3 @c Copyright (C) 1991-2020 Free Software Foundation, Inc.
6 @include configdoc.texi
7 @c (configdoc.texi is generated by the Makefile)
13 @macro gcctabopt{body}
19 @c Configure for the generation of man pages
47 @dircategory Software development
49 * Ld: (ld). The GNU linker.
54 This file documents the @sc{gnu} linker LD
55 @ifset VERSION_PACKAGE
56 @value{VERSION_PACKAGE}
58 version @value{VERSION}.
60 Copyright @copyright{} 1991-2020 Free Software Foundation, Inc.
62 Permission is granted to copy, distribute and/or modify this document
63 under the terms of the GNU Free Documentation License, Version 1.3
64 or any later version published by the Free Software Foundation;
65 with no Invariant Sections, with no Front-Cover Texts, and with no
66 Back-Cover Texts. A copy of the license is included in the
67 section entitled ``GNU Free Documentation License''.
71 @setchapternewpage odd
72 @settitle The GNU linker
77 @ifset VERSION_PACKAGE
78 @subtitle @value{VERSION_PACKAGE}
80 @subtitle Version @value{VERSION}
81 @author Steve Chamberlain
82 @author Ian Lance Taylor
87 \hfill Red Hat Inc\par
88 \hfill nickc\@credhat.com, doc\@redhat.com\par
89 \hfill {\it The GNU linker}\par
90 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
92 \global\parindent=0pt % Steve likes it this way.
95 @vskip 0pt plus 1filll
96 @c man begin COPYRIGHT
97 Copyright @copyright{} 1991-2020 Free Software Foundation, Inc.
99 Permission is granted to copy, distribute and/or modify this document
100 under the terms of the GNU Free Documentation License, Version 1.3
101 or any later version published by the Free Software Foundation;
102 with no Invariant Sections, with no Front-Cover Texts, and with no
103 Back-Cover Texts. A copy of the license is included in the
104 section entitled ``GNU Free Documentation License''.
110 @c FIXME: Talk about importance of *order* of args, cmds to linker!
115 This file documents the @sc{gnu} linker ld
116 @ifset VERSION_PACKAGE
117 @value{VERSION_PACKAGE}
119 version @value{VERSION}.
121 This document is distributed under the terms of the GNU Free
122 Documentation License version 1.3. A copy of the license is included
123 in the section entitled ``GNU Free Documentation License''.
126 * Overview:: Overview
127 * Invocation:: Invocation
128 * Scripts:: Linker Scripts
130 * Machine Dependent:: Machine Dependent Features
134 * H8/300:: ld and the H8/300
137 * Renesas:: ld and other Renesas micros
140 * ARM:: ld and the ARM family
143 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
146 * HPPA ELF32:: ld and HPPA 32-bit ELF
149 * M68K:: ld and Motorola 68K family
152 * MIPS:: ld and MIPS family
155 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
158 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
161 * S/390 ELF:: ld and S/390 ELF Support
164 * SPU ELF:: ld and SPU ELF Support
167 * TI COFF:: ld and the TI COFF
170 * Win32:: ld and WIN32 (cygwin/mingw)
173 * Xtensa:: ld and Xtensa Processors
176 @ifclear SingleFormat
179 @c Following blank line required for remaining bug in makeinfo conds/menus
181 * Reporting Bugs:: Reporting Bugs
182 * MRI:: MRI Compatible Script Files
183 * GNU Free Documentation License:: GNU Free Documentation License
184 * LD Index:: LD Index
191 @cindex @sc{gnu} linker
192 @cindex what is this?
195 @c man begin SYNOPSIS
196 ld [@b{options}] @var{objfile} @dots{}
200 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
201 the Info entries for @file{binutils} and
206 @c man begin DESCRIPTION
208 @command{ld} combines a number of object and archive files, relocates
209 their data and ties up symbol references. Usually the last step in
210 compiling a program is to run @command{ld}.
212 @command{ld} accepts Linker Command Language files written in
213 a superset of AT&T's Link Editor Command Language syntax,
214 to provide explicit and total control over the linking process.
218 This man page does not describe the command language; see the
219 @command{ld} entry in @code{info} for full details on the command
220 language and on other aspects of the GNU linker.
223 @ifclear SingleFormat
224 This version of @command{ld} uses the general purpose BFD libraries
225 to operate on object files. This allows @command{ld} to read, combine, and
226 write object files in many different formats---for example, COFF or
227 @code{a.out}. Different formats may be linked together to produce any
228 available kind of object file. @xref{BFD}, for more information.
231 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
232 linkers in providing diagnostic information. Many linkers abandon
233 execution immediately upon encountering an error; whenever possible,
234 @command{ld} continues executing, allowing you to identify other errors
235 (or, in some cases, to get an output file in spite of the error).
242 @c man begin DESCRIPTION
244 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
245 and to be as compatible as possible with other linkers. As a result,
246 you have many choices to control its behavior.
252 * Options:: Command-line Options
253 * Environment:: Environment Variables
257 @section Command-line Options
265 The linker supports a plethora of command-line options, but in actual
266 practice few of them are used in any particular context.
267 @cindex standard Unix system
268 For instance, a frequent use of @command{ld} is to link standard Unix
269 object files on a standard, supported Unix system. On such a system, to
270 link a file @code{hello.o}:
273 ld -o @var{output} /lib/crt0.o hello.o -lc
276 This tells @command{ld} to produce a file called @var{output} as the
277 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
278 the library @code{libc.a}, which will come from the standard search
279 directories. (See the discussion of the @samp{-l} option below.)
281 Some of the command-line options to @command{ld} may be specified at any
282 point in the command line. However, options which refer to files, such
283 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
284 which the option appears in the command line, relative to the object
285 files and other file options. Repeating non-file options with a
286 different argument will either have no further effect, or override prior
287 occurrences (those further to the left on the command line) of that
288 option. Options which may be meaningfully specified more than once are
289 noted in the descriptions below.
292 Non-option arguments are object files or archives which are to be linked
293 together. They may follow, precede, or be mixed in with command-line
294 options, except that an object file argument may not be placed between
295 an option and its argument.
297 Usually the linker is invoked with at least one object file, but you can
298 specify other forms of binary input files using @samp{-l}, @samp{-R},
299 and the script command language. If @emph{no} binary input files at all
300 are specified, the linker does not produce any output, and issues the
301 message @samp{No input files}.
303 If the linker cannot recognize the format of an object file, it will
304 assume that it is a linker script. A script specified in this way
305 augments the main linker script used for the link (either the default
306 linker script or the one specified by using @samp{-T}). This feature
307 permits the linker to link against a file which appears to be an object
308 or an archive, but actually merely defines some symbol values, or uses
309 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
310 script in this way merely augments the main linker script, with the
311 extra commands placed after the main script; use the @samp{-T} option
312 to replace the default linker script entirely, but note the effect of
313 the @code{INSERT} command. @xref{Scripts}.
315 For options whose names are a single letter,
316 option arguments must either follow the option letter without intervening
317 whitespace, or be given as separate arguments immediately following the
318 option that requires them.
320 For options whose names are multiple letters, either one dash or two can
321 precede the option name; for example, @samp{-trace-symbol} and
322 @samp{--trace-symbol} are equivalent. Note---there is one exception to
323 this rule. Multiple letter options that start with a lower case 'o' can
324 only be preceded by two dashes. This is to reduce confusion with the
325 @samp{-o} option. So for example @samp{-omagic} sets the output file
326 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
329 Arguments to multiple-letter options must either be separated from the
330 option name by an equals sign, or be given as separate arguments
331 immediately following the option that requires them. For example,
332 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
333 Unique abbreviations of the names of multiple-letter options are
336 Note---if the linker is being invoked indirectly, via a compiler driver
337 (e.g. @samp{gcc}) then all the linker command-line options should be
338 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
339 compiler driver) like this:
342 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
345 This is important, because otherwise the compiler driver program may
346 silently drop the linker options, resulting in a bad link. Confusion
347 may also arise when passing options that require values through a
348 driver, as the use of a space between option and argument acts as
349 a separator, and causes the driver to pass only the option to the linker
350 and the argument to the compiler. In this case, it is simplest to use
351 the joined forms of both single- and multiple-letter options, such as:
354 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
357 Here is a table of the generic command-line switches accepted by the GNU
361 @include at-file.texi
363 @kindex -a @var{keyword}
364 @item -a @var{keyword}
365 This option is supported for HP/UX compatibility. The @var{keyword}
366 argument must be one of the strings @samp{archive}, @samp{shared}, or
367 @samp{default}. @samp{-aarchive} is functionally equivalent to
368 @samp{-Bstatic}, and the other two keywords are functionally equivalent
369 to @samp{-Bdynamic}. This option may be used any number of times.
371 @kindex --audit @var{AUDITLIB}
372 @item --audit @var{AUDITLIB}
373 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
374 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
375 specified in the library. If specified multiple times @code{DT_AUDIT}
376 will contain a colon separated list of audit interfaces to use. If the linker
377 finds an object with an audit entry while searching for shared libraries,
378 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
379 This option is only meaningful on ELF platforms supporting the rtld-audit
382 @ifclear SingleFormat
383 @cindex binary input format
384 @kindex -b @var{format}
385 @kindex --format=@var{format}
388 @item -b @var{input-format}
389 @itemx --format=@var{input-format}
390 @command{ld} may be configured to support more than one kind of object
391 file. If your @command{ld} is configured this way, you can use the
392 @samp{-b} option to specify the binary format for input object files
393 that follow this option on the command line. Even when @command{ld} is
394 configured to support alternative object formats, you don't usually need
395 to specify this, as @command{ld} should be configured to expect as a
396 default input format the most usual format on each machine.
397 @var{input-format} is a text string, the name of a particular format
398 supported by the BFD libraries. (You can list the available binary
399 formats with @samp{objdump -i}.)
402 You may want to use this option if you are linking files with an unusual
403 binary format. You can also use @samp{-b} to switch formats explicitly (when
404 linking object files of different formats), by including
405 @samp{-b @var{input-format}} before each group of object files in a
408 The default format is taken from the environment variable
413 You can also define the input format from a script, using the command
416 see @ref{Format Commands}.
420 @kindex -c @var{MRI-cmdfile}
421 @kindex --mri-script=@var{MRI-cmdfile}
422 @cindex compatibility, MRI
423 @item -c @var{MRI-commandfile}
424 @itemx --mri-script=@var{MRI-commandfile}
425 For compatibility with linkers produced by MRI, @command{ld} accepts script
426 files written in an alternate, restricted command language, described in
428 @ref{MRI,,MRI Compatible Script Files}.
431 the MRI Compatible Script Files section of GNU ld documentation.
433 Introduce MRI script files with
434 the option @samp{-c}; use the @samp{-T} option to run linker
435 scripts written in the general-purpose @command{ld} scripting language.
436 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
437 specified by any @samp{-L} options.
439 @cindex common allocation
446 These three options are equivalent; multiple forms are supported for
447 compatibility with other linkers. They assign space to common symbols
448 even if a relocatable output file is specified (with @samp{-r}). The
449 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
450 @xref{Miscellaneous Commands}.
452 @kindex --depaudit @var{AUDITLIB}
453 @kindex -P @var{AUDITLIB}
454 @item --depaudit @var{AUDITLIB}
455 @itemx -P @var{AUDITLIB}
456 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
457 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
458 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
459 will contain a colon separated list of audit interfaces to use. This
460 option is only meaningful on ELF platforms supporting the rtld-audit interface.
461 The -P option is provided for Solaris compatibility.
463 @kindex --enable-non-contiguous-regions
464 @item --enable-non-contiguous-regions
465 This option avoids generating an error if an input section does not
466 fit a matching output section. The linker tries to allocate the input
467 section to subseque nt matching output sections, and generates an
468 error only if no output section is large enough. This is useful when
469 several non-contiguous memory regions are available and the input
470 section does not require a particular one. The order in which input
471 sections are evaluated does not change, for instance:
475 MEM1 (rwx) : ORIGIN : 0x1000, LENGTH = 0x14
476 MEM2 (rwx) : ORIGIN : 0x1000, LENGTH = 0x40
477 MEM3 (rwx) : ORIGIN : 0x2000, LENGTH = 0x40
480 mem1 : @{ *(.data.*); @} > MEM1
481 mem2 : @{ *(.data.*); @} > MEM2
482 mem3 : @{ *(.data.*); @} > MEM2
490 results in .data.1 affected to mem1, and .data.2 and .data.3
491 affected to mem2, even though .data.3 would fit in mem3.
494 This option is incompatible with INSERT statements because it changes
495 the way input sections are mapped to output sections.
497 @kindex --enable-non-contiguous-regions-warnings
498 @item --enable-non-contiguous-regions-warnings
499 This option enables warnings when
500 @code{--enable-non-contiguous-regions} allows possibly unexpected
501 matches in sections mapping, potentially leading to silently
502 discarding a section instead of failing because it does not fit any
505 @cindex entry point, from command line
506 @kindex -e @var{entry}
507 @kindex --entry=@var{entry}
509 @itemx --entry=@var{entry}
510 Use @var{entry} as the explicit symbol for beginning execution of your
511 program, rather than the default entry point. If there is no symbol
512 named @var{entry}, the linker will try to parse @var{entry} as a number,
513 and use that as the entry address (the number will be interpreted in
514 base 10; you may use a leading @samp{0x} for base 16, or a leading
515 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
516 and other ways of specifying the entry point.
518 @kindex --exclude-libs
519 @item --exclude-libs @var{lib},@var{lib},...
520 Specifies a list of archive libraries from which symbols should not be automatically
521 exported. The library names may be delimited by commas or colons. Specifying
522 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
523 automatic export. This option is available only for the i386 PE targeted
524 port of the linker and for ELF targeted ports. For i386 PE, symbols
525 explicitly listed in a .def file are still exported, regardless of this
526 option. For ELF targeted ports, symbols affected by this option will
527 be treated as hidden.
529 @kindex --exclude-modules-for-implib
530 @item --exclude-modules-for-implib @var{module},@var{module},...
531 Specifies a list of object files or archive members, from which symbols
532 should not be automatically exported, but which should be copied wholesale
533 into the import library being generated during the link. The module names
534 may be delimited by commas or colons, and must match exactly the filenames
535 used by @command{ld} to open the files; for archive members, this is simply
536 the member name, but for object files the name listed must include and
537 match precisely any path used to specify the input file on the linker's
538 command-line. This option is available only for the i386 PE targeted port
539 of the linker. Symbols explicitly listed in a .def file are still exported,
540 regardless of this option.
542 @cindex dynamic symbol table
544 @kindex --export-dynamic
545 @kindex --no-export-dynamic
547 @itemx --export-dynamic
548 @itemx --no-export-dynamic
549 When creating a dynamically linked executable, using the @option{-E}
550 option or the @option{--export-dynamic} option causes the linker to add
551 all symbols to the dynamic symbol table. The dynamic symbol table is the
552 set of symbols which are visible from dynamic objects at run time.
554 If you do not use either of these options (or use the
555 @option{--no-export-dynamic} option to restore the default behavior), the
556 dynamic symbol table will normally contain only those symbols which are
557 referenced by some dynamic object mentioned in the link.
559 If you use @code{dlopen} to load a dynamic object which needs to refer
560 back to the symbols defined by the program, rather than some other
561 dynamic object, then you will probably need to use this option when
562 linking the program itself.
564 You can also use the dynamic list to control what symbols should
565 be added to the dynamic symbol table if the output format supports it.
566 See the description of @samp{--dynamic-list}.
568 Note that this option is specific to ELF targeted ports. PE targets
569 support a similar function to export all symbols from a DLL or EXE; see
570 the description of @samp{--export-all-symbols} below.
572 @ifclear SingleFormat
573 @cindex big-endian objects
577 Link big-endian objects. This affects the default output format.
579 @cindex little-endian objects
582 Link little-endian objects. This affects the default output format.
585 @kindex -f @var{name}
586 @kindex --auxiliary=@var{name}
588 @itemx --auxiliary=@var{name}
589 When creating an ELF shared object, set the internal DT_AUXILIARY field
590 to the specified name. This tells the dynamic linker that the symbol
591 table of the shared object should be used as an auxiliary filter on the
592 symbol table of the shared object @var{name}.
594 If you later link a program against this filter object, then, when you
595 run the program, the dynamic linker will see the DT_AUXILIARY field. If
596 the dynamic linker resolves any symbols from the filter object, it will
597 first check whether there is a definition in the shared object
598 @var{name}. If there is one, it will be used instead of the definition
599 in the filter object. The shared object @var{name} need not exist.
600 Thus the shared object @var{name} may be used to provide an alternative
601 implementation of certain functions, perhaps for debugging or for
602 machine-specific performance.
604 This option may be specified more than once. The DT_AUXILIARY entries
605 will be created in the order in which they appear on the command line.
607 @kindex -F @var{name}
608 @kindex --filter=@var{name}
610 @itemx --filter=@var{name}
611 When creating an ELF shared object, set the internal DT_FILTER field to
612 the specified name. This tells the dynamic linker that the symbol table
613 of the shared object which is being created should be used as a filter
614 on the symbol table of the shared object @var{name}.
616 If you later link a program against this filter object, then, when you
617 run the program, the dynamic linker will see the DT_FILTER field. The
618 dynamic linker will resolve symbols according to the symbol table of the
619 filter object as usual, but it will actually link to the definitions
620 found in the shared object @var{name}. Thus the filter object can be
621 used to select a subset of the symbols provided by the object
624 Some older linkers used the @option{-F} option throughout a compilation
625 toolchain for specifying object-file format for both input and output
627 @ifclear SingleFormat
628 The @sc{gnu} linker uses other mechanisms for this purpose: the
629 @option{-b}, @option{--format}, @option{--oformat} options, the
630 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
631 environment variable.
633 The @sc{gnu} linker will ignore the @option{-F} option when not
634 creating an ELF shared object.
636 @cindex finalization function
637 @kindex -fini=@var{name}
638 @item -fini=@var{name}
639 When creating an ELF executable or shared object, call NAME when the
640 executable or shared object is unloaded, by setting DT_FINI to the
641 address of the function. By default, the linker uses @code{_fini} as
642 the function to call.
646 Ignored. Provided for compatibility with other tools.
648 @kindex -G @var{value}
649 @kindex --gpsize=@var{value}
652 @itemx --gpsize=@var{value}
653 Set the maximum size of objects to be optimized using the GP register to
654 @var{size}. This is only meaningful for object file formats such as
655 MIPS ELF that support putting large and small objects into different
656 sections. This is ignored for other object file formats.
658 @cindex runtime library name
659 @kindex -h @var{name}
660 @kindex -soname=@var{name}
662 @itemx -soname=@var{name}
663 When creating an ELF shared object, set the internal DT_SONAME field to
664 the specified name. When an executable is linked with a shared object
665 which has a DT_SONAME field, then when the executable is run the dynamic
666 linker will attempt to load the shared object specified by the DT_SONAME
667 field rather than the using the file name given to the linker.
670 @cindex incremental link
672 Perform an incremental link (same as option @samp{-r}).
674 @cindex initialization function
675 @kindex -init=@var{name}
676 @item -init=@var{name}
677 When creating an ELF executable or shared object, call NAME when the
678 executable or shared object is loaded, by setting DT_INIT to the address
679 of the function. By default, the linker uses @code{_init} as the
682 @cindex archive files, from cmd line
683 @kindex -l @var{namespec}
684 @kindex --library=@var{namespec}
685 @item -l @var{namespec}
686 @itemx --library=@var{namespec}
687 Add the archive or object file specified by @var{namespec} to the
688 list of files to link. This option may be used any number of times.
689 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
690 will search the library path for a file called @var{filename}, otherwise it
691 will search the library path for a file called @file{lib@var{namespec}.a}.
693 On systems which support shared libraries, @command{ld} may also search for
694 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
695 and SunOS systems, @command{ld} will search a directory for a library
696 called @file{lib@var{namespec}.so} before searching for one called
697 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
698 indicates a shared library.) Note that this behavior does not apply
699 to @file{:@var{filename}}, which always specifies a file called
702 The linker will search an archive only once, at the location where it is
703 specified on the command line. If the archive defines a symbol which
704 was undefined in some object which appeared before the archive on the
705 command line, the linker will include the appropriate file(s) from the
706 archive. However, an undefined symbol in an object appearing later on
707 the command line will not cause the linker to search the archive again.
709 See the @option{-(} option for a way to force the linker to search
710 archives multiple times.
712 You may list the same archive multiple times on the command line.
715 This type of archive searching is standard for Unix linkers. However,
716 if you are using @command{ld} on AIX, note that it is different from the
717 behaviour of the AIX linker.
720 @cindex search directory, from cmd line
722 @kindex --library-path=@var{dir}
723 @item -L @var{searchdir}
724 @itemx --library-path=@var{searchdir}
725 Add path @var{searchdir} to the list of paths that @command{ld} will search
726 for archive libraries and @command{ld} control scripts. You may use this
727 option any number of times. The directories are searched in the order
728 in which they are specified on the command line. Directories specified
729 on the command line are searched before the default directories. All
730 @option{-L} options apply to all @option{-l} options, regardless of the
731 order in which the options appear. @option{-L} options do not affect
732 how @command{ld} searches for a linker script unless @option{-T}
735 If @var{searchdir} begins with @code{=} or @code{$SYSROOT}, then this
736 prefix will be replaced by the @dfn{sysroot prefix}, controlled by the
737 @samp{--sysroot} option, or specified when the linker is configured.
740 The default set of paths searched (without being specified with
741 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
742 some cases also on how it was configured. @xref{Environment}.
745 The paths can also be specified in a link script with the
746 @code{SEARCH_DIR} command. Directories specified this way are searched
747 at the point in which the linker script appears in the command line.
750 @kindex -m @var{emulation}
751 @item -m @var{emulation}
752 Emulate the @var{emulation} linker. You can list the available
753 emulations with the @samp{--verbose} or @samp{-V} options.
755 If the @samp{-m} option is not used, the emulation is taken from the
756 @code{LDEMULATION} environment variable, if that is defined.
758 Otherwise, the default emulation depends upon how the linker was
766 Print a link map to the standard output. A link map provides
767 information about the link, including the following:
771 Where object files are mapped into memory.
773 How common symbols are allocated.
775 All archive members included in the link, with a mention of the symbol
776 which caused the archive member to be brought in.
778 The values assigned to symbols.
780 Note - symbols whose values are computed by an expression which
781 involves a reference to a previous value of the same symbol may not
782 have correct result displayed in the link map. This is because the
783 linker discards intermediate results and only retains the final value
784 of an expression. Under such circumstances the linker will display
785 the final value enclosed by square brackets. Thus for example a
786 linker script containing:
794 will produce the following output in the link map if the @option{-M}
799 [0x0000000c] foo = (foo * 0x4)
800 [0x0000000c] foo = (foo + 0x8)
803 See @ref{Expressions} for more information about expressions in linker
807 How GNU properties are merged.
809 When the linker merges input .note.gnu.property sections into one output
810 .note.gnu.property section, some properties are removed or updated.
811 These actions are reported in the link map. For example:
814 Removed property 0xc0000002 to merge foo.o (0x1) and bar.o (not found)
817 This indicates that property 0xc0000002 is removed from output when
818 merging properties in @file{foo.o}, whose property 0xc0000002 value
819 is 0x1, and @file{bar.o}, which doesn't have property 0xc0000002.
822 Updated property 0xc0010001 (0x1) to merge foo.o (0x1) and bar.o (0x1)
825 This indicates that property 0xc0010001 value is updated to 0x1 in output
826 when merging properties in @file{foo.o}, whose 0xc0010001 property value
827 is 0x1, and @file{bar.o}, whose 0xc0010001 property value is 0x1.
830 @cindex link map discarded
831 @kindex --print-map-discarded
832 @kindex --no-print-map-discarded
833 @item --print-map-discarded
834 @itemx --no-print-map-discarded
835 Print (or do not print) the list of discarded and garbage collected sections
836 in the link map. Enabled by default.
839 @cindex read-only text
844 Turn off page alignment of sections, and disable linking against shared
845 libraries. If the output format supports Unix style magic numbers,
846 mark the output as @code{NMAGIC}.
850 @cindex read/write from cmd line
854 Set the text and data sections to be readable and writable. Also, do
855 not page-align the data segment, and disable linking against shared
856 libraries. If the output format supports Unix style magic numbers,
857 mark the output as @code{OMAGIC}. Note: Although a writable text section
858 is allowed for PE-COFF targets, it does not conform to the format
859 specification published by Microsoft.
864 This option negates most of the effects of the @option{-N} option. It
865 sets the text section to be read-only, and forces the data segment to
866 be page-aligned. Note - this option does not enable linking against
867 shared libraries. Use @option{-Bdynamic} for this.
869 @kindex -o @var{output}
870 @kindex --output=@var{output}
871 @cindex naming the output file
872 @item -o @var{output}
873 @itemx --output=@var{output}
874 Use @var{output} as the name for the program produced by @command{ld}; if this
875 option is not specified, the name @file{a.out} is used by default. The
876 script command @code{OUTPUT} can also specify the output file name.
878 @kindex -O @var{level}
879 @cindex generating optimized output
881 If @var{level} is a numeric values greater than zero @command{ld} optimizes
882 the output. This might take significantly longer and therefore probably
883 should only be enabled for the final binary. At the moment this
884 option only affects ELF shared library generation. Future releases of
885 the linker may make more use of this option. Also currently there is
886 no difference in the linker's behaviour for different non-zero values
887 of this option. Again this may change with future releases.
889 @kindex -plugin @var{name}
890 @item -plugin @var{name}
891 Involve a plugin in the linking process. The @var{name} parameter is
892 the absolute filename of the plugin. Usually this parameter is
893 automatically added by the complier, when using link time
894 optimization, but users can also add their own plugins if they so
897 Note that the location of the compiler originated plugins is different
898 from the place where the @command{ar}, @command{nm} and
899 @command{ranlib} programs search for their plugins. In order for
900 those commands to make use of a compiler based plugin it must first be
901 copied into the @file{$@{libdir@}/bfd-plugins} directory. All gcc
902 based linker plugins are backward compatible, so it is sufficient to
903 just copy in the newest one.
906 @cindex push state governing input file handling
908 The @option{--push-state} allows to preserve the current state of the
909 flags which govern the input file handling so that they can all be
910 restored with one corresponding @option{--pop-state} option.
912 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
913 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
914 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
915 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
916 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
917 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
919 One target for this option are specifications for @file{pkg-config}. When
920 used with the @option{--libs} option all possibly needed libraries are
921 listed and then possibly linked with all the time. It is better to return
922 something as follows:
925 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
929 @cindex pop state governing input file handling
931 Undoes the effect of --push-state, restores the previous values of the
932 flags governing input file handling.
935 @kindex --emit-relocs
936 @cindex retain relocations in final executable
939 Leave relocation sections and contents in fully linked executables.
940 Post link analysis and optimization tools may need this information in
941 order to perform correct modifications of executables. This results
942 in larger executables.
944 This option is currently only supported on ELF platforms.
946 @kindex --force-dynamic
947 @cindex forcing the creation of dynamic sections
948 @item --force-dynamic
949 Force the output file to have dynamic sections. This option is specific
953 @cindex relocatable output
955 @kindex --relocatable
958 Generate relocatable output---i.e., generate an output file that can in
959 turn serve as input to @command{ld}. This is often called @dfn{partial
960 linking}. As a side effect, in environments that support standard Unix
961 magic numbers, this option also sets the output file's magic number to
963 @c ; see @option{-N}.
964 If this option is not specified, an absolute file is produced. When
965 linking C++ programs, this option @emph{will not} resolve references to
966 constructors; to do that, use @samp{-Ur}.
968 When an input file does not have the same format as the output file,
969 partial linking is only supported if that input file does not contain any
970 relocations. Different output formats can have further restrictions; for
971 example some @code{a.out}-based formats do not support partial linking
972 with input files in other formats at all.
974 This option does the same thing as @samp{-i}.
976 @kindex -R @var{file}
977 @kindex --just-symbols=@var{file}
978 @cindex symbol-only input
979 @item -R @var{filename}
980 @itemx --just-symbols=@var{filename}
981 Read symbol names and their addresses from @var{filename}, but do not
982 relocate it or include it in the output. This allows your output file
983 to refer symbolically to absolute locations of memory defined in other
984 programs. You may use this option more than once.
986 For compatibility with other ELF linkers, if the @option{-R} option is
987 followed by a directory name, rather than a file name, it is treated as
988 the @option{-rpath} option.
992 @cindex strip all symbols
995 Omit all symbol information from the output file.
998 @kindex --strip-debug
999 @cindex strip debugger symbols
1001 @itemx --strip-debug
1002 Omit debugger symbol information (but not all symbols) from the output file.
1004 @kindex --strip-discarded
1005 @kindex --no-strip-discarded
1006 @item --strip-discarded
1007 @itemx --no-strip-discarded
1008 Omit (or do not omit) global symbols defined in discarded sections.
1013 @cindex input files, displaying
1016 Print the names of the input files as @command{ld} processes them. If
1017 @samp{-t} is given twice then members within archives are also printed.
1018 @samp{-t} output is useful to generate a list of all the object files
1019 and scripts involved in linking, for example, when packaging files for
1020 a linker bug report.
1022 @kindex -T @var{script}
1023 @kindex --script=@var{script}
1024 @cindex script files
1025 @item -T @var{scriptfile}
1026 @itemx --script=@var{scriptfile}
1027 Use @var{scriptfile} as the linker script. This script replaces
1028 @command{ld}'s default linker script (rather than adding to it), so
1029 @var{commandfile} must specify everything necessary to describe the
1030 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
1031 the current directory, @code{ld} looks for it in the directories
1032 specified by any preceding @samp{-L} options. Multiple @samp{-T}
1035 @kindex -dT @var{script}
1036 @kindex --default-script=@var{script}
1037 @cindex script files
1038 @item -dT @var{scriptfile}
1039 @itemx --default-script=@var{scriptfile}
1040 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
1042 This option is similar to the @option{--script} option except that
1043 processing of the script is delayed until after the rest of the
1044 command line has been processed. This allows options placed after the
1045 @option{--default-script} option on the command line to affect the
1046 behaviour of the linker script, which can be important when the linker
1047 command line cannot be directly controlled by the user. (eg because
1048 the command line is being constructed by another tool, such as
1051 @kindex -u @var{symbol}
1052 @kindex --undefined=@var{symbol}
1053 @cindex undefined symbol
1054 @item -u @var{symbol}
1055 @itemx --undefined=@var{symbol}
1056 Force @var{symbol} to be entered in the output file as an undefined
1057 symbol. Doing this may, for example, trigger linking of additional
1058 modules from standard libraries. @samp{-u} may be repeated with
1059 different option arguments to enter additional undefined symbols. This
1060 option is equivalent to the @code{EXTERN} linker script command.
1062 If this option is being used to force additional modules to be pulled
1063 into the link, and if it is an error for the symbol to remain
1064 undefined, then the option @option{--require-defined} should be used
1067 @kindex --require-defined=@var{symbol}
1068 @cindex symbols, require defined
1069 @cindex defined symbol
1070 @item --require-defined=@var{symbol}
1071 Require that @var{symbol} is defined in the output file. This option
1072 is the same as option @option{--undefined} except that if @var{symbol}
1073 is not defined in the output file then the linker will issue an error
1074 and exit. The same effect can be achieved in a linker script by using
1075 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
1076 can be used multiple times to require additional symbols.
1079 @cindex constructors
1081 For anything other than C++ programs, this option is equivalent to
1082 @samp{-r}: it generates relocatable output---i.e., an output file that can in
1083 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
1084 @emph{does} resolve references to constructors, unlike @samp{-r}.
1085 It does not work to use @samp{-Ur} on files that were themselves linked
1086 with @samp{-Ur}; once the constructor table has been built, it cannot
1087 be added to. Use @samp{-Ur} only for the last partial link, and
1088 @samp{-r} for the others.
1090 @kindex --orphan-handling=@var{MODE}
1091 @cindex orphan sections
1092 @cindex sections, orphan
1093 @item --orphan-handling=@var{MODE}
1094 Control how orphan sections are handled. An orphan section is one not
1095 specifically mentioned in a linker script. @xref{Orphan Sections}.
1097 @var{MODE} can have any of the following values:
1101 Orphan sections are placed into a suitable output section following
1102 the strategy described in @ref{Orphan Sections}. The option
1103 @samp{--unique} also affects how sections are placed.
1106 All orphan sections are discarded, by placing them in the
1107 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1110 The linker will place the orphan section as for @code{place} and also
1114 The linker will exit with an error if any orphan section is found.
1117 The default if @samp{--orphan-handling} is not given is @code{place}.
1119 @kindex --unique[=@var{SECTION}]
1120 @item --unique[=@var{SECTION}]
1121 Creates a separate output section for every input section matching
1122 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1123 missing, for every orphan input section. An orphan section is one not
1124 specifically mentioned in a linker script. You may use this option
1125 multiple times on the command line; It prevents the normal merging of
1126 input sections with the same name, overriding output section assignments
1136 Display the version number for @command{ld}. The @option{-V} option also
1137 lists the supported emulations.
1140 @kindex --discard-all
1141 @cindex deleting local symbols
1143 @itemx --discard-all
1144 Delete all local symbols.
1147 @kindex --discard-locals
1148 @cindex local symbols, deleting
1150 @itemx --discard-locals
1151 Delete all temporary local symbols. (These symbols start with
1152 system-specific local label prefixes, typically @samp{.L} for ELF systems
1153 or @samp{L} for traditional a.out systems.)
1155 @kindex -y @var{symbol}
1156 @kindex --trace-symbol=@var{symbol}
1157 @cindex symbol tracing
1158 @item -y @var{symbol}
1159 @itemx --trace-symbol=@var{symbol}
1160 Print the name of each linked file in which @var{symbol} appears. This
1161 option may be given any number of times. On many systems it is necessary
1162 to prepend an underscore.
1164 This option is useful when you have an undefined symbol in your link but
1165 don't know where the reference is coming from.
1167 @kindex -Y @var{path}
1169 Add @var{path} to the default library search path. This option exists
1170 for Solaris compatibility.
1172 @kindex -z @var{keyword}
1173 @item -z @var{keyword}
1174 The recognized keywords are:
1178 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1180 @item call-nop=prefix-addr
1181 @itemx call-nop=suffix-nop
1182 @itemx call-nop=prefix-@var{byte}
1183 @itemx call-nop=suffix-@var{byte}
1184 Specify the 1-byte @code{NOP} padding when transforming indirect call
1185 to a locally defined function, foo, via its GOT slot.
1186 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1187 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1188 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1189 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1190 Supported for i386 and x86_64.
