3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007 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
43 * Ld: (ld). The GNU linker.
49 This file documents the @sc{gnu} linker LD
50 @ifset VERSION_PACKAGE
51 @value{VERSION_PACKAGE}
53 version @value{VERSION}.
55 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
56 2001, 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
58 Permission is granted to copy, distribute and/or modify this document
59 under the terms of the GNU Free Documentation License, Version 1.1
60 or any later version published by the Free Software Foundation;
61 with no Invariant Sections, with no Front-Cover Texts, and with no
62 Back-Cover Texts. A copy of the license is included in the
63 section entitled ``GNU Free Documentation License''.
67 @setchapternewpage odd
68 @settitle The GNU linker
73 @ifset VERSION_PACKAGE
74 @subtitle @value{VERSION_PACKAGE}
76 @subtitle Version @value{VERSION}
77 @author Steve Chamberlain
78 @author Ian Lance Taylor
83 \hfill Red Hat Inc\par
84 \hfill nickc\@credhat.com, doc\@redhat.com\par
85 \hfill {\it The GNU linker}\par
86 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
88 \global\parindent=0pt % Steve likes it this way.
91 @vskip 0pt plus 1filll
92 @c man begin COPYRIGHT
93 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
94 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
96 Permission is granted to copy, distribute and/or modify this document
97 under the terms of the GNU Free Documentation License, Version 1.1
98 or any later version published by the Free Software Foundation;
99 with no Invariant Sections, with no Front-Cover Texts, and with no
100 Back-Cover Texts. A copy of the license is included in the
101 section entitled ``GNU Free Documentation License''.
107 @c FIXME: Talk about importance of *order* of args, cmds to linker!
112 This file documents the @sc{gnu} linker ld
113 @ifset VERSION_PACKAGE
114 @value{VERSION_PACKAGE}
116 version @value{VERSION}.
118 This document is distributed under the terms of the GNU Free
119 Documentation License. A copy of the license is included in the
120 section entitled ``GNU Free Documentation License''.
123 * Overview:: Overview
124 * Invocation:: Invocation
125 * Scripts:: Linker Scripts
127 * Machine Dependent:: Machine Dependent Features
131 * H8/300:: ld and the H8/300
134 * Renesas:: ld and other Renesas micros
137 * i960:: ld and the Intel 960 family
140 * ARM:: ld and the ARM family
143 * HPPA ELF32:: ld and HPPA 32-bit ELF
146 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
149 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
152 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
155 * SPU ELF:: ld and SPU ELF Support
158 * TI COFF:: ld and the TI COFF
161 * Win32:: ld and WIN32 (cygwin/mingw)
164 * Xtensa:: ld and Xtensa Processors
167 @ifclear SingleFormat
170 @c Following blank line required for remaining bug in makeinfo conds/menus
172 * Reporting Bugs:: Reporting Bugs
173 * MRI:: MRI Compatible Script Files
174 * GNU Free Documentation License:: GNU Free Documentation License
175 * LD Index:: LD Index
182 @cindex @sc{gnu} linker
183 @cindex what is this?
186 @c man begin SYNOPSIS
187 ld [@b{options}] @var{objfile} @dots{}
191 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
192 the Info entries for @file{binutils} and
197 @c man begin DESCRIPTION
199 @command{ld} combines a number of object and archive files, relocates
200 their data and ties up symbol references. Usually the last step in
201 compiling a program is to run @command{ld}.
203 @command{ld} accepts Linker Command Language files written in
204 a superset of AT&T's Link Editor Command Language syntax,
205 to provide explicit and total control over the linking process.
209 This man page does not describe the command language; see the
210 @command{ld} entry in @code{info} for full details on the command
211 language and on other aspects of the GNU linker.
214 @ifclear SingleFormat
215 This version of @command{ld} uses the general purpose BFD libraries
216 to operate on object files. This allows @command{ld} to read, combine, and
217 write object files in many different formats---for example, COFF or
218 @code{a.out}. Different formats may be linked together to produce any
219 available kind of object file. @xref{BFD}, for more information.
222 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
223 linkers in providing diagnostic information. Many linkers abandon
224 execution immediately upon encountering an error; whenever possible,
225 @command{ld} continues executing, allowing you to identify other errors
226 (or, in some cases, to get an output file in spite of the error).
233 @c man begin DESCRIPTION
235 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
236 and to be as compatible as possible with other linkers. As a result,
237 you have many choices to control its behavior.
243 * Options:: Command Line Options
244 * Environment:: Environment Variables
248 @section Command Line Options
256 The linker supports a plethora of command-line options, but in actual
257 practice few of them are used in any particular context.
258 @cindex standard Unix system
259 For instance, a frequent use of @command{ld} is to link standard Unix
260 object files on a standard, supported Unix system. On such a system, to
261 link a file @code{hello.o}:
264 ld -o @var{output} /lib/crt0.o hello.o -lc
267 This tells @command{ld} to produce a file called @var{output} as the
268 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
269 the library @code{libc.a}, which will come from the standard search
270 directories. (See the discussion of the @samp{-l} option below.)
272 Some of the command-line options to @command{ld} may be specified at any
273 point in the command line. However, options which refer to files, such
274 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
275 which the option appears in the command line, relative to the object
276 files and other file options. Repeating non-file options with a
277 different argument will either have no further effect, or override prior
278 occurrences (those further to the left on the command line) of that
279 option. Options which may be meaningfully specified more than once are
280 noted in the descriptions below.
283 Non-option arguments are object files or archives which are to be linked
284 together. They may follow, precede, or be mixed in with command-line
285 options, except that an object file argument may not be placed between
286 an option and its argument.
288 Usually the linker is invoked with at least one object file, but you can
289 specify other forms of binary input files using @samp{-l}, @samp{-R},
290 and the script command language. If @emph{no} binary input files at all
291 are specified, the linker does not produce any output, and issues the
292 message @samp{No input files}.
294 If the linker cannot recognize the format of an object file, it will
295 assume that it is a linker script. A script specified in this way
296 augments the main linker script used for the link (either the default
297 linker script or the one specified by using @samp{-T}). This feature
298 permits the linker to link against a file which appears to be an object
299 or an archive, but actually merely defines some symbol values, or uses
300 @code{INPUT} or @code{GROUP} to load other objects. Note that
301 specifying a script in this way merely augments the main linker script;
302 use the @samp{-T} option to replace the default linker script entirely.
305 For options whose names are a single letter,
306 option arguments must either follow the option letter without intervening
307 whitespace, or be given as separate arguments immediately following the
308 option that requires them.
310 For options whose names are multiple letters, either one dash or two can
311 precede the option name; for example, @samp{-trace-symbol} and
312 @samp{--trace-symbol} are equivalent. Note---there is one exception to
313 this rule. Multiple letter options that start with a lower case 'o' can
314 only be preceded by two dashes. This is to reduce confusion with the
315 @samp{-o} option. So for example @samp{-omagic} sets the output file
316 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
319 Arguments to multiple-letter options must either be separated from the
320 option name by an equals sign, or be given as separate arguments
321 immediately following the option that requires them. For example,
322 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
323 Unique abbreviations of the names of multiple-letter options are
326 Note---if the linker is being invoked indirectly, via a compiler driver
327 (e.g. @samp{gcc}) then all the linker command line options should be
328 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
329 compiler driver) like this:
332 gcc -Wl,--startgroup foo.o bar.o -Wl,--endgroup
335 This is important, because otherwise the compiler driver program may
336 silently drop the linker options, resulting in a bad link.
338 Here is a table of the generic command line switches accepted by the GNU
342 @include at-file.texi
344 @kindex -a@var{keyword}
345 @item -a@var{keyword}
346 This option is supported for HP/UX compatibility. The @var{keyword}
347 argument must be one of the strings @samp{archive}, @samp{shared}, or
348 @samp{default}. @samp{-aarchive} is functionally equivalent to
349 @samp{-Bstatic}, and the other two keywords are functionally equivalent
350 to @samp{-Bdynamic}. This option may be used any number of times.
353 @cindex architectures
355 @item -A@var{architecture}
356 @kindex --architecture=@var{arch}
357 @itemx --architecture=@var{architecture}
358 In the current release of @command{ld}, this option is useful only for the
359 Intel 960 family of architectures. In that @command{ld} configuration, the
360 @var{architecture} argument identifies the particular architecture in
361 the 960 family, enabling some safeguards and modifying the
362 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
363 family}, for details.
365 Future releases of @command{ld} may support similar functionality for
366 other architecture families.
369 @ifclear SingleFormat
370 @cindex binary input format
371 @kindex -b @var{format}
372 @kindex --format=@var{format}
375 @item -b @var{input-format}
376 @itemx --format=@var{input-format}
377 @command{ld} may be configured to support more than one kind of object
378 file. If your @command{ld} is configured this way, you can use the
379 @samp{-b} option to specify the binary format for input object files
380 that follow this option on the command line. Even when @command{ld} is
381 configured to support alternative object formats, you don't usually need
382 to specify this, as @command{ld} should be configured to expect as a
383 default input format the most usual format on each machine.
384 @var{input-format} is a text string, the name of a particular format
385 supported by the BFD libraries. (You can list the available binary
386 formats with @samp{objdump -i}.)
389 You may want to use this option if you are linking files with an unusual
390 binary format. You can also use @samp{-b} to switch formats explicitly (when
391 linking object files of different formats), by including
392 @samp{-b @var{input-format}} before each group of object files in a
395 The default format is taken from the environment variable
400 You can also define the input format from a script, using the command
403 see @ref{Format Commands}.
407 @kindex -c @var{MRI-cmdfile}
408 @kindex --mri-script=@var{MRI-cmdfile}
409 @cindex compatibility, MRI
410 @item -c @var{MRI-commandfile}
411 @itemx --mri-script=@var{MRI-commandfile}
412 For compatibility with linkers produced by MRI, @command{ld} accepts script
413 files written in an alternate, restricted command language, described in
415 @ref{MRI,,MRI Compatible Script Files}.
