3 @c Copyright (C) 1991-2018 Free Software Foundation, Inc.
6 @include configdoc.texi
7 @c (configdoc.texi is generated by the Makefile)
13 @macro gcctabopt{body}
19 @c Configure for the generation of man pages
46 @dircategory Software development
48 * Ld: (ld). The GNU linker.
53 This file documents the @sc{gnu} linker LD
54 @ifset VERSION_PACKAGE
55 @value{VERSION_PACKAGE}
57 version @value{VERSION}.
59 Copyright @copyright{} 1991-2018 Free Software Foundation, Inc.
61 Permission is granted to copy, distribute and/or modify this document
62 under the terms of the GNU Free Documentation License, Version 1.3
63 or any later version published by the Free Software Foundation;
64 with no Invariant Sections, with no Front-Cover Texts, and with no
65 Back-Cover Texts. A copy of the license is included in the
66 section entitled ``GNU Free Documentation License''.
70 @setchapternewpage odd
71 @settitle The GNU linker
76 @ifset VERSION_PACKAGE
77 @subtitle @value{VERSION_PACKAGE}
79 @subtitle Version @value{VERSION}
80 @author Steve Chamberlain
81 @author Ian Lance Taylor
86 \hfill Red Hat Inc\par
87 \hfill nickc\@credhat.com, doc\@redhat.com\par
88 \hfill {\it The GNU linker}\par
89 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
91 \global\parindent=0pt % Steve likes it this way.
94 @vskip 0pt plus 1filll
95 @c man begin COPYRIGHT
96 Copyright @copyright{} 1991-2018 Free Software Foundation, Inc.
98 Permission is granted to copy, distribute and/or modify this document
99 under the terms of the GNU Free Documentation License, Version 1.3
100 or any later version published by the Free Software Foundation;
101 with no Invariant Sections, with no Front-Cover Texts, and with no
102 Back-Cover Texts. A copy of the license is included in the
103 section entitled ``GNU Free Documentation License''.
109 @c FIXME: Talk about importance of *order* of args, cmds to linker!
114 This file documents the @sc{gnu} linker ld
115 @ifset VERSION_PACKAGE
116 @value{VERSION_PACKAGE}
118 version @value{VERSION}.
120 This document is distributed under the terms of the GNU Free
121 Documentation License version 1.3. A copy of the license is included
122 in the section entitled ``GNU Free Documentation License''.
125 * Overview:: Overview
126 * Invocation:: Invocation
127 * Scripts:: Linker Scripts
129 * Machine Dependent:: Machine Dependent Features
133 * H8/300:: ld and the H8/300
136 * Renesas:: ld and other Renesas micros
139 * i960:: ld and the Intel 960 family
142 * ARM:: ld and the ARM family
145 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
148 * HPPA ELF32:: ld and HPPA 32-bit ELF
151 * M68K:: ld and Motorola 68K family
154 * MIPS:: ld and MIPS family
157 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
160 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
163 * S/390 ELF:: ld and S/390 ELF Support
166 * SPU ELF:: ld and SPU ELF Support
169 * TI COFF:: ld and the TI COFF
172 * Win32:: ld and WIN32 (cygwin/mingw)
175 * Xtensa:: ld and Xtensa Processors
178 @ifclear SingleFormat
181 @c Following blank line required for remaining bug in makeinfo conds/menus
183 * Reporting Bugs:: Reporting Bugs
184 * MRI:: MRI Compatible Script Files
185 * GNU Free Documentation License:: GNU Free Documentation License
186 * LD Index:: LD Index
193 @cindex @sc{gnu} linker
194 @cindex what is this?
197 @c man begin SYNOPSIS
198 ld [@b{options}] @var{objfile} @dots{}
202 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
203 the Info entries for @file{binutils} and
208 @c man begin DESCRIPTION
210 @command{ld} combines a number of object and archive files, relocates
211 their data and ties up symbol references. Usually the last step in
212 compiling a program is to run @command{ld}.
214 @command{ld} accepts Linker Command Language files written in
215 a superset of AT&T's Link Editor Command Language syntax,
216 to provide explicit and total control over the linking process.
220 This man page does not describe the command language; see the
221 @command{ld} entry in @code{info} for full details on the command
222 language and on other aspects of the GNU linker.
225 @ifclear SingleFormat
226 This version of @command{ld} uses the general purpose BFD libraries
227 to operate on object files. This allows @command{ld} to read, combine, and
228 write object files in many different formats---for example, COFF or
229 @code{a.out}. Different formats may be linked together to produce any
230 available kind of object file. @xref{BFD}, for more information.
233 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
234 linkers in providing diagnostic information. Many linkers abandon
235 execution immediately upon encountering an error; whenever possible,
236 @command{ld} continues executing, allowing you to identify other errors
237 (or, in some cases, to get an output file in spite of the error).
244 @c man begin DESCRIPTION
246 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
247 and to be as compatible as possible with other linkers. As a result,
248 you have many choices to control its behavior.
254 * Options:: Command Line Options
255 * Environment:: Environment Variables
259 @section Command Line Options
267 The linker supports a plethora of command-line options, but in actual
268 practice few of them are used in any particular context.
269 @cindex standard Unix system
270 For instance, a frequent use of @command{ld} is to link standard Unix
271 object files on a standard, supported Unix system. On such a system, to
272 link a file @code{hello.o}:
275 ld -o @var{output} /lib/crt0.o hello.o -lc
278 This tells @command{ld} to produce a file called @var{output} as the
279 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
280 the library @code{libc.a}, which will come from the standard search
281 directories. (See the discussion of the @samp{-l} option below.)
283 Some of the command-line options to @command{ld} may be specified at any
284 point in the command line. However, options which refer to files, such
285 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
286 which the option appears in the command line, relative to the object
287 files and other file options. Repeating non-file options with a
288 different argument will either have no further effect, or override prior
289 occurrences (those further to the left on the command line) of that
290 option. Options which may be meaningfully specified more than once are
291 noted in the descriptions below.
294 Non-option arguments are object files or archives which are to be linked
295 together. They may follow, precede, or be mixed in with command-line
296 options, except that an object file argument may not be placed between
297 an option and its argument.
299 Usually the linker is invoked with at least one object file, but you can
300 specify other forms of binary input files using @samp{-l}, @samp{-R},
301 and the script command language. If @emph{no} binary input files at all
302 are specified, the linker does not produce any output, and issues the
303 message @samp{No input files}.
305 If the linker cannot recognize the format of an object file, it will
306 assume that it is a linker script. A script specified in this way
307 augments the main linker script used for the link (either the default
308 linker script or the one specified by using @samp{-T}). This feature
309 permits the linker to link against a file which appears to be an object
310 or an archive, but actually merely defines some symbol values, or uses
311 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
312 script in this way merely augments the main linker script, with the
313 extra commands placed after the main script; use the @samp{-T} option
314 to replace the default linker script entirely, but note the effect of
315 the @code{INSERT} command. @xref{Scripts}.
317 For options whose names are a single letter,
318 option arguments must either follow the option letter without intervening
319 whitespace, or be given as separate arguments immediately following the
320 option that requires them.
322 For options whose names are multiple letters, either one dash or two can
323 precede the option name; for example, @samp{-trace-symbol} and
324 @samp{--trace-symbol} are equivalent. Note---there is one exception to
325 this rule. Multiple letter options that start with a lower case 'o' can
326 only be preceded by two dashes. This is to reduce confusion with the
327 @samp{-o} option. So for example @samp{-omagic} sets the output file
328 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
331 Arguments to multiple-letter options must either be separated from the
332 option name by an equals sign, or be given as separate arguments
333 immediately following the option that requires them. For example,
334 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
335 Unique abbreviations of the names of multiple-letter options are
338 Note---if the linker is being invoked indirectly, via a compiler driver
339 (e.g. @samp{gcc}) then all the linker command line options should be
340 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
341 compiler driver) like this:
344 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
347 This is important, because otherwise the compiler driver program may
348 silently drop the linker options, resulting in a bad link. Confusion
349 may also arise when passing options that require values through a
350 driver, as the use of a space between option and argument acts as
351 a separator, and causes the driver to pass only the option to the linker
352 and the argument to the compiler. In this case, it is simplest to use
353 the joined forms of both single- and multiple-letter options, such as:
356 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
359 Here is a table of the generic command line switches accepted by the GNU
363 @include at-file.texi
365 @kindex -a @var{keyword}
366 @item -a @var{keyword}
367 This option is supported for HP/UX compatibility. The @var{keyword}
368 argument must be one of the strings @samp{archive}, @samp{shared}, or
369 @samp{default}. @samp{-aarchive} is functionally equivalent to
370 @samp{-Bstatic}, and the other two keywords are functionally equivalent
371 to @samp{-Bdynamic}. This option may be used any number of times.
373 @kindex --audit @var{AUDITLIB}
374 @item --audit @var{AUDITLIB}
375 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
376 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
377 specified in the library. If specified multiple times @code{DT_AUDIT}
378 will contain a colon separated list of audit interfaces to use. If the linker
379 finds an object with an audit entry while searching for shared libraries,
380 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
381 This option is only meaningful on ELF platforms supporting the rtld-audit
385 @cindex architectures
386 @kindex -A @var{arch}
387 @item -A @var{architecture}
388 @kindex --architecture=@var{arch}
389 @itemx --architecture=@var{architecture}
390 In the current release of @command{ld}, this option is useful only for the
391 Intel 960 family of architectures. In that @command{ld} configuration, the
392 @var{architecture} argument identifies the particular architecture in
393 the 960 family, enabling some safeguards and modifying the
394 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
395 family}, for details.
397 Future releases of @command{ld} may support similar functionality for
398 other architecture families.
401 @ifclear SingleFormat
402 @cindex binary input format
403 @kindex -b @var{format}
404 @kindex --format=@var{format}
407 @item -b @var{input-format}
408 @itemx --format=@var{input-format}
409 @command{ld} may be configured to support more than one kind of object
410 file. If your @command{ld} is configured this way, you can use the
411 @samp{-b} option to specify the binary format for input object files
412 that follow this option on the command line. Even when @command{ld} is
413 configured to support alternative object formats, you don't usually need
414 to specify this, as @command{ld} should be configured to expect as a
415 default input format the most usual format on each machine.
416 @var{input-format} is a text string, the name of a particular format
417 supported by the BFD libraries. (You can list the available binary
418 formats with @samp{objdump -i}.)
421 You may want to use this option if you are linking files with an unusual
422 binary format. You can also use @samp{-b} to switch formats explicitly (when
423 linking object files of different formats), by including
424 @samp{-b @var{input-format}} before each group of object files in a
427 The default format is taken from the environment variable
432 You can also define the input format from a script, using the command
435 see @ref{Format Commands}.
439 @kindex -c @var{MRI-cmdfile}
440 @kindex --mri-script=@var{MRI-cmdfile}
441 @cindex compatibility, MRI
442 @item -c @var{MRI-commandfile}
443 @itemx --mri-script=@var{MRI-commandfile}
444 For compatibility with linkers produced by MRI, @command{ld} accepts script
445 files written in an alternate, restricted command language, described in
447 @ref{MRI,,MRI Compatible Script Files}.
450 the MRI Compatible Script Files section of GNU ld documentation.
452 Introduce MRI script files with
453 the option @samp{-c}; use the @samp{-T} option to run linker
454 scripts written in the general-purpose @command{ld} scripting language.
455 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
456 specified by any @samp{-L} options.
458 @cindex common allocation
465 These three options are equivalent; multiple forms are supported for
466 compatibility with other linkers. They assign space to common symbols
467 even if a relocatable output file is specified (with @samp{-r}). The
468 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
469 @xref{Miscellaneous Commands}.
471 @kindex --depaudit @var{AUDITLIB}
472 @kindex -P @var{AUDITLIB}
473 @item --depaudit @var{AUDITLIB}
474 @itemx -P @var{AUDITLIB}
475 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
476 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
477 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
478 will contain a colon separated list of audit interfaces to use. This
479 option is only meaningful on ELF platforms supporting the rtld-audit interface.
480 The -P option is provided for Solaris compatibility.
482 @cindex entry point, from command line
483 @kindex -e @var{entry}
484 @kindex --entry=@var{entry}
486 @itemx --entry=@var{entry}
487 Use @var{entry} as the explicit symbol for beginning execution of your
488 program, rather than the default entry point. If there is no symbol
489 named @var{entry}, the linker will try to parse @var{entry} as a number,
490 and use that as the entry address (the number will be interpreted in
491 base 10; you may use a leading @samp{0x} for base 16, or a leading
492 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
493 and other ways of specifying the entry point.
495 @kindex --exclude-libs
496 @item --exclude-libs @var{lib},@var{lib},...
497 Specifies a list of archive libraries from which symbols should not be automatically
498 exported. The library names may be delimited by commas or colons. Specifying
499 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
500 automatic export. This option is available only for the i386 PE targeted
501 port of the linker and for ELF targeted ports. For i386 PE, symbols
502 explicitly listed in a .def file are still exported, regardless of this
503 option. For ELF targeted ports, symbols affected by this option will
504 be treated as hidden.
506 @kindex --exclude-modules-for-implib
507 @item --exclude-modules-for-implib @var{module},@var{module},...
508 Specifies a list of object files or archive members, from which symbols
509 should not be automatically exported, but which should be copied wholesale
510 into the import library being generated during the link. The module names
511 may be delimited by commas or colons, and must match exactly the filenames
512 used by @command{ld} to open the files; for archive members, this is simply
513 the member name, but for object files the name listed must include and
514 match precisely any path used to specify the input file on the linker's
515 command-line. This option is available only for the i386 PE targeted port
516 of the linker. Symbols explicitly listed in a .def file are still exported,
517 regardless of this option.
519 @cindex dynamic symbol table
521 @kindex --export-dynamic
522 @kindex --no-export-dynamic
524 @itemx --export-dynamic
525 @itemx --no-export-dynamic
526 When creating a dynamically linked executable, using the @option{-E}
527 option or the @option{--export-dynamic} option causes the linker to add
528 all symbols to the dynamic symbol table. The dynamic symbol table is the
529 set of symbols which are visible from dynamic objects at run time.
531 If you do not use either of these options (or use the
532 @option{--no-export-dynamic} option to restore the default behavior), the
533 dynamic symbol table will normally contain only those symbols which are
534 referenced by some dynamic object mentioned in the link.
536 If you use @code{dlopen} to load a dynamic object which needs to refer
537 back to the symbols defined by the program, rather than some other
538 dynamic object, then you will probably need to use this option when
539 linking the program itself.
541 You can also use the dynamic list to control what symbols should
542 be added to the dynamic symbol table if the output format supports it.
543 See the description of @samp{--dynamic-list}.
545 Note that this option is specific to ELF targeted ports. PE targets
546 support a similar function to export all symbols from a DLL or EXE; see
547 the description of @samp{--export-all-symbols} below.
549 @ifclear SingleFormat
550 @cindex big-endian objects
554 Link big-endian objects. This affects the default output format.
556 @cindex little-endian objects
559 Link little-endian objects. This affects the default output format.
562 @kindex -f @var{name}
563 @kindex --auxiliary=@var{name}
565 @itemx --auxiliary=@var{name}
566 When creating an ELF shared object, set the internal DT_AUXILIARY field
567 to the specified name. This tells the dynamic linker that the symbol
568 table of the shared object should be used as an auxiliary filter on the
569 symbol table of the shared object @var{name}.
571 If you later link a program against this filter object, then, when you
572 run the program, the dynamic linker will see the DT_AUXILIARY field. If
573 the dynamic linker resolves any symbols from the filter object, it will
574 first check whether there is a definition in the shared object
575 @var{name}. If there is one, it will be used instead of the definition
576 in the filter object. The shared object @var{name} need not exist.
577 Thus the shared object @var{name} may be used to provide an alternative
578 implementation of certain functions, perhaps for debugging or for
579 machine specific performance.
581 This option may be specified more than once. The DT_AUXILIARY entries
582 will be created in the order in which they appear on the command line.
584 @kindex -F @var{name}
585 @kindex --filter=@var{name}
587 @itemx --filter=@var{name}
588 When creating an ELF shared object, set the internal DT_FILTER field to
589 the specified name. This tells the dynamic linker that the symbol table
590 of the shared object which is being created should be used as a filter
591 on the symbol table of the shared object @var{name}.
593 If you later link a program against this filter object, then, when you
594 run the program, the dynamic linker will see the DT_FILTER field. The
595 dynamic linker will resolve symbols according to the symbol table of the
596 filter object as usual, but it will actually link to the definitions
597 found in the shared object @var{name}. Thus the filter object can be
598 used to select a subset of the symbols provided by the object
601 Some older linkers used the @option{-F} option throughout a compilation
602 toolchain for specifying object-file format for both input and output
604 @ifclear SingleFormat
605 The @sc{gnu} linker uses other mechanisms for this purpose: the
606 @option{-b}, @option{--format}, @option{--oformat} options, the
607 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
608 environment variable.
610 The @sc{gnu} linker will ignore the @option{-F} option when not
611 creating an ELF shared object.
613 @cindex finalization function
614 @kindex -fini=@var{name}
615 @item -fini=@var{name}
616 When creating an ELF executable or shared object, call NAME when the
617 executable or shared object is unloaded, by setting DT_FINI to the
618 address of the function. By default, the linker uses @code{_fini} as
619 the function to call.
623 Ignored. Provided for compatibility with other tools.
625 @kindex -G @var{value}
626 @kindex --gpsize=@var{value}
629 @itemx --gpsize=@var{value}
630 Set the maximum size of objects to be optimized using the GP register to
631 @var{size}. This is only meaningful for object file formats such as
632 MIPS ELF that support putting large and small objects into different
633 sections. This is ignored for other object file formats.
635 @cindex runtime library name
636 @kindex -h @var{name}
637 @kindex -soname=@var{name}
639 @itemx -soname=@var{name}
640 When creating an ELF shared object, set the internal DT_SONAME field to
641 the specified name. When an executable is linked with a shared object
642 which has a DT_SONAME field, then when the executable is run the dynamic
643 linker will attempt to load the shared object specified by the DT_SONAME
644 field rather than the using the file name given to the linker.
647 @cindex incremental link
649 Perform an incremental link (same as option @samp{-r}).
651 @cindex initialization function
652 @kindex -init=@var{name}
653 @item -init=@var{name}
654 When creating an ELF executable or shared object, call NAME when the
655 executable or shared object is loaded, by setting DT_INIT to the address
656 of the function. By default, the linker uses @code{_init} as the
659 @cindex archive files, from cmd line
660 @kindex -l @var{namespec}
661 @kindex --library=@var{namespec}
662 @item -l @var{namespec}
663 @itemx --library=@var{namespec}
664 Add the archive or object file specified by @var{namespec} to the
665 list of files to link. This option may be used any number of times.
666 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
667 will search the library path for a file called @var{filename}, otherwise it
668 will search the library path for a file called @file{lib@var{namespec}.a}.
670 On systems which support shared libraries, @command{ld} may also search for
671 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
672 and SunOS systems, @command{ld} will search a directory for a library
673 called @file{lib@var{namespec}.so} before searching for one called
674 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
675 indicates a shared library.) Note that this behavior does not apply
676 to @file{:@var{filename}}, which always specifies a file called
679 The linker will search an archive only once, at the location where it is
680 specified on the command line. If the archive defines a symbol which
681 was undefined in some object which appeared before the archive on the
682 command line, the linker will include the appropriate file(s) from the
683 archive. However, an undefined symbol in an object appearing later on
684 the command line will not cause the linker to search the archive again.
686 See the @option{-(} option for a way to force the linker to search
687 archives multiple times.
689 You may list the same archive multiple times on the command line.
692 This type of archive searching is standard for Unix linkers. However,
693 if you are using @command{ld} on AIX, note that it is different from the
694 behaviour of the AIX linker.
697 @cindex search directory, from cmd line
699 @kindex --library-path=@var{dir}
700 @item -L @var{searchdir}
701 @itemx --library-path=@var{searchdir}
702 Add path @var{searchdir} to the list of paths that @command{ld} will search
703 for archive libraries and @command{ld} control scripts. You may use this
704 option any number of times. The directories are searched in the order
705 in which they are specified on the command line. Directories specified
706 on the command line are searched before the default directories. All
707 @option{-L} options apply to all @option{-l} options, regardless of the
708 order in which the options appear. @option{-L} options do not affect
709 how @command{ld} searches for a linker script unless @option{-T}
712 If @var{searchdir} begins with @code{=} or @code{$SYSROOT}, then this
713 prefix will be replaced by the @dfn{sysroot prefix}, controlled by the
714 @samp{--sysroot} option, or specified when the linker is configured.
717 The default set of paths searched (without being specified with
718 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
719 some cases also on how it was configured. @xref{Environment}.
722 The paths can also be specified in a link script with the
723 @code{SEARCH_DIR} command. Directories specified this way are searched
724 at the point in which the linker script appears in the command line.
727 @kindex -m @var{emulation}
728 @item -m @var{emulation}
729 Emulate the @var{emulation} linker. You can list the available
730 emulations with the @samp{--verbose} or @samp{-V} options.
732 If the @samp{-m} option is not used, the emulation is taken from the
733 @code{LDEMULATION} environment variable, if that is defined.
735 Otherwise, the default emulation depends upon how the linker was
743 Print a link map to the standard output. A link map provides
744 information about the link, including the following:
748 Where object files are mapped into memory.
750 How common symbols are allocated.
752 All archive members included in the link, with a mention of the symbol
753 which caused the archive member to be brought in.
755 The values assigned to symbols.
757 Note - symbols whose values are computed by an expression which
758 involves a reference to a previous value of the same symbol may not
759 have correct result displayed in the link map. This is because the
760 linker discards intermediate results and only retains the final value
761 of an expression. Under such circumstances the linker will display
762 the final value enclosed by square brackets. Thus for example a
763 linker script containing:
771 will produce the following output in the link map if the @option{-M}
776 [0x0000000c] foo = (foo * 0x4)
777 [0x0000000c] foo = (foo + 0x8)
780 See @ref{Expressions} for more information about expressions in linker
785 @cindex read-only text
790 Turn off page alignment of sections, and disable linking against shared
791 libraries. If the output format supports Unix style magic numbers,
792 mark the output as @code{NMAGIC}.
796 @cindex read/write from cmd line
800 Set the text and data sections to be readable and writable. Also, do
801 not page-align the data segment, and disable linking against shared
802 libraries. If the output format supports Unix style magic numbers,
803 mark the output as @code{OMAGIC}. Note: Although a writable text section
804 is allowed for PE-COFF targets, it does not conform to the format
805 specification published by Microsoft.
810 This option negates most of the effects of the @option{-N} option. It
811 sets the text section to be read-only, and forces the data segment to
812 be page-aligned. Note - this option does not enable linking against
813 shared libraries. Use @option{-Bdynamic} for this.
815 @kindex -o @var{output}
816 @kindex --output=@var{output}
817 @cindex naming the output file
818 @item -o @var{output}
819 @itemx --output=@var{output}
820 Use @var{output} as the name for the program produced by @command{ld}; if this
821 option is not specified, the name @file{a.out} is used by default. The
822 script command @code{OUTPUT} can also specify the output file name.
824 @kindex -O @var{level}
825 @cindex generating optimized output
827 If @var{level} is a numeric values greater than zero @command{ld} optimizes
828 the output. This might take significantly longer and therefore probably
829 should only be enabled for the final binary. At the moment this
830 option only affects ELF shared library generation. Future releases of
831 the linker may make more use of this option. Also currently there is
832 no difference in the linker's behaviour for different non-zero values
833 of this option. Again this may change with future releases.
835 @kindex -plugin @var{name}
836 @item -plugin @var{name}
837 Involve a plugin in the linking process. The @var{name} parameter is
838 the absolute filename of the plugin. Usually this parameter is
839 automatically added by the complier, when using link time
840 optimization, but users can also add their own plugins if they so
843 Note that the location of the compiler originated plugins is different
844 from the place where the @command{ar}, @command{nm} and
845 @command{ranlib} programs search for their plugins. In order for
846 those commands to make use of a compiler based plugin it must first be
847 copied into the @file{$@{libdir@}/bfd-plugins} directory. All gcc
848 based linker plugins are backward compatible, so it is sufficient to
849 just copy in the newest one.
852 @cindex push state governing input file handling
854 The @option{--push-state} allows to preserve the current state of the
855 flags which govern the input file handling so that they can all be
856 restored with one corresponding @option{--pop-state} option.
858 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
859 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
860 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
861 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
862 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
863 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
865 One target for this option are specifications for @file{pkg-config}. When
866 used with the @option{--libs} option all possibly needed libraries are
867 listed and then possibly linked with all the time. It is better to return
868 something as follows:
871 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
875 @cindex pop state governing input file handling
877 Undoes the effect of --push-state, restores the previous values of the
878 flags governing input file handling.
881 @kindex --emit-relocs
882 @cindex retain relocations in final executable
885 Leave relocation sections and contents in fully linked executables.
886 Post link analysis and optimization tools may need this information in
887 order to perform correct modifications of executables. This results
888 in larger executables.
890 This option is currently only supported on ELF platforms.
892 @kindex --force-dynamic
893 @cindex forcing the creation of dynamic sections
894 @item --force-dynamic
895 Force the output file to have dynamic sections. This option is specific
899 @cindex relocatable output
901 @kindex --relocatable
904 Generate relocatable output---i.e., generate an output file that can in
905 turn serve as input to @command{ld}. This is often called @dfn{partial
906 linking}. As a side effect, in environments that support standard Unix
907 magic numbers, this option also sets the output file's magic number to
909 @c ; see @option{-N}.
910 If this option is not specified, an absolute file is produced. When
911 linking C++ programs, this option @emph{will not} resolve references to
912 constructors; to do that, use @samp{-Ur}.
914 When an input file does not have the same format as the output file,
915 partial linking is only supported if that input file does not contain any
916 relocations. Different output formats can have further restrictions; for
917 example some @code{a.out}-based formats do not support partial linking
918 with input files in other formats at all.
920 This option does the same thing as @samp{-i}.
922 @kindex -R @var{file}
923 @kindex --just-symbols=@var{file}
924 @cindex symbol-only input
925 @item -R @var{filename}
926 @itemx --just-symbols=@var{filename}
927 Read symbol names and their addresses from @var{filename}, but do not
928 relocate it or include it in the output. This allows your output file
929 to refer symbolically to absolute locations of memory defined in other
930 programs. You may use this option more than once.
932 For compatibility with other ELF linkers, if the @option{-R} option is
933 followed by a directory name, rather than a file name, it is treated as
934 the @option{-rpath} option.
938 @cindex strip all symbols
941 Omit all symbol information from the output file.
944 @kindex --strip-debug
945 @cindex strip debugger symbols
948 Omit debugger symbol information (but not all symbols) from the output file.
950 @kindex --strip-discarded
951 @kindex --no-strip-discarded
952 @item --strip-discarded
953 @itemx --no-strip-discarded
954 Omit (or do not omit) global symbols defined in discarded sections.
959 @cindex input files, displaying
962 Print the names of the input files as @command{ld} processes them.
964 @kindex -T @var{script}
965 @kindex --script=@var{script}
967 @item -T @var{scriptfile}
968 @itemx --script=@var{scriptfile}
969 Use @var{scriptfile} as the linker script. This script replaces
970 @command{ld}'s default linker script (rather than adding to it), so
971 @var{commandfile} must specify everything necessary to describe the
972 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
973 the current directory, @code{ld} looks for it in the directories
974 specified by any preceding @samp{-L} options. Multiple @samp{-T}
977 @kindex -dT @var{script}
978 @kindex --default-script=@var{script}
980 @item -dT @var{scriptfile}
981 @itemx --default-script=@var{scriptfile}
982 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
984 This option is similar to the @option{--script} option except that
985 processing of the script is delayed until after the rest of the
986 command line has been processed. This allows options placed after the
987 @option{--default-script} option on the command line to affect the
988 behaviour of the linker script, which can be important when the linker
989 command line cannot be directly controlled by the user. (eg because
990 the command line is being constructed by another tool, such as
993 @kindex -u @var{symbol}
994 @kindex --undefined=@var{symbol}
995 @cindex undefined symbol
996 @item -u @var{symbol}
997 @itemx --undefined=@var{symbol}
998 Force @var{symbol} to be entered in the output file as an undefined
999 symbol. Doing this may, for example, trigger linking of additional
1000 modules from standard libraries. @samp{-u} may be repeated with
1001 different option arguments to enter additional undefined symbols. This
1002 option is equivalent to the @code{EXTERN} linker script command.
1004 If this option is being used to force additional modules to be pulled
1005 into the link, and if it is an error for the symbol to remain
1006 undefined, then the option @option{--require-defined} should be used
1009 @kindex --require-defined=@var{symbol}
1010 @cindex symbols, require defined
1011 @cindex defined symbol
1012 @item --require-defined=@var{symbol}
1013 Require that @var{symbol} is defined in the output file. This option
1014 is the same as option @option{--undefined} except that if @var{symbol}
1015 is not defined in the output file then the linker will issue an error
1016 and exit. The same effect can be achieved in a linker script by using
1017 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
1018 can be used multiple times to require additional symbols.
1021 @cindex constructors
1023 For anything other than C++ programs, this option is equivalent to
1024 @samp{-r}: it generates relocatable output---i.e., an output file that can in
1025 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
1026 @emph{does} resolve references to constructors, unlike @samp{-r}.
1027 It does not work to use @samp{-Ur} on files that were themselves linked
1028 with @samp{-Ur}; once the constructor table has been built, it cannot
1029 be added to. Use @samp{-Ur} only for the last partial link, and
1030 @samp{-r} for the others.
1032 @kindex --orphan-handling=@var{MODE}
1033 @cindex orphan sections
1034 @cindex sections, orphan
1035 @item --orphan-handling=@var{MODE}
1036 Control how orphan sections are handled. An orphan section is one not
1037 specifically mentioned in a linker script. @xref{Orphan Sections}.
1039 @var{MODE} can have any of the following values:
1043 Orphan sections are placed into a suitable output section following
1044 the strategy described in @ref{Orphan Sections}. The option
1045 @samp{--unique} also affects how sections are placed.
1048 All orphan sections are discarded, by placing them in the
1049 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1052 The linker will place the orphan section as for @code{place} and also
1056 The linker will exit with an error if any orphan section is found.
1059 The default if @samp{--orphan-handling} is not given is @code{place}.
1061 @kindex --unique[=@var{SECTION}]
1062 @item --unique[=@var{SECTION}]
1063 Creates a separate output section for every input section matching
1064 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1065 missing, for every orphan input section. An orphan section is one not
1066 specifically mentioned in a linker script. You may use this option
1067 multiple times on the command line; It prevents the normal merging of
1068 input sections with the same name, overriding output section assignments
1078 Display the version number for @command{ld}. The @option{-V} option also
1079 lists the supported emulations.
1082 @kindex --discard-all
1083 @cindex deleting local symbols
1085 @itemx --discard-all
1086 Delete all local symbols.
1089 @kindex --discard-locals
1090 @cindex local symbols, deleting
1092 @itemx --discard-locals
1093 Delete all temporary local symbols. (These symbols start with
1094 system-specific local label prefixes, typically @samp{.L} for ELF systems
1095 or @samp{L} for traditional a.out systems.)
1097 @kindex -y @var{symbol}
1098 @kindex --trace-symbol=@var{symbol}
1099 @cindex symbol tracing
1100 @item -y @var{symbol}
1101 @itemx --trace-symbol=@var{symbol}
1102 Print the name of each linked file in which @var{symbol} appears. This
1103 option may be given any number of times. On many systems it is necessary
1104 to prepend an underscore.
1106 This option is useful when you have an undefined symbol in your link but
1107 don't know where the reference is coming from.
1109 @kindex -Y @var{path}
1111 Add @var{path} to the default library search path. This option exists
1112 for Solaris compatibility.
1114 @kindex -z @var{keyword}
1115 @item -z @var{keyword}
1116 The recognized keywords are:
1120 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1122 @item call-nop=prefix-addr
1123 @itemx call-nop=suffix-nop
1124 @itemx call-nop=prefix-@var{byte}
1125 @itemx call-nop=suffix-@var{byte}
1126 Specify the 1-byte @code{NOP} padding when transforming indirect call
1127 to a locally defined function, foo, via its GOT slot.
1128 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1129 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1130 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1131 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1132 Supported for i386 and x86_64.
1136 Combine multiple dynamic relocation sections and sort to improve
1137 dynamic symbol lookup caching. Do not do this if @samp{nocombreloc}.
1141 Generate common symbols with STT_COMMON type during a relocatable
1142 link. Use STT_OBJECT type if @samp{nocommon}.
1144 @item common-page-size=@var{value}
1145 Set the page size most commonly used to @var{value}. Memory image
1146 layout will be optimized to minimize memory pages if the system is
1147 using pages of this size.
1150 Disallows undefined symbols in object files. Undefined symbols in
1151 shared libraries are still allowed.
1153 @item dynamic-undefined-weak
1154 @itemx nodynamic-undefined-weak
1155 Make undefined weak symbols dynamic when building a dynamic object,
1156 if they are referenced from a regular object file and not forced local
1157 by symbol visibility or versioning. Do not make them dynamic if
1158 @samp{nodynamic-undefined-weak}. If neither option is given, a target
1159 may default to either option being in force, or make some other
1160 selection of undefined weak symbols dynamic. Not all targets support
1164 Marks the object as requiring executable stack.
1167 This option is only meaningful when building a shared object. It makes
1168 the symbols defined by this shared object available for symbol resolution
1169 of subsequently loaded libraries.