1192 @item cet-report=none
1193 @itemx cet-report=warning
1194 @itemx cet-report=error
1195 Specify how to report the missing GNU_PROPERTY_X86_FEATURE_1_IBT and
1196 GNU_PROPERTY_X86_FEATURE_1_SHSTK properties in input .note.gnu.property
1197 section. @option{cet-report=none}, which is the default, will make the
1198 linker not report missing properties in input files.
1199 @option{cet-report=warning} will make the linker issue a warning for
1200 missing properties in input files. @option{cet-report=error} will make
1201 the linker issue an error for missing properties in input files.
1202 Note that @option{ibt} will turn off the missing
1203 GNU_PROPERTY_X86_FEATURE_1_IBT property report and @option{shstk} will
1204 turn off the missing GNU_PROPERTY_X86_FEATURE_1_SHSTK property report.
1205 Supported for Linux/i386 and Linux/x86_64.
1209 Combine multiple dynamic relocation sections and sort to improve
1210 dynamic symbol lookup caching. Do not do this if @samp{nocombreloc}.
1214 Generate common symbols with STT_COMMON type during a relocatable
1215 link. Use STT_OBJECT type if @samp{nocommon}.
1217 @item common-page-size=@var{value}
1218 Set the page size most commonly used to @var{value}. Memory image
1219 layout will be optimized to minimize memory pages if the system is
1220 using pages of this size.
1223 Report unresolved symbol references from regular object files. This
1224 is done even if the linker is creating a non-symbolic shared library.
1225 This option is the inverse of @samp{-z undefs}.
1227 @item dynamic-undefined-weak
1228 @itemx nodynamic-undefined-weak
1229 Make undefined weak symbols dynamic when building a dynamic object,
1230 if they are referenced from a regular object file and not forced local
1231 by symbol visibility or versioning. Do not make them dynamic if
1232 @samp{nodynamic-undefined-weak}. If neither option is given, a target
1233 may default to either option being in force, or make some other
1234 selection of undefined weak symbols dynamic. Not all targets support
1238 Marks the object as requiring executable stack.
1241 This option is only meaningful when building a shared object. It makes
1242 the symbols defined by this shared object available for symbol resolution
1243 of subsequently loaded libraries.
1246 This option is only meaningful when building a dynamic executable.
1247 This option marks the executable as requiring global auditing by
1248 setting the @code{DF_1_GLOBAUDIT} bit in the @code{DT_FLAGS_1} dynamic
1249 tag. Global auditing requires that any auditing library defined via
1250 the @option{--depaudit} or @option{-P} command-line options be run for
1251 all dynamic objects loaded by the application.
1254 Generate Intel Indirect Branch Tracking (IBT) enabled PLT entries.
1255 Supported for Linux/i386 and Linux/x86_64.
1258 Generate GNU_PROPERTY_X86_FEATURE_1_IBT in .note.gnu.property section
1259 to indicate compatibility with IBT. This also implies @option{ibtplt}.
1260 Supported for Linux/i386 and Linux/x86_64.
1263 This option is only meaningful when building a shared object.
1264 It marks the object so that its runtime initialization will occur
1265 before the runtime initialization of any other objects brought into
1266 the process at the same time. Similarly the runtime finalization of
1267 the object will occur after the runtime finalization of any other
1271 Specify that the dynamic loader should modify its symbol search order
1272 so that symbols in this shared library interpose all other shared
1273 libraries not so marked.
1276 When generating an executable or shared library, mark it to tell the
1277 dynamic linker to defer function call resolution to the point when
1278 the function is called (lazy binding), rather than at load time.
1279 Lazy binding is the default.
1282 Specify that the object's filters be processed immediately at runtime.
1284 @item max-page-size=@var{value}
1285 Set the maximum memory page size supported to @var{value}.
1288 Allow multiple definitions.
1291 Disable linker generated .dynbss variables used in place of variables
1292 defined in shared libraries. May result in dynamic text relocations.
1295 Specify that the dynamic loader search for dependencies of this object
1296 should ignore any default library search paths.
1299 Specify that the object shouldn't be unloaded at runtime.
1302 Specify that the object is not available to @code{dlopen}.
1305 Specify that the object can not be dumped by @code{dldump}.
1308 Marks the object as not requiring executable stack.
1310 @item noextern-protected-data
1311 Don't treat protected data symbols as external when building a shared
1312 library. This option overrides the linker backend default. It can be
1313 used to work around incorrect relocations against protected data symbols
1314 generated by compiler. Updates on protected data symbols by another
1315 module aren't visible to the resulting shared library. Supported for
1318 @item noreloc-overflow
1319 Disable relocation overflow check. This can be used to disable
1320 relocation overflow check if there will be no dynamic relocation
1321 overflow at run-time. Supported for x86_64.
1324 When generating an executable or shared library, mark it to tell the
1325 dynamic linker to resolve all symbols when the program is started, or
1326 when the shared library is loaded by dlopen, instead of deferring
1327 function call resolution to the point when the function is first
1331 Specify that the object requires @samp{$ORIGIN} handling in paths.
1335 Create an ELF @code{PT_GNU_RELRO} segment header in the object. This
1336 specifies a memory segment that should be made read-only after
1337 relocation, if supported. Specifying @samp{common-page-size} smaller
1338 than the system page size will render this protection ineffective.
1339 Don't create an ELF @code{PT_GNU_RELRO} segment if @samp{norelro}.
1342 @itemx noseparate-code
1343 Create separate code @code{PT_LOAD} segment header in the object. This
1344 specifies a memory segment that should contain only instructions and must
1345 be in wholly disjoint pages from any other data. Don't create separate
1346 code @code{PT_LOAD} segment if @samp{noseparate-code} is used.
1349 Generate GNU_PROPERTY_X86_FEATURE_1_SHSTK in .note.gnu.property section
1350 to indicate compatibility with Intel Shadow Stack. Supported for
1351 Linux/i386 and Linux/x86_64.
1353 @item stack-size=@var{value}
1354 Specify a stack size for an ELF @code{PT_GNU_STACK} segment.
1355 Specifying zero will override any default non-zero sized
1356 @code{PT_GNU_STACK} segment creation.
1361 Report an error if DT_TEXTREL is set, i.e., if the binary has dynamic
1362 relocations in read-only sections. Don't report an error if
1363 @samp{notext} or @samp{textoff}.
1366 Do not report unresolved symbol references from regular object files,
1367 either when creating an executable, or when creating a shared library.
1368 This option is the inverse of @samp{-z defs}.
1372 Other keywords are ignored for Solaris compatibility.
1375 @cindex groups of archives
1376 @item -( @var{archives} -)
1377 @itemx --start-group @var{archives} --end-group
1378 The @var{archives} should be a list of archive files. They may be
1379 either explicit file names, or @samp{-l} options.
1381 The specified archives are searched repeatedly until no new undefined
1382 references are created. Normally, an archive is searched only once in
1383 the order that it is specified on the command line. If a symbol in that
1384 archive is needed to resolve an undefined symbol referred to by an
1385 object in an archive that appears later on the command line, the linker
1386 would not be able to resolve that reference. By grouping the archives,
1387 they will all be searched repeatedly until all possible references are
1390 Using this option has a significant performance cost. It is best to use
1391 it only when there are unavoidable circular references between two or
1394 @kindex --accept-unknown-input-arch
1395 @kindex --no-accept-unknown-input-arch
1396 @item --accept-unknown-input-arch
1397 @itemx --no-accept-unknown-input-arch
1398 Tells the linker to accept input files whose architecture cannot be
1399 recognised. The assumption is that the user knows what they are doing
1400 and deliberately wants to link in these unknown input files. This was
1401 the default behaviour of the linker, before release 2.14. The default
1402 behaviour from release 2.14 onwards is to reject such input files, and
1403 so the @samp{--accept-unknown-input-arch} option has been added to
1404 restore the old behaviour.
1407 @kindex --no-as-needed
1409 @itemx --no-as-needed
1410 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1411 on the command line after the @option{--as-needed} option. Normally
1412 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1413 on the command line, regardless of whether the library is actually
1414 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1415 emitted for a library that @emph{at that point in the link} satisfies a
1416 non-weak undefined symbol reference from a regular object file or, if
1417 the library is not found in the DT_NEEDED lists of other needed libraries, a
1418 non-weak undefined symbol reference from another needed dynamic library.
1419 Object files or libraries appearing on the command line @emph{after}
1420 the library in question do not affect whether the library is seen as
1421 needed. This is similar to the rules for extraction of object files
1422 from archives. @option{--no-as-needed} restores the default behaviour.
1424 @kindex --add-needed
1425 @kindex --no-add-needed
1427 @itemx --no-add-needed
1428 These two options have been deprecated because of the similarity of
1429 their names to the @option{--as-needed} and @option{--no-as-needed}
1430 options. They have been replaced by @option{--copy-dt-needed-entries}
1431 and @option{--no-copy-dt-needed-entries}.
1433 @kindex -assert @var{keyword}
1434 @item -assert @var{keyword}
1435 This option is ignored for SunOS compatibility.
1439 @kindex -call_shared
1443 Link against dynamic libraries. This is only meaningful on platforms
1444 for which shared libraries are supported. This option is normally the
1445 default on such platforms. The different variants of this option are
1446 for compatibility with various systems. You may use this option
1447 multiple times on the command line: it affects library searching for
1448 @option{-l} options which follow it.
1452 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1453 section. This causes the runtime linker to handle lookups in this
1454 object and its dependencies to be performed only inside the group.
1455 @option{--unresolved-symbols=report-all} is implied. This option is
1456 only meaningful on ELF platforms which support shared libraries.
1466 Do not link against shared libraries. This is only meaningful on
1467 platforms for which shared libraries are supported. The different
1468 variants of this option are for compatibility with various systems. You
1469 may use this option multiple times on the command line: it affects
1470 library searching for @option{-l} options which follow it. This
1471 option also implies @option{--unresolved-symbols=report-all}. This
1472 option can be used with @option{-shared}. Doing so means that a
1473 shared library is being created but that all of the library's external
1474 references must be resolved by pulling in entries from static
1479 When creating a shared library, bind references to global symbols to the
1480 definition within the shared library, if any. Normally, it is possible
1481 for a program linked against a shared library to override the definition
1482 within the shared library. This option is only meaningful on ELF
1483 platforms which support shared libraries.
1485 @kindex -Bsymbolic-functions
1486 @item -Bsymbolic-functions
1487 When creating a shared library, bind references to global function
1488 symbols to the definition within the shared library, if any.
1489 This option is only meaningful on ELF platforms which support shared
1492 @kindex --dynamic-list=@var{dynamic-list-file}
1493 @item --dynamic-list=@var{dynamic-list-file}
1494 Specify the name of a dynamic list file to the linker. This is
1495 typically used when creating shared libraries to specify a list of
1496 global symbols whose references shouldn't be bound to the definition
1497 within the shared library, or creating dynamically linked executables
1498 to specify a list of symbols which should be added to the symbol table
1499 in the executable. This option is only meaningful on ELF platforms
1500 which support shared libraries.
1502 The format of the dynamic list is the same as the version node without
1503 scope and node name. See @ref{VERSION} for more information.
1505 @kindex --dynamic-list-data
1506 @item --dynamic-list-data
1507 Include all global data symbols to the dynamic list.
1509 @kindex --dynamic-list-cpp-new
1510 @item --dynamic-list-cpp-new
1511 Provide the builtin dynamic list for C++ operator new and delete. It
1512 is mainly useful for building shared libstdc++.
1514 @kindex --dynamic-list-cpp-typeinfo
1515 @item --dynamic-list-cpp-typeinfo
1516 Provide the builtin dynamic list for C++ runtime type identification.
1518 @kindex --check-sections
1519 @kindex --no-check-sections
1520 @item --check-sections
1521 @itemx --no-check-sections
1522 Asks the linker @emph{not} to check section addresses after they have
1523 been assigned to see if there are any overlaps. Normally the linker will
1524 perform this check, and if it finds any overlaps it will produce
1525 suitable error messages. The linker does know about, and does make
1526 allowances for sections in overlays. The default behaviour can be
1527 restored by using the command-line switch @option{--check-sections}.
1528 Section overlap is not usually checked for relocatable links. You can
1529 force checking in that case by using the @option{--check-sections}
1532 @kindex --copy-dt-needed-entries
1533 @kindex --no-copy-dt-needed-entries
1534 @item --copy-dt-needed-entries
1535 @itemx --no-copy-dt-needed-entries
1536 This option affects the treatment of dynamic libraries referred to
1537 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1538 command line. Normally the linker won't add a DT_NEEDED tag to the
1539 output binary for each library mentioned in a DT_NEEDED tag in an
1540 input dynamic library. With @option{--copy-dt-needed-entries}
1541 specified on the command line however any dynamic libraries that
1542 follow it will have their DT_NEEDED entries added. The default
1543 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1545 This option also has an effect on the resolution of symbols in dynamic
1546 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1547 mentioned on the command line will be recursively searched, following
1548 their DT_NEEDED tags to other libraries, in order to resolve symbols
1549 required by the output binary. With the default setting however
1550 the searching of dynamic libraries that follow it will stop with the
1551 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1554 @cindex cross reference table
1557 Output a cross reference table. If a linker map file is being
1558 generated, the cross reference table is printed to the map file.
1559 Otherwise, it is printed on the standard output.
1561 The format of the table is intentionally simple, so that it may be
1562 easily processed by a script if necessary. The symbols are printed out,
1563 sorted by name. For each symbol, a list of file names is given. If the
1564 symbol is defined, the first file listed is the location of the
1565 definition. If the symbol is defined as a common value then any files
1566 where this happens appear next. Finally any files that reference the
1569 @cindex common allocation
1570 @kindex --no-define-common
1571 @item --no-define-common
1572 This option inhibits the assignment of addresses to common symbols.
1573 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1574 @xref{Miscellaneous Commands}.
1576 The @samp{--no-define-common} option allows decoupling
1577 the decision to assign addresses to Common symbols from the choice
1578 of the output file type; otherwise a non-Relocatable output type
1579 forces assigning addresses to Common symbols.
1580 Using @samp{--no-define-common} allows Common symbols that are referenced
1581 from a shared library to be assigned addresses only in the main program.
1582 This eliminates the unused duplicate space in the shared library,
1583 and also prevents any possible confusion over resolving to the wrong
1584 duplicate when there are many dynamic modules with specialized search
1585 paths for runtime symbol resolution.
1587 @cindex group allocation in linker script
1588 @cindex section groups
1590 @kindex --force-group-allocation
1591 @item --force-group-allocation
1592 This option causes the linker to place section group members like
1593 normal input sections, and to delete the section groups. This is the
1594 default behaviour for a final link but this option can be used to
1595 change the behaviour of a relocatable link (@samp{-r}). The script
1596 command @code{FORCE_GROUP_ALLOCATION} has the same
1597 effect. @xref{Miscellaneous Commands}.
1599 @cindex symbols, from command line
1600 @kindex --defsym=@var{symbol}=@var{exp}
1601 @item --defsym=@var{symbol}=@var{expression}
1602 Create a global symbol in the output file, containing the absolute
1603 address given by @var{expression}. You may use this option as many
1604 times as necessary to define multiple symbols in the command line. A
1605 limited form of arithmetic is supported for the @var{expression} in this
1606 context: you may give a hexadecimal constant or the name of an existing
1607 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1608 constants or symbols. If you need more elaborate expressions, consider
1609 using the linker command language from a script (@pxref{Assignments}).
1610 @emph{Note:} there should be no white space between @var{symbol}, the
1611 equals sign (``@key{=}''), and @var{expression}.
1613 @cindex demangling, from command line
1614 @kindex --demangle[=@var{style}]
1615 @kindex --no-demangle
1616 @item --demangle[=@var{style}]
1617 @itemx --no-demangle
1618 These options control whether to demangle symbol names in error messages
1619 and other output. When the linker is told to demangle, it tries to
1620 present symbol names in a readable fashion: it strips leading
1621 underscores if they are used by the object file format, and converts C++
1622 mangled symbol names into user readable names. Different compilers have
1623 different mangling styles. The optional demangling style argument can be used
1624 to choose an appropriate demangling style for your compiler. The linker will
1625 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1626 is set. These options may be used to override the default.
1628 @cindex dynamic linker, from command line
1629 @kindex -I@var{file}
1630 @kindex --dynamic-linker=@var{file}
1632 @itemx --dynamic-linker=@var{file}
1633 Set the name of the dynamic linker. This is only meaningful when
1634 generating dynamically linked ELF executables. The default dynamic
1635 linker is normally correct; don't use this unless you know what you are
1638 @kindex --no-dynamic-linker
1639 @item --no-dynamic-linker
1640 When producing an executable file, omit the request for a dynamic
1641 linker to be used at load-time. This is only meaningful for ELF
1642 executables that contain dynamic relocations, and usually requires
1643 entry point code that is capable of processing these relocations.
1645 @kindex --embedded-relocs
1646 @item --embedded-relocs
1647 This option is similar to the @option{--emit-relocs} option except
1648 that the relocs are stored in a target-specific section. This option
1649 is only supported by the @samp{BFIN}, @samp{CR16} and @emph{M68K}
1652 @kindex --disable-multiple-abs-defs
1653 @item --disable-multiple-abs-defs
1654 Do not allow multiple definitions with symbols included
1655 in filename invoked by -R or --just-symbols
1657 @kindex --fatal-warnings
1658 @kindex --no-fatal-warnings
1659 @item --fatal-warnings
1660 @itemx --no-fatal-warnings
1661 Treat all warnings as errors. The default behaviour can be restored
1662 with the option @option{--no-fatal-warnings}.
1664 @kindex --force-exe-suffix
1665 @item --force-exe-suffix
1666 Make sure that an output file has a .exe suffix.
1668 If a successfully built fully linked output file does not have a
1669 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1670 the output file to one of the same name with a @code{.exe} suffix. This
1671 option is useful when using unmodified Unix makefiles on a Microsoft
1672 Windows host, since some versions of Windows won't run an image unless
1673 it ends in a @code{.exe} suffix.
1675 @kindex --gc-sections
1676 @kindex --no-gc-sections
1677 @cindex garbage collection
1679 @itemx --no-gc-sections
1680 Enable garbage collection of unused input sections. It is ignored on
1681 targets that do not support this option. The default behaviour (of not
1682 performing this garbage collection) can be restored by specifying
1683 @samp{--no-gc-sections} on the command line. Note that garbage
1684 collection for COFF and PE format targets is supported, but the
1685 implementation is currently considered to be experimental.
1687 @samp{--gc-sections} decides which input sections are used by
1688 examining symbols and relocations. The section containing the entry
1689 symbol and all sections containing symbols undefined on the
1690 command-line will be kept, as will sections containing symbols
1691 referenced by dynamic objects. Note that when building shared
1692 libraries, the linker must assume that any visible symbol is
1693 referenced. Once this initial set of sections has been determined,
1694 the linker recursively marks as used any section referenced by their
1695 relocations. See @samp{--entry}, @samp{--undefined}, and
1696 @samp{--gc-keep-exported}.
1698 This option can be set when doing a partial link (enabled with option
1699 @samp{-r}). In this case the root of symbols kept must be explicitly
1700 specified either by one of the options @samp{--entry},
1701 @samp{--undefined}, or @samp{--gc-keep-exported} or by a @code{ENTRY}
1702 command in the linker script.
1704 @kindex --print-gc-sections
1705 @kindex --no-print-gc-sections
1706 @cindex garbage collection
1707 @item --print-gc-sections
1708 @itemx --no-print-gc-sections
1709 List all sections removed by garbage collection. The listing is
1710 printed on stderr. This option is only effective if garbage
1711 collection has been enabled via the @samp{--gc-sections}) option. The
1712 default behaviour (of not listing the sections that are removed) can
1713 be restored by specifying @samp{--no-print-gc-sections} on the command
1716 @kindex --gc-keep-exported
1717 @cindex garbage collection
1718 @item --gc-keep-exported
1719 When @samp{--gc-sections} is enabled, this option prevents garbage
1720 collection of unused input sections that contain global symbols having
1721 default or protected visibility. This option is intended to be used for
1722 executables where unreferenced sections would otherwise be garbage
1723 collected regardless of the external visibility of contained symbols.
1724 Note that this option has no effect when linking shared objects since
1725 it is already the default behaviour. This option is only supported for
1728 @kindex --print-output-format
1729 @cindex output format
1730 @item --print-output-format
1731 Print the name of the default output format (perhaps influenced by
1732 other command-line options). This is the string that would appear
1733 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1735 @kindex --print-memory-usage
1736 @cindex memory usage
1737 @item --print-memory-usage
1738 Print used size, total size and used size of memory regions created with
1739 the @ref{MEMORY} command. This is useful on embedded targets to have a
1740 quick view of amount of free memory. The format of the output has one
1741 headline and one line per region. It is both human readable and easily
1742 parsable by tools. Here is an example of an output:
1745 Memory region Used Size Region Size %age Used
1746 ROM: 256 KB 1 MB 25.00%
1747 RAM: 32 B 2 GB 0.00%
1754 Print a summary of the command-line options on the standard output and exit.
1756 @kindex --target-help
1758 Print a summary of all target-specific options on the standard output and exit.
1760 @kindex -Map=@var{mapfile}
1761 @item -Map=@var{mapfile}
1762 Print a link map to the file @var{mapfile}. See the description of the
1763 @option{-M} option, above.
1765 @cindex memory usage
1766 @kindex --no-keep-memory
1767 @item --no-keep-memory
1768 @command{ld} normally optimizes for speed over memory usage by caching the
1769 symbol tables of input files in memory. This option tells @command{ld} to
1770 instead optimize for memory usage, by rereading the symbol tables as
1771 necessary. This may be required if @command{ld} runs out of memory space
1772 while linking a large executable.
1774 @kindex --no-undefined
1777 @item --no-undefined
1779 Report unresolved symbol references from regular object files. This
1780 is done even if the linker is creating a non-symbolic shared library.
1781 The switch @option{--[no-]allow-shlib-undefined} controls the
1782 behaviour for reporting unresolved references found in shared
1783 libraries being linked in.
1785 The effects of this option can be reverted by using @code{-z undefs}.
1787 @kindex --allow-multiple-definition
1789 @item --allow-multiple-definition
1791 Normally when a symbol is defined multiple times, the linker will
1792 report a fatal error. These options allow multiple definitions and the
1793 first definition will be used.
1795 @kindex --allow-shlib-undefined
1796 @kindex --no-allow-shlib-undefined
1797 @item --allow-shlib-undefined
1798 @itemx --no-allow-shlib-undefined
1799 Allows or disallows undefined symbols in shared libraries.
1800 This switch is similar to @option{--no-undefined} except that it
1801 determines the behaviour when the undefined symbols are in a
1802 shared library rather than a regular object file. It does not affect
1803 how undefined symbols in regular object files are handled.
1805 The default behaviour is to report errors for any undefined symbols
1806 referenced in shared libraries if the linker is being used to create
1807 an executable, but to allow them if the linker is being used to create
1810 The reasons for allowing undefined symbol references in shared
1811 libraries specified at link time are that:
1815 A shared library specified at link time may not be the same as the one
1816 that is available at load time, so the symbol might actually be
1817 resolvable at load time.
1819 There are some operating systems, eg BeOS and HPPA, where undefined
1820 symbols in shared libraries are normal.
1822 The BeOS kernel for example patches shared libraries at load time to
1823 select whichever function is most appropriate for the current
1824 architecture. This is used, for example, to dynamically select an
1825 appropriate memset function.
1828 @kindex --no-undefined-version
1829 @item --no-undefined-version
1830 Normally when a symbol has an undefined version, the linker will ignore
1831 it. This option disallows symbols with undefined version and a fatal error
1832 will be issued instead.
1834 @kindex --default-symver
1835 @item --default-symver
1836 Create and use a default symbol version (the soname) for unversioned
1839 @kindex --default-imported-symver
1840 @item --default-imported-symver
1841 Create and use a default symbol version (the soname) for unversioned
1844 @kindex --no-warn-mismatch
1845 @item --no-warn-mismatch
1846 Normally @command{ld} will give an error if you try to link together input
1847 files that are mismatched for some reason, perhaps because they have
1848 been compiled for different processors or for different endiannesses.
1849 This option tells @command{ld} that it should silently permit such possible
1850 errors. This option should only be used with care, in cases when you
1851 have taken some special action that ensures that the linker errors are
1854 @kindex --no-warn-search-mismatch
1855 @item --no-warn-search-mismatch
1856 Normally @command{ld} will give a warning if it finds an incompatible
1857 library during a library search. This option silences the warning.
1859 @kindex --no-whole-archive
1860 @item --no-whole-archive
1861 Turn off the effect of the @option{--whole-archive} option for subsequent
1864 @cindex output file after errors
1865 @kindex --noinhibit-exec
1866 @item --noinhibit-exec
1867 Retain the executable output file whenever it is still usable.
1868 Normally, the linker will not produce an output file if it encounters
1869 errors during the link process; it exits without writing an output file
1870 when it issues any error whatsoever.
1874 Only search library directories explicitly specified on the
1875 command line. Library directories specified in linker scripts
1876 (including linker scripts specified on the command line) are ignored.
1878 @ifclear SingleFormat
1879 @kindex --oformat=@var{output-format}
1880 @item --oformat=@var{output-format}
1881 @command{ld} may be configured to support more than one kind of object
1882 file. If your @command{ld} is configured this way, you can use the
1883 @samp{--oformat} option to specify the binary format for the output
1884 object file. Even when @command{ld} is configured to support alternative
1885 object formats, you don't usually need to specify this, as @command{ld}
1886 should be configured to produce as a default output format the most
1887 usual format on each machine. @var{output-format} is a text string, the
1888 name of a particular format supported by the BFD libraries. (You can
1889 list the available binary formats with @samp{objdump -i}.) The script
1890 command @code{OUTPUT_FORMAT} can also specify the output format, but
1891 this option overrides it. @xref{BFD}.
1894 @kindex --out-implib
1895 @item --out-implib @var{file}
1896 Create an import library in @var{file} corresponding to the executable
1897 the linker is generating (eg. a DLL or ELF program). This import
1898 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
1899 may be used to link clients against the generated executable; this
1900 behaviour makes it possible to skip a separate import library creation
1901 step (eg. @code{dlltool} for DLLs). This option is only available for
1902 the i386 PE and ELF targetted ports of the linker.
1905 @kindex --pic-executable
1907 @itemx --pic-executable
1908 @cindex position independent executables
1909 Create a position independent executable. This is currently only supported on
1910 ELF platforms. Position independent executables are similar to shared
1911 libraries in that they are relocated by the dynamic linker to the virtual
1912 address the OS chooses for them (which can vary between invocations). Like
1913 normal dynamically linked executables they can be executed and symbols
1914 defined in the executable cannot be overridden by shared libraries.
1918 This option is ignored for Linux compatibility.
1922 This option is ignored for SVR4 compatibility.
1925 @cindex synthesizing linker
1926 @cindex relaxing addressing modes
1930 An option with machine dependent effects.
1932 This option is only supported on a few targets.
1935 @xref{H8/300,,@command{ld} and the H8/300}.
1938 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1941 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1944 @xref{Nios II,,@command{ld} and the Altera Nios II}.
1947 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1950 On some platforms the @samp{--relax} option performs target-specific,
1951 global optimizations that become possible when the linker resolves
1952 addressing in the program, such as relaxing address modes,
1953 synthesizing new instructions, selecting shorter version of current
1954 instructions, and combining constant values.
1956 On some platforms these link time global optimizations may make symbolic
1957 debugging of the resulting executable impossible.
1959 This is known to be the case for the Matsushita MN10200 and MN10300
1960 family of processors.
1964 On platforms where this is not supported, @samp{--relax} is accepted,
1968 On platforms where @samp{--relax} is accepted the option
1969 @samp{--no-relax} can be used to disable the feature.
1971 @cindex retaining specified symbols
1972 @cindex stripping all but some symbols
1973 @cindex symbols, retaining selectively
1974 @kindex --retain-symbols-file=@var{filename}
1975 @item --retain-symbols-file=@var{filename}
1976 Retain @emph{only} the symbols listed in the file @var{filename},
1977 discarding all others. @var{filename} is simply a flat file, with one
1978 symbol name per line. This option is especially useful in environments
1982 where a large global symbol table is accumulated gradually, to conserve
1985 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1986 or symbols needed for relocations.
1988 You may only specify @samp{--retain-symbols-file} once in the command
1989 line. It overrides @samp{-s} and @samp{-S}.
1992 @item -rpath=@var{dir}
1993 @cindex runtime library search path
1994 @kindex -rpath=@var{dir}
1995 Add a directory to the runtime library search path. This is used when
1996 linking an ELF executable with shared objects. All @option{-rpath}
1997 arguments are concatenated and passed to the runtime linker, which uses
1998 them to locate shared objects at runtime.
2000 The @option{-rpath} option is also used when locating shared objects which
2001 are needed by shared objects explicitly included in the link; see the
2002 description of the @option{-rpath-link} option. Searching @option{-rpath}
2003 in this way is only supported by native linkers and cross linkers which
2004 have been configured with the @option{--with-sysroot} option.
2006 If @option{-rpath} is not used when linking an ELF executable, the
2007 contents of the environment variable @code{LD_RUN_PATH} will be used if it
2010 The @option{-rpath} option may also be used on SunOS. By default, on
2011 SunOS, the linker will form a runtime search path out of all the
2012 @option{-L} options it is given. If a @option{-rpath} option is used, the
2013 runtime search path will be formed exclusively using the @option{-rpath}
2014 options, ignoring the @option{-L} options. This can be useful when using
2015 gcc, which adds many @option{-L} options which may be on NFS mounted
2018 For compatibility with other ELF linkers, if the @option{-R} option is
2019 followed by a directory name, rather than a file name, it is treated as
2020 the @option{-rpath} option.
2024 @cindex link-time runtime library search path
2025 @kindex -rpath-link=@var{dir}
2026 @item -rpath-link=@var{dir}
2027 When using ELF or SunOS, one shared library may require another. This
2028 happens when an @code{ld -shared} link includes a shared library as one
2031 When the linker encounters such a dependency when doing a non-shared,
2032 non-relocatable link, it will automatically try to locate the required
2033 shared library and include it in the link, if it is not included
2034 explicitly. In such a case, the @option{-rpath-link} option
2035 specifies the first set of directories to search. The
2036 @option{-rpath-link} option may specify a sequence of directory names
2037 either by specifying a list of names separated by colons, or by
2038 appearing multiple times.
2040 The tokens @var{$ORIGIN} and @var{$LIB} can appear in these search
2041 directories. They will be replaced by the full path to the directory
2042 containing the program or shared object in the case of @var{$ORIGIN}
2043 and either @samp{lib} - for 32-bit binaries - or @samp{lib64} - for
2044 64-bit binaries - in the case of @var{$LIB}.
2046 The alternative form of these tokens - @var{$@{ORIGIN@}} and
2047 @var{$@{LIB@}} can also be used. The token @var{$PLATFORM} is not
2050 This option should be used with caution as it overrides the search path
2051 that may have been hard compiled into a shared library. In such a case it
2052 is possible to use unintentionally a different search path than the
2053 runtime linker would do.
2055 The linker uses the following search paths to locate required shared
2060 Any directories specified by @option{-rpath-link} options.
2062 Any directories specified by @option{-rpath} options. The difference
2063 between @option{-rpath} and @option{-rpath-link} is that directories
2064 specified by @option{-rpath} options are included in the executable and
2065 used at runtime, whereas the @option{-rpath-link} option is only effective
2066 at link time. Searching @option{-rpath} in this way is only supported
2067 by native linkers and cross linkers which have been configured with
2068 the @option{--with-sysroot} option.
2070 On an ELF system, for native linkers, if the @option{-rpath} and
2071 @option{-rpath-link} options were not used, search the contents of the
2072 environment variable @code{LD_RUN_PATH}.
2074 On SunOS, if the @option{-rpath} option was not used, search any
2075 directories specified using @option{-L} options.
2077 For a native linker, search the contents of the environment
2078 variable @code{LD_LIBRARY_PATH}.
2080 For a native ELF linker, the directories in @code{DT_RUNPATH} or
2081 @code{DT_RPATH} of a shared library are searched for shared
2082 libraries needed by it. The @code{DT_RPATH} entries are ignored if
2083 @code{DT_RUNPATH} entries exist.
2085 The default directories, normally @file{/lib} and @file{/usr/lib}.
2087 For a linker for a Linux system, if the file @file{/etc/ld.so.conf}
2088 exists, the list of directories found in that file. Note: the path
2089 to this file is prefixed with the @code{sysroot} value, if that is
2090 defined, and then any @code{prefix} string if the linker was
2091 configured with the @command{--prefix=<path>} option.
2093 For a native linker on a FreeBSD system, any directories specified by
2094 the @code{_PATH_ELF_HINTS} macro defined in the @file{elf-hints.h}
2097 Any directories specifed by a @code{SEARCH_DIR} command in the
2098 linker script being used.
2101 If the required shared library is not found, the linker will issue a
2102 warning and continue with the link.
2109 @cindex shared libraries
2110 Create a shared library. This is currently only supported on ELF, XCOFF
2111 and SunOS platforms. On SunOS, the linker will automatically create a
2112 shared library if the @option{-e} option is not used and there are
2113 undefined symbols in the link.
2115 @kindex --sort-common
2117 @itemx --sort-common=ascending
2118 @itemx --sort-common=descending
2119 This option tells @command{ld} to sort the common symbols by alignment in
2120 ascending or descending order when it places them in the appropriate output
2121 sections. The symbol alignments considered are sixteen-byte or larger,
2122 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
2123 between symbols due to alignment constraints. If no sorting order is
2124 specified, then descending order is assumed.
2126 @kindex --sort-section=name
2127 @item --sort-section=name
2128 This option will apply @code{SORT_BY_NAME} to all wildcard section
2129 patterns in the linker script.
2131 @kindex --sort-section=alignment
2132 @item --sort-section=alignment
2133 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
2134 patterns in the linker script.
2136 @kindex --spare-dynamic-tags
2137 @item --spare-dynamic-tags=@var{count}
2138 This option specifies the number of empty slots to leave in the
2139 .dynamic section of ELF shared objects. Empty slots may be needed by
2140 post processing tools, such as the prelinker. The default is 5.
2142 @kindex --split-by-file
2143 @item --split-by-file[=@var{size}]
2144 Similar to @option{--split-by-reloc} but creates a new output section for
2145 each input file when @var{size} is reached. @var{size} defaults to a
2146 size of 1 if not given.
