3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011,
6 @c Free Software Foundation, Inc.
9 @include configdoc.texi
10 @c (configdoc.texi is generated by the Makefile)
16 @macro gcctabopt{body}
22 @c Configure for the generation of man pages
45 @dircategory Software development
47 * Ld: (ld). The GNU linker.
52 This file documents the @sc{gnu} linker LD
53 @ifset VERSION_PACKAGE
54 @value{VERSION_PACKAGE}
56 version @value{VERSION}.
58 Copyright @copyright{} 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
59 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2012, 2013
60 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, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
98 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009,
99 2012, 2013 Free Software Foundation, Inc.
101 Permission is granted to copy, distribute and/or modify this document
102 under the terms of the GNU Free Documentation License, Version 1.3
103 or any later version published by the Free Software Foundation;
104 with no Invariant Sections, with no Front-Cover Texts, and with no
105 Back-Cover Texts. A copy of the license is included in the
106 section entitled ``GNU Free Documentation License''.
112 @c FIXME: Talk about importance of *order* of args, cmds to linker!
117 This file documents the @sc{gnu} linker ld
118 @ifset VERSION_PACKAGE
119 @value{VERSION_PACKAGE}
121 version @value{VERSION}.
123 This document is distributed under the terms of the GNU Free
124 Documentation License version 1.3. A copy of the license is included
125 in the section entitled ``GNU Free Documentation License''.
128 * Overview:: Overview
129 * Invocation:: Invocation
130 * Scripts:: Linker Scripts
132 * Machine Dependent:: Machine Dependent Features
136 * H8/300:: ld and the H8/300
139 * Renesas:: ld and other Renesas micros
142 * i960:: ld and the Intel 960 family
145 * ARM:: ld and the ARM family
148 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
151 * HPPA ELF32:: ld and HPPA 32-bit ELF
154 * M68K:: ld and Motorola 68K family
157 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
160 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
163 * SPU ELF:: ld and SPU ELF Support
166 * TI COFF:: ld and the TI COFF
169 * Win32:: ld and WIN32 (cygwin/mingw)
172 * Xtensa:: ld and Xtensa Processors
175 @ifclear SingleFormat
178 @c Following blank line required for remaining bug in makeinfo conds/menus
180 * Reporting Bugs:: Reporting Bugs
181 * MRI:: MRI Compatible Script Files
182 * GNU Free Documentation License:: GNU Free Documentation License
183 * LD Index:: LD Index
190 @cindex @sc{gnu} linker
191 @cindex what is this?
194 @c man begin SYNOPSIS
195 ld [@b{options}] @var{objfile} @dots{}
199 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
200 the Info entries for @file{binutils} and
205 @c man begin DESCRIPTION
207 @command{ld} combines a number of object and archive files, relocates
208 their data and ties up symbol references. Usually the last step in
209 compiling a program is to run @command{ld}.
211 @command{ld} accepts Linker Command Language files written in
212 a superset of AT&T's Link Editor Command Language syntax,
213 to provide explicit and total control over the linking process.
217 This man page does not describe the command language; see the
218 @command{ld} entry in @code{info} for full details on the command
219 language and on other aspects of the GNU linker.
222 @ifclear SingleFormat
223 This version of @command{ld} uses the general purpose BFD libraries
224 to operate on object files. This allows @command{ld} to read, combine, and
225 write object files in many different formats---for example, COFF or
226 @code{a.out}. Different formats may be linked together to produce any
227 available kind of object file. @xref{BFD}, for more information.
230 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
231 linkers in providing diagnostic information. Many linkers abandon
232 execution immediately upon encountering an error; whenever possible,
233 @command{ld} continues executing, allowing you to identify other errors
234 (or, in some cases, to get an output file in spite of the error).
241 @c man begin DESCRIPTION
243 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
244 and to be as compatible as possible with other linkers. As a result,
245 you have many choices to control its behavior.
251 * Options:: Command Line Options
252 * Environment:: Environment Variables
256 @section Command Line Options
264 The linker supports a plethora of command-line options, but in actual
265 practice few of them are used in any particular context.
266 @cindex standard Unix system
267 For instance, a frequent use of @command{ld} is to link standard Unix
268 object files on a standard, supported Unix system. On such a system, to
269 link a file @code{hello.o}:
272 ld -o @var{output} /lib/crt0.o hello.o -lc
275 This tells @command{ld} to produce a file called @var{output} as the
276 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
277 the library @code{libc.a}, which will come from the standard search
278 directories. (See the discussion of the @samp{-l} option below.)
280 Some of the command-line options to @command{ld} may be specified at any
281 point in the command line. However, options which refer to files, such
282 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
283 which the option appears in the command line, relative to the object
284 files and other file options. Repeating non-file options with a
285 different argument will either have no further effect, or override prior
286 occurrences (those further to the left on the command line) of that
287 option. Options which may be meaningfully specified more than once are
288 noted in the descriptions below.
291 Non-option arguments are object files or archives which are to be linked
292 together. They may follow, precede, or be mixed in with command-line
293 options, except that an object file argument may not be placed between
294 an option and its argument.
296 Usually the linker is invoked with at least one object file, but you can
297 specify other forms of binary input files using @samp{-l}, @samp{-R},
298 and the script command language. If @emph{no} binary input files at all
299 are specified, the linker does not produce any output, and issues the
300 message @samp{No input files}.
302 If the linker cannot recognize the format of an object file, it will
303 assume that it is a linker script. A script specified in this way
304 augments the main linker script used for the link (either the default
305 linker script or the one specified by using @samp{-T}). This feature
306 permits the linker to link against a file which appears to be an object
307 or an archive, but actually merely defines some symbol values, or uses
308 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
309 script in this way merely augments the main linker script, with the
310 extra commands placed after the main script; use the @samp{-T} option
311 to replace the default linker script entirely, but note the effect of
312 the @code{INSERT} command. @xref{Scripts}.
314 For options whose names are a single letter,
315 option arguments must either follow the option letter without intervening
316 whitespace, or be given as separate arguments immediately following the
317 option that requires them.
319 For options whose names are multiple letters, either one dash or two can
320 precede the option name; for example, @samp{-trace-symbol} and
321 @samp{--trace-symbol} are equivalent. Note---there is one exception to
322 this rule. Multiple letter options that start with a lower case 'o' can
323 only be preceded by two dashes. This is to reduce confusion with the
324 @samp{-o} option. So for example @samp{-omagic} sets the output file
325 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
328 Arguments to multiple-letter options must either be separated from the
329 option name by an equals sign, or be given as separate arguments
330 immediately following the option that requires them. For example,
331 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
332 Unique abbreviations of the names of multiple-letter options are
335 Note---if the linker is being invoked indirectly, via a compiler driver
336 (e.g. @samp{gcc}) then all the linker command line options should be
337 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
338 compiler driver) like this:
341 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
344 This is important, because otherwise the compiler driver program may
345 silently drop the linker options, resulting in a bad link. Confusion
346 may also arise when passing options that require values through a
347 driver, as the use of a space between option and argument acts as
348 a separator, and causes the driver to pass only the option to the linker
349 and the argument to the compiler. In this case, it is simplest to use
350 the joined forms of both single- and multiple-letter options, such as:
353 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
356 Here is a table of the generic command line switches accepted by the GNU
360 @include at-file.texi
362 @kindex -a @var{keyword}
363 @item -a @var{keyword}
364 This option is supported for HP/UX compatibility. The @var{keyword}
365 argument must be one of the strings @samp{archive}, @samp{shared}, or
366 @samp{default}. @samp{-aarchive} is functionally equivalent to
367 @samp{-Bstatic}, and the other two keywords are functionally equivalent
368 to @samp{-Bdynamic}. This option may be used any number of times.
370 @kindex --audit @var{AUDITLIB}
371 @item --audit @var{AUDITLIB}
372 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
373 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
374 specified in the library. If specified multiple times @code{DT_AUDIT}
375 will contain a colon separated list of audit interfaces to use. If the linker
376 finds an object with an audit entry while searching for shared libraries,
377 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
378 This option is only meaningful on ELF platforms supporting the rtld-audit
382 @cindex architectures
383 @kindex -A @var{arch}
384 @item -A @var{architecture}
385 @kindex --architecture=@var{arch}
386 @itemx --architecture=@var{architecture}
387 In the current release of @command{ld}, this option is useful only for the
388 Intel 960 family of architectures. In that @command{ld} configuration, the
389 @var{architecture} argument identifies the particular architecture in
390 the 960 family, enabling some safeguards and modifying the
391 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
392 family}, for details.
394 Future releases of @command{ld} may support similar functionality for
395 other architecture families.
398 @ifclear SingleFormat
399 @cindex binary input format
400 @kindex -b @var{format}
401 @kindex --format=@var{format}
404 @item -b @var{input-format}
405 @itemx --format=@var{input-format}
406 @command{ld} may be configured to support more than one kind of object
407 file. If your @command{ld} is configured this way, you can use the
408 @samp{-b} option to specify the binary format for input object files
409 that follow this option on the command line. Even when @command{ld} is
410 configured to support alternative object formats, you don't usually need
411 to specify this, as @command{ld} should be configured to expect as a
412 default input format the most usual format on each machine.
413 @var{input-format} is a text string, the name of a particular format
414 supported by the BFD libraries. (You can list the available binary
415 formats with @samp{objdump -i}.)
418 You may want to use this option if you are linking files with an unusual
419 binary format. You can also use @samp{-b} to switch formats explicitly (when
420 linking object files of different formats), by including
421 @samp{-b @var{input-format}} before each group of object files in a
424 The default format is taken from the environment variable
429 You can also define the input format from a script, using the command
432 see @ref{Format Commands}.
436 @kindex -c @var{MRI-cmdfile}
437 @kindex --mri-script=@var{MRI-cmdfile}
438 @cindex compatibility, MRI
439 @item -c @var{MRI-commandfile}
440 @itemx --mri-script=@var{MRI-commandfile}
441 For compatibility with linkers produced by MRI, @command{ld} accepts script
442 files written in an alternate, restricted command language, described in
444 @ref{MRI,,MRI Compatible Script Files}.
447 the MRI Compatible Script Files section of GNU ld documentation.
449 Introduce MRI script files with
450 the option @samp{-c}; use the @samp{-T} option to run linker
451 scripts written in the general-purpose @command{ld} scripting language.
452 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
453 specified by any @samp{-L} options.
455 @cindex common allocation
462 These three options are equivalent; multiple forms are supported for
463 compatibility with other linkers. They assign space to common symbols
464 even if a relocatable output file is specified (with @samp{-r}). The
465 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
466 @xref{Miscellaneous Commands}.
468 @kindex --depaudit @var{AUDITLIB}
469 @kindex -P @var{AUDITLIB}
470 @item --depaudit @var{AUDITLIB}
471 @itemx -P @var{AUDITLIB}
472 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
473 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
474 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
475 will contain a colon separated list of audit interfaces to use. This
476 option is only meaningful on ELF platforms supporting the rtld-audit interface.
477 The -P option is provided for Solaris compatibility.
479 @cindex entry point, from command line
480 @kindex -e @var{entry}
481 @kindex --entry=@var{entry}
483 @itemx --entry=@var{entry}
484 Use @var{entry} as the explicit symbol for beginning execution of your
485 program, rather than the default entry point. If there is no symbol
486 named @var{entry}, the linker will try to parse @var{entry} as a number,
487 and use that as the entry address (the number will be interpreted in
488 base 10; you may use a leading @samp{0x} for base 16, or a leading
489 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
490 and other ways of specifying the entry point.
492 @kindex --exclude-libs
493 @item --exclude-libs @var{lib},@var{lib},...
494 Specifies a list of archive libraries from which symbols should not be automatically
495 exported. The library names may be delimited by commas or colons. Specifying
496 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
497 automatic export. This option is available only for the i386 PE targeted
498 port of the linker and for ELF targeted ports. For i386 PE, symbols
499 explicitly listed in a .def file are still exported, regardless of this
500 option. For ELF targeted ports, symbols affected by this option will
501 be treated as hidden.
503 @kindex --exclude-modules-for-implib
504 @item --exclude-modules-for-implib @var{module},@var{module},...
505 Specifies a list of object files or archive members, from which symbols
506 should not be automatically exported, but which should be copied wholesale
507 into the import library being generated during the link. The module names
508 may be delimited by commas or colons, and must match exactly the filenames
509 used by @command{ld} to open the files; for archive members, this is simply
510 the member name, but for object files the name listed must include and
511 match precisely any path used to specify the input file on the linker's
512 command-line. This option is available only for the i386 PE targeted port
513 of the linker. Symbols explicitly listed in a .def file are still exported,
514 regardless of this option.
516 @cindex dynamic symbol table
518 @kindex --export-dynamic
519 @kindex --no-export-dynamic
521 @itemx --export-dynamic
522 @itemx --no-export-dynamic
523 When creating a dynamically linked executable, using the @option{-E}
524 option or the @option{--export-dynamic} option causes the linker to add
525 all symbols to the dynamic symbol table. The dynamic symbol table is the
526 set of symbols which are visible from dynamic objects at run time.
528 If you do not use either of these options (or use the
529 @option{--no-export-dynamic} option to restore the default behavior), the
530 dynamic symbol table will normally contain only those symbols which are
531 referenced by some dynamic object mentioned in the link.
533 If you use @code{dlopen} to load a dynamic object which needs to refer
534 back to the symbols defined by the program, rather than some other
535 dynamic object, then you will probably need to use this option when
536 linking the program itself.
538 You can also use the dynamic list to control what symbols should
539 be added to the dynamic symbol table if the output format supports it.
540 See the description of @samp{--dynamic-list}.
542 Note that this option is specific to ELF targeted ports. PE targets
543 support a similar function to export all symbols from a DLL or EXE; see
544 the description of @samp{--export-all-symbols} below.
546 @ifclear SingleFormat
547 @cindex big-endian objects
551 Link big-endian objects. This affects the default output format.
553 @cindex little-endian objects
556 Link little-endian objects. This affects the default output format.
559 @kindex -f @var{name}
560 @kindex --auxiliary=@var{name}
562 @itemx --auxiliary=@var{name}
563 When creating an ELF shared object, set the internal DT_AUXILIARY field
564 to the specified name. This tells the dynamic linker that the symbol
565 table of the shared object should be used as an auxiliary filter on the
566 symbol table of the shared object @var{name}.
568 If you later link a program against this filter object, then, when you
569 run the program, the dynamic linker will see the DT_AUXILIARY field. If
570 the dynamic linker resolves any symbols from the filter object, it will
571 first check whether there is a definition in the shared object
572 @var{name}. If there is one, it will be used instead of the definition
573 in the filter object. The shared object @var{name} need not exist.
574 Thus the shared object @var{name} may be used to provide an alternative
575 implementation of certain functions, perhaps for debugging or for
576 machine specific performance.
578 This option may be specified more than once. The DT_AUXILIARY entries
579 will be created in the order in which they appear on the command line.
581 @kindex -F @var{name}
582 @kindex --filter=@var{name}
584 @itemx --filter=@var{name}
585 When creating an ELF shared object, set the internal DT_FILTER field to
586 the specified name. This tells the dynamic linker that the symbol table
587 of the shared object which is being created should be used as a filter
588 on the symbol table of the shared object @var{name}.
590 If you later link a program against this filter object, then, when you
591 run the program, the dynamic linker will see the DT_FILTER field. The
592 dynamic linker will resolve symbols according to the symbol table of the
593 filter object as usual, but it will actually link to the definitions
594 found in the shared object @var{name}. Thus the filter object can be
595 used to select a subset of the symbols provided by the object
598 Some older linkers used the @option{-F} option throughout a compilation
599 toolchain for specifying object-file format for both input and output
601 @ifclear SingleFormat
602 The @sc{gnu} linker uses other mechanisms for this purpose: the
603 @option{-b}, @option{--format}, @option{--oformat} options, the
604 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
605 environment variable.
607 The @sc{gnu} linker will ignore the @option{-F} option when not
608 creating an ELF shared object.
610 @cindex finalization function
611 @kindex -fini=@var{name}
612 @item -fini=@var{name}
613 When creating an ELF executable or shared object, call NAME when the
614 executable or shared object is unloaded, by setting DT_FINI to the
615 address of the function. By default, the linker uses @code{_fini} as
616 the function to call.
620 Ignored. Provided for compatibility with other tools.
622 @kindex -G @var{value}
623 @kindex --gpsize=@var{value}
626 @itemx --gpsize=@var{value}
627 Set the maximum size of objects to be optimized using the GP register to
628 @var{size}. This is only meaningful for object file formats such as
629 MIPS ECOFF which supports putting large and small objects into different
630 sections. This is ignored for other object file formats.
632 @cindex runtime library name
633 @kindex -h @var{name}
634 @kindex -soname=@var{name}
636 @itemx -soname=@var{name}
637 When creating an ELF shared object, set the internal DT_SONAME field to
638 the specified name. When an executable is linked with a shared object
639 which has a DT_SONAME field, then when the executable is run the dynamic
640 linker will attempt to load the shared object specified by the DT_SONAME
641 field rather than the using the file name given to the linker.
644 @cindex incremental link
646 Perform an incremental link (same as option @samp{-r}).
648 @cindex initialization function
649 @kindex -init=@var{name}
650 @item -init=@var{name}
651 When creating an ELF executable or shared object, call NAME when the
652 executable or shared object is loaded, by setting DT_INIT to the address
653 of the function. By default, the linker uses @code{_init} as the
656 @cindex archive files, from cmd line
657 @kindex -l @var{namespec}
658 @kindex --library=@var{namespec}
659 @item -l @var{namespec}
660 @itemx --library=@var{namespec}
661 Add the archive or object file specified by @var{namespec} to the
662 list of files to link. This option may be used any number of times.
663 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
664 will search the library path for a file called @var{filename}, otherwise it
665 will search the library path for a file called @file{lib@var{namespec}.a}.
667 On systems which support shared libraries, @command{ld} may also search for
668 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
669 and SunOS systems, @command{ld} will search a directory for a library
670 called @file{lib@var{namespec}.so} before searching for one called
671 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
672 indicates a shared library.) Note that this behavior does not apply
673 to @file{:@var{filename}}, which always specifies a file called
676 The linker will search an archive only once, at the location where it is
677 specified on the command line. If the archive defines a symbol which
678 was undefined in some object which appeared before the archive on the
679 command line, the linker will include the appropriate file(s) from the
680 archive. However, an undefined symbol in an object appearing later on
681 the command line will not cause the linker to search the archive again.
683 See the @option{-(} option for a way to force the linker to search
684 archives multiple times.
686 You may list the same archive multiple times on the command line.
689 This type of archive searching is standard for Unix linkers. However,
690 if you are using @command{ld} on AIX, note that it is different from the
691 behaviour of the AIX linker.
694 @cindex search directory, from cmd line
696 @kindex --library-path=@var{dir}
697 @item -L @var{searchdir}
698 @itemx --library-path=@var{searchdir}
699 Add path @var{searchdir} to the list of paths that @command{ld} will search
700 for archive libraries and @command{ld} control scripts. You may use this
701 option any number of times. The directories are searched in the order
702 in which they are specified on the command line. Directories specified
703 on the command line are searched before the default directories. All
704 @option{-L} options apply to all @option{-l} options, regardless of the
705 order in which the options appear. @option{-L} options do not affect
706 how @command{ld} searches for a linker script unless @option{-T}
709 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
710 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
713 The default set of paths searched (without being specified with
714 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
715 some cases also on how it was configured. @xref{Environment}.
718 The paths can also be specified in a link script with the
719 @code{SEARCH_DIR} command. Directories specified this way are searched
720 at the point in which the linker script appears in the command line.
723 @kindex -m @var{emulation}
724 @item -m @var{emulation}
725 Emulate the @var{emulation} linker. You can list the available
726 emulations with the @samp{--verbose} or @samp{-V} options.
728 If the @samp{-m} option is not used, the emulation is taken from the
729 @code{LDEMULATION} environment variable, if that is defined.
731 Otherwise, the default emulation depends upon how the linker was
739 Print a link map to the standard output. A link map provides
740 information about the link, including the following:
744 Where object files are mapped into memory.
746 How common symbols are allocated.
748 All archive members included in the link, with a mention of the symbol
749 which caused the archive member to be brought in.
751 The values assigned to symbols.
753 Note - symbols whose values are computed by an expression which
754 involves a reference to a previous value of the same symbol may not
755 have correct result displayed in the link map. This is because the
756 linker discards intermediate results and only retains the final value
757 of an expression. Under such circumstances the linker will display
758 the final value enclosed by square brackets. Thus for example a
759 linker script containing:
767 will produce the following output in the link map if the @option{-M}
772 [0x0000000c] foo = (foo * 0x4)
773 [0x0000000c] foo = (foo + 0x8)
776 See @ref{Expressions} for more information about expressions in linker
781 @cindex read-only text
786 Turn off page alignment of sections, and disable linking against shared
787 libraries. If the output format supports Unix style magic numbers,
788 mark the output as @code{NMAGIC}.
792 @cindex read/write from cmd line
796 Set the text and data sections to be readable and writable. Also, do
797 not page-align the data segment, and disable linking against shared
798 libraries. If the output format supports Unix style magic numbers,
799 mark the output as @code{OMAGIC}. Note: Although a writable text section
800 is allowed for PE-COFF targets, it does not conform to the format
801 specification published by Microsoft.
806 This option negates most of the effects of the @option{-N} option. It
807 sets the text section to be read-only, and forces the data segment to
808 be page-aligned. Note - this option does not enable linking against
809 shared libraries. Use @option{-Bdynamic} for this.
811 @kindex -o @var{output}
812 @kindex --output=@var{output}
813 @cindex naming the output file
814 @item -o @var{output}
815 @itemx --output=@var{output}
816 Use @var{output} as the name for the program produced by @command{ld}; if this
817 option is not specified, the name @file{a.out} is used by default. The
818 script command @code{OUTPUT} can also specify the output file name.
820 @kindex -O @var{level}
821 @cindex generating optimized output
823 If @var{level} is a numeric values greater than zero @command{ld} optimizes
824 the output. This might take significantly longer and therefore probably
825 should only be enabled for the final binary. At the moment this
826 option only affects ELF shared library generation. Future releases of
827 the linker may make more use of this option. Also currently there is
828 no difference in the linker's behaviour for different non-zero values
829 of this option. Again this may change with future releases.
832 @kindex --emit-relocs
833 @cindex retain relocations in final executable
836 Leave relocation sections and contents in fully linked executables.
837 Post link analysis and optimization tools may need this information in
838 order to perform correct modifications of executables. This results
839 in larger executables.
841 This option is currently only supported on ELF platforms.
843 @kindex --force-dynamic
844 @cindex forcing the creation of dynamic sections
845 @item --force-dynamic
846 Force the output file to have dynamic sections. This option is specific
850 @cindex relocatable output
852 @kindex --relocatable
855 Generate relocatable output---i.e., generate an output file that can in
856 turn serve as input to @command{ld}. This is often called @dfn{partial
857 linking}. As a side effect, in environments that support standard Unix
858 magic numbers, this option also sets the output file's magic number to
860 @c ; see @option{-N}.
861 If this option is not specified, an absolute file is produced. When
862 linking C++ programs, this option @emph{will not} resolve references to
863 constructors; to do that, use @samp{-Ur}.
865 When an input file does not have the same format as the output file,
866 partial linking is only supported if that input file does not contain any
867 relocations. Different output formats can have further restrictions; for
868 example some @code{a.out}-based formats do not support partial linking
869 with input files in other formats at all.
871 This option does the same thing as @samp{-i}.
873 @kindex -R @var{file}
874 @kindex --just-symbols=@var{file}
875 @cindex symbol-only input
876 @item -R @var{filename}
877 @itemx --just-symbols=@var{filename}
878 Read symbol names and their addresses from @var{filename}, but do not
879 relocate it or include it in the output. This allows your output file
880 to refer symbolically to absolute locations of memory defined in other
881 programs. You may use this option more than once.
883 For compatibility with other ELF linkers, if the @option{-R} option is
884 followed by a directory name, rather than a file name, it is treated as
885 the @option{-rpath} option.
889 @cindex strip all symbols
892 Omit all symbol information from the output file.
895 @kindex --strip-debug
896 @cindex strip debugger symbols
899 Omit debugger symbol information (but not all symbols) from the output file.
903 @cindex input files, displaying
906 Print the names of the input files as @command{ld} processes them.
908 @kindex -T @var{script}
909 @kindex --script=@var{script}
911 @item -T @var{scriptfile}
912 @itemx --script=@var{scriptfile}
913 Use @var{scriptfile} as the linker script. This script replaces
914 @command{ld}'s default linker script (rather than adding to it), so
915 @var{commandfile} must specify everything necessary to describe the
916 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
917 the current directory, @code{ld} looks for it in the directories
918 specified by any preceding @samp{-L} options. Multiple @samp{-T}
921 @kindex -dT @var{script}
922 @kindex --default-script=@var{script}
924 @item -dT @var{scriptfile}
925 @itemx --default-script=@var{scriptfile}
926 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
928 This option is similar to the @option{--script} option except that
929 processing of the script is delayed until after the rest of the
930 command line has been processed. This allows options placed after the
931 @option{--default-script} option on the command line to affect the
932 behaviour of the linker script, which can be important when the linker
933 command line cannot be directly controlled by the user. (eg because
934 the command line is being constructed by another tool, such as
937 @kindex -u @var{symbol}
938 @kindex --undefined=@var{symbol}
939 @cindex undefined symbol
940 @item -u @var{symbol}
941 @itemx --undefined=@var{symbol}
942 Force @var{symbol} to be entered in the output file as an undefined
943 symbol. Doing this may, for example, trigger linking of additional
944 modules from standard libraries. @samp{-u} may be repeated with
945 different option arguments to enter additional undefined symbols. This
946 option is equivalent to the @code{EXTERN} linker script command.
951 For anything other than C++ programs, this option is equivalent to
952 @samp{-r}: it generates relocatable output---i.e., an output file that can in
953 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
954 @emph{does} resolve references to constructors, unlike @samp{-r}.
955 It does not work to use @samp{-Ur} on files that were themselves linked
956 with @samp{-Ur}; once the constructor table has been built, it cannot
957 be added to. Use @samp{-Ur} only for the last partial link, and
958 @samp{-r} for the others.
960 @kindex --unique[=@var{SECTION}]
961 @item --unique[=@var{SECTION}]
962 Creates a separate output section for every input section matching
963 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
964 missing, for every orphan input section. An orphan section is one not
965 specifically mentioned in a linker script. You may use this option
966 multiple times on the command line; It prevents the normal merging of
967 input sections with the same name, overriding output section assignments
977 Display the version number for @command{ld}. The @option{-V} option also
978 lists the supported emulations.
981 @kindex --discard-all
982 @cindex deleting local symbols
985 Delete all local symbols.
988 @kindex --discard-locals
989 @cindex local symbols, deleting
991 @itemx --discard-locals
992 Delete all temporary local symbols. (These symbols start with
993 system-specific local label prefixes, typically @samp{.L} for ELF systems
994 or @samp{L} for traditional a.out systems.)
996 @kindex -y @var{symbol}
997 @kindex --trace-symbol=@var{symbol}
998 @cindex symbol tracing
999 @item -y @var{symbol}
1000 @itemx --trace-symbol=@var{symbol}
1001 Print the name of each linked file in which @var{symbol} appears. This
1002 option may be given any number of times. On many systems it is necessary
1003 to prepend an underscore.
1005 This option is useful when you have an undefined symbol in your link but
1006 don't know where the reference is coming from.
1008 @kindex -Y @var{path}
1010 Add @var{path} to the default library search path. This option exists
1011 for Solaris compatibility.
1013 @kindex -z @var{keyword}
1014 @item -z @var{keyword}
1015 The recognized keywords are:
1019 Combines multiple reloc sections and sorts them to make dynamic symbol
1020 lookup caching possible.
1023 Disallows undefined symbols in object files. Undefined symbols in
1024 shared libraries are still allowed.
1027 Marks the object as requiring executable stack.
1030 This option is only meaningful when building a shared object. It makes
1031 the symbols defined by this shared object available for symbol resolution
1032 of subsequently loaded libraries.
1035 This option is only meaningful when building a shared object.
1036 It marks the object so that its runtime initialization will occur
1037 before the runtime initialization of any other objects brought into
1038 the process at the same time. Similarly the runtime finalization of
1039 the object will occur after the runtime finalization of any other
1043 Marks the object that its symbol table interposes before all symbols
1044 but the primary executable.
1047 When generating an executable or shared library, mark it to tell the
1048 dynamic linker to defer function call resolution to the point when
1049 the function is called (lazy binding), rather than at load time.
1050 Lazy binding is the default.
1053 Marks the object that its filters be processed immediately at
1057 Allows multiple definitions.
1060 Disables multiple reloc sections combining.
1063 Disables production of copy relocs.
1066 Marks the object that the search for dependencies of this object will
1067 ignore any default library search paths.
1070 Marks the object shouldn't be unloaded at runtime.
1073 Marks the object not available to @code{dlopen}.
1076 Marks the object can not be dumped by @code{dldump}.
1079 Marks the object as not requiring executable stack.
1082 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1085 When generating an executable or shared library, mark it to tell the
1086 dynamic linker to resolve all symbols when the program is started, or
1087 when the shared library is linked to using dlopen, instead of
1088 deferring function call resolution to the point when the function is
1092 Marks the object may contain $ORIGIN.
1095 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1097 @item max-page-size=@var{value}
1098 Set the emulation maximum page size to @var{value}.
1100 @item common-page-size=@var{value}
1101 Set the emulation common page size to @var{value}.
1103 @item stack-size=@var{value}
1104 Specify a stack size for in an ELF @code{PT_GNU_STACK} segment.
1105 Specifying zero will override any default non-zero sized
1106 @code{PT_GNU_STACK} segment creation.