1172 This option is only meaningful when building a dynamic executable.
1173 This option marks the executable as requiring global auditing by
1174 setting the @code{DF_1_GLOBAUDIT} bit in the @code{DT_FLAGS_1} dynamic
1175 tag. Global auditing requires that any auditing library defined via
1176 the @option{--depaudit} or @option{-P} command line options be run for
1177 all dynamic objects loaded by the application.
1180 Generate Intel Indirect Branch Tracking (IBT) enabled PLT entries.
1181 Supported for Linux/i386 and Linux/x86_64.
1184 Generate GNU_PROPERTY_X86_FEATURE_1_IBT in .note.gnu.property section
1185 to indicate compatibility with IBT. This also implies @option{ibtplt}.
1186 Supported for Linux/i386 and Linux/x86_64.
1189 This option is only meaningful when building a shared object.
1190 It marks the object so that its runtime initialization will occur
1191 before the runtime initialization of any other objects brought into
1192 the process at the same time. Similarly the runtime finalization of
1193 the object will occur after the runtime finalization of any other
1197 Specify that the dynamic loader should modify its symbol search order
1198 so that symbols in this shared library interpose all other shared
1199 libraries not so marked.
1202 When generating an executable or shared library, mark it to tell the
1203 dynamic linker to defer function call resolution to the point when
1204 the function is called (lazy binding), rather than at load time.
1205 Lazy binding is the default.
1208 Specify that the object's filters be processed immediately at runtime.
1210 @item max-page-size=@var{value}
1211 Set the maximum memory page size supported to @var{value}.
1214 Allow multiple definitions.
1217 Disable linker generated .dynbss variables used in place of variables
1218 defined in shared libraries. May result in dynamic text relocations.
1221 Specify that the dynamic loader search for dependencies of this object
1222 should ignore any default library search paths.
1225 Specify that the object shouldn't be unloaded at runtime.
1228 Specify that the object is not available to @code{dlopen}.
1231 Specify that the object can not be dumped by @code{dldump}.
1234 Marks the object as not requiring executable stack.
1236 @item noextern-protected-data
1237 Don't treat protected data symbols as external when building a shared
1238 library. This option overrides the linker backend default. It can be
1239 used to work around incorrect relocations against protected data symbols
1240 generated by compiler. Updates on protected data symbols by another
1241 module aren't visible to the resulting shared library. Supported for
1244 @item noreloc-overflow
1245 Disable relocation overflow check. This can be used to disable
1246 relocation overflow check if there will be no dynamic relocation
1247 overflow at run-time. Supported for x86_64.
1250 When generating an executable or shared library, mark it to tell the
1251 dynamic linker to resolve all symbols when the program is started, or
1252 when the shared library is loaded by dlopen, instead of deferring
1253 function call resolution to the point when the function is first
1257 Specify that the object requires @samp{$ORIGIN} handling in paths.
1261 Create an ELF @code{PT_GNU_RELRO} segment header in the object. This
1262 specifies a memory segment that should be made read-only after
1263 relocation, if supported. Specifying @samp{common-page-size} smaller
1264 than the system page size will render this protection ineffective.
1265 Don't create an ELF @code{PT_GNU_RELRO} segment if @samp{norelro}.
1268 Generate GNU_PROPERTY_X86_FEATURE_1_SHSTK in .note.gnu.property section
1269 to indicate compatibility with Intel Shadow Stack. Supported for
1270 Linux/i386 and Linux/x86_64.
1272 @item stack-size=@var{value}
1273 Specify a stack size for an ELF @code{PT_GNU_STACK} segment.
1274 Specifying zero will override any default non-zero sized
1275 @code{PT_GNU_STACK} segment creation.
1280 Report an error if DT_TEXTREL is set, i.e., if the binary has dynamic
1281 relocations in read-only sections. Don't report an error if
1282 @samp{notext} or @samp{textoff}.
1286 Other keywords are ignored for Solaris compatibility.
1289 @cindex groups of archives
1290 @item -( @var{archives} -)
1291 @itemx --start-group @var{archives} --end-group
1292 The @var{archives} should be a list of archive files. They may be
1293 either explicit file names, or @samp{-l} options.
1295 The specified archives are searched repeatedly until no new undefined
1296 references are created. Normally, an archive is searched only once in
1297 the order that it is specified on the command line. If a symbol in that
1298 archive is needed to resolve an undefined symbol referred to by an
1299 object in an archive that appears later on the command line, the linker
1300 would not be able to resolve that reference. By grouping the archives,
1301 they all be searched repeatedly until all possible references are
1304 Using this option has a significant performance cost. It is best to use
1305 it only when there are unavoidable circular references between two or
1308 @kindex --accept-unknown-input-arch
1309 @kindex --no-accept-unknown-input-arch
1310 @item --accept-unknown-input-arch
1311 @itemx --no-accept-unknown-input-arch
1312 Tells the linker to accept input files whose architecture cannot be
1313 recognised. The assumption is that the user knows what they are doing
1314 and deliberately wants to link in these unknown input files. This was
1315 the default behaviour of the linker, before release 2.14. The default
1316 behaviour from release 2.14 onwards is to reject such input files, and
1317 so the @samp{--accept-unknown-input-arch} option has been added to
1318 restore the old behaviour.
1321 @kindex --no-as-needed
1323 @itemx --no-as-needed
1324 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1325 on the command line after the @option{--as-needed} option. Normally
1326 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1327 on the command line, regardless of whether the library is actually
1328 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1329 emitted for a library that @emph{at that point in the link} satisfies a
1330 non-weak undefined symbol reference from a regular object file or, if
1331 the library is not found in the DT_NEEDED lists of other needed libraries, a
1332 non-weak undefined symbol reference from another needed dynamic library.
1333 Object files or libraries appearing on the command line @emph{after}
1334 the library in question do not affect whether the library is seen as
1335 needed. This is similar to the rules for extraction of object files
1336 from archives. @option{--no-as-needed} restores the default behaviour.
1338 @kindex --add-needed
1339 @kindex --no-add-needed
1341 @itemx --no-add-needed
1342 These two options have been deprecated because of the similarity of
1343 their names to the @option{--as-needed} and @option{--no-as-needed}
1344 options. They have been replaced by @option{--copy-dt-needed-entries}
1345 and @option{--no-copy-dt-needed-entries}.
1347 @kindex -assert @var{keyword}
1348 @item -assert @var{keyword}
1349 This option is ignored for SunOS compatibility.
1353 @kindex -call_shared
1357 Link against dynamic libraries. This is only meaningful on platforms
1358 for which shared libraries are supported. This option is normally the
1359 default on such platforms. The different variants of this option are
1360 for compatibility with various systems. You may use this option
1361 multiple times on the command line: it affects library searching for
1362 @option{-l} options which follow it.
1366 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1367 section. This causes the runtime linker to handle lookups in this
1368 object and its dependencies to be performed only inside the group.
1369 @option{--unresolved-symbols=report-all} is implied. This option is
1370 only meaningful on ELF platforms which support shared libraries.
1380 Do not link against shared libraries. This is only meaningful on
1381 platforms for which shared libraries are supported. The different
1382 variants of this option are for compatibility with various systems. You
1383 may use this option multiple times on the command line: it affects
1384 library searching for @option{-l} options which follow it. This
1385 option also implies @option{--unresolved-symbols=report-all}. This
1386 option can be used with @option{-shared}. Doing so means that a
1387 shared library is being created but that all of the library's external
1388 references must be resolved by pulling in entries from static
1393 When creating a shared library, bind references to global symbols to the
1394 definition within the shared library, if any. Normally, it is possible
1395 for a program linked against a shared library to override the definition
1396 within the shared library. This option can also be used with the
1397 @option{--export-dynamic} option, when creating a position independent
1398 executable, to bind references to global symbols to the definition within
1399 the executable. This option is only meaningful on ELF platforms which
1400 support shared libraries and position independent executables.
1402 @kindex -Bsymbolic-functions
1403 @item -Bsymbolic-functions
1404 When creating a shared library, bind references to global function
1405 symbols to the definition within the shared library, if any.
1406 This option can also be used with the @option{--export-dynamic} option,
1407 when creating a position independent executable, to bind references
1408 to global function symbols to the definition within the executable.
1409 This option is only meaningful on ELF platforms which support shared
1410 libraries and position independent executables.
1412 @kindex --dynamic-list=@var{dynamic-list-file}
1413 @item --dynamic-list=@var{dynamic-list-file}
1414 Specify the name of a dynamic list file to the linker. This is
1415 typically used when creating shared libraries to specify a list of
1416 global symbols whose references shouldn't be bound to the definition
1417 within the shared library, or creating dynamically linked executables
1418 to specify a list of symbols which should be added to the symbol table
1419 in the executable. This option is only meaningful on ELF platforms
1420 which support shared libraries.
1422 The format of the dynamic list is the same as the version node without
1423 scope and node name. See @ref{VERSION} for more information.
1425 @kindex --dynamic-list-data
1426 @item --dynamic-list-data
1427 Include all global data symbols to the dynamic list.
1429 @kindex --dynamic-list-cpp-new
1430 @item --dynamic-list-cpp-new
1431 Provide the builtin dynamic list for C++ operator new and delete. It
1432 is mainly useful for building shared libstdc++.
1434 @kindex --dynamic-list-cpp-typeinfo
1435 @item --dynamic-list-cpp-typeinfo
1436 Provide the builtin dynamic list for C++ runtime type identification.
1438 @kindex --check-sections
1439 @kindex --no-check-sections
1440 @item --check-sections
1441 @itemx --no-check-sections
1442 Asks the linker @emph{not} to check section addresses after they have
1443 been assigned to see if there are any overlaps. Normally the linker will
1444 perform this check, and if it finds any overlaps it will produce
1445 suitable error messages. The linker does know about, and does make
1446 allowances for sections in overlays. The default behaviour can be
1447 restored by using the command line switch @option{--check-sections}.
1448 Section overlap is not usually checked for relocatable links. You can
1449 force checking in that case by using the @option{--check-sections}
1452 @kindex --copy-dt-needed-entries
1453 @kindex --no-copy-dt-needed-entries
1454 @item --copy-dt-needed-entries
1455 @itemx --no-copy-dt-needed-entries
1456 This option affects the treatment of dynamic libraries referred to
1457 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1458 command line. Normally the linker won't add a DT_NEEDED tag to the
1459 output binary for each library mentioned in a DT_NEEDED tag in an
1460 input dynamic library. With @option{--copy-dt-needed-entries}
1461 specified on the command line however any dynamic libraries that
1462 follow it will have their DT_NEEDED entries added. The default
1463 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1465 This option also has an effect on the resolution of symbols in dynamic
1466 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1467 mentioned on the command line will be recursively searched, following
1468 their DT_NEEDED tags to other libraries, in order to resolve symbols
1469 required by the output binary. With the default setting however
1470 the searching of dynamic libraries that follow it will stop with the
1471 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1474 @cindex cross reference table
1477 Output a cross reference table. If a linker map file is being
1478 generated, the cross reference table is printed to the map file.
1479 Otherwise, it is printed on the standard output.
1481 The format of the table is intentionally simple, so that it may be
1482 easily processed by a script if necessary. The symbols are printed out,
1483 sorted by name. For each symbol, a list of file names is given. If the
1484 symbol is defined, the first file listed is the location of the
1485 definition. If the symbol is defined as a common value then any files
1486 where this happens appear next. Finally any files that reference the
1489 @cindex common allocation
1490 @kindex --no-define-common
1491 @item --no-define-common
1492 This option inhibits the assignment of addresses to common symbols.
1493 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1494 @xref{Miscellaneous Commands}.
1496 The @samp{--no-define-common} option allows decoupling
1497 the decision to assign addresses to Common symbols from the choice
1498 of the output file type; otherwise a non-Relocatable output type
1499 forces assigning addresses to Common symbols.
1500 Using @samp{--no-define-common} allows Common symbols that are referenced
1501 from a shared library to be assigned addresses only in the main program.
1502 This eliminates the unused duplicate space in the shared library,
1503 and also prevents any possible confusion over resolving to the wrong
1504 duplicate when there are many dynamic modules with specialized search
1505 paths for runtime symbol resolution.
1507 @cindex group allocation in linker script
1508 @cindex section groups
1510 @kindex --force-group-allocation
1511 @item --force-group-allocation
1512 This option causes the linker to place section group members like
1513 normal input sections, and to delete the section groups. This is the
1514 default behaviour for a final link but this option can be used to
1515 change the behaviour of a relocatable link (@samp{-r}). The script
1516 command @code{FORCE_GROUP_ALLOCATION} has the same
1517 effect. @xref{Miscellaneous Commands}.
1519 @cindex symbols, from command line
1520 @kindex --defsym=@var{symbol}=@var{exp}
1521 @item --defsym=@var{symbol}=@var{expression}
1522 Create a global symbol in the output file, containing the absolute
1523 address given by @var{expression}. You may use this option as many
1524 times as necessary to define multiple symbols in the command line. A
1525 limited form of arithmetic is supported for the @var{expression} in this
1526 context: you may give a hexadecimal constant or the name of an existing
1527 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1528 constants or symbols. If you need more elaborate expressions, consider
1529 using the linker command language from a script (@pxref{Assignments}).
1530 @emph{Note:} there should be no white space between @var{symbol}, the
1531 equals sign (``@key{=}''), and @var{expression}.
1533 @cindex demangling, from command line
1534 @kindex --demangle[=@var{style}]
1535 @kindex --no-demangle
1536 @item --demangle[=@var{style}]
1537 @itemx --no-demangle
1538 These options control whether to demangle symbol names in error messages
1539 and other output. When the linker is told to demangle, it tries to
1540 present symbol names in a readable fashion: it strips leading
1541 underscores if they are used by the object file format, and converts C++
1542 mangled symbol names into user readable names. Different compilers have
1543 different mangling styles. The optional demangling style argument can be used
1544 to choose an appropriate demangling style for your compiler. The linker will
1545 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1546 is set. These options may be used to override the default.
1548 @cindex dynamic linker, from command line
1549 @kindex -I@var{file}
1550 @kindex --dynamic-linker=@var{file}
1552 @itemx --dynamic-linker=@var{file}
1553 Set the name of the dynamic linker. This is only meaningful when
1554 generating dynamically linked ELF executables. The default dynamic
1555 linker is normally correct; don't use this unless you know what you are
1558 @kindex --no-dynamic-linker
1559 @item --no-dynamic-linker
1560 When producing an executable file, omit the request for a dynamic
1561 linker to be used at load-time. This is only meaningful for ELF
1562 executables that contain dynamic relocations, and usually requires
1563 entry point code that is capable of processing these relocations.
1565 @kindex --embedded-relocs
1566 @item --embedded-relocs
1567 This option is similar to the @option{--emit-relocs} option except
1568 that the relocs are stored in a target specific section. This option
1569 is only supported by the @samp{BFIN}, @samp{CR16} and @emph{M68K}
1572 @kindex --fatal-warnings
1573 @kindex --no-fatal-warnings
1574 @item --fatal-warnings
1575 @itemx --no-fatal-warnings
1576 Treat all warnings as errors. The default behaviour can be restored
1577 with the option @option{--no-fatal-warnings}.
1579 @kindex --force-exe-suffix
1580 @item --force-exe-suffix
1581 Make sure that an output file has a .exe suffix.
1583 If a successfully built fully linked output file does not have a
1584 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1585 the output file to one of the same name with a @code{.exe} suffix. This
1586 option is useful when using unmodified Unix makefiles on a Microsoft
1587 Windows host, since some versions of Windows won't run an image unless
1588 it ends in a @code{.exe} suffix.
1590 @kindex --gc-sections
1591 @kindex --no-gc-sections
1592 @cindex garbage collection
1594 @itemx --no-gc-sections
1595 Enable garbage collection of unused input sections. It is ignored on
1596 targets that do not support this option. The default behaviour (of not
1597 performing this garbage collection) can be restored by specifying
1598 @samp{--no-gc-sections} on the command line. Note that garbage
1599 collection for COFF and PE format targets is supported, but the
1600 implementation is currently considered to be experimental.
1602 @samp{--gc-sections} decides which input sections are used by
1603 examining symbols and relocations. The section containing the entry
1604 symbol and all sections containing symbols undefined on the
1605 command-line will be kept, as will sections containing symbols
1606 referenced by dynamic objects. Note that when building shared
1607 libraries, the linker must assume that any visible symbol is
1608 referenced. Once this initial set of sections has been determined,
1609 the linker recursively marks as used any section referenced by their
1610 relocations. See @samp{--entry} and @samp{--undefined}.
1612 This option can be set when doing a partial link (enabled with option
1613 @samp{-r}). In this case the root of symbols kept must be explicitly
1614 specified either by an @samp{--entry} or @samp{--undefined} option or by
1615 a @code{ENTRY} command in the linker script.
1617 @kindex --print-gc-sections
1618 @kindex --no-print-gc-sections
1619 @cindex garbage collection
1620 @item --print-gc-sections
1621 @itemx --no-print-gc-sections
1622 List all sections removed by garbage collection. The listing is
1623 printed on stderr. This option is only effective if garbage
1624 collection has been enabled via the @samp{--gc-sections}) option. The
1625 default behaviour (of not listing the sections that are removed) can
1626 be restored by specifying @samp{--no-print-gc-sections} on the command
1629 @kindex --gc-keep-exported
1630 @cindex garbage collection
1631 @item --gc-keep-exported
1632 When @samp{--gc-sections} is enabled, this option prevents garbage
1633 collection of unused input sections that contain global symbols having
1634 default or protected visibility. This option is intended to be used for
1635 executables where unreferenced sections would otherwise be garbage
1636 collected regardless of the external visibility of contained symbols.
1637 Note that this option has no effect when linking shared objects since
1638 it is already the default behaviour. This option is only supported for
1641 @kindex --print-output-format
1642 @cindex output format
1643 @item --print-output-format
1644 Print the name of the default output format (perhaps influenced by
1645 other command-line options). This is the string that would appear
1646 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1648 @kindex --print-memory-usage
1649 @cindex memory usage
1650 @item --print-memory-usage
1651 Print used size, total size and used size of memory regions created with
1652 the @ref{MEMORY} command. This is useful on embedded targets to have a
1653 quick view of amount of free memory. The format of the output has one
1654 headline and one line per region. It is both human readable and easily
1655 parsable by tools. Here is an example of an output:
1658 Memory region Used Size Region Size %age Used
1659 ROM: 256 KB 1 MB 25.00%
1660 RAM: 32 B 2 GB 0.00%
1667 Print a summary of the command-line options on the standard output and exit.
1669 @kindex --target-help
1671 Print a summary of all target specific options on the standard output and exit.
1673 @kindex -Map=@var{mapfile}
1674 @item -Map=@var{mapfile}
1675 Print a link map to the file @var{mapfile}. See the description of the
1676 @option{-M} option, above.
1678 @cindex memory usage
1679 @kindex --no-keep-memory
1680 @item --no-keep-memory
1681 @command{ld} normally optimizes for speed over memory usage by caching the
1682 symbol tables of input files in memory. This option tells @command{ld} to
1683 instead optimize for memory usage, by rereading the symbol tables as
1684 necessary. This may be required if @command{ld} runs out of memory space
1685 while linking a large executable.
1687 @kindex --no-undefined
1689 @item --no-undefined
1691 Report unresolved symbol references from regular object files. This
1692 is done even if the linker is creating a non-symbolic shared library.
1693 The switch @option{--[no-]allow-shlib-undefined} controls the
1694 behaviour for reporting unresolved references found in shared
1695 libraries being linked in.
1697 @kindex --allow-multiple-definition
1699 @item --allow-multiple-definition
1701 Normally when a symbol is defined multiple times, the linker will
1702 report a fatal error. These options allow multiple definitions and the
1703 first definition will be used.
1705 @kindex --allow-shlib-undefined
1706 @kindex --no-allow-shlib-undefined
1707 @item --allow-shlib-undefined
1708 @itemx --no-allow-shlib-undefined
1709 Allows or disallows undefined symbols in shared libraries.
1710 This switch is similar to @option{--no-undefined} except that it
1711 determines the behaviour when the undefined symbols are in a
1712 shared library rather than a regular object file. It does not affect
1713 how undefined symbols in regular object files are handled.
1715 The default behaviour is to report errors for any undefined symbols
1716 referenced in shared libraries if the linker is being used to create
1717 an executable, but to allow them if the linker is being used to create
1720 The reasons for allowing undefined symbol references in shared
1721 libraries specified at link time are that:
1725 A shared library specified at link time may not be the same as the one
1726 that is available at load time, so the symbol might actually be
1727 resolvable at load time.
1729 There are some operating systems, eg BeOS and HPPA, where undefined
1730 symbols in shared libraries are normal.
1732 The BeOS kernel for example patches shared libraries at load time to
1733 select whichever function is most appropriate for the current
1734 architecture. This is used, for example, to dynamically select an
1735 appropriate memset function.
1738 @kindex --no-undefined-version
1739 @item --no-undefined-version
1740 Normally when a symbol has an undefined version, the linker will ignore
1741 it. This option disallows symbols with undefined version and a fatal error
1742 will be issued instead.
1744 @kindex --default-symver
1745 @item --default-symver
1746 Create and use a default symbol version (the soname) for unversioned
1749 @kindex --default-imported-symver
1750 @item --default-imported-symver
1751 Create and use a default symbol version (the soname) for unversioned
1754 @kindex --no-warn-mismatch
1755 @item --no-warn-mismatch
1756 Normally @command{ld} will give an error if you try to link together input
1757 files that are mismatched for some reason, perhaps because they have
1758 been compiled for different processors or for different endiannesses.
1759 This option tells @command{ld} that it should silently permit such possible
1760 errors. This option should only be used with care, in cases when you
1761 have taken some special action that ensures that the linker errors are
1764 @kindex --no-warn-search-mismatch
1765 @item --no-warn-search-mismatch
1766 Normally @command{ld} will give a warning if it finds an incompatible
1767 library during a library search. This option silences the warning.
1769 @kindex --no-whole-archive
1770 @item --no-whole-archive
1771 Turn off the effect of the @option{--whole-archive} option for subsequent
1774 @cindex output file after errors
1775 @kindex --noinhibit-exec
1776 @item --noinhibit-exec
1777 Retain the executable output file whenever it is still usable.
1778 Normally, the linker will not produce an output file if it encounters
1779 errors during the link process; it exits without writing an output file
1780 when it issues any error whatsoever.
1784 Only search library directories explicitly specified on the
1785 command line. Library directories specified in linker scripts
1786 (including linker scripts specified on the command line) are ignored.
1788 @ifclear SingleFormat
1789 @kindex --oformat=@var{output-format}
1790 @item --oformat=@var{output-format}
1791 @command{ld} may be configured to support more than one kind of object
1792 file. If your @command{ld} is configured this way, you can use the
1793 @samp{--oformat} option to specify the binary format for the output
1794 object file. Even when @command{ld} is configured to support alternative
1795 object formats, you don't usually need to specify this, as @command{ld}
1796 should be configured to produce as a default output format the most
1797 usual format on each machine. @var{output-format} is a text string, the
1798 name of a particular format supported by the BFD libraries. (You can
1799 list the available binary formats with @samp{objdump -i}.) The script
1800 command @code{OUTPUT_FORMAT} can also specify the output format, but
1801 this option overrides it. @xref{BFD}.
1804 @kindex --out-implib
1805 @item --out-implib @var{file}
1806 Create an import library in @var{file} corresponding to the executable
1807 the linker is generating (eg. a DLL or ELF program). This import
1808 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
1809 may be used to link clients against the generated executable; this
1810 behaviour makes it possible to skip a separate import library creation
1811 step (eg. @code{dlltool} for DLLs). This option is only available for
1812 the i386 PE and ELF targetted ports of the linker.
1815 @kindex --pic-executable
1817 @itemx --pic-executable
1818 @cindex position independent executables
1819 Create a position independent executable. This is currently only supported on
1820 ELF platforms. Position independent executables are similar to shared
1821 libraries in that they are relocated by the dynamic linker to the virtual
1822 address the OS chooses for them (which can vary between invocations). Like
1823 normal dynamically linked executables they can be executed and symbols
1824 defined in the executable cannot be overridden by shared libraries.
1828 This option is ignored for Linux compatibility.
1832 This option is ignored for SVR4 compatibility.
1835 @cindex synthesizing linker
1836 @cindex relaxing addressing modes
1840 An option with machine dependent effects.
1842 This option is only supported on a few targets.
1845 @xref{H8/300,,@command{ld} and the H8/300}.
1848 @xref{i960,, @command{ld} and the Intel 960 family}.
1851 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1854 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1857 @xref{Nios II,,@command{ld} and the Altera Nios II}.
1860 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1863 On some platforms the @samp{--relax} option performs target specific,
1864 global optimizations that become possible when the linker resolves
1865 addressing in the program, such as relaxing address modes,
1866 synthesizing new instructions, selecting shorter version of current
1867 instructions, and combining constant values.
1869 On some platforms these link time global optimizations may make symbolic
1870 debugging of the resulting executable impossible.
1872 This is known to be the case for the Matsushita MN10200 and MN10300
1873 family of processors.
1877 On platforms where this is not supported, @samp{--relax} is accepted,
1881 On platforms where @samp{--relax} is accepted the option
1882 @samp{--no-relax} can be used to disable the feature.
1884 @cindex retaining specified symbols
1885 @cindex stripping all but some symbols
1886 @cindex symbols, retaining selectively
1887 @kindex --retain-symbols-file=@var{filename}
1888 @item --retain-symbols-file=@var{filename}
1889 Retain @emph{only} the symbols listed in the file @var{filename},
1890 discarding all others. @var{filename} is simply a flat file, with one
1891 symbol name per line. This option is especially useful in environments
1895 where a large global symbol table is accumulated gradually, to conserve
1898 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1899 or symbols needed for relocations.
1901 You may only specify @samp{--retain-symbols-file} once in the command
1902 line. It overrides @samp{-s} and @samp{-S}.
1905 @item -rpath=@var{dir}
1906 @cindex runtime library search path
1907 @kindex -rpath=@var{dir}
1908 Add a directory to the runtime library search path. This is used when
1909 linking an ELF executable with shared objects. All @option{-rpath}
1910 arguments are concatenated and passed to the runtime linker, which uses
1911 them to locate shared objects at runtime. The @option{-rpath} option is
1912 also used when locating shared objects which are needed by shared
1913 objects explicitly included in the link; see the description of the
1914 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1915 ELF executable, the contents of the environment variable
1916 @code{LD_RUN_PATH} will be used if it is defined.
1918 The @option{-rpath} option may also be used on SunOS. By default, on
1919 SunOS, the linker will form a runtime search path out of all the
1920 @option{-L} options it is given. If a @option{-rpath} option is used, the
1921 runtime search path will be formed exclusively using the @option{-rpath}
1922 options, ignoring the @option{-L} options. This can be useful when using
1923 gcc, which adds many @option{-L} options which may be on NFS mounted
1926 For compatibility with other ELF linkers, if the @option{-R} option is
1927 followed by a directory name, rather than a file name, it is treated as
1928 the @option{-rpath} option.
1932 @cindex link-time runtime library search path
1933 @kindex -rpath-link=@var{dir}
1934 @item -rpath-link=@var{dir}
1935 When using ELF or SunOS, one shared library may require another. This
1936 happens when an @code{ld -shared} link includes a shared library as one
1939 When the linker encounters such a dependency when doing a non-shared,
1940 non-relocatable link, it will automatically try to locate the required
1941 shared library and include it in the link, if it is not included
1942 explicitly. In such a case, the @option{-rpath-link} option
1943 specifies the first set of directories to search. The
1944 @option{-rpath-link} option may specify a sequence of directory names
1945 either by specifying a list of names separated by colons, or by
1946 appearing multiple times.
1948 The tokens @var{$ORIGIN} and @var{$LIB} can appear in these search
1949 directories. They will be replaced by the full path to the directory
1950 containing the program or shared object in the case of @var{$ORIGIN}
1951 and either @samp{lib} - for 32-bit binaries - or @samp{lib64} - for
1952 64-bit binaries - in the case of @var{$LIB}.
1954 The alternative form of these tokens - @var{$@{ORIGIN@}} and
1955 @var{$@{LIB@}} can also be used. The token @var{$PLATFORM} is not
1958 This option should be used with caution as it overrides the search path
1959 that may have been hard compiled into a shared library. In such a case it
1960 is possible to use unintentionally a different search path than the
1961 runtime linker would do.
1963 The linker uses the following search paths to locate required shared
1967 Any directories specified by @option{-rpath-link} options.
1969 Any directories specified by @option{-rpath} options. The difference
1970 between @option{-rpath} and @option{-rpath-link} is that directories
1971 specified by @option{-rpath} options are included in the executable and
1972 used at runtime, whereas the @option{-rpath-link} option is only effective
1973 at link time. Searching @option{-rpath} in this way is only supported
1974 by native linkers and cross linkers which have been configured with
1975 the @option{--with-sysroot} option.
1977 On an ELF system, for native linkers, if the @option{-rpath} and
1978 @option{-rpath-link} options were not used, search the contents of the
1979 environment variable @code{LD_RUN_PATH}.
1981 On SunOS, if the @option{-rpath} option was not used, search any
1982 directories specified using @option{-L} options.
1984 For a native linker, search the contents of the environment
1985 variable @code{LD_LIBRARY_PATH}.
1987 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1988 @code{DT_RPATH} of a shared library are searched for shared
1989 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1990 @code{DT_RUNPATH} entries exist.
1992 The default directories, normally @file{/lib} and @file{/usr/lib}.
1994 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1995 exists, the list of directories found in that file.
1998 If the required shared library is not found, the linker will issue a
1999 warning and continue with the link.
2006 @cindex shared libraries
2007 Create a shared library. This is currently only supported on ELF, XCOFF
2008 and SunOS platforms. On SunOS, the linker will automatically create a
2009 shared library if the @option{-e} option is not used and there are
2010 undefined symbols in the link.
2012 @kindex --sort-common
2014 @itemx --sort-common=ascending
2015 @itemx --sort-common=descending
2016 This option tells @command{ld} to sort the common symbols by alignment in
2017 ascending or descending order when it places them in the appropriate output
2018 sections. The symbol alignments considered are sixteen-byte or larger,
2019 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
2020 between symbols due to alignment constraints. If no sorting order is
2021 specified, then descending order is assumed.
2023 @kindex --sort-section=name
2024 @item --sort-section=name
2025 This option will apply @code{SORT_BY_NAME} to all wildcard section
2026 patterns in the linker script.
2028 @kindex --sort-section=alignment
2029 @item --sort-section=alignment
2030 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
2031 patterns in the linker script.
2033 @kindex --spare-dynamic-tags
2034 @item --spare-dynamic-tags=@var{count}
2035 This option specifies the number of empty slots to leave in the
2036 .dynamic section of ELF shared objects. Empty slots may be needed by
2037 post processing tools, such as the prelinker. The default is 5.
2039 @kindex --split-by-file
2040 @item --split-by-file[=@var{size}]
2041 Similar to @option{--split-by-reloc} but creates a new output section for
2042 each input file when @var{size} is reached. @var{size} defaults to a
2043 size of 1 if not given.
2045 @kindex --split-by-reloc
2046 @item --split-by-reloc[=@var{count}]
2047 Tries to creates extra sections in the output file so that no single
2048 output section in the file contains more than @var{count} relocations.
2049 This is useful when generating huge relocatable files for downloading into
2050 certain real time kernels with the COFF object file format; since COFF
2051 cannot represent more than 65535 relocations in a single section. Note
2052 that this will fail to work with object file formats which do not
2053 support arbitrary sections. The linker will not split up individual
2054 input sections for redistribution, so if a single input section contains
2055 more than @var{count} relocations one output section will contain that
2056 many relocations. @var{count} defaults to a value of 32768.
2060 Compute and display statistics about the operation of the linker, such
2061 as execution time and memory usage.
2063 @kindex --sysroot=@var{directory}
2064 @item --sysroot=@var{directory}
2065 Use @var{directory} as the location of the sysroot, overriding the
2066 configure-time default. This option is only supported by linkers
2067 that were configured using @option{--with-sysroot}.
2071 This is used by COFF/PE based targets to create a task-linked object
2072 file where all of the global symbols have been converted to statics.
2074 @kindex --traditional-format
2075 @cindex traditional format
2076 @item --traditional-format
2077 For some targets, the output of @command{ld} is different in some ways from
2078 the output of some existing linker. This switch requests @command{ld} to
2079 use the traditional format instead.
2082 For example, on SunOS, @command{ld} combines duplicate entries in the
2083 symbol string table. This can reduce the size of an output file with
2084 full debugging information by over 30 percent. Unfortunately, the SunOS
2085 @code{dbx} program can not read the resulting program (@code{gdb} has no
2086 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
2087 combine duplicate entries.
2089 @kindex --section-start=@var{sectionname}=@var{org}
2090 @item --section-start=@var{sectionname}=@var{org}
2091 Locate a section in the output file at the absolute
2092 address given by @var{org}. You may use this option as many
2093 times as necessary to locate multiple sections in the command
2095 @var{org} must be a single hexadecimal integer;
2096 for compatibility with other linkers, you may omit the leading
2097 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
2098 should be no white space between @var{sectionname}, the equals
2099 sign (``@key{=}''), and @var{org}.
