3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
7 @include configdoc.texi
8 @c (configdoc.texi is generated by the Makefile)
14 @macro gcctabopt{body}
20 @c Configure for the generation of man pages
45 * Ld: (ld). The GNU linker.
51 This file documents the @sc{gnu} linker LD
52 @ifset VERSION_PACKAGE
53 @value{VERSION_PACKAGE}
55 version @value{VERSION}.
57 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
58 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
60 Permission is granted to copy, distribute and/or modify this document
61 under the terms of the GNU Free Documentation License, Version 1.1
62 or any later version published by the Free Software Foundation;
63 with no Invariant Sections, with no Front-Cover Texts, and with no
64 Back-Cover Texts. A copy of the license is included in the
65 section entitled ``GNU Free Documentation License''.
69 @setchapternewpage odd
70 @settitle The GNU linker
75 @ifset VERSION_PACKAGE
76 @subtitle @value{VERSION_PACKAGE}
78 @subtitle Version @value{VERSION}
79 @author Steve Chamberlain
80 @author Ian Lance Taylor
85 \hfill Red Hat Inc\par
86 \hfill nickc\@credhat.com, doc\@redhat.com\par
87 \hfill {\it The GNU linker}\par
88 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
90 \global\parindent=0pt % Steve likes it this way.
93 @vskip 0pt plus 1filll
94 @c man begin COPYRIGHT
95 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
96 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
98 Permission is granted to copy, distribute and/or modify this document
99 under the terms of the GNU Free Documentation License, Version 1.1
100 or any later version published by the Free Software Foundation;
101 with no Invariant Sections, with no Front-Cover Texts, and with no
102 Back-Cover Texts. A copy of the license is included in the
103 section entitled ``GNU Free Documentation License''.
109 @c FIXME: Talk about importance of *order* of args, cmds to linker!
114 This file documents the @sc{gnu} linker ld
115 @ifset VERSION_PACKAGE
116 @value{VERSION_PACKAGE}
118 version @value{VERSION}.
120 This document is distributed under the terms of the GNU Free
121 Documentation License. A copy of the license is included in the
122 section entitled ``GNU Free Documentation License''.
125 * Overview:: Overview
126 * Invocation:: Invocation
127 * Scripts:: Linker Scripts
129 * Machine Dependent:: Machine Dependent Features
133 * H8/300:: ld and the H8/300
136 * Renesas:: ld and other Renesas micros
139 * i960:: ld and the Intel 960 family
142 * ARM:: ld and the ARM family
145 * Blackfin:: ld and the Blackfin family
148 * HPPA ELF32:: ld and HPPA 32-bit ELF
151 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
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,--startgroup foo.o bar.o -Wl,--endgroup
344 This is important, because otherwise the compiler driver program may
345 silently drop the linker options, resulting in a bad link.
347 Here is a table of the generic command line switches accepted by the GNU
351 @include at-file.texi
353 @kindex -a@var{keyword}
354 @item -a@var{keyword}
355 This option is supported for HP/UX compatibility. The @var{keyword}
356 argument must be one of the strings @samp{archive}, @samp{shared}, or
357 @samp{default}. @samp{-aarchive} is functionally equivalent to
358 @samp{-Bstatic}, and the other two keywords are functionally equivalent
359 to @samp{-Bdynamic}. This option may be used any number of times.
362 @cindex architectures
364 @item -A@var{architecture}
365 @kindex --architecture=@var{arch}
366 @itemx --architecture=@var{architecture}
367 In the current release of @command{ld}, this option is useful only for the
368 Intel 960 family of architectures. In that @command{ld} configuration, the
369 @var{architecture} argument identifies the particular architecture in
370 the 960 family, enabling some safeguards and modifying the
371 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
372 family}, for details.
374 Future releases of @command{ld} may support similar functionality for
375 other architecture families.
378 @ifclear SingleFormat
379 @cindex binary input format
380 @kindex -b @var{format}
381 @kindex --format=@var{format}
384 @item -b @var{input-format}
385 @itemx --format=@var{input-format}
386 @command{ld} may be configured to support more than one kind of object
387 file. If your @command{ld} is configured this way, you can use the
388 @samp{-b} option to specify the binary format for input object files
389 that follow this option on the command line. Even when @command{ld} is
390 configured to support alternative object formats, you don't usually need
391 to specify this, as @command{ld} should be configured to expect as a
392 default input format the most usual format on each machine.
393 @var{input-format} is a text string, the name of a particular format
394 supported by the BFD libraries. (You can list the available binary
395 formats with @samp{objdump -i}.)
398 You may want to use this option if you are linking files with an unusual
399 binary format. You can also use @samp{-b} to switch formats explicitly (when
400 linking object files of different formats), by including
401 @samp{-b @var{input-format}} before each group of object files in a
404 The default format is taken from the environment variable
409 You can also define the input format from a script, using the command
412 see @ref{Format Commands}.
416 @kindex -c @var{MRI-cmdfile}
417 @kindex --mri-script=@var{MRI-cmdfile}
418 @cindex compatibility, MRI
419 @item -c @var{MRI-commandfile}
420 @itemx --mri-script=@var{MRI-commandfile}
421 For compatibility with linkers produced by MRI, @command{ld} accepts script
422 files written in an alternate, restricted command language, described in
424 @ref{MRI,,MRI Compatible Script Files}.
427 the MRI Compatible Script Files section of GNU ld documentation.
429 Introduce MRI script files with
430 the option @samp{-c}; use the @samp{-T} option to run linker
431 scripts written in the general-purpose @command{ld} scripting language.
432 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
433 specified by any @samp{-L} options.
435 @cindex common allocation
442 These three options are equivalent; multiple forms are supported for
443 compatibility with other linkers. They assign space to common symbols
444 even if a relocatable output file is specified (with @samp{-r}). The
445 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
446 @xref{Miscellaneous Commands}.
448 @cindex entry point, from command line
449 @kindex -e @var{entry}
450 @kindex --entry=@var{entry}
452 @itemx --entry=@var{entry}
453 Use @var{entry} as the explicit symbol for beginning execution of your
454 program, rather than the default entry point. If there is no symbol
455 named @var{entry}, the linker will try to parse @var{entry} as a number,
456 and use that as the entry address (the number will be interpreted in
457 base 10; you may use a leading @samp{0x} for base 16, or a leading
458 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
459 and other ways of specifying the entry point.
461 @kindex --exclude-libs
462 @item --exclude-libs @var{lib},@var{lib},...
463 Specifies a list of archive libraries from which symbols should not be automatically
464 exported. The library names may be delimited by commas or colons. Specifying
465 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
466 automatic export. This option is available only for the i386 PE targeted
467 port of the linker and for ELF targeted ports. For i386 PE, symbols
468 explicitly listed in a .def file are still exported, regardless of this
469 option. For ELF targeted ports, symbols affected by this option will
470 be treated as hidden.
472 @cindex dynamic symbol table
474 @kindex --export-dynamic
476 @itemx --export-dynamic
477 When creating a dynamically linked executable, add all symbols to the
478 dynamic symbol table. The dynamic symbol table is the set of symbols
479 which are visible from dynamic objects at run time.
481 If you do not use this option, the dynamic symbol table will normally
482 contain only those symbols which are referenced by some dynamic object
483 mentioned in the link.
485 If you use @code{dlopen} to load a dynamic object which needs to refer
486 back to the symbols defined by the program, rather than some other
487 dynamic object, then you will probably need to use this option when
488 linking the program itself.
490 You can also use the dynamic list to control what symbols should
491 be added to the dynamic symbol table if the output format supports it.
492 See the description of @samp{--dynamic-list}.
494 @ifclear SingleFormat
495 @cindex big-endian objects
499 Link big-endian objects. This affects the default output format.
501 @cindex little-endian objects
504 Link little-endian objects. This affects the default output format.
510 @itemx --auxiliary @var{name}
511 When creating an ELF shared object, set the internal DT_AUXILIARY field
512 to the specified name. This tells the dynamic linker that the symbol
513 table of the shared object should be used as an auxiliary filter on the
514 symbol table of the shared object @var{name}.
516 If you later link a program against this filter object, then, when you
517 run the program, the dynamic linker will see the DT_AUXILIARY field. If
518 the dynamic linker resolves any symbols from the filter object, it will
519 first check whether there is a definition in the shared object
520 @var{name}. If there is one, it will be used instead of the definition
521 in the filter object. The shared object @var{name} need not exist.
522 Thus the shared object @var{name} may be used to provide an alternative
523 implementation of certain functions, perhaps for debugging or for
524 machine specific performance.
526 This option may be specified more than once. The DT_AUXILIARY entries
527 will be created in the order in which they appear on the command line.
532 @itemx --filter @var{name}
533 When creating an ELF shared object, set the internal DT_FILTER field to
534 the specified name. This tells the dynamic linker that the symbol table
535 of the shared object which is being created should be used as a filter
536 on the symbol table of the shared object @var{name}.
538 If you later link a program against this filter object, then, when you
539 run the program, the dynamic linker will see the DT_FILTER field. The
540 dynamic linker will resolve symbols according to the symbol table of the
541 filter object as usual, but it will actually link to the definitions
542 found in the shared object @var{name}. Thus the filter object can be
543 used to select a subset of the symbols provided by the object
546 Some older linkers used the @option{-F} option throughout a compilation
547 toolchain for specifying object-file format for both input and output
549 @ifclear SingleFormat
550 The @sc{gnu} linker uses other mechanisms for this purpose: the
551 @option{-b}, @option{--format}, @option{--oformat} options, the
552 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
553 environment variable.
555 The @sc{gnu} linker will ignore the @option{-F} option when not
556 creating an ELF shared object.
558 @cindex finalization function
560 @item -fini @var{name}
561 When creating an ELF executable or shared object, call NAME when the
562 executable or shared object is unloaded, by setting DT_FINI to the
563 address of the function. By default, the linker uses @code{_fini} as
564 the function to call.
568 Ignored. Provided for compatibility with other tools.
574 @itemx --gpsize=@var{value}
575 Set the maximum size of objects to be optimized using the GP register to
576 @var{size}. This is only meaningful for object file formats such as
577 MIPS ECOFF which supports putting large and small objects into different
578 sections. This is ignored for other object file formats.
580 @cindex runtime library name
582 @kindex -soname=@var{name}
584 @itemx -soname=@var{name}
585 When creating an ELF shared object, set the internal DT_SONAME field to
586 the specified name. When an executable is linked with a shared object
587 which has a DT_SONAME field, then when the executable is run the dynamic
588 linker will attempt to load the shared object specified by the DT_SONAME
589 field rather than the using the file name given to the linker.
592 @cindex incremental link
594 Perform an incremental link (same as option @samp{-r}).
596 @cindex initialization function
598 @item -init @var{name}
599 When creating an ELF executable or shared object, call NAME when the
600 executable or shared object is loaded, by setting DT_INIT to the address
601 of the function. By default, the linker uses @code{_init} as the
604 @cindex archive files, from cmd line
605 @kindex -l@var{namespec}
606 @kindex --library=@var{namespec}
607 @item -l@var{namespec}
608 @itemx --library=@var{namespec}
609 Add the archive or object file specified by @var{namespec} to the
610 list of files to link. This option may be used any number of times.
611 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
612 will search the library path for a file called @var{filename}, otherise it
613 will search the library path for a file called @file{lib@var{namespec}.a}.
615 On systems which support shared libraries, @command{ld} may also search for
616 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
617 and SunOS systems, @command{ld} will search a directory for a library
618 called @file{lib@var{namespec}.so} before searching for one called
619 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
620 indicates a shared library.) Note that this behavior does not apply
621 to @file{:@var{filename}}, which always specifies a file called
624 The linker will search an archive only once, at the location where it is
625 specified on the command line. If the archive defines a symbol which
626 was undefined in some object which appeared before the archive on the
627 command line, the linker will include the appropriate file(s) from the
628 archive. However, an undefined symbol in an object appearing later on
629 the command line will not cause the linker to search the archive again.
631 See the @option{-(} option for a way to force the linker to search
632 archives multiple times.
634 You may list the same archive multiple times on the command line.
637 This type of archive searching is standard for Unix linkers. However,
638 if you are using @command{ld} on AIX, note that it is different from the
639 behaviour of the AIX linker.
642 @cindex search directory, from cmd line
644 @kindex --library-path=@var{dir}
645 @item -L@var{searchdir}
646 @itemx --library-path=@var{searchdir}
647 Add path @var{searchdir} to the list of paths that @command{ld} will search
648 for archive libraries and @command{ld} control scripts. You may use this
649 option any number of times. The directories are searched in the order
650 in which they are specified on the command line. Directories specified
651 on the command line are searched before the default directories. All
652 @option{-L} options apply to all @option{-l} options, regardless of the
653 order in which the options appear.
655 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
656 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
659 The default set of paths searched (without being specified with
660 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
661 some cases also on how it was configured. @xref{Environment}.
664 The paths can also be specified in a link script with the
665 @code{SEARCH_DIR} command. Directories specified this way are searched
666 at the point in which the linker script appears in the command line.
669 @kindex -m @var{emulation}
670 @item -m@var{emulation}
671 Emulate the @var{emulation} linker. You can list the available
672 emulations with the @samp{--verbose} or @samp{-V} options.
674 If the @samp{-m} option is not used, the emulation is taken from the
675 @code{LDEMULATION} environment variable, if that is defined.
677 Otherwise, the default emulation depends upon how the linker was
685 Print a link map to the standard output. A link map provides
686 information about the link, including the following:
690 Where object files are mapped into memory.
692 How common symbols are allocated.
694 All archive members included in the link, with a mention of the symbol
695 which caused the archive member to be brought in.
697 The values assigned to symbols.
699 Note - symbols whose values are computed by an expression which
700 involves a reference to a previous value of the same symbol may not
701 have correct result displayed in the link map. This is because the
702 linker discards intermediate results and only retains the final value
703 of an expression. Under such circumstances the linker will display
704 the final value enclosed by square brackets. Thus for example a
705 linker script containing:
713 will produce the following output in the link map if the @option{-M}
718 [0x0000000c] foo = (foo * 0x4)
719 [0x0000000c] foo = (foo + 0x8)
722 See @ref{Expressions} for more information about expressions in linker
727 @cindex read-only text
732 Turn off page alignment of sections, and mark the output as
733 @code{NMAGIC} if possible.
737 @cindex read/write from cmd line
741 Set the text and data sections to be readable and writable. Also, do
742 not page-align the data segment, and disable linking against shared
743 libraries. If the output format supports Unix style magic numbers,
744 mark the output as @code{OMAGIC}. Note: Although a writable text section
745 is allowed for PE-COFF targets, it does not conform to the format
746 specification published by Microsoft.
751 This option negates most of the effects of the @option{-N} option. It
752 sets the text section to be read-only, and forces the data segment to
753 be page-aligned. Note - this option does not enable linking against
754 shared libraries. Use @option{-Bdynamic} for this.
756 @kindex -o @var{output}
757 @kindex --output=@var{output}
758 @cindex naming the output file
759 @item -o @var{output}
760 @itemx --output=@var{output}
761 Use @var{output} as the name for the program produced by @command{ld}; if this
762 option is not specified, the name @file{a.out} is used by default. The
763 script command @code{OUTPUT} can also specify the output file name.
765 @kindex -O @var{level}
766 @cindex generating optimized output
768 If @var{level} is a numeric values greater than zero @command{ld} optimizes
769 the output. This might take significantly longer and therefore probably
770 should only be enabled for the final binary. At the moment this
771 option only affects ELF shared library generation. Future releases of
772 the linker may make more use of this option. Also currently there is
773 no difference in the linker's behaviour for different non-zero values
774 of this option. Again this may change with future releases.
777 @kindex --emit-relocs
778 @cindex retain relocations in final executable
781 Leave relocation sections and contents in fully linked executables.
782 Post link analysis and optimization tools may need this information in
783 order to perform correct modifications of executables. This results
784 in larger executables.
786 This option is currently only supported on ELF platforms.
788 @kindex --force-dynamic
789 @cindex forcing the creation of dynamic sections
790 @item --force-dynamic
791 Force the output file to have dynamic sections. This option is specific
795 @cindex relocatable output
797 @kindex --relocatable
800 Generate relocatable output---i.e., generate an output file that can in
801 turn serve as input to @command{ld}. This is often called @dfn{partial
802 linking}. As a side effect, in environments that support standard Unix
803 magic numbers, this option also sets the output file's magic number to
805 @c ; see @option{-N}.
806 If this option is not specified, an absolute file is produced. When
807 linking C++ programs, this option @emph{will not} resolve references to
808 constructors; to do that, use @samp{-Ur}.
810 When an input file does not have the same format as the output file,
811 partial linking is only supported if that input file does not contain any
812 relocations. Different output formats can have further restrictions; for
813 example some @code{a.out}-based formats do not support partial linking
814 with input files in other formats at all.
816 This option does the same thing as @samp{-i}.
818 @kindex -R @var{file}
819 @kindex --just-symbols=@var{file}
820 @cindex symbol-only input
821 @item -R @var{filename}
822 @itemx --just-symbols=@var{filename}
823 Read symbol names and their addresses from @var{filename}, but do not
824 relocate it or include it in the output. This allows your output file
825 to refer symbolically to absolute locations of memory defined in other
826 programs. You may use this option more than once.
828 For compatibility with other ELF linkers, if the @option{-R} option is
829 followed by a directory name, rather than a file name, it is treated as
830 the @option{-rpath} option.
833 @cindex input files, displaying
835 Put code in a seperate segment, not along with other read only data.
839 @cindex strip all symbols
842 Omit all symbol information from the output file.
845 @kindex --strip-debug
846 @cindex strip debugger symbols
849 Omit debugger symbol information (but not all symbols) from the output file.
853 @cindex input files, displaying
856 Print the names of the input files as @command{ld} processes them.
858 @kindex -T @var{script}
859 @kindex --script=@var{script}
861 @item -T @var{scriptfile}
862 @itemx --script=@var{scriptfile}
863 Use @var{scriptfile} as the linker script. This script replaces
864 @command{ld}'s default linker script (rather than adding to it), so
865 @var{commandfile} must specify everything necessary to describe the
866 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
867 the current directory, @code{ld} looks for it in the directories
868 specified by any preceding @samp{-L} options. Multiple @samp{-T}
871 @kindex -dT @var{script}
872 @kindex --default-script=@var{script}
874 @item -dT @var{scriptfile}
875 @itemx --default-script=@var{scriptfile}
876 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
878 This option is similar to the @option{--script} option except that
879 processing of the script is delayed until after the rest of the
880 command line has been processed. This allows options placed after the
881 @option{--default-script} option on the command line to affect the
882 behaviour of the linker script, which can be important when the linker
883 command line cannot be directly controlled by the user. (eg because
884 the command line is being constructed by another tool, such as
887 @kindex -u @var{symbol}
888 @kindex --undefined=@var{symbol}
889 @cindex undefined symbol
890 @item -u @var{symbol}
891 @itemx --undefined=@var{symbol}
892 Force @var{symbol} to be entered in the output file as an undefined
893 symbol. Doing this may, for example, trigger linking of additional
894 modules from standard libraries. @samp{-u} may be repeated with
895 different option arguments to enter additional undefined symbols. This
896 option is equivalent to the @code{EXTERN} linker script command.
901 For anything other than C++ programs, this option is equivalent to
902 @samp{-r}: it generates relocatable output---i.e., an output file that can in
903 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
904 @emph{does} resolve references to constructors, unlike @samp{-r}.
905 It does not work to use @samp{-Ur} on files that were themselves linked
906 with @samp{-Ur}; once the constructor table has been built, it cannot
907 be added to. Use @samp{-Ur} only for the last partial link, and
908 @samp{-r} for the others.
910 @kindex --unique[=@var{SECTION}]
911 @item --unique[=@var{SECTION}]
912 Creates a separate output section for every input section matching
913 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
914 missing, for every orphan input section. An orphan section is one not
915 specifically mentioned in a linker script. You may use this option
916 multiple times on the command line; It prevents the normal merging of
917 input sections with the same name, overriding output section assignments
927 Display the version number for @command{ld}. The @option{-V} option also
928 lists the supported emulations.
931 @kindex --discard-all
932 @cindex deleting local symbols
935 Delete all local symbols.
938 @kindex --discard-locals
939 @cindex local symbols, deleting
941 @itemx --discard-locals
942 Delete all temporary local symbols. (These symbols start with
943 system-specific local label prefixes, typically @samp{.L} for ELF systems
944 or @samp{L} for traditional a.out systems.)
946 @kindex -y @var{symbol}
947 @kindex --trace-symbol=@var{symbol}
948 @cindex symbol tracing
949 @item -y @var{symbol}
950 @itemx --trace-symbol=@var{symbol}
951 Print the name of each linked file in which @var{symbol} appears. This
952 option may be given any number of times. On many systems it is necessary
953 to prepend an underscore.
955 This option is useful when you have an undefined symbol in your link but
956 don't know where the reference is coming from.
958 @kindex -Y @var{path}
960 Add @var{path} to the default library search path. This option exists
961 for Solaris compatibility.
963 @kindex -z @var{keyword}
964 @item -z @var{keyword}
965 The recognized keywords are:
969 Combines multiple reloc sections and sorts them to make dynamic symbol
970 lookup caching possible.
973 Disallows undefined symbols in object files. Undefined symbols in
974 shared libraries are still allowed.
977 Marks the object as requiring executable stack.
980 This option is only meaningful when building a shared object.
981 It marks the object so that its runtime initialization will occur
982 before the runtime initialization of any other objects brought into
983 the process at the same time. Similarly the runtime finalization of
984 the object will occur after the runtime finalization of any other
988 Marks the object that its symbol table interposes before all symbols
989 but the primary executable.
992 When generating an executable or shared library, mark it to tell the
993 dynamic linker to defer function call resolution to the point when
994 the function is called (lazy binding), rather than at load time.
995 Lazy binding is the default.
998 Marks the object that its filters be processed immediately at
1002 Allows multiple definitions.
1005 Disables multiple reloc sections combining.
1008 Disables production of copy relocs.
1011 Marks the object that the search for dependencies of this object will
1012 ignore any default library search paths.
1015 Marks the object shouldn't be unloaded at runtime.
1018 Marks the object not available to @code{dlopen}.
1021 Marks the object can not be dumped by @code{dldump}.
1024 Marks the object as not requiring executable stack.
1027 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1030 When generating an executable or shared library, mark it to tell the
1031 dynamic linker to resolve all symbols when the program is started, or
1032 when the shared library is linked to using dlopen, instead of
1033 deferring function call resolution to the point when the function is
1037 Marks the object may contain $ORIGIN.
1040 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1042 @item max-page-size=@var{value}
1043 Set the emulation maximum page size to @var{value}.
1045 @item common-page-size=@var{value}
1046 Set the emulation common page size to @var{value}.
1050 Other keywords are ignored for Solaris compatibility.
1053 @cindex groups of archives
1054 @item -( @var{archives} -)
1055 @itemx --start-group @var{archives} --end-group
1056 The @var{archives} should be a list of archive files. They may be
1057 either explicit file names, or @samp{-l} options.
1059 The specified archives are searched repeatedly until no new undefined
1060 references are created. Normally, an archive is searched only once in
1061 the order that it is specified on the command line. If a symbol in that
1062 archive is needed to resolve an undefined symbol referred to by an
1063 object in an archive that appears later on the command line, the linker
1064 would not be able to resolve that reference. By grouping the archives,
1065 they all be searched repeatedly until all possible references are
1068 Using this option has a significant performance cost. It is best to use
1069 it only when there are unavoidable circular references between two or
1072 @kindex --accept-unknown-input-arch
1073 @kindex --no-accept-unknown-input-arch
1074 @item --accept-unknown-input-arch
1075 @itemx --no-accept-unknown-input-arch
1076 Tells the linker to accept input files whose architecture cannot be
1077 recognised. The assumption is that the user knows what they are doing
1078 and deliberately wants to link in these unknown input files. This was
1079 the default behaviour of the linker, before release 2.14. The default
1080 behaviour from release 2.14 onwards is to reject such input files, and
1081 so the @samp{--accept-unknown-input-arch} option has been added to
1082 restore the old behaviour.
1085 @kindex --no-as-needed
1087 @itemx --no-as-needed
1088 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1089 on the command line after the @option{--as-needed} option. Normally,
1090 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1091 on the command line, regardless of whether the library is actually
1092 needed. @option{--as-needed} causes DT_NEEDED tags to only be emitted
1093 for libraries that satisfy some symbol reference from regular objects
1094 which is undefined at the point that the library was linked.
1095 @option{--no-as-needed} restores the default behaviour.
1097 @kindex --add-needed
1098 @kindex --no-add-needed
1100 @itemx --no-add-needed
1101 This option affects the treatment of dynamic libraries from ELF
1102 DT_NEEDED tags in dynamic libraries mentioned on the command line after
1103 the @option{--no-add-needed} option. Normally, the linker will add
1104 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1105 @option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1106 for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1107 the default behaviour.
1109 @kindex -assert @var{keyword}
1110 @item -assert @var{keyword}
1111 This option is ignored for SunOS compatibility.
1115 @kindex -call_shared
1119 Link against dynamic libraries. This is only meaningful on platforms
1120 for which shared libraries are supported. This option is normally the
1121 default on such platforms. The different variants of this option are
1122 for compatibility with various systems. You may use this option
1123 multiple times on the command line: it affects library searching for
1124 @option{-l} options which follow it.
1128 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1129 section. This causes the runtime linker to handle lookups in this
1130 object and its dependencies to be performed only inside the group.
1131 @option{--unresolved-symbols=report-all} is implied. This option is
1132 only meaningful on ELF platforms which support shared libraries.
1142 Do not link against shared libraries. This is only meaningful on
1143 platforms for which shared libraries are supported. The different
1144 variants of this option are for compatibility with various systems. You
1145 may use this option multiple times on the command line: it affects
1146 library searching for @option{-l} options which follow it. This
1147 option also implies @option{--unresolved-symbols=report-all}. This
1148 option can be used with @option{-shared}. Doing so means that a
1149 shared library is being created but that all of the library's external
1150 references must be resolved by pulling in entries from static
1155 When creating a shared library, bind references to global symbols to the
1156 definition within the shared library, if any. Normally, it is possible
1157 for a program linked against a shared library to override the definition
1158 within the shared library. This option is only meaningful on ELF
1159 platforms which support shared libraries.