1110 Other keywords are ignored for Solaris compatibility.
1113 @cindex groups of archives
1114 @item -( @var{archives} -)
1115 @itemx --start-group @var{archives} --end-group
1116 The @var{archives} should be a list of archive files. They may be
1117 either explicit file names, or @samp{-l} options.
1119 The specified archives are searched repeatedly until no new undefined
1120 references are created. Normally, an archive is searched only once in
1121 the order that it is specified on the command line. If a symbol in that
1122 archive is needed to resolve an undefined symbol referred to by an
1123 object in an archive that appears later on the command line, the linker
1124 would not be able to resolve that reference. By grouping the archives,
1125 they all be searched repeatedly until all possible references are
1128 Using this option has a significant performance cost. It is best to use
1129 it only when there are unavoidable circular references between two or
1132 @kindex --accept-unknown-input-arch
1133 @kindex --no-accept-unknown-input-arch
1134 @item --accept-unknown-input-arch
1135 @itemx --no-accept-unknown-input-arch
1136 Tells the linker to accept input files whose architecture cannot be
1137 recognised. The assumption is that the user knows what they are doing
1138 and deliberately wants to link in these unknown input files. This was
1139 the default behaviour of the linker, before release 2.14. The default
1140 behaviour from release 2.14 onwards is to reject such input files, and
1141 so the @samp{--accept-unknown-input-arch} option has been added to
1142 restore the old behaviour.
1145 @kindex --no-as-needed
1147 @itemx --no-as-needed
1148 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1149 on the command line after the @option{--as-needed} option. Normally
1150 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1151 on the command line, regardless of whether the library is actually
1152 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1153 emitted for a library that satisfies an undefined symbol reference
1154 from a regular object file or, if the library is not found in the
1155 DT_NEEDED lists of other libraries linked up to that point, an
1156 undefined symbol reference from another dynamic library.
1157 @option{--no-as-needed} restores the default behaviour.
1159 @kindex --add-needed
1160 @kindex --no-add-needed
1162 @itemx --no-add-needed
1163 These two options have been deprecated because of the similarity of
1164 their names to the @option{--as-needed} and @option{--no-as-needed}
1165 options. They have been replaced by @option{--copy-dt-needed-entries}
1166 and @option{--no-copy-dt-needed-entries}.
1168 @kindex -assert @var{keyword}
1169 @item -assert @var{keyword}
1170 This option is ignored for SunOS compatibility.
1174 @kindex -call_shared
1178 Link against dynamic libraries. This is only meaningful on platforms
1179 for which shared libraries are supported. This option is normally the
1180 default on such platforms. The different variants of this option are
1181 for compatibility with various systems. You may use this option
1182 multiple times on the command line: it affects library searching for
1183 @option{-l} options which follow it.
1187 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1188 section. This causes the runtime linker to handle lookups in this
1189 object and its dependencies to be performed only inside the group.
1190 @option{--unresolved-symbols=report-all} is implied. This option is
1191 only meaningful on ELF platforms which support shared libraries.
1201 Do not link against shared libraries. This is only meaningful on
1202 platforms for which shared libraries are supported. The different
1203 variants of this option are for compatibility with various systems. You
1204 may use this option multiple times on the command line: it affects
1205 library searching for @option{-l} options which follow it. This
1206 option also implies @option{--unresolved-symbols=report-all}. This
1207 option can be used with @option{-shared}. Doing so means that a
1208 shared library is being created but that all of the library's external
1209 references must be resolved by pulling in entries from static
1214 When creating a shared library, bind references to global symbols to the
1215 definition within the shared library, if any. Normally, it is possible
1216 for a program linked against a shared library to override the definition
1217 within the shared library. This option is only meaningful on ELF
1218 platforms which support shared libraries.
1220 @kindex -Bsymbolic-functions
1221 @item -Bsymbolic-functions
1222 When creating a shared library, bind references to global function
1223 symbols to the definition within the shared library, if any.
1224 This option is only meaningful on ELF platforms which support shared
1227 @kindex --dynamic-list=@var{dynamic-list-file}
1228 @item --dynamic-list=@var{dynamic-list-file}
1229 Specify the name of a dynamic list file to the linker. This is
1230 typically used when creating shared libraries to specify a list of
1231 global symbols whose references shouldn't be bound to the definition
1232 within the shared library, or creating dynamically linked executables
1233 to specify a list of symbols which should be added to the symbol table
1234 in the executable. This option is only meaningful on ELF platforms
1235 which support shared libraries.
1237 The format of the dynamic list is the same as the version node without
1238 scope and node name. See @ref{VERSION} for more information.
1240 @kindex --dynamic-list-data
1241 @item --dynamic-list-data
1242 Include all global data symbols to the dynamic list.
1244 @kindex --dynamic-list-cpp-new
1245 @item --dynamic-list-cpp-new
1246 Provide the builtin dynamic list for C++ operator new and delete. It
1247 is mainly useful for building shared libstdc++.
1249 @kindex --dynamic-list-cpp-typeinfo
1250 @item --dynamic-list-cpp-typeinfo
1251 Provide the builtin dynamic list for C++ runtime type identification.
1253 @kindex --check-sections
1254 @kindex --no-check-sections
1255 @item --check-sections
1256 @itemx --no-check-sections
1257 Asks the linker @emph{not} to check section addresses after they have
1258 been assigned to see if there are any overlaps. Normally the linker will
1259 perform this check, and if it finds any overlaps it will produce
1260 suitable error messages. The linker does know about, and does make
1261 allowances for sections in overlays. The default behaviour can be
1262 restored by using the command line switch @option{--check-sections}.
1263 Section overlap is not usually checked for relocatable links. You can
1264 force checking in that case by using the @option{--check-sections}
1267 @kindex --copy-dt-needed-entries
1268 @kindex --no-copy-dt-needed-entries
1269 @item --copy-dt-needed-entries
1270 @itemx --no-copy-dt-needed-entries
1271 This option affects the treatment of dynamic libraries referred to
1272 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1273 command line. Normally the linker won't add a DT_NEEDED tag to the
1274 output binary for each library mentioned in a DT_NEEDED tag in an
1275 input dynamic library. With @option{--copy-dt-needed-entries}
1276 specified on the command line however any dynamic libraries that
1277 follow it will have their DT_NEEDED entries added. The default
1278 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1280 This option also has an effect on the resolution of symbols in dynamic
1281 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1282 mentioned on the command line will be recursively searched, following
1283 their DT_NEEDED tags to other libraries, in order to resolve symbols
1284 required by the output binary. With the default setting however
1285 the searching of dynamic libraries that follow it will stop with the
1286 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1289 @cindex cross reference table
1292 Output a cross reference table. If a linker map file is being
1293 generated, the cross reference table is printed to the map file.
1294 Otherwise, it is printed on the standard output.
1296 The format of the table is intentionally simple, so that it may be
1297 easily processed by a script if necessary. The symbols are printed out,
1298 sorted by name. For each symbol, a list of file names is given. If the
1299 symbol is defined, the first file listed is the location of the
1300 definition. The remaining files contain references to the symbol.
1302 @cindex common allocation
1303 @kindex --no-define-common
1304 @item --no-define-common
1305 This option inhibits the assignment of addresses to common symbols.
1306 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1307 @xref{Miscellaneous Commands}.
1309 The @samp{--no-define-common} option allows decoupling
1310 the decision to assign addresses to Common symbols from the choice
1311 of the output file type; otherwise a non-Relocatable output type
1312 forces assigning addresses to Common symbols.
1313 Using @samp{--no-define-common} allows Common symbols that are referenced
1314 from a shared library to be assigned addresses only in the main program.
1315 This eliminates the unused duplicate space in the shared library,
1316 and also prevents any possible confusion over resolving to the wrong
1317 duplicate when there are many dynamic modules with specialized search
1318 paths for runtime symbol resolution.
1320 @cindex symbols, from command line
1321 @kindex --defsym=@var{symbol}=@var{exp}
1322 @item --defsym=@var{symbol}=@var{expression}
1323 Create a global symbol in the output file, containing the absolute
1324 address given by @var{expression}. You may use this option as many
1325 times as necessary to define multiple symbols in the command line. A
1326 limited form of arithmetic is supported for the @var{expression} in this
1327 context: you may give a hexadecimal constant or the name of an existing
1328 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1329 constants or symbols. If you need more elaborate expressions, consider
1330 using the linker command language from a script (@pxref{Assignments,,
1331 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1332 space between @var{symbol}, the equals sign (``@key{=}''), and
1335 @cindex demangling, from command line
1336 @kindex --demangle[=@var{style}]
1337 @kindex --no-demangle
1338 @item --demangle[=@var{style}]
1339 @itemx --no-demangle
1340 These options control whether to demangle symbol names in error messages
1341 and other output. When the linker is told to demangle, it tries to
1342 present symbol names in a readable fashion: it strips leading
1343 underscores if they are used by the object file format, and converts C++
1344 mangled symbol names into user readable names. Different compilers have
1345 different mangling styles. The optional demangling style argument can be used
1346 to choose an appropriate demangling style for your compiler. The linker will
1347 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1348 is set. These options may be used to override the default.
1350 @cindex dynamic linker, from command line
1351 @kindex -I@var{file}
1352 @kindex --dynamic-linker=@var{file}
1354 @itemx --dynamic-linker=@var{file}
1355 Set the name of the dynamic linker. This is only meaningful when
1356 generating dynamically linked ELF executables. The default dynamic
1357 linker is normally correct; don't use this unless you know what you are
1360 @kindex --fatal-warnings
1361 @kindex --no-fatal-warnings
1362 @item --fatal-warnings
1363 @itemx --no-fatal-warnings
1364 Treat all warnings as errors. The default behaviour can be restored
1365 with the option @option{--no-fatal-warnings}.
1367 @kindex --force-exe-suffix
1368 @item --force-exe-suffix
1369 Make sure that an output file has a .exe suffix.
1371 If a successfully built fully linked output file does not have a
1372 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1373 the output file to one of the same name with a @code{.exe} suffix. This
1374 option is useful when using unmodified Unix makefiles on a Microsoft
1375 Windows host, since some versions of Windows won't run an image unless
1376 it ends in a @code{.exe} suffix.
1378 @kindex --gc-sections
1379 @kindex --no-gc-sections
1380 @cindex garbage collection
1382 @itemx --no-gc-sections
1383 Enable garbage collection of unused input sections. It is ignored on
1384 targets that do not support this option. The default behaviour (of not
1385 performing this garbage collection) can be restored by specifying
1386 @samp{--no-gc-sections} on the command line.
1388 @samp{--gc-sections} decides which input sections are used by
1389 examining symbols and relocations. The section containing the entry
1390 symbol and all sections containing symbols undefined on the
1391 command-line will be kept, as will sections containing symbols
1392 referenced by dynamic objects. Note that when building shared
1393 libraries, the linker must assume that any visible symbol is
1394 referenced. Once this initial set of sections has been determined,
1395 the linker recursively marks as used any section referenced by their
1396 relocations. See @samp{--entry} and @samp{--undefined}.
1398 This option can be set when doing a partial link (enabled with option
1399 @samp{-r}). In this case the root of symbols kept must be explicitly
1400 specified either by an @samp{--entry} or @samp{--undefined} option or by
1401 a @code{ENTRY} command in the linker script.
1403 @kindex --print-gc-sections
1404 @kindex --no-print-gc-sections
1405 @cindex garbage collection
1406 @item --print-gc-sections
1407 @itemx --no-print-gc-sections
1408 List all sections removed by garbage collection. The listing is
1409 printed on stderr. This option is only effective if garbage
1410 collection has been enabled via the @samp{--gc-sections}) option. The
1411 default behaviour (of not listing the sections that are removed) can
1412 be restored by specifying @samp{--no-print-gc-sections} on the command
1415 @kindex --print-output-format
1416 @cindex output format
1417 @item --print-output-format
1418 Print the name of the default output format (perhaps influenced by
1419 other command-line options). This is the string that would appear
1420 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1426 Print a summary of the command-line options on the standard output and exit.
1428 @kindex --target-help
1430 Print a summary of all target specific options on the standard output and exit.
1432 @kindex -Map=@var{mapfile}
1433 @item -Map=@var{mapfile}
1434 Print a link map to the file @var{mapfile}. See the description of the
1435 @option{-M} option, above.
1437 @cindex memory usage
1438 @kindex --no-keep-memory
1439 @item --no-keep-memory
1440 @command{ld} normally optimizes for speed over memory usage by caching the
1441 symbol tables of input files in memory. This option tells @command{ld} to
1442 instead optimize for memory usage, by rereading the symbol tables as
1443 necessary. This may be required if @command{ld} runs out of memory space
1444 while linking a large executable.
1446 @kindex --no-undefined
1448 @item --no-undefined
1450 Report unresolved symbol references from regular object files. This
1451 is done even if the linker is creating a non-symbolic shared library.
1452 The switch @option{--[no-]allow-shlib-undefined} controls the
1453 behaviour for reporting unresolved references found in shared
1454 libraries being linked in.
1456 @kindex --allow-multiple-definition
1458 @item --allow-multiple-definition
1460 Normally when a symbol is defined multiple times, the linker will
1461 report a fatal error. These options allow multiple definitions and the
1462 first definition will be used.
1464 @kindex --allow-shlib-undefined
1465 @kindex --no-allow-shlib-undefined
1466 @item --allow-shlib-undefined
1467 @itemx --no-allow-shlib-undefined
1468 Allows or disallows undefined symbols in shared libraries.
1469 This switch is similar to @option{--no-undefined} except that it
1470 determines the behaviour when the undefined symbols are in a
1471 shared library rather than a regular object file. It does not affect
1472 how undefined symbols in regular object files are handled.
1474 The default behaviour is to report errors for any undefined symbols
1475 referenced in shared libraries if the linker is being used to create
1476 an executable, but to allow them if the linker is being used to create
1479 The reasons for allowing undefined symbol references in shared
1480 libraries specified at link time are that:
1484 A shared library specified at link time may not be the same as the one
1485 that is available at load time, so the symbol might actually be
1486 resolvable at load time.
1488 There are some operating systems, eg BeOS and HPPA, where undefined
1489 symbols in shared libraries are normal.
1491 The BeOS kernel for example patches shared libraries at load time to
1492 select whichever function is most appropriate for the current
1493 architecture. This is used, for example, to dynamically select an
1494 appropriate memset function.
1497 @kindex --no-undefined-version
1498 @item --no-undefined-version
1499 Normally when a symbol has an undefined version, the linker will ignore
1500 it. This option disallows symbols with undefined version and a fatal error
1501 will be issued instead.
1503 @kindex --default-symver
1504 @item --default-symver
1505 Create and use a default symbol version (the soname) for unversioned
1508 @kindex --default-imported-symver
1509 @item --default-imported-symver
1510 Create and use a default symbol version (the soname) for unversioned
1513 @kindex --no-warn-mismatch
1514 @item --no-warn-mismatch
1515 Normally @command{ld} will give an error if you try to link together input
1516 files that are mismatched for some reason, perhaps because they have
1517 been compiled for different processors or for different endiannesses.
1518 This option tells @command{ld} that it should silently permit such possible
1519 errors. This option should only be used with care, in cases when you
1520 have taken some special action that ensures that the linker errors are
1523 @kindex --no-warn-search-mismatch
1524 @item --no-warn-search-mismatch
1525 Normally @command{ld} will give a warning if it finds an incompatible
1526 library during a library search. This option silences the warning.
1528 @kindex --no-whole-archive
1529 @item --no-whole-archive
1530 Turn off the effect of the @option{--whole-archive} option for subsequent
1533 @cindex output file after errors
1534 @kindex --noinhibit-exec
1535 @item --noinhibit-exec
1536 Retain the executable output file whenever it is still usable.
1537 Normally, the linker will not produce an output file if it encounters
1538 errors during the link process; it exits without writing an output file
1539 when it issues any error whatsoever.
1543 Only search library directories explicitly specified on the
1544 command line. Library directories specified in linker scripts
1545 (including linker scripts specified on the command line) are ignored.
1547 @ifclear SingleFormat
1548 @kindex --oformat=@var{output-format}
1549 @item --oformat=@var{output-format}
1550 @command{ld} may be configured to support more than one kind of object
1551 file. If your @command{ld} is configured this way, you can use the
1552 @samp{--oformat} option to specify the binary format for the output
1553 object file. Even when @command{ld} is configured to support alternative
1554 object formats, you don't usually need to specify this, as @command{ld}
1555 should be configured to produce as a default output format the most
1556 usual format on each machine. @var{output-format} is a text string, the
1557 name of a particular format supported by the BFD libraries. (You can
1558 list the available binary formats with @samp{objdump -i}.) The script
1559 command @code{OUTPUT_FORMAT} can also specify the output format, but
1560 this option overrides it. @xref{BFD}.
1564 @kindex --pic-executable
1566 @itemx --pic-executable
1567 @cindex position independent executables
1568 Create a position independent executable. This is currently only supported on
1569 ELF platforms. Position independent executables are similar to shared
1570 libraries in that they are relocated by the dynamic linker to the virtual
1571 address the OS chooses for them (which can vary between invocations). Like
1572 normal dynamically linked executables they can be executed and symbols
1573 defined in the executable cannot be overridden by shared libraries.
1577 This option is ignored for Linux compatibility.
1581 This option is ignored for SVR4 compatibility.
1584 @cindex synthesizing linker
1585 @cindex relaxing addressing modes
1589 An option with machine dependent effects.
1591 This option is only supported on a few targets.
1594 @xref{H8/300,,@command{ld} and the H8/300}.
1597 @xref{i960,, @command{ld} and the Intel 960 family}.
1600 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1603 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1606 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1609 On some platforms the @samp{--relax} option performs target specific,
1610 global optimizations that become possible when the linker resolves
1611 addressing in the program, such as relaxing address modes,
1612 synthesizing new instructions, selecting shorter version of current
1613 instructions, and combining constant values.
1615 On some platforms these link time global optimizations may make symbolic
1616 debugging of the resulting executable impossible.
1618 This is known to be the case for the Matsushita MN10200 and MN10300
1619 family of processors.
1623 On platforms where this is not supported, @samp{--relax} is accepted,
1627 On platforms where @samp{--relax} is accepted the option
1628 @samp{--no-relax} can be used to disable the feature.
1630 @cindex retaining specified symbols
1631 @cindex stripping all but some symbols
1632 @cindex symbols, retaining selectively
1633 @kindex --retain-symbols-file=@var{filename}
1634 @item --retain-symbols-file=@var{filename}
1635 Retain @emph{only} the symbols listed in the file @var{filename},
1636 discarding all others. @var{filename} is simply a flat file, with one
1637 symbol name per line. This option is especially useful in environments
1641 where a large global symbol table is accumulated gradually, to conserve
1644 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1645 or symbols needed for relocations.
1647 You may only specify @samp{--retain-symbols-file} once in the command
1648 line. It overrides @samp{-s} and @samp{-S}.
1651 @item -rpath=@var{dir}
1652 @cindex runtime library search path
1653 @kindex -rpath=@var{dir}
1654 Add a directory to the runtime library search path. This is used when
1655 linking an ELF executable with shared objects. All @option{-rpath}
1656 arguments are concatenated and passed to the runtime linker, which uses
1657 them to locate shared objects at runtime. The @option{-rpath} option is
1658 also used when locating shared objects which are needed by shared
1659 objects explicitly included in the link; see the description of the
1660 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1661 ELF executable, the contents of the environment variable
1662 @code{LD_RUN_PATH} will be used if it is defined.
1664 The @option{-rpath} option may also be used on SunOS. By default, on
1665 SunOS, the linker will form a runtime search patch out of all the
1666 @option{-L} options it is given. If a @option{-rpath} option is used, the
1667 runtime search path will be formed exclusively using the @option{-rpath}
1668 options, ignoring the @option{-L} options. This can be useful when using
1669 gcc, which adds many @option{-L} options which may be on NFS mounted
1672 For compatibility with other ELF linkers, if the @option{-R} option is
1673 followed by a directory name, rather than a file name, it is treated as
1674 the @option{-rpath} option.
1678 @cindex link-time runtime library search path
1679 @kindex -rpath-link=@var{dir}
1680 @item -rpath-link=@var{dir}
1681 When using ELF or SunOS, one shared library may require another. This
1682 happens when an @code{ld -shared} link includes a shared library as one
1685 When the linker encounters such a dependency when doing a non-shared,
1686 non-relocatable link, it will automatically try to locate the required
1687 shared library and include it in the link, if it is not included
1688 explicitly. In such a case, the @option{-rpath-link} option
1689 specifies the first set of directories to search. The
1690 @option{-rpath-link} option may specify a sequence of directory names
1691 either by specifying a list of names separated by colons, or by
1692 appearing multiple times.
1694 This option should be used with caution as it overrides the search path
1695 that may have been hard compiled into a shared library. In such a case it
1696 is possible to use unintentionally a different search path than the
1697 runtime linker would do.
1699 The linker uses the following search paths to locate required shared
1703 Any directories specified by @option{-rpath-link} options.
1705 Any directories specified by @option{-rpath} options. The difference
1706 between @option{-rpath} and @option{-rpath-link} is that directories
1707 specified by @option{-rpath} options are included in the executable and
1708 used at runtime, whereas the @option{-rpath-link} option is only effective
1709 at link time. Searching @option{-rpath} in this way is only supported
1710 by native linkers and cross linkers which have been configured with
1711 the @option{--with-sysroot} option.
1713 On an ELF system, for native linkers, if the @option{-rpath} and
1714 @option{-rpath-link} options were not used, search the contents of the
1715 environment variable @code{LD_RUN_PATH}.
1717 On SunOS, if the @option{-rpath} option was not used, search any
1718 directories specified using @option{-L} options.
1720 For a native linker, search the contents of the environment
1721 variable @code{LD_LIBRARY_PATH}.
1723 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1724 @code{DT_RPATH} of a shared library are searched for shared
1725 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1726 @code{DT_RUNPATH} entries exist.
1728 The default directories, normally @file{/lib} and @file{/usr/lib}.
1730 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1731 exists, the list of directories found in that file.
1734 If the required shared library is not found, the linker will issue a
1735 warning and continue with the link.
1742 @cindex shared libraries
1743 Create a shared library. This is currently only supported on ELF, XCOFF
1744 and SunOS platforms. On SunOS, the linker will automatically create a
1745 shared library if the @option{-e} option is not used and there are
1746 undefined symbols in the link.
1748 @kindex --sort-common
1750 @itemx --sort-common=ascending
1751 @itemx --sort-common=descending
1752 This option tells @command{ld} to sort the common symbols by alignment in
1753 ascending or descending order when it places them in the appropriate output
1754 sections. The symbol alignments considered are sixteen-byte or larger,
1755 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1756 between symbols due to alignment constraints. If no sorting order is
1757 specified, then descending order is assumed.
1759 @kindex --sort-section=name
1760 @item --sort-section=name
1761 This option will apply @code{SORT_BY_NAME} to all wildcard section
1762 patterns in the linker script.
1764 @kindex --sort-section=alignment
1765 @item --sort-section=alignment
1766 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1767 patterns in the linker script.
1769 @kindex --split-by-file
1770 @item --split-by-file[=@var{size}]
1771 Similar to @option{--split-by-reloc} but creates a new output section for
1772 each input file when @var{size} is reached. @var{size} defaults to a
1773 size of 1 if not given.
1775 @kindex --split-by-reloc
1776 @item --split-by-reloc[=@var{count}]
1777 Tries to creates extra sections in the output file so that no single
1778 output section in the file contains more than @var{count} relocations.
1779 This is useful when generating huge relocatable files for downloading into
1780 certain real time kernels with the COFF object file format; since COFF
1781 cannot represent more than 65535 relocations in a single section. Note
1782 that this will fail to work with object file formats which do not
1783 support arbitrary sections. The linker will not split up individual
1784 input sections for redistribution, so if a single input section contains
1785 more than @var{count} relocations one output section will contain that
1786 many relocations. @var{count} defaults to a value of 32768.
1790 Compute and display statistics about the operation of the linker, such
1791 as execution time and memory usage.
1793 @kindex --sysroot=@var{directory}
1794 @item --sysroot=@var{directory}
1795 Use @var{directory} as the location of the sysroot, overriding the
1796 configure-time default. This option is only supported by linkers
1797 that were configured using @option{--with-sysroot}.
1799 @kindex --traditional-format
1800 @cindex traditional format
1801 @item --traditional-format
1802 For some targets, the output of @command{ld} is different in some ways from
1803 the output of some existing linker. This switch requests @command{ld} to
1804 use the traditional format instead.
1807 For example, on SunOS, @command{ld} combines duplicate entries in the
1808 symbol string table. This can reduce the size of an output file with
1809 full debugging information by over 30 percent. Unfortunately, the SunOS
1810 @code{dbx} program can not read the resulting program (@code{gdb} has no
1811 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1812 combine duplicate entries.
1814 @kindex --section-start=@var{sectionname}=@var{org}
1815 @item --section-start=@var{sectionname}=@var{org}
1816 Locate a section in the output file at the absolute
1817 address given by @var{org}. You may use this option as many
1818 times as necessary to locate multiple sections in the command
1820 @var{org} must be a single hexadecimal integer;
1821 for compatibility with other linkers, you may omit the leading
1822 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1823 should be no white space between @var{sectionname}, the equals
1824 sign (``@key{=}''), and @var{org}.
1826 @kindex -Tbss=@var{org}
1827 @kindex -Tdata=@var{org}
1828 @kindex -Ttext=@var{org}
1829 @cindex segment origins, cmd line
1830 @item -Tbss=@var{org}
1831 @itemx -Tdata=@var{org}
1832 @itemx -Ttext=@var{org}
1833 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
1834 @code{.text} as the @var{sectionname}.
1836 @kindex -Ttext-segment=@var{org}
1837 @item -Ttext-segment=@var{org}
1838 @cindex text segment origin, cmd line
1839 When creating an ELF executable or shared object, it will set the address
1840 of the first byte of the text segment.
1842 @kindex -Trodata-segment=@var{org}
1843 @item -Trodata-segment=@var{org}
1844 @cindex rodata segment origin, cmd line
1845 When creating an ELF executable or shared object for a target where
1846 the read-only data is in its own segment separate from the executable
1847 text, it will set the address of the first byte of the read-only data segment.
1849 @kindex -Tldata-segment=@var{org}
1850 @item -Tldata-segment=@var{org}
1851 @cindex ldata segment origin, cmd line
1852 When creating an ELF executable or shared object for x86-64 medium memory
1853 model, it will set the address of the first byte of the ldata segment.
1855 @kindex --unresolved-symbols
1856 @item --unresolved-symbols=@var{method}
1857 Determine how to handle unresolved symbols. There are four possible
1858 values for @samp{method}:
1862 Do not report any unresolved symbols.
1865 Report all unresolved symbols. This is the default.
1867 @item ignore-in-object-files
1868 Report unresolved symbols that are contained in shared libraries, but
1869 ignore them if they come from regular object files.
1871 @item ignore-in-shared-libs
1872 Report unresolved symbols that come from regular object files, but
1873 ignore them if they come from shared libraries. This can be useful
1874 when creating a dynamic binary and it is known that all the shared
1875 libraries that it should be referencing are included on the linker's
1879 The behaviour for shared libraries on their own can also be controlled
1880 by the @option{--[no-]allow-shlib-undefined} option.
1882 Normally the linker will generate an error message for each reported
1883 unresolved symbol but the option @option{--warn-unresolved-symbols}
1884 can change this to a warning.
1886 @kindex --verbose[=@var{NUMBER}]
1887 @cindex verbose[=@var{NUMBER}]
1889 @itemx --verbose[=@var{NUMBER}]
1890 Display the version number for @command{ld} and list the linker emulations
1891 supported. Display which input files can and cannot be opened. Display
1892 the linker script being used by the linker. If the optional @var{NUMBER}
1893 argument > 1, plugin symbol status will also be displayed.
1895 @kindex --version-script=@var{version-scriptfile}
1896 @cindex version script, symbol versions
1897 @item --version-script=@var{version-scriptfile}
1898 Specify the name of a version script to the linker. This is typically
1899 used when creating shared libraries to specify additional information
1900 about the version hierarchy for the library being created. This option
1901 is only fully supported on ELF platforms which support shared libraries;
1902 see @ref{VERSION}. It is partially supported on PE platforms, which can
1903 use version scripts to filter symbol visibility in auto-export mode: any
1904 symbols marked @samp{local} in the version script will not be exported.
1907 @kindex --warn-common
1908 @cindex warnings, on combining symbols
1909 @cindex combining symbols, warnings on
1911 Warn when a common symbol is combined with another common symbol or with
1912 a symbol definition. Unix linkers allow this somewhat sloppy practice,
1913 but linkers on some other operating systems do not. This option allows
1914 you to find potential problems from combining global symbols.
1915 Unfortunately, some C libraries use this practice, so you may get some
1916 warnings about symbols in the libraries as well as in your programs.
1918 There are three kinds of global symbols, illustrated here by C examples:
1922 A definition, which goes in the initialized data section of the output
1926 An undefined reference, which does not allocate space.
1927 There must be either a definition or a common symbol for the
1931 A common symbol. If there are only (one or more) common symbols for a
1932 variable, it goes in the uninitialized data area of the output file.
1933 The linker merges multiple common symbols for the same variable into a
1934 single symbol. If they are of different sizes, it picks the largest
1935 size. The linker turns a common symbol into a declaration, if there is
1936 a definition of the same variable.