418 the MRI Compatible Script Files section of GNU ld documentation.
420 Introduce MRI script files with
421 the option @samp{-c}; use the @samp{-T} option to run linker
422 scripts written in the general-purpose @command{ld} scripting language.
423 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
424 specified by any @samp{-L} options.
426 @cindex common allocation
433 These three options are equivalent; multiple forms are supported for
434 compatibility with other linkers. They assign space to common symbols
435 even if a relocatable output file is specified (with @samp{-r}). The
436 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
437 @xref{Miscellaneous Commands}.
439 @cindex entry point, from command line
440 @kindex -e @var{entry}
441 @kindex --entry=@var{entry}
443 @itemx --entry=@var{entry}
444 Use @var{entry} as the explicit symbol for beginning execution of your
445 program, rather than the default entry point. If there is no symbol
446 named @var{entry}, the linker will try to parse @var{entry} as a number,
447 and use that as the entry address (the number will be interpreted in
448 base 10; you may use a leading @samp{0x} for base 16, or a leading
449 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
450 and other ways of specifying the entry point.
452 @kindex --exclude-libs
453 @item --exclude-libs @var{lib},@var{lib},...
454 Specifies a list of archive libraries from which symbols should not be automatically
455 exported. The library names may be delimited by commas or colons. Specifying
456 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
457 automatic export. This option is available only for the i386 PE targeted
458 port of the linker and for ELF targeted ports. For i386 PE, symbols
459 explicitly listed in a .def file are still exported, regardless of this
460 option. For ELF targeted ports, symbols affected by this option will
461 be treated as hidden.
463 @cindex dynamic symbol table
465 @kindex --export-dynamic
467 @itemx --export-dynamic
468 When creating a dynamically linked executable, add all symbols to the
469 dynamic symbol table. The dynamic symbol table is the set of symbols
470 which are visible from dynamic objects at run time.
472 If you do not use this option, the dynamic symbol table will normally
473 contain only those symbols which are referenced by some dynamic object
474 mentioned in the link.
476 If you use @code{dlopen} to load a dynamic object which needs to refer
477 back to the symbols defined by the program, rather than some other
478 dynamic object, then you will probably need to use this option when
479 linking the program itself.
481 You can also use the dynamic list to control what symbols should
482 be added to the dynamic symbol table if the output format supports it.
483 See the description of @samp{--dynamic-list}.
485 @ifclear SingleFormat
486 @cindex big-endian objects
490 Link big-endian objects. This affects the default output format.
492 @cindex little-endian objects
495 Link little-endian objects. This affects the default output format.
501 @itemx --auxiliary @var{name}
502 When creating an ELF shared object, set the internal DT_AUXILIARY field
503 to the specified name. This tells the dynamic linker that the symbol
504 table of the shared object should be used as an auxiliary filter on the
505 symbol table of the shared object @var{name}.
507 If you later link a program against this filter object, then, when you
508 run the program, the dynamic linker will see the DT_AUXILIARY field. If
509 the dynamic linker resolves any symbols from the filter object, it will
510 first check whether there is a definition in the shared object
511 @var{name}. If there is one, it will be used instead of the definition
512 in the filter object. The shared object @var{name} need not exist.
513 Thus the shared object @var{name} may be used to provide an alternative
514 implementation of certain functions, perhaps for debugging or for
515 machine specific performance.
517 This option may be specified more than once. The DT_AUXILIARY entries
518 will be created in the order in which they appear on the command line.
523 @itemx --filter @var{name}
524 When creating an ELF shared object, set the internal DT_FILTER field to
525 the specified name. This tells the dynamic linker that the symbol table
526 of the shared object which is being created should be used as a filter
527 on the symbol table of the shared object @var{name}.
529 If you later link a program against this filter object, then, when you
530 run the program, the dynamic linker will see the DT_FILTER field. The
531 dynamic linker will resolve symbols according to the symbol table of the
532 filter object as usual, but it will actually link to the definitions
533 found in the shared object @var{name}. Thus the filter object can be
534 used to select a subset of the symbols provided by the object
537 Some older linkers used the @option{-F} option throughout a compilation
538 toolchain for specifying object-file format for both input and output
540 @ifclear SingleFormat
541 The @sc{gnu} linker uses other mechanisms for this purpose: the
542 @option{-b}, @option{--format}, @option{--oformat} options, the
543 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
544 environment variable.
546 The @sc{gnu} linker will ignore the @option{-F} option when not
547 creating an ELF shared object.
549 @cindex finalization function
551 @item -fini @var{name}
552 When creating an ELF executable or shared object, call NAME when the
553 executable or shared object is unloaded, by setting DT_FINI to the
554 address of the function. By default, the linker uses @code{_fini} as
555 the function to call.
559 Ignored. Provided for compatibility with other tools.
565 @itemx --gpsize=@var{value}
566 Set the maximum size of objects to be optimized using the GP register to
567 @var{size}. This is only meaningful for object file formats such as
568 MIPS ECOFF which supports putting large and small objects into different
569 sections. This is ignored for other object file formats.
571 @cindex runtime library name
573 @kindex -soname=@var{name}
575 @itemx -soname=@var{name}
576 When creating an ELF shared object, set the internal DT_SONAME field to
577 the specified name. When an executable is linked with a shared object
578 which has a DT_SONAME field, then when the executable is run the dynamic
579 linker will attempt to load the shared object specified by the DT_SONAME
580 field rather than the using the file name given to the linker.
583 @cindex incremental link
585 Perform an incremental link (same as option @samp{-r}).
587 @cindex initialization function
589 @item -init @var{name}
590 When creating an ELF executable or shared object, call NAME when the
591 executable or shared object is loaded, by setting DT_INIT to the address
592 of the function. By default, the linker uses @code{_init} as the
595 @cindex archive files, from cmd line
596 @kindex -l@var{namespec}
597 @kindex --library=@var{namespec}
598 @item -l@var{namespec}
599 @itemx --library=@var{namespec}
600 Add the archive or object file specified by @var{namespec} to the
601 list of files to link. This option may be used any number of times.
602 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
603 will search the library path for a file called @var{filename}, otherise it
604 will search the library path for a file called @file{lib@var{namespec}.a}.
606 On systems which support shared libraries, @command{ld} may also search for
607 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
608 and SunOS systems, @command{ld} will search a directory for a library
609 called @file{lib@var{namespec}.so} before searching for one called
610 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
611 indicates a shared library.) Note that this behavior does not apply
612 to @file{:@var{filename}}, which always specifies a file called
615 The linker will search an archive only once, at the location where it is
616 specified on the command line. If the archive defines a symbol which
617 was undefined in some object which appeared before the archive on the
618 command line, the linker will include the appropriate file(s) from the
619 archive. However, an undefined symbol in an object appearing later on
620 the command line will not cause the linker to search the archive again.
622 See the @option{-(} option for a way to force the linker to search
623 archives multiple times.
625 You may list the same archive multiple times on the command line.
628 This type of archive searching is standard for Unix linkers. However,
629 if you are using @command{ld} on AIX, note that it is different from the
630 behaviour of the AIX linker.
633 @cindex search directory, from cmd line
635 @kindex --library-path=@var{dir}
636 @item -L@var{searchdir}
637 @itemx --library-path=@var{searchdir}
638 Add path @var{searchdir} to the list of paths that @command{ld} will search
639 for archive libraries and @command{ld} control scripts. You may use this
640 option any number of times. The directories are searched in the order
641 in which they are specified on the command line. Directories specified
642 on the command line are searched before the default directories. All
643 @option{-L} options apply to all @option{-l} options, regardless of the
644 order in which the options appear.
646 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
647 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
650 The default set of paths searched (without being specified with
651 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
652 some cases also on how it was configured. @xref{Environment}.
655 The paths can also be specified in a link script with the
656 @code{SEARCH_DIR} command. Directories specified this way are searched
657 at the point in which the linker script appears in the command line.
660 @kindex -m @var{emulation}
661 @item -m@var{emulation}
662 Emulate the @var{emulation} linker. You can list the available
663 emulations with the @samp{--verbose} or @samp{-V} options.
665 If the @samp{-m} option is not used, the emulation is taken from the
666 @code{LDEMULATION} environment variable, if that is defined.
668 Otherwise, the default emulation depends upon how the linker was
676 Print a link map to the standard output. A link map provides
677 information about the link, including the following:
681 Where object files are mapped into memory.
683 How common symbols are allocated.
685 All archive members included in the link, with a mention of the symbol
686 which caused the archive member to be brought in.
688 The values assigned to symbols.
690 Note - symbols whose values are computed by an expression which
691 involves a reference to a previous value of the same symbol may not
692 have correct result displayed in the link map. This is because the
693 linker discards intermediate results and only retains the final value
694 of an expression. Under such circumstances the linker will display
695 the final value enclosed by square brackets. Thus for example a
696 linker script containing:
704 will produce the following output in the link map if the @option{-M}
709 [0x0000000c] foo = (foo * 0x4)
710 [0x0000000c] foo = (foo + 0x8)
713 See @ref{Expressions} for more information about expressions in linker
718 @cindex read-only text
723 Turn off page alignment of sections, and mark the output as
724 @code{NMAGIC} if possible.
728 @cindex read/write from cmd line
732 Set the text and data sections to be readable and writable. Also, do
733 not page-align the data segment, and disable linking against shared
734 libraries. If the output format supports Unix style magic numbers,
735 mark the output as @code{OMAGIC}. Note: Although a writable text section
736 is allowed for PE-COFF targets, it does not conform to the format
737 specification published by Microsoft.
742 This option negates most of the effects of the @option{-N} option. It
743 sets the text section to be read-only, and forces the data segment to
744 be page-aligned. Note - this option does not enable linking against
745 shared libraries. Use @option{-Bdynamic} for this.