2101 @kindex -Tbss=@var{org}
2102 @kindex -Tdata=@var{org}
2103 @kindex -Ttext=@var{org}
2104 @cindex segment origins, cmd line
2105 @item -Tbss=@var{org}
2106 @itemx -Tdata=@var{org}
2107 @itemx -Ttext=@var{org}
2108 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2109 @code{.text} as the @var{sectionname}.
2111 @kindex -Ttext-segment=@var{org}
2112 @item -Ttext-segment=@var{org}
2113 @cindex text segment origin, cmd line
2114 When creating an ELF executable, it will set the address of the first
2115 byte of the text segment.
2117 @kindex -Trodata-segment=@var{org}
2118 @item -Trodata-segment=@var{org}
2119 @cindex rodata segment origin, cmd line
2120 When creating an ELF executable or shared object for a target where
2121 the read-only data is in its own segment separate from the executable
2122 text, it will set the address of the first byte of the read-only data segment.
2124 @kindex -Tldata-segment=@var{org}
2125 @item -Tldata-segment=@var{org}
2126 @cindex ldata segment origin, cmd line
2127 When creating an ELF executable or shared object for x86-64 medium memory
2128 model, it will set the address of the first byte of the ldata segment.
2130 @kindex --unresolved-symbols
2131 @item --unresolved-symbols=@var{method}
2132 Determine how to handle unresolved symbols. There are four possible
2133 values for @samp{method}:
2137 Do not report any unresolved symbols.
2140 Report all unresolved symbols. This is the default.
2142 @item ignore-in-object-files
2143 Report unresolved symbols that are contained in shared libraries, but
2144 ignore them if they come from regular object files.
2146 @item ignore-in-shared-libs
2147 Report unresolved symbols that come from regular object files, but
2148 ignore them if they come from shared libraries. This can be useful
2149 when creating a dynamic binary and it is known that all the shared
2150 libraries that it should be referencing are included on the linker's
2154 The behaviour for shared libraries on their own can also be controlled
2155 by the @option{--[no-]allow-shlib-undefined} option.
2157 Normally the linker will generate an error message for each reported
2158 unresolved symbol but the option @option{--warn-unresolved-symbols}
2159 can change this to a warning.
2161 @kindex --verbose[=@var{NUMBER}]
2162 @cindex verbose[=@var{NUMBER}]
2164 @itemx --verbose[=@var{NUMBER}]
2165 Display the version number for @command{ld} and list the linker emulations
2166 supported. Display which input files can and cannot be opened. Display
2167 the linker script being used by the linker. If the optional @var{NUMBER}
2168 argument > 1, plugin symbol status will also be displayed.
2170 @kindex --version-script=@var{version-scriptfile}
2171 @cindex version script, symbol versions
2172 @item --version-script=@var{version-scriptfile}
2173 Specify the name of a version script to the linker. This is typically
2174 used when creating shared libraries to specify additional information
2175 about the version hierarchy for the library being created. This option
2176 is only fully supported on ELF platforms which support shared libraries;
2177 see @ref{VERSION}. It is partially supported on PE platforms, which can
2178 use version scripts to filter symbol visibility in auto-export mode: any
2179 symbols marked @samp{local} in the version script will not be exported.
2182 @kindex --warn-common
2183 @cindex warnings, on combining symbols
2184 @cindex combining symbols, warnings on
2186 Warn when a common symbol is combined with another common symbol or with
2187 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2188 but linkers on some other operating systems do not. This option allows
2189 you to find potential problems from combining global symbols.
2190 Unfortunately, some C libraries use this practice, so you may get some
2191 warnings about symbols in the libraries as well as in your programs.
2193 There are three kinds of global symbols, illustrated here by C examples:
2197 A definition, which goes in the initialized data section of the output
2201 An undefined reference, which does not allocate space.
2202 There must be either a definition or a common symbol for the
2206 A common symbol. If there are only (one or more) common symbols for a
2207 variable, it goes in the uninitialized data area of the output file.
2208 The linker merges multiple common symbols for the same variable into a
2209 single symbol. If they are of different sizes, it picks the largest
2210 size. The linker turns a common symbol into a declaration, if there is
2211 a definition of the same variable.
2214 The @samp{--warn-common} option can produce five kinds of warnings.
2215 Each warning consists of a pair of lines: the first describes the symbol
2216 just encountered, and the second describes the previous symbol
2217 encountered with the same name. One or both of the two symbols will be
2222 Turning a common symbol into a reference, because there is already a
2223 definition for the symbol.
2225 @var{file}(@var{section}): warning: common of `@var{symbol}'
2226 overridden by definition
2227 @var{file}(@var{section}): warning: defined here
2231 Turning a common symbol into a reference, because a later definition for
2232 the symbol is encountered. This is the same as the previous case,
2233 except that the symbols are encountered in a different order.
2235 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2237 @var{file}(@var{section}): warning: common is here
2241 Merging a common symbol with a previous same-sized common symbol.
2243 @var{file}(@var{section}): warning: multiple common
2245 @var{file}(@var{section}): warning: previous common is here
2249 Merging a common symbol with a previous larger common symbol.
2251 @var{file}(@var{section}): warning: common of `@var{symbol}'
2252 overridden by larger common
2253 @var{file}(@var{section}): warning: larger common is here
2257 Merging a common symbol with a previous smaller common symbol. This is
2258 the same as the previous case, except that the symbols are
2259 encountered in a different order.
2261 @var{file}(@var{section}): warning: common of `@var{symbol}'
2262 overriding smaller common
2263 @var{file}(@var{section}): warning: smaller common is here
2267 @kindex --warn-constructors
2268 @item --warn-constructors
2269 Warn if any global constructors are used. This is only useful for a few
2270 object file formats. For formats like COFF or ELF, the linker can not
2271 detect the use of global constructors.
2273 @kindex --warn-multiple-gp
2274 @item --warn-multiple-gp
2275 Warn if multiple global pointer values are required in the output file.
2276 This is only meaningful for certain processors, such as the Alpha.
2277 Specifically, some processors put large-valued constants in a special
2278 section. A special register (the global pointer) points into the middle
2279 of this section, so that constants can be loaded efficiently via a
2280 base-register relative addressing mode. Since the offset in
2281 base-register relative mode is fixed and relatively small (e.g., 16
2282 bits), this limits the maximum size of the constant pool. Thus, in
2283 large programs, it is often necessary to use multiple global pointer
2284 values in order to be able to address all possible constants. This
2285 option causes a warning to be issued whenever this case occurs.
2288 @cindex warnings, on undefined symbols
2289 @cindex undefined symbols, warnings on
2291 Only warn once for each undefined symbol, rather than once per module
2294 @kindex --warn-section-align
2295 @cindex warnings, on section alignment
2296 @cindex section alignment, warnings on
2297 @item --warn-section-align
2298 Warn if the address of an output section is changed because of
2299 alignment. Typically, the alignment will be set by an input section.
2300 The address will only be changed if it not explicitly specified; that
2301 is, if the @code{SECTIONS} command does not specify a start address for
2302 the section (@pxref{SECTIONS}).
2304 @kindex --warn-shared-textrel
2305 @item --warn-shared-textrel
2306 Warn if the linker adds a DT_TEXTREL to a shared object.
2308 @kindex --warn-alternate-em
2309 @item --warn-alternate-em
2310 Warn if an object has alternate ELF machine code.
2312 @kindex --warn-unresolved-symbols
2313 @item --warn-unresolved-symbols
2314 If the linker is going to report an unresolved symbol (see the option
2315 @option{--unresolved-symbols}) it will normally generate an error.
2316 This option makes it generate a warning instead.
2318 @kindex --error-unresolved-symbols
2319 @item --error-unresolved-symbols
2320 This restores the linker's default behaviour of generating errors when
2321 it is reporting unresolved symbols.
2323 @kindex --whole-archive
2324 @cindex including an entire archive
2325 @item --whole-archive
2326 For each archive mentioned on the command line after the
2327 @option{--whole-archive} option, include every object file in the archive
2328 in the link, rather than searching the archive for the required object
2329 files. This is normally used to turn an archive file into a shared
2330 library, forcing every object to be included in the resulting shared
2331 library. This option may be used more than once.
2333 Two notes when using this option from gcc: First, gcc doesn't know
2334 about this option, so you have to use @option{-Wl,-whole-archive}.
2335 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2336 list of archives, because gcc will add its own list of archives to
2337 your link and you may not want this flag to affect those as well.
2339 @kindex --wrap=@var{symbol}
2340 @item --wrap=@var{symbol}
2341 Use a wrapper function for @var{symbol}. Any undefined reference to
2342 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2343 undefined reference to @code{__real_@var{symbol}} will be resolved to
2346 This can be used to provide a wrapper for a system function. The
2347 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2348 wishes to call the system function, it should call
2349 @code{__real_@var{symbol}}.
2351 Here is a trivial example:
2355 __wrap_malloc (size_t c)
2357 printf ("malloc called with %zu\n", c);
2358 return __real_malloc (c);
2362 If you link other code with this file using @option{--wrap malloc}, then
2363 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2364 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2365 call the real @code{malloc} function.
2367 You may wish to provide a @code{__real_malloc} function as well, so that
2368 links without the @option{--wrap} option will succeed. If you do this,
2369 you should not put the definition of @code{__real_malloc} in the same
2370 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2371 call before the linker has a chance to wrap it to @code{malloc}.
2373 @kindex --eh-frame-hdr
2374 @kindex --no-eh-frame-hdr
2375 @item --eh-frame-hdr
2376 @itemx --no-eh-frame-hdr
2377 Request (@option{--eh-frame-hdr}) or suppress
2378 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2379 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2381 @kindex --ld-generated-unwind-info
2382 @item --no-ld-generated-unwind-info
2383 Request creation of @code{.eh_frame} unwind info for linker
2384 generated code sections like PLT. This option is on by default
2385 if linker generated unwind info is supported.
2387 @kindex --enable-new-dtags
2388 @kindex --disable-new-dtags
2389 @item --enable-new-dtags
2390 @itemx --disable-new-dtags
2391 This linker can create the new dynamic tags in ELF. But the older ELF
2392 systems may not understand them. If you specify
2393 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2394 and older dynamic tags will be omitted.
2395 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2396 created. By default, the new dynamic tags are not created. Note that
2397 those options are only available for ELF systems.
2399 @kindex --hash-size=@var{number}
2400 @item --hash-size=@var{number}
2401 Set the default size of the linker's hash tables to a prime number
2402 close to @var{number}. Increasing this value can reduce the length of
2403 time it takes the linker to perform its tasks, at the expense of
2404 increasing the linker's memory requirements. Similarly reducing this
2405 value can reduce the memory requirements at the expense of speed.
2407 @kindex --hash-style=@var{style}
2408 @item --hash-style=@var{style}
2409 Set the type of linker's hash table(s). @var{style} can be either
2410 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2411 new style GNU @code{.gnu.hash} section or @code{both} for both
2412 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2413 hash tables. The default is @code{sysv}.
2415 @kindex --compress-debug-sections=none
2416 @kindex --compress-debug-sections=zlib
2417 @kindex --compress-debug-sections=zlib-gnu
2418 @kindex --compress-debug-sections=zlib-gabi
2419 @item --compress-debug-sections=none
2420 @itemx --compress-debug-sections=zlib
2421 @itemx --compress-debug-sections=zlib-gnu
2422 @itemx --compress-debug-sections=zlib-gabi
2423 On ELF platforms, these options control how DWARF debug sections are
2424 compressed using zlib.
2426 @option{--compress-debug-sections=none} doesn't compress DWARF debug
2427 sections. @option{--compress-debug-sections=zlib-gnu} compresses
2428 DWARF debug sections and renames them to begin with @samp{.zdebug}
2429 instead of @samp{.debug}. @option{--compress-debug-sections=zlib-gabi}
2430 also compresses DWARF debug sections, but rather than renaming them it
2431 sets the SHF_COMPRESSED flag in the sections' headers.
2433 The @option{--compress-debug-sections=zlib} option is an alias for
2434 @option{--compress-debug-sections=zlib-gabi}.
2436 Note that this option overrides any compression in input debug
2437 sections, so if a binary is linked with @option{--compress-debug-sections=none}
2438 for example, then any compressed debug sections in input files will be
2439 uncompressed before they are copied into the output binary.
2441 The default compression behaviour varies depending upon the target
2442 involved and the configure options used to build the toolchain. The
2443 default can be determined by examining the output from the linker's
2444 @option{--help} option.
2446 @kindex --reduce-memory-overheads
2447 @item --reduce-memory-overheads
2448 This option reduces memory requirements at ld runtime, at the expense of
2449 linking speed. This was introduced to select the old O(n^2) algorithm
2450 for link map file generation, rather than the new O(n) algorithm which uses
2451 about 40% more memory for symbol storage.
2453 Another effect of the switch is to set the default hash table size to
2454 1021, which again saves memory at the cost of lengthening the linker's
2455 run time. This is not done however if the @option{--hash-size} switch
2458 The @option{--reduce-memory-overheads} switch may be also be used to
2459 enable other tradeoffs in future versions of the linker.
2462 @kindex --build-id=@var{style}
2464 @itemx --build-id=@var{style}
2465 Request the creation of a @code{.note.gnu.build-id} ELF note section
2466 or a @code{.buildid} COFF section. The contents of the note are
2467 unique bits identifying this linked file. @var{style} can be
2468 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2469 @sc{SHA1} hash on the normative parts of the output contents,
2470 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2471 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2472 string specified as an even number of hexadecimal digits (@code{-} and
2473 @code{:} characters between digit pairs are ignored). If @var{style}
2474 is omitted, @code{sha1} is used.
2476 The @code{md5} and @code{sha1} styles produces an identifier
2477 that is always the same in an identical output file, but will be
2478 unique among all nonidentical output files. It is not intended
2479 to be compared as a checksum for the file's contents. A linked
2480 file may be changed later by other tools, but the build ID bit
2481 string identifying the original linked file does not change.
2483 Passing @code{none} for @var{style} disables the setting from any
2484 @code{--build-id} options earlier on the command line.
2489 @subsection Options Specific to i386 PE Targets
2491 @c man begin OPTIONS
2493 The i386 PE linker supports the @option{-shared} option, which causes
2494 the output to be a dynamically linked library (DLL) instead of a
2495 normal executable. You should name the output @code{*.dll} when you
2496 use this option. In addition, the linker fully supports the standard
2497 @code{*.def} files, which may be specified on the linker command line
2498 like an object file (in fact, it should precede archives it exports
2499 symbols from, to ensure that they get linked in, just like a normal
2502 In addition to the options common to all targets, the i386 PE linker
2503 support additional command line options that are specific to the i386
2504 PE target. Options that take values may be separated from their
2505 values by either a space or an equals sign.
2509 @kindex --add-stdcall-alias
2510 @item --add-stdcall-alias
2511 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2512 as-is and also with the suffix stripped.
2513 [This option is specific to the i386 PE targeted port of the linker]
2516 @item --base-file @var{file}
2517 Use @var{file} as the name of a file in which to save the base
2518 addresses of all the relocations needed for generating DLLs with
2520 [This is an i386 PE specific option]
2524 Create a DLL instead of a regular executable. You may also use
2525 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2527 [This option is specific to the i386 PE targeted port of the linker]
2529 @kindex --enable-long-section-names
2530 @kindex --disable-long-section-names
2531 @item --enable-long-section-names
2532 @itemx --disable-long-section-names
2533 The PE variants of the COFF object format add an extension that permits
2534 the use of section names longer than eight characters, the normal limit
2535 for COFF. By default, these names are only allowed in object files, as
2536 fully-linked executable images do not carry the COFF string table required
2537 to support the longer names. As a GNU extension, it is possible to
2538 allow their use in executable images as well, or to (probably pointlessly!)
2539 disallow it in object files, by using these two options. Executable images
2540 generated with these long section names are slightly non-standard, carrying
2541 as they do a string table, and may generate confusing output when examined
2542 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2543 GDB relies on the use of PE long section names to find Dwarf-2 debug
2544 information sections in an executable image at runtime, and so if neither
2545 option is specified on the command-line, @command{ld} will enable long
2546 section names, overriding the default and technically correct behaviour,
2547 when it finds the presence of debug information while linking an executable
2548 image and not stripping symbols.
2549 [This option is valid for all PE targeted ports of the linker]
2551 @kindex --enable-stdcall-fixup
2552 @kindex --disable-stdcall-fixup
2553 @item --enable-stdcall-fixup
2554 @itemx --disable-stdcall-fixup
2555 If the link finds a symbol that it cannot resolve, it will attempt to
2556 do ``fuzzy linking'' by looking for another defined symbol that differs
2557 only in the format of the symbol name (cdecl vs stdcall) and will
2558 resolve that symbol by linking to the match. For example, the
2559 undefined symbol @code{_foo} might be linked to the function
2560 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2561 to the function @code{_bar}. When the linker does this, it prints a
2562 warning, since it normally should have failed to link, but sometimes
2563 import libraries generated from third-party dlls may need this feature
2564 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2565 feature is fully enabled and warnings are not printed. If you specify
2566 @option{--disable-stdcall-fixup}, this feature is disabled and such
2567 mismatches are considered to be errors.
2568 [This option is specific to the i386 PE targeted port of the linker]
2570 @kindex --leading-underscore
2571 @kindex --no-leading-underscore
2572 @item --leading-underscore
2573 @itemx --no-leading-underscore
2574 For most targets default symbol-prefix is an underscore and is defined
2575 in target's description. By this option it is possible to
2576 disable/enable the default underscore symbol-prefix.
2578 @cindex DLLs, creating
2579 @kindex --export-all-symbols
2580 @item --export-all-symbols
2581 If given, all global symbols in the objects used to build a DLL will
2582 be exported by the DLL. Note that this is the default if there
2583 otherwise wouldn't be any exported symbols. When symbols are
2584 explicitly exported via DEF files or implicitly exported via function
2585 attributes, the default is to not export anything else unless this
2586 option is given. Note that the symbols @code{DllMain@@12},
2587 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2588 @code{impure_ptr} will not be automatically
2589 exported. Also, symbols imported from other DLLs will not be
2590 re-exported, nor will symbols specifying the DLL's internal layout
2591 such as those beginning with @code{_head_} or ending with
2592 @code{_iname}. In addition, no symbols from @code{libgcc},
2593 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2594 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2595 not be exported, to help with C++ DLLs. Finally, there is an
2596 extensive list of cygwin-private symbols that are not exported
2597 (obviously, this applies on when building DLLs for cygwin targets).
2598 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2599 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2600 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2601 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2602 @code{cygwin_premain3}, and @code{environ}.
2603 [This option is specific to the i386 PE targeted port of the linker]
2605 @kindex --exclude-symbols
2606 @item --exclude-symbols @var{symbol},@var{symbol},...
2607 Specifies a list of symbols which should not be automatically
2608 exported. The symbol names may be delimited by commas or colons.
2609 [This option is specific to the i386 PE targeted port of the linker]
2611 @kindex --exclude-all-symbols
2612 @item --exclude-all-symbols
2613 Specifies no symbols should be automatically exported.
2614 [This option is specific to the i386 PE targeted port of the linker]
2616 @kindex --file-alignment
2617 @item --file-alignment
2618 Specify the file alignment. Sections in the file will always begin at
2619 file offsets which are multiples of this number. This defaults to
2621 [This option is specific to the i386 PE targeted port of the linker]
2625 @item --heap @var{reserve}
2626 @itemx --heap @var{reserve},@var{commit}
2627 Specify the number of bytes of memory to reserve (and optionally commit)
2628 to be used as heap for this program. The default is 1MB reserved, 4K
2630 [This option is specific to the i386 PE targeted port of the linker]
2633 @kindex --image-base
2634 @item --image-base @var{value}
2635 Use @var{value} as the base address of your program or dll. This is
2636 the lowest memory location that will be used when your program or dll
2637 is loaded. To reduce the need to relocate and improve performance of
2638 your dlls, each should have a unique base address and not overlap any
2639 other dlls. The default is 0x400000 for executables, and 0x10000000
2641 [This option is specific to the i386 PE targeted port of the linker]
2645 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2646 symbols before they are exported.
2647 [This option is specific to the i386 PE targeted port of the linker]
2649 @kindex --large-address-aware
2650 @item --large-address-aware
2651 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2652 header is set to indicate that this executable supports virtual addresses
2653 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2654 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2655 section of the BOOT.INI. Otherwise, this bit has no effect.
2656 [This option is specific to PE targeted ports of the linker]
2658 @kindex --disable-large-address-aware
2659 @item --disable-large-address-aware
2660 Reverts the effect of a previous @samp{--large-address-aware} option.
2661 This is useful if @samp{--large-address-aware} is always set by the compiler
2662 driver (e.g. Cygwin gcc) and the executable does not support virtual
2663 addresses greater than 2 gigabytes.
2664 [This option is specific to PE targeted ports of the linker]
2666 @kindex --major-image-version
2667 @item --major-image-version @var{value}
2668 Sets the major number of the ``image version''. Defaults to 1.
2669 [This option is specific to the i386 PE targeted port of the linker]
2671 @kindex --major-os-version
2672 @item --major-os-version @var{value}
2673 Sets the major number of the ``os version''. Defaults to 4.
2674 [This option is specific to the i386 PE targeted port of the linker]
2676 @kindex --major-subsystem-version
2677 @item --major-subsystem-version @var{value}
2678 Sets the major number of the ``subsystem version''. Defaults to 4.
2679 [This option is specific to the i386 PE targeted port of the linker]
2681 @kindex --minor-image-version
2682 @item --minor-image-version @var{value}
2683 Sets the minor number of the ``image version''. Defaults to 0.
2684 [This option is specific to the i386 PE targeted port of the linker]
2686 @kindex --minor-os-version
2687 @item --minor-os-version @var{value}
2688 Sets the minor number of the ``os version''. Defaults to 0.
2689 [This option is specific to the i386 PE targeted port of the linker]
2691 @kindex --minor-subsystem-version
2692 @item --minor-subsystem-version @var{value}
2693 Sets the minor number of the ``subsystem version''. Defaults to 0.
2694 [This option is specific to the i386 PE targeted port of the linker]
2696 @cindex DEF files, creating
2697 @cindex DLLs, creating
2698 @kindex --output-def
2699 @item --output-def @var{file}
2700 The linker will create the file @var{file} which will contain a DEF
2701 file corresponding to the DLL the linker is generating. This DEF file
2702 (which should be called @code{*.def}) may be used to create an import
2703 library with @code{dlltool} or may be used as a reference to
2704 automatically or implicitly exported symbols.
2705 [This option is specific to the i386 PE targeted port of the linker]
2707 @cindex DLLs, creating
2708 @kindex --enable-auto-image-base
2709 @item --enable-auto-image-base
2710 @itemx --enable-auto-image-base=@var{value}
2711 Automatically choose the image base for DLLs, optionally starting with base
2712 @var{value}, unless one is specified using the @code{--image-base} argument.
2713 By using a hash generated from the dllname to create unique image bases
2714 for each DLL, in-memory collisions and relocations which can delay program
2715 execution are avoided.
2716 [This option is specific to the i386 PE targeted port of the linker]
2718 @kindex --disable-auto-image-base
2719 @item --disable-auto-image-base
2720 Do not automatically generate a unique image base. If there is no
2721 user-specified image base (@code{--image-base}) then use the platform
2723 [This option is specific to the i386 PE targeted port of the linker]
2725 @cindex DLLs, linking to
2726 @kindex --dll-search-prefix
2727 @item --dll-search-prefix @var{string}
2728 When linking dynamically to a dll without an import library,
2729 search for @code{<string><basename>.dll} in preference to
2730 @code{lib<basename>.dll}. This behaviour allows easy distinction
2731 between DLLs built for the various "subplatforms": native, cygwin,
2732 uwin, pw, etc. For instance, cygwin DLLs typically use
2733 @code{--dll-search-prefix=cyg}.
2734 [This option is specific to the i386 PE targeted port of the linker]
2736 @kindex --enable-auto-import
2737 @item --enable-auto-import
2738 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2739 DATA imports from DLLs, and create the necessary thunking symbols when
2740 building the import libraries with those DATA exports. Note: Use of the
2741 'auto-import' extension will cause the text section of the image file
2742 to be made writable. This does not conform to the PE-COFF format
2743 specification published by Microsoft.
2745 Note - use of the 'auto-import' extension will also cause read only
2746 data which would normally be placed into the .rdata section to be
2747 placed into the .data section instead. This is in order to work
2748 around a problem with consts that is described here:
2749 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2751 Using 'auto-import' generally will 'just work' -- but sometimes you may
2754 "variable '<var>' can't be auto-imported. Please read the
2755 documentation for ld's @code{--enable-auto-import} for details."
2757 This message occurs when some (sub)expression accesses an address
2758 ultimately given by the sum of two constants (Win32 import tables only
2759 allow one). Instances where this may occur include accesses to member
2760 fields of struct variables imported from a DLL, as well as using a
2761 constant index into an array variable imported from a DLL. Any
2762 multiword variable (arrays, structs, long long, etc) may trigger
2763 this error condition. However, regardless of the exact data type
2764 of the offending exported variable, ld will always detect it, issue
2765 the warning, and exit.
2767 There are several ways to address this difficulty, regardless of the
2768 data type of the exported variable:
2770 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2771 of adjusting references in your client code for runtime environment, so
2772 this method works only when runtime environment supports this feature.
2774 A second solution is to force one of the 'constants' to be a variable --
2775 that is, unknown and un-optimizable at compile time. For arrays,
2776 there are two possibilities: a) make the indexee (the array's address)
2777 a variable, or b) make the 'constant' index a variable. Thus:
2780 extern type extern_array[];
2782 @{ volatile type *t=extern_array; t[1] @}
2788 extern type extern_array[];
2790 @{ volatile int t=1; extern_array[t] @}
2793 For structs (and most other multiword data types) the only option
2794 is to make the struct itself (or the long long, or the ...) variable:
2797 extern struct s extern_struct;
2798 extern_struct.field -->
2799 @{ volatile struct s *t=&extern_struct; t->field @}
2805 extern long long extern_ll;
2807 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2810 A third method of dealing with this difficulty is to abandon
2811 'auto-import' for the offending symbol and mark it with
2812 @code{__declspec(dllimport)}. However, in practice that
2813 requires using compile-time #defines to indicate whether you are
2814 building a DLL, building client code that will link to the DLL, or
2815 merely building/linking to a static library. In making the choice
2816 between the various methods of resolving the 'direct address with
2817 constant offset' problem, you should consider typical real-world usage:
2825 void main(int argc, char **argv)@{
2826 printf("%d\n",arr[1]);
2836 void main(int argc, char **argv)@{
2837 /* This workaround is for win32 and cygwin; do not "optimize" */
2838 volatile int *parr = arr;
2839 printf("%d\n",parr[1]);
2846 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2847 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2848 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2849 #define FOO_IMPORT __declspec(dllimport)
2853 extern FOO_IMPORT int arr[];
2856 void main(int argc, char **argv)@{
2857 printf("%d\n",arr[1]);
2861 A fourth way to avoid this problem is to re-code your
2862 library to use a functional interface rather than a data interface
2863 for the offending variables (e.g. set_foo() and get_foo() accessor
2865 [This option is specific to the i386 PE targeted port of the linker]
2867 @kindex --disable-auto-import
2868 @item --disable-auto-import
2869 Do not attempt to do sophisticated linking of @code{_symbol} to
2870 @code{__imp__symbol} for DATA imports from DLLs.
2871 [This option is specific to the i386 PE targeted port of the linker]
2873 @kindex --enable-runtime-pseudo-reloc
2874 @item --enable-runtime-pseudo-reloc
2875 If your code contains expressions described in --enable-auto-import section,
2876 that is, DATA imports from DLL with non-zero offset, this switch will create
2877 a vector of 'runtime pseudo relocations' which can be used by runtime
2878 environment to adjust references to such data in your client code.
2879 [This option is specific to the i386 PE targeted port of the linker]
2881 @kindex --disable-runtime-pseudo-reloc
2882 @item --disable-runtime-pseudo-reloc
2883 Do not create pseudo relocations for non-zero offset DATA imports from
2885 [This option is specific to the i386 PE targeted port of the linker]
2887 @kindex --enable-extra-pe-debug
2888 @item --enable-extra-pe-debug
2889 Show additional debug info related to auto-import symbol thunking.
2890 [This option is specific to the i386 PE targeted port of the linker]
2892 @kindex --section-alignment
2893 @item --section-alignment
2894 Sets the section alignment. Sections in memory will always begin at
2895 addresses which are a multiple of this number. Defaults to 0x1000.
2896 [This option is specific to the i386 PE targeted port of the linker]
2900 @item --stack @var{reserve}
2901 @itemx --stack @var{reserve},@var{commit}
2902 Specify the number of bytes of memory to reserve (and optionally commit)
2903 to be used as stack for this program. The default is 2MB reserved, 4K
2905 [This option is specific to the i386 PE targeted port of the linker]
2908 @item --subsystem @var{which}
2909 @itemx --subsystem @var{which}:@var{major}
2910 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2911 Specifies the subsystem under which your program will execute. The
2912 legal values for @var{which} are @code{native}, @code{windows},
2913 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2914 the subsystem version also. Numeric values are also accepted for
2916 [This option is specific to the i386 PE targeted port of the linker]
2918 The following options set flags in the @code{DllCharacteristics} field
2919 of the PE file header:
2920 [These options are specific to PE targeted ports of the linker]
2922 @kindex --high-entropy-va
2923 @item --high-entropy-va
2924 Image is compatible with 64-bit address space layout randomization
2927 @kindex --dynamicbase
2929 The image base address may be relocated using address space layout
2930 randomization (ASLR). This feature was introduced with MS Windows
2931 Vista for i386 PE targets.
2933 @kindex --forceinteg
2935 Code integrity checks are enforced.
2939 The image is compatible with the Data Execution Prevention.
2940 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2942 @kindex --no-isolation
2943 @item --no-isolation
2944 Although the image understands isolation, do not isolate the image.
2948 The image does not use SEH. No SE handler may be called from
2953 Do not bind this image.
2957 The driver uses the MS Windows Driver Model.
2961 The image is Terminal Server aware.
2963 @kindex --insert-timestamp
2964 @item --insert-timestamp
2965 @itemx --no-insert-timestamp
2966 Insert a real timestamp into the image. This is the default behaviour
2967 as it matches legacy code and it means that the image will work with
2968 other, proprietary tools. The problem with this default is that it
2969 will result in slightly different images being produced each time the
2970 same sources are linked. The option @option{--no-insert-timestamp}
2971 can be used to insert a zero value for the timestamp, this ensuring
2972 that binaries produced from identical sources will compare
2979 @subsection Options specific to C6X uClinux targets
2981 @c man begin OPTIONS
2983 The C6X uClinux target uses a binary format called DSBT to support shared
2984 libraries. Each shared library in the system needs to have a unique index;
2985 all executables use an index of 0.
2990 @item --dsbt-size @var{size}
2991 This option sets the number of entries in the DSBT of the current executable
2992 or shared library to @var{size}. The default is to create a table with 64
2995 @kindex --dsbt-index
2996 @item --dsbt-index @var{index}
2997 This option sets the DSBT index of the current executable or shared library
2998 to @var{index}. The default is 0, which is appropriate for generating
2999 executables. If a shared library is generated with a DSBT index of 0, the
3000 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
3002 @kindex --no-merge-exidx-entries
3003 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
3004 exidx entries in frame unwind info.
3012 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
3014 @c man begin OPTIONS
3016 The 68HC11 and 68HC12 linkers support specific options to control the
3017 memory bank switching mapping and trampoline code generation.
3021 @kindex --no-trampoline
3022 @item --no-trampoline
3023 This option disables the generation of trampoline. By default a trampoline
3024 is generated for each far function which is called using a @code{jsr}
3025 instruction (this happens when a pointer to a far function is taken).
3027 @kindex --bank-window
3028 @item --bank-window @var{name}
3029 This option indicates to the linker the name of the memory region in
3030 the @samp{MEMORY} specification that describes the memory bank window.
3031 The definition of such region is then used by the linker to compute
3032 paging and addresses within the memory window.
3040 @subsection Options specific to Motorola 68K target
3042 @c man begin OPTIONS
3044 The following options are supported to control handling of GOT generation
3045 when linking for 68K targets.
3050 @item --got=@var{type}
3051 This option tells the linker which GOT generation scheme to use.
3052 @var{type} should be one of @samp{single}, @samp{negative},
3053 @samp{multigot} or @samp{target}. For more information refer to the
3054 Info entry for @file{ld}.
3062 @subsection Options specific to MIPS targets
3064 @c man begin OPTIONS
3066 The following options are supported to control microMIPS instruction
3067 generation and branch relocation checks for ISA mode transitions when
3068 linking for MIPS targets.