1939 The @samp{--warn-common} option can produce five kinds of warnings.
1940 Each warning consists of a pair of lines: the first describes the symbol
1941 just encountered, and the second describes the previous symbol
1942 encountered with the same name. One or both of the two symbols will be
1947 Turning a common symbol into a reference, because there is already a
1948 definition for the symbol.
1950 @var{file}(@var{section}): warning: common of `@var{symbol}'
1951 overridden by definition
1952 @var{file}(@var{section}): warning: defined here
1956 Turning a common symbol into a reference, because a later definition for
1957 the symbol is encountered. This is the same as the previous case,
1958 except that the symbols are encountered in a different order.
1960 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1962 @var{file}(@var{section}): warning: common is here
1966 Merging a common symbol with a previous same-sized common symbol.
1968 @var{file}(@var{section}): warning: multiple common
1970 @var{file}(@var{section}): warning: previous common is here
1974 Merging a common symbol with a previous larger common symbol.
1976 @var{file}(@var{section}): warning: common of `@var{symbol}'
1977 overridden by larger common
1978 @var{file}(@var{section}): warning: larger common is here
1982 Merging a common symbol with a previous smaller common symbol. This is
1983 the same as the previous case, except that the symbols are
1984 encountered in a different order.
1986 @var{file}(@var{section}): warning: common of `@var{symbol}'
1987 overriding smaller common
1988 @var{file}(@var{section}): warning: smaller common is here
1992 @kindex --warn-constructors
1993 @item --warn-constructors
1994 Warn if any global constructors are used. This is only useful for a few
1995 object file formats. For formats like COFF or ELF, the linker can not
1996 detect the use of global constructors.
1998 @kindex --warn-multiple-gp
1999 @item --warn-multiple-gp
2000 Warn if multiple global pointer values are required in the output file.
2001 This is only meaningful for certain processors, such as the Alpha.
2002 Specifically, some processors put large-valued constants in a special
2003 section. A special register (the global pointer) points into the middle
2004 of this section, so that constants can be loaded efficiently via a
2005 base-register relative addressing mode. Since the offset in
2006 base-register relative mode is fixed and relatively small (e.g., 16
2007 bits), this limits the maximum size of the constant pool. Thus, in
2008 large programs, it is often necessary to use multiple global pointer
2009 values in order to be able to address all possible constants. This
2010 option causes a warning to be issued whenever this case occurs.
2013 @cindex warnings, on undefined symbols
2014 @cindex undefined symbols, warnings on
2016 Only warn once for each undefined symbol, rather than once per module
2019 @kindex --warn-section-align
2020 @cindex warnings, on section alignment
2021 @cindex section alignment, warnings on
2022 @item --warn-section-align
2023 Warn if the address of an output section is changed because of
2024 alignment. Typically, the alignment will be set by an input section.
2025 The address will only be changed if it not explicitly specified; that
2026 is, if the @code{SECTIONS} command does not specify a start address for
2027 the section (@pxref{SECTIONS}).
2029 @kindex --warn-shared-textrel
2030 @item --warn-shared-textrel
2031 Warn if the linker adds a DT_TEXTREL to a shared object.
2033 @kindex --warn-alternate-em
2034 @item --warn-alternate-em
2035 Warn if an object has alternate ELF machine code.
2037 @kindex --warn-unresolved-symbols
2038 @item --warn-unresolved-symbols
2039 If the linker is going to report an unresolved symbol (see the option
2040 @option{--unresolved-symbols}) it will normally generate an error.
2041 This option makes it generate a warning instead.
2043 @kindex --error-unresolved-symbols
2044 @item --error-unresolved-symbols
2045 This restores the linker's default behaviour of generating errors when
2046 it is reporting unresolved symbols.
2048 @kindex --whole-archive
2049 @cindex including an entire archive
2050 @item --whole-archive
2051 For each archive mentioned on the command line after the
2052 @option{--whole-archive} option, include every object file in the archive
2053 in the link, rather than searching the archive for the required object
2054 files. This is normally used to turn an archive file into a shared
2055 library, forcing every object to be included in the resulting shared
2056 library. This option may be used more than once.
2058 Two notes when using this option from gcc: First, gcc doesn't know
2059 about this option, so you have to use @option{-Wl,-whole-archive}.
2060 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2061 list of archives, because gcc will add its own list of archives to
2062 your link and you may not want this flag to affect those as well.
2064 @kindex --wrap=@var{symbol}
2065 @item --wrap=@var{symbol}
2066 Use a wrapper function for @var{symbol}. Any undefined reference to
2067 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2068 undefined reference to @code{__real_@var{symbol}} will be resolved to
2071 This can be used to provide a wrapper for a system function. The
2072 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2073 wishes to call the system function, it should call
2074 @code{__real_@var{symbol}}.
2076 Here is a trivial example:
2080 __wrap_malloc (size_t c)
2082 printf ("malloc called with %zu\n", c);
2083 return __real_malloc (c);
2087 If you link other code with this file using @option{--wrap malloc}, then
2088 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2089 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2090 call the real @code{malloc} function.
2092 You may wish to provide a @code{__real_malloc} function as well, so that
2093 links without the @option{--wrap} option will succeed. If you do this,
2094 you should not put the definition of @code{__real_malloc} in the same
2095 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2096 call before the linker has a chance to wrap it to @code{malloc}.
2098 @kindex --eh-frame-hdr
2099 @item --eh-frame-hdr
2100 Request creation of @code{.eh_frame_hdr} section and ELF
2101 @code{PT_GNU_EH_FRAME} segment header.
2103 @kindex --ld-generated-unwind-info
2104 @item --no-ld-generated-unwind-info
2105 Request creation of @code{.eh_frame} unwind info for linker
2106 generated code sections like PLT. This option is on by default
2107 if linker generated unwind info is supported.
2109 @kindex --enable-new-dtags
2110 @kindex --disable-new-dtags
2111 @item --enable-new-dtags
2112 @itemx --disable-new-dtags
2113 This linker can create the new dynamic tags in ELF. But the older ELF
2114 systems may not understand them. If you specify
2115 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2116 and older dynamic tags will be omitted.
2117 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2118 created. By default, the new dynamic tags are not created. Note that
2119 those options are only available for ELF systems.
2121 @kindex --hash-size=@var{number}
2122 @item --hash-size=@var{number}
2123 Set the default size of the linker's hash tables to a prime number
2124 close to @var{number}. Increasing this value can reduce the length of
2125 time it takes the linker to perform its tasks, at the expense of
2126 increasing the linker's memory requirements. Similarly reducing this
2127 value can reduce the memory requirements at the expense of speed.
2129 @kindex --hash-style=@var{style}
2130 @item --hash-style=@var{style}
2131 Set the type of linker's hash table(s). @var{style} can be either
2132 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2133 new style GNU @code{.gnu.hash} section or @code{both} for both
2134 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2135 hash tables. The default is @code{sysv}.
2137 @kindex --reduce-memory-overheads
2138 @item --reduce-memory-overheads
2139 This option reduces memory requirements at ld runtime, at the expense of
2140 linking speed. This was introduced to select the old O(n^2) algorithm
2141 for link map file generation, rather than the new O(n) algorithm which uses
2142 about 40% more memory for symbol storage.
2144 Another effect of the switch is to set the default hash table size to
2145 1021, which again saves memory at the cost of lengthening the linker's
2146 run time. This is not done however if the @option{--hash-size} switch
2149 The @option{--reduce-memory-overheads} switch may be also be used to
2150 enable other tradeoffs in future versions of the linker.
2153 @kindex --build-id=@var{style}
2155 @itemx --build-id=@var{style}
2156 Request creation of @code{.note.gnu.build-id} ELF note section.
2157 The contents of the note are unique bits identifying this linked
2158 file. @var{style} can be @code{uuid} to use 128 random bits,
2159 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
2160 parts of the output contents, @code{md5} to use a 128-bit
2161 @sc{MD5} hash on the normative parts of the output contents, or
2162 @code{0x@var{hexstring}} to use a chosen bit string specified as
2163 an even number of hexadecimal digits (@code{-} and @code{:}
2164 characters between digit pairs are ignored). If @var{style} is
2165 omitted, @code{sha1} is used.
2167 The @code{md5} and @code{sha1} styles produces an identifier
2168 that is always the same in an identical output file, but will be
2169 unique among all nonidentical output files. It is not intended
2170 to be compared as a checksum for the file's contents. A linked
2171 file may be changed later by other tools, but the build ID bit
2172 string identifying the original linked file does not change.
2174 Passing @code{none} for @var{style} disables the setting from any
2175 @code{--build-id} options earlier on the command line.
2180 @subsection Options Specific to i386 PE Targets
2182 @c man begin OPTIONS
2184 The i386 PE linker supports the @option{-shared} option, which causes
2185 the output to be a dynamically linked library (DLL) instead of a
2186 normal executable. You should name the output @code{*.dll} when you
2187 use this option. In addition, the linker fully supports the standard
2188 @code{*.def} files, which may be specified on the linker command line
2189 like an object file (in fact, it should precede archives it exports
2190 symbols from, to ensure that they get linked in, just like a normal
2193 In addition to the options common to all targets, the i386 PE linker
2194 support additional command line options that are specific to the i386
2195 PE target. Options that take values may be separated from their
2196 values by either a space or an equals sign.
2200 @kindex --add-stdcall-alias
2201 @item --add-stdcall-alias
2202 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2203 as-is and also with the suffix stripped.
2204 [This option is specific to the i386 PE targeted port of the linker]
2207 @item --base-file @var{file}
2208 Use @var{file} as the name of a file in which to save the base
2209 addresses of all the relocations needed for generating DLLs with
2211 [This is an i386 PE specific option]
2215 Create a DLL instead of a regular executable. You may also use
2216 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2218 [This option is specific to the i386 PE targeted port of the linker]
2220 @kindex --enable-long-section-names
2221 @kindex --disable-long-section-names
2222 @item --enable-long-section-names
2223 @itemx --disable-long-section-names
2224 The PE variants of the Coff object format add an extension that permits
2225 the use of section names longer than eight characters, the normal limit
2226 for Coff. By default, these names are only allowed in object files, as
2227 fully-linked executable images do not carry the Coff string table required
2228 to support the longer names. As a GNU extension, it is possible to
2229 allow their use in executable images as well, or to (probably pointlessly!)
2230 disallow it in object files, by using these two options. Executable images
2231 generated with these long section names are slightly non-standard, carrying
2232 as they do a string table, and may generate confusing output when examined
2233 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2234 GDB relies on the use of PE long section names to find Dwarf-2 debug
2235 information sections in an executable image at runtime, and so if neither
2236 option is specified on the command-line, @command{ld} will enable long
2237 section names, overriding the default and technically correct behaviour,
2238 when it finds the presence of debug information while linking an executable
2239 image and not stripping symbols.
2240 [This option is valid for all PE targeted ports of the linker]
2242 @kindex --enable-stdcall-fixup
2243 @kindex --disable-stdcall-fixup
2244 @item --enable-stdcall-fixup
2245 @itemx --disable-stdcall-fixup
2246 If the link finds a symbol that it cannot resolve, it will attempt to
2247 do ``fuzzy linking'' by looking for another defined symbol that differs
2248 only in the format of the symbol name (cdecl vs stdcall) and will
2249 resolve that symbol by linking to the match. For example, the
2250 undefined symbol @code{_foo} might be linked to the function
2251 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2252 to the function @code{_bar}. When the linker does this, it prints a
2253 warning, since it normally should have failed to link, but sometimes
2254 import libraries generated from third-party dlls may need this feature
2255 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2256 feature is fully enabled and warnings are not printed. If you specify
2257 @option{--disable-stdcall-fixup}, this feature is disabled and such
2258 mismatches are considered to be errors.
2259 [This option is specific to the i386 PE targeted port of the linker]
2261 @kindex --leading-underscore
2262 @kindex --no-leading-underscore
2263 @item --leading-underscore
2264 @itemx --no-leading-underscore
2265 For most targets default symbol-prefix is an underscore and is defined
2266 in target's description. By this option it is possible to
2267 disable/enable the default underscore symbol-prefix.
2269 @cindex DLLs, creating
2270 @kindex --export-all-symbols
2271 @item --export-all-symbols
2272 If given, all global symbols in the objects used to build a DLL will
2273 be exported by the DLL. Note that this is the default if there
2274 otherwise wouldn't be any exported symbols. When symbols are
2275 explicitly exported via DEF files or implicitly exported via function
2276 attributes, the default is to not export anything else unless this
2277 option is given. Note that the symbols @code{DllMain@@12},
2278 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2279 @code{impure_ptr} will not be automatically
2280 exported. Also, symbols imported from other DLLs will not be
2281 re-exported, nor will symbols specifying the DLL's internal layout
2282 such as those beginning with @code{_head_} or ending with
2283 @code{_iname}. In addition, no symbols from @code{libgcc},
2284 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2285 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2286 not be exported, to help with C++ DLLs. Finally, there is an
2287 extensive list of cygwin-private symbols that are not exported
2288 (obviously, this applies on when building DLLs for cygwin targets).
2289 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2290 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2291 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2292 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2293 @code{cygwin_premain3}, and @code{environ}.
2294 [This option is specific to the i386 PE targeted port of the linker]
2296 @kindex --exclude-symbols
2297 @item --exclude-symbols @var{symbol},@var{symbol},...
2298 Specifies a list of symbols which should not be automatically
2299 exported. The symbol names may be delimited by commas or colons.
2300 [This option is specific to the i386 PE targeted port of the linker]
2302 @kindex --exclude-all-symbols
2303 @item --exclude-all-symbols
2304 Specifies no symbols should be automatically exported.
2305 [This option is specific to the i386 PE targeted port of the linker]
2307 @kindex --file-alignment
2308 @item --file-alignment
2309 Specify the file alignment. Sections in the file will always begin at
2310 file offsets which are multiples of this number. This defaults to
2312 [This option is specific to the i386 PE targeted port of the linker]
2316 @item --heap @var{reserve}
2317 @itemx --heap @var{reserve},@var{commit}
2318 Specify the number of bytes of memory to reserve (and optionally commit)
2319 to be used as heap for this program. The default is 1Mb reserved, 4K
2321 [This option is specific to the i386 PE targeted port of the linker]
2324 @kindex --image-base
2325 @item --image-base @var{value}
2326 Use @var{value} as the base address of your program or dll. This is
2327 the lowest memory location that will be used when your program or dll
2328 is loaded. To reduce the need to relocate and improve performance of
2329 your dlls, each should have a unique base address and not overlap any
2330 other dlls. The default is 0x400000 for executables, and 0x10000000
2332 [This option is specific to the i386 PE targeted port of the linker]
2336 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2337 symbols before they are exported.
2338 [This option is specific to the i386 PE targeted port of the linker]
2340 @kindex --large-address-aware
2341 @item --large-address-aware
2342 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2343 header is set to indicate that this executable supports virtual addresses
2344 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2345 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2346 section of the BOOT.INI. Otherwise, this bit has no effect.
2347 [This option is specific to PE targeted ports of the linker]
2349 @kindex --major-image-version
2350 @item --major-image-version @var{value}
2351 Sets the major number of the ``image version''. Defaults to 1.
2352 [This option is specific to the i386 PE targeted port of the linker]
2354 @kindex --major-os-version
2355 @item --major-os-version @var{value}
2356 Sets the major number of the ``os version''. Defaults to 4.
2357 [This option is specific to the i386 PE targeted port of the linker]
2359 @kindex --major-subsystem-version
2360 @item --major-subsystem-version @var{value}
2361 Sets the major number of the ``subsystem version''. Defaults to 4.
2362 [This option is specific to the i386 PE targeted port of the linker]
2364 @kindex --minor-image-version
2365 @item --minor-image-version @var{value}
2366 Sets the minor number of the ``image version''. Defaults to 0.
2367 [This option is specific to the i386 PE targeted port of the linker]
2369 @kindex --minor-os-version
2370 @item --minor-os-version @var{value}
2371 Sets the minor number of the ``os version''. Defaults to 0.
2372 [This option is specific to the i386 PE targeted port of the linker]
2374 @kindex --minor-subsystem-version
2375 @item --minor-subsystem-version @var{value}
2376 Sets the minor number of the ``subsystem version''. Defaults to 0.
2377 [This option is specific to the i386 PE targeted port of the linker]
2379 @cindex DEF files, creating
2380 @cindex DLLs, creating
2381 @kindex --output-def
2382 @item --output-def @var{file}
2383 The linker will create the file @var{file} which will contain a DEF
2384 file corresponding to the DLL the linker is generating. This DEF file
2385 (which should be called @code{*.def}) may be used to create an import
2386 library with @code{dlltool} or may be used as a reference to
2387 automatically or implicitly exported symbols.
2388 [This option is specific to the i386 PE targeted port of the linker]
2390 @cindex DLLs, creating
2391 @kindex --out-implib
2392 @item --out-implib @var{file}
2393 The linker will create the file @var{file} which will contain an
2394 import lib corresponding to the DLL the linker is generating. This
2395 import lib (which should be called @code{*.dll.a} or @code{*.a}
2396 may be used to link clients against the generated DLL; this behaviour
2397 makes it possible to skip a separate @code{dlltool} import library
2399 [This option is specific to the i386 PE targeted port of the linker]
2401 @kindex --enable-auto-image-base
2402 @item --enable-auto-image-base
2403 Automatically choose the image base for DLLs, unless one is specified
2404 using the @code{--image-base} argument. By using a hash generated
2405 from the dllname to create unique image bases for each DLL, in-memory
2406 collisions and relocations which can delay program execution are
2408 [This option is specific to the i386 PE targeted port of the linker]
2410 @kindex --disable-auto-image-base
2411 @item --disable-auto-image-base
2412 Do not automatically generate a unique image base. If there is no
2413 user-specified image base (@code{--image-base}) then use the platform
2415 [This option is specific to the i386 PE targeted port of the linker]
2417 @cindex DLLs, linking to
2418 @kindex --dll-search-prefix
2419 @item --dll-search-prefix @var{string}
2420 When linking dynamically to a dll without an import library,
2421 search for @code{<string><basename>.dll} in preference to
2422 @code{lib<basename>.dll}. This behaviour allows easy distinction
2423 between DLLs built for the various "subplatforms": native, cygwin,
2424 uwin, pw, etc. For instance, cygwin DLLs typically use
2425 @code{--dll-search-prefix=cyg}.
2426 [This option is specific to the i386 PE targeted port of the linker]
2428 @kindex --enable-auto-import
2429 @item --enable-auto-import
2430 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2431 DATA imports from DLLs, and create the necessary thunking symbols when
2432 building the import libraries with those DATA exports. Note: Use of the
2433 'auto-import' extension will cause the text section of the image file
2434 to be made writable. This does not conform to the PE-COFF format
2435 specification published by Microsoft.
2437 Note - use of the 'auto-import' extension will also cause read only
2438 data which would normally be placed into the .rdata section to be
2439 placed into the .data section instead. This is in order to work
2440 around a problem with consts that is described here:
2441 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2443 Using 'auto-import' generally will 'just work' -- but sometimes you may
2446 "variable '<var>' can't be auto-imported. Please read the
2447 documentation for ld's @code{--enable-auto-import} for details."
2449 This message occurs when some (sub)expression accesses an address
2450 ultimately given by the sum of two constants (Win32 import tables only
2451 allow one). Instances where this may occur include accesses to member
2452 fields of struct variables imported from a DLL, as well as using a
2453 constant index into an array variable imported from a DLL. Any
2454 multiword variable (arrays, structs, long long, etc) may trigger
2455 this error condition. However, regardless of the exact data type
2456 of the offending exported variable, ld will always detect it, issue
2457 the warning, and exit.
2459 There are several ways to address this difficulty, regardless of the
2460 data type of the exported variable:
2462 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2463 of adjusting references in your client code for runtime environment, so
2464 this method works only when runtime environment supports this feature.
2466 A second solution is to force one of the 'constants' to be a variable --
2467 that is, unknown and un-optimizable at compile time. For arrays,
2468 there are two possibilities: a) make the indexee (the array's address)
2469 a variable, or b) make the 'constant' index a variable. Thus:
2472 extern type extern_array[];
2474 @{ volatile type *t=extern_array; t[1] @}
2480 extern type extern_array[];
2482 @{ volatile int t=1; extern_array[t] @}
2485 For structs (and most other multiword data types) the only option
2486 is to make the struct itself (or the long long, or the ...) variable:
2489 extern struct s extern_struct;
2490 extern_struct.field -->
2491 @{ volatile struct s *t=&extern_struct; t->field @}
2497 extern long long extern_ll;
2499 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2502 A third method of dealing with this difficulty is to abandon
2503 'auto-import' for the offending symbol and mark it with
2504 @code{__declspec(dllimport)}. However, in practice that
2505 requires using compile-time #defines to indicate whether you are
2506 building a DLL, building client code that will link to the DLL, or
2507 merely building/linking to a static library. In making the choice
2508 between the various methods of resolving the 'direct address with
2509 constant offset' problem, you should consider typical real-world usage:
2517 void main(int argc, char **argv)@{
2518 printf("%d\n",arr[1]);
2528 void main(int argc, char **argv)@{
2529 /* This workaround is for win32 and cygwin; do not "optimize" */
2530 volatile int *parr = arr;
2531 printf("%d\n",parr[1]);
2538 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2539 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2540 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2541 #define FOO_IMPORT __declspec(dllimport)
2545 extern FOO_IMPORT int arr[];
2548 void main(int argc, char **argv)@{
2549 printf("%d\n",arr[1]);
2553 A fourth way to avoid this problem is to re-code your
2554 library to use a functional interface rather than a data interface
2555 for the offending variables (e.g. set_foo() and get_foo() accessor
2557 [This option is specific to the i386 PE targeted port of the linker]
2559 @kindex --disable-auto-import
2560 @item --disable-auto-import
2561 Do not attempt to do sophisticated linking of @code{_symbol} to
2562 @code{__imp__symbol} for DATA imports from DLLs.
2563 [This option is specific to the i386 PE targeted port of the linker]
2565 @kindex --enable-runtime-pseudo-reloc
2566 @item --enable-runtime-pseudo-reloc
2567 If your code contains expressions described in --enable-auto-import section,
2568 that is, DATA imports from DLL with non-zero offset, this switch will create
2569 a vector of 'runtime pseudo relocations' which can be used by runtime
2570 environment to adjust references to such data in your client code.
2571 [This option is specific to the i386 PE targeted port of the linker]
2573 @kindex --disable-runtime-pseudo-reloc
2574 @item --disable-runtime-pseudo-reloc
2575 Do not create pseudo relocations for non-zero offset DATA imports from
2576 DLLs. This is the default.
2577 [This option is specific to the i386 PE targeted port of the linker]
2579 @kindex --enable-extra-pe-debug
2580 @item --enable-extra-pe-debug
2581 Show additional debug info related to auto-import symbol thunking.
2582 [This option is specific to the i386 PE targeted port of the linker]
2584 @kindex --section-alignment
2585 @item --section-alignment
2586 Sets the section alignment. Sections in memory will always begin at
2587 addresses which are a multiple of this number. Defaults to 0x1000.
2588 [This option is specific to the i386 PE targeted port of the linker]
2592 @item --stack @var{reserve}
2593 @itemx --stack @var{reserve},@var{commit}
2594 Specify the number of bytes of memory to reserve (and optionally commit)
2595 to be used as stack for this program. The default is 2Mb reserved, 4K
2597 [This option is specific to the i386 PE targeted port of the linker]
2600 @item --subsystem @var{which}
2601 @itemx --subsystem @var{which}:@var{major}
2602 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2603 Specifies the subsystem under which your program will execute. The
2604 legal values for @var{which} are @code{native}, @code{windows},
2605 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2606 the subsystem version also. Numeric values are also accepted for
2608 [This option is specific to the i386 PE targeted port of the linker]
2610 The following options set flags in the @code{DllCharacteristics} field
2611 of the PE file header:
2612 [These options are specific to PE targeted ports of the linker]
2614 @kindex --dynamicbase
2616 The image base address may be relocated using address space layout
2617 randomization (ASLR). This feature was introduced with MS Windows
2618 Vista for i386 PE targets.
2620 @kindex --forceinteg
2622 Code integrity checks are enforced.
2626 The image is compatible with the Data Execution Prevention.
2627 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2629 @kindex --no-isolation
2630 @item --no-isolation
2631 Although the image understands isolation, do not isolate the image.
2635 The image does not use SEH. No SE handler may be called from
2640 Do not bind this image.
2644 The driver uses the MS Windows Driver Model.
2648 The image is Terminal Server aware.
2655 @subsection Options specific to C6X uClinux targets
2657 @c man begin OPTIONS
2659 The C6X uClinux target uses a binary format called DSBT to support shared
2660 libraries. Each shared library in the system needs to have a unique index;
2661 all executables use an index of 0.
2666 @item --dsbt-size @var{size}
2667 This option sets the number of entires in the DSBT of the current executable
2668 or shared library to @var{size}. The default is to create a table with 64
2671 @kindex --dsbt-index
2672 @item --dsbt-index @var{index}
2673 This option sets the DSBT index of the current executable or shared library
2674 to @var{index}. The default is 0, which is appropriate for generating
2675 executables. If a shared library is generated with a DSBT index of 0, the
2676 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
2678 @kindex --no-merge-exidx-entries
2679 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
2680 exidx entries in frame unwind info.
2688 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2690 @c man begin OPTIONS
2692 The 68HC11 and 68HC12 linkers support specific options to control the
2693 memory bank switching mapping and trampoline code generation.
2697 @kindex --no-trampoline
2698 @item --no-trampoline
2699 This option disables the generation of trampoline. By default a trampoline
2700 is generated for each far function which is called using a @code{jsr}
2701 instruction (this happens when a pointer to a far function is taken).
2703 @kindex --bank-window
2704 @item --bank-window @var{name}
2705 This option indicates to the linker the name of the memory region in
2706 the @samp{MEMORY} specification that describes the memory bank window.
2707 The definition of such region is then used by the linker to compute
2708 paging and addresses within the memory window.
2716 @subsection Options specific to Motorola 68K target
2718 @c man begin OPTIONS
2720 The following options are supported to control handling of GOT generation
2721 when linking for 68K targets.
2726 @item --got=@var{type}
2727 This option tells the linker which GOT generation scheme to use.
2728 @var{type} should be one of @samp{single}, @samp{negative},
2729 @samp{multigot} or @samp{target}. For more information refer to the
2730 Info entry for @file{ld}.
2739 @section Environment Variables
2741 @c man begin ENVIRONMENT
2743 You can change the behaviour of @command{ld} with the environment variables
2744 @ifclear SingleFormat
2747 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2749 @ifclear SingleFormat
2751 @cindex default input format
2752 @code{GNUTARGET} determines the input-file object format if you don't
2753 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2754 of the BFD names for an input format (@pxref{BFD}). If there is no
2755 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2756 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2757 attempts to discover the input format by examining binary input files;
2758 this method often succeeds, but there are potential ambiguities, since
2759 there is no method of ensuring that the magic number used to specify
2760 object-file formats is unique. However, the configuration procedure for
2761 BFD on each system places the conventional format for that system first
2762 in the search-list, so ambiguities are resolved in favor of convention.
2766 @cindex default emulation
2767 @cindex emulation, default
2768 @code{LDEMULATION} determines the default emulation if you don't use the
2769 @samp{-m} option. The emulation can affect various aspects of linker
2770 behaviour, particularly the default linker script. You can list the
2771 available emulations with the @samp{--verbose} or @samp{-V} options. If
2772 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2773 variable is not defined, the default emulation depends upon how the
2774 linker was configured.
2776 @kindex COLLECT_NO_DEMANGLE
2777 @cindex demangling, default
2778 Normally, the linker will default to demangling symbols. However, if
2779 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2780 default to not demangling symbols. This environment variable is used in
2781 a similar fashion by the @code{gcc} linker wrapper program. The default
2782 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2789 @chapter Linker Scripts
2792 @cindex linker scripts
2793 @cindex command files
2794 Every link is controlled by a @dfn{linker script}. This script is
2795 written in the linker command language.
2797 The main purpose of the linker script is to describe how the sections in
2798 the input files should be mapped into the output file, and to control
2799 the memory layout of the output file. Most linker scripts do nothing
2800 more than this. However, when necessary, the linker script can also
2801 direct the linker to perform many other operations, using the commands
2804 The linker always uses a linker script. If you do not supply one
2805 yourself, the linker will use a default script that is compiled into the
2806 linker executable. You can use the @samp{--verbose} command line option
2807 to display the default linker script. Certain command line options,
2808 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2810 You may supply your own linker script by using the @samp{-T} command
2811 line option. When you do this, your linker script will replace the
2812 default linker script.
2814 You may also use linker scripts implicitly by naming them as input files
2815 to the linker, as though they were files to be linked. @xref{Implicit
2819 * Basic Script Concepts:: Basic Linker Script Concepts
2820 * Script Format:: Linker Script Format
2821 * Simple Example:: Simple Linker Script Example
2822 * Simple Commands:: Simple Linker Script Commands
2823 * Assignments:: Assigning Values to Symbols
2824 * SECTIONS:: SECTIONS Command
2825 * MEMORY:: MEMORY Command
2826 * PHDRS:: PHDRS Command
2827 * VERSION:: VERSION Command
2828 * Expressions:: Expressions in Linker Scripts
2829 * Implicit Linker Scripts:: Implicit Linker Scripts
2832 @node Basic Script Concepts
2833 @section Basic Linker Script Concepts
2834 @cindex linker script concepts
2835 We need to define some basic concepts and vocabulary in order to
2836 describe the linker script language.