2148 @kindex --split-by-reloc
2149 @item --split-by-reloc[=@var{count}]
2150 Tries to creates extra sections in the output file so that no single
2151 output section in the file contains more than @var{count} relocations.
2152 This is useful when generating huge relocatable files for downloading into
2153 certain real time kernels with the COFF object file format; since COFF
2154 cannot represent more than 65535 relocations in a single section. Note
2155 that this will fail to work with object file formats which do not
2156 support arbitrary sections. The linker will not split up individual
2157 input sections for redistribution, so if a single input section contains
2158 more than @var{count} relocations one output section will contain that
2159 many relocations. @var{count} defaults to a value of 32768.
2163 Compute and display statistics about the operation of the linker, such
2164 as execution time and memory usage.
2166 @kindex --sysroot=@var{directory}
2167 @item --sysroot=@var{directory}
2168 Use @var{directory} as the location of the sysroot, overriding the
2169 configure-time default. This option is only supported by linkers
2170 that were configured using @option{--with-sysroot}.
2174 This is used by COFF/PE based targets to create a task-linked object
2175 file where all of the global symbols have been converted to statics.
2177 @kindex --traditional-format
2178 @cindex traditional format
2179 @item --traditional-format
2180 For some targets, the output of @command{ld} is different in some ways from
2181 the output of some existing linker. This switch requests @command{ld} to
2182 use the traditional format instead.
2185 For example, on SunOS, @command{ld} combines duplicate entries in the
2186 symbol string table. This can reduce the size of an output file with
2187 full debugging information by over 30 percent. Unfortunately, the SunOS
2188 @code{dbx} program can not read the resulting program (@code{gdb} has no
2189 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
2190 combine duplicate entries.
2192 @kindex --section-start=@var{sectionname}=@var{org}
2193 @item --section-start=@var{sectionname}=@var{org}
2194 Locate a section in the output file at the absolute
2195 address given by @var{org}. You may use this option as many
2196 times as necessary to locate multiple sections in the command
2198 @var{org} must be a single hexadecimal integer;
2199 for compatibility with other linkers, you may omit the leading
2200 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
2201 should be no white space between @var{sectionname}, the equals
2202 sign (``@key{=}''), and @var{org}.
2204 @kindex -Tbss=@var{org}
2205 @kindex -Tdata=@var{org}
2206 @kindex -Ttext=@var{org}
2207 @cindex segment origins, cmd line
2208 @item -Tbss=@var{org}
2209 @itemx -Tdata=@var{org}
2210 @itemx -Ttext=@var{org}
2211 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2212 @code{.text} as the @var{sectionname}.
2214 @kindex -Ttext-segment=@var{org}
2215 @item -Ttext-segment=@var{org}
2216 @cindex text segment origin, cmd line
2217 When creating an ELF executable, it will set the address of the first
2218 byte of the text segment.
2220 @kindex -Trodata-segment=@var{org}
2221 @item -Trodata-segment=@var{org}
2222 @cindex rodata segment origin, cmd line
2223 When creating an ELF executable or shared object for a target where
2224 the read-only data is in its own segment separate from the executable
2225 text, it will set the address of the first byte of the read-only data segment.
2227 @kindex -Tldata-segment=@var{org}
2228 @item -Tldata-segment=@var{org}
2229 @cindex ldata segment origin, cmd line
2230 When creating an ELF executable or shared object for x86-64 medium memory
2231 model, it will set the address of the first byte of the ldata segment.
2233 @kindex --unresolved-symbols
2234 @item --unresolved-symbols=@var{method}
2235 Determine how to handle unresolved symbols. There are four possible
2236 values for @samp{method}:
2240 Do not report any unresolved symbols.
2243 Report all unresolved symbols. This is the default.
2245 @item ignore-in-object-files
2246 Report unresolved symbols that are contained in shared libraries, but
2247 ignore them if they come from regular object files.
2249 @item ignore-in-shared-libs
2250 Report unresolved symbols that come from regular object files, but
2251 ignore them if they come from shared libraries. This can be useful
2252 when creating a dynamic binary and it is known that all the shared
2253 libraries that it should be referencing are included on the linker's
2257 The behaviour for shared libraries on their own can also be controlled
2258 by the @option{--[no-]allow-shlib-undefined} option.
2260 Normally the linker will generate an error message for each reported
2261 unresolved symbol but the option @option{--warn-unresolved-symbols}
2262 can change this to a warning.
2264 @kindex --verbose[=@var{NUMBER}]
2265 @cindex verbose[=@var{NUMBER}]
2267 @itemx --verbose[=@var{NUMBER}]
2268 Display the version number for @command{ld} and list the linker emulations
2269 supported. Display which input files can and cannot be opened. Display
2270 the linker script being used by the linker. If the optional @var{NUMBER}
2271 argument > 1, plugin symbol status will also be displayed.
2273 @kindex --version-script=@var{version-scriptfile}
2274 @cindex version script, symbol versions
2275 @item --version-script=@var{version-scriptfile}
2276 Specify the name of a version script to the linker. This is typically
2277 used when creating shared libraries to specify additional information
2278 about the version hierarchy for the library being created. This option
2279 is only fully supported on ELF platforms which support shared libraries;
2280 see @ref{VERSION}. It is partially supported on PE platforms, which can
2281 use version scripts to filter symbol visibility in auto-export mode: any
2282 symbols marked @samp{local} in the version script will not be exported.
2285 @kindex --warn-common
2286 @cindex warnings, on combining symbols
2287 @cindex combining symbols, warnings on
2289 Warn when a common symbol is combined with another common symbol or with
2290 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2291 but linkers on some other operating systems do not. This option allows
2292 you to find potential problems from combining global symbols.
2293 Unfortunately, some C libraries use this practice, so you may get some
2294 warnings about symbols in the libraries as well as in your programs.
2296 There are three kinds of global symbols, illustrated here by C examples:
2300 A definition, which goes in the initialized data section of the output
2304 An undefined reference, which does not allocate space.
2305 There must be either a definition or a common symbol for the
2309 A common symbol. If there are only (one or more) common symbols for a
2310 variable, it goes in the uninitialized data area of the output file.
2311 The linker merges multiple common symbols for the same variable into a
2312 single symbol. If they are of different sizes, it picks the largest
2313 size. The linker turns a common symbol into a declaration, if there is
2314 a definition of the same variable.
2317 The @samp{--warn-common} option can produce five kinds of warnings.
2318 Each warning consists of a pair of lines: the first describes the symbol
2319 just encountered, and the second describes the previous symbol
2320 encountered with the same name. One or both of the two symbols will be
2325 Turning a common symbol into a reference, because there is already a
2326 definition for the symbol.
2328 @var{file}(@var{section}): warning: common of `@var{symbol}'
2329 overridden by definition
2330 @var{file}(@var{section}): warning: defined here
2334 Turning a common symbol into a reference, because a later definition for
2335 the symbol is encountered. This is the same as the previous case,
2336 except that the symbols are encountered in a different order.
2338 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2340 @var{file}(@var{section}): warning: common is here
2344 Merging a common symbol with a previous same-sized common symbol.
2346 @var{file}(@var{section}): warning: multiple common
2348 @var{file}(@var{section}): warning: previous common is here
2352 Merging a common symbol with a previous larger common symbol.
2354 @var{file}(@var{section}): warning: common of `@var{symbol}'
2355 overridden by larger common
2356 @var{file}(@var{section}): warning: larger common is here
2360 Merging a common symbol with a previous smaller common symbol. This is
2361 the same as the previous case, except that the symbols are
2362 encountered in a different order.
2364 @var{file}(@var{section}): warning: common of `@var{symbol}'
2365 overriding smaller common
2366 @var{file}(@var{section}): warning: smaller common is here
2370 @kindex --warn-constructors
2371 @item --warn-constructors
2372 Warn if any global constructors are used. This is only useful for a few
2373 object file formats. For formats like COFF or ELF, the linker can not
2374 detect the use of global constructors.
2376 @kindex --warn-multiple-gp
2377 @item --warn-multiple-gp
2378 Warn if multiple global pointer values are required in the output file.
2379 This is only meaningful for certain processors, such as the Alpha.
2380 Specifically, some processors put large-valued constants in a special
2381 section. A special register (the global pointer) points into the middle
2382 of this section, so that constants can be loaded efficiently via a
2383 base-register relative addressing mode. Since the offset in
2384 base-register relative mode is fixed and relatively small (e.g., 16
2385 bits), this limits the maximum size of the constant pool. Thus, in
2386 large programs, it is often necessary to use multiple global pointer
2387 values in order to be able to address all possible constants. This
2388 option causes a warning to be issued whenever this case occurs.
2391 @cindex warnings, on undefined symbols
2392 @cindex undefined symbols, warnings on
2394 Only warn once for each undefined symbol, rather than once per module
2397 @kindex --warn-section-align
2398 @cindex warnings, on section alignment
2399 @cindex section alignment, warnings on
2400 @item --warn-section-align
2401 Warn if the address of an output section is changed because of
2402 alignment. Typically, the alignment will be set by an input section.
2403 The address will only be changed if it not explicitly specified; that
2404 is, if the @code{SECTIONS} command does not specify a start address for
2405 the section (@pxref{SECTIONS}).
2407 @kindex --warn-shared-textrel
2408 @item --warn-shared-textrel
2409 Warn if the linker adds a DT_TEXTREL to a shared object.
2411 @kindex --warn-alternate-em
2412 @item --warn-alternate-em
2413 Warn if an object has alternate ELF machine code.
2415 @kindex --warn-unresolved-symbols
2416 @item --warn-unresolved-symbols
2417 If the linker is going to report an unresolved symbol (see the option
2418 @option{--unresolved-symbols}) it will normally generate an error.
2419 This option makes it generate a warning instead.
2421 @kindex --error-unresolved-symbols
2422 @item --error-unresolved-symbols
2423 This restores the linker's default behaviour of generating errors when
2424 it is reporting unresolved symbols.
2426 @kindex --whole-archive
2427 @cindex including an entire archive
2428 @item --whole-archive
2429 For each archive mentioned on the command line after the
2430 @option{--whole-archive} option, include every object file in the archive
2431 in the link, rather than searching the archive for the required object
2432 files. This is normally used to turn an archive file into a shared
2433 library, forcing every object to be included in the resulting shared
2434 library. This option may be used more than once.
2436 Two notes when using this option from gcc: First, gcc doesn't know
2437 about this option, so you have to use @option{-Wl,-whole-archive}.
2438 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2439 list of archives, because gcc will add its own list of archives to
2440 your link and you may not want this flag to affect those as well.
2442 @kindex --wrap=@var{symbol}
2443 @item --wrap=@var{symbol}
2444 Use a wrapper function for @var{symbol}. Any undefined reference to
2445 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2446 undefined reference to @code{__real_@var{symbol}} will be resolved to
2449 This can be used to provide a wrapper for a system function. The
2450 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2451 wishes to call the system function, it should call
2452 @code{__real_@var{symbol}}.
2454 Here is a trivial example:
2458 __wrap_malloc (size_t c)
2460 printf ("malloc called with %zu\n", c);
2461 return __real_malloc (c);
2465 If you link other code with this file using @option{--wrap malloc}, then
2466 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2467 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2468 call the real @code{malloc} function.
2470 You may wish to provide a @code{__real_malloc} function as well, so that
2471 links without the @option{--wrap} option will succeed. If you do this,
2472 you should not put the definition of @code{__real_malloc} in the same
2473 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2474 call before the linker has a chance to wrap it to @code{malloc}.
2476 Only undefined references are replaced by the linker. So, translation unit
2477 internal references to @var{symbol} are not resolved to
2478 @code{__wrap_@var{symbol}}. In the next example, the call to @code{f} in
2479 @code{g} is not resolved to @code{__wrap_f}.
2495 @kindex --eh-frame-hdr
2496 @kindex --no-eh-frame-hdr
2497 @item --eh-frame-hdr
2498 @itemx --no-eh-frame-hdr
2499 Request (@option{--eh-frame-hdr}) or suppress
2500 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2501 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2503 @kindex --ld-generated-unwind-info
2504 @item --no-ld-generated-unwind-info
2505 Request creation of @code{.eh_frame} unwind info for linker
2506 generated code sections like PLT. This option is on by default
2507 if linker generated unwind info is supported.
2509 @kindex --enable-new-dtags
2510 @kindex --disable-new-dtags
2511 @item --enable-new-dtags
2512 @itemx --disable-new-dtags
2513 This linker can create the new dynamic tags in ELF. But the older ELF
2514 systems may not understand them. If you specify
2515 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2516 and older dynamic tags will be omitted.
2517 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2518 created. By default, the new dynamic tags are not created. Note that
2519 those options are only available for ELF systems.
2521 @kindex --hash-size=@var{number}
2522 @item --hash-size=@var{number}
2523 Set the default size of the linker's hash tables to a prime number
2524 close to @var{number}. Increasing this value can reduce the length of
2525 time it takes the linker to perform its tasks, at the expense of
2526 increasing the linker's memory requirements. Similarly reducing this
2527 value can reduce the memory requirements at the expense of speed.
2529 @kindex --hash-style=@var{style}
2530 @item --hash-style=@var{style}
2531 Set the type of linker's hash table(s). @var{style} can be either
2532 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2533 new style GNU @code{.gnu.hash} section or @code{both} for both
2534 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2535 hash tables. The default depends upon how the linker was configured,
2536 but for most Linux based systems it will be @code{both}.
2538 @kindex --compress-debug-sections=none
2539 @kindex --compress-debug-sections=zlib
2540 @kindex --compress-debug-sections=zlib-gnu
2541 @kindex --compress-debug-sections=zlib-gabi
2542 @item --compress-debug-sections=none
2543 @itemx --compress-debug-sections=zlib
2544 @itemx --compress-debug-sections=zlib-gnu
2545 @itemx --compress-debug-sections=zlib-gabi
2546 On ELF platforms, these options control how DWARF debug sections are
2547 compressed using zlib.
2549 @option{--compress-debug-sections=none} doesn't compress DWARF debug
2550 sections. @option{--compress-debug-sections=zlib-gnu} compresses
2551 DWARF debug sections and renames them to begin with @samp{.zdebug}
2552 instead of @samp{.debug}. @option{--compress-debug-sections=zlib-gabi}
2553 also compresses DWARF debug sections, but rather than renaming them it
2554 sets the SHF_COMPRESSED flag in the sections' headers.
2556 The @option{--compress-debug-sections=zlib} option is an alias for
2557 @option{--compress-debug-sections=zlib-gabi}.
2559 Note that this option overrides any compression in input debug
2560 sections, so if a binary is linked with @option{--compress-debug-sections=none}
2561 for example, then any compressed debug sections in input files will be
2562 uncompressed before they are copied into the output binary.
2564 The default compression behaviour varies depending upon the target
2565 involved and the configure options used to build the toolchain. The
2566 default can be determined by examining the output from the linker's
2567 @option{--help} option.
2569 @kindex --reduce-memory-overheads
2570 @item --reduce-memory-overheads
2571 This option reduces memory requirements at ld runtime, at the expense of
2572 linking speed. This was introduced to select the old O(n^2) algorithm
2573 for link map file generation, rather than the new O(n) algorithm which uses
2574 about 40% more memory for symbol storage.
2576 Another effect of the switch is to set the default hash table size to
2577 1021, which again saves memory at the cost of lengthening the linker's
2578 run time. This is not done however if the @option{--hash-size} switch
2581 The @option{--reduce-memory-overheads} switch may be also be used to
2582 enable other tradeoffs in future versions of the linker.
2585 @kindex --build-id=@var{style}
2587 @itemx --build-id=@var{style}
2588 Request the creation of a @code{.note.gnu.build-id} ELF note section
2589 or a @code{.buildid} COFF section. The contents of the note are
2590 unique bits identifying this linked file. @var{style} can be
2591 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2592 @sc{SHA1} hash on the normative parts of the output contents,
2593 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2594 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2595 string specified as an even number of hexadecimal digits (@code{-} and
2596 @code{:} characters between digit pairs are ignored). If @var{style}
2597 is omitted, @code{sha1} is used.
2599 The @code{md5} and @code{sha1} styles produces an identifier
2600 that is always the same in an identical output file, but will be
2601 unique among all nonidentical output files. It is not intended
2602 to be compared as a checksum for the file's contents. A linked
2603 file may be changed later by other tools, but the build ID bit
2604 string identifying the original linked file does not change.
2606 Passing @code{none} for @var{style} disables the setting from any
2607 @code{--build-id} options earlier on the command line.
2612 @subsection Options Specific to i386 PE Targets
2614 @c man begin OPTIONS
2616 The i386 PE linker supports the @option{-shared} option, which causes
2617 the output to be a dynamically linked library (DLL) instead of a
2618 normal executable. You should name the output @code{*.dll} when you
2619 use this option. In addition, the linker fully supports the standard
2620 @code{*.def} files, which may be specified on the linker command line
2621 like an object file (in fact, it should precede archives it exports
2622 symbols from, to ensure that they get linked in, just like a normal
2625 In addition to the options common to all targets, the i386 PE linker
2626 support additional command-line options that are specific to the i386
2627 PE target. Options that take values may be separated from their
2628 values by either a space or an equals sign.
2632 @kindex --add-stdcall-alias
2633 @item --add-stdcall-alias
2634 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2635 as-is and also with the suffix stripped.
2636 [This option is specific to the i386 PE targeted port of the linker]
2639 @item --base-file @var{file}
2640 Use @var{file} as the name of a file in which to save the base
2641 addresses of all the relocations needed for generating DLLs with
2643 [This is an i386 PE specific option]
2647 Create a DLL instead of a regular executable. You may also use
2648 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2650 [This option is specific to the i386 PE targeted port of the linker]
2652 @kindex --enable-long-section-names
2653 @kindex --disable-long-section-names
2654 @item --enable-long-section-names
2655 @itemx --disable-long-section-names
2656 The PE variants of the COFF object format add an extension that permits
2657 the use of section names longer than eight characters, the normal limit
2658 for COFF. By default, these names are only allowed in object files, as
2659 fully-linked executable images do not carry the COFF string table required
2660 to support the longer names. As a GNU extension, it is possible to
2661 allow their use in executable images as well, or to (probably pointlessly!)
2662 disallow it in object files, by using these two options. Executable images
2663 generated with these long section names are slightly non-standard, carrying
2664 as they do a string table, and may generate confusing output when examined
2665 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2666 GDB relies on the use of PE long section names to find Dwarf-2 debug
2667 information sections in an executable image at runtime, and so if neither
2668 option is specified on the command-line, @command{ld} will enable long
2669 section names, overriding the default and technically correct behaviour,
2670 when it finds the presence of debug information while linking an executable
2671 image and not stripping symbols.
2672 [This option is valid for all PE targeted ports of the linker]
2674 @kindex --enable-stdcall-fixup
2675 @kindex --disable-stdcall-fixup
2676 @item --enable-stdcall-fixup
2677 @itemx --disable-stdcall-fixup
2678 If the link finds a symbol that it cannot resolve, it will attempt to
2679 do ``fuzzy linking'' by looking for another defined symbol that differs
2680 only in the format of the symbol name (cdecl vs stdcall) and will
2681 resolve that symbol by linking to the match. For example, the
2682 undefined symbol @code{_foo} might be linked to the function
2683 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2684 to the function @code{_bar}. When the linker does this, it prints a
2685 warning, since it normally should have failed to link, but sometimes
2686 import libraries generated from third-party dlls may need this feature
2687 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2688 feature is fully enabled and warnings are not printed. If you specify
2689 @option{--disable-stdcall-fixup}, this feature is disabled and such
2690 mismatches are considered to be errors.
2691 [This option is specific to the i386 PE targeted port of the linker]
2693 @kindex --leading-underscore
2694 @kindex --no-leading-underscore
2695 @item --leading-underscore
2696 @itemx --no-leading-underscore
2697 For most targets default symbol-prefix is an underscore and is defined
2698 in target's description. By this option it is possible to
2699 disable/enable the default underscore symbol-prefix.
2701 @cindex DLLs, creating
2702 @kindex --export-all-symbols
2703 @item --export-all-symbols
2704 If given, all global symbols in the objects used to build a DLL will
2705 be exported by the DLL. Note that this is the default if there
2706 otherwise wouldn't be any exported symbols. When symbols are
2707 explicitly exported via DEF files or implicitly exported via function
2708 attributes, the default is to not export anything else unless this
2709 option is given. Note that the symbols @code{DllMain@@12},
2710 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2711 @code{impure_ptr} will not be automatically
2712 exported. Also, symbols imported from other DLLs will not be
2713 re-exported, nor will symbols specifying the DLL's internal layout
2714 such as those beginning with @code{_head_} or ending with
2715 @code{_iname}. In addition, no symbols from @code{libgcc},
2716 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2717 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2718 not be exported, to help with C++ DLLs. Finally, there is an
2719 extensive list of cygwin-private symbols that are not exported
2720 (obviously, this applies on when building DLLs for cygwin targets).
2721 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2722 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2723 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2724 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2725 @code{cygwin_premain3}, and @code{environ}.
2726 [This option is specific to the i386 PE targeted port of the linker]
2728 @kindex --exclude-symbols
2729 @item --exclude-symbols @var{symbol},@var{symbol},...
2730 Specifies a list of symbols which should not be automatically
2731 exported. The symbol names may be delimited by commas or colons.
2732 [This option is specific to the i386 PE targeted port of the linker]
2734 @kindex --exclude-all-symbols
2735 @item --exclude-all-symbols
2736 Specifies no symbols should be automatically exported.
2737 [This option is specific to the i386 PE targeted port of the linker]
2739 @kindex --file-alignment
2740 @item --file-alignment
2741 Specify the file alignment. Sections in the file will always begin at
2742 file offsets which are multiples of this number. This defaults to
2744 [This option is specific to the i386 PE targeted port of the linker]
2748 @item --heap @var{reserve}
2749 @itemx --heap @var{reserve},@var{commit}
2750 Specify the number of bytes of memory to reserve (and optionally commit)
2751 to be used as heap for this program. The default is 1MB reserved, 4K
2753 [This option is specific to the i386 PE targeted port of the linker]
2756 @kindex --image-base
2757 @item --image-base @var{value}
2758 Use @var{value} as the base address of your program or dll. This is
2759 the lowest memory location that will be used when your program or dll
2760 is loaded. To reduce the need to relocate and improve performance of
2761 your dlls, each should have a unique base address and not overlap any
2762 other dlls. The default is 0x400000 for executables, and 0x10000000
2764 [This option is specific to the i386 PE targeted port of the linker]
2768 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2769 symbols before they are exported.
2770 [This option is specific to the i386 PE targeted port of the linker]
2772 @kindex --large-address-aware
2773 @item --large-address-aware
2774 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2775 header is set to indicate that this executable supports virtual addresses
2776 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2777 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2778 section of the BOOT.INI. Otherwise, this bit has no effect.
2779 [This option is specific to PE targeted ports of the linker]
2781 @kindex --disable-large-address-aware
2782 @item --disable-large-address-aware
2783 Reverts the effect of a previous @samp{--large-address-aware} option.
2784 This is useful if @samp{--large-address-aware} is always set by the compiler
2785 driver (e.g. Cygwin gcc) and the executable does not support virtual
2786 addresses greater than 2 gigabytes.
2787 [This option is specific to PE targeted ports of the linker]
2789 @kindex --major-image-version
2790 @item --major-image-version @var{value}
2791 Sets the major number of the ``image version''. Defaults to 1.
2792 [This option is specific to the i386 PE targeted port of the linker]
2794 @kindex --major-os-version
2795 @item --major-os-version @var{value}
2796 Sets the major number of the ``os version''. Defaults to 4.
2797 [This option is specific to the i386 PE targeted port of the linker]
2799 @kindex --major-subsystem-version
2800 @item --major-subsystem-version @var{value}
2801 Sets the major number of the ``subsystem version''. Defaults to 4.
2802 [This option is specific to the i386 PE targeted port of the linker]
2804 @kindex --minor-image-version
2805 @item --minor-image-version @var{value}
2806 Sets the minor number of the ``image version''. Defaults to 0.
2807 [This option is specific to the i386 PE targeted port of the linker]
2809 @kindex --minor-os-version
2810 @item --minor-os-version @var{value}
2811 Sets the minor number of the ``os version''. Defaults to 0.
2812 [This option is specific to the i386 PE targeted port of the linker]
2814 @kindex --minor-subsystem-version
2815 @item --minor-subsystem-version @var{value}
2816 Sets the minor number of the ``subsystem version''. Defaults to 0.
2817 [This option is specific to the i386 PE targeted port of the linker]
2819 @cindex DEF files, creating
2820 @cindex DLLs, creating
2821 @kindex --output-def
2822 @item --output-def @var{file}
2823 The linker will create the file @var{file} which will contain a DEF
2824 file corresponding to the DLL the linker is generating. This DEF file
2825 (which should be called @code{*.def}) may be used to create an import
2826 library with @code{dlltool} or may be used as a reference to
2827 automatically or implicitly exported symbols.
2828 [This option is specific to the i386 PE targeted port of the linker]
2830 @cindex DLLs, creating
2831 @kindex --enable-auto-image-base
2832 @item --enable-auto-image-base
2833 @itemx --enable-auto-image-base=@var{value}
2834 Automatically choose the image base for DLLs, optionally starting with base
2835 @var{value}, unless one is specified using the @code{--image-base} argument.
2836 By using a hash generated from the dllname to create unique image bases
2837 for each DLL, in-memory collisions and relocations which can delay program
2838 execution are avoided.
2839 [This option is specific to the i386 PE targeted port of the linker]
2841 @kindex --disable-auto-image-base
2842 @item --disable-auto-image-base
2843 Do not automatically generate a unique image base. If there is no
2844 user-specified image base (@code{--image-base}) then use the platform
2846 [This option is specific to the i386 PE targeted port of the linker]
2848 @cindex DLLs, linking to
2849 @kindex --dll-search-prefix
2850 @item --dll-search-prefix @var{string}
2851 When linking dynamically to a dll without an import library,
2852 search for @code{<string><basename>.dll} in preference to
2853 @code{lib<basename>.dll}. This behaviour allows easy distinction
2854 between DLLs built for the various "subplatforms": native, cygwin,
2855 uwin, pw, etc. For instance, cygwin DLLs typically use
2856 @code{--dll-search-prefix=cyg}.
2857 [This option is specific to the i386 PE targeted port of the linker]
2859 @kindex --enable-auto-import
2860 @item --enable-auto-import
2861 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2862 DATA imports from DLLs, thus making it possible to bypass the dllimport
2863 mechanism on the user side and to reference unmangled symbol names.
2864 [This option is specific to the i386 PE targeted port of the linker]
2866 The following remarks pertain to the original implementation of the
2867 feature and are obsolete nowadays for Cygwin and MinGW targets.
2869 Note: Use of the 'auto-import' extension will cause the text section
2870 of the image file to be made writable. This does not conform to the
2871 PE-COFF format specification published by Microsoft.
2873 Note - use of the 'auto-import' extension will also cause read only
2874 data which would normally be placed into the .rdata section to be
2875 placed into the .data section instead. This is in order to work
2876 around a problem with consts that is described here:
2877 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2879 Using 'auto-import' generally will 'just work' -- but sometimes you may
2882 "variable '<var>' can't be auto-imported. Please read the
2883 documentation for ld's @code{--enable-auto-import} for details."
2885 This message occurs when some (sub)expression accesses an address
2886 ultimately given by the sum of two constants (Win32 import tables only
2887 allow one). Instances where this may occur include accesses to member
2888 fields of struct variables imported from a DLL, as well as using a
2889 constant index into an array variable imported from a DLL. Any
2890 multiword variable (arrays, structs, long long, etc) may trigger
2891 this error condition. However, regardless of the exact data type
2892 of the offending exported variable, ld will always detect it, issue
2893 the warning, and exit.
2895 There are several ways to address this difficulty, regardless of the
2896 data type of the exported variable:
2898 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2899 of adjusting references in your client code for runtime environment, so
2900 this method works only when runtime environment supports this feature.
2902 A second solution is to force one of the 'constants' to be a variable --
2903 that is, unknown and un-optimizable at compile time. For arrays,
2904 there are two possibilities: a) make the indexee (the array's address)
2905 a variable, or b) make the 'constant' index a variable. Thus:
2908 extern type extern_array[];
2910 @{ volatile type *t=extern_array; t[1] @}
2916 extern type extern_array[];
2918 @{ volatile int t=1; extern_array[t] @}
2921 For structs (and most other multiword data types) the only option
2922 is to make the struct itself (or the long long, or the ...) variable:
2925 extern struct s extern_struct;
2926 extern_struct.field -->
2927 @{ volatile struct s *t=&extern_struct; t->field @}
2933 extern long long extern_ll;
2935 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2938 A third method of dealing with this difficulty is to abandon
2939 'auto-import' for the offending symbol and mark it with
2940 @code{__declspec(dllimport)}. However, in practice that
2941 requires using compile-time #defines to indicate whether you are
2942 building a DLL, building client code that will link to the DLL, or
2943 merely building/linking to a static library. In making the choice
2944 between the various methods of resolving the 'direct address with
2945 constant offset' problem, you should consider typical real-world usage:
2953 void main(int argc, char **argv)@{
2954 printf("%d\n",arr[1]);
2964 void main(int argc, char **argv)@{
2965 /* This workaround is for win32 and cygwin; do not "optimize" */
2966 volatile int *parr = arr;
2967 printf("%d\n",parr[1]);
2974 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2975 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2976 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2977 #define FOO_IMPORT __declspec(dllimport)
2981 extern FOO_IMPORT int arr[];
2984 void main(int argc, char **argv)@{
2985 printf("%d\n",arr[1]);
2989 A fourth way to avoid this problem is to re-code your
2990 library to use a functional interface rather than a data interface
2991 for the offending variables (e.g. set_foo() and get_foo() accessor
2994 @kindex --disable-auto-import
2995 @item --disable-auto-import
2996 Do not attempt to do sophisticated linking of @code{_symbol} to
2997 @code{__imp__symbol} for DATA imports from DLLs.
2998 [This option is specific to the i386 PE targeted port of the linker]
3000 @kindex --enable-runtime-pseudo-reloc
3001 @item --enable-runtime-pseudo-reloc
3002 If your code contains expressions described in --enable-auto-import section,
3003 that is, DATA imports from DLL with non-zero offset, this switch will create
3004 a vector of 'runtime pseudo relocations' which can be used by runtime
3005 environment to adjust references to such data in your client code.
3006 [This option is specific to the i386 PE targeted port of the linker]
3008 @kindex --disable-runtime-pseudo-reloc
3009 @item --disable-runtime-pseudo-reloc
3010 Do not create pseudo relocations for non-zero offset DATA imports from DLLs.
3011 [This option is specific to the i386 PE targeted port of the linker]
3013 @kindex --enable-extra-pe-debug
3014 @item --enable-extra-pe-debug
3015 Show additional debug info related to auto-import symbol thunking.
3016 [This option is specific to the i386 PE targeted port of the linker]
3018 @kindex --section-alignment
3019 @item --section-alignment
3020 Sets the section alignment. Sections in memory will always begin at
3021 addresses which are a multiple of this number. Defaults to 0x1000.
3022 [This option is specific to the i386 PE targeted port of the linker]
3026 @item --stack @var{reserve}
3027 @itemx --stack @var{reserve},@var{commit}
3028 Specify the number of bytes of memory to reserve (and optionally commit)
3029 to be used as stack for this program. The default is 2MB reserved, 4K
3031 [This option is specific to the i386 PE targeted port of the linker]
3034 @item --subsystem @var{which}
3035 @itemx --subsystem @var{which}:@var{major}
3036 @itemx --subsystem @var{which}:@var{major}.@var{minor}
3037 Specifies the subsystem under which your program will execute. The
3038 legal values for @var{which} are @code{native}, @code{windows},
3039 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
3040 the subsystem version also. Numeric values are also accepted for
3042 [This option is specific to the i386 PE targeted port of the linker]
3044 The following options set flags in the @code{DllCharacteristics} field
3045 of the PE file header:
3046 [These options are specific to PE targeted ports of the linker]
3048 @kindex --high-entropy-va
3049 @item --high-entropy-va
3050 Image is compatible with 64-bit address space layout randomization
3052 This option also implies @option{--dynamicbase} and
3053 @option{--enable-reloc-section}.
3055 @kindex --dynamicbase
3057 The image base address may be relocated using address space layout
3058 randomization (ASLR). This feature was introduced with MS Windows
3059 Vista for i386 PE targets.
3060 This option also implies @option{--enable-reloc-section}.
3062 @kindex --forceinteg
3064 Code integrity checks are enforced.
3068 The image is compatible with the Data Execution Prevention.
3069 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
3071 @kindex --no-isolation
3072 @item --no-isolation
3073 Although the image understands isolation, do not isolate the image.
3077 The image does not use SEH. No SE handler may be called from
3082 Do not bind this image.
3086 The driver uses the MS Windows Driver Model.
3090 The image is Terminal Server aware.
3092 @kindex --insert-timestamp
3093 @item --insert-timestamp
3094 @itemx --no-insert-timestamp
3095 Insert a real timestamp into the image. This is the default behaviour
3096 as it matches legacy code and it means that the image will work with
3097 other, proprietary tools. The problem with this default is that it
3098 will result in slightly different images being produced each time the
3099 same sources are linked. The option @option{--no-insert-timestamp}
3100 can be used to insert a zero value for the timestamp, this ensuring
3101 that binaries produced from identical sources will compare
3104 @kindex --enable-reloc-section
3105 @item --enable-reloc-section
3106 Create the base relocation table, which is necessary if the image
3107 is loaded at a different image base than specified in the PE header.
3113 @subsection Options specific to C6X uClinux targets
3115 @c man begin OPTIONS
3117 The C6X uClinux target uses a binary format called DSBT to support shared
3118 libraries. Each shared library in the system needs to have a unique index;
3119 all executables use an index of 0.
3124 @item --dsbt-size @var{size}
3125 This option sets the number of entries in the DSBT of the current executable
3126 or shared library to @var{size}. The default is to create a table with 64
3129 @kindex --dsbt-index
3130 @item --dsbt-index @var{index}
3131 This option sets the DSBT index of the current executable or shared library
3132 to @var{index}. The default is 0, which is appropriate for generating
3133 executables. If a shared library is generated with a DSBT index of 0, the
3134 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
3136 @kindex --no-merge-exidx-entries
3137 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
3138 exidx entries in frame unwind info.
3146 @subsection Options specific to C-SKY targets
3148 @c man begin OPTIONS
3152 @kindex --branch-stub on C-SKY
3154 This option enables linker branch relaxation by inserting branch stub
3155 sections when needed to extend the range of branches. This option is
3156 usually not required since C-SKY supports branch and call instructions that
3157 can access the full memory range and branch relaxation is normally handled by
3158 the compiler or assembler.