1161 @kindex -Bsymbolic-functions
1162 @item -Bsymbolic-functions
1163 When creating a shared library, bind references to global function
1164 symbols to the definition within the shared library, if any.
1165 This option is only meaningful on ELF platforms which support shared
1168 @kindex --dynamic-list=@var{dynamic-list-file}
1169 @item --dynamic-list=@var{dynamic-list-file}
1170 Specify the name of a dynamic list file to the linker. This is
1171 typically used when creating shared libraries to specify a list of
1172 global symbols whose references shouldn't be bound to the definition
1173 within the shared library, or creating dynamically linked executables
1174 to specify a list of symbols which should be added to the symbol table
1175 in the executable. This option is only meaningful on ELF platforms
1176 which support shared libraries.
1178 The format of the dynamic list is the same as the version node without
1179 scope and node name. See @ref{VERSION} for more information.
1181 @kindex --dynamic-list-data
1182 @item --dynamic-list-data
1183 Include all global data symbols to the dynamic list.
1185 @kindex --dynamic-list-cpp-new
1186 @item --dynamic-list-cpp-new
1187 Provide the builtin dynamic list for C++ operator new and delete. It
1188 is mainly useful for building shared libstdc++.
1190 @kindex --dynamic-list-cpp-typeinfo
1191 @item --dynamic-list-cpp-typeinfo
1192 Provide the builtin dynamic list for C++ runtime type identification.
1194 @kindex --check-sections
1195 @kindex --no-check-sections
1196 @item --check-sections
1197 @itemx --no-check-sections
1198 Asks the linker @emph{not} to check section addresses after they have
1199 been assigned to see if there are any overlaps. Normally the linker will
1200 perform this check, and if it finds any overlaps it will produce
1201 suitable error messages. The linker does know about, and does make
1202 allowances for sections in overlays. The default behaviour can be
1203 restored by using the command line switch @option{--check-sections}.
1205 @cindex cross reference table
1208 Output a cross reference table. If a linker map file is being
1209 generated, the cross reference table is printed to the map file.
1210 Otherwise, it is printed on the standard output.
1212 The format of the table is intentionally simple, so that it may be
1213 easily processed by a script if necessary. The symbols are printed out,
1214 sorted by name. For each symbol, a list of file names is given. If the
1215 symbol is defined, the first file listed is the location of the
1216 definition. The remaining files contain references to the symbol.
1218 @cindex common allocation
1219 @kindex --no-define-common
1220 @item --no-define-common
1221 This option inhibits the assignment of addresses to common symbols.
1222 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1223 @xref{Miscellaneous Commands}.
1225 The @samp{--no-define-common} option allows decoupling
1226 the decision to assign addresses to Common symbols from the choice
1227 of the output file type; otherwise a non-Relocatable output type
1228 forces assigning addresses to Common symbols.
1229 Using @samp{--no-define-common} allows Common symbols that are referenced
1230 from a shared library to be assigned addresses only in the main program.
1231 This eliminates the unused duplicate space in the shared library,
1232 and also prevents any possible confusion over resolving to the wrong
1233 duplicate when there are many dynamic modules with specialized search
1234 paths for runtime symbol resolution.
1236 @cindex symbols, from command line
1237 @kindex --defsym @var{symbol}=@var{exp}
1238 @item --defsym @var{symbol}=@var{expression}
1239 Create a global symbol in the output file, containing the absolute
1240 address given by @var{expression}. You may use this option as many
1241 times as necessary to define multiple symbols in the command line. A
1242 limited form of arithmetic is supported for the @var{expression} in this
1243 context: you may give a hexadecimal constant or the name of an existing
1244 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1245 constants or symbols. If you need more elaborate expressions, consider
1246 using the linker command language from a script (@pxref{Assignments,,
1247 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1248 space between @var{symbol}, the equals sign (``@key{=}''), and
1251 @cindex demangling, from command line
1252 @kindex --demangle[=@var{style}]
1253 @kindex --no-demangle
1254 @item --demangle[=@var{style}]
1255 @itemx --no-demangle
1256 These options control whether to demangle symbol names in error messages
1257 and other output. When the linker is told to demangle, it tries to
1258 present symbol names in a readable fashion: it strips leading
1259 underscores if they are used by the object file format, and converts C++
1260 mangled symbol names into user readable names. Different compilers have
1261 different mangling styles. The optional demangling style argument can be used
1262 to choose an appropriate demangling style for your compiler. The linker will
1263 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1264 is set. These options may be used to override the default.
1266 @cindex dynamic linker, from command line
1267 @kindex -I@var{file}
1268 @kindex --dynamic-linker @var{file}
1269 @item --dynamic-linker @var{file}
1270 Set the name of the dynamic linker. This is only meaningful when
1271 generating dynamically linked ELF executables. The default dynamic
1272 linker is normally correct; don't use this unless you know what you are
1275 @kindex --fatal-warnings
1276 @kindex --no-fatal-warnings
1277 @item --fatal-warnings
1278 @itemx --no-fatal-warnings
1279 Treat all warnings as errors. The default behaviour can be restored
1280 with the option @option{--no-fatal-warnings}.
1282 @kindex --force-exe-suffix
1283 @item --force-exe-suffix
1284 Make sure that an output file has a .exe suffix.
1286 If a successfully built fully linked output file does not have a
1287 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1288 the output file to one of the same name with a @code{.exe} suffix. This
1289 option is useful when using unmodified Unix makefiles on a Microsoft
1290 Windows host, since some versions of Windows won't run an image unless
1291 it ends in a @code{.exe} suffix.
1293 @kindex --gc-sections
1294 @kindex --no-gc-sections
1295 @cindex garbage collection
1297 @itemx --no-gc-sections
1298 Enable garbage collection of unused input sections. It is ignored on
1299 targets that do not support this option. The default behaviour (of not
1300 performing this garbage collection) can be restored by specifying
1301 @samp{--no-gc-sections} on the command line.
1303 @samp{--gc-sections} decides which input sections are used by
1304 examining symbols and relocations. The section containing the entry
1305 symbol and all sections containing symbols undefined on the
1306 command-line will be kept, as will sections containing symbols
1307 referenced by dynamic objects. Note that when building shared
1308 libraries, the linker must assume that any visible symbol is
1309 referenced. Once this initial set of sections has been determined,
1310 the linker recursively marks as used any section referenced by their
1311 relocations. See @samp{--entry} and @samp{--undefined}.
1313 This option can be set when doing a partial link (enabled with option
1314 @samp{-r}). In this case the root of symbols kept must be explicitely
1315 specified either by an @samp{--entry} or @samp{--undefined} option or by
1316 a @code{ENTRY} command in the linker script.
1318 @kindex --print-gc-sections
1319 @kindex --no-print-gc-sections
1320 @cindex garbage collection
1321 @item --print-gc-sections
1322 @itemx --no-print-gc-sections
1323 List all sections removed by garbage collection. The listing is
1324 printed on stderr. This option is only effective if garbage
1325 collection has been enabled via the @samp{--gc-sections}) option. The
1326 default behaviour (of not listing the sections that are removed) can
1327 be restored by specifying @samp{--no-print-gc-sections} on the command
1334 Print a summary of the command-line options on the standard output and exit.
1336 @kindex --target-help
1338 Print a summary of all target specific options on the standard output and exit.
1341 @item -Map @var{mapfile}
1342 Print a link map to the file @var{mapfile}. See the description of the
1343 @option{-M} option, above.
1345 @cindex memory usage
1346 @kindex --no-keep-memory
1347 @item --no-keep-memory
1348 @command{ld} normally optimizes for speed over memory usage by caching the
1349 symbol tables of input files in memory. This option tells @command{ld} to
1350 instead optimize for memory usage, by rereading the symbol tables as
1351 necessary. This may be required if @command{ld} runs out of memory space
1352 while linking a large executable.
1354 @kindex --no-undefined
1356 @item --no-undefined
1358 Report unresolved symbol references from regular object files. This
1359 is done even if the linker is creating a non-symbolic shared library.
1360 The switch @option{--[no-]allow-shlib-undefined} controls the
1361 behaviour for reporting unresolved references found in shared
1362 libraries being linked in.
1364 @kindex --allow-multiple-definition
1366 @item --allow-multiple-definition
1368 Normally when a symbol is defined multiple times, the linker will
1369 report a fatal error. These options allow multiple definitions and the
1370 first definition will be used.
1372 @kindex --allow-shlib-undefined
1373 @kindex --no-allow-shlib-undefined
1374 @item --allow-shlib-undefined
1375 @itemx --no-allow-shlib-undefined
1376 Allows (the default) or disallows undefined symbols in shared libraries.
1377 This switch is similar to @option{--no-undefined} except that it
1378 determines the behaviour when the undefined symbols are in a
1379 shared library rather than a regular object file. It does not affect
1380 how undefined symbols in regular object files are handled.
1382 The reason that @option{--allow-shlib-undefined} is the default is that
1383 the shared library being specified at link time may not be the same as
1384 the one that is available at load time, so the symbols might actually be
1385 resolvable at load time. Plus there are some systems, (eg BeOS) where
1386 undefined symbols in shared libraries is normal. (The kernel patches
1387 them at load time to select which function is most appropriate
1388 for the current architecture. This is used for example to dynamically
1389 select an appropriate memset function). Apparently it is also normal
1390 for HPPA shared libraries to have undefined symbols.
1392 @kindex --no-undefined-version
1393 @item --no-undefined-version
1394 Normally when a symbol has an undefined version, the linker will ignore
1395 it. This option disallows symbols with undefined version and a fatal error
1396 will be issued instead.
1398 @kindex --default-symver
1399 @item --default-symver
1400 Create and use a default symbol version (the soname) for unversioned
1403 @kindex --default-imported-symver
1404 @item --default-imported-symver
1405 Create and use a default symbol version (the soname) for unversioned
1408 @kindex --no-warn-mismatch
1409 @item --no-warn-mismatch
1410 Normally @command{ld} will give an error if you try to link together input
1411 files that are mismatched for some reason, perhaps because they have
1412 been compiled for different processors or for different endiannesses.
1413 This option tells @command{ld} that it should silently permit such possible
1414 errors. This option should only be used with care, in cases when you
1415 have taken some special action that ensures that the linker errors are
1418 @kindex --no-warn-search-mismatch
1419 @item --no-warn-search-mismatch
1420 Normally @command{ld} will give a warning if it finds an incompatible
1421 library during a library search. This option silences the warning.
1423 @kindex --no-whole-archive
1424 @item --no-whole-archive
1425 Turn off the effect of the @option{--whole-archive} option for subsequent
1428 @cindex output file after errors
1429 @kindex --noinhibit-exec
1430 @item --noinhibit-exec
1431 Retain the executable output file whenever it is still usable.
1432 Normally, the linker will not produce an output file if it encounters
1433 errors during the link process; it exits without writing an output file
1434 when it issues any error whatsoever.
1438 Only search library directories explicitly specified on the
1439 command line. Library directories specified in linker scripts
1440 (including linker scripts specified on the command line) are ignored.
1442 @ifclear SingleFormat
1444 @item --oformat @var{output-format}
1445 @command{ld} may be configured to support more than one kind of object
1446 file. If your @command{ld} is configured this way, you can use the
1447 @samp{--oformat} option to specify the binary format for the output
1448 object file. Even when @command{ld} is configured to support alternative
1449 object formats, you don't usually need to specify this, as @command{ld}
1450 should be configured to produce as a default output format the most
1451 usual format on each machine. @var{output-format} is a text string, the
1452 name of a particular format supported by the BFD libraries. (You can
1453 list the available binary formats with @samp{objdump -i}.) The script
1454 command @code{OUTPUT_FORMAT} can also specify the output format, but
1455 this option overrides it. @xref{BFD}.
1459 @kindex --pic-executable
1461 @itemx --pic-executable
1462 @cindex position independent executables
1463 Create a position independent executable. This is currently only supported on
1464 ELF platforms. Position independent executables are similar to shared
1465 libraries in that they are relocated by the dynamic linker to the virtual
1466 address the OS chooses for them (which can vary between invocations). Like
1467 normal dynamically linked executables they can be executed and symbols
1468 defined in the executable cannot be overridden by shared libraries.
1472 This option is ignored for Linux compatibility.
1476 This option is ignored for SVR4 compatibility.
1479 @cindex synthesizing linker
1480 @cindex relaxing addressing modes
1482 An option with machine dependent effects.
1484 This option is only supported on a few targets.
1487 @xref{H8/300,,@command{ld} and the H8/300}.
1490 @xref{i960,, @command{ld} and the Intel 960 family}.
1493 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1496 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1499 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1502 On some platforms, the @samp{--relax} option performs global
1503 optimizations that become possible when the linker resolves addressing
1504 in the program, such as relaxing address modes and synthesizing new
1505 instructions in the output object file.
1507 On some platforms these link time global optimizations may make symbolic
1508 debugging of the resulting executable impossible.
1511 the case for the Matsushita MN10200 and MN10300 family of processors.
1515 On platforms where this is not supported, @samp{--relax} is accepted,
1519 @cindex retaining specified symbols
1520 @cindex stripping all but some symbols
1521 @cindex symbols, retaining selectively
1522 @item --retain-symbols-file @var{filename}
1523 Retain @emph{only} the symbols listed in the file @var{filename},
1524 discarding all others. @var{filename} is simply a flat file, with one
1525 symbol name per line. This option is especially useful in environments
1529 where a large global symbol table is accumulated gradually, to conserve
1532 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1533 or symbols needed for relocations.
1535 You may only specify @samp{--retain-symbols-file} once in the command
1536 line. It overrides @samp{-s} and @samp{-S}.
1539 @item -rpath @var{dir}
1540 @cindex runtime library search path
1542 Add a directory to the runtime library search path. This is used when
1543 linking an ELF executable with shared objects. All @option{-rpath}
1544 arguments are concatenated and passed to the runtime linker, which uses
1545 them to locate shared objects at runtime. The @option{-rpath} option is
1546 also used when locating shared objects which are needed by shared
1547 objects explicitly included in the link; see the description of the
1548 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1549 ELF executable, the contents of the environment variable
1550 @code{LD_RUN_PATH} will be used if it is defined.
1552 The @option{-rpath} option may also be used on SunOS. By default, on
1553 SunOS, the linker will form a runtime search patch out of all the
1554 @option{-L} options it is given. If a @option{-rpath} option is used, the
1555 runtime search path will be formed exclusively using the @option{-rpath}
1556 options, ignoring the @option{-L} options. This can be useful when using
1557 gcc, which adds many @option{-L} options which may be on NFS mounted
1560 For compatibility with other ELF linkers, if the @option{-R} option is
1561 followed by a directory name, rather than a file name, it is treated as
1562 the @option{-rpath} option.
1566 @cindex link-time runtime library search path
1568 @item -rpath-link @var{DIR}
1569 When using ELF or SunOS, one shared library may require another. This
1570 happens when an @code{ld -shared} link includes a shared library as one
1573 When the linker encounters such a dependency when doing a non-shared,
1574 non-relocatable link, it will automatically try to locate the required
1575 shared library and include it in the link, if it is not included
1576 explicitly. In such a case, the @option{-rpath-link} option
1577 specifies the first set of directories to search. The
1578 @option{-rpath-link} option may specify a sequence of directory names
1579 either by specifying a list of names separated by colons, or by
1580 appearing multiple times.
1582 This option should be used with caution as it overrides the search path
1583 that may have been hard compiled into a shared library. In such a case it
1584 is possible to use unintentionally a different search path than the
1585 runtime linker would do.
1587 The linker uses the following search paths to locate required shared
1591 Any directories specified by @option{-rpath-link} options.
1593 Any directories specified by @option{-rpath} options. The difference
1594 between @option{-rpath} and @option{-rpath-link} is that directories
1595 specified by @option{-rpath} options are included in the executable and
1596 used at runtime, whereas the @option{-rpath-link} option is only effective
1597 at link time. Searching @option{-rpath} in this way is only supported
1598 by native linkers and cross linkers which have been configured with
1599 the @option{--with-sysroot} option.
1601 On an ELF system, for native linkers, if the @option{-rpath} and
1602 @option{-rpath-link} options were not used, search the contents of the
1603 environment variable @code{LD_RUN_PATH}.
1605 On SunOS, if the @option{-rpath} option was not used, search any
1606 directories specified using @option{-L} options.
1608 For a native linker, the search the contents of the environment
1609 variable @code{LD_LIBRARY_PATH}.
1611 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1612 @code{DT_RPATH} of a shared library are searched for shared
1613 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1614 @code{DT_RUNPATH} entries exist.
1616 The default directories, normally @file{/lib} and @file{/usr/lib}.
1618 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1619 exists, the list of directories found in that file.
1622 If the required shared library is not found, the linker will issue a
1623 warning and continue with the link.
1630 @cindex shared libraries
1631 Create a shared library. This is currently only supported on ELF, XCOFF
1632 and SunOS platforms. On SunOS, the linker will automatically create a
1633 shared library if the @option{-e} option is not used and there are
1634 undefined symbols in the link.
1636 @item --sort-common [= ascending | descending]
1637 @kindex --sort-common
1638 This option tells @command{ld} to sort the common symbols by alignment in
1639 ascending or descending order when it places them in the appropriate output
1640 sections. The symbol alignments considered are sixteen-byte or larger,
1641 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1642 between symbols due to alignment constraints. If no sorting order is
1643 specified, then descending order is assumed.
1645 @kindex --sort-section name
1646 @item --sort-section name
1647 This option will apply @code{SORT_BY_NAME} to all wildcard section
1648 patterns in the linker script.
1650 @kindex --sort-section alignment
1651 @item --sort-section alignment
1652 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1653 patterns in the linker script.
1655 @kindex --split-by-file
1656 @item --split-by-file [@var{size}]
1657 Similar to @option{--split-by-reloc} but creates a new output section for
1658 each input file when @var{size} is reached. @var{size} defaults to a
1659 size of 1 if not given.
1661 @kindex --split-by-reloc
1662 @item --split-by-reloc [@var{count}]
1663 Tries to creates extra sections in the output file so that no single
1664 output section in the file contains more than @var{count} relocations.
1665 This is useful when generating huge relocatable files for downloading into
1666 certain real time kernels with the COFF object file format; since COFF
1667 cannot represent more than 65535 relocations in a single section. Note
1668 that this will fail to work with object file formats which do not
1669 support arbitrary sections. The linker will not split up individual
1670 input sections for redistribution, so if a single input section contains
1671 more than @var{count} relocations one output section will contain that
1672 many relocations. @var{count} defaults to a value of 32768.
1676 Compute and display statistics about the operation of the linker, such
1677 as execution time and memory usage.
1680 @item --sysroot=@var{directory}
1681 Use @var{directory} as the location of the sysroot, overriding the
1682 configure-time default. This option is only supported by linkers
1683 that were configured using @option{--with-sysroot}.
1685 @kindex --traditional-format
1686 @cindex traditional format
1687 @item --traditional-format
1688 For some targets, the output of @command{ld} is different in some ways from
1689 the output of some existing linker. This switch requests @command{ld} to
1690 use the traditional format instead.
1693 For example, on SunOS, @command{ld} combines duplicate entries in the
1694 symbol string table. This can reduce the size of an output file with
1695 full debugging information by over 30 percent. Unfortunately, the SunOS
1696 @code{dbx} program can not read the resulting program (@code{gdb} has no
1697 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1698 combine duplicate entries.
1700 @kindex --section-start @var{sectionname}=@var{org}
1701 @item --section-start @var{sectionname}=@var{org}
1702 Locate a section in the output file at the absolute
1703 address given by @var{org}. You may use this option as many
1704 times as necessary to locate multiple sections in the command
1706 @var{org} must be a single hexadecimal integer;
1707 for compatibility with other linkers, you may omit the leading
1708 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1709 should be no white space between @var{sectionname}, the equals
1710 sign (``@key{=}''), and @var{org}.
1712 @kindex -Tbss @var{org}
1713 @kindex -Tdata @var{org}
1714 @kindex -Ttext @var{org}
1715 @cindex segment origins, cmd line
1716 @item -Tbss @var{org}
1717 @itemx -Tdata @var{org}
1718 @itemx -Ttext @var{org}
1719 Same as --section-start, with @code{.bss}, @code{.data} or
1720 @code{.text} as the @var{sectionname}.
1722 @kindex --unresolved-symbols
1723 @item --unresolved-symbols=@var{method}
1724 Determine how to handle unresolved symbols. There are four possible
1725 values for @samp{method}:
1729 Do not report any unresolved symbols.
1732 Report all unresolved symbols. This is the default.
1734 @item ignore-in-object-files
1735 Report unresolved symbols that are contained in shared libraries, but
1736 ignore them if they come from regular object files.
1738 @item ignore-in-shared-libs
1739 Report unresolved symbols that come from regular object files, but
1740 ignore them if they come from shared libraries. This can be useful
1741 when creating a dynamic binary and it is known that all the shared
1742 libraries that it should be referencing are included on the linker's
1746 The behaviour for shared libraries on their own can also be controlled
1747 by the @option{--[no-]allow-shlib-undefined} option.
1749 Normally the linker will generate an error message for each reported
1750 unresolved symbol but the option @option{--warn-unresolved-symbols}
1751 can change this to a warning.
1757 Display the version number for @command{ld} and list the linker emulations
1758 supported. Display which input files can and cannot be opened. Display
1759 the linker script being used by the linker.
1761 @kindex --version-script=@var{version-scriptfile}
1762 @cindex version script, symbol versions
1763 @itemx --version-script=@var{version-scriptfile}
1764 Specify the name of a version script to the linker. This is typically
1765 used when creating shared libraries to specify additional information
1766 about the version hierarchy for the library being created. This option
1767 is only meaningful on ELF platforms which support shared libraries.
1770 @kindex --warn-common
1771 @cindex warnings, on combining symbols
1772 @cindex combining symbols, warnings on
1774 Warn when a common symbol is combined with another common symbol or with
1775 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1776 but linkers on some other operating systems do not. This option allows
1777 you to find potential problems from combining global symbols.
1778 Unfortunately, some C libraries use this practise, so you may get some
1779 warnings about symbols in the libraries as well as in your programs.
1781 There are three kinds of global symbols, illustrated here by C examples:
1785 A definition, which goes in the initialized data section of the output
1789 An undefined reference, which does not allocate space.
1790 There must be either a definition or a common symbol for the
1794 A common symbol. If there are only (one or more) common symbols for a
1795 variable, it goes in the uninitialized data area of the output file.
1796 The linker merges multiple common symbols for the same variable into a
1797 single symbol. If they are of different sizes, it picks the largest
1798 size. The linker turns a common symbol into a declaration, if there is
1799 a definition of the same variable.
1802 The @samp{--warn-common} option can produce five kinds of warnings.
1803 Each warning consists of a pair of lines: the first describes the symbol
1804 just encountered, and the second describes the previous symbol
1805 encountered with the same name. One or both of the two symbols will be
1810 Turning a common symbol into a reference, because there is already a
1811 definition for the symbol.
1813 @var{file}(@var{section}): warning: common of `@var{symbol}'
1814 overridden by definition
1815 @var{file}(@var{section}): warning: defined here
1819 Turning a common symbol into a reference, because a later definition for
1820 the symbol is encountered. This is the same as the previous case,
1821 except that the symbols are encountered in a different order.
1823 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1825 @var{file}(@var{section}): warning: common is here
1829 Merging a common symbol with a previous same-sized common symbol.
1831 @var{file}(@var{section}): warning: multiple common
1833 @var{file}(@var{section}): warning: previous common is here
1837 Merging a common symbol with a previous larger common symbol.
1839 @var{file}(@var{section}): warning: common of `@var{symbol}'
1840 overridden by larger common
1841 @var{file}(@var{section}): warning: larger common is here
1845 Merging a common symbol with a previous smaller common symbol. This is
1846 the same as the previous case, except that the symbols are
1847 encountered in a different order.
1849 @var{file}(@var{section}): warning: common of `@var{symbol}'
1850 overriding smaller common
1851 @var{file}(@var{section}): warning: smaller common is here
1855 @kindex --warn-constructors
1856 @item --warn-constructors
1857 Warn if any global constructors are used. This is only useful for a few
1858 object file formats. For formats like COFF or ELF, the linker can not
1859 detect the use of global constructors.
1861 @kindex --warn-multiple-gp
1862 @item --warn-multiple-gp
1863 Warn if multiple global pointer values are required in the output file.
1864 This is only meaningful for certain processors, such as the Alpha.
1865 Specifically, some processors put large-valued constants in a special
1866 section. A special register (the global pointer) points into the middle
1867 of this section, so that constants can be loaded efficiently via a
1868 base-register relative addressing mode. Since the offset in
1869 base-register relative mode is fixed and relatively small (e.g., 16
1870 bits), this limits the maximum size of the constant pool. Thus, in
1871 large programs, it is often necessary to use multiple global pointer
1872 values in order to be able to address all possible constants. This
1873 option causes a warning to be issued whenever this case occurs.
1876 @cindex warnings, on undefined symbols
1877 @cindex undefined symbols, warnings on
1879 Only warn once for each undefined symbol, rather than once per module
1882 @kindex --warn-section-align
1883 @cindex warnings, on section alignment
1884 @cindex section alignment, warnings on
1885 @item --warn-section-align
1886 Warn if the address of an output section is changed because of
1887 alignment. Typically, the alignment will be set by an input section.
1888 The address will only be changed if it not explicitly specified; that
1889 is, if the @code{SECTIONS} command does not specify a start address for
1890 the section (@pxref{SECTIONS}).
1892 @kindex --warn-shared-textrel
1893 @item --warn-shared-textrel
1894 Warn if the linker adds a DT_TEXTREL to a shared object.
1896 @kindex --warn-unresolved-symbols
1897 @item --warn-unresolved-symbols
1898 If the linker is going to report an unresolved symbol (see the option
1899 @option{--unresolved-symbols}) it will normally generate an error.
1900 This option makes it generate a warning instead.
1902 @kindex --error-unresolved-symbols
1903 @item --error-unresolved-symbols
1904 This restores the linker's default behaviour of generating errors when
1905 it is reporting unresolved symbols.
1907 @kindex --whole-archive
1908 @cindex including an entire archive
1909 @item --whole-archive
1910 For each archive mentioned on the command line after the
1911 @option{--whole-archive} option, include every object file in the archive
1912 in the link, rather than searching the archive for the required object
1913 files. This is normally used to turn an archive file into a shared
1914 library, forcing every object to be included in the resulting shared
1915 library. This option may be used more than once.