2838 The linker combines input files into a single output file. The output
2839 file and each input file are in a special data format known as an
2840 @dfn{object file format}. Each file is called an @dfn{object file}.
2841 The output file is often called an @dfn{executable}, but for our
2842 purposes we will also call it an object file. Each object file has,
2843 among other things, a list of @dfn{sections}. We sometimes refer to a
2844 section in an input file as an @dfn{input section}; similarly, a section
2845 in the output file is an @dfn{output section}.
2847 Each section in an object file has a name and a size. Most sections
2848 also have an associated block of data, known as the @dfn{section
2849 contents}. A section may be marked as @dfn{loadable}, which mean that
2850 the contents should be loaded into memory when the output file is run.
2851 A section with no contents may be @dfn{allocatable}, which means that an
2852 area in memory should be set aside, but nothing in particular should be
2853 loaded there (in some cases this memory must be zeroed out). A section
2854 which is neither loadable nor allocatable typically contains some sort
2855 of debugging information.
2857 Every loadable or allocatable output section has two addresses. The
2858 first is the @dfn{VMA}, or virtual memory address. This is the address
2859 the section will have when the output file is run. The second is the
2860 @dfn{LMA}, or load memory address. This is the address at which the
2861 section will be loaded. In most cases the two addresses will be the
2862 same. An example of when they might be different is when a data section
2863 is loaded into ROM, and then copied into RAM when the program starts up
2864 (this technique is often used to initialize global variables in a ROM
2865 based system). In this case the ROM address would be the LMA, and the
2866 RAM address would be the VMA.
2868 You can see the sections in an object file by using the @code{objdump}
2869 program with the @samp{-h} option.
2871 Every object file also has a list of @dfn{symbols}, known as the
2872 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2873 has a name, and each defined symbol has an address, among other
2874 information. If you compile a C or C++ program into an object file, you
2875 will get a defined symbol for every defined function and global or
2876 static variable. Every undefined function or global variable which is
2877 referenced in the input file will become an undefined symbol.
2879 You can see the symbols in an object file by using the @code{nm}
2880 program, or by using the @code{objdump} program with the @samp{-t}
2884 @section Linker Script Format
2885 @cindex linker script format
2886 Linker scripts are text files.
2888 You write a linker script as a series of commands. Each command is
2889 either a keyword, possibly followed by arguments, or an assignment to a
2890 symbol. You may separate commands using semicolons. Whitespace is
2893 Strings such as file or format names can normally be entered directly.
2894 If the file name contains a character such as a comma which would
2895 otherwise serve to separate file names, you may put the file name in
2896 double quotes. There is no way to use a double quote character in a
2899 You may include comments in linker scripts just as in C, delimited by
2900 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2903 @node Simple Example
2904 @section Simple Linker Script Example
2905 @cindex linker script example
2906 @cindex example of linker script
2907 Many linker scripts are fairly simple.
2909 The simplest possible linker script has just one command:
2910 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2911 memory layout of the output file.
2913 The @samp{SECTIONS} command is a powerful command. Here we will
2914 describe a simple use of it. Let's assume your program consists only of
2915 code, initialized data, and uninitialized data. These will be in the
2916 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2917 Let's assume further that these are the only sections which appear in
2920 For this example, let's say that the code should be loaded at address
2921 0x10000, and that the data should start at address 0x8000000. Here is a
2922 linker script which will do that:
2927 .text : @{ *(.text) @}
2929 .data : @{ *(.data) @}
2930 .bss : @{ *(.bss) @}
2934 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2935 followed by a series of symbol assignments and output section
2936 descriptions enclosed in curly braces.
2938 The first line inside the @samp{SECTIONS} command of the above example
2939 sets the value of the special symbol @samp{.}, which is the location
2940 counter. If you do not specify the address of an output section in some
2941 other way (other ways are described later), the address is set from the
2942 current value of the location counter. The location counter is then
2943 incremented by the size of the output section. At the start of the
2944 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2946 The second line defines an output section, @samp{.text}. The colon is
2947 required syntax which may be ignored for now. Within the curly braces
2948 after the output section name, you list the names of the input sections
2949 which should be placed into this output section. The @samp{*} is a
2950 wildcard which matches any file name. The expression @samp{*(.text)}
2951 means all @samp{.text} input sections in all input files.
2953 Since the location counter is @samp{0x10000} when the output section
2954 @samp{.text} is defined, the linker will set the address of the
2955 @samp{.text} section in the output file to be @samp{0x10000}.
2957 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2958 the output file. The linker will place the @samp{.data} output section
2959 at address @samp{0x8000000}. After the linker places the @samp{.data}
2960 output section, the value of the location counter will be
2961 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2962 effect is that the linker will place the @samp{.bss} output section
2963 immediately after the @samp{.data} output section in memory.
2965 The linker will ensure that each output section has the required
2966 alignment, by increasing the location counter if necessary. In this
2967 example, the specified addresses for the @samp{.text} and @samp{.data}
2968 sections will probably satisfy any alignment constraints, but the linker
2969 may have to create a small gap between the @samp{.data} and @samp{.bss}
2972 That's it! That's a simple and complete linker script.
2974 @node Simple Commands
2975 @section Simple Linker Script Commands
2976 @cindex linker script simple commands
2977 In this section we describe the simple linker script commands.
2980 * Entry Point:: Setting the entry point
2981 * File Commands:: Commands dealing with files
2982 @ifclear SingleFormat
2983 * Format Commands:: Commands dealing with object file formats
2986 * REGION_ALIAS:: Assign alias names to memory regions
2987 * Miscellaneous Commands:: Other linker script commands
2991 @subsection Setting the Entry Point
2992 @kindex ENTRY(@var{symbol})
2993 @cindex start of execution
2994 @cindex first instruction
2996 The first instruction to execute in a program is called the @dfn{entry
2997 point}. You can use the @code{ENTRY} linker script command to set the
2998 entry point. The argument is a symbol name:
3003 There are several ways to set the entry point. The linker will set the
3004 entry point by trying each of the following methods in order, and
3005 stopping when one of them succeeds:
3008 the @samp{-e} @var{entry} command-line option;
3010 the @code{ENTRY(@var{symbol})} command in a linker script;
3012 the value of a target specific symbol, if it is defined; For many
3013 targets this is @code{start}, but PE and BeOS based systems for example
3014 check a list of possible entry symbols, matching the first one found.
3016 the address of the first byte of the @samp{.text} section, if present;
3018 The address @code{0}.
3022 @subsection Commands Dealing with Files
3023 @cindex linker script file commands
3024 Several linker script commands deal with files.
3027 @item INCLUDE @var{filename}
3028 @kindex INCLUDE @var{filename}
3029 @cindex including a linker script
3030 Include the linker script @var{filename} at this point. The file will
3031 be searched for in the current directory, and in any directory specified
3032 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3035 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3036 @code{SECTIONS} commands, or in output section descriptions.
3038 @item INPUT(@var{file}, @var{file}, @dots{})
3039 @itemx INPUT(@var{file} @var{file} @dots{})
3040 @kindex INPUT(@var{files})
3041 @cindex input files in linker scripts
3042 @cindex input object files in linker scripts
3043 @cindex linker script input object files
3044 The @code{INPUT} command directs the linker to include the named files
3045 in the link, as though they were named on the command line.
3047 For example, if you always want to include @file{subr.o} any time you do
3048 a link, but you can't be bothered to put it on every link command line,
3049 then you can put @samp{INPUT (subr.o)} in your linker script.
3051 In fact, if you like, you can list all of your input files in the linker
3052 script, and then invoke the linker with nothing but a @samp{-T} option.
3054 In case a @dfn{sysroot prefix} is configured, and the filename starts
3055 with the @samp{/} character, and the script being processed was
3056 located inside the @dfn{sysroot prefix}, the filename will be looked
3057 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3058 open the file in the current directory. If it is not found, the
3059 linker will search through the archive library search path. See the
3060 description of @samp{-L} in @ref{Options,,Command Line Options}.
3062 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3063 name to @code{lib@var{file}.a}, as with the command line argument
3066 When you use the @code{INPUT} command in an implicit linker script, the
3067 files will be included in the link at the point at which the linker
3068 script file is included. This can affect archive searching.
3070 @item GROUP(@var{file}, @var{file}, @dots{})
3071 @itemx GROUP(@var{file} @var{file} @dots{})
3072 @kindex GROUP(@var{files})
3073 @cindex grouping input files
3074 The @code{GROUP} command is like @code{INPUT}, except that the named
3075 files should all be archives, and they are searched repeatedly until no
3076 new undefined references are created. See the description of @samp{-(}
3077 in @ref{Options,,Command Line Options}.
3079 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3080 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3081 @kindex AS_NEEDED(@var{files})
3082 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3083 commands, among other filenames. The files listed will be handled
3084 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3085 with the exception of ELF shared libraries, that will be added only
3086 when they are actually needed. This construct essentially enables
3087 @option{--as-needed} option for all the files listed inside of it
3088 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3091 @item OUTPUT(@var{filename})
3092 @kindex OUTPUT(@var{filename})
3093 @cindex output file name in linker script
3094 The @code{OUTPUT} command names the output file. Using
3095 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3096 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3097 Line Options}). If both are used, the command line option takes
3100 You can use the @code{OUTPUT} command to define a default name for the
3101 output file other than the usual default of @file{a.out}.
3103 @item SEARCH_DIR(@var{path})
3104 @kindex SEARCH_DIR(@var{path})
3105 @cindex library search path in linker script
3106 @cindex archive search path in linker script
3107 @cindex search path in linker script
3108 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3109 @command{ld} looks for archive libraries. Using
3110 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3111 on the command line (@pxref{Options,,Command Line Options}). If both
3112 are used, then the linker will search both paths. Paths specified using
3113 the command line option are searched first.
3115 @item STARTUP(@var{filename})
3116 @kindex STARTUP(@var{filename})
3117 @cindex first input file
3118 The @code{STARTUP} command is just like the @code{INPUT} command, except
3119 that @var{filename} will become the first input file to be linked, as
3120 though it were specified first on the command line. This may be useful
3121 when using a system in which the entry point is always the start of the
3125 @ifclear SingleFormat
3126 @node Format Commands
3127 @subsection Commands Dealing with Object File Formats
3128 A couple of linker script commands deal with object file formats.
3131 @item OUTPUT_FORMAT(@var{bfdname})
3132 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3133 @kindex OUTPUT_FORMAT(@var{bfdname})
3134 @cindex output file format in linker script
3135 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3136 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3137 exactly like using @samp{--oformat @var{bfdname}} on the command line
3138 (@pxref{Options,,Command Line Options}). If both are used, the command
3139 line option takes precedence.
3141 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3142 formats based on the @samp{-EB} and @samp{-EL} command line options.
3143 This permits the linker script to set the output format based on the
3146 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3147 will be the first argument, @var{default}. If @samp{-EB} is used, the
3148 output format will be the second argument, @var{big}. If @samp{-EL} is
3149 used, the output format will be the third argument, @var{little}.
3151 For example, the default linker script for the MIPS ELF target uses this
3154 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3156 This says that the default format for the output file is
3157 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3158 option, the output file will be created in the @samp{elf32-littlemips}
3161 @item TARGET(@var{bfdname})
3162 @kindex TARGET(@var{bfdname})
3163 @cindex input file format in linker script
3164 The @code{TARGET} command names the BFD format to use when reading input
3165 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3166 This command is like using @samp{-b @var{bfdname}} on the command line
3167 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3168 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3169 command is also used to set the format for the output file. @xref{BFD}.
3174 @subsection Assign alias names to memory regions
3175 @kindex REGION_ALIAS(@var{alias}, @var{region})
3176 @cindex region alias
3177 @cindex region names
3179 Alias names can be added to existing memory regions created with the
3180 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3183 REGION_ALIAS(@var{alias}, @var{region})
3186 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3187 memory region @var{region}. This allows a flexible mapping of output sections
3188 to memory regions. An example follows.
3190 Suppose we have an application for embedded systems which come with various
3191 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3192 that allows code execution or data storage. Some may have a read-only,
3193 non-volatile memory @code{ROM} that allows code execution and read-only data
3194 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3195 read-only data access and no code execution capability. We have four output
3200 @code{.text} program code;
3202 @code{.rodata} read-only data;
3204 @code{.data} read-write initialized data;
3206 @code{.bss} read-write zero initialized data.
3209 The goal is to provide a linker command file that contains a system independent
3210 part defining the output sections and a system dependent part mapping the
3211 output sections to the memory regions available on the system. Our embedded
3212 systems come with three different memory setups @code{A}, @code{B} and
3214 @multitable @columnfractions .25 .25 .25 .25
3215 @item Section @tab Variant A @tab Variant B @tab Variant C
3216 @item .text @tab RAM @tab ROM @tab ROM
3217 @item .rodata @tab RAM @tab ROM @tab ROM2
3218 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3219 @item .bss @tab RAM @tab RAM @tab RAM
3221 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3222 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3223 the load address of the @code{.data} section starts in all three variants at
3224 the end of the @code{.rodata} section.
3226 The base linker script that deals with the output sections follows. It
3227 includes the system dependent @code{linkcmds.memory} file that describes the
3230 INCLUDE linkcmds.memory
3243 .data : AT (rodata_end)
3248 data_size = SIZEOF(.data);
3249 data_load_start = LOADADDR(.data);
3257 Now we need three different @code{linkcmds.memory} files to define memory
3258 regions and alias names. The content of @code{linkcmds.memory} for the three
3259 variants @code{A}, @code{B} and @code{C}:
3262 Here everything goes into the @code{RAM}.
3266 RAM : ORIGIN = 0, LENGTH = 4M
3269 REGION_ALIAS("REGION_TEXT", RAM);
3270 REGION_ALIAS("REGION_RODATA", RAM);
3271 REGION_ALIAS("REGION_DATA", RAM);
3272 REGION_ALIAS("REGION_BSS", RAM);
3275 Program code and read-only data go into the @code{ROM}. Read-write data goes
3276 into the @code{RAM}. An image of the initialized data is loaded into the
3277 @code{ROM} and will be copied during system start into the @code{RAM}.
3281 ROM : ORIGIN = 0, LENGTH = 3M
3282 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3285 REGION_ALIAS("REGION_TEXT", ROM);
3286 REGION_ALIAS("REGION_RODATA", ROM);
3287 REGION_ALIAS("REGION_DATA", RAM);
3288 REGION_ALIAS("REGION_BSS", RAM);
3291 Program code goes into the @code{ROM}. Read-only data goes into the
3292 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3293 initialized data is loaded into the @code{ROM2} and will be copied during
3294 system start into the @code{RAM}.
3298 ROM : ORIGIN = 0, LENGTH = 2M
3299 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3300 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3303 REGION_ALIAS("REGION_TEXT", ROM);
3304 REGION_ALIAS("REGION_RODATA", ROM2);
3305 REGION_ALIAS("REGION_DATA", RAM);
3306 REGION_ALIAS("REGION_BSS", RAM);
3310 It is possible to write a common system initialization routine to copy the
3311 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3316 extern char data_start [];
3317 extern char data_size [];
3318 extern char data_load_start [];
3320 void copy_data(void)
3322 if (data_start != data_load_start)
3324 memcpy(data_start, data_load_start, (size_t) data_size);
3329 @node Miscellaneous Commands
3330 @subsection Other Linker Script Commands
3331 There are a few other linker scripts commands.
3334 @item ASSERT(@var{exp}, @var{message})
3336 @cindex assertion in linker script
3337 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3338 with an error code, and print @var{message}.
3340 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3342 @cindex undefined symbol in linker script
3343 Force @var{symbol} to be entered in the output file as an undefined
3344 symbol. Doing this may, for example, trigger linking of additional
3345 modules from standard libraries. You may list several @var{symbol}s for
3346 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3347 command has the same effect as the @samp{-u} command-line option.
3349 @item FORCE_COMMON_ALLOCATION
3350 @kindex FORCE_COMMON_ALLOCATION
3351 @cindex common allocation in linker script
3352 This command has the same effect as the @samp{-d} command-line option:
3353 to make @command{ld} assign space to common symbols even if a relocatable
3354 output file is specified (@samp{-r}).
3356 @item INHIBIT_COMMON_ALLOCATION
3357 @kindex INHIBIT_COMMON_ALLOCATION
3358 @cindex common allocation in linker script
3359 This command has the same effect as the @samp{--no-define-common}
3360 command-line option: to make @code{ld} omit the assignment of addresses
3361 to common symbols even for a non-relocatable output file.
3363 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3365 @cindex insert user script into default script
3366 This command is typically used in a script specified by @samp{-T} to
3367 augment the default @code{SECTIONS} with, for example, overlays. It
3368 inserts all prior linker script statements after (or before)
3369 @var{output_section}, and also causes @samp{-T} to not override the
3370 default linker script. The exact insertion point is as for orphan
3371 sections. @xref{Location Counter}. The insertion happens after the
3372 linker has mapped input sections to output sections. Prior to the
3373 insertion, since @samp{-T} scripts are parsed before the default
3374 linker script, statements in the @samp{-T} script occur before the
3375 default linker script statements in the internal linker representation
3376 of the script. In particular, input section assignments will be made
3377 to @samp{-T} output sections before those in the default script. Here
3378 is an example of how a @samp{-T} script using @code{INSERT} might look:
3385 .ov1 @{ ov1*(.text) @}
3386 .ov2 @{ ov2*(.text) @}
3392 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3393 @kindex NOCROSSREFS(@var{sections})
3394 @cindex cross references
3395 This command may be used to tell @command{ld} to issue an error about any
3396 references among certain output sections.
3398 In certain types of programs, particularly on embedded systems when
3399 using overlays, when one section is loaded into memory, another section
3400 will not be. Any direct references between the two sections would be
3401 errors. For example, it would be an error if code in one section called
3402 a function defined in the other section.
3404 The @code{NOCROSSREFS} command takes a list of output section names. If
3405 @command{ld} detects any cross references between the sections, it reports
3406 an error and returns a non-zero exit status. Note that the
3407 @code{NOCROSSREFS} command uses output section names, not input section
3410 @ifclear SingleFormat
3411 @item OUTPUT_ARCH(@var{bfdarch})
3412 @kindex OUTPUT_ARCH(@var{bfdarch})
3413 @cindex machine architecture
3414 @cindex architecture
3415 Specify a particular output machine architecture. The argument is one
3416 of the names used by the BFD library (@pxref{BFD}). You can see the
3417 architecture of an object file by using the @code{objdump} program with
3418 the @samp{-f} option.
3421 @item LD_FEATURE(@var{string})
3422 @kindex LD_FEATURE(@var{string})
3423 This command may be used to modify @command{ld} behavior. If
3424 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3425 in a script are simply treated as numbers everywhere.
3426 @xref{Expression Section}.
3430 @section Assigning Values to Symbols
3431 @cindex assignment in scripts
3432 @cindex symbol definition, scripts
3433 @cindex variables, defining
3434 You may assign a value to a symbol in a linker script. This will define
3435 the symbol and place it into the symbol table with a global scope.
3438 * Simple Assignments:: Simple Assignments
3441 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3442 * Source Code Reference:: How to use a linker script defined symbol in source code
3445 @node Simple Assignments
3446 @subsection Simple Assignments
3448 You may assign to a symbol using any of the C assignment operators:
3451 @item @var{symbol} = @var{expression} ;
3452 @itemx @var{symbol} += @var{expression} ;
3453 @itemx @var{symbol} -= @var{expression} ;
3454 @itemx @var{symbol} *= @var{expression} ;
3455 @itemx @var{symbol} /= @var{expression} ;
3456 @itemx @var{symbol} <<= @var{expression} ;
3457 @itemx @var{symbol} >>= @var{expression} ;
3458 @itemx @var{symbol} &= @var{expression} ;
3459 @itemx @var{symbol} |= @var{expression} ;
3462 The first case will define @var{symbol} to the value of
3463 @var{expression}. In the other cases, @var{symbol} must already be
3464 defined, and the value will be adjusted accordingly.
3466 The special symbol name @samp{.} indicates the location counter. You
3467 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3469 The semicolon after @var{expression} is required.
3471 Expressions are defined below; see @ref{Expressions}.
3473 You may write symbol assignments as commands in their own right, or as
3474 statements within a @code{SECTIONS} command, or as part of an output
3475 section description in a @code{SECTIONS} command.
3477 The section of the symbol will be set from the section of the
3478 expression; for more information, see @ref{Expression Section}.
3480 Here is an example showing the three different places that symbol
3481 assignments may be used:
3492 _bdata = (. + 3) & ~ 3;
3493 .data : @{ *(.data) @}
3497 In this example, the symbol @samp{floating_point} will be defined as
3498 zero. The symbol @samp{_etext} will be defined as the address following
3499 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3500 defined as the address following the @samp{.text} output section aligned
3501 upward to a 4 byte boundary.
3506 For ELF targeted ports, define a symbol that will be hidden and won't be
3507 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
3509 Here is the example from @ref{Simple Assignments}, rewritten to use
3513 HIDDEN(floating_point = 0);
3521 HIDDEN(_bdata = (. + 3) & ~ 3);
3522 .data : @{ *(.data) @}
3526 In this case none of the three symbols will be visible outside this module.
3531 In some cases, it is desirable for a linker script to define a symbol
3532 only if it is referenced and is not defined by any object included in
3533 the link. For example, traditional linkers defined the symbol
3534 @samp{etext}. However, ANSI C requires that the user be able to use
3535 @samp{etext} as a function name without encountering an error. The
3536 @code{PROVIDE} keyword may be used to define a symbol, such as
3537 @samp{etext}, only if it is referenced but not defined. The syntax is
3538 @code{PROVIDE(@var{symbol} = @var{expression})}.
3540 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3553 In this example, if the program defines @samp{_etext} (with a leading
3554 underscore), the linker will give a multiple definition error. If, on
3555 the other hand, the program defines @samp{etext} (with no leading
3556 underscore), the linker will silently use the definition in the program.
3557 If the program references @samp{etext} but does not define it, the
3558 linker will use the definition in the linker script.
3560 @node PROVIDE_HIDDEN
3561 @subsection PROVIDE_HIDDEN
3562 @cindex PROVIDE_HIDDEN
3563 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3564 hidden and won't be exported.
3566 @node Source Code Reference
3567 @subsection Source Code Reference
3569 Accessing a linker script defined variable from source code is not
3570 intuitive. In particular a linker script symbol is not equivalent to
3571 a variable declaration in a high level language, it is instead a
3572 symbol that does not have a value.
3574 Before going further, it is important to note that compilers often
3575 transform names in the source code into different names when they are
3576 stored in the symbol table. For example, Fortran compilers commonly
3577 prepend or append an underscore, and C++ performs extensive @samp{name
3578 mangling}. Therefore there might be a discrepancy between the name
3579 of a variable as it is used in source code and the name of the same
3580 variable as it is defined in a linker script. For example in C a
3581 linker script variable might be referred to as:
3587 But in the linker script it might be defined as:
3593 In the remaining examples however it is assumed that no name
3594 transformation has taken place.
3596 When a symbol is declared in a high level language such as C, two
3597 things happen. The first is that the compiler reserves enough space
3598 in the program's memory to hold the @emph{value} of the symbol. The
3599 second is that the compiler creates an entry in the program's symbol
3600 table which holds the symbol's @emph{address}. ie the symbol table
3601 contains the address of the block of memory holding the symbol's
3602 value. So for example the following C declaration, at file scope:
3608 creates a entry called @samp{foo} in the symbol table. This entry
3609 holds the address of an @samp{int} sized block of memory where the
3610 number 1000 is initially stored.
3612 When a program references a symbol the compiler generates code that
3613 first accesses the symbol table to find the address of the symbol's
3614 memory block and then code to read the value from that memory block.
3621 looks up the symbol @samp{foo} in the symbol table, gets the address
3622 associated with this symbol and then writes the value 1 into that
3629 looks up the symbol @samp{foo} in the symbol table, gets it address
3630 and then copies this address into the block of memory associated with
3631 the variable @samp{a}.
3633 Linker scripts symbol declarations, by contrast, create an entry in
3634 the symbol table but do not assign any memory to them. Thus they are
3635 an address without a value. So for example the linker script definition:
3641 creates an entry in the symbol table called @samp{foo} which holds
3642 the address of memory location 1000, but nothing special is stored at
3643 address 1000. This means that you cannot access the @emph{value} of a
3644 linker script defined symbol - it has no value - all you can do is
3645 access the @emph{address} of a linker script defined symbol.
3647 Hence when you are using a linker script defined symbol in source code
3648 you should always take the address of the symbol, and never attempt to
3649 use its value. For example suppose you want to copy the contents of a
3650 section of memory called .ROM into a section called .FLASH and the
3651 linker script contains these declarations:
3655 start_of_ROM = .ROM;
3656 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3657 start_of_FLASH = .FLASH;
3661 Then the C source code to perform the copy would be:
3665 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3667 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3671 Note the use of the @samp{&} operators. These are correct.
3674 @section SECTIONS Command
3676 The @code{SECTIONS} command tells the linker how to map input sections
3677 into output sections, and how to place the output sections in memory.
3679 The format of the @code{SECTIONS} command is:
3683 @var{sections-command}
3684 @var{sections-command}
3689 Each @var{sections-command} may of be one of the following:
3693 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3695 a symbol assignment (@pxref{Assignments})
3697 an output section description
3699 an overlay description
3702 The @code{ENTRY} command and symbol assignments are permitted inside the
3703 @code{SECTIONS} command for convenience in using the location counter in
3704 those commands. This can also make the linker script easier to
3705 understand because you can use those commands at meaningful points in
3706 the layout of the output file.
3708 Output section descriptions and overlay descriptions are described
3711 If you do not use a @code{SECTIONS} command in your linker script, the
3712 linker will place each input section into an identically named output
3713 section in the order that the sections are first encountered in the
3714 input files. If all input sections are present in the first file, for
3715 example, the order of sections in the output file will match the order
3716 in the first input file. The first section will be at address zero.
3719 * Output Section Description:: Output section description
3720 * Output Section Name:: Output section name
3721 * Output Section Address:: Output section address
3722 * Input Section:: Input section description
3723 * Output Section Data:: Output section data
3724 * Output Section Keywords:: Output section keywords
3725 * Output Section Discarding:: Output section discarding
3726 * Output Section Attributes:: Output section attributes
3727 * Overlay Description:: Overlay description
3730 @node Output Section Description
3731 @subsection Output Section Description
3732 The full description of an output section looks like this:
3735 @var{section} [@var{address}] [(@var{type})] :
3737 [ALIGN(@var{section_align})]
3738 [SUBALIGN(@var{subsection_align})]
3741 @var{output-section-command}
3742 @var{output-section-command}
3744 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3748 Most output sections do not use most of the optional section attributes.
3750 The whitespace around @var{section} is required, so that the section
3751 name is unambiguous. The colon and the curly braces are also required.
3752 The line breaks and other white space are optional.
3754 Each @var{output-section-command} may be one of the following:
3758 a symbol assignment (@pxref{Assignments})
3760 an input section description (@pxref{Input Section})
3762 data values to include directly (@pxref{Output Section Data})
3764 a special output section keyword (@pxref{Output Section Keywords})
3767 @node Output Section Name
3768 @subsection Output Section Name
3769 @cindex name, section
3770 @cindex section name
3771 The name of the output section is @var{section}. @var{section} must
3772 meet the constraints of your output format. In formats which only
3773 support a limited number of sections, such as @code{a.out}, the name
3774 must be one of the names supported by the format (@code{a.out}, for
3775 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3776 output format supports any number of sections, but with numbers and not
3777 names (as is the case for Oasys), the name should be supplied as a
3778 quoted numeric string. A section name may consist of any sequence of
3779 characters, but a name which contains any unusual characters such as
3780 commas must be quoted.
3782 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3785 @node Output Section Address
3786 @subsection Output Section Address
3787 @cindex address, section
3788 @cindex section address
3789 The @var{address} is an expression for the VMA (the virtual memory
3790 address) of the output section. This address is optional, but if it
3791 is provided then the output address will be set exactly as specified.
3793 If the output address is not specified then one will be chosen for the
3794 section, based on the heuristic below. This address will be adjusted
3795 to fit the alignment requirement of the output section. The
3796 alignment requirement is the strictest alignment of any input section
3797 contained within the output section.
3799 The output section address heuristic is as follows:
3803 If an output memory @var{region} is set for the section then it
3804 is added to this region and its address will be the next free address
3808 If the MEMORY command has been used to create a list of memory
3809 regions then the first region which has attributes compatible with the
3810 section is selected to contain it. The section's output address will
3811 be the next free address in that region; @ref{MEMORY}.
3814 If no memory regions were specified, or none match the section then
3815 the output address will be based on the current value of the location
3823 .text . : @{ *(.text) @}
3830 .text : @{ *(.text) @}
3834 are subtly different. The first will set the address of the
3835 @samp{.text} output section to the current value of the location
3836 counter. The second will set it to the current value of the location
3837 counter aligned to the strictest alignment of any of the @samp{.text}
3840 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3841 For example, if you want to align the section on a 0x10 byte boundary,
3842 so that the lowest four bits of the section address are zero, you could
3843 do something like this:
3845 .text ALIGN(0x10) : @{ *(.text) @}
3848 This works because @code{ALIGN} returns the current location counter
3849 aligned upward to the specified value.
3851 Specifying @var{address} for a section will change the value of the
3852 location counter, provided that the section is non-empty. (Empty
3853 sections are ignored).
