=@var{version-scriptfile}
2323 @cindex version script, symbol versions
2324 @item --version-script=@var{version-scriptfile}
2325 Specify the name of a version script to the linker. This is typically
2326 used when creating shared libraries to specify additional information
2327 about the version hierarchy for the library being created. This option
2328 is only fully supported on ELF platforms which support shared libraries;
2329 see @ref{VERSION}. It is partially supported on PE platforms, which can
2330 use version scripts to filter symbol visibility in auto-export mode: any
2331 symbols marked @samp{local} in the version script will not be exported.
2334 @kindex --warn-common
2335 @cindex warnings, on combining symbols
2336 @cindex combining symbols, warnings on
2338 Warn when a common symbol is combined with another common symbol or with
2339 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2340 but linkers on some other operating systems do not. This option allows
2341 you to find potential problems from combining global symbols.
2342 Unfortunately, some C libraries use this practice, so you may get some
2343 warnings about symbols in the libraries as well as in your programs.
2345 There are three kinds of global symbols, illustrated here by C examples:
2349 A definition, which goes in the initialized data section of the output
2353 An undefined reference, which does not allocate space.
2354 There must be either a definition or a common symbol for the
2358 A common symbol. If there are only (one or more) common symbols for a
2359 variable, it goes in the uninitialized data area of the output file.
2360 The linker merges multiple common symbols for the same variable into a
2361 single symbol. If they are of different sizes, it picks the largest
2362 size. The linker turns a common symbol into a declaration, if there is
2363 a definition of the same variable.
2366 The @samp{--warn-common} option can produce five kinds of warnings.
2367 Each warning consists of a pair of lines: the first describes the symbol
2368 just encountered, and the second describes the previous symbol
2369 encountered with the same name. One or both of the two symbols will be
2374 Turning a common symbol into a reference, because there is already a
2375 definition for the symbol.
2377 @var{file}(@var{section}): warning: common of `@var{symbol}'
2378 overridden by definition
2379 @var{file}(@var{section}): warning: defined here
2383 Turning a common symbol into a reference, because a later definition for
2384 the symbol is encountered. This is the same as the previous case,
2385 except that the symbols are encountered in a different order.
2387 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2389 @var{file}(@var{section}): warning: common is here
2393 Merging a common symbol with a previous same-sized common symbol.
2395 @var{file}(@var{section}): warning: multiple common
2397 @var{file}(@var{section}): warning: previous common is here
2401 Merging a common symbol with a previous larger common symbol.
2403 @var{file}(@var{section}): warning: common of `@var{symbol}'
2404 overridden by larger common
2405 @var{file}(@var{section}): warning: larger common is here
2409 Merging a common symbol with a previous smaller common symbol. This is
2410 the same as the previous case, except that the symbols are
2411 encountered in a different order.
2413 @var{file}(@var{section}): warning: common of `@var{symbol}'
2414 overriding smaller common
2415 @var{file}(@var{section}): warning: smaller common is here
2419 @kindex --warn-constructors
2420 @item --warn-constructors
2421 Warn if any global constructors are used. This is only useful for a few
2422 object file formats. For formats like COFF or ELF, the linker can not
2423 detect the use of global constructors.
2425 @kindex --warn-multiple-gp
2426 @item --warn-multiple-gp
2427 Warn if multiple global pointer values are required in the output file.
2428 This is only meaningful for certain processors, such as the Alpha.
2429 Specifically, some processors put large-valued constants in a special
2430 section. A special register (the global pointer) points into the middle
2431 of this section, so that constants can be loaded efficiently via a
2432 base-register relative addressing mode. Since the offset in
2433 base-register relative mode is fixed and relatively small (e.g., 16
2434 bits), this limits the maximum size of the constant pool. Thus, in
2435 large programs, it is often necessary to use multiple global pointer
2436 values in order to be able to address all possible constants. This
2437 option causes a warning to be issued whenever this case occurs.
2440 @cindex warnings, on undefined symbols
2441 @cindex undefined symbols, warnings on
2443 Only warn once for each undefined symbol, rather than once per module
2446 @kindex --warn-section-align
2447 @cindex warnings, on section alignment
2448 @cindex section alignment, warnings on
2449 @item --warn-section-align
2450 Warn if the address of an output section is changed because of
2451 alignment. Typically, the alignment will be set by an input section.
2452 The address will only be changed if it not explicitly specified; that
2453 is, if the @code{SECTIONS} command does not specify a start address for
2454 the section (@pxref{SECTIONS}).
2456 @kindex --warn-textrel
2457 @item --warn-textrel
2458 Warn if the linker adds DT_TEXTREL to a position-independent executable
2461 @kindex --warn-alternate-em
2462 @item --warn-alternate-em
2463 Warn if an object has alternate ELF machine code.
2465 @kindex --warn-unresolved-symbols
2466 @item --warn-unresolved-symbols
2467 If the linker is going to report an unresolved symbol (see the option
2468 @option{--unresolved-symbols}) it will normally generate an error.
2469 This option makes it generate a warning instead.
2471 @kindex --error-unresolved-symbols
2472 @item --error-unresolved-symbols
2473 This restores the linker's default behaviour of generating errors when
2474 it is reporting unresolved symbols.
2476 @kindex --whole-archive
2477 @cindex including an entire archive
2478 @item --whole-archive
2479 For each archive mentioned on the command line after the
2480 @option{--whole-archive} option, include every object file in the archive
2481 in the link, rather than searching the archive for the required object
2482 files. This is normally used to turn an archive file into a shared
2483 library, forcing every object to be included in the resulting shared
2484 library. This option may be used more than once.
2486 Two notes when using this option from gcc: First, gcc doesn't know
2487 about this option, so you have to use @option{-Wl,-whole-archive}.
2488 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2489 list of archives, because gcc will add its own list of archives to
2490 your link and you may not want this flag to affect those as well.
2492 @kindex --wrap=@var{symbol}
2493 @item --wrap=@var{symbol}
2494 Use a wrapper function for @var{symbol}. Any undefined reference to
2495 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2496 undefined reference to @code{__real_@var{symbol}} will be resolved to
2499 This can be used to provide a wrapper for a system function. The
2500 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2501 wishes to call the system function, it should call
2502 @code{__real_@var{symbol}}.
2504 Here is a trivial example:
2508 __wrap_malloc (size_t c)
2510 printf ("malloc called with %zu\n", c);
2511 return __real_malloc (c);
2515 If you link other code with this file using @option{--wrap malloc}, then
2516 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2517 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2518 call the real @code{malloc} function.
2520 You may wish to provide a @code{__real_malloc} function as well, so that
2521 links without the @option{--wrap} option will succeed. If you do this,
2522 you should not put the definition of @code{__real_malloc} in the same
2523 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2524 call before the linker has a chance to wrap it to @code{malloc}.
2526 Only undefined references are replaced by the linker. So, translation unit
2527 internal references to @var{symbol} are not resolved to
2528 @code{__wrap_@var{symbol}}. In the next example, the call to @code{f} in
2529 @code{g} is not resolved to @code{__wrap_f}.
2545 @kindex --eh-frame-hdr
2546 @kindex --no-eh-frame-hdr
2547 @item --eh-frame-hdr
2548 @itemx --no-eh-frame-hdr
2549 Request (@option{--eh-frame-hdr}) or suppress
2550 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2551 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2553 @kindex --ld-generated-unwind-info
2554 @item --no-ld-generated-unwind-info
2555 Request creation of @code{.eh_frame} unwind info for linker
2556 generated code sections like PLT. This option is on by default
2557 if linker generated unwind info is supported.
2559 @kindex --enable-new-dtags
2560 @kindex --disable-new-dtags
2561 @item --enable-new-dtags
2562 @itemx --disable-new-dtags
2563 This linker can create the new dynamic tags in ELF. But the older ELF
2564 systems may not understand them. If you specify
2565 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2566 and older dynamic tags will be omitted.
2567 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2568 created. By default, the new dynamic tags are not created. Note that
2569 those options are only available for ELF systems.
2571 @kindex --hash-size=@var{number}
2572 @item --hash-size=@var{number}
2573 Set the default size of the linker's hash tables to a prime number
2574 close to @var{number}. Increasing this value can reduce the length of
2575 time it takes the linker to perform its tasks, at the expense of
2576 increasing the linker's memory requirements. Similarly reducing this
2577 value can reduce the memory requirements at the expense of speed.
2579 @kindex --hash-style=@var{style}
2580 @item --hash-style=@var{style}
2581 Set the type of linker's hash table(s). @var{style} can be either
2582 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2583 new style GNU @code{.gnu.hash} section or @code{both} for both
2584 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2585 hash tables. The default depends upon how the linker was configured,
2586 but for most Linux based systems it will be @code{both}.
2588 @kindex --compress-debug-sections=none
2589 @kindex --compress-debug-sections=zlib
2590 @kindex --compress-debug-sections=zlib-gnu
2591 @kindex --compress-debug-sections=zlib-gabi
2592 @item --compress-debug-sections=none
2593 @itemx --compress-debug-sections=zlib
2594 @itemx --compress-debug-sections=zlib-gnu
2595 @itemx --compress-debug-sections=zlib-gabi
2596 On ELF platforms, these options control how DWARF debug sections are
2597 compressed using zlib.
2599 @option{--compress-debug-sections=none} doesn't compress DWARF debug
2600 sections. @option{--compress-debug-sections=zlib-gnu} compresses
2601 DWARF debug sections and renames them to begin with @samp{.zdebug}
2602 instead of @samp{.debug}. @option{--compress-debug-sections=zlib-gabi}
2603 also compresses DWARF debug sections, but rather than renaming them it
2604 sets the SHF_COMPRESSED flag in the sections' headers.
2606 The @option{--compress-debug-sections=zlib} option is an alias for
2607 @option{--compress-debug-sections=zlib-gabi}.
2609 Note that this option overrides any compression in input debug
2610 sections, so if a binary is linked with @option{--compress-debug-sections=none}
2611 for example, then any compressed debug sections in input files will be
2612 uncompressed before they are copied into the output binary.
2614 The default compression behaviour varies depending upon the target
2615 involved and the configure options used to build the toolchain. The
2616 default can be determined by examining the output from the linker's
2617 @option{--help} option.
2619 @kindex --reduce-memory-overheads
2620 @item --reduce-memory-overheads
2621 This option reduces memory requirements at ld runtime, at the expense of
2622 linking speed. This was introduced to select the old O(n^2) algorithm
2623 for link map file generation, rather than the new O(n) algorithm which uses
2624 about 40% more memory for symbol storage.
2626 Another effect of the switch is to set the default hash table size to
2627 1021, which again saves memory at the cost of lengthening the linker's
2628 run time. This is not done however if the @option{--hash-size} switch
2631 The @option{--reduce-memory-overheads} switch may be also be used to
2632 enable other tradeoffs in future versions of the linker.
2635 @kindex --build-id=@var{style}
2637 @itemx --build-id=@var{style}
2638 Request the creation of a @code{.note.gnu.build-id} ELF note section
2639 or a @code{.buildid} COFF section. The contents of the note are
2640 unique bits identifying this linked file. @var{style} can be
2641 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2642 @sc{SHA1} hash on the normative parts of the output contents,
2643 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2644 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2645 string specified as an even number of hexadecimal digits (@code{-} and
2646 @code{:} characters between digit pairs are ignored). If @var{style}
2647 is omitted, @code{sha1} is used.
2649 The @code{md5} and @code{sha1} styles produces an identifier
2650 that is always the same in an identical output file, but will be
2651 unique among all nonidentical output files. It is not intended
2652 to be compared as a checksum for the file's contents. A linked
2653 file may be changed later by other tools, but the build ID bit
2654 string identifying the original linked file does not change.
2656 Passing @code{none} for @var{style} disables the setting from any
2657 @code{--build-id} options earlier on the command line.
2662 @subsection Options Specific to i386 PE Targets
2664 @c man begin OPTIONS
2666 The i386 PE linker supports the @option{-shared} option, which causes
2667 the output to be a dynamically linked library (DLL) instead of a
2668 normal executable. You should name the output @code{*.dll} when you
2669 use this option. In addition, the linker fully supports the standard
2670 @code{*.def} files, which may be specified on the linker command line
2671 like an object file (in fact, it should precede archives it exports
2672 symbols from, to ensure that they get linked in, just like a normal
2675 In addition to the options common to all targets, the i386 PE linker
2676 support additional command-line options that are specific to the i386
2677 PE target. Options that take values may be separated from their
2678 values by either a space or an equals sign.
2682 @kindex --add-stdcall-alias
2683 @item --add-stdcall-alias
2684 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2685 as-is and also with the suffix stripped.
2686 [This option is specific to the i386 PE targeted port of the linker]
2689 @item --base-file @var{file}
2690 Use @var{file} as the name of a file in which to save the base
2691 addresses of all the relocations needed for generating DLLs with
2693 [This is an i386 PE specific option]
2697 Create a DLL instead of a regular executable. You may also use
2698 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2700 [This option is specific to the i386 PE targeted port of the linker]
2702 @kindex --enable-long-section-names
2703 @kindex --disable-long-section-names
2704 @item --enable-long-section-names
2705 @itemx --disable-long-section-names
2706 The PE variants of the COFF object format add an extension that permits
2707 the use of section names longer than eight characters, the normal limit
2708 for COFF. By default, these names are only allowed in object files, as
2709 fully-linked executable images do not carry the COFF string table required
2710 to support the longer names. As a GNU extension, it is possible to
2711 allow their use in executable images as well, or to (probably pointlessly!)
2712 disallow it in object files, by using these two options. Executable images
2713 generated with these long section names are slightly non-standard, carrying
2714 as they do a string table, and may generate confusing output when examined
2715 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2716 GDB relies on the use of PE long section names to find Dwarf-2 debug
2717 information sections in an executable image at runtime, and so if neither
2718 option is specified on the command-line, @command{ld} will enable long
2719 section names, overriding the default and technically correct behaviour,
2720 when it finds the presence of debug information while linking an executable
2721 image and not stripping symbols.
2722 [This option is valid for all PE targeted ports of the linker]
2724 @kindex --enable-stdcall-fixup
2725 @kindex --disable-stdcall-fixup
2726 @item --enable-stdcall-fixup
2727 @itemx --disable-stdcall-fixup
2728 If the link finds a symbol that it cannot resolve, it will attempt to
2729 do ``fuzzy linking'' by looking for another defined symbol that differs
2730 only in the format of the symbol name (cdecl vs stdcall) and will
2731 resolve that symbol by linking to the match. For example, the
2732 undefined symbol @code{_foo} might be linked to the function
2733 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2734 to the function @code{_bar}. When the linker does this, it prints a
2735 warning, since it normally should have failed to link, but sometimes
2736 import libraries generated from third-party dlls may need this feature
2737 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2738 feature is fully enabled and warnings are not printed. If you specify
2739 @option{--disable-stdcall-fixup}, this feature is disabled and such
2740 mismatches are considered to be errors.
2741 [This option is specific to the i386 PE targeted port of the linker]
2743 @kindex --leading-underscore
2744 @kindex --no-leading-underscore
2745 @item --leading-underscore
2746 @itemx --no-leading-underscore
2747 For most targets default symbol-prefix is an underscore and is defined
2748 in target's description. By this option it is possible to
2749 disable/enable the default underscore symbol-prefix.
2751 @cindex DLLs, creating
2752 @kindex --export-all-symbols
2753 @item --export-all-symbols
2754 If given, all global symbols in the objects used to build a DLL will
2755 be exported by the DLL. Note that this is the default if there
2756 otherwise wouldn't be any exported symbols. When symbols are
2757 explicitly exported via DEF files or implicitly exported via function
2758 attributes, the default is to not export anything else unless this
2759 option is given. Note that the symbols @code{DllMain@@12},
2760 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2761 @code{impure_ptr} will not be automatically
2762 exported. Also, symbols imported from other DLLs will not be
2763 re-exported, nor will symbols specifying the DLL's internal layout
2764 such as those beginning with @code{_head_} or ending with
2765 @code{_iname}. In addition, no symbols from @code{libgcc},
2766 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2767 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2768 not be exported, to help with C++ DLLs. Finally, there is an
2769 extensive list of cygwin-private symbols that are not exported
2770 (obviously, this applies on when building DLLs for cygwin targets).
2771 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2772 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2773 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2774 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2775 @code{cygwin_premain3}, and @code{environ}.
2776 [This option is specific to the i386 PE targeted port of the linker]
2778 @kindex --exclude-symbols
2779 @item --exclude-symbols @var{symbol},@var{symbol},...
2780 Specifies a list of symbols which should not be automatically
2781 exported. The symbol names may be delimited by commas or colons.
2782 [This option is specific to the i386 PE targeted port of the linker]
2784 @kindex --exclude-all-symbols
2785 @item --exclude-all-symbols
2786 Specifies no symbols should be automatically exported.
2787 [This option is specific to the i386 PE targeted port of the linker]
2789 @kindex --file-alignment
2790 @item --file-alignment
2791 Specify the file alignment. Sections in the file will always begin at
2792 file offsets which are multiples of this number. This defaults to
2794 [This option is specific to the i386 PE targeted port of the linker]
2798 @item --heap @var{reserve}
2799 @itemx --heap @var{reserve},@var{commit}
2800 Specify the number of bytes of memory to reserve (and optionally commit)
2801 to be used as heap for this program. The default is 1MB reserved, 4K
2803 [This option is specific to the i386 PE targeted port of the linker]
2806 @kindex --image-base
2807 @item --image-base @var{value}
2808 Use @var{value} as the base address of your program or dll. This is
2809 the lowest memory location that will be used when your program or dll
2810 is loaded. To reduce the need to relocate and improve performance of
2811 your dlls, each should have a unique base address and not overlap any
2812 other dlls. The default is 0x400000 for executables, and 0x10000000
2814 [This option is specific to the i386 PE targeted port of the linker]
2818 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2819 symbols before they are exported.
2820 [This option is specific to the i386 PE targeted port of the linker]
2822 @kindex --large-address-aware
2823 @item --large-address-aware
2824 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2825 header is set to indicate that this executable supports virtual addresses
2826 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2827 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2828 section of the BOOT.INI. Otherwise, this bit has no effect.
2829 [This option is specific to PE targeted ports of the linker]
2831 @kindex --disable-large-address-aware
2832 @item --disable-large-address-aware
2833 Reverts the effect of a previous @samp{--large-address-aware} option.
2834 This is useful if @samp{--large-address-aware} is always set by the compiler
2835 driver (e.g. Cygwin gcc) and the executable does not support virtual
2836 addresses greater than 2 gigabytes.
2837 [This option is specific to PE targeted ports of the linker]
2839 @kindex --major-image-version
2840 @item --major-image-version @var{value}
2841 Sets the major number of the ``image version''. Defaults to 1.
2842 [This option is specific to the i386 PE targeted port of the linker]
2844 @kindex --major-os-version
2845 @item --major-os-version @var{value}
2846 Sets the major number of the ``os version''. Defaults to 4.
2847 [This option is specific to the i386 PE targeted port of the linker]
2849 @kindex --major-subsystem-version
2850 @item --major-subsystem-version @var{value}
2851 Sets the major number of the ``subsystem version''. Defaults to 4.
2852 [This option is specific to the i386 PE targeted port of the linker]
2854 @kindex --minor-image-version
2855 @item --minor-image-version @var{value}
2856 Sets the minor number of the ``image version''. Defaults to 0.
2857 [This option is specific to the i386 PE targeted port of the linker]
2859 @kindex --minor-os-version
2860 @item --minor-os-version @var{value}
2861 Sets the minor number of the ``os version''. Defaults to 0.
2862 [This option is specific to the i386 PE targeted port of the linker]
2864 @kindex --minor-subsystem-version
2865 @item --minor-subsystem-version @var{value}
2866 Sets the minor number of the ``subsystem version''. Defaults to 0.
2867 [This option is specific to the i386 PE targeted port of the linker]
2869 @cindex DEF files, creating
2870 @cindex DLLs, creating
2871 @kindex --output-def
2872 @item --output-def @var{file}
2873 The linker will create the file @var{file} which will contain a DEF
2874 file corresponding to the DLL the linker is generating. This DEF file
2875 (which should be called @code{*.def}) may be used to create an import
2876 library with @code{dlltool} or may be used as a reference to
2877 automatically or implicitly exported symbols.
2878 [This option is specific to the i386 PE targeted port of the linker]
2880 @cindex DLLs, creating
2881 @kindex --enable-auto-image-base
2882 @item --enable-auto-image-base
2883 @itemx --enable-auto-image-base=@var{value}
2884 Automatically choose the image base for DLLs, optionally starting with base
2885 @var{value}, unless one is specified using the @code{--image-base} argument.
2886 By using a hash generated from the dllname to create unique image bases
2887 for each DLL, in-memory collisions and relocations which can delay program
2888 execution are avoided.
2889 [This option is specific to the i386 PE targeted port of the linker]
2891 @kindex --disable-auto-image-base
2892 @item --disable-auto-image-base
2893 Do not automatically generate a unique image base. If there is no
2894 user-specified image base (@code{--image-base}) then use the platform
2896 [This option is specific to the i386 PE targeted port of the linker]
2898 @cindex DLLs, linking to
2899 @kindex --dll-search-prefix
2900 @item --dll-search-prefix @var{string}
2901 When linking dynamically to a dll without an import library,
2902 search for @code{<string><basename>.dll} in preference to
2903 @code{lib<basename>.dll}. This behaviour allows easy distinction
2904 between DLLs built for the various "subplatforms": native, cygwin,
2905 uwin, pw, etc. For instance, cygwin DLLs typically use
2906 @code{--dll-search-prefix=cyg}.
2907 [This option is specific to the i386 PE targeted port of the linker]
2909 @kindex --enable-auto-import
2910 @item --enable-auto-import
2911 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2912 DATA imports from DLLs, thus making it possible to bypass the dllimport
2913 mechanism on the user side and to reference unmangled symbol names.
2914 [This option is specific to the i386 PE targeted port of the linker]
2916 The following remarks pertain to the original implementation of the
2917 feature and are obsolete nowadays for Cygwin and MinGW targets.
2919 Note: Use of the 'auto-import' extension will cause the text section
2920 of the image file to be made writable. This does not conform to the
2921 PE-COFF format specification published by Microsoft.
2923 Note - use of the 'auto-import' extension will also cause read only
2924 data which would normally be placed into the .rdata section to be
2925 placed into the .data section instead. This is in order to work
2926 around a problem with consts that is described here:
2927 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2929 Using 'auto-import' generally will 'just work' -- but sometimes you may
2932 "variable '<var>' can't be auto-imported. Please read the
2933 documentation for ld's @code{--enable-auto-import} for details."
2935 This message occurs when some (sub)expression accesses an address
2936 ultimately given by the sum of two constants (Win32 import tables only
2937 allow one). Instances where this may occur include accesses to member
2938 fields of struct variables imported from a DLL, as well as using a
2939 constant index into an array variable imported from a DLL. Any
2940 multiword variable (arrays, structs, long long, etc) may trigger
2941 this error condition. However, regardless of the exact data type
2942 of the offending exported variable, ld will always detect it, issue
2943 the warning, and exit.
2945 There are several ways to address this difficulty, regardless of the
2946 data type of the exported variable:
2948 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2949 of adjusting references in your client code for runtime environment, so
2950 this method works only when runtime environment supports this feature.
2952 A second solution is to force one of the 'constants' to be a variable --
2953 that is, unknown and un-optimizable at compile time. For arrays,
2954 there are two possibilities: a) make the indexee (the array's address)
2955 a variable, or b) make the 'constant' index a variable. Thus:
2958 extern type extern_array[];
2960 @{ volatile type *t=extern_array; t[1] @}
2966 extern type extern_array[];
2968 @{ volatile int t=1; extern_array[t] @}
2971 For structs (and most other multiword data types) the only option
2972 is to make the struct itself (or the long long, or the ...) variable:
2975 extern struct s extern_struct;
2976 extern_struct.field -->
2977 @{ volatile struct s *t=&extern_struct; t->field @}
2983 extern long long extern_ll;
2985 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2988 A third method of dealing with this difficulty is to abandon
2989 'auto-import' for the offending symbol and mark it with
2990 @code{__declspec(dllimport)}. However, in practice that
2991 requires using compile-time #defines to indicate whether you are
2992 building a DLL, building client code that will link to the DLL, or
2993 merely building/linking to a static library. In making the choice
2994 between the various methods of resolving the 'direct address with
2995 constant offset' problem, you should consider typical real-world usage:
3003 void main(int argc, char **argv)@{
3004 printf("%d\n",arr[1]);
3014 void main(int argc, char **argv)@{
3015 /* This workaround is for win32 and cygwin; do not "optimize" */
3016 volatile int *parr = arr;
3017 printf("%d\n",parr[1]);
3024 /* Note: auto-export is assumed (no __declspec(dllexport)) */
3025 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
3026 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
3027 #define FOO_IMPORT __declspec(dllimport)
3031 extern FOO_IMPORT int arr[];
3034 void main(int argc, char **argv)@{
3035 printf("%d\n",arr[1]);
3039 A fourth way to avoid this problem is to re-code your
3040 library to use a functional interface rather than a data interface
3041 for the offending variables (e.g. set_foo() and get_foo() accessor
3044 @kindex --disable-auto-import
3045 @item --disable-auto-import
3046 Do not attempt to do sophisticated linking of @code{_symbol} to
3047 @code{__imp__symbol} for DATA imports from DLLs.