747 @kindex -o @var{output}
748 @kindex --output=@var{output}
749 @cindex naming the output file
750 @item -o @var{output}
751 @itemx --output=@var{output}
752 Use @var{output} as the name for the program produced by @command{ld}; if this
753 option is not specified, the name @file{a.out} is used by default. The
754 script command @code{OUTPUT} can also specify the output file name.
756 @kindex -O @var{level}
757 @cindex generating optimized output
759 If @var{level} is a numeric values greater than zero @command{ld} optimizes
760 the output. This might take significantly longer and therefore probably
761 should only be enabled for the final binary. At the moment this
762 option only affects ELF shared library generation. Future releases of
763 the linker may make more use of this option. Also currently there is
764 no difference in the linker's behaviour for different non-zero values
765 of this option. Again this may change with future releases.
768 @kindex --emit-relocs
769 @cindex retain relocations in final executable
772 Leave relocation sections and contents in fully linked executables.
773 Post link analysis and optimization tools may need this information in
774 order to perform correct modifications of executables. This results
775 in larger executables.
777 This option is currently only supported on ELF platforms.
779 @kindex --force-dynamic
780 @cindex forcing the creation of dynamic sections
781 @item --force-dynamic
782 Force the output file to have dynamic sections. This option is specific
786 @cindex relocatable output
788 @kindex --relocatable
791 Generate relocatable output---i.e., generate an output file that can in
792 turn serve as input to @command{ld}. This is often called @dfn{partial
793 linking}. As a side effect, in environments that support standard Unix
794 magic numbers, this option also sets the output file's magic number to
796 @c ; see @option{-N}.
797 If this option is not specified, an absolute file is produced. When
798 linking C++ programs, this option @emph{will not} resolve references to
799 constructors; to do that, use @samp{-Ur}.
801 When an input file does not have the same format as the output file,
802 partial linking is only supported if that input file does not contain any
803 relocations. Different output formats can have further restrictions; for
804 example some @code{a.out}-based formats do not support partial linking
805 with input files in other formats at all.
807 This option does the same thing as @samp{-i}.
809 @kindex -R @var{file}
810 @kindex --just-symbols=@var{file}
811 @cindex symbol-only input
812 @item -R @var{filename}
813 @itemx --just-symbols=@var{filename}
814 Read symbol names and their addresses from @var{filename}, but do not
815 relocate it or include it in the output. This allows your output file
816 to refer symbolically to absolute locations of memory defined in other
817 programs. You may use this option more than once.
819 For compatibility with other ELF linkers, if the @option{-R} option is
820 followed by a directory name, rather than a file name, it is treated as
821 the @option{-rpath} option.
825 @cindex strip all symbols
828 Omit all symbol information from the output file.
831 @kindex --strip-debug
832 @cindex strip debugger symbols
835 Omit debugger symbol information (but not all symbols) from the output file.
839 @cindex input files, displaying
842 Print the names of the input files as @command{ld} processes them.
844 @kindex -T @var{script}
845 @kindex --script=@var{script}
847 @item -T @var{scriptfile}
848 @itemx --script=@var{scriptfile}
849 Use @var{scriptfile} as the linker script. This script replaces
850 @command{ld}'s default linker script (rather than adding to it), so
851 @var{commandfile} must specify everything necessary to describe the
852 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
853 the current directory, @code{ld} looks for it in the directories
854 specified by any preceding @samp{-L} options. Multiple @samp{-T}
857 @kindex -dT @var{script}
858 @kindex --default-script=@var{script}
860 @item -dT @var{scriptfile}
861 @itemx --default-script=@var{scriptfile}
862 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
864 This option is similar to the @option{--script} option except that
865 processing of the script is delayed until after the rest of the
866 command line has been processed. This allows options placed after the
867 @option{--default-script} option on the command line to affect the
868 behaviour of the linker script, which can be important when the linker
869 command line cannot be directly controlled by the user. (eg because
870 the command line is being constructed by another tool, such as
873 @kindex -u @var{symbol}
874 @kindex --undefined=@var{symbol}
875 @cindex undefined symbol
876 @item -u @var{symbol}
877 @itemx --undefined=@var{symbol}
878 Force @var{symbol} to be entered in the output file as an undefined
879 symbol. Doing this may, for example, trigger linking of additional
880 modules from standard libraries. @samp{-u} may be repeated with
881 different option arguments to enter additional undefined symbols. This
882 option is equivalent to the @code{EXTERN} linker script command.
887 For anything other than C++ programs, this option is equivalent to
888 @samp{-r}: it generates relocatable output---i.e., an output file that can in
889 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
890 @emph{does} resolve references to constructors, unlike @samp{-r}.
891 It does not work to use @samp{-Ur} on files that were themselves linked
892 with @samp{-Ur}; once the constructor table has been built, it cannot
893 be added to. Use @samp{-Ur} only for the last partial link, and
894 @samp{-r} for the others.
896 @kindex --unique[=@var{SECTION}]
897 @item --unique[=@var{SECTION}]
898 Creates a separate output section for every input section matching
899 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
900 missing, for every orphan input section. An orphan section is one not
901 specifically mentioned in a linker script. You may use this option
902 multiple times on the command line; It prevents the normal merging of
903 input sections with the same name, overriding output section assignments
913 Display the version number for @command{ld}. The @option{-V} option also
914 lists the supported emulations.
917 @kindex --discard-all
918 @cindex deleting local symbols
921 Delete all local symbols.
924 @kindex --discard-locals
925 @cindex local symbols, deleting
927 @itemx --discard-locals
928 Delete all temporary local symbols. (These symbols start with
929 system-specific local label prefixes, typically @samp{.L} for ELF systems
930 or @samp{L} for traditional a.out systems.)
932 @kindex -y @var{symbol}
933 @kindex --trace-symbol=@var{symbol}
934 @cindex symbol tracing
935 @item -y @var{symbol}
936 @itemx --trace-symbol=@var{symbol}
937 Print the name of each linked file in which @var{symbol} appears. This
938 option may be given any number of times. On many systems it is necessary
939 to prepend an underscore.
941 This option is useful when you have an undefined symbol in your link but
942 don't know where the reference is coming from.
944 @kindex -Y @var{path}
946 Add @var{path} to the default library search path. This option exists
947 for Solaris compatibility.
949 @kindex -z @var{keyword}
950 @item -z @var{keyword}
951 The recognized keywords are:
955 Combines multiple reloc sections and sorts them to make dynamic symbol
956 lookup caching possible.
959 Disallows undefined symbols in object files. Undefined symbols in
960 shared libraries are still allowed.
963 Marks the object as requiring executable stack.
966 This option is only meaningful when building a shared object.
967 It marks the object so that its runtime initialization will occur
968 before the runtime initialization of any other objects brought into
969 the process at the same time. Similarly the runtime finalization of
970 the object will occur after the runtime finalization of any other
974 Marks the object that its symbol table interposes before all symbols
975 but the primary executable.
978 When generating an executable or shared library, mark it to tell the
979 dynamic linker to defer function call resolution to the point when
980 the function is called (lazy binding), rather than at load time.
981 Lazy binding is the default.
984 Marks the object that its filters be processed immediately at
988 Allows multiple definitions.
991 Disables multiple reloc sections combining.
994 Disables production of copy relocs.
997 Marks the object that the search for dependencies of this object will
998 ignore any default library search paths.
1001 Marks the object shouldn't be unloaded at runtime.
1004 Marks the object not available to @code{dlopen}.
1007 Marks the object can not be dumped by @code{dldump}.
1010 Marks the object as not requiring executable stack.
1013 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1016 When generating an executable or shared library, mark it to tell the
1017 dynamic linker to resolve all symbols when the program is started, or
1018 when the shared library is linked to using dlopen, instead of
1019 deferring function call resolution to the point when the function is
1023 Marks the object may contain $ORIGIN.
1026 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1028 @item max-page-size=@var{value}
1029 Set the emulation maximum page size to @var{value}.
1031 @item common-page-size=@var{value}
1032 Set the emulation common page size to @var{value}.
1036 Other keywords are ignored for Solaris compatibility.
1039 @cindex groups of archives
1040 @item -( @var{archives} -)
1041 @itemx --start-group @var{archives} --end-group
1042 The @var{archives} should be a list of archive files. They may be
1043 either explicit file names, or @samp{-l} options.
1045 The specified archives are searched repeatedly until no new undefined
1046 references are created. Normally, an archive is searched only once in
1047 the order that it is specified on the command line. If a symbol in that
1048 archive is needed to resolve an undefined symbol referred to by an
1049 object in an archive that appears later on the command line, the linker
1050 would not be able to resolve that reference. By grouping the archives,
1051 they all be searched repeatedly until all possible references are
1054 Using this option has a significant performance cost. It is best to use
1055 it only when there are unavoidable circular references between two or
1058 @kindex --accept-unknown-input-arch
1059 @kindex --no-accept-unknown-input-arch
1060 @item --accept-unknown-input-arch
1061 @itemx --no-accept-unknown-input-arch
1062 Tells the linker to accept input files whose architecture cannot be
1063 recognised. The assumption is that the user knows what they are doing
1064 and deliberately wants to link in these unknown input files. This was
1065 the default behaviour of the linker, before release 2.14. The default
1066 behaviour from release 2.14 onwards is to reject such input files, and
1067 so the @samp{--accept-unknown-input-arch} option has been added to
1068 restore the old behaviour.
1071 @kindex --no-as-needed
1073 @itemx --no-as-needed
1074 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1075 on the command line after the @option{--as-needed} option. Normally,
1076 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1077 on the command line, regardless of whether the library is actually
1078 needed. @option{--as-needed} causes DT_NEEDED tags to only be emitted
1079 for libraries that satisfy some symbol reference from regular objects
1080 which is undefined at the point that the library was linked.
1081 @option{--no-as-needed} restores the default behaviour.