3856 @subsection Input Section Description
3857 @cindex input sections
3858 @cindex mapping input sections to output sections
3859 The most common output section command is an input section description.
3861 The input section description is the most basic linker script operation.
3862 You use output sections to tell the linker how to lay out your program
3863 in memory. You use input section descriptions to tell the linker how to
3864 map the input files into your memory layout.
3867 * Input Section Basics:: Input section basics
3868 * Input Section Wildcards:: Input section wildcard patterns
3869 * Input Section Common:: Input section for common symbols
3870 * Input Section Keep:: Input section and garbage collection
3871 * Input Section Example:: Input section example
3874 @node Input Section Basics
3875 @subsubsection Input Section Basics
3876 @cindex input section basics
3877 An input section description consists of a file name optionally followed
3878 by a list of section names in parentheses.
3880 The file name and the section name may be wildcard patterns, which we
3881 describe further below (@pxref{Input Section Wildcards}).
3883 The most common input section description is to include all input
3884 sections with a particular name in the output section. For example, to
3885 include all input @samp{.text} sections, you would write:
3890 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3891 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3892 match all files except the ones specified in the EXCLUDE_FILE list. For
3895 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3897 will cause all .ctors sections from all files except @file{crtend.o} and
3898 @file{otherfile.o} to be included.
3900 There are two ways to include more than one section:
3906 The difference between these is the order in which the @samp{.text} and
3907 @samp{.rdata} input sections will appear in the output section. In the
3908 first example, they will be intermingled, appearing in the same order as
3909 they are found in the linker input. In the second example, all
3910 @samp{.text} input sections will appear first, followed by all
3911 @samp{.rdata} input sections.
3913 You can specify a file name to include sections from a particular file.
3914 You would do this if one or more of your files contain special data that
3915 needs to be at a particular location in memory. For example:
3920 To refine the sections that are included based on the section flags
3921 of an input section, INPUT_SECTION_FLAGS may be used.
3923 Here is a simple example for using Section header flags for ELF sections:
3928 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
3929 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
3934 In this example, the output section @samp{.text} will be comprised of any
3935 input section matching the name *(.text) whose section header flags
3936 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
3937 @samp{.text2} will be comprised of any input section matching the name *(.text)
3938 whose section header flag @code{SHF_WRITE} is clear.
3940 You can also specify files within archives by writing a pattern
3941 matching the archive, a colon, then the pattern matching the file,
3942 with no whitespace around the colon.
3946 matches file within archive
3948 matches the whole archive
3950 matches file but not one in an archive
3953 Either one or both of @samp{archive} and @samp{file} can contain shell
3954 wildcards. On DOS based file systems, the linker will assume that a
3955 single letter followed by a colon is a drive specifier, so
3956 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
3957 within an archive called @samp{c}. @samp{archive:file} filespecs may
3958 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
3959 other linker script contexts. For instance, you cannot extract a file
3960 from an archive by using @samp{archive:file} in an @code{INPUT}
3963 If you use a file name without a list of sections, then all sections in
3964 the input file will be included in the output section. This is not
3965 commonly done, but it may by useful on occasion. For example:
3970 When you use a file name which is not an @samp{archive:file} specifier
3971 and does not contain any wild card
3972 characters, the linker will first see if you also specified the file
3973 name on the linker command line or in an @code{INPUT} command. If you
3974 did not, the linker will attempt to open the file as an input file, as
3975 though it appeared on the command line. Note that this differs from an
3976 @code{INPUT} command, because the linker will not search for the file in
3977 the archive search path.
3979 @node Input Section Wildcards
3980 @subsubsection Input Section Wildcard Patterns
3981 @cindex input section wildcards
3982 @cindex wildcard file name patterns
3983 @cindex file name wildcard patterns
3984 @cindex section name wildcard patterns
3985 In an input section description, either the file name or the section
3986 name or both may be wildcard patterns.
3988 The file name of @samp{*} seen in many examples is a simple wildcard
3989 pattern for the file name.
3991 The wildcard patterns are like those used by the Unix shell.
3995 matches any number of characters
3997 matches any single character
3999 matches a single instance of any of the @var{chars}; the @samp{-}
4000 character may be used to specify a range of characters, as in
4001 @samp{[a-z]} to match any lower case letter
4003 quotes the following character
4006 When a file name is matched with a wildcard, the wildcard characters
4007 will not match a @samp{/} character (used to separate directory names on
4008 Unix). A pattern consisting of a single @samp{*} character is an
4009 exception; it will always match any file name, whether it contains a
4010 @samp{/} or not. In a section name, the wildcard characters will match
4011 a @samp{/} character.
4013 File name wildcard patterns only match files which are explicitly
4014 specified on the command line or in an @code{INPUT} command. The linker
4015 does not search directories to expand wildcards.
4017 If a file name matches more than one wildcard pattern, or if a file name
4018 appears explicitly and is also matched by a wildcard pattern, the linker
4019 will use the first match in the linker script. For example, this
4020 sequence of input section descriptions is probably in error, because the
4021 @file{data.o} rule will not be used:
4023 .data : @{ *(.data) @}
4024 .data1 : @{ data.o(.data) @}
4027 @cindex SORT_BY_NAME
4028 Normally, the linker will place files and sections matched by wildcards
4029 in the order in which they are seen during the link. You can change
4030 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4031 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4032 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4033 into ascending order by name before placing them in the output file.
4035 @cindex SORT_BY_ALIGNMENT
4036 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4037 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4038 ascending order by alignment before placing them in the output file.
4040 @cindex SORT_BY_INIT_PRIORITY
4041 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4042 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4043 ascending order by numerical value of the GCC init_priority attribute
4044 encoded in the section name before placing them in the output file.
4047 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4049 When there are nested section sorting commands in linker script, there
4050 can be at most 1 level of nesting for section sorting commands.
4054 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4055 It will sort the input sections by name first, then by alignment if 2
4056 sections have the same name.
4058 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4059 It will sort the input sections by alignment first, then by name if 2
4060 sections have the same alignment.
4062 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4063 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4065 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4066 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4068 All other nested section sorting commands are invalid.
4071 When both command line section sorting option and linker script
4072 section sorting command are used, section sorting command always
4073 takes precedence over the command line option.
4075 If the section sorting command in linker script isn't nested, the
4076 command line option will make the section sorting command to be
4077 treated as nested sorting command.
4081 @code{SORT_BY_NAME} (wildcard section pattern ) with
4082 @option{--sort-sections alignment} is equivalent to
4083 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4085 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4086 @option{--sort-section name} is equivalent to
4087 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4090 If the section sorting command in linker script is nested, the
4091 command line option will be ignored.
4094 @code{SORT_NONE} disables section sorting by ignoring the command line
4095 section sorting option.
4097 If you ever get confused about where input sections are going, use the
4098 @samp{-M} linker option to generate a map file. The map file shows
4099 precisely how input sections are mapped to output sections.
4101 This example shows how wildcard patterns might be used to partition
4102 files. This linker script directs the linker to place all @samp{.text}
4103 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4104 The linker will place the @samp{.data} section from all files beginning
4105 with an upper case character in @samp{.DATA}; for all other files, the
4106 linker will place the @samp{.data} section in @samp{.data}.
4110 .text : @{ *(.text) @}
4111 .DATA : @{ [A-Z]*(.data) @}
4112 .data : @{ *(.data) @}
4113 .bss : @{ *(.bss) @}
4118 @node Input Section Common
4119 @subsubsection Input Section for Common Symbols
4120 @cindex common symbol placement
4121 @cindex uninitialized data placement
4122 A special notation is needed for common symbols, because in many object
4123 file formats common symbols do not have a particular input section. The
4124 linker treats common symbols as though they are in an input section
4125 named @samp{COMMON}.
4127 You may use file names with the @samp{COMMON} section just as with any
4128 other input sections. You can use this to place common symbols from a
4129 particular input file in one section while common symbols from other
4130 input files are placed in another section.
4132 In most cases, common symbols in input files will be placed in the
4133 @samp{.bss} section in the output file. For example:
4135 .bss @{ *(.bss) *(COMMON) @}
4138 @cindex scommon section
4139 @cindex small common symbols
4140 Some object file formats have more than one type of common symbol. For
4141 example, the MIPS ELF object file format distinguishes standard common
4142 symbols and small common symbols. In this case, the linker will use a
4143 different special section name for other types of common symbols. In
4144 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4145 symbols and @samp{.scommon} for small common symbols. This permits you
4146 to map the different types of common symbols into memory at different
4150 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4151 notation is now considered obsolete. It is equivalent to
4154 @node Input Section Keep
4155 @subsubsection Input Section and Garbage Collection
4157 @cindex garbage collection
4158 When link-time garbage collection is in use (@samp{--gc-sections}),
4159 it is often useful to mark sections that should not be eliminated.
4160 This is accomplished by surrounding an input section's wildcard entry
4161 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4162 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4164 @node Input Section Example
4165 @subsubsection Input Section Example
4166 The following example is a complete linker script. It tells the linker
4167 to read all of the sections from file @file{all.o} and place them at the
4168 start of output section @samp{outputa} which starts at location
4169 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4170 follows immediately, in the same output section. All of section
4171 @samp{.input2} from @file{foo.o} goes into output section
4172 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4173 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4174 files are written to output section @samp{outputc}.
4202 @node Output Section Data
4203 @subsection Output Section Data
4205 @cindex section data
4206 @cindex output section data
4207 @kindex BYTE(@var{expression})
4208 @kindex SHORT(@var{expression})
4209 @kindex LONG(@var{expression})
4210 @kindex QUAD(@var{expression})
4211 @kindex SQUAD(@var{expression})
4212 You can include explicit bytes of data in an output section by using
4213 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4214 an output section command. Each keyword is followed by an expression in
4215 parentheses providing the value to store (@pxref{Expressions}). The
4216 value of the expression is stored at the current value of the location
4219 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4220 store one, two, four, and eight bytes (respectively). After storing the
4221 bytes, the location counter is incremented by the number of bytes
4224 For example, this will store the byte 1 followed by the four byte value
4225 of the symbol @samp{addr}:
4231 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4232 same; they both store an 8 byte, or 64 bit, value. When both host and
4233 target are 32 bits, an expression is computed as 32 bits. In this case
4234 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4235 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4237 If the object file format of the output file has an explicit endianness,
4238 which is the normal case, the value will be stored in that endianness.
4239 When the object file format does not have an explicit endianness, as is
4240 true of, for example, S-records, the value will be stored in the
4241 endianness of the first input object file.
4243 Note---these commands only work inside a section description and not
4244 between them, so the following will produce an error from the linker:
4246 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4248 whereas this will work:
4250 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4253 @kindex FILL(@var{expression})
4254 @cindex holes, filling
4255 @cindex unspecified memory
4256 You may use the @code{FILL} command to set the fill pattern for the
4257 current section. It is followed by an expression in parentheses. Any
4258 otherwise unspecified regions of memory within the section (for example,
4259 gaps left due to the required alignment of input sections) are filled
4260 with the value of the expression, repeated as
4261 necessary. A @code{FILL} statement covers memory locations after the
4262 point at which it occurs in the section definition; by including more
4263 than one @code{FILL} statement, you can have different fill patterns in
4264 different parts of an output section.
4266 This example shows how to fill unspecified regions of memory with the
4272 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4273 section attribute, but it only affects the
4274 part of the section following the @code{FILL} command, rather than the
4275 entire section. If both are used, the @code{FILL} command takes
4276 precedence. @xref{Output Section Fill}, for details on the fill
4279 @node Output Section Keywords
4280 @subsection Output Section Keywords
4281 There are a couple of keywords which can appear as output section
4285 @kindex CREATE_OBJECT_SYMBOLS
4286 @cindex input filename symbols
4287 @cindex filename symbols
4288 @item CREATE_OBJECT_SYMBOLS
4289 The command tells the linker to create a symbol for each input file.
4290 The name of each symbol will be the name of the corresponding input
4291 file. The section of each symbol will be the output section in which
4292 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4294 This is conventional for the a.out object file format. It is not
4295 normally used for any other object file format.
4297 @kindex CONSTRUCTORS
4298 @cindex C++ constructors, arranging in link
4299 @cindex constructors, arranging in link
4301 When linking using the a.out object file format, the linker uses an
4302 unusual set construct to support C++ global constructors and
4303 destructors. When linking object file formats which do not support
4304 arbitrary sections, such as ECOFF and XCOFF, the linker will
4305 automatically recognize C++ global constructors and destructors by name.
4306 For these object file formats, the @code{CONSTRUCTORS} command tells the
4307 linker to place constructor information in the output section where the
4308 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4309 ignored for other object file formats.
4311 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4312 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4313 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4314 the start and end of the global destructors. The
4315 first word in the list is the number of entries, followed by the address
4316 of each constructor or destructor, followed by a zero word. The
4317 compiler must arrange to actually run the code. For these object file
4318 formats @sc{gnu} C++ normally calls constructors from a subroutine
4319 @code{__main}; a call to @code{__main} is automatically inserted into
4320 the startup code for @code{main}. @sc{gnu} C++ normally runs
4321 destructors either by using @code{atexit}, or directly from the function
4324 For object file formats such as @code{COFF} or @code{ELF} which support
4325 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4326 addresses of global constructors and destructors into the @code{.ctors}
4327 and @code{.dtors} sections. Placing the following sequence into your
4328 linker script will build the sort of table which the @sc{gnu} C++
4329 runtime code expects to see.
4333 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4338 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4344 If you are using the @sc{gnu} C++ support for initialization priority,
4345 which provides some control over the order in which global constructors
4346 are run, you must sort the constructors at link time to ensure that they
4347 are executed in the correct order. When using the @code{CONSTRUCTORS}
4348 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4349 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4350 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4353 Normally the compiler and linker will handle these issues automatically,
4354 and you will not need to concern yourself with them. However, you may
4355 need to consider this if you are using C++ and writing your own linker
4360 @node Output Section Discarding
4361 @subsection Output Section Discarding
4362 @cindex discarding sections
4363 @cindex sections, discarding
4364 @cindex removing sections
4365 The linker will not create output sections with no contents. This is
4366 for convenience when referring to input sections that may or may not
4367 be present in any of the input files. For example:
4369 .foo : @{ *(.foo) @}
4372 will only create a @samp{.foo} section in the output file if there is a
4373 @samp{.foo} section in at least one input file, and if the input
4374 sections are not all empty. Other link script directives that allocate
4375 space in an output section will also create the output section.
4377 The linker will ignore address assignments (@pxref{Output Section Address})
4378 on discarded output sections, except when the linker script defines
4379 symbols in the output section. In that case the linker will obey
4380 the address assignments, possibly advancing dot even though the
4381 section is discarded.
4384 The special output section name @samp{/DISCARD/} may be used to discard
4385 input sections. Any input sections which are assigned to an output
4386 section named @samp{/DISCARD/} are not included in the output file.
4388 @node Output Section Attributes
4389 @subsection Output Section Attributes
4390 @cindex output section attributes
4391 We showed above that the full description of an output section looked
4396 @var{section} [@var{address}] [(@var{type})] :
4398 [ALIGN(@var{section_align})]
4399 [SUBALIGN(@var{subsection_align})]
4402 @var{output-section-command}
4403 @var{output-section-command}
4405 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4409 We've already described @var{section}, @var{address}, and
4410 @var{output-section-command}. In this section we will describe the
4411 remaining section attributes.
4414 * Output Section Type:: Output section type
4415 * Output Section LMA:: Output section LMA
4416 * Forced Output Alignment:: Forced Output Alignment
4417 * Forced Input Alignment:: Forced Input Alignment
4418 * Output Section Constraint:: Output section constraint
4419 * Output Section Region:: Output section region
4420 * Output Section Phdr:: Output section phdr
4421 * Output Section Fill:: Output section fill
4424 @node Output Section Type
4425 @subsubsection Output Section Type
4426 Each output section may have a type. The type is a keyword in
4427 parentheses. The following types are defined:
4431 The section should be marked as not loadable, so that it will not be
4432 loaded into memory when the program is run.
4437 These type names are supported for backward compatibility, and are
4438 rarely used. They all have the same effect: the section should be
4439 marked as not allocatable, so that no memory is allocated for the
4440 section when the program is run.
4444 @cindex prevent unnecessary loading
4445 @cindex loading, preventing
4446 The linker normally sets the attributes of an output section based on
4447 the input sections which map into it. You can override this by using
4448 the section type. For example, in the script sample below, the
4449 @samp{ROM} section is addressed at memory location @samp{0} and does not
4450 need to be loaded when the program is run.
4454 ROM 0 (NOLOAD) : @{ @dots{} @}
4460 @node Output Section LMA
4461 @subsubsection Output Section LMA
4462 @kindex AT>@var{lma_region}
4463 @kindex AT(@var{lma})
4464 @cindex load address
4465 @cindex section load address
4466 Every section has a virtual address (VMA) and a load address (LMA); see
4467 @ref{Basic Script Concepts}. The virtual address is specified by the
4468 @pxref{Output Section Address} described earlier. The load address is
4469 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
4470 address is optional.
4472 The @code{AT} keyword takes an expression as an argument. This
4473 specifies the exact load address of the section. The @code{AT>} keyword
4474 takes the name of a memory region as an argument. @xref{MEMORY}. The
4475 load address of the section is set to the next free address in the
4476 region, aligned to the section's alignment requirements.
4478 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4479 section, the linker will use the following heuristic to determine the
4484 If the section has a specific VMA address, then this is used as
4485 the LMA address as well.
4488 If the section is not allocatable then its LMA is set to its VMA.
4491 Otherwise if a memory region can be found that is compatible
4492 with the current section, and this region contains at least one
4493 section, then the LMA is set so the difference between the
4494 VMA and LMA is the same as the difference between the VMA and LMA of
4495 the last section in the located region.
4498 If no memory regions have been declared then a default region
4499 that covers the entire address space is used in the previous step.
4502 If no suitable region could be found, or there was no previous
4503 section then the LMA is set equal to the VMA.
4506 @cindex ROM initialized data
4507 @cindex initialized data in ROM
4508 This feature is designed to make it easy to build a ROM image. For
4509 example, the following linker script creates three output sections: one
4510 called @samp{.text}, which starts at @code{0x1000}, one called
4511 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
4512 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
4513 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
4514 defined with the value @code{0x2000}, which shows that the location
4515 counter holds the VMA value, not the LMA value.
4521 .text 0x1000 : @{ *(.text) _etext = . ; @}
4523 AT ( ADDR (.text) + SIZEOF (.text) )
4524 @{ _data = . ; *(.data); _edata = . ; @}
4526 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
4531 The run-time initialization code for use with a program generated with
4532 this linker script would include something like the following, to copy
4533 the initialized data from the ROM image to its runtime address. Notice
4534 how this code takes advantage of the symbols defined by the linker
4539 extern char _etext, _data, _edata, _bstart, _bend;
4540 char *src = &_etext;
4543 /* ROM has data at end of text; copy it. */
4544 while (dst < &_edata)
4548 for (dst = &_bstart; dst< &_bend; dst++)
4553 @node Forced Output Alignment
4554 @subsubsection Forced Output Alignment
4555 @kindex ALIGN(@var{section_align})
4556 @cindex forcing output section alignment
4557 @cindex output section alignment
4558 You can increase an output section's alignment by using ALIGN.
4560 @node Forced Input Alignment
4561 @subsubsection Forced Input Alignment
4562 @kindex SUBALIGN(@var{subsection_align})
4563 @cindex forcing input section alignment
4564 @cindex input section alignment
4565 You can force input section alignment within an output section by using
4566 SUBALIGN. The value specified overrides any alignment given by input
4567 sections, whether larger or smaller.
4569 @node Output Section Constraint
4570 @subsubsection Output Section Constraint
4573 @cindex constraints on output sections
4574 You can specify that an output section should only be created if all
4575 of its input sections are read-only or all of its input sections are
4576 read-write by using the keyword @code{ONLY_IF_RO} and
4577 @code{ONLY_IF_RW} respectively.
4579 @node Output Section Region
4580 @subsubsection Output Section Region
4581 @kindex >@var{region}
4582 @cindex section, assigning to memory region
4583 @cindex memory regions and sections
4584 You can assign a section to a previously defined region of memory by
4585 using @samp{>@var{region}}. @xref{MEMORY}.
4587 Here is a simple example:
4590 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4591 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4595 @node Output Section Phdr
4596 @subsubsection Output Section Phdr
4598 @cindex section, assigning to program header
4599 @cindex program headers and sections
4600 You can assign a section to a previously defined program segment by
4601 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4602 one or more segments, then all subsequent allocated sections will be
4603 assigned to those segments as well, unless they use an explicitly
4604 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4605 linker to not put the section in any segment at all.
4607 Here is a simple example:
4610 PHDRS @{ text PT_LOAD ; @}
4611 SECTIONS @{ .text : @{ *(.text) @} :text @}
4615 @node Output Section Fill
4616 @subsubsection Output Section Fill
4617 @kindex =@var{fillexp}
4618 @cindex section fill pattern
4619 @cindex fill pattern, entire section
4620 You can set the fill pattern for an entire section by using
4621 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4622 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4623 within the output section (for example, gaps left due to the required
4624 alignment of input sections) will be filled with the value, repeated as
4625 necessary. If the fill expression is a simple hex number, ie. a string
4626 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4627 an arbitrarily long sequence of hex digits can be used to specify the
4628 fill pattern; Leading zeros become part of the pattern too. For all
4629 other cases, including extra parentheses or a unary @code{+}, the fill
4630 pattern is the four least significant bytes of the value of the
4631 expression. In all cases, the number is big-endian.
4633 You can also change the fill value with a @code{FILL} command in the
4634 output section commands; (@pxref{Output Section Data}).
4636 Here is a simple example:
4639 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4643 @node Overlay Description
4644 @subsection Overlay Description
4647 An overlay description provides an easy way to describe sections which
4648 are to be loaded as part of a single memory image but are to be run at
4649 the same memory address. At run time, some sort of overlay manager will
4650 copy the overlaid sections in and out of the runtime memory address as
4651 required, perhaps by simply manipulating addressing bits. This approach
4652 can be useful, for example, when a certain region of memory is faster
4655 Overlays are described using the @code{OVERLAY} command. The
4656 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4657 output section description. The full syntax of the @code{OVERLAY}
4658 command is as follows:
4661 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4665 @var{output-section-command}
4666 @var{output-section-command}
4668 @} [:@var{phdr}@dots{}] [=@var{fill}]
4671 @var{output-section-command}
4672 @var{output-section-command}
4674 @} [:@var{phdr}@dots{}] [=@var{fill}]
4676 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4680 Everything is optional except @code{OVERLAY} (a keyword), and each
4681 section must have a name (@var{secname1} and @var{secname2} above). The
4682 section definitions within the @code{OVERLAY} construct are identical to
4683 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
4684 except that no addresses and no memory regions may be defined for
4685 sections within an @code{OVERLAY}.
4687 The sections are all defined with the same starting address. The load
4688 addresses of the sections are arranged such that they are consecutive in
4689 memory starting at the load address used for the @code{OVERLAY} as a
4690 whole (as with normal section definitions, the load address is optional,
4691 and defaults to the start address; the start address is also optional,
4692 and defaults to the current value of the location counter).
4694 If the @code{NOCROSSREFS} keyword is used, and there any references
4695 among the sections, the linker will report an error. Since the sections
4696 all run at the same address, it normally does not make sense for one
4697 section to refer directly to another. @xref{Miscellaneous Commands,
4700 For each section within the @code{OVERLAY}, the linker automatically
4701 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4702 defined as the starting load address of the section. The symbol
4703 @code{__load_stop_@var{secname}} is defined as the final load address of
4704 the section. Any characters within @var{secname} which are not legal
4705 within C identifiers are removed. C (or assembler) code may use these
4706 symbols to move the overlaid sections around as necessary.
4708 At the end of the overlay, the value of the location counter is set to
4709 the start address of the overlay plus the size of the largest section.
4711 Here is an example. Remember that this would appear inside a
4712 @code{SECTIONS} construct.
4715 OVERLAY 0x1000 : AT (0x4000)
4717 .text0 @{ o1/*.o(.text) @}
4718 .text1 @{ o2/*.o(.text) @}
4723 This will define both @samp{.text0} and @samp{.text1} to start at
4724 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4725 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4726 following symbols will be defined if referenced: @code{__load_start_text0},
4727 @code{__load_stop_text0}, @code{__load_start_text1},
4728 @code{__load_stop_text1}.
4730 C code to copy overlay @code{.text1} into the overlay area might look
4735 extern char __load_start_text1, __load_stop_text1;
4736 memcpy ((char *) 0x1000, &__load_start_text1,
4737 &__load_stop_text1 - &__load_start_text1);
4741 Note that the @code{OVERLAY} command is just syntactic sugar, since
4742 everything it does can be done using the more basic commands. The above
4743 example could have been written identically as follows.
4747 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4748 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4749 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4750 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4751 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4752 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4753 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4758 @section MEMORY Command
4760 @cindex memory regions
4761 @cindex regions of memory
4762 @cindex allocating memory
4763 @cindex discontinuous memory
4764 The linker's default configuration permits allocation of all available
4765 memory. You can override this by using the @code{MEMORY} command.
4767 The @code{MEMORY} command describes the location and size of blocks of
4768 memory in the target. You can use it to describe which memory regions
4769 may be used by the linker, and which memory regions it must avoid. You
4770 can then assign sections to particular memory regions. The linker will
4771 set section addresses based on the memory regions, and will warn about
4772 regions that become too full. The linker will not shuffle sections
4773 around to fit into the available regions.
4775 A linker script may contain at most one use of the @code{MEMORY}
4776 command. However, you can define as many blocks of memory within it as
4777 you wish. The syntax is:
4782 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4788 The @var{name} is a name used in the linker script to refer to the
4789 region. The region name has no meaning outside of the linker script.
4790 Region names are stored in a separate name space, and will not conflict
4791 with symbol names, file names, or section names. Each memory region
4792 must have a distinct name within the @code{MEMORY} command. However you can
4793 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
4796 @cindex memory region attributes
4797 The @var{attr} string is an optional list of attributes that specify
4798 whether to use a particular memory region for an input section which is
4799 not explicitly mapped in the linker script. As described in
4800 @ref{SECTIONS}, if you do not specify an output section for some input
4801 section, the linker will create an output section with the same name as
4802 the input section. If you define region attributes, the linker will use
4803 them to select the memory region for the output section that it creates.
4805 The @var{attr} string must consist only of the following characters:
4820 Invert the sense of any of the attributes that follow
4823 If a unmapped section matches any of the listed attributes other than
4824 @samp{!}, it will be placed in the memory region. The @samp{!}
4825 attribute reverses this test, so that an unmapped section will be placed
4826 in the memory region only if it does not match any of the listed
4832 The @var{origin} is an numerical expression for the start address of
4833 the memory region. The expression must evaluate to a constant and it
4834 cannot involve any symbols. The keyword @code{ORIGIN} may be
4835 abbreviated to @code{org} or @code{o} (but not, for example,
4841 The @var{len} is an expression for the size in bytes of the memory
4842 region. As with the @var{origin} expression, the expression must
4843 be numerical only and must evaluate to a constant. The keyword
4844 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4846 In the following example, we specify that there are two memory regions
4847 available for allocation: one starting at @samp{0} for 256 kilobytes,
4848 and the other starting at @samp{0x40000000} for four megabytes. The
4849 linker will place into the @samp{rom} memory region every section which
4850 is not explicitly mapped into a memory region, and is either read-only
4851 or executable. The linker will place other sections which are not
4852 explicitly mapped into a memory region into the @samp{ram} memory
4859 rom (rx) : ORIGIN = 0, LENGTH = 256K
4860 ram (!rx) : org = 0x40000000, l = 4M
4865 Once you define a memory region, you can direct the linker to place
4866 specific output sections into that memory region by using the
4867 @samp{>@var{region}} output section attribute. For example, if you have
4868 a memory region named @samp{mem}, you would use @samp{>mem} in the
4869 output section definition. @xref{Output Section Region}. If no address
4870 was specified for the output section, the linker will set the address to
4871 the next available address within the memory region. If the combined
4872 output sections directed to a memory region are too large for the
4873 region, the linker will issue an error message.
4875 It is possible to access the origin and length of a memory in an
4876 expression via the @code{ORIGIN(@var{memory})} and
4877 @code{LENGTH(@var{memory})} functions:
4881 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4886 @section PHDRS Command
4888 @cindex program headers
4889 @cindex ELF program headers
4890 @cindex program segments
4891 @cindex segments, ELF
4892 The ELF object file format uses @dfn{program headers}, also knows as
4893 @dfn{segments}. The program headers describe how the program should be
4894 loaded into memory. You can print them out by using the @code{objdump}
4895 program with the @samp{-p} option.
4897 When you run an ELF program on a native ELF system, the system loader
4898 reads the program headers in order to figure out how to load the
4899 program. This will only work if the program headers are set correctly.
4900 This manual does not describe the details of how the system loader
4901 interprets program headers; for more information, see the ELF ABI.
4903 The linker will create reasonable program headers by default. However,
4904 in some cases, you may need to specify the program headers more
4905 precisely. You may use the @code{PHDRS} command for this purpose. When
4906 the linker sees the @code{PHDRS} command in the linker script, it will
4907 not create any program headers other than the ones specified.
4909 The linker only pays attention to the @code{PHDRS} command when
4910 generating an ELF output file. In other cases, the linker will simply
4911 ignore @code{PHDRS}.
4913 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4914 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4920 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4921 [ FLAGS ( @var{flags} ) ] ;
4926 The @var{name} is used only for reference in the @code{SECTIONS} command
4927 of the linker script. It is not put into the output file. Program
4928 header names are stored in a separate name space, and will not conflict
4929 with symbol names, file names, or section names. Each program header
4930 must have a distinct name. The headers are processed in order and it
4931 is usual for them to map to sections in ascending load address order.