1917 Two notes when using this option from gcc: First, gcc doesn't know
1918 about this option, so you have to use @option{-Wl,-whole-archive}.
1919 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1920 list of archives, because gcc will add its own list of archives to
1921 your link and you may not want this flag to affect those as well.
1924 @item --wrap @var{symbol}
1925 Use a wrapper function for @var{symbol}. Any undefined reference to
1926 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1927 undefined reference to @code{__real_@var{symbol}} will be resolved to
1930 This can be used to provide a wrapper for a system function. The
1931 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1932 wishes to call the system function, it should call
1933 @code{__real_@var{symbol}}.
1935 Here is a trivial example:
1939 __wrap_malloc (size_t c)
1941 printf ("malloc called with %zu\n", c);
1942 return __real_malloc (c);
1946 If you link other code with this file using @option{--wrap malloc}, then
1947 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1948 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1949 call the real @code{malloc} function.
1951 You may wish to provide a @code{__real_malloc} function as well, so that
1952 links without the @option{--wrap} option will succeed. If you do this,
1953 you should not put the definition of @code{__real_malloc} in the same
1954 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
1955 call before the linker has a chance to wrap it to @code{malloc}.
1957 @kindex --eh-frame-hdr
1958 @item --eh-frame-hdr
1959 Request creation of @code{.eh_frame_hdr} section and ELF
1960 @code{PT_GNU_EH_FRAME} segment header.
1962 @kindex --enable-new-dtags
1963 @kindex --disable-new-dtags
1964 @item --enable-new-dtags
1965 @itemx --disable-new-dtags
1966 This linker can create the new dynamic tags in ELF. But the older ELF
1967 systems may not understand them. If you specify
1968 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
1969 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
1970 created. By default, the new dynamic tags are not created. Note that
1971 those options are only available for ELF systems.
1973 @kindex --hash-size=@var{number}
1974 @item --hash-size=@var{number}
1975 Set the default size of the linker's hash tables to a prime number
1976 close to @var{number}. Increasing this value can reduce the length of
1977 time it takes the linker to perform its tasks, at the expense of
1978 increasing the linker's memory requirements. Similarly reducing this
1979 value can reduce the memory requirements at the expense of speed.
1981 @kindex --hash-style=@var{style}
1982 @item --hash-style=@var{style}
1983 Set the type of linker's hash table(s). @var{style} can be either
1984 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
1985 new style GNU @code{.gnu.hash} section or @code{both} for both
1986 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
1987 hash tables. The default is @code{sysv}.
1989 @kindex --reduce-memory-overheads
1990 @item --reduce-memory-overheads
1991 This option reduces memory requirements at ld runtime, at the expense of
1992 linking speed. This was introduced to select the old O(n^2) algorithm
1993 for link map file generation, rather than the new O(n) algorithm which uses
1994 about 40% more memory for symbol storage.
1996 Another effect of the switch is to set the default hash table size to
1997 1021, which again saves memory at the cost of lengthening the linker's
1998 run time. This is not done however if the @option{--hash-size} switch
2001 The @option{--reduce-memory-overheads} switch may be also be used to
2002 enable other tradeoffs in future versions of the linker.
2005 @kindex --build-id=@var{style}
2007 @itemx --build-id=@var{style}
2008 Request creation of @code{.note.gnu.build-id} ELF note section.
2009 The contents of the note are unique bits identifying this linked
2010 file. @var{style} can be @code{uuid} to use 128 random bits,
2011 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
2012 parts of the output contents, @code{md5} to use a 128-bit
2013 @sc{MD5} hash on the normative parts of the output contents, or
2014 @code{0x@var{hexstring}} to use a chosen bit string specified as
2015 an even number of hexadecimal digits (@code{-} and @code{:}
2016 characters between digit pairs are ignored). If @var{style} is
2017 omitted, @code{sha1} is used.
2019 The @code{md5} and @code{sha1} styles produces an identifier
2020 that is always the same in an identical output file, but will be
2021 unique among all nonidentical output files. It is not intended
2022 to be compared as a checksum for the file's contents. A linked
2023 file may be changed later by other tools, but the build ID bit
2024 string identifying the original linked file does not change.
2026 Passing @code{none} for @var{style} disables the setting from any
2027 @code{--build-id} options earlier on the command line.
2032 @subsection Options Specific to i386 PE Targets
2034 @c man begin OPTIONS
2036 The i386 PE linker supports the @option{-shared} option, which causes
2037 the output to be a dynamically linked library (DLL) instead of a
2038 normal executable. You should name the output @code{*.dll} when you
2039 use this option. In addition, the linker fully supports the standard
2040 @code{*.def} files, which may be specified on the linker command line
2041 like an object file (in fact, it should precede archives it exports
2042 symbols from, to ensure that they get linked in, just like a normal
2045 In addition to the options common to all targets, the i386 PE linker
2046 support additional command line options that are specific to the i386
2047 PE target. Options that take values may be separated from their
2048 values by either a space or an equals sign.
2052 @kindex --add-stdcall-alias
2053 @item --add-stdcall-alias
2054 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2055 as-is and also with the suffix stripped.
2056 [This option is specific to the i386 PE targeted port of the linker]
2059 @item --base-file @var{file}
2060 Use @var{file} as the name of a file in which to save the base
2061 addresses of all the relocations needed for generating DLLs with
2063 [This is an i386 PE specific option]
2067 Create a DLL instead of a regular executable. You may also use
2068 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2070 [This option is specific to the i386 PE targeted port of the linker]
2072 @kindex --enable-stdcall-fixup
2073 @kindex --disable-stdcall-fixup
2074 @item --enable-stdcall-fixup
2075 @itemx --disable-stdcall-fixup
2076 If the link finds a symbol that it cannot resolve, it will attempt to
2077 do ``fuzzy linking'' by looking for another defined symbol that differs
2078 only in the format of the symbol name (cdecl vs stdcall) and will
2079 resolve that symbol by linking to the match. For example, the
2080 undefined symbol @code{_foo} might be linked to the function
2081 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2082 to the function @code{_bar}. When the linker does this, it prints a
2083 warning, since it normally should have failed to link, but sometimes
2084 import libraries generated from third-party dlls may need this feature
2085 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2086 feature is fully enabled and warnings are not printed. If you specify
2087 @option{--disable-stdcall-fixup}, this feature is disabled and such
2088 mismatches are considered to be errors.
2089 [This option is specific to the i386 PE targeted port of the linker]
2091 @cindex DLLs, creating
2092 @kindex --export-all-symbols
2093 @item --export-all-symbols
2094 If given, all global symbols in the objects used to build a DLL will
2095 be exported by the DLL. Note that this is the default if there
2096 otherwise wouldn't be any exported symbols. When symbols are
2097 explicitly exported via DEF files or implicitly exported via function
2098 attributes, the default is to not export anything else unless this
2099 option is given. Note that the symbols @code{DllMain@@12},
2100 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2101 @code{impure_ptr} will not be automatically
2102 exported. Also, symbols imported from other DLLs will not be
2103 re-exported, nor will symbols specifying the DLL's internal layout
2104 such as those beginning with @code{_head_} or ending with
2105 @code{_iname}. In addition, no symbols from @code{libgcc},
2106 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2107 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2108 not be exported, to help with C++ DLLs. Finally, there is an
2109 extensive list of cygwin-private symbols that are not exported
2110 (obviously, this applies on when building DLLs for cygwin targets).
2111 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2112 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2113 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2114 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2115 @code{cygwin_premain3}, and @code{environ}.
2116 [This option is specific to the i386 PE targeted port of the linker]
2118 @kindex --exclude-symbols
2119 @item --exclude-symbols @var{symbol},@var{symbol},...
2120 Specifies a list of symbols which should not be automatically
2121 exported. The symbol names may be delimited by commas or colons.
2122 [This option is specific to the i386 PE targeted port of the linker]
2124 @kindex --file-alignment
2125 @item --file-alignment
2126 Specify the file alignment. Sections in the file will always begin at
2127 file offsets which are multiples of this number. This defaults to
2129 [This option is specific to the i386 PE targeted port of the linker]
2133 @item --heap @var{reserve}
2134 @itemx --heap @var{reserve},@var{commit}
2135 Specify the number of bytes of memory to reserve (and optionally commit)
2136 to be used as heap for this program. The default is 1Mb reserved, 4K
2138 [This option is specific to the i386 PE targeted port of the linker]
2141 @kindex --image-base
2142 @item --image-base @var{value}
2143 Use @var{value} as the base address of your program or dll. This is
2144 the lowest memory location that will be used when your program or dll
2145 is loaded. To reduce the need to relocate and improve performance of
2146 your dlls, each should have a unique base address and not overlap any
2147 other dlls. The default is 0x400000 for executables, and 0x10000000
2149 [This option is specific to the i386 PE targeted port of the linker]
2153 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2154 symbols before they are exported.
2155 [This option is specific to the i386 PE targeted port of the linker]
2157 @kindex --large-address-aware
2158 @item --large-address-aware
2159 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2160 header is set to indicate that this executable supports virtual addresses
2161 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2162 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2163 section of the BOOT.INI. Otherwise, this bit has no effect.
2164 [This option is specific to PE targeted ports of the linker]
2166 @kindex --major-image-version
2167 @item --major-image-version @var{value}
2168 Sets the major number of the ``image version''. Defaults to 1.
2169 [This option is specific to the i386 PE targeted port of the linker]
2171 @kindex --major-os-version
2172 @item --major-os-version @var{value}
2173 Sets the major number of the ``os version''. Defaults to 4.
2174 [This option is specific to the i386 PE targeted port of the linker]
2176 @kindex --major-subsystem-version
2177 @item --major-subsystem-version @var{value}
2178 Sets the major number of the ``subsystem version''. Defaults to 4.
2179 [This option is specific to the i386 PE targeted port of the linker]
2181 @kindex --minor-image-version
2182 @item --minor-image-version @var{value}
2183 Sets the minor number of the ``image version''. Defaults to 0.
2184 [This option is specific to the i386 PE targeted port of the linker]
2186 @kindex --minor-os-version
2187 @item --minor-os-version @var{value}
2188 Sets the minor number of the ``os version''. Defaults to 0.
2189 [This option is specific to the i386 PE targeted port of the linker]
2191 @kindex --minor-subsystem-version
2192 @item --minor-subsystem-version @var{value}
2193 Sets the minor number of the ``subsystem version''. Defaults to 0.
2194 [This option is specific to the i386 PE targeted port of the linker]
2196 @cindex DEF files, creating
2197 @cindex DLLs, creating
2198 @kindex --output-def
2199 @item --output-def @var{file}
2200 The linker will create the file @var{file} which will contain a DEF
2201 file corresponding to the DLL the linker is generating. This DEF file
2202 (which should be called @code{*.def}) may be used to create an import
2203 library with @code{dlltool} or may be used as a reference to
2204 automatically or implicitly exported symbols.
2205 [This option is specific to the i386 PE targeted port of the linker]
2207 @cindex DLLs, creating
2208 @kindex --out-implib
2209 @item --out-implib @var{file}
2210 The linker will create the file @var{file} which will contain an
2211 import lib corresponding to the DLL the linker is generating. This
2212 import lib (which should be called @code{*.dll.a} or @code{*.a}
2213 may be used to link clients against the generated DLL; this behaviour
2214 makes it possible to skip a separate @code{dlltool} import library
2216 [This option is specific to the i386 PE targeted port of the linker]
2218 @kindex --enable-auto-image-base
2219 @item --enable-auto-image-base
2220 Automatically choose the image base for DLLs, unless one is specified
2221 using the @code{--image-base} argument. By using a hash generated
2222 from the dllname to create unique image bases for each DLL, in-memory
2223 collisions and relocations which can delay program execution are
2225 [This option is specific to the i386 PE targeted port of the linker]
2227 @kindex --disable-auto-image-base
2228 @item --disable-auto-image-base
2229 Do not automatically generate a unique image base. If there is no
2230 user-specified image base (@code{--image-base}) then use the platform
2232 [This option is specific to the i386 PE targeted port of the linker]
2234 @cindex DLLs, linking to
2235 @kindex --dll-search-prefix
2236 @item --dll-search-prefix @var{string}
2237 When linking dynamically to a dll without an import library,
2238 search for @code{<string><basename>.dll} in preference to
2239 @code{lib<basename>.dll}. This behaviour allows easy distinction
2240 between DLLs built for the various "subplatforms": native, cygwin,
2241 uwin, pw, etc. For instance, cygwin DLLs typically use
2242 @code{--dll-search-prefix=cyg}.
2243 [This option is specific to the i386 PE targeted port of the linker]
2245 @kindex --enable-auto-import
2246 @item --enable-auto-import
2247 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2248 DATA imports from DLLs, and create the necessary thunking symbols when
2249 building the import libraries with those DATA exports. Note: Use of the
2250 'auto-import' extension will cause the text section of the image file
2251 to be made writable. This does not conform to the PE-COFF format
2252 specification published by Microsoft.
2254 Note - use of the 'auto-import' extension will also cause read only
2255 data which would normally be placed into the .rdata section to be
2256 placed into the .data section instead. This is in order to work
2257 around a problem with consts that is described here:
2258 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2260 Using 'auto-import' generally will 'just work' -- but sometimes you may
2263 "variable '<var>' can't be auto-imported. Please read the
2264 documentation for ld's @code{--enable-auto-import} for details."
2266 This message occurs when some (sub)expression accesses an address
2267 ultimately given by the sum of two constants (Win32 import tables only
2268 allow one). Instances where this may occur include accesses to member
2269 fields of struct variables imported from a DLL, as well as using a
2270 constant index into an array variable imported from a DLL. Any
2271 multiword variable (arrays, structs, long long, etc) may trigger
2272 this error condition. However, regardless of the exact data type
2273 of the offending exported variable, ld will always detect it, issue
2274 the warning, and exit.
2276 There are several ways to address this difficulty, regardless of the
2277 data type of the exported variable:
2279 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2280 of adjusting references in your client code for runtime environment, so
2281 this method works only when runtime environment supports this feature.
2283 A second solution is to force one of the 'constants' to be a variable --
2284 that is, unknown and un-optimizable at compile time. For arrays,
2285 there are two possibilities: a) make the indexee (the array's address)
2286 a variable, or b) make the 'constant' index a variable. Thus:
2289 extern type extern_array[];
2291 @{ volatile type *t=extern_array; t[1] @}
2297 extern type extern_array[];
2299 @{ volatile int t=1; extern_array[t] @}
2302 For structs (and most other multiword data types) the only option
2303 is to make the struct itself (or the long long, or the ...) variable:
2306 extern struct s extern_struct;
2307 extern_struct.field -->
2308 @{ volatile struct s *t=&extern_struct; t->field @}
2314 extern long long extern_ll;
2316 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2319 A third method of dealing with this difficulty is to abandon
2320 'auto-import' for the offending symbol and mark it with
2321 @code{__declspec(dllimport)}. However, in practise that
2322 requires using compile-time #defines to indicate whether you are
2323 building a DLL, building client code that will link to the DLL, or
2324 merely building/linking to a static library. In making the choice
2325 between the various methods of resolving the 'direct address with
2326 constant offset' problem, you should consider typical real-world usage:
2334 void main(int argc, char **argv)@{
2335 printf("%d\n",arr[1]);
2345 void main(int argc, char **argv)@{
2346 /* This workaround is for win32 and cygwin; do not "optimize" */
2347 volatile int *parr = arr;
2348 printf("%d\n",parr[1]);
2355 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2356 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2357 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2358 #define FOO_IMPORT __declspec(dllimport)
2362 extern FOO_IMPORT int arr[];
2365 void main(int argc, char **argv)@{
2366 printf("%d\n",arr[1]);
2370 A fourth way to avoid this problem is to re-code your
2371 library to use a functional interface rather than a data interface
2372 for the offending variables (e.g. set_foo() and get_foo() accessor
2374 [This option is specific to the i386 PE targeted port of the linker]
2376 @kindex --disable-auto-import
2377 @item --disable-auto-import
2378 Do not attempt to do sophisticated linking of @code{_symbol} to
2379 @code{__imp__symbol} for DATA imports from DLLs.
2380 [This option is specific to the i386 PE targeted port of the linker]
2382 @kindex --enable-runtime-pseudo-reloc
2383 @item --enable-runtime-pseudo-reloc
2384 If your code contains expressions described in --enable-auto-import section,
2385 that is, DATA imports from DLL with non-zero offset, this switch will create
2386 a vector of 'runtime pseudo relocations' which can be used by runtime
2387 environment to adjust references to such data in your client code.
2388 [This option is specific to the i386 PE targeted port of the linker]
2390 @kindex --disable-runtime-pseudo-reloc
2391 @item --disable-runtime-pseudo-reloc
2392 Do not create pseudo relocations for non-zero offset DATA imports from
2393 DLLs. This is the default.
2394 [This option is specific to the i386 PE targeted port of the linker]
2396 @kindex --enable-extra-pe-debug
2397 @item --enable-extra-pe-debug
2398 Show additional debug info related to auto-import symbol thunking.
2399 [This option is specific to the i386 PE targeted port of the linker]
2401 @kindex --section-alignment
2402 @item --section-alignment
2403 Sets the section alignment. Sections in memory will always begin at
2404 addresses which are a multiple of this number. Defaults to 0x1000.
2405 [This option is specific to the i386 PE targeted port of the linker]
2409 @item --stack @var{reserve}
2410 @itemx --stack @var{reserve},@var{commit}
2411 Specify the number of bytes of memory to reserve (and optionally commit)
2412 to be used as stack for this program. The default is 2Mb reserved, 4K
2414 [This option is specific to the i386 PE targeted port of the linker]
2417 @item --subsystem @var{which}
2418 @itemx --subsystem @var{which}:@var{major}
2419 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2420 Specifies the subsystem under which your program will execute. The
2421 legal values for @var{which} are @code{native}, @code{windows},
2422 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2423 the subsystem version also. Numeric values are also accepted for
2425 [This option is specific to the i386 PE targeted port of the linker]
2432 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2434 @c man begin OPTIONS
2436 The 68HC11 and 68HC12 linkers support specific options to control the
2437 memory bank switching mapping and trampoline code generation.
2441 @kindex --no-trampoline
2442 @item --no-trampoline
2443 This option disables the generation of trampoline. By default a trampoline
2444 is generated for each far function which is called using a @code{jsr}
2445 instruction (this happens when a pointer to a far function is taken).
2447 @kindex --bank-window
2448 @item --bank-window @var{name}
2449 This option indicates to the linker the name of the memory region in
2450 the @samp{MEMORY} specification that describes the memory bank window.
2451 The definition of such region is then used by the linker to compute
2452 paging and addresses within the memory window.
2460 @subsection Options specific to Motorola 68K target
2462 @c man begin OPTIONS
2464 The following options are supported to control handling of GOT generation
2465 when linking for 68K targets.
2470 @item --got=@var{type}
2471 This option tells the linker which GOT generation scheme to use.
2472 @var{type} should be one of @samp{single}, @samp{negative},
2473 @samp{multigot} or @samp{target}. For more information refer to the
2474 Info entry for @file{ld}.
2483 @section Environment Variables
2485 @c man begin ENVIRONMENT
2487 You can change the behaviour of @command{ld} with the environment variables
2488 @ifclear SingleFormat
2491 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2493 @ifclear SingleFormat
2495 @cindex default input format
2496 @code{GNUTARGET} determines the input-file object format if you don't
2497 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2498 of the BFD names for an input format (@pxref{BFD}). If there is no
2499 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2500 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2501 attempts to discover the input format by examining binary input files;
2502 this method often succeeds, but there are potential ambiguities, since
2503 there is no method of ensuring that the magic number used to specify
2504 object-file formats is unique. However, the configuration procedure for
2505 BFD on each system places the conventional format for that system first
2506 in the search-list, so ambiguities are resolved in favor of convention.
2510 @cindex default emulation
2511 @cindex emulation, default
2512 @code{LDEMULATION} determines the default emulation if you don't use the
2513 @samp{-m} option. The emulation can affect various aspects of linker
2514 behaviour, particularly the default linker script. You can list the
2515 available emulations with the @samp{--verbose} or @samp{-V} options. If
2516 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2517 variable is not defined, the default emulation depends upon how the
2518 linker was configured.
2520 @kindex COLLECT_NO_DEMANGLE
2521 @cindex demangling, default
2522 Normally, the linker will default to demangling symbols. However, if
2523 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2524 default to not demangling symbols. This environment variable is used in
2525 a similar fashion by the @code{gcc} linker wrapper program. The default
2526 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2533 @chapter Linker Scripts
2536 @cindex linker scripts
2537 @cindex command files
2538 Every link is controlled by a @dfn{linker script}. This script is
2539 written in the linker command language.
2541 The main purpose of the linker script is to describe how the sections in
2542 the input files should be mapped into the output file, and to control
2543 the memory layout of the output file. Most linker scripts do nothing
2544 more than this. However, when necessary, the linker script can also
2545 direct the linker to perform many other operations, using the commands
2548 The linker always uses a linker script. If you do not supply one
2549 yourself, the linker will use a default script that is compiled into the
2550 linker executable. You can use the @samp{--verbose} command line option
2551 to display the default linker script. Certain command line options,
2552 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2554 You may supply your own linker script by using the @samp{-T} command
2555 line option. When you do this, your linker script will replace the
2556 default linker script.
2558 You may also use linker scripts implicitly by naming them as input files
2559 to the linker, as though they were files to be linked. @xref{Implicit
2563 * Basic Script Concepts:: Basic Linker Script Concepts
2564 * Script Format:: Linker Script Format
2565 * Simple Example:: Simple Linker Script Example
2566 * Simple Commands:: Simple Linker Script Commands
2567 * Assignments:: Assigning Values to Symbols
2568 * SECTIONS:: SECTIONS Command
2569 * MEMORY:: MEMORY Command
2570 * PHDRS:: PHDRS Command
2571 * VERSION:: VERSION Command
2572 * Expressions:: Expressions in Linker Scripts
2573 * Implicit Linker Scripts:: Implicit Linker Scripts
2576 @node Basic Script Concepts
2577 @section Basic Linker Script Concepts
2578 @cindex linker script concepts
2579 We need to define some basic concepts and vocabulary in order to
2580 describe the linker script language.
2582 The linker combines input files into a single output file. The output
2583 file and each input file are in a special data format known as an
2584 @dfn{object file format}. Each file is called an @dfn{object file}.
2585 The output file is often called an @dfn{executable}, but for our
2586 purposes we will also call it an object file. Each object file has,
2587 among other things, a list of @dfn{sections}. We sometimes refer to a
2588 section in an input file as an @dfn{input section}; similarly, a section
2589 in the output file is an @dfn{output section}.
2591 Each section in an object file has a name and a size. Most sections
2592 also have an associated block of data, known as the @dfn{section
2593 contents}. A section may be marked as @dfn{loadable}, which mean that
2594 the contents should be loaded into memory when the output file is run.
2595 A section with no contents may be @dfn{allocatable}, which means that an
2596 area in memory should be set aside, but nothing in particular should be
2597 loaded there (in some cases this memory must be zeroed out). A section
2598 which is neither loadable nor allocatable typically contains some sort
2599 of debugging information.
2601 Every loadable or allocatable output section has two addresses. The
2602 first is the @dfn{VMA}, or virtual memory address. This is the address
2603 the section will have when the output file is run. The second is the
2604 @dfn{LMA}, or load memory address. This is the address at which the
2605 section will be loaded. In most cases the two addresses will be the
2606 same. An example of when they might be different is when a data section
2607 is loaded into ROM, and then copied into RAM when the program starts up
2608 (this technique is often used to initialize global variables in a ROM
2609 based system). In this case the ROM address would be the LMA, and the
2610 RAM address would be the VMA.
2612 You can see the sections in an object file by using the @code{objdump}
2613 program with the @samp{-h} option.
2615 Every object file also has a list of @dfn{symbols}, known as the
2616 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2617 has a name, and each defined symbol has an address, among other
2618 information. If you compile a C or C++ program into an object file, you
2619 will get a defined symbol for every defined function and global or
2620 static variable. Every undefined function or global variable which is
2621 referenced in the input file will become an undefined symbol.
2623 You can see the symbols in an object file by using the @code{nm}
2624 program, or by using the @code{objdump} program with the @samp{-t}
2628 @section Linker Script Format
2629 @cindex linker script format
2630 Linker scripts are text files.
2632 You write a linker script as a series of commands. Each command is
2633 either a keyword, possibly followed by arguments, or an assignment to a
2634 symbol. You may separate commands using semicolons. Whitespace is
2637 Strings such as file or format names can normally be entered directly.
2638 If the file name contains a character such as a comma which would
2639 otherwise serve to separate file names, you may put the file name in
2640 double quotes. There is no way to use a double quote character in a
2643 You may include comments in linker scripts just as in C, delimited by
2644 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2647 @node Simple Example
2648 @section Simple Linker Script Example
2649 @cindex linker script example
2650 @cindex example of linker script
2651 Many linker scripts are fairly simple.
2653 The simplest possible linker script has just one command:
2654 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2655 memory layout of the output file.
2657 The @samp{SECTIONS} command is a powerful command. Here we will
2658 describe a simple use of it. Let's assume your program consists only of
2659 code, initialized data, and uninitialized data. These will be in the
2660 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2661 Let's assume further that these are the only sections which appear in
2664 For this example, let's say that the code should be loaded at address
2665 0x10000, and that the data should start at address 0x8000000. Here is a
2666 linker script which will do that:
2671 .text : @{ *(.text) @}
2673 .data : @{ *(.data) @}
2674 .bss : @{ *(.bss) @}
2678 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2679 followed by a series of symbol assignments and output section
2680 descriptions enclosed in curly braces.
2682 The first line inside the @samp{SECTIONS} command of the above example
2683 sets the value of the special symbol @samp{.}, which is the location
2684 counter. If you do not specify the address of an output section in some
2685 other way (other ways are described later), the address is set from the
2686 current value of the location counter. The location counter is then
2687 incremented by the size of the output section. At the start of the
2688 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2690 The second line defines an output section, @samp{.text}. The colon is
2691 required syntax which may be ignored for now. Within the curly braces
2692 after the output section name, you list the names of the input sections
2693 which should be placed into this output section. The @samp{*} is a
2694 wildcard which matches any file name. The expression @samp{*(.text)}
2695 means all @samp{.text} input sections in all input files.