3076 These options control the choice of microMIPS instructions used in code
3077 generated by the linker, such as that in the PLT or lazy binding stubs,
3078 or in relaxation. If @samp{--insn32} is used, then the linker only uses
3079 32-bit instruction encodings. By default or if @samp{--no-insn32} is
3080 used, all instruction encodings are used, including 16-bit ones where
3083 @kindex --ignore-branch-isa
3084 @item --ignore-branch-isa
3085 @kindex --no-ignore-branch-isa
3086 @itemx --no-ignore-branch-isa
3087 These options control branch relocation checks for invalid ISA mode
3088 transitions. If @samp{--ignore-branch-isa} is used, then the linker
3089 accepts any branch relocations and any ISA mode transition required
3090 is lost in relocation calculation, except for some cases of @code{BAL}
3091 instructions which meet relaxation conditions and are converted to
3092 equivalent @code{JALX} instructions as the associated relocation is
3093 calculated. By default or if @samp{--no-ignore-branch-isa} is used
3094 a check is made causing the loss of an ISA mode transition to produce
3104 @section Environment Variables
3106 @c man begin ENVIRONMENT
3108 You can change the behaviour of @command{ld} with the environment variables
3109 @ifclear SingleFormat
3112 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
3114 @ifclear SingleFormat
3116 @cindex default input format
3117 @code{GNUTARGET} determines the input-file object format if you don't
3118 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
3119 of the BFD names for an input format (@pxref{BFD}). If there is no
3120 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
3121 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
3122 attempts to discover the input format by examining binary input files;
3123 this method often succeeds, but there are potential ambiguities, since
3124 there is no method of ensuring that the magic number used to specify
3125 object-file formats is unique. However, the configuration procedure for
3126 BFD on each system places the conventional format for that system first
3127 in the search-list, so ambiguities are resolved in favor of convention.
3131 @cindex default emulation
3132 @cindex emulation, default
3133 @code{LDEMULATION} determines the default emulation if you don't use the
3134 @samp{-m} option. The emulation can affect various aspects of linker
3135 behaviour, particularly the default linker script. You can list the
3136 available emulations with the @samp{--verbose} or @samp{-V} options. If
3137 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3138 variable is not defined, the default emulation depends upon how the
3139 linker was configured.
3141 @kindex COLLECT_NO_DEMANGLE
3142 @cindex demangling, default
3143 Normally, the linker will default to demangling symbols. However, if
3144 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3145 default to not demangling symbols. This environment variable is used in
3146 a similar fashion by the @code{gcc} linker wrapper program. The default
3147 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3154 @chapter Linker Scripts
3157 @cindex linker scripts
3158 @cindex command files
3159 Every link is controlled by a @dfn{linker script}. This script is
3160 written in the linker command language.
3162 The main purpose of the linker script is to describe how the sections in
3163 the input files should be mapped into the output file, and to control
3164 the memory layout of the output file. Most linker scripts do nothing
3165 more than this. However, when necessary, the linker script can also
3166 direct the linker to perform many other operations, using the commands
3169 The linker always uses a linker script. If you do not supply one
3170 yourself, the linker will use a default script that is compiled into the
3171 linker executable. You can use the @samp{--verbose} command line option
3172 to display the default linker script. Certain command line options,
3173 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3175 You may supply your own linker script by using the @samp{-T} command
3176 line option. When you do this, your linker script will replace the
3177 default linker script.
3179 You may also use linker scripts implicitly by naming them as input files
3180 to the linker, as though they were files to be linked. @xref{Implicit
3184 * Basic Script Concepts:: Basic Linker Script Concepts
3185 * Script Format:: Linker Script Format
3186 * Simple Example:: Simple Linker Script Example
3187 * Simple Commands:: Simple Linker Script Commands
3188 * Assignments:: Assigning Values to Symbols
3189 * SECTIONS:: SECTIONS Command
3190 * MEMORY:: MEMORY Command
3191 * PHDRS:: PHDRS Command
3192 * VERSION:: VERSION Command
3193 * Expressions:: Expressions in Linker Scripts
3194 * Implicit Linker Scripts:: Implicit Linker Scripts
3197 @node Basic Script Concepts
3198 @section Basic Linker Script Concepts
3199 @cindex linker script concepts
3200 We need to define some basic concepts and vocabulary in order to
3201 describe the linker script language.
3203 The linker combines input files into a single output file. The output
3204 file and each input file are in a special data format known as an
3205 @dfn{object file format}. Each file is called an @dfn{object file}.
3206 The output file is often called an @dfn{executable}, but for our
3207 purposes we will also call it an object file. Each object file has,
3208 among other things, a list of @dfn{sections}. We sometimes refer to a
3209 section in an input file as an @dfn{input section}; similarly, a section
3210 in the output file is an @dfn{output section}.
3212 Each section in an object file has a name and a size. Most sections
3213 also have an associated block of data, known as the @dfn{section
3214 contents}. A section may be marked as @dfn{loadable}, which means that
3215 the contents should be loaded into memory when the output file is run.
3216 A section with no contents may be @dfn{allocatable}, which means that an
3217 area in memory should be set aside, but nothing in particular should be
3218 loaded there (in some cases this memory must be zeroed out). A section
3219 which is neither loadable nor allocatable typically contains some sort
3220 of debugging information.
3222 Every loadable or allocatable output section has two addresses. The
3223 first is the @dfn{VMA}, or virtual memory address. This is the address
3224 the section will have when the output file is run. The second is the
3225 @dfn{LMA}, or load memory address. This is the address at which the
3226 section will be loaded. In most cases the two addresses will be the
3227 same. An example of when they might be different is when a data section
3228 is loaded into ROM, and then copied into RAM when the program starts up
3229 (this technique is often used to initialize global variables in a ROM
3230 based system). In this case the ROM address would be the LMA, and the
3231 RAM address would be the VMA.
3233 You can see the sections in an object file by using the @code{objdump}
3234 program with the @samp{-h} option.
3236 Every object file also has a list of @dfn{symbols}, known as the
3237 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3238 has a name, and each defined symbol has an address, among other
3239 information. If you compile a C or C++ program into an object file, you
3240 will get a defined symbol for every defined function and global or
3241 static variable. Every undefined function or global variable which is
3242 referenced in the input file will become an undefined symbol.
3244 You can see the symbols in an object file by using the @code{nm}
3245 program, or by using the @code{objdump} program with the @samp{-t}
3249 @section Linker Script Format
3250 @cindex linker script format
3251 Linker scripts are text files.
3253 You write a linker script as a series of commands. Each command is
3254 either a keyword, possibly followed by arguments, or an assignment to a
3255 symbol. You may separate commands using semicolons. Whitespace is
3258 Strings such as file or format names can normally be entered directly.
3259 If the file name contains a character such as a comma which would
3260 otherwise serve to separate file names, you may put the file name in
3261 double quotes. There is no way to use a double quote character in a
3264 You may include comments in linker scripts just as in C, delimited by
3265 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3268 @node Simple Example
3269 @section Simple Linker Script Example
3270 @cindex linker script example
3271 @cindex example of linker script
3272 Many linker scripts are fairly simple.
3274 The simplest possible linker script has just one command:
3275 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3276 memory layout of the output file.
3278 The @samp{SECTIONS} command is a powerful command. Here we will
3279 describe a simple use of it. Let's assume your program consists only of
3280 code, initialized data, and uninitialized data. These will be in the
3281 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3282 Let's assume further that these are the only sections which appear in
3285 For this example, let's say that the code should be loaded at address
3286 0x10000, and that the data should start at address 0x8000000. Here is a
3287 linker script which will do that:
3292 .text : @{ *(.text) @}
3294 .data : @{ *(.data) @}
3295 .bss : @{ *(.bss) @}
3299 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3300 followed by a series of symbol assignments and output section
3301 descriptions enclosed in curly braces.
3303 The first line inside the @samp{SECTIONS} command of the above example
3304 sets the value of the special symbol @samp{.}, which is the location
3305 counter. If you do not specify the address of an output section in some
3306 other way (other ways are described later), the address is set from the
3307 current value of the location counter. The location counter is then
3308 incremented by the size of the output section. At the start of the
3309 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3311 The second line defines an output section, @samp{.text}. The colon is
3312 required syntax which may be ignored for now. Within the curly braces
3313 after the output section name, you list the names of the input sections
3314 which should be placed into this output section. The @samp{*} is a
3315 wildcard which matches any file name. The expression @samp{*(.text)}
3316 means all @samp{.text} input sections in all input files.
3318 Since the location counter is @samp{0x10000} when the output section
3319 @samp{.text} is defined, the linker will set the address of the
3320 @samp{.text} section in the output file to be @samp{0x10000}.
3322 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3323 the output file. The linker will place the @samp{.data} output section
3324 at address @samp{0x8000000}. After the linker places the @samp{.data}
3325 output section, the value of the location counter will be
3326 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3327 effect is that the linker will place the @samp{.bss} output section
3328 immediately after the @samp{.data} output section in memory.
3330 The linker will ensure that each output section has the required
3331 alignment, by increasing the location counter if necessary. In this
3332 example, the specified addresses for the @samp{.text} and @samp{.data}
3333 sections will probably satisfy any alignment constraints, but the linker
3334 may have to create a small gap between the @samp{.data} and @samp{.bss}
3337 That's it! That's a simple and complete linker script.
3339 @node Simple Commands
3340 @section Simple Linker Script Commands
3341 @cindex linker script simple commands
3342 In this section we describe the simple linker script commands.
3345 * Entry Point:: Setting the entry point
3346 * File Commands:: Commands dealing with files
3347 @ifclear SingleFormat
3348 * Format Commands:: Commands dealing with object file formats
3351 * REGION_ALIAS:: Assign alias names to memory regions
3352 * Miscellaneous Commands:: Other linker script commands
3356 @subsection Setting the Entry Point
3357 @kindex ENTRY(@var{symbol})
3358 @cindex start of execution
3359 @cindex first instruction
3361 The first instruction to execute in a program is called the @dfn{entry
3362 point}. You can use the @code{ENTRY} linker script command to set the
3363 entry point. The argument is a symbol name:
3368 There are several ways to set the entry point. The linker will set the
3369 entry point by trying each of the following methods in order, and
3370 stopping when one of them succeeds:
3373 the @samp{-e} @var{entry} command-line option;
3375 the @code{ENTRY(@var{symbol})} command in a linker script;
3377 the value of a target specific symbol, if it is defined; For many
3378 targets this is @code{start}, but PE and BeOS based systems for example
3379 check a list of possible entry symbols, matching the first one found.
3381 the address of the first byte of the @samp{.text} section, if present;
3383 The address @code{0}.
3387 @subsection Commands Dealing with Files
3388 @cindex linker script file commands
3389 Several linker script commands deal with files.
3392 @item INCLUDE @var{filename}
3393 @kindex INCLUDE @var{filename}
3394 @cindex including a linker script
3395 Include the linker script @var{filename} at this point. The file will
3396 be searched for in the current directory, and in any directory specified
3397 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3400 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3401 @code{SECTIONS} commands, or in output section descriptions.
3403 @item INPUT(@var{file}, @var{file}, @dots{})
3404 @itemx INPUT(@var{file} @var{file} @dots{})
3405 @kindex INPUT(@var{files})
3406 @cindex input files in linker scripts
3407 @cindex input object files in linker scripts
3408 @cindex linker script input object files
3409 The @code{INPUT} command directs the linker to include the named files
3410 in the link, as though they were named on the command line.
3412 For example, if you always want to include @file{subr.o} any time you do
3413 a link, but you can't be bothered to put it on every link command line,
3414 then you can put @samp{INPUT (subr.o)} in your linker script.
3416 In fact, if you like, you can list all of your input files in the linker
3417 script, and then invoke the linker with nothing but a @samp{-T} option.
3419 In case a @dfn{sysroot prefix} is configured, and the filename starts
3420 with the @samp{/} character, and the script being processed was
3421 located inside the @dfn{sysroot prefix}, the filename will be looked
3422 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3423 open the file in the current directory. If it is not found, the
3424 linker will search through the archive library search path.
3425 The @dfn{sysroot prefix} can also be forced by specifying @code{=}
3426 as the first character in the filename path, or prefixing the filename
3427 path with @code{$SYSROOT}. See also the description of @samp{-L} in
3428 @ref{Options,,Command Line Options}.
3430 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3431 name to @code{lib@var{file}.a}, as with the command line argument
3434 When you use the @code{INPUT} command in an implicit linker script, the
3435 files will be included in the link at the point at which the linker
3436 script file is included. This can affect archive searching.
3438 @item GROUP(@var{file}, @var{file}, @dots{})
3439 @itemx GROUP(@var{file} @var{file} @dots{})
3440 @kindex GROUP(@var{files})
3441 @cindex grouping input files
3442 The @code{GROUP} command is like @code{INPUT}, except that the named
3443 files should all be archives, and they are searched repeatedly until no
3444 new undefined references are created. See the description of @samp{-(}
3445 in @ref{Options,,Command Line Options}.
3447 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3448 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3449 @kindex AS_NEEDED(@var{files})
3450 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3451 commands, among other filenames. The files listed will be handled
3452 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3453 with the exception of ELF shared libraries, that will be added only
3454 when they are actually needed. This construct essentially enables
3455 @option{--as-needed} option for all the files listed inside of it
3456 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3459 @item OUTPUT(@var{filename})
3460 @kindex OUTPUT(@var{filename})
3461 @cindex output file name in linker script
3462 The @code{OUTPUT} command names the output file. Using
3463 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3464 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3465 Line Options}). If both are used, the command line option takes
3468 You can use the @code{OUTPUT} command to define a default name for the
3469 output file other than the usual default of @file{a.out}.
3471 @item SEARCH_DIR(@var{path})
3472 @kindex SEARCH_DIR(@var{path})
3473 @cindex library search path in linker script
3474 @cindex archive search path in linker script
3475 @cindex search path in linker script
3476 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3477 @command{ld} looks for archive libraries. Using
3478 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3479 on the command line (@pxref{Options,,Command Line Options}). If both
3480 are used, then the linker will search both paths. Paths specified using
3481 the command line option are searched first.
3483 @item STARTUP(@var{filename})
3484 @kindex STARTUP(@var{filename})
3485 @cindex first input file
3486 The @code{STARTUP} command is just like the @code{INPUT} command, except
3487 that @var{filename} will become the first input file to be linked, as
3488 though it were specified first on the command line. This may be useful
3489 when using a system in which the entry point is always the start of the
3493 @ifclear SingleFormat
3494 @node Format Commands
3495 @subsection Commands Dealing with Object File Formats
3496 A couple of linker script commands deal with object file formats.
3499 @item OUTPUT_FORMAT(@var{bfdname})
3500 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3501 @kindex OUTPUT_FORMAT(@var{bfdname})
3502 @cindex output file format in linker script
3503 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3504 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3505 exactly like using @samp{--oformat @var{bfdname}} on the command line
3506 (@pxref{Options,,Command Line Options}). If both are used, the command
3507 line option takes precedence.
3509 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3510 formats based on the @samp{-EB} and @samp{-EL} command line options.
3511 This permits the linker script to set the output format based on the
3514 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3515 will be the first argument, @var{default}. If @samp{-EB} is used, the
3516 output format will be the second argument, @var{big}. If @samp{-EL} is
3517 used, the output format will be the third argument, @var{little}.
3519 For example, the default linker script for the MIPS ELF target uses this
3522 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3524 This says that the default format for the output file is
3525 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3526 option, the output file will be created in the @samp{elf32-littlemips}
3529 @item TARGET(@var{bfdname})
3530 @kindex TARGET(@var{bfdname})
3531 @cindex input file format in linker script
3532 The @code{TARGET} command names the BFD format to use when reading input
3533 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3534 This command is like using @samp{-b @var{bfdname}} on the command line
3535 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3536 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3537 command is also used to set the format for the output file. @xref{BFD}.
3542 @subsection Assign alias names to memory regions
3543 @kindex REGION_ALIAS(@var{alias}, @var{region})
3544 @cindex region alias
3545 @cindex region names
3547 Alias names can be added to existing memory regions created with the
3548 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3551 REGION_ALIAS(@var{alias}, @var{region})
3554 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3555 memory region @var{region}. This allows a flexible mapping of output sections
3556 to memory regions. An example follows.
3558 Suppose we have an application for embedded systems which come with various
3559 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3560 that allows code execution or data storage. Some may have a read-only,
3561 non-volatile memory @code{ROM} that allows code execution and read-only data
3562 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3563 read-only data access and no code execution capability. We have four output
3568 @code{.text} program code;
3570 @code{.rodata} read-only data;
3572 @code{.data} read-write initialized data;
3574 @code{.bss} read-write zero initialized data.
3577 The goal is to provide a linker command file that contains a system independent
3578 part defining the output sections and a system dependent part mapping the
3579 output sections to the memory regions available on the system. Our embedded
3580 systems come with three different memory setups @code{A}, @code{B} and
3582 @multitable @columnfractions .25 .25 .25 .25
3583 @item Section @tab Variant A @tab Variant B @tab Variant C
3584 @item .text @tab RAM @tab ROM @tab ROM
3585 @item .rodata @tab RAM @tab ROM @tab ROM2
3586 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3587 @item .bss @tab RAM @tab RAM @tab RAM
3589 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3590 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3591 the load address of the @code{.data} section starts in all three variants at
3592 the end of the @code{.rodata} section.
3594 The base linker script that deals with the output sections follows. It
3595 includes the system dependent @code{linkcmds.memory} file that describes the
3598 INCLUDE linkcmds.memory
3611 .data : AT (rodata_end)
3616 data_size = SIZEOF(.data);
3617 data_load_start = LOADADDR(.data);
3625 Now we need three different @code{linkcmds.memory} files to define memory
3626 regions and alias names. The content of @code{linkcmds.memory} for the three
3627 variants @code{A}, @code{B} and @code{C}:
3630 Here everything goes into the @code{RAM}.
3634 RAM : ORIGIN = 0, LENGTH = 4M
3637 REGION_ALIAS("REGION_TEXT", RAM);
3638 REGION_ALIAS("REGION_RODATA", RAM);
3639 REGION_ALIAS("REGION_DATA", RAM);
3640 REGION_ALIAS("REGION_BSS", RAM);
3643 Program code and read-only data go into the @code{ROM}. Read-write data goes
3644 into the @code{RAM}. An image of the initialized data is loaded into the
3645 @code{ROM} and will be copied during system start into the @code{RAM}.
3649 ROM : ORIGIN = 0, LENGTH = 3M
3650 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3653 REGION_ALIAS("REGION_TEXT", ROM);
3654 REGION_ALIAS("REGION_RODATA", ROM);
3655 REGION_ALIAS("REGION_DATA", RAM);
3656 REGION_ALIAS("REGION_BSS", RAM);
3659 Program code goes into the @code{ROM}. Read-only data goes into the
3660 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3661 initialized data is loaded into the @code{ROM2} and will be copied during
3662 system start into the @code{RAM}.
3666 ROM : ORIGIN = 0, LENGTH = 2M
3667 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3668 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3671 REGION_ALIAS("REGION_TEXT", ROM);
3672 REGION_ALIAS("REGION_RODATA", ROM2);
3673 REGION_ALIAS("REGION_DATA", RAM);
3674 REGION_ALIAS("REGION_BSS", RAM);
3678 It is possible to write a common system initialization routine to copy the
3679 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3684 extern char data_start [];
3685 extern char data_size [];
3686 extern char data_load_start [];
3688 void copy_data(void)
3690 if (data_start != data_load_start)
3692 memcpy(data_start, data_load_start, (size_t) data_size);
3697 @node Miscellaneous Commands
3698 @subsection Other Linker Script Commands
3699 There are a few other linker scripts commands.
3702 @item ASSERT(@var{exp}, @var{message})
3704 @cindex assertion in linker script
3705 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3706 with an error code, and print @var{message}.
3708 Note that assertions are checked before the final stages of linking
3709 take place. This means that expressions involving symbols PROVIDEd
3710 inside section definitions will fail if the user has not set values
3711 for those symbols. The only exception to this rule is PROVIDEd
3712 symbols that just reference dot. Thus an assertion like this:
3717 PROVIDE (__stack = .);
3718 PROVIDE (__stack_size = 0x100);
3719 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3723 will fail if @code{__stack_size} is not defined elsewhere. Symbols
3724 PROVIDEd outside of section definitions are evaluated earlier, so they
3725 can be used inside ASSERTions. Thus:
3728 PROVIDE (__stack_size = 0x100);
3731 PROVIDE (__stack = .);
3732 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3738 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3740 @cindex undefined symbol in linker script
3741 Force @var{symbol} to be entered in the output file as an undefined
3742 symbol. Doing this may, for example, trigger linking of additional
3743 modules from standard libraries. You may list several @var{symbol}s for
3744 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3745 command has the same effect as the @samp{-u} command-line option.
3747 @item FORCE_COMMON_ALLOCATION
3748 @kindex FORCE_COMMON_ALLOCATION
3749 @cindex common allocation in linker script
3750 This command has the same effect as the @samp{-d} command-line option:
3751 to make @command{ld} assign space to common symbols even if a relocatable
3752 output file is specified (@samp{-r}).
3754 @item INHIBIT_COMMON_ALLOCATION
3755 @kindex INHIBIT_COMMON_ALLOCATION
3756 @cindex common allocation in linker script
3757 This command has the same effect as the @samp{--no-define-common}
3758 command-line option: to make @code{ld} omit the assignment of addresses
3759 to common symbols even for a non-relocatable output file.
3761 @item FORCE_GROUP_ALLOCATION
3762 @kindex FORCE_GROUP_ALLOCATION
3763 @cindex group allocation in linker script
3764 @cindex section groups
3766 This command has the same effect as the
3767 @samp{--force-group-allocation} command-line option: to make
3768 @command{ld} place section group members like normal input sections,
3769 and to delete the section groups even if a relocatable output file is
3770 specified (@samp{-r}).
3772 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3774 @cindex insert user script into default script
3775 This command is typically used in a script specified by @samp{-T} to
3776 augment the default @code{SECTIONS} with, for example, overlays. It
3777 inserts all prior linker script statements after (or before)
3778 @var{output_section}, and also causes @samp{-T} to not override the
3779 default linker script. The exact insertion point is as for orphan
3780 sections. @xref{Location Counter}. The insertion happens after the
3781 linker has mapped input sections to output sections. Prior to the
3782 insertion, since @samp{-T} scripts are parsed before the default
3783 linker script, statements in the @samp{-T} script occur before the
3784 default linker script statements in the internal linker representation
3785 of the script. In particular, input section assignments will be made
3786 to @samp{-T} output sections before those in the default script. Here
3787 is an example of how a @samp{-T} script using @code{INSERT} might look:
3794 .ov1 @{ ov1*(.text) @}
3795 .ov2 @{ ov2*(.text) @}
3801 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3802 @kindex NOCROSSREFS(@var{sections})
3803 @cindex cross references
3804 This command may be used to tell @command{ld} to issue an error about any
3805 references among certain output sections.
3807 In certain types of programs, particularly on embedded systems when
3808 using overlays, when one section is loaded into memory, another section
3809 will not be. Any direct references between the two sections would be
3810 errors. For example, it would be an error if code in one section called
3811 a function defined in the other section.
3813 The @code{NOCROSSREFS} command takes a list of output section names. If
3814 @command{ld} detects any cross references between the sections, it reports
3815 an error and returns a non-zero exit status. Note that the
3816 @code{NOCROSSREFS} command uses output section names, not input section
3819 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
3820 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
3821 @cindex cross references
3822 This command may be used to tell @command{ld} to issue an error about any
3823 references to one section from a list of other sections.
3825 The @code{NOCROSSREFS} command is useful when ensuring that two or more
3826 output sections are entirely independent but there are situations where
3827 a one-way dependency is needed. For example, in a multi-core application
3828 there may be shared code that can be called from each core but for safety
3829 must never call back.
3831 The @code{NOCROSSREFS_TO} command takes a list of output section names.
3832 The first section can not be referenced from any of the other sections.
3833 If @command{ld} detects any references to the first section from any of
3834 the other sections, it reports an error and returns a non-zero exit
3835 status. Note that the @code{NOCROSSREFS_TO} command uses output section
3836 names, not input section names.
3838 @ifclear SingleFormat
3839 @item OUTPUT_ARCH(@var{bfdarch})
3840 @kindex OUTPUT_ARCH(@var{bfdarch})
3841 @cindex machine architecture
3842 @cindex architecture
3843 Specify a particular output machine architecture. The argument is one
3844 of the names used by the BFD library (@pxref{BFD}). You can see the
3845 architecture of an object file by using the @code{objdump} program with
3846 the @samp{-f} option.
3849 @item LD_FEATURE(@var{string})
3850 @kindex LD_FEATURE(@var{string})
3851 This command may be used to modify @command{ld} behavior. If
3852 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3853 in a script are simply treated as numbers everywhere.
3854 @xref{Expression Section}.
3858 @section Assigning Values to Symbols
3859 @cindex assignment in scripts
3860 @cindex symbol definition, scripts
3861 @cindex variables, defining
3862 You may assign a value to a symbol in a linker script. This will define
3863 the symbol and place it into the symbol table with a global scope.
3866 * Simple Assignments:: Simple Assignments
3869 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3870 * Source Code Reference:: How to use a linker script defined symbol in source code
3873 @node Simple Assignments
3874 @subsection Simple Assignments
3876 You may assign to a symbol using any of the C assignment operators:
3879 @item @var{symbol} = @var{expression} ;
3880 @itemx @var{symbol} += @var{expression} ;
3881 @itemx @var{symbol} -= @var{expression} ;
3882 @itemx @var{symbol} *= @var{expression} ;
3883 @itemx @var{symbol} /= @var{expression} ;
3884 @itemx @var{symbol} <<= @var{expression} ;
3885 @itemx @var{symbol} >>= @var{expression} ;
3886 @itemx @var{symbol} &= @var{expression} ;
3887 @itemx @var{symbol} |= @var{expression} ;
3890 The first case will define @var{symbol} to the value of
3891 @var{expression}. In the other cases, @var{symbol} must already be
3892 defined, and the value will be adjusted accordingly.
3894 The special symbol name @samp{.} indicates the location counter. You
3895 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3897 The semicolon after @var{expression} is required.
3899 Expressions are defined below; see @ref{Expressions}.
3901 You may write symbol assignments as commands in their own right, or as
3902 statements within a @code{SECTIONS} command, or as part of an output
3903 section description in a @code{SECTIONS} command.
3905 The section of the symbol will be set from the section of the
3906 expression; for more information, see @ref{Expression Section}.
3908 Here is an example showing the three different places that symbol
3909 assignments may be used:
3920 _bdata = (. + 3) & ~ 3;
3921 .data : @{ *(.data) @}
3925 In this example, the symbol @samp{floating_point} will be defined as
3926 zero. The symbol @samp{_etext} will be defined as the address following
3927 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3928 defined as the address following the @samp{.text} output section aligned
3929 upward to a 4 byte boundary.
3934 For ELF targeted ports, define a symbol that will be hidden and won't be
3935 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
3937 Here is the example from @ref{Simple Assignments}, rewritten to use
3941 HIDDEN(floating_point = 0);
3949 HIDDEN(_bdata = (. + 3) & ~ 3);
3950 .data : @{ *(.data) @}
3954 In this case none of the three symbols will be visible outside this module.
3959 In some cases, it is desirable for a linker script to define a symbol
3960 only if it is referenced and is not defined by any object included in
3961 the link. For example, traditional linkers defined the symbol
3962 @samp{etext}. However, ANSI C requires that the user be able to use
3963 @samp{etext} as a function name without encountering an error. The
3964 @code{PROVIDE} keyword may be used to define a symbol, such as
3965 @samp{etext}, only if it is referenced but not defined. The syntax is
3966 @code{PROVIDE(@var{symbol} = @var{expression})}.
3968 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3981 In this example, if the program defines @samp{_etext} (with a leading
3982 underscore), the linker will give a multiple definition error. If, on
3983 the other hand, the program defines @samp{etext} (with no leading
3984 underscore), the linker will silently use the definition in the program.
3985 If the program references @samp{etext} but does not define it, the
3986 linker will use the definition in the linker script.
3988 @node PROVIDE_HIDDEN
3989 @subsection PROVIDE_HIDDEN
3990 @cindex PROVIDE_HIDDEN
3991 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3992 hidden and won't be exported.
3994 @node Source Code Reference
3995 @subsection Source Code Reference
3997 Accessing a linker script defined variable from source code is not
3998 intuitive. In particular a linker script symbol is not equivalent to
3999 a variable declaration in a high level language, it is instead a
4000 symbol that does not have a value.
4002 Before going further, it is important to note that compilers often
4003 transform names in the source code into different names when they are
4004 stored in the symbol table. For example, Fortran compilers commonly
4005 prepend or append an underscore, and C++ performs extensive @samp{name
4006 mangling}. Therefore there might be a discrepancy between the name
4007 of a variable as it is used in source code and the name of the same
4008 variable as it is defined in a linker script. For example in C a
4009 linker script variable might be referred to as:
4015 But in the linker script it might be defined as:
4021 In the remaining examples however it is assumed that no name
4022 transformation has taken place.
4024 When a symbol is declared in a high level language such as C, two
4025 things happen. The first is that the compiler reserves enough space
4026 in the program's memory to hold the @emph{value} of the symbol. The
4027 second is that the compiler creates an entry in the program's symbol
4028 table which holds the symbol's @emph{address}. ie the symbol table
4029 contains the address of the block of memory holding the symbol's
4030 value. So for example the following C declaration, at file scope:
4036 creates an entry called @samp{foo} in the symbol table. This entry
4037 holds the address of an @samp{int} sized block of memory where the
4038 number 1000 is initially stored.
4040 When a program references a symbol the compiler generates code that
4041 first accesses the symbol table to find the address of the symbol's
4042 memory block and then code to read the value from that memory block.
4049 looks up the symbol @samp{foo} in the symbol table, gets the address
4050 associated with this symbol and then writes the value 1 into that
4057 looks up the symbol @samp{foo} in the symbol table, gets its address
4058 and then copies this address into the block of memory associated with
4059 the variable @samp{a}.
4061 Linker scripts symbol declarations, by contrast, create an entry in
4062 the symbol table but do not assign any memory to them. Thus they are
4063 an address without a value. So for example the linker script definition:
4069 creates an entry in the symbol table called @samp{foo} which holds
4070 the address of memory location 1000, but nothing special is stored at
4071 address 1000. This means that you cannot access the @emph{value} of a
4072 linker script defined symbol - it has no value - all you can do is
4073 access the @emph{address} of a linker script defined symbol.
4075 Hence when you are using a linker script defined symbol in source code
4076 you should always take the address of the symbol, and never attempt to
4077 use its value. For example suppose you want to copy the contents of a
4078 section of memory called .ROM into a section called .FLASH and the
4079 linker script contains these declarations:
4083 start_of_ROM = .ROM;
4084 end_of_ROM = .ROM + sizeof (.ROM);
4085 start_of_FLASH = .FLASH;
4089 Then the C source code to perform the copy would be:
4093 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
4095 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
4099 Note the use of the @samp{&} operators. These are correct.
4100 Alternatively the symbols can be treated as the names of vectors or
4101 arrays and then the code will again work as expected:
4105 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
4107 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
4111 Note how using this method does not require the use of @samp{&}
4115 @section SECTIONS Command
4117 The @code{SECTIONS} command tells the linker how to map input sections
4118 into output sections, and how to place the output sections in memory.
4120 The format of the @code{SECTIONS} command is:
4124 @var{sections-command}
4125 @var{sections-command}
4130 Each @var{sections-command} may of be one of the following:
4134 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
4136 a symbol assignment (@pxref{Assignments})
4138 an output section description
4140 an overlay description
4143 The @code{ENTRY} command and symbol assignments are permitted inside the
4144 @code{SECTIONS} command for convenience in using the location counter in
4145 those commands. This can also make the linker script easier to
4146 understand because you can use those commands at meaningful points in
4147 the layout of the output file.
4149 Output section descriptions and overlay descriptions are described
4152 If you do not use a @code{SECTIONS} command in your linker script, the
4153 linker will place each input section into an identically named output
4154 section in the order that the sections are first encountered in the
4155 input files. If all input sections are present in the first file, for
4156 example, the order of sections in the output file will match the order
4157 in the first input file. The first section will be at address zero.
4160 * Output Section Description:: Output section description
4161 * Output Section Name:: Output section name
4162 * Output Section Address:: Output section address
4163 * Input Section:: Input section description
4164 * Output Section Data:: Output section data
4165 * Output Section Keywords:: Output section keywords
4166 * Output Section Discarding:: Output section discarding
4167 * Output Section Attributes:: Output section attributes
4168 * Overlay Description:: Overlay description
4171 @node Output Section Description
4172 @subsection Output Section Description
4173 The full description of an output section looks like this:
4176 @var{section} [@var{address}] [(@var{type})] :
4178 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4179 [SUBALIGN(@var{subsection_align})]
4182 @var{output-section-command}
4183 @var{output-section-command}
4185 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4189 Most output sections do not use most of the optional section attributes.
4191 The whitespace around @var{section} is required, so that the section
4192 name is unambiguous. The colon and the curly braces are also required.
4193 The comma at the end may be required if a @var{fillexp} is used and
4194 the next @var{sections-command} looks like a continuation of the expression.
4195 The line breaks and other white space are optional.
4197 Each @var{output-section-command} may be one of the following:
4201 a symbol assignment (@pxref{Assignments})
4203 an input section description (@pxref{Input Section})
4205 data values to include directly (@pxref{Output Section Data})
4207 a special output section keyword (@pxref{Output Section Keywords})
4210 @node Output Section Name
4211 @subsection Output Section Name
4212 @cindex name, section
4213 @cindex section name
4214 The name of the output section is @var{section}. @var{section} must
4215 meet the constraints of your output format. In formats which only
4216 support a limited number of sections, such as @code{a.out}, the name
4217 must be one of the names supported by the format (@code{a.out}, for
4218 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4219 output format supports any number of sections, but with numbers and not
4220 names (as is the case for Oasys), the name should be supplied as a
4221 quoted numeric string. A section name may consist of any sequence of
4222 characters, but a name which contains any unusual characters such as
4223 commas must be quoted.