3048 [This option is specific to the i386 PE targeted port of the linker]
3050 @kindex --enable-runtime-pseudo-reloc
3051 @item --enable-runtime-pseudo-reloc
3052 If your code contains expressions described in --enable-auto-import section,
3053 that is, DATA imports from DLL with non-zero offset, this switch will create
3054 a vector of 'runtime pseudo relocations' which can be used by runtime
3055 environment to adjust references to such data in your client code.
3056 [This option is specific to the i386 PE targeted port of the linker]
3058 @kindex --disable-runtime-pseudo-reloc
3059 @item --disable-runtime-pseudo-reloc
3060 Do not create pseudo relocations for non-zero offset DATA imports from DLLs.
3061 [This option is specific to the i386 PE targeted port of the linker]
3063 @kindex --enable-extra-pe-debug
3064 @item --enable-extra-pe-debug
3065 Show additional debug info related to auto-import symbol thunking.
3066 [This option is specific to the i386 PE targeted port of the linker]
3068 @kindex --section-alignment
3069 @item --section-alignment
3070 Sets the section alignment. Sections in memory will always begin at
3071 addresses which are a multiple of this number. Defaults to 0x1000.
3072 [This option is specific to the i386 PE targeted port of the linker]
3076 @item --stack @var{reserve}
3077 @itemx --stack @var{reserve},@var{commit}
3078 Specify the number of bytes of memory to reserve (and optionally commit)
3079 to be used as stack for this program. The default is 2MB reserved, 4K
3081 [This option is specific to the i386 PE targeted port of the linker]
3084 @item --subsystem @var{which}
3085 @itemx --subsystem @var{which}:@var{major}
3086 @itemx --subsystem @var{which}:@var{major}.@var{minor}
3087 Specifies the subsystem under which your program will execute. The
3088 legal values for @var{which} are @code{native}, @code{windows},
3089 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
3090 the subsystem version also. Numeric values are also accepted for
3092 [This option is specific to the i386 PE targeted port of the linker]
3094 The following options set flags in the @code{DllCharacteristics} field
3095 of the PE file header:
3096 [These options are specific to PE targeted ports of the linker]
3098 @kindex --high-entropy-va
3099 @item --high-entropy-va
3100 Image is compatible with 64-bit address space layout randomization
3102 This option also implies @option{--dynamicbase} and
3103 @option{--enable-reloc-section}.
3105 @kindex --dynamicbase
3107 The image base address may be relocated using address space layout
3108 randomization (ASLR). This feature was introduced with MS Windows
3109 Vista for i386 PE targets.
3110 This option also implies @option{--enable-reloc-section}.
3112 @kindex --forceinteg
3114 Code integrity checks are enforced.
3118 The image is compatible with the Data Execution Prevention.
3119 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
3121 @kindex --no-isolation
3122 @item --no-isolation
3123 Although the image understands isolation, do not isolate the image.
3127 The image does not use SEH. No SE handler may be called from
3132 Do not bind this image.
3136 The driver uses the MS Windows Driver Model.
3140 The image is Terminal Server aware.
3142 @kindex --insert-timestamp
3143 @item --insert-timestamp
3144 @itemx --no-insert-timestamp
3145 Insert a real timestamp into the image. This is the default behaviour
3146 as it matches legacy code and it means that the image will work with
3147 other, proprietary tools. The problem with this default is that it
3148 will result in slightly different images being produced each time the
3149 same sources are linked. The option @option{--no-insert-timestamp}
3150 can be used to insert a zero value for the timestamp, this ensuring
3151 that binaries produced from identical sources will compare
3154 @kindex --enable-reloc-section
3155 @item --enable-reloc-section
3156 Create the base relocation table, which is necessary if the image
3157 is loaded at a different image base than specified in the PE header.
3163 @subsection Options specific to C6X uClinux targets
3165 @c man begin OPTIONS
3167 The C6X uClinux target uses a binary format called DSBT to support shared
3168 libraries. Each shared library in the system needs to have a unique index;
3169 all executables use an index of 0.
3174 @item --dsbt-size @var{size}
3175 This option sets the number of entries in the DSBT of the current executable
3176 or shared library to @var{size}. The default is to create a table with 64
3179 @kindex --dsbt-index
3180 @item --dsbt-index @var{index}
3181 This option sets the DSBT index of the current executable or shared library
3182 to @var{index}. The default is 0, which is appropriate for generating
3183 executables. If a shared library is generated with a DSBT index of 0, the
3184 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
3186 @kindex --no-merge-exidx-entries
3187 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
3188 exidx entries in frame unwind info.
3196 @subsection Options specific to C-SKY targets
3198 @c man begin OPTIONS
3202 @kindex --branch-stub on C-SKY
3204 This option enables linker branch relaxation by inserting branch stub
3205 sections when needed to extend the range of branches. This option is
3206 usually not required since C-SKY supports branch and call instructions that
3207 can access the full memory range and branch relaxation is normally handled by
3208 the compiler or assembler.
3210 @kindex --stub-group-size on C-SKY
3211 @item --stub-group-size=@var{N}
3212 This option allows finer control of linker branch stub creation.
3213 It sets the maximum size of a group of input sections that can
3214 be handled by one stub section. A negative value of @var{N} locates
3215 stub sections after their branches, while a positive value allows stub
3216 sections to appear either before or after the branches. Values of
3217 @samp{1} or @samp{-1} indicate that the
3218 linker should choose suitable defaults.
3226 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
3228 @c man begin OPTIONS
3230 The 68HC11 and 68HC12 linkers support specific options to control the
3231 memory bank switching mapping and trampoline code generation.
3235 @kindex --no-trampoline
3236 @item --no-trampoline
3237 This option disables the generation of trampoline. By default a trampoline
3238 is generated for each far function which is called using a @code{jsr}
3239 instruction (this happens when a pointer to a far function is taken).
3241 @kindex --bank-window
3242 @item --bank-window @var{name}
3243 This option indicates to the linker the name of the memory region in
3244 the @samp{MEMORY} specification that describes the memory bank window.
3245 The definition of such region is then used by the linker to compute
3246 paging and addresses within the memory window.
3254 @subsection Options specific to Motorola 68K target
3256 @c man begin OPTIONS
3258 The following options are supported to control handling of GOT generation
3259 when linking for 68K targets.
3264 @item --got=@var{type}
3265 This option tells the linker which GOT generation scheme to use.
3266 @var{type} should be one of @samp{single}, @samp{negative},
3267 @samp{multigot} or @samp{target}. For more information refer to the
3268 Info entry for @file{ld}.
3276 @subsection Options specific to MIPS targets
3278 @c man begin OPTIONS
3280 The following options are supported to control microMIPS instruction
3281 generation and branch relocation checks for ISA mode transitions when
3282 linking for MIPS targets.
3290 These options control the choice of microMIPS instructions used in code
3291 generated by the linker, such as that in the PLT or lazy binding stubs,
3292 or in relaxation. If @samp{--insn32} is used, then the linker only uses
3293 32-bit instruction encodings. By default or if @samp{--no-insn32} is
3294 used, all instruction encodings are used, including 16-bit ones where
3297 @kindex --ignore-branch-isa
3298 @item --ignore-branch-isa
3299 @kindex --no-ignore-branch-isa
3300 @itemx --no-ignore-branch-isa
3301 These options control branch relocation checks for invalid ISA mode
3302 transitions. If @samp{--ignore-branch-isa} is used, then the linker
3303 accepts any branch relocations and any ISA mode transition required
3304 is lost in relocation calculation, except for some cases of @code{BAL}
3305 instructions which meet relaxation conditions and are converted to
3306 equivalent @code{JALX} instructions as the associated relocation is
3307 calculated. By default or if @samp{--no-ignore-branch-isa} is used
3308 a check is made causing the loss of an ISA mode transition to produce
3311 @kindex --compact-branches
3312 @item --compact-branches
3313 @kindex --no-compact-branches
3314 @itemx --no-compact-branches
3315 These options control the generation of compact instructions by the linker
3316 in the PLT entries for MIPS R6.
3325 @subsection Options specific to PDP11 targets
3327 @c man begin OPTIONS
3329 For the pdp11-aout target, three variants of the output format can be
3330 produced as selected by the following options. The default variant
3331 for pdp11-aout is the @samp{--omagic} option, whereas for other
3332 targets @samp{--nmagic} is the default. The @samp{--imagic} option is
3333 defined only for the pdp11-aout target, while the others are described
3334 here as they apply to the pdp11-aout target.
3343 Mark the output as @code{OMAGIC} (0407) in the @file{a.out} header to
3344 indicate that the text segment is not to be write-protected and
3345 shared. Since the text and data sections are both readable and
3346 writable, the data section is allocated immediately contiguous after
3347 the text segment. This is the oldest format for PDP11 executable
3348 programs and is the default for @command{ld} on PDP11 Unix systems
3349 from the beginning through 2.11BSD.
3356 Mark the output as @code{NMAGIC} (0410) in the @file{a.out} header to
3357 indicate that when the output file is executed, the text portion will
3358 be read-only and shareable among all processes executing the same
3359 file. This involves moving the data areas up to the first possible 8K
3360 byte page boundary following the end of the text. This option creates
3361 a @emph{pure executable} format.
3368 Mark the output as @code{IMAGIC} (0411) in the @file{a.out} header to
3369 indicate that when the output file is executed, the program text and
3370 data areas will be loaded into separate address spaces using the split
3371 instruction and data space feature of the memory management unit in
3372 larger models of the PDP11. This doubles the address space available
3373 to the program. The text segment is again pure, write-protected, and
3374 shareable. The only difference in the output format between this
3375 option and the others, besides the magic number, is that both the text
3376 and data sections start at location 0. The @samp{-z} option selected
3377 this format in 2.11BSD. This option creates a @emph{separate
3383 Equivalent to @samp{--nmagic} for pdp11-aout.
3392 @section Environment Variables
3394 @c man begin ENVIRONMENT
3396 You can change the behaviour of @command{ld} with the environment variables
3397 @ifclear SingleFormat
3400 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
3402 @ifclear SingleFormat
3404 @cindex default input format
3405 @code{GNUTARGET} determines the input-file object format if you don't
3406 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
3407 of the BFD names for an input format (@pxref{BFD}). If there is no
3408 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
3409 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
3410 attempts to discover the input format by examining binary input files;
3411 this method often succeeds, but there are potential ambiguities, since
3412 there is no method of ensuring that the magic number used to specify
3413 object-file formats is unique. However, the configuration procedure for
3414 BFD on each system places the conventional format for that system first
3415 in the search-list, so ambiguities are resolved in favor of convention.
3419 @cindex default emulation
3420 @cindex emulation, default
3421 @code{LDEMULATION} determines the default emulation if you don't use the
3422 @samp{-m} option. The emulation can affect various aspects of linker
3423 behaviour, particularly the default linker script. You can list the
3424 available emulations with the @samp{--verbose} or @samp{-V} options. If
3425 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3426 variable is not defined, the default emulation depends upon how the
3427 linker was configured.
3429 @kindex COLLECT_NO_DEMANGLE
3430 @cindex demangling, default
3431 Normally, the linker will default to demangling symbols. However, if
3432 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3433 default to not demangling symbols. This environment variable is used in
3434 a similar fashion by the @code{gcc} linker wrapper program. The default
3435 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3442 @chapter Linker Scripts
3445 @cindex linker scripts
3446 @cindex command files
3447 Every link is controlled by a @dfn{linker script}. This script is
3448 written in the linker command language.
3450 The main purpose of the linker script is to describe how the sections in
3451 the input files should be mapped into the output file, and to control
3452 the memory layout of the output file. Most linker scripts do nothing
3453 more than this. However, when necessary, the linker script can also
3454 direct the linker to perform many other operations, using the commands
3457 The linker always uses a linker script. If you do not supply one
3458 yourself, the linker will use a default script that is compiled into the
3459 linker executable. You can use the @samp{--verbose} command-line option
3460 to display the default linker script. Certain command-line options,
3461 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3463 You may supply your own linker script by using the @samp{-T} command
3464 line option. When you do this, your linker script will replace the
3465 default linker script.
3467 You may also use linker scripts implicitly by naming them as input files
3468 to the linker, as though they were files to be linked. @xref{Implicit
3472 * Basic Script Concepts:: Basic Linker Script Concepts
3473 * Script Format:: Linker Script Format
3474 * Simple Example:: Simple Linker Script Example
3475 * Simple Commands:: Simple Linker Script Commands
3476 * Assignments:: Assigning Values to Symbols
3477 * SECTIONS:: SECTIONS Command
3478 * MEMORY:: MEMORY Command
3479 * PHDRS:: PHDRS Command
3480 * VERSION:: VERSION Command
3481 * Expressions:: Expressions in Linker Scripts
3482 * Implicit Linker Scripts:: Implicit Linker Scripts
3485 @node Basic Script Concepts
3486 @section Basic Linker Script Concepts
3487 @cindex linker script concepts
3488 We need to define some basic concepts and vocabulary in order to
3489 describe the linker script language.
3491 The linker combines input files into a single output file. The output
3492 file and each input file are in a special data format known as an
3493 @dfn{object file format}. Each file is called an @dfn{object file}.
3494 The output file is often called an @dfn{executable}, but for our
3495 purposes we will also call it an object file. Each object file has,
3496 among other things, a list of @dfn{sections}. We sometimes refer to a
3497 section in an input file as an @dfn{input section}; similarly, a section
3498 in the output file is an @dfn{output section}.
3500 Each section in an object file has a name and a size. Most sections
3501 also have an associated block of data, known as the @dfn{section
3502 contents}. A section may be marked as @dfn{loadable}, which means that
3503 the contents should be loaded into memory when the output file is run.
3504 A section with no contents may be @dfn{allocatable}, which means that an
3505 area in memory should be set aside, but nothing in particular should be
3506 loaded there (in some cases this memory must be zeroed out). A section
3507 which is neither loadable nor allocatable typically contains some sort
3508 of debugging information.
3510 Every loadable or allocatable output section has two addresses. The
3511 first is the @dfn{VMA}, or virtual memory address. This is the address
3512 the section will have when the output file is run. The second is the
3513 @dfn{LMA}, or load memory address. This is the address at which the
3514 section will be loaded. In most cases the two addresses will be the
3515 same. An example of when they might be different is when a data section
3516 is loaded into ROM, and then copied into RAM when the program starts up
3517 (this technique is often used to initialize global variables in a ROM
3518 based system). In this case the ROM address would be the LMA, and the
3519 RAM address would be the VMA.
3521 You can see the sections in an object file by using the @code{objdump}
3522 program with the @samp{-h} option.
3524 Every object file also has a list of @dfn{symbols}, known as the
3525 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3526 has a name, and each defined symbol has an address, among other
3527 information. If you compile a C or C++ program into an object file, you
3528 will get a defined symbol for every defined function and global or
3529 static variable. Every undefined function or global variable which is
3530 referenced in the input file will become an undefined symbol.
3532 You can see the symbols in an object file by using the @code{nm}
3533 program, or by using the @code{objdump} program with the @samp{-t}
3537 @section Linker Script Format
3538 @cindex linker script format
3539 Linker scripts are text files.
3541 You write a linker script as a series of commands. Each command is
3542 either a keyword, possibly followed by arguments, or an assignment to a
3543 symbol. You may separate commands using semicolons. Whitespace is
3546 Strings such as file or format names can normally be entered directly.
3547 If the file name contains a character such as a comma which would
3548 otherwise serve to separate file names, you may put the file name in
3549 double quotes. There is no way to use a double quote character in a
3552 You may include comments in linker scripts just as in C, delimited by
3553 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3556 @node Simple Example
3557 @section Simple Linker Script Example
3558 @cindex linker script example
3559 @cindex example of linker script
3560 Many linker scripts are fairly simple.
3562 The simplest possible linker script has just one command:
3563 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3564 memory layout of the output file.
3566 The @samp{SECTIONS} command is a powerful command. Here we will
3567 describe a simple use of it. Let's assume your program consists only of
3568 code, initialized data, and uninitialized data. These will be in the
3569 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3570 Let's assume further that these are the only sections which appear in
3573 For this example, let's say that the code should be loaded at address
3574 0x10000, and that the data should start at address 0x8000000. Here is a
3575 linker script which will do that:
3580 .text : @{ *(.text) @}
3582 .data : @{ *(.data) @}
3583 .bss : @{ *(.bss) @}
3587 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3588 followed by a series of symbol assignments and output section
3589 descriptions enclosed in curly braces.
3591 The first line inside the @samp{SECTIONS} command of the above example
3592 sets the value of the special symbol @samp{.}, which is the location
3593 counter. If you do not specify the address of an output section in some
3594 other way (other ways are described later), the address is set from the
3595 current value of the location counter. The location counter is then
3596 incremented by the size of the output section. At the start of the
3597 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3599 The second line defines an output section, @samp{.text}. The colon is
3600 required syntax which may be ignored for now. Within the curly braces
3601 after the output section name, you list the names of the input sections
3602 which should be placed into this output section. The @samp{*} is a
3603 wildcard which matches any file name. The expression @samp{*(.text)}
3604 means all @samp{.text} input sections in all input files.
3606 Since the location counter is @samp{0x10000} when the output section
3607 @samp{.text} is defined, the linker will set the address of the
3608 @samp{.text} section in the output file to be @samp{0x10000}.
3610 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3611 the output file. The linker will place the @samp{.data} output section
3612 at address @samp{0x8000000}. After the linker places the @samp{.data}
3613 output section, the value of the location counter will be
3614 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3615 effect is that the linker will place the @samp{.bss} output section
3616 immediately after the @samp{.data} output section in memory.
3618 The linker will ensure that each output section has the required
3619 alignment, by increasing the location counter if necessary. In this
3620 example, the specified addresses for the @samp{.text} and @samp{.data}
3621 sections will probably satisfy any alignment constraints, but the linker
3622 may have to create a small gap between the @samp{.data} and @samp{.bss}
3625 That's it! That's a simple and complete linker script.
3627 @node Simple Commands
3628 @section Simple Linker Script Commands
3629 @cindex linker script simple commands
3630 In this section we describe the simple linker script commands.
3633 * Entry Point:: Setting the entry point
3634 * File Commands:: Commands dealing with files
3635 @ifclear SingleFormat
3636 * Format Commands:: Commands dealing with object file formats
3639 * REGION_ALIAS:: Assign alias names to memory regions
3640 * Miscellaneous Commands:: Other linker script commands
3644 @subsection Setting the Entry Point
3645 @kindex ENTRY(@var{symbol})
3646 @cindex start of execution
3647 @cindex first instruction
3649 The first instruction to execute in a program is called the @dfn{entry
3650 point}. You can use the @code{ENTRY} linker script command to set the
3651 entry point. The argument is a symbol name:
3656 There are several ways to set the entry point. The linker will set the
3657 entry point by trying each of the following methods in order, and
3658 stopping when one of them succeeds:
3661 the @samp{-e} @var{entry} command-line option;
3663 the @code{ENTRY(@var{symbol})} command in a linker script;
3665 the value of a target-specific symbol, if it is defined; For many
3666 targets this is @code{start}, but PE- and BeOS-based systems for example
3667 check a list of possible entry symbols, matching the first one found.
3669 the address of the first byte of the @samp{.text} section, if present;
3671 The address @code{0}.
3675 @subsection Commands Dealing with Files
3676 @cindex linker script file commands
3677 Several linker script commands deal with files.
3680 @item INCLUDE @var{filename}
3681 @kindex INCLUDE @var{filename}
3682 @cindex including a linker script
3683 Include the linker script @var{filename} at this point. The file will
3684 be searched for in the current directory, and in any directory specified
3685 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3688 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3689 @code{SECTIONS} commands, or in output section descriptions.
3691 @item INPUT(@var{file}, @var{file}, @dots{})
3692 @itemx INPUT(@var{file} @var{file} @dots{})
3693 @kindex INPUT(@var{files})
3694 @cindex input files in linker scripts
3695 @cindex input object files in linker scripts
3696 @cindex linker script input object files
3697 The @code{INPUT} command directs the linker to include the named files
3698 in the link, as though they were named on the command line.
3700 For example, if you always want to include @file{subr.o} any time you do
3701 a link, but you can't be bothered to put it on every link command line,
3702 then you can put @samp{INPUT (subr.o)} in your linker script.
3704 In fact, if you like, you can list all of your input files in the linker
3705 script, and then invoke the linker with nothing but a @samp{-T} option.
3707 In case a @dfn{sysroot prefix} is configured, and the filename starts
3708 with the @samp{/} character, and the script being processed was
3709 located inside the @dfn{sysroot prefix}, the filename will be looked
3710 for in the @dfn{sysroot prefix}. The @dfn{sysroot prefix} can also be forced by specifying
3711 @code{=} as the first character in the filename path, or prefixing the
3712 filename path with @code{$SYSROOT}. See also the description of
3713 @samp{-L} in @ref{Options,,Command-line Options}.
3715 If a @dfn{sysroot prefix} is not used then the linker will try to open
3716 the file in the directory containing the linker script. If it is not
3717 found the linker will then search the current directory. If it is still
3718 not found the linker will search through the archive library search
3721 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3722 name to @code{lib@var{file}.a}, as with the command-line argument
3725 When you use the @code{INPUT} command in an implicit linker script, the
3726 files will be included in the link at the point at which the linker
3727 script file is included. This can affect archive searching.
3729 @item GROUP(@var{file}, @var{file}, @dots{})
3730 @itemx GROUP(@var{file} @var{file} @dots{})
3731 @kindex GROUP(@var{files})
3732 @cindex grouping input files
3733 The @code{GROUP} command is like @code{INPUT}, except that the named
3734 files should all be archives, and they are searched repeatedly until no
3735 new undefined references are created. See the description of @samp{-(}
3736 in @ref{Options,,Command-line Options}.
3738 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3739 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3740 @kindex AS_NEEDED(@var{files})
3741 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3742 commands, among other filenames. The files listed will be handled
3743 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3744 with the exception of ELF shared libraries, that will be added only
3745 when they are actually needed. This construct essentially enables
3746 @option{--as-needed} option for all the files listed inside of it
3747 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3750 @item OUTPUT(@var{filename})
3751 @kindex OUTPUT(@var{filename})
3752 @cindex output file name in linker script
3753 The @code{OUTPUT} command names the output file. Using
3754 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3755 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3756 Line Options}). If both are used, the command-line option takes
3759 You can use the @code{OUTPUT} command to define a default name for the
3760 output file other than the usual default of @file{a.out}.
3762 @item SEARCH_DIR(@var{path})
3763 @kindex SEARCH_DIR(@var{path})
3764 @cindex library search path in linker script
3765 @cindex archive search path in linker script
3766 @cindex search path in linker script
3767 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3768 @command{ld} looks for archive libraries. Using
3769 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3770 on the command line (@pxref{Options,,Command-line Options}). If both
3771 are used, then the linker will search both paths. Paths specified using
3772 the command-line option are searched first.
3774 @item STARTUP(@var{filename})
3775 @kindex STARTUP(@var{filename})
3776 @cindex first input file
3777 The @code{STARTUP} command is just like the @code{INPUT} command, except
3778 that @var{filename} will become the first input file to be linked, as
3779 though it were specified first on the command line. This may be useful
3780 when using a system in which the entry point is always the start of the
3784 @ifclear SingleFormat
3785 @node Format Commands
3786 @subsection Commands Dealing with Object File Formats
3787 A couple of linker script commands deal with object file formats.
3790 @item OUTPUT_FORMAT(@var{bfdname})
3791 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3792 @kindex OUTPUT_FORMAT(@var{bfdname})
3793 @cindex output file format in linker script
3794 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3795 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3796 exactly like using @samp{--oformat @var{bfdname}} on the command line
3797 (@pxref{Options,,Command-line Options}). If both are used, the command
3798 line option takes precedence.
3800 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3801 formats based on the @samp{-EB} and @samp{-EL} command-line options.
3802 This permits the linker script to set the output format based on the
3805 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3806 will be the first argument, @var{default}. If @samp{-EB} is used, the
3807 output format will be the second argument, @var{big}. If @samp{-EL} is
3808 used, the output format will be the third argument, @var{little}.
3810 For example, the default linker script for the MIPS ELF target uses this
3813 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3815 This says that the default format for the output file is
3816 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command-line
3817 option, the output file will be created in the @samp{elf32-littlemips}
3820 @item TARGET(@var{bfdname})
3821 @kindex TARGET(@var{bfdname})
3822 @cindex input file format in linker script
3823 The @code{TARGET} command names the BFD format to use when reading input
3824 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3825 This command is like using @samp{-b @var{bfdname}} on the command line
3826 (@pxref{Options,,Command-line Options}). If the @code{TARGET} command
3827 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3828 command is also used to set the format for the output file. @xref{BFD}.
3833 @subsection Assign alias names to memory regions
3834 @kindex REGION_ALIAS(@var{alias}, @var{region})
3835 @cindex region alias
3836 @cindex region names
3838 Alias names can be added to existing memory regions created with the
3839 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3842 REGION_ALIAS(@var{alias}, @var{region})
3845 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3846 memory region @var{region}. This allows a flexible mapping of output sections
3847 to memory regions. An example follows.
3849 Suppose we have an application for embedded systems which come with various
3850 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3851 that allows code execution or data storage. Some may have a read-only,
3852 non-volatile memory @code{ROM} that allows code execution and read-only data
3853 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3854 read-only data access and no code execution capability. We have four output
3859 @code{.text} program code;
3861 @code{.rodata} read-only data;
3863 @code{.data} read-write initialized data;
3865 @code{.bss} read-write zero initialized data.