2697 Since the location counter is @samp{0x10000} when the output section
2698 @samp{.text} is defined, the linker will set the address of the
2699 @samp{.text} section in the output file to be @samp{0x10000}.
2701 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2702 the output file. The linker will place the @samp{.data} output section
2703 at address @samp{0x8000000}. After the linker places the @samp{.data}
2704 output section, the value of the location counter will be
2705 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2706 effect is that the linker will place the @samp{.bss} output section
2707 immediately after the @samp{.data} output section in memory.
2709 The linker will ensure that each output section has the required
2710 alignment, by increasing the location counter if necessary. In this
2711 example, the specified addresses for the @samp{.text} and @samp{.data}
2712 sections will probably satisfy any alignment constraints, but the linker
2713 may have to create a small gap between the @samp{.data} and @samp{.bss}
2716 That's it! That's a simple and complete linker script.
2718 @node Simple Commands
2719 @section Simple Linker Script Commands
2720 @cindex linker script simple commands
2721 In this section we describe the simple linker script commands.
2724 * Entry Point:: Setting the entry point
2725 * File Commands:: Commands dealing with files
2726 @ifclear SingleFormat
2727 * Format Commands:: Commands dealing with object file formats
2730 * Miscellaneous Commands:: Other linker script commands
2734 @subsection Setting the Entry Point
2735 @kindex ENTRY(@var{symbol})
2736 @cindex start of execution
2737 @cindex first instruction
2739 The first instruction to execute in a program is called the @dfn{entry
2740 point}. You can use the @code{ENTRY} linker script command to set the
2741 entry point. The argument is a symbol name:
2746 There are several ways to set the entry point. The linker will set the
2747 entry point by trying each of the following methods in order, and
2748 stopping when one of them succeeds:
2751 the @samp{-e} @var{entry} command-line option;
2753 the @code{ENTRY(@var{symbol})} command in a linker script;
2755 the value of the symbol @code{start}, if defined;
2757 the address of the first byte of the @samp{.text} section, if present;
2759 The address @code{0}.
2763 @subsection Commands Dealing with Files
2764 @cindex linker script file commands
2765 Several linker script commands deal with files.
2768 @item INCLUDE @var{filename}
2769 @kindex INCLUDE @var{filename}
2770 @cindex including a linker script
2771 Include the linker script @var{filename} at this point. The file will
2772 be searched for in the current directory, and in any directory specified
2773 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2776 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
2777 @code{SECTIONS} commands, or in output section descriptions.
2779 @item INPUT(@var{file}, @var{file}, @dots{})
2780 @itemx INPUT(@var{file} @var{file} @dots{})
2781 @kindex INPUT(@var{files})
2782 @cindex input files in linker scripts
2783 @cindex input object files in linker scripts
2784 @cindex linker script input object files
2785 The @code{INPUT} command directs the linker to include the named files
2786 in the link, as though they were named on the command line.
2788 For example, if you always want to include @file{subr.o} any time you do
2789 a link, but you can't be bothered to put it on every link command line,
2790 then you can put @samp{INPUT (subr.o)} in your linker script.
2792 In fact, if you like, you can list all of your input files in the linker
2793 script, and then invoke the linker with nothing but a @samp{-T} option.
2795 In case a @dfn{sysroot prefix} is configured, and the filename starts
2796 with the @samp{/} character, and the script being processed was
2797 located inside the @dfn{sysroot prefix}, the filename will be looked
2798 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2799 open the file in the current directory. If it is not found, the
2800 linker will search through the archive library search path. See the
2801 description of @samp{-L} in @ref{Options,,Command Line Options}.
2803 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2804 name to @code{lib@var{file}.a}, as with the command line argument
2807 When you use the @code{INPUT} command in an implicit linker script, the
2808 files will be included in the link at the point at which the linker
2809 script file is included. This can affect archive searching.
2811 @item GROUP(@var{file}, @var{file}, @dots{})
2812 @itemx GROUP(@var{file} @var{file} @dots{})
2813 @kindex GROUP(@var{files})
2814 @cindex grouping input files
2815 The @code{GROUP} command is like @code{INPUT}, except that the named
2816 files should all be archives, and they are searched repeatedly until no
2817 new undefined references are created. See the description of @samp{-(}
2818 in @ref{Options,,Command Line Options}.
2820 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
2821 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
2822 @kindex AS_NEEDED(@var{files})
2823 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2824 commands, among other filenames. The files listed will be handled
2825 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2826 with the exception of ELF shared libraries, that will be added only
2827 when they are actually needed. This construct essentially enables
2828 @option{--as-needed} option for all the files listed inside of it
2829 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2832 @item OUTPUT(@var{filename})
2833 @kindex OUTPUT(@var{filename})
2834 @cindex output file name in linker script
2835 The @code{OUTPUT} command names the output file. Using
2836 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2837 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2838 Line Options}). If both are used, the command line option takes
2841 You can use the @code{OUTPUT} command to define a default name for the
2842 output file other than the usual default of @file{a.out}.
2844 @item SEARCH_DIR(@var{path})
2845 @kindex SEARCH_DIR(@var{path})
2846 @cindex library search path in linker script
2847 @cindex archive search path in linker script
2848 @cindex search path in linker script
2849 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2850 @command{ld} looks for archive libraries. Using
2851 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2852 on the command line (@pxref{Options,,Command Line Options}). If both
2853 are used, then the linker will search both paths. Paths specified using
2854 the command line option are searched first.
2856 @item STARTUP(@var{filename})
2857 @kindex STARTUP(@var{filename})
2858 @cindex first input file
2859 The @code{STARTUP} command is just like the @code{INPUT} command, except
2860 that @var{filename} will become the first input file to be linked, as
2861 though it were specified first on the command line. This may be useful
2862 when using a system in which the entry point is always the start of the
2866 @ifclear SingleFormat
2867 @node Format Commands
2868 @subsection Commands Dealing with Object File Formats
2869 A couple of linker script commands deal with object file formats.
2872 @item OUTPUT_FORMAT(@var{bfdname})
2873 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2874 @kindex OUTPUT_FORMAT(@var{bfdname})
2875 @cindex output file format in linker script
2876 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2877 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2878 exactly like using @samp{--oformat @var{bfdname}} on the command line
2879 (@pxref{Options,,Command Line Options}). If both are used, the command
2880 line option takes precedence.
2882 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2883 formats based on the @samp{-EB} and @samp{-EL} command line options.
2884 This permits the linker script to set the output format based on the
2887 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2888 will be the first argument, @var{default}. If @samp{-EB} is used, the
2889 output format will be the second argument, @var{big}. If @samp{-EL} is
2890 used, the output format will be the third argument, @var{little}.
2892 For example, the default linker script for the MIPS ELF target uses this
2895 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2897 This says that the default format for the output file is
2898 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
2899 option, the output file will be created in the @samp{elf32-littlemips}
2902 @item TARGET(@var{bfdname})
2903 @kindex TARGET(@var{bfdname})
2904 @cindex input file format in linker script
2905 The @code{TARGET} command names the BFD format to use when reading input
2906 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
2907 This command is like using @samp{-b @var{bfdname}} on the command line
2908 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
2909 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
2910 command is also used to set the format for the output file. @xref{BFD}.
2914 @node Miscellaneous Commands
2915 @subsection Other Linker Script Commands
2916 There are a few other linker scripts commands.
2919 @item ASSERT(@var{exp}, @var{message})
2921 @cindex assertion in linker script
2922 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
2923 with an error code, and print @var{message}.
2925 @item EXTERN(@var{symbol} @var{symbol} @dots{})
2927 @cindex undefined symbol in linker script
2928 Force @var{symbol} to be entered in the output file as an undefined
2929 symbol. Doing this may, for example, trigger linking of additional
2930 modules from standard libraries. You may list several @var{symbol}s for
2931 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
2932 command has the same effect as the @samp{-u} command-line option.
2934 @item FORCE_COMMON_ALLOCATION
2935 @kindex FORCE_COMMON_ALLOCATION
2936 @cindex common allocation in linker script
2937 This command has the same effect as the @samp{-d} command-line option:
2938 to make @command{ld} assign space to common symbols even if a relocatable
2939 output file is specified (@samp{-r}).
2941 @item INHIBIT_COMMON_ALLOCATION
2942 @kindex INHIBIT_COMMON_ALLOCATION
2943 @cindex common allocation in linker script
2944 This command has the same effect as the @samp{--no-define-common}
2945 command-line option: to make @code{ld} omit the assignment of addresses
2946 to common symbols even for a non-relocatable output file.
2948 @item INSERT [ AFTER | BEFORE ] @var{output_section}
2950 @cindex insert user script into default script
2951 This command is typically used in a script specified by @samp{-T} to
2952 augment the default @code{SECTIONS} with, for example, overlays. It
2953 inserts all prior linker script statements after (or before)
2954 @var{output_section}, and also causes @samp{-T} to not override the
2955 default linker script. The exact insertion point is as for orphan
2956 sections. @xref{Location Counter}. The insertion happens after the
2957 linker has mapped input sections to output sections. Prior to the
2958 insertion, since @samp{-T} scripts are parsed before the default
2959 linker script, statements in the @samp{-T} script occur before the
2960 default linker script statements in the internal linker representation
2961 of the script. In particular, input section assignments will be made
2962 to @samp{-T} output sections before those in the default script. Here
2963 is an example of how a @samp{-T} script using @code{INSERT} might look:
2970 .ov1 @{ ov1*(.text) @}
2971 .ov2 @{ ov2*(.text) @}
2977 @item NOCROSSREFS(@var{section} @var{section} @dots{})
2978 @kindex NOCROSSREFS(@var{sections})
2979 @cindex cross references
2980 This command may be used to tell @command{ld} to issue an error about any
2981 references among certain output sections.
2983 In certain types of programs, particularly on embedded systems when
2984 using overlays, when one section is loaded into memory, another section
2985 will not be. Any direct references between the two sections would be
2986 errors. For example, it would be an error if code in one section called
2987 a function defined in the other section.
2989 The @code{NOCROSSREFS} command takes a list of output section names. If
2990 @command{ld} detects any cross references between the sections, it reports
2991 an error and returns a non-zero exit status. Note that the
2992 @code{NOCROSSREFS} command uses output section names, not input section
2995 @ifclear SingleFormat
2996 @item OUTPUT_ARCH(@var{bfdarch})
2997 @kindex OUTPUT_ARCH(@var{bfdarch})
2998 @cindex machine architecture
2999 @cindex architecture
3000 Specify a particular output machine architecture. The argument is one
3001 of the names used by the BFD library (@pxref{BFD}). You can see the
3002 architecture of an object file by using the @code{objdump} program with
3003 the @samp{-f} option.
3008 @section Assigning Values to Symbols
3009 @cindex assignment in scripts
3010 @cindex symbol definition, scripts
3011 @cindex variables, defining
3012 You may assign a value to a symbol in a linker script. This will define
3013 the symbol and place it into the symbol table with a global scope.
3016 * Simple Assignments:: Simple Assignments
3018 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3019 * Source Code Reference:: How to use a linker script defined symbol in source code
3022 @node Simple Assignments
3023 @subsection Simple Assignments
3025 You may assign to a symbol using any of the C assignment operators:
3028 @item @var{symbol} = @var{expression} ;
3029 @itemx @var{symbol} += @var{expression} ;
3030 @itemx @var{symbol} -= @var{expression} ;
3031 @itemx @var{symbol} *= @var{expression} ;
3032 @itemx @var{symbol} /= @var{expression} ;
3033 @itemx @var{symbol} <<= @var{expression} ;
3034 @itemx @var{symbol} >>= @var{expression} ;
3035 @itemx @var{symbol} &= @var{expression} ;
3036 @itemx @var{symbol} |= @var{expression} ;
3039 The first case will define @var{symbol} to the value of
3040 @var{expression}. In the other cases, @var{symbol} must already be
3041 defined, and the value will be adjusted accordingly.
3043 The special symbol name @samp{.} indicates the location counter. You
3044 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3046 The semicolon after @var{expression} is required.
3048 Expressions are defined below; see @ref{Expressions}.
3050 You may write symbol assignments as commands in their own right, or as
3051 statements within a @code{SECTIONS} command, or as part of an output
3052 section description in a @code{SECTIONS} command.
3054 The section of the symbol will be set from the section of the
3055 expression; for more information, see @ref{Expression Section}.
3057 Here is an example showing the three different places that symbol
3058 assignments may be used:
3069 _bdata = (. + 3) & ~ 3;
3070 .data : @{ *(.data) @}
3074 In this example, the symbol @samp{floating_point} will be defined as
3075 zero. The symbol @samp{_etext} will be defined as the address following
3076 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3077 defined as the address following the @samp{.text} output section aligned
3078 upward to a 4 byte boundary.
3083 In some cases, it is desirable for a linker script to define a symbol
3084 only if it is referenced and is not defined by any object included in
3085 the link. For example, traditional linkers defined the symbol
3086 @samp{etext}. However, ANSI C requires that the user be able to use
3087 @samp{etext} as a function name without encountering an error. The
3088 @code{PROVIDE} keyword may be used to define a symbol, such as
3089 @samp{etext}, only if it is referenced but not defined. The syntax is
3090 @code{PROVIDE(@var{symbol} = @var{expression})}.
3092 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3105 In this example, if the program defines @samp{_etext} (with a leading
3106 underscore), the linker will give a multiple definition error. If, on
3107 the other hand, the program defines @samp{etext} (with no leading
3108 underscore), the linker will silently use the definition in the program.
3109 If the program references @samp{etext} but does not define it, the
3110 linker will use the definition in the linker script.
3112 @node PROVIDE_HIDDEN
3113 @subsection PROVIDE_HIDDEN
3114 @cindex PROVIDE_HIDDEN
3115 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3116 hidden and won't be exported.
3118 @node Source Code Reference
3119 @subsection Source Code Reference
3121 Accessing a linker script defined variable from source code is not
3122 intuitive. In particular a linker script symbol is not equivalent to
3123 a variable declaration in a high level language, it is instead a
3124 symbol that does not have a value.
3126 Before going further, it is important to note that compilers often
3127 transform names in the source code into different names when they are
3128 stored in the symbol table. For example, Fortran compilers commonly
3129 prepend or append an underscore, and C++ performs extensive @samp{name
3130 mangling}. Therefore there might be a discrepancy between the name
3131 of a variable as it is used in source code and the name of the same
3132 variable as it is defined in a linker script. For example in C a
3133 linker script variable might be referred to as:
3139 But in the linker script it might be defined as:
3145 In the remaining examples however it is assumed that no name
3146 transformation has taken place.
3148 When a symbol is declared in a high level language such as C, two
3149 things happen. The first is that the compiler reserves enough space
3150 in the program's memory to hold the @emph{value} of the symbol. The
3151 second is that the compiler creates an entry in the program's symbol
3152 table which holds the symbol's @emph{address}. ie the symbol table
3153 contains the address of the block of memory holding the symbol's
3154 value. So for example the following C declaration, at file scope:
3160 creates a entry called @samp{foo} in the symbol table. This entry
3161 holds the address of an @samp{int} sized block of memory where the
3162 number 1000 is initially stored.
3164 When a program references a symbol the compiler generates code that
3165 first accesses the symbol table to find the address of the symbol's
3166 memory block and then code to read the value from that memory block.
3173 looks up the symbol @samp{foo} in the symbol table, gets the address
3174 associated with this symbol and then writes the value 1 into that
3181 looks up the symbol @samp{foo} in the symbol table, gets it address
3182 and then copies this address into the block of memory associated with
3183 the variable @samp{a}.
3185 Linker scripts symbol declarations, by contrast, create an entry in
3186 the symbol table but do not assign any memory to them. Thus they are
3187 an address without a value. So for example the linker script definition:
3193 creates an entry in the symbol table called @samp{foo} which holds
3194 the address of memory location 1000, but nothing special is stored at
3195 address 1000. This means that you cannot access the @emph{value} of a
3196 linker script defined symbol - it has no value - all you can do is
3197 access the @emph{address} of a linker script defined symbol.
3199 Hence when you are using a linker script defined symbol in source code
3200 you should always take the address of the symbol, and never attempt to
3201 use its value. For example suppose you want to copy the contents of a
3202 section of memory called .ROM into a section called .FLASH and the
3203 linker script contains these declarations:
3207 start_of_ROM = .ROM;
3208 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3209 start_of_FLASH = .FLASH;
3213 Then the C source code to perform the copy would be:
3217 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3219 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3223 Note the use of the @samp{&} operators. These are correct.
3226 @section SECTIONS Command
3228 The @code{SECTIONS} command tells the linker how to map input sections
3229 into output sections, and how to place the output sections in memory.
3231 The format of the @code{SECTIONS} command is:
3235 @var{sections-command}
3236 @var{sections-command}
3241 Each @var{sections-command} may of be one of the following:
3245 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3247 a symbol assignment (@pxref{Assignments})
3249 an output section description
3251 an overlay description
3254 The @code{ENTRY} command and symbol assignments are permitted inside the
3255 @code{SECTIONS} command for convenience in using the location counter in
3256 those commands. This can also make the linker script easier to
3257 understand because you can use those commands at meaningful points in
3258 the layout of the output file.
3260 Output section descriptions and overlay descriptions are described
3263 If you do not use a @code{SECTIONS} command in your linker script, the
3264 linker will place each input section into an identically named output
3265 section in the order that the sections are first encountered in the
3266 input files. If all input sections are present in the first file, for
3267 example, the order of sections in the output file will match the order
3268 in the first input file. The first section will be at address zero.
3271 * Output Section Description:: Output section description
3272 * Output Section Name:: Output section name
3273 * Output Section Address:: Output section address
3274 * Input Section:: Input section description
3275 * Output Section Data:: Output section data
3276 * Output Section Keywords:: Output section keywords
3277 * Output Section Discarding:: Output section discarding
3278 * Output Section Attributes:: Output section attributes
3279 * Overlay Description:: Overlay description
3282 @node Output Section Description
3283 @subsection Output Section Description
3284 The full description of an output section looks like this:
3287 @var{section} [@var{address}] [(@var{type})] :
3288 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3290 @var{output-section-command}
3291 @var{output-section-command}
3293 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3297 Most output sections do not use most of the optional section attributes.
3299 The whitespace around @var{section} is required, so that the section
3300 name is unambiguous. The colon and the curly braces are also required.
3301 The line breaks and other white space are optional.
3303 Each @var{output-section-command} may be one of the following:
3307 a symbol assignment (@pxref{Assignments})
3309 an input section description (@pxref{Input Section})
3311 data values to include directly (@pxref{Output Section Data})
3313 a special output section keyword (@pxref{Output Section Keywords})
3316 @node Output Section Name
3317 @subsection Output Section Name
3318 @cindex name, section
3319 @cindex section name
3320 The name of the output section is @var{section}. @var{section} must
3321 meet the constraints of your output format. In formats which only
3322 support a limited number of sections, such as @code{a.out}, the name
3323 must be one of the names supported by the format (@code{a.out}, for
3324 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3325 output format supports any number of sections, but with numbers and not
3326 names (as is the case for Oasys), the name should be supplied as a
3327 quoted numeric string. A section name may consist of any sequence of
3328 characters, but a name which contains any unusual characters such as
3329 commas must be quoted.
3331 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3334 @node Output Section Address
3335 @subsection Output Section Address
3336 @cindex address, section
3337 @cindex section address
3338 The @var{address} is an expression for the VMA (the virtual memory
3339 address) of the output section. If you do not provide @var{address},
3340 the linker will set it based on @var{region} if present, or otherwise
3341 based on the current value of the location counter.
3343 If you provide @var{address}, the address of the output section will be
3344 set to precisely that. If you provide neither @var{address} nor
3345 @var{region}, then the address of the output section will be set to the
3346 current value of the location counter aligned to the alignment
3347 requirements of the output section. The alignment requirement of the
3348 output section is the strictest alignment of any input section contained
3349 within the output section.
3353 .text . : @{ *(.text) @}
3358 .text : @{ *(.text) @}
3361 are subtly different. The first will set the address of the
3362 @samp{.text} output section to the current value of the location
3363 counter. The second will set it to the current value of the location
3364 counter aligned to the strictest alignment of a @samp{.text} input
3367 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3368 For example, if you want to align the section on a 0x10 byte boundary,
3369 so that the lowest four bits of the section address are zero, you could
3370 do something like this:
3372 .text ALIGN(0x10) : @{ *(.text) @}
3375 This works because @code{ALIGN} returns the current location counter
3376 aligned upward to the specified value.
3378 Specifying @var{address} for a section will change the value of the
3382 @subsection Input Section Description
3383 @cindex input sections
3384 @cindex mapping input sections to output sections
3385 The most common output section command is an input section description.
3387 The input section description is the most basic linker script operation.
3388 You use output sections to tell the linker how to lay out your program
3389 in memory. You use input section descriptions to tell the linker how to
3390 map the input files into your memory layout.
3393 * Input Section Basics:: Input section basics
3394 * Input Section Wildcards:: Input section wildcard patterns
3395 * Input Section Common:: Input section for common symbols
3396 * Input Section Keep:: Input section and garbage collection
3397 * Input Section Example:: Input section example
3400 @node Input Section Basics
3401 @subsubsection Input Section Basics
3402 @cindex input section basics
3403 An input section description consists of a file name optionally followed
3404 by a list of section names in parentheses.
3406 The file name and the section name may be wildcard patterns, which we
3407 describe further below (@pxref{Input Section Wildcards}).
3409 The most common input section description is to include all input
3410 sections with a particular name in the output section. For example, to
3411 include all input @samp{.text} sections, you would write:
3416 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3417 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3418 match all files except the ones specified in the EXCLUDE_FILE list. For
3421 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3423 will cause all .ctors sections from all files except @file{crtend.o} and
3424 @file{otherfile.o} to be included.
3426 There are two ways to include more than one section:
3432 The difference between these is the order in which the @samp{.text} and
3433 @samp{.rdata} input sections will appear in the output section. In the
3434 first example, they will be intermingled, appearing in the same order as
3435 they are found in the linker input. In the second example, all
3436 @samp{.text} input sections will appear first, followed by all
3437 @samp{.rdata} input sections.
3439 You can specify a file name to include sections from a particular file.
3440 You would do this if one or more of your files contain special data that
3441 needs to be at a particular location in memory. For example:
3446 You can also specify files within archives by writing a pattern
3447 matching the archive, a colon, then the pattern matching the file,
3448 with no whitespace around the colon.
3452 matches file within archive
3454 matches the whole archive
3456 matches file but not one in an archive
3459 Either one or both of @samp{archive} and @samp{file} can contain shell
3460 wildcards. On DOS based file systems, the linker will assume that a
3461 single letter followed by a colon is a drive specifier, so
3462 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
3463 within an archive called @samp{c}. @samp{archive:file} filespecs may
3464 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
3465 other linker script contexts. For instance, you cannot extract a file
3466 from an archive by using @samp{archive:file} in an @code{INPUT}
3469 If you use a file name without a list of sections, then all sections in
3470 the input file will be included in the output section. This is not
3471 commonly done, but it may by useful on occasion. For example:
3476 When you use a file name which is not an @samp{archive:file} specifier
3477 and does not contain any wild card
3478 characters, the linker will first see if you also specified the file
3479 name on the linker command line or in an @code{INPUT} command. If you
3480 did not, the linker will attempt to open the file as an input file, as
3481 though it appeared on the command line. Note that this differs from an
3482 @code{INPUT} command, because the linker will not search for the file in
3483 the archive search path.
3485 @node Input Section Wildcards
3486 @subsubsection Input Section Wildcard Patterns
3487 @cindex input section wildcards
3488 @cindex wildcard file name patterns
3489 @cindex file name wildcard patterns
3490 @cindex section name wildcard patterns
3491 In an input section description, either the file name or the section
3492 name or both may be wildcard patterns.
3494 The file name of @samp{*} seen in many examples is a simple wildcard
3495 pattern for the file name.
3497 The wildcard patterns are like those used by the Unix shell.
3501 matches any number of characters
3503 matches any single character
3505 matches a single instance of any of the @var{chars}; the @samp{-}
3506 character may be used to specify a range of characters, as in
3507 @samp{[a-z]} to match any lower case letter
3509 quotes the following character
3512 When a file name is matched with a wildcard, the wildcard characters
3513 will not match a @samp{/} character (used to separate directory names on
3514 Unix). A pattern consisting of a single @samp{*} character is an
3515 exception; it will always match any file name, whether it contains a
3516 @samp{/} or not. In a section name, the wildcard characters will match
3517 a @samp{/} character.
3519 File name wildcard patterns only match files which are explicitly
3520 specified on the command line or in an @code{INPUT} command. The linker
3521 does not search directories to expand wildcards.
3523 If a file name matches more than one wildcard pattern, or if a file name
3524 appears explicitly and is also matched by a wildcard pattern, the linker
3525 will use the first match in the linker script. For example, this
3526 sequence of input section descriptions is probably in error, because the
3527 @file{data.o} rule will not be used:
3529 .data : @{ *(.data) @}
3530 .data1 : @{ data.o(.data) @}
3533 @cindex SORT_BY_NAME
3534 Normally, the linker will place files and sections matched by wildcards
3535 in the order in which they are seen during the link. You can change
3536 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3537 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
3538 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3539 into ascending order by name before placing them in the output file.
3541 @cindex SORT_BY_ALIGNMENT
3542 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3543 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3544 ascending order by alignment before placing them in the output file.
3547 @code{SORT} is an alias for @code{SORT_BY_NAME}.
3549 When there are nested section sorting commands in linker script, there
3550 can be at most 1 level of nesting for section sorting commands.
3554 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3555 It will sort the input sections by name first, then by alignment if 2
3556 sections have the same name.
3558 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3559 It will sort the input sections by alignment first, then by name if 2
3560 sections have the same alignment.
3562 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3563 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3565 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3566 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3568 All other nested section sorting commands are invalid.
3571 When both command line section sorting option and linker script
3572 section sorting command are used, section sorting command always
3573 takes precedence over the command line option.
3575 If the section sorting command in linker script isn't nested, the
3576 command line option will make the section sorting command to be
3577 treated as nested sorting command.