4225 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4228 @node Output Section Address
4229 @subsection Output Section Address
4230 @cindex address, section
4231 @cindex section address
4232 The @var{address} is an expression for the VMA (the virtual memory
4233 address) of the output section. This address is optional, but if it
4234 is provided then the output address will be set exactly as specified.
4236 If the output address is not specified then one will be chosen for the
4237 section, based on the heuristic below. This address will be adjusted
4238 to fit the alignment requirement of the output section. The
4239 alignment requirement is the strictest alignment of any input section
4240 contained within the output section.
4242 The output section address heuristic is as follows:
4246 If an output memory @var{region} is set for the section then it
4247 is added to this region and its address will be the next free address
4251 If the MEMORY command has been used to create a list of memory
4252 regions then the first region which has attributes compatible with the
4253 section is selected to contain it. The section's output address will
4254 be the next free address in that region; @ref{MEMORY}.
4257 If no memory regions were specified, or none match the section then
4258 the output address will be based on the current value of the location
4266 .text . : @{ *(.text) @}
4273 .text : @{ *(.text) @}
4277 are subtly different. The first will set the address of the
4278 @samp{.text} output section to the current value of the location
4279 counter. The second will set it to the current value of the location
4280 counter aligned to the strictest alignment of any of the @samp{.text}
4283 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4284 For example, if you want to align the section on a 0x10 byte boundary,
4285 so that the lowest four bits of the section address are zero, you could
4286 do something like this:
4288 .text ALIGN(0x10) : @{ *(.text) @}
4291 This works because @code{ALIGN} returns the current location counter
4292 aligned upward to the specified value.
4294 Specifying @var{address} for a section will change the value of the
4295 location counter, provided that the section is non-empty. (Empty
4296 sections are ignored).
4299 @subsection Input Section Description
4300 @cindex input sections
4301 @cindex mapping input sections to output sections
4302 The most common output section command is an input section description.
4304 The input section description is the most basic linker script operation.
4305 You use output sections to tell the linker how to lay out your program
4306 in memory. You use input section descriptions to tell the linker how to
4307 map the input files into your memory layout.
4310 * Input Section Basics:: Input section basics
4311 * Input Section Wildcards:: Input section wildcard patterns
4312 * Input Section Common:: Input section for common symbols
4313 * Input Section Keep:: Input section and garbage collection
4314 * Input Section Example:: Input section example
4317 @node Input Section Basics
4318 @subsubsection Input Section Basics
4319 @cindex input section basics
4320 An input section description consists of a file name optionally followed
4321 by a list of section names in parentheses.
4323 The file name and the section name may be wildcard patterns, which we
4324 describe further below (@pxref{Input Section Wildcards}).
4326 The most common input section description is to include all input
4327 sections with a particular name in the output section. For example, to
4328 include all input @samp{.text} sections, you would write:
4333 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4334 @cindex EXCLUDE_FILE
4335 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4336 match all files except the ones specified in the EXCLUDE_FILE list. For
4339 EXCLUDE_FILE (*crtend.o *otherfile.o) *(.ctors)
4342 will cause all .ctors sections from all files except @file{crtend.o}
4343 and @file{otherfile.o} to be included. The EXCLUDE_FILE can also be
4344 placed inside the section list, for example:
4346 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4349 The result of this is identically to the previous example. Supporting
4350 two syntaxes for EXCLUDE_FILE is useful if the section list contains
4351 more than one section, as described below.
4353 There are two ways to include more than one section:
4359 The difference between these is the order in which the @samp{.text} and
4360 @samp{.rdata} input sections will appear in the output section. In the
4361 first example, they will be intermingled, appearing in the same order as
4362 they are found in the linker input. In the second example, all
4363 @samp{.text} input sections will appear first, followed by all
4364 @samp{.rdata} input sections.
4366 When using EXCLUDE_FILE with more than one section, if the exclusion
4367 is within the section list then the exclusion only applies to the
4368 immediately following section, for example:
4370 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4373 will cause all @samp{.text} sections from all files except
4374 @file{somefile.o} to be included, while all @samp{.rdata} sections
4375 from all files, including @file{somefile.o}, will be included. To
4376 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4377 could be modified to:
4379 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4382 Alternatively, placing the EXCLUDE_FILE outside of the section list,
4383 before the input file selection, will cause the exclusion to apply for
4384 all sections. Thus the previous example can be rewritten as:
4386 EXCLUDE_FILE (*somefile.o) *(.text .rdata)
4389 You can specify a file name to include sections from a particular file.
4390 You would do this if one or more of your files contain special data that
4391 needs to be at a particular location in memory. For example:
4396 To refine the sections that are included based on the section flags
4397 of an input section, INPUT_SECTION_FLAGS may be used.
4399 Here is a simple example for using Section header flags for ELF sections:
4404 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4405 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4410 In this example, the output section @samp{.text} will be comprised of any
4411 input section matching the name *(.text) whose section header flags
4412 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4413 @samp{.text2} will be comprised of any input section matching the name *(.text)
4414 whose section header flag @code{SHF_WRITE} is clear.
4416 You can also specify files within archives by writing a pattern
4417 matching the archive, a colon, then the pattern matching the file,
4418 with no whitespace around the colon.
4422 matches file within archive
4424 matches the whole archive
4426 matches file but not one in an archive
4429 Either one or both of @samp{archive} and @samp{file} can contain shell
4430 wildcards. On DOS based file systems, the linker will assume that a
4431 single letter followed by a colon is a drive specifier, so
4432 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4433 within an archive called @samp{c}. @samp{archive:file} filespecs may
4434 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4435 other linker script contexts. For instance, you cannot extract a file
4436 from an archive by using @samp{archive:file} in an @code{INPUT}
4439 If you use a file name without a list of sections, then all sections in
4440 the input file will be included in the output section. This is not
4441 commonly done, but it may by useful on occasion. For example:
4446 When you use a file name which is not an @samp{archive:file} specifier
4447 and does not contain any wild card
4448 characters, the linker will first see if you also specified the file
4449 name on the linker command line or in an @code{INPUT} command. If you
4450 did not, the linker will attempt to open the file as an input file, as
4451 though it appeared on the command line. Note that this differs from an
4452 @code{INPUT} command, because the linker will not search for the file in
4453 the archive search path.
4455 @node Input Section Wildcards
4456 @subsubsection Input Section Wildcard Patterns
4457 @cindex input section wildcards
4458 @cindex wildcard file name patterns
4459 @cindex file name wildcard patterns
4460 @cindex section name wildcard patterns
4461 In an input section description, either the file name or the section
4462 name or both may be wildcard patterns.
4464 The file name of @samp{*} seen in many examples is a simple wildcard
4465 pattern for the file name.
4467 The wildcard patterns are like those used by the Unix shell.
4471 matches any number of characters
4473 matches any single character
4475 matches a single instance of any of the @var{chars}; the @samp{-}
4476 character may be used to specify a range of characters, as in
4477 @samp{[a-z]} to match any lower case letter
4479 quotes the following character
4482 When a file name is matched with a wildcard, the wildcard characters
4483 will not match a @samp{/} character (used to separate directory names on
4484 Unix). A pattern consisting of a single @samp{*} character is an
4485 exception; it will always match any file name, whether it contains a
4486 @samp{/} or not. In a section name, the wildcard characters will match
4487 a @samp{/} character.
4489 File name wildcard patterns only match files which are explicitly
4490 specified on the command line or in an @code{INPUT} command. The linker
4491 does not search directories to expand wildcards.
4493 If a file name matches more than one wildcard pattern, or if a file name
4494 appears explicitly and is also matched by a wildcard pattern, the linker
4495 will use the first match in the linker script. For example, this
4496 sequence of input section descriptions is probably in error, because the
4497 @file{data.o} rule will not be used:
4499 .data : @{ *(.data) @}
4500 .data1 : @{ data.o(.data) @}
4503 @cindex SORT_BY_NAME
4504 Normally, the linker will place files and sections matched by wildcards
4505 in the order in which they are seen during the link. You can change
4506 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4507 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4508 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4509 into ascending order by name before placing them in the output file.
4511 @cindex SORT_BY_ALIGNMENT
4512 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4513 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4514 descending order by alignment before placing them in the output file.
4515 Larger alignments are placed before smaller alignments in order to
4516 reduce the amount of padding necessary.
4518 @cindex SORT_BY_INIT_PRIORITY
4519 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4520 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4521 ascending order by numerical value of the GCC init_priority attribute
4522 encoded in the section name before placing them in the output file.
4525 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4527 When there are nested section sorting commands in linker script, there
4528 can be at most 1 level of nesting for section sorting commands.
4532 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4533 It will sort the input sections by name first, then by alignment if two
4534 sections have the same name.
4536 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4537 It will sort the input sections by alignment first, then by name if two
4538 sections have the same alignment.
4540 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4541 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4543 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4544 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4546 All other nested section sorting commands are invalid.
4549 When both command line section sorting option and linker script
4550 section sorting command are used, section sorting command always
4551 takes precedence over the command line option.
4553 If the section sorting command in linker script isn't nested, the
4554 command line option will make the section sorting command to be
4555 treated as nested sorting command.
4559 @code{SORT_BY_NAME} (wildcard section pattern ) with
4560 @option{--sort-sections alignment} is equivalent to
4561 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4563 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4564 @option{--sort-section name} is equivalent to
4565 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4568 If the section sorting command in linker script is nested, the
4569 command line option will be ignored.
4572 @code{SORT_NONE} disables section sorting by ignoring the command line
4573 section sorting option.
4575 If you ever get confused about where input sections are going, use the
4576 @samp{-M} linker option to generate a map file. The map file shows
4577 precisely how input sections are mapped to output sections.
4579 This example shows how wildcard patterns might be used to partition
4580 files. This linker script directs the linker to place all @samp{.text}
4581 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4582 The linker will place the @samp{.data} section from all files beginning
4583 with an upper case character in @samp{.DATA}; for all other files, the
4584 linker will place the @samp{.data} section in @samp{.data}.
4588 .text : @{ *(.text) @}
4589 .DATA : @{ [A-Z]*(.data) @}
4590 .data : @{ *(.data) @}
4591 .bss : @{ *(.bss) @}
4596 @node Input Section Common
4597 @subsubsection Input Section for Common Symbols
4598 @cindex common symbol placement
4599 @cindex uninitialized data placement
4600 A special notation is needed for common symbols, because in many object
4601 file formats common symbols do not have a particular input section. The
4602 linker treats common symbols as though they are in an input section
4603 named @samp{COMMON}.
4605 You may use file names with the @samp{COMMON} section just as with any
4606 other input sections. You can use this to place common symbols from a
4607 particular input file in one section while common symbols from other
4608 input files are placed in another section.
4610 In most cases, common symbols in input files will be placed in the
4611 @samp{.bss} section in the output file. For example:
4613 .bss @{ *(.bss) *(COMMON) @}
4616 @cindex scommon section
4617 @cindex small common symbols
4618 Some object file formats have more than one type of common symbol. For
4619 example, the MIPS ELF object file format distinguishes standard common
4620 symbols and small common symbols. In this case, the linker will use a
4621 different special section name for other types of common symbols. In
4622 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4623 symbols and @samp{.scommon} for small common symbols. This permits you
4624 to map the different types of common symbols into memory at different
4628 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4629 notation is now considered obsolete. It is equivalent to
4632 @node Input Section Keep
4633 @subsubsection Input Section and Garbage Collection
4635 @cindex garbage collection
4636 When link-time garbage collection is in use (@samp{--gc-sections}),
4637 it is often useful to mark sections that should not be eliminated.
4638 This is accomplished by surrounding an input section's wildcard entry
4639 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4640 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4642 @node Input Section Example
4643 @subsubsection Input Section Example
4644 The following example is a complete linker script. It tells the linker
4645 to read all of the sections from file @file{all.o} and place them at the
4646 start of output section @samp{outputa} which starts at location
4647 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4648 follows immediately, in the same output section. All of section
4649 @samp{.input2} from @file{foo.o} goes into output section
4650 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4651 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4652 files are written to output section @samp{outputc}.
4680 If an output section's name is the same as the input section's name
4681 and is representable as a C identifier, then the linker will
4682 automatically @pxref{PROVIDE} two symbols: __start_SECNAME and
4683 __stop_SECNAME, where SECNAME is the name of the section. These
4684 indicate the start address and end address of the output section
4685 respectively. Note: most section names are not representable as
4686 C identifiers because they contain a @samp{.} character.
4688 @node Output Section Data
4689 @subsection Output Section Data
4691 @cindex section data
4692 @cindex output section data
4693 @kindex BYTE(@var{expression})
4694 @kindex SHORT(@var{expression})
4695 @kindex LONG(@var{expression})
4696 @kindex QUAD(@var{expression})
4697 @kindex SQUAD(@var{expression})
4698 You can include explicit bytes of data in an output section by using
4699 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4700 an output section command. Each keyword is followed by an expression in
4701 parentheses providing the value to store (@pxref{Expressions}). The
4702 value of the expression is stored at the current value of the location
4705 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4706 store one, two, four, and eight bytes (respectively). After storing the
4707 bytes, the location counter is incremented by the number of bytes
4710 For example, this will store the byte 1 followed by the four byte value
4711 of the symbol @samp{addr}:
4717 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4718 same; they both store an 8 byte, or 64 bit, value. When both host and
4719 target are 32 bits, an expression is computed as 32 bits. In this case
4720 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4721 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4723 If the object file format of the output file has an explicit endianness,
4724 which is the normal case, the value will be stored in that endianness.
4725 When the object file format does not have an explicit endianness, as is
4726 true of, for example, S-records, the value will be stored in the
4727 endianness of the first input object file.
4729 Note---these commands only work inside a section description and not
4730 between them, so the following will produce an error from the linker:
4732 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4734 whereas this will work:
4736 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4739 @kindex FILL(@var{expression})
4740 @cindex holes, filling
4741 @cindex unspecified memory
4742 You may use the @code{FILL} command to set the fill pattern for the
4743 current section. It is followed by an expression in parentheses. Any
4744 otherwise unspecified regions of memory within the section (for example,
4745 gaps left due to the required alignment of input sections) are filled
4746 with the value of the expression, repeated as
4747 necessary. A @code{FILL} statement covers memory locations after the
4748 point at which it occurs in the section definition; by including more
4749 than one @code{FILL} statement, you can have different fill patterns in
4750 different parts of an output section.
4752 This example shows how to fill unspecified regions of memory with the
4758 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4759 section attribute, but it only affects the
4760 part of the section following the @code{FILL} command, rather than the
4761 entire section. If both are used, the @code{FILL} command takes
4762 precedence. @xref{Output Section Fill}, for details on the fill
4765 @node Output Section Keywords
4766 @subsection Output Section Keywords
4767 There are a couple of keywords which can appear as output section
4771 @kindex CREATE_OBJECT_SYMBOLS
4772 @cindex input filename symbols
4773 @cindex filename symbols
4774 @item CREATE_OBJECT_SYMBOLS
4775 The command tells the linker to create a symbol for each input file.
4776 The name of each symbol will be the name of the corresponding input
4777 file. The section of each symbol will be the output section in which
4778 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4780 This is conventional for the a.out object file format. It is not
4781 normally used for any other object file format.
4783 @kindex CONSTRUCTORS
4784 @cindex C++ constructors, arranging in link
4785 @cindex constructors, arranging in link
4787 When linking using the a.out object file format, the linker uses an
4788 unusual set construct to support C++ global constructors and
4789 destructors. When linking object file formats which do not support
4790 arbitrary sections, such as ECOFF and XCOFF, the linker will
4791 automatically recognize C++ global constructors and destructors by name.
4792 For these object file formats, the @code{CONSTRUCTORS} command tells the
4793 linker to place constructor information in the output section where the
4794 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4795 ignored for other object file formats.
4797 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4798 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4799 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4800 the start and end of the global destructors. The
4801 first word in the list is the number of entries, followed by the address
4802 of each constructor or destructor, followed by a zero word. The
4803 compiler must arrange to actually run the code. For these object file
4804 formats @sc{gnu} C++ normally calls constructors from a subroutine
4805 @code{__main}; a call to @code{__main} is automatically inserted into
4806 the startup code for @code{main}. @sc{gnu} C++ normally runs
4807 destructors either by using @code{atexit}, or directly from the function
4810 For object file formats such as @code{COFF} or @code{ELF} which support
4811 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4812 addresses of global constructors and destructors into the @code{.ctors}
4813 and @code{.dtors} sections. Placing the following sequence into your
4814 linker script will build the sort of table which the @sc{gnu} C++
4815 runtime code expects to see.
4819 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4824 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4830 If you are using the @sc{gnu} C++ support for initialization priority,
4831 which provides some control over the order in which global constructors
4832 are run, you must sort the constructors at link time to ensure that they
4833 are executed in the correct order. When using the @code{CONSTRUCTORS}
4834 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4835 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4836 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4839 Normally the compiler and linker will handle these issues automatically,
4840 and you will not need to concern yourself with them. However, you may
4841 need to consider this if you are using C++ and writing your own linker
4846 @node Output Section Discarding
4847 @subsection Output Section Discarding
4848 @cindex discarding sections
4849 @cindex sections, discarding
4850 @cindex removing sections
4851 The linker will not normally create output sections with no contents.
4852 This is for convenience when referring to input sections that may or
4853 may not be present in any of the input files. For example:
4855 .foo : @{ *(.foo) @}
4858 will only create a @samp{.foo} section in the output file if there is a
4859 @samp{.foo} section in at least one input file, and if the input
4860 sections are not all empty. Other link script directives that allocate
4861 space in an output section will also create the output section. So
4862 too will assignments to dot even if the assignment does not create
4863 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
4864 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
4865 @samp{sym} is an absolute symbol of value 0 defined in the script.
4866 This allows you to force output of an empty section with @samp{. = .}.
4868 The linker will ignore address assignments (@pxref{Output Section Address})
4869 on discarded output sections, except when the linker script defines
4870 symbols in the output section. In that case the linker will obey
4871 the address assignments, possibly advancing dot even though the
4872 section is discarded.
4875 The special output section name @samp{/DISCARD/} may be used to discard
4876 input sections. Any input sections which are assigned to an output
4877 section named @samp{/DISCARD/} are not included in the output file.
4879 @node Output Section Attributes
4880 @subsection Output Section Attributes
4881 @cindex output section attributes
4882 We showed above that the full description of an output section looked
4887 @var{section} [@var{address}] [(@var{type})] :
4889 [ALIGN(@var{section_align})]
4890 [SUBALIGN(@var{subsection_align})]
4893 @var{output-section-command}
4894 @var{output-section-command}
4896 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4900 We've already described @var{section}, @var{address}, and
4901 @var{output-section-command}. In this section we will describe the
4902 remaining section attributes.
4905 * Output Section Type:: Output section type
4906 * Output Section LMA:: Output section LMA
4907 * Forced Output Alignment:: Forced Output Alignment
4908 * Forced Input Alignment:: Forced Input Alignment
4909 * Output Section Constraint:: Output section constraint
4910 * Output Section Region:: Output section region
4911 * Output Section Phdr:: Output section phdr
4912 * Output Section Fill:: Output section fill
4915 @node Output Section Type
4916 @subsubsection Output Section Type
4917 Each output section may have a type. The type is a keyword in
4918 parentheses. The following types are defined:
4922 The section should be marked as not loadable, so that it will not be
4923 loaded into memory when the program is run.
4928 These type names are supported for backward compatibility, and are
4929 rarely used. They all have the same effect: the section should be
4930 marked as not allocatable, so that no memory is allocated for the
4931 section when the program is run.
4935 @cindex prevent unnecessary loading
4936 @cindex loading, preventing
4937 The linker normally sets the attributes of an output section based on
4938 the input sections which map into it. You can override this by using
4939 the section type. For example, in the script sample below, the
4940 @samp{ROM} section is addressed at memory location @samp{0} and does not
4941 need to be loaded when the program is run.
4945 ROM 0 (NOLOAD) : @{ @dots{} @}
4951 @node Output Section LMA
4952 @subsubsection Output Section LMA
4953 @kindex AT>@var{lma_region}
4954 @kindex AT(@var{lma})
4955 @cindex load address
4956 @cindex section load address
4957 Every section has a virtual address (VMA) and a load address (LMA); see
4958 @ref{Basic Script Concepts}. The virtual address is specified by the
4959 @pxref{Output Section Address} described earlier. The load address is
4960 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
4961 address is optional.
4963 The @code{AT} keyword takes an expression as an argument. This
4964 specifies the exact load address of the section. The @code{AT>} keyword
4965 takes the name of a memory region as an argument. @xref{MEMORY}. The
4966 load address of the section is set to the next free address in the
4967 region, aligned to the section's alignment requirements.
4969 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4970 section, the linker will use the following heuristic to determine the
4975 If the section has a specific VMA address, then this is used as
4976 the LMA address as well.
4979 If the section is not allocatable then its LMA is set to its VMA.
4982 Otherwise if a memory region can be found that is compatible
4983 with the current section, and this region contains at least one
4984 section, then the LMA is set so the difference between the
4985 VMA and LMA is the same as the difference between the VMA and LMA of
4986 the last section in the located region.
4989 If no memory regions have been declared then a default region
4990 that covers the entire address space is used in the previous step.
4993 If no suitable region could be found, or there was no previous
4994 section then the LMA is set equal to the VMA.
4997 @cindex ROM initialized data
4998 @cindex initialized data in ROM
4999 This feature is designed to make it easy to build a ROM image. For
5000 example, the following linker script creates three output sections: one
5001 called @samp{.text}, which starts at @code{0x1000}, one called
5002 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
5003 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
5004 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
5005 defined with the value @code{0x2000}, which shows that the location
5006 counter holds the VMA value, not the LMA value.
5012 .text 0x1000 : @{ *(.text) _etext = . ; @}
5014 AT ( ADDR (.text) + SIZEOF (.text) )
5015 @{ _data = . ; *(.data); _edata = . ; @}
5017 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
5022 The run-time initialization code for use with a program generated with
5023 this linker script would include something like the following, to copy
5024 the initialized data from the ROM image to its runtime address. Notice
5025 how this code takes advantage of the symbols defined by the linker
5030 extern char _etext, _data, _edata, _bstart, _bend;
5031 char *src = &_etext;
5034 /* ROM has data at end of text; copy it. */
5035 while (dst < &_edata)
5039 for (dst = &_bstart; dst< &_bend; dst++)
5044 @node Forced Output Alignment
5045 @subsubsection Forced Output Alignment
5046 @kindex ALIGN(@var{section_align})
5047 @cindex forcing output section alignment
5048 @cindex output section alignment
5049 You can increase an output section's alignment by using ALIGN. As an
5050 alternative you can enforce that the difference between the VMA and LMA remains
5051 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
5053 @node Forced Input Alignment
5054 @subsubsection Forced Input Alignment
5055 @kindex SUBALIGN(@var{subsection_align})
5056 @cindex forcing input section alignment
5057 @cindex input section alignment
5058 You can force input section alignment within an output section by using
5059 SUBALIGN. The value specified overrides any alignment given by input
5060 sections, whether larger or smaller.
5062 @node Output Section Constraint
5063 @subsubsection Output Section Constraint
5066 @cindex constraints on output sections
5067 You can specify that an output section should only be created if all
5068 of its input sections are read-only or all of its input sections are
5069 read-write by using the keyword @code{ONLY_IF_RO} and
5070 @code{ONLY_IF_RW} respectively.
5072 @node Output Section Region
5073 @subsubsection Output Section Region
5074 @kindex >@var{region}
5075 @cindex section, assigning to memory region
5076 @cindex memory regions and sections
5077 You can assign a section to a previously defined region of memory by
5078 using @samp{>@var{region}}. @xref{MEMORY}.
5080 Here is a simple example:
5083 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
5084 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
5088 @node Output Section Phdr
5089 @subsubsection Output Section Phdr
5091 @cindex section, assigning to program header
5092 @cindex program headers and sections
5093 You can assign a section to a previously defined program segment by
5094 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
5095 one or more segments, then all subsequent allocated sections will be
5096 assigned to those segments as well, unless they use an explicitly
5097 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
5098 linker to not put the section in any segment at all.
5100 Here is a simple example:
5103 PHDRS @{ text PT_LOAD ; @}
5104 SECTIONS @{ .text : @{ *(.text) @} :text @}
5108 @node Output Section Fill
5109 @subsubsection Output Section Fill
5110 @kindex =@var{fillexp}
5111 @cindex section fill pattern
5112 @cindex fill pattern, entire section
5113 You can set the fill pattern for an entire section by using
5114 @samp{=@var{fillexp}}. @var{fillexp} is an expression
5115 (@pxref{Expressions}). Any otherwise unspecified regions of memory
5116 within the output section (for example, gaps left due to the required
5117 alignment of input sections) will be filled with the value, repeated as
5118 necessary. If the fill expression is a simple hex number, ie. a string
5119 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
5120 an arbitrarily long sequence of hex digits can be used to specify the
5121 fill pattern; Leading zeros become part of the pattern too. For all
5122 other cases, including extra parentheses or a unary @code{+}, the fill
5123 pattern is the four least significant bytes of the value of the
5124 expression. In all cases, the number is big-endian.
5126 You can also change the fill value with a @code{FILL} command in the
5127 output section commands; (@pxref{Output Section Data}).
5129 Here is a simple example:
5132 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
5136 @node Overlay Description
5137 @subsection Overlay Description
5140 An overlay description provides an easy way to describe sections which
5141 are to be loaded as part of a single memory image but are to be run at
5142 the same memory address. At run time, some sort of overlay manager will
5143 copy the overlaid sections in and out of the runtime memory address as
5144 required, perhaps by simply manipulating addressing bits. This approach
5145 can be useful, for example, when a certain region of memory is faster
5148 Overlays are described using the @code{OVERLAY} command. The
5149 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
5150 output section description. The full syntax of the @code{OVERLAY}
5151 command is as follows:
5154 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
5158 @var{output-section-command}
5159 @var{output-section-command}
5161 @} [:@var{phdr}@dots{}] [=@var{fill}]
5164 @var{output-section-command}
5165 @var{output-section-command}
5167 @} [:@var{phdr}@dots{}] [=@var{fill}]
5169 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5173 Everything is optional except @code{OVERLAY} (a keyword), and each
5174 section must have a name (@var{secname1} and @var{secname2} above). The
5175 section definitions within the @code{OVERLAY} construct are identical to
5176 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5177 except that no addresses and no memory regions may be defined for
5178 sections within an @code{OVERLAY}.
5180 The comma at the end may be required if a @var{fill} is used and
5181 the next @var{sections-command} looks like a continuation of the expression.
5183 The sections are all defined with the same starting address. The load
5184 addresses of the sections are arranged such that they are consecutive in
5185 memory starting at the load address used for the @code{OVERLAY} as a
5186 whole (as with normal section definitions, the load address is optional,
5187 and defaults to the start address; the start address is also optional,
5188 and defaults to the current value of the location counter).
5190 If the @code{NOCROSSREFS} keyword is used, and there are any
5191 references among the sections, the linker will report an error. Since
5192 the sections all run at the same address, it normally does not make
5193 sense for one section to refer directly to another.
5194 @xref{Miscellaneous Commands, NOCROSSREFS}.
5196 For each section within the @code{OVERLAY}, the linker automatically
5197 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5198 defined as the starting load address of the section. The symbol
5199 @code{__load_stop_@var{secname}} is defined as the final load address of
5200 the section. Any characters within @var{secname} which are not legal
5201 within C identifiers are removed. C (or assembler) code may use these
5202 symbols to move the overlaid sections around as necessary.
5204 At the end of the overlay, the value of the location counter is set to
5205 the start address of the overlay plus the size of the largest section.
5207 Here is an example. Remember that this would appear inside a
5208 @code{SECTIONS} construct.
5211 OVERLAY 0x1000 : AT (0x4000)
5213 .text0 @{ o1/*.o(.text) @}
5214 .text1 @{ o2/*.o(.text) @}
5219 This will define both @samp{.text0} and @samp{.text1} to start at
5220 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5221 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5222 following symbols will be defined if referenced: @code{__load_start_text0},
5223 @code{__load_stop_text0}, @code{__load_start_text1},
5224 @code{__load_stop_text1}.
5226 C code to copy overlay @code{.text1} into the overlay area might look
5231 extern char __load_start_text1, __load_stop_text1;
5232 memcpy ((char *) 0x1000, &__load_start_text1,
5233 &__load_stop_text1 - &__load_start_text1);
5237 Note that the @code{OVERLAY} command is just syntactic sugar, since
5238 everything it does can be done using the more basic commands. The above
5239 example could have been written identically as follows.
5243 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5244 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5245 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5246 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5247 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5248 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5249 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5254 @section MEMORY Command
5256 @cindex memory regions
5257 @cindex regions of memory
5258 @cindex allocating memory
5259 @cindex discontinuous memory
5260 The linker's default configuration permits allocation of all available
5261 memory. You can override this by using the @code{MEMORY} command.
5263 The @code{MEMORY} command describes the location and size of blocks of
5264 memory in the target. You can use it to describe which memory regions
5265 may be used by the linker, and which memory regions it must avoid. You
5266 can then assign sections to particular memory regions. The linker will
5267 set section addresses based on the memory regions, and will warn about
5268 regions that become too full. The linker will not shuffle sections
5269 around to fit into the available regions.
5271 A linker script may contain many uses of the @code{MEMORY} command,
5272 however, all memory blocks defined are treated as if they were
5273 specified inside a single @code{MEMORY} command. The syntax for
5279 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5285 The @var{name} is a name used in the linker script to refer to the
5286 region. The region name has no meaning outside of the linker script.
5287 Region names are stored in a separate name space, and will not conflict
5288 with symbol names, file names, or section names. Each memory region
5289 must have a distinct name within the @code{MEMORY} command. However you can
5290 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5293 @cindex memory region attributes
5294 The @var{attr} string is an optional list of attributes that specify
5295 whether to use a particular memory region for an input section which is
5296 not explicitly mapped in the linker script. As described in
5297 @ref{SECTIONS}, if you do not specify an output section for some input
5298 section, the linker will create an output section with the same name as
5299 the input section. If you define region attributes, the linker will use
5300 them to select the memory region for the output section that it creates.
5302 The @var{attr} string must consist only of the following characters:
5317 Invert the sense of any of the attributes that follow
5320 If a unmapped section matches any of the listed attributes other than
5321 @samp{!}, it will be placed in the memory region. The @samp{!}
5322 attribute reverses this test, so that an unmapped section will be placed
5323 in the memory region only if it does not match any of the listed
5329 The @var{origin} is an numerical expression for the start address of
5330 the memory region. The expression must evaluate to a constant and it
5331 cannot involve any symbols. The keyword @code{ORIGIN} may be
5332 abbreviated to @code{org} or @code{o} (but not, for example,
5338 The @var{len} is an expression for the size in bytes of the memory
5339 region. As with the @var{origin} expression, the expression must
5340 be numerical only and must evaluate to a constant. The keyword
5341 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5343 In the following example, we specify that there are two memory regions
5344 available for allocation: one starting at @samp{0} for 256 kilobytes,
5345 and the other starting at @samp{0x40000000} for four megabytes. The
5346 linker will place into the @samp{rom} memory region every section which
5347 is not explicitly mapped into a memory region, and is either read-only
5348 or executable. The linker will place other sections which are not
5349 explicitly mapped into a memory region into the @samp{ram} memory
5356 rom (rx) : ORIGIN = 0, LENGTH = 256K
5357 ram (!rx) : org = 0x40000000, l = 4M
5362 Once you define a memory region, you can direct the linker to place
5363 specific output sections into that memory region by using the
5364 @samp{>@var{region}} output section attribute. For example, if you have
5365 a memory region named @samp{mem}, you would use @samp{>mem} in the
5366 output section definition. @xref{Output Section Region}. If no address
5367 was specified for the output section, the linker will set the address to
5368 the next available address within the memory region. If the combined
5369 output sections directed to a memory region are too large for the
5370 region, the linker will issue an error message.
5372 It is possible to access the origin and length of a memory in an
5373 expression via the @code{ORIGIN(@var{memory})} and
5374 @code{LENGTH(@var{memory})} functions:
5378 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5383 @section PHDRS Command
5385 @cindex program headers
5386 @cindex ELF program headers
5387 @cindex program segments
5388 @cindex segments, ELF
5389 The ELF object file format uses @dfn{program headers}, also knows as
5390 @dfn{segments}. The program headers describe how the program should be
5391 loaded into memory. You can print them out by using the @code{objdump}
5392 program with the @samp{-p} option.
5394 When you run an ELF program on a native ELF system, the system loader
5395 reads the program headers in order to figure out how to load the
5396 program. This will only work if the program headers are set correctly.
5397 This manual does not describe the details of how the system loader
5398 interprets program headers; for more information, see the ELF ABI.
5400 The linker will create reasonable program headers by default. However,
5401 in some cases, you may need to specify the program headers more
5402 precisely. You may use the @code{PHDRS} command for this purpose. When
5403 the linker sees the @code{PHDRS} command in the linker script, it will
5404 not create any program headers other than the ones specified.
5406 The linker only pays attention to the @code{PHDRS} command when
5407 generating an ELF output file. In other cases, the linker will simply
5408 ignore @code{PHDRS}.