3868 The goal is to provide a linker command file that contains a system independent
3869 part defining the output sections and a system dependent part mapping the
3870 output sections to the memory regions available on the system. Our embedded
3871 systems come with three different memory setups @code{A}, @code{B} and
3873 @multitable @columnfractions .25 .25 .25 .25
3874 @item Section @tab Variant A @tab Variant B @tab Variant C
3875 @item .text @tab RAM @tab ROM @tab ROM
3876 @item .rodata @tab RAM @tab ROM @tab ROM2
3877 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3878 @item .bss @tab RAM @tab RAM @tab RAM
3880 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3881 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3882 the load address of the @code{.data} section starts in all three variants at
3883 the end of the @code{.rodata} section.
3885 The base linker script that deals with the output sections follows. It
3886 includes the system dependent @code{linkcmds.memory} file that describes the
3889 INCLUDE linkcmds.memory
3902 .data : AT (rodata_end)
3907 data_size = SIZEOF(.data);
3908 data_load_start = LOADADDR(.data);
3916 Now we need three different @code{linkcmds.memory} files to define memory
3917 regions and alias names. The content of @code{linkcmds.memory} for the three
3918 variants @code{A}, @code{B} and @code{C}:
3921 Here everything goes into the @code{RAM}.
3925 RAM : ORIGIN = 0, LENGTH = 4M
3928 REGION_ALIAS("REGION_TEXT", RAM);
3929 REGION_ALIAS("REGION_RODATA", RAM);
3930 REGION_ALIAS("REGION_DATA", RAM);
3931 REGION_ALIAS("REGION_BSS", RAM);
3934 Program code and read-only data go into the @code{ROM}. Read-write data goes
3935 into the @code{RAM}. An image of the initialized data is loaded into the
3936 @code{ROM} and will be copied during system start into the @code{RAM}.
3940 ROM : ORIGIN = 0, LENGTH = 3M
3941 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3944 REGION_ALIAS("REGION_TEXT", ROM);
3945 REGION_ALIAS("REGION_RODATA", ROM);
3946 REGION_ALIAS("REGION_DATA", RAM);
3947 REGION_ALIAS("REGION_BSS", RAM);
3950 Program code goes into the @code{ROM}. Read-only data goes into the
3951 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3952 initialized data is loaded into the @code{ROM2} and will be copied during
3953 system start into the @code{RAM}.
3957 ROM : ORIGIN = 0, LENGTH = 2M
3958 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3959 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3962 REGION_ALIAS("REGION_TEXT", ROM);
3963 REGION_ALIAS("REGION_RODATA", ROM2);
3964 REGION_ALIAS("REGION_DATA", RAM);
3965 REGION_ALIAS("REGION_BSS", RAM);
3969 It is possible to write a common system initialization routine to copy the
3970 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3975 extern char data_start [];
3976 extern char data_size [];
3977 extern char data_load_start [];
3979 void copy_data(void)
3981 if (data_start != data_load_start)
3983 memcpy(data_start, data_load_start, (size_t) data_size);
3988 @node Miscellaneous Commands
3989 @subsection Other Linker Script Commands
3990 There are a few other linker scripts commands.
3993 @item ASSERT(@var{exp}, @var{message})
3995 @cindex assertion in linker script
3996 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3997 with an error code, and print @var{message}.
3999 Note that assertions are checked before the final stages of linking
4000 take place. This means that expressions involving symbols PROVIDEd
4001 inside section definitions will fail if the user has not set values
4002 for those symbols. The only exception to this rule is PROVIDEd
4003 symbols that just reference dot. Thus an assertion like this:
4008 PROVIDE (__stack = .);
4009 PROVIDE (__stack_size = 0x100);
4010 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
4014 will fail if @code{__stack_size} is not defined elsewhere. Symbols
4015 PROVIDEd outside of section definitions are evaluated earlier, so they
4016 can be used inside ASSERTions. Thus:
4019 PROVIDE (__stack_size = 0x100);
4022 PROVIDE (__stack = .);
4023 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
4029 @item EXTERN(@var{symbol} @var{symbol} @dots{})
4031 @cindex undefined symbol in linker script
4032 Force @var{symbol} to be entered in the output file as an undefined
4033 symbol. Doing this may, for example, trigger linking of additional
4034 modules from standard libraries. You may list several @var{symbol}s for
4035 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
4036 command has the same effect as the @samp{-u} command-line option.
4038 @item FORCE_COMMON_ALLOCATION
4039 @kindex FORCE_COMMON_ALLOCATION
4040 @cindex common allocation in linker script
4041 This command has the same effect as the @samp{-d} command-line option:
4042 to make @command{ld} assign space to common symbols even if a relocatable
4043 output file is specified (@samp{-r}).
4045 @item INHIBIT_COMMON_ALLOCATION
4046 @kindex INHIBIT_COMMON_ALLOCATION
4047 @cindex common allocation in linker script
4048 This command has the same effect as the @samp{--no-define-common}
4049 command-line option: to make @code{ld} omit the assignment of addresses
4050 to common symbols even for a non-relocatable output file.
4052 @item FORCE_GROUP_ALLOCATION
4053 @kindex FORCE_GROUP_ALLOCATION
4054 @cindex group allocation in linker script
4055 @cindex section groups
4057 This command has the same effect as the
4058 @samp{--force-group-allocation} command-line option: to make
4059 @command{ld} place section group members like normal input sections,
4060 and to delete the section groups even if a relocatable output file is
4061 specified (@samp{-r}).
4063 @item INSERT [ AFTER | BEFORE ] @var{output_section}
4065 @cindex insert user script into default script
4066 This command is typically used in a script specified by @samp{-T} to
4067 augment the default @code{SECTIONS} with, for example, overlays. It
4068 inserts all prior linker script statements after (or before)
4069 @var{output_section}, and also causes @samp{-T} to not override the
4070 default linker script. The exact insertion point is as for orphan
4071 sections. @xref{Location Counter}. The insertion happens after the
4072 linker has mapped input sections to output sections. Prior to the
4073 insertion, since @samp{-T} scripts are parsed before the default
4074 linker script, statements in the @samp{-T} script occur before the
4075 default linker script statements in the internal linker representation
4076 of the script. In particular, input section assignments will be made
4077 to @samp{-T} output sections before those in the default script. Here
4078 is an example of how a @samp{-T} script using @code{INSERT} might look:
4085 .ov1 @{ ov1*(.text) @}
4086 .ov2 @{ ov2*(.text) @}
4092 @item NOCROSSREFS(@var{section} @var{section} @dots{})
4093 @kindex NOCROSSREFS(@var{sections})
4094 @cindex cross references
4095 This command may be used to tell @command{ld} to issue an error about any
4096 references among certain output sections.
4098 In certain types of programs, particularly on embedded systems when
4099 using overlays, when one section is loaded into memory, another section
4100 will not be. Any direct references between the two sections would be
4101 errors. For example, it would be an error if code in one section called
4102 a function defined in the other section.
4104 The @code{NOCROSSREFS} command takes a list of output section names. If
4105 @command{ld} detects any cross references between the sections, it reports
4106 an error and returns a non-zero exit status. Note that the
4107 @code{NOCROSSREFS} command uses output section names, not input section
4110 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
4111 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
4112 @cindex cross references
4113 This command may be used to tell @command{ld} to issue an error about any
4114 references to one section from a list of other sections.
4116 The @code{NOCROSSREFS} command is useful when ensuring that two or more
4117 output sections are entirely independent but there are situations where
4118 a one-way dependency is needed. For example, in a multi-core application
4119 there may be shared code that can be called from each core but for safety
4120 must never call back.
4122 The @code{NOCROSSREFS_TO} command takes a list of output section names.
4123 The first section can not be referenced from any of the other sections.
4124 If @command{ld} detects any references to the first section from any of
4125 the other sections, it reports an error and returns a non-zero exit
4126 status. Note that the @code{NOCROSSREFS_TO} command uses output section
4127 names, not input section names.
4129 @ifclear SingleFormat
4130 @item OUTPUT_ARCH(@var{bfdarch})
4131 @kindex OUTPUT_ARCH(@var{bfdarch})
4132 @cindex machine architecture
4133 @cindex architecture
4134 Specify a particular output machine architecture. The argument is one
4135 of the names used by the BFD library (@pxref{BFD}). You can see the
4136 architecture of an object file by using the @code{objdump} program with
4137 the @samp{-f} option.
4140 @item LD_FEATURE(@var{string})
4141 @kindex LD_FEATURE(@var{string})
4142 This command may be used to modify @command{ld} behavior. If
4143 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
4144 in a script are simply treated as numbers everywhere.
4145 @xref{Expression Section}.
4149 @section Assigning Values to Symbols
4150 @cindex assignment in scripts
4151 @cindex symbol definition, scripts
4152 @cindex variables, defining
4153 You may assign a value to a symbol in a linker script. This will define
4154 the symbol and place it into the symbol table with a global scope.
4157 * Simple Assignments:: Simple Assignments
4160 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
4161 * Source Code Reference:: How to use a linker script defined symbol in source code
4164 @node Simple Assignments
4165 @subsection Simple Assignments
4167 You may assign to a symbol using any of the C assignment operators:
4170 @item @var{symbol} = @var{expression} ;
4171 @itemx @var{symbol} += @var{expression} ;
4172 @itemx @var{symbol} -= @var{expression} ;
4173 @itemx @var{symbol} *= @var{expression} ;
4174 @itemx @var{symbol} /= @var{expression} ;
4175 @itemx @var{symbol} <<= @var{expression} ;
4176 @itemx @var{symbol} >>= @var{expression} ;
4177 @itemx @var{symbol} &= @var{expression} ;
4178 @itemx @var{symbol} |= @var{expression} ;
4181 The first case will define @var{symbol} to the value of
4182 @var{expression}. In the other cases, @var{symbol} must already be
4183 defined, and the value will be adjusted accordingly.
4185 The special symbol name @samp{.} indicates the location counter. You
4186 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
4188 The semicolon after @var{expression} is required.
4190 Expressions are defined below; see @ref{Expressions}.
4192 You may write symbol assignments as commands in their own right, or as
4193 statements within a @code{SECTIONS} command, or as part of an output
4194 section description in a @code{SECTIONS} command.
4196 The section of the symbol will be set from the section of the
4197 expression; for more information, see @ref{Expression Section}.
4199 Here is an example showing the three different places that symbol
4200 assignments may be used:
4211 _bdata = (. + 3) & ~ 3;
4212 .data : @{ *(.data) @}
4216 In this example, the symbol @samp{floating_point} will be defined as
4217 zero. The symbol @samp{_etext} will be defined as the address following
4218 the last @samp{.text} input section. The symbol @samp{_bdata} will be
4219 defined as the address following the @samp{.text} output section aligned
4220 upward to a 4 byte boundary.
4225 For ELF targeted ports, define a symbol that will be hidden and won't be
4226 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
4228 Here is the example from @ref{Simple Assignments}, rewritten to use
4232 HIDDEN(floating_point = 0);
4240 HIDDEN(_bdata = (. + 3) & ~ 3);
4241 .data : @{ *(.data) @}
4245 In this case none of the three symbols will be visible outside this module.
4250 In some cases, it is desirable for a linker script to define a symbol
4251 only if it is referenced and is not defined by any object included in
4252 the link. For example, traditional linkers defined the symbol
4253 @samp{etext}. However, ANSI C requires that the user be able to use
4254 @samp{etext} as a function name without encountering an error. The
4255 @code{PROVIDE} keyword may be used to define a symbol, such as
4256 @samp{etext}, only if it is referenced but not defined. The syntax is
4257 @code{PROVIDE(@var{symbol} = @var{expression})}.
4259 Here is an example of using @code{PROVIDE} to define @samp{etext}:
4272 In this example, if the program defines @samp{_etext} (with a leading
4273 underscore), the linker will give a multiple definition error. If, on
4274 the other hand, the program defines @samp{etext} (with no leading
4275 underscore), the linker will silently use the definition in the program.
4276 If the program references @samp{etext} but does not define it, the
4277 linker will use the definition in the linker script.
4279 Note - the @code{PROVIDE} directive considers a common symbol to be
4280 defined, even though such a symbol could be combined with the symbol
4281 that the @code{PROVIDE} would create. This is particularly important
4282 when considering constructor and destructor list symbols such as
4283 @samp{__CTOR_LIST__} as these are often defined as common symbols.
4285 @node PROVIDE_HIDDEN
4286 @subsection PROVIDE_HIDDEN
4287 @cindex PROVIDE_HIDDEN
4288 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
4289 hidden and won't be exported.
4291 @node Source Code Reference
4292 @subsection Source Code Reference
4294 Accessing a linker script defined variable from source code is not
4295 intuitive. In particular a linker script symbol is not equivalent to
4296 a variable declaration in a high level language, it is instead a
4297 symbol that does not have a value.
4299 Before going further, it is important to note that compilers often
4300 transform names in the source code into different names when they are
4301 stored in the symbol table. For example, Fortran compilers commonly
4302 prepend or append an underscore, and C++ performs extensive @samp{name
4303 mangling}. Therefore there might be a discrepancy between the name
4304 of a variable as it is used in source code and the name of the same
4305 variable as it is defined in a linker script. For example in C a
4306 linker script variable might be referred to as:
4312 But in the linker script it might be defined as:
4318 In the remaining examples however it is assumed that no name
4319 transformation has taken place.
4321 When a symbol is declared in a high level language such as C, two
4322 things happen. The first is that the compiler reserves enough space
4323 in the program's memory to hold the @emph{value} of the symbol. The
4324 second is that the compiler creates an entry in the program's symbol
4325 table which holds the symbol's @emph{address}. ie the symbol table
4326 contains the address of the block of memory holding the symbol's
4327 value. So for example the following C declaration, at file scope:
4333 creates an entry called @samp{foo} in the symbol table. This entry
4334 holds the address of an @samp{int} sized block of memory where the
4335 number 1000 is initially stored.
4337 When a program references a symbol the compiler generates code that
4338 first accesses the symbol table to find the address of the symbol's
4339 memory block and then code to read the value from that memory block.
4346 looks up the symbol @samp{foo} in the symbol table, gets the address
4347 associated with this symbol and then writes the value 1 into that
4354 looks up the symbol @samp{foo} in the symbol table, gets its address
4355 and then copies this address into the block of memory associated with
4356 the variable @samp{a}.
4358 Linker scripts symbol declarations, by contrast, create an entry in
4359 the symbol table but do not assign any memory to them. Thus they are
4360 an address without a value. So for example the linker script definition:
4366 creates an entry in the symbol table called @samp{foo} which holds
4367 the address of memory location 1000, but nothing special is stored at
4368 address 1000. This means that you cannot access the @emph{value} of a
4369 linker script defined symbol - it has no value - all you can do is
4370 access the @emph{address} of a linker script defined symbol.
4372 Hence when you are using a linker script defined symbol in source code
4373 you should always take the address of the symbol, and never attempt to
4374 use its value. For example suppose you want to copy the contents of a
4375 section of memory called .ROM into a section called .FLASH and the
4376 linker script contains these declarations:
4380 start_of_ROM = .ROM;
4381 end_of_ROM = .ROM + sizeof (.ROM);
4382 start_of_FLASH = .FLASH;
4386 Then the C source code to perform the copy would be:
4390 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
4392 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
4396 Note the use of the @samp{&} operators. These are correct.
4397 Alternatively the symbols can be treated as the names of vectors or
4398 arrays and then the code will again work as expected:
4402 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
4404 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
4408 Note how using this method does not require the use of @samp{&}
4412 @section SECTIONS Command
4414 The @code{SECTIONS} command tells the linker how to map input sections
4415 into output sections, and how to place the output sections in memory.
4417 The format of the @code{SECTIONS} command is:
4421 @var{sections-command}
4422 @var{sections-command}
4427 Each @var{sections-command} may of be one of the following:
4431 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
4433 a symbol assignment (@pxref{Assignments})
4435 an output section description
4437 an overlay description
4440 The @code{ENTRY} command and symbol assignments are permitted inside the
4441 @code{SECTIONS} command for convenience in using the location counter in
4442 those commands. This can also make the linker script easier to
4443 understand because you can use those commands at meaningful points in
4444 the layout of the output file.
4446 Output section descriptions and overlay descriptions are described
4449 If you do not use a @code{SECTIONS} command in your linker script, the
4450 linker will place each input section into an identically named output
4451 section in the order that the sections are first encountered in the
4452 input files. If all input sections are present in the first file, for
4453 example, the order of sections in the output file will match the order
4454 in the first input file. The first section will be at address zero.
4457 * Output Section Description:: Output section description
4458 * Output Section Name:: Output section name
4459 * Output Section Address:: Output section address
4460 * Input Section:: Input section description
4461 * Output Section Data:: Output section data
4462 * Output Section Keywords:: Output section keywords
4463 * Output Section Discarding:: Output section discarding
4464 * Output Section Attributes:: Output section attributes
4465 * Overlay Description:: Overlay description
4468 @node Output Section Description
4469 @subsection Output Section Description
4470 The full description of an output section looks like this:
4473 @var{section} [@var{address}] [(@var{type})] :
4475 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4476 [SUBALIGN(@var{subsection_align})]
4479 @var{output-section-command}
4480 @var{output-section-command}
4482 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4486 Most output sections do not use most of the optional section attributes.
4488 The whitespace around @var{section} is required, so that the section
4489 name is unambiguous. The colon and the curly braces are also required.
4490 The comma at the end may be required if a @var{fillexp} is used and
4491 the next @var{sections-command} looks like a continuation of the expression.
4492 The line breaks and other white space are optional.
4494 Each @var{output-section-command} may be one of the following:
4498 a symbol assignment (@pxref{Assignments})
4500 an input section description (@pxref{Input Section})
4502 data values to include directly (@pxref{Output Section Data})
4504 a special output section keyword (@pxref{Output Section Keywords})
4507 @node Output Section Name
4508 @subsection Output Section Name
4509 @cindex name, section
4510 @cindex section name
4511 The name of the output section is @var{section}. @var{section} must
4512 meet the constraints of your output format. In formats which only
4513 support a limited number of sections, such as @code{a.out}, the name
4514 must be one of the names supported by the format (@code{a.out}, for
4515 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4516 output format supports any number of sections, but with numbers and not
4517 names (as is the case for Oasys), the name should be supplied as a
4518 quoted numeric string. A section name may consist of any sequence of
4519 characters, but a name which contains any unusual characters such as
4520 commas must be quoted.
4522 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4525 @node Output Section Address
4526 @subsection Output Section Address
4527 @cindex address, section
4528 @cindex section address
4529 The @var{address} is an expression for the VMA (the virtual memory
4530 address) of the output section. This address is optional, but if it
4531 is provided then the output address will be set exactly as specified.
4533 If the output address is not specified then one will be chosen for the
4534 section, based on the heuristic below. This address will be adjusted
4535 to fit the alignment requirement of the output section. The
4536 alignment requirement is the strictest alignment of any input section
4537 contained within the output section.
4539 The output section address heuristic is as follows:
4543 If an output memory @var{region} is set for the section then it
4544 is added to this region and its address will be the next free address
4548 If the MEMORY command has been used to create a list of memory
4549 regions then the first region which has attributes compatible with the
4550 section is selected to contain it. The section's output address will
4551 be the next free address in that region; @ref{MEMORY}.
4554 If no memory regions were specified, or none match the section then
4555 the output address will be based on the current value of the location
4563 .text . : @{ *(.text) @}
4570 .text : @{ *(.text) @}
4574 are subtly different. The first will set the address of the
4575 @samp{.text} output section to the current value of the location
4576 counter. The second will set it to the current value of the location
4577 counter aligned to the strictest alignment of any of the @samp{.text}
4580 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4581 For example, if you want to align the section on a 0x10 byte boundary,
4582 so that the lowest four bits of the section address are zero, you could
4583 do something like this:
4585 .text ALIGN(0x10) : @{ *(.text) @}
4588 This works because @code{ALIGN} returns the current location counter
4589 aligned upward to the specified value.
4591 Specifying @var{address} for a section will change the value of the
4592 location counter, provided that the section is non-empty. (Empty
4593 sections are ignored).
4596 @subsection Input Section Description
4597 @cindex input sections
4598 @cindex mapping input sections to output sections
4599 The most common output section command is an input section description.
4601 The input section description is the most basic linker script operation.
4602 You use output sections to tell the linker how to lay out your program
4603 in memory. You use input section descriptions to tell the linker how to
4604 map the input files into your memory layout.
4607 * Input Section Basics:: Input section basics
4608 * Input Section Wildcards:: Input section wildcard patterns
4609 * Input Section Common:: Input section for common symbols
4610 * Input Section Keep:: Input section and garbage collection
4611 * Input Section Example:: Input section example
4614 @node Input Section Basics
4615 @subsubsection Input Section Basics
4616 @cindex input section basics
4617 An input section description consists of a file name optionally followed
4618 by a list of section names in parentheses.
4620 The file name and the section name may be wildcard patterns, which we
4621 describe further below (@pxref{Input Section Wildcards}).
4623 The most common input section description is to include all input
4624 sections with a particular name in the output section. For example, to
4625 include all input @samp{.text} sections, you would write:
4630 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4631 @cindex EXCLUDE_FILE
4632 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4633 match all files except the ones specified in the EXCLUDE_FILE list. For
4636 EXCLUDE_FILE (*crtend.o *otherfile.o) *(.ctors)
4639 will cause all .ctors sections from all files except @file{crtend.o}
4640 and @file{otherfile.o} to be included. The EXCLUDE_FILE can also be
4641 placed inside the section list, for example:
4643 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4646 The result of this is identically to the previous example. Supporting
4647 two syntaxes for EXCLUDE_FILE is useful if the section list contains
4648 more than one section, as described below.
4650 There are two ways to include more than one section:
4656 The difference between these is the order in which the @samp{.text} and
4657 @samp{.rdata} input sections will appear in the output section. In the
4658 first example, they will be intermingled, appearing in the same order as
4659 they are found in the linker input. In the second example, all
4660 @samp{.text} input sections will appear first, followed by all
4661 @samp{.rdata} input sections.
4663 When using EXCLUDE_FILE with more than one section, if the exclusion
4664 is within the section list then the exclusion only applies to the
4665 immediately following section, for example:
4667 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4670 will cause all @samp{.text} sections from all files except
4671 @file{somefile.o} to be included, while all @samp{.rdata} sections
4672 from all files, including @file{somefile.o}, will be included. To
4673 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4674 could be modified to:
4676 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4679 Alternatively, placing the EXCLUDE_FILE outside of the section list,
4680 before the input file selection, will cause the exclusion to apply for
4681 all sections. Thus the previous example can be rewritten as:
4683 EXCLUDE_FILE (*somefile.o) *(.text .rdata)
4686 You can specify a file name to include sections from a particular file.
4687 You would do this if one or more of your files contain special data that
4688 needs to be at a particular location in memory. For example:
4693 To refine the sections that are included based on the section flags
4694 of an input section, INPUT_SECTION_FLAGS may be used.
4696 Here is a simple example for using Section header flags for ELF sections:
4701 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4702 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4707 In this example, the output section @samp{.text} will be comprised of any
4708 input section matching the name *(.text) whose section header flags
4709 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4710 @samp{.text2} will be comprised of any input section matching the name *(.text)
4711 whose section header flag @code{SHF_WRITE} is clear.
4713 You can also specify files within archives by writing a pattern
4714 matching the archive, a colon, then the pattern matching the file,
4715 with no whitespace around the colon.
4719 matches file within archive
4721 matches the whole archive
4723 matches file but not one in an archive
4726 Either one or both of @samp{archive} and @samp{file} can contain shell
4727 wildcards. On DOS based file systems, the linker will assume that a
4728 single letter followed by a colon is a drive specifier, so
4729 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4730 within an archive called @samp{c}. @samp{archive:file} filespecs may
4731 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4732 other linker script contexts. For instance, you cannot extract a file
4733 from an archive by using @samp{archive:file} in an @code{INPUT}
4736 If you use a file name without a list of sections, then all sections in
4737 the input file will be included in the output section. This is not
4738 commonly done, but it may by useful on occasion. For example:
4743 When you use a file name which is not an @samp{archive:file} specifier
4744 and does not contain any wild card
4745 characters, the linker will first see if you also specified the file
4746 name on the linker command line or in an @code{INPUT} command. If you
4747 did not, the linker will attempt to open the file as an input file, as
4748 though it appeared on the command line. Note that this differs from an
4749 @code{INPUT} command, because the linker will not search for the file in
4750 the archive search path.
4752 @node Input Section Wildcards
4753 @subsubsection Input Section Wildcard Patterns
4754 @cindex input section wildcards
4755 @cindex wildcard file name patterns
4756 @cindex file name wildcard patterns
4757 @cindex section name wildcard patterns
4758 In an input section description, either the file name or the section
4759 name or both may be wildcard patterns.
4761 The file name of @samp{*} seen in many examples is a simple wildcard
4762 pattern for the file name.
4764 The wildcard patterns are like those used by the Unix shell.
4768 matches any number of characters
4770 matches any single character
4772 matches a single instance of any of the @var{chars}; the @samp{-}
4773 character may be used to specify a range of characters, as in
4774 @samp{[a-z]} to match any lower case letter
4776 quotes the following character
4779 When a file name is matched with a wildcard, the wildcard characters
4780 will not match a @samp{/} character (used to separate directory names on
4781 Unix). A pattern consisting of a single @samp{*} character is an
4782 exception; it will always match any file name, whether it contains a
4783 @samp{/} or not. In a section name, the wildcard characters will match
4784 a @samp{/} character.
4786 File name wildcard patterns only match files which are explicitly
4787 specified on the command line or in an @code{INPUT} command. The linker
4788 does not search directories to expand wildcards.