3160 @kindex --stub-group-size on C-SKY
3161 @item --stub-group-size=@var{N}
3162 This option allows finer control of linker branch stub creation.
3163 It sets the maximum size of a group of input sections that can
3164 be handled by one stub section. A negative value of @var{N} locates
3165 stub sections after their branches, while a positive value allows stub
3166 sections to appear either before or after the branches. Values of
3167 @samp{1} or @samp{-1} indicate that the
3168 linker should choose suitable defaults.
3176 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
3178 @c man begin OPTIONS
3180 The 68HC11 and 68HC12 linkers support specific options to control the
3181 memory bank switching mapping and trampoline code generation.
3185 @kindex --no-trampoline
3186 @item --no-trampoline
3187 This option disables the generation of trampoline. By default a trampoline
3188 is generated for each far function which is called using a @code{jsr}
3189 instruction (this happens when a pointer to a far function is taken).
3191 @kindex --bank-window
3192 @item --bank-window @var{name}
3193 This option indicates to the linker the name of the memory region in
3194 the @samp{MEMORY} specification that describes the memory bank window.
3195 The definition of such region is then used by the linker to compute
3196 paging and addresses within the memory window.
3204 @subsection Options specific to Motorola 68K target
3206 @c man begin OPTIONS
3208 The following options are supported to control handling of GOT generation
3209 when linking for 68K targets.
3214 @item --got=@var{type}
3215 This option tells the linker which GOT generation scheme to use.
3216 @var{type} should be one of @samp{single}, @samp{negative},
3217 @samp{multigot} or @samp{target}. For more information refer to the
3218 Info entry for @file{ld}.
3226 @subsection Options specific to MIPS targets
3228 @c man begin OPTIONS
3230 The following options are supported to control microMIPS instruction
3231 generation and branch relocation checks for ISA mode transitions when
3232 linking for MIPS targets.
3240 These options control the choice of microMIPS instructions used in code
3241 generated by the linker, such as that in the PLT or lazy binding stubs,
3242 or in relaxation. If @samp{--insn32} is used, then the linker only uses
3243 32-bit instruction encodings. By default or if @samp{--no-insn32} is
3244 used, all instruction encodings are used, including 16-bit ones where
3247 @kindex --ignore-branch-isa
3248 @item --ignore-branch-isa
3249 @kindex --no-ignore-branch-isa
3250 @itemx --no-ignore-branch-isa
3251 These options control branch relocation checks for invalid ISA mode
3252 transitions. If @samp{--ignore-branch-isa} is used, then the linker
3253 accepts any branch relocations and any ISA mode transition required
3254 is lost in relocation calculation, except for some cases of @code{BAL}
3255 instructions which meet relaxation conditions and are converted to
3256 equivalent @code{JALX} instructions as the associated relocation is
3257 calculated. By default or if @samp{--no-ignore-branch-isa} is used
3258 a check is made causing the loss of an ISA mode transition to produce
3261 @kindex --compact-branches
3262 @item --compact-branches
3263 @kindex --no-compact-branches
3264 @itemx --no-compact-branches
3265 These options control the generation of compact instructions by the linker
3266 in the PLT entries for MIPS R6.
3275 @subsection Options specific to PDP11 targets
3277 @c man begin OPTIONS
3279 For the pdp11-aout target, three variants of the output format can be
3280 produced as selected by the following options. The default variant
3281 for pdp11-aout is the @samp{--omagic} option, whereas for other
3282 targets @samp{--nmagic} is the default. The @samp{--imagic} option is
3283 defined only for the pdp11-aout target, while the others are described
3284 here as they apply to the pdp11-aout target.
3293 Mark the output as @code{OMAGIC} (0407) in the @file{a.out} header to
3294 indicate that the text segment is not to be write-protected and
3295 shared. Since the text and data sections are both readable and
3296 writable, the data section is allocated immediately contiguous after
3297 the text segment. This is the oldest format for PDP11 executable
3298 programs and is the default for @command{ld} on PDP11 Unix systems
3299 from the beginning through 2.11BSD.
3306 Mark the output as @code{NMAGIC} (0410) in the @file{a.out} header to
3307 indicate that when the output file is executed, the text portion will
3308 be read-only and shareable among all processes executing the same
3309 file. This involves moving the data areas up to the first possible 8K
3310 byte page boundary following the end of the text. This option creates
3311 a @emph{pure executable} format.
3318 Mark the output as @code{IMAGIC} (0411) in the @file{a.out} header to
3319 indicate that when the output file is executed, the program text and
3320 data areas will be loaded into separate address spaces using the split
3321 instruction and data space feature of the memory management unit in
3322 larger models of the PDP11. This doubles the address space available
3323 to the program. The text segment is again pure, write-protected, and
3324 shareable. The only difference in the output format between this
3325 option and the others, besides the magic number, is that both the text
3326 and data sections start at location 0. The @samp{-z} option selected
3327 this format in 2.11BSD. This option creates a @emph{separate
3333 Equivalent to @samp{--nmagic} for pdp11-aout.
3342 @section Environment Variables
3344 @c man begin ENVIRONMENT
3346 You can change the behaviour of @command{ld} with the environment variables
3347 @ifclear SingleFormat
3350 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
3352 @ifclear SingleFormat
3354 @cindex default input format
3355 @code{GNUTARGET} determines the input-file object format if you don't
3356 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
3357 of the BFD names for an input format (@pxref{BFD}). If there is no
3358 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
3359 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
3360 attempts to discover the input format by examining binary input files;
3361 this method often succeeds, but there are potential ambiguities, since
3362 there is no method of ensuring that the magic number used to specify
3363 object-file formats is unique. However, the configuration procedure for
3364 BFD on each system places the conventional format for that system first
3365 in the search-list, so ambiguities are resolved in favor of convention.
3369 @cindex default emulation
3370 @cindex emulation, default
3371 @code{LDEMULATION} determines the default emulation if you don't use the
3372 @samp{-m} option. The emulation can affect various aspects of linker
3373 behaviour, particularly the default linker script. You can list the
3374 available emulations with the @samp{--verbose} or @samp{-V} options. If
3375 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3376 variable is not defined, the default emulation depends upon how the
3377 linker was configured.
3379 @kindex COLLECT_NO_DEMANGLE
3380 @cindex demangling, default
3381 Normally, the linker will default to demangling symbols. However, if
3382 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3383 default to not demangling symbols. This environment variable is used in
3384 a similar fashion by the @code{gcc} linker wrapper program. The default
3385 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3392 @chapter Linker Scripts
3395 @cindex linker scripts
3396 @cindex command files
3397 Every link is controlled by a @dfn{linker script}. This script is
3398 written in the linker command language.
3400 The main purpose of the linker script is to describe how the sections in
3401 the input files should be mapped into the output file, and to control
3402 the memory layout of the output file. Most linker scripts do nothing
3403 more than this. However, when necessary, the linker script can also
3404 direct the linker to perform many other operations, using the commands
3407 The linker always uses a linker script. If you do not supply one
3408 yourself, the linker will use a default script that is compiled into the
3409 linker executable. You can use the @samp{--verbose} command-line option
3410 to display the default linker script. Certain command-line options,
3411 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3413 You may supply your own linker script by using the @samp{-T} command
3414 line option. When you do this, your linker script will replace the
3415 default linker script.
3417 You may also use linker scripts implicitly by naming them as input files
3418 to the linker, as though they were files to be linked. @xref{Implicit
3422 * Basic Script Concepts:: Basic Linker Script Concepts
3423 * Script Format:: Linker Script Format
3424 * Simple Example:: Simple Linker Script Example
3425 * Simple Commands:: Simple Linker Script Commands
3426 * Assignments:: Assigning Values to Symbols
3427 * SECTIONS:: SECTIONS Command
3428 * MEMORY:: MEMORY Command
3429 * PHDRS:: PHDRS Command
3430 * VERSION:: VERSION Command
3431 * Expressions:: Expressions in Linker Scripts
3432 * Implicit Linker Scripts:: Implicit Linker Scripts
3435 @node Basic Script Concepts
3436 @section Basic Linker Script Concepts
3437 @cindex linker script concepts
3438 We need to define some basic concepts and vocabulary in order to
3439 describe the linker script language.
3441 The linker combines input files into a single output file. The output
3442 file and each input file are in a special data format known as an
3443 @dfn{object file format}. Each file is called an @dfn{object file}.
3444 The output file is often called an @dfn{executable}, but for our
3445 purposes we will also call it an object file. Each object file has,
3446 among other things, a list of @dfn{sections}. We sometimes refer to a
3447 section in an input file as an @dfn{input section}; similarly, a section
3448 in the output file is an @dfn{output section}.
3450 Each section in an object file has a name and a size. Most sections
3451 also have an associated block of data, known as the @dfn{section
3452 contents}. A section may be marked as @dfn{loadable}, which means that
3453 the contents should be loaded into memory when the output file is run.
3454 A section with no contents may be @dfn{allocatable}, which means that an
3455 area in memory should be set aside, but nothing in particular should be
3456 loaded there (in some cases this memory must be zeroed out). A section
3457 which is neither loadable nor allocatable typically contains some sort
3458 of debugging information.
3460 Every loadable or allocatable output section has two addresses. The
3461 first is the @dfn{VMA}, or virtual memory address. This is the address
3462 the section will have when the output file is run. The second is the
3463 @dfn{LMA}, or load memory address. This is the address at which the
3464 section will be loaded. In most cases the two addresses will be the
3465 same. An example of when they might be different is when a data section
3466 is loaded into ROM, and then copied into RAM when the program starts up
3467 (this technique is often used to initialize global variables in a ROM
3468 based system). In this case the ROM address would be the LMA, and the
3469 RAM address would be the VMA.
3471 You can see the sections in an object file by using the @code{objdump}
3472 program with the @samp{-h} option.
3474 Every object file also has a list of @dfn{symbols}, known as the
3475 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3476 has a name, and each defined symbol has an address, among other
3477 information. If you compile a C or C++ program into an object file, you
3478 will get a defined symbol for every defined function and global or
3479 static variable. Every undefined function or global variable which is
3480 referenced in the input file will become an undefined symbol.
3482 You can see the symbols in an object file by using the @code{nm}
3483 program, or by using the @code{objdump} program with the @samp{-t}
3487 @section Linker Script Format
3488 @cindex linker script format
3489 Linker scripts are text files.
3491 You write a linker script as a series of commands. Each command is
3492 either a keyword, possibly followed by arguments, or an assignment to a
3493 symbol. You may separate commands using semicolons. Whitespace is
3496 Strings such as file or format names can normally be entered directly.
3497 If the file name contains a character such as a comma which would
3498 otherwise serve to separate file names, you may put the file name in
3499 double quotes. There is no way to use a double quote character in a
3502 You may include comments in linker scripts just as in C, delimited by
3503 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3506 @node Simple Example
3507 @section Simple Linker Script Example
3508 @cindex linker script example
3509 @cindex example of linker script
3510 Many linker scripts are fairly simple.
3512 The simplest possible linker script has just one command:
3513 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3514 memory layout of the output file.
3516 The @samp{SECTIONS} command is a powerful command. Here we will
3517 describe a simple use of it. Let's assume your program consists only of
3518 code, initialized data, and uninitialized data. These will be in the
3519 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3520 Let's assume further that these are the only sections which appear in
3523 For this example, let's say that the code should be loaded at address
3524 0x10000, and that the data should start at address 0x8000000. Here is a
3525 linker script which will do that:
3530 .text : @{ *(.text) @}
3532 .data : @{ *(.data) @}
3533 .bss : @{ *(.bss) @}
3537 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3538 followed by a series of symbol assignments and output section
3539 descriptions enclosed in curly braces.
3541 The first line inside the @samp{SECTIONS} command of the above example
3542 sets the value of the special symbol @samp{.}, which is the location
3543 counter. If you do not specify the address of an output section in some
3544 other way (other ways are described later), the address is set from the
3545 current value of the location counter. The location counter is then
3546 incremented by the size of the output section. At the start of the
3547 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3549 The second line defines an output section, @samp{.text}. The colon is
3550 required syntax which may be ignored for now. Within the curly braces
3551 after the output section name, you list the names of the input sections
3552 which should be placed into this output section. The @samp{*} is a
3553 wildcard which matches any file name. The expression @samp{*(.text)}
3554 means all @samp{.text} input sections in all input files.
3556 Since the location counter is @samp{0x10000} when the output section
3557 @samp{.text} is defined, the linker will set the address of the
3558 @samp{.text} section in the output file to be @samp{0x10000}.
3560 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3561 the output file. The linker will place the @samp{.data} output section
3562 at address @samp{0x8000000}. After the linker places the @samp{.data}
3563 output section, the value of the location counter will be
3564 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3565 effect is that the linker will place the @samp{.bss} output section
3566 immediately after the @samp{.data} output section in memory.
3568 The linker will ensure that each output section has the required
3569 alignment, by increasing the location counter if necessary. In this
3570 example, the specified addresses for the @samp{.text} and @samp{.data}
3571 sections will probably satisfy any alignment constraints, but the linker
3572 may have to create a small gap between the @samp{.data} and @samp{.bss}
3575 That's it! That's a simple and complete linker script.
3577 @node Simple Commands
3578 @section Simple Linker Script Commands
3579 @cindex linker script simple commands
3580 In this section we describe the simple linker script commands.
3583 * Entry Point:: Setting the entry point
3584 * File Commands:: Commands dealing with files
3585 @ifclear SingleFormat
3586 * Format Commands:: Commands dealing with object file formats
3589 * REGION_ALIAS:: Assign alias names to memory regions
3590 * Miscellaneous Commands:: Other linker script commands
3594 @subsection Setting the Entry Point
3595 @kindex ENTRY(@var{symbol})
3596 @cindex start of execution
3597 @cindex first instruction
3599 The first instruction to execute in a program is called the @dfn{entry
3600 point}. You can use the @code{ENTRY} linker script command to set the
3601 entry point. The argument is a symbol name:
3606 There are several ways to set the entry point. The linker will set the
3607 entry point by trying each of the following methods in order, and
3608 stopping when one of them succeeds:
3611 the @samp{-e} @var{entry} command-line option;
3613 the @code{ENTRY(@var{symbol})} command in a linker script;
3615 the value of a target-specific symbol, if it is defined; For many
3616 targets this is @code{start}, but PE- and BeOS-based systems for example
3617 check a list of possible entry symbols, matching the first one found.
3619 the address of the first byte of the @samp{.text} section, if present;
3621 The address @code{0}.
3625 @subsection Commands Dealing with Files
3626 @cindex linker script file commands
3627 Several linker script commands deal with files.
3630 @item INCLUDE @var{filename}
3631 @kindex INCLUDE @var{filename}
3632 @cindex including a linker script
3633 Include the linker script @var{filename} at this point. The file will
3634 be searched for in the current directory, and in any directory specified
3635 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3638 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3639 @code{SECTIONS} commands, or in output section descriptions.
3641 @item INPUT(@var{file}, @var{file}, @dots{})
3642 @itemx INPUT(@var{file} @var{file} @dots{})
3643 @kindex INPUT(@var{files})
3644 @cindex input files in linker scripts
3645 @cindex input object files in linker scripts
3646 @cindex linker script input object files
3647 The @code{INPUT} command directs the linker to include the named files
3648 in the link, as though they were named on the command line.
3650 For example, if you always want to include @file{subr.o} any time you do
3651 a link, but you can't be bothered to put it on every link command line,
3652 then you can put @samp{INPUT (subr.o)} in your linker script.
3654 In fact, if you like, you can list all of your input files in the linker
3655 script, and then invoke the linker with nothing but a @samp{-T} option.
3657 In case a @dfn{sysroot prefix} is configured, and the filename starts
3658 with the @samp{/} character, and the script being processed was
3659 located inside the @dfn{sysroot prefix}, the filename will be looked
3660 for in the @dfn{sysroot prefix}. The @dfn{sysroot prefix} can also be forced by specifying
3661 @code{=} as the first character in the filename path, or prefixing the
3662 filename path with @code{$SYSROOT}. See also the description of
3663 @samp{-L} in @ref{Options,,Command-line Options}.
3665 If a @dfn{sysroot prefix} is not used then the linker will try to open
3666 the file in the directory containing the linker script. If it is not
3667 found the linker will then search the current directory. If it is still
3668 not found the linker will search through the archive library search
3671 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3672 name to @code{lib@var{file}.a}, as with the command-line argument
3675 When you use the @code{INPUT} command in an implicit linker script, the
3676 files will be included in the link at the point at which the linker
3677 script file is included. This can affect archive searching.
3679 @item GROUP(@var{file}, @var{file}, @dots{})
3680 @itemx GROUP(@var{file} @var{file} @dots{})
3681 @kindex GROUP(@var{files})
3682 @cindex grouping input files
3683 The @code{GROUP} command is like @code{INPUT}, except that the named
3684 files should all be archives, and they are searched repeatedly until no
3685 new undefined references are created. See the description of @samp{-(}
3686 in @ref{Options,,Command-line Options}.
3688 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3689 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3690 @kindex AS_NEEDED(@var{files})
3691 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3692 commands, among other filenames. The files listed will be handled
3693 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3694 with the exception of ELF shared libraries, that will be added only
3695 when they are actually needed. This construct essentially enables
3696 @option{--as-needed} option for all the files listed inside of it
3697 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3700 @item OUTPUT(@var{filename})
3701 @kindex OUTPUT(@var{filename})
3702 @cindex output file name in linker script
3703 The @code{OUTPUT} command names the output file. Using
3704 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3705 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3706 Line Options}). If both are used, the command-line option takes
3709 You can use the @code{OUTPUT} command to define a default name for the
3710 output file other than the usual default of @file{a.out}.
3712 @item SEARCH_DIR(@var{path})
3713 @kindex SEARCH_DIR(@var{path})
3714 @cindex library search path in linker script
3715 @cindex archive search path in linker script
3716 @cindex search path in linker script
3717 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3718 @command{ld} looks for archive libraries. Using
3719 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3720 on the command line (@pxref{Options,,Command-line Options}). If both
3721 are used, then the linker will search both paths. Paths specified using
3722 the command-line option are searched first.
3724 @item STARTUP(@var{filename})
3725 @kindex STARTUP(@var{filename})
3726 @cindex first input file
3727 The @code{STARTUP} command is just like the @code{INPUT} command, except
3728 that @var{filename} will become the first input file to be linked, as
3729 though it were specified first on the command line. This may be useful
3730 when using a system in which the entry point is always the start of the
3734 @ifclear SingleFormat
3735 @node Format Commands
3736 @subsection Commands Dealing with Object File Formats
3737 A couple of linker script commands deal with object file formats.
3740 @item OUTPUT_FORMAT(@var{bfdname})
3741 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3742 @kindex OUTPUT_FORMAT(@var{bfdname})
3743 @cindex output file format in linker script
3744 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3745 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3746 exactly like using @samp{--oformat @var{bfdname}} on the command line
3747 (@pxref{Options,,Command-line Options}). If both are used, the command
3748 line option takes precedence.
3750 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3751 formats based on the @samp{-EB} and @samp{-EL} command-line options.
3752 This permits the linker script to set the output format based on the
3755 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3756 will be the first argument, @var{default}. If @samp{-EB} is used, the
3757 output format will be the second argument, @var{big}. If @samp{-EL} is
3758 used, the output format will be the third argument, @var{little}.
3760 For example, the default linker script for the MIPS ELF target uses this
3763 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3765 This says that the default format for the output file is
3766 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command-line
3767 option, the output file will be created in the @samp{elf32-littlemips}
3770 @item TARGET(@var{bfdname})
3771 @kindex TARGET(@var{bfdname})
3772 @cindex input file format in linker script
3773 The @code{TARGET} command names the BFD format to use when reading input
3774 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3775 This command is like using @samp{-b @var{bfdname}} on the command line
3776 (@pxref{Options,,Command-line Options}). If the @code{TARGET} command
3777 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3778 command is also used to set the format for the output file. @xref{BFD}.
3783 @subsection Assign alias names to memory regions
3784 @kindex REGION_ALIAS(@var{alias}, @var{region})
3785 @cindex region alias
3786 @cindex region names
3788 Alias names can be added to existing memory regions created with the
3789 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3792 REGION_ALIAS(@var{alias}, @var{region})
3795 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3796 memory region @var{region}. This allows a flexible mapping of output sections
3797 to memory regions. An example follows.
3799 Suppose we have an application for embedded systems which come with various
3800 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3801 that allows code execution or data storage. Some may have a read-only,
3802 non-volatile memory @code{ROM} that allows code execution and read-only data
3803 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3804 read-only data access and no code execution capability. We have four output
3809 @code{.text} program code;
3811 @code{.rodata} read-only data;
3813 @code{.data} read-write initialized data;
3815 @code{.bss} read-write zero initialized data.
3818 The goal is to provide a linker command file that contains a system independent
3819 part defining the output sections and a system dependent part mapping the
3820 output sections to the memory regions available on the system. Our embedded
3821 systems come with three different memory setups @code{A}, @code{B} and
3823 @multitable @columnfractions .25 .25 .25 .25
3824 @item Section @tab Variant A @tab Variant B @tab Variant C
3825 @item .text @tab RAM @tab ROM @tab ROM
3826 @item .rodata @tab RAM @tab ROM @tab ROM2
3827 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3828 @item .bss @tab RAM @tab RAM @tab RAM
3830 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3831 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3832 the load address of the @code{.data} section starts in all three variants at
3833 the end of the @code{.rodata} section.
3835 The base linker script that deals with the output sections follows. It
3836 includes the system dependent @code{linkcmds.memory} file that describes the
3839 INCLUDE linkcmds.memory
3852 .data : AT (rodata_end)
3857 data_size = SIZEOF(.data);
3858 data_load_start = LOADADDR(.data);
3866 Now we need three different @code{linkcmds.memory} files to define memory
3867 regions and alias names. The content of @code{linkcmds.memory} for the three
3868 variants @code{A}, @code{B} and @code{C}:
3871 Here everything goes into the @code{RAM}.
3875 RAM : ORIGIN = 0, LENGTH = 4M
3878 REGION_ALIAS("REGION_TEXT", RAM);
3879 REGION_ALIAS("REGION_RODATA", RAM);
3880 REGION_ALIAS("REGION_DATA", RAM);
3881 REGION_ALIAS("REGION_BSS", RAM);
3884 Program code and read-only data go into the @code{ROM}. Read-write data goes
3885 into the @code{RAM}. An image of the initialized data is loaded into the
3886 @code{ROM} and will be copied during system start into the @code{RAM}.
3890 ROM : ORIGIN = 0, LENGTH = 3M
3891 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3894 REGION_ALIAS("REGION_TEXT", ROM);
3895 REGION_ALIAS("REGION_RODATA", ROM);
3896 REGION_ALIAS("REGION_DATA", RAM);
3897 REGION_ALIAS("REGION_BSS", RAM);
3900 Program code goes into the @code{ROM}. Read-only data goes into the
3901 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3902 initialized data is loaded into the @code{ROM2} and will be copied during
3903 system start into the @code{RAM}.
3907 ROM : ORIGIN = 0, LENGTH = 2M
3908 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3909 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3912 REGION_ALIAS("REGION_TEXT", ROM);
3913 REGION_ALIAS("REGION_RODATA", ROM2);
3914 REGION_ALIAS("REGION_DATA", RAM);
3915 REGION_ALIAS("REGION_BSS", RAM);
3919 It is possible to write a common system initialization routine to copy the
3920 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3925 extern char data_start [];
3926 extern char data_size [];
3927 extern char data_load_start [];
3929 void copy_data(void)
3931 if (data_start != data_load_start)
3933 memcpy(data_start, data_load_start, (size_t) data_size);
3938 @node Miscellaneous Commands
3939 @subsection Other Linker Script Commands
3940 There are a few other linker scripts commands.
3943 @item ASSERT(@var{exp}, @var{message})
3945 @cindex assertion in linker script
3946 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3947 with an error code, and print @var{message}.
3949 Note that assertions are checked before the final stages of linking
3950 take place. This means that expressions involving symbols PROVIDEd
3951 inside section definitions will fail if the user has not set values
3952 for those symbols. The only exception to this rule is PROVIDEd
3953 symbols that just reference dot. Thus an assertion like this:
3958 PROVIDE (__stack = .);
3959 PROVIDE (__stack_size = 0x100);
3960 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3964 will fail if @code{__stack_size} is not defined elsewhere. Symbols
3965 PROVIDEd outside of section definitions are evaluated earlier, so they
3966 can be used inside ASSERTions. Thus:
3969 PROVIDE (__stack_size = 0x100);
3972 PROVIDE (__stack = .);
3973 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3979 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3981 @cindex undefined symbol in linker script
3982 Force @var{symbol} to be entered in the output file as an undefined
3983 symbol. Doing this may, for example, trigger linking of additional
3984 modules from standard libraries. You may list several @var{symbol}s for
3985 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3986 command has the same effect as the @samp{-u} command-line option.
3988 @item FORCE_COMMON_ALLOCATION
3989 @kindex FORCE_COMMON_ALLOCATION
3990 @cindex common allocation in linker script
3991 This command has the same effect as the @samp{-d} command-line option:
3992 to make @command{ld} assign space to common symbols even if a relocatable
3993 output file is specified (@samp{-r}).
3995 @item INHIBIT_COMMON_ALLOCATION
3996 @kindex INHIBIT_COMMON_ALLOCATION
3997 @cindex common allocation in linker script
3998 This command has the same effect as the @samp{--no-define-common}
3999 command-line option: to make @code{ld} omit the assignment of addresses
4000 to common symbols even for a non-relocatable output file.
4002 @item FORCE_GROUP_ALLOCATION
4003 @kindex FORCE_GROUP_ALLOCATION
4004 @cindex group allocation in linker script
4005 @cindex section groups
4007 This command has the same effect as the
4008 @samp{--force-group-allocation} command-line option: to make
4009 @command{ld} place section group members like normal input sections,
4010 and to delete the section groups even if a relocatable output file is
4011 specified (@samp{-r}).
4013 @item INSERT [ AFTER | BEFORE ] @var{output_section}
4015 @cindex insert user script into default script
4016 This command is typically used in a script specified by @samp{-T} to
4017 augment the default @code{SECTIONS} with, for example, overlays. It
4018 inserts all prior linker script statements after (or before)
4019 @var{output_section}, and also causes @samp{-T} to not override the
4020 default linker script. The exact insertion point is as for orphan
4021 sections. @xref{Location Counter}. The insertion happens after the
4022 linker has mapped input sections to output sections. Prior to the
4023 insertion, since @samp{-T} scripts are parsed before the default
4024 linker script, statements in the @samp{-T} script occur before the
4025 default linker script statements in the internal linker representation
4026 of the script. In particular, input section assignments will be made
4027 to @samp{-T} output sections before those in the default script. Here
4028 is an example of how a @samp{-T} script using @code{INSERT} might look:
4035 .ov1 @{ ov1*(.text) @}
4036 .ov2 @{ ov2*(.text) @}
4042 @item NOCROSSREFS(@var{section} @var{section} @dots{})
4043 @kindex NOCROSSREFS(@var{sections})
4044 @cindex cross references
4045 This command may be used to tell @command{ld} to issue an error about any
4046 references among certain output sections.
4048 In certain types of programs, particularly on embedded systems when
4049 using overlays, when one section is loaded into memory, another section
4050 will not be. Any direct references between the two sections would be
4051 errors. For example, it would be an error if code in one section called
4052 a function defined in the other section.
4054 The @code{NOCROSSREFS} command takes a list of output section names. If
4055 @command{ld} detects any cross references between the sections, it reports
4056 an error and returns a non-zero exit status. Note that the
4057 @code{NOCROSSREFS} command uses output section names, not input section
4060 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
4061 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
4062 @cindex cross references
4063 This command may be used to tell @command{ld} to issue an error about any
4064 references to one section from a list of other sections.
4066 The @code{NOCROSSREFS} command is useful when ensuring that two or more
4067 output sections are entirely independent but there are situations where
4068 a one-way dependency is needed. For example, in a multi-core application
4069 there may be shared code that can be called from each core but for safety
4070 must never call back.
4072 The @code{NOCROSSREFS_TO} command takes a list of output section names.
4073 The first section can not be referenced from any of the other sections.
4074 If @command{ld} detects any references to the first section from any of
4075 the other sections, it reports an error and returns a non-zero exit
4076 status. Note that the @code{NOCROSSREFS_TO} command uses output section
4077 names, not input section names.
4079 @ifclear SingleFormat
4080 @item OUTPUT_ARCH(@var{bfdarch})
4081 @kindex OUTPUT_ARCH(@var{bfdarch})
4082 @cindex machine architecture
4083 @cindex architecture
4084 Specify a particular output machine architecture. The argument is one
4085 of the names used by the BFD library (@pxref{BFD}). You can see the
4086 architecture of an object file by using the @code{objdump} program with
4087 the @samp{-f} option.
4090 @item LD_FEATURE(@var{string})
4091 @kindex LD_FEATURE(@var{string})
4092 This command may be used to modify @command{ld} behavior. If
4093 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
4094 in a script are simply treated as numbers everywhere.
4095 @xref{Expression Section}.
4099 @section Assigning Values to Symbols
4100 @cindex assignment in scripts
4101 @cindex symbol definition, scripts
4102 @cindex variables, defining
4103 You may assign a value to a symbol in a linker script. This will define
4104 the symbol and place it into the symbol table with a global scope.
4107 * Simple Assignments:: Simple Assignments
4110 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
4111 * Source Code Reference:: How to use a linker script defined symbol in source code
4114 @node Simple Assignments
4115 @subsection Simple Assignments
4117 You may assign to a symbol using any of the C assignment operators:
4120 @item @var{symbol} = @var{expression} ;
4121 @itemx @var{symbol} += @var{expression} ;
4122 @itemx @var{symbol} -= @var{expression} ;
4123 @itemx @var{symbol} *= @var{expression} ;
4124 @itemx @var{symbol} /= @var{expression} ;
4125 @itemx @var{symbol} <<= @var{expression} ;
4126 @itemx @var{symbol} >>= @var{expression} ;
4127 @itemx @var{symbol} &= @var{expression} ;
4128 @itemx @var{symbol} |= @var{expression} ;
4131 The first case will define @var{symbol} to the value of
4132 @var{expression}. In the other cases, @var{symbol} must already be
4133 defined, and the value will be adjusted accordingly.
4135 The special symbol name @samp{.} indicates the location counter. You
4136 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
4138 The semicolon after @var{expression} is required.
4140 Expressions are defined below; see @ref{Expressions}.
4142 You may write symbol assignments as commands in their own right, or as
4143 statements within a @code{SECTIONS} command, or as part of an output
4144 section description in a @code{SECTIONS} command.
4146 The section of the symbol will be set from the section of the
4147 expression; for more information, see @ref{Expression Section}.
4149 Here is an example showing the three different places that symbol
4150 assignments may be used:
4161 _bdata = (. + 3) & ~ 3;
4162 .data : @{ *(.data) @}
4166 In this example, the symbol @samp{floating_point} will be defined as
4167 zero. The symbol @samp{_etext} will be defined as the address following
4168 the last @samp{.text} input section. The symbol @samp{_bdata} will be
4169 defined as the address following the @samp{.text} output section aligned
4170 upward to a 4 byte boundary.
4175 For ELF targeted ports, define a symbol that will be hidden and won't be
4176 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
4178 Here is the example from @ref{Simple Assignments}, rewritten to use
4182 HIDDEN(floating_point = 0);
4190 HIDDEN(_bdata = (. + 3) & ~ 3);
4191 .data : @{ *(.data) @}
4195 In this case none of the three symbols will be visible outside this module.
4200 In some cases, it is desirable for a linker script to define a symbol
4201 only if it is referenced and is not defined by any object included in
4202 the link. For example, traditional linkers defined the symbol
4203 @samp{etext}. However, ANSI C requires that the user be able to use
4204 @samp{etext} as a function name without encountering an error. The
4205 @code{PROVIDE} keyword may be used to define a symbol, such as
4206 @samp{etext}, only if it is referenced but not defined. The syntax is
4207 @code{PROVIDE(@var{symbol} = @var{expression})}.
4209 Here is an example of using @code{PROVIDE} to define @samp{etext}:
4222 In this example, if the program defines @samp{_etext} (with a leading
4223 underscore), the linker will give a multiple definition error. If, on
4224 the other hand, the program defines @samp{etext} (with no leading
4225 underscore), the linker will silently use the definition in the program.
4226 If the program references @samp{etext} but does not define it, the
4227 linker will use the definition in the linker script.
4229 Note - the @code{PROVIDE} directive considers a common symbol to be
4230 defined, even though such a symbol could be combined with the symbol
4231 that the @code{PROVIDE} would create. This is particularly important
4232 when considering constructor and destructor list symbols such as
4233 @samp{__CTOR_LIST__} as these are often defined as common symbols.
4235 @node PROVIDE_HIDDEN
4236 @subsection PROVIDE_HIDDEN
4237 @cindex PROVIDE_HIDDEN
4238 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
4239 hidden and won't be exported.
4241 @node Source Code Reference
4242 @subsection Source Code Reference
4244 Accessing a linker script defined variable from source code is not
4245 intuitive. In particular a linker script symbol is not equivalent to
4246 a variable declaration in a high level language, it is instead a
4247 symbol that does not have a value.
4249 Before going further, it is important to note that compilers often
4250 transform names in the source code into different names when they are
4251 stored in the symbol table. For example, Fortran compilers commonly
4252 prepend or append an underscore, and C++ performs extensive @samp{name
4253 mangling}. Therefore there might be a discrepancy between the name
4254 of a variable as it is used in source code and the name of the same
4255 variable as it is defined in a linker script. For example in C a
4256 linker script variable might be referred to as:
4262 But in the linker script it might be defined as:
4268 In the remaining examples however it is assumed that no name
4269 transformation has taken place.
4271 When a symbol is declared in a high level language such as C, two
4272 things happen. The first is that the compiler reserves enough space
4273 in the program's memory to hold the @emph{value} of the symbol. The
4274 second is that the compiler creates an entry in the program's symbol
4275 table which holds the symbol's @emph{address}. ie the symbol table
4276 contains the address of the block of memory holding the symbol's
4277 value. So for example the following C declaration, at file scope:
4283 creates an entry called @samp{foo} in the symbol table. This entry
4284 holds the address of an @samp{int} sized block of memory where the
4285 number 1000 is initially stored.
4287 When a program references a symbol the compiler generates code that
4288 first accesses the symbol table to find the address of the symbol's
4289 memory block and then code to read the value from that memory block.