5410 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5411 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5417 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5418 [ FLAGS ( @var{flags} ) ] ;
5423 The @var{name} is used only for reference in the @code{SECTIONS} command
5424 of the linker script. It is not put into the output file. Program
5425 header names are stored in a separate name space, and will not conflict
5426 with symbol names, file names, or section names. Each program header
5427 must have a distinct name. The headers are processed in order and it
5428 is usual for them to map to sections in ascending load address order.
5430 Certain program header types describe segments of memory which the
5431 system loader will load from the file. In the linker script, you
5432 specify the contents of these segments by placing allocatable output
5433 sections in the segments. You use the @samp{:@var{phdr}} output section
5434 attribute to place a section in a particular segment. @xref{Output
5437 It is normal to put certain sections in more than one segment. This
5438 merely implies that one segment of memory contains another. You may
5439 repeat @samp{:@var{phdr}}, using it once for each segment which should
5440 contain the section.
5442 If you place a section in one or more segments using @samp{:@var{phdr}},
5443 then the linker will place all subsequent allocatable sections which do
5444 not specify @samp{:@var{phdr}} in the same segments. This is for
5445 convenience, since generally a whole set of contiguous sections will be
5446 placed in a single segment. You can use @code{:NONE} to override the
5447 default segment and tell the linker to not put the section in any
5452 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
5453 the program header type to further describe the contents of the segment.
5454 The @code{FILEHDR} keyword means that the segment should include the ELF
5455 file header. The @code{PHDRS} keyword means that the segment should
5456 include the ELF program headers themselves. If applied to a loadable
5457 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5460 The @var{type} may be one of the following. The numbers indicate the
5461 value of the keyword.
5464 @item @code{PT_NULL} (0)
5465 Indicates an unused program header.
5467 @item @code{PT_LOAD} (1)
5468 Indicates that this program header describes a segment to be loaded from
5471 @item @code{PT_DYNAMIC} (2)
5472 Indicates a segment where dynamic linking information can be found.
5474 @item @code{PT_INTERP} (3)
5475 Indicates a segment where the name of the program interpreter may be
5478 @item @code{PT_NOTE} (4)
5479 Indicates a segment holding note information.
5481 @item @code{PT_SHLIB} (5)
5482 A reserved program header type, defined but not specified by the ELF
5485 @item @code{PT_PHDR} (6)
5486 Indicates a segment where the program headers may be found.
5488 @item @code{PT_TLS} (7)
5489 Indicates a segment containing thread local storage.
5491 @item @var{expression}
5492 An expression giving the numeric type of the program header. This may
5493 be used for types not defined above.
5496 You can specify that a segment should be loaded at a particular address
5497 in memory by using an @code{AT} expression. This is identical to the
5498 @code{AT} command used as an output section attribute (@pxref{Output
5499 Section LMA}). The @code{AT} command for a program header overrides the
5500 output section attribute.
5502 The linker will normally set the segment flags based on the sections
5503 which comprise the segment. You may use the @code{FLAGS} keyword to
5504 explicitly specify the segment flags. The value of @var{flags} must be
5505 an integer. It is used to set the @code{p_flags} field of the program
5508 Here is an example of @code{PHDRS}. This shows a typical set of program
5509 headers used on a native ELF system.
5515 headers PT_PHDR PHDRS ;
5517 text PT_LOAD FILEHDR PHDRS ;
5519 dynamic PT_DYNAMIC ;
5525 .interp : @{ *(.interp) @} :text :interp
5526 .text : @{ *(.text) @} :text
5527 .rodata : @{ *(.rodata) @} /* defaults to :text */
5529 . = . + 0x1000; /* move to a new page in memory */
5530 .data : @{ *(.data) @} :data
5531 .dynamic : @{ *(.dynamic) @} :data :dynamic
5538 @section VERSION Command
5539 @kindex VERSION @{script text@}
5540 @cindex symbol versions
5541 @cindex version script
5542 @cindex versions of symbols
5543 The linker supports symbol versions when using ELF. Symbol versions are
5544 only useful when using shared libraries. The dynamic linker can use
5545 symbol versions to select a specific version of a function when it runs
5546 a program that may have been linked against an earlier version of the
5549 You can include a version script directly in the main linker script, or
5550 you can supply the version script as an implicit linker script. You can
5551 also use the @samp{--version-script} linker option.
5553 The syntax of the @code{VERSION} command is simply
5555 VERSION @{ version-script-commands @}
5558 The format of the version script commands is identical to that used by
5559 Sun's linker in Solaris 2.5. The version script defines a tree of
5560 version nodes. You specify the node names and interdependencies in the
5561 version script. You can specify which symbols are bound to which
5562 version nodes, and you can reduce a specified set of symbols to local
5563 scope so that they are not globally visible outside of the shared
5566 The easiest way to demonstrate the version script language is with a few
5592 This example version script defines three version nodes. The first
5593 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5594 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5595 a number of symbols to local scope so that they are not visible outside
5596 of the shared library; this is done using wildcard patterns, so that any
5597 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5598 is matched. The wildcard patterns available are the same as those used
5599 in the shell when matching filenames (also known as ``globbing'').
5600 However, if you specify the symbol name inside double quotes, then the
5601 name is treated as literal, rather than as a glob pattern.
5603 Next, the version script defines node @samp{VERS_1.2}. This node
5604 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5605 to the version node @samp{VERS_1.2}.
5607 Finally, the version script defines node @samp{VERS_2.0}. This node
5608 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5609 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5611 When the linker finds a symbol defined in a library which is not
5612 specifically bound to a version node, it will effectively bind it to an
5613 unspecified base version of the library. You can bind all otherwise
5614 unspecified symbols to a given version node by using @samp{global: *;}
5615 somewhere in the version script. Note that it's slightly crazy to use
5616 wildcards in a global spec except on the last version node. Global
5617 wildcards elsewhere run the risk of accidentally adding symbols to the
5618 set exported for an old version. That's wrong since older versions
5619 ought to have a fixed set of symbols.
5621 The names of the version nodes have no specific meaning other than what
5622 they might suggest to the person reading them. The @samp{2.0} version
5623 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5624 However, this would be a confusing way to write a version script.
5626 Node name can be omitted, provided it is the only version node
5627 in the version script. Such version script doesn't assign any versions to
5628 symbols, only selects which symbols will be globally visible out and which
5632 @{ global: foo; bar; local: *; @};
5635 When you link an application against a shared library that has versioned
5636 symbols, the application itself knows which version of each symbol it
5637 requires, and it also knows which version nodes it needs from each
5638 shared library it is linked against. Thus at runtime, the dynamic
5639 loader can make a quick check to make sure that the libraries you have
5640 linked against do in fact supply all of the version nodes that the
5641 application will need to resolve all of the dynamic symbols. In this
5642 way it is possible for the dynamic linker to know with certainty that
5643 all external symbols that it needs will be resolvable without having to
5644 search for each symbol reference.
5646 The symbol versioning is in effect a much more sophisticated way of
5647 doing minor version checking that SunOS does. The fundamental problem
5648 that is being addressed here is that typically references to external
5649 functions are bound on an as-needed basis, and are not all bound when
5650 the application starts up. If a shared library is out of date, a
5651 required interface may be missing; when the application tries to use
5652 that interface, it may suddenly and unexpectedly fail. With symbol
5653 versioning, the user will get a warning when they start their program if
5654 the libraries being used with the application are too old.
5656 There are several GNU extensions to Sun's versioning approach. The
5657 first of these is the ability to bind a symbol to a version node in the
5658 source file where the symbol is defined instead of in the versioning
5659 script. This was done mainly to reduce the burden on the library
5660 maintainer. You can do this by putting something like:
5662 __asm__(".symver original_foo,foo@@VERS_1.1");
5665 in the C source file. This renames the function @samp{original_foo} to
5666 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5667 The @samp{local:} directive can be used to prevent the symbol
5668 @samp{original_foo} from being exported. A @samp{.symver} directive
5669 takes precedence over a version script.
5671 The second GNU extension is to allow multiple versions of the same
5672 function to appear in a given shared library. In this way you can make
5673 an incompatible change to an interface without increasing the major
5674 version number of the shared library, while still allowing applications
5675 linked against the old interface to continue to function.
5677 To do this, you must use multiple @samp{.symver} directives in the
5678 source file. Here is an example:
5681 __asm__(".symver original_foo,foo@@");
5682 __asm__(".symver old_foo,foo@@VERS_1.1");
5683 __asm__(".symver old_foo1,foo@@VERS_1.2");
5684 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5687 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5688 unspecified base version of the symbol. The source file that contains this
5689 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5690 @samp{old_foo1}, and @samp{new_foo}.
5692 When you have multiple definitions of a given symbol, there needs to be
5693 some way to specify a default version to which external references to
5694 this symbol will be bound. You can do this with the
5695 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5696 declare one version of a symbol as the default in this manner; otherwise
5697 you would effectively have multiple definitions of the same symbol.
5699 If you wish to bind a reference to a specific version of the symbol
5700 within the shared library, you can use the aliases of convenience
5701 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5702 specifically bind to an external version of the function in question.
5704 You can also specify the language in the version script:
5707 VERSION extern "lang" @{ version-script-commands @}
5710 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5711 The linker will iterate over the list of symbols at the link time and
5712 demangle them according to @samp{lang} before matching them to the
5713 patterns specified in @samp{version-script-commands}. The default
5714 @samp{lang} is @samp{C}.
5716 Demangled names may contains spaces and other special characters. As
5717 described above, you can use a glob pattern to match demangled names,
5718 or you can use a double-quoted string to match the string exactly. In
5719 the latter case, be aware that minor differences (such as differing
5720 whitespace) between the version script and the demangler output will
5721 cause a mismatch. As the exact string generated by the demangler
5722 might change in the future, even if the mangled name does not, you
5723 should check that all of your version directives are behaving as you
5724 expect when you upgrade.
5727 @section Expressions in Linker Scripts
5730 The syntax for expressions in the linker script language is identical to
5731 that of C expressions. All expressions are evaluated as integers. All
5732 expressions are evaluated in the same size, which is 32 bits if both the
5733 host and target are 32 bits, and is otherwise 64 bits.
5735 You can use and set symbol values in expressions.
5737 The linker defines several special purpose builtin functions for use in
5741 * Constants:: Constants
5742 * Symbolic Constants:: Symbolic constants
5743 * Symbols:: Symbol Names
5744 * Orphan Sections:: Orphan Sections
5745 * Location Counter:: The Location Counter
5746 * Operators:: Operators
5747 * Evaluation:: Evaluation
5748 * Expression Section:: The Section of an Expression
5749 * Builtin Functions:: Builtin Functions
5753 @subsection Constants
5754 @cindex integer notation
5755 @cindex constants in linker scripts
5756 All constants are integers.
5758 As in C, the linker considers an integer beginning with @samp{0} to be
5759 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5760 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5761 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5762 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5763 value without a prefix or a suffix is considered to be decimal.
5765 @cindex scaled integers
5766 @cindex K and M integer suffixes
5767 @cindex M and K integer suffixes
5768 @cindex suffixes for integers
5769 @cindex integer suffixes
5770 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5774 @c END TEXI2ROFF-KILL
5775 @code{1024} or @code{1024*1024}
5779 ${\rm 1024}$ or ${\rm 1024}^2$
5781 @c END TEXI2ROFF-KILL
5782 respectively. For example, the following
5783 all refer to the same quantity:
5792 Note - the @code{K} and @code{M} suffixes cannot be used in
5793 conjunction with the base suffixes mentioned above.
5795 @node Symbolic Constants
5796 @subsection Symbolic Constants
5797 @cindex symbolic constants
5799 It is possible to refer to target specific constants via the use of
5800 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5805 The target's maximum page size.
5807 @item COMMONPAGESIZE
5808 @kindex COMMONPAGESIZE
5809 The target's default page size.
5815 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5818 will create a text section aligned to the largest page boundary
5819 supported by the target.
5822 @subsection Symbol Names
5823 @cindex symbol names
5825 @cindex quoted symbol names
5827 Unless quoted, symbol names start with a letter, underscore, or period
5828 and may include letters, digits, underscores, periods, and hyphens.
5829 Unquoted symbol names must not conflict with any keywords. You can
5830 specify a symbol which contains odd characters or has the same name as a
5831 keyword by surrounding the symbol name in double quotes:
5834 "with a space" = "also with a space" + 10;
5837 Since symbols can contain many non-alphabetic characters, it is safest
5838 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5839 whereas @samp{A - B} is an expression involving subtraction.
5841 @node Orphan Sections
5842 @subsection Orphan Sections
5844 Orphan sections are sections present in the input files which
5845 are not explicitly placed into the output file by the linker
5846 script. The linker will still copy these sections into the
5847 output file by either finding, or creating a suitable output section
5848 in which to place the orphaned input section.
5850 If the name of an orphaned input section exactly matches the name of
5851 an existing output section, then the orphaned input section will be
5852 placed at the end of that output section.
5854 If there is no output section with a matching name then new output
5855 sections will be created. Each new output section will have the same
5856 name as the orphan section placed within it. If there are multiple
5857 orphan sections with the same name, these will all be combined into
5858 one new output section.
5860 If new output sections are created to hold orphaned input sections,
5861 then the linker must decide where to place these new output sections
5862 in relation to existing output sections. On most modern targets, the
5863 linker attempts to place orphan sections after sections of the same
5864 attribute, such as code vs data, loadable vs non-loadable, etc. If no
5865 sections with matching attributes are found, or your target lacks this
5866 support, the orphan section is placed at the end of the file.
5868 The command line options @samp{--orphan-handling} and @samp{--unique}
5869 (@pxref{Options,,Command Line Options}) can be used to control which
5870 output sections an orphan is placed in.
5872 @node Location Counter
5873 @subsection The Location Counter
5876 @cindex location counter
5877 @cindex current output location
5878 The special linker variable @dfn{dot} @samp{.} always contains the
5879 current output location counter. Since the @code{.} always refers to a
5880 location in an output section, it may only appear in an expression
5881 within a @code{SECTIONS} command. The @code{.} symbol may appear
5882 anywhere that an ordinary symbol is allowed in an expression.
5885 Assigning a value to @code{.} will cause the location counter to be
5886 moved. This may be used to create holes in the output section. The
5887 location counter may not be moved backwards inside an output section,
5888 and may not be moved backwards outside of an output section if so
5889 doing creates areas with overlapping LMAs.
5905 In the previous example, the @samp{.text} section from @file{file1} is
5906 located at the beginning of the output section @samp{output}. It is
5907 followed by a 1000 byte gap. Then the @samp{.text} section from
5908 @file{file2} appears, also with a 1000 byte gap following before the
5909 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5910 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5912 @cindex dot inside sections
5913 Note: @code{.} actually refers to the byte offset from the start of the
5914 current containing object. Normally this is the @code{SECTIONS}
5915 statement, whose start address is 0, hence @code{.} can be used as an
5916 absolute address. If @code{.} is used inside a section description
5917 however, it refers to the byte offset from the start of that section,
5918 not an absolute address. Thus in a script like this:
5936 The @samp{.text} section will be assigned a starting address of 0x100
5937 and a size of exactly 0x200 bytes, even if there is not enough data in
5938 the @samp{.text} input sections to fill this area. (If there is too
5939 much data, an error will be produced because this would be an attempt to
5940 move @code{.} backwards). The @samp{.data} section will start at 0x500
5941 and it will have an extra 0x600 bytes worth of space after the end of
5942 the values from the @samp{.data} input sections and before the end of
5943 the @samp{.data} output section itself.
5945 @cindex dot outside sections
5946 Setting symbols to the value of the location counter outside of an
5947 output section statement can result in unexpected values if the linker
5948 needs to place orphan sections. For example, given the following:
5954 .text: @{ *(.text) @}
5958 .data: @{ *(.data) @}
5963 If the linker needs to place some input section, e.g. @code{.rodata},
5964 not mentioned in the script, it might choose to place that section
5965 between @code{.text} and @code{.data}. You might think the linker
5966 should place @code{.rodata} on the blank line in the above script, but
5967 blank lines are of no particular significance to the linker. As well,
5968 the linker doesn't associate the above symbol names with their
5969 sections. Instead, it assumes that all assignments or other
5970 statements belong to the previous output section, except for the
5971 special case of an assignment to @code{.}. I.e., the linker will
5972 place the orphan @code{.rodata} section as if the script was written
5979 .text: @{ *(.text) @}
5983 .rodata: @{ *(.rodata) @}
5984 .data: @{ *(.data) @}
5989 This may or may not be the script author's intention for the value of
5990 @code{start_of_data}. One way to influence the orphan section
5991 placement is to assign the location counter to itself, as the linker
5992 assumes that an assignment to @code{.} is setting the start address of
5993 a following output section and thus should be grouped with that
5994 section. So you could write:
6000 .text: @{ *(.text) @}
6005 .data: @{ *(.data) @}
6010 Now, the orphan @code{.rodata} section will be placed between
6011 @code{end_of_text} and @code{start_of_data}.
6015 @subsection Operators
6016 @cindex operators for arithmetic
6017 @cindex arithmetic operators
6018 @cindex precedence in expressions
6019 The linker recognizes the standard C set of arithmetic operators, with
6020 the standard bindings and precedence levels:
6023 @c END TEXI2ROFF-KILL
6025 precedence associativity Operators Notes
6031 5 left == != > < <= >=
6037 11 right &= += -= *= /= (2)
6041 (1) Prefix operators
6042 (2) @xref{Assignments}.
6046 \vskip \baselineskip
6047 %"lispnarrowing" is the extra indent used generally for smallexample
6048 \hskip\lispnarrowing\vbox{\offinterlineskip
6051 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
6052 height2pt&\omit&&\omit&&\omit&\cr
6053 &Precedence&& Associativity &&{\rm Operators}&\cr
6054 height2pt&\omit&&\omit&&\omit&\cr
6056 height2pt&\omit&&\omit&&\omit&\cr
6058 % '176 is tilde, '~' in tt font
6059 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
6060 &2&&left&&* / \%&\cr
6063 &5&&left&&== != > < <= >=&\cr
6066 &8&&left&&{\&\&}&\cr
6069 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
6071 height2pt&\omit&&\omit&&\omit&\cr}
6076 @obeylines@parskip=0pt@parindent=0pt
6077 @dag@quad Prefix operators.
6078 @ddag@quad @xref{Assignments}.
6081 @c END TEXI2ROFF-KILL
6084 @subsection Evaluation
6085 @cindex lazy evaluation
6086 @cindex expression evaluation order
6087 The linker evaluates expressions lazily. It only computes the value of
6088 an expression when absolutely necessary.
6090 The linker needs some information, such as the value of the start
6091 address of the first section, and the origins and lengths of memory
6092 regions, in order to do any linking at all. These values are computed
6093 as soon as possible when the linker reads in the linker script.
6095 However, other values (such as symbol values) are not known or needed
6096 until after storage allocation. Such values are evaluated later, when
6097 other information (such as the sizes of output sections) is available
6098 for use in the symbol assignment expression.
6100 The sizes of sections cannot be known until after allocation, so
6101 assignments dependent upon these are not performed until after
6104 Some expressions, such as those depending upon the location counter
6105 @samp{.}, must be evaluated during section allocation.
6107 If the result of an expression is required, but the value is not
6108 available, then an error results. For example, a script like the
6114 .text 9+this_isnt_constant :
6120 will cause the error message @samp{non constant expression for initial
6123 @node Expression Section
6124 @subsection The Section of an Expression
6125 @cindex expression sections
6126 @cindex absolute expressions
6127 @cindex relative expressions
6128 @cindex absolute and relocatable symbols
6129 @cindex relocatable and absolute symbols
6130 @cindex symbols, relocatable and absolute
6131 Addresses and symbols may be section relative, or absolute. A section
6132 relative symbol is relocatable. If you request relocatable output
6133 using the @samp{-r} option, a further link operation may change the
6134 value of a section relative symbol. On the other hand, an absolute
6135 symbol will retain the same value throughout any further link
6138 Some terms in linker expressions are addresses. This is true of
6139 section relative symbols and for builtin functions that return an
6140 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
6141 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
6142 functions that return a non-address value, such as @code{LENGTH}.
6143 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
6144 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
6145 differently depending on their location, for compatibility with older
6146 versions of @code{ld}. Expressions appearing outside an output
6147 section definition treat all numbers as absolute addresses.
6148 Expressions appearing inside an output section definition treat
6149 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
6150 given, then absolute symbols and numbers are simply treated as numbers
6153 In the following simple example,
6160 __executable_start = 0x100;
6164 __data_start = 0x10;
6172 both @code{.} and @code{__executable_start} are set to the absolute
6173 address 0x100 in the first two assignments, then both @code{.} and
6174 @code{__data_start} are set to 0x10 relative to the @code{.data}
6175 section in the second two assignments.
6177 For expressions involving numbers, relative addresses and absolute
6178 addresses, ld follows these rules to evaluate terms:
6182 Unary operations on an absolute address or number, and binary
6183 operations on two absolute addresses or two numbers, or between one
6184 absolute address and a number, apply the operator to the value(s).
6186 Unary operations on a relative address, and binary operations on two
6187 relative addresses in the same section or between one relative address
6188 and a number, apply the operator to the offset part of the address(es).
6190 Other binary operations, that is, between two relative addresses not
6191 in the same section, or between a relative address and an absolute
6192 address, first convert any non-absolute term to an absolute address
6193 before applying the operator.
6196 The result section of each sub-expression is as follows:
6200 An operation involving only numbers results in a number.
6202 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6204 The result of other binary arithmetic and logical operations on two
6205 relative addresses in the same section or two absolute addresses
6206 (after above conversions) is also a number when
6207 @code{LD_FEATURE ("SANE_EXPR")} or inside an output section definition
6208 but an absolute address otherwise.
6210 The result of other operations on relative addresses or one
6211 relative address and a number, is a relative address in the same
6212 section as the relative operand(s).
6214 The result of other operations on absolute addresses (after above
6215 conversions) is an absolute address.
6218 You can use the builtin function @code{ABSOLUTE} to force an expression
6219 to be absolute when it would otherwise be relative. For example, to
6220 create an absolute symbol set to the address of the end of the output
6221 section @samp{.data}:
6225 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6229 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6230 @samp{.data} section.
6232 Using @code{LOADADDR} also forces an expression absolute, since this
6233 particular builtin function returns an absolute address.
6235 @node Builtin Functions
6236 @subsection Builtin Functions
6237 @cindex functions in expressions
6238 The linker script language includes a number of builtin functions for
6239 use in linker script expressions.
6242 @item ABSOLUTE(@var{exp})
6243 @kindex ABSOLUTE(@var{exp})
6244 @cindex expression, absolute
6245 Return the absolute (non-relocatable, as opposed to non-negative) value
6246 of the expression @var{exp}. Primarily useful to assign an absolute
6247 value to a symbol within a section definition, where symbol values are
6248 normally section relative. @xref{Expression Section}.
6250 @item ADDR(@var{section})
6251 @kindex ADDR(@var{section})
6252 @cindex section address in expression
6253 Return the address (VMA) of the named @var{section}. Your
6254 script must previously have defined the location of that section. In
6255 the following example, @code{start_of_output_1}, @code{symbol_1} and
6256 @code{symbol_2} are assigned equivalent values, except that
6257 @code{symbol_1} will be relative to the @code{.output1} section while
6258 the other two will be absolute:
6264 start_of_output_1 = ABSOLUTE(.);
6269 symbol_1 = ADDR(.output1);
6270 symbol_2 = start_of_output_1;
6276 @item ALIGN(@var{align})
6277 @itemx ALIGN(@var{exp},@var{align})
6278 @kindex ALIGN(@var{align})
6279 @kindex ALIGN(@var{exp},@var{align})
6280 @cindex round up location counter
6281 @cindex align location counter
6282 @cindex round up expression
6283 @cindex align expression
6284 Return the location counter (@code{.}) or arbitrary expression aligned
6285 to the next @var{align} boundary. The single operand @code{ALIGN}
6286 doesn't change the value of the location counter---it just does
6287 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6288 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6289 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6291 Here is an example which aligns the output @code{.data} section to the
6292 next @code{0x2000} byte boundary after the preceding section and sets a
6293 variable within the section to the next @code{0x8000} boundary after the
6298 .data ALIGN(0x2000): @{
6300 variable = ALIGN(0x8000);
6306 The first use of @code{ALIGN} in this example specifies the location of
6307 a section because it is used as the optional @var{address} attribute of
6308 a section definition (@pxref{Output Section Address}). The second use
6309 of @code{ALIGN} is used to defines the value of a symbol.
6311 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6313 @item ALIGNOF(@var{section})
6314 @kindex ALIGNOF(@var{section})
6315 @cindex section alignment
6316 Return the alignment in bytes of the named @var{section}, if that section has
6317 been allocated. If the section has not been allocated when this is
6318 evaluated, the linker will report an error. In the following example,
6319 the alignment of the @code{.output} section is stored as the first
6320 value in that section.
6325 LONG (ALIGNOF (.output))
6332 @item BLOCK(@var{exp})
6333 @kindex BLOCK(@var{exp})
6334 This is a synonym for @code{ALIGN}, for compatibility with older linker
6335 scripts. It is most often seen when setting the address of an output
6338 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6339 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6340 This is equivalent to either
6342 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6346 (ALIGN(@var{maxpagesize})
6347 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6350 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6351 for the data segment (area between the result of this expression and
6352 @code{DATA_SEGMENT_END}) than the former or not.
6353 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6354 memory will be saved at the expense of up to @var{commonpagesize} wasted
6355 bytes in the on-disk file.
6357 This expression can only be used directly in @code{SECTIONS} commands, not in
6358 any output section descriptions and only once in the linker script.
6359 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6360 be the system page size the object wants to be optimized for while still
6361 running on system page sizes up to @var{maxpagesize}. Note however
6362 that @samp{-z relro} protection will not be effective if the system
6363 page size is larger than @var{commonpagesize}.
6368 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6371 @item DATA_SEGMENT_END(@var{exp})
6372 @kindex DATA_SEGMENT_END(@var{exp})
6373 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6374 evaluation purposes.
6377 . = DATA_SEGMENT_END(.);
6380 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6381 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6382 This defines the end of the @code{PT_GNU_RELRO} segment when
6383 @samp{-z relro} option is used.
6384 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6385 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6386 @var{exp} + @var{offset} is aligned to the @var{commonpagesize}
6387 argument given to @code{DATA_SEGMENT_ALIGN}. If present in the linker
6388 script, it must be placed between @code{DATA_SEGMENT_ALIGN} and
6389 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6390 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6394 . = DATA_SEGMENT_RELRO_END(24, .);
6397 @item DEFINED(@var{symbol})
6398 @kindex DEFINED(@var{symbol})
6399 @cindex symbol defaults
6400 Return 1 if @var{symbol} is in the linker global symbol table and is
6401 defined before the statement using DEFINED in the script, otherwise
6402 return 0. You can use this function to provide
6403 default values for symbols. For example, the following script fragment
6404 shows how to set a global symbol @samp{begin} to the first location in
6405 the @samp{.text} section---but if a symbol called @samp{begin} already
6406 existed, its value is preserved:
6412 begin = DEFINED(begin) ? begin : . ;
6420 @item LENGTH(@var{memory})
6421 @kindex LENGTH(@var{memory})
6422 Return the length of the memory region named @var{memory}.
6424 @item LOADADDR(@var{section})
6425 @kindex LOADADDR(@var{section})
6426 @cindex section load address in expression
6427 Return the absolute LMA of the named @var{section}. (@pxref{Output
6430 @item LOG2CEIL(@var{exp})
6431 @kindex LOG2CEIL(@var{exp})
6432 Return the binary logarithm of @var{exp} rounded towards infinity.
6433 @code{LOG2CEIL(0)} returns 0.
6436 @item MAX(@var{exp1}, @var{exp2})
6437 Returns the maximum of @var{exp1} and @var{exp2}.
6440 @item MIN(@var{exp1}, @var{exp2})
6441 Returns the minimum of @var{exp1} and @var{exp2}.
6443 @item NEXT(@var{exp})
6444 @kindex NEXT(@var{exp})
6445 @cindex unallocated address, next
6446 Return the next unallocated address that is a multiple of @var{exp}.
6447 This function is closely related to @code{ALIGN(@var{exp})}; unless you
6448 use the @code{MEMORY} command to define discontinuous memory for the
6449 output file, the two functions are equivalent.
6451 @item ORIGIN(@var{memory})
6452 @kindex ORIGIN(@var{memory})
6453 Return the origin of the memory region named @var{memory}.
6455 @item SEGMENT_START(@var{segment}, @var{default})
6456 @kindex SEGMENT_START(@var{segment}, @var{default})
6457 Return the base address of the named @var{segment}. If an explicit
6458 value has already been given for this segment (with a command-line
6459 @samp{-T} option) then that value will be returned otherwise the value
6460 will be @var{default}. At present, the @samp{-T} command-line option
6461 can only be used to set the base address for the ``text'', ``data'', and
6462 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
6465 @item SIZEOF(@var{section})
6466 @kindex SIZEOF(@var{section})
6467 @cindex section size
6468 Return the size in bytes of the named @var{section}, if that section has
6469 been allocated. If the section has not been allocated when this is
6470 evaluated, the linker will report an error. In the following example,
6471 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6480 symbol_1 = .end - .start ;
6481 symbol_2 = SIZEOF(.output);
6486 @item SIZEOF_HEADERS
6487 @itemx sizeof_headers
6488 @kindex SIZEOF_HEADERS
6490 Return the size in bytes of the output file's headers. This is
6491 information which appears at the start of the output file. You can use
6492 this number when setting the start address of the first section, if you
6493 choose, to facilitate paging.
6495 @cindex not enough room for program headers
6496 @cindex program headers, not enough room
6497 When producing an ELF output file, if the linker script uses the
6498 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6499 number of program headers before it has determined all the section
6500 addresses and sizes. If the linker later discovers that it needs
6501 additional program headers, it will report an error @samp{not enough
6502 room for program headers}. To avoid this error, you must avoid using
6503 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6504 script to avoid forcing the linker to use additional program headers, or
6505 you must define the program headers yourself using the @code{PHDRS}
6506 command (@pxref{PHDRS}).
6509 @node Implicit Linker Scripts
6510 @section Implicit Linker Scripts
6511 @cindex implicit linker scripts
6512 If you specify a linker input file which the linker can not recognize as
6513 an object file or an archive file, it will try to read the file as a
6514 linker script. If the file can not be parsed as a linker script, the
6515 linker will report an error.
6517 An implicit linker script will not replace the default linker script.
6519 Typically an implicit linker script would contain only symbol
6520 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6523 Any input files read because of an implicit linker script will be read
6524 at the position in the command line where the implicit linker script was
6525 read. This can affect archive searching.
6528 @node Machine Dependent
6529 @chapter Machine Dependent Features
6531 @cindex machine dependencies
6532 @command{ld} has additional features on some platforms; the following
6533 sections describe them. Machines where @command{ld} has no additional
6534 functionality are not listed.
6538 * H8/300:: @command{ld} and the H8/300
6541 * i960:: @command{ld} and the Intel 960 family
6544 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6547 * ARM:: @command{ld} and the ARM family
6550 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6553 * M68K:: @command{ld} and the Motorola 68K family
6556 * MIPS:: @command{ld} and the MIPS family
6559 * MMIX:: @command{ld} and MMIX
6562 * MSP430:: @command{ld} and MSP430
6565 * NDS32:: @command{ld} and NDS32
6568 * Nios II:: @command{ld} and the Altera Nios II
6571 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6574 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6577 * S/390 ELF:: @command{ld} and S/390 ELF Support
6580 * SPU ELF:: @command{ld} and SPU ELF Support
6583 * TI COFF:: @command{ld} and TI COFF
6586 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6589 * Xtensa:: @command{ld} and Xtensa Processors
6600 @section @command{ld} and the H8/300
6602 @cindex H8/300 support
6603 For the H8/300, @command{ld} can perform these global optimizations when
6604 you specify the @samp{--relax} command-line option.
6607 @cindex relaxing on H8/300
6608 @item relaxing address modes
6609 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6610 targets are within eight bits, and turns them into eight-bit
6611 program-counter relative @code{bsr} and @code{bra} instructions,
6614 @cindex synthesizing on H8/300
6615 @item synthesizing instructions
6616 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6617 @command{ld} finds all @code{mov.b} instructions which use the
6618 sixteen-bit absolute address form, but refer to the top
6619 page of memory, and changes them to use the eight-bit address form.
6620 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6621 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6622 top page of memory).
6624 @command{ld} finds all @code{mov} instructions which use the register
6625 indirect with 32-bit displacement addressing mode, but use a small
6626 displacement inside 16-bit displacement range, and changes them to use
6627 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6628 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6629 whenever the displacement @var{d} is in the 16 bit signed integer
6630 range. Only implemented in ELF-format ld).
6632 @item bit manipulation instructions
6633 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6634 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6635 which use 32 bit and 16 bit absolute address form, but refer to the top
6636 page of memory, and changes them to use the 8 bit address form.
6637 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6638 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6639 the top page of memory).
6641 @item system control instructions
6642 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6643 32 bit absolute address form, but refer to the top page of memory, and
6644 changes them to use 16 bit address form.
6645 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6646 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6647 the top page of memory).
6657 @c This stuff is pointless to say unless you're especially concerned
6658 @c with Renesas chips; don't enable it for generic case, please.
6660 @chapter @command{ld} and Other Renesas Chips
6662 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6663 H8/500, and SH chips. No special features, commands, or command-line
6664 options are required for these chips.