4933 Certain program header types describe segments of memory which the
4934 system loader will load from the file. In the linker script, you
4935 specify the contents of these segments by placing allocatable output
4936 sections in the segments. You use the @samp{:@var{phdr}} output section
4937 attribute to place a section in a particular segment. @xref{Output
4940 It is normal to put certain sections in more than one segment. This
4941 merely implies that one segment of memory contains another. You may
4942 repeat @samp{:@var{phdr}}, using it once for each segment which should
4943 contain the section.
4945 If you place a section in one or more segments using @samp{:@var{phdr}},
4946 then the linker will place all subsequent allocatable sections which do
4947 not specify @samp{:@var{phdr}} in the same segments. This is for
4948 convenience, since generally a whole set of contiguous sections will be
4949 placed in a single segment. You can use @code{:NONE} to override the
4950 default segment and tell the linker to not put the section in any
4955 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
4956 the program header type to further describe the contents of the segment.
4957 The @code{FILEHDR} keyword means that the segment should include the ELF
4958 file header. The @code{PHDRS} keyword means that the segment should
4959 include the ELF program headers themselves. If applied to a loadable
4960 segment (@code{PT_LOAD}), all prior loadable segments must have one of
4963 The @var{type} may be one of the following. The numbers indicate the
4964 value of the keyword.
4967 @item @code{PT_NULL} (0)
4968 Indicates an unused program header.
4970 @item @code{PT_LOAD} (1)
4971 Indicates that this program header describes a segment to be loaded from
4974 @item @code{PT_DYNAMIC} (2)
4975 Indicates a segment where dynamic linking information can be found.
4977 @item @code{PT_INTERP} (3)
4978 Indicates a segment where the name of the program interpreter may be
4981 @item @code{PT_NOTE} (4)
4982 Indicates a segment holding note information.
4984 @item @code{PT_SHLIB} (5)
4985 A reserved program header type, defined but not specified by the ELF
4988 @item @code{PT_PHDR} (6)
4989 Indicates a segment where the program headers may be found.
4991 @item @var{expression}
4992 An expression giving the numeric type of the program header. This may
4993 be used for types not defined above.
4996 You can specify that a segment should be loaded at a particular address
4997 in memory by using an @code{AT} expression. This is identical to the
4998 @code{AT} command used as an output section attribute (@pxref{Output
4999 Section LMA}). The @code{AT} command for a program header overrides the
5000 output section attribute.
5002 The linker will normally set the segment flags based on the sections
5003 which comprise the segment. You may use the @code{FLAGS} keyword to
5004 explicitly specify the segment flags. The value of @var{flags} must be
5005 an integer. It is used to set the @code{p_flags} field of the program
5008 Here is an example of @code{PHDRS}. This shows a typical set of program
5009 headers used on a native ELF system.
5015 headers PT_PHDR PHDRS ;
5017 text PT_LOAD FILEHDR PHDRS ;
5019 dynamic PT_DYNAMIC ;
5025 .interp : @{ *(.interp) @} :text :interp
5026 .text : @{ *(.text) @} :text
5027 .rodata : @{ *(.rodata) @} /* defaults to :text */
5029 . = . + 0x1000; /* move to a new page in memory */
5030 .data : @{ *(.data) @} :data
5031 .dynamic : @{ *(.dynamic) @} :data :dynamic
5038 @section VERSION Command
5039 @kindex VERSION @{script text@}
5040 @cindex symbol versions
5041 @cindex version script
5042 @cindex versions of symbols
5043 The linker supports symbol versions when using ELF. Symbol versions are
5044 only useful when using shared libraries. The dynamic linker can use
5045 symbol versions to select a specific version of a function when it runs
5046 a program that may have been linked against an earlier version of the
5049 You can include a version script directly in the main linker script, or
5050 you can supply the version script as an implicit linker script. You can
5051 also use the @samp{--version-script} linker option.
5053 The syntax of the @code{VERSION} command is simply
5055 VERSION @{ version-script-commands @}
5058 The format of the version script commands is identical to that used by
5059 Sun's linker in Solaris 2.5. The version script defines a tree of
5060 version nodes. You specify the node names and interdependencies in the
5061 version script. You can specify which symbols are bound to which
5062 version nodes, and you can reduce a specified set of symbols to local
5063 scope so that they are not globally visible outside of the shared
5066 The easiest way to demonstrate the version script language is with a few
5092 This example version script defines three version nodes. The first
5093 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5094 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5095 a number of symbols to local scope so that they are not visible outside
5096 of the shared library; this is done using wildcard patterns, so that any
5097 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5098 is matched. The wildcard patterns available are the same as those used
5099 in the shell when matching filenames (also known as ``globbing'').
5100 However, if you specify the symbol name inside double quotes, then the
5101 name is treated as literal, rather than as a glob pattern.
5103 Next, the version script defines node @samp{VERS_1.2}. This node
5104 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5105 to the version node @samp{VERS_1.2}.
5107 Finally, the version script defines node @samp{VERS_2.0}. This node
5108 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5109 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5111 When the linker finds a symbol defined in a library which is not
5112 specifically bound to a version node, it will effectively bind it to an
5113 unspecified base version of the library. You can bind all otherwise
5114 unspecified symbols to a given version node by using @samp{global: *;}
5115 somewhere in the version script. Note that it's slightly crazy to use
5116 wildcards in a global spec except on the last version node. Global
5117 wildcards elsewhere run the risk of accidentally adding symbols to the
5118 set exported for an old version. That's wrong since older versions
5119 ought to have a fixed set of symbols.
5121 The names of the version nodes have no specific meaning other than what
5122 they might suggest to the person reading them. The @samp{2.0} version
5123 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5124 However, this would be a confusing way to write a version script.
5126 Node name can be omitted, provided it is the only version node
5127 in the version script. Such version script doesn't assign any versions to
5128 symbols, only selects which symbols will be globally visible out and which
5132 @{ global: foo; bar; local: *; @};
5135 When you link an application against a shared library that has versioned
5136 symbols, the application itself knows which version of each symbol it
5137 requires, and it also knows which version nodes it needs from each
5138 shared library it is linked against. Thus at runtime, the dynamic
5139 loader can make a quick check to make sure that the libraries you have
5140 linked against do in fact supply all of the version nodes that the
5141 application will need to resolve all of the dynamic symbols. In this
5142 way it is possible for the dynamic linker to know with certainty that
5143 all external symbols that it needs will be resolvable without having to
5144 search for each symbol reference.
5146 The symbol versioning is in effect a much more sophisticated way of
5147 doing minor version checking that SunOS does. The fundamental problem
5148 that is being addressed here is that typically references to external
5149 functions are bound on an as-needed basis, and are not all bound when
5150 the application starts up. If a shared library is out of date, a
5151 required interface may be missing; when the application tries to use
5152 that interface, it may suddenly and unexpectedly fail. With symbol
5153 versioning, the user will get a warning when they start their program if
5154 the libraries being used with the application are too old.
5156 There are several GNU extensions to Sun's versioning approach. The
5157 first of these is the ability to bind a symbol to a version node in the
5158 source file where the symbol is defined instead of in the versioning
5159 script. This was done mainly to reduce the burden on the library
5160 maintainer. You can do this by putting something like:
5162 __asm__(".symver original_foo,foo@@VERS_1.1");
5165 in the C source file. This renames the function @samp{original_foo} to
5166 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5167 The @samp{local:} directive can be used to prevent the symbol
5168 @samp{original_foo} from being exported. A @samp{.symver} directive
5169 takes precedence over a version script.
5171 The second GNU extension is to allow multiple versions of the same
5172 function to appear in a given shared library. In this way you can make
5173 an incompatible change to an interface without increasing the major
5174 version number of the shared library, while still allowing applications
5175 linked against the old interface to continue to function.
5177 To do this, you must use multiple @samp{.symver} directives in the
5178 source file. Here is an example:
5181 __asm__(".symver original_foo,foo@@");
5182 __asm__(".symver old_foo,foo@@VERS_1.1");
5183 __asm__(".symver old_foo1,foo@@VERS_1.2");
5184 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5187 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5188 unspecified base version of the symbol. The source file that contains this
5189 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5190 @samp{old_foo1}, and @samp{new_foo}.
5192 When you have multiple definitions of a given symbol, there needs to be
5193 some way to specify a default version to which external references to
5194 this symbol will be bound. You can do this with the
5195 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5196 declare one version of a symbol as the default in this manner; otherwise
5197 you would effectively have multiple definitions of the same symbol.
5199 If you wish to bind a reference to a specific version of the symbol
5200 within the shared library, you can use the aliases of convenience
5201 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5202 specifically bind to an external version of the function in question.
5204 You can also specify the language in the version script:
5207 VERSION extern "lang" @{ version-script-commands @}
5210 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5211 The linker will iterate over the list of symbols at the link time and
5212 demangle them according to @samp{lang} before matching them to the
5213 patterns specified in @samp{version-script-commands}. The default
5214 @samp{lang} is @samp{C}.
5216 Demangled names may contains spaces and other special characters. As
5217 described above, you can use a glob pattern to match demangled names,
5218 or you can use a double-quoted string to match the string exactly. In
5219 the latter case, be aware that minor differences (such as differing
5220 whitespace) between the version script and the demangler output will
5221 cause a mismatch. As the exact string generated by the demangler
5222 might change in the future, even if the mangled name does not, you
5223 should check that all of your version directives are behaving as you
5224 expect when you upgrade.
5227 @section Expressions in Linker Scripts
5230 The syntax for expressions in the linker script language is identical to
5231 that of C expressions. All expressions are evaluated as integers. All
5232 expressions are evaluated in the same size, which is 32 bits if both the
5233 host and target are 32 bits, and is otherwise 64 bits.
5235 You can use and set symbol values in expressions.
5237 The linker defines several special purpose builtin functions for use in
5241 * Constants:: Constants
5242 * Symbolic Constants:: Symbolic constants
5243 * Symbols:: Symbol Names
5244 * Orphan Sections:: Orphan Sections
5245 * Location Counter:: The Location Counter
5246 * Operators:: Operators
5247 * Evaluation:: Evaluation
5248 * Expression Section:: The Section of an Expression
5249 * Builtin Functions:: Builtin Functions
5253 @subsection Constants
5254 @cindex integer notation
5255 @cindex constants in linker scripts
5256 All constants are integers.
5258 As in C, the linker considers an integer beginning with @samp{0} to be
5259 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5260 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5261 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5262 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5263 value without a prefix or a suffix is considered to be decimal.
5265 @cindex scaled integers
5266 @cindex K and M integer suffixes
5267 @cindex M and K integer suffixes
5268 @cindex suffixes for integers
5269 @cindex integer suffixes
5270 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5274 @c END TEXI2ROFF-KILL
5275 @code{1024} or @code{1024*1024}
5279 ${\rm 1024}$ or ${\rm 1024}^2$
5281 @c END TEXI2ROFF-KILL
5282 respectively. For example, the following
5283 all refer to the same quantity:
5292 Note - the @code{K} and @code{M} suffixes cannot be used in
5293 conjunction with the base suffixes mentioned above.
5295 @node Symbolic Constants
5296 @subsection Symbolic Constants
5297 @cindex symbolic constants
5299 It is possible to refer to target specific constants via the use of
5300 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5305 The target's maximum page size.
5307 @item COMMONPAGESIZE
5308 @kindex COMMONPAGESIZE
5309 The target's default page size.
5315 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5318 will create a text section aligned to the largest page boundary
5319 supported by the target.
5322 @subsection Symbol Names
5323 @cindex symbol names
5325 @cindex quoted symbol names
5327 Unless quoted, symbol names start with a letter, underscore, or period
5328 and may include letters, digits, underscores, periods, and hyphens.
5329 Unquoted symbol names must not conflict with any keywords. You can
5330 specify a symbol which contains odd characters or has the same name as a
5331 keyword by surrounding the symbol name in double quotes:
5334 "with a space" = "also with a space" + 10;
5337 Since symbols can contain many non-alphabetic characters, it is safest
5338 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5339 whereas @samp{A - B} is an expression involving subtraction.
5341 @node Orphan Sections
5342 @subsection Orphan Sections
5344 Orphan sections are sections present in the input files which
5345 are not explicitly placed into the output file by the linker
5346 script. The linker will still copy these sections into the
5347 output file, but it has to guess as to where they should be
5348 placed. The linker uses a simple heuristic to do this. It
5349 attempts to place orphan sections after non-orphan sections of the
5350 same attribute, such as code vs data, loadable vs non-loadable, etc.
5351 If there is not enough room to do this then it places
5352 at the end of the file.
5354 For ELF targets, the attribute of the section includes section type as
5355 well as section flag.
5357 If an orphaned section's name is representable as a C identifier then
5358 the linker will automatically @pxref{PROVIDE} two symbols:
5359 __start_SECNAME and __stop_SECNAME, where SECNAME is the name of the
5360 section. These indicate the start address and end address of the
5361 orphaned section respectively. Note: most section names are not
5362 representable as C identifiers because they contain a @samp{.}
5365 @node Location Counter
5366 @subsection The Location Counter
5369 @cindex location counter
5370 @cindex current output location
5371 The special linker variable @dfn{dot} @samp{.} always contains the
5372 current output location counter. Since the @code{.} always refers to a
5373 location in an output section, it may only appear in an expression
5374 within a @code{SECTIONS} command. The @code{.} symbol may appear
5375 anywhere that an ordinary symbol is allowed in an expression.
5378 Assigning a value to @code{.} will cause the location counter to be
5379 moved. This may be used to create holes in the output section. The
5380 location counter may not be moved backwards inside an output section,
5381 and may not be moved backwards outside of an output section if so
5382 doing creates areas with overlapping LMAs.
5398 In the previous example, the @samp{.text} section from @file{file1} is
5399 located at the beginning of the output section @samp{output}. It is
5400 followed by a 1000 byte gap. Then the @samp{.text} section from
5401 @file{file2} appears, also with a 1000 byte gap following before the
5402 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5403 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5405 @cindex dot inside sections
5406 Note: @code{.} actually refers to the byte offset from the start of the
5407 current containing object. Normally this is the @code{SECTIONS}
5408 statement, whose start address is 0, hence @code{.} can be used as an
5409 absolute address. If @code{.} is used inside a section description
5410 however, it refers to the byte offset from the start of that section,
5411 not an absolute address. Thus in a script like this:
5429 The @samp{.text} section will be assigned a starting address of 0x100
5430 and a size of exactly 0x200 bytes, even if there is not enough data in
5431 the @samp{.text} input sections to fill this area. (If there is too
5432 much data, an error will be produced because this would be an attempt to
5433 move @code{.} backwards). The @samp{.data} section will start at 0x500
5434 and it will have an extra 0x600 bytes worth of space after the end of
5435 the values from the @samp{.data} input sections and before the end of
5436 the @samp{.data} output section itself.
5438 @cindex dot outside sections
5439 Setting symbols to the value of the location counter outside of an
5440 output section statement can result in unexpected values if the linker
5441 needs to place orphan sections. For example, given the following:
5447 .text: @{ *(.text) @}
5451 .data: @{ *(.data) @}
5456 If the linker needs to place some input section, e.g. @code{.rodata},
5457 not mentioned in the script, it might choose to place that section
5458 between @code{.text} and @code{.data}. You might think the linker
5459 should place @code{.rodata} on the blank line in the above script, but
5460 blank lines are of no particular significance to the linker. As well,
5461 the linker doesn't associate the above symbol names with their
5462 sections. Instead, it assumes that all assignments or other
5463 statements belong to the previous output section, except for the
5464 special case of an assignment to @code{.}. I.e., the linker will
5465 place the orphan @code{.rodata} section as if the script was written
5472 .text: @{ *(.text) @}
5476 .rodata: @{ *(.rodata) @}
5477 .data: @{ *(.data) @}
5482 This may or may not be the script author's intention for the value of
5483 @code{start_of_data}. One way to influence the orphan section
5484 placement is to assign the location counter to itself, as the linker
5485 assumes that an assignment to @code{.} is setting the start address of
5486 a following output section and thus should be grouped with that
5487 section. So you could write:
5493 .text: @{ *(.text) @}
5498 .data: @{ *(.data) @}
5503 Now, the orphan @code{.rodata} section will be placed between
5504 @code{end_of_text} and @code{start_of_data}.
5508 @subsection Operators
5509 @cindex operators for arithmetic
5510 @cindex arithmetic operators
5511 @cindex precedence in expressions
5512 The linker recognizes the standard C set of arithmetic operators, with
5513 the standard bindings and precedence levels:
5516 @c END TEXI2ROFF-KILL
5518 precedence associativity Operators Notes
5524 5 left == != > < <= >=
5530 11 right &= += -= *= /= (2)
5534 (1) Prefix operators
5535 (2) @xref{Assignments}.
5539 \vskip \baselineskip
5540 %"lispnarrowing" is the extra indent used generally for smallexample
5541 \hskip\lispnarrowing\vbox{\offinterlineskip
5544 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
5545 height2pt&\omit&&\omit&&\omit&\cr
5546 &Precedence&& Associativity &&{\rm Operators}&\cr
5547 height2pt&\omit&&\omit&&\omit&\cr
5549 height2pt&\omit&&\omit&&\omit&\cr
5551 % '176 is tilde, '~' in tt font
5552 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
5553 &2&&left&&* / \%&\cr
5556 &5&&left&&== != > < <= >=&\cr
5559 &8&&left&&{\&\&}&\cr
5562 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
5564 height2pt&\omit&&\omit&&\omit&\cr}
5569 @obeylines@parskip=0pt@parindent=0pt
5570 @dag@quad Prefix operators.
5571 @ddag@quad @xref{Assignments}.
5574 @c END TEXI2ROFF-KILL
5577 @subsection Evaluation
5578 @cindex lazy evaluation
5579 @cindex expression evaluation order
5580 The linker evaluates expressions lazily. It only computes the value of
5581 an expression when absolutely necessary.
5583 The linker needs some information, such as the value of the start
5584 address of the first section, and the origins and lengths of memory
5585 regions, in order to do any linking at all. These values are computed
5586 as soon as possible when the linker reads in the linker script.
5588 However, other values (such as symbol values) are not known or needed
5589 until after storage allocation. Such values are evaluated later, when
5590 other information (such as the sizes of output sections) is available
5591 for use in the symbol assignment expression.
5593 The sizes of sections cannot be known until after allocation, so
5594 assignments dependent upon these are not performed until after
5597 Some expressions, such as those depending upon the location counter
5598 @samp{.}, must be evaluated during section allocation.
5600 If the result of an expression is required, but the value is not
5601 available, then an error results. For example, a script like the
5607 .text 9+this_isnt_constant :
5613 will cause the error message @samp{non constant expression for initial
5616 @node Expression Section
5617 @subsection The Section of an Expression
5618 @cindex expression sections
5619 @cindex absolute expressions
5620 @cindex relative expressions
5621 @cindex absolute and relocatable symbols
5622 @cindex relocatable and absolute symbols
5623 @cindex symbols, relocatable and absolute
5624 Addresses and symbols may be section relative, or absolute. A section
5625 relative symbol is relocatable. If you request relocatable output
5626 using the @samp{-r} option, a further link operation may change the
5627 value of a section relative symbol. On the other hand, an absolute
5628 symbol will retain the same value throughout any further link
5631 Some terms in linker expressions are addresses. This is true of
5632 section relative symbols and for builtin functions that return an
5633 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
5634 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
5635 functions that return a non-address value, such as @code{LENGTH}.
5636 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
5637 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
5638 differently depending on their location, for compatibility with older
5639 versions of @code{ld}. Expressions appearing outside an output
5640 section definition treat all numbers as absolute addresses.
5641 Expressions appearing inside an output section definition treat
5642 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
5643 given, then absolute symbols and numbers are simply treated as numbers
5646 In the following simple example,
5653 __executable_start = 0x100;
5657 __data_start = 0x10;
5665 both @code{.} and @code{__executable_start} are set to the absolute
5666 address 0x100 in the first two assignments, then both @code{.} and
5667 @code{__data_start} are set to 0x10 relative to the @code{.data}
5668 section in the second two assignments.
5670 For expressions involving numbers, relative addresses and absolute
5671 addresses, ld follows these rules to evaluate terms:
5675 Unary operations on an absolute address or number, and binary
5676 operations on two absolute addresses or two numbers, or between one
5677 absolute address and a number, apply the operator to the value(s).
5679 Unary operations on a relative address, and binary operations on two
5680 relative addresses in the same section or between one relative address
5681 and a number, apply the operator to the offset part of the address(es).
5683 Other binary operations, that is, between two relative addresses not
5684 in the same section, or between a relative address and an absolute
5685 address, first convert any non-absolute term to an absolute address
5686 before applying the operator.
5689 The result section of each sub-expression is as follows:
5693 An operation involving only numbers results in a number.
5695 The result of comparisons, @samp{&&} and @samp{||} is also a number.
5697 The result of other binary arithmetic and logical operations on two
5698 relative addresses in the same section or two absolute addresses
5699 (after above conversions) is also a number.
5701 The result of other operations on relative addresses or one
5702 relative address and a number, is a relative address in the same
5703 section as the relative operand(s).
5705 The result of other operations on absolute addresses (after above
5706 conversions) is an absolute address.
5709 You can use the builtin function @code{ABSOLUTE} to force an expression
5710 to be absolute when it would otherwise be relative. For example, to
5711 create an absolute symbol set to the address of the end of the output
5712 section @samp{.data}:
5716 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
5720 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
5721 @samp{.data} section.
5723 Using @code{LOADADDR} also forces an expression absolute, since this
5724 particular builtin function returns an absolute address.
5726 @node Builtin Functions
5727 @subsection Builtin Functions
5728 @cindex functions in expressions
5729 The linker script language includes a number of builtin functions for
5730 use in linker script expressions.
5733 @item ABSOLUTE(@var{exp})
5734 @kindex ABSOLUTE(@var{exp})
5735 @cindex expression, absolute
5736 Return the absolute (non-relocatable, as opposed to non-negative) value
5737 of the expression @var{exp}. Primarily useful to assign an absolute
5738 value to a symbol within a section definition, where symbol values are
5739 normally section relative. @xref{Expression Section}.
5741 @item ADDR(@var{section})
5742 @kindex ADDR(@var{section})
5743 @cindex section address in expression
5744 Return the address (VMA) of the named @var{section}. Your
5745 script must previously have defined the location of that section. In
5746 the following example, @code{start_of_output_1}, @code{symbol_1} and
5747 @code{symbol_2} are assigned equivalent values, except that
5748 @code{symbol_1} will be relative to the @code{.output1} section while
5749 the other two will be absolute:
5755 start_of_output_1 = ABSOLUTE(.);
5760 symbol_1 = ADDR(.output1);
5761 symbol_2 = start_of_output_1;
5767 @item ALIGN(@var{align})
5768 @itemx ALIGN(@var{exp},@var{align})
5769 @kindex ALIGN(@var{align})
5770 @kindex ALIGN(@var{exp},@var{align})
5771 @cindex round up location counter
5772 @cindex align location counter
5773 @cindex round up expression
5774 @cindex align expression
5775 Return the location counter (@code{.}) or arbitrary expression aligned
5776 to the next @var{align} boundary. The single operand @code{ALIGN}
5777 doesn't change the value of the location counter---it just does
5778 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5779 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5780 equivalent to @code{ALIGN(., @var{align})}).
5782 Here is an example which aligns the output @code{.data} section to the
5783 next @code{0x2000} byte boundary after the preceding section and sets a
5784 variable within the section to the next @code{0x8000} boundary after the
5789 .data ALIGN(0x2000): @{
5791 variable = ALIGN(0x8000);
5797 The first use of @code{ALIGN} in this example specifies the location of
5798 a section because it is used as the optional @var{address} attribute of
5799 a section definition (@pxref{Output Section Address}). The second use
5800 of @code{ALIGN} is used to defines the value of a symbol.
5802 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5804 @item ALIGNOF(@var{section})
5805 @kindex ALIGNOF(@var{section})
5806 @cindex section alignment
5807 Return the alignment in bytes of the named @var{section}, if that section has
5808 been allocated. If the section has not been allocated when this is
5809 evaluated, the linker will report an error. In the following example,
5810 the alignment of the @code{.output} section is stored as the first
5811 value in that section.
5816 LONG (ALIGNOF (.output))
5823 @item BLOCK(@var{exp})
5824 @kindex BLOCK(@var{exp})
5825 This is a synonym for @code{ALIGN}, for compatibility with older linker
5826 scripts. It is most often seen when setting the address of an output
5829 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5830 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5831 This is equivalent to either
5833 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5837 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5840 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5841 for the data segment (area between the result of this expression and
5842 @code{DATA_SEGMENT_END}) than the former or not.
5843 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5844 memory will be saved at the expense of up to @var{commonpagesize} wasted
5845 bytes in the on-disk file.
5847 This expression can only be used directly in @code{SECTIONS} commands, not in
5848 any output section descriptions and only once in the linker script.
5849 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5850 be the system page size the object wants to be optimized for (while still
5851 working on system page sizes up to @var{maxpagesize}).
5856 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5859 @item DATA_SEGMENT_END(@var{exp})
5860 @kindex DATA_SEGMENT_END(@var{exp})
5861 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5862 evaluation purposes.
5865 . = DATA_SEGMENT_END(.);
5868 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5869 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5870 This defines the end of the @code{PT_GNU_RELRO} segment when
5871 @samp{-z relro} option is used. Second argument is returned.
5872 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5873 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5874 @var{exp} + @var{offset} is aligned to the most commonly used page
5875 boundary for particular target. If present in the linker script,
5876 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5877 @code{DATA_SEGMENT_END}.
5880 . = DATA_SEGMENT_RELRO_END(24, .);
5883 @item DEFINED(@var{symbol})
5884 @kindex DEFINED(@var{symbol})
5885 @cindex symbol defaults
5886 Return 1 if @var{symbol} is in the linker global symbol table and is
5887 defined before the statement using DEFINED in the script, otherwise
5888 return 0. You can use this function to provide
5889 default values for symbols. For example, the following script fragment
5890 shows how to set a global symbol @samp{begin} to the first location in
5891 the @samp{.text} section---but if a symbol called @samp{begin} already
5892 existed, its value is preserved:
5898 begin = DEFINED(begin) ? begin : . ;
5906 @item LENGTH(@var{memory})
5907 @kindex LENGTH(@var{memory})
5908 Return the length of the memory region named @var{memory}.
5910 @item LOADADDR(@var{section})
5911 @kindex LOADADDR(@var{section})
5912 @cindex section load address in expression
5913 Return the absolute LMA of the named @var{section}. (@pxref{Output
5917 @item MAX(@var{exp1}, @var{exp2})
5918 Returns the maximum of @var{exp1} and @var{exp2}.
5921 @item MIN(@var{exp1}, @var{exp2})
5922 Returns the minimum of @var{exp1} and @var{exp2}.
5924 @item NEXT(@var{exp})
5925 @kindex NEXT(@var{exp})
5926 @cindex unallocated address, next
5927 Return the next unallocated address that is a multiple of @var{exp}.
5928 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5929 use the @code{MEMORY} command to define discontinuous memory for the
5930 output file, the two functions are equivalent.
5932 @item ORIGIN(@var{memory})
5933 @kindex ORIGIN(@var{memory})
5934 Return the origin of the memory region named @var{memory}.
5936 @item SEGMENT_START(@var{segment}, @var{default})
5937 @kindex SEGMENT_START(@var{segment}, @var{default})
5938 Return the base address of the named @var{segment}. If an explicit
5939 value has been given for this segment (with a command-line @samp{-T}
5940 option) that value will be returned; otherwise the value will be
5941 @var{default}. At present, the @samp{-T} command-line option can only
5942 be used to set the base address for the ``text'', ``data'', and
5943 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
5946 @item SIZEOF(@var{section})
5947 @kindex SIZEOF(@var{section})
5948 @cindex section size
5949 Return the size in bytes of the named @var{section}, if that section has
5950 been allocated. If the section has not been allocated when this is
5951 evaluated, the linker will report an error. In the following example,
5952 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5961 symbol_1 = .end - .start ;
5962 symbol_2 = SIZEOF(.output);
5967 @item SIZEOF_HEADERS
5968 @itemx sizeof_headers
5969 @kindex SIZEOF_HEADERS
5971 Return the size in bytes of the output file's headers. This is
5972 information which appears at the start of the output file. You can use
5973 this number when setting the start address of the first section, if you
5974 choose, to facilitate paging.
5976 @cindex not enough room for program headers
5977 @cindex program headers, not enough room
5978 When producing an ELF output file, if the linker script uses the
5979 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5980 number of program headers before it has determined all the section
5981 addresses and sizes. If the linker later discovers that it needs
5982 additional program headers, it will report an error @samp{not enough
5983 room for program headers}. To avoid this error, you must avoid using
5984 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5985 script to avoid forcing the linker to use additional program headers, or
5986 you must define the program headers yourself using the @code{PHDRS}
5987 command (@pxref{PHDRS}).
5990 @node Implicit Linker Scripts
5991 @section Implicit Linker Scripts
5992 @cindex implicit linker scripts
5993 If you specify a linker input file which the linker can not recognize as
5994 an object file or an archive file, it will try to read the file as a
5995 linker script. If the file can not be parsed as a linker script, the
5996 linker will report an error.
5998 An implicit linker script will not replace the default linker script.
6000 Typically an implicit linker script would contain only symbol
6001 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6004 Any input files read because of an implicit linker script will be read
6005 at the position in the command line where the implicit linker script was
6006 read. This can affect archive searching.