3581 @code{SORT_BY_NAME} (wildcard section pattern ) with
3582 @option{--sort-sections alignment} is equivalent to
3583 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3585 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3586 @option{--sort-section name} is equivalent to
3587 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3590 If the section sorting command in linker script is nested, the
3591 command line option will be ignored.
3593 If you ever get confused about where input sections are going, use the
3594 @samp{-M} linker option to generate a map file. The map file shows
3595 precisely how input sections are mapped to output sections.
3597 This example shows how wildcard patterns might be used to partition
3598 files. This linker script directs the linker to place all @samp{.text}
3599 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3600 The linker will place the @samp{.data} section from all files beginning
3601 with an upper case character in @samp{.DATA}; for all other files, the
3602 linker will place the @samp{.data} section in @samp{.data}.
3606 .text : @{ *(.text) @}
3607 .DATA : @{ [A-Z]*(.data) @}
3608 .data : @{ *(.data) @}
3609 .bss : @{ *(.bss) @}
3614 @node Input Section Common
3615 @subsubsection Input Section for Common Symbols
3616 @cindex common symbol placement
3617 @cindex uninitialized data placement
3618 A special notation is needed for common symbols, because in many object
3619 file formats common symbols do not have a particular input section. The
3620 linker treats common symbols as though they are in an input section
3621 named @samp{COMMON}.
3623 You may use file names with the @samp{COMMON} section just as with any
3624 other input sections. You can use this to place common symbols from a
3625 particular input file in one section while common symbols from other
3626 input files are placed in another section.
3628 In most cases, common symbols in input files will be placed in the
3629 @samp{.bss} section in the output file. For example:
3631 .bss @{ *(.bss) *(COMMON) @}
3634 @cindex scommon section
3635 @cindex small common symbols
3636 Some object file formats have more than one type of common symbol. For
3637 example, the MIPS ELF object file format distinguishes standard common
3638 symbols and small common symbols. In this case, the linker will use a
3639 different special section name for other types of common symbols. In
3640 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3641 symbols and @samp{.scommon} for small common symbols. This permits you
3642 to map the different types of common symbols into memory at different
3646 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3647 notation is now considered obsolete. It is equivalent to
3650 @node Input Section Keep
3651 @subsubsection Input Section and Garbage Collection
3653 @cindex garbage collection
3654 When link-time garbage collection is in use (@samp{--gc-sections}),
3655 it is often useful to mark sections that should not be eliminated.
3656 This is accomplished by surrounding an input section's wildcard entry
3657 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3658 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3660 @node Input Section Example
3661 @subsubsection Input Section Example
3662 The following example is a complete linker script. It tells the linker
3663 to read all of the sections from file @file{all.o} and place them at the
3664 start of output section @samp{outputa} which starts at location
3665 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3666 follows immediately, in the same output section. All of section
3667 @samp{.input2} from @file{foo.o} goes into output section
3668 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3669 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3670 files are written to output section @samp{outputc}.
3698 @node Output Section Data
3699 @subsection Output Section Data
3701 @cindex section data
3702 @cindex output section data
3703 @kindex BYTE(@var{expression})
3704 @kindex SHORT(@var{expression})
3705 @kindex LONG(@var{expression})
3706 @kindex QUAD(@var{expression})
3707 @kindex SQUAD(@var{expression})
3708 You can include explicit bytes of data in an output section by using
3709 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3710 an output section command. Each keyword is followed by an expression in
3711 parentheses providing the value to store (@pxref{Expressions}). The
3712 value of the expression is stored at the current value of the location
3715 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3716 store one, two, four, and eight bytes (respectively). After storing the
3717 bytes, the location counter is incremented by the number of bytes
3720 For example, this will store the byte 1 followed by the four byte value
3721 of the symbol @samp{addr}:
3727 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3728 same; they both store an 8 byte, or 64 bit, value. When both host and
3729 target are 32 bits, an expression is computed as 32 bits. In this case
3730 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3731 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
3733 If the object file format of the output file has an explicit endianness,
3734 which is the normal case, the value will be stored in that endianness.
3735 When the object file format does not have an explicit endianness, as is
3736 true of, for example, S-records, the value will be stored in the
3737 endianness of the first input object file.
3739 Note---these commands only work inside a section description and not
3740 between them, so the following will produce an error from the linker:
3742 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
3744 whereas this will work:
3746 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
3749 @kindex FILL(@var{expression})
3750 @cindex holes, filling
3751 @cindex unspecified memory
3752 You may use the @code{FILL} command to set the fill pattern for the
3753 current section. It is followed by an expression in parentheses. Any
3754 otherwise unspecified regions of memory within the section (for example,
3755 gaps left due to the required alignment of input sections) are filled
3756 with the value of the expression, repeated as
3757 necessary. A @code{FILL} statement covers memory locations after the
3758 point at which it occurs in the section definition; by including more
3759 than one @code{FILL} statement, you can have different fill patterns in
3760 different parts of an output section.
3762 This example shows how to fill unspecified regions of memory with the
3768 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
3769 section attribute, but it only affects the
3770 part of the section following the @code{FILL} command, rather than the
3771 entire section. If both are used, the @code{FILL} command takes
3772 precedence. @xref{Output Section Fill}, for details on the fill
3775 @node Output Section Keywords
3776 @subsection Output Section Keywords
3777 There are a couple of keywords which can appear as output section
3781 @kindex CREATE_OBJECT_SYMBOLS
3782 @cindex input filename symbols
3783 @cindex filename symbols
3784 @item CREATE_OBJECT_SYMBOLS
3785 The command tells the linker to create a symbol for each input file.
3786 The name of each symbol will be the name of the corresponding input
3787 file. The section of each symbol will be the output section in which
3788 the @code{CREATE_OBJECT_SYMBOLS} command appears.
3790 This is conventional for the a.out object file format. It is not
3791 normally used for any other object file format.
3793 @kindex CONSTRUCTORS
3794 @cindex C++ constructors, arranging in link
3795 @cindex constructors, arranging in link
3797 When linking using the a.out object file format, the linker uses an
3798 unusual set construct to support C++ global constructors and
3799 destructors. When linking object file formats which do not support
3800 arbitrary sections, such as ECOFF and XCOFF, the linker will
3801 automatically recognize C++ global constructors and destructors by name.
3802 For these object file formats, the @code{CONSTRUCTORS} command tells the
3803 linker to place constructor information in the output section where the
3804 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
3805 ignored for other object file formats.
3807 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
3808 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
3809 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
3810 the start and end of the global destructors. The
3811 first word in the list is the number of entries, followed by the address
3812 of each constructor or destructor, followed by a zero word. The
3813 compiler must arrange to actually run the code. For these object file
3814 formats @sc{gnu} C++ normally calls constructors from a subroutine
3815 @code{__main}; a call to @code{__main} is automatically inserted into
3816 the startup code for @code{main}. @sc{gnu} C++ normally runs
3817 destructors either by using @code{atexit}, or directly from the function
3820 For object file formats such as @code{COFF} or @code{ELF} which support
3821 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
3822 addresses of global constructors and destructors into the @code{.ctors}
3823 and @code{.dtors} sections. Placing the following sequence into your
3824 linker script will build the sort of table which the @sc{gnu} C++
3825 runtime code expects to see.
3829 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3834 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3840 If you are using the @sc{gnu} C++ support for initialization priority,
3841 which provides some control over the order in which global constructors
3842 are run, you must sort the constructors at link time to ensure that they
3843 are executed in the correct order. When using the @code{CONSTRUCTORS}
3844 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
3845 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
3846 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
3849 Normally the compiler and linker will handle these issues automatically,
3850 and you will not need to concern yourself with them. However, you may
3851 need to consider this if you are using C++ and writing your own linker
3856 @node Output Section Discarding
3857 @subsection Output Section Discarding
3858 @cindex discarding sections
3859 @cindex sections, discarding
3860 @cindex removing sections
3861 The linker will not create output sections with no contents. This is
3862 for convenience when referring to input sections that may or may not
3863 be present in any of the input files. For example:
3865 .foo : @{ *(.foo) @}
3868 will only create a @samp{.foo} section in the output file if there is a
3869 @samp{.foo} section in at least one input file, and if the input
3870 sections are not all empty. Other link script directives that allocate
3871 space in an output section will also create the output section.
3873 The linker will ignore address assignments (@pxref{Output Section Address})
3874 on discarded output sections, except when the linker script defines
3875 symbols in the output section. In that case the linker will obey
3876 the address assignments, possibly advancing dot even though the
3877 section is discarded.
3880 The special output section name @samp{/DISCARD/} may be used to discard
3881 input sections. Any input sections which are assigned to an output
3882 section named @samp{/DISCARD/} are not included in the output file.
3884 @node Output Section Attributes
3885 @subsection Output Section Attributes
3886 @cindex output section attributes
3887 We showed above that the full description of an output section looked
3891 @var{section} [@var{address}] [(@var{type})] :
3892 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3894 @var{output-section-command}
3895 @var{output-section-command}
3897 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3900 We've already described @var{section}, @var{address}, and
3901 @var{output-section-command}. In this section we will describe the
3902 remaining section attributes.
3905 * Output Section Type:: Output section type
3906 * Output Section LMA:: Output section LMA
3907 * Forced Output Alignment:: Forced Output Alignment
3908 * Forced Input Alignment:: Forced Input Alignment
3909 * Output Section Region:: Output section region
3910 * Output Section Phdr:: Output section phdr
3911 * Output Section Fill:: Output section fill
3914 @node Output Section Type
3915 @subsubsection Output Section Type
3916 Each output section may have a type. The type is a keyword in
3917 parentheses. The following types are defined:
3921 The section should be marked as not loadable, so that it will not be
3922 loaded into memory when the program is run.
3927 These type names are supported for backward compatibility, and are
3928 rarely used. They all have the same effect: the section should be
3929 marked as not allocatable, so that no memory is allocated for the
3930 section when the program is run.
3934 @cindex prevent unnecessary loading
3935 @cindex loading, preventing
3936 The linker normally sets the attributes of an output section based on
3937 the input sections which map into it. You can override this by using
3938 the section type. For example, in the script sample below, the
3939 @samp{ROM} section is addressed at memory location @samp{0} and does not
3940 need to be loaded when the program is run. The contents of the
3941 @samp{ROM} section will appear in the linker output file as usual.
3945 ROM 0 (NOLOAD) : @{ @dots{} @}
3951 @node Output Section LMA
3952 @subsubsection Output Section LMA
3953 @kindex AT>@var{lma_region}
3954 @kindex AT(@var{lma})
3955 @cindex load address
3956 @cindex section load address
3957 Every section has a virtual address (VMA) and a load address (LMA); see
3958 @ref{Basic Script Concepts}. The address expression which may appear in
3959 an output section description sets the VMA (@pxref{Output Section
3962 The expression @var{lma} that follows the @code{AT} keyword specifies
3963 the load address of the section.
3965 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
3966 specify a memory region for the section's load address. @xref{MEMORY}.
3967 Note that if the section has not had a VMA assigned to it then the
3968 linker will use the @var{lma_region} as the VMA region as well.
3970 If neither @code{AT} nor @code{AT>} is specified for an allocatable
3971 section, the linker will set the LMA such that the difference between
3972 VMA and LMA for the section is the same as the preceding output
3973 section in the same region. If there is no preceding output section
3974 or the section is not allocatable, the linker will set the LMA equal
3976 @xref{Output Section Region}.
3978 @cindex ROM initialized data
3979 @cindex initialized data in ROM
3980 This feature is designed to make it easy to build a ROM image. For
3981 example, the following linker script creates three output sections: one
3982 called @samp{.text}, which starts at @code{0x1000}, one called
3983 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
3984 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
3985 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
3986 defined with the value @code{0x2000}, which shows that the location
3987 counter holds the VMA value, not the LMA value.
3993 .text 0x1000 : @{ *(.text) _etext = . ; @}
3995 AT ( ADDR (.text) + SIZEOF (.text) )
3996 @{ _data = . ; *(.data); _edata = . ; @}
3998 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
4003 The run-time initialization code for use with a program generated with
4004 this linker script would include something like the following, to copy
4005 the initialized data from the ROM image to its runtime address. Notice
4006 how this code takes advantage of the symbols defined by the linker
4011 extern char _etext, _data, _edata, _bstart, _bend;
4012 char *src = &_etext;
4015 /* ROM has data at end of text; copy it. */
4016 while (dst < &_edata) @{
4021 for (dst = &_bstart; dst< &_bend; dst++)
4026 @node Forced Output Alignment
4027 @subsubsection Forced Output Alignment
4028 @kindex ALIGN(@var{section_align})
4029 @cindex forcing output section alignment
4030 @cindex output section alignment
4031 You can increase an output section's alignment by using ALIGN.
4033 @node Forced Input Alignment
4034 @subsubsection Forced Input Alignment
4035 @kindex SUBALIGN(@var{subsection_align})
4036 @cindex forcing input section alignment
4037 @cindex input section alignment
4038 You can force input section alignment within an output section by using
4039 SUBALIGN. The value specified overrides any alignment given by input
4040 sections, whether larger or smaller.
4042 @node Output Section Region
4043 @subsubsection Output Section Region
4044 @kindex >@var{region}
4045 @cindex section, assigning to memory region
4046 @cindex memory regions and sections
4047 You can assign a section to a previously defined region of memory by
4048 using @samp{>@var{region}}. @xref{MEMORY}.
4050 Here is a simple example:
4053 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4054 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4058 @node Output Section Phdr
4059 @subsubsection Output Section Phdr
4061 @cindex section, assigning to program header
4062 @cindex program headers and sections
4063 You can assign a section to a previously defined program segment by
4064 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4065 one or more segments, then all subsequent allocated sections will be
4066 assigned to those segments as well, unless they use an explicitly
4067 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4068 linker to not put the section in any segment at all.
4070 Here is a simple example:
4073 PHDRS @{ text PT_LOAD ; @}
4074 SECTIONS @{ .text : @{ *(.text) @} :text @}
4078 @node Output Section Fill
4079 @subsubsection Output Section Fill
4080 @kindex =@var{fillexp}
4081 @cindex section fill pattern
4082 @cindex fill pattern, entire section
4083 You can set the fill pattern for an entire section by using
4084 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4085 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4086 within the output section (for example, gaps left due to the required
4087 alignment of input sections) will be filled with the value, repeated as
4088 necessary. If the fill expression is a simple hex number, ie. a string
4089 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4090 an arbitrarily long sequence of hex digits can be used to specify the
4091 fill pattern; Leading zeros become part of the pattern too. For all
4092 other cases, including extra parentheses or a unary @code{+}, the fill
4093 pattern is the four least significant bytes of the value of the
4094 expression. In all cases, the number is big-endian.
4096 You can also change the fill value with a @code{FILL} command in the
4097 output section commands; (@pxref{Output Section Data}).
4099 Here is a simple example:
4102 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4106 @node Overlay Description
4107 @subsection Overlay Description
4110 An overlay description provides an easy way to describe sections which
4111 are to be loaded as part of a single memory image but are to be run at
4112 the same memory address. At run time, some sort of overlay manager will
4113 copy the overlaid sections in and out of the runtime memory address as
4114 required, perhaps by simply manipulating addressing bits. This approach
4115 can be useful, for example, when a certain region of memory is faster
4118 Overlays are described using the @code{OVERLAY} command. The
4119 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4120 output section description. The full syntax of the @code{OVERLAY}
4121 command is as follows:
4124 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4128 @var{output-section-command}
4129 @var{output-section-command}
4131 @} [:@var{phdr}@dots{}] [=@var{fill}]
4134 @var{output-section-command}
4135 @var{output-section-command}
4137 @} [:@var{phdr}@dots{}] [=@var{fill}]
4139 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4143 Everything is optional except @code{OVERLAY} (a keyword), and each
4144 section must have a name (@var{secname1} and @var{secname2} above). The
4145 section definitions within the @code{OVERLAY} construct are identical to
4146 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
4147 except that no addresses and no memory regions may be defined for
4148 sections within an @code{OVERLAY}.
4150 The sections are all defined with the same starting address. The load
4151 addresses of the sections are arranged such that they are consecutive in
4152 memory starting at the load address used for the @code{OVERLAY} as a
4153 whole (as with normal section definitions, the load address is optional,
4154 and defaults to the start address; the start address is also optional,
4155 and defaults to the current value of the location counter).
4157 If the @code{NOCROSSREFS} keyword is used, and there any references
4158 among the sections, the linker will report an error. Since the sections
4159 all run at the same address, it normally does not make sense for one
4160 section to refer directly to another. @xref{Miscellaneous Commands,
4163 For each section within the @code{OVERLAY}, the linker automatically
4164 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4165 defined as the starting load address of the section. The symbol
4166 @code{__load_stop_@var{secname}} is defined as the final load address of
4167 the section. Any characters within @var{secname} which are not legal
4168 within C identifiers are removed. C (or assembler) code may use these
4169 symbols to move the overlaid sections around as necessary.
4171 At the end of the overlay, the value of the location counter is set to
4172 the start address of the overlay plus the size of the largest section.
4174 Here is an example. Remember that this would appear inside a
4175 @code{SECTIONS} construct.
4178 OVERLAY 0x1000 : AT (0x4000)
4180 .text0 @{ o1/*.o(.text) @}
4181 .text1 @{ o2/*.o(.text) @}
4186 This will define both @samp{.text0} and @samp{.text1} to start at
4187 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4188 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4189 following symbols will be defined if referenced: @code{__load_start_text0},
4190 @code{__load_stop_text0}, @code{__load_start_text1},
4191 @code{__load_stop_text1}.
4193 C code to copy overlay @code{.text1} into the overlay area might look
4198 extern char __load_start_text1, __load_stop_text1;
4199 memcpy ((char *) 0x1000, &__load_start_text1,
4200 &__load_stop_text1 - &__load_start_text1);
4204 Note that the @code{OVERLAY} command is just syntactic sugar, since
4205 everything it does can be done using the more basic commands. The above
4206 example could have been written identically as follows.
4210 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4211 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4212 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4213 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4214 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4215 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4216 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4221 @section MEMORY Command
4223 @cindex memory regions
4224 @cindex regions of memory
4225 @cindex allocating memory
4226 @cindex discontinuous memory
4227 The linker's default configuration permits allocation of all available
4228 memory. You can override this by using the @code{MEMORY} command.
4230 The @code{MEMORY} command describes the location and size of blocks of
4231 memory in the target. You can use it to describe which memory regions
4232 may be used by the linker, and which memory regions it must avoid. You
4233 can then assign sections to particular memory regions. The linker will
4234 set section addresses based on the memory regions, and will warn about
4235 regions that become too full. The linker will not shuffle sections
4236 around to fit into the available regions.
4238 A linker script may contain at most one use of the @code{MEMORY}
4239 command. However, you can define as many blocks of memory within it as
4240 you wish. The syntax is:
4245 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4251 The @var{name} is a name used in the linker script to refer to the
4252 region. The region name has no meaning outside of the linker script.
4253 Region names are stored in a separate name space, and will not conflict
4254 with symbol names, file names, or section names. Each memory region
4255 must have a distinct name.
4257 @cindex memory region attributes
4258 The @var{attr} string is an optional list of attributes that specify
4259 whether to use a particular memory region for an input section which is
4260 not explicitly mapped in the linker script. As described in
4261 @ref{SECTIONS}, if you do not specify an output section for some input
4262 section, the linker will create an output section with the same name as
4263 the input section. If you define region attributes, the linker will use
4264 them to select the memory region for the output section that it creates.
4266 The @var{attr} string must consist only of the following characters:
4281 Invert the sense of any of the preceding attributes
4284 If a unmapped section matches any of the listed attributes other than
4285 @samp{!}, it will be placed in the memory region. The @samp{!}
4286 attribute reverses this test, so that an unmapped section will be placed
4287 in the memory region only if it does not match any of the listed
4293 The @var{origin} is an numerical expression for the start address of
4294 the memory region. The expression must evaluate to a constant and it
4295 cannot involve any symbols. The keyword @code{ORIGIN} may be
4296 abbreviated to @code{org} or @code{o} (but not, for example,
4302 The @var{len} is an expression for the size in bytes of the memory
4303 region. As with the @var{origin} expression, the expression must
4304 be numerical only and must evaluate to a constant. The keyword
4305 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4307 In the following example, we specify that there are two memory regions
4308 available for allocation: one starting at @samp{0} for 256 kilobytes,
4309 and the other starting at @samp{0x40000000} for four megabytes. The
4310 linker will place into the @samp{rom} memory region every section which
4311 is not explicitly mapped into a memory region, and is either read-only
4312 or executable. The linker will place other sections which are not
4313 explicitly mapped into a memory region into the @samp{ram} memory
4320 rom (rx) : ORIGIN = 0, LENGTH = 256K
4321 ram (!rx) : org = 0x40000000, l = 4M
4326 Once you define a memory region, you can direct the linker to place
4327 specific output sections into that memory region by using the
4328 @samp{>@var{region}} output section attribute. For example, if you have
4329 a memory region named @samp{mem}, you would use @samp{>mem} in the
4330 output section definition. @xref{Output Section Region}. If no address
4331 was specified for the output section, the linker will set the address to
4332 the next available address within the memory region. If the combined
4333 output sections directed to a memory region are too large for the
4334 region, the linker will issue an error message.
4336 It is possible to access the origin and length of a memory in an
4337 expression via the @code{ORIGIN(@var{memory})} and
4338 @code{LENGTH(@var{memory})} functions:
4342 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4347 @section PHDRS Command
4349 @cindex program headers
4350 @cindex ELF program headers
4351 @cindex program segments
4352 @cindex segments, ELF
4353 The ELF object file format uses @dfn{program headers}, also knows as
4354 @dfn{segments}. The program headers describe how the program should be
4355 loaded into memory. You can print them out by using the @code{objdump}
4356 program with the @samp{-p} option.
4358 When you run an ELF program on a native ELF system, the system loader
4359 reads the program headers in order to figure out how to load the
4360 program. This will only work if the program headers are set correctly.
4361 This manual does not describe the details of how the system loader
4362 interprets program headers; for more information, see the ELF ABI.
4364 The linker will create reasonable program headers by default. However,
4365 in some cases, you may need to specify the program headers more
4366 precisely. You may use the @code{PHDRS} command for this purpose. When
4367 the linker sees the @code{PHDRS} command in the linker script, it will
4368 not create any program headers other than the ones specified.
4370 The linker only pays attention to the @code{PHDRS} command when
4371 generating an ELF output file. In other cases, the linker will simply
4372 ignore @code{PHDRS}.
4374 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4375 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4381 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4382 [ FLAGS ( @var{flags} ) ] ;
4387 The @var{name} is used only for reference in the @code{SECTIONS} command
4388 of the linker script. It is not put into the output file. Program
4389 header names are stored in a separate name space, and will not conflict
4390 with symbol names, file names, or section names. Each program header
4391 must have a distinct name.
4393 Certain program header types describe segments of memory which the
4394 system loader will load from the file. In the linker script, you
4395 specify the contents of these segments by placing allocatable output
4396 sections in the segments. You use the @samp{:@var{phdr}} output section
4397 attribute to place a section in a particular segment. @xref{Output
4400 It is normal to put certain sections in more than one segment. This
4401 merely implies that one segment of memory contains another. You may
4402 repeat @samp{:@var{phdr}}, using it once for each segment which should
4403 contain the section.
4405 If you place a section in one or more segments using @samp{:@var{phdr}},
4406 then the linker will place all subsequent allocatable sections which do
4407 not specify @samp{:@var{phdr}} in the same segments. This is for
4408 convenience, since generally a whole set of contiguous sections will be
4409 placed in a single segment. You can use @code{:NONE} to override the
4410 default segment and tell the linker to not put the section in any
4415 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4416 the program header type to further describe the contents of the segment.
4417 The @code{FILEHDR} keyword means that the segment should include the ELF
4418 file header. The @code{PHDRS} keyword means that the segment should
4419 include the ELF program headers themselves.
4421 The @var{type} may be one of the following. The numbers indicate the
4422 value of the keyword.
4425 @item @code{PT_NULL} (0)
4426 Indicates an unused program header.
4428 @item @code{PT_LOAD} (1)
4429 Indicates that this program header describes a segment to be loaded from
4432 @item @code{PT_DYNAMIC} (2)
4433 Indicates a segment where dynamic linking information can be found.
4435 @item @code{PT_INTERP} (3)
4436 Indicates a segment where the name of the program interpreter may be
4439 @item @code{PT_NOTE} (4)
4440 Indicates a segment holding note information.
4442 @item @code{PT_SHLIB} (5)
4443 A reserved program header type, defined but not specified by the ELF
4446 @item @code{PT_PHDR} (6)
4447 Indicates a segment where the program headers may be found.
4449 @item @var{expression}
4450 An expression giving the numeric type of the program header. This may
4451 be used for types not defined above.
4454 You can specify that a segment should be loaded at a particular address
4455 in memory by using an @code{AT} expression. This is identical to the
4456 @code{AT} command used as an output section attribute (@pxref{Output
4457 Section LMA}). The @code{AT} command for a program header overrides the
4458 output section attribute.
4460 The linker will normally set the segment flags based on the sections
4461 which comprise the segment. You may use the @code{FLAGS} keyword to
4462 explicitly specify the segment flags. The value of @var{flags} must be
4463 an integer. It is used to set the @code{p_flags} field of the program
4466 Here is an example of @code{PHDRS}. This shows a typical set of program
4467 headers used on a native ELF system.
4473 headers PT_PHDR PHDRS ;
4475 text PT_LOAD FILEHDR PHDRS ;
4477 dynamic PT_DYNAMIC ;
4483 .interp : @{ *(.interp) @} :text :interp
4484 .text : @{ *(.text) @} :text
4485 .rodata : @{ *(.rodata) @} /* defaults to :text */
4487 . = . + 0x1000; /* move to a new page in memory */
4488 .data : @{ *(.data) @} :data
4489 .dynamic : @{ *(.dynamic) @} :data :dynamic
4496 @section VERSION Command
4497 @kindex VERSION @{script text@}
4498 @cindex symbol versions
4499 @cindex version script
4500 @cindex versions of symbols
4501 The linker supports symbol versions when using ELF. Symbol versions are
4502 only useful when using shared libraries. The dynamic linker can use
4503 symbol versions to select a specific version of a function when it runs
4504 a program that may have been linked against an earlier version of the
4507 You can include a version script directly in the main linker script, or
4508 you can supply the version script as an implicit linker script. You can
4509 also use the @samp{--version-script} linker option.