6674 @section @command{ld} and the Intel 960 Family
6676 @cindex i960 support
6678 You can use the @samp{-A@var{architecture}} command line option to
6679 specify one of the two-letter names identifying members of the 960
6680 family; the option specifies the desired output target, and warns of any
6681 incompatible instructions in the input files. It also modifies the
6682 linker's search strategy for archive libraries, to support the use of
6683 libraries specific to each particular architecture, by including in the
6684 search loop names suffixed with the string identifying the architecture.
6686 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
6687 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
6688 paths, and in any paths you specify with @samp{-L}) for a library with
6701 The first two possibilities would be considered in any event; the last
6702 two are due to the use of @w{@samp{-ACA}}.
6704 You can meaningfully use @samp{-A} more than once on a command line, since
6705 the 960 architecture family allows combination of target architectures; each
6706 use will add another pair of name variants to search for when @w{@samp{-l}}
6707 specifies a library.
6709 @cindex @option{--relax} on i960
6710 @cindex relaxing on i960
6711 @command{ld} supports the @samp{--relax} option for the i960 family. If
6712 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
6713 @code{calx} instructions whose targets are within 24 bits, and turns
6714 them into 24-bit program-counter relative @code{bal} and @code{cal}
6715 instructions, respectively. @command{ld} also turns @code{cal}
6716 instructions into @code{bal} instructions when it determines that the
6717 target subroutine is a leaf routine (that is, the target subroutine does
6718 not itself call any subroutines).
6735 @node M68HC11/68HC12
6736 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6738 @cindex M68HC11 and 68HC12 support
6740 @subsection Linker Relaxation
6742 For the Motorola 68HC11, @command{ld} can perform these global
6743 optimizations when you specify the @samp{--relax} command-line option.
6746 @cindex relaxing on M68HC11
6747 @item relaxing address modes
6748 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6749 targets are within eight bits, and turns them into eight-bit
6750 program-counter relative @code{bsr} and @code{bra} instructions,
6753 @command{ld} also looks at all 16-bit extended addressing modes and
6754 transforms them in a direct addressing mode when the address is in
6755 page 0 (between 0 and 0x0ff).
6757 @item relaxing gcc instruction group
6758 When @command{gcc} is called with @option{-mrelax}, it can emit group
6759 of instructions that the linker can optimize to use a 68HC11 direct
6760 addressing mode. These instructions consists of @code{bclr} or
6761 @code{bset} instructions.
6765 @subsection Trampoline Generation
6767 @cindex trampoline generation on M68HC11
6768 @cindex trampoline generation on M68HC12
6769 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6770 call a far function using a normal @code{jsr} instruction. The linker
6771 will also change the relocation to some far function to use the
6772 trampoline address instead of the function address. This is typically the
6773 case when a pointer to a function is taken. The pointer will in fact
6774 point to the function trampoline.
6782 @section @command{ld} and the ARM family
6784 @cindex ARM interworking support
6785 @kindex --support-old-code
6786 For the ARM, @command{ld} will generate code stubs to allow functions calls
6787 between ARM and Thumb code. These stubs only work with code that has
6788 been compiled and assembled with the @samp{-mthumb-interwork} command
6789 line option. If it is necessary to link with old ARM object files or
6790 libraries, which have not been compiled with the -mthumb-interwork
6791 option then the @samp{--support-old-code} command line switch should be
6792 given to the linker. This will make it generate larger stub functions
6793 which will work with non-interworking aware ARM code. Note, however,
6794 the linker does not support generating stubs for function calls to
6795 non-interworking aware Thumb code.
6797 @cindex thumb entry point
6798 @cindex entry point, thumb
6799 @kindex --thumb-entry=@var{entry}
6800 The @samp{--thumb-entry} switch is a duplicate of the generic
6801 @samp{--entry} switch, in that it sets the program's starting address.
6802 But it also sets the bottom bit of the address, so that it can be
6803 branched to using a BX instruction, and the program will start
6804 executing in Thumb mode straight away.
6806 @cindex PE import table prefixing
6807 @kindex --use-nul-prefixed-import-tables
6808 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6809 the import tables idata4 and idata5 have to be generated with a zero
6810 element prefix for import libraries. This is the old style to generate
6811 import tables. By default this option is turned off.
6815 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6816 executables. This option is only valid when linking big-endian
6817 objects - ie ones which have been assembled with the @option{-EB}
6818 option. The resulting image will contain big-endian data and
6822 @kindex --target1-rel
6823 @kindex --target1-abs
6824 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6825 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6826 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6827 and @samp{--target1-abs} switches override the default.
6830 @kindex --target2=@var{type}
6831 The @samp{--target2=type} switch overrides the default definition of the
6832 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6833 meanings, and target defaults are as follows:
6836 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6838 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6840 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6845 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6846 specification) enables objects compiled for the ARMv4 architecture to be
6847 interworking-safe when linked with other objects compiled for ARMv4t, but
6848 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6850 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6851 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6852 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6854 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6855 relocations are ignored.
6857 @cindex FIX_V4BX_INTERWORKING
6858 @kindex --fix-v4bx-interworking
6859 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6860 relocations with a branch to the following veneer:
6868 This allows generation of libraries/applications that work on ARMv4 cores
6869 and are still interworking safe. Note that the above veneer clobbers the
6870 condition flags, so may cause incorrect program behavior in rare cases.
6874 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6875 BLX instructions (available on ARMv5t and above) in various
6876 situations. Currently it is used to perform calls via the PLT from Thumb
6877 code using BLX rather than using BX and a mode-switching stub before
6878 each PLT entry. This should lead to such calls executing slightly faster.
6880 This option is enabled implicitly for SymbianOS, so there is no need to
6881 specify it if you are using that target.
6883 @cindex VFP11_DENORM_FIX
6884 @kindex --vfp11-denorm-fix
6885 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6886 bug in certain VFP11 coprocessor hardware, which sometimes allows
6887 instructions with denorm operands (which must be handled by support code)
6888 to have those operands overwritten by subsequent instructions before
6889 the support code can read the intended values.
6891 The bug may be avoided in scalar mode if you allow at least one
6892 intervening instruction between a VFP11 instruction which uses a register
6893 and another instruction which writes to the same register, or at least two
6894 intervening instructions if vector mode is in use. The bug only affects
6895 full-compliance floating-point mode: you do not need this workaround if
6896 you are using "runfast" mode. Please contact ARM for further details.
6898 If you know you are using buggy VFP11 hardware, you can
6899 enable this workaround by specifying the linker option
6900 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6901 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6902 vector mode (the latter also works for scalar code). The default is
6903 @samp{--vfp-denorm-fix=none}.
6905 If the workaround is enabled, instructions are scanned for
6906 potentially-troublesome sequences, and a veneer is created for each
6907 such sequence which may trigger the erratum. The veneer consists of the
6908 first instruction of the sequence and a branch back to the subsequent
6909 instruction. The original instruction is then replaced with a branch to
6910 the veneer. The extra cycles required to call and return from the veneer
6911 are sufficient to avoid the erratum in both the scalar and vector cases.
6913 @cindex ARM1176 erratum workaround
6914 @kindex --fix-arm1176
6915 @kindex --no-fix-arm1176
6916 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6917 in certain ARM1176 processors. The workaround is enabled by default if you
6918 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6919 unconditionally by specifying @samp{--no-fix-arm1176}.
6921 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6922 Programmer Advice Notice'' available on the ARM documentation website at:
6923 http://infocenter.arm.com/.
6925 @cindex STM32L4xx erratum workaround
6926 @kindex --fix-stm32l4xx-629360
6928 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
6929 workaround for a bug in the bus matrix / memory controller for some of
6930 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
6931 off-chip memory via the affected bus for bus reads of 9 words or more,
6932 the bus can generate corrupt data and/or abort. These are only
6933 core-initiated accesses (not DMA), and might affect any access:
6934 integer loads such as LDM, POP and floating-point loads such as VLDM,
6935 VPOP. Stores are not affected.
6937 The bug can be avoided by splitting memory accesses into the
6938 necessary chunks to keep bus reads below 8 words.
6940 The workaround is not enabled by default, this is equivalent to use
6941 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
6942 STM32L4xx hardware, you can enable the workaround by specifying the
6943 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
6944 @samp{--fix-stm32l4xx-629360=default}.
6946 If the workaround is enabled, instructions are scanned for
6947 potentially-troublesome sequences, and a veneer is created for each
6948 such sequence which may trigger the erratum. The veneer consists in a
6949 replacement sequence emulating the behaviour of the original one and a
6950 branch back to the subsequent instruction. The original instruction is
6951 then replaced with a branch to the veneer.
6953 The workaround does not always preserve the memory access order for
6954 the LDMDB instruction, when the instruction loads the PC.
6956 The workaround is not able to handle problematic instructions when
6957 they are in the middle of an IT block, since a branch is not allowed
6958 there. In that case, the linker reports a warning and no replacement
6961 The workaround is not able to replace problematic instructions with a
6962 PC-relative branch instruction if the @samp{.text} section is too
6963 large. In that case, when the branch that replaces the original code
6964 cannot be encoded, the linker reports a warning and no replacement
6967 @cindex NO_ENUM_SIZE_WARNING
6968 @kindex --no-enum-size-warning
6969 The @option{--no-enum-size-warning} switch prevents the linker from
6970 warning when linking object files that specify incompatible EABI
6971 enumeration size attributes. For example, with this switch enabled,
6972 linking of an object file using 32-bit enumeration values with another
6973 using enumeration values fitted into the smallest possible space will
6976 @cindex NO_WCHAR_SIZE_WARNING
6977 @kindex --no-wchar-size-warning
6978 The @option{--no-wchar-size-warning} switch prevents the linker from
6979 warning when linking object files that specify incompatible EABI
6980 @code{wchar_t} size attributes. For example, with this switch enabled,
6981 linking of an object file using 32-bit @code{wchar_t} values with another
6982 using 16-bit @code{wchar_t} values will not be diagnosed.
6985 @kindex --pic-veneer
6986 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6987 ARM/Thumb interworking veneers, even if the rest of the binary
6988 is not PIC. This avoids problems on uClinux targets where
6989 @samp{--emit-relocs} is used to generate relocatable binaries.
6991 @cindex STUB_GROUP_SIZE
6992 @kindex --stub-group-size=@var{N}
6993 The linker will automatically generate and insert small sequences of
6994 code into a linked ARM ELF executable whenever an attempt is made to
6995 perform a function call to a symbol that is too far away. The
6996 placement of these sequences of instructions - called stubs - is
6997 controlled by the command line option @option{--stub-group-size=N}.
6998 The placement is important because a poor choice can create a need for
6999 duplicate stubs, increasing the code size. The linker will try to
7000 group stubs together in order to reduce interruptions to the flow of
7001 code, but it needs guidance as to how big these groups should be and
7002 where they should be placed.
7004 The value of @samp{N}, the parameter to the
7005 @option{--stub-group-size=} option controls where the stub groups are
7006 placed. If it is negative then all stubs are placed after the first
7007 branch that needs them. If it is positive then the stubs can be
7008 placed either before or after the branches that need them. If the
7009 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
7010 exactly where to place groups of stubs, using its built in heuristics.
7011 A value of @samp{N} greater than 1 (or smaller than -1) tells the
7012 linker that a single group of stubs can service at most @samp{N} bytes
7013 from the input sections.
7015 The default, if @option{--stub-group-size=} is not specified, is
7018 Farcalls stubs insertion is fully supported for the ARM-EABI target
7019 only, because it relies on object files properties not present
7022 @cindex Cortex-A8 erratum workaround
7023 @kindex --fix-cortex-a8
7024 @kindex --no-fix-cortex-a8
7025 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}.
7027 The erratum only affects Thumb-2 code. Please contact ARM for further details.
7029 @cindex Cortex-A53 erratum 835769 workaround
7030 @kindex --fix-cortex-a53-835769
7031 @kindex --no-fix-cortex-a53-835769
7032 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}.
7034 Please contact ARM for further details.
7036 @kindex --merge-exidx-entries
7037 @kindex --no-merge-exidx-entries
7038 @cindex Merging exidx entries
7039 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
7042 @cindex 32-bit PLT entries
7043 The @samp{--long-plt} option enables the use of 16 byte PLT entries
7044 which support up to 4Gb of code. The default is to use 12 byte PLT
7045 entries which only support 512Mb of code.
7047 @kindex --no-apply-dynamic-relocs
7048 @cindex AArch64 rela addend
7049 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
7050 link-time values for dynamic relocations.
7052 @cindex Placement of SG veneers
7053 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
7054 Its start address must be set, either with the command line option
7055 @samp{--section-start} or in a linker script, to indicate where to place these
7058 @kindex --cmse-implib
7059 @cindex Secure gateway import library
7060 The @samp{--cmse-implib} option requests that the import libraries
7061 specified by the @samp{--out-implib} and @samp{--in-implib} options are
7062 secure gateway import libraries, suitable for linking a non-secure
7063 executable against secure code as per ARMv8-M Security Extensions.
7065 @kindex --in-implib=@var{file}
7066 @cindex Input import library
7067 The @samp{--in-implib=file} specifies an input import library whose symbols
7068 must keep the same address in the executable being produced. A warning is
7069 given if no @samp{--out-implib} is given but new symbols have been introduced
7070 in the executable that should be listed in its import library. Otherwise, if
7071 @samp{--out-implib} is specified, the symbols are added to the output import
7072 library. A warning is also given if some symbols present in the input import
7073 library have disappeared from the executable. This option is only effective
7074 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
7088 @section @command{ld} and HPPA 32-bit ELF Support
7089 @cindex HPPA multiple sub-space stubs
7090 @kindex --multi-subspace
7091 When generating a shared library, @command{ld} will by default generate
7092 import stubs suitable for use with a single sub-space application.
7093 The @samp{--multi-subspace} switch causes @command{ld} to generate export
7094 stubs, and different (larger) import stubs suitable for use with
7095 multiple sub-spaces.
7097 @cindex HPPA stub grouping
7098 @kindex --stub-group-size=@var{N}
7099 Long branch stubs and import/export stubs are placed by @command{ld} in
7100 stub sections located between groups of input sections.
7101 @samp{--stub-group-size} specifies the maximum size of a group of input
7102 sections handled by one stub section. Since branch offsets are signed,
7103 a stub section may serve two groups of input sections, one group before
7104 the stub section, and one group after it. However, when using
7105 conditional branches that require stubs, it may be better (for branch
7106 prediction) that stub sections only serve one group of input sections.
7107 A negative value for @samp{N} chooses this scheme, ensuring that
7108 branches to stubs always use a negative offset. Two special values of
7109 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7110 @command{ld} to automatically size input section groups for the branch types
7111 detected, with the same behaviour regarding stub placement as other
7112 positive or negative values of @samp{N} respectively.
7114 Note that @samp{--stub-group-size} does not split input sections. A
7115 single input section larger than the group size specified will of course
7116 create a larger group (of one section). If input sections are too
7117 large, it may not be possible for a branch to reach its stub.
7130 @section @command{ld} and the Motorola 68K family
7132 @cindex Motorola 68K GOT generation
7133 @kindex --got=@var{type}
7134 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
7135 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
7136 @samp{target}. When @samp{target} is selected the linker chooses
7137 the default GOT generation scheme for the current target.
7138 @samp{single} tells the linker to generate a single GOT with
7139 entries only at non-negative offsets.
7140 @samp{negative} instructs the linker to generate a single GOT with
7141 entries at both negative and positive offsets. Not all environments
7143 @samp{multigot} allows the linker to generate several GOTs in the
7144 output file. All GOT references from a single input object
7145 file access the same GOT, but references from different input object
7146 files might access different GOTs. Not all environments support such GOTs.
7159 @section @command{ld} and the MIPS family
7161 @cindex MIPS microMIPS instruction choice selection
7164 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
7165 microMIPS instructions used in code generated by the linker, such as that
7166 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
7167 used, then the linker only uses 32-bit instruction encodings. By default
7168 or if @samp{--no-insn32} is used, all instruction encodings are used,
7169 including 16-bit ones where possible.
7171 @cindex MIPS branch relocation check control
7172 @kindex --ignore-branch-isa
7173 @kindex --no-ignore-branch-isa
7174 The @samp{--ignore-branch-isa} and @samp{--no-ignore-branch-isa} options
7175 control branch relocation checks for invalid ISA mode transitions. If
7176 @samp{--ignore-branch-isa} is used, then the linker accepts any branch
7177 relocations and any ISA mode transition required is lost in relocation
7178 calculation, except for some cases of @code{BAL} instructions which meet
7179 relaxation conditions and are converted to equivalent @code{JALX}
7180 instructions as the associated relocation is calculated. By default
7181 or if @samp{--no-ignore-branch-isa} is used a check is made causing
7182 the loss of an ISA mode transition to produce an error.
7195 @section @code{ld} and MMIX
7196 For MMIX, there is a choice of generating @code{ELF} object files or
7197 @code{mmo} object files when linking. The simulator @code{mmix}
7198 understands the @code{mmo} format. The binutils @code{objcopy} utility
7199 can translate between the two formats.
7201 There is one special section, the @samp{.MMIX.reg_contents} section.
7202 Contents in this section is assumed to correspond to that of global
7203 registers, and symbols referring to it are translated to special symbols,
7204 equal to registers. In a final link, the start address of the
7205 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7206 global register multiplied by 8. Register @code{$255} is not included in
7207 this section; it is always set to the program entry, which is at the
7208 symbol @code{Main} for @code{mmo} files.
7210 Global symbols with the prefix @code{__.MMIX.start.}, for example
7211 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7212 The default linker script uses these to set the default start address
7215 Initial and trailing multiples of zero-valued 32-bit words in a section,
7216 are left out from an mmo file.
7229 @section @code{ld} and MSP430
7230 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7231 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7232 just pass @samp{-m help} option to the linker).
7234 @cindex MSP430 extra sections
7235 The linker will recognize some extra sections which are MSP430 specific:
7238 @item @samp{.vectors}
7239 Defines a portion of ROM where interrupt vectors located.
7241 @item @samp{.bootloader}
7242 Defines the bootloader portion of the ROM (if applicable). Any code
7243 in this section will be uploaded to the MPU.
7245 @item @samp{.infomem}
7246 Defines an information memory section (if applicable). Any code in
7247 this section will be uploaded to the MPU.
7249 @item @samp{.infomemnobits}
7250 This is the same as the @samp{.infomem} section except that any code
7251 in this section will not be uploaded to the MPU.
7253 @item @samp{.noinit}
7254 Denotes a portion of RAM located above @samp{.bss} section.
7256 The last two sections are used by gcc.
7260 @cindex MSP430 Options
7261 @kindex --code-region
7262 @item --code-region=[either,lower,upper,none]
7263 This will transform .text* sections to [either,lower,upper].text* sections. The
7264 argument passed to GCC for -mcode-region is propagated to the linker
7267 @kindex --data-region
7268 @item --data-region=[either,lower,upper,none]
7269 This will transform .data*, .bss* and .rodata* sections to
7270 [either,lower,upper].[data,bss,rodata]* sections. The argument passed to GCC
7271 for -mdata-region is propagated to the linker using this option.
7273 @kindex --disable-sec-transformation
7274 @item --disable-sec-transformation
7275 Prevent the transformation of sections as specified by the @code{--code-region}
7276 and @code{--data-region} options.
7277 This is useful if you are compiling and linking using a single call to the GCC
7278 wrapper, and want to compile the source files using -m[code,data]-region but
7279 not transform the sections for prebuilt libraries and objects.
7293 @section @code{ld} and NDS32
7294 @kindex relaxing on NDS32
7295 For NDS32, there are some options to select relaxation behavior. The linker
7296 relaxes objects according to these options.
7299 @item @samp{--m[no-]fp-as-gp}
7300 Disable/enable fp-as-gp relaxation.
7302 @item @samp{--mexport-symbols=FILE}
7303 Exporting symbols and their address into FILE as linker script.
7305 @item @samp{--m[no-]ex9}
7306 Disable/enable link-time EX9 relaxation.
7308 @item @samp{--mexport-ex9=FILE}
7309 Export the EX9 table after linking.
7311 @item @samp{--mimport-ex9=FILE}
7312 Import the Ex9 table for EX9 relaxation.
7314 @item @samp{--mupdate-ex9}
7315 Update the existing EX9 table.
7317 @item @samp{--mex9-limit=NUM}
7318 Maximum number of entries in the ex9 table.
7320 @item @samp{--mex9-loop-aware}
7321 Avoid generating the EX9 instruction inside the loop.
7323 @item @samp{--m[no-]ifc}
7324 Disable/enable the link-time IFC optimization.
7326 @item @samp{--mifc-loop-aware}
7327 Avoid generating the IFC instruction inside the loop.
7341 @section @command{ld} and the Altera Nios II
7342 @cindex Nios II call relaxation
7343 @kindex --relax on Nios II
7345 Call and immediate jump instructions on Nios II processors are limited to
7346 transferring control to addresses in the same 256MB memory segment,
7347 which may result in @command{ld} giving
7348 @samp{relocation truncated to fit} errors with very large programs.
7349 The command-line option @option{--relax} enables the generation of
7350 trampolines that can access the entire 32-bit address space for calls
7351 outside the normal @code{call} and @code{jmpi} address range. These
7352 trampolines are inserted at section boundaries, so may not themselves
7353 be reachable if an input section and its associated call trampolines are
7356 The @option{--relax} option is enabled by default unless @option{-r}
7357 is also specified. You can disable trampoline generation by using the
7358 @option{--no-relax} linker option. You can also disable this optimization
7359 locally by using the @samp{set .noat} directive in assembly-language
7360 source files, as the linker-inserted trampolines use the @code{at}
7361 register as a temporary.
7363 Note that the linker @option{--relax} option is independent of assembler
7364 relaxation options, and that using the GNU assembler's @option{-relax-all}
7365 option interferes with the linker's more selective call instruction relaxation.
7378 @section @command{ld} and PowerPC 32-bit ELF Support
7379 @cindex PowerPC long branches
7380 @kindex --relax on PowerPC
7381 Branches on PowerPC processors are limited to a signed 26-bit
7382 displacement, which may result in @command{ld} giving
7383 @samp{relocation truncated to fit} errors with very large programs.
7384 @samp{--relax} enables the generation of trampolines that can access
7385 the entire 32-bit address space. These trampolines are inserted at
7386 section boundaries, so may not themselves be reachable if an input
7387 section exceeds 33M in size. You may combine @samp{-r} and
7388 @samp{--relax} to add trampolines in a partial link. In that case
7389 both branches to undefined symbols and inter-section branches are also
7390 considered potentially out of range, and trampolines inserted.
7392 @cindex PowerPC ELF32 options
7397 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7398 generates code capable of using a newer PLT and GOT layout that has
7399 the security advantage of no executable section ever needing to be
7400 writable and no writable section ever being executable. PowerPC
7401 @command{ld} will generate this layout, including stubs to access the
7402 PLT, if all input files (including startup and static libraries) were
7403 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7404 BSS PLT (and GOT layout) which can give slightly better performance.
7406 @kindex --secure-plt
7408 @command{ld} will use the new PLT and GOT layout if it is linking new
7409 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7410 when linking non-PIC code. This option requests the new PLT and GOT
7411 layout. A warning will be given if some object file requires the old
7417 The new secure PLT and GOT are placed differently relative to other
7418 sections compared to older BSS PLT and GOT placement. The location of
7419 @code{.plt} must change because the new secure PLT is an initialized
7420 section while the old PLT is uninitialized. The reason for the
7421 @code{.got} change is more subtle: The new placement allows
7422 @code{.got} to be read-only in applications linked with
7423 @samp{-z relro -z now}. However, this placement means that
7424 @code{.sdata} cannot always be used in shared libraries, because the
7425 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7426 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7427 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7428 really only useful for other compilers that may do so.
7430 @cindex PowerPC stub symbols
7431 @kindex --emit-stub-syms
7432 @item --emit-stub-syms
7433 This option causes @command{ld} to label linker stubs with a local
7434 symbol that encodes the stub type and destination.
7436 @cindex PowerPC TLS optimization
7437 @kindex --no-tls-optimize
7438 @item --no-tls-optimize
7439 PowerPC @command{ld} normally performs some optimization of code
7440 sequences used to access Thread-Local Storage. Use this option to
7441 disable the optimization.
7454 @node PowerPC64 ELF64
7455 @section @command{ld} and PowerPC64 64-bit ELF Support
7457 @cindex PowerPC64 ELF64 options
7459 @cindex PowerPC64 stub grouping
7460 @kindex --stub-group-size
7461 @item --stub-group-size
7462 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
7463 by @command{ld} in stub sections located between groups of input sections.
7464 @samp{--stub-group-size} specifies the maximum size of a group of input
7465 sections handled by one stub section. Since branch offsets are signed,
7466 a stub section may serve two groups of input sections, one group before
7467 the stub section, and one group after it. However, when using
7468 conditional branches that require stubs, it may be better (for branch
7469 prediction) that stub sections only serve one group of input sections.
7470 A negative value for @samp{N} chooses this scheme, ensuring that
7471 branches to stubs always use a negative offset. Two special values of
7472 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7473 @command{ld} to automatically size input section groups for the branch types
7474 detected, with the same behaviour regarding stub placement as other
7475 positive or negative values of @samp{N} respectively.
7477 Note that @samp{--stub-group-size} does not split input sections. A
7478 single input section larger than the group size specified will of course
7479 create a larger group (of one section). If input sections are too
7480 large, it may not be possible for a branch to reach its stub.
7482 @cindex PowerPC64 stub symbols
7483 @kindex --emit-stub-syms
7484 @item --emit-stub-syms
7485 This option causes @command{ld} to label linker stubs with a local
7486 symbol that encodes the stub type and destination.
7488 @cindex PowerPC64 dot symbols
7490 @kindex --no-dotsyms
7493 These two options control how @command{ld} interprets version patterns
7494 in a version script. Older PowerPC64 compilers emitted both a
7495 function descriptor symbol with the same name as the function, and a
7496 code entry symbol with the name prefixed by a dot (@samp{.}). To
7497 properly version a function @samp{foo}, the version script thus needs
7498 to control both @samp{foo} and @samp{.foo}. The option
7499 @samp{--dotsyms}, on by default, automatically adds the required
7500 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
7503 @cindex PowerPC64 register save/restore functions
7504 @kindex --save-restore-funcs
7505 @kindex --no-save-restore-funcs
7506 @item --save-restore-funcs
7507 @itemx --no-save-restore-funcs
7508 These two options control whether PowerPC64 @command{ld} automatically
7509 provides out-of-line register save and restore functions used by
7510 @samp{-Os} code. The default is to provide any such referenced
7511 function for a normal final link, and to not do so for a relocatable
7514 @cindex PowerPC64 TLS optimization
7515 @kindex --no-tls-optimize
7516 @item --no-tls-optimize
7517 PowerPC64 @command{ld} normally performs some optimization of code
7518 sequences used to access Thread-Local Storage. Use this option to
7519 disable the optimization.
7521 @cindex PowerPC64 __tls_get_addr optimization
7522 @kindex --tls-get-addr-optimize
7523 @kindex --no-tls-get-addr-optimize
7524 @item --tls-get-addr-optimize
7525 @itemx --no-tls-get-addr-optimize
7526 These options control whether PowerPC64 @command{ld} uses a special
7527 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
7528 an optimization that allows the second and subsequent calls to
7529 @code{__tls_get_addr} for a given symbol to be resolved by the special
7530 stub without calling in to glibc. By default the linker enables this
7531 option when glibc advertises the availability of __tls_get_addr_opt.
7532 Forcing this option on when using an older glibc won't do much besides
7533 slow down your applications, but may be useful if linking an
7534 application against an older glibc with the expectation that it will
7535 normally be used on systems having a newer glibc.
7537 @cindex PowerPC64 OPD optimization
7538 @kindex --no-opd-optimize
7539 @item --no-opd-optimize
7540 PowerPC64 @command{ld} normally removes @code{.opd} section entries
7541 corresponding to deleted link-once functions, or functions removed by
7542 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
7543 Use this option to disable @code{.opd} optimization.
7545 @cindex PowerPC64 OPD spacing
7546 @kindex --non-overlapping-opd
7547 @item --non-overlapping-opd
7548 Some PowerPC64 compilers have an option to generate compressed
7549 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
7550 the static chain pointer (unused in C) with the first word of the next
7551 entry. This option expands such entries to the full 24 bytes.
7553 @cindex PowerPC64 TOC optimization
7554 @kindex --no-toc-optimize
7555 @item --no-toc-optimize
7556 PowerPC64 @command{ld} normally removes unused @code{.toc} section
7557 entries. Such entries are detected by examining relocations that
7558 reference the TOC in code sections. A reloc in a deleted code section
7559 marks a TOC word as unneeded, while a reloc in a kept code section
7560 marks a TOC word as needed. Since the TOC may reference itself, TOC
7561 relocs are also examined. TOC words marked as both needed and
7562 unneeded will of course be kept. TOC words without any referencing
7563 reloc are assumed to be part of a multi-word entry, and are kept or
7564 discarded as per the nearest marked preceding word. This works
7565 reliably for compiler generated code, but may be incorrect if assembly
7566 code is used to insert TOC entries. Use this option to disable the
7569 @cindex PowerPC64 multi-TOC
7570 @kindex --no-multi-toc
7571 @item --no-multi-toc
7572 If given any toc option besides @code{-mcmodel=medium} or
7573 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
7575 entries are accessed with a 16-bit offset from r2. This limits the
7576 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
7577 grouping code sections such that each group uses less than 64K for its
7578 TOC entries, then inserts r2 adjusting stubs between inter-group
7579 calls. @command{ld} does not split apart input sections, so cannot
7580 help if a single input file has a @code{.toc} section that exceeds
7581 64K, most likely from linking multiple files with @command{ld -r}.
7582 Use this option to turn off this feature.
7584 @cindex PowerPC64 TOC sorting
7585 @kindex --no-toc-sort
7587 By default, @command{ld} sorts TOC sections so that those whose file
7588 happens to have a section called @code{.init} or @code{.fini} are
7589 placed first, followed by TOC sections referenced by code generated
7590 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
7591 referenced only by code generated with PowerPC64 gcc's
7592 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
7593 results in better TOC grouping for multi-TOC. Use this option to turn
7596 @cindex PowerPC64 PLT stub alignment
7598 @kindex --no-plt-align
7600 @itemx --no-plt-align
7601 Use these options to control whether individual PLT call stubs are
7602 aligned to a 32-byte boundary, or to the specified power of two
7603 boundary when using @code{--plt-align=}. A negative value may be
7604 specified to pad PLT call stubs so that they do not cross the
7605 specified power of two boundary (or the minimum number of boundaries
7606 if a PLT stub is so large that it must cross a boundary). By default
7607 PLT call stubs are aligned to 32-byte boundaries.
7609 @cindex PowerPC64 PLT call stub static chain
7610 @kindex --plt-static-chain
7611 @kindex --no-plt-static-chain
7612 @item --plt-static-chain
7613 @itemx --no-plt-static-chain
7614 Use these options to control whether PLT call stubs load the static
7615 chain pointer (r11). @code{ld} defaults to not loading the static
7616 chain since there is never any need to do so on a PLT call.
7618 @cindex PowerPC64 PLT call stub thread safety
7619 @kindex --plt-thread-safe
7620 @kindex --no-plt-thread-safe
7621 @item --plt-thread-safe
7622 @itemx --no-thread-safe
7623 With power7's weakly ordered memory model, it is possible when using
7624 lazy binding for ld.so to update a plt entry in one thread and have
7625 another thread see the individual plt entry words update in the wrong
7626 order, despite ld.so carefully writing in the correct order and using
7627 memory write barriers. To avoid this we need some sort of read
7628 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
7629 looks for calls to commonly used functions that create threads, and if
7630 seen, adds the necessary barriers. Use these options to change the
7633 @cindex PowerPC64 ELFv2 PLT localentry optimization
7634 @kindex --plt-localentry
7635 @kindex --no-plt-localentry
7636 @item --plt-localentry
7637 @itemx --no-localentry
7638 ELFv2 functions with localentry:0 are those with a single entry point,
7639 ie. global entry == local entry, and that have no requirement on r2
7640 (the TOC/GOT pointer) or r12, and guarantee r2 is unchanged on return.
7641 Such an external function can be called via the PLT without saving r2
7642 or restoring it on return, avoiding a common load-hit-store for small
7643 functions. The optimization is attractive, with up to 40% reduction
7644 in execution time for a small function, but can result in symbol
7645 interposition failures. Also, minor changes in a shared library,
7646 including system libraries, can cause a function that was localentry:0
7647 to become localentry:8. This will result in a dynamic loader
7648 complaint and failure to run. The option is experimental, use with
7649 care. @option{--no-plt-localentry} is the default.
7663 @section @command{ld} and S/390 ELF Support
7665 @cindex S/390 ELF options
7669 @kindex --s390-pgste
7671 This option marks the result file with a @code{PT_S390_PGSTE}
7672 segment. The Linux kernel is supposed to allocate 4k page tables for
7673 binaries marked that way.
7687 @section @command{ld} and SPU ELF Support
7689 @cindex SPU ELF options
7695 This option marks an executable as a PIC plugin module.
7697 @cindex SPU overlays
7698 @kindex --no-overlays
7700 Normally, @command{ld} recognizes calls to functions within overlay
7701 regions, and redirects such calls to an overlay manager via a stub.
7702 @command{ld} also provides a built-in overlay manager. This option
7703 turns off all this special overlay handling.