1083 @kindex --add-needed
1084 @kindex --no-add-needed
1086 @itemx --no-add-needed
1087 This option affects the treatment of dynamic libraries from ELF
1088 DT_NEEDED tags in dynamic libraries mentioned on the command line after
1089 the @option{--no-add-needed} option. Normally, the linker will add
1090 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1091 @option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1092 for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1093 the default behaviour.
1095 @kindex -assert @var{keyword}
1096 @item -assert @var{keyword}
1097 This option is ignored for SunOS compatibility.
1101 @kindex -call_shared
1105 Link against dynamic libraries. This is only meaningful on platforms
1106 for which shared libraries are supported. This option is normally the
1107 default on such platforms. The different variants of this option are
1108 for compatibility with various systems. You may use this option
1109 multiple times on the command line: it affects library searching for
1110 @option{-l} options which follow it.
1114 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1115 section. This causes the runtime linker to handle lookups in this
1116 object and its dependencies to be performed only inside the group.
1117 @option{--unresolved-symbols=report-all} is implied. This option is
1118 only meaningful on ELF platforms which support shared libraries.
1128 Do not link against shared libraries. This is only meaningful on
1129 platforms for which shared libraries are supported. The different
1130 variants of this option are for compatibility with various systems. You
1131 may use this option multiple times on the command line: it affects
1132 library searching for @option{-l} options which follow it. This
1133 option also implies @option{--unresolved-symbols=report-all}. This
1134 option can be used with @option{-shared}. Doing so means that a
1135 shared library is being created but that all of the library's external
1136 references must be resolved by pulling in entries from static
1141 When creating a shared library, bind references to global symbols to the
1142 definition within the shared library, if any. Normally, it is possible
1143 for a program linked against a shared library to override the definition
1144 within the shared library. This option is only meaningful on ELF
1145 platforms which support shared libraries.
1147 @kindex -Bsymbolic-functions
1148 @item -Bsymbolic-functions
1149 When creating a shared library, bind references to global function
1150 symbols to the definition within the shared library, if any.
1151 This option is only meaningful on ELF platforms which support shared
1154 @kindex --dynamic-list=@var{dynamic-list-file}
1155 @item --dynamic-list=@var{dynamic-list-file}
1156 Specify the name of a dynamic list file to the linker. This is
1157 typically used when creating shared libraries to specify a list of
1158 global symbols whose references shouldn't be bound to the definition
1159 within the shared library, or creating dynamically linked executables
1160 to specify a list of symbols which should be added to the symbol table
1161 in the executable. This option is only meaningful on ELF platforms
1162 which support shared libraries.
1164 The format of the dynamic list is the same as the version node without
1165 scope and node name. See @ref{VERSION} for more information.
1167 @kindex --dynamic-list-data
1168 @item --dynamic-list-data
1169 Include all global data symbols to the dynamic list.
1171 @kindex --dynamic-list-cpp-new
1172 @item --dynamic-list-cpp-new
1173 Provide the builtin dynamic list for C++ operator new and delete. It
1174 is mainly useful for building shared libstdc++.
1176 @kindex --dynamic-list-cpp-typeinfo
1177 @item --dynamic-list-cpp-typeinfo
1178 Provide the builtin dynamic list for C++ runtime type identification.
1180 @kindex --check-sections
1181 @kindex --no-check-sections
1182 @item --check-sections
1183 @itemx --no-check-sections
1184 Asks the linker @emph{not} to check section addresses after they have
1185 been assigned to see if there are any overlaps. Normally the linker will
1186 perform this check, and if it finds any overlaps it will produce
1187 suitable error messages. The linker does know about, and does make
1188 allowances for sections in overlays. The default behaviour can be
1189 restored by using the command line switch @option{--check-sections}.
1191 @cindex cross reference table
1194 Output a cross reference table. If a linker map file is being
1195 generated, the cross reference table is printed to the map file.
1196 Otherwise, it is printed on the standard output.
1198 The format of the table is intentionally simple, so that it may be
1199 easily processed by a script if necessary. The symbols are printed out,
1200 sorted by name. For each symbol, a list of file names is given. If the
1201 symbol is defined, the first file listed is the location of the
1202 definition. The remaining files contain references to the symbol.
1204 @cindex common allocation
1205 @kindex --no-define-common
1206 @item --no-define-common
1207 This option inhibits the assignment of addresses to common symbols.
1208 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1209 @xref{Miscellaneous Commands}.
1211 The @samp{--no-define-common} option allows decoupling
1212 the decision to assign addresses to Common symbols from the choice
1213 of the output file type; otherwise a non-Relocatable output type
1214 forces assigning addresses to Common symbols.
1215 Using @samp{--no-define-common} allows Common symbols that are referenced
1216 from a shared library to be assigned addresses only in the main program.
1217 This eliminates the unused duplicate space in the shared library,
1218 and also prevents any possible confusion over resolving to the wrong
1219 duplicate when there are many dynamic modules with specialized search
1220 paths for runtime symbol resolution.
1222 @cindex symbols, from command line
1223 @kindex --defsym @var{symbol}=@var{exp}
1224 @item --defsym @var{symbol}=@var{expression}
1225 Create a global symbol in the output file, containing the absolute
1226 address given by @var{expression}. You may use this option as many
1227 times as necessary to define multiple symbols in the command line. A
1228 limited form of arithmetic is supported for the @var{expression} in this
1229 context: you may give a hexadecimal constant or the name of an existing
1230 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1231 constants or symbols. If you need more elaborate expressions, consider
1232 using the linker command language from a script (@pxref{Assignments,,
1233 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1234 space between @var{symbol}, the equals sign (``@key{=}''), and
1237 @cindex demangling, from command line
1238 @kindex --demangle[=@var{style}]
1239 @kindex --no-demangle
1240 @item --demangle[=@var{style}]
1241 @itemx --no-demangle
1242 These options control whether to demangle symbol names in error messages
1243 and other output. When the linker is told to demangle, it tries to
1244 present symbol names in a readable fashion: it strips leading
1245 underscores if they are used by the object file format, and converts C++
1246 mangled symbol names into user readable names. Different compilers have
1247 different mangling styles. The optional demangling style argument can be used
1248 to choose an appropriate demangling style for your compiler. The linker will
1249 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1250 is set. These options may be used to override the default.
1252 @cindex dynamic linker, from command line
1253 @kindex -I@var{file}
1254 @kindex --dynamic-linker @var{file}
1255 @item --dynamic-linker @var{file}
1256 Set the name of the dynamic linker. This is only meaningful when
1257 generating dynamically linked ELF executables. The default dynamic
1258 linker is normally correct; don't use this unless you know what you are
1262 @kindex --fatal-warnings
1263 @item --fatal-warnings
1264 Treat all warnings as errors.
1266 @kindex --force-exe-suffix
1267 @item --force-exe-suffix
1268 Make sure that an output file has a .exe suffix.
1270 If a successfully built fully linked output file does not have a
1271 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1272 the output file to one of the same name with a @code{.exe} suffix. This
1273 option is useful when using unmodified Unix makefiles on a Microsoft
1274 Windows host, since some versions of Windows won't run an image unless
1275 it ends in a @code{.exe} suffix.
1277 @kindex --gc-sections
1278 @kindex --no-gc-sections
1279 @cindex garbage collection
1281 @itemx --no-gc-sections
1282 Enable garbage collection of unused input sections. It is ignored on
1283 targets that do not support this option. This option is not compatible
1284 with @samp{-r} or @samp{--emit-relocs}. The default behaviour (of not
1285 performing this garbage collection) can be restored by specifying
1286 @samp{--no-gc-sections} on the command line.
1288 @samp{--gc-sections} decides which input sections are used by
1289 examining symbols and relocations. The section containing the entry
1290 symbol and all sections containing symbols undefined on the
1291 command-line will be kept, as will sections containing symbols
1292 referenced by dynamic objects. Note that when building shared
1293 libraries, the linker must assume that any visible symbol is
1294 referenced. Once this initial set of sections has been determined,
1295 the linker recursively marks as used any section referenced by their
1296 relocations. See @samp{--entry} and @samp{--undefined}.
1298 @kindex --print-gc-sections
1299 @kindex --no-print-gc-sections
1300 @cindex garbage collection
1301 @item --print-gc-sections
1302 @itemx --no-print-gc-sections
1303 List all sections removed by garbage collection. The listing is
1304 printed on stderr. This option is only effective if garbage
1305 collection has been enabled via the @samp{--gc-sections}) option. The
1306 default behaviour (of not listing the sections that are removed) can
1307 be restored by specifying @samp{--no-print-gc-sections} on the command
1314 Print a summary of the command-line options on the standard output and exit.
1316 @kindex --target-help
1318 Print a summary of all target specific options on the standard output and exit.
1321 @item -Map @var{mapfile}
1322 Print a link map to the file @var{mapfile}. See the description of the
1323 @option{-M} option, above.
1325 @cindex memory usage
1326 @kindex --no-keep-memory
1327 @item --no-keep-memory
1328 @command{ld} normally optimizes for speed over memory usage by caching the
1329 symbol tables of input files in memory. This option tells @command{ld} to
1330 instead optimize for memory usage, by rereading the symbol tables as
1331 necessary. This may be required if @command{ld} runs out of memory space
1332 while linking a large executable.
1334 @kindex --no-undefined
1336 @item --no-undefined
1338 Report unresolved symbol references from regular object files. This
1339 is done even if the linker is creating a non-symbolic shared library.
1340 The switch @option{--[no-]allow-shlib-undefined} controls the
1341 behaviour for reporting unresolved references found in shared
1342 libraries being linked in.
1344 @kindex --allow-multiple-definition
1346 @item --allow-multiple-definition
1348 Normally when a symbol is defined multiple times, the linker will
1349 report a fatal error. These options allow multiple definitions and the
1350 first definition will be used.