4511 The syntax of the @code{VERSION} command is simply
4513 VERSION @{ version-script-commands @}
4516 The format of the version script commands is identical to that used by
4517 Sun's linker in Solaris 2.5. The version script defines a tree of
4518 version nodes. You specify the node names and interdependencies in the
4519 version script. You can specify which symbols are bound to which
4520 version nodes, and you can reduce a specified set of symbols to local
4521 scope so that they are not globally visible outside of the shared
4524 The easiest way to demonstrate the version script language is with a few
4545 "int f(int, double)";
4550 This example version script defines three version nodes. The first
4551 version node defined is @samp{VERS_1.1}; it has no other dependencies.
4552 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
4553 a number of symbols to local scope so that they are not visible outside
4554 of the shared library; this is done using wildcard patterns, so that any
4555 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4556 is matched. The wildcard patterns available are the same as those used
4557 in the shell when matching filenames (also known as ``globbing'').
4558 However, if you specify the symbol name inside double quotes, then the
4559 name is treated as literal, rather than as a glob pattern.
4561 Next, the version script defines node @samp{VERS_1.2}. This node
4562 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
4563 to the version node @samp{VERS_1.2}.
4565 Finally, the version script defines node @samp{VERS_2.0}. This node
4566 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
4567 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4569 When the linker finds a symbol defined in a library which is not
4570 specifically bound to a version node, it will effectively bind it to an
4571 unspecified base version of the library. You can bind all otherwise
4572 unspecified symbols to a given version node by using @samp{global: *;}
4573 somewhere in the version script.
4575 The names of the version nodes have no specific meaning other than what
4576 they might suggest to the person reading them. The @samp{2.0} version
4577 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4578 However, this would be a confusing way to write a version script.
4580 Node name can be omitted, provided it is the only version node
4581 in the version script. Such version script doesn't assign any versions to
4582 symbols, only selects which symbols will be globally visible out and which
4586 @{ global: foo; bar; local: *; @};
4589 When you link an application against a shared library that has versioned
4590 symbols, the application itself knows which version of each symbol it
4591 requires, and it also knows which version nodes it needs from each
4592 shared library it is linked against. Thus at runtime, the dynamic
4593 loader can make a quick check to make sure that the libraries you have
4594 linked against do in fact supply all of the version nodes that the
4595 application will need to resolve all of the dynamic symbols. In this
4596 way it is possible for the dynamic linker to know with certainty that
4597 all external symbols that it needs will be resolvable without having to
4598 search for each symbol reference.
4600 The symbol versioning is in effect a much more sophisticated way of
4601 doing minor version checking that SunOS does. The fundamental problem
4602 that is being addressed here is that typically references to external
4603 functions are bound on an as-needed basis, and are not all bound when
4604 the application starts up. If a shared library is out of date, a
4605 required interface may be missing; when the application tries to use
4606 that interface, it may suddenly and unexpectedly fail. With symbol
4607 versioning, the user will get a warning when they start their program if
4608 the libraries being used with the application are too old.
4610 There are several GNU extensions to Sun's versioning approach. The
4611 first of these is the ability to bind a symbol to a version node in the
4612 source file where the symbol is defined instead of in the versioning
4613 script. This was done mainly to reduce the burden on the library
4614 maintainer. You can do this by putting something like:
4616 __asm__(".symver original_foo,foo@@VERS_1.1");
4619 in the C source file. This renames the function @samp{original_foo} to
4620 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4621 The @samp{local:} directive can be used to prevent the symbol
4622 @samp{original_foo} from being exported. A @samp{.symver} directive
4623 takes precedence over a version script.
4625 The second GNU extension is to allow multiple versions of the same
4626 function to appear in a given shared library. In this way you can make
4627 an incompatible change to an interface without increasing the major
4628 version number of the shared library, while still allowing applications
4629 linked against the old interface to continue to function.
4631 To do this, you must use multiple @samp{.symver} directives in the
4632 source file. Here is an example:
4635 __asm__(".symver original_foo,foo@@");
4636 __asm__(".symver old_foo,foo@@VERS_1.1");
4637 __asm__(".symver old_foo1,foo@@VERS_1.2");
4638 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4641 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4642 unspecified base version of the symbol. The source file that contains this
4643 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4644 @samp{old_foo1}, and @samp{new_foo}.
4646 When you have multiple definitions of a given symbol, there needs to be
4647 some way to specify a default version to which external references to
4648 this symbol will be bound. You can do this with the
4649 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4650 declare one version of a symbol as the default in this manner; otherwise
4651 you would effectively have multiple definitions of the same symbol.
4653 If you wish to bind a reference to a specific version of the symbol
4654 within the shared library, you can use the aliases of convenience
4655 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4656 specifically bind to an external version of the function in question.
4658 You can also specify the language in the version script:
4661 VERSION extern "lang" @{ version-script-commands @}
4664 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4665 The linker will iterate over the list of symbols at the link time and
4666 demangle them according to @samp{lang} before matching them to the
4667 patterns specified in @samp{version-script-commands}.
4669 Demangled names may contains spaces and other special characters. As
4670 described above, you can use a glob pattern to match demangled names,
4671 or you can use a double-quoted string to match the string exactly. In
4672 the latter case, be aware that minor differences (such as differing
4673 whitespace) between the version script and the demangler output will
4674 cause a mismatch. As the exact string generated by the demangler
4675 might change in the future, even if the mangled name does not, you
4676 should check that all of your version directives are behaving as you
4677 expect when you upgrade.
4680 @section Expressions in Linker Scripts
4683 The syntax for expressions in the linker script language is identical to
4684 that of C expressions. All expressions are evaluated as integers. All
4685 expressions are evaluated in the same size, which is 32 bits if both the
4686 host and target are 32 bits, and is otherwise 64 bits.
4688 You can use and set symbol values in expressions.
4690 The linker defines several special purpose builtin functions for use in
4694 * Constants:: Constants
4695 * Symbols:: Symbol Names
4696 * Orphan Sections:: Orphan Sections
4697 * Location Counter:: The Location Counter
4698 * Operators:: Operators
4699 * Evaluation:: Evaluation
4700 * Expression Section:: The Section of an Expression
4701 * Builtin Functions:: Builtin Functions
4705 @subsection Constants
4706 @cindex integer notation
4707 @cindex constants in linker scripts
4708 All constants are integers.
4710 As in C, the linker considers an integer beginning with @samp{0} to be
4711 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4712 hexadecimal. The linker considers other integers to be decimal.
4714 @cindex scaled integers
4715 @cindex K and M integer suffixes
4716 @cindex M and K integer suffixes
4717 @cindex suffixes for integers
4718 @cindex integer suffixes
4719 In addition, you can use the suffixes @code{K} and @code{M} to scale a
4723 @c END TEXI2ROFF-KILL
4724 @code{1024} or @code{1024*1024}
4728 ${\rm 1024}$ or ${\rm 1024}^2$
4730 @c END TEXI2ROFF-KILL
4731 respectively. For example, the following all refer to the same quantity:
4739 @subsection Symbol Names
4740 @cindex symbol names
4742 @cindex quoted symbol names
4744 Unless quoted, symbol names start with a letter, underscore, or period
4745 and may include letters, digits, underscores, periods, and hyphens.
4746 Unquoted symbol names must not conflict with any keywords. You can
4747 specify a symbol which contains odd characters or has the same name as a
4748 keyword by surrounding the symbol name in double quotes:
4751 "with a space" = "also with a space" + 10;
4754 Since symbols can contain many non-alphabetic characters, it is safest
4755 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
4756 whereas @samp{A - B} is an expression involving subtraction.
4758 @node Orphan Sections
4759 @subsection Orphan Sections
4761 Orphan sections are sections present in the input files which
4762 are not explicitly placed into the output file by the linker
4763 script. The linker will still copy these sections into the
4764 output file, but it has to guess as to where they should be
4765 placed. The linker uses a simple heuristic to do this. It
4766 attempts to place orphan sections after non-orphan sections of the
4767 same attribute, such as code vs data, loadable vs non-loadable, etc.
4768 If there is not enough room to do this then it places
4769 at the end of the file.
4771 For ELF targets, the attribute of the section includes section type as
4772 well as section flag.
4774 If an orphaned section's name is representable as a C identifier then
4775 the linker will automatically @pxref{PROVIDE} two symbols:
4776 __start_SECNAME and __end_SECNAME, where SECNAME is the name of the
4777 section. These indicate the start address and end address of the
4778 orphaned section respectively. Note: most section names are not
4779 representable as C identifiers because they contain a @samp{.}
4782 @node Location Counter
4783 @subsection The Location Counter
4786 @cindex location counter
4787 @cindex current output location
4788 The special linker variable @dfn{dot} @samp{.} always contains the
4789 current output location counter. Since the @code{.} always refers to a
4790 location in an output section, it may only appear in an expression
4791 within a @code{SECTIONS} command. The @code{.} symbol may appear
4792 anywhere that an ordinary symbol is allowed in an expression.
4795 Assigning a value to @code{.} will cause the location counter to be
4796 moved. This may be used to create holes in the output section. The
4797 location counter may not be moved backwards inside an output section,
4798 and may not be moved backwards outside of an output section if so
4799 doing creates areas with overlapping LMAs.
4815 In the previous example, the @samp{.text} section from @file{file1} is
4816 located at the beginning of the output section @samp{output}. It is
4817 followed by a 1000 byte gap. Then the @samp{.text} section from
4818 @file{file2} appears, also with a 1000 byte gap following before the
4819 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
4820 specifies what data to write in the gaps (@pxref{Output Section Fill}).
4822 @cindex dot inside sections
4823 Note: @code{.} actually refers to the byte offset from the start of the
4824 current containing object. Normally this is the @code{SECTIONS}
4825 statement, whose start address is 0, hence @code{.} can be used as an
4826 absolute address. If @code{.} is used inside a section description
4827 however, it refers to the byte offset from the start of that section,
4828 not an absolute address. Thus in a script like this:
4846 The @samp{.text} section will be assigned a starting address of 0x100
4847 and a size of exactly 0x200 bytes, even if there is not enough data in
4848 the @samp{.text} input sections to fill this area. (If there is too
4849 much data, an error will be produced because this would be an attempt to
4850 move @code{.} backwards). The @samp{.data} section will start at 0x500
4851 and it will have an extra 0x600 bytes worth of space after the end of
4852 the values from the @samp{.data} input sections and before the end of
4853 the @samp{.data} output section itself.
4855 @cindex dot outside sections
4856 Setting symbols to the value of the location counter outside of an
4857 output section statement can result in unexpected values if the linker
4858 needs to place orphan sections. For example, given the following:
4864 .text: @{ *(.text) @}
4868 .data: @{ *(.data) @}
4873 If the linker needs to place some input section, e.g. @code{.rodata},
4874 not mentioned in the script, it might choose to place that section
4875 between @code{.text} and @code{.data}. You might think the linker
4876 should place @code{.rodata} on the blank line in the above script, but
4877 blank lines are of no particular significance to the linker. As well,
4878 the linker doesn't associate the above symbol names with their
4879 sections. Instead, it assumes that all assignments or other
4880 statements belong to the previous output section, except for the
4881 special case of an assignment to @code{.}. I.e., the linker will
4882 place the orphan @code{.rodata} section as if the script was written
4889 .text: @{ *(.text) @}
4893 .rodata: @{ *(.rodata) @}
4894 .data: @{ *(.data) @}
4899 This may or may not be the script author's intention for the value of
4900 @code{start_of_data}. One way to influence the orphan section
4901 placement is to assign the location counter to itself, as the linker
4902 assumes that an assignment to @code{.} is setting the start address of
4903 a following output section and thus should be grouped with that
4904 section. So you could write:
4910 .text: @{ *(.text) @}
4915 .data: @{ *(.data) @}
4920 Now, the orphan @code{.rodata} section will be placed between
4921 @code{end_of_text} and @code{start_of_data}.
4925 @subsection Operators
4926 @cindex operators for arithmetic
4927 @cindex arithmetic operators
4928 @cindex precedence in expressions
4929 The linker recognizes the standard C set of arithmetic operators, with
4930 the standard bindings and precedence levels:
4933 @c END TEXI2ROFF-KILL
4935 precedence associativity Operators Notes
4941 5 left == != > < <= >=
4947 11 right &= += -= *= /= (2)
4951 (1) Prefix operators
4952 (2) @xref{Assignments}.
4956 \vskip \baselineskip
4957 %"lispnarrowing" is the extra indent used generally for smallexample
4958 \hskip\lispnarrowing\vbox{\offinterlineskip
4961 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
4962 height2pt&\omit&&\omit&&\omit&\cr
4963 &Precedence&& Associativity &&{\rm Operators}&\cr
4964 height2pt&\omit&&\omit&&\omit&\cr
4966 height2pt&\omit&&\omit&&\omit&\cr
4968 % '176 is tilde, '~' in tt font
4969 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
4970 &2&&left&&* / \%&\cr
4973 &5&&left&&== != > < <= >=&\cr
4976 &8&&left&&{\&\&}&\cr
4979 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
4981 height2pt&\omit&&\omit&&\omit&\cr}
4986 @obeylines@parskip=0pt@parindent=0pt
4987 @dag@quad Prefix operators.
4988 @ddag@quad @xref{Assignments}.
4991 @c END TEXI2ROFF-KILL
4994 @subsection Evaluation
4995 @cindex lazy evaluation
4996 @cindex expression evaluation order
4997 The linker evaluates expressions lazily. It only computes the value of
4998 an expression when absolutely necessary.
5000 The linker needs some information, such as the value of the start
5001 address of the first section, and the origins and lengths of memory
5002 regions, in order to do any linking at all. These values are computed
5003 as soon as possible when the linker reads in the linker script.
5005 However, other values (such as symbol values) are not known or needed
5006 until after storage allocation. Such values are evaluated later, when
5007 other information (such as the sizes of output sections) is available
5008 for use in the symbol assignment expression.
5010 The sizes of sections cannot be known until after allocation, so
5011 assignments dependent upon these are not performed until after
5014 Some expressions, such as those depending upon the location counter
5015 @samp{.}, must be evaluated during section allocation.
5017 If the result of an expression is required, but the value is not
5018 available, then an error results. For example, a script like the
5024 .text 9+this_isnt_constant :
5030 will cause the error message @samp{non constant expression for initial
5033 @node Expression Section
5034 @subsection The Section of an Expression
5035 @cindex expression sections
5036 @cindex absolute expressions
5037 @cindex relative expressions
5038 @cindex absolute and relocatable symbols
5039 @cindex relocatable and absolute symbols
5040 @cindex symbols, relocatable and absolute
5041 When the linker evaluates an expression, the result is either absolute
5042 or relative to some section. A relative expression is expressed as a
5043 fixed offset from the base of a section.
5045 The position of the expression within the linker script determines
5046 whether it is absolute or relative. An expression which appears within
5047 an output section definition is relative to the base of the output
5048 section. An expression which appears elsewhere will be absolute.
5050 A symbol set to a relative expression will be relocatable if you request
5051 relocatable output using the @samp{-r} option. That means that a
5052 further link operation may change the value of the symbol. The symbol's
5053 section will be the section of the relative expression.
5055 A symbol set to an absolute expression will retain the same value
5056 through any further link operation. The symbol will be absolute, and
5057 will not have any particular associated section.
5059 You can use the builtin function @code{ABSOLUTE} to force an expression
5060 to be absolute when it would otherwise be relative. For example, to
5061 create an absolute symbol set to the address of the end of the output
5062 section @samp{.data}:
5066 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
5070 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
5071 @samp{.data} section.
5073 @node Builtin Functions
5074 @subsection Builtin Functions
5075 @cindex functions in expressions
5076 The linker script language includes a number of builtin functions for
5077 use in linker script expressions.
5080 @item ABSOLUTE(@var{exp})
5081 @kindex ABSOLUTE(@var{exp})
5082 @cindex expression, absolute
5083 Return the absolute (non-relocatable, as opposed to non-negative) value
5084 of the expression @var{exp}. Primarily useful to assign an absolute
5085 value to a symbol within a section definition, where symbol values are
5086 normally section relative. @xref{Expression Section}.
5088 @item ADDR(@var{section})
5089 @kindex ADDR(@var{section})
5090 @cindex section address in expression
5091 Return the absolute address (the VMA) of the named @var{section}. Your
5092 script must previously have defined the location of that section. In
5093 the following example, @code{symbol_1} and @code{symbol_2} are assigned
5100 start_of_output_1 = ABSOLUTE(.);
5105 symbol_1 = ADDR(.output1);
5106 symbol_2 = start_of_output_1;
5112 @item ALIGN(@var{align})
5113 @itemx ALIGN(@var{exp},@var{align})
5114 @kindex ALIGN(@var{align})
5115 @kindex ALIGN(@var{exp},@var{align})
5116 @cindex round up location counter
5117 @cindex align location counter
5118 @cindex round up expression
5119 @cindex align expression
5120 Return the location counter (@code{.}) or arbitrary expression aligned
5121 to the next @var{align} boundary. The single operand @code{ALIGN}
5122 doesn't change the value of the location counter---it just does
5123 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5124 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5125 equivalent to @code{ALIGN(., @var{align})}).
5127 Here is an example which aligns the output @code{.data} section to the
5128 next @code{0x2000} byte boundary after the preceding section and sets a
5129 variable within the section to the next @code{0x8000} boundary after the
5134 .data ALIGN(0x2000): @{
5136 variable = ALIGN(0x8000);
5142 The first use of @code{ALIGN} in this example specifies the location of
5143 a section because it is used as the optional @var{address} attribute of
5144 a section definition (@pxref{Output Section Address}). The second use
5145 of @code{ALIGN} is used to defines the value of a symbol.
5147 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5149 @item ALIGNOF(@var{section})
5150 @kindex ALIGNOF(@var{section})
5151 @cindex section alignment
5152 Return the alignment in bytes of the named @var{section}, if that section has
5153 been allocated. If the section has not been allocated when this is
5154 evaluated, the linker will report an error. In the following example,
5155 the alignment of the @code{.output} section is stored as the first
5156 value in that section.
5161 LONG (ALIGNOF (.output))
5168 @item BLOCK(@var{exp})
5169 @kindex BLOCK(@var{exp})
5170 This is a synonym for @code{ALIGN}, for compatibility with older linker
5171 scripts. It is most often seen when setting the address of an output
5174 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5175 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5176 This is equivalent to either
5178 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5182 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5185 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5186 for the data segment (area between the result of this expression and
5187 @code{DATA_SEGMENT_END}) than the former or not.
5188 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5189 memory will be saved at the expense of up to @var{commonpagesize} wasted
5190 bytes in the on-disk file.
5192 This expression can only be used directly in @code{SECTIONS} commands, not in
5193 any output section descriptions and only once in the linker script.
5194 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5195 be the system page size the object wants to be optimized for (while still
5196 working on system page sizes up to @var{maxpagesize}).
5201 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5204 @item DATA_SEGMENT_END(@var{exp})
5205 @kindex DATA_SEGMENT_END(@var{exp})
5206 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5207 evaluation purposes.
5210 . = DATA_SEGMENT_END(.);
5213 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5214 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5215 This defines the end of the @code{PT_GNU_RELRO} segment when
5216 @samp{-z relro} option is used. Second argument is returned.
5217 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5218 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5219 @var{exp} + @var{offset} is aligned to the most commonly used page
5220 boundary for particular target. If present in the linker script,
5221 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5222 @code{DATA_SEGMENT_END}.
5225 . = DATA_SEGMENT_RELRO_END(24, .);
5228 @item DEFINED(@var{symbol})
5229 @kindex DEFINED(@var{symbol})
5230 @cindex symbol defaults
5231 Return 1 if @var{symbol} is in the linker global symbol table and is
5232 defined before the statement using DEFINED in the script, otherwise
5233 return 0. You can use this function to provide
5234 default values for symbols. For example, the following script fragment
5235 shows how to set a global symbol @samp{begin} to the first location in
5236 the @samp{.text} section---but if a symbol called @samp{begin} already
5237 existed, its value is preserved:
5243 begin = DEFINED(begin) ? begin : . ;
5251 @item LENGTH(@var{memory})
5252 @kindex LENGTH(@var{memory})
5253 Return the length of the memory region named @var{memory}.
5255 @item LOADADDR(@var{section})
5256 @kindex LOADADDR(@var{section})
5257 @cindex section load address in expression
5258 Return the absolute LMA of the named @var{section}. This is normally
5259 the same as @code{ADDR}, but it may be different if the @code{AT}
5260 attribute is used in the output section definition (@pxref{Output
5264 @item MAX(@var{exp1}, @var{exp2})
5265 Returns the maximum of @var{exp1} and @var{exp2}.
5268 @item MIN(@var{exp1}, @var{exp2})
5269 Returns the minimum of @var{exp1} and @var{exp2}.
5271 @item NEXT(@var{exp})
5272 @kindex NEXT(@var{exp})
5273 @cindex unallocated address, next
5274 Return the next unallocated address that is a multiple of @var{exp}.
5275 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5276 use the @code{MEMORY} command to define discontinuous memory for the
5277 output file, the two functions are equivalent.
5279 @item ORIGIN(@var{memory})
5280 @kindex ORIGIN(@var{memory})
5281 Return the origin of the memory region named @var{memory}.
5283 @item SEGMENT_START(@var{segment}, @var{default})
5284 @kindex SEGMENT_START(@var{segment}, @var{default})
5285 Return the base address of the named @var{segment}. If an explicit
5286 value has been given for this segment (with a command-line @samp{-T}
5287 option) that value will be returned; otherwise the value will be
5288 @var{default}. At present, the @samp{-T} command-line option can only
5289 be used to set the base address for the ``text'', ``data'', and
5290 ``bss'' sections, but you use @code{SEGMENT_START} with any segment
5293 @item SIZEOF(@var{section})
5294 @kindex SIZEOF(@var{section})
5295 @cindex section size
5296 Return the size in bytes of the named @var{section}, if that section has
5297 been allocated. If the section has not been allocated when this is
5298 evaluated, the linker will report an error. In the following example,
5299 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5308 symbol_1 = .end - .start ;
5309 symbol_2 = SIZEOF(.output);
5314 @item SIZEOF_HEADERS
5315 @itemx sizeof_headers
5316 @kindex SIZEOF_HEADERS
5318 Return the size in bytes of the output file's headers. This is
5319 information which appears at the start of the output file. You can use
5320 this number when setting the start address of the first section, if you
5321 choose, to facilitate paging.
5323 @cindex not enough room for program headers
5324 @cindex program headers, not enough room
5325 When producing an ELF output file, if the linker script uses the
5326 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5327 number of program headers before it has determined all the section
5328 addresses and sizes. If the linker later discovers that it needs
5329 additional program headers, it will report an error @samp{not enough
5330 room for program headers}. To avoid this error, you must avoid using
5331 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5332 script to avoid forcing the linker to use additional program headers, or
5333 you must define the program headers yourself using the @code{PHDRS}
5334 command (@pxref{PHDRS}).
5337 @node Implicit Linker Scripts
5338 @section Implicit Linker Scripts
5339 @cindex implicit linker scripts
5340 If you specify a linker input file which the linker can not recognize as
5341 an object file or an archive file, it will try to read the file as a
5342 linker script. If the file can not be parsed as a linker script, the
5343 linker will report an error.
5345 An implicit linker script will not replace the default linker script.
5347 Typically an implicit linker script would contain only symbol
5348 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5351 Any input files read because of an implicit linker script will be read
5352 at the position in the command line where the implicit linker script was
5353 read. This can affect archive searching.
5356 @node Machine Dependent
5357 @chapter Machine Dependent Features
5359 @cindex machine dependencies
5360 @command{ld} has additional features on some platforms; the following
5361 sections describe them. Machines where @command{ld} has no additional
5362 functionality are not listed.
5366 * H8/300:: @command{ld} and the H8/300
5369 * i960:: @command{ld} and the Intel 960 family
5372 * ARM:: @command{ld} and the ARM family
5375 * Blackfin:: @command{ld} and the Blackfin family
5378 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
5381 * M68K:: @command{ld} and the Motorola 68K family
5384 * MMIX:: @command{ld} and MMIX
5387 * MSP430:: @command{ld} and MSP430
5390 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
5393 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
5396 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
5399 * SPU ELF:: @command{ld} and SPU ELF Support
5402 * TI COFF:: @command{ld} and TI COFF
5405 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
5408 * Xtensa:: @command{ld} and Xtensa Processors
5419 @section @command{ld} and the H8/300
5421 @cindex H8/300 support
5422 For the H8/300, @command{ld} can perform these global optimizations when
5423 you specify the @samp{--relax} command-line option.
5426 @cindex relaxing on H8/300
5427 @item relaxing address modes
5428 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5429 targets are within eight bits, and turns them into eight-bit
5430 program-counter relative @code{bsr} and @code{bra} instructions,
5433 @cindex synthesizing on H8/300
5434 @item synthesizing instructions
5435 @c FIXME: specifically mov.b, or any mov instructions really?
5436 @command{ld} finds all @code{mov.b} instructions which use the
5437 sixteen-bit absolute address form, but refer to the top
5438 page of memory, and changes them to use the eight-bit address form.
5439 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5440 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5441 top page of memory).
5443 @item bit manipulation instructions
5444 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
5445 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5446 which use 32 bit and 16 bit absolute address form, but refer to the top
5447 page of memory, and changes them to use the 8 bit address form.
5448 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5449 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5450 the top page of memory).
5452 @item system control instructions
5453 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
5454 32 bit absolute address form, but refer to the top page of memory, and
5455 changes them to use 16 bit address form.
5456 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5457 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5458 the top page of memory).
5468 @c This stuff is pointless to say unless you're especially concerned
5469 @c with Renesas chips; don't enable it for generic case, please.
5471 @chapter @command{ld} and Other Renesas Chips
5473 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5474 H8/500, and SH chips. No special features, commands, or command-line
5475 options are required for these chips.
5485 @section @command{ld} and the Intel 960 Family
5487 @cindex i960 support
5489 You can use the @samp{-A@var{architecture}} command line option to
5490 specify one of the two-letter names identifying members of the 960
5491 family; the option specifies the desired output target, and warns of any
5492 incompatible instructions in the input files. It also modifies the
5493 linker's search strategy for archive libraries, to support the use of
5494 libraries specific to each particular architecture, by including in the
5495 search loop names suffixed with the string identifying the architecture.