4790 If a file name matches more than one wildcard pattern, or if a file name
4791 appears explicitly and is also matched by a wildcard pattern, the linker
4792 will use the first match in the linker script. For example, this
4793 sequence of input section descriptions is probably in error, because the
4794 @file{data.o} rule will not be used:
4796 .data : @{ *(.data) @}
4797 .data1 : @{ data.o(.data) @}
4800 @cindex SORT_BY_NAME
4801 Normally, the linker will place files and sections matched by wildcards
4802 in the order in which they are seen during the link. You can change
4803 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4804 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4805 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4806 into ascending order by name before placing them in the output file.
4808 @cindex SORT_BY_ALIGNMENT
4809 @code{SORT_BY_ALIGNMENT} is similar to @code{SORT_BY_NAME}.
4810 @code{SORT_BY_ALIGNMENT} will sort sections into descending order of
4811 alignment before placing them in the output file. Placing larger
4812 alignments before smaller alignments can reduce the amount of padding
4815 @cindex SORT_BY_INIT_PRIORITY
4816 @code{SORT_BY_INIT_PRIORITY} is also similar to @code{SORT_BY_NAME}.
4817 @code{SORT_BY_INIT_PRIORITY} will sort sections into ascending
4818 numerical order of the GCC init_priority attribute encoded in the
4819 section name before placing them in the output file. In
4820 @code{.init_array.NNNNN} and @code{.fini_array.NNNNN}, @code{NNNNN} is
4821 the init_priority. In @code{.ctors.NNNNN} and @code{.dtors.NNNNN},
4822 @code{NNNNN} is 65535 minus the init_priority.
4825 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4827 When there are nested section sorting commands in linker script, there
4828 can be at most 1 level of nesting for section sorting commands.
4832 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4833 It will sort the input sections by name first, then by alignment if two
4834 sections have the same name.
4836 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4837 It will sort the input sections by alignment first, then by name if two
4838 sections have the same alignment.
4840 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4841 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4843 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4844 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4846 All other nested section sorting commands are invalid.
4849 When both command-line section sorting option and linker script
4850 section sorting command are used, section sorting command always
4851 takes precedence over the command-line option.
4853 If the section sorting command in linker script isn't nested, the
4854 command-line option will make the section sorting command to be
4855 treated as nested sorting command.
4859 @code{SORT_BY_NAME} (wildcard section pattern ) with
4860 @option{--sort-sections alignment} is equivalent to
4861 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4863 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4864 @option{--sort-section name} is equivalent to
4865 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4868 If the section sorting command in linker script is nested, the
4869 command-line option will be ignored.
4872 @code{SORT_NONE} disables section sorting by ignoring the command-line
4873 section sorting option.
4875 If you ever get confused about where input sections are going, use the
4876 @samp{-M} linker option to generate a map file. The map file shows
4877 precisely how input sections are mapped to output sections.
4879 This example shows how wildcard patterns might be used to partition
4880 files. This linker script directs the linker to place all @samp{.text}
4881 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4882 The linker will place the @samp{.data} section from all files beginning
4883 with an upper case character in @samp{.DATA}; for all other files, the
4884 linker will place the @samp{.data} section in @samp{.data}.
4888 .text : @{ *(.text) @}
4889 .DATA : @{ [A-Z]*(.data) @}
4890 .data : @{ *(.data) @}
4891 .bss : @{ *(.bss) @}
4896 @node Input Section Common
4897 @subsubsection Input Section for Common Symbols
4898 @cindex common symbol placement
4899 @cindex uninitialized data placement
4900 A special notation is needed for common symbols, because in many object
4901 file formats common symbols do not have a particular input section. The
4902 linker treats common symbols as though they are in an input section
4903 named @samp{COMMON}.
4905 You may use file names with the @samp{COMMON} section just as with any
4906 other input sections. You can use this to place common symbols from a
4907 particular input file in one section while common symbols from other
4908 input files are placed in another section.
4910 In most cases, common symbols in input files will be placed in the
4911 @samp{.bss} section in the output file. For example:
4913 .bss @{ *(.bss) *(COMMON) @}
4916 @cindex scommon section
4917 @cindex small common symbols
4918 Some object file formats have more than one type of common symbol. For
4919 example, the MIPS ELF object file format distinguishes standard common
4920 symbols and small common symbols. In this case, the linker will use a
4921 different special section name for other types of common symbols. In
4922 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4923 symbols and @samp{.scommon} for small common symbols. This permits you
4924 to map the different types of common symbols into memory at different
4928 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4929 notation is now considered obsolete. It is equivalent to
4932 @node Input Section Keep
4933 @subsubsection Input Section and Garbage Collection
4935 @cindex garbage collection
4936 When link-time garbage collection is in use (@samp{--gc-sections}),
4937 it is often useful to mark sections that should not be eliminated.
4938 This is accomplished by surrounding an input section's wildcard entry
4939 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4940 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4942 @node Input Section Example
4943 @subsubsection Input Section Example
4944 The following example is a complete linker script. It tells the linker
4945 to read all of the sections from file @file{all.o} and place them at the
4946 start of output section @samp{outputa} which starts at location
4947 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4948 follows immediately, in the same output section. All of section
4949 @samp{.input2} from @file{foo.o} goes into output section
4950 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4951 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4952 files are written to output section @samp{outputc}.
4980 If an output section's name is the same as the input section's name
4981 and is representable as a C identifier, then the linker will
4982 automatically @pxref{PROVIDE} two symbols: __start_SECNAME and
4983 __stop_SECNAME, where SECNAME is the name of the section. These
4984 indicate the start address and end address of the output section
4985 respectively. Note: most section names are not representable as
4986 C identifiers because they contain a @samp{.} character.
4988 @node Output Section Data
4989 @subsection Output Section Data
4991 @cindex section data
4992 @cindex output section data
4993 @kindex BYTE(@var{expression})
4994 @kindex SHORT(@var{expression})
4995 @kindex LONG(@var{expression})
4996 @kindex QUAD(@var{expression})
4997 @kindex SQUAD(@var{expression})
4998 You can include explicit bytes of data in an output section by using
4999 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
5000 an output section command. Each keyword is followed by an expression in
5001 parentheses providing the value to store (@pxref{Expressions}). The
5002 value of the expression is stored at the current value of the location
5005 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
5006 store one, two, four, and eight bytes (respectively). After storing the
5007 bytes, the location counter is incremented by the number of bytes
5010 For example, this will store the byte 1 followed by the four byte value
5011 of the symbol @samp{addr}:
5017 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
5018 same; they both store an 8 byte, or 64 bit, value. When both host and
5019 target are 32 bits, an expression is computed as 32 bits. In this case
5020 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
5021 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
5023 If the object file format of the output file has an explicit endianness,
5024 which is the normal case, the value will be stored in that endianness.
5025 When the object file format does not have an explicit endianness, as is
5026 true of, for example, S-records, the value will be stored in the
5027 endianness of the first input object file.
5029 Note---these commands only work inside a section description and not
5030 between them, so the following will produce an error from the linker:
5032 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
5034 whereas this will work:
5036 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
5039 @kindex FILL(@var{expression})
5040 @cindex holes, filling
5041 @cindex unspecified memory
5042 You may use the @code{FILL} command to set the fill pattern for the
5043 current section. It is followed by an expression in parentheses. Any
5044 otherwise unspecified regions of memory within the section (for example,
5045 gaps left due to the required alignment of input sections) are filled
5046 with the value of the expression, repeated as
5047 necessary. A @code{FILL} statement covers memory locations after the
5048 point at which it occurs in the section definition; by including more
5049 than one @code{FILL} statement, you can have different fill patterns in
5050 different parts of an output section.
5052 This example shows how to fill unspecified regions of memory with the
5058 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
5059 section attribute, but it only affects the
5060 part of the section following the @code{FILL} command, rather than the
5061 entire section. If both are used, the @code{FILL} command takes
5062 precedence. @xref{Output Section Fill}, for details on the fill
5065 @node Output Section Keywords
5066 @subsection Output Section Keywords
5067 There are a couple of keywords which can appear as output section
5071 @kindex CREATE_OBJECT_SYMBOLS
5072 @cindex input filename symbols
5073 @cindex filename symbols
5074 @item CREATE_OBJECT_SYMBOLS
5075 The command tells the linker to create a symbol for each input file.
5076 The name of each symbol will be the name of the corresponding input
5077 file. The section of each symbol will be the output section in which
5078 the @code{CREATE_OBJECT_SYMBOLS} command appears.
5080 This is conventional for the a.out object file format. It is not
5081 normally used for any other object file format.
5083 @kindex CONSTRUCTORS
5084 @cindex C++ constructors, arranging in link
5085 @cindex constructors, arranging in link
5087 When linking using the a.out object file format, the linker uses an
5088 unusual set construct to support C++ global constructors and
5089 destructors. When linking object file formats which do not support
5090 arbitrary sections, such as ECOFF and XCOFF, the linker will
5091 automatically recognize C++ global constructors and destructors by name.
5092 For these object file formats, the @code{CONSTRUCTORS} command tells the
5093 linker to place constructor information in the output section where the
5094 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
5095 ignored for other object file formats.
5097 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
5098 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
5099 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
5100 the start and end of the global destructors. The
5101 first word in the list is the number of entries, followed by the address
5102 of each constructor or destructor, followed by a zero word. The
5103 compiler must arrange to actually run the code. For these object file
5104 formats @sc{gnu} C++ normally calls constructors from a subroutine
5105 @code{__main}; a call to @code{__main} is automatically inserted into
5106 the startup code for @code{main}. @sc{gnu} C++ normally runs
5107 destructors either by using @code{atexit}, or directly from the function
5110 For object file formats such as @code{COFF} or @code{ELF} which support
5111 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
5112 addresses of global constructors and destructors into the @code{.ctors}
5113 and @code{.dtors} sections. Placing the following sequence into your
5114 linker script will build the sort of table which the @sc{gnu} C++
5115 runtime code expects to see.
5119 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
5124 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
5130 If you are using the @sc{gnu} C++ support for initialization priority,
5131 which provides some control over the order in which global constructors
5132 are run, you must sort the constructors at link time to ensure that they
5133 are executed in the correct order. When using the @code{CONSTRUCTORS}
5134 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
5135 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
5136 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
5139 Normally the compiler and linker will handle these issues automatically,
5140 and you will not need to concern yourself with them. However, you may
5141 need to consider this if you are using C++ and writing your own linker
5146 @node Output Section Discarding
5147 @subsection Output Section Discarding
5148 @cindex discarding sections
5149 @cindex sections, discarding
5150 @cindex removing sections
5151 The linker will not normally create output sections with no contents.
5152 This is for convenience when referring to input sections that may or
5153 may not be present in any of the input files. For example:
5155 .foo : @{ *(.foo) @}
5158 will only create a @samp{.foo} section in the output file if there is a
5159 @samp{.foo} section in at least one input file, and if the input
5160 sections are not all empty. Other link script directives that allocate
5161 space in an output section will also create the output section. So
5162 too will assignments to dot even if the assignment does not create
5163 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
5164 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
5165 @samp{sym} is an absolute symbol of value 0 defined in the script.
5166 This allows you to force output of an empty section with @samp{. = .}.
5168 The linker will ignore address assignments (@pxref{Output Section Address})
5169 on discarded output sections, except when the linker script defines
5170 symbols in the output section. In that case the linker will obey
5171 the address assignments, possibly advancing dot even though the
5172 section is discarded.
5175 The special output section name @samp{/DISCARD/} may be used to discard
5176 input sections. Any input sections which are assigned to an output
5177 section named @samp{/DISCARD/} are not included in the output file.
5179 Note, sections that match the @samp{/DISCARD/} output section will be
5180 discarded even if they are in an ELF section group which has other
5181 members which are not being discarded. This is deliberate.
5182 Discarding takes precedence over grouping.
5184 @node Output Section Attributes
5185 @subsection Output Section Attributes
5186 @cindex output section attributes
5187 We showed above that the full description of an output section looked
5192 @var{section} [@var{address}] [(@var{type})] :
5194 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
5195 [SUBALIGN(@var{subsection_align})]
5198 @var{output-section-command}
5199 @var{output-section-command}
5201 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
5205 We've already described @var{section}, @var{address}, and
5206 @var{output-section-command}. In this section we will describe the
5207 remaining section attributes.
5210 * Output Section Type:: Output section type
5211 * Output Section LMA:: Output section LMA
5212 * Forced Output Alignment:: Forced Output Alignment
5213 * Forced Input Alignment:: Forced Input Alignment
5214 * Output Section Constraint:: Output section constraint
5215 * Output Section Region:: Output section region
5216 * Output Section Phdr:: Output section phdr
5217 * Output Section Fill:: Output section fill
5220 @node Output Section Type
5221 @subsubsection Output Section Type
5222 Each output section may have a type. The type is a keyword in
5223 parentheses. The following types are defined:
5227 The section should be marked as not loadable, so that it will not be
5228 loaded into memory when the program is run.
5233 These type names are supported for backward compatibility, and are
5234 rarely used. They all have the same effect: the section should be
5235 marked as not allocatable, so that no memory is allocated for the
5236 section when the program is run.
5240 @cindex prevent unnecessary loading
5241 @cindex loading, preventing
5242 The linker normally sets the attributes of an output section based on
5243 the input sections which map into it. You can override this by using
5244 the section type. For example, in the script sample below, the
5245 @samp{ROM} section is addressed at memory location @samp{0} and does not
5246 need to be loaded when the program is run.
5250 ROM 0 (NOLOAD) : @{ @dots{} @}
5256 @node Output Section LMA
5257 @subsubsection Output Section LMA
5258 @kindex AT>@var{lma_region}
5259 @kindex AT(@var{lma})
5260 @cindex load address
5261 @cindex section load address
5262 Every section has a virtual address (VMA) and a load address (LMA); see
5263 @ref{Basic Script Concepts}. The virtual address is specified by the
5264 @pxref{Output Section Address} described earlier. The load address is
5265 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
5266 address is optional.
5268 The @code{AT} keyword takes an expression as an argument. This
5269 specifies the exact load address of the section. The @code{AT>} keyword
5270 takes the name of a memory region as an argument. @xref{MEMORY}. The
5271 load address of the section is set to the next free address in the
5272 region, aligned to the section's alignment requirements.
5274 If neither @code{AT} nor @code{AT>} is specified for an allocatable
5275 section, the linker will use the following heuristic to determine the
5280 If the section has a specific VMA address, then this is used as
5281 the LMA address as well.
5284 If the section is not allocatable then its LMA is set to its VMA.
5287 Otherwise if a memory region can be found that is compatible
5288 with the current section, and this region contains at least one
5289 section, then the LMA is set so the difference between the
5290 VMA and LMA is the same as the difference between the VMA and LMA of
5291 the last section in the located region.
5294 If no memory regions have been declared then a default region
5295 that covers the entire address space is used in the previous step.
5298 If no suitable region could be found, or there was no previous
5299 section then the LMA is set equal to the VMA.
5302 @cindex ROM initialized data
5303 @cindex initialized data in ROM
5304 This feature is designed to make it easy to build a ROM image. For
5305 example, the following linker script creates three output sections: one
5306 called @samp{.text}, which starts at @code{0x1000}, one called
5307 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
5308 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
5309 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
5310 defined with the value @code{0x2000}, which shows that the location
5311 counter holds the VMA value, not the LMA value.
5317 .text 0x1000 : @{ *(.text) _etext = . ; @}
5319 AT ( ADDR (.text) + SIZEOF (.text) )
5320 @{ _data = . ; *(.data); _edata = . ; @}
5322 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
5327 The run-time initialization code for use with a program generated with
5328 this linker script would include something like the following, to copy
5329 the initialized data from the ROM image to its runtime address. Notice
5330 how this code takes advantage of the symbols defined by the linker
5335 extern char _etext, _data, _edata, _bstart, _bend;
5336 char *src = &_etext;
5339 /* ROM has data at end of text; copy it. */
5340 while (dst < &_edata)
5344 for (dst = &_bstart; dst< &_bend; dst++)
5349 @node Forced Output Alignment
5350 @subsubsection Forced Output Alignment
5351 @kindex ALIGN(@var{section_align})
5352 @cindex forcing output section alignment
5353 @cindex output section alignment
5354 You can increase an output section's alignment by using ALIGN. As an
5355 alternative you can enforce that the difference between the VMA and LMA remains
5356 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
5358 @node Forced Input Alignment
5359 @subsubsection Forced Input Alignment
5360 @kindex SUBALIGN(@var{subsection_align})
5361 @cindex forcing input section alignment
5362 @cindex input section alignment
5363 You can force input section alignment within an output section by using
5364 SUBALIGN. The value specified overrides any alignment given by input
5365 sections, whether larger or smaller.
5367 @node Output Section Constraint
5368 @subsubsection Output Section Constraint
5371 @cindex constraints on output sections
5372 You can specify that an output section should only be created if all
5373 of its input sections are read-only or all of its input sections are
5374 read-write by using the keyword @code{ONLY_IF_RO} and
5375 @code{ONLY_IF_RW} respectively.
5377 @node Output Section Region
5378 @subsubsection Output Section Region
5379 @kindex >@var{region}
5380 @cindex section, assigning to memory region
5381 @cindex memory regions and sections
5382 You can assign a section to a previously defined region of memory by
5383 using @samp{>@var{region}}. @xref{MEMORY}.
5385 Here is a simple example:
5388 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
5389 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
5393 @node Output Section Phdr
5394 @subsubsection Output Section Phdr
5396 @cindex section, assigning to program header
5397 @cindex program headers and sections
5398 You can assign a section to a previously defined program segment by
5399 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
5400 one or more segments, then all subsequent allocated sections will be
5401 assigned to those segments as well, unless they use an explicitly
5402 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
5403 linker to not put the section in any segment at all.
5405 Here is a simple example:
5408 PHDRS @{ text PT_LOAD ; @}
5409 SECTIONS @{ .text : @{ *(.text) @} :text @}
5413 @node Output Section Fill
5414 @subsubsection Output Section Fill
5415 @kindex =@var{fillexp}
5416 @cindex section fill pattern
5417 @cindex fill pattern, entire section
5418 You can set the fill pattern for an entire section by using
5419 @samp{=@var{fillexp}}. @var{fillexp} is an expression
5420 (@pxref{Expressions}). Any otherwise unspecified regions of memory
5421 within the output section (for example, gaps left due to the required
5422 alignment of input sections) will be filled with the value, repeated as
5423 necessary. If the fill expression is a simple hex number, ie. a string
5424 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
5425 an arbitrarily long sequence of hex digits can be used to specify the
5426 fill pattern; Leading zeros become part of the pattern too. For all
5427 other cases, including extra parentheses or a unary @code{+}, the fill
5428 pattern is the four least significant bytes of the value of the
5429 expression. In all cases, the number is big-endian.
5431 You can also change the fill value with a @code{FILL} command in the
5432 output section commands; (@pxref{Output Section Data}).
5434 Here is a simple example:
5437 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
5441 @node Overlay Description
5442 @subsection Overlay Description
5445 An overlay description provides an easy way to describe sections which
5446 are to be loaded as part of a single memory image but are to be run at
5447 the same memory address. At run time, some sort of overlay manager will
5448 copy the overlaid sections in and out of the runtime memory address as
5449 required, perhaps by simply manipulating addressing bits. This approach
5450 can be useful, for example, when a certain region of memory is faster
5453 Overlays are described using the @code{OVERLAY} command. The
5454 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
5455 output section description. The full syntax of the @code{OVERLAY}
5456 command is as follows:
5459 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
5463 @var{output-section-command}
5464 @var{output-section-command}
5466 @} [:@var{phdr}@dots{}] [=@var{fill}]
5469 @var{output-section-command}
5470 @var{output-section-command}
5472 @} [:@var{phdr}@dots{}] [=@var{fill}]
5474 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5478 Everything is optional except @code{OVERLAY} (a keyword), and each
5479 section must have a name (@var{secname1} and @var{secname2} above). The
5480 section definitions within the @code{OVERLAY} construct are identical to
5481 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5482 except that no addresses and no memory regions may be defined for
5483 sections within an @code{OVERLAY}.
5485 The comma at the end may be required if a @var{fill} is used and
5486 the next @var{sections-command} looks like a continuation of the expression.
5488 The sections are all defined with the same starting address. The load
5489 addresses of the sections are arranged such that they are consecutive in
5490 memory starting at the load address used for the @code{OVERLAY} as a
5491 whole (as with normal section definitions, the load address is optional,
5492 and defaults to the start address; the start address is also optional,
5493 and defaults to the current value of the location counter).
5495 If the @code{NOCROSSREFS} keyword is used, and there are any
5496 references among the sections, the linker will report an error. Since
5497 the sections all run at the same address, it normally does not make
5498 sense for one section to refer directly to another.
5499 @xref{Miscellaneous Commands, NOCROSSREFS}.
5501 For each section within the @code{OVERLAY}, the linker automatically
5502 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5503 defined as the starting load address of the section. The symbol
5504 @code{__load_stop_@var{secname}} is defined as the final load address of
5505 the section. Any characters within @var{secname} which are not legal
5506 within C identifiers are removed. C (or assembler) code may use these
5507 symbols to move the overlaid sections around as necessary.
5509 At the end of the overlay, the value of the location counter is set to
5510 the start address of the overlay plus the size of the largest section.
5512 Here is an example. Remember that this would appear inside a
5513 @code{SECTIONS} construct.
5516 OVERLAY 0x1000 : AT (0x4000)
5518 .text0 @{ o1/*.o(.text) @}
5519 .text1 @{ o2/*.o(.text) @}
5524 This will define both @samp{.text0} and @samp{.text1} to start at
5525 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5526 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5527 following symbols will be defined if referenced: @code{__load_start_text0},
5528 @code{__load_stop_text0}, @code{__load_start_text1},
5529 @code{__load_stop_text1}.
5531 C code to copy overlay @code{.text1} into the overlay area might look
5536 extern char __load_start_text1, __load_stop_text1;
5537 memcpy ((char *) 0x1000, &__load_start_text1,
5538 &__load_stop_text1 - &__load_start_text1);
5542 Note that the @code{OVERLAY} command is just syntactic sugar, since
5543 everything it does can be done using the more basic commands. The above
5544 example could have been written identically as follows.
5548 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5549 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5550 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5551 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5552 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5553 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5554 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5559 @section MEMORY Command
5561 @cindex memory regions
5562 @cindex regions of memory
5563 @cindex allocating memory
5564 @cindex discontinuous memory
5565 The linker's default configuration permits allocation of all available
5566 memory. You can override this by using the @code{MEMORY} command.
5568 The @code{MEMORY} command describes the location and size of blocks of
5569 memory in the target. You can use it to describe which memory regions
5570 may be used by the linker, and which memory regions it must avoid. You
5571 can then assign sections to particular memory regions. The linker will
5572 set section addresses based on the memory regions, and will warn about
5573 regions that become too full. The linker will not shuffle sections
5574 around to fit into the available regions.
5576 A linker script may contain many uses of the @code{MEMORY} command,
5577 however, all memory blocks defined are treated as if they were
5578 specified inside a single @code{MEMORY} command. The syntax for
5584 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5590 The @var{name} is a name used in the linker script to refer to the
5591 region. The region name has no meaning outside of the linker script.
5592 Region names are stored in a separate name space, and will not conflict
5593 with symbol names, file names, or section names. Each memory region
5594 must have a distinct name within the @code{MEMORY} command. However you can
5595 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5598 @cindex memory region attributes
5599 The @var{attr} string is an optional list of attributes that specify
5600 whether to use a particular memory region for an input section which is
5601 not explicitly mapped in the linker script. As described in
5602 @ref{SECTIONS}, if you do not specify an output section for some input
5603 section, the linker will create an output section with the same name as
5604 the input section. If you define region attributes, the linker will use
5605 them to select the memory region for the output section that it creates.
5607 The @var{attr} string must consist only of the following characters:
5622 Invert the sense of any of the attributes that follow
5625 If an unmapped section matches any of the listed attributes other than
5626 @samp{!}, it will be placed in the memory region. The @samp{!}
5627 attribute reverses the test for the characters that follow, so that an
5628 unmapped section will be placed in the memory region only if it does
5629 not match any of the attributes listed afterwards. Thus an attribute
5630 string of @samp{RW!X} will match any unmapped section that has either
5631 or both of the @samp{R} and @samp{W} attributes, but only as long as
5632 the section does not also have the @samp{X} attribute.
5637 The @var{origin} is an numerical expression for the start address of
5638 the memory region. The expression must evaluate to a constant and it
5639 cannot involve any symbols. The keyword @code{ORIGIN} may be
5640 abbreviated to @code{org} or @code{o} (but not, for example,
5646 The @var{len} is an expression for the size in bytes of the memory
5647 region. As with the @var{origin} expression, the expression must
5648 be numerical only and must evaluate to a constant. The keyword
5649 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5651 In the following example, we specify that there are two memory regions
5652 available for allocation: one starting at @samp{0} for 256 kilobytes,
5653 and the other starting at @samp{0x40000000} for four megabytes. The
5654 linker will place into the @samp{rom} memory region every section which
5655 is not explicitly mapped into a memory region, and is either read-only
5656 or executable. The linker will place other sections which are not
5657 explicitly mapped into a memory region into the @samp{ram} memory
5664 rom (rx) : ORIGIN = 0, LENGTH = 256K
5665 ram (!rx) : org = 0x40000000, l = 4M
5670 Once you define a memory region, you can direct the linker to place
5671 specific output sections into that memory region by using the
5672 @samp{>@var{region}} output section attribute. For example, if you have
5673 a memory region named @samp{mem}, you would use @samp{>mem} in the
5674 output section definition. @xref{Output Section Region}. If no address
5675 was specified for the output section, the linker will set the address to
5676 the next available address within the memory region. If the combined
5677 output sections directed to a memory region are too large for the
5678 region, the linker will issue an error message.