1352 @kindex --allow-shlib-undefined
1353 @kindex --no-allow-shlib-undefined
1354 @item --allow-shlib-undefined
1355 @itemx --no-allow-shlib-undefined
1356 Allows (the default) or disallows undefined symbols in shared libraries.
1357 This switch is similar to @option{--no-undefined} except that it
1358 determines the behaviour when the undefined symbols are in a
1359 shared library rather than a regular object file. It does not affect
1360 how undefined symbols in regular object files are handled.
1362 The reason that @option{--allow-shlib-undefined} is the default is that
1363 the shared library being specified at link time may not be the same as
1364 the one that is available at load time, so the symbols might actually be
1365 resolvable at load time. Plus there are some systems, (eg BeOS) where
1366 undefined symbols in shared libraries is normal. (The kernel patches
1367 them at load time to select which function is most appropriate
1368 for the current architecture. This is used for example to dynamically
1369 select an appropriate memset function). Apparently it is also normal
1370 for HPPA shared libraries to have undefined symbols.
1372 @kindex --no-undefined-version
1373 @item --no-undefined-version
1374 Normally when a symbol has an undefined version, the linker will ignore
1375 it. This option disallows symbols with undefined version and a fatal error
1376 will be issued instead.
1378 @kindex --default-symver
1379 @item --default-symver
1380 Create and use a default symbol version (the soname) for unversioned
1383 @kindex --default-imported-symver
1384 @item --default-imported-symver
1385 Create and use a default symbol version (the soname) for unversioned
1388 @kindex --no-warn-mismatch
1389 @item --no-warn-mismatch
1390 Normally @command{ld} will give an error if you try to link together input
1391 files that are mismatched for some reason, perhaps because they have
1392 been compiled for different processors or for different endiannesses.
1393 This option tells @command{ld} that it should silently permit such possible
1394 errors. This option should only be used with care, in cases when you
1395 have taken some special action that ensures that the linker errors are
1398 @kindex --no-warn-search-mismatch
1399 @item --no-warn-search-mismatch
1400 Normally @command{ld} will give a warning if it finds an incompatible
1401 library during a library search. This option silences the warning.
1403 @kindex --no-whole-archive
1404 @item --no-whole-archive
1405 Turn off the effect of the @option{--whole-archive} option for subsequent
1408 @cindex output file after errors
1409 @kindex --noinhibit-exec
1410 @item --noinhibit-exec
1411 Retain the executable output file whenever it is still usable.
1412 Normally, the linker will not produce an output file if it encounters
1413 errors during the link process; it exits without writing an output file
1414 when it issues any error whatsoever.
1418 Only search library directories explicitly specified on the
1419 command line. Library directories specified in linker scripts
1420 (including linker scripts specified on the command line) are ignored.
1422 @ifclear SingleFormat
1424 @item --oformat @var{output-format}
1425 @command{ld} may be configured to support more than one kind of object
1426 file. If your @command{ld} is configured this way, you can use the
1427 @samp{--oformat} option to specify the binary format for the output
1428 object file. Even when @command{ld} is configured to support alternative
1429 object formats, you don't usually need to specify this, as @command{ld}
1430 should be configured to produce as a default output format the most
1431 usual format on each machine. @var{output-format} is a text string, the
1432 name of a particular format supported by the BFD libraries. (You can
1433 list the available binary formats with @samp{objdump -i}.) The script
1434 command @code{OUTPUT_FORMAT} can also specify the output format, but
1435 this option overrides it. @xref{BFD}.
1439 @kindex --pic-executable
1441 @itemx --pic-executable
1442 @cindex position independent executables
1443 Create a position independent executable. This is currently only supported on
1444 ELF platforms. Position independent executables are similar to shared
1445 libraries in that they are relocated by the dynamic linker to the virtual
1446 address the OS chooses for them (which can vary between invocations). Like
1447 normal dynamically linked executables they can be executed and symbols
1448 defined in the executable cannot be overridden by shared libraries.
1452 This option is ignored for Linux compatibility.
1456 This option is ignored for SVR4 compatibility.
1459 @cindex synthesizing linker
1460 @cindex relaxing addressing modes
1462 An option with machine dependent effects.
1464 This option is only supported on a few targets.
1467 @xref{H8/300,,@command{ld} and the H8/300}.
1470 @xref{i960,, @command{ld} and the Intel 960 family}.
1473 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1476 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1479 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1482 On some platforms, the @samp{--relax} option performs global
1483 optimizations that become possible when the linker resolves addressing
1484 in the program, such as relaxing address modes and synthesizing new
1485 instructions in the output object file.
1487 On some platforms these link time global optimizations may make symbolic
1488 debugging of the resulting executable impossible.
1491 the case for the Matsushita MN10200 and MN10300 family of processors.
1495 On platforms where this is not supported, @samp{--relax} is accepted,
1499 @cindex retaining specified symbols
1500 @cindex stripping all but some symbols
1501 @cindex symbols, retaining selectively
1502 @item --retain-symbols-file @var{filename}
1503 Retain @emph{only} the symbols listed in the file @var{filename},
1504 discarding all others. @var{filename} is simply a flat file, with one
1505 symbol name per line. This option is especially useful in environments
1509 where a large global symbol table is accumulated gradually, to conserve
1512 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1513 or symbols needed for relocations.
1515 You may only specify @samp{--retain-symbols-file} once in the command
1516 line. It overrides @samp{-s} and @samp{-S}.
1519 @item -rpath @var{dir}
1520 @cindex runtime library search path
1522 Add a directory to the runtime library search path. This is used when
1523 linking an ELF executable with shared objects. All @option{-rpath}
1524 arguments are concatenated and passed to the runtime linker, which uses
1525 them to locate shared objects at runtime. The @option{-rpath} option is
1526 also used when locating shared objects which are needed by shared
1527 objects explicitly included in the link; see the description of the
1528 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1529 ELF executable, the contents of the environment variable
1530 @code{LD_RUN_PATH} will be used if it is defined.
1532 The @option{-rpath} option may also be used on SunOS. By default, on
1533 SunOS, the linker will form a runtime search patch out of all the
1534 @option{-L} options it is given. If a @option{-rpath} option is used, the
1535 runtime search path will be formed exclusively using the @option{-rpath}
1536 options, ignoring the @option{-L} options. This can be useful when using
1537 gcc, which adds many @option{-L} options which may be on NFS mounted
1540 For compatibility with other ELF linkers, if the @option{-R} option is
1541 followed by a directory name, rather than a file name, it is treated as
1542 the @option{-rpath} option.
1546 @cindex link-time runtime library search path
1548 @item -rpath-link @var{DIR}
1549 When using ELF or SunOS, one shared library may require another. This
1550 happens when an @code{ld -shared} link includes a shared library as one
1553 When the linker encounters such a dependency when doing a non-shared,
1554 non-relocatable link, it will automatically try to locate the required
1555 shared library and include it in the link, if it is not included
1556 explicitly. In such a case, the @option{-rpath-link} option
1557 specifies the first set of directories to search. The
1558 @option{-rpath-link} option may specify a sequence of directory names
1559 either by specifying a list of names separated by colons, or by
1560 appearing multiple times.
1562 This option should be used with caution as it overrides the search path
1563 that may have been hard compiled into a shared library. In such a case it
1564 is possible to use unintentionally a different search path than the
1565 runtime linker would do.
1567 The linker uses the following search paths to locate required shared
1571 Any directories specified by @option{-rpath-link} options.
1573 Any directories specified by @option{-rpath} options. The difference
1574 between @option{-rpath} and @option{-rpath-link} is that directories
1575 specified by @option{-rpath} options are included in the executable and
1576 used at runtime, whereas the @option{-rpath-link} option is only effective
1577 at link time. Searching @option{-rpath} in this way is only supported
1578 by native linkers and cross linkers which have been configured with
1579 the @option{--with-sysroot} option.
1581 On an ELF system, for native linkers, if the @option{-rpath} and
1582 @option{-rpath-link} options were not used, search the contents of the
1583 environment variable @code{LD_RUN_PATH}.
1585 On SunOS, if the @option{-rpath} option was not used, search any
1586 directories specified using @option{-L} options.
1588 For a native linker, the search the contents of the environment
1589 variable @code{LD_LIBRARY_PATH}.
1591 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1592 @code{DT_RPATH} of a shared library are searched for shared
1593 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1594 @code{DT_RUNPATH} entries exist.
1596 The default directories, normally @file{/lib} and @file{/usr/lib}.
1598 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1599 exists, the list of directories found in that file.
1602 If the required shared library is not found, the linker will issue a
1603 warning and continue with the link.
1610 @cindex shared libraries
1611 Create a shared library. This is currently only supported on ELF, XCOFF
1612 and SunOS platforms. On SunOS, the linker will automatically create a
1613 shared library if the @option{-e} option is not used and there are
1614 undefined symbols in the link.
1617 @kindex --sort-common
1618 This option tells @command{ld} to sort the common symbols by size when it
1619 places them in the appropriate output sections. First come all the one
1620 byte symbols, then all the two byte, then all the four byte, and then
1621 everything else. This is to prevent gaps between symbols due to
1622 alignment constraints.
1624 @kindex --sort-section name
1625 @item --sort-section name
1626 This option will apply @code{SORT_BY_NAME} to all wildcard section
1627 patterns in the linker script.
1629 @kindex --sort-section alignment
1630 @item --sort-section alignment
1631 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1632 patterns in the linker script.
1634 @kindex --split-by-file
1635 @item --split-by-file [@var{size}]
1636 Similar to @option{--split-by-reloc} but creates a new output section for
1637 each input file when @var{size} is reached. @var{size} defaults to a
1638 size of 1 if not given.
1640 @kindex --split-by-reloc
1641 @item --split-by-reloc [@var{count}]
1642 Tries to creates extra sections in the output file so that no single
1643 output section in the file contains more than @var{count} relocations.