4296 looks up the symbol @samp{foo} in the symbol table, gets the address
4297 associated with this symbol and then writes the value 1 into that
4304 looks up the symbol @samp{foo} in the symbol table, gets its address
4305 and then copies this address into the block of memory associated with
4306 the variable @samp{a}.
4308 Linker scripts symbol declarations, by contrast, create an entry in
4309 the symbol table but do not assign any memory to them. Thus they are
4310 an address without a value. So for example the linker script definition:
4316 creates an entry in the symbol table called @samp{foo} which holds
4317 the address of memory location 1000, but nothing special is stored at
4318 address 1000. This means that you cannot access the @emph{value} of a
4319 linker script defined symbol - it has no value - all you can do is
4320 access the @emph{address} of a linker script defined symbol.
4322 Hence when you are using a linker script defined symbol in source code
4323 you should always take the address of the symbol, and never attempt to
4324 use its value. For example suppose you want to copy the contents of a
4325 section of memory called .ROM into a section called .FLASH and the
4326 linker script contains these declarations:
4330 start_of_ROM = .ROM;
4331 end_of_ROM = .ROM + sizeof (.ROM);
4332 start_of_FLASH = .FLASH;
4336 Then the C source code to perform the copy would be:
4340 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
4342 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
4346 Note the use of the @samp{&} operators. These are correct.
4347 Alternatively the symbols can be treated as the names of vectors or
4348 arrays and then the code will again work as expected:
4352 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
4354 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
4358 Note how using this method does not require the use of @samp{&}
4362 @section SECTIONS Command
4364 The @code{SECTIONS} command tells the linker how to map input sections
4365 into output sections, and how to place the output sections in memory.
4367 The format of the @code{SECTIONS} command is:
4371 @var{sections-command}
4372 @var{sections-command}
4377 Each @var{sections-command} may of be one of the following:
4381 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
4383 a symbol assignment (@pxref{Assignments})
4385 an output section description
4387 an overlay description
4390 The @code{ENTRY} command and symbol assignments are permitted inside the
4391 @code{SECTIONS} command for convenience in using the location counter in
4392 those commands. This can also make the linker script easier to
4393 understand because you can use those commands at meaningful points in
4394 the layout of the output file.
4396 Output section descriptions and overlay descriptions are described
4399 If you do not use a @code{SECTIONS} command in your linker script, the
4400 linker will place each input section into an identically named output
4401 section in the order that the sections are first encountered in the
4402 input files. If all input sections are present in the first file, for
4403 example, the order of sections in the output file will match the order
4404 in the first input file. The first section will be at address zero.
4407 * Output Section Description:: Output section description
4408 * Output Section Name:: Output section name
4409 * Output Section Address:: Output section address
4410 * Input Section:: Input section description
4411 * Output Section Data:: Output section data
4412 * Output Section Keywords:: Output section keywords
4413 * Output Section Discarding:: Output section discarding
4414 * Output Section Attributes:: Output section attributes
4415 * Overlay Description:: Overlay description
4418 @node Output Section Description
4419 @subsection Output Section Description
4420 The full description of an output section looks like this:
4423 @var{section} [@var{address}] [(@var{type})] :
4425 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4426 [SUBALIGN(@var{subsection_align})]
4429 @var{output-section-command}
4430 @var{output-section-command}
4432 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4436 Most output sections do not use most of the optional section attributes.
4438 The whitespace around @var{section} is required, so that the section
4439 name is unambiguous. The colon and the curly braces are also required.
4440 The comma at the end may be required if a @var{fillexp} is used and
4441 the next @var{sections-command} looks like a continuation of the expression.
4442 The line breaks and other white space are optional.
4444 Each @var{output-section-command} may be one of the following:
4448 a symbol assignment (@pxref{Assignments})
4450 an input section description (@pxref{Input Section})
4452 data values to include directly (@pxref{Output Section Data})
4454 a special output section keyword (@pxref{Output Section Keywords})
4457 @node Output Section Name
4458 @subsection Output Section Name
4459 @cindex name, section
4460 @cindex section name
4461 The name of the output section is @var{section}. @var{section} must
4462 meet the constraints of your output format. In formats which only
4463 support a limited number of sections, such as @code{a.out}, the name
4464 must be one of the names supported by the format (@code{a.out}, for
4465 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4466 output format supports any number of sections, but with numbers and not
4467 names (as is the case for Oasys), the name should be supplied as a
4468 quoted numeric string. A section name may consist of any sequence of
4469 characters, but a name which contains any unusual characters such as
4470 commas must be quoted.
4472 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4475 @node Output Section Address
4476 @subsection Output Section Address
4477 @cindex address, section
4478 @cindex section address
4479 The @var{address} is an expression for the VMA (the virtual memory
4480 address) of the output section. This address is optional, but if it
4481 is provided then the output address will be set exactly as specified.
4483 If the output address is not specified then one will be chosen for the
4484 section, based on the heuristic below. This address will be adjusted
4485 to fit the alignment requirement of the output section. The
4486 alignment requirement is the strictest alignment of any input section
4487 contained within the output section.
4489 The output section address heuristic is as follows:
4493 If an output memory @var{region} is set for the section then it
4494 is added to this region and its address will be the next free address
4498 If the MEMORY command has been used to create a list of memory
4499 regions then the first region which has attributes compatible with the
4500 section is selected to contain it. The section's output address will
4501 be the next free address in that region; @ref{MEMORY}.
4504 If no memory regions were specified, or none match the section then
4505 the output address will be based on the current value of the location
4513 .text . : @{ *(.text) @}
4520 .text : @{ *(.text) @}
4524 are subtly different. The first will set the address of the
4525 @samp{.text} output section to the current value of the location
4526 counter. The second will set it to the current value of the location
4527 counter aligned to the strictest alignment of any of the @samp{.text}
4530 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4531 For example, if you want to align the section on a 0x10 byte boundary,
4532 so that the lowest four bits of the section address are zero, you could
4533 do something like this:
4535 .text ALIGN(0x10) : @{ *(.text) @}
4538 This works because @code{ALIGN} returns the current location counter
4539 aligned upward to the specified value.
4541 Specifying @var{address} for a section will change the value of the
4542 location counter, provided that the section is non-empty. (Empty
4543 sections are ignored).
4546 @subsection Input Section Description
4547 @cindex input sections
4548 @cindex mapping input sections to output sections
4549 The most common output section command is an input section description.
4551 The input section description is the most basic linker script operation.
4552 You use output sections to tell the linker how to lay out your program
4553 in memory. You use input section descriptions to tell the linker how to
4554 map the input files into your memory layout.
4557 * Input Section Basics:: Input section basics
4558 * Input Section Wildcards:: Input section wildcard patterns
4559 * Input Section Common:: Input section for common symbols
4560 * Input Section Keep:: Input section and garbage collection
4561 * Input Section Example:: Input section example
4564 @node Input Section Basics
4565 @subsubsection Input Section Basics
4566 @cindex input section basics
4567 An input section description consists of a file name optionally followed
4568 by a list of section names in parentheses.
4570 The file name and the section name may be wildcard patterns, which we
4571 describe further below (@pxref{Input Section Wildcards}).
4573 The most common input section description is to include all input
4574 sections with a particular name in the output section. For example, to
4575 include all input @samp{.text} sections, you would write:
4580 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4581 @cindex EXCLUDE_FILE
4582 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4583 match all files except the ones specified in the EXCLUDE_FILE list. For
4586 EXCLUDE_FILE (*crtend.o *otherfile.o) *(.ctors)
4589 will cause all .ctors sections from all files except @file{crtend.o}
4590 and @file{otherfile.o} to be included. The EXCLUDE_FILE can also be
4591 placed inside the section list, for example:
4593 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4596 The result of this is identically to the previous example. Supporting
4597 two syntaxes for EXCLUDE_FILE is useful if the section list contains
4598 more than one section, as described below.
4600 There are two ways to include more than one section:
4606 The difference between these is the order in which the @samp{.text} and
4607 @samp{.rdata} input sections will appear in the output section. In the
4608 first example, they will be intermingled, appearing in the same order as
4609 they are found in the linker input. In the second example, all
4610 @samp{.text} input sections will appear first, followed by all
4611 @samp{.rdata} input sections.
4613 When using EXCLUDE_FILE with more than one section, if the exclusion
4614 is within the section list then the exclusion only applies to the
4615 immediately following section, for example:
4617 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4620 will cause all @samp{.text} sections from all files except
4621 @file{somefile.o} to be included, while all @samp{.rdata} sections
4622 from all files, including @file{somefile.o}, will be included. To
4623 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4624 could be modified to:
4626 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4629 Alternatively, placing the EXCLUDE_FILE outside of the section list,
4630 before the input file selection, will cause the exclusion to apply for
4631 all sections. Thus the previous example can be rewritten as:
4633 EXCLUDE_FILE (*somefile.o) *(.text .rdata)
4636 You can specify a file name to include sections from a particular file.
4637 You would do this if one or more of your files contain special data that
4638 needs to be at a particular location in memory. For example:
4643 To refine the sections that are included based on the section flags
4644 of an input section, INPUT_SECTION_FLAGS may be used.
4646 Here is a simple example for using Section header flags for ELF sections:
4651 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4652 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4657 In this example, the output section @samp{.text} will be comprised of any
4658 input section matching the name *(.text) whose section header flags
4659 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4660 @samp{.text2} will be comprised of any input section matching the name *(.text)
4661 whose section header flag @code{SHF_WRITE} is clear.
4663 You can also specify files within archives by writing a pattern
4664 matching the archive, a colon, then the pattern matching the file,
4665 with no whitespace around the colon.
4669 matches file within archive
4671 matches the whole archive
4673 matches file but not one in an archive
4676 Either one or both of @samp{archive} and @samp{file} can contain shell
4677 wildcards. On DOS based file systems, the linker will assume that a
4678 single letter followed by a colon is a drive specifier, so
4679 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4680 within an archive called @samp{c}. @samp{archive:file} filespecs may
4681 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4682 other linker script contexts. For instance, you cannot extract a file
4683 from an archive by using @samp{archive:file} in an @code{INPUT}
4686 If you use a file name without a list of sections, then all sections in
4687 the input file will be included in the output section. This is not
4688 commonly done, but it may by useful on occasion. For example:
4693 When you use a file name which is not an @samp{archive:file} specifier
4694 and does not contain any wild card
4695 characters, the linker will first see if you also specified the file
4696 name on the linker command line or in an @code{INPUT} command. If you
4697 did not, the linker will attempt to open the file as an input file, as
4698 though it appeared on the command line. Note that this differs from an
4699 @code{INPUT} command, because the linker will not search for the file in
4700 the archive search path.
4702 @node Input Section Wildcards
4703 @subsubsection Input Section Wildcard Patterns
4704 @cindex input section wildcards
4705 @cindex wildcard file name patterns
4706 @cindex file name wildcard patterns
4707 @cindex section name wildcard patterns
4708 In an input section description, either the file name or the section
4709 name or both may be wildcard patterns.
4711 The file name of @samp{*} seen in many examples is a simple wildcard
4712 pattern for the file name.
4714 The wildcard patterns are like those used by the Unix shell.
4718 matches any number of characters
4720 matches any single character
4722 matches a single instance of any of the @var{chars}; the @samp{-}
4723 character may be used to specify a range of characters, as in
4724 @samp{[a-z]} to match any lower case letter
4726 quotes the following character
4729 When a file name is matched with a wildcard, the wildcard characters
4730 will not match a @samp{/} character (used to separate directory names on
4731 Unix). A pattern consisting of a single @samp{*} character is an
4732 exception; it will always match any file name, whether it contains a
4733 @samp{/} or not. In a section name, the wildcard characters will match
4734 a @samp{/} character.
4736 File name wildcard patterns only match files which are explicitly
4737 specified on the command line or in an @code{INPUT} command. The linker
4738 does not search directories to expand wildcards.
4740 If a file name matches more than one wildcard pattern, or if a file name
4741 appears explicitly and is also matched by a wildcard pattern, the linker
4742 will use the first match in the linker script. For example, this
4743 sequence of input section descriptions is probably in error, because the
4744 @file{data.o} rule will not be used:
4746 .data : @{ *(.data) @}
4747 .data1 : @{ data.o(.data) @}
4750 @cindex SORT_BY_NAME
4751 Normally, the linker will place files and sections matched by wildcards
4752 in the order in which they are seen during the link. You can change
4753 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4754 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4755 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4756 into ascending order by name before placing them in the output file.
4758 @cindex SORT_BY_ALIGNMENT
4759 @code{SORT_BY_ALIGNMENT} is similar to @code{SORT_BY_NAME}.
4760 @code{SORT_BY_ALIGNMENT} will sort sections into descending order of
4761 alignment before placing them in the output file. Placing larger
4762 alignments before smaller alignments can reduce the amount of padding
4765 @cindex SORT_BY_INIT_PRIORITY
4766 @code{SORT_BY_INIT_PRIORITY} is also similar to @code{SORT_BY_NAME}.
4767 @code{SORT_BY_INIT_PRIORITY} will sort sections into ascending
4768 numerical order of the GCC init_priority attribute encoded in the
4769 section name before placing them in the output file. In
4770 @code{.init_array.NNNNN} and @code{.fini_array.NNNNN}, @code{NNNNN} is
4771 the init_priority. In @code{.ctors.NNNNN} and @code{.dtors.NNNNN},
4772 @code{NNNNN} is 65535 minus the init_priority.
4775 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4777 When there are nested section sorting commands in linker script, there
4778 can be at most 1 level of nesting for section sorting commands.
4782 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4783 It will sort the input sections by name first, then by alignment if two
4784 sections have the same name.
4786 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4787 It will sort the input sections by alignment first, then by name if two
4788 sections have the same alignment.
4790 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4791 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4793 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4794 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4796 All other nested section sorting commands are invalid.
4799 When both command-line section sorting option and linker script
4800 section sorting command are used, section sorting command always
4801 takes precedence over the command-line option.
4803 If the section sorting command in linker script isn't nested, the
4804 command-line option will make the section sorting command to be
4805 treated as nested sorting command.
4809 @code{SORT_BY_NAME} (wildcard section pattern ) with
4810 @option{--sort-sections alignment} is equivalent to
4811 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4813 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4814 @option{--sort-section name} is equivalent to
4815 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4818 If the section sorting command in linker script is nested, the
4819 command-line option will be ignored.
4822 @code{SORT_NONE} disables section sorting by ignoring the command-line
4823 section sorting option.
4825 If you ever get confused about where input sections are going, use the
4826 @samp{-M} linker option to generate a map file. The map file shows
4827 precisely how input sections are mapped to output sections.
4829 This example shows how wildcard patterns might be used to partition
4830 files. This linker script directs the linker to place all @samp{.text}
4831 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4832 The linker will place the @samp{.data} section from all files beginning
4833 with an upper case character in @samp{.DATA}; for all other files, the
4834 linker will place the @samp{.data} section in @samp{.data}.
4838 .text : @{ *(.text) @}
4839 .DATA : @{ [A-Z]*(.data) @}
4840 .data : @{ *(.data) @}
4841 .bss : @{ *(.bss) @}
4846 @node Input Section Common
4847 @subsubsection Input Section for Common Symbols
4848 @cindex common symbol placement
4849 @cindex uninitialized data placement
4850 A special notation is needed for common symbols, because in many object
4851 file formats common symbols do not have a particular input section. The
4852 linker treats common symbols as though they are in an input section
4853 named @samp{COMMON}.
4855 You may use file names with the @samp{COMMON} section just as with any
4856 other input sections. You can use this to place common symbols from a
4857 particular input file in one section while common symbols from other
4858 input files are placed in another section.
4860 In most cases, common symbols in input files will be placed in the
4861 @samp{.bss} section in the output file. For example:
4863 .bss @{ *(.bss) *(COMMON) @}
4866 @cindex scommon section
4867 @cindex small common symbols
4868 Some object file formats have more than one type of common symbol. For
4869 example, the MIPS ELF object file format distinguishes standard common
4870 symbols and small common symbols. In this case, the linker will use a
4871 different special section name for other types of common symbols. In
4872 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4873 symbols and @samp{.scommon} for small common symbols. This permits you
4874 to map the different types of common symbols into memory at different
4878 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4879 notation is now considered obsolete. It is equivalent to
4882 @node Input Section Keep
4883 @subsubsection Input Section and Garbage Collection
4885 @cindex garbage collection
4886 When link-time garbage collection is in use (@samp{--gc-sections}),
4887 it is often useful to mark sections that should not be eliminated.
4888 This is accomplished by surrounding an input section's wildcard entry
4889 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4890 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4892 @node Input Section Example
4893 @subsubsection Input Section Example
4894 The following example is a complete linker script. It tells the linker
4895 to read all of the sections from file @file{all.o} and place them at the
4896 start of output section @samp{outputa} which starts at location
4897 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4898 follows immediately, in the same output section. All of section
4899 @samp{.input2} from @file{foo.o} goes into output section
4900 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4901 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4902 files are written to output section @samp{outputc}.
4930 If an output section's name is the same as the input section's name
4931 and is representable as a C identifier, then the linker will
4932 automatically @pxref{PROVIDE} two symbols: __start_SECNAME and
4933 __stop_SECNAME, where SECNAME is the name of the section. These
4934 indicate the start address and end address of the output section
4935 respectively. Note: most section names are not representable as
4936 C identifiers because they contain a @samp{.} character.
4938 @node Output Section Data
4939 @subsection Output Section Data
4941 @cindex section data
4942 @cindex output section data
4943 @kindex BYTE(@var{expression})
4944 @kindex SHORT(@var{expression})
4945 @kindex LONG(@var{expression})
4946 @kindex QUAD(@var{expression})
4947 @kindex SQUAD(@var{expression})
4948 You can include explicit bytes of data in an output section by using
4949 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4950 an output section command. Each keyword is followed by an expression in
4951 parentheses providing the value to store (@pxref{Expressions}). The
4952 value of the expression is stored at the current value of the location
4955 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4956 store one, two, four, and eight bytes (respectively). After storing the
4957 bytes, the location counter is incremented by the number of bytes
4960 For example, this will store the byte 1 followed by the four byte value
4961 of the symbol @samp{addr}:
4967 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4968 same; they both store an 8 byte, or 64 bit, value. When both host and
4969 target are 32 bits, an expression is computed as 32 bits. In this case
4970 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4971 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4973 If the object file format of the output file has an explicit endianness,
4974 which is the normal case, the value will be stored in that endianness.
4975 When the object file format does not have an explicit endianness, as is
4976 true of, for example, S-records, the value will be stored in the
4977 endianness of the first input object file.
4979 Note---these commands only work inside a section description and not
4980 between them, so the following will produce an error from the linker:
4982 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4984 whereas this will work:
4986 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4989 @kindex FILL(@var{expression})
4990 @cindex holes, filling
4991 @cindex unspecified memory
4992 You may use the @code{FILL} command to set the fill pattern for the
4993 current section. It is followed by an expression in parentheses. Any
4994 otherwise unspecified regions of memory within the section (for example,
4995 gaps left due to the required alignment of input sections) are filled
4996 with the value of the expression, repeated as
4997 necessary. A @code{FILL} statement covers memory locations after the
4998 point at which it occurs in the section definition; by including more
4999 than one @code{FILL} statement, you can have different fill patterns in
5000 different parts of an output section.
5002 This example shows how to fill unspecified regions of memory with the
5008 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
5009 section attribute, but it only affects the
5010 part of the section following the @code{FILL} command, rather than the
5011 entire section. If both are used, the @code{FILL} command takes
5012 precedence. @xref{Output Section Fill}, for details on the fill
5015 @node Output Section Keywords
5016 @subsection Output Section Keywords
5017 There are a couple of keywords which can appear as output section
5021 @kindex CREATE_OBJECT_SYMBOLS
5022 @cindex input filename symbols
5023 @cindex filename symbols
5024 @item CREATE_OBJECT_SYMBOLS
5025 The command tells the linker to create a symbol for each input file.
5026 The name of each symbol will be the name of the corresponding input
5027 file. The section of each symbol will be the output section in which
5028 the @code{CREATE_OBJECT_SYMBOLS} command appears.
5030 This is conventional for the a.out object file format. It is not
5031 normally used for any other object file format.
5033 @kindex CONSTRUCTORS
5034 @cindex C++ constructors, arranging in link
5035 @cindex constructors, arranging in link
5037 When linking using the a.out object file format, the linker uses an
5038 unusual set construct to support C++ global constructors and
5039 destructors. When linking object file formats which do not support
5040 arbitrary sections, such as ECOFF and XCOFF, the linker will
5041 automatically recognize C++ global constructors and destructors by name.
5042 For these object file formats, the @code{CONSTRUCTORS} command tells the
5043 linker to place constructor information in the output section where the
5044 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
5045 ignored for other object file formats.
5047 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
5048 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
5049 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
5050 the start and end of the global destructors. The
5051 first word in the list is the number of entries, followed by the address
5052 of each constructor or destructor, followed by a zero word. The
5053 compiler must arrange to actually run the code. For these object file
5054 formats @sc{gnu} C++ normally calls constructors from a subroutine
5055 @code{__main}; a call to @code{__main} is automatically inserted into
5056 the startup code for @code{main}. @sc{gnu} C++ normally runs
5057 destructors either by using @code{atexit}, or directly from the function
5060 For object file formats such as @code{COFF} or @code{ELF} which support
5061 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
5062 addresses of global constructors and destructors into the @code{.ctors}
5063 and @code{.dtors} sections. Placing the following sequence into your
5064 linker script will build the sort of table which the @sc{gnu} C++
5065 runtime code expects to see.
5069 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
5074 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
5080 If you are using the @sc{gnu} C++ support for initialization priority,
5081 which provides some control over the order in which global constructors
5082 are run, you must sort the constructors at link time to ensure that they
5083 are executed in the correct order. When using the @code{CONSTRUCTORS}
5084 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
5085 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
5086 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
5089 Normally the compiler and linker will handle these issues automatically,
5090 and you will not need to concern yourself with them. However, you may
5091 need to consider this if you are using C++ and writing your own linker
5096 @node Output Section Discarding
5097 @subsection Output Section Discarding
5098 @cindex discarding sections
5099 @cindex sections, discarding
5100 @cindex removing sections
5101 The linker will not normally create output sections with no contents.
5102 This is for convenience when referring to input sections that may or
5103 may not be present in any of the input files. For example:
5105 .foo : @{ *(.foo) @}
5108 will only create a @samp{.foo} section in the output file if there is a
5109 @samp{.foo} section in at least one input file, and if the input
5110 sections are not all empty. Other link script directives that allocate
5111 space in an output section will also create the output section. So
5112 too will assignments to dot even if the assignment does not create
5113 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
5114 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
5115 @samp{sym} is an absolute symbol of value 0 defined in the script.
5116 This allows you to force output of an empty section with @samp{. = .}.
5118 The linker will ignore address assignments (@pxref{Output Section Address})
5119 on discarded output sections, except when the linker script defines
5120 symbols in the output section. In that case the linker will obey
5121 the address assignments, possibly advancing dot even though the
5122 section is discarded.
5125 The special output section name @samp{/DISCARD/} may be used to discard
5126 input sections. Any input sections which are assigned to an output
5127 section named @samp{/DISCARD/} are not included in the output file.
5129 Note, sections that match the @samp{/DISCARD/} output section will be
5130 discarded even if they are in an ELF section group which has other
5131 members which are not being discarded. This is deliberate.
5132 Discarding takes precedence over grouping.
5134 @node Output Section Attributes
5135 @subsection Output Section Attributes
5136 @cindex output section attributes
5137 We showed above that the full description of an output section looked
5142 @var{section} [@var{address}] [(@var{type})] :
5144 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
5145 [SUBALIGN(@var{subsection_align})]
5148 @var{output-section-command}
5149 @var{output-section-command}
5151 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
5155 We've already described @var{section}, @var{address}, and
5156 @var{output-section-command}. In this section we will describe the
5157 remaining section attributes.
5160 * Output Section Type:: Output section type
5161 * Output Section LMA:: Output section LMA
5162 * Forced Output Alignment:: Forced Output Alignment
5163 * Forced Input Alignment:: Forced Input Alignment
5164 * Output Section Constraint:: Output section constraint
5165 * Output Section Region:: Output section region
5166 * Output Section Phdr:: Output section phdr
5167 * Output Section Fill:: Output section fill
5170 @node Output Section Type
5171 @subsubsection Output Section Type
5172 Each output section may have a type. The type is a keyword in
5173 parentheses. The following types are defined:
5177 The section should be marked as not loadable, so that it will not be
5178 loaded into memory when the program is run.
5183 These type names are supported for backward compatibility, and are
5184 rarely used. They all have the same effect: the section should be
5185 marked as not allocatable, so that no memory is allocated for the
5186 section when the program is run.
5190 @cindex prevent unnecessary loading
5191 @cindex loading, preventing
5192 The linker normally sets the attributes of an output section based on
5193 the input sections which map into it. You can override this by using
5194 the section type. For example, in the script sample below, the
5195 @samp{ROM} section is addressed at memory location @samp{0} and does not
5196 need to be loaded when the program is run.
5200 ROM 0 (NOLOAD) : @{ @dots{} @}
5206 @node Output Section LMA
5207 @subsubsection Output Section LMA
5208 @kindex AT>@var{lma_region}
5209 @kindex AT(@var{lma})
5210 @cindex load address
5211 @cindex section load address
5212 Every section has a virtual address (VMA) and a load address (LMA); see
5213 @ref{Basic Script Concepts}. The virtual address is specified by the
5214 @pxref{Output Section Address} described earlier. The load address is
5215 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
5216 address is optional.
5218 The @code{AT} keyword takes an expression as an argument. This
5219 specifies the exact load address of the section. The @code{AT>} keyword
5220 takes the name of a memory region as an argument. @xref{MEMORY}. The
5221 load address of the section is set to the next free address in the
5222 region, aligned to the section's alignment requirements.
5224 If neither @code{AT} nor @code{AT>} is specified for an allocatable
5225 section, the linker will use the following heuristic to determine the
5230 If the section has a specific VMA address, then this is used as
5231 the LMA address as well.
5234 If the section is not allocatable then its LMA is set to its VMA.
5237 Otherwise if a memory region can be found that is compatible
5238 with the current section, and this region contains at least one
5239 section, then the LMA is set so the difference between the
5240 VMA and LMA is the same as the difference between the VMA and LMA of
5241 the last section in the located region.
5244 If no memory regions have been declared then a default region
5245 that covers the entire address space is used in the previous step.
5248 If no suitable region could be found, or there was no previous
5249 section then the LMA is set equal to the VMA.
5252 @cindex ROM initialized data
5253 @cindex initialized data in ROM
5254 This feature is designed to make it easy to build a ROM image. For
5255 example, the following linker script creates three output sections: one
5256 called @samp{.text}, which starts at @code{0x1000}, one called
5257 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
5258 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
5259 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
5260 defined with the value @code{0x2000}, which shows that the location
5261 counter holds the VMA value, not the LMA value.
5267 .text 0x1000 : @{ *(.text) _etext = . ; @}
5269 AT ( ADDR (.text) + SIZEOF (.text) )
5270 @{ _data = . ; *(.data); _edata = . ; @}
5272 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
5277 The run-time initialization code for use with a program generated with
5278 this linker script would include something like the following, to copy
5279 the initialized data from the ROM image to its runtime address. Notice
5280 how this code takes advantage of the symbols defined by the linker
5285 extern char _etext, _data, _edata, _bstart, _bend;
5286 char *src = &_etext;
5289 /* ROM has data at end of text; copy it. */
5290 while (dst < &_edata)
5294 for (dst = &_bstart; dst< &_bend; dst++)
5299 @node Forced Output Alignment
5300 @subsubsection Forced Output Alignment
5301 @kindex ALIGN(@var{section_align})
5302 @cindex forcing output section alignment
5303 @cindex output section alignment
5304 You can increase an output section's alignment by using ALIGN. As an
5305 alternative you can enforce that the difference between the VMA and LMA remains
5306 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
5308 @node Forced Input Alignment
5309 @subsubsection Forced Input Alignment
5310 @kindex SUBALIGN(@var{subsection_align})
5311 @cindex forcing input section alignment
5312 @cindex input section alignment
5313 You can force input section alignment within an output section by using
5314 SUBALIGN. The value specified overrides any alignment given by input
5315 sections, whether larger or smaller.
5317 @node Output Section Constraint
5318 @subsubsection Output Section Constraint
5321 @cindex constraints on output sections
5322 You can specify that an output section should only be created if all
5323 of its input sections are read-only or all of its input sections are
5324 read-write by using the keyword @code{ONLY_IF_RO} and
5325 @code{ONLY_IF_RW} respectively.
5327 @node Output Section Region
5328 @subsubsection Output Section Region
5329 @kindex >@var{region}
5330 @cindex section, assigning to memory region
5331 @cindex memory regions and sections
5332 You can assign a section to a previously defined region of memory by
5333 using @samp{>@var{region}}. @xref{MEMORY}.
5335 Here is a simple example:
5338 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
5339 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
5343 @node Output Section Phdr
5344 @subsubsection Output Section Phdr
5346 @cindex section, assigning to program header
5347 @cindex program headers and sections
5348 You can assign a section to a previously defined program segment by
5349 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
5350 one or more segments, then all subsequent allocated sections will be
5351 assigned to those segments as well, unless they use an explicitly
5352 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
5353 linker to not put the section in any segment at all.
5355 Here is a simple example:
5358 PHDRS @{ text PT_LOAD ; @}
5359 SECTIONS @{ .text : @{ *(.text) @} :text @}
5363 @node Output Section Fill
5364 @subsubsection Output Section Fill
5365 @kindex =@var{fillexp}
5366 @cindex section fill pattern
5367 @cindex fill pattern, entire section
5368 You can set the fill pattern for an entire section by using
5369 @samp{=@var{fillexp}}. @var{fillexp} is an expression
5370 (@pxref{Expressions}). Any otherwise unspecified regions of memory
5371 within the output section (for example, gaps left due to the required
5372 alignment of input sections) will be filled with the value, repeated as
5373 necessary. If the fill expression is a simple hex number, ie. a string
5374 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
5375 an arbitrarily long sequence of hex digits can be used to specify the
5376 fill pattern; Leading zeros become part of the pattern too. For all
5377 other cases, including extra parentheses or a unary @code{+}, the fill
5378 pattern is the four least significant bytes of the value of the
5379 expression. In all cases, the number is big-endian.
5381 You can also change the fill value with a @code{FILL} command in the
5382 output section commands; (@pxref{Output Section Data}).
5384 Here is a simple example:
5387 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
5391 @node Overlay Description
5392 @subsection Overlay Description
5395 An overlay description provides an easy way to describe sections which
5396 are to be loaded as part of a single memory image but are to be run at
5397 the same memory address. At run time, some sort of overlay manager will
5398 copy the overlaid sections in and out of the runtime memory address as
5399 required, perhaps by simply manipulating addressing bits. This approach
5400 can be useful, for example, when a certain region of memory is faster
5403 Overlays are described using the @code{OVERLAY} command. The
5404 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
5405 output section description. The full syntax of the @code{OVERLAY}
5406 command is as follows:
5409 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
5413 @var{output-section-command}
5414 @var{output-section-command}
5416 @} [:@var{phdr}@dots{}] [=@var{fill}]
5419 @var{output-section-command}
5420 @var{output-section-command}
5422 @} [:@var{phdr}@dots{}] [=@var{fill}]
5424 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5428 Everything is optional except @code{OVERLAY} (a keyword), and each
5429 section must have a name (@var{secname1} and @var{secname2} above). The
5430 section definitions within the @code{OVERLAY} construct are identical to
5431 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5432 except that no addresses and no memory regions may be defined for
5433 sections within an @code{OVERLAY}.
5435 The comma at the end may be required if a @var{fill} is used and
5436 the next @var{sections-command} looks like a continuation of the expression.
5438 The sections are all defined with the same starting address. The load
5439 addresses of the sections are arranged such that they are consecutive in
5440 memory starting at the load address used for the @code{OVERLAY} as a
5441 whole (as with normal section definitions, the load address is optional,
5442 and defaults to the start address; the start address is also optional,
5443 and defaults to the current value of the location counter).
5445 If the @code{NOCROSSREFS} keyword is used, and there are any
5446 references among the sections, the linker will report an error. Since
5447 the sections all run at the same address, it normally does not make
5448 sense for one section to refer directly to another.
5449 @xref{Miscellaneous Commands, NOCROSSREFS}.
5451 For each section within the @code{OVERLAY}, the linker automatically
5452 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5453 defined as the starting load address of the section. The symbol
5454 @code{__load_stop_@var{secname}} is defined as the final load address of
5455 the section. Any characters within @var{secname} which are not legal
5456 within C identifiers are removed. C (or assembler) code may use these
5457 symbols to move the overlaid sections around as necessary.
5459 At the end of the overlay, the value of the location counter is set to
5460 the start address of the overlay plus the size of the largest section.
5462 Here is an example. Remember that this would appear inside a
5463 @code{SECTIONS} construct.
5466 OVERLAY 0x1000 : AT (0x4000)
5468 .text0 @{ o1/*.o(.text) @}
5469 .text1 @{ o2/*.o(.text) @}
5474 This will define both @samp{.text0} and @samp{.text1} to start at
5475 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5476 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5477 following symbols will be defined if referenced: @code{__load_start_text0},
5478 @code{__load_stop_text0}, @code{__load_start_text1},
5479 @code{__load_stop_text1}.
5481 C code to copy overlay @code{.text1} into the overlay area might look
5486 extern char __load_start_text1, __load_stop_text1;
5487 memcpy ((char *) 0x1000, &__load_start_text1,
5488 &__load_stop_text1 - &__load_start_text1);
5492 Note that the @code{OVERLAY} command is just syntactic sugar, since
5493 everything it does can be done using the more basic commands. The above
5494 example could have been written identically as follows.
5498 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5499 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5500 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5501 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5502 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5503 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5504 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5509 @section MEMORY Command
5511 @cindex memory regions
5512 @cindex regions of memory
5513 @cindex allocating memory
5514 @cindex discontinuous memory
5515 The linker's default configuration permits allocation of all available
5516 memory. You can override this by using the @code{MEMORY} command.
5518 The @code{MEMORY} command describes the location and size of blocks of
5519 memory in the target. You can use it to describe which memory regions
5520 may be used by the linker, and which memory regions it must avoid. You
5521 can then assign sections to particular memory regions. The linker will
5522 set section addresses based on the memory regions, and will warn about
5523 regions that become too full. The linker will not shuffle sections
5524 around to fit into the available regions.