5680 It is possible to access the origin and length of a memory in an
5681 expression via the @code{ORIGIN(@var{memory})} and
5682 @code{LENGTH(@var{memory})} functions:
5686 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5691 @section PHDRS Command
5693 @cindex program headers
5694 @cindex ELF program headers
5695 @cindex program segments
5696 @cindex segments, ELF
5697 The ELF object file format uses @dfn{program headers}, also knows as
5698 @dfn{segments}. The program headers describe how the program should be
5699 loaded into memory. You can print them out by using the @code{objdump}
5700 program with the @samp{-p} option.
5702 When you run an ELF program on a native ELF system, the system loader
5703 reads the program headers in order to figure out how to load the
5704 program. This will only work if the program headers are set correctly.
5705 This manual does not describe the details of how the system loader
5706 interprets program headers; for more information, see the ELF ABI.
5708 The linker will create reasonable program headers by default. However,
5709 in some cases, you may need to specify the program headers more
5710 precisely. You may use the @code{PHDRS} command for this purpose. When
5711 the linker sees the @code{PHDRS} command in the linker script, it will
5712 not create any program headers other than the ones specified.
5714 The linker only pays attention to the @code{PHDRS} command when
5715 generating an ELF output file. In other cases, the linker will simply
5716 ignore @code{PHDRS}.
5718 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5719 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5725 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5726 [ FLAGS ( @var{flags} ) ] ;
5731 The @var{name} is used only for reference in the @code{SECTIONS} command
5732 of the linker script. It is not put into the output file. Program
5733 header names are stored in a separate name space, and will not conflict
5734 with symbol names, file names, or section names. Each program header
5735 must have a distinct name. The headers are processed in order and it
5736 is usual for them to map to sections in ascending load address order.
5738 Certain program header types describe segments of memory which the
5739 system loader will load from the file. In the linker script, you
5740 specify the contents of these segments by placing allocatable output
5741 sections in the segments. You use the @samp{:@var{phdr}} output section
5742 attribute to place a section in a particular segment. @xref{Output
5745 It is normal to put certain sections in more than one segment. This
5746 merely implies that one segment of memory contains another. You may
5747 repeat @samp{:@var{phdr}}, using it once for each segment which should
5748 contain the section.
5750 If you place a section in one or more segments using @samp{:@var{phdr}},
5751 then the linker will place all subsequent allocatable sections which do
5752 not specify @samp{:@var{phdr}} in the same segments. This is for
5753 convenience, since generally a whole set of contiguous sections will be
5754 placed in a single segment. You can use @code{:NONE} to override the
5755 default segment and tell the linker to not put the section in any
5760 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
5761 the program header type to further describe the contents of the segment.
5762 The @code{FILEHDR} keyword means that the segment should include the ELF
5763 file header. The @code{PHDRS} keyword means that the segment should
5764 include the ELF program headers themselves. If applied to a loadable
5765 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5768 The @var{type} may be one of the following. The numbers indicate the
5769 value of the keyword.
5772 @item @code{PT_NULL} (0)
5773 Indicates an unused program header.
5775 @item @code{PT_LOAD} (1)
5776 Indicates that this program header describes a segment to be loaded from
5779 @item @code{PT_DYNAMIC} (2)
5780 Indicates a segment where dynamic linking information can be found.
5782 @item @code{PT_INTERP} (3)
5783 Indicates a segment where the name of the program interpreter may be
5786 @item @code{PT_NOTE} (4)
5787 Indicates a segment holding note information.
5789 @item @code{PT_SHLIB} (5)
5790 A reserved program header type, defined but not specified by the ELF
5793 @item @code{PT_PHDR} (6)
5794 Indicates a segment where the program headers may be found.
5796 @item @code{PT_TLS} (7)
5797 Indicates a segment containing thread local storage.
5799 @item @var{expression}
5800 An expression giving the numeric type of the program header. This may
5801 be used for types not defined above.
5804 You can specify that a segment should be loaded at a particular address
5805 in memory by using an @code{AT} expression. This is identical to the
5806 @code{AT} command used as an output section attribute (@pxref{Output
5807 Section LMA}). The @code{AT} command for a program header overrides the
5808 output section attribute.
5810 The linker will normally set the segment flags based on the sections
5811 which comprise the segment. You may use the @code{FLAGS} keyword to
5812 explicitly specify the segment flags. The value of @var{flags} must be
5813 an integer. It is used to set the @code{p_flags} field of the program
5816 Here is an example of @code{PHDRS}. This shows a typical set of program
5817 headers used on a native ELF system.
5823 headers PT_PHDR PHDRS ;
5825 text PT_LOAD FILEHDR PHDRS ;
5827 dynamic PT_DYNAMIC ;
5833 .interp : @{ *(.interp) @} :text :interp
5834 .text : @{ *(.text) @} :text
5835 .rodata : @{ *(.rodata) @} /* defaults to :text */
5837 . = . + 0x1000; /* move to a new page in memory */
5838 .data : @{ *(.data) @} :data
5839 .dynamic : @{ *(.dynamic) @} :data :dynamic
5846 @section VERSION Command
5847 @kindex VERSION @{script text@}
5848 @cindex symbol versions
5849 @cindex version script
5850 @cindex versions of symbols
5851 The linker supports symbol versions when using ELF. Symbol versions are
5852 only useful when using shared libraries. The dynamic linker can use
5853 symbol versions to select a specific version of a function when it runs
5854 a program that may have been linked against an earlier version of the
5857 You can include a version script directly in the main linker script, or
5858 you can supply the version script as an implicit linker script. You can
5859 also use the @samp{--version-script} linker option.
5861 The syntax of the @code{VERSION} command is simply
5863 VERSION @{ version-script-commands @}
5866 The format of the version script commands is identical to that used by
5867 Sun's linker in Solaris 2.5. The version script defines a tree of
5868 version nodes. You specify the node names and interdependencies in the
5869 version script. You can specify which symbols are bound to which
5870 version nodes, and you can reduce a specified set of symbols to local
5871 scope so that they are not globally visible outside of the shared
5874 The easiest way to demonstrate the version script language is with a few
5900 This example version script defines three version nodes. The first
5901 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5902 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5903 a number of symbols to local scope so that they are not visible outside
5904 of the shared library; this is done using wildcard patterns, so that any
5905 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5906 is matched. The wildcard patterns available are the same as those used
5907 in the shell when matching filenames (also known as ``globbing'').
5908 However, if you specify the symbol name inside double quotes, then the
5909 name is treated as literal, rather than as a glob pattern.
5911 Next, the version script defines node @samp{VERS_1.2}. This node
5912 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5913 to the version node @samp{VERS_1.2}.
5915 Finally, the version script defines node @samp{VERS_2.0}. This node
5916 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5917 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5919 When the linker finds a symbol defined in a library which is not
5920 specifically bound to a version node, it will effectively bind it to an
5921 unspecified base version of the library. You can bind all otherwise
5922 unspecified symbols to a given version node by using @samp{global: *;}
5923 somewhere in the version script. Note that it's slightly crazy to use
5924 wildcards in a global spec except on the last version node. Global
5925 wildcards elsewhere run the risk of accidentally adding symbols to the
5926 set exported for an old version. That's wrong since older versions
5927 ought to have a fixed set of symbols.
5929 The names of the version nodes have no specific meaning other than what
5930 they might suggest to the person reading them. The @samp{2.0} version
5931 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5932 However, this would be a confusing way to write a version script.
5934 Node name can be omitted, provided it is the only version node
5935 in the version script. Such version script doesn't assign any versions to
5936 symbols, only selects which symbols will be globally visible out and which
5940 @{ global: foo; bar; local: *; @};
5943 When you link an application against a shared library that has versioned
5944 symbols, the application itself knows which version of each symbol it
5945 requires, and it also knows which version nodes it needs from each
5946 shared library it is linked against. Thus at runtime, the dynamic
5947 loader can make a quick check to make sure that the libraries you have
5948 linked against do in fact supply all of the version nodes that the
5949 application will need to resolve all of the dynamic symbols. In this
5950 way it is possible for the dynamic linker to know with certainty that
5951 all external symbols that it needs will be resolvable without having to
5952 search for each symbol reference.
5954 The symbol versioning is in effect a much more sophisticated way of
5955 doing minor version checking that SunOS does. The fundamental problem
5956 that is being addressed here is that typically references to external
5957 functions are bound on an as-needed basis, and are not all bound when
5958 the application starts up. If a shared library is out of date, a
5959 required interface may be missing; when the application tries to use
5960 that interface, it may suddenly and unexpectedly fail. With symbol
5961 versioning, the user will get a warning when they start their program if
5962 the libraries being used with the application are too old.
5964 There are several GNU extensions to Sun's versioning approach. The
5965 first of these is the ability to bind a symbol to a version node in the
5966 source file where the symbol is defined instead of in the versioning
5967 script. This was done mainly to reduce the burden on the library
5968 maintainer. You can do this by putting something like:
5970 __asm__(".symver original_foo,foo@@VERS_1.1");
5973 in the C source file. This renames the function @samp{original_foo} to
5974 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5975 The @samp{local:} directive can be used to prevent the symbol
5976 @samp{original_foo} from being exported. A @samp{.symver} directive
5977 takes precedence over a version script.
5979 The second GNU extension is to allow multiple versions of the same
5980 function to appear in a given shared library. In this way you can make
5981 an incompatible change to an interface without increasing the major
5982 version number of the shared library, while still allowing applications
5983 linked against the old interface to continue to function.
5985 To do this, you must use multiple @samp{.symver} directives in the
5986 source file. Here is an example:
5989 __asm__(".symver original_foo,foo@@");
5990 __asm__(".symver old_foo,foo@@VERS_1.1");
5991 __asm__(".symver old_foo1,foo@@VERS_1.2");
5992 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5995 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5996 unspecified base version of the symbol. The source file that contains this
5997 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5998 @samp{old_foo1}, and @samp{new_foo}.
6000 When you have multiple definitions of a given symbol, there needs to be
6001 some way to specify a default version to which external references to
6002 this symbol will be bound. You can do this with the
6003 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
6004 declare one version of a symbol as the default in this manner; otherwise
6005 you would effectively have multiple definitions of the same symbol.
6007 If you wish to bind a reference to a specific version of the symbol
6008 within the shared library, you can use the aliases of convenience
6009 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
6010 specifically bind to an external version of the function in question.
6012 You can also specify the language in the version script:
6015 VERSION extern "lang" @{ version-script-commands @}
6018 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
6019 The linker will iterate over the list of symbols at the link time and
6020 demangle them according to @samp{lang} before matching them to the
6021 patterns specified in @samp{version-script-commands}. The default
6022 @samp{lang} is @samp{C}.
6024 Demangled names may contains spaces and other special characters. As
6025 described above, you can use a glob pattern to match demangled names,
6026 or you can use a double-quoted string to match the string exactly. In
6027 the latter case, be aware that minor differences (such as differing
6028 whitespace) between the version script and the demangler output will
6029 cause a mismatch. As the exact string generated by the demangler
6030 might change in the future, even if the mangled name does not, you
6031 should check that all of your version directives are behaving as you
6032 expect when you upgrade.
6035 @section Expressions in Linker Scripts
6038 The syntax for expressions in the linker script language is identical to
6039 that of C expressions. All expressions are evaluated as integers. All
6040 expressions are evaluated in the same size, which is 32 bits if both the
6041 host and target are 32 bits, and is otherwise 64 bits.
6043 You can use and set symbol values in expressions.
6045 The linker defines several special purpose builtin functions for use in
6049 * Constants:: Constants
6050 * Symbolic Constants:: Symbolic constants
6051 * Symbols:: Symbol Names
6052 * Orphan Sections:: Orphan Sections
6053 * Location Counter:: The Location Counter
6054 * Operators:: Operators
6055 * Evaluation:: Evaluation
6056 * Expression Section:: The Section of an Expression
6057 * Builtin Functions:: Builtin Functions
6061 @subsection Constants
6062 @cindex integer notation
6063 @cindex constants in linker scripts
6064 All constants are integers.
6066 As in C, the linker considers an integer beginning with @samp{0} to be
6067 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
6068 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
6069 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
6070 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
6071 value without a prefix or a suffix is considered to be decimal.
6073 @cindex scaled integers
6074 @cindex K and M integer suffixes
6075 @cindex M and K integer suffixes
6076 @cindex suffixes for integers
6077 @cindex integer suffixes
6078 In addition, you can use the suffixes @code{K} and @code{M} to scale a
6082 @c END TEXI2ROFF-KILL
6083 @code{1024} or @code{1024*1024}
6087 ${\rm 1024}$ or ${\rm 1024}^2$
6089 @c END TEXI2ROFF-KILL
6090 respectively. For example, the following
6091 all refer to the same quantity:
6100 Note - the @code{K} and @code{M} suffixes cannot be used in
6101 conjunction with the base suffixes mentioned above.
6103 @node Symbolic Constants
6104 @subsection Symbolic Constants
6105 @cindex symbolic constants
6107 It is possible to refer to target-specific constants via the use of
6108 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
6113 The target's maximum page size.
6115 @item COMMONPAGESIZE
6116 @kindex COMMONPAGESIZE
6117 The target's default page size.
6123 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
6126 will create a text section aligned to the largest page boundary
6127 supported by the target.
6130 @subsection Symbol Names
6131 @cindex symbol names
6133 @cindex quoted symbol names
6135 Unless quoted, symbol names start with a letter, underscore, or period
6136 and may include letters, digits, underscores, periods, and hyphens.
6137 Unquoted symbol names must not conflict with any keywords. You can
6138 specify a symbol which contains odd characters or has the same name as a
6139 keyword by surrounding the symbol name in double quotes:
6142 "with a space" = "also with a space" + 10;
6145 Since symbols can contain many non-alphabetic characters, it is safest
6146 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
6147 whereas @samp{A - B} is an expression involving subtraction.
6149 @node Orphan Sections
6150 @subsection Orphan Sections
6152 Orphan sections are sections present in the input files which
6153 are not explicitly placed into the output file by the linker
6154 script. The linker will still copy these sections into the
6155 output file by either finding, or creating a suitable output section
6156 in which to place the orphaned input section.
6158 If the name of an orphaned input section exactly matches the name of
6159 an existing output section, then the orphaned input section will be
6160 placed at the end of that output section.
6162 If there is no output section with a matching name then new output
6163 sections will be created. Each new output section will have the same
6164 name as the orphan section placed within it. If there are multiple
6165 orphan sections with the same name, these will all be combined into
6166 one new output section.
6168 If new output sections are created to hold orphaned input sections,
6169 then the linker must decide where to place these new output sections
6170 in relation to existing output sections. On most modern targets, the
6171 linker attempts to place orphan sections after sections of the same
6172 attribute, such as code vs data, loadable vs non-loadable, etc. If no
6173 sections with matching attributes are found, or your target lacks this
6174 support, the orphan section is placed at the end of the file.
6176 The command-line options @samp{--orphan-handling} and @samp{--unique}
6177 (@pxref{Options,,Command-line Options}) can be used to control which
6178 output sections an orphan is placed in.
6180 @node Location Counter
6181 @subsection The Location Counter
6184 @cindex location counter
6185 @cindex current output location
6186 The special linker variable @dfn{dot} @samp{.} always contains the
6187 current output location counter. Since the @code{.} always refers to a
6188 location in an output section, it may only appear in an expression
6189 within a @code{SECTIONS} command. The @code{.} symbol may appear
6190 anywhere that an ordinary symbol is allowed in an expression.
6193 Assigning a value to @code{.} will cause the location counter to be
6194 moved. This may be used to create holes in the output section. The
6195 location counter may not be moved backwards inside an output section,
6196 and may not be moved backwards outside of an output section if so
6197 doing creates areas with overlapping LMAs.
6213 In the previous example, the @samp{.text} section from @file{file1} is
6214 located at the beginning of the output section @samp{output}. It is
6215 followed by a 1000 byte gap. Then the @samp{.text} section from
6216 @file{file2} appears, also with a 1000 byte gap following before the
6217 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
6218 specifies what data to write in the gaps (@pxref{Output Section Fill}).
6220 @cindex dot inside sections
6221 Note: @code{.} actually refers to the byte offset from the start of the
6222 current containing object. Normally this is the @code{SECTIONS}
6223 statement, whose start address is 0, hence @code{.} can be used as an
6224 absolute address. If @code{.} is used inside a section description
6225 however, it refers to the byte offset from the start of that section,
6226 not an absolute address. Thus in a script like this:
6244 The @samp{.text} section will be assigned a starting address of 0x100
6245 and a size of exactly 0x200 bytes, even if there is not enough data in
6246 the @samp{.text} input sections to fill this area. (If there is too
6247 much data, an error will be produced because this would be an attempt to
6248 move @code{.} backwards). The @samp{.data} section will start at 0x500
6249 and it will have an extra 0x600 bytes worth of space after the end of
6250 the values from the @samp{.data} input sections and before the end of
6251 the @samp{.data} output section itself.
6253 @cindex dot outside sections
6254 Setting symbols to the value of the location counter outside of an
6255 output section statement can result in unexpected values if the linker
6256 needs to place orphan sections. For example, given the following:
6262 .text: @{ *(.text) @}
6266 .data: @{ *(.data) @}
6271 If the linker needs to place some input section, e.g. @code{.rodata},
6272 not mentioned in the script, it might choose to place that section
6273 between @code{.text} and @code{.data}. You might think the linker
6274 should place @code{.rodata} on the blank line in the above script, but
6275 blank lines are of no particular significance to the linker. As well,
6276 the linker doesn't associate the above symbol names with their
6277 sections. Instead, it assumes that all assignments or other
6278 statements belong to the previous output section, except for the
6279 special case of an assignment to @code{.}. I.e., the linker will
6280 place the orphan @code{.rodata} section as if the script was written
6287 .text: @{ *(.text) @}
6291 .rodata: @{ *(.rodata) @}
6292 .data: @{ *(.data) @}
6297 This may or may not be the script author's intention for the value of
6298 @code{start_of_data}. One way to influence the orphan section
6299 placement is to assign the location counter to itself, as the linker
6300 assumes that an assignment to @code{.} is setting the start address of
6301 a following output section and thus should be grouped with that
6302 section. So you could write:
6308 .text: @{ *(.text) @}
6313 .data: @{ *(.data) @}
6318 Now, the orphan @code{.rodata} section will be placed between
6319 @code{end_of_text} and @code{start_of_data}.
6323 @subsection Operators
6324 @cindex operators for arithmetic
6325 @cindex arithmetic operators
6326 @cindex precedence in expressions
6327 The linker recognizes the standard C set of arithmetic operators, with
6328 the standard bindings and precedence levels:
6331 @c END TEXI2ROFF-KILL
6333 precedence associativity Operators Notes
6339 5 left == != > < <= >=
6345 11 right &= += -= *= /= (2)
6349 (1) Prefix operators
6350 (2) @xref{Assignments}.
6354 \vskip \baselineskip
6355 %"lispnarrowing" is the extra indent used generally for smallexample
6356 \hskip\lispnarrowing\vbox{\offinterlineskip
6359 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
6360 height2pt&\omit&&\omit&&\omit&\cr
6361 &Precedence&& Associativity &&{\rm Operators}&\cr
6362 height2pt&\omit&&\omit&&\omit&\cr
6364 height2pt&\omit&&\omit&&\omit&\cr
6366 % '176 is tilde, '~' in tt font
6367 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
6368 &2&&left&&* / \%&\cr
6371 &5&&left&&== != > < <= >=&\cr
6374 &8&&left&&{\&\&}&\cr
6377 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
6379 height2pt&\omit&&\omit&&\omit&\cr}
6384 @obeylines@parskip=0pt@parindent=0pt
6385 @dag@quad Prefix operators.
6386 @ddag@quad @xref{Assignments}.
6389 @c END TEXI2ROFF-KILL
6392 @subsection Evaluation
6393 @cindex lazy evaluation
6394 @cindex expression evaluation order
6395 The linker evaluates expressions lazily. It only computes the value of
6396 an expression when absolutely necessary.
6398 The linker needs some information, such as the value of the start
6399 address of the first section, and the origins and lengths of memory
6400 regions, in order to do any linking at all. These values are computed
6401 as soon as possible when the linker reads in the linker script.
6403 However, other values (such as symbol values) are not known or needed
6404 until after storage allocation. Such values are evaluated later, when
6405 other information (such as the sizes of output sections) is available
6406 for use in the symbol assignment expression.
6408 The sizes of sections cannot be known until after allocation, so
6409 assignments dependent upon these are not performed until after
6412 Some expressions, such as those depending upon the location counter
6413 @samp{.}, must be evaluated during section allocation.
6415 If the result of an expression is required, but the value is not
6416 available, then an error results. For example, a script like the
6422 .text 9+this_isnt_constant :
6428 will cause the error message @samp{non constant expression for initial
6431 @node Expression Section
6432 @subsection The Section of an Expression
6433 @cindex expression sections
6434 @cindex absolute expressions
6435 @cindex relative expressions
6436 @cindex absolute and relocatable symbols
6437 @cindex relocatable and absolute symbols
6438 @cindex symbols, relocatable and absolute
6439 Addresses and symbols may be section relative, or absolute. A section
6440 relative symbol is relocatable. If you request relocatable output
6441 using the @samp{-r} option, a further link operation may change the
6442 value of a section relative symbol. On the other hand, an absolute
6443 symbol will retain the same value throughout any further link
6446 Some terms in linker expressions are addresses. This is true of
6447 section relative symbols and for builtin functions that return an
6448 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
6449 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
6450 functions that return a non-address value, such as @code{LENGTH}.
6451 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
6452 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
6453 differently depending on their location, for compatibility with older
6454 versions of @code{ld}. Expressions appearing outside an output
6455 section definition treat all numbers as absolute addresses.
6456 Expressions appearing inside an output section definition treat
6457 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
6458 given, then absolute symbols and numbers are simply treated as numbers
6461 In the following simple example,
6468 __executable_start = 0x100;
6472 __data_start = 0x10;
6480 both @code{.} and @code{__executable_start} are set to the absolute
6481 address 0x100 in the first two assignments, then both @code{.} and
6482 @code{__data_start} are set to 0x10 relative to the @code{.data}
6483 section in the second two assignments.
6485 For expressions involving numbers, relative addresses and absolute
6486 addresses, ld follows these rules to evaluate terms:
6490 Unary operations on an absolute address or number, and binary
6491 operations on two absolute addresses or two numbers, or between one
6492 absolute address and a number, apply the operator to the value(s).
6494 Unary operations on a relative address, and binary operations on two
6495 relative addresses in the same section or between one relative address
6496 and a number, apply the operator to the offset part of the address(es).
6498 Other binary operations, that is, between two relative addresses not
6499 in the same section, or between a relative address and an absolute
6500 address, first convert any non-absolute term to an absolute address
6501 before applying the operator.
6504 The result section of each sub-expression is as follows:
6508 An operation involving only numbers results in a number.
6510 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6512 The result of other binary arithmetic and logical operations on two
6513 relative addresses in the same section or two absolute addresses
6514 (after above conversions) is also a number when
6515 @code{LD_FEATURE ("SANE_EXPR")} or inside an output section definition
6516 but an absolute address otherwise.
6518 The result of other operations on relative addresses or one
6519 relative address and a number, is a relative address in the same
6520 section as the relative operand(s).
6522 The result of other operations on absolute addresses (after above
6523 conversions) is an absolute address.
6526 You can use the builtin function @code{ABSOLUTE} to force an expression
6527 to be absolute when it would otherwise be relative. For example, to
6528 create an absolute symbol set to the address of the end of the output
6529 section @samp{.data}:
6533 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6537 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6538 @samp{.data} section.
6540 Using @code{LOADADDR} also forces an expression absolute, since this
6541 particular builtin function returns an absolute address.
6543 @node Builtin Functions
6544 @subsection Builtin Functions
6545 @cindex functions in expressions
6546 The linker script language includes a number of builtin functions for
6547 use in linker script expressions.
6550 @item ABSOLUTE(@var{exp})
6551 @kindex ABSOLUTE(@var{exp})
6552 @cindex expression, absolute
6553 Return the absolute (non-relocatable, as opposed to non-negative) value
6554 of the expression @var{exp}. Primarily useful to assign an absolute
6555 value to a symbol within a section definition, where symbol values are
6556 normally section relative. @xref{Expression Section}.
6558 @item ADDR(@var{section})
6559 @kindex ADDR(@var{section})
6560 @cindex section address in expression
6561 Return the address (VMA) of the named @var{section}. Your
6562 script must previously have defined the location of that section. In
6563 the following example, @code{start_of_output_1}, @code{symbol_1} and
6564 @code{symbol_2} are assigned equivalent values, except that
6565 @code{symbol_1} will be relative to the @code{.output1} section while
6566 the other two will be absolute:
6572 start_of_output_1 = ABSOLUTE(.);
6577 symbol_1 = ADDR(.output1);
6578 symbol_2 = start_of_output_1;
6584 @item ALIGN(@var{align})
6585 @itemx ALIGN(@var{exp},@var{align})
6586 @kindex ALIGN(@var{align})
6587 @kindex ALIGN(@var{exp},@var{align})
6588 @cindex round up location counter
6589 @cindex align location counter
6590 @cindex round up expression
6591 @cindex align expression
6592 Return the location counter (@code{.}) or arbitrary expression aligned
6593 to the next @var{align} boundary. The single operand @code{ALIGN}
6594 doesn't change the value of the location counter---it just does
6595 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6596 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6597 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6599 Here is an example which aligns the output @code{.data} section to the
6600 next @code{0x2000} byte boundary after the preceding section and sets a
6601 variable within the section to the next @code{0x8000} boundary after the
6606 .data ALIGN(0x2000): @{
6608 variable = ALIGN(0x8000);
6614 The first use of @code{ALIGN} in this example specifies the location of
6615 a section because it is used as the optional @var{address} attribute of
6616 a section definition (@pxref{Output Section Address}). The second use
6617 of @code{ALIGN} is used to defines the value of a symbol.