1644 This is useful when generating huge relocatable files for downloading into
1645 certain real time kernels with the COFF object file format; since COFF
1646 cannot represent more than 65535 relocations in a single section. Note
1647 that this will fail to work with object file formats which do not
1648 support arbitrary sections. The linker will not split up individual
1649 input sections for redistribution, so if a single input section contains
1650 more than @var{count} relocations one output section will contain that
1651 many relocations. @var{count} defaults to a value of 32768.
1655 Compute and display statistics about the operation of the linker, such
1656 as execution time and memory usage.
1659 @item --sysroot=@var{directory}
1660 Use @var{directory} as the location of the sysroot, overriding the
1661 configure-time default. This option is only supported by linkers
1662 that were configured using @option{--with-sysroot}.
1664 @kindex --traditional-format
1665 @cindex traditional format
1666 @item --traditional-format
1667 For some targets, the output of @command{ld} is different in some ways from
1668 the output of some existing linker. This switch requests @command{ld} to
1669 use the traditional format instead.
1672 For example, on SunOS, @command{ld} combines duplicate entries in the
1673 symbol string table. This can reduce the size of an output file with
1674 full debugging information by over 30 percent. Unfortunately, the SunOS
1675 @code{dbx} program can not read the resulting program (@code{gdb} has no
1676 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1677 combine duplicate entries.
1679 @kindex --section-start @var{sectionname}=@var{org}
1680 @item --section-start @var{sectionname}=@var{org}
1681 Locate a section in the output file at the absolute
1682 address given by @var{org}. You may use this option as many
1683 times as necessary to locate multiple sections in the command
1685 @var{org} must be a single hexadecimal integer;
1686 for compatibility with other linkers, you may omit the leading
1687 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1688 should be no white space between @var{sectionname}, the equals
1689 sign (``@key{=}''), and @var{org}.
1691 @kindex -Tbss @var{org}
1692 @kindex -Tdata @var{org}
1693 @kindex -Ttext @var{org}
1694 @cindex segment origins, cmd line
1695 @item -Tbss @var{org}
1696 @itemx -Tdata @var{org}
1697 @itemx -Ttext @var{org}
1698 Same as --section-start, with @code{.bss}, @code{.data} or
1699 @code{.text} as the @var{sectionname}.
1701 @kindex --unresolved-symbols
1702 @item --unresolved-symbols=@var{method}
1703 Determine how to handle unresolved symbols. There are four possible
1704 values for @samp{method}:
1708 Do not report any unresolved symbols.
1711 Report all unresolved symbols. This is the default.
1713 @item ignore-in-object-files
1714 Report unresolved symbols that are contained in shared libraries, but
1715 ignore them if they come from regular object files.
1717 @item ignore-in-shared-libs
1718 Report unresolved symbols that come from regular object files, but
1719 ignore them if they come from shared libraries. This can be useful
1720 when creating a dynamic binary and it is known that all the shared
1721 libraries that it should be referencing are included on the linker's
1725 The behaviour for shared libraries on their own can also be controlled
1726 by the @option{--[no-]allow-shlib-undefined} option.
1728 Normally the linker will generate an error message for each reported
1729 unresolved symbol but the option @option{--warn-unresolved-symbols}
1730 can change this to a warning.
1736 Display the version number for @command{ld} and list the linker emulations
1737 supported. Display which input files can and cannot be opened. Display
1738 the linker script being used by the linker.
1740 @kindex --version-script=@var{version-scriptfile}
1741 @cindex version script, symbol versions
1742 @itemx --version-script=@var{version-scriptfile}
1743 Specify the name of a version script to the linker. This is typically
1744 used when creating shared libraries to specify additional information
1745 about the version hierarchy for the library being created. This option
1746 is only meaningful on ELF platforms which support shared libraries.
1749 @kindex --warn-common
1750 @cindex warnings, on combining symbols
1751 @cindex combining symbols, warnings on
1753 Warn when a common symbol is combined with another common symbol or with
1754 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1755 but linkers on some other operating systems do not. This option allows
1756 you to find potential problems from combining global symbols.
1757 Unfortunately, some C libraries use this practise, so you may get some
1758 warnings about symbols in the libraries as well as in your programs.
1760 There are three kinds of global symbols, illustrated here by C examples:
1764 A definition, which goes in the initialized data section of the output
1768 An undefined reference, which does not allocate space.
1769 There must be either a definition or a common symbol for the
1773 A common symbol. If there are only (one or more) common symbols for a
1774 variable, it goes in the uninitialized data area of the output file.
1775 The linker merges multiple common symbols for the same variable into a
1776 single symbol. If they are of different sizes, it picks the largest
1777 size. The linker turns a common symbol into a declaration, if there is
1778 a definition of the same variable.
1781 The @samp{--warn-common} option can produce five kinds of warnings.
1782 Each warning consists of a pair of lines: the first describes the symbol
1783 just encountered, and the second describes the previous symbol
1784 encountered with the same name. One or both of the two symbols will be
1789 Turning a common symbol into a reference, because there is already a
1790 definition for the symbol.
1792 @var{file}(@var{section}): warning: common of `@var{symbol}'
1793 overridden by definition
1794 @var{file}(@var{section}): warning: defined here
1798 Turning a common symbol into a reference, because a later definition for
1799 the symbol is encountered. This is the same as the previous case,
1800 except that the symbols are encountered in a different order.
1802 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1804 @var{file}(@var{section}): warning: common is here
1808 Merging a common symbol with a previous same-sized common symbol.
1810 @var{file}(@var{section}): warning: multiple common
1812 @var{file}(@var{section}): warning: previous common is here
1816 Merging a common symbol with a previous larger common symbol.
1818 @var{file}(@var{section}): warning: common of `@var{symbol}'
1819 overridden by larger common
1820 @var{file}(@var{section}): warning: larger common is here
1824 Merging a common symbol with a previous smaller common symbol. This is
1825 the same as the previous case, except that the symbols are
1826 encountered in a different order.
1828 @var{file}(@var{section}): warning: common of `@var{symbol}'
1829 overriding smaller common
1830 @var{file}(@var{section}): warning: smaller common is here
1834 @kindex --warn-constructors
1835 @item --warn-constructors
1836 Warn if any global constructors are used. This is only useful for a few
1837 object file formats. For formats like COFF or ELF, the linker can not
1838 detect the use of global constructors.
1840 @kindex --warn-multiple-gp
1841 @item --warn-multiple-gp
1842 Warn if multiple global pointer values are required in the output file.
1843 This is only meaningful for certain processors, such as the Alpha.
1844 Specifically, some processors put large-valued constants in a special
1845 section. A special register (the global pointer) points into the middle
1846 of this section, so that constants can be loaded efficiently via a
1847 base-register relative addressing mode. Since the offset in
1848 base-register relative mode is fixed and relatively small (e.g., 16
1849 bits), this limits the maximum size of the constant pool. Thus, in
1850 large programs, it is often necessary to use multiple global pointer
1851 values in order to be able to address all possible constants. This
1852 option causes a warning to be issued whenever this case occurs.
1855 @cindex warnings, on undefined symbols
1856 @cindex undefined symbols, warnings on
1858 Only warn once for each undefined symbol, rather than once per module
1861 @kindex --warn-section-align
1862 @cindex warnings, on section alignment
1863 @cindex section alignment, warnings on
1864 @item --warn-section-align
1865 Warn if the address of an output section is changed because of
1866 alignment. Typically, the alignment will be set by an input section.
1867 The address will only be changed if it not explicitly specified; that
1868 is, if the @code{SECTIONS} command does not specify a start address for
1869 the section (@pxref{SECTIONS}).
1871 @kindex --warn-shared-textrel
1872 @item --warn-shared-textrel
1873 Warn if the linker adds a DT_TEXTREL to a shared object.
1875 @kindex --warn-unresolved-symbols
1876 @item --warn-unresolved-symbols
1877 If the linker is going to report an unresolved symbol (see the option
1878 @option{--unresolved-symbols}) it will normally generate an error.
1879 This option makes it generate a warning instead.
1881 @kindex --error-unresolved-symbols
1882 @item --error-unresolved-symbols
1883 This restores the linker's default behaviour of generating errors when
1884 it is reporting unresolved symbols.
1886 @kindex --whole-archive
1887 @cindex including an entire archive
1888 @item --whole-archive
1889 For each archive mentioned on the command line after the
1890 @option{--whole-archive} option, include every object file in the archive
1891 in the link, rather than searching the archive for the required object
1892 files. This is normally used to turn an archive file into a shared
1893 library, forcing every object to be included in the resulting shared
1894 library. This option may be used more than once.
1896 Two notes when using this option from gcc: First, gcc doesn't know
1897 about this option, so you have to use @option{-Wl,-whole-archive}.
1898 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1899 list of archives, because gcc will add its own list of archives to
1900 your link and you may not want this flag to affect those as well.
1903 @item --wrap @var{symbol}
1904 Use a wrapper function for @var{symbol}. Any undefined reference to
1905 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1906 undefined reference to @code{__real_@var{symbol}} will be resolved to
1909 This can be used to provide a wrapper for a system function. The
1910 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1911 wishes to call the system function, it should call
1912 @code{__real_@var{symbol}}.
1914 Here is a trivial example:
1918 __wrap_malloc (size_t c)
1920 printf ("malloc called with %zu\n", c);
1921 return __real_malloc (c);
1925 If you link other code with this file using @option{--wrap malloc}, then
1926 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1927 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1928 call the real @code{malloc} function.
1930 You may wish to provide a @code{__real_malloc} function as well, so that
1931 links without the @option{--wrap} option will succeed. If you do this,
1932 you should not put the definition of @code{__real_malloc} in the same
1933 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
1934 call before the linker has a chance to wrap it to @code{malloc}.
1936 @kindex --eh-frame-hdr
1937 @item --eh-frame-hdr
1938 Request creation of @code{.eh_frame_hdr} section and ELF
1939 @code{PT_GNU_EH_FRAME} segment header.