7705 @cindex SPU overlay stub symbols
7706 @kindex --emit-stub-syms
7707 @item --emit-stub-syms
7708 This option causes @command{ld} to label overlay stubs with a local
7709 symbol that encodes the stub type and destination.
7711 @cindex SPU extra overlay stubs
7712 @kindex --extra-overlay-stubs
7713 @item --extra-overlay-stubs
7714 This option causes @command{ld} to add overlay call stubs on all
7715 function calls out of overlay regions. Normally stubs are not added
7716 on calls to non-overlay regions.
7718 @cindex SPU local store size
7719 @kindex --local-store=lo:hi
7720 @item --local-store=lo:hi
7721 @command{ld} usually checks that a final executable for SPU fits in
7722 the address range 0 to 256k. This option may be used to change the
7723 range. Disable the check entirely with @option{--local-store=0:0}.
7726 @kindex --stack-analysis
7727 @item --stack-analysis
7728 SPU local store space is limited. Over-allocation of stack space
7729 unnecessarily limits space available for code and data, while
7730 under-allocation results in runtime failures. If given this option,
7731 @command{ld} will provide an estimate of maximum stack usage.
7732 @command{ld} does this by examining symbols in code sections to
7733 determine the extents of functions, and looking at function prologues
7734 for stack adjusting instructions. A call-graph is created by looking
7735 for relocations on branch instructions. The graph is then searched
7736 for the maximum stack usage path. Note that this analysis does not
7737 find calls made via function pointers, and does not handle recursion
7738 and other cycles in the call graph. Stack usage may be
7739 under-estimated if your code makes such calls. Also, stack usage for
7740 dynamic allocation, e.g. alloca, will not be detected. If a link map
7741 is requested, detailed information about each function's stack usage
7742 and calls will be given.
7745 @kindex --emit-stack-syms
7746 @item --emit-stack-syms
7747 This option, if given along with @option{--stack-analysis} will result
7748 in @command{ld} emitting stack sizing symbols for each function.
7749 These take the form @code{__stack_<function_name>} for global
7750 functions, and @code{__stack_<number>_<function_name>} for static
7751 functions. @code{<number>} is the section id in hex. The value of
7752 such symbols is the stack requirement for the corresponding function.
7753 The symbol size will be zero, type @code{STT_NOTYPE}, binding
7754 @code{STB_LOCAL}, and section @code{SHN_ABS}.
7768 @section @command{ld}'s Support for Various TI COFF Versions
7769 @cindex TI COFF versions
7770 @kindex --format=@var{version}
7771 The @samp{--format} switch allows selection of one of the various
7772 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
7773 also supported. The TI COFF versions also vary in header byte-order
7774 format; @command{ld} will read any version or byte order, but the output
7775 header format depends on the default specified by the specific target.
7788 @section @command{ld} and WIN32 (cygwin/mingw)
7790 This section describes some of the win32 specific @command{ld} issues.
7791 See @ref{Options,,Command Line Options} for detailed description of the
7792 command line options mentioned here.
7795 @cindex import libraries
7796 @item import libraries
7797 The standard Windows linker creates and uses so-called import
7798 libraries, which contains information for linking to dll's. They are
7799 regular static archives and are handled as any other static
7800 archive. The cygwin and mingw ports of @command{ld} have specific
7801 support for creating such libraries provided with the
7802 @samp{--out-implib} command line option.
7804 @item exporting DLL symbols
7805 @cindex exporting DLL symbols
7806 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
7809 @item using auto-export functionality
7810 @cindex using auto-export functionality
7811 By default @command{ld} exports symbols with the auto-export functionality,
7812 which is controlled by the following command line options:
7815 @item --export-all-symbols [This is the default]
7816 @item --exclude-symbols
7817 @item --exclude-libs
7818 @item --exclude-modules-for-implib
7819 @item --version-script
7822 When auto-export is in operation, @command{ld} will export all the non-local
7823 (global and common) symbols it finds in a DLL, with the exception of a few
7824 symbols known to belong to the system's runtime and libraries. As it will
7825 often not be desirable to export all of a DLL's symbols, which may include
7826 private functions that are not part of any public interface, the command-line
7827 options listed above may be used to filter symbols out from the list for
7828 exporting. The @samp{--output-def} option can be used in order to see the
7829 final list of exported symbols with all exclusions taken into effect.
7831 If @samp{--export-all-symbols} is not given explicitly on the
7832 command line, then the default auto-export behavior will be @emph{disabled}
7833 if either of the following are true:
7836 @item A DEF file is used.
7837 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
7840 @item using a DEF file
7841 @cindex using a DEF file
7842 Another way of exporting symbols is using a DEF file. A DEF file is
7843 an ASCII file containing definitions of symbols which should be
7844 exported when a dll is created. Usually it is named @samp{<dll
7845 name>.def} and is added as any other object file to the linker's
7846 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7849 gcc -o <output> <objectfiles> <dll name>.def
7852 Using a DEF file turns off the normal auto-export behavior, unless the
7853 @samp{--export-all-symbols} option is also used.
7855 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7858 LIBRARY "xyz.dll" BASE=0x20000000
7864 another_foo = abc.dll.afoo
7870 This example defines a DLL with a non-default base address and seven
7871 symbols in the export table. The third exported symbol @code{_bar} is an
7872 alias for the second. The fourth symbol, @code{another_foo} is resolved
7873 by "forwarding" to another module and treating it as an alias for
7874 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7875 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7876 export library is an alias of @samp{foo}, which gets the string name
7877 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7878 symbol, which gets in export table the name @samp{var1}.
7880 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7881 name of the output DLL. If @samp{<name>} does not include a suffix,
7882 the default library suffix, @samp{.DLL} is appended.
7884 When the .DEF file is used to build an application, rather than a
7885 library, the @code{NAME <name>} command should be used instead of
7886 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7887 executable suffix, @samp{.EXE} is appended.
7889 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7890 specification @code{BASE = <number>} may be used to specify a
7891 non-default base address for the image.
7893 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7894 or they specify an empty string, the internal name is the same as the
7895 filename specified on the command line.
7897 The complete specification of an export symbol is:
7901 ( ( ( <name1> [ = <name2> ] )
7902 | ( <name1> = <module-name> . <external-name>))
7903 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7906 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7907 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7908 @samp{<name1>} as a "forward" alias for the symbol
7909 @samp{<external-name>} in the DLL @samp{<module-name>}.
7910 Optionally, the symbol may be exported by the specified ordinal
7911 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7912 string in import/export table for the symbol.
7914 The optional keywords that follow the declaration indicate:
7916 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7917 will still be exported by its ordinal alias (either the value specified
7918 by the .def specification or, otherwise, the value assigned by the
7919 linker). The symbol name, however, does remain visible in the import
7920 library (if any), unless @code{PRIVATE} is also specified.
7922 @code{DATA}: The symbol is a variable or object, rather than a function.
7923 The import lib will export only an indirect reference to @code{foo} as
7924 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7927 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7928 well as @code{_imp__foo} into the import library. Both refer to the
7929 read-only import address table's pointer to the variable, not to the
7930 variable itself. This can be dangerous. If the user code fails to add
7931 the @code{dllimport} attribute and also fails to explicitly add the
7932 extra indirection that the use of the attribute enforces, the
7933 application will behave unexpectedly.
7935 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7936 it into the static import library used to resolve imports at link time. The
7937 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7938 API at runtime or by using the GNU ld extension of linking directly to
7939 the DLL without an import library.
7941 See ld/deffilep.y in the binutils sources for the full specification of
7942 other DEF file statements
7944 @cindex creating a DEF file
7945 While linking a shared dll, @command{ld} is able to create a DEF file
7946 with the @samp{--output-def <file>} command line option.
7948 @item Using decorations
7949 @cindex Using decorations
7950 Another way of marking symbols for export is to modify the source code
7951 itself, so that when building the DLL each symbol to be exported is
7955 __declspec(dllexport) int a_variable
7956 __declspec(dllexport) void a_function(int with_args)
7959 All such symbols will be exported from the DLL. If, however,
7960 any of the object files in the DLL contain symbols decorated in
7961 this way, then the normal auto-export behavior is disabled, unless
7962 the @samp{--export-all-symbols} option is also used.
7964 Note that object files that wish to access these symbols must @emph{not}
7965 decorate them with dllexport. Instead, they should use dllimport,
7969 __declspec(dllimport) int a_variable
7970 __declspec(dllimport) void a_function(int with_args)
7973 This complicates the structure of library header files, because
7974 when included by the library itself the header must declare the
7975 variables and functions as dllexport, but when included by client
7976 code the header must declare them as dllimport. There are a number
7977 of idioms that are typically used to do this; often client code can
7978 omit the __declspec() declaration completely. See
7979 @samp{--enable-auto-import} and @samp{automatic data imports} for more
7983 @cindex automatic data imports
7984 @item automatic data imports
7985 The standard Windows dll format supports data imports from dlls only
7986 by adding special decorations (dllimport/dllexport), which let the
7987 compiler produce specific assembler instructions to deal with this
7988 issue. This increases the effort necessary to port existing Un*x
7989 code to these platforms, especially for large
7990 c++ libraries and applications. The auto-import feature, which was
7991 initially provided by Paul Sokolovsky, allows one to omit the
7992 decorations to achieve a behavior that conforms to that on POSIX/Un*x
7993 platforms. This feature is enabled with the @samp{--enable-auto-import}
7994 command-line option, although it is enabled by default on cygwin/mingw.
7995 The @samp{--enable-auto-import} option itself now serves mainly to
7996 suppress any warnings that are ordinarily emitted when linked objects
7997 trigger the feature's use.
7999 auto-import of variables does not always work flawlessly without
8000 additional assistance. Sometimes, you will see this message
8002 "variable '<var>' can't be auto-imported. Please read the
8003 documentation for ld's @code{--enable-auto-import} for details."
8005 The @samp{--enable-auto-import} documentation explains why this error
8006 occurs, and several methods that can be used to overcome this difficulty.
8007 One of these methods is the @emph{runtime pseudo-relocs} feature, described
8010 @cindex runtime pseudo-relocation
8011 For complex variables imported from DLLs (such as structs or classes),
8012 object files typically contain a base address for the variable and an
8013 offset (@emph{addend}) within the variable--to specify a particular
8014 field or public member, for instance. Unfortunately, the runtime loader used
8015 in win32 environments is incapable of fixing these references at runtime
8016 without the additional information supplied by dllimport/dllexport decorations.
8017 The standard auto-import feature described above is unable to resolve these
8020 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
8021 be resolved without error, while leaving the task of adjusting the references
8022 themselves (with their non-zero addends) to specialized code provided by the
8023 runtime environment. Recent versions of the cygwin and mingw environments and
8024 compilers provide this runtime support; older versions do not. However, the
8025 support is only necessary on the developer's platform; the compiled result will
8026 run without error on an older system.
8028 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
8031 @cindex direct linking to a dll
8032 @item direct linking to a dll
8033 The cygwin/mingw ports of @command{ld} support the direct linking,
8034 including data symbols, to a dll without the usage of any import
8035 libraries. This is much faster and uses much less memory than does the
8036 traditional import library method, especially when linking large
8037 libraries or applications. When @command{ld} creates an import lib, each
8038 function or variable exported from the dll is stored in its own bfd, even
8039 though a single bfd could contain many exports. The overhead involved in
8040 storing, loading, and processing so many bfd's is quite large, and explains the
8041 tremendous time, memory, and storage needed to link against particularly
8042 large or complex libraries when using import libs.
8044 Linking directly to a dll uses no extra command-line switches other than
8045 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
8046 of names to match each library. All that is needed from the developer's
8047 perspective is an understanding of this search, in order to force ld to
8048 select the dll instead of an import library.
8051 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
8052 to find, in the first directory of its search path,
8064 before moving on to the next directory in the search path.
8066 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
8067 where @samp{<prefix>} is set by the @command{ld} option
8068 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
8069 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
8072 Other win32-based unix environments, such as mingw or pw32, may use other
8073 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
8074 was originally intended to help avoid name conflicts among dll's built for the
8075 various win32/un*x environments, so that (for example) two versions of a zlib dll
8076 could coexist on the same machine.
8078 The generic cygwin/mingw path layout uses a @samp{bin} directory for
8079 applications and dll's and a @samp{lib} directory for the import
8080 libraries (using cygwin nomenclature):
8086 libxxx.dll.a (in case of dll's)
8087 libxxx.a (in case of static archive)
8090 Linking directly to a dll without using the import library can be
8093 1. Use the dll directly by adding the @samp{bin} path to the link line
8095 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
8098 However, as the dll's often have version numbers appended to their names
8099 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
8100 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
8101 not versioned, and do not have this difficulty.
8103 2. Create a symbolic link from the dll to a file in the @samp{lib}
8104 directory according to the above mentioned search pattern. This
8105 should be used to avoid unwanted changes in the tools needed for
8109 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
8112 Then you can link without any make environment changes.
8115 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
8118 This technique also avoids the version number problems, because the following is
8125 libxxx.dll.a -> ../bin/cygxxx-5.dll
8128 Linking directly to a dll without using an import lib will work
8129 even when auto-import features are exercised, and even when
8130 @samp{--enable-runtime-pseudo-relocs} is used.
8132 Given the improvements in speed and memory usage, one might justifiably
8133 wonder why import libraries are used at all. There are three reasons:
8135 1. Until recently, the link-directly-to-dll functionality did @emph{not}
8136 work with auto-imported data.
8138 2. Sometimes it is necessary to include pure static objects within the
8139 import library (which otherwise contains only bfd's for indirection
8140 symbols that point to the exports of a dll). Again, the import lib
8141 for the cygwin kernel makes use of this ability, and it is not
8142 possible to do this without an import lib.
8144 3. Symbol aliases can only be resolved using an import lib. This is
8145 critical when linking against OS-supplied dll's (eg, the win32 API)
8146 in which symbols are usually exported as undecorated aliases of their
8147 stdcall-decorated assembly names.
8149 So, import libs are not going away. But the ability to replace
8150 true import libs with a simple symbolic link to (or a copy of)
8151 a dll, in many cases, is a useful addition to the suite of tools
8152 binutils makes available to the win32 developer. Given the
8153 massive improvements in memory requirements during linking, storage
8154 requirements, and linking speed, we expect that many developers
8155 will soon begin to use this feature whenever possible.
8157 @item symbol aliasing
8159 @item adding additional names
8160 Sometimes, it is useful to export symbols with additional names.
8161 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
8162 exported as @samp{_foo} by using special directives in the DEF file
8163 when creating the dll. This will affect also the optional created
8164 import library. Consider the following DEF file:
8167 LIBRARY "xyz.dll" BASE=0x61000000
8174 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
8176 Another method for creating a symbol alias is to create it in the
8177 source code using the "weak" attribute:
8180 void foo () @{ /* Do something. */; @}
8181 void _foo () __attribute__ ((weak, alias ("foo")));
8184 See the gcc manual for more information about attributes and weak
8187 @item renaming symbols
8188 Sometimes it is useful to rename exports. For instance, the cygwin
8189 kernel does this regularly. A symbol @samp{_foo} can be exported as
8190 @samp{foo} but not as @samp{_foo} by using special directives in the
8191 DEF file. (This will also affect the import library, if it is
8192 created). In the following example:
8195 LIBRARY "xyz.dll" BASE=0x61000000
8201 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
8205 Note: using a DEF file disables the default auto-export behavior,
8206 unless the @samp{--export-all-symbols} command line option is used.
8207 If, however, you are trying to rename symbols, then you should list
8208 @emph{all} desired exports in the DEF file, including the symbols
8209 that are not being renamed, and do @emph{not} use the
8210 @samp{--export-all-symbols} option. If you list only the
8211 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
8212 to handle the other symbols, then the both the new names @emph{and}
8213 the original names for the renamed symbols will be exported.
8214 In effect, you'd be aliasing those symbols, not renaming them,
8215 which is probably not what you wanted.
8217 @cindex weak externals
8218 @item weak externals
8219 The Windows object format, PE, specifies a form of weak symbols called
8220 weak externals. When a weak symbol is linked and the symbol is not
8221 defined, the weak symbol becomes an alias for some other symbol. There
8222 are three variants of weak externals:
8224 @item Definition is searched for in objects and libraries, historically
8225 called lazy externals.
8226 @item Definition is searched for only in other objects, not in libraries.
8227 This form is not presently implemented.
8228 @item No search; the symbol is an alias. This form is not presently
8231 As a GNU extension, weak symbols that do not specify an alternate symbol
8232 are supported. If the symbol is undefined when linking, the symbol
8233 uses a default value.
8235 @cindex aligned common symbols
8236 @item aligned common symbols
8237 As a GNU extension to the PE file format, it is possible to specify the
8238 desired alignment for a common symbol. This information is conveyed from
8239 the assembler or compiler to the linker by means of GNU-specific commands
8240 carried in the object file's @samp{.drectve} section, which are recognized
8241 by @command{ld} and respected when laying out the common symbols. Native
8242 tools will be able to process object files employing this GNU extension,
8243 but will fail to respect the alignment instructions, and may issue noisy
8244 warnings about unknown linker directives.
8259 @section @code{ld} and Xtensa Processors
8261 @cindex Xtensa processors
8262 The default @command{ld} behavior for Xtensa processors is to interpret
8263 @code{SECTIONS} commands so that lists of explicitly named sections in a
8264 specification with a wildcard file will be interleaved when necessary to
8265 keep literal pools within the range of PC-relative load offsets. For
8266 example, with the command:
8278 @command{ld} may interleave some of the @code{.literal}
8279 and @code{.text} sections from different object files to ensure that the
8280 literal pools are within the range of PC-relative load offsets. A valid
8281 interleaving might place the @code{.literal} sections from an initial
8282 group of files followed by the @code{.text} sections of that group of
8283 files. Then, the @code{.literal} sections from the rest of the files
8284 and the @code{.text} sections from the rest of the files would follow.
8286 @cindex @option{--relax} on Xtensa
8287 @cindex relaxing on Xtensa
8288 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8289 provides two important link-time optimizations. The first optimization
8290 is to combine identical literal values to reduce code size. A redundant
8291 literal will be removed and all the @code{L32R} instructions that use it
8292 will be changed to reference an identical literal, as long as the
8293 location of the replacement literal is within the offset range of all
8294 the @code{L32R} instructions. The second optimization is to remove
8295 unnecessary overhead from assembler-generated ``longcall'' sequences of
8296 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8297 range of direct @code{CALL@var{n}} instructions.
8299 For each of these cases where an indirect call sequence can be optimized
8300 to a direct call, the linker will change the @code{CALLX@var{n}}
8301 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8302 instruction, and remove the literal referenced by the @code{L32R}
8303 instruction if it is not used for anything else. Removing the
8304 @code{L32R} instruction always reduces code size but can potentially
8305 hurt performance by changing the alignment of subsequent branch targets.
8306 By default, the linker will always preserve alignments, either by
8307 switching some instructions between 24-bit encodings and the equivalent
8308 density instructions or by inserting a no-op in place of the @code{L32R}
8309 instruction that was removed. If code size is more important than
8310 performance, the @option{--size-opt} option can be used to prevent the
8311 linker from widening density instructions or inserting no-ops, except in
8312 a few cases where no-ops are required for correctness.
8314 The following Xtensa-specific command-line options can be used to
8317 @cindex Xtensa options
8320 When optimizing indirect calls to direct calls, optimize for code size
8321 more than performance. With this option, the linker will not insert
8322 no-ops or widen density instructions to preserve branch target
8323 alignment. There may still be some cases where no-ops are required to
8324 preserve the correctness of the code.
8332 @ifclear SingleFormat
8337 @cindex object file management
8338 @cindex object formats available
8340 The linker accesses object and archive files using the BFD libraries.
8341 These libraries allow the linker to use the same routines to operate on
8342 object files whatever the object file format. A different object file
8343 format can be supported simply by creating a new BFD back end and adding
8344 it to the library. To conserve runtime memory, however, the linker and
8345 associated tools are usually configured to support only a subset of the
8346 object file formats available. You can use @code{objdump -i}
8347 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8348 list all the formats available for your configuration.
8350 @cindex BFD requirements
8351 @cindex requirements for BFD
8352 As with most implementations, BFD is a compromise between
8353 several conflicting requirements. The major factor influencing
8354 BFD design was efficiency: any time used converting between
8355 formats is time which would not have been spent had BFD not
8356 been involved. This is partly offset by abstraction payback; since
8357 BFD simplifies applications and back ends, more time and care
8358 may be spent optimizing algorithms for a greater speed.
8360 One minor artifact of the BFD solution which you should bear in
8361 mind is the potential for information loss. There are two places where
8362 useful information can be lost using the BFD mechanism: during
8363 conversion and during output. @xref{BFD information loss}.
8366 * BFD outline:: How it works: an outline of BFD
8370 @section How It Works: An Outline of BFD
8371 @cindex opening object files
8372 @include bfdsumm.texi
8375 @node Reporting Bugs
8376 @chapter Reporting Bugs
8377 @cindex bugs in @command{ld}
8378 @cindex reporting bugs in @command{ld}
8380 Your bug reports play an essential role in making @command{ld} reliable.
8382 Reporting a bug may help you by bringing a solution to your problem, or
8383 it may not. But in any case the principal function of a bug report is
8384 to help the entire community by making the next version of @command{ld}
8385 work better. Bug reports are your contribution to the maintenance of
8388 In order for a bug report to serve its purpose, you must include the
8389 information that enables us to fix the bug.
8392 * Bug Criteria:: Have you found a bug?
8393 * Bug Reporting:: How to report bugs
8397 @section Have You Found a Bug?
8398 @cindex bug criteria
8400 If you are not sure whether you have found a bug, here are some guidelines:
8403 @cindex fatal signal
8404 @cindex linker crash
8405 @cindex crash of linker
8407 If the linker gets a fatal signal, for any input whatever, that is a
8408 @command{ld} bug. Reliable linkers never crash.
8410 @cindex error on valid input
8412 If @command{ld} produces an error message for valid input, that is a bug.
8414 @cindex invalid input
8416 If @command{ld} does not produce an error message for invalid input, that
8417 may be a bug. In the general case, the linker can not verify that
8418 object files are correct.
8421 If you are an experienced user of linkers, your suggestions for
8422 improvement of @command{ld} are welcome in any case.
8426 @section How to Report Bugs
8428 @cindex @command{ld} bugs, reporting
8430 A number of companies and individuals offer support for @sc{gnu}
8431 products. If you obtained @command{ld} from a support organization, we
8432 recommend you contact that organization first.
8434 You can find contact information for many support companies and
8435 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
8439 Otherwise, send bug reports for @command{ld} to
8443 The fundamental principle of reporting bugs usefully is this:
8444 @strong{report all the facts}. If you are not sure whether to state a
8445 fact or leave it out, state it!
8447 Often people omit facts because they think they know what causes the
8448 problem and assume that some details do not matter. Thus, you might
8449 assume that the name of a symbol you use in an example does not
8450 matter. Well, probably it does not, but one cannot be sure. Perhaps
8451 the bug is a stray memory reference which happens to fetch from the
8452 location where that name is stored in memory; perhaps, if the name
8453 were different, the contents of that location would fool the linker
8454 into doing the right thing despite the bug. Play it safe and give a
8455 specific, complete example. That is the easiest thing for you to do,
8456 and the most helpful.
8458 Keep in mind that the purpose of a bug report is to enable us to fix
8459 the bug if it is new to us. Therefore, always write your bug reports
8460 on the assumption that the bug has not been reported previously.
8462 Sometimes people give a few sketchy facts and ask, ``Does this ring a
8463 bell?'' This cannot help us fix a bug, so it is basically useless. We
8464 respond by asking for enough details to enable us to investigate.
8465 You might as well expedite matters by sending them to begin with.
8467 To enable us to fix the bug, you should include all these things:
8471 The version of @command{ld}. @command{ld} announces it if you start it with
8472 the @samp{--version} argument.
8474 Without this, we will not know whether there is any point in looking for
8475 the bug in the current version of @command{ld}.
8478 Any patches you may have applied to the @command{ld} source, including any
8479 patches made to the @code{BFD} library.
8482 The type of machine you are using, and the operating system name and
8486 What compiler (and its version) was used to compile @command{ld}---e.g.
8490 The command arguments you gave the linker to link your example and
8491 observe the bug. To guarantee you will not omit something important,
8492 list them all. A copy of the Makefile (or the output from make) is
8495 If we were to try to guess the arguments, we would probably guess wrong
8496 and then we might not encounter the bug.
8499 A complete input file, or set of input files, that will reproduce the
8500 bug. It is generally most helpful to send the actual object files
8501 provided that they are reasonably small. Say no more than 10K. For
8502 bigger files you can either make them available by FTP or HTTP or else
8503 state that you are willing to send the object file(s) to whomever
8504 requests them. (Note - your email will be going to a mailing list, so
8505 we do not want to clog it up with large attachments). But small
8506 attachments are best.
8508 If the source files were assembled using @code{gas} or compiled using
8509 @code{gcc}, then it may be OK to send the source files rather than the
8510 object files. In this case, be sure to say exactly what version of
8511 @code{gas} or @code{gcc} was used to produce the object files. Also say
8512 how @code{gas} or @code{gcc} were configured.
8515 A description of what behavior you observe that you believe is
8516 incorrect. For example, ``It gets a fatal signal.''
8518 Of course, if the bug is that @command{ld} gets a fatal signal, then we
8519 will certainly notice it. But if the bug is incorrect output, we might
8520 not notice unless it is glaringly wrong. You might as well not give us
8521 a chance to make a mistake.
8523 Even if the problem you experience is a fatal signal, you should still
8524 say so explicitly. Suppose something strange is going on, such as, your
8525 copy of @command{ld} is out of sync, or you have encountered a bug in the
8526 C library on your system. (This has happened!) Your copy might crash
8527 and ours would not. If you told us to expect a crash, then when ours
8528 fails to crash, we would know that the bug was not happening for us. If
8529 you had not told us to expect a crash, then we would not be able to draw
8530 any conclusion from our observations.
8533 If you wish to suggest changes to the @command{ld} source, send us context
8534 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
8535 @samp{-p} option. Always send diffs from the old file to the new file.
8536 If you even discuss something in the @command{ld} source, refer to it by
8537 context, not by line number.
8539 The line numbers in our development sources will not match those in your
8540 sources. Your line numbers would convey no useful information to us.
8543 Here are some things that are not necessary:
8547 A description of the envelope of the bug.
8549 Often people who encounter a bug spend a lot of time investigating
8550 which changes to the input file will make the bug go away and which
8551 changes will not affect it.
8553 This is often time consuming and not very useful, because the way we
8554 will find the bug is by running a single example under the debugger
8555 with breakpoints, not by pure deduction from a series of examples.
8556 We recommend that you save your time for something else.
8558 Of course, if you can find a simpler example to report @emph{instead}
8559 of the original one, that is a convenience for us. Errors in the
8560 output will be easier to spot, running under the debugger will take
8561 less time, and so on.
8563 However, simplification is not vital; if you do not want to do this,
8564 report the bug anyway and send us the entire test case you used.
8567 A patch for the bug.
8569 A patch for the bug does help us if it is a good one. But do not omit
8570 the necessary information, such as the test case, on the assumption that
8571 a patch is all we need. We might see problems with your patch and decide
8572 to fix the problem another way, or we might not understand it at all.
8574 Sometimes with a program as complicated as @command{ld} it is very hard to
8575 construct an example that will make the program follow a certain path
8576 through the code. If you do not send us the example, we will not be
8577 able to construct one, so we will not be able to verify that the bug is
8580 And if we cannot understand what bug you are trying to fix, or why your
8581 patch should be an improvement, we will not install it. A test case will
8582 help us to understand.
8585 A guess about what the bug is or what it depends on.
8587 Such guesses are usually wrong. Even we cannot guess right about such
8588 things without first using the debugger to find the facts.
8592 @appendix MRI Compatible Script Files
8593 @cindex MRI compatibility
8594 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
8595 linker, @command{ld} can use MRI compatible linker scripts as an
8596 alternative to the more general-purpose linker scripting language
8597 described in @ref{Scripts}. MRI compatible linker scripts have a much
8598 simpler command set than the scripting language otherwise used with
8599 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
8600 linker commands; these commands are described here.
8602 In general, MRI scripts aren't of much use with the @code{a.out} object
8603 file format, since it only has three sections and MRI scripts lack some
8604 features to make use of them.
8606 You can specify a file containing an MRI-compatible script using the
8607 @samp{-c} command-line option.
8609 Each command in an MRI-compatible script occupies its own line; each
8610 command line starts with the keyword that identifies the command (though
8611 blank lines are also allowed for punctuation). If a line of an
8612 MRI-compatible script begins with an unrecognized keyword, @command{ld}
8613 issues a warning message, but continues processing the script.
8615 Lines beginning with @samp{*} are comments.
8617 You can write these commands using all upper-case letters, or all
8618 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
8619 The following list shows only the upper-case form of each command.
8622 @cindex @code{ABSOLUTE} (MRI)
8623 @item ABSOLUTE @var{secname}
8624 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
8625 Normally, @command{ld} includes in the output file all sections from all
8626 the input files. However, in an MRI-compatible script, you can use the
8627 @code{ABSOLUTE} command to restrict the sections that will be present in
8628 your output program. If the @code{ABSOLUTE} command is used at all in a
8629 script, then only the sections named explicitly in @code{ABSOLUTE}
8630 commands will appear in the linker output. You can still use other
8631 input sections (whatever you select on the command line, or using
8632 @code{LOAD}) to resolve addresses in the output file.
8634 @cindex @code{ALIAS} (MRI)
8635 @item ALIAS @var{out-secname}, @var{in-secname}
8636 Use this command to place the data from input section @var{in-secname}
8637 in a section called @var{out-secname} in the linker output file.
8639 @var{in-secname} may be an integer.
8641 @cindex @code{ALIGN} (MRI)
8642 @item ALIGN @var{secname} = @var{expression}
8643 Align the section called @var{secname} to @var{expression}. The
8644 @var{expression} should be a power of two.
8646 @cindex @code{BASE} (MRI)
8647 @item BASE @var{expression}
8648 Use the value of @var{expression} as the lowest address (other than
8649 absolute addresses) in the output file.
8651 @cindex @code{CHIP} (MRI)
8652 @item CHIP @var{expression}
8653 @itemx CHIP @var{expression}, @var{expression}
8654 This command does nothing; it is accepted only for compatibility.
8656 @cindex @code{END} (MRI)
8658 This command does nothing whatever; it's only accepted for compatibility.
8660 @cindex @code{FORMAT} (MRI)
8661 @item FORMAT @var{output-format}
8662 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
8663 language, but restricted to one of these output formats:
8667 S-records, if @var{output-format} is @samp{S}
8670 IEEE, if @var{output-format} is @samp{IEEE}
8673 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
8677 @cindex @code{LIST} (MRI)
8678 @item LIST @var{anything}@dots{}
8679 Print (to the standard output file) a link map, as produced by the
8680 @command{ld} command-line option @samp{-M}.
8682 The keyword @code{LIST} may be followed by anything on the
8683 same line, with no change in its effect.
8685 @cindex @code{LOAD} (MRI)
8686 @item LOAD @var{filename}
8687 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
8688 Include one or more object file @var{filename} in the link; this has the
8689 same effect as specifying @var{filename} directly on the @command{ld}
8692 @cindex @code{NAME} (MRI)
8693 @item NAME @var{output-name}
8694 @var{output-name} is the name for the program produced by @command{ld}; the
8695 MRI-compatible command @code{NAME} is equivalent to the command-line
8696 option @samp{-o} or the general script language command @code{OUTPUT}.
8698 @cindex @code{ORDER} (MRI)
8699 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
8700 @itemx ORDER @var{secname} @var{secname} @var{secname}
8701 Normally, @command{ld} orders the sections in its output file in the
8702 order in which they first appear in the input files. In an MRI-compatible
8703 script, you can override this ordering with the @code{ORDER} command. The
8704 sections you list with @code{ORDER} will appear first in your output
8705 file, in the order specified.
8707 @cindex @code{PUBLIC} (MRI)
8708 @item PUBLIC @var{name}=@var{expression}
8709 @itemx PUBLIC @var{name},@var{expression}
8710 @itemx PUBLIC @var{name} @var{expression}
8711 Supply a value (@var{expression}) for external symbol
8712 @var{name} used in the linker input files.
8714 @cindex @code{SECT} (MRI)
8715 @item SECT @var{secname}, @var{expression}
8716 @itemx SECT @var{secname}=@var{expression}
8717 @itemx SECT @var{secname} @var{expression}
8718 You can use any of these three forms of the @code{SECT} command to
8719 specify the start address (@var{expression}) for section @var{secname}.
8720 If you have more than one @code{SECT} statement for the same
8721 @var{secname}, only the @emph{first} sets the start address.
8724 @node GNU Free Documentation License
8725 @appendix GNU Free Documentation License
8729 @unnumbered LD Index
8734 % I think something like @@colophon should be in texinfo. In the
8736 \long\def\colophon{\hbox to0pt{}\vfill
8737 \centerline{The body of this manual is set in}
8738 \centerline{\fontname\tenrm,}
8739 \centerline{with headings in {\bf\fontname\tenbf}}
8740 \centerline{and examples in {\tt\fontname\tentt}.}
8741 \centerline{{\it\fontname\tenit\/} and}
8742 \centerline{{\sl\fontname\tensl\/}}
8743 \centerline{are used for emphasis.}\vfill}
8745 % Blame: doc@@cygnus.com, 28mar91.