6619 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6621 @item ALIGNOF(@var{section})
6622 @kindex ALIGNOF(@var{section})
6623 @cindex section alignment
6624 Return the alignment in bytes of the named @var{section}, if that section has
6625 been allocated. If the section has not been allocated when this is
6626 evaluated, the linker will report an error. In the following example,
6627 the alignment of the @code{.output} section is stored as the first
6628 value in that section.
6633 LONG (ALIGNOF (.output))
6640 @item BLOCK(@var{exp})
6641 @kindex BLOCK(@var{exp})
6642 This is a synonym for @code{ALIGN}, for compatibility with older linker
6643 scripts. It is most often seen when setting the address of an output
6646 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6647 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6648 This is equivalent to either
6650 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6654 (ALIGN(@var{maxpagesize})
6655 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6658 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6659 for the data segment (area between the result of this expression and
6660 @code{DATA_SEGMENT_END}) than the former or not.
6661 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6662 memory will be saved at the expense of up to @var{commonpagesize} wasted
6663 bytes in the on-disk file.
6665 This expression can only be used directly in @code{SECTIONS} commands, not in
6666 any output section descriptions and only once in the linker script.
6667 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6668 be the system page size the object wants to be optimized for while still
6669 running on system page sizes up to @var{maxpagesize}. Note however
6670 that @samp{-z relro} protection will not be effective if the system
6671 page size is larger than @var{commonpagesize}.
6676 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6679 @item DATA_SEGMENT_END(@var{exp})
6680 @kindex DATA_SEGMENT_END(@var{exp})
6681 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6682 evaluation purposes.
6685 . = DATA_SEGMENT_END(.);
6688 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6689 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6690 This defines the end of the @code{PT_GNU_RELRO} segment when
6691 @samp{-z relro} option is used.
6692 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6693 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6694 @var{exp} + @var{offset} is aligned to the @var{commonpagesize}
6695 argument given to @code{DATA_SEGMENT_ALIGN}. If present in the linker
6696 script, it must be placed between @code{DATA_SEGMENT_ALIGN} and
6697 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6698 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6702 . = DATA_SEGMENT_RELRO_END(24, .);
6705 @item DEFINED(@var{symbol})
6706 @kindex DEFINED(@var{symbol})
6707 @cindex symbol defaults
6708 Return 1 if @var{symbol} is in the linker global symbol table and is
6709 defined before the statement using DEFINED in the script, otherwise
6710 return 0. You can use this function to provide
6711 default values for symbols. For example, the following script fragment
6712 shows how to set a global symbol @samp{begin} to the first location in
6713 the @samp{.text} section---but if a symbol called @samp{begin} already
6714 existed, its value is preserved:
6720 begin = DEFINED(begin) ? begin : . ;
6728 @item LENGTH(@var{memory})
6729 @kindex LENGTH(@var{memory})
6730 Return the length of the memory region named @var{memory}.
6732 @item LOADADDR(@var{section})
6733 @kindex LOADADDR(@var{section})
6734 @cindex section load address in expression
6735 Return the absolute LMA of the named @var{section}. (@pxref{Output
6738 @item LOG2CEIL(@var{exp})
6739 @kindex LOG2CEIL(@var{exp})
6740 Return the binary logarithm of @var{exp} rounded towards infinity.
6741 @code{LOG2CEIL(0)} returns 0.
6744 @item MAX(@var{exp1}, @var{exp2})
6745 Returns the maximum of @var{exp1} and @var{exp2}.
6748 @item MIN(@var{exp1}, @var{exp2})
6749 Returns the minimum of @var{exp1} and @var{exp2}.
6751 @item NEXT(@var{exp})
6752 @kindex NEXT(@var{exp})
6753 @cindex unallocated address, next
6754 Return the next unallocated address that is a multiple of @var{exp}.
6755 This function is closely related to @code{ALIGN(@var{exp})}; unless you
6756 use the @code{MEMORY} command to define discontinuous memory for the
6757 output file, the two functions are equivalent.
6759 @item ORIGIN(@var{memory})
6760 @kindex ORIGIN(@var{memory})
6761 Return the origin of the memory region named @var{memory}.
6763 @item SEGMENT_START(@var{segment}, @var{default})
6764 @kindex SEGMENT_START(@var{segment}, @var{default})
6765 Return the base address of the named @var{segment}. If an explicit
6766 value has already been given for this segment (with a command-line
6767 @samp{-T} option) then that value will be returned otherwise the value
6768 will be @var{default}. At present, the @samp{-T} command-line option
6769 can only be used to set the base address for the ``text'', ``data'', and
6770 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
6773 @item SIZEOF(@var{section})
6774 @kindex SIZEOF(@var{section})
6775 @cindex section size
6776 Return the size in bytes of the named @var{section}, if that section has
6777 been allocated. If the section has not been allocated when this is
6778 evaluated, the linker will report an error. In the following example,
6779 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6788 symbol_1 = .end - .start ;
6789 symbol_2 = SIZEOF(.output);
6794 @item SIZEOF_HEADERS
6795 @itemx sizeof_headers
6796 @kindex SIZEOF_HEADERS
6798 Return the size in bytes of the output file's headers. This is
6799 information which appears at the start of the output file. You can use
6800 this number when setting the start address of the first section, if you
6801 choose, to facilitate paging.
6803 @cindex not enough room for program headers
6804 @cindex program headers, not enough room
6805 When producing an ELF output file, if the linker script uses the
6806 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6807 number of program headers before it has determined all the section
6808 addresses and sizes. If the linker later discovers that it needs
6809 additional program headers, it will report an error @samp{not enough
6810 room for program headers}. To avoid this error, you must avoid using
6811 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6812 script to avoid forcing the linker to use additional program headers, or
6813 you must define the program headers yourself using the @code{PHDRS}
6814 command (@pxref{PHDRS}).
6817 @node Implicit Linker Scripts
6818 @section Implicit Linker Scripts
6819 @cindex implicit linker scripts
6820 If you specify a linker input file which the linker can not recognize as
6821 an object file or an archive file, it will try to read the file as a
6822 linker script. If the file can not be parsed as a linker script, the
6823 linker will report an error.
6825 An implicit linker script will not replace the default linker script.
6827 Typically an implicit linker script would contain only symbol
6828 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6831 Any input files read because of an implicit linker script will be read
6832 at the position in the command line where the implicit linker script was
6833 read. This can affect archive searching.
6836 @node Machine Dependent
6837 @chapter Machine Dependent Features
6839 @cindex machine dependencies
6840 @command{ld} has additional features on some platforms; the following
6841 sections describe them. Machines where @command{ld} has no additional
6842 functionality are not listed.
6846 * H8/300:: @command{ld} and the H8/300
6849 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6852 * ARM:: @command{ld} and the ARM family
6855 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6858 * M68K:: @command{ld} and the Motorola 68K family
6861 * MIPS:: @command{ld} and the MIPS family
6864 * MMIX:: @command{ld} and MMIX
6867 * MSP430:: @command{ld} and MSP430
6870 * NDS32:: @command{ld} and NDS32
6873 * Nios II:: @command{ld} and the Altera Nios II
6876 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6879 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6882 * S/390 ELF:: @command{ld} and S/390 ELF Support
6885 * SPU ELF:: @command{ld} and SPU ELF Support
6888 * TI COFF:: @command{ld} and TI COFF
6891 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6894 * Xtensa:: @command{ld} and Xtensa Processors
6905 @section @command{ld} and the H8/300
6907 @cindex H8/300 support
6908 For the H8/300, @command{ld} can perform these global optimizations when
6909 you specify the @samp{--relax} command-line option.
6912 @cindex relaxing on H8/300
6913 @item relaxing address modes
6914 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6915 targets are within eight bits, and turns them into eight-bit
6916 program-counter relative @code{bsr} and @code{bra} instructions,
6919 @cindex synthesizing on H8/300
6920 @item synthesizing instructions
6921 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6922 @command{ld} finds all @code{mov.b} instructions which use the
6923 sixteen-bit absolute address form, but refer to the top
6924 page of memory, and changes them to use the eight-bit address form.
6925 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6926 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6927 top page of memory).
6929 @command{ld} finds all @code{mov} instructions which use the register
6930 indirect with 32-bit displacement addressing mode, but use a small
6931 displacement inside 16-bit displacement range, and changes them to use
6932 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6933 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6934 whenever the displacement @var{d} is in the 16 bit signed integer
6935 range. Only implemented in ELF-format ld).
6937 @item bit manipulation instructions
6938 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6939 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6940 which use 32 bit and 16 bit absolute address form, but refer to the top
6941 page of memory, and changes them to use the 8 bit address form.
6942 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6943 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6944 the top page of memory).
6946 @item system control instructions
6947 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6948 32 bit absolute address form, but refer to the top page of memory, and
6949 changes them to use 16 bit address form.
6950 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6951 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6952 the top page of memory).
6962 @c This stuff is pointless to say unless you're especially concerned
6963 @c with Renesas chips; don't enable it for generic case, please.
6965 @chapter @command{ld} and Other Renesas Chips
6967 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6968 H8/500, and SH chips. No special features, commands, or command-line
6969 options are required for these chips.
6983 @node M68HC11/68HC12
6984 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6986 @cindex M68HC11 and 68HC12 support
6988 @subsection Linker Relaxation
6990 For the Motorola 68HC11, @command{ld} can perform these global
6991 optimizations when you specify the @samp{--relax} command-line option.
6994 @cindex relaxing on M68HC11
6995 @item relaxing address modes
6996 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6997 targets are within eight bits, and turns them into eight-bit
6998 program-counter relative @code{bsr} and @code{bra} instructions,
7001 @command{ld} also looks at all 16-bit extended addressing modes and
7002 transforms them in a direct addressing mode when the address is in
7003 page 0 (between 0 and 0x0ff).
7005 @item relaxing gcc instruction group
7006 When @command{gcc} is called with @option{-mrelax}, it can emit group
7007 of instructions that the linker can optimize to use a 68HC11 direct
7008 addressing mode. These instructions consists of @code{bclr} or
7009 @code{bset} instructions.
7013 @subsection Trampoline Generation
7015 @cindex trampoline generation on M68HC11
7016 @cindex trampoline generation on M68HC12
7017 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
7018 call a far function using a normal @code{jsr} instruction. The linker
7019 will also change the relocation to some far function to use the
7020 trampoline address instead of the function address. This is typically the
7021 case when a pointer to a function is taken. The pointer will in fact
7022 point to the function trampoline.
7030 @section @command{ld} and the ARM family
7032 @cindex ARM interworking support
7033 @kindex --support-old-code
7034 For the ARM, @command{ld} will generate code stubs to allow functions calls
7035 between ARM and Thumb code. These stubs only work with code that has
7036 been compiled and assembled with the @samp{-mthumb-interwork} command
7037 line option. If it is necessary to link with old ARM object files or
7038 libraries, which have not been compiled with the -mthumb-interwork
7039 option then the @samp{--support-old-code} command-line switch should be
7040 given to the linker. This will make it generate larger stub functions
7041 which will work with non-interworking aware ARM code. Note, however,
7042 the linker does not support generating stubs for function calls to
7043 non-interworking aware Thumb code.
7045 @cindex thumb entry point
7046 @cindex entry point, thumb
7047 @kindex --thumb-entry=@var{entry}
7048 The @samp{--thumb-entry} switch is a duplicate of the generic
7049 @samp{--entry} switch, in that it sets the program's starting address.
7050 But it also sets the bottom bit of the address, so that it can be
7051 branched to using a BX instruction, and the program will start
7052 executing in Thumb mode straight away.
7054 @cindex PE import table prefixing
7055 @kindex --use-nul-prefixed-import-tables
7056 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
7057 the import tables idata4 and idata5 have to be generated with a zero
7058 element prefix for import libraries. This is the old style to generate
7059 import tables. By default this option is turned off.
7063 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
7064 executables. This option is only valid when linking big-endian
7065 objects - ie ones which have been assembled with the @option{-EB}
7066 option. The resulting image will contain big-endian data and
7070 @kindex --target1-rel
7071 @kindex --target1-abs
7072 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
7073 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
7074 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
7075 and @samp{--target1-abs} switches override the default.
7078 @kindex --target2=@var{type}
7079 The @samp{--target2=type} switch overrides the default definition of the
7080 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
7081 meanings, and target defaults are as follows:
7084 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
7086 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
7088 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
7093 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
7094 specification) enables objects compiled for the ARMv4 architecture to be
7095 interworking-safe when linked with other objects compiled for ARMv4t, but
7096 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
7098 In the latter case, the switch @option{--fix-v4bx} must be passed to the
7099 linker, which causes v4t @code{BX rM} instructions to be rewritten as
7100 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
7102 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
7103 relocations are ignored.
7105 @cindex FIX_V4BX_INTERWORKING
7106 @kindex --fix-v4bx-interworking
7107 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
7108 relocations with a branch to the following veneer:
7116 This allows generation of libraries/applications that work on ARMv4 cores
7117 and are still interworking safe. Note that the above veneer clobbers the
7118 condition flags, so may cause incorrect program behavior in rare cases.
7122 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
7123 BLX instructions (available on ARMv5t and above) in various
7124 situations. Currently it is used to perform calls via the PLT from Thumb
7125 code using BLX rather than using BX and a mode-switching stub before
7126 each PLT entry. This should lead to such calls executing slightly faster.
7128 This option is enabled implicitly for SymbianOS, so there is no need to
7129 specify it if you are using that target.
7131 @cindex VFP11_DENORM_FIX
7132 @kindex --vfp11-denorm-fix
7133 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
7134 bug in certain VFP11 coprocessor hardware, which sometimes allows
7135 instructions with denorm operands (which must be handled by support code)
7136 to have those operands overwritten by subsequent instructions before
7137 the support code can read the intended values.
7139 The bug may be avoided in scalar mode if you allow at least one
7140 intervening instruction between a VFP11 instruction which uses a register
7141 and another instruction which writes to the same register, or at least two
7142 intervening instructions if vector mode is in use. The bug only affects
7143 full-compliance floating-point mode: you do not need this workaround if
7144 you are using "runfast" mode. Please contact ARM for further details.
7146 If you know you are using buggy VFP11 hardware, you can
7147 enable this workaround by specifying the linker option
7148 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
7149 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
7150 vector mode (the latter also works for scalar code). The default is
7151 @samp{--vfp-denorm-fix=none}.
7153 If the workaround is enabled, instructions are scanned for
7154 potentially-troublesome sequences, and a veneer is created for each
7155 such sequence which may trigger the erratum. The veneer consists of the
7156 first instruction of the sequence and a branch back to the subsequent
7157 instruction. The original instruction is then replaced with a branch to
7158 the veneer. The extra cycles required to call and return from the veneer
7159 are sufficient to avoid the erratum in both the scalar and vector cases.
7161 @cindex ARM1176 erratum workaround
7162 @kindex --fix-arm1176
7163 @kindex --no-fix-arm1176
7164 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
7165 in certain ARM1176 processors. The workaround is enabled by default if you
7166 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
7167 unconditionally by specifying @samp{--no-fix-arm1176}.
7169 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
7170 Programmer Advice Notice'' available on the ARM documentation website at:
7171 http://infocenter.arm.com/.
7173 @cindex STM32L4xx erratum workaround
7174 @kindex --fix-stm32l4xx-629360
7176 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
7177 workaround for a bug in the bus matrix / memory controller for some of
7178 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
7179 off-chip memory via the affected bus for bus reads of 9 words or more,
7180 the bus can generate corrupt data and/or abort. These are only
7181 core-initiated accesses (not DMA), and might affect any access:
7182 integer loads such as LDM, POP and floating-point loads such as VLDM,
7183 VPOP. Stores are not affected.
7185 The bug can be avoided by splitting memory accesses into the
7186 necessary chunks to keep bus reads below 8 words.
7188 The workaround is not enabled by default, this is equivalent to use
7189 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
7190 STM32L4xx hardware, you can enable the workaround by specifying the
7191 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
7192 @samp{--fix-stm32l4xx-629360=default}.
7194 If the workaround is enabled, instructions are scanned for
7195 potentially-troublesome sequences, and a veneer is created for each
7196 such sequence which may trigger the erratum. The veneer consists in a
7197 replacement sequence emulating the behaviour of the original one and a
7198 branch back to the subsequent instruction. The original instruction is
7199 then replaced with a branch to the veneer.
7201 The workaround does not always preserve the memory access order for
7202 the LDMDB instruction, when the instruction loads the PC.
7204 The workaround is not able to handle problematic instructions when
7205 they are in the middle of an IT block, since a branch is not allowed
7206 there. In that case, the linker reports a warning and no replacement
7209 The workaround is not able to replace problematic instructions with a
7210 PC-relative branch instruction if the @samp{.text} section is too
7211 large. In that case, when the branch that replaces the original code
7212 cannot be encoded, the linker reports a warning and no replacement
7215 @cindex NO_ENUM_SIZE_WARNING
7216 @kindex --no-enum-size-warning
7217 The @option{--no-enum-size-warning} switch prevents the linker from
7218 warning when linking object files that specify incompatible EABI
7219 enumeration size attributes. For example, with this switch enabled,
7220 linking of an object file using 32-bit enumeration values with another
7221 using enumeration values fitted into the smallest possible space will
7224 @cindex NO_WCHAR_SIZE_WARNING
7225 @kindex --no-wchar-size-warning
7226 The @option{--no-wchar-size-warning} switch prevents the linker from
7227 warning when linking object files that specify incompatible EABI
7228 @code{wchar_t} size attributes. For example, with this switch enabled,
7229 linking of an object file using 32-bit @code{wchar_t} values with another
7230 using 16-bit @code{wchar_t} values will not be diagnosed.
7233 @kindex --pic-veneer
7234 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
7235 ARM/Thumb interworking veneers, even if the rest of the binary
7236 is not PIC. This avoids problems on uClinux targets where
7237 @samp{--emit-relocs} is used to generate relocatable binaries.
7239 @cindex STUB_GROUP_SIZE
7240 @kindex --stub-group-size=@var{N}
7241 The linker will automatically generate and insert small sequences of
7242 code into a linked ARM ELF executable whenever an attempt is made to
7243 perform a function call to a symbol that is too far away. The
7244 placement of these sequences of instructions - called stubs - is
7245 controlled by the command-line option @option{--stub-group-size=N}.
7246 The placement is important because a poor choice can create a need for
7247 duplicate stubs, increasing the code size. The linker will try to
7248 group stubs together in order to reduce interruptions to the flow of
7249 code, but it needs guidance as to how big these groups should be and
7250 where they should be placed.
7252 The value of @samp{N}, the parameter to the
7253 @option{--stub-group-size=} option controls where the stub groups are
7254 placed. If it is negative then all stubs are placed after the first
7255 branch that needs them. If it is positive then the stubs can be
7256 placed either before or after the branches that need them. If the
7257 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
7258 exactly where to place groups of stubs, using its built in heuristics.
7259 A value of @samp{N} greater than 1 (or smaller than -1) tells the
7260 linker that a single group of stubs can service at most @samp{N} bytes
7261 from the input sections.
7263 The default, if @option{--stub-group-size=} is not specified, is
7266 Farcalls stubs insertion is fully supported for the ARM-EABI target
7267 only, because it relies on object files properties not present
7270 @cindex Cortex-A8 erratum workaround
7271 @kindex --fix-cortex-a8
7272 @kindex --no-fix-cortex-a8
7273 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}.
7275 The erratum only affects Thumb-2 code. Please contact ARM for further details.
7277 @cindex Cortex-A53 erratum 835769 workaround
7278 @kindex --fix-cortex-a53-835769
7279 @kindex --no-fix-cortex-a53-835769
7280 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}.
7282 Please contact ARM for further details.
7284 @kindex --merge-exidx-entries
7285 @kindex --no-merge-exidx-entries
7286 @cindex Merging exidx entries
7287 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
7290 @cindex 32-bit PLT entries
7291 The @samp{--long-plt} option enables the use of 16 byte PLT entries
7292 which support up to 4Gb of code. The default is to use 12 byte PLT
7293 entries which only support 512Mb of code.
7295 @kindex --no-apply-dynamic-relocs
7296 @cindex AArch64 rela addend
7297 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
7298 link-time values for dynamic relocations.
7300 @cindex Placement of SG veneers
7301 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
7302 Its start address must be set, either with the command-line option
7303 @samp{--section-start} or in a linker script, to indicate where to place these
7306 @kindex --cmse-implib
7307 @cindex Secure gateway import library
7308 The @samp{--cmse-implib} option requests that the import libraries
7309 specified by the @samp{--out-implib} and @samp{--in-implib} options are
7310 secure gateway import libraries, suitable for linking a non-secure
7311 executable against secure code as per ARMv8-M Security Extensions.
7313 @kindex --in-implib=@var{file}
7314 @cindex Input import library
7315 The @samp{--in-implib=file} specifies an input import library whose symbols
7316 must keep the same address in the executable being produced. A warning is
7317 given if no @samp{--out-implib} is given but new symbols have been introduced
7318 in the executable that should be listed in its import library. Otherwise, if
7319 @samp{--out-implib} is specified, the symbols are added to the output import
7320 library. A warning is also given if some symbols present in the input import
7321 library have disappeared from the executable. This option is only effective
7322 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
7336 @section @command{ld} and HPPA 32-bit ELF Support
7337 @cindex HPPA multiple sub-space stubs
7338 @kindex --multi-subspace
7339 When generating a shared library, @command{ld} will by default generate
7340 import stubs suitable for use with a single sub-space application.
7341 The @samp{--multi-subspace} switch causes @command{ld} to generate export
7342 stubs, and different (larger) import stubs suitable for use with
7343 multiple sub-spaces.
7345 @cindex HPPA stub grouping
7346 @kindex --stub-group-size=@var{N}
7347 Long branch stubs and import/export stubs are placed by @command{ld} in
7348 stub sections located between groups of input sections.
7349 @samp{--stub-group-size} specifies the maximum size of a group of input
7350 sections handled by one stub section. Since branch offsets are signed,
7351 a stub section may serve two groups of input sections, one group before
7352 the stub section, and one group after it. However, when using
7353 conditional branches that require stubs, it may be better (for branch
7354 prediction) that stub sections only serve one group of input sections.
7355 A negative value for @samp{N} chooses this scheme, ensuring that
7356 branches to stubs always use a negative offset. Two special values of
7357 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7358 @command{ld} to automatically size input section groups for the branch types
7359 detected, with the same behaviour regarding stub placement as other
7360 positive or negative values of @samp{N} respectively.
7362 Note that @samp{--stub-group-size} does not split input sections. A
7363 single input section larger than the group size specified will of course
7364 create a larger group (of one section). If input sections are too
7365 large, it may not be possible for a branch to reach its stub.
7378 @section @command{ld} and the Motorola 68K family
7380 @cindex Motorola 68K GOT generation
7381 @kindex --got=@var{type}
7382 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
7383 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
7384 @samp{target}. When @samp{target} is selected the linker chooses
7385 the default GOT generation scheme for the current target.
7386 @samp{single} tells the linker to generate a single GOT with
7387 entries only at non-negative offsets.
7388 @samp{negative} instructs the linker to generate a single GOT with
7389 entries at both negative and positive offsets. Not all environments
7391 @samp{multigot} allows the linker to generate several GOTs in the
7392 output file. All GOT references from a single input object
7393 file access the same GOT, but references from different input object
7394 files might access different GOTs. Not all environments support such GOTs.
7407 @section @command{ld} and the MIPS family
7409 @cindex MIPS microMIPS instruction choice selection
7412 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
7413 microMIPS instructions used in code generated by the linker, such as that
7414 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
7415 used, then the linker only uses 32-bit instruction encodings. By default
7416 or if @samp{--no-insn32} is used, all instruction encodings are used,
7417 including 16-bit ones where possible.
7419 @cindex MIPS branch relocation check control
7420 @kindex --ignore-branch-isa
7421 @kindex --no-ignore-branch-isa
7422 The @samp{--ignore-branch-isa} and @samp{--no-ignore-branch-isa} options
7423 control branch relocation checks for invalid ISA mode transitions. If
7424 @samp{--ignore-branch-isa} is used, then the linker accepts any branch
7425 relocations and any ISA mode transition required is lost in relocation
7426 calculation, except for some cases of @code{BAL} instructions which meet
7427 relaxation conditions and are converted to equivalent @code{JALX}
7428 instructions as the associated relocation is calculated. By default
7429 or if @samp{--no-ignore-branch-isa} is used a check is made causing
7430 the loss of an ISA mode transition to produce an error.
7443 @section @code{ld} and MMIX
7444 For MMIX, there is a choice of generating @code{ELF} object files or
7445 @code{mmo} object files when linking. The simulator @code{mmix}
7446 understands the @code{mmo} format. The binutils @code{objcopy} utility
7447 can translate between the two formats.