1941 @kindex --enable-new-dtags
1942 @kindex --disable-new-dtags
1943 @item --enable-new-dtags
1944 @itemx --disable-new-dtags
1945 This linker can create the new dynamic tags in ELF. But the older ELF
1946 systems may not understand them. If you specify
1947 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
1948 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
1949 created. By default, the new dynamic tags are not created. Note that
1950 those options are only available for ELF systems.
1952 @kindex --hash-size=@var{number}
1953 @item --hash-size=@var{number}
1954 Set the default size of the linker's hash tables to a prime number
1955 close to @var{number}. Increasing this value can reduce the length of
1956 time it takes the linker to perform its tasks, at the expense of
1957 increasing the linker's memory requirements. Similarly reducing this
1958 value can reduce the memory requirements at the expense of speed.
1960 @kindex --hash-style=@var{style}
1961 @item --hash-style=@var{style}
1962 Set the type of linker's hash table(s). @var{style} can be either
1963 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
1964 new style GNU @code{.gnu.hash} section or @code{both} for both
1965 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
1966 hash tables. The default is @code{sysv}.
1968 @kindex --reduce-memory-overheads
1969 @item --reduce-memory-overheads
1970 This option reduces memory requirements at ld runtime, at the expense of
1971 linking speed. This was introduced to select the old O(n^2) algorithm
1972 for link map file generation, rather than the new O(n) algorithm which uses
1973 about 40% more memory for symbol storage.
1975 Another effect of the switch is to set the default hash table size to
1976 1021, which again saves memory at the cost of lengthening the linker's
1977 run time. This is not done however if the @option{--hash-size} switch
1980 The @option{--reduce-memory-overheads} switch may be also be used to
1981 enable other tradeoffs in future versions of the linker.
1984 @kindex --build-id=@var{style}
1986 @itemx --build-id=@var{style}
1987 Request creation of @code{.note.gnu.build-id} ELF note section.
1988 The contents of the note are unique bits identifying this linked
1989 file. @var{style} can be @code{uuid} to use 128 random bits,
1990 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
1991 parts of the output contents, @code{md5} to use a 128-bit
1992 @sc{MD5} hash on the normative parts of the output contents, or
1993 @code{0x@var{hexstring}} to use a chosen bit string specified as
1994 an even number of hexadecimal digits (@code{-} and @code{:}
1995 characters between digit pairs are ignored). If @var{style} is
1996 omitted, @code{sha1} is used.
1998 The @code{md5} and @code{sha1} styles produces an identifier
1999 that is always the same in an identical output file, but will be
2000 unique among all nonidentical output files. It is not intended
2001 to be compared as a checksum for the file's contents. A linked
2002 file may be changed later by other tools, but the build ID bit
2003 string identifying the original linked file does not change.
2005 Passing @code{none} for @var{style} disables the setting from any
2006 @code{--build-id} options earlier on the command line.
2011 @subsection Options Specific to i386 PE Targets
2013 @c man begin OPTIONS
2015 The i386 PE linker supports the @option{-shared} option, which causes
2016 the output to be a dynamically linked library (DLL) instead of a
2017 normal executable. You should name the output @code{*.dll} when you
2018 use this option. In addition, the linker fully supports the standard
2019 @code{*.def} files, which may be specified on the linker command line
2020 like an object file (in fact, it should precede archives it exports
2021 symbols from, to ensure that they get linked in, just like a normal
2024 In addition to the options common to all targets, the i386 PE linker
2025 support additional command line options that are specific to the i386
2026 PE target. Options that take values may be separated from their
2027 values by either a space or an equals sign.
2031 @kindex --add-stdcall-alias
2032 @item --add-stdcall-alias
2033 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2034 as-is and also with the suffix stripped.
2035 [This option is specific to the i386 PE targeted port of the linker]
2038 @item --base-file @var{file}
2039 Use @var{file} as the name of a file in which to save the base
2040 addresses of all the relocations needed for generating DLLs with
2042 [This is an i386 PE specific option]
2046 Create a DLL instead of a regular executable. You may also use
2047 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2049 [This option is specific to the i386 PE targeted port of the linker]
2051 @kindex --enable-stdcall-fixup
2052 @kindex --disable-stdcall-fixup
2053 @item --enable-stdcall-fixup
2054 @itemx --disable-stdcall-fixup
2055 If the link finds a symbol that it cannot resolve, it will attempt to
2056 do ``fuzzy linking'' by looking for another defined symbol that differs
2057 only in the format of the symbol name (cdecl vs stdcall) and will
2058 resolve that symbol by linking to the match. For example, the
2059 undefined symbol @code{_foo} might be linked to the function
2060 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2061 to the function @code{_bar}. When the linker does this, it prints a
2062 warning, since it normally should have failed to link, but sometimes
2063 import libraries generated from third-party dlls may need this feature
2064 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2065 feature is fully enabled and warnings are not printed. If you specify
2066 @option{--disable-stdcall-fixup}, this feature is disabled and such
2067 mismatches are considered to be errors.
2068 [This option is specific to the i386 PE targeted port of the linker]
2070 @cindex DLLs, creating
2071 @kindex --export-all-symbols
2072 @item --export-all-symbols
2073 If given, all global symbols in the objects used to build a DLL will
2074 be exported by the DLL. Note that this is the default if there
2075 otherwise wouldn't be any exported symbols. When symbols are
2076 explicitly exported via DEF files or implicitly exported via function
2077 attributes, the default is to not export anything else unless this
2078 option is given. Note that the symbols @code{DllMain@@12},
2079 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2080 @code{impure_ptr} will not be automatically
2081 exported. Also, symbols imported from other DLLs will not be
2082 re-exported, nor will symbols specifying the DLL's internal layout
2083 such as those beginning with @code{_head_} or ending with
2084 @code{_iname}. In addition, no symbols from @code{libgcc},
2085 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2086 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2087 not be exported, to help with C++ DLLs. Finally, there is an
2088 extensive list of cygwin-private symbols that are not exported
2089 (obviously, this applies on when building DLLs for cygwin targets).
2090 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2091 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2092 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2093 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2094 @code{cygwin_premain3}, and @code{environ}.
2095 [This option is specific to the i386 PE targeted port of the linker]
2097 @kindex --exclude-symbols
2098 @item --exclude-symbols @var{symbol},@var{symbol},...
2099 Specifies a list of symbols which should not be automatically
2100 exported. The symbol names may be delimited by commas or colons.
2101 [This option is specific to the i386 PE targeted port of the linker]
2103 @kindex --file-alignment
2104 @item --file-alignment
2105 Specify the file alignment. Sections in the file will always begin at
2106 file offsets which are multiples of this number. This defaults to
2108 [This option is specific to the i386 PE targeted port of the linker]
2112 @item --heap @var{reserve}
2113 @itemx --heap @var{reserve},@var{commit}
2114 Specify the number of bytes of memory to reserve (and optionally commit)
2115 to be used as heap for this program. The default is 1Mb reserved, 4K
2117 [This option is specific to the i386 PE targeted port of the linker]
2120 @kindex --image-base
2121 @item --image-base @var{value}
2122 Use @var{value} as the base address of your program or dll. This is
2123 the lowest memory location that will be used when your program or dll
2124 is loaded. To reduce the need to relocate and improve performance of
2125 your dlls, each should have a unique base address and not overlap any
2126 other dlls. The default is 0x400000 for executables, and 0x10000000
2128 [This option is specific to the i386 PE targeted port of the linker]
2132 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2133 symbols before they are exported.
2134 [This option is specific to the i386 PE targeted port of the linker]
2136 @kindex --large-address-aware
2137 @item --large-address-aware
2138 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2139 header is set to indicate that this executable supports virtual addresses
2140 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2141 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2142 section of the BOOT.INI. Otherwise, this bit has no effect.
2143 [This option is specific to PE targeted ports of the linker]
2145 @kindex --major-image-version
2146 @item --major-image-version @var{value}
2147 Sets the major number of the ``image version''. Defaults to 1.
2148 [This option is specific to the i386 PE targeted port of the linker]
2150 @kindex --major-os-version
2151 @item --major-os-version @var{value}
2152 Sets the major number of the ``os version''. Defaults to 4.
2153 [This option is specific to the i386 PE targeted port of the linker]
2155 @kindex --major-subsystem-version
2156 @item --major-subsystem-version @var{value}
2157 Sets the major number of the ``subsystem version''. Defaults to 4.
2158 [This option is specific to the i386 PE targeted port of the linker]
2160 @kindex --minor-image-version
2161 @item --minor-image-version @var{value}
2162 Sets the minor number of the ``image version''. Defaults to 0.
2163 [This option is specific to the i386 PE targeted port of the linker]
2165 @kindex --minor-os-version
2166 @item --minor-os-version @var{value}
2167 Sets the minor number of the ``os version''. Defaults to 0.
2168 [This option is specific to the i386 PE targeted port of the linker]
2170 @kindex --minor-subsystem-version
2171 @item --minor-subsystem-version @var{value}
2172 Sets the minor number of the ``subsystem version''. Defaults to 0.
2173 [This option is specific to the i386 PE targeted port of the linker]
2175 @cindex DEF files, creating
2176 @cindex DLLs, creating
2177 @kindex --output-def
2178 @item --output-def @var{file}
2179 The linker will create the file @var{file} which will contain a DEF
2180 file corresponding to the DLL the linker is generating. This DEF file
2181 (which should be called @code{*.def}) may be used to create an import
2182 library with @code{dlltool} or may be used as a reference to
2183 automatically or implicitly exported symbols.