5526 A linker script may contain many uses of the @code{MEMORY} command,
5527 however, all memory blocks defined are treated as if they were
5528 specified inside a single @code{MEMORY} command. The syntax for
5534 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5540 The @var{name} is a name used in the linker script to refer to the
5541 region. The region name has no meaning outside of the linker script.
5542 Region names are stored in a separate name space, and will not conflict
5543 with symbol names, file names, or section names. Each memory region
5544 must have a distinct name within the @code{MEMORY} command. However you can
5545 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5548 @cindex memory region attributes
5549 The @var{attr} string is an optional list of attributes that specify
5550 whether to use a particular memory region for an input section which is
5551 not explicitly mapped in the linker script. As described in
5552 @ref{SECTIONS}, if you do not specify an output section for some input
5553 section, the linker will create an output section with the same name as
5554 the input section. If you define region attributes, the linker will use
5555 them to select the memory region for the output section that it creates.
5557 The @var{attr} string must consist only of the following characters:
5572 Invert the sense of any of the attributes that follow
5575 If an unmapped section matches any of the listed attributes other than
5576 @samp{!}, it will be placed in the memory region. The @samp{!}
5577 attribute reverses the test for the characters that follow, so that an
5578 unmapped section will be placed in the memory region only if it does
5579 not match any of the attributes listed afterwards. Thus an attribute
5580 string of @samp{RW!X} will match any unmapped section that has either
5581 or both of the @samp{R} and @samp{W} attributes, but only as long as
5582 the section does not also have the @samp{X} attribute.
5587 The @var{origin} is an numerical expression for the start address of
5588 the memory region. The expression must evaluate to a constant and it
5589 cannot involve any symbols. The keyword @code{ORIGIN} may be
5590 abbreviated to @code{org} or @code{o} (but not, for example,
5596 The @var{len} is an expression for the size in bytes of the memory
5597 region. As with the @var{origin} expression, the expression must
5598 be numerical only and must evaluate to a constant. The keyword
5599 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5601 In the following example, we specify that there are two memory regions
5602 available for allocation: one starting at @samp{0} for 256 kilobytes,
5603 and the other starting at @samp{0x40000000} for four megabytes. The
5604 linker will place into the @samp{rom} memory region every section which
5605 is not explicitly mapped into a memory region, and is either read-only
5606 or executable. The linker will place other sections which are not
5607 explicitly mapped into a memory region into the @samp{ram} memory
5614 rom (rx) : ORIGIN = 0, LENGTH = 256K
5615 ram (!rx) : org = 0x40000000, l = 4M
5620 Once you define a memory region, you can direct the linker to place
5621 specific output sections into that memory region by using the
5622 @samp{>@var{region}} output section attribute. For example, if you have
5623 a memory region named @samp{mem}, you would use @samp{>mem} in the
5624 output section definition. @xref{Output Section Region}. If no address
5625 was specified for the output section, the linker will set the address to
5626 the next available address within the memory region. If the combined
5627 output sections directed to a memory region are too large for the
5628 region, the linker will issue an error message.
5630 It is possible to access the origin and length of a memory in an
5631 expression via the @code{ORIGIN(@var{memory})} and
5632 @code{LENGTH(@var{memory})} functions:
5636 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5641 @section PHDRS Command
5643 @cindex program headers
5644 @cindex ELF program headers
5645 @cindex program segments
5646 @cindex segments, ELF
5647 The ELF object file format uses @dfn{program headers}, also knows as
5648 @dfn{segments}. The program headers describe how the program should be
5649 loaded into memory. You can print them out by using the @code{objdump}
5650 program with the @samp{-p} option.
5652 When you run an ELF program on a native ELF system, the system loader
5653 reads the program headers in order to figure out how to load the
5654 program. This will only work if the program headers are set correctly.
5655 This manual does not describe the details of how the system loader
5656 interprets program headers; for more information, see the ELF ABI.
5658 The linker will create reasonable program headers by default. However,
5659 in some cases, you may need to specify the program headers more
5660 precisely. You may use the @code{PHDRS} command for this purpose. When
5661 the linker sees the @code{PHDRS} command in the linker script, it will
5662 not create any program headers other than the ones specified.
5664 The linker only pays attention to the @code{PHDRS} command when
5665 generating an ELF output file. In other cases, the linker will simply
5666 ignore @code{PHDRS}.
5668 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5669 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5675 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5676 [ FLAGS ( @var{flags} ) ] ;
5681 The @var{name} is used only for reference in the @code{SECTIONS} command
5682 of the linker script. It is not put into the output file. Program
5683 header names are stored in a separate name space, and will not conflict
5684 with symbol names, file names, or section names. Each program header
5685 must have a distinct name. The headers are processed in order and it
5686 is usual for them to map to sections in ascending load address order.
5688 Certain program header types describe segments of memory which the
5689 system loader will load from the file. In the linker script, you
5690 specify the contents of these segments by placing allocatable output
5691 sections in the segments. You use the @samp{:@var{phdr}} output section
5692 attribute to place a section in a particular segment. @xref{Output
5695 It is normal to put certain sections in more than one segment. This
5696 merely implies that one segment of memory contains another. You may
5697 repeat @samp{:@var{phdr}}, using it once for each segment which should
5698 contain the section.
5700 If you place a section in one or more segments using @samp{:@var{phdr}},
5701 then the linker will place all subsequent allocatable sections which do
5702 not specify @samp{:@var{phdr}} in the same segments. This is for
5703 convenience, since generally a whole set of contiguous sections will be
5704 placed in a single segment. You can use @code{:NONE} to override the
5705 default segment and tell the linker to not put the section in any
5710 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
5711 the program header type to further describe the contents of the segment.
5712 The @code{FILEHDR} keyword means that the segment should include the ELF
5713 file header. The @code{PHDRS} keyword means that the segment should
5714 include the ELF program headers themselves. If applied to a loadable
5715 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5718 The @var{type} may be one of the following. The numbers indicate the
5719 value of the keyword.
5722 @item @code{PT_NULL} (0)
5723 Indicates an unused program header.
5725 @item @code{PT_LOAD} (1)
5726 Indicates that this program header describes a segment to be loaded from
5729 @item @code{PT_DYNAMIC} (2)
5730 Indicates a segment where dynamic linking information can be found.
5732 @item @code{PT_INTERP} (3)
5733 Indicates a segment where the name of the program interpreter may be
5736 @item @code{PT_NOTE} (4)
5737 Indicates a segment holding note information.
5739 @item @code{PT_SHLIB} (5)
5740 A reserved program header type, defined but not specified by the ELF
5743 @item @code{PT_PHDR} (6)
5744 Indicates a segment where the program headers may be found.
5746 @item @code{PT_TLS} (7)
5747 Indicates a segment containing thread local storage.
5749 @item @var{expression}
5750 An expression giving the numeric type of the program header. This may
5751 be used for types not defined above.
5754 You can specify that a segment should be loaded at a particular address
5755 in memory by using an @code{AT} expression. This is identical to the
5756 @code{AT} command used as an output section attribute (@pxref{Output
5757 Section LMA}). The @code{AT} command for a program header overrides the
5758 output section attribute.
5760 The linker will normally set the segment flags based on the sections
5761 which comprise the segment. You may use the @code{FLAGS} keyword to
5762 explicitly specify the segment flags. The value of @var{flags} must be
5763 an integer. It is used to set the @code{p_flags} field of the program
5766 Here is an example of @code{PHDRS}. This shows a typical set of program
5767 headers used on a native ELF system.
5773 headers PT_PHDR PHDRS ;
5775 text PT_LOAD FILEHDR PHDRS ;
5777 dynamic PT_DYNAMIC ;
5783 .interp : @{ *(.interp) @} :text :interp
5784 .text : @{ *(.text) @} :text
5785 .rodata : @{ *(.rodata) @} /* defaults to :text */
5787 . = . + 0x1000; /* move to a new page in memory */
5788 .data : @{ *(.data) @} :data
5789 .dynamic : @{ *(.dynamic) @} :data :dynamic
5796 @section VERSION Command
5797 @kindex VERSION @{script text@}
5798 @cindex symbol versions
5799 @cindex version script
5800 @cindex versions of symbols
5801 The linker supports symbol versions when using ELF. Symbol versions are
5802 only useful when using shared libraries. The dynamic linker can use
5803 symbol versions to select a specific version of a function when it runs
5804 a program that may have been linked against an earlier version of the
5807 You can include a version script directly in the main linker script, or
5808 you can supply the version script as an implicit linker script. You can
5809 also use the @samp{--version-script} linker option.
5811 The syntax of the @code{VERSION} command is simply
5813 VERSION @{ version-script-commands @}
5816 The format of the version script commands is identical to that used by
5817 Sun's linker in Solaris 2.5. The version script defines a tree of
5818 version nodes. You specify the node names and interdependencies in the
5819 version script. You can specify which symbols are bound to which
5820 version nodes, and you can reduce a specified set of symbols to local
5821 scope so that they are not globally visible outside of the shared
5824 The easiest way to demonstrate the version script language is with a few
5850 This example version script defines three version nodes. The first
5851 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5852 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5853 a number of symbols to local scope so that they are not visible outside
5854 of the shared library; this is done using wildcard patterns, so that any
5855 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5856 is matched. The wildcard patterns available are the same as those used
5857 in the shell when matching filenames (also known as ``globbing'').
5858 However, if you specify the symbol name inside double quotes, then the
5859 name is treated as literal, rather than as a glob pattern.
5861 Next, the version script defines node @samp{VERS_1.2}. This node
5862 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5863 to the version node @samp{VERS_1.2}.
5865 Finally, the version script defines node @samp{VERS_2.0}. This node
5866 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5867 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5869 When the linker finds a symbol defined in a library which is not
5870 specifically bound to a version node, it will effectively bind it to an
5871 unspecified base version of the library. You can bind all otherwise
5872 unspecified symbols to a given version node by using @samp{global: *;}
5873 somewhere in the version script. Note that it's slightly crazy to use
5874 wildcards in a global spec except on the last version node. Global
5875 wildcards elsewhere run the risk of accidentally adding symbols to the
5876 set exported for an old version. That's wrong since older versions
5877 ought to have a fixed set of symbols.
5879 The names of the version nodes have no specific meaning other than what
5880 they might suggest to the person reading them. The @samp{2.0} version
5881 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5882 However, this would be a confusing way to write a version script.
5884 Node name can be omitted, provided it is the only version node
5885 in the version script. Such version script doesn't assign any versions to
5886 symbols, only selects which symbols will be globally visible out and which
5890 @{ global: foo; bar; local: *; @};
5893 When you link an application against a shared library that has versioned
5894 symbols, the application itself knows which version of each symbol it
5895 requires, and it also knows which version nodes it needs from each
5896 shared library it is linked against. Thus at runtime, the dynamic
5897 loader can make a quick check to make sure that the libraries you have
5898 linked against do in fact supply all of the version nodes that the
5899 application will need to resolve all of the dynamic symbols. In this
5900 way it is possible for the dynamic linker to know with certainty that
5901 all external symbols that it needs will be resolvable without having to
5902 search for each symbol reference.
5904 The symbol versioning is in effect a much more sophisticated way of
5905 doing minor version checking that SunOS does. The fundamental problem
5906 that is being addressed here is that typically references to external
5907 functions are bound on an as-needed basis, and are not all bound when
5908 the application starts up. If a shared library is out of date, a
5909 required interface may be missing; when the application tries to use
5910 that interface, it may suddenly and unexpectedly fail. With symbol
5911 versioning, the user will get a warning when they start their program if
5912 the libraries being used with the application are too old.
5914 There are several GNU extensions to Sun's versioning approach. The
5915 first of these is the ability to bind a symbol to a version node in the
5916 source file where the symbol is defined instead of in the versioning
5917 script. This was done mainly to reduce the burden on the library
5918 maintainer. You can do this by putting something like:
5920 __asm__(".symver original_foo,foo@@VERS_1.1");
5923 in the C source file. This renames the function @samp{original_foo} to
5924 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5925 The @samp{local:} directive can be used to prevent the symbol
5926 @samp{original_foo} from being exported. A @samp{.symver} directive
5927 takes precedence over a version script.
5929 The second GNU extension is to allow multiple versions of the same
5930 function to appear in a given shared library. In this way you can make
5931 an incompatible change to an interface without increasing the major
5932 version number of the shared library, while still allowing applications
5933 linked against the old interface to continue to function.
5935 To do this, you must use multiple @samp{.symver} directives in the
5936 source file. Here is an example:
5939 __asm__(".symver original_foo,foo@@");
5940 __asm__(".symver old_foo,foo@@VERS_1.1");
5941 __asm__(".symver old_foo1,foo@@VERS_1.2");
5942 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5945 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5946 unspecified base version of the symbol. The source file that contains this
5947 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5948 @samp{old_foo1}, and @samp{new_foo}.
5950 When you have multiple definitions of a given symbol, there needs to be
5951 some way to specify a default version to which external references to
5952 this symbol will be bound. You can do this with the
5953 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5954 declare one version of a symbol as the default in this manner; otherwise
5955 you would effectively have multiple definitions of the same symbol.
5957 If you wish to bind a reference to a specific version of the symbol
5958 within the shared library, you can use the aliases of convenience
5959 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5960 specifically bind to an external version of the function in question.
5962 You can also specify the language in the version script:
5965 VERSION extern "lang" @{ version-script-commands @}
5968 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5969 The linker will iterate over the list of symbols at the link time and
5970 demangle them according to @samp{lang} before matching them to the
5971 patterns specified in @samp{version-script-commands}. The default
5972 @samp{lang} is @samp{C}.
5974 Demangled names may contains spaces and other special characters. As
5975 described above, you can use a glob pattern to match demangled names,
5976 or you can use a double-quoted string to match the string exactly. In
5977 the latter case, be aware that minor differences (such as differing
5978 whitespace) between the version script and the demangler output will
5979 cause a mismatch. As the exact string generated by the demangler
5980 might change in the future, even if the mangled name does not, you
5981 should check that all of your version directives are behaving as you
5982 expect when you upgrade.
5985 @section Expressions in Linker Scripts
5988 The syntax for expressions in the linker script language is identical to
5989 that of C expressions. All expressions are evaluated as integers. All
5990 expressions are evaluated in the same size, which is 32 bits if both the
5991 host and target are 32 bits, and is otherwise 64 bits.
5993 You can use and set symbol values in expressions.
5995 The linker defines several special purpose builtin functions for use in
5999 * Constants:: Constants
6000 * Symbolic Constants:: Symbolic constants
6001 * Symbols:: Symbol Names
6002 * Orphan Sections:: Orphan Sections
6003 * Location Counter:: The Location Counter
6004 * Operators:: Operators
6005 * Evaluation:: Evaluation
6006 * Expression Section:: The Section of an Expression
6007 * Builtin Functions:: Builtin Functions
6011 @subsection Constants
6012 @cindex integer notation
6013 @cindex constants in linker scripts
6014 All constants are integers.
6016 As in C, the linker considers an integer beginning with @samp{0} to be
6017 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
6018 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
6019 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
6020 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
6021 value without a prefix or a suffix is considered to be decimal.
6023 @cindex scaled integers
6024 @cindex K and M integer suffixes
6025 @cindex M and K integer suffixes
6026 @cindex suffixes for integers
6027 @cindex integer suffixes
6028 In addition, you can use the suffixes @code{K} and @code{M} to scale a
6032 @c END TEXI2ROFF-KILL
6033 @code{1024} or @code{1024*1024}
6037 ${\rm 1024}$ or ${\rm 1024}^2$
6039 @c END TEXI2ROFF-KILL
6040 respectively. For example, the following
6041 all refer to the same quantity:
6050 Note - the @code{K} and @code{M} suffixes cannot be used in
6051 conjunction with the base suffixes mentioned above.
6053 @node Symbolic Constants
6054 @subsection Symbolic Constants
6055 @cindex symbolic constants
6057 It is possible to refer to target-specific constants via the use of
6058 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
6063 The target's maximum page size.
6065 @item COMMONPAGESIZE
6066 @kindex COMMONPAGESIZE
6067 The target's default page size.
6073 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
6076 will create a text section aligned to the largest page boundary
6077 supported by the target.
6080 @subsection Symbol Names
6081 @cindex symbol names
6083 @cindex quoted symbol names
6085 Unless quoted, symbol names start with a letter, underscore, or period
6086 and may include letters, digits, underscores, periods, and hyphens.
6087 Unquoted symbol names must not conflict with any keywords. You can
6088 specify a symbol which contains odd characters or has the same name as a
6089 keyword by surrounding the symbol name in double quotes:
6092 "with a space" = "also with a space" + 10;
6095 Since symbols can contain many non-alphabetic characters, it is safest
6096 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
6097 whereas @samp{A - B} is an expression involving subtraction.
6099 @node Orphan Sections
6100 @subsection Orphan Sections
6102 Orphan sections are sections present in the input files which
6103 are not explicitly placed into the output file by the linker
6104 script. The linker will still copy these sections into the
6105 output file by either finding, or creating a suitable output section
6106 in which to place the orphaned input section.
6108 If the name of an orphaned input section exactly matches the name of
6109 an existing output section, then the orphaned input section will be
6110 placed at the end of that output section.
6112 If there is no output section with a matching name then new output
6113 sections will be created. Each new output section will have the same
6114 name as the orphan section placed within it. If there are multiple
6115 orphan sections with the same name, these will all be combined into
6116 one new output section.
6118 If new output sections are created to hold orphaned input sections,
6119 then the linker must decide where to place these new output sections
6120 in relation to existing output sections. On most modern targets, the
6121 linker attempts to place orphan sections after sections of the same
6122 attribute, such as code vs data, loadable vs non-loadable, etc. If no
6123 sections with matching attributes are found, or your target lacks this
6124 support, the orphan section is placed at the end of the file.
6126 The command-line options @samp{--orphan-handling} and @samp{--unique}
6127 (@pxref{Options,,Command-line Options}) can be used to control which
6128 output sections an orphan is placed in.
6130 @node Location Counter
6131 @subsection The Location Counter
6134 @cindex location counter
6135 @cindex current output location
6136 The special linker variable @dfn{dot} @samp{.} always contains the
6137 current output location counter. Since the @code{.} always refers to a
6138 location in an output section, it may only appear in an expression
6139 within a @code{SECTIONS} command. The @code{.} symbol may appear
6140 anywhere that an ordinary symbol is allowed in an expression.
6143 Assigning a value to @code{.} will cause the location counter to be
6144 moved. This may be used to create holes in the output section. The
6145 location counter may not be moved backwards inside an output section,
6146 and may not be moved backwards outside of an output section if so
6147 doing creates areas with overlapping LMAs.
6163 In the previous example, the @samp{.text} section from @file{file1} is
6164 located at the beginning of the output section @samp{output}. It is
6165 followed by a 1000 byte gap. Then the @samp{.text} section from
6166 @file{file2} appears, also with a 1000 byte gap following before the
6167 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
6168 specifies what data to write in the gaps (@pxref{Output Section Fill}).
6170 @cindex dot inside sections
6171 Note: @code{.} actually refers to the byte offset from the start of the
6172 current containing object. Normally this is the @code{SECTIONS}
6173 statement, whose start address is 0, hence @code{.} can be used as an
6174 absolute address. If @code{.} is used inside a section description
6175 however, it refers to the byte offset from the start of that section,
6176 not an absolute address. Thus in a script like this:
6194 The @samp{.text} section will be assigned a starting address of 0x100
6195 and a size of exactly 0x200 bytes, even if there is not enough data in
6196 the @samp{.text} input sections to fill this area. (If there is too
6197 much data, an error will be produced because this would be an attempt to
6198 move @code{.} backwards). The @samp{.data} section will start at 0x500
6199 and it will have an extra 0x600 bytes worth of space after the end of
6200 the values from the @samp{.data} input sections and before the end of
6201 the @samp{.data} output section itself.
6203 @cindex dot outside sections
6204 Setting symbols to the value of the location counter outside of an
6205 output section statement can result in unexpected values if the linker
6206 needs to place orphan sections. For example, given the following:
6212 .text: @{ *(.text) @}
6216 .data: @{ *(.data) @}
6221 If the linker needs to place some input section, e.g. @code{.rodata},
6222 not mentioned in the script, it might choose to place that section
6223 between @code{.text} and @code{.data}. You might think the linker
6224 should place @code{.rodata} on the blank line in the above script, but
6225 blank lines are of no particular significance to the linker. As well,
6226 the linker doesn't associate the above symbol names with their
6227 sections. Instead, it assumes that all assignments or other
6228 statements belong to the previous output section, except for the
6229 special case of an assignment to @code{.}. I.e., the linker will
6230 place the orphan @code{.rodata} section as if the script was written
6237 .text: @{ *(.text) @}
6241 .rodata: @{ *(.rodata) @}
6242 .data: @{ *(.data) @}
6247 This may or may not be the script author's intention for the value of
6248 @code{start_of_data}. One way to influence the orphan section
6249 placement is to assign the location counter to itself, as the linker
6250 assumes that an assignment to @code{.} is setting the start address of
6251 a following output section and thus should be grouped with that
6252 section. So you could write:
6258 .text: @{ *(.text) @}
6263 .data: @{ *(.data) @}
6268 Now, the orphan @code{.rodata} section will be placed between
6269 @code{end_of_text} and @code{start_of_data}.
6273 @subsection Operators
6274 @cindex operators for arithmetic
6275 @cindex arithmetic operators
6276 @cindex precedence in expressions
6277 The linker recognizes the standard C set of arithmetic operators, with
6278 the standard bindings and precedence levels:
6281 @c END TEXI2ROFF-KILL
6283 precedence associativity Operators Notes
6289 5 left == != > < <= >=
6295 11 right &= += -= *= /= (2)
6299 (1) Prefix operators
6300 (2) @xref{Assignments}.
6304 \vskip \baselineskip
6305 %"lispnarrowing" is the extra indent used generally for smallexample
6306 \hskip\lispnarrowing\vbox{\offinterlineskip
6309 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
6310 height2pt&\omit&&\omit&&\omit&\cr
6311 &Precedence&& Associativity &&{\rm Operators}&\cr
6312 height2pt&\omit&&\omit&&\omit&\cr
6314 height2pt&\omit&&\omit&&\omit&\cr
6316 % '176 is tilde, '~' in tt font
6317 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
6318 &2&&left&&* / \%&\cr
6321 &5&&left&&== != > < <= >=&\cr
6324 &8&&left&&{\&\&}&\cr
6327 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
6329 height2pt&\omit&&\omit&&\omit&\cr}
6334 @obeylines@parskip=0pt@parindent=0pt
6335 @dag@quad Prefix operators.
6336 @ddag@quad @xref{Assignments}.
6339 @c END TEXI2ROFF-KILL
6342 @subsection Evaluation
6343 @cindex lazy evaluation
6344 @cindex expression evaluation order
6345 The linker evaluates expressions lazily. It only computes the value of
6346 an expression when absolutely necessary.
6348 The linker needs some information, such as the value of the start
6349 address of the first section, and the origins and lengths of memory
6350 regions, in order to do any linking at all. These values are computed
6351 as soon as possible when the linker reads in the linker script.
6353 However, other values (such as symbol values) are not known or needed
6354 until after storage allocation. Such values are evaluated later, when
6355 other information (such as the sizes of output sections) is available
6356 for use in the symbol assignment expression.
6358 The sizes of sections cannot be known until after allocation, so
6359 assignments dependent upon these are not performed until after
6362 Some expressions, such as those depending upon the location counter
6363 @samp{.}, must be evaluated during section allocation.
6365 If the result of an expression is required, but the value is not
6366 available, then an error results. For example, a script like the
6372 .text 9+this_isnt_constant :
6378 will cause the error message @samp{non constant expression for initial
6381 @node Expression Section
6382 @subsection The Section of an Expression
6383 @cindex expression sections
6384 @cindex absolute expressions
6385 @cindex relative expressions
6386 @cindex absolute and relocatable symbols
6387 @cindex relocatable and absolute symbols
6388 @cindex symbols, relocatable and absolute
6389 Addresses and symbols may be section relative, or absolute. A section
6390 relative symbol is relocatable. If you request relocatable output
6391 using the @samp{-r} option, a further link operation may change the
6392 value of a section relative symbol. On the other hand, an absolute
6393 symbol will retain the same value throughout any further link
6396 Some terms in linker expressions are addresses. This is true of
6397 section relative symbols and for builtin functions that return an
6398 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
6399 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
6400 functions that return a non-address value, such as @code{LENGTH}.
6401 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
6402 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
6403 differently depending on their location, for compatibility with older
6404 versions of @code{ld}. Expressions appearing outside an output
6405 section definition treat all numbers as absolute addresses.
6406 Expressions appearing inside an output section definition treat
6407 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
6408 given, then absolute symbols and numbers are simply treated as numbers
6411 In the following simple example,
6418 __executable_start = 0x100;
6422 __data_start = 0x10;
6430 both @code{.} and @code{__executable_start} are set to the absolute
6431 address 0x100 in the first two assignments, then both @code{.} and
6432 @code{__data_start} are set to 0x10 relative to the @code{.data}
6433 section in the second two assignments.
6435 For expressions involving numbers, relative addresses and absolute
6436 addresses, ld follows these rules to evaluate terms:
6440 Unary operations on an absolute address or number, and binary
6441 operations on two absolute addresses or two numbers, or between one
6442 absolute address and a number, apply the operator to the value(s).
6444 Unary operations on a relative address, and binary operations on two
6445 relative addresses in the same section or between one relative address
6446 and a number, apply the operator to the offset part of the address(es).
6448 Other binary operations, that is, between two relative addresses not
6449 in the same section, or between a relative address and an absolute
6450 address, first convert any non-absolute term to an absolute address
6451 before applying the operator.
6454 The result section of each sub-expression is as follows:
6458 An operation involving only numbers results in a number.
6460 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6462 The result of other binary arithmetic and logical operations on two
6463 relative addresses in the same section or two absolute addresses
6464 (after above conversions) is also a number when
6465 @code{LD_FEATURE ("SANE_EXPR")} or inside an output section definition
6466 but an absolute address otherwise.
6468 The result of other operations on relative addresses or one
6469 relative address and a number, is a relative address in the same
6470 section as the relative operand(s).
6472 The result of other operations on absolute addresses (after above
6473 conversions) is an absolute address.
6476 You can use the builtin function @code{ABSOLUTE} to force an expression
6477 to be absolute when it would otherwise be relative. For example, to
6478 create an absolute symbol set to the address of the end of the output
6479 section @samp{.data}:
6483 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6487 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6488 @samp{.data} section.
6490 Using @code{LOADADDR} also forces an expression absolute, since this
6491 particular builtin function returns an absolute address.
6493 @node Builtin Functions
6494 @subsection Builtin Functions
6495 @cindex functions in expressions
6496 The linker script language includes a number of builtin functions for
6497 use in linker script expressions.
6500 @item ABSOLUTE(@var{exp})
6501 @kindex ABSOLUTE(@var{exp})
6502 @cindex expression, absolute
6503 Return the absolute (non-relocatable, as opposed to non-negative) value
6504 of the expression @var{exp}. Primarily useful to assign an absolute
6505 value to a symbol within a section definition, where symbol values are
6506 normally section relative. @xref{Expression Section}.
6508 @item ADDR(@var{section})
6509 @kindex ADDR(@var{section})
6510 @cindex section address in expression
6511 Return the address (VMA) of the named @var{section}. Your
6512 script must previously have defined the location of that section. In
6513 the following example, @code{start_of_output_1}, @code{symbol_1} and
6514 @code{symbol_2} are assigned equivalent values, except that
6515 @code{symbol_1} will be relative to the @code{.output1} section while
6516 the other two will be absolute:
6522 start_of_output_1 = ABSOLUTE(.);
6527 symbol_1 = ADDR(.output1);
6528 symbol_2 = start_of_output_1;
6534 @item ALIGN(@var{align})
6535 @itemx ALIGN(@var{exp},@var{align})
6536 @kindex ALIGN(@var{align})
6537 @kindex ALIGN(@var{exp},@var{align})
6538 @cindex round up location counter
6539 @cindex align location counter
6540 @cindex round up expression
6541 @cindex align expression
6542 Return the location counter (@code{.}) or arbitrary expression aligned
6543 to the next @var{align} boundary. The single operand @code{ALIGN}
6544 doesn't change the value of the location counter---it just does
6545 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6546 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6547 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6549 Here is an example which aligns the output @code{.data} section to the
6550 next @code{0x2000} byte boundary after the preceding section and sets a
6551 variable within the section to the next @code{0x8000} boundary after the
6556 .data ALIGN(0x2000): @{
6558 variable = ALIGN(0x8000);
6564 The first use of @code{ALIGN} in this example specifies the location of
6565 a section because it is used as the optional @var{address} attribute of
6566 a section definition (@pxref{Output Section Address}). The second use
6567 of @code{ALIGN} is used to defines the value of a symbol.
6569 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6571 @item ALIGNOF(@var{section})
6572 @kindex ALIGNOF(@var{section})
6573 @cindex section alignment
6574 Return the alignment in bytes of the named @var{section}, if that section has
6575 been allocated. If the section has not been allocated when this is
6576 evaluated, the linker will report an error. In the following example,
6577 the alignment of the @code{.output} section is stored as the first
6578 value in that section.
6583 LONG (ALIGNOF (.output))
6590 @item BLOCK(@var{exp})
6591 @kindex BLOCK(@var{exp})
6592 This is a synonym for @code{ALIGN}, for compatibility with older linker
6593 scripts. It is most often seen when setting the address of an output
6596 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6597 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6598 This is equivalent to either
6600 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6604 (ALIGN(@var{maxpagesize})
6605 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6608 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6609 for the data segment (area between the result of this expression and
6610 @code{DATA_SEGMENT_END}) than the former or not.
6611 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6612 memory will be saved at the expense of up to @var{commonpagesize} wasted
6613 bytes in the on-disk file.
6615 This expression can only be used directly in @code{SECTIONS} commands, not in
6616 any output section descriptions and only once in the linker script.
6617 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6618 be the system page size the object wants to be optimized for while still
6619 running on system page sizes up to @var{maxpagesize}. Note however
6620 that @samp{-z relro} protection will not be effective if the system
6621 page size is larger than @var{commonpagesize}.
6626 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6629 @item DATA_SEGMENT_END(@var{exp})
6630 @kindex DATA_SEGMENT_END(@var{exp})
6631 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6632 evaluation purposes.
6635 . = DATA_SEGMENT_END(.);
6638 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6639 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6640 This defines the end of the @code{PT_GNU_RELRO} segment when
6641 @samp{-z relro} option is used.
6642 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6643 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6644 @var{exp} + @var{offset} is aligned to the @var{commonpagesize}
6645 argument given to @code{DATA_SEGMENT_ALIGN}. If present in the linker
6646 script, it must be placed between @code{DATA_SEGMENT_ALIGN} and
6647 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6648 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6652 . = DATA_SEGMENT_RELRO_END(24, .);
6655 @item DEFINED(@var{symbol})
6656 @kindex DEFINED(@var{symbol})
6657 @cindex symbol defaults
6658 Return 1 if @var{symbol} is in the linker global symbol table and is
6659 defined before the statement using DEFINED in the script, otherwise
6660 return 0. You can use this function to provide
6661 default values for symbols. For example, the following script fragment
6662 shows how to set a global symbol @samp{begin} to the first location in
6663 the @samp{.text} section---but if a symbol called @samp{begin} already
6664 existed, its value is preserved:
6670 begin = DEFINED(begin) ? begin : . ;
6678 @item LENGTH(@var{memory})
6679 @kindex LENGTH(@var{memory})
6680 Return the length of the memory region named @var{memory}.
6682 @item LOADADDR(@var{section})
6683 @kindex LOADADDR(@var{section})
6684 @cindex section load address in expression
6685 Return the absolute LMA of the named @var{section}. (@pxref{Output
6688 @item LOG2CEIL(@var{exp})
6689 @kindex LOG2CEIL(@var{exp})
6690 Return the binary logarithm of @var{exp} rounded towards infinity.
6691 @code{LOG2CEIL(0)} returns 0.
6694 @item MAX(@var{exp1}, @var{exp2})
6695 Returns the maximum of @var{exp1} and @var{exp2}.
6698 @item MIN(@var{exp1}, @var{exp2})
6699 Returns the minimum of @var{exp1} and @var{exp2}.
6701 @item NEXT(@var{exp})
6702 @kindex NEXT(@var{exp})
6703 @cindex unallocated address, next
6704 Return the next unallocated address that is a multiple of @var{exp}.
6705 This function is closely related to @code{ALIGN(@var{exp})}; unless you
6706 use the @code{MEMORY} command to define discontinuous memory for the
6707 output file, the two functions are equivalent.
6709 @item ORIGIN(@var{memory})
6710 @kindex ORIGIN(@var{memory})
6711 Return the origin of the memory region named @var{memory}.
6713 @item SEGMENT_START(@var{segment}, @var{default})
6714 @kindex SEGMENT_START(@var{segment}, @var{default})
6715 Return the base address of the named @var{segment}. If an explicit
6716 value has already been given for this segment (with a command-line
6717 @samp{-T} option) then that value will be returned otherwise the value
6718 will be @var{default}. At present, the @samp{-T} command-line option
6719 can only be used to set the base address for the ``text'', ``data'', and
6720 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
6723 @item SIZEOF(@var{section})
6724 @kindex SIZEOF(@var{section})
6725 @cindex section size
6726 Return the size in bytes of the named @var{section}, if that section has
6727 been allocated. If the section has not been allocated when this is
6728 evaluated, the linker will report an error. In the following example,
6729 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6738 symbol_1 = .end - .start ;
6739 symbol_2 = SIZEOF(.output);
6744 @item SIZEOF_HEADERS
6745 @itemx sizeof_headers
6746 @kindex SIZEOF_HEADERS
6748 Return the size in bytes of the output file's headers. This is
6749 information which appears at the start of the output file. You can use
6750 this number when setting the start address of the first section, if you
6751 choose, to facilitate paging.
6753 @cindex not enough room for program headers
6754 @cindex program headers, not enough room
6755 When producing an ELF output file, if the linker script uses the
6756 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6757 number of program headers before it has determined all the section
6758 addresses and sizes. If the linker later discovers that it needs
6759 additional program headers, it will report an error @samp{not enough
6760 room for program headers}. To avoid this error, you must avoid using
6761 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6762 script to avoid forcing the linker to use additional program headers, or
6763 you must define the program headers yourself using the @code{PHDRS}
6764 command (@pxref{PHDRS}).
6767 @node Implicit Linker Scripts
6768 @section Implicit Linker Scripts
6769 @cindex implicit linker scripts
6770 If you specify a linker input file which the linker can not recognize as
6771 an object file or an archive file, it will try to read the file as a
6772 linker script. If the file can not be parsed as a linker script, the
6773 linker will report an error.