7449 There is one special section, the @samp{.MMIX.reg_contents} section.
7450 Contents in this section is assumed to correspond to that of global
7451 registers, and symbols referring to it are translated to special symbols,
7452 equal to registers. In a final link, the start address of the
7453 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7454 global register multiplied by 8. Register @code{$255} is not included in
7455 this section; it is always set to the program entry, which is at the
7456 symbol @code{Main} for @code{mmo} files.
7458 Global symbols with the prefix @code{__.MMIX.start.}, for example
7459 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7460 The default linker script uses these to set the default start address
7463 Initial and trailing multiples of zero-valued 32-bit words in a section,
7464 are left out from an mmo file.
7477 @section @code{ld} and MSP430
7478 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7479 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7480 just pass @samp{-m help} option to the linker).
7482 @cindex MSP430 extra sections
7483 The linker will recognize some extra sections which are MSP430 specific:
7486 @item @samp{.vectors}
7487 Defines a portion of ROM where interrupt vectors located.
7489 @item @samp{.bootloader}
7490 Defines the bootloader portion of the ROM (if applicable). Any code
7491 in this section will be uploaded to the MPU.
7493 @item @samp{.infomem}
7494 Defines an information memory section (if applicable). Any code in
7495 this section will be uploaded to the MPU.
7497 @item @samp{.infomemnobits}
7498 This is the same as the @samp{.infomem} section except that any code
7499 in this section will not be uploaded to the MPU.
7501 @item @samp{.noinit}
7502 Denotes a portion of RAM located above @samp{.bss} section.
7504 The last two sections are used by gcc.
7508 @cindex MSP430 Options
7509 @kindex --code-region
7510 @item --code-region=[either,lower,upper,none]
7511 This will transform .text* sections to [either,lower,upper].text* sections. The
7512 argument passed to GCC for -mcode-region is propagated to the linker
7515 @kindex --data-region
7516 @item --data-region=[either,lower,upper,none]
7517 This will transform .data*, .bss* and .rodata* sections to
7518 [either,lower,upper].[data,bss,rodata]* sections. The argument passed to GCC
7519 for -mdata-region is propagated to the linker using this option.
7521 @kindex --disable-sec-transformation
7522 @item --disable-sec-transformation
7523 Prevent the transformation of sections as specified by the @code{--code-region}
7524 and @code{--data-region} options.
7525 This is useful if you are compiling and linking using a single call to the GCC
7526 wrapper, and want to compile the source files using -m[code,data]-region but
7527 not transform the sections for prebuilt libraries and objects.
7541 @section @code{ld} and NDS32
7542 @kindex relaxing on NDS32
7543 For NDS32, there are some options to select relaxation behavior. The linker
7544 relaxes objects according to these options.
7547 @item @samp{--m[no-]fp-as-gp}
7548 Disable/enable fp-as-gp relaxation.
7550 @item @samp{--mexport-symbols=FILE}
7551 Exporting symbols and their address into FILE as linker script.
7553 @item @samp{--m[no-]ex9}
7554 Disable/enable link-time EX9 relaxation.
7556 @item @samp{--mexport-ex9=FILE}
7557 Export the EX9 table after linking.
7559 @item @samp{--mimport-ex9=FILE}
7560 Import the Ex9 table for EX9 relaxation.
7562 @item @samp{--mupdate-ex9}
7563 Update the existing EX9 table.
7565 @item @samp{--mex9-limit=NUM}
7566 Maximum number of entries in the ex9 table.
7568 @item @samp{--mex9-loop-aware}
7569 Avoid generating the EX9 instruction inside the loop.
7571 @item @samp{--m[no-]ifc}
7572 Disable/enable the link-time IFC optimization.
7574 @item @samp{--mifc-loop-aware}
7575 Avoid generating the IFC instruction inside the loop.
7589 @section @command{ld} and the Altera Nios II
7590 @cindex Nios II call relaxation
7591 @kindex --relax on Nios II
7593 Call and immediate jump instructions on Nios II processors are limited to
7594 transferring control to addresses in the same 256MB memory segment,
7595 which may result in @command{ld} giving
7596 @samp{relocation truncated to fit} errors with very large programs.
7597 The command-line option @option{--relax} enables the generation of
7598 trampolines that can access the entire 32-bit address space for calls
7599 outside the normal @code{call} and @code{jmpi} address range. These
7600 trampolines are inserted at section boundaries, so may not themselves
7601 be reachable if an input section and its associated call trampolines are
7604 The @option{--relax} option is enabled by default unless @option{-r}
7605 is also specified. You can disable trampoline generation by using the
7606 @option{--no-relax} linker option. You can also disable this optimization
7607 locally by using the @samp{set .noat} directive in assembly-language
7608 source files, as the linker-inserted trampolines use the @code{at}
7609 register as a temporary.
7611 Note that the linker @option{--relax} option is independent of assembler
7612 relaxation options, and that using the GNU assembler's @option{-relax-all}
7613 option interferes with the linker's more selective call instruction relaxation.
7626 @section @command{ld} and PowerPC 32-bit ELF Support
7627 @cindex PowerPC long branches
7628 @kindex --relax on PowerPC
7629 Branches on PowerPC processors are limited to a signed 26-bit
7630 displacement, which may result in @command{ld} giving
7631 @samp{relocation truncated to fit} errors with very large programs.
7632 @samp{--relax} enables the generation of trampolines that can access
7633 the entire 32-bit address space. These trampolines are inserted at
7634 section boundaries, so may not themselves be reachable if an input
7635 section exceeds 33M in size. You may combine @samp{-r} and
7636 @samp{--relax} to add trampolines in a partial link. In that case
7637 both branches to undefined symbols and inter-section branches are also
7638 considered potentially out of range, and trampolines inserted.
7640 @cindex PowerPC ELF32 options
7645 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7646 generates code capable of using a newer PLT and GOT layout that has
7647 the security advantage of no executable section ever needing to be
7648 writable and no writable section ever being executable. PowerPC
7649 @command{ld} will generate this layout, including stubs to access the
7650 PLT, if all input files (including startup and static libraries) were
7651 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7652 BSS PLT (and GOT layout) which can give slightly better performance.
7654 @kindex --secure-plt
7656 @command{ld} will use the new PLT and GOT layout if it is linking new
7657 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7658 when linking non-PIC code. This option requests the new PLT and GOT
7659 layout. A warning will be given if some object file requires the old
7665 The new secure PLT and GOT are placed differently relative to other
7666 sections compared to older BSS PLT and GOT placement. The location of
7667 @code{.plt} must change because the new secure PLT is an initialized
7668 section while the old PLT is uninitialized. The reason for the
7669 @code{.got} change is more subtle: The new placement allows
7670 @code{.got} to be read-only in applications linked with
7671 @samp{-z relro -z now}. However, this placement means that
7672 @code{.sdata} cannot always be used in shared libraries, because the
7673 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7674 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7675 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7676 really only useful for other compilers that may do so.
7678 @cindex PowerPC stub symbols
7679 @kindex --emit-stub-syms
7680 @item --emit-stub-syms
7681 This option causes @command{ld} to label linker stubs with a local
7682 symbol that encodes the stub type and destination.
7684 @cindex PowerPC TLS optimization
7685 @kindex --no-tls-optimize
7686 @item --no-tls-optimize
7687 PowerPC @command{ld} normally performs some optimization of code
7688 sequences used to access Thread-Local Storage. Use this option to
7689 disable the optimization.
7702 @node PowerPC64 ELF64
7703 @section @command{ld} and PowerPC64 64-bit ELF Support
7705 @cindex PowerPC64 ELF64 options
7707 @cindex PowerPC64 stub grouping
7708 @kindex --stub-group-size
7709 @item --stub-group-size
7710 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
7711 by @command{ld} in stub sections located between groups of input sections.
7712 @samp{--stub-group-size} specifies the maximum size of a group of input
7713 sections handled by one stub section. Since branch offsets are signed,
7714 a stub section may serve two groups of input sections, one group before
7715 the stub section, and one group after it. However, when using
7716 conditional branches that require stubs, it may be better (for branch
7717 prediction) that stub sections only serve one group of input sections.
7718 A negative value for @samp{N} chooses this scheme, ensuring that
7719 branches to stubs always use a negative offset. Two special values of
7720 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7721 @command{ld} to automatically size input section groups for the branch types
7722 detected, with the same behaviour regarding stub placement as other
7723 positive or negative values of @samp{N} respectively.
7725 Note that @samp{--stub-group-size} does not split input sections. A
7726 single input section larger than the group size specified will of course
7727 create a larger group (of one section). If input sections are too
7728 large, it may not be possible for a branch to reach its stub.
7730 @cindex PowerPC64 stub symbols
7731 @kindex --emit-stub-syms
7732 @item --emit-stub-syms
7733 This option causes @command{ld} to label linker stubs with a local
7734 symbol that encodes the stub type and destination.
7736 @cindex PowerPC64 dot symbols
7738 @kindex --no-dotsyms
7741 These two options control how @command{ld} interprets version patterns
7742 in a version script. Older PowerPC64 compilers emitted both a
7743 function descriptor symbol with the same name as the function, and a
7744 code entry symbol with the name prefixed by a dot (@samp{.}). To
7745 properly version a function @samp{foo}, the version script thus needs
7746 to control both @samp{foo} and @samp{.foo}. The option
7747 @samp{--dotsyms}, on by default, automatically adds the required
7748 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
7751 @cindex PowerPC64 register save/restore functions
7752 @kindex --save-restore-funcs
7753 @kindex --no-save-restore-funcs
7754 @item --save-restore-funcs
7755 @itemx --no-save-restore-funcs
7756 These two options control whether PowerPC64 @command{ld} automatically
7757 provides out-of-line register save and restore functions used by
7758 @samp{-Os} code. The default is to provide any such referenced
7759 function for a normal final link, and to not do so for a relocatable
7762 @cindex PowerPC64 TLS optimization
7763 @kindex --no-tls-optimize
7764 @item --no-tls-optimize
7765 PowerPC64 @command{ld} normally performs some optimization of code
7766 sequences used to access Thread-Local Storage. Use this option to
7767 disable the optimization.
7769 @cindex PowerPC64 __tls_get_addr optimization
7770 @kindex --tls-get-addr-optimize
7771 @kindex --no-tls-get-addr-optimize
7772 @kindex --tls-get-addr-regsave
7773 @kindex --no-tls-get-addr-regsave
7774 @item --tls-get-addr-optimize
7775 @itemx --no-tls-get-addr-optimize
7776 These options control how PowerPC64 @command{ld} uses a special
7777 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
7778 an optimization that allows the second and subsequent calls to
7779 @code{__tls_get_addr} for a given symbol to be resolved by the special
7780 stub without calling in to glibc. By default the linker enables
7781 generation of the stub when glibc advertises the availability of
7783 Using @option{--tls-get-addr-optimize} with an older glibc won't do
7784 much besides slow down your applications, but may be useful if linking
7785 an application against an older glibc with the expectation that it
7786 will normally be used on systems having a newer glibc.
7787 @option{--tls-get-addr-regsave} forces generation of a stub that saves
7788 and restores volatile registers around the call into glibc. Normally,
7789 this is done when the linker detects a call to __tls_get_addr_desc.
7790 Such calls then go via the register saving stub to __tls_get_addr_opt.
7791 @option{--no-tls-get-addr-regsave} disables generation of the
7794 @cindex PowerPC64 OPD optimization
7795 @kindex --no-opd-optimize
7796 @item --no-opd-optimize
7797 PowerPC64 @command{ld} normally removes @code{.opd} section entries
7798 corresponding to deleted link-once functions, or functions removed by
7799 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
7800 Use this option to disable @code{.opd} optimization.
7802 @cindex PowerPC64 OPD spacing
7803 @kindex --non-overlapping-opd
7804 @item --non-overlapping-opd
7805 Some PowerPC64 compilers have an option to generate compressed
7806 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
7807 the static chain pointer (unused in C) with the first word of the next
7808 entry. This option expands such entries to the full 24 bytes.
7810 @cindex PowerPC64 TOC optimization
7811 @kindex --no-toc-optimize
7812 @item --no-toc-optimize
7813 PowerPC64 @command{ld} normally removes unused @code{.toc} section
7814 entries. Such entries are detected by examining relocations that
7815 reference the TOC in code sections. A reloc in a deleted code section
7816 marks a TOC word as unneeded, while a reloc in a kept code section
7817 marks a TOC word as needed. Since the TOC may reference itself, TOC
7818 relocs are also examined. TOC words marked as both needed and
7819 unneeded will of course be kept. TOC words without any referencing
7820 reloc are assumed to be part of a multi-word entry, and are kept or
7821 discarded as per the nearest marked preceding word. This works
7822 reliably for compiler generated code, but may be incorrect if assembly
7823 code is used to insert TOC entries. Use this option to disable the
7826 @cindex PowerPC64 multi-TOC
7827 @kindex --no-multi-toc
7828 @item --no-multi-toc
7829 If given any toc option besides @code{-mcmodel=medium} or
7830 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
7832 entries are accessed with a 16-bit offset from r2. This limits the
7833 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
7834 grouping code sections such that each group uses less than 64K for its
7835 TOC entries, then inserts r2 adjusting stubs between inter-group
7836 calls. @command{ld} does not split apart input sections, so cannot
7837 help if a single input file has a @code{.toc} section that exceeds
7838 64K, most likely from linking multiple files with @command{ld -r}.
7839 Use this option to turn off this feature.
7841 @cindex PowerPC64 TOC sorting
7842 @kindex --no-toc-sort
7844 By default, @command{ld} sorts TOC sections so that those whose file
7845 happens to have a section called @code{.init} or @code{.fini} are
7846 placed first, followed by TOC sections referenced by code generated
7847 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
7848 referenced only by code generated with PowerPC64 gcc's
7849 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
7850 results in better TOC grouping for multi-TOC. Use this option to turn
7853 @cindex PowerPC64 PLT stub alignment
7855 @kindex --no-plt-align
7857 @itemx --no-plt-align
7858 Use these options to control whether individual PLT call stubs are
7859 aligned to a 32-byte boundary, or to the specified power of two
7860 boundary when using @code{--plt-align=}. A negative value may be
7861 specified to pad PLT call stubs so that they do not cross the
7862 specified power of two boundary (or the minimum number of boundaries
7863 if a PLT stub is so large that it must cross a boundary). By default
7864 PLT call stubs are aligned to 32-byte boundaries.
7866 @cindex PowerPC64 PLT call stub static chain
7867 @kindex --plt-static-chain
7868 @kindex --no-plt-static-chain
7869 @item --plt-static-chain
7870 @itemx --no-plt-static-chain
7871 Use these options to control whether PLT call stubs load the static
7872 chain pointer (r11). @code{ld} defaults to not loading the static
7873 chain since there is never any need to do so on a PLT call.
7875 @cindex PowerPC64 PLT call stub thread safety
7876 @kindex --plt-thread-safe
7877 @kindex --no-plt-thread-safe
7878 @item --plt-thread-safe
7879 @itemx --no-plt-thread-safe
7880 With power7's weakly ordered memory model, it is possible when using
7881 lazy binding for ld.so to update a plt entry in one thread and have
7882 another thread see the individual plt entry words update in the wrong
7883 order, despite ld.so carefully writing in the correct order and using
7884 memory write barriers. To avoid this we need some sort of read
7885 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
7886 looks for calls to commonly used functions that create threads, and if
7887 seen, adds the necessary barriers. Use these options to change the
7890 @cindex PowerPC64 ELFv2 PLT localentry optimization
7891 @kindex --plt-localentry
7892 @kindex --no-plt-localentry
7893 @item --plt-localentry
7894 @itemx --no-localentry
7895 ELFv2 functions with localentry:0 are those with a single entry point,
7896 ie. global entry == local entry, and that have no requirement on r2
7897 (the TOC/GOT pointer) or r12, and guarantee r2 is unchanged on return.
7898 Such an external function can be called via the PLT without saving r2
7899 or restoring it on return, avoiding a common load-hit-store for small
7900 functions. The optimization is attractive, with up to 40% reduction
7901 in execution time for a small function, but can result in symbol
7902 interposition failures. Also, minor changes in a shared library,
7903 including system libraries, can cause a function that was localentry:0
7904 to become localentry:8. This will result in a dynamic loader
7905 complaint and failure to run. The option is experimental, use with
7906 care. @option{--no-plt-localentry} is the default.
7920 @section @command{ld} and S/390 ELF Support
7922 @cindex S/390 ELF options
7926 @kindex --s390-pgste
7928 This option marks the result file with a @code{PT_S390_PGSTE}
7929 segment. The Linux kernel is supposed to allocate 4k page tables for
7930 binaries marked that way.
7944 @section @command{ld} and SPU ELF Support
7946 @cindex SPU ELF options
7952 This option marks an executable as a PIC plugin module.
7954 @cindex SPU overlays
7955 @kindex --no-overlays
7957 Normally, @command{ld} recognizes calls to functions within overlay
7958 regions, and redirects such calls to an overlay manager via a stub.
7959 @command{ld} also provides a built-in overlay manager. This option
7960 turns off all this special overlay handling.
7962 @cindex SPU overlay stub symbols
7963 @kindex --emit-stub-syms
7964 @item --emit-stub-syms
7965 This option causes @command{ld} to label overlay stubs with a local
7966 symbol that encodes the stub type and destination.
7968 @cindex SPU extra overlay stubs
7969 @kindex --extra-overlay-stubs
7970 @item --extra-overlay-stubs
7971 This option causes @command{ld} to add overlay call stubs on all
7972 function calls out of overlay regions. Normally stubs are not added
7973 on calls to non-overlay regions.
7975 @cindex SPU local store size
7976 @kindex --local-store=lo:hi
7977 @item --local-store=lo:hi
7978 @command{ld} usually checks that a final executable for SPU fits in
7979 the address range 0 to 256k. This option may be used to change the
7980 range. Disable the check entirely with @option{--local-store=0:0}.
7983 @kindex --stack-analysis
7984 @item --stack-analysis
7985 SPU local store space is limited. Over-allocation of stack space
7986 unnecessarily limits space available for code and data, while
7987 under-allocation results in runtime failures. If given this option,
7988 @command{ld} will provide an estimate of maximum stack usage.
7989 @command{ld} does this by examining symbols in code sections to
7990 determine the extents of functions, and looking at function prologues
7991 for stack adjusting instructions. A call-graph is created by looking
7992 for relocations on branch instructions. The graph is then searched
7993 for the maximum stack usage path. Note that this analysis does not
7994 find calls made via function pointers, and does not handle recursion
7995 and other cycles in the call graph. Stack usage may be
7996 under-estimated if your code makes such calls. Also, stack usage for
7997 dynamic allocation, e.g. alloca, will not be detected. If a link map
7998 is requested, detailed information about each function's stack usage
7999 and calls will be given.
8002 @kindex --emit-stack-syms
8003 @item --emit-stack-syms
8004 This option, if given along with @option{--stack-analysis} will result
8005 in @command{ld} emitting stack sizing symbols for each function.
8006 These take the form @code{__stack_<function_name>} for global
8007 functions, and @code{__stack_<number>_<function_name>} for static
8008 functions. @code{<number>} is the section id in hex. The value of
8009 such symbols is the stack requirement for the corresponding function.
8010 The symbol size will be zero, type @code{STT_NOTYPE}, binding
8011 @code{STB_LOCAL}, and section @code{SHN_ABS}.
8025 @section @command{ld}'s Support for Various TI COFF Versions
8026 @cindex TI COFF versions
8027 @kindex --format=@var{version}
8028 The @samp{--format} switch allows selection of one of the various
8029 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
8030 also supported. The TI COFF versions also vary in header byte-order
8031 format; @command{ld} will read any version or byte order, but the output
8032 header format depends on the default specified by the specific target.
8045 @section @command{ld} and WIN32 (cygwin/mingw)
8047 This section describes some of the win32 specific @command{ld} issues.
8048 See @ref{Options,,Command-line Options} for detailed description of the
8049 command-line options mentioned here.
8052 @cindex import libraries
8053 @item import libraries
8054 The standard Windows linker creates and uses so-called import
8055 libraries, which contains information for linking to dll's. They are
8056 regular static archives and are handled as any other static
8057 archive. The cygwin and mingw ports of @command{ld} have specific
8058 support for creating such libraries provided with the
8059 @samp{--out-implib} command-line option.
8061 @item exporting DLL symbols
8062 @cindex exporting DLL symbols
8063 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
8066 @item using auto-export functionality
8067 @cindex using auto-export functionality
8068 By default @command{ld} exports symbols with the auto-export functionality,
8069 which is controlled by the following command-line options:
8072 @item --export-all-symbols [This is the default]
8073 @item --exclude-symbols
8074 @item --exclude-libs
8075 @item --exclude-modules-for-implib
8076 @item --version-script
8079 When auto-export is in operation, @command{ld} will export all the non-local
8080 (global and common) symbols it finds in a DLL, with the exception of a few
8081 symbols known to belong to the system's runtime and libraries. As it will
8082 often not be desirable to export all of a DLL's symbols, which may include
8083 private functions that are not part of any public interface, the command-line
8084 options listed above may be used to filter symbols out from the list for
8085 exporting. The @samp{--output-def} option can be used in order to see the
8086 final list of exported symbols with all exclusions taken into effect.
8088 If @samp{--export-all-symbols} is not given explicitly on the
8089 command line, then the default auto-export behavior will be @emph{disabled}
8090 if either of the following are true:
8093 @item A DEF file is used.
8094 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
8097 @item using a DEF file
8098 @cindex using a DEF file
8099 Another way of exporting symbols is using a DEF file. A DEF file is
8100 an ASCII file containing definitions of symbols which should be
8101 exported when a dll is created. Usually it is named @samp{<dll
8102 name>.def} and is added as any other object file to the linker's
8103 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
8106 gcc -o <output> <objectfiles> <dll name>.def
8109 Using a DEF file turns off the normal auto-export behavior, unless the
8110 @samp{--export-all-symbols} option is also used.
8112 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
8115 LIBRARY "xyz.dll" BASE=0x20000000
8121 another_foo = abc.dll.afoo
8127 This example defines a DLL with a non-default base address and seven
8128 symbols in the export table. The third exported symbol @code{_bar} is an
8129 alias for the second. The fourth symbol, @code{another_foo} is resolved
8130 by "forwarding" to another module and treating it as an alias for
8131 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
8132 @code{var1} is declared to be a data object. The @samp{doo} symbol in
8133 export library is an alias of @samp{foo}, which gets the string name
8134 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
8135 symbol, which gets in export table the name @samp{var1}.
8137 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
8138 name of the output DLL. If @samp{<name>} does not include a suffix,
8139 the default library suffix, @samp{.DLL} is appended.
8141 When the .DEF file is used to build an application, rather than a
8142 library, the @code{NAME <name>} command should be used instead of
8143 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
8144 executable suffix, @samp{.EXE} is appended.
8146 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
8147 specification @code{BASE = <number>} may be used to specify a
8148 non-default base address for the image.
8150 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
8151 or they specify an empty string, the internal name is the same as the
8152 filename specified on the command line.
8154 The complete specification of an export symbol is:
8158 ( ( ( <name1> [ = <name2> ] )
8159 | ( <name1> = <module-name> . <external-name>))
8160 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
8163 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
8164 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
8165 @samp{<name1>} as a "forward" alias for the symbol
8166 @samp{<external-name>} in the DLL @samp{<module-name>}.
8167 Optionally, the symbol may be exported by the specified ordinal
8168 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
8169 string in import/export table for the symbol.
8171 The optional keywords that follow the declaration indicate:
8173 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
8174 will still be exported by its ordinal alias (either the value specified
8175 by the .def specification or, otherwise, the value assigned by the
8176 linker). The symbol name, however, does remain visible in the import
8177 library (if any), unless @code{PRIVATE} is also specified.
8179 @code{DATA}: The symbol is a variable or object, rather than a function.
8180 The import lib will export only an indirect reference to @code{foo} as
8181 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
8184 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
8185 well as @code{_imp__foo} into the import library. Both refer to the
8186 read-only import address table's pointer to the variable, not to the
8187 variable itself. This can be dangerous. If the user code fails to add
8188 the @code{dllimport} attribute and also fails to explicitly add the
8189 extra indirection that the use of the attribute enforces, the
8190 application will behave unexpectedly.
8192 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
8193 it into the static import library used to resolve imports at link time. The
8194 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
8195 API at runtime or by using the GNU ld extension of linking directly to
8196 the DLL without an import library.
8198 See ld/deffilep.y in the binutils sources for the full specification of
8199 other DEF file statements
8201 @cindex creating a DEF file
8202 While linking a shared dll, @command{ld} is able to create a DEF file
8203 with the @samp{--output-def <file>} command-line option.
8205 @item Using decorations
8206 @cindex Using decorations
8207 Another way of marking symbols for export is to modify the source code
8208 itself, so that when building the DLL each symbol to be exported is
8212 __declspec(dllexport) int a_variable
8213 __declspec(dllexport) void a_function(int with_args)
8216 All such symbols will be exported from the DLL. If, however,
8217 any of the object files in the DLL contain symbols decorated in
8218 this way, then the normal auto-export behavior is disabled, unless
8219 the @samp{--export-all-symbols} option is also used.