6009 @node Machine Dependent
6010 @chapter Machine Dependent Features
6012 @cindex machine dependencies
6013 @command{ld} has additional features on some platforms; the following
6014 sections describe them. Machines where @command{ld} has no additional
6015 functionality are not listed.
6019 * H8/300:: @command{ld} and the H8/300
6022 * i960:: @command{ld} and the Intel 960 family
6025 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6028 * ARM:: @command{ld} and the ARM family
6031 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6034 * M68K:: @command{ld} and the Motorola 68K family
6037 * MMIX:: @command{ld} and MMIX
6040 * MSP430:: @command{ld} and MSP430
6043 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6046 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6049 * SPU ELF:: @command{ld} and SPU ELF Support
6052 * TI COFF:: @command{ld} and TI COFF
6055 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6058 * Xtensa:: @command{ld} and Xtensa Processors
6069 @section @command{ld} and the H8/300
6071 @cindex H8/300 support
6072 For the H8/300, @command{ld} can perform these global optimizations when
6073 you specify the @samp{--relax} command-line option.
6076 @cindex relaxing on H8/300
6077 @item relaxing address modes
6078 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6079 targets are within eight bits, and turns them into eight-bit
6080 program-counter relative @code{bsr} and @code{bra} instructions,
6083 @cindex synthesizing on H8/300
6084 @item synthesizing instructions
6085 @c FIXME: specifically mov.b, or any mov instructions really?
6086 @command{ld} finds all @code{mov.b} instructions which use the
6087 sixteen-bit absolute address form, but refer to the top
6088 page of memory, and changes them to use the eight-bit address form.
6089 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6090 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6091 top page of memory).
6093 @item bit manipulation instructions
6094 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6095 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6096 which use 32 bit and 16 bit absolute address form, but refer to the top
6097 page of memory, and changes them to use the 8 bit address form.
6098 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6099 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6100 the top page of memory).
6102 @item system control instructions
6103 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6104 32 bit absolute address form, but refer to the top page of memory, and
6105 changes them to use 16 bit address form.
6106 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6107 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6108 the top page of memory).
6118 @c This stuff is pointless to say unless you're especially concerned
6119 @c with Renesas chips; don't enable it for generic case, please.
6121 @chapter @command{ld} and Other Renesas Chips
6123 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6124 H8/500, and SH chips. No special features, commands, or command-line
6125 options are required for these chips.
6135 @section @command{ld} and the Intel 960 Family
6137 @cindex i960 support
6139 You can use the @samp{-A@var{architecture}} command line option to
6140 specify one of the two-letter names identifying members of the 960
6141 family; the option specifies the desired output target, and warns of any
6142 incompatible instructions in the input files. It also modifies the
6143 linker's search strategy for archive libraries, to support the use of
6144 libraries specific to each particular architecture, by including in the
6145 search loop names suffixed with the string identifying the architecture.
6147 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
6148 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
6149 paths, and in any paths you specify with @samp{-L}) for a library with
6162 The first two possibilities would be considered in any event; the last
6163 two are due to the use of @w{@samp{-ACA}}.
6165 You can meaningfully use @samp{-A} more than once on a command line, since
6166 the 960 architecture family allows combination of target architectures; each
6167 use will add another pair of name variants to search for when @w{@samp{-l}}
6168 specifies a library.
6170 @cindex @option{--relax} on i960
6171 @cindex relaxing on i960
6172 @command{ld} supports the @samp{--relax} option for the i960 family. If
6173 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
6174 @code{calx} instructions whose targets are within 24 bits, and turns
6175 them into 24-bit program-counter relative @code{bal} and @code{cal}
6176 instructions, respectively. @command{ld} also turns @code{cal}
6177 instructions into @code{bal} instructions when it determines that the
6178 target subroutine is a leaf routine (that is, the target subroutine does
6179 not itself call any subroutines).
6181 @cindex Cortex-A8 erratum workaround
6182 @kindex --fix-cortex-a8
6183 @kindex --no-fix-cortex-a8
6184 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}.
6186 The erratum only affects Thumb-2 code. Please contact ARM for further details.
6188 @kindex --merge-exidx-entries
6189 @kindex --no-merge-exidx-entries
6190 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
6207 @node M68HC11/68HC12
6208 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6210 @cindex M68HC11 and 68HC12 support
6212 @subsection Linker Relaxation
6214 For the Motorola 68HC11, @command{ld} can perform these global
6215 optimizations when you specify the @samp{--relax} command-line option.
6218 @cindex relaxing on M68HC11
6219 @item relaxing address modes
6220 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6221 targets are within eight bits, and turns them into eight-bit
6222 program-counter relative @code{bsr} and @code{bra} instructions,
6225 @command{ld} also looks at all 16-bit extended addressing modes and
6226 transforms them in a direct addressing mode when the address is in
6227 page 0 (between 0 and 0x0ff).
6229 @item relaxing gcc instruction group
6230 When @command{gcc} is called with @option{-mrelax}, it can emit group
6231 of instructions that the linker can optimize to use a 68HC11 direct
6232 addressing mode. These instructions consists of @code{bclr} or
6233 @code{bset} instructions.
6237 @subsection Trampoline Generation
6239 @cindex trampoline generation on M68HC11
6240 @cindex trampoline generation on M68HC12
6241 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6242 call a far function using a normal @code{jsr} instruction. The linker
6243 will also change the relocation to some far function to use the
6244 trampoline address instead of the function address. This is typically the
6245 case when a pointer to a function is taken. The pointer will in fact
6246 point to the function trampoline.
6254 @section @command{ld} and the ARM family
6256 @cindex ARM interworking support
6257 @kindex --support-old-code
6258 For the ARM, @command{ld} will generate code stubs to allow functions calls
6259 between ARM and Thumb code. These stubs only work with code that has
6260 been compiled and assembled with the @samp{-mthumb-interwork} command
6261 line option. If it is necessary to link with old ARM object files or
6262 libraries, which have not been compiled with the -mthumb-interwork
6263 option then the @samp{--support-old-code} command line switch should be
6264 given to the linker. This will make it generate larger stub functions
6265 which will work with non-interworking aware ARM code. Note, however,
6266 the linker does not support generating stubs for function calls to
6267 non-interworking aware Thumb code.
6269 @cindex thumb entry point
6270 @cindex entry point, thumb
6271 @kindex --thumb-entry=@var{entry}
6272 The @samp{--thumb-entry} switch is a duplicate of the generic
6273 @samp{--entry} switch, in that it sets the program's starting address.
6274 But it also sets the bottom bit of the address, so that it can be
6275 branched to using a BX instruction, and the program will start
6276 executing in Thumb mode straight away.
6278 @cindex PE import table prefixing
6279 @kindex --use-nul-prefixed-import-tables
6280 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6281 the import tables idata4 and idata5 have to be generated with a zero
6282 element prefix for import libraries. This is the old style to generate
6283 import tables. By default this option is turned off.
6287 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6288 executables. This option is only valid when linking big-endian objects.
6289 The resulting image will contain big-endian data and little-endian code.
6292 @kindex --target1-rel
6293 @kindex --target1-abs
6294 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6295 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6296 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6297 and @samp{--target1-abs} switches override the default.
6300 @kindex --target2=@var{type}
6301 The @samp{--target2=type} switch overrides the default definition of the
6302 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6303 meanings, and target defaults are as follows:
6306 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6308 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6310 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6315 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6316 specification) enables objects compiled for the ARMv4 architecture to be
6317 interworking-safe when linked with other objects compiled for ARMv4t, but
6318 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6320 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6321 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6322 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6324 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6325 relocations are ignored.
6327 @cindex FIX_V4BX_INTERWORKING
6328 @kindex --fix-v4bx-interworking
6329 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6330 relocations with a branch to the following veneer:
6338 This allows generation of libraries/applications that work on ARMv4 cores
6339 and are still interworking safe. Note that the above veneer clobbers the
6340 condition flags, so may cause incorrect program behavior in rare cases.
6344 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6345 BLX instructions (available on ARMv5t and above) in various
6346 situations. Currently it is used to perform calls via the PLT from Thumb
6347 code using BLX rather than using BX and a mode-switching stub before
6348 each PLT entry. This should lead to such calls executing slightly faster.
6350 This option is enabled implicitly for SymbianOS, so there is no need to
6351 specify it if you are using that target.
6353 @cindex VFP11_DENORM_FIX
6354 @kindex --vfp11-denorm-fix
6355 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6356 bug in certain VFP11 coprocessor hardware, which sometimes allows
6357 instructions with denorm operands (which must be handled by support code)
6358 to have those operands overwritten by subsequent instructions before
6359 the support code can read the intended values.
6361 The bug may be avoided in scalar mode if you allow at least one
6362 intervening instruction between a VFP11 instruction which uses a register
6363 and another instruction which writes to the same register, or at least two
6364 intervening instructions if vector mode is in use. The bug only affects
6365 full-compliance floating-point mode: you do not need this workaround if
6366 you are using "runfast" mode. Please contact ARM for further details.
6368 If you know you are using buggy VFP11 hardware, you can
6369 enable this workaround by specifying the linker option
6370 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6371 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6372 vector mode (the latter also works for scalar code). The default is
6373 @samp{--vfp-denorm-fix=none}.
6375 If the workaround is enabled, instructions are scanned for
6376 potentially-troublesome sequences, and a veneer is created for each
6377 such sequence which may trigger the erratum. The veneer consists of the
6378 first instruction of the sequence and a branch back to the subsequent
6379 instruction. The original instruction is then replaced with a branch to
6380 the veneer. The extra cycles required to call and return from the veneer
6381 are sufficient to avoid the erratum in both the scalar and vector cases.
6383 @cindex ARM1176 erratum workaround
6384 @kindex --fix-arm1176
6385 @kindex --no-fix-arm1176
6386 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6387 in certain ARM1176 processors. The workaround is enabled by default if you
6388 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6389 unconditionally by specifying @samp{--no-fix-arm1176}.
6391 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6392 Programmer Advice Notice'' available on the ARM documentation website at:
6393 http://infocenter.arm.com/.
6395 @cindex NO_ENUM_SIZE_WARNING
6396 @kindex --no-enum-size-warning
6397 The @option{--no-enum-size-warning} switch prevents the linker from
6398 warning when linking object files that specify incompatible EABI
6399 enumeration size attributes. For example, with this switch enabled,
6400 linking of an object file using 32-bit enumeration values with another
6401 using enumeration values fitted into the smallest possible space will
6404 @cindex NO_WCHAR_SIZE_WARNING
6405 @kindex --no-wchar-size-warning
6406 The @option{--no-wchar-size-warning} switch prevents the linker from
6407 warning when linking object files that specify incompatible EABI
6408 @code{wchar_t} size attributes. For example, with this switch enabled,
6409 linking of an object file using 32-bit @code{wchar_t} values with another
6410 using 16-bit @code{wchar_t} values will not be diagnosed.
6413 @kindex --pic-veneer
6414 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6415 ARM/Thumb interworking veneers, even if the rest of the binary
6416 is not PIC. This avoids problems on uClinux targets where
6417 @samp{--emit-relocs} is used to generate relocatable binaries.
6419 @cindex STUB_GROUP_SIZE
6420 @kindex --stub-group-size=@var{N}
6421 The linker will automatically generate and insert small sequences of
6422 code into a linked ARM ELF executable whenever an attempt is made to
6423 perform a function call to a symbol that is too far away. The
6424 placement of these sequences of instructions - called stubs - is
6425 controlled by the command line option @option{--stub-group-size=N}.
6426 The placement is important because a poor choice can create a need for
6427 duplicate stubs, increasing the code size. The linker will try to
6428 group stubs together in order to reduce interruptions to the flow of
6429 code, but it needs guidance as to how big these groups should be and
6430 where they should be placed.
6432 The value of @samp{N}, the parameter to the
6433 @option{--stub-group-size=} option controls where the stub groups are
6434 placed. If it is negative then all stubs are placed after the first
6435 branch that needs them. If it is positive then the stubs can be
6436 placed either before or after the branches that need them. If the
6437 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6438 exactly where to place groups of stubs, using its built in heuristics.
6439 A value of @samp{N} greater than 1 (or smaller than -1) tells the
6440 linker that a single group of stubs can service at most @samp{N} bytes
6441 from the input sections.
6443 The default, if @option{--stub-group-size=} is not specified, is
6446 Farcalls stubs insertion is fully supported for the ARM-EABI target
6447 only, because it relies on object files properties not present
6461 @section @command{ld} and HPPA 32-bit ELF Support
6462 @cindex HPPA multiple sub-space stubs
6463 @kindex --multi-subspace
6464 When generating a shared library, @command{ld} will by default generate
6465 import stubs suitable for use with a single sub-space application.
6466 The @samp{--multi-subspace} switch causes @command{ld} to generate export
6467 stubs, and different (larger) import stubs suitable for use with
6468 multiple sub-spaces.
6470 @cindex HPPA stub grouping
6471 @kindex --stub-group-size=@var{N}
6472 Long branch stubs and import/export stubs are placed by @command{ld} in
6473 stub sections located between groups of input sections.
6474 @samp{--stub-group-size} specifies the maximum size of a group of input
6475 sections handled by one stub section. Since branch offsets are signed,
6476 a stub section may serve two groups of input sections, one group before
6477 the stub section, and one group after it. However, when using
6478 conditional branches that require stubs, it may be better (for branch
6479 prediction) that stub sections only serve one group of input sections.
6480 A negative value for @samp{N} chooses this scheme, ensuring that
6481 branches to stubs always use a negative offset. Two special values of
6482 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6483 @command{ld} to automatically size input section groups for the branch types
6484 detected, with the same behaviour regarding stub placement as other
6485 positive or negative values of @samp{N} respectively.
6487 Note that @samp{--stub-group-size} does not split input sections. A
6488 single input section larger than the group size specified will of course
6489 create a larger group (of one section). If input sections are too
6490 large, it may not be possible for a branch to reach its stub.
6503 @section @command{ld} and the Motorola 68K family
6505 @cindex Motorola 68K GOT generation
6506 @kindex --got=@var{type}
6507 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
6508 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
6509 @samp{target}. When @samp{target} is selected the linker chooses
6510 the default GOT generation scheme for the current target.
6511 @samp{single} tells the linker to generate a single GOT with
6512 entries only at non-negative offsets.
6513 @samp{negative} instructs the linker to generate a single GOT with
6514 entries at both negative and positive offsets. Not all environments
6516 @samp{multigot} allows the linker to generate several GOTs in the
6517 output file. All GOT references from a single input object
6518 file access the same GOT, but references from different input object
6519 files might access different GOTs. Not all environments support such GOTs.
6532 @section @code{ld} and MMIX
6533 For MMIX, there is a choice of generating @code{ELF} object files or
6534 @code{mmo} object files when linking. The simulator @code{mmix}
6535 understands the @code{mmo} format. The binutils @code{objcopy} utility
6536 can translate between the two formats.
6538 There is one special section, the @samp{.MMIX.reg_contents} section.
6539 Contents in this section is assumed to correspond to that of global
6540 registers, and symbols referring to it are translated to special symbols,
6541 equal to registers. In a final link, the start address of the
6542 @samp{.MMIX.reg_contents} section corresponds to the first allocated
6543 global register multiplied by 8. Register @code{$255} is not included in
6544 this section; it is always set to the program entry, which is at the
6545 symbol @code{Main} for @code{mmo} files.
6547 Global symbols with the prefix @code{__.MMIX.start.}, for example
6548 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
6549 The default linker script uses these to set the default start address
6552 Initial and trailing multiples of zero-valued 32-bit words in a section,
6553 are left out from an mmo file.
6566 @section @code{ld} and MSP430
6567 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
6568 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
6569 just pass @samp{-m help} option to the linker).
6571 @cindex MSP430 extra sections
6572 The linker will recognize some extra sections which are MSP430 specific:
6575 @item @samp{.vectors}
6576 Defines a portion of ROM where interrupt vectors located.
6578 @item @samp{.bootloader}
6579 Defines the bootloader portion of the ROM (if applicable). Any code
6580 in this section will be uploaded to the MPU.
6582 @item @samp{.infomem}
6583 Defines an information memory section (if applicable). Any code in
6584 this section will be uploaded to the MPU.
6586 @item @samp{.infomemnobits}
6587 This is the same as the @samp{.infomem} section except that any code
6588 in this section will not be uploaded to the MPU.
6590 @item @samp{.noinit}
6591 Denotes a portion of RAM located above @samp{.bss} section.
6593 The last two sections are used by gcc.
6607 @section @command{ld} and PowerPC 32-bit ELF Support
6608 @cindex PowerPC long branches
6609 @kindex --relax on PowerPC
6610 Branches on PowerPC processors are limited to a signed 26-bit
6611 displacement, which may result in @command{ld} giving
6612 @samp{relocation truncated to fit} errors with very large programs.
6613 @samp{--relax} enables the generation of trampolines that can access
6614 the entire 32-bit address space. These trampolines are inserted at
6615 section boundaries, so may not themselves be reachable if an input
6616 section exceeds 33M in size. You may combine @samp{-r} and
6617 @samp{--relax} to add trampolines in a partial link. In that case
6618 both branches to undefined symbols and inter-section branches are also
6619 considered potentially out of range, and trampolines inserted.
6621 @cindex PowerPC ELF32 options
6626 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
6627 generates code capable of using a newer PLT and GOT layout that has
6628 the security advantage of no executable section ever needing to be
6629 writable and no writable section ever being executable. PowerPC
6630 @command{ld} will generate this layout, including stubs to access the
6631 PLT, if all input files (including startup and static libraries) were
6632 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
6633 BSS PLT (and GOT layout) which can give slightly better performance.
6635 @kindex --secure-plt
6637 @command{ld} will use the new PLT and GOT layout if it is linking new
6638 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
6639 when linking non-PIC code. This option requests the new PLT and GOT
6640 layout. A warning will be given if some object file requires the old
6646 The new secure PLT and GOT are placed differently relative to other
6647 sections compared to older BSS PLT and GOT placement. The location of
6648 @code{.plt} must change because the new secure PLT is an initialized
6649 section while the old PLT is uninitialized. The reason for the
6650 @code{.got} change is more subtle: The new placement allows
6651 @code{.got} to be read-only in applications linked with
6652 @samp{-z relro -z now}. However, this placement means that
6653 @code{.sdata} cannot always be used in shared libraries, because the
6654 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
6655 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
6656 GCC doesn't use @code{.sdata} in shared libraries, so this option is
6657 really only useful for other compilers that may do so.
6659 @cindex PowerPC stub symbols
6660 @kindex --emit-stub-syms
6661 @item --emit-stub-syms
6662 This option causes @command{ld} to label linker stubs with a local
6663 symbol that encodes the stub type and destination.
6665 @cindex PowerPC TLS optimization
6666 @kindex --no-tls-optimize
6667 @item --no-tls-optimize
6668 PowerPC @command{ld} normally performs some optimization of code
6669 sequences used to access Thread-Local Storage. Use this option to
6670 disable the optimization.
6683 @node PowerPC64 ELF64
6684 @section @command{ld} and PowerPC64 64-bit ELF Support
6686 @cindex PowerPC64 ELF64 options
6688 @cindex PowerPC64 stub grouping
6689 @kindex --stub-group-size
6690 @item --stub-group-size
6691 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
6692 by @command{ld} in stub sections located between groups of input sections.
6693 @samp{--stub-group-size} specifies the maximum size of a group of input
6694 sections handled by one stub section. Since branch offsets are signed,
6695 a stub section may serve two groups of input sections, one group before
6696 the stub section, and one group after it. However, when using
6697 conditional branches that require stubs, it may be better (for branch
6698 prediction) that stub sections only serve one group of input sections.
6699 A negative value for @samp{N} chooses this scheme, ensuring that
6700 branches to stubs always use a negative offset. Two special values of
6701 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6702 @command{ld} to automatically size input section groups for the branch types
6703 detected, with the same behaviour regarding stub placement as other
6704 positive or negative values of @samp{N} respectively.
6706 Note that @samp{--stub-group-size} does not split input sections. A
6707 single input section larger than the group size specified will of course
6708 create a larger group (of one section). If input sections are too
6709 large, it may not be possible for a branch to reach its stub.
6711 @cindex PowerPC64 stub symbols
6712 @kindex --emit-stub-syms
6713 @item --emit-stub-syms
6714 This option causes @command{ld} to label linker stubs with a local
6715 symbol that encodes the stub type and destination.
6717 @cindex PowerPC64 dot symbols
6719 @kindex --no-dotsyms
6720 @item --dotsyms, --no-dotsyms
6721 These two options control how @command{ld} interprets version patterns
6722 in a version script. Older PowerPC64 compilers emitted both a
6723 function descriptor symbol with the same name as the function, and a
6724 code entry symbol with the name prefixed by a dot (@samp{.}). To
6725 properly version a function @samp{foo}, the version script thus needs
6726 to control both @samp{foo} and @samp{.foo}. The option
6727 @samp{--dotsyms}, on by default, automatically adds the required
6728 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
6731 @cindex PowerPC64 TLS optimization
6732 @kindex --no-tls-optimize
6733 @item --no-tls-optimize
6734 PowerPC64 @command{ld} normally performs some optimization of code
6735 sequences used to access Thread-Local Storage. Use this option to
6736 disable the optimization.
6738 @cindex PowerPC64 OPD optimization
6739 @kindex --no-opd-optimize
6740 @item --no-opd-optimize
6741 PowerPC64 @command{ld} normally removes @code{.opd} section entries
6742 corresponding to deleted link-once functions, or functions removed by
6743 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
6744 Use this option to disable @code{.opd} optimization.
6746 @cindex PowerPC64 OPD spacing
6747 @kindex --non-overlapping-opd
6748 @item --non-overlapping-opd
6749 Some PowerPC64 compilers have an option to generate compressed
6750 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
6751 the static chain pointer (unused in C) with the first word of the next
6752 entry. This option expands such entries to the full 24 bytes.
6754 @cindex PowerPC64 TOC optimization
6755 @kindex --no-toc-optimize
6756 @item --no-toc-optimize
6757 PowerPC64 @command{ld} normally removes unused @code{.toc} section
6758 entries. Such entries are detected by examining relocations that
6759 reference the TOC in code sections. A reloc in a deleted code section
6760 marks a TOC word as unneeded, while a reloc in a kept code section
6761 marks a TOC word as needed. Since the TOC may reference itself, TOC
6762 relocs are also examined. TOC words marked as both needed and
6763 unneeded will of course be kept. TOC words without any referencing
6764 reloc are assumed to be part of a multi-word entry, and are kept or
6765 discarded as per the nearest marked preceding word. This works
6766 reliably for compiler generated code, but may be incorrect if assembly
6767 code is used to insert TOC entries. Use this option to disable the
6770 @cindex PowerPC64 multi-TOC
6771 @kindex --no-multi-toc
6772 @item --no-multi-toc
6773 If given any toc option besides @code{-mcmodel=medium} or
6774 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
6776 entries are accessed with a 16-bit offset from r2. This limits the
6777 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
6778 grouping code sections such that each group uses less than 64K for its
6779 TOC entries, then inserts r2 adjusting stubs between inter-group
6780 calls. @command{ld} does not split apart input sections, so cannot
6781 help if a single input file has a @code{.toc} section that exceeds
6782 64K, most likely from linking multiple files with @command{ld -r}.
6783 Use this option to turn off this feature.
6785 @cindex PowerPC64 TOC sorting
6786 @kindex --no-toc-sort
6788 By default, @command{ld} sorts TOC sections so that those whose file
6789 happens to have a section called @code{.init} or @code{.fini} are
6790 placed first, followed by TOC sections referenced by code generated
6791 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
6792 referenced only by code generated with PowerPC64 gcc's
6793 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
6794 results in better TOC grouping for multi-TOC. Use this option to turn
6797 @cindex PowerPC64 PLT stub alignment
6799 @kindex --no-plt-align
6801 @itemx --no-plt-align
6802 Use these options to control whether individual PLT call stubs are
6803 aligned to a 32-byte boundary, or to the specified power of two
6804 boundary when using @code{--plt-align=}. By default PLT call stubs
6807 @cindex PowerPC64 PLT call stub static chain
6808 @kindex --plt-static-chain
6809 @kindex --no-plt-static-chain
6810 @item --plt-static-chain
6811 @itemx --no-plt-static-chain
6812 Use these options to control whether PLT call stubs load the static
6813 chain pointer (r11). @code{ld} defaults to not loading the static
6814 chain since there is never any need to do so on a PLT call.
6816 @cindex PowerPC64 PLT call stub thread safety
6817 @kindex --plt-thread-safe
6818 @kindex --no-plt-thread-safe
6819 @item --plt-thread-safe
6820 @itemx --no-thread-safe
6821 With power7's weakly ordered memory model, it is possible when using
6822 lazy binding for ld.so to update a plt entry in one thread and have
6823 another thread see the individual plt entry words update in the wrong
6824 order, despite ld.so carefully writing in the correct order and using
6825 memory write barriers. To avoid this we need some sort of read
6826 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
6827 looks for calls to commonly used functions that create threads, and if
6828 seen, adds the necessary barriers. Use these options to change the
6843 @section @command{ld} and SPU ELF Support
6845 @cindex SPU ELF options
6851 This option marks an executable as a PIC plugin module.
6853 @cindex SPU overlays
6854 @kindex --no-overlays
6856 Normally, @command{ld} recognizes calls to functions within overlay
6857 regions, and redirects such calls to an overlay manager via a stub.
6858 @command{ld} also provides a built-in overlay manager. This option
6859 turns off all this special overlay handling.
6861 @cindex SPU overlay stub symbols
6862 @kindex --emit-stub-syms
6863 @item --emit-stub-syms
6864 This option causes @command{ld} to label overlay stubs with a local
6865 symbol that encodes the stub type and destination.
6867 @cindex SPU extra overlay stubs
6868 @kindex --extra-overlay-stubs
6869 @item --extra-overlay-stubs
6870 This option causes @command{ld} to add overlay call stubs on all
6871 function calls out of overlay regions. Normally stubs are not added
6872 on calls to non-overlay regions.
6874 @cindex SPU local store size
6875 @kindex --local-store=lo:hi
6876 @item --local-store=lo:hi
6877 @command{ld} usually checks that a final executable for SPU fits in
6878 the address range 0 to 256k. This option may be used to change the
6879 range. Disable the check entirely with @option{--local-store=0:0}.
6882 @kindex --stack-analysis
6883 @item --stack-analysis
6884 SPU local store space is limited. Over-allocation of stack space
6885 unnecessarily limits space available for code and data, while
6886 under-allocation results in runtime failures. If given this option,
6887 @command{ld} will provide an estimate of maximum stack usage.
6888 @command{ld} does this by examining symbols in code sections to
6889 determine the extents of functions, and looking at function prologues
6890 for stack adjusting instructions. A call-graph is created by looking
6891 for relocations on branch instructions. The graph is then searched
6892 for the maximum stack usage path. Note that this analysis does not
6893 find calls made via function pointers, and does not handle recursion
6894 and other cycles in the call graph. Stack usage may be
6895 under-estimated if your code makes such calls. Also, stack usage for
6896 dynamic allocation, e.g. alloca, will not be detected. If a link map
6897 is requested, detailed information about each function's stack usage
6898 and calls will be given.
6901 @kindex --emit-stack-syms
6902 @item --emit-stack-syms
6903 This option, if given along with @option{--stack-analysis} will result
6904 in @command{ld} emitting stack sizing symbols for each function.
6905 These take the form @code{__stack_<function_name>} for global
6906 functions, and @code{__stack_<number>_<function_name>} for static
6907 functions. @code{<number>} is the section id in hex. The value of
6908 such symbols is the stack requirement for the corresponding function.
6909 The symbol size will be zero, type @code{STT_NOTYPE}, binding
6910 @code{STB_LOCAL}, and section @code{SHN_ABS}.
6924 @section @command{ld}'s Support for Various TI COFF Versions
6925 @cindex TI COFF versions
6926 @kindex --format=@var{version}
6927 The @samp{--format} switch allows selection of one of the various
6928 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
6929 also supported. The TI COFF versions also vary in header byte-order
6930 format; @command{ld} will read any version or byte order, but the output
6931 header format depends on the default specified by the specific target.
6944 @section @command{ld} and WIN32 (cygwin/mingw)
6946 This section describes some of the win32 specific @command{ld} issues.
6947 See @ref{Options,,Command Line Options} for detailed description of the
6948 command line options mentioned here.
6951 @cindex import libraries
6952 @item import libraries
6953 The standard Windows linker creates and uses so-called import
6954 libraries, which contains information for linking to dll's. They are
6955 regular static archives and are handled as any other static
6956 archive. The cygwin and mingw ports of @command{ld} have specific
6957 support for creating such libraries provided with the
6958 @samp{--out-implib} command line option.
6960 @item exporting DLL symbols
6961 @cindex exporting DLL symbols
6962 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6965 @item using auto-export functionality
6966 @cindex using auto-export functionality
6967 By default @command{ld} exports symbols with the auto-export functionality,
6968 which is controlled by the following command line options:
6971 @item --export-all-symbols [This is the default]
6972 @item --exclude-symbols
6973 @item --exclude-libs
6974 @item --exclude-modules-for-implib
6975 @item --version-script
6978 When auto-export is in operation, @command{ld} will export all the non-local
6979 (global and common) symbols it finds in a DLL, with the exception of a few
6980 symbols known to belong to the system's runtime and libraries. As it will
6981 often not be desirable to export all of a DLL's symbols, which may include
6982 private functions that are not part of any public interface, the command-line
6983 options listed above may be used to filter symbols out from the list for
6984 exporting. The @samp{--output-def} option can be used in order to see the
6985 final list of exported symbols with all exclusions taken into effect.