5497 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5498 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5499 paths, and in any paths you specify with @samp{-L}) for a library with
5512 The first two possibilities would be considered in any event; the last
5513 two are due to the use of @w{@samp{-ACA}}.
5515 You can meaningfully use @samp{-A} more than once on a command line, since
5516 the 960 architecture family allows combination of target architectures; each
5517 use will add another pair of name variants to search for when @w{@samp{-l}}
5518 specifies a library.
5520 @cindex @option{--relax} on i960
5521 @cindex relaxing on i960
5522 @command{ld} supports the @samp{--relax} option for the i960 family. If
5523 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5524 @code{calx} instructions whose targets are within 24 bits, and turns
5525 them into 24-bit program-counter relative @code{bal} and @code{cal}
5526 instructions, respectively. @command{ld} also turns @code{cal}
5527 instructions into @code{bal} instructions when it determines that the
5528 target subroutine is a leaf routine (that is, the target subroutine does
5529 not itself call any subroutines).
5546 @node M68HC11/68HC12
5547 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
5549 @cindex M68HC11 and 68HC12 support
5551 @subsection Linker Relaxation
5553 For the Motorola 68HC11, @command{ld} can perform these global
5554 optimizations when you specify the @samp{--relax} command-line option.
5557 @cindex relaxing on M68HC11
5558 @item relaxing address modes
5559 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5560 targets are within eight bits, and turns them into eight-bit
5561 program-counter relative @code{bsr} and @code{bra} instructions,
5564 @command{ld} also looks at all 16-bit extended addressing modes and
5565 transforms them in a direct addressing mode when the address is in
5566 page 0 (between 0 and 0x0ff).
5568 @item relaxing gcc instruction group
5569 When @command{gcc} is called with @option{-mrelax}, it can emit group
5570 of instructions that the linker can optimize to use a 68HC11 direct
5571 addressing mode. These instructions consists of @code{bclr} or
5572 @code{bset} instructions.
5576 @subsection Trampoline Generation
5578 @cindex trampoline generation on M68HC11
5579 @cindex trampoline generation on M68HC12
5580 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5581 call a far function using a normal @code{jsr} instruction. The linker
5582 will also change the relocation to some far function to use the
5583 trampoline address instead of the function address. This is typically the
5584 case when a pointer to a function is taken. The pointer will in fact
5585 point to the function trampoline.
5593 @section @command{ld} and the ARM family
5595 @cindex ARM interworking support
5596 @kindex --support-old-code
5597 For the ARM, @command{ld} will generate code stubs to allow functions calls
5598 between ARM and Thumb code. These stubs only work with code that has
5599 been compiled and assembled with the @samp{-mthumb-interwork} command
5600 line option. If it is necessary to link with old ARM object files or
5601 libraries, which have not been compiled with the -mthumb-interwork
5602 option then the @samp{--support-old-code} command line switch should be
5603 given to the linker. This will make it generate larger stub functions
5604 which will work with non-interworking aware ARM code. Note, however,
5605 the linker does not support generating stubs for function calls to
5606 non-interworking aware Thumb code.
5608 @cindex thumb entry point
5609 @cindex entry point, thumb
5610 @kindex --thumb-entry=@var{entry}
5611 The @samp{--thumb-entry} switch is a duplicate of the generic
5612 @samp{--entry} switch, in that it sets the program's starting address.
5613 But it also sets the bottom bit of the address, so that it can be
5614 branched to using a BX instruction, and the program will start
5615 executing in Thumb mode straight away.
5619 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5620 executables. This option is only valid when linking big-endian objects.
5621 The resulting image will contain big-endian data and little-endian code.
5624 @kindex --target1-rel
5625 @kindex --target1-abs
5626 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5627 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
5628 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
5629 and @samp{--target1-abs} switches override the default.
5632 @kindex --target2=@var{type}
5633 The @samp{--target2=type} switch overrides the default definition of the
5634 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
5635 meanings, and target defaults are as follows:
5638 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5640 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
5642 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5647 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5648 specification) enables objects compiled for the ARMv4 architecture to be
5649 interworking-safe when linked with other objects compiled for ARMv4t, but
5650 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
5652 In the latter case, the switch @option{--fix-v4bx} must be passed to the
5653 linker, which causes v4t @code{BX rM} instructions to be rewritten as
5654 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
5656 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
5657 relocations are ignored.
5659 @cindex FIX_V4BX_INTERWORKING
5660 @kindex --fix-v4bx-interworking
5661 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
5662 relocations with a branch to the following veneer:
5670 This allows generation of libraries/applications that work on ARMv4 cores
5671 and are still interworking safe. Note that the above veneer clobbers the
5672 condition flags, so may cause incorrect progrm behavior in rare cases.
5676 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
5677 BLX instructions (available on ARMv5t and above) in various
5678 situations. Currently it is used to perform calls via the PLT from Thumb
5679 code using BLX rather than using BX and a mode-switching stub before
5680 each PLT entry. This should lead to such calls executing slightly faster.
5682 This option is enabled implicitly for SymbianOS, so there is no need to
5683 specify it if you are using that target.
5685 @cindex VFP11_DENORM_FIX
5686 @kindex --vfp11-denorm-fix
5687 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
5688 bug in certain VFP11 coprocessor hardware, which sometimes allows
5689 instructions with denorm operands (which must be handled by support code)
5690 to have those operands overwritten by subsequent instructions before
5691 the support code can read the intended values.
5693 The bug may be avoided in scalar mode if you allow at least one
5694 intervening instruction between a VFP11 instruction which uses a register
5695 and another instruction which writes to the same register, or at least two
5696 intervening instructions if vector mode is in use. The bug only affects
5697 full-compliance floating-point mode: you do not need this workaround if
5698 you are using "runfast" mode. Please contact ARM for further details.
5700 If you know you are using buggy VFP11 hardware, you can
5701 enable this workaround by specifying the linker option
5702 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
5703 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
5704 vector mode (the latter also works for scalar code). The default is
5705 @samp{--vfp-denorm-fix=none}.
5707 If the workaround is enabled, instructions are scanned for
5708 potentially-troublesome sequences, and a veneer is created for each
5709 such sequence which may trigger the erratum. The veneer consists of the
5710 first instruction of the sequence and a branch back to the subsequent
5711 instruction. The original instruction is then replaced with a branch to
5712 the veneer. The extra cycles required to call and return from the veneer
5713 are sufficient to avoid the erratum in both the scalar and vector cases.
5715 @cindex NO_ENUM_SIZE_WARNING
5716 @kindex --no-enum-size-warning
5717 The @option{--no-enum-size-warning} switch prevents the linker from
5718 warning when linking object files that specify incompatible EABI
5719 enumeration size attributes. For example, with this switch enabled,
5720 linking of an object file using 32-bit enumeration values with another
5721 using enumeration values fitted into the smallest possible space will
5725 @kindex --pic-veneer
5726 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
5727 ARM/Thumb interworking veneers, even if the rest of the binary
5728 is not PIC. This avoids problems on uClinux targets where
5729 @samp{--emit-relocs} is used to generate relocatable binaries.
5731 @cindex STUB_GROUP_SIZE
5732 @kindex --stub-group-size=@var{N}
5733 The linker will automatically generate and insert small sequences of
5734 code into a linked ARM ELF executable whenever an attempt is made to
5735 perform a function call to a symbol that is too far away. The
5736 placement of these sequences of instructions - called stubs - is
5737 controlled by the command line option @option{--stub-group-size=N}.
5738 The placement is important because a poor choice can create a need for
5739 duplicate stubs, increasing the code sizw. The linker will try to
5740 group stubs together in order to reduce interruptions to the flow of
5741 code, but it needs guidance as to how big these groups should be and
5742 where they should be placed.
5744 The value of @samp{N}, the parameter to the
5745 @option{--stub-group-size=} option controls where the stub groups are
5746 placed. If it is negative then all stubs are placed before the first
5747 branch that needs them. If it is positive then the stubs can be
5748 placed either before or after the branches that need them. If the
5749 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
5750 exactly where to place groups of stubs, using its built in heuristics.
5751 A value of @samp{N} greater than 1 (or smaller than -1) tells the
5752 linker that a single group of stubs can service at most @samp{N} bytes
5753 from the input sections.
5755 The default, if @option{--stub-group-size=} is not specified, is
5758 Farcalls stubs insertion is fully supported for the ARM-EABI target
5759 only, because it relies on object files properties not present
5773 @section @command{ld} and the Blackfin family
5775 @cindex Put code in L1 instruction SRAM
5776 @kindex --code-in-l1
5777 The @samp{--code-in-l1} option adds a specific flag in the ELF header.
5778 This flag tells loader to put the code segments into Blackfin
5779 L1 instruction SRAM.
5781 @cindex Put code in L1 data SRAM
5782 @kindex --data-in-l1
5783 The @samp{--data-in-l1} option adds a specific flag in the ELF header.
5784 This flag tells loader to put the data segments into Blackfin L1 data SRAM.
5797 @section @command{ld} and HPPA 32-bit ELF Support
5798 @cindex HPPA multiple sub-space stubs
5799 @kindex --multi-subspace
5800 When generating a shared library, @command{ld} will by default generate
5801 import stubs suitable for use with a single sub-space application.
5802 The @samp{--multi-subspace} switch causes @command{ld} to generate export
5803 stubs, and different (larger) import stubs suitable for use with
5804 multiple sub-spaces.
5806 @cindex HPPA stub grouping
5807 @kindex --stub-group-size=@var{N}
5808 Long branch stubs and import/export stubs are placed by @command{ld} in
5809 stub sections located between groups of input sections.
5810 @samp{--stub-group-size} specifies the maximum size of a group of input
5811 sections handled by one stub section. Since branch offsets are signed,
5812 a stub section may serve two groups of input sections, one group before
5813 the stub section, and one group after it. However, when using
5814 conditional branches that require stubs, it may be better (for branch
5815 prediction) that stub sections only serve one group of input sections.
5816 A negative value for @samp{N} chooses this scheme, ensuring that
5817 branches to stubs always use a negative offset. Two special values of
5818 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5819 @command{ld} to automatically size input section groups for the branch types
5820 detected, with the same behaviour regarding stub placement as other
5821 positive or negative values of @samp{N} respectively.
5823 Note that @samp{--stub-group-size} does not split input sections. A
5824 single input section larger than the group size specified will of course
5825 create a larger group (of one section). If input sections are too
5826 large, it may not be possible for a branch to reach its stub.
5839 @section @command{ld} and the Motorola 68K family
5841 @cindex Motorola 68K GOT generation
5842 @kindex --got=@var{type}
5843 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
5844 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
5845 @samp{target}. When @samp{target} is selected the linker chooses
5846 the default GOT generation scheme for the current target.
5847 @samp{single} tells the linker to generate a single GOT with
5848 entries only at non-negative offsets.
5849 @samp{negative} instructs the linker to generate a single GOT with
5850 entries at both negative and positive offsets. Not all environments
5852 @samp{multigot} allows the linker to generate several GOTs in the
5853 output file. All GOT references from a single input object
5854 file access the same GOT, but references from different input object
5855 files might access different GOTs. Not all environments support such GOTs.
5868 @section @code{ld} and MMIX
5869 For MMIX, there is a choice of generating @code{ELF} object files or
5870 @code{mmo} object files when linking. The simulator @code{mmix}
5871 understands the @code{mmo} format. The binutils @code{objcopy} utility
5872 can translate between the two formats.
5874 There is one special section, the @samp{.MMIX.reg_contents} section.
5875 Contents in this section is assumed to correspond to that of global
5876 registers, and symbols referring to it are translated to special symbols,
5877 equal to registers. In a final link, the start address of the
5878 @samp{.MMIX.reg_contents} section corresponds to the first allocated
5879 global register multiplied by 8. Register @code{$255} is not included in
5880 this section; it is always set to the program entry, which is at the
5881 symbol @code{Main} for @code{mmo} files.
5883 Global symbols with the prefix @code{__.MMIX.start.}, for example
5884 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
5885 The default linker script uses these to set the default start address
5888 Initial and trailing multiples of zero-valued 32-bit words in a section,
5889 are left out from an mmo file.
5902 @section @code{ld} and MSP430
5903 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
5904 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
5905 just pass @samp{-m help} option to the linker).
5907 @cindex MSP430 extra sections
5908 The linker will recognize some extra sections which are MSP430 specific:
5911 @item @samp{.vectors}
5912 Defines a portion of ROM where interrupt vectors located.
5914 @item @samp{.bootloader}
5915 Defines the bootloader portion of the ROM (if applicable). Any code
5916 in this section will be uploaded to the MPU.
5918 @item @samp{.infomem}
5919 Defines an information memory section (if applicable). Any code in
5920 this section will be uploaded to the MPU.
5922 @item @samp{.infomemnobits}
5923 This is the same as the @samp{.infomem} section except that any code
5924 in this section will not be uploaded to the MPU.
5926 @item @samp{.noinit}
5927 Denotes a portion of RAM located above @samp{.bss} section.
5929 The last two sections are used by gcc.
5943 @section @command{ld} and PowerPC 32-bit ELF Support
5944 @cindex PowerPC long branches
5945 @kindex --relax on PowerPC
5946 Branches on PowerPC processors are limited to a signed 26-bit
5947 displacement, which may result in @command{ld} giving
5948 @samp{relocation truncated to fit} errors with very large programs.
5949 @samp{--relax} enables the generation of trampolines that can access
5950 the entire 32-bit address space. These trampolines are inserted at
5951 section boundaries, so may not themselves be reachable if an input
5952 section exceeds 33M in size.
5954 @cindex PowerPC ELF32 options
5959 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
5960 generates code capable of using a newer PLT and GOT layout that has
5961 the security advantage of no executable section ever needing to be
5962 writable and no writable section ever being executable. PowerPC
5963 @command{ld} will generate this layout, including stubs to access the
5964 PLT, if all input files (including startup and static libraries) were
5965 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
5966 BSS PLT (and GOT layout) which can give slightly better performance.
5968 @kindex --secure-plt
5970 @command{ld} will use the new PLT and GOT layout if it is linking new
5971 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
5972 when linking non-PIC code. This option requests the new PLT and GOT
5973 layout. A warning will be given if some object file requires the old
5979 The new secure PLT and GOT are placed differently relative to other
5980 sections compared to older BSS PLT and GOT placement. The location of
5981 @code{.plt} must change because the new secure PLT is an initialized
5982 section while the old PLT is uninitialized. The reason for the
5983 @code{.got} change is more subtle: The new placement allows
5984 @code{.got} to be read-only in applications linked with
5985 @samp{-z relro -z now}. However, this placement means that
5986 @code{.sdata} cannot always be used in shared libraries, because the
5987 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
5988 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
5989 GCC doesn't use @code{.sdata} in shared libraries, so this option is
5990 really only useful for other compilers that may do so.
5992 @cindex PowerPC stub symbols
5993 @kindex --emit-stub-syms
5994 @item --emit-stub-syms
5995 This option causes @command{ld} to label linker stubs with a local
5996 symbol that encodes the stub type and destination.
5998 @cindex PowerPC TLS optimization
5999 @kindex --no-tls-optimize
6000 @item --no-tls-optimize
6001 PowerPC @command{ld} normally performs some optimization of code
6002 sequences used to access Thread-Local Storage. Use this option to
6003 disable the optimization.
6016 @node PowerPC64 ELF64
6017 @section @command{ld} and PowerPC64 64-bit ELF Support
6019 @cindex PowerPC64 ELF64 options
6021 @cindex PowerPC64 stub grouping
6022 @kindex --stub-group-size
6023 @item --stub-group-size
6024 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
6025 by @command{ld} in stub sections located between groups of input sections.
6026 @samp{--stub-group-size} specifies the maximum size of a group of input
6027 sections handled by one stub section. Since branch offsets are signed,
6028 a stub section may serve two groups of input sections, one group before
6029 the stub section, and one group after it. However, when using
6030 conditional branches that require stubs, it may be better (for branch
6031 prediction) that stub sections only serve one group of input sections.
6032 A negative value for @samp{N} chooses this scheme, ensuring that
6033 branches to stubs always use a negative offset. Two special values of
6034 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6035 @command{ld} to automatically size input section groups for the branch types
6036 detected, with the same behaviour regarding stub placement as other
6037 positive or negative values of @samp{N} respectively.
6039 Note that @samp{--stub-group-size} does not split input sections. A
6040 single input section larger than the group size specified will of course
6041 create a larger group (of one section). If input sections are too
6042 large, it may not be possible for a branch to reach its stub.
6044 @cindex PowerPC64 stub symbols
6045 @kindex --emit-stub-syms
6046 @item --emit-stub-syms
6047 This option causes @command{ld} to label linker stubs with a local
6048 symbol that encodes the stub type and destination.
6050 @cindex PowerPC64 dot symbols
6052 @kindex --no-dotsyms
6053 @item --dotsyms, --no-dotsyms
6054 These two options control how @command{ld} interprets version patterns
6055 in a version script. Older PowerPC64 compilers emitted both a
6056 function descriptor symbol with the same name as the function, and a
6057 code entry symbol with the name prefixed by a dot (@samp{.}). To
6058 properly version a function @samp{foo}, the version script thus needs
6059 to control both @samp{foo} and @samp{.foo}. The option
6060 @samp{--dotsyms}, on by default, automatically adds the required
6061 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
6064 @cindex PowerPC64 TLS optimization
6065 @kindex --no-tls-optimize
6066 @item --no-tls-optimize
6067 PowerPC64 @command{ld} normally performs some optimization of code
6068 sequences used to access Thread-Local Storage. Use this option to
6069 disable the optimization.
6071 @cindex PowerPC64 OPD optimization
6072 @kindex --no-opd-optimize
6073 @item --no-opd-optimize
6074 PowerPC64 @command{ld} normally removes @code{.opd} section entries
6075 corresponding to deleted link-once functions, or functions removed by
6076 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
6077 Use this option to disable @code{.opd} optimization.
6079 @cindex PowerPC64 OPD spacing
6080 @kindex --non-overlapping-opd
6081 @item --non-overlapping-opd
6082 Some PowerPC64 compilers have an option to generate compressed
6083 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
6084 the static chain pointer (unused in C) with the first word of the next
6085 entry. This option expands such entries to the full 24 bytes.
6087 @cindex PowerPC64 TOC optimization
6088 @kindex --no-toc-optimize
6089 @item --no-toc-optimize
6090 PowerPC64 @command{ld} normally removes unused @code{.toc} section
6091 entries. Such entries are detected by examining relocations that
6092 reference the TOC in code sections. A reloc in a deleted code section
6093 marks a TOC word as unneeded, while a reloc in a kept code section
6094 marks a TOC word as needed. Since the TOC may reference itself, TOC
6095 relocs are also examined. TOC words marked as both needed and
6096 unneeded will of course be kept. TOC words without any referencing
6097 reloc are assumed to be part of a multi-word entry, and are kept or
6098 discarded as per the nearest marked preceding word. This works
6099 reliably for compiler generated code, but may be incorrect if assembly
6100 code is used to insert TOC entries. Use this option to disable the
6103 @cindex PowerPC64 multi-TOC
6104 @kindex --no-multi-toc
6105 @item --no-multi-toc
6106 By default, PowerPC64 GCC generates code for a TOC model where TOC
6107 entries are accessed with a 16-bit offset from r2. This limits the
6108 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
6109 grouping code sections such that each group uses less than 64K for its
6110 TOC entries, then inserts r2 adjusting stubs between inter-group
6111 calls. @command{ld} does not split apart input sections, so cannot
6112 help if a single input file has a @code{.toc} section that exceeds
6113 64K, most likely from linking multiple files with @command{ld -r}.
6114 Use this option to turn off this feature.
6128 @section @command{ld} and SPU ELF Support
6130 @cindex SPU ELF options
6136 This option marks an executable as a PIC plugin module.
6138 @cindex SPU overlays
6139 @kindex --no-overlays
6141 Normally, @command{ld} recognizes calls to functions within overlay
6142 regions, and redirects such calls to an overlay manager via a stub.
6143 @command{ld} also provides a built-in overlay manager. This option
6144 turns off all this special overlay handling.
6146 @cindex SPU overlay stub symbols
6147 @kindex --emit-stub-syms
6148 @item --emit-stub-syms
6149 This option causes @command{ld} to label overlay stubs with a local
6150 symbol that encodes the stub type and destination.
6152 @cindex SPU extra overlay stubs
6153 @kindex --extra-overlay-stubs
6154 @item --extra-overlay-stubs
6155 This option causes @command{ld} to add overlay call stubs on all
6156 function calls out of overlay regions. Normally stubs are not added
6157 on calls to non-overlay regions.
6159 @cindex SPU local store size
6160 @kindex --local-store=lo:hi
6161 @item --local-store=lo:hi
6162 @command{ld} usually checks that a final executable for SPU fits in
6163 the address range 0 to 256k. This option may be used to change the
6164 range. Disable the check entirely with @option{--local-store=0:0}.
6167 @kindex --stack-analysis
6168 @item --stack-analysis
6169 SPU local store space is limited. Over-allocation of stack space
6170 unnecessarily limits space available for code and data, while
6171 under-allocation results in runtime failures. If given this option,
6172 @command{ld} will provide an estimate of maximum stack usage.
6173 @command{ld} does this by examining symbols in code sections to
6174 determine the extents of functions, and looking at function prologues
6175 for stack adjusting instructions. A call-graph is created by looking
6176 for relocations on branch instructions. The graph is then searched
6177 for the maximum stack usage path. Note that this analysis does not
6178 find calls made via function pointers, and does not handle recursion
6179 and other cycles in the call graph. Stack usage may be
6180 under-estimated if your code makes such calls. Also, stack usage for
6181 dynamic allocation, e.g. alloca, will not be detected. If a link map
6182 is requested, detailed information about each function's stack usage
6183 and calls will be given.
6186 @kindex --emit-stack-syms
6187 @item --emit-stack-syms
6188 This option, if given along with @option{--stack-analysis} will result
6189 in @command{ld} emitting stack sizing symbols for each function.
6190 These take the form @code{__stack_<function_name>} for global
6191 functions, and @code{__stack_<number>_<function_name>} for static
6192 functions. @code{<number>} is the section id in hex. The value of
6193 such symbols is the stack requirement for the corresponding function.
6194 The symbol size will be zero, type @code{STT_NOTYPE}, binding
6195 @code{STB_LOCAL}, and section @code{SHN_ABS}.
6209 @section @command{ld}'s Support for Various TI COFF Versions
6210 @cindex TI COFF versions
6211 @kindex --format=@var{version}
6212 The @samp{--format} switch allows selection of one of the various
6213 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
6214 also supported. The TI COFF versions also vary in header byte-order
6215 format; @command{ld} will read any version or byte order, but the output
6216 header format depends on the default specified by the specific target.
6229 @section @command{ld} and WIN32 (cygwin/mingw)
6231 This section describes some of the win32 specific @command{ld} issues.
6232 See @ref{Options,,Command Line Options} for detailed description of the
6233 command line options mentioned here.
6236 @cindex import libraries
6237 @item import libraries
6238 The standard Windows linker creates and uses so-called import
6239 libraries, which contains information for linking to dll's. They are
6240 regular static archives and are handled as any other static
6241 archive. The cygwin and mingw ports of @command{ld} have specific
6242 support for creating such libraries provided with the
6243 @samp{--out-implib} command line option.
6245 @item exporting DLL symbols
6246 @cindex exporting DLL symbols
6247 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6250 @item using auto-export functionality
6251 @cindex using auto-export functionality
6252 By default @command{ld} exports symbols with the auto-export functionality,
6253 which is controlled by the following command line options:
6256 @item --export-all-symbols [This is the default]
6257 @item --exclude-symbols
6258 @item --exclude-libs
6261 If, however, @samp{--export-all-symbols} is not given explicitly on the
6262 command line, then the default auto-export behavior will be @emph{disabled}
6263 if either of the following are true:
6266 @item A DEF file is used.
6267 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
6270 @item using a DEF file
6271 @cindex using a DEF file
6272 Another way of exporting symbols is using a DEF file. A DEF file is
6273 an ASCII file containing definitions of symbols which should be
6274 exported when a dll is created. Usually it is named @samp{<dll
6275 name>.def} and is added as any other object file to the linker's
6276 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
6279 gcc -o <output> <objectfiles> <dll name>.def
6282 Using a DEF file turns off the normal auto-export behavior, unless the
6283 @samp{--export-all-symbols} option is also used.
6285 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
6288 LIBRARY "xyz.dll" BASE=0x20000000
6294 another_foo = abc.dll.afoo
6298 This example defines a DLL with a non-default base address and five
6299 symbols in the export table. The third exported symbol @code{_bar} is an
6300 alias for the second. The fourth symbol, @code{another_foo} is resolved
6301 by "forwarding" to another module and treating it as an alias for
6302 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
6303 @code{var1} is declared to be a data object.
6305 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
6306 name of the output DLL. If @samp{<name>} does not include a suffix,
6307 the default library suffix, @samp{.DLL} is appended.
6309 When the .DEF file is used to build an application, rather than a
6310 library, the @code{NAME <name>} command should be used instead of
6311 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
6312 executable suffix, @samp{.EXE} is appended.
6314 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
6315 specification @code{BASE = <number>} may be used to specify a
6316 non-default base address for the image.
6318 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
6319 or they specify an empty string, the internal name is the same as the
6320 filename specified on the command line.
6322 The complete specification of an export symbol is:
6326 ( ( ( <name1> [ = <name2> ] )
6327 | ( <name1> = <module-name> . <external-name>))
6328 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
6331 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
6332 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
6333 @samp{<name1>} as a "forward" alias for the symbol
6334 @samp{<external-name>} in the DLL @samp{<module-name>}.
6335 Optionally, the symbol may be exported by the specified ordinal
6336 @samp{<integer>} alias.
6338 The optional keywords that follow the declaration indicate:
6340 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
6341 will still be exported by its ordinal alias (either the value specified
6342 by the .def specification or, otherwise, the value assigned by the
6343 linker). The symbol name, however, does remain visible in the import
6344 library (if any), unless @code{PRIVATE} is also specified.
6346 @code{DATA}: The symbol is a variable or object, rather than a function.