6775 An implicit linker script will not replace the default linker script.
6777 Typically an implicit linker script would contain only symbol
6778 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6781 Any input files read because of an implicit linker script will be read
6782 at the position in the command line where the implicit linker script was
6783 read. This can affect archive searching.
6786 @node Machine Dependent
6787 @chapter Machine Dependent Features
6789 @cindex machine dependencies
6790 @command{ld} has additional features on some platforms; the following
6791 sections describe them. Machines where @command{ld} has no additional
6792 functionality are not listed.
6796 * H8/300:: @command{ld} and the H8/300
6799 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6802 * ARM:: @command{ld} and the ARM family
6805 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6808 * M68K:: @command{ld} and the Motorola 68K family
6811 * MIPS:: @command{ld} and the MIPS family
6814 * MMIX:: @command{ld} and MMIX
6817 * MSP430:: @command{ld} and MSP430
6820 * NDS32:: @command{ld} and NDS32
6823 * Nios II:: @command{ld} and the Altera Nios II
6826 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6829 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6832 * S/390 ELF:: @command{ld} and S/390 ELF Support
6835 * SPU ELF:: @command{ld} and SPU ELF Support
6838 * TI COFF:: @command{ld} and TI COFF
6841 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6844 * Xtensa:: @command{ld} and Xtensa Processors
6855 @section @command{ld} and the H8/300
6857 @cindex H8/300 support
6858 For the H8/300, @command{ld} can perform these global optimizations when
6859 you specify the @samp{--relax} command-line option.
6862 @cindex relaxing on H8/300
6863 @item relaxing address modes
6864 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6865 targets are within eight bits, and turns them into eight-bit
6866 program-counter relative @code{bsr} and @code{bra} instructions,
6869 @cindex synthesizing on H8/300
6870 @item synthesizing instructions
6871 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6872 @command{ld} finds all @code{mov.b} instructions which use the
6873 sixteen-bit absolute address form, but refer to the top
6874 page of memory, and changes them to use the eight-bit address form.
6875 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6876 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6877 top page of memory).
6879 @command{ld} finds all @code{mov} instructions which use the register
6880 indirect with 32-bit displacement addressing mode, but use a small
6881 displacement inside 16-bit displacement range, and changes them to use
6882 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6883 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6884 whenever the displacement @var{d} is in the 16 bit signed integer
6885 range. Only implemented in ELF-format ld).
6887 @item bit manipulation instructions
6888 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6889 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6890 which use 32 bit and 16 bit absolute address form, but refer to the top
6891 page of memory, and changes them to use the 8 bit address form.
6892 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6893 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6894 the top page of memory).
6896 @item system control instructions
6897 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6898 32 bit absolute address form, but refer to the top page of memory, and
6899 changes them to use 16 bit address form.
6900 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6901 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6902 the top page of memory).
6912 @c This stuff is pointless to say unless you're especially concerned
6913 @c with Renesas chips; don't enable it for generic case, please.
6915 @chapter @command{ld} and Other Renesas Chips
6917 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6918 H8/500, and SH chips. No special features, commands, or command-line
6919 options are required for these chips.
6933 @node M68HC11/68HC12
6934 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6936 @cindex M68HC11 and 68HC12 support
6938 @subsection Linker Relaxation
6940 For the Motorola 68HC11, @command{ld} can perform these global
6941 optimizations when you specify the @samp{--relax} command-line option.
6944 @cindex relaxing on M68HC11
6945 @item relaxing address modes
6946 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6947 targets are within eight bits, and turns them into eight-bit
6948 program-counter relative @code{bsr} and @code{bra} instructions,
6951 @command{ld} also looks at all 16-bit extended addressing modes and
6952 transforms them in a direct addressing mode when the address is in
6953 page 0 (between 0 and 0x0ff).
6955 @item relaxing gcc instruction group
6956 When @command{gcc} is called with @option{-mrelax}, it can emit group
6957 of instructions that the linker can optimize to use a 68HC11 direct
6958 addressing mode. These instructions consists of @code{bclr} or
6959 @code{bset} instructions.
6963 @subsection Trampoline Generation
6965 @cindex trampoline generation on M68HC11
6966 @cindex trampoline generation on M68HC12
6967 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6968 call a far function using a normal @code{jsr} instruction. The linker
6969 will also change the relocation to some far function to use the
6970 trampoline address instead of the function address. This is typically the
6971 case when a pointer to a function is taken. The pointer will in fact
6972 point to the function trampoline.
6980 @section @command{ld} and the ARM family
6982 @cindex ARM interworking support
6983 @kindex --support-old-code
6984 For the ARM, @command{ld} will generate code stubs to allow functions calls
6985 between ARM and Thumb code. These stubs only work with code that has
6986 been compiled and assembled with the @samp{-mthumb-interwork} command
6987 line option. If it is necessary to link with old ARM object files or
6988 libraries, which have not been compiled with the -mthumb-interwork
6989 option then the @samp{--support-old-code} command-line switch should be
6990 given to the linker. This will make it generate larger stub functions
6991 which will work with non-interworking aware ARM code. Note, however,
6992 the linker does not support generating stubs for function calls to
6993 non-interworking aware Thumb code.
6995 @cindex thumb entry point
6996 @cindex entry point, thumb
6997 @kindex --thumb-entry=@var{entry}
6998 The @samp{--thumb-entry} switch is a duplicate of the generic
6999 @samp{--entry} switch, in that it sets the program's starting address.
7000 But it also sets the bottom bit of the address, so that it can be
7001 branched to using a BX instruction, and the program will start
7002 executing in Thumb mode straight away.
7004 @cindex PE import table prefixing
7005 @kindex --use-nul-prefixed-import-tables
7006 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
7007 the import tables idata4 and idata5 have to be generated with a zero
7008 element prefix for import libraries. This is the old style to generate
7009 import tables. By default this option is turned off.
7013 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
7014 executables. This option is only valid when linking big-endian
7015 objects - ie ones which have been assembled with the @option{-EB}
7016 option. The resulting image will contain big-endian data and
7020 @kindex --target1-rel
7021 @kindex --target1-abs
7022 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
7023 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
7024 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
7025 and @samp{--target1-abs} switches override the default.
7028 @kindex --target2=@var{type}
7029 The @samp{--target2=type} switch overrides the default definition of the
7030 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
7031 meanings, and target defaults are as follows:
7034 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
7036 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
7038 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
7043 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
7044 specification) enables objects compiled for the ARMv4 architecture to be
7045 interworking-safe when linked with other objects compiled for ARMv4t, but
7046 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
7048 In the latter case, the switch @option{--fix-v4bx} must be passed to the
7049 linker, which causes v4t @code{BX rM} instructions to be rewritten as
7050 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
7052 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
7053 relocations are ignored.
7055 @cindex FIX_V4BX_INTERWORKING
7056 @kindex --fix-v4bx-interworking
7057 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
7058 relocations with a branch to the following veneer:
7066 This allows generation of libraries/applications that work on ARMv4 cores
7067 and are still interworking safe. Note that the above veneer clobbers the
7068 condition flags, so may cause incorrect program behavior in rare cases.
7072 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
7073 BLX instructions (available on ARMv5t and above) in various
7074 situations. Currently it is used to perform calls via the PLT from Thumb
7075 code using BLX rather than using BX and a mode-switching stub before
7076 each PLT entry. This should lead to such calls executing slightly faster.
7078 This option is enabled implicitly for SymbianOS, so there is no need to
7079 specify it if you are using that target.
7081 @cindex VFP11_DENORM_FIX
7082 @kindex --vfp11-denorm-fix
7083 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
7084 bug in certain VFP11 coprocessor hardware, which sometimes allows
7085 instructions with denorm operands (which must be handled by support code)
7086 to have those operands overwritten by subsequent instructions before
7087 the support code can read the intended values.
7089 The bug may be avoided in scalar mode if you allow at least one
7090 intervening instruction between a VFP11 instruction which uses a register
7091 and another instruction which writes to the same register, or at least two
7092 intervening instructions if vector mode is in use. The bug only affects
7093 full-compliance floating-point mode: you do not need this workaround if
7094 you are using "runfast" mode. Please contact ARM for further details.
7096 If you know you are using buggy VFP11 hardware, you can
7097 enable this workaround by specifying the linker option
7098 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
7099 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
7100 vector mode (the latter also works for scalar code). The default is
7101 @samp{--vfp-denorm-fix=none}.
7103 If the workaround is enabled, instructions are scanned for
7104 potentially-troublesome sequences, and a veneer is created for each
7105 such sequence which may trigger the erratum. The veneer consists of the
7106 first instruction of the sequence and a branch back to the subsequent
7107 instruction. The original instruction is then replaced with a branch to
7108 the veneer. The extra cycles required to call and return from the veneer
7109 are sufficient to avoid the erratum in both the scalar and vector cases.
7111 @cindex ARM1176 erratum workaround
7112 @kindex --fix-arm1176
7113 @kindex --no-fix-arm1176
7114 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
7115 in certain ARM1176 processors. The workaround is enabled by default if you
7116 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
7117 unconditionally by specifying @samp{--no-fix-arm1176}.
7119 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
7120 Programmer Advice Notice'' available on the ARM documentation website at:
7121 http://infocenter.arm.com/.
7123 @cindex STM32L4xx erratum workaround
7124 @kindex --fix-stm32l4xx-629360
7126 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
7127 workaround for a bug in the bus matrix / memory controller for some of
7128 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
7129 off-chip memory via the affected bus for bus reads of 9 words or more,
7130 the bus can generate corrupt data and/or abort. These are only
7131 core-initiated accesses (not DMA), and might affect any access:
7132 integer loads such as LDM, POP and floating-point loads such as VLDM,
7133 VPOP. Stores are not affected.
7135 The bug can be avoided by splitting memory accesses into the
7136 necessary chunks to keep bus reads below 8 words.
7138 The workaround is not enabled by default, this is equivalent to use
7139 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
7140 STM32L4xx hardware, you can enable the workaround by specifying the
7141 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
7142 @samp{--fix-stm32l4xx-629360=default}.
7144 If the workaround is enabled, instructions are scanned for
7145 potentially-troublesome sequences, and a veneer is created for each
7146 such sequence which may trigger the erratum. The veneer consists in a
7147 replacement sequence emulating the behaviour of the original one and a
7148 branch back to the subsequent instruction. The original instruction is
7149 then replaced with a branch to the veneer.
7151 The workaround does not always preserve the memory access order for
7152 the LDMDB instruction, when the instruction loads the PC.
7154 The workaround is not able to handle problematic instructions when
7155 they are in the middle of an IT block, since a branch is not allowed
7156 there. In that case, the linker reports a warning and no replacement
7159 The workaround is not able to replace problematic instructions with a
7160 PC-relative branch instruction if the @samp{.text} section is too
7161 large. In that case, when the branch that replaces the original code
7162 cannot be encoded, the linker reports a warning and no replacement
7165 @cindex NO_ENUM_SIZE_WARNING
7166 @kindex --no-enum-size-warning
7167 The @option{--no-enum-size-warning} switch prevents the linker from
7168 warning when linking object files that specify incompatible EABI
7169 enumeration size attributes. For example, with this switch enabled,
7170 linking of an object file using 32-bit enumeration values with another
7171 using enumeration values fitted into the smallest possible space will
7174 @cindex NO_WCHAR_SIZE_WARNING
7175 @kindex --no-wchar-size-warning
7176 The @option{--no-wchar-size-warning} switch prevents the linker from
7177 warning when linking object files that specify incompatible EABI
7178 @code{wchar_t} size attributes. For example, with this switch enabled,
7179 linking of an object file using 32-bit @code{wchar_t} values with another
7180 using 16-bit @code{wchar_t} values will not be diagnosed.
7183 @kindex --pic-veneer
7184 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
7185 ARM/Thumb interworking veneers, even if the rest of the binary
7186 is not PIC. This avoids problems on uClinux targets where
7187 @samp{--emit-relocs} is used to generate relocatable binaries.
7189 @cindex STUB_GROUP_SIZE
7190 @kindex --stub-group-size=@var{N}
7191 The linker will automatically generate and insert small sequences of
7192 code into a linked ARM ELF executable whenever an attempt is made to
7193 perform a function call to a symbol that is too far away. The
7194 placement of these sequences of instructions - called stubs - is
7195 controlled by the command-line option @option{--stub-group-size=N}.
7196 The placement is important because a poor choice can create a need for
7197 duplicate stubs, increasing the code size. The linker will try to
7198 group stubs together in order to reduce interruptions to the flow of
7199 code, but it needs guidance as to how big these groups should be and
7200 where they should be placed.
7202 The value of @samp{N}, the parameter to the
7203 @option{--stub-group-size=} option controls where the stub groups are
7204 placed. If it is negative then all stubs are placed after the first
7205 branch that needs them. If it is positive then the stubs can be
7206 placed either before or after the branches that need them. If the
7207 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
7208 exactly where to place groups of stubs, using its built in heuristics.
7209 A value of @samp{N} greater than 1 (or smaller than -1) tells the
7210 linker that a single group of stubs can service at most @samp{N} bytes
7211 from the input sections.
7213 The default, if @option{--stub-group-size=} is not specified, is
7216 Farcalls stubs insertion is fully supported for the ARM-EABI target
7217 only, because it relies on object files properties not present
7220 @cindex Cortex-A8 erratum workaround
7221 @kindex --fix-cortex-a8
7222 @kindex --no-fix-cortex-a8
7223 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}.
7225 The erratum only affects Thumb-2 code. Please contact ARM for further details.
7227 @cindex Cortex-A53 erratum 835769 workaround
7228 @kindex --fix-cortex-a53-835769
7229 @kindex --no-fix-cortex-a53-835769
7230 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}.
7232 Please contact ARM for further details.
7234 @kindex --merge-exidx-entries
7235 @kindex --no-merge-exidx-entries
7236 @cindex Merging exidx entries
7237 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
7240 @cindex 32-bit PLT entries
7241 The @samp{--long-plt} option enables the use of 16 byte PLT entries
7242 which support up to 4Gb of code. The default is to use 12 byte PLT
7243 entries which only support 512Mb of code.
7245 @kindex --no-apply-dynamic-relocs
7246 @cindex AArch64 rela addend
7247 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
7248 link-time values for dynamic relocations.
7250 @cindex Placement of SG veneers
7251 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
7252 Its start address must be set, either with the command-line option
7253 @samp{--section-start} or in a linker script, to indicate where to place these
7256 @kindex --cmse-implib
7257 @cindex Secure gateway import library
7258 The @samp{--cmse-implib} option requests that the import libraries
7259 specified by the @samp{--out-implib} and @samp{--in-implib} options are
7260 secure gateway import libraries, suitable for linking a non-secure
7261 executable against secure code as per ARMv8-M Security Extensions.
7263 @kindex --in-implib=@var{file}
7264 @cindex Input import library
7265 The @samp{--in-implib=file} specifies an input import library whose symbols
7266 must keep the same address in the executable being produced. A warning is
7267 given if no @samp{--out-implib} is given but new symbols have been introduced
7268 in the executable that should be listed in its import library. Otherwise, if
7269 @samp{--out-implib} is specified, the symbols are added to the output import
7270 library. A warning is also given if some symbols present in the input import
7271 library have disappeared from the executable. This option is only effective
7272 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
7286 @section @command{ld} and HPPA 32-bit ELF Support
7287 @cindex HPPA multiple sub-space stubs
7288 @kindex --multi-subspace
7289 When generating a shared library, @command{ld} will by default generate
7290 import stubs suitable for use with a single sub-space application.
7291 The @samp{--multi-subspace} switch causes @command{ld} to generate export
7292 stubs, and different (larger) import stubs suitable for use with
7293 multiple sub-spaces.
7295 @cindex HPPA stub grouping
7296 @kindex --stub-group-size=@var{N}
7297 Long branch stubs and import/export stubs are placed by @command{ld} in
7298 stub sections located between groups of input sections.
7299 @samp{--stub-group-size} specifies the maximum size of a group of input
7300 sections handled by one stub section. Since branch offsets are signed,
7301 a stub section may serve two groups of input sections, one group before
7302 the stub section, and one group after it. However, when using
7303 conditional branches that require stubs, it may be better (for branch
7304 prediction) that stub sections only serve one group of input sections.
7305 A negative value for @samp{N} chooses this scheme, ensuring that
7306 branches to stubs always use a negative offset. Two special values of
7307 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7308 @command{ld} to automatically size input section groups for the branch types
7309 detected, with the same behaviour regarding stub placement as other
7310 positive or negative values of @samp{N} respectively.
7312 Note that @samp{--stub-group-size} does not split input sections. A
7313 single input section larger than the group size specified will of course
7314 create a larger group (of one section). If input sections are too
7315 large, it may not be possible for a branch to reach its stub.
7328 @section @command{ld} and the Motorola 68K family
7330 @cindex Motorola 68K GOT generation
7331 @kindex --got=@var{type}
7332 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
7333 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
7334 @samp{target}. When @samp{target} is selected the linker chooses
7335 the default GOT generation scheme for the current target.
7336 @samp{single} tells the linker to generate a single GOT with
7337 entries only at non-negative offsets.
7338 @samp{negative} instructs the linker to generate a single GOT with
7339 entries at both negative and positive offsets. Not all environments
7341 @samp{multigot} allows the linker to generate several GOTs in the
7342 output file. All GOT references from a single input object
7343 file access the same GOT, but references from different input object
7344 files might access different GOTs. Not all environments support such GOTs.
7357 @section @command{ld} and the MIPS family
7359 @cindex MIPS microMIPS instruction choice selection
7362 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
7363 microMIPS instructions used in code generated by the linker, such as that
7364 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
7365 used, then the linker only uses 32-bit instruction encodings. By default
7366 or if @samp{--no-insn32} is used, all instruction encodings are used,
7367 including 16-bit ones where possible.
7369 @cindex MIPS branch relocation check control
7370 @kindex --ignore-branch-isa
7371 @kindex --no-ignore-branch-isa
7372 The @samp{--ignore-branch-isa} and @samp{--no-ignore-branch-isa} options
7373 control branch relocation checks for invalid ISA mode transitions. If
7374 @samp{--ignore-branch-isa} is used, then the linker accepts any branch
7375 relocations and any ISA mode transition required is lost in relocation
7376 calculation, except for some cases of @code{BAL} instructions which meet
7377 relaxation conditions and are converted to equivalent @code{JALX}
7378 instructions as the associated relocation is calculated. By default
7379 or if @samp{--no-ignore-branch-isa} is used a check is made causing
7380 the loss of an ISA mode transition to produce an error.
7393 @section @code{ld} and MMIX
7394 For MMIX, there is a choice of generating @code{ELF} object files or
7395 @code{mmo} object files when linking. The simulator @code{mmix}
7396 understands the @code{mmo} format. The binutils @code{objcopy} utility
7397 can translate between the two formats.
7399 There is one special section, the @samp{.MMIX.reg_contents} section.
7400 Contents in this section is assumed to correspond to that of global
7401 registers, and symbols referring to it are translated to special symbols,
7402 equal to registers. In a final link, the start address of the
7403 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7404 global register multiplied by 8. Register @code{$255} is not included in
7405 this section; it is always set to the program entry, which is at the
7406 symbol @code{Main} for @code{mmo} files.
7408 Global symbols with the prefix @code{__.MMIX.start.}, for example
7409 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7410 The default linker script uses these to set the default start address
7413 Initial and trailing multiples of zero-valued 32-bit words in a section,
7414 are left out from an mmo file.
7427 @section @code{ld} and MSP430
7428 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7429 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7430 just pass @samp{-m help} option to the linker).
7432 @cindex MSP430 extra sections
7433 The linker will recognize some extra sections which are MSP430 specific:
7436 @item @samp{.vectors}
7437 Defines a portion of ROM where interrupt vectors located.
7439 @item @samp{.bootloader}
7440 Defines the bootloader portion of the ROM (if applicable). Any code
7441 in this section will be uploaded to the MPU.
7443 @item @samp{.infomem}
7444 Defines an information memory section (if applicable). Any code in
7445 this section will be uploaded to the MPU.
7447 @item @samp{.infomemnobits}
7448 This is the same as the @samp{.infomem} section except that any code
7449 in this section will not be uploaded to the MPU.
7451 @item @samp{.noinit}
7452 Denotes a portion of RAM located above @samp{.bss} section.
7454 The last two sections are used by gcc.
7458 @cindex MSP430 Options
7459 @kindex --code-region
7460 @item --code-region=[either,lower,upper,none]
7461 This will transform .text* sections to [either,lower,upper].text* sections. The
7462 argument passed to GCC for -mcode-region is propagated to the linker
7465 @kindex --data-region
7466 @item --data-region=[either,lower,upper,none]
7467 This will transform .data*, .bss* and .rodata* sections to
7468 [either,lower,upper].[data,bss,rodata]* sections. The argument passed to GCC
7469 for -mdata-region is propagated to the linker using this option.
7471 @kindex --disable-sec-transformation
7472 @item --disable-sec-transformation
7473 Prevent the transformation of sections as specified by the @code{--code-region}
7474 and @code{--data-region} options.
7475 This is useful if you are compiling and linking using a single call to the GCC
7476 wrapper, and want to compile the source files using -m[code,data]-region but
7477 not transform the sections for prebuilt libraries and objects.
7491 @section @code{ld} and NDS32
7492 @kindex relaxing on NDS32
7493 For NDS32, there are some options to select relaxation behavior. The linker
7494 relaxes objects according to these options.
7497 @item @samp{--m[no-]fp-as-gp}
7498 Disable/enable fp-as-gp relaxation.
7500 @item @samp{--mexport-symbols=FILE}
7501 Exporting symbols and their address into FILE as linker script.
7503 @item @samp{--m[no-]ex9}
7504 Disable/enable link-time EX9 relaxation.
7506 @item @samp{--mexport-ex9=FILE}
7507 Export the EX9 table after linking.
7509 @item @samp{--mimport-ex9=FILE}
7510 Import the Ex9 table for EX9 relaxation.
7512 @item @samp{--mupdate-ex9}
7513 Update the existing EX9 table.
7515 @item @samp{--mex9-limit=NUM}
7516 Maximum number of entries in the ex9 table.
7518 @item @samp{--mex9-loop-aware}
7519 Avoid generating the EX9 instruction inside the loop.
7521 @item @samp{--m[no-]ifc}
7522 Disable/enable the link-time IFC optimization.
7524 @item @samp{--mifc-loop-aware}
7525 Avoid generating the IFC instruction inside the loop.
7539 @section @command{ld} and the Altera Nios II
7540 @cindex Nios II call relaxation
7541 @kindex --relax on Nios II
7543 Call and immediate jump instructions on Nios II processors are limited to
7544 transferring control to addresses in the same 256MB memory segment,
7545 which may result in @command{ld} giving
7546 @samp{relocation truncated to fit} errors with very large programs.
7547 The command-line option @option{--relax} enables the generation of
7548 trampolines that can access the entire 32-bit address space for calls
7549 outside the normal @code{call} and @code{jmpi} address range. These
7550 trampolines are inserted at section boundaries, so may not themselves
7551 be reachable if an input section and its associated call trampolines are
7554 The @option{--relax} option is enabled by default unless @option{-r}
7555 is also specified. You can disable trampoline generation by using the
7556 @option{--no-relax} linker option. You can also disable this optimization
7557 locally by using the @samp{set .noat} directive in assembly-language
7558 source files, as the linker-inserted trampolines use the @code{at}
7559 register as a temporary.
7561 Note that the linker @option{--relax} option is independent of assembler
7562 relaxation options, and that using the GNU assembler's @option{-relax-all}
7563 option interferes with the linker's more selective call instruction relaxation.
7576 @section @command{ld} and PowerPC 32-bit ELF Support
7577 @cindex PowerPC long branches
7578 @kindex --relax on PowerPC
7579 Branches on PowerPC processors are limited to a signed 26-bit
7580 displacement, which may result in @command{ld} giving
7581 @samp{relocation truncated to fit} errors with very large programs.
7582 @samp{--relax} enables the generation of trampolines that can access
7583 the entire 32-bit address space. These trampolines are inserted at
7584 section boundaries, so may not themselves be reachable if an input
7585 section exceeds 33M in size. You may combine @samp{-r} and
7586 @samp{--relax} to add trampolines in a partial link. In that case
7587 both branches to undefined symbols and inter-section branches are also
7588 considered potentially out of range, and trampolines inserted.
7590 @cindex PowerPC ELF32 options
7595 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7596 generates code capable of using a newer PLT and GOT layout that has
7597 the security advantage of no executable section ever needing to be
7598 writable and no writable section ever being executable. PowerPC
7599 @command{ld} will generate this layout, including stubs to access the
7600 PLT, if all input files (including startup and static libraries) were
7601 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7602 BSS PLT (and GOT layout) which can give slightly better performance.
7604 @kindex --secure-plt
7606 @command{ld} will use the new PLT and GOT layout if it is linking new
7607 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7608 when linking non-PIC code. This option requests the new PLT and GOT
7609 layout. A warning will be given if some object file requires the old
7615 The new secure PLT and GOT are placed differently relative to other
7616 sections compared to older BSS PLT and GOT placement. The location of
7617 @code{.plt} must change because the new secure PLT is an initialized
7618 section while the old PLT is uninitialized. The reason for the
7619 @code{.got} change is more subtle: The new placement allows
7620 @code{.got} to be read-only in applications linked with
7621 @samp{-z relro -z now}. However, this placement means that
7622 @code{.sdata} cannot always be used in shared libraries, because the
7623 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7624 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7625 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7626 really only useful for other compilers that may do so.
7628 @cindex PowerPC stub symbols
7629 @kindex --emit-stub-syms
7630 @item --emit-stub-syms
7631 This option causes @command{ld} to label linker stubs with a local
7632 symbol that encodes the stub type and destination.
7634 @cindex PowerPC TLS optimization
7635 @kindex --no-tls-optimize
7636 @item --no-tls-optimize
7637 PowerPC @command{ld} normally performs some optimization of code
7638 sequences used to access Thread-Local Storage. Use this option to
7639 disable the optimization.
7652 @node PowerPC64 ELF64
7653 @section @command{ld} and PowerPC64 64-bit ELF Support
7655 @cindex PowerPC64 ELF64 options
7657 @cindex PowerPC64 stub grouping
7658 @kindex --stub-group-size
7659 @item --stub-group-size
7660 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
7661 by @command{ld} in stub sections located between groups of input sections.
7662 @samp{--stub-group-size} specifies the maximum size of a group of input
7663 sections handled by one stub section. Since branch offsets are signed,
7664 a stub section may serve two groups of input sections, one group before
7665 the stub section, and one group after it. However, when using
7666 conditional branches that require stubs, it may be better (for branch
7667 prediction) that stub sections only serve one group of input sections.
7668 A negative value for @samp{N} chooses this scheme, ensuring that
7669 branches to stubs always use a negative offset. Two special values of
7670 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7671 @command{ld} to automatically size input section groups for the branch types
7672 detected, with the same behaviour regarding stub placement as other
7673 positive or negative values of @samp{N} respectively.
7675 Note that @samp{--stub-group-size} does not split input sections. A
7676 single input section larger than the group size specified will of course
7677 create a larger group (of one section). If input sections are too
7678 large, it may not be possible for a branch to reach its stub.
7680 @cindex PowerPC64 stub symbols
7681 @kindex --emit-stub-syms
7682 @item --emit-stub-syms
7683 This option causes @command{ld} to label linker stubs with a local
7684 symbol that encodes the stub type and destination.
7686 @cindex PowerPC64 dot symbols
7688 @kindex --no-dotsyms
7691 These two options control how @command{ld} interprets version patterns
7692 in a version script. Older PowerPC64 compilers emitted both a
7693 function descriptor symbol with the same name as the function, and a
7694 code entry symbol with the name prefixed by a dot (@samp{.}). To
7695 properly version a function @samp{foo}, the version script thus needs
7696 to control both @samp{foo} and @samp{.foo}. The option
7697 @samp{--dotsyms}, on by default, automatically adds the required
7698 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
7701 @cindex PowerPC64 register save/restore functions
7702 @kindex --save-restore-funcs
7703 @kindex --no-save-restore-funcs
7704 @item --save-restore-funcs
7705 @itemx --no-save-restore-funcs
7706 These two options control whether PowerPC64 @command{ld} automatically
7707 provides out-of-line register save and restore functions used by
7708 @samp{-Os} code. The default is to provide any such referenced
7709 function for a normal final link, and to not do so for a relocatable
7712 @cindex PowerPC64 TLS optimization
7713 @kindex --no-tls-optimize
7714 @item --no-tls-optimize
7715 PowerPC64 @command{ld} normally performs some optimization of code
7716 sequences used to access Thread-Local Storage. Use this option to
7717 disable the optimization.
7719 @cindex PowerPC64 __tls_get_addr optimization
7720 @kindex --tls-get-addr-optimize
7721 @kindex --no-tls-get-addr-optimize
7722 @kindex --tls-get-addr-regsave
7723 @kindex --no-tls-get-addr-regsave
7724 @item --tls-get-addr-optimize
7725 @itemx --no-tls-get-addr-optimize
7726 These options control how PowerPC64 @command{ld} uses a special
7727 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
7728 an optimization that allows the second and subsequent calls to
7729 @code{__tls_get_addr} for a given symbol to be resolved by the special
7730 stub without calling in to glibc. By default the linker enables
7731 generation of the stub when glibc advertises the availability of
7733 Using @option{--tls-get-addr-optimize} with an older glibc won't do
7734 much besides slow down your applications, but may be useful if linking
7735 an application against an older glibc with the expectation that it
7736 will normally be used on systems having a newer glibc.
7737 @option{--tls-get-addr-regsave} forces generation of a stub that saves
7738 and restores volatile registers around the call into glibc. Normally,
7739 this is done when the linker detects a call to __tls_get_addr_desc.
7740 Such calls then go via the register saving stub to __tls_get_addr_opt.
7741 @option{--no-tls-get-addr-regsave} disables generation of the
7744 @cindex PowerPC64 OPD optimization
7745 @kindex --no-opd-optimize
7746 @item --no-opd-optimize
7747 PowerPC64 @command{ld} normally removes @code{.opd} section entries
7748 corresponding to deleted link-once functions, or functions removed by
7749 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
7750 Use this option to disable @code{.opd} optimization.
7752 @cindex PowerPC64 OPD spacing
7753 @kindex --non-overlapping-opd
7754 @item --non-overlapping-opd
7755 Some PowerPC64 compilers have an option to generate compressed
7756 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
7757 the static chain pointer (unused in C) with the first word of the next
7758 entry. This option expands such entries to the full 24 bytes.
7760 @cindex PowerPC64 TOC optimization
7761 @kindex --no-toc-optimize
7762 @item --no-toc-optimize
7763 PowerPC64 @command{ld} normally removes unused @code{.toc} section
7764 entries. Such entries are detected by examining relocations that
7765 reference the TOC in code sections. A reloc in a deleted code section
7766 marks a TOC word as unneeded, while a reloc in a kept code section
7767 marks a TOC word as needed. Since the TOC may reference itself, TOC
7768 relocs are also examined. TOC words marked as both needed and
7769 unneeded will of course be kept. TOC words without any referencing
7770 reloc are assumed to be part of a multi-word entry, and are kept or
7771 discarded as per the nearest marked preceding word. This works
7772 reliably for compiler generated code, but may be incorrect if assembly
7773 code is used to insert TOC entries. Use this option to disable the
7776 @cindex PowerPC64 multi-TOC
7777 @kindex --no-multi-toc
7778 @item --no-multi-toc
7779 If given any toc option besides @code{-mcmodel=medium} or
7780 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
7782 entries are accessed with a 16-bit offset from r2. This limits the
7783 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
7784 grouping code sections such that each group uses less than 64K for its
7785 TOC entries, then inserts r2 adjusting stubs between inter-group
7786 calls. @command{ld} does not split apart input sections, so cannot
7787 help if a single input file has a @code{.toc} section that exceeds
7788 64K, most likely from linking multiple files with @command{ld -r}.
7789 Use this option to turn off this feature.
7791 @cindex PowerPC64 TOC sorting
7792 @kindex --no-toc-sort
7794 By default, @command{ld} sorts TOC sections so that those whose file
7795 happens to have a section called @code{.init} or @code{.fini} are
7796 placed first, followed by TOC sections referenced by code generated
7797 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
7798 referenced only by code generated with PowerPC64 gcc's
7799 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
7800 results in better TOC grouping for multi-TOC. Use this option to turn
7803 @cindex PowerPC64 PLT stub alignment
7805 @kindex --no-plt-align
7807 @itemx --no-plt-align
7808 Use these options to control whether individual PLT call stubs are
7809 aligned to a 32-byte boundary, or to the specified power of two
7810 boundary when using @code{--plt-align=}. A negative value may be
7811 specified to pad PLT call stubs so that they do not cross the
7812 specified power of two boundary (or the minimum number of boundaries
7813 if a PLT stub is so large that it must cross a boundary). By default
7814 PLT call stubs are aligned to 32-byte boundaries.
7816 @cindex PowerPC64 PLT call stub static chain
7817 @kindex --plt-static-chain
7818 @kindex --no-plt-static-chain
7819 @item --plt-static-chain
7820 @itemx --no-plt-static-chain
7821 Use these options to control whether PLT call stubs load the static
7822 chain pointer (r11). @code{ld} defaults to not loading the static
7823 chain since there is never any need to do so on a PLT call.
7825 @cindex PowerPC64 PLT call stub thread safety
7826 @kindex --plt-thread-safe
7827 @kindex --no-plt-thread-safe
7828 @item --plt-thread-safe
7829 @itemx --no-plt-thread-safe
7830 With power7's weakly ordered memory model, it is possible when using
7831 lazy binding for ld.so to update a plt entry in one thread and have
7832 another thread see the individual plt entry words update in the wrong
7833 order, despite ld.so carefully writing in the correct order and using
7834 memory write barriers. To avoid this we need some sort of read
7835 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
7836 looks for calls to commonly used functions that create threads, and if
7837 seen, adds the necessary barriers. Use these options to change the
7840 @cindex PowerPC64 ELFv2 PLT localentry optimization
7841 @kindex --plt-localentry
7842 @kindex --no-plt-localentry
7843 @item --plt-localentry
7844 @itemx --no-localentry
7845 ELFv2 functions with localentry:0 are those with a single entry point,
7846 ie. global entry == local entry, and that have no requirement on r2
7847 (the TOC/GOT pointer) or r12, and guarantee r2 is unchanged on return.