6987 If @samp{--export-all-symbols} is not given explicitly on the
6988 command line, then the default auto-export behavior will be @emph{disabled}
6989 if either of the following are true:
6992 @item A DEF file is used.
6993 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
6996 @item using a DEF file
6997 @cindex using a DEF file
6998 Another way of exporting symbols is using a DEF file. A DEF file is
6999 an ASCII file containing definitions of symbols which should be
7000 exported when a dll is created. Usually it is named @samp{<dll
7001 name>.def} and is added as any other object file to the linker's
7002 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7005 gcc -o <output> <objectfiles> <dll name>.def
7008 Using a DEF file turns off the normal auto-export behavior, unless the
7009 @samp{--export-all-symbols} option is also used.
7011 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7014 LIBRARY "xyz.dll" BASE=0x20000000
7020 another_foo = abc.dll.afoo
7026 This example defines a DLL with a non-default base address and seven
7027 symbols in the export table. The third exported symbol @code{_bar} is an
7028 alias for the second. The fourth symbol, @code{another_foo} is resolved
7029 by "forwarding" to another module and treating it as an alias for
7030 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7031 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7032 export library is an alias of @samp{foo}, which gets the string name
7033 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7034 symbol, which gets in export table the name @samp{var1}.
7036 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7037 name of the output DLL. If @samp{<name>} does not include a suffix,
7038 the default library suffix, @samp{.DLL} is appended.
7040 When the .DEF file is used to build an application, rather than a
7041 library, the @code{NAME <name>} command should be used instead of
7042 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7043 executable suffix, @samp{.EXE} is appended.
7045 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7046 specification @code{BASE = <number>} may be used to specify a
7047 non-default base address for the image.
7049 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7050 or they specify an empty string, the internal name is the same as the
7051 filename specified on the command line.
7053 The complete specification of an export symbol is:
7057 ( ( ( <name1> [ = <name2> ] )
7058 | ( <name1> = <module-name> . <external-name>))
7059 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7062 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7063 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7064 @samp{<name1>} as a "forward" alias for the symbol
7065 @samp{<external-name>} in the DLL @samp{<module-name>}.
7066 Optionally, the symbol may be exported by the specified ordinal
7067 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7068 string in import/export table for the symbol.
7070 The optional keywords that follow the declaration indicate:
7072 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7073 will still be exported by its ordinal alias (either the value specified
7074 by the .def specification or, otherwise, the value assigned by the
7075 linker). The symbol name, however, does remain visible in the import
7076 library (if any), unless @code{PRIVATE} is also specified.
7078 @code{DATA}: The symbol is a variable or object, rather than a function.
7079 The import lib will export only an indirect reference to @code{foo} as
7080 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7083 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7084 well as @code{_imp__foo} into the import library. Both refer to the
7085 read-only import address table's pointer to the variable, not to the
7086 variable itself. This can be dangerous. If the user code fails to add
7087 the @code{dllimport} attribute and also fails to explicitly add the
7088 extra indirection that the use of the attribute enforces, the
7089 application will behave unexpectedly.
7091 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7092 it into the static import library used to resolve imports at link time. The
7093 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7094 API at runtime or by by using the GNU ld extension of linking directly to
7095 the DLL without an import library.
7097 See ld/deffilep.y in the binutils sources for the full specification of
7098 other DEF file statements
7100 @cindex creating a DEF file
7101 While linking a shared dll, @command{ld} is able to create a DEF file
7102 with the @samp{--output-def <file>} command line option.
7104 @item Using decorations
7105 @cindex Using decorations
7106 Another way of marking symbols for export is to modify the source code
7107 itself, so that when building the DLL each symbol to be exported is
7111 __declspec(dllexport) int a_variable
7112 __declspec(dllexport) void a_function(int with_args)
7115 All such symbols will be exported from the DLL. If, however,
7116 any of the object files in the DLL contain symbols decorated in
7117 this way, then the normal auto-export behavior is disabled, unless
7118 the @samp{--export-all-symbols} option is also used.
7120 Note that object files that wish to access these symbols must @emph{not}
7121 decorate them with dllexport. Instead, they should use dllimport,
7125 __declspec(dllimport) int a_variable
7126 __declspec(dllimport) void a_function(int with_args)
7129 This complicates the structure of library header files, because
7130 when included by the library itself the header must declare the
7131 variables and functions as dllexport, but when included by client
7132 code the header must declare them as dllimport. There are a number
7133 of idioms that are typically used to do this; often client code can
7134 omit the __declspec() declaration completely. See
7135 @samp{--enable-auto-import} and @samp{automatic data imports} for more
7139 @cindex automatic data imports
7140 @item automatic data imports
7141 The standard Windows dll format supports data imports from dlls only
7142 by adding special decorations (dllimport/dllexport), which let the
7143 compiler produce specific assembler instructions to deal with this
7144 issue. This increases the effort necessary to port existing Un*x
7145 code to these platforms, especially for large
7146 c++ libraries and applications. The auto-import feature, which was
7147 initially provided by Paul Sokolovsky, allows one to omit the
7148 decorations to achieve a behavior that conforms to that on POSIX/Un*x
7149 platforms. This feature is enabled with the @samp{--enable-auto-import}
7150 command-line option, although it is enabled by default on cygwin/mingw.
7151 The @samp{--enable-auto-import} option itself now serves mainly to
7152 suppress any warnings that are ordinarily emitted when linked objects
7153 trigger the feature's use.
7155 auto-import of variables does not always work flawlessly without
7156 additional assistance. Sometimes, you will see this message
7158 "variable '<var>' can't be auto-imported. Please read the
7159 documentation for ld's @code{--enable-auto-import} for details."
7161 The @samp{--enable-auto-import} documentation explains why this error
7162 occurs, and several methods that can be used to overcome this difficulty.
7163 One of these methods is the @emph{runtime pseudo-relocs} feature, described
7166 @cindex runtime pseudo-relocation
7167 For complex variables imported from DLLs (such as structs or classes),
7168 object files typically contain a base address for the variable and an
7169 offset (@emph{addend}) within the variable--to specify a particular
7170 field or public member, for instance. Unfortunately, the runtime loader used
7171 in win32 environments is incapable of fixing these references at runtime
7172 without the additional information supplied by dllimport/dllexport decorations.
7173 The standard auto-import feature described above is unable to resolve these
7176 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
7177 be resolved without error, while leaving the task of adjusting the references
7178 themselves (with their non-zero addends) to specialized code provided by the
7179 runtime environment. Recent versions of the cygwin and mingw environments and
7180 compilers provide this runtime support; older versions do not. However, the
7181 support is only necessary on the developer's platform; the compiled result will
7182 run without error on an older system.
7184 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
7187 @cindex direct linking to a dll
7188 @item direct linking to a dll
7189 The cygwin/mingw ports of @command{ld} support the direct linking,
7190 including data symbols, to a dll without the usage of any import
7191 libraries. This is much faster and uses much less memory than does the
7192 traditional import library method, especially when linking large
7193 libraries or applications. When @command{ld} creates an import lib, each
7194 function or variable exported from the dll is stored in its own bfd, even
7195 though a single bfd could contain many exports. The overhead involved in
7196 storing, loading, and processing so many bfd's is quite large, and explains the
7197 tremendous time, memory, and storage needed to link against particularly
7198 large or complex libraries when using import libs.
7200 Linking directly to a dll uses no extra command-line switches other than
7201 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
7202 of names to match each library. All that is needed from the developer's
7203 perspective is an understanding of this search, in order to force ld to
7204 select the dll instead of an import library.
7207 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
7208 to find, in the first directory of its search path,
7220 before moving on to the next directory in the search path.
7222 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
7223 where @samp{<prefix>} is set by the @command{ld} option
7224 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
7225 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
7228 Other win32-based unix environments, such as mingw or pw32, may use other
7229 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
7230 was originally intended to help avoid name conflicts among dll's built for the
7231 various win32/un*x environments, so that (for example) two versions of a zlib dll
7232 could coexist on the same machine.
7234 The generic cygwin/mingw path layout uses a @samp{bin} directory for
7235 applications and dll's and a @samp{lib} directory for the import
7236 libraries (using cygwin nomenclature):
7242 libxxx.dll.a (in case of dll's)
7243 libxxx.a (in case of static archive)
7246 Linking directly to a dll without using the import library can be
7249 1. Use the dll directly by adding the @samp{bin} path to the link line
7251 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
7254 However, as the dll's often have version numbers appended to their names
7255 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
7256 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
7257 not versioned, and do not have this difficulty.
7259 2. Create a symbolic link from the dll to a file in the @samp{lib}
7260 directory according to the above mentioned search pattern. This
7261 should be used to avoid unwanted changes in the tools needed for
7265 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
7268 Then you can link without any make environment changes.
7271 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
7274 This technique also avoids the version number problems, because the following is
7281 libxxx.dll.a -> ../bin/cygxxx-5.dll
7284 Linking directly to a dll without using an import lib will work
7285 even when auto-import features are exercised, and even when
7286 @samp{--enable-runtime-pseudo-relocs} is used.
7288 Given the improvements in speed and memory usage, one might justifiably
7289 wonder why import libraries are used at all. There are three reasons:
7291 1. Until recently, the link-directly-to-dll functionality did @emph{not}
7292 work with auto-imported data.
7294 2. Sometimes it is necessary to include pure static objects within the
7295 import library (which otherwise contains only bfd's for indirection
7296 symbols that point to the exports of a dll). Again, the import lib
7297 for the cygwin kernel makes use of this ability, and it is not
7298 possible to do this without an import lib.
7300 3. Symbol aliases can only be resolved using an import lib. This is
7301 critical when linking against OS-supplied dll's (eg, the win32 API)
7302 in which symbols are usually exported as undecorated aliases of their
7303 stdcall-decorated assembly names.
7305 So, import libs are not going away. But the ability to replace
7306 true import libs with a simple symbolic link to (or a copy of)
7307 a dll, in many cases, is a useful addition to the suite of tools
7308 binutils makes available to the win32 developer. Given the
7309 massive improvements in memory requirements during linking, storage
7310 requirements, and linking speed, we expect that many developers
7311 will soon begin to use this feature whenever possible.
7313 @item symbol aliasing
7315 @item adding additional names
7316 Sometimes, it is useful to export symbols with additional names.
7317 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
7318 exported as @samp{_foo} by using special directives in the DEF file
7319 when creating the dll. This will affect also the optional created
7320 import library. Consider the following DEF file:
7323 LIBRARY "xyz.dll" BASE=0x61000000
7330 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
7332 Another method for creating a symbol alias is to create it in the
7333 source code using the "weak" attribute:
7336 void foo () @{ /* Do something. */; @}
7337 void _foo () __attribute__ ((weak, alias ("foo")));
7340 See the gcc manual for more information about attributes and weak
7343 @item renaming symbols
7344 Sometimes it is useful to rename exports. For instance, the cygwin
7345 kernel does this regularly. A symbol @samp{_foo} can be exported as
7346 @samp{foo} but not as @samp{_foo} by using special directives in the
7347 DEF file. (This will also affect the import library, if it is
7348 created). In the following example:
7351 LIBRARY "xyz.dll" BASE=0x61000000
7357 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
7361 Note: using a DEF file disables the default auto-export behavior,
7362 unless the @samp{--export-all-symbols} command line option is used.
7363 If, however, you are trying to rename symbols, then you should list
7364 @emph{all} desired exports in the DEF file, including the symbols
7365 that are not being renamed, and do @emph{not} use the
7366 @samp{--export-all-symbols} option. If you list only the
7367 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
7368 to handle the other symbols, then the both the new names @emph{and}
7369 the original names for the renamed symbols will be exported.
7370 In effect, you'd be aliasing those symbols, not renaming them,
7371 which is probably not what you wanted.
7373 @cindex weak externals
7374 @item weak externals
7375 The Windows object format, PE, specifies a form of weak symbols called
7376 weak externals. When a weak symbol is linked and the symbol is not
7377 defined, the weak symbol becomes an alias for some other symbol. There
7378 are three variants of weak externals:
7380 @item Definition is searched for in objects and libraries, historically
7381 called lazy externals.
7382 @item Definition is searched for only in other objects, not in libraries.
7383 This form is not presently implemented.
7384 @item No search; the symbol is an alias. This form is not presently
7387 As a GNU extension, weak symbols that do not specify an alternate symbol
7388 are supported. If the symbol is undefined when linking, the symbol
7389 uses a default value.
7391 @cindex aligned common symbols
7392 @item aligned common symbols
7393 As a GNU extension to the PE file format, it is possible to specify the
7394 desired alignment for a common symbol. This information is conveyed from
7395 the assembler or compiler to the linker by means of GNU-specific commands
7396 carried in the object file's @samp{.drectve} section, which are recognized
7397 by @command{ld} and respected when laying out the common symbols. Native
7398 tools will be able to process object files employing this GNU extension,
7399 but will fail to respect the alignment instructions, and may issue noisy
7400 warnings about unknown linker directives.
7414 @section @code{ld} and Xtensa Processors
7416 @cindex Xtensa processors
7417 The default @command{ld} behavior for Xtensa processors is to interpret
7418 @code{SECTIONS} commands so that lists of explicitly named sections in a
7419 specification with a wildcard file will be interleaved when necessary to
7420 keep literal pools within the range of PC-relative load offsets. For
7421 example, with the command:
7433 @command{ld} may interleave some of the @code{.literal}
7434 and @code{.text} sections from different object files to ensure that the
7435 literal pools are within the range of PC-relative load offsets. A valid
7436 interleaving might place the @code{.literal} sections from an initial
7437 group of files followed by the @code{.text} sections of that group of
7438 files. Then, the @code{.literal} sections from the rest of the files
7439 and the @code{.text} sections from the rest of the files would follow.
7441 @cindex @option{--relax} on Xtensa
7442 @cindex relaxing on Xtensa
7443 Relaxation is enabled by default for the Xtensa version of @command{ld} and
7444 provides two important link-time optimizations. The first optimization
7445 is to combine identical literal values to reduce code size. A redundant
7446 literal will be removed and all the @code{L32R} instructions that use it
7447 will be changed to reference an identical literal, as long as the
7448 location of the replacement literal is within the offset range of all
7449 the @code{L32R} instructions. The second optimization is to remove
7450 unnecessary overhead from assembler-generated ``longcall'' sequences of
7451 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
7452 range of direct @code{CALL@var{n}} instructions.
7454 For each of these cases where an indirect call sequence can be optimized
7455 to a direct call, the linker will change the @code{CALLX@var{n}}
7456 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
7457 instruction, and remove the literal referenced by the @code{L32R}
7458 instruction if it is not used for anything else. Removing the
7459 @code{L32R} instruction always reduces code size but can potentially
7460 hurt performance by changing the alignment of subsequent branch targets.
7461 By default, the linker will always preserve alignments, either by
7462 switching some instructions between 24-bit encodings and the equivalent
7463 density instructions or by inserting a no-op in place of the @code{L32R}
7464 instruction that was removed. If code size is more important than
7465 performance, the @option{--size-opt} option can be used to prevent the
7466 linker from widening density instructions or inserting no-ops, except in
7467 a few cases where no-ops are required for correctness.
7469 The following Xtensa-specific command-line options can be used to
7472 @cindex Xtensa options
7475 When optimizing indirect calls to direct calls, optimize for code size
7476 more than performance. With this option, the linker will not insert
7477 no-ops or widen density instructions to preserve branch target
7478 alignment. There may still be some cases where no-ops are required to
7479 preserve the correctness of the code.
7487 @ifclear SingleFormat
7492 @cindex object file management
7493 @cindex object formats available
7495 The linker accesses object and archive files using the BFD libraries.
7496 These libraries allow the linker to use the same routines to operate on
7497 object files whatever the object file format. A different object file
7498 format can be supported simply by creating a new BFD back end and adding
7499 it to the library. To conserve runtime memory, however, the linker and
7500 associated tools are usually configured to support only a subset of the
7501 object file formats available. You can use @code{objdump -i}
7502 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
7503 list all the formats available for your configuration.
7505 @cindex BFD requirements
7506 @cindex requirements for BFD
7507 As with most implementations, BFD is a compromise between
7508 several conflicting requirements. The major factor influencing
7509 BFD design was efficiency: any time used converting between
7510 formats is time which would not have been spent had BFD not
7511 been involved. This is partly offset by abstraction payback; since
7512 BFD simplifies applications and back ends, more time and care
7513 may be spent optimizing algorithms for a greater speed.
7515 One minor artifact of the BFD solution which you should bear in
7516 mind is the potential for information loss. There are two places where
7517 useful information can be lost using the BFD mechanism: during
7518 conversion and during output. @xref{BFD information loss}.
7521 * BFD outline:: How it works: an outline of BFD
7525 @section How It Works: An Outline of BFD
7526 @cindex opening object files
7527 @include bfdsumm.texi
7530 @node Reporting Bugs
7531 @chapter Reporting Bugs
7532 @cindex bugs in @command{ld}
7533 @cindex reporting bugs in @command{ld}
7535 Your bug reports play an essential role in making @command{ld} reliable.
7537 Reporting a bug may help you by bringing a solution to your problem, or
7538 it may not. But in any case the principal function of a bug report is
7539 to help the entire community by making the next version of @command{ld}
7540 work better. Bug reports are your contribution to the maintenance of
7543 In order for a bug report to serve its purpose, you must include the
7544 information that enables us to fix the bug.
7547 * Bug Criteria:: Have you found a bug?
7548 * Bug Reporting:: How to report bugs
7552 @section Have You Found a Bug?
7553 @cindex bug criteria
7555 If you are not sure whether you have found a bug, here are some guidelines:
7558 @cindex fatal signal
7559 @cindex linker crash
7560 @cindex crash of linker
7562 If the linker gets a fatal signal, for any input whatever, that is a
7563 @command{ld} bug. Reliable linkers never crash.
7565 @cindex error on valid input
7567 If @command{ld} produces an error message for valid input, that is a bug.
7569 @cindex invalid input
7571 If @command{ld} does not produce an error message for invalid input, that
7572 may be a bug. In the general case, the linker can not verify that
7573 object files are correct.
7576 If you are an experienced user of linkers, your suggestions for
7577 improvement of @command{ld} are welcome in any case.
7581 @section How to Report Bugs
7583 @cindex @command{ld} bugs, reporting
7585 A number of companies and individuals offer support for @sc{gnu}
7586 products. If you obtained @command{ld} from a support organization, we
7587 recommend you contact that organization first.
7589 You can find contact information for many support companies and
7590 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
7594 Otherwise, send bug reports for @command{ld} to
7598 The fundamental principle of reporting bugs usefully is this:
7599 @strong{report all the facts}. If you are not sure whether to state a
7600 fact or leave it out, state it!
7602 Often people omit facts because they think they know what causes the
7603 problem and assume that some details do not matter. Thus, you might
7604 assume that the name of a symbol you use in an example does not
7605 matter. Well, probably it does not, but one cannot be sure. Perhaps
7606 the bug is a stray memory reference which happens to fetch from the
7607 location where that name is stored in memory; perhaps, if the name
7608 were different, the contents of that location would fool the linker
7609 into doing the right thing despite the bug. Play it safe and give a
7610 specific, complete example. That is the easiest thing for you to do,
7611 and the most helpful.
7613 Keep in mind that the purpose of a bug report is to enable us to fix
7614 the bug if it is new to us. Therefore, always write your bug reports
7615 on the assumption that the bug has not been reported previously.
7617 Sometimes people give a few sketchy facts and ask, ``Does this ring a
7618 bell?'' This cannot help us fix a bug, so it is basically useless. We
7619 respond by asking for enough details to enable us to investigate.
7620 You might as well expedite matters by sending them to begin with.
7622 To enable us to fix the bug, you should include all these things:
7626 The version of @command{ld}. @command{ld} announces it if you start it with
7627 the @samp{--version} argument.
7629 Without this, we will not know whether there is any point in looking for
7630 the bug in the current version of @command{ld}.
7633 Any patches you may have applied to the @command{ld} source, including any
7634 patches made to the @code{BFD} library.
7637 The type of machine you are using, and the operating system name and
7641 What compiler (and its version) was used to compile @command{ld}---e.g.
7645 The command arguments you gave the linker to link your example and
7646 observe the bug. To guarantee you will not omit something important,
7647 list them all. A copy of the Makefile (or the output from make) is
7650 If we were to try to guess the arguments, we would probably guess wrong
7651 and then we might not encounter the bug.
7654 A complete input file, or set of input files, that will reproduce the
7655 bug. It is generally most helpful to send the actual object files
7656 provided that they are reasonably small. Say no more than 10K. For
7657 bigger files you can either make them available by FTP or HTTP or else
7658 state that you are willing to send the object file(s) to whomever
7659 requests them. (Note - your email will be going to a mailing list, so
7660 we do not want to clog it up with large attachments). But small
7661 attachments are best.
7663 If the source files were assembled using @code{gas} or compiled using
7664 @code{gcc}, then it may be OK to send the source files rather than the
7665 object files. In this case, be sure to say exactly what version of
7666 @code{gas} or @code{gcc} was used to produce the object files. Also say
7667 how @code{gas} or @code{gcc} were configured.
7670 A description of what behavior you observe that you believe is
7671 incorrect. For example, ``It gets a fatal signal.''
7673 Of course, if the bug is that @command{ld} gets a fatal signal, then we
7674 will certainly notice it. But if the bug is incorrect output, we might
7675 not notice unless it is glaringly wrong. You might as well not give us
7676 a chance to make a mistake.
7678 Even if the problem you experience is a fatal signal, you should still
7679 say so explicitly. Suppose something strange is going on, such as, your
7680 copy of @command{ld} is out of sync, or you have encountered a bug in the
7681 C library on your system. (This has happened!) Your copy might crash
7682 and ours would not. If you told us to expect a crash, then when ours
7683 fails to crash, we would know that the bug was not happening for us. If
7684 you had not told us to expect a crash, then we would not be able to draw
7685 any conclusion from our observations.
7688 If you wish to suggest changes to the @command{ld} source, send us context
7689 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
7690 @samp{-p} option. Always send diffs from the old file to the new file.
7691 If you even discuss something in the @command{ld} source, refer to it by
7692 context, not by line number.
7694 The line numbers in our development sources will not match those in your
7695 sources. Your line numbers would convey no useful information to us.
7698 Here are some things that are not necessary:
7702 A description of the envelope of the bug.
7704 Often people who encounter a bug spend a lot of time investigating
7705 which changes to the input file will make the bug go away and which
7706 changes will not affect it.
7708 This is often time consuming and not very useful, because the way we
7709 will find the bug is by running a single example under the debugger
7710 with breakpoints, not by pure deduction from a series of examples.
7711 We recommend that you save your time for something else.
7713 Of course, if you can find a simpler example to report @emph{instead}
7714 of the original one, that is a convenience for us. Errors in the
7715 output will be easier to spot, running under the debugger will take
7716 less time, and so on.
7718 However, simplification is not vital; if you do not want to do this,
7719 report the bug anyway and send us the entire test case you used.
7722 A patch for the bug.
7724 A patch for the bug does help us if it is a good one. But do not omit
7725 the necessary information, such as the test case, on the assumption that
7726 a patch is all we need. We might see problems with your patch and decide
7727 to fix the problem another way, or we might not understand it at all.
7729 Sometimes with a program as complicated as @command{ld} it is very hard to
7730 construct an example that will make the program follow a certain path
7731 through the code. If you do not send us the example, we will not be
7732 able to construct one, so we will not be able to verify that the bug is
7735 And if we cannot understand what bug you are trying to fix, or why your
7736 patch should be an improvement, we will not install it. A test case will
7737 help us to understand.
7740 A guess about what the bug is or what it depends on.
7742 Such guesses are usually wrong. Even we cannot guess right about such
7743 things without first using the debugger to find the facts.
7747 @appendix MRI Compatible Script Files
7748 @cindex MRI compatibility
7749 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
7750 linker, @command{ld} can use MRI compatible linker scripts as an
7751 alternative to the more general-purpose linker scripting language
7752 described in @ref{Scripts}. MRI compatible linker scripts have a much
7753 simpler command set than the scripting language otherwise used with
7754 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
7755 linker commands; these commands are described here.
7757 In general, MRI scripts aren't of much use with the @code{a.out} object
7758 file format, since it only has three sections and MRI scripts lack some
7759 features to make use of them.
7761 You can specify a file containing an MRI-compatible script using the
7762 @samp{-c} command-line option.
7764 Each command in an MRI-compatible script occupies its own line; each
7765 command line starts with the keyword that identifies the command (though
7766 blank lines are also allowed for punctuation). If a line of an
7767 MRI-compatible script begins with an unrecognized keyword, @command{ld}
7768 issues a warning message, but continues processing the script.
7770 Lines beginning with @samp{*} are comments.
7772 You can write these commands using all upper-case letters, or all
7773 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
7774 The following list shows only the upper-case form of each command.
7777 @cindex @code{ABSOLUTE} (MRI)
7778 @item ABSOLUTE @var{secname}
7779 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
7780 Normally, @command{ld} includes in the output file all sections from all
7781 the input files. However, in an MRI-compatible script, you can use the
7782 @code{ABSOLUTE} command to restrict the sections that will be present in
7783 your output program. If the @code{ABSOLUTE} command is used at all in a
7784 script, then only the sections named explicitly in @code{ABSOLUTE}
7785 commands will appear in the linker output. You can still use other
7786 input sections (whatever you select on the command line, or using
7787 @code{LOAD}) to resolve addresses in the output file.
7789 @cindex @code{ALIAS} (MRI)
7790 @item ALIAS @var{out-secname}, @var{in-secname}
7791 Use this command to place the data from input section @var{in-secname}
7792 in a section called @var{out-secname} in the linker output file.
7794 @var{in-secname} may be an integer.
7796 @cindex @code{ALIGN} (MRI)
7797 @item ALIGN @var{secname} = @var{expression}
7798 Align the section called @var{secname} to @var{expression}. The
7799 @var{expression} should be a power of two.
7801 @cindex @code{BASE} (MRI)
7802 @item BASE @var{expression}
7803 Use the value of @var{expression} as the lowest address (other than
7804 absolute addresses) in the output file.
7806 @cindex @code{CHIP} (MRI)
7807 @item CHIP @var{expression}
7808 @itemx CHIP @var{expression}, @var{expression}
7809 This command does nothing; it is accepted only for compatibility.
7811 @cindex @code{END} (MRI)
7813 This command does nothing whatever; it's only accepted for compatibility.
7815 @cindex @code{FORMAT} (MRI)
7816 @item FORMAT @var{output-format}
7817 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
7818 language, but restricted to one of these output formats:
7822 S-records, if @var{output-format} is @samp{S}
7825 IEEE, if @var{output-format} is @samp{IEEE}
7828 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
7832 @cindex @code{LIST} (MRI)
7833 @item LIST @var{anything}@dots{}
7834 Print (to the standard output file) a link map, as produced by the
7835 @command{ld} command-line option @samp{-M}.
7837 The keyword @code{LIST} may be followed by anything on the
7838 same line, with no change in its effect.
7840 @cindex @code{LOAD} (MRI)
7841 @item LOAD @var{filename}
7842 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
7843 Include one or more object file @var{filename} in the link; this has the
7844 same effect as specifying @var{filename} directly on the @command{ld}
7847 @cindex @code{NAME} (MRI)
7848 @item NAME @var{output-name}
7849 @var{output-name} is the name for the program produced by @command{ld}; the
7850 MRI-compatible command @code{NAME} is equivalent to the command-line
7851 option @samp{-o} or the general script language command @code{OUTPUT}.
7853 @cindex @code{ORDER} (MRI)
7854 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
7855 @itemx ORDER @var{secname} @var{secname} @var{secname}
7856 Normally, @command{ld} orders the sections in its output file in the
7857 order in which they first appear in the input files. In an MRI-compatible
7858 script, you can override this ordering with the @code{ORDER} command. The
7859 sections you list with @code{ORDER} will appear first in your output
7860 file, in the order specified.
7862 @cindex @code{PUBLIC} (MRI)
7863 @item PUBLIC @var{name}=@var{expression}
7864 @itemx PUBLIC @var{name},@var{expression}
7865 @itemx PUBLIC @var{name} @var{expression}
7866 Supply a value (@var{expression}) for external symbol
7867 @var{name} used in the linker input files.
7869 @cindex @code{SECT} (MRI)
7870 @item SECT @var{secname}, @var{expression}
7871 @itemx SECT @var{secname}=@var{expression}
7872 @itemx SECT @var{secname} @var{expression}
7873 You can use any of these three forms of the @code{SECT} command to
7874 specify the start address (@var{expression}) for section @var{secname}.
7875 If you have more than one @code{SECT} statement for the same
7876 @var{secname}, only the @emph{first} sets the start address.
7879 @node GNU Free Documentation License
7880 @appendix GNU Free Documentation License
7884 @unnumbered LD Index
7889 % I think something like @@colophon should be in texinfo. In the
7891 \long\def\colophon{\hbox to0pt{}\vfill
7892 \centerline{The body of this manual is set in}
7893 \centerline{\fontname\tenrm,}
7894 \centerline{with headings in {\bf\fontname\tenbf}}
7895 \centerline{and examples in {\tt\fontname\tentt}.}
7896 \centerline{{\it\fontname\tenit\/} and}
7897 \centerline{{\sl\fontname\tensl\/}}
7898 \centerline{are used for emphasis.}\vfill}
7900 % Blame: doc@@cygnus.com, 28mar91.