8221 Note that object files that wish to access these symbols must @emph{not}
8222 decorate them with dllexport. Instead, they should use dllimport,
8226 __declspec(dllimport) int a_variable
8227 __declspec(dllimport) void a_function(int with_args)
8230 This complicates the structure of library header files, because
8231 when included by the library itself the header must declare the
8232 variables and functions as dllexport, but when included by client
8233 code the header must declare them as dllimport. There are a number
8234 of idioms that are typically used to do this; often client code can
8235 omit the __declspec() declaration completely. See
8236 @samp{--enable-auto-import} and @samp{automatic data imports} for more
8240 @cindex automatic data imports
8241 @item automatic data imports
8242 The standard Windows dll format supports data imports from dlls only
8243 by adding special decorations (dllimport/dllexport), which let the
8244 compiler produce specific assembler instructions to deal with this
8245 issue. This increases the effort necessary to port existing Un*x
8246 code to these platforms, especially for large
8247 c++ libraries and applications. The auto-import feature, which was
8248 initially provided by Paul Sokolovsky, allows one to omit the
8249 decorations to achieve a behavior that conforms to that on POSIX/Un*x
8250 platforms. This feature is enabled with the @samp{--enable-auto-import}
8251 command-line option, although it is enabled by default on cygwin/mingw.
8252 The @samp{--enable-auto-import} option itself now serves mainly to
8253 suppress any warnings that are ordinarily emitted when linked objects
8254 trigger the feature's use.
8256 auto-import of variables does not always work flawlessly without
8257 additional assistance. Sometimes, you will see this message
8259 "variable '<var>' can't be auto-imported. Please read the
8260 documentation for ld's @code{--enable-auto-import} for details."
8262 The @samp{--enable-auto-import} documentation explains why this error
8263 occurs, and several methods that can be used to overcome this difficulty.
8264 One of these methods is the @emph{runtime pseudo-relocs} feature, described
8267 @cindex runtime pseudo-relocation
8268 For complex variables imported from DLLs (such as structs or classes),
8269 object files typically contain a base address for the variable and an
8270 offset (@emph{addend}) within the variable--to specify a particular
8271 field or public member, for instance. Unfortunately, the runtime loader used
8272 in win32 environments is incapable of fixing these references at runtime
8273 without the additional information supplied by dllimport/dllexport decorations.
8274 The standard auto-import feature described above is unable to resolve these
8277 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
8278 be resolved without error, while leaving the task of adjusting the references
8279 themselves (with their non-zero addends) to specialized code provided by the
8280 runtime environment. Recent versions of the cygwin and mingw environments and
8281 compilers provide this runtime support; older versions do not. However, the
8282 support is only necessary on the developer's platform; the compiled result will
8283 run without error on an older system.
8285 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
8288 @cindex direct linking to a dll
8289 @item direct linking to a dll
8290 The cygwin/mingw ports of @command{ld} support the direct linking,
8291 including data symbols, to a dll without the usage of any import
8292 libraries. This is much faster and uses much less memory than does the
8293 traditional import library method, especially when linking large
8294 libraries or applications. When @command{ld} creates an import lib, each
8295 function or variable exported from the dll is stored in its own bfd, even
8296 though a single bfd could contain many exports. The overhead involved in
8297 storing, loading, and processing so many bfd's is quite large, and explains the
8298 tremendous time, memory, and storage needed to link against particularly
8299 large or complex libraries when using import libs.
8301 Linking directly to a dll uses no extra command-line switches other than
8302 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
8303 of names to match each library. All that is needed from the developer's
8304 perspective is an understanding of this search, in order to force ld to
8305 select the dll instead of an import library.
8308 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
8309 to find, in the first directory of its search path,
8322 before moving on to the next directory in the search path.
8324 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
8325 where @samp{<prefix>} is set by the @command{ld} option
8326 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
8327 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
8330 Other win32-based unix environments, such as mingw or pw32, may use other
8331 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
8332 was originally intended to help avoid name conflicts among dll's built for the
8333 various win32/un*x environments, so that (for example) two versions of a zlib dll
8334 could coexist on the same machine.
8336 The generic cygwin/mingw path layout uses a @samp{bin} directory for
8337 applications and dll's and a @samp{lib} directory for the import
8338 libraries (using cygwin nomenclature):
8344 libxxx.dll.a (in case of dll's)
8345 libxxx.a (in case of static archive)
8348 Linking directly to a dll without using the import library can be
8351 1. Use the dll directly by adding the @samp{bin} path to the link line
8353 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
8356 However, as the dll's often have version numbers appended to their names
8357 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
8358 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
8359 not versioned, and do not have this difficulty.
8361 2. Create a symbolic link from the dll to a file in the @samp{lib}
8362 directory according to the above mentioned search pattern. This
8363 should be used to avoid unwanted changes in the tools needed for
8367 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
8370 Then you can link without any make environment changes.
8373 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
8376 This technique also avoids the version number problems, because the following is
8383 libxxx.dll.a -> ../bin/cygxxx-5.dll
8386 Linking directly to a dll without using an import lib will work
8387 even when auto-import features are exercised, and even when
8388 @samp{--enable-runtime-pseudo-relocs} is used.
8390 Given the improvements in speed and memory usage, one might justifiably
8391 wonder why import libraries are used at all. There are three reasons:
8393 1. Until recently, the link-directly-to-dll functionality did @emph{not}
8394 work with auto-imported data.
8396 2. Sometimes it is necessary to include pure static objects within the
8397 import library (which otherwise contains only bfd's for indirection
8398 symbols that point to the exports of a dll). Again, the import lib
8399 for the cygwin kernel makes use of this ability, and it is not
8400 possible to do this without an import lib.
8402 3. Symbol aliases can only be resolved using an import lib. This is
8403 critical when linking against OS-supplied dll's (eg, the win32 API)
8404 in which symbols are usually exported as undecorated aliases of their
8405 stdcall-decorated assembly names.
8407 So, import libs are not going away. But the ability to replace
8408 true import libs with a simple symbolic link to (or a copy of)
8409 a dll, in many cases, is a useful addition to the suite of tools
8410 binutils makes available to the win32 developer. Given the
8411 massive improvements in memory requirements during linking, storage
8412 requirements, and linking speed, we expect that many developers
8413 will soon begin to use this feature whenever possible.
8415 @item symbol aliasing
8417 @item adding additional names
8418 Sometimes, it is useful to export symbols with additional names.
8419 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
8420 exported as @samp{_foo} by using special directives in the DEF file
8421 when creating the dll. This will affect also the optional created
8422 import library. Consider the following DEF file:
8425 LIBRARY "xyz.dll" BASE=0x61000000
8432 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
8434 Another method for creating a symbol alias is to create it in the
8435 source code using the "weak" attribute:
8438 void foo () @{ /* Do something. */; @}
8439 void _foo () __attribute__ ((weak, alias ("foo")));
8442 See the gcc manual for more information about attributes and weak
8445 @item renaming symbols
8446 Sometimes it is useful to rename exports. For instance, the cygwin
8447 kernel does this regularly. A symbol @samp{_foo} can be exported as
8448 @samp{foo} but not as @samp{_foo} by using special directives in the
8449 DEF file. (This will also affect the import library, if it is
8450 created). In the following example:
8453 LIBRARY "xyz.dll" BASE=0x61000000
8459 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
8463 Note: using a DEF file disables the default auto-export behavior,
8464 unless the @samp{--export-all-symbols} command-line option is used.
8465 If, however, you are trying to rename symbols, then you should list
8466 @emph{all} desired exports in the DEF file, including the symbols
8467 that are not being renamed, and do @emph{not} use the
8468 @samp{--export-all-symbols} option. If you list only the
8469 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
8470 to handle the other symbols, then the both the new names @emph{and}
8471 the original names for the renamed symbols will be exported.
8472 In effect, you'd be aliasing those symbols, not renaming them,
8473 which is probably not what you wanted.
8475 @cindex weak externals
8476 @item weak externals
8477 The Windows object format, PE, specifies a form of weak symbols called
8478 weak externals. When a weak symbol is linked and the symbol is not
8479 defined, the weak symbol becomes an alias for some other symbol. There
8480 are three variants of weak externals:
8482 @item Definition is searched for in objects and libraries, historically
8483 called lazy externals.
8484 @item Definition is searched for only in other objects, not in libraries.
8485 This form is not presently implemented.
8486 @item No search; the symbol is an alias. This form is not presently
8489 As a GNU extension, weak symbols that do not specify an alternate symbol
8490 are supported. If the symbol is undefined when linking, the symbol
8491 uses a default value.
8493 @cindex aligned common symbols
8494 @item aligned common symbols
8495 As a GNU extension to the PE file format, it is possible to specify the
8496 desired alignment for a common symbol. This information is conveyed from
8497 the assembler or compiler to the linker by means of GNU-specific commands
8498 carried in the object file's @samp{.drectve} section, which are recognized
8499 by @command{ld} and respected when laying out the common symbols. Native
8500 tools will be able to process object files employing this GNU extension,
8501 but will fail to respect the alignment instructions, and may issue noisy
8502 warnings about unknown linker directives.
8517 @section @code{ld} and Xtensa Processors
8519 @cindex Xtensa processors
8520 The default @command{ld} behavior for Xtensa processors is to interpret
8521 @code{SECTIONS} commands so that lists of explicitly named sections in a
8522 specification with a wildcard file will be interleaved when necessary to
8523 keep literal pools within the range of PC-relative load offsets. For
8524 example, with the command:
8536 @command{ld} may interleave some of the @code{.literal}
8537 and @code{.text} sections from different object files to ensure that the
8538 literal pools are within the range of PC-relative load offsets. A valid
8539 interleaving might place the @code{.literal} sections from an initial
8540 group of files followed by the @code{.text} sections of that group of
8541 files. Then, the @code{.literal} sections from the rest of the files
8542 and the @code{.text} sections from the rest of the files would follow.
8544 @cindex @option{--relax} on Xtensa
8545 @cindex relaxing on Xtensa
8546 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8547 provides two important link-time optimizations. The first optimization
8548 is to combine identical literal values to reduce code size. A redundant
8549 literal will be removed and all the @code{L32R} instructions that use it
8550 will be changed to reference an identical literal, as long as the
8551 location of the replacement literal is within the offset range of all
8552 the @code{L32R} instructions. The second optimization is to remove
8553 unnecessary overhead from assembler-generated ``longcall'' sequences of
8554 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8555 range of direct @code{CALL@var{n}} instructions.
8557 For each of these cases where an indirect call sequence can be optimized
8558 to a direct call, the linker will change the @code{CALLX@var{n}}
8559 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8560 instruction, and remove the literal referenced by the @code{L32R}
8561 instruction if it is not used for anything else. Removing the
8562 @code{L32R} instruction always reduces code size but can potentially
8563 hurt performance by changing the alignment of subsequent branch targets.
8564 By default, the linker will always preserve alignments, either by
8565 switching some instructions between 24-bit encodings and the equivalent
8566 density instructions or by inserting a no-op in place of the @code{L32R}
8567 instruction that was removed. If code size is more important than
8568 performance, the @option{--size-opt} option can be used to prevent the
8569 linker from widening density instructions or inserting no-ops, except in
8570 a few cases where no-ops are required for correctness.
8572 The following Xtensa-specific command-line options can be used to
8575 @cindex Xtensa options
8578 When optimizing indirect calls to direct calls, optimize for code size
8579 more than performance. With this option, the linker will not insert
8580 no-ops or widen density instructions to preserve branch target
8581 alignment. There may still be some cases where no-ops are required to
8582 preserve the correctness of the code.
8584 @item --abi-windowed
8586 Choose ABI for the output object and for the generated PLT code.
8587 PLT code inserted by the linker must match ABI of the output object
8588 because windowed and call0 ABI use incompatible function call
8590 Default ABI is chosen by the ABI tag in the @code{.xtensa.info} section
8591 of the first input object.
8592 A warning is issued if ABI tags of input objects do not match each other
8593 or the chosen output object ABI.
8601 @ifclear SingleFormat
8606 @cindex object file management
8607 @cindex object formats available
8609 The linker accesses object and archive files using the BFD libraries.
8610 These libraries allow the linker to use the same routines to operate on
8611 object files whatever the object file format. A different object file
8612 format can be supported simply by creating a new BFD back end and adding
8613 it to the library. To conserve runtime memory, however, the linker and
8614 associated tools are usually configured to support only a subset of the
8615 object file formats available. You can use @code{objdump -i}
8616 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8617 list all the formats available for your configuration.
8619 @cindex BFD requirements
8620 @cindex requirements for BFD
8621 As with most implementations, BFD is a compromise between
8622 several conflicting requirements. The major factor influencing
8623 BFD design was efficiency: any time used converting between
8624 formats is time which would not have been spent had BFD not
8625 been involved. This is partly offset by abstraction payback; since
8626 BFD simplifies applications and back ends, more time and care
8627 may be spent optimizing algorithms for a greater speed.
8629 One minor artifact of the BFD solution which you should bear in
8630 mind is the potential for information loss. There are two places where
8631 useful information can be lost using the BFD mechanism: during
8632 conversion and during output. @xref{BFD information loss}.
8635 * BFD outline:: How it works: an outline of BFD
8639 @section How It Works: An Outline of BFD
8640 @cindex opening object files
8641 @include bfdsumm.texi
8644 @node Reporting Bugs
8645 @chapter Reporting Bugs
8646 @cindex bugs in @command{ld}
8647 @cindex reporting bugs in @command{ld}
8649 Your bug reports play an essential role in making @command{ld} reliable.
8651 Reporting a bug may help you by bringing a solution to your problem, or
8652 it may not. But in any case the principal function of a bug report is
8653 to help the entire community by making the next version of @command{ld}
8654 work better. Bug reports are your contribution to the maintenance of
8657 In order for a bug report to serve its purpose, you must include the
8658 information that enables us to fix the bug.
8661 * Bug Criteria:: Have you found a bug?
8662 * Bug Reporting:: How to report bugs
8666 @section Have You Found a Bug?
8667 @cindex bug criteria
8669 If you are not sure whether you have found a bug, here are some guidelines:
8672 @cindex fatal signal
8673 @cindex linker crash
8674 @cindex crash of linker
8676 If the linker gets a fatal signal, for any input whatever, that is a
8677 @command{ld} bug. Reliable linkers never crash.
8679 @cindex error on valid input
8681 If @command{ld} produces an error message for valid input, that is a bug.
8683 @cindex invalid input
8685 If @command{ld} does not produce an error message for invalid input, that
8686 may be a bug. In the general case, the linker can not verify that
8687 object files are correct.
8690 If you are an experienced user of linkers, your suggestions for
8691 improvement of @command{ld} are welcome in any case.
8695 @section How to Report Bugs
8697 @cindex @command{ld} bugs, reporting
8699 A number of companies and individuals offer support for @sc{gnu}
8700 products. If you obtained @command{ld} from a support organization, we
8701 recommend you contact that organization first.
8703 You can find contact information for many support companies and
8704 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
8708 Otherwise, send bug reports for @command{ld} to
8712 The fundamental principle of reporting bugs usefully is this:
8713 @strong{report all the facts}. If you are not sure whether to state a
8714 fact or leave it out, state it!
8716 Often people omit facts because they think they know what causes the
8717 problem and assume that some details do not matter. Thus, you might
8718 assume that the name of a symbol you use in an example does not
8719 matter. Well, probably it does not, but one cannot be sure. Perhaps
8720 the bug is a stray memory reference which happens to fetch from the
8721 location where that name is stored in memory; perhaps, if the name
8722 were different, the contents of that location would fool the linker
8723 into doing the right thing despite the bug. Play it safe and give a
8724 specific, complete example. That is the easiest thing for you to do,
8725 and the most helpful.
8727 Keep in mind that the purpose of a bug report is to enable us to fix
8728 the bug if it is new to us. Therefore, always write your bug reports
8729 on the assumption that the bug has not been reported previously.
8731 Sometimes people give a few sketchy facts and ask, ``Does this ring a
8732 bell?'' This cannot help us fix a bug, so it is basically useless. We
8733 respond by asking for enough details to enable us to investigate.
8734 You might as well expedite matters by sending them to begin with.
8736 To enable us to fix the bug, you should include all these things:
8740 The version of @command{ld}. @command{ld} announces it if you start it with
8741 the @samp{--version} argument.
8743 Without this, we will not know whether there is any point in looking for
8744 the bug in the current version of @command{ld}.
8747 Any patches you may have applied to the @command{ld} source, including any
8748 patches made to the @code{BFD} library.
8751 The type of machine you are using, and the operating system name and
8755 What compiler (and its version) was used to compile @command{ld}---e.g.
8759 The command arguments you gave the linker to link your example and
8760 observe the bug. To guarantee you will not omit something important,
8761 list them all. A copy of the Makefile (or the output from make) is
8764 If we were to try to guess the arguments, we would probably guess wrong
8765 and then we might not encounter the bug.
8768 A complete input file, or set of input files, that will reproduce the
8769 bug. It is generally most helpful to send the actual object files
8770 provided that they are reasonably small. Say no more than 10K. For
8771 bigger files you can either make them available by FTP or HTTP or else
8772 state that you are willing to send the object file(s) to whomever
8773 requests them. (Note - your email will be going to a mailing list, so
8774 we do not want to clog it up with large attachments). But small
8775 attachments are best.
8777 If the source files were assembled using @code{gas} or compiled using
8778 @code{gcc}, then it may be OK to send the source files rather than the
8779 object files. In this case, be sure to say exactly what version of
8780 @code{gas} or @code{gcc} was used to produce the object files. Also say
8781 how @code{gas} or @code{gcc} were configured.
8784 A description of what behavior you observe that you believe is
8785 incorrect. For example, ``It gets a fatal signal.''
8787 Of course, if the bug is that @command{ld} gets a fatal signal, then we
8788 will certainly notice it. But if the bug is incorrect output, we might
8789 not notice unless it is glaringly wrong. You might as well not give us
8790 a chance to make a mistake.
8792 Even if the problem you experience is a fatal signal, you should still
8793 say so explicitly. Suppose something strange is going on, such as, your
8794 copy of @command{ld} is out of sync, or you have encountered a bug in the
8795 C library on your system. (This has happened!) Your copy might crash
8796 and ours would not. If you told us to expect a crash, then when ours
8797 fails to crash, we would know that the bug was not happening for us. If
8798 you had not told us to expect a crash, then we would not be able to draw
8799 any conclusion from our observations.
8802 If you wish to suggest changes to the @command{ld} source, send us context
8803 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
8804 @samp{-p} option. Always send diffs from the old file to the new file.
8805 If you even discuss something in the @command{ld} source, refer to it by
8806 context, not by line number.
8808 The line numbers in our development sources will not match those in your
8809 sources. Your line numbers would convey no useful information to us.
8812 Here are some things that are not necessary:
8816 A description of the envelope of the bug.
8818 Often people who encounter a bug spend a lot of time investigating
8819 which changes to the input file will make the bug go away and which
8820 changes will not affect it.
8822 This is often time consuming and not very useful, because the way we
8823 will find the bug is by running a single example under the debugger
8824 with breakpoints, not by pure deduction from a series of examples.
8825 We recommend that you save your time for something else.
8827 Of course, if you can find a simpler example to report @emph{instead}
8828 of the original one, that is a convenience for us. Errors in the
8829 output will be easier to spot, running under the debugger will take
8830 less time, and so on.
8832 However, simplification is not vital; if you do not want to do this,
8833 report the bug anyway and send us the entire test case you used.
8836 A patch for the bug.
8838 A patch for the bug does help us if it is a good one. But do not omit
8839 the necessary information, such as the test case, on the assumption that
8840 a patch is all we need. We might see problems with your patch and decide
8841 to fix the problem another way, or we might not understand it at all.
8843 Sometimes with a program as complicated as @command{ld} it is very hard to
8844 construct an example that will make the program follow a certain path
8845 through the code. If you do not send us the example, we will not be
8846 able to construct one, so we will not be able to verify that the bug is
8849 And if we cannot understand what bug you are trying to fix, or why your
8850 patch should be an improvement, we will not install it. A test case will
8851 help us to understand.
8854 A guess about what the bug is or what it depends on.
8856 Such guesses are usually wrong. Even we cannot guess right about such
8857 things without first using the debugger to find the facts.
8861 @appendix MRI Compatible Script Files
8862 @cindex MRI compatibility
8863 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
8864 linker, @command{ld} can use MRI compatible linker scripts as an
8865 alternative to the more general-purpose linker scripting language
8866 described in @ref{Scripts}. MRI compatible linker scripts have a much
8867 simpler command set than the scripting language otherwise used with
8868 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
8869 linker commands; these commands are described here.
8871 In general, MRI scripts aren't of much use with the @code{a.out} object
8872 file format, since it only has three sections and MRI scripts lack some
8873 features to make use of them.
8875 You can specify a file containing an MRI-compatible script using the
8876 @samp{-c} command-line option.
8878 Each command in an MRI-compatible script occupies its own line; each
8879 command line starts with the keyword that identifies the command (though
8880 blank lines are also allowed for punctuation). If a line of an
8881 MRI-compatible script begins with an unrecognized keyword, @command{ld}
8882 issues a warning message, but continues processing the script.
8884 Lines beginning with @samp{*} are comments.
8886 You can write these commands using all upper-case letters, or all
8887 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
8888 The following list shows only the upper-case form of each command.
8891 @cindex @code{ABSOLUTE} (MRI)
8892 @item ABSOLUTE @var{secname}
8893 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
8894 Normally, @command{ld} includes in the output file all sections from all
8895 the input files. However, in an MRI-compatible script, you can use the
8896 @code{ABSOLUTE} command to restrict the sections that will be present in
8897 your output program. If the @code{ABSOLUTE} command is used at all in a
8898 script, then only the sections named explicitly in @code{ABSOLUTE}
8899 commands will appear in the linker output. You can still use other
8900 input sections (whatever you select on the command line, or using
8901 @code{LOAD}) to resolve addresses in the output file.
8903 @cindex @code{ALIAS} (MRI)
8904 @item ALIAS @var{out-secname}, @var{in-secname}
8905 Use this command to place the data from input section @var{in-secname}
8906 in a section called @var{out-secname} in the linker output file.
8908 @var{in-secname} may be an integer.
8910 @cindex @code{ALIGN} (MRI)
8911 @item ALIGN @var{secname} = @var{expression}
8912 Align the section called @var{secname} to @var{expression}. The
8913 @var{expression} should be a power of two.
8915 @cindex @code{BASE} (MRI)
8916 @item BASE @var{expression}
8917 Use the value of @var{expression} as the lowest address (other than
8918 absolute addresses) in the output file.
8920 @cindex @code{CHIP} (MRI)
8921 @item CHIP @var{expression}
8922 @itemx CHIP @var{expression}, @var{expression}
8923 This command does nothing; it is accepted only for compatibility.
8925 @cindex @code{END} (MRI)
8927 This command does nothing whatever; it's only accepted for compatibility.
8929 @cindex @code{FORMAT} (MRI)
8930 @item FORMAT @var{output-format}
8931 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
8932 language, but restricted to S-records, if @var{output-format} is @samp{S}
8934 @cindex @code{LIST} (MRI)
8935 @item LIST @var{anything}@dots{}
8936 Print (to the standard output file) a link map, as produced by the
8937 @command{ld} command-line option @samp{-M}.
8939 The keyword @code{LIST} may be followed by anything on the
8940 same line, with no change in its effect.
8942 @cindex @code{LOAD} (MRI)
8943 @item LOAD @var{filename}
8944 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
8945 Include one or more object file @var{filename} in the link; this has the
8946 same effect as specifying @var{filename} directly on the @command{ld}
8949 @cindex @code{NAME} (MRI)
8950 @item NAME @var{output-name}
8951 @var{output-name} is the name for the program produced by @command{ld}; the
8952 MRI-compatible command @code{NAME} is equivalent to the command-line
8953 option @samp{-o} or the general script language command @code{OUTPUT}.
8955 @cindex @code{ORDER} (MRI)
8956 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
8957 @itemx ORDER @var{secname} @var{secname} @var{secname}
8958 Normally, @command{ld} orders the sections in its output file in the
8959 order in which they first appear in the input files. In an MRI-compatible
8960 script, you can override this ordering with the @code{ORDER} command. The
8961 sections you list with @code{ORDER} will appear first in your output
8962 file, in the order specified.
8964 @cindex @code{PUBLIC} (MRI)
8965 @item PUBLIC @var{name}=@var{expression}
8966 @itemx PUBLIC @var{name},@var{expression}
8967 @itemx PUBLIC @var{name} @var{expression}
8968 Supply a value (@var{expression}) for external symbol
8969 @var{name} used in the linker input files.
8971 @cindex @code{SECT} (MRI)
8972 @item SECT @var{secname}, @var{expression}
8973 @itemx SECT @var{secname}=@var{expression}
8974 @itemx SECT @var{secname} @var{expression}
8975 You can use any of these three forms of the @code{SECT} command to
8976 specify the start address (@var{expression}) for section @var{secname}.
8977 If you have more than one @code{SECT} statement for the same
8978 @var{secname}, only the @emph{first} sets the start address.
8981 @node GNU Free Documentation License
8982 @appendix GNU Free Documentation License
8986 @unnumbered LD Index
8991 % I think something like @@colophon should be in texinfo. In the
8993 \long\def\colophon{\hbox to0pt{}\vfill
8994 \centerline{The body of this manual is set in}
8995 \centerline{\fontname\tenrm,}
8996 \centerline{with headings in {\bf\fontname\tenbf}}
8997 \centerline{and examples in {\tt\fontname\tentt}.}
8998 \centerline{{\it\fontname\tenit\/} and}
8999 \centerline{{\sl\fontname\tensl\/}}
9000 \centerline{are used for emphasis.}\vfill}
9002 % Blame: doc@@cygnus.com, 28mar91.