7848 Such an external function can be called via the PLT without saving r2
7849 or restoring it on return, avoiding a common load-hit-store for small
7850 functions. The optimization is attractive, with up to 40% reduction
7851 in execution time for a small function, but can result in symbol
7852 interposition failures. Also, minor changes in a shared library,
7853 including system libraries, can cause a function that was localentry:0
7854 to become localentry:8. This will result in a dynamic loader
7855 complaint and failure to run. The option is experimental, use with
7856 care. @option{--no-plt-localentry} is the default.
7870 @section @command{ld} and S/390 ELF Support
7872 @cindex S/390 ELF options
7876 @kindex --s390-pgste
7878 This option marks the result file with a @code{PT_S390_PGSTE}
7879 segment. The Linux kernel is supposed to allocate 4k page tables for
7880 binaries marked that way.
7894 @section @command{ld} and SPU ELF Support
7896 @cindex SPU ELF options
7902 This option marks an executable as a PIC plugin module.
7904 @cindex SPU overlays
7905 @kindex --no-overlays
7907 Normally, @command{ld} recognizes calls to functions within overlay
7908 regions, and redirects such calls to an overlay manager via a stub.
7909 @command{ld} also provides a built-in overlay manager. This option
7910 turns off all this special overlay handling.
7912 @cindex SPU overlay stub symbols
7913 @kindex --emit-stub-syms
7914 @item --emit-stub-syms
7915 This option causes @command{ld} to label overlay stubs with a local
7916 symbol that encodes the stub type and destination.
7918 @cindex SPU extra overlay stubs
7919 @kindex --extra-overlay-stubs
7920 @item --extra-overlay-stubs
7921 This option causes @command{ld} to add overlay call stubs on all
7922 function calls out of overlay regions. Normally stubs are not added
7923 on calls to non-overlay regions.
7925 @cindex SPU local store size
7926 @kindex --local-store=lo:hi
7927 @item --local-store=lo:hi
7928 @command{ld} usually checks that a final executable for SPU fits in
7929 the address range 0 to 256k. This option may be used to change the
7930 range. Disable the check entirely with @option{--local-store=0:0}.
7933 @kindex --stack-analysis
7934 @item --stack-analysis
7935 SPU local store space is limited. Over-allocation of stack space
7936 unnecessarily limits space available for code and data, while
7937 under-allocation results in runtime failures. If given this option,
7938 @command{ld} will provide an estimate of maximum stack usage.
7939 @command{ld} does this by examining symbols in code sections to
7940 determine the extents of functions, and looking at function prologues
7941 for stack adjusting instructions. A call-graph is created by looking
7942 for relocations on branch instructions. The graph is then searched
7943 for the maximum stack usage path. Note that this analysis does not
7944 find calls made via function pointers, and does not handle recursion
7945 and other cycles in the call graph. Stack usage may be
7946 under-estimated if your code makes such calls. Also, stack usage for
7947 dynamic allocation, e.g. alloca, will not be detected. If a link map
7948 is requested, detailed information about each function's stack usage
7949 and calls will be given.
7952 @kindex --emit-stack-syms
7953 @item --emit-stack-syms
7954 This option, if given along with @option{--stack-analysis} will result
7955 in @command{ld} emitting stack sizing symbols for each function.
7956 These take the form @code{__stack_<function_name>} for global
7957 functions, and @code{__stack_<number>_<function_name>} for static
7958 functions. @code{<number>} is the section id in hex. The value of
7959 such symbols is the stack requirement for the corresponding function.
7960 The symbol size will be zero, type @code{STT_NOTYPE}, binding
7961 @code{STB_LOCAL}, and section @code{SHN_ABS}.
7975 @section @command{ld}'s Support for Various TI COFF Versions
7976 @cindex TI COFF versions
7977 @kindex --format=@var{version}
7978 The @samp{--format} switch allows selection of one of the various
7979 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
7980 also supported. The TI COFF versions also vary in header byte-order
7981 format; @command{ld} will read any version or byte order, but the output
7982 header format depends on the default specified by the specific target.
7995 @section @command{ld} and WIN32 (cygwin/mingw)
7997 This section describes some of the win32 specific @command{ld} issues.
7998 See @ref{Options,,Command-line Options} for detailed description of the
7999 command-line options mentioned here.
8002 @cindex import libraries
8003 @item import libraries
8004 The standard Windows linker creates and uses so-called import
8005 libraries, which contains information for linking to dll's. They are
8006 regular static archives and are handled as any other static
8007 archive. The cygwin and mingw ports of @command{ld} have specific
8008 support for creating such libraries provided with the
8009 @samp{--out-implib} command-line option.
8011 @item exporting DLL symbols
8012 @cindex exporting DLL symbols
8013 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
8016 @item using auto-export functionality
8017 @cindex using auto-export functionality
8018 By default @command{ld} exports symbols with the auto-export functionality,
8019 which is controlled by the following command-line options:
8022 @item --export-all-symbols [This is the default]
8023 @item --exclude-symbols
8024 @item --exclude-libs
8025 @item --exclude-modules-for-implib
8026 @item --version-script
8029 When auto-export is in operation, @command{ld} will export all the non-local
8030 (global and common) symbols it finds in a DLL, with the exception of a few
8031 symbols known to belong to the system's runtime and libraries. As it will
8032 often not be desirable to export all of a DLL's symbols, which may include
8033 private functions that are not part of any public interface, the command-line
8034 options listed above may be used to filter symbols out from the list for
8035 exporting. The @samp{--output-def} option can be used in order to see the
8036 final list of exported symbols with all exclusions taken into effect.
8038 If @samp{--export-all-symbols} is not given explicitly on the
8039 command line, then the default auto-export behavior will be @emph{disabled}
8040 if either of the following are true:
8043 @item A DEF file is used.
8044 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
8047 @item using a DEF file
8048 @cindex using a DEF file
8049 Another way of exporting symbols is using a DEF file. A DEF file is
8050 an ASCII file containing definitions of symbols which should be
8051 exported when a dll is created. Usually it is named @samp{<dll
8052 name>.def} and is added as any other object file to the linker's
8053 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
8056 gcc -o <output> <objectfiles> <dll name>.def
8059 Using a DEF file turns off the normal auto-export behavior, unless the
8060 @samp{--export-all-symbols} option is also used.
8062 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
8065 LIBRARY "xyz.dll" BASE=0x20000000
8071 another_foo = abc.dll.afoo
8077 This example defines a DLL with a non-default base address and seven
8078 symbols in the export table. The third exported symbol @code{_bar} is an
8079 alias for the second. The fourth symbol, @code{another_foo} is resolved
8080 by "forwarding" to another module and treating it as an alias for
8081 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
8082 @code{var1} is declared to be a data object. The @samp{doo} symbol in
8083 export library is an alias of @samp{foo}, which gets the string name
8084 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
8085 symbol, which gets in export table the name @samp{var1}.
8087 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
8088 name of the output DLL. If @samp{<name>} does not include a suffix,
8089 the default library suffix, @samp{.DLL} is appended.
8091 When the .DEF file is used to build an application, rather than a
8092 library, the @code{NAME <name>} command should be used instead of
8093 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
8094 executable suffix, @samp{.EXE} is appended.
8096 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
8097 specification @code{BASE = <number>} may be used to specify a
8098 non-default base address for the image.
8100 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
8101 or they specify an empty string, the internal name is the same as the
8102 filename specified on the command line.
8104 The complete specification of an export symbol is:
8108 ( ( ( <name1> [ = <name2> ] )
8109 | ( <name1> = <module-name> . <external-name>))
8110 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
8113 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
8114 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
8115 @samp{<name1>} as a "forward" alias for the symbol
8116 @samp{<external-name>} in the DLL @samp{<module-name>}.
8117 Optionally, the symbol may be exported by the specified ordinal
8118 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
8119 string in import/export table for the symbol.
8121 The optional keywords that follow the declaration indicate:
8123 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
8124 will still be exported by its ordinal alias (either the value specified
8125 by the .def specification or, otherwise, the value assigned by the
8126 linker). The symbol name, however, does remain visible in the import
8127 library (if any), unless @code{PRIVATE} is also specified.
8129 @code{DATA}: The symbol is a variable or object, rather than a function.
8130 The import lib will export only an indirect reference to @code{foo} as
8131 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
8134 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
8135 well as @code{_imp__foo} into the import library. Both refer to the
8136 read-only import address table's pointer to the variable, not to the
8137 variable itself. This can be dangerous. If the user code fails to add
8138 the @code{dllimport} attribute and also fails to explicitly add the
8139 extra indirection that the use of the attribute enforces, the
8140 application will behave unexpectedly.
8142 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
8143 it into the static import library used to resolve imports at link time. The
8144 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
8145 API at runtime or by using the GNU ld extension of linking directly to
8146 the DLL without an import library.
8148 See ld/deffilep.y in the binutils sources for the full specification of
8149 other DEF file statements
8151 @cindex creating a DEF file
8152 While linking a shared dll, @command{ld} is able to create a DEF file
8153 with the @samp{--output-def <file>} command-line option.
8155 @item Using decorations
8156 @cindex Using decorations
8157 Another way of marking symbols for export is to modify the source code
8158 itself, so that when building the DLL each symbol to be exported is
8162 __declspec(dllexport) int a_variable
8163 __declspec(dllexport) void a_function(int with_args)
8166 All such symbols will be exported from the DLL. If, however,
8167 any of the object files in the DLL contain symbols decorated in
8168 this way, then the normal auto-export behavior is disabled, unless
8169 the @samp{--export-all-symbols} option is also used.
8171 Note that object files that wish to access these symbols must @emph{not}
8172 decorate them with dllexport. Instead, they should use dllimport,
8176 __declspec(dllimport) int a_variable
8177 __declspec(dllimport) void a_function(int with_args)
8180 This complicates the structure of library header files, because
8181 when included by the library itself the header must declare the
8182 variables and functions as dllexport, but when included by client
8183 code the header must declare them as dllimport. There are a number
8184 of idioms that are typically used to do this; often client code can
8185 omit the __declspec() declaration completely. See
8186 @samp{--enable-auto-import} and @samp{automatic data imports} for more
8190 @cindex automatic data imports
8191 @item automatic data imports
8192 The standard Windows dll format supports data imports from dlls only
8193 by adding special decorations (dllimport/dllexport), which let the
8194 compiler produce specific assembler instructions to deal with this
8195 issue. This increases the effort necessary to port existing Un*x
8196 code to these platforms, especially for large
8197 c++ libraries and applications. The auto-import feature, which was
8198 initially provided by Paul Sokolovsky, allows one to omit the
8199 decorations to achieve a behavior that conforms to that on POSIX/Un*x
8200 platforms. This feature is enabled with the @samp{--enable-auto-import}
8201 command-line option, although it is enabled by default on cygwin/mingw.
8202 The @samp{--enable-auto-import} option itself now serves mainly to
8203 suppress any warnings that are ordinarily emitted when linked objects
8204 trigger the feature's use.
8206 auto-import of variables does not always work flawlessly without
8207 additional assistance. Sometimes, you will see this message
8209 "variable '<var>' can't be auto-imported. Please read the
8210 documentation for ld's @code{--enable-auto-import} for details."
8212 The @samp{--enable-auto-import} documentation explains why this error
8213 occurs, and several methods that can be used to overcome this difficulty.
8214 One of these methods is the @emph{runtime pseudo-relocs} feature, described
8217 @cindex runtime pseudo-relocation
8218 For complex variables imported from DLLs (such as structs or classes),
8219 object files typically contain a base address for the variable and an
8220 offset (@emph{addend}) within the variable--to specify a particular
8221 field or public member, for instance. Unfortunately, the runtime loader used
8222 in win32 environments is incapable of fixing these references at runtime
8223 without the additional information supplied by dllimport/dllexport decorations.
8224 The standard auto-import feature described above is unable to resolve these
8227 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
8228 be resolved without error, while leaving the task of adjusting the references
8229 themselves (with their non-zero addends) to specialized code provided by the
8230 runtime environment. Recent versions of the cygwin and mingw environments and
8231 compilers provide this runtime support; older versions do not. However, the
8232 support is only necessary on the developer's platform; the compiled result will
8233 run without error on an older system.
8235 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
8238 @cindex direct linking to a dll
8239 @item direct linking to a dll
8240 The cygwin/mingw ports of @command{ld} support the direct linking,
8241 including data symbols, to a dll without the usage of any import
8242 libraries. This is much faster and uses much less memory than does the
8243 traditional import library method, especially when linking large
8244 libraries or applications. When @command{ld} creates an import lib, each
8245 function or variable exported from the dll is stored in its own bfd, even
8246 though a single bfd could contain many exports. The overhead involved in
8247 storing, loading, and processing so many bfd's is quite large, and explains the
8248 tremendous time, memory, and storage needed to link against particularly
8249 large or complex libraries when using import libs.
8251 Linking directly to a dll uses no extra command-line switches other than
8252 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
8253 of names to match each library. All that is needed from the developer's
8254 perspective is an understanding of this search, in order to force ld to
8255 select the dll instead of an import library.
8258 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
8259 to find, in the first directory of its search path,
8272 before moving on to the next directory in the search path.
8274 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
8275 where @samp{<prefix>} is set by the @command{ld} option
8276 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
8277 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
8280 Other win32-based unix environments, such as mingw or pw32, may use other
8281 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
8282 was originally intended to help avoid name conflicts among dll's built for the
8283 various win32/un*x environments, so that (for example) two versions of a zlib dll
8284 could coexist on the same machine.
8286 The generic cygwin/mingw path layout uses a @samp{bin} directory for
8287 applications and dll's and a @samp{lib} directory for the import
8288 libraries (using cygwin nomenclature):
8294 libxxx.dll.a (in case of dll's)
8295 libxxx.a (in case of static archive)
8298 Linking directly to a dll without using the import library can be
8301 1. Use the dll directly by adding the @samp{bin} path to the link line
8303 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
8306 However, as the dll's often have version numbers appended to their names
8307 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
8308 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
8309 not versioned, and do not have this difficulty.
8311 2. Create a symbolic link from the dll to a file in the @samp{lib}
8312 directory according to the above mentioned search pattern. This
8313 should be used to avoid unwanted changes in the tools needed for
8317 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
8320 Then you can link without any make environment changes.
8323 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
8326 This technique also avoids the version number problems, because the following is
8333 libxxx.dll.a -> ../bin/cygxxx-5.dll
8336 Linking directly to a dll without using an import lib will work
8337 even when auto-import features are exercised, and even when
8338 @samp{--enable-runtime-pseudo-relocs} is used.
8340 Given the improvements in speed and memory usage, one might justifiably
8341 wonder why import libraries are used at all. There are three reasons:
8343 1. Until recently, the link-directly-to-dll functionality did @emph{not}
8344 work with auto-imported data.
8346 2. Sometimes it is necessary to include pure static objects within the
8347 import library (which otherwise contains only bfd's for indirection
8348 symbols that point to the exports of a dll). Again, the import lib
8349 for the cygwin kernel makes use of this ability, and it is not
8350 possible to do this without an import lib.
8352 3. Symbol aliases can only be resolved using an import lib. This is
8353 critical when linking against OS-supplied dll's (eg, the win32 API)
8354 in which symbols are usually exported as undecorated aliases of their
8355 stdcall-decorated assembly names.
8357 So, import libs are not going away. But the ability to replace
8358 true import libs with a simple symbolic link to (or a copy of)
8359 a dll, in many cases, is a useful addition to the suite of tools
8360 binutils makes available to the win32 developer. Given the
8361 massive improvements in memory requirements during linking, storage
8362 requirements, and linking speed, we expect that many developers
8363 will soon begin to use this feature whenever possible.
8365 @item symbol aliasing
8367 @item adding additional names
8368 Sometimes, it is useful to export symbols with additional names.
8369 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
8370 exported as @samp{_foo} by using special directives in the DEF file
8371 when creating the dll. This will affect also the optional created
8372 import library. Consider the following DEF file:
8375 LIBRARY "xyz.dll" BASE=0x61000000
8382 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
8384 Another method for creating a symbol alias is to create it in the
8385 source code using the "weak" attribute:
8388 void foo () @{ /* Do something. */; @}
8389 void _foo () __attribute__ ((weak, alias ("foo")));
8392 See the gcc manual for more information about attributes and weak
8395 @item renaming symbols
8396 Sometimes it is useful to rename exports. For instance, the cygwin
8397 kernel does this regularly. A symbol @samp{_foo} can be exported as
8398 @samp{foo} but not as @samp{_foo} by using special directives in the
8399 DEF file. (This will also affect the import library, if it is
8400 created). In the following example:
8403 LIBRARY "xyz.dll" BASE=0x61000000
8409 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
8413 Note: using a DEF file disables the default auto-export behavior,
8414 unless the @samp{--export-all-symbols} command-line option is used.
8415 If, however, you are trying to rename symbols, then you should list
8416 @emph{all} desired exports in the DEF file, including the symbols
8417 that are not being renamed, and do @emph{not} use the
8418 @samp{--export-all-symbols} option. If you list only the
8419 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
8420 to handle the other symbols, then the both the new names @emph{and}
8421 the original names for the renamed symbols will be exported.
8422 In effect, you'd be aliasing those symbols, not renaming them,
8423 which is probably not what you wanted.
8425 @cindex weak externals
8426 @item weak externals
8427 The Windows object format, PE, specifies a form of weak symbols called
8428 weak externals. When a weak symbol is linked and the symbol is not
8429 defined, the weak symbol becomes an alias for some other symbol. There
8430 are three variants of weak externals:
8432 @item Definition is searched for in objects and libraries, historically
8433 called lazy externals.
8434 @item Definition is searched for only in other objects, not in libraries.
8435 This form is not presently implemented.
8436 @item No search; the symbol is an alias. This form is not presently
8439 As a GNU extension, weak symbols that do not specify an alternate symbol
8440 are supported. If the symbol is undefined when linking, the symbol
8441 uses a default value.
8443 @cindex aligned common symbols
8444 @item aligned common symbols
8445 As a GNU extension to the PE file format, it is possible to specify the
8446 desired alignment for a common symbol. This information is conveyed from
8447 the assembler or compiler to the linker by means of GNU-specific commands
8448 carried in the object file's @samp{.drectve} section, which are recognized
8449 by @command{ld} and respected when laying out the common symbols. Native
8450 tools will be able to process object files employing this GNU extension,
8451 but will fail to respect the alignment instructions, and may issue noisy
8452 warnings about unknown linker directives.
8467 @section @code{ld} and Xtensa Processors
8469 @cindex Xtensa processors
8470 The default @command{ld} behavior for Xtensa processors is to interpret
8471 @code{SECTIONS} commands so that lists of explicitly named sections in a
8472 specification with a wildcard file will be interleaved when necessary to
8473 keep literal pools within the range of PC-relative load offsets. For
8474 example, with the command:
8486 @command{ld} may interleave some of the @code{.literal}
8487 and @code{.text} sections from different object files to ensure that the
8488 literal pools are within the range of PC-relative load offsets. A valid
8489 interleaving might place the @code{.literal} sections from an initial
8490 group of files followed by the @code{.text} sections of that group of
8491 files. Then, the @code{.literal} sections from the rest of the files
8492 and the @code{.text} sections from the rest of the files would follow.
8494 @cindex @option{--relax} on Xtensa
8495 @cindex relaxing on Xtensa
8496 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8497 provides two important link-time optimizations. The first optimization
8498 is to combine identical literal values to reduce code size. A redundant
8499 literal will be removed and all the @code{L32R} instructions that use it
8500 will be changed to reference an identical literal, as long as the
8501 location of the replacement literal is within the offset range of all
8502 the @code{L32R} instructions. The second optimization is to remove
8503 unnecessary overhead from assembler-generated ``longcall'' sequences of
8504 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8505 range of direct @code{CALL@var{n}} instructions.
8507 For each of these cases where an indirect call sequence can be optimized
8508 to a direct call, the linker will change the @code{CALLX@var{n}}
8509 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8510 instruction, and remove the literal referenced by the @code{L32R}
8511 instruction if it is not used for anything else. Removing the
8512 @code{L32R} instruction always reduces code size but can potentially
8513 hurt performance by changing the alignment of subsequent branch targets.
8514 By default, the linker will always preserve alignments, either by
8515 switching some instructions between 24-bit encodings and the equivalent
8516 density instructions or by inserting a no-op in place of the @code{L32R}
8517 instruction that was removed. If code size is more important than
8518 performance, the @option{--size-opt} option can be used to prevent the
8519 linker from widening density instructions or inserting no-ops, except in
8520 a few cases where no-ops are required for correctness.
8522 The following Xtensa-specific command-line options can be used to
8525 @cindex Xtensa options
8528 When optimizing indirect calls to direct calls, optimize for code size
8529 more than performance. With this option, the linker will not insert
8530 no-ops or widen density instructions to preserve branch target
8531 alignment. There may still be some cases where no-ops are required to
8532 preserve the correctness of the code.
8540 @ifclear SingleFormat
8545 @cindex object file management
8546 @cindex object formats available
8548 The linker accesses object and archive files using the BFD libraries.
8549 These libraries allow the linker to use the same routines to operate on
8550 object files whatever the object file format. A different object file
8551 format can be supported simply by creating a new BFD back end and adding
8552 it to the library. To conserve runtime memory, however, the linker and
8553 associated tools are usually configured to support only a subset of the
8554 object file formats available. You can use @code{objdump -i}
8555 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8556 list all the formats available for your configuration.
8558 @cindex BFD requirements
8559 @cindex requirements for BFD
8560 As with most implementations, BFD is a compromise between
8561 several conflicting requirements. The major factor influencing
8562 BFD design was efficiency: any time used converting between
8563 formats is time which would not have been spent had BFD not
8564 been involved. This is partly offset by abstraction payback; since
8565 BFD simplifies applications and back ends, more time and care
8566 may be spent optimizing algorithms for a greater speed.
8568 One minor artifact of the BFD solution which you should bear in
8569 mind is the potential for information loss. There are two places where
8570 useful information can be lost using the BFD mechanism: during
8571 conversion and during output. @xref{BFD information loss}.
8574 * BFD outline:: How it works: an outline of BFD
8578 @section How It Works: An Outline of BFD
8579 @cindex opening object files
8580 @include bfdsumm.texi
8583 @node Reporting Bugs
8584 @chapter Reporting Bugs
8585 @cindex bugs in @command{ld}
8586 @cindex reporting bugs in @command{ld}
8588 Your bug reports play an essential role in making @command{ld} reliable.
8590 Reporting a bug may help you by bringing a solution to your problem, or
8591 it may not. But in any case the principal function of a bug report is
8592 to help the entire community by making the next version of @command{ld}
8593 work better. Bug reports are your contribution to the maintenance of
8596 In order for a bug report to serve its purpose, you must include the
8597 information that enables us to fix the bug.
8600 * Bug Criteria:: Have you found a bug?
8601 * Bug Reporting:: How to report bugs
8605 @section Have You Found a Bug?
8606 @cindex bug criteria
8608 If you are not sure whether you have found a bug, here are some guidelines:
8611 @cindex fatal signal
8612 @cindex linker crash
8613 @cindex crash of linker
8615 If the linker gets a fatal signal, for any input whatever, that is a
8616 @command{ld} bug. Reliable linkers never crash.
8618 @cindex error on valid input
8620 If @command{ld} produces an error message for valid input, that is a bug.
8622 @cindex invalid input
8624 If @command{ld} does not produce an error message for invalid input, that
8625 may be a bug. In the general case, the linker can not verify that
8626 object files are correct.
8629 If you are an experienced user of linkers, your suggestions for
8630 improvement of @command{ld} are welcome in any case.
8634 @section How to Report Bugs
8636 @cindex @command{ld} bugs, reporting
8638 A number of companies and individuals offer support for @sc{gnu}
8639 products. If you obtained @command{ld} from a support organization, we
8640 recommend you contact that organization first.
8642 You can find contact information for many support companies and
8643 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
8647 Otherwise, send bug reports for @command{ld} to
8651 The fundamental principle of reporting bugs usefully is this:
8652 @strong{report all the facts}. If you are not sure whether to state a
8653 fact or leave it out, state it!
8655 Often people omit facts because they think they know what causes the
8656 problem and assume that some details do not matter. Thus, you might
8657 assume that the name of a symbol you use in an example does not
8658 matter. Well, probably it does not, but one cannot be sure. Perhaps
8659 the bug is a stray memory reference which happens to fetch from the
8660 location where that name is stored in memory; perhaps, if the name
8661 were different, the contents of that location would fool the linker
8662 into doing the right thing despite the bug. Play it safe and give a
8663 specific, complete example. That is the easiest thing for you to do,
8664 and the most helpful.
8666 Keep in mind that the purpose of a bug report is to enable us to fix
8667 the bug if it is new to us. Therefore, always write your bug reports
8668 on the assumption that the bug has not been reported previously.
8670 Sometimes people give a few sketchy facts and ask, ``Does this ring a
8671 bell?'' This cannot help us fix a bug, so it is basically useless. We
8672 respond by asking for enough details to enable us to investigate.
8673 You might as well expedite matters by sending them to begin with.
8675 To enable us to fix the bug, you should include all these things:
8679 The version of @command{ld}. @command{ld} announces it if you start it with
8680 the @samp{--version} argument.
8682 Without this, we will not know whether there is any point in looking for
8683 the bug in the current version of @command{ld}.
8686 Any patches you may have applied to the @command{ld} source, including any
8687 patches made to the @code{BFD} library.
8690 The type of machine you are using, and the operating system name and
8694 What compiler (and its version) was used to compile @command{ld}---e.g.
8698 The command arguments you gave the linker to link your example and
8699 observe the bug. To guarantee you will not omit something important,
8700 list them all. A copy of the Makefile (or the output from make) is
8703 If we were to try to guess the arguments, we would probably guess wrong
8704 and then we might not encounter the bug.
8707 A complete input file, or set of input files, that will reproduce the
8708 bug. It is generally most helpful to send the actual object files
8709 provided that they are reasonably small. Say no more than 10K. For
8710 bigger files you can either make them available by FTP or HTTP or else
8711 state that you are willing to send the object file(s) to whomever
8712 requests them. (Note - your email will be going to a mailing list, so
8713 we do not want to clog it up with large attachments). But small
8714 attachments are best.
8716 If the source files were assembled using @code{gas} or compiled using
8717 @code{gcc}, then it may be OK to send the source files rather than the
8718 object files. In this case, be sure to say exactly what version of
8719 @code{gas} or @code{gcc} was used to produce the object files. Also say
8720 how @code{gas} or @code{gcc} were configured.
8723 A description of what behavior you observe that you believe is
8724 incorrect. For example, ``It gets a fatal signal.''
8726 Of course, if the bug is that @command{ld} gets a fatal signal, then we
8727 will certainly notice it. But if the bug is incorrect output, we might
8728 not notice unless it is glaringly wrong. You might as well not give us
8729 a chance to make a mistake.
8731 Even if the problem you experience is a fatal signal, you should still
8732 say so explicitly. Suppose something strange is going on, such as, your
8733 copy of @command{ld} is out of sync, or you have encountered a bug in the
8734 C library on your system. (This has happened!) Your copy might crash
8735 and ours would not. If you told us to expect a crash, then when ours
8736 fails to crash, we would know that the bug was not happening for us. If
8737 you had not told us to expect a crash, then we would not be able to draw
8738 any conclusion from our observations.
8741 If you wish to suggest changes to the @command{ld} source, send us context
8742 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
8743 @samp{-p} option. Always send diffs from the old file to the new file.
8744 If you even discuss something in the @command{ld} source, refer to it by
8745 context, not by line number.
8747 The line numbers in our development sources will not match those in your
8748 sources. Your line numbers would convey no useful information to us.
8751 Here are some things that are not necessary:
8755 A description of the envelope of the bug.
8757 Often people who encounter a bug spend a lot of time investigating
8758 which changes to the input file will make the bug go away and which
8759 changes will not affect it.
8761 This is often time consuming and not very useful, because the way we
8762 will find the bug is by running a single example under the debugger
8763 with breakpoints, not by pure deduction from a series of examples.
8764 We recommend that you save your time for something else.
8766 Of course, if you can find a simpler example to report @emph{instead}
8767 of the original one, that is a convenience for us. Errors in the
8768 output will be easier to spot, running under the debugger will take
8769 less time, and so on.
8771 However, simplification is not vital; if you do not want to do this,
8772 report the bug anyway and send us the entire test case you used.
8775 A patch for the bug.
8777 A patch for the bug does help us if it is a good one. But do not omit
8778 the necessary information, such as the test case, on the assumption that
8779 a patch is all we need. We might see problems with your patch and decide
8780 to fix the problem another way, or we might not understand it at all.
8782 Sometimes with a program as complicated as @command{ld} it is very hard to
8783 construct an example that will make the program follow a certain path
8784 through the code. If you do not send us the example, we will not be
8785 able to construct one, so we will not be able to verify that the bug is
8788 And if we cannot understand what bug you are trying to fix, or why your
8789 patch should be an improvement, we will not install it. A test case will
8790 help us to understand.
8793 A guess about what the bug is or what it depends on.
8795 Such guesses are usually wrong. Even we cannot guess right about such
8796 things without first using the debugger to find the facts.
8800 @appendix MRI Compatible Script Files
8801 @cindex MRI compatibility
8802 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
8803 linker, @command{ld} can use MRI compatible linker scripts as an
8804 alternative to the more general-purpose linker scripting language
8805 described in @ref{Scripts}. MRI compatible linker scripts have a much
8806 simpler command set than the scripting language otherwise used with
8807 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
8808 linker commands; these commands are described here.
8810 In general, MRI scripts aren't of much use with the @code{a.out} object
8811 file format, since it only has three sections and MRI scripts lack some
8812 features to make use of them.
8814 You can specify a file containing an MRI-compatible script using the
8815 @samp{-c} command-line option.
8817 Each command in an MRI-compatible script occupies its own line; each
8818 command line starts with the keyword that identifies the command (though
8819 blank lines are also allowed for punctuation). If a line of an
8820 MRI-compatible script begins with an unrecognized keyword, @command{ld}
8821 issues a warning message, but continues processing the script.
8823 Lines beginning with @samp{*} are comments.
8825 You can write these commands using all upper-case letters, or all
8826 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
8827 The following list shows only the upper-case form of each command.
8830 @cindex @code{ABSOLUTE} (MRI)
8831 @item ABSOLUTE @var{secname}
8832 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
8833 Normally, @command{ld} includes in the output file all sections from all
8834 the input files. However, in an MRI-compatible script, you can use the
8835 @code{ABSOLUTE} command to restrict the sections that will be present in
8836 your output program. If the @code{ABSOLUTE} command is used at all in a
8837 script, then only the sections named explicitly in @code{ABSOLUTE}
8838 commands will appear in the linker output. You can still use other
8839 input sections (whatever you select on the command line, or using
8840 @code{LOAD}) to resolve addresses in the output file.
8842 @cindex @code{ALIAS} (MRI)
8843 @item ALIAS @var{out-secname}, @var{in-secname}
8844 Use this command to place the data from input section @var{in-secname}
8845 in a section called @var{out-secname} in the linker output file.
8847 @var{in-secname} may be an integer.
8849 @cindex @code{ALIGN} (MRI)
8850 @item ALIGN @var{secname} = @var{expression}
8851 Align the section called @var{secname} to @var{expression}. The
8852 @var{expression} should be a power of two.
8854 @cindex @code{BASE} (MRI)
8855 @item BASE @var{expression}
8856 Use the value of @var{expression} as the lowest address (other than
8857 absolute addresses) in the output file.
8859 @cindex @code{CHIP} (MRI)
8860 @item CHIP @var{expression}
8861 @itemx CHIP @var{expression}, @var{expression}
8862 This command does nothing; it is accepted only for compatibility.
8864 @cindex @code{END} (MRI)
8866 This command does nothing whatever; it's only accepted for compatibility.
8868 @cindex @code{FORMAT} (MRI)
8869 @item FORMAT @var{output-format}
8870 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
8871 language, but restricted to S-records, if @var{output-format} is @samp{S}
8873 @cindex @code{LIST} (MRI)
8874 @item LIST @var{anything}@dots{}
8875 Print (to the standard output file) a link map, as produced by the
8876 @command{ld} command-line option @samp{-M}.
8878 The keyword @code{LIST} may be followed by anything on the
8879 same line, with no change in its effect.
8881 @cindex @code{LOAD} (MRI)
8882 @item LOAD @var{filename}
8883 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
8884 Include one or more object file @var{filename} in the link; this has the
8885 same effect as specifying @var{filename} directly on the @command{ld}
8888 @cindex @code{NAME} (MRI)
8889 @item NAME @var{output-name}
8890 @var{output-name} is the name for the program produced by @command{ld}; the
8891 MRI-compatible command @code{NAME} is equivalent to the command-line
8892 option @samp{-o} or the general script language command @code{OUTPUT}.
8894 @cindex @code{ORDER} (MRI)
8895 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
8896 @itemx ORDER @var{secname} @var{secname} @var{secname}
8897 Normally, @command{ld} orders the sections in its output file in the
8898 order in which they first appear in the input files. In an MRI-compatible
8899 script, you can override this ordering with the @code{ORDER} command. The
8900 sections you list with @code{ORDER} will appear first in your output
8901 file, in the order specified.
8903 @cindex @code{PUBLIC} (MRI)
8904 @item PUBLIC @var{name}=@var{expression}
8905 @itemx PUBLIC @var{name},@var{expression}
8906 @itemx PUBLIC @var{name} @var{expression}
8907 Supply a value (@var{expression}) for external symbol
8908 @var{name} used in the linker input files.
8910 @cindex @code{SECT} (MRI)
8911 @item SECT @var{secname}, @var{expression}
8912 @itemx SECT @var{secname}=@var{expression}
8913 @itemx SECT @var{secname} @var{expression}
8914 You can use any of these three forms of the @code{SECT} command to
8915 specify the start address (@var{expression}) for section @var{secname}.
8916 If you have more than one @code{SECT} statement for the same
8917 @var{secname}, only the @emph{first} sets the start address.
8920 @node GNU Free Documentation License
8921 @appendix GNU Free Documentation License
8925 @unnumbered LD Index
8930 % I think something like @@colophon should be in texinfo. In the
8932 \long\def\colophon{\hbox to0pt{}\vfill
8933 \centerline{The body of this manual is set in}
8934 \centerline{\fontname\tenrm,}
8935 \centerline{with headings in {\bf\fontname\tenbf}}
8936 \centerline{and examples in {\tt\fontname\tentt}.}
8937 \centerline{{\it\fontname\tenit\/} and}
8938 \centerline{{\sl\fontname\tensl\/}}
8939 \centerline{are used for emphasis.}\vfill}
8941 % Blame: doc@@cygnus.com, 28mar91.