6347 The import lib will export only an indirect reference to @code{foo} as
6348 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
6351 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
6352 well as @code{_imp__foo} into the import library. Both refer to the
6353 read-only import address table's pointer to the variable, not to the
6354 variable itself. This can be dangerous. If the user code fails to add
6355 the @code{dllimport} attribute and also fails to explicitly add the
6356 extra indirection that the use of the attribute enforces, the
6357 application will behave unexpectedly.
6359 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
6360 it into the static import library used to resolve imports at link time. The
6361 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
6362 API at runtime or by by using the GNU ld extension of linking directly to
6363 the DLL without an import library.
6365 See ld/deffilep.y in the binutils sources for the full specification of
6366 other DEF file statements
6368 @cindex creating a DEF file
6369 While linking a shared dll, @command{ld} is able to create a DEF file
6370 with the @samp{--output-def <file>} command line option.
6372 @item Using decorations
6373 @cindex Using decorations
6374 Another way of marking symbols for export is to modify the source code
6375 itself, so that when building the DLL each symbol to be exported is
6379 __declspec(dllexport) int a_variable
6380 __declspec(dllexport) void a_function(int with_args)
6383 All such symbols will be exported from the DLL. If, however,
6384 any of the object files in the DLL contain symbols decorated in
6385 this way, then the normal auto-export behavior is disabled, unless
6386 the @samp{--export-all-symbols} option is also used.
6388 Note that object files that wish to access these symbols must @emph{not}
6389 decorate them with dllexport. Instead, they should use dllimport,
6393 __declspec(dllimport) int a_variable
6394 __declspec(dllimport) void a_function(int with_args)
6397 This complicates the structure of library header files, because
6398 when included by the library itself the header must declare the
6399 variables and functions as dllexport, but when included by client
6400 code the header must declare them as dllimport. There are a number
6401 of idioms that are typically used to do this; often client code can
6402 omit the __declspec() declaration completely. See
6403 @samp{--enable-auto-import} and @samp{automatic data imports} for more
6407 @cindex automatic data imports
6408 @item automatic data imports
6409 The standard Windows dll format supports data imports from dlls only
6410 by adding special decorations (dllimport/dllexport), which let the
6411 compiler produce specific assembler instructions to deal with this
6412 issue. This increases the effort necessary to port existing Un*x
6413 code to these platforms, especially for large
6414 c++ libraries and applications. The auto-import feature, which was
6415 initially provided by Paul Sokolovsky, allows one to omit the
6416 decorations to achieve a behavior that conforms to that on POSIX/Un*x
6417 platforms. This feature is enabled with the @samp{--enable-auto-import}
6418 command-line option, although it is enabled by default on cygwin/mingw.
6419 The @samp{--enable-auto-import} option itself now serves mainly to
6420 suppress any warnings that are ordinarily emitted when linked objects
6421 trigger the feature's use.
6423 auto-import of variables does not always work flawlessly without
6424 additional assistance. Sometimes, you will see this message
6426 "variable '<var>' can't be auto-imported. Please read the
6427 documentation for ld's @code{--enable-auto-import} for details."
6429 The @samp{--enable-auto-import} documentation explains why this error
6430 occurs, and several methods that can be used to overcome this difficulty.
6431 One of these methods is the @emph{runtime pseudo-relocs} feature, described
6434 @cindex runtime pseudo-relocation
6435 For complex variables imported from DLLs (such as structs or classes),
6436 object files typically contain a base address for the variable and an
6437 offset (@emph{addend}) within the variable--to specify a particular
6438 field or public member, for instance. Unfortunately, the runtime loader used
6439 in win32 environments is incapable of fixing these references at runtime
6440 without the additional information supplied by dllimport/dllexport decorations.
6441 The standard auto-import feature described above is unable to resolve these
6444 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
6445 be resolved without error, while leaving the task of adjusting the references
6446 themselves (with their non-zero addends) to specialized code provided by the
6447 runtime environment. Recent versions of the cygwin and mingw environments and
6448 compilers provide this runtime support; older versions do not. However, the
6449 support is only necessary on the developer's platform; the compiled result will
6450 run without error on an older system.
6452 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
6455 @cindex direct linking to a dll
6456 @item direct linking to a dll
6457 The cygwin/mingw ports of @command{ld} support the direct linking,
6458 including data symbols, to a dll without the usage of any import
6459 libraries. This is much faster and uses much less memory than does the
6460 traditional import library method, especially when linking large
6461 libraries or applications. When @command{ld} creates an import lib, each
6462 function or variable exported from the dll is stored in its own bfd, even
6463 though a single bfd could contain many exports. The overhead involved in
6464 storing, loading, and processing so many bfd's is quite large, and explains the
6465 tremendous time, memory, and storage needed to link against particularly
6466 large or complex libraries when using import libs.
6468 Linking directly to a dll uses no extra command-line switches other than
6469 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
6470 of names to match each library. All that is needed from the developer's
6471 perspective is an understanding of this search, in order to force ld to
6472 select the dll instead of an import library.
6475 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
6476 to find, in the first directory of its search path,
6488 before moving on to the next directory in the search path.
6490 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
6491 where @samp{<prefix>} is set by the @command{ld} option
6492 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6493 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6496 Other win32-based unix environments, such as mingw or pw32, may use other
6497 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
6498 was originally intended to help avoid name conflicts among dll's built for the
6499 various win32/un*x environments, so that (for example) two versions of a zlib dll
6500 could coexist on the same machine.
6502 The generic cygwin/mingw path layout uses a @samp{bin} directory for
6503 applications and dll's and a @samp{lib} directory for the import
6504 libraries (using cygwin nomenclature):
6510 libxxx.dll.a (in case of dll's)
6511 libxxx.a (in case of static archive)
6514 Linking directly to a dll without using the import library can be
6517 1. Use the dll directly by adding the @samp{bin} path to the link line
6519 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6522 However, as the dll's often have version numbers appended to their names
6523 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6524 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
6525 not versioned, and do not have this difficulty.
6527 2. Create a symbolic link from the dll to a file in the @samp{lib}
6528 directory according to the above mentioned search pattern. This
6529 should be used to avoid unwanted changes in the tools needed for
6533 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6536 Then you can link without any make environment changes.
6539 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6542 This technique also avoids the version number problems, because the following is
6549 libxxx.dll.a -> ../bin/cygxxx-5.dll
6552 Linking directly to a dll without using an import lib will work
6553 even when auto-import features are exercised, and even when
6554 @samp{--enable-runtime-pseudo-relocs} is used.
6556 Given the improvements in speed and memory usage, one might justifiably
6557 wonder why import libraries are used at all. There are three reasons:
6559 1. Until recently, the link-directly-to-dll functionality did @emph{not}
6560 work with auto-imported data.
6562 2. Sometimes it is necessary to include pure static objects within the
6563 import library (which otherwise contains only bfd's for indirection
6564 symbols that point to the exports of a dll). Again, the import lib
6565 for the cygwin kernel makes use of this ability, and it is not
6566 possible to do this without an import lib.
6568 3. Symbol aliases can only be resolved using an import lib. This is
6569 critical when linking against OS-supplied dll's (eg, the win32 API)
6570 in which symbols are usually exported as undecorated aliases of their
6571 stdcall-decorated assembly names.
6573 So, import libs are not going away. But the ability to replace
6574 true import libs with a simple symbolic link to (or a copy of)
6575 a dll, in many cases, is a useful addition to the suite of tools
6576 binutils makes available to the win32 developer. Given the
6577 massive improvements in memory requirements during linking, storage
6578 requirements, and linking speed, we expect that many developers
6579 will soon begin to use this feature whenever possible.
6581 @item symbol aliasing
6583 @item adding additional names
6584 Sometimes, it is useful to export symbols with additional names.
6585 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6586 exported as @samp{_foo} by using special directives in the DEF file
6587 when creating the dll. This will affect also the optional created
6588 import library. Consider the following DEF file:
6591 LIBRARY "xyz.dll" BASE=0x61000000
6598 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6600 Another method for creating a symbol alias is to create it in the
6601 source code using the "weak" attribute:
6604 void foo () @{ /* Do something. */; @}
6605 void _foo () __attribute__ ((weak, alias ("foo")));
6608 See the gcc manual for more information about attributes and weak
6611 @item renaming symbols
6612 Sometimes it is useful to rename exports. For instance, the cygwin
6613 kernel does this regularly. A symbol @samp{_foo} can be exported as
6614 @samp{foo} but not as @samp{_foo} by using special directives in the
6615 DEF file. (This will also affect the import library, if it is
6616 created). In the following example:
6619 LIBRARY "xyz.dll" BASE=0x61000000
6625 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6629 Note: using a DEF file disables the default auto-export behavior,
6630 unless the @samp{--export-all-symbols} command line option is used.
6631 If, however, you are trying to rename symbols, then you should list
6632 @emph{all} desired exports in the DEF file, including the symbols
6633 that are not being renamed, and do @emph{not} use the
6634 @samp{--export-all-symbols} option. If you list only the
6635 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6636 to handle the other symbols, then the both the new names @emph{and}
6637 the original names for the renamed symbols will be exported.
6638 In effect, you'd be aliasing those symbols, not renaming them,
6639 which is probably not what you wanted.
6641 @cindex weak externals
6642 @item weak externals
6643 The Windows object format, PE, specifies a form of weak symbols called
6644 weak externals. When a weak symbol is linked and the symbol is not
6645 defined, the weak symbol becomes an alias for some other symbol. There
6646 are three variants of weak externals:
6648 @item Definition is searched for in objects and libraries, historically
6649 called lazy externals.
6650 @item Definition is searched for only in other objects, not in libraries.
6651 This form is not presently implemented.
6652 @item No search; the symbol is an alias. This form is not presently
6655 As a GNU extension, weak symbols that do not specify an alternate symbol
6656 are supported. If the symbol is undefined when linking, the symbol
6657 uses a default value.
6671 @section @code{ld} and Xtensa Processors
6673 @cindex Xtensa processors
6674 The default @command{ld} behavior for Xtensa processors is to interpret
6675 @code{SECTIONS} commands so that lists of explicitly named sections in a
6676 specification with a wildcard file will be interleaved when necessary to
6677 keep literal pools within the range of PC-relative load offsets. For
6678 example, with the command:
6690 @command{ld} may interleave some of the @code{.literal}
6691 and @code{.text} sections from different object files to ensure that the
6692 literal pools are within the range of PC-relative load offsets. A valid
6693 interleaving might place the @code{.literal} sections from an initial
6694 group of files followed by the @code{.text} sections of that group of
6695 files. Then, the @code{.literal} sections from the rest of the files
6696 and the @code{.text} sections from the rest of the files would follow.
6698 @cindex @option{--relax} on Xtensa
6699 @cindex relaxing on Xtensa
6700 Relaxation is enabled by default for the Xtensa version of @command{ld} and
6701 provides two important link-time optimizations. The first optimization
6702 is to combine identical literal values to reduce code size. A redundant
6703 literal will be removed and all the @code{L32R} instructions that use it
6704 will be changed to reference an identical literal, as long as the
6705 location of the replacement literal is within the offset range of all
6706 the @code{L32R} instructions. The second optimization is to remove
6707 unnecessary overhead from assembler-generated ``longcall'' sequences of
6708 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
6709 range of direct @code{CALL@var{n}} instructions.
6711 For each of these cases where an indirect call sequence can be optimized
6712 to a direct call, the linker will change the @code{CALLX@var{n}}
6713 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
6714 instruction, and remove the literal referenced by the @code{L32R}
6715 instruction if it is not used for anything else. Removing the
6716 @code{L32R} instruction always reduces code size but can potentially
6717 hurt performance by changing the alignment of subsequent branch targets.
6718 By default, the linker will always preserve alignments, either by
6719 switching some instructions between 24-bit encodings and the equivalent
6720 density instructions or by inserting a no-op in place of the @code{L32R}
6721 instruction that was removed. If code size is more important than
6722 performance, the @option{--size-opt} option can be used to prevent the
6723 linker from widening density instructions or inserting no-ops, except in
6724 a few cases where no-ops are required for correctness.
6726 The following Xtensa-specific command-line options can be used to
6729 @cindex Xtensa options
6733 Since the Xtensa version of @code{ld} enables the @option{--relax} option
6734 by default, the @option{--no-relax} option is provided to disable
6738 When optimizing indirect calls to direct calls, optimize for code size
6739 more than performance. With this option, the linker will not insert
6740 no-ops or widen density instructions to preserve branch target
6741 alignment. There may still be some cases where no-ops are required to
6742 preserve the correctness of the code.
6750 @ifclear SingleFormat
6755 @cindex object file management
6756 @cindex object formats available
6758 The linker accesses object and archive files using the BFD libraries.
6759 These libraries allow the linker to use the same routines to operate on
6760 object files whatever the object file format. A different object file
6761 format can be supported simply by creating a new BFD back end and adding
6762 it to the library. To conserve runtime memory, however, the linker and
6763 associated tools are usually configured to support only a subset of the
6764 object file formats available. You can use @code{objdump -i}
6765 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
6766 list all the formats available for your configuration.
6768 @cindex BFD requirements
6769 @cindex requirements for BFD
6770 As with most implementations, BFD is a compromise between
6771 several conflicting requirements. The major factor influencing
6772 BFD design was efficiency: any time used converting between
6773 formats is time which would not have been spent had BFD not
6774 been involved. This is partly offset by abstraction payback; since
6775 BFD simplifies applications and back ends, more time and care
6776 may be spent optimizing algorithms for a greater speed.
6778 One minor artifact of the BFD solution which you should bear in
6779 mind is the potential for information loss. There are two places where
6780 useful information can be lost using the BFD mechanism: during
6781 conversion and during output. @xref{BFD information loss}.
6784 * BFD outline:: How it works: an outline of BFD
6788 @section How It Works: An Outline of BFD
6789 @cindex opening object files
6790 @include bfdsumm.texi
6793 @node Reporting Bugs
6794 @chapter Reporting Bugs
6795 @cindex bugs in @command{ld}
6796 @cindex reporting bugs in @command{ld}
6798 Your bug reports play an essential role in making @command{ld} reliable.
6800 Reporting a bug may help you by bringing a solution to your problem, or
6801 it may not. But in any case the principal function of a bug report is
6802 to help the entire community by making the next version of @command{ld}
6803 work better. Bug reports are your contribution to the maintenance of
6806 In order for a bug report to serve its purpose, you must include the
6807 information that enables us to fix the bug.
6810 * Bug Criteria:: Have you found a bug?
6811 * Bug Reporting:: How to report bugs
6815 @section Have You Found a Bug?
6816 @cindex bug criteria
6818 If you are not sure whether you have found a bug, here are some guidelines:
6821 @cindex fatal signal
6822 @cindex linker crash
6823 @cindex crash of linker
6825 If the linker gets a fatal signal, for any input whatever, that is a
6826 @command{ld} bug. Reliable linkers never crash.
6828 @cindex error on valid input
6830 If @command{ld} produces an error message for valid input, that is a bug.
6832 @cindex invalid input
6834 If @command{ld} does not produce an error message for invalid input, that
6835 may be a bug. In the general case, the linker can not verify that
6836 object files are correct.
6839 If you are an experienced user of linkers, your suggestions for
6840 improvement of @command{ld} are welcome in any case.
6844 @section How to Report Bugs
6846 @cindex @command{ld} bugs, reporting
6848 A number of companies and individuals offer support for @sc{gnu}
6849 products. If you obtained @command{ld} from a support organization, we
6850 recommend you contact that organization first.
6852 You can find contact information for many support companies and
6853 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
6857 Otherwise, send bug reports for @command{ld} to
6861 The fundamental principle of reporting bugs usefully is this:
6862 @strong{report all the facts}. If you are not sure whether to state a
6863 fact or leave it out, state it!
6865 Often people omit facts because they think they know what causes the
6866 problem and assume that some details do not matter. Thus, you might
6867 assume that the name of a symbol you use in an example does not
6868 matter. Well, probably it does not, but one cannot be sure. Perhaps
6869 the bug is a stray memory reference which happens to fetch from the
6870 location where that name is stored in memory; perhaps, if the name
6871 were different, the contents of that location would fool the linker
6872 into doing the right thing despite the bug. Play it safe and give a
6873 specific, complete example. That is the easiest thing for you to do,
6874 and the most helpful.
6876 Keep in mind that the purpose of a bug report is to enable us to fix
6877 the bug if it is new to us. Therefore, always write your bug reports
6878 on the assumption that the bug has not been reported previously.
6880 Sometimes people give a few sketchy facts and ask, ``Does this ring a
6881 bell?'' This cannot help us fix a bug, so it is basically useless. We
6882 respond by asking for enough details to enable us to investigate.
6883 You might as well expedite matters by sending them to begin with.
6885 To enable us to fix the bug, you should include all these things:
6889 The version of @command{ld}. @command{ld} announces it if you start it with
6890 the @samp{--version} argument.
6892 Without this, we will not know whether there is any point in looking for
6893 the bug in the current version of @command{ld}.
6896 Any patches you may have applied to the @command{ld} source, including any
6897 patches made to the @code{BFD} library.
6900 The type of machine you are using, and the operating system name and
6904 What compiler (and its version) was used to compile @command{ld}---e.g.
6908 The command arguments you gave the linker to link your example and
6909 observe the bug. To guarantee you will not omit something important,
6910 list them all. A copy of the Makefile (or the output from make) is
6913 If we were to try to guess the arguments, we would probably guess wrong
6914 and then we might not encounter the bug.
6917 A complete input file, or set of input files, that will reproduce the
6918 bug. It is generally most helpful to send the actual object files
6919 provided that they are reasonably small. Say no more than 10K. For
6920 bigger files you can either make them available by FTP or HTTP or else
6921 state that you are willing to send the object file(s) to whomever
6922 requests them. (Note - your email will be going to a mailing list, so
6923 we do not want to clog it up with large attachments). But small
6924 attachments are best.
6926 If the source files were assembled using @code{gas} or compiled using
6927 @code{gcc}, then it may be OK to send the source files rather than the
6928 object files. In this case, be sure to say exactly what version of
6929 @code{gas} or @code{gcc} was used to produce the object files. Also say
6930 how @code{gas} or @code{gcc} were configured.
6933 A description of what behavior you observe that you believe is
6934 incorrect. For example, ``It gets a fatal signal.''
6936 Of course, if the bug is that @command{ld} gets a fatal signal, then we
6937 will certainly notice it. But if the bug is incorrect output, we might
6938 not notice unless it is glaringly wrong. You might as well not give us
6939 a chance to make a mistake.
6941 Even if the problem you experience is a fatal signal, you should still
6942 say so explicitly. Suppose something strange is going on, such as, your
6943 copy of @command{ld} is out of sync, or you have encountered a bug in the
6944 C library on your system. (This has happened!) Your copy might crash
6945 and ours would not. If you told us to expect a crash, then when ours
6946 fails to crash, we would know that the bug was not happening for us. If
6947 you had not told us to expect a crash, then we would not be able to draw
6948 any conclusion from our observations.
6951 If you wish to suggest changes to the @command{ld} source, send us context
6952 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
6953 @samp{-p} option. Always send diffs from the old file to the new file.
6954 If you even discuss something in the @command{ld} source, refer to it by
6955 context, not by line number.
6957 The line numbers in our development sources will not match those in your
6958 sources. Your line numbers would convey no useful information to us.
6961 Here are some things that are not necessary:
6965 A description of the envelope of the bug.
6967 Often people who encounter a bug spend a lot of time investigating
6968 which changes to the input file will make the bug go away and which
6969 changes will not affect it.
6971 This is often time consuming and not very useful, because the way we
6972 will find the bug is by running a single example under the debugger
6973 with breakpoints, not by pure deduction from a series of examples.
6974 We recommend that you save your time for something else.
6976 Of course, if you can find a simpler example to report @emph{instead}
6977 of the original one, that is a convenience for us. Errors in the
6978 output will be easier to spot, running under the debugger will take
6979 less time, and so on.
6981 However, simplification is not vital; if you do not want to do this,
6982 report the bug anyway and send us the entire test case you used.
6985 A patch for the bug.
6987 A patch for the bug does help us if it is a good one. But do not omit
6988 the necessary information, such as the test case, on the assumption that
6989 a patch is all we need. We might see problems with your patch and decide
6990 to fix the problem another way, or we might not understand it at all.
6992 Sometimes with a program as complicated as @command{ld} it is very hard to
6993 construct an example that will make the program follow a certain path
6994 through the code. If you do not send us the example, we will not be
6995 able to construct one, so we will not be able to verify that the bug is
6998 And if we cannot understand what bug you are trying to fix, or why your
6999 patch should be an improvement, we will not install it. A test case will
7000 help us to understand.
7003 A guess about what the bug is or what it depends on.
7005 Such guesses are usually wrong. Even we cannot guess right about such
7006 things without first using the debugger to find the facts.
7010 @appendix MRI Compatible Script Files
7011 @cindex MRI compatibility
7012 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
7013 linker, @command{ld} can use MRI compatible linker scripts as an
7014 alternative to the more general-purpose linker scripting language
7015 described in @ref{Scripts}. MRI compatible linker scripts have a much
7016 simpler command set than the scripting language otherwise used with
7017 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
7018 linker commands; these commands are described here.
7020 In general, MRI scripts aren't of much use with the @code{a.out} object
7021 file format, since it only has three sections and MRI scripts lack some
7022 features to make use of them.
7024 You can specify a file containing an MRI-compatible script using the
7025 @samp{-c} command-line option.
7027 Each command in an MRI-compatible script occupies its own line; each
7028 command line starts with the keyword that identifies the command (though
7029 blank lines are also allowed for punctuation). If a line of an
7030 MRI-compatible script begins with an unrecognized keyword, @command{ld}
7031 issues a warning message, but continues processing the script.
7033 Lines beginning with @samp{*} are comments.
7035 You can write these commands using all upper-case letters, or all
7036 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
7037 The following list shows only the upper-case form of each command.
7040 @cindex @code{ABSOLUTE} (MRI)
7041 @item ABSOLUTE @var{secname}
7042 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
7043 Normally, @command{ld} includes in the output file all sections from all
7044 the input files. However, in an MRI-compatible script, you can use the
7045 @code{ABSOLUTE} command to restrict the sections that will be present in
7046 your output program. If the @code{ABSOLUTE} command is used at all in a
7047 script, then only the sections named explicitly in @code{ABSOLUTE}
7048 commands will appear in the linker output. You can still use other
7049 input sections (whatever you select on the command line, or using
7050 @code{LOAD}) to resolve addresses in the output file.
7052 @cindex @code{ALIAS} (MRI)
7053 @item ALIAS @var{out-secname}, @var{in-secname}
7054 Use this command to place the data from input section @var{in-secname}
7055 in a section called @var{out-secname} in the linker output file.
7057 @var{in-secname} may be an integer.
7059 @cindex @code{ALIGN} (MRI)
7060 @item ALIGN @var{secname} = @var{expression}
7061 Align the section called @var{secname} to @var{expression}. The
7062 @var{expression} should be a power of two.
7064 @cindex @code{BASE} (MRI)
7065 @item BASE @var{expression}
7066 Use the value of @var{expression} as the lowest address (other than
7067 absolute addresses) in the output file.
7069 @cindex @code{CHIP} (MRI)
7070 @item CHIP @var{expression}
7071 @itemx CHIP @var{expression}, @var{expression}
7072 This command does nothing; it is accepted only for compatibility.
7074 @cindex @code{END} (MRI)
7076 This command does nothing whatever; it's only accepted for compatibility.
7078 @cindex @code{FORMAT} (MRI)
7079 @item FORMAT @var{output-format}
7080 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
7081 language, but restricted to one of these output formats:
7085 S-records, if @var{output-format} is @samp{S}
7088 IEEE, if @var{output-format} is @samp{IEEE}
7091 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
7095 @cindex @code{LIST} (MRI)
7096 @item LIST @var{anything}@dots{}
7097 Print (to the standard output file) a link map, as produced by the
7098 @command{ld} command-line option @samp{-M}.
7100 The keyword @code{LIST} may be followed by anything on the
7101 same line, with no change in its effect.
7103 @cindex @code{LOAD} (MRI)
7104 @item LOAD @var{filename}
7105 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
7106 Include one or more object file @var{filename} in the link; this has the
7107 same effect as specifying @var{filename} directly on the @command{ld}
7110 @cindex @code{NAME} (MRI)
7111 @item NAME @var{output-name}
7112 @var{output-name} is the name for the program produced by @command{ld}; the
7113 MRI-compatible command @code{NAME} is equivalent to the command-line
7114 option @samp{-o} or the general script language command @code{OUTPUT}.
7116 @cindex @code{ORDER} (MRI)
7117 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
7118 @itemx ORDER @var{secname} @var{secname} @var{secname}
7119 Normally, @command{ld} orders the sections in its output file in the
7120 order in which they first appear in the input files. In an MRI-compatible
7121 script, you can override this ordering with the @code{ORDER} command. The
7122 sections you list with @code{ORDER} will appear first in your output
7123 file, in the order specified.
7125 @cindex @code{PUBLIC} (MRI)
7126 @item PUBLIC @var{name}=@var{expression}
7127 @itemx PUBLIC @var{name},@var{expression}
7128 @itemx PUBLIC @var{name} @var{expression}
7129 Supply a value (@var{expression}) for external symbol
7130 @var{name} used in the linker input files.
7132 @cindex @code{SECT} (MRI)
7133 @item SECT @var{secname}, @var{expression}
7134 @itemx SECT @var{secname}=@var{expression}
7135 @itemx SECT @var{secname} @var{expression}
7136 You can use any of these three forms of the @code{SECT} command to
7137 specify the start address (@var{expression}) for section @var{secname}.
7138 If you have more than one @code{SECT} statement for the same
7139 @var{secname}, only the @emph{first} sets the start address.
7145 @unnumbered LD Index
7150 % I think something like @colophon should be in texinfo. In the
7152 \long\def\colophon{\hbox to0pt{}\vfill
7153 \centerline{The body of this manual is set in}
7154 \centerline{\fontname\tenrm,}
7155 \centerline{with headings in {\bf\fontname\tenbf}}
7156 \centerline{and examples in {\tt\fontname\tentt}.}
7157 \centerline{{\it\fontname\tenit\/} and}
7158 \centerline{{\sl\fontname\tensl\/}}
7159 \centerline{are used for emphasis.}\vfill}
7161 % Blame: doc@cygnus.com, 28mar91.