2184 [This option is specific to the i386 PE targeted port of the linker]
2186 @cindex DLLs, creating
2187 @kindex --out-implib
2188 @item --out-implib @var{file}
2189 The linker will create the file @var{file} which will contain an
2190 import lib corresponding to the DLL the linker is generating. This
2191 import lib (which should be called @code{*.dll.a} or @code{*.a}
2192 may be used to link clients against the generated DLL; this behaviour
2193 makes it possible to skip a separate @code{dlltool} import library
2195 [This option is specific to the i386 PE targeted port of the linker]
2197 @kindex --enable-auto-image-base
2198 @item --enable-auto-image-base
2199 Automatically choose the image base for DLLs, unless one is specified
2200 using the @code{--image-base} argument. By using a hash generated
2201 from the dllname to create unique image bases for each DLL, in-memory
2202 collisions and relocations which can delay program execution are
2204 [This option is specific to the i386 PE targeted port of the linker]
2206 @kindex --disable-auto-image-base
2207 @item --disable-auto-image-base
2208 Do not automatically generate a unique image base. If there is no
2209 user-specified image base (@code{--image-base}) then use the platform
2211 [This option is specific to the i386 PE targeted port of the linker]
2213 @cindex DLLs, linking to
2214 @kindex --dll-search-prefix
2215 @item --dll-search-prefix @var{string}
2216 When linking dynamically to a dll without an import library,
2217 search for @code{<string><basename>.dll} in preference to
2218 @code{lib<basename>.dll}. This behaviour allows easy distinction
2219 between DLLs built for the various "subplatforms": native, cygwin,
2220 uwin, pw, etc. For instance, cygwin DLLs typically use
2221 @code{--dll-search-prefix=cyg}.
2222 [This option is specific to the i386 PE targeted port of the linker]
2224 @kindex --enable-auto-import
2225 @item --enable-auto-import
2226 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2227 DATA imports from DLLs, and create the necessary thunking symbols when
2228 building the import libraries with those DATA exports. Note: Use of the
2229 'auto-import' extension will cause the text section of the image file
2230 to be made writable. This does not conform to the PE-COFF format
2231 specification published by Microsoft.
2233 Note - use of the 'auto-import' extension will also cause read only
2234 data which would normally be placed into the .rdata section to be
2235 placed into the .data section instead. This is in order to work
2236 around a problem with consts that is described here:
2237 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2239 Using 'auto-import' generally will 'just work' -- but sometimes you may
2242 "variable '<var>' can't be auto-imported. Please read the
2243 documentation for ld's @code{--enable-auto-import} for details."
2245 This message occurs when some (sub)expression accesses an address
2246 ultimately given by the sum of two constants (Win32 import tables only
2247 allow one). Instances where this may occur include accesses to member
2248 fields of struct variables imported from a DLL, as well as using a
2249 constant index into an array variable imported from a DLL. Any
2250 multiword variable (arrays, structs, long long, etc) may trigger
2251 this error condition. However, regardless of the exact data type
2252 of the offending exported variable, ld will always detect it, issue
2253 the warning, and exit.
2255 There are several ways to address this difficulty, regardless of the
2256 data type of the exported variable:
2258 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2259 of adjusting references in your client code for runtime environment, so
2260 this method works only when runtime environment supports this feature.
2262 A second solution is to force one of the 'constants' to be a variable --
2263 that is, unknown and un-optimizable at compile time. For arrays,
2264 there are two possibilities: a) make the indexee (the array's address)
2265 a variable, or b) make the 'constant' index a variable. Thus:
2268 extern type extern_array[];
2270 @{ volatile type *t=extern_array; t[1] @}
2276 extern type extern_array[];
2278 @{ volatile int t=1; extern_array[t] @}
2281 For structs (and most other multiword data types) the only option
2282 is to make the struct itself (or the long long, or the ...) variable:
2285 extern struct s extern_struct;
2286 extern_struct.field -->
2287 @{ volatile struct s *t=&extern_struct; t->field @}
2293 extern long long extern_ll;
2295 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2298 A third method of dealing with this difficulty is to abandon
2299 'auto-import' for the offending symbol and mark it with
2300 @code{__declspec(dllimport)}. However, in practise that
2301 requires using compile-time #defines to indicate whether you are
2302 building a DLL, building client code that will link to the DLL, or
2303 merely building/linking to a static library. In making the choice
2304 between the various methods of resolving the 'direct address with
2305 constant offset' problem, you should consider typical real-world usage:
2313 void main(int argc, char **argv)@{
2314 printf("%d\n",arr[1]);
2324 void main(int argc, char **argv)@{
2325 /* This workaround is for win32 and cygwin; do not "optimize" */
2326 volatile int *parr = arr;
2327 printf("%d\n",parr[1]);
2334 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2335 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2336 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2337 #define FOO_IMPORT __declspec(dllimport)
2341 extern FOO_IMPORT int arr[];
2344 void main(int argc, char **argv)@{
2345 printf("%d\n",arr[1]);
2349 A fourth way to avoid this problem is to re-code your
2350 library to use a functional interface rather than a data interface
2351 for the offending variables (e.g. set_foo() and get_foo() accessor
2353 [This option is specific to the i386 PE targeted port of the linker]
2355 @kindex --disable-auto-import
2356 @item --disable-auto-import
2357 Do not attempt to do sophisticated linking of @code{_symbol} to
2358 @code{__imp__symbol} for DATA imports from DLLs.
2359 [This option is specific to the i386 PE targeted port of the linker]
2361 @kindex --enable-runtime-pseudo-reloc
2362 @item --enable-runtime-pseudo-reloc
2363 If your code contains expressions described in --enable-auto-import section,
2364 that is, DATA imports from DLL with non-zero offset, this switch will create
2365 a vector of 'runtime pseudo relocations' which can be used by runtime
2366 environment to adjust references to such data in your client code.
2367 [This option is specific to the i386 PE targeted port of the linker]
2369 @kindex --disable-runtime-pseudo-reloc
2370 @item --disable-runtime-pseudo-reloc
2371 Do not create pseudo relocations for non-zero offset DATA imports from
2372 DLLs. This is the default.
2373 [This option is specific to the i386 PE targeted port of the linker]
2375 @kindex --enable-extra-pe-debug
2376 @item --enable-extra-pe-debug
2377 Show additional debug info related to auto-import symbol thunking.
2378 [This option is specific to the i386 PE targeted port of the linker]
2380 @kindex --section-alignment
2381 @item --section-alignment
2382 Sets the section alignment. Sections in memory will always begin at
2383 addresses which are a multiple of this number. Defaults to 0x1000.
2384 [This option is specific to the i386 PE targeted port of the linker]
2388 @item --stack @var{reserve}
2389 @itemx --stack @var{reserve},@var{commit}
2390 Specify the number of bytes of memory to reserve (and optionally commit)
2391 to be used as stack for this program. The default is 2Mb reserved, 4K
2393 [This option is specific to the i386 PE targeted port of the linker]
2396 @item --subsystem @var{which}
2397 @itemx --subsystem @var{which}:@var{major}
2398 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2399 Specifies the subsystem under which your program will execute. The
2400 legal values for @var{which} are @code{native}, @code{windows},
2401 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2402 the subsystem version also. Numeric values are also accepted for
2404 [This option is specific to the i386 PE targeted port of the linker]
2411 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2413 @c man begin OPTIONS
2415 The 68HC11 and 68HC12 linkers support specific options to control the
2416 memory bank switching mapping and trampoline code generation.
2420 @kindex --no-trampoline
2421 @item --no-trampoline
2422 This option disables the generation of trampoline. By default a trampoline
2423 is generated for each far function which is called using a @code{jsr}
2424 instruction (this happens when a pointer to a far function is taken).
2426 @kindex --bank-window
2427 @item --bank-window @var{name}
2428 This option indicates to the linker the name of the memory region in
2429 the @samp{MEMORY} specification that describes the memory bank window.
2430 The definition of such region is then used by the linker to compute
2431 paging and addresses within the memory window.
2440 @section Environment Variables
2442 @c man begin ENVIRONMENT
2444 You can change the behaviour of @command{ld} with the environment variables
2445 @ifclear SingleFormat
2448 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2450 @ifclear SingleFormat
2452 @cindex default input format
2453 @code{GNUTARGET} determines the input-file object format if you don't
2454 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2455 of the BFD names for an input format (@pxref{BFD}). If there is no
2456 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2457 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2458 attempts to discover the input format by examining binary input files;
2459 this method often succeeds, but there are potential ambiguities, since
2460 there is no method of ensuring that the magic number used to specify
2461 object-file formats is unique. However, the configuration procedure for
2462 BFD on each system places the conventional format for that system first
2463 in the search-list, so ambiguities are resolved in favor of convention.
2467 @cindex default emulation
2468 @cindex emulation, default
2469 @code{LDEMULATION} determines the default emulation if you don't use the
2470 @samp{-m} option. The emulation can affect various aspects of linker
2471 behaviour, particularly the default linker script. You can list the
2472 available emulations with the @samp{--verbose} or @samp{-V} options. If
2473 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2474 variable is not defined, the default emulation depends upon how the
2475 linker was configured.
2477 @kindex COLLECT_NO_DEMANGLE
2478 @cindex demangling, default
2479 Normally, the linker will default to demangling symbols. However, if
2480 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2481 default to not demangling symbols. This environment variable is used in
2482 a similar fashion by the @code{gcc} linker wrapper program. The default
2483 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2490 @chapter Linker Scripts
2493 @cindex linker scripts
2494 @cindex command files
2495 Every link is controlled by a @dfn{linker script}. This script is
2496 written in the linker command language.
2498 The main purpose of the linker script is to describe how the sections in
2499 the input files should be mapped into the output file, and to control
2500 the memory layout of the output file. Most linker scripts do nothing
2501 more than this. However, when necessary, the linker script can also
2502 direct the linker to perform many other operations, using the commands
2505 The linker always uses a linker script. If you do not supply one
2506 yourself, the linker will use a default script that is compiled into the
2507 linker executable. You can use the @samp{--verbose} command line option
2508 to display the default linker script. Certain command line options,
2509 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2511 You may supply your own linker script by using the @samp{-T} command
2512 line option. When you do this, your linker script will replace the
2513 default linker script.
2515 You may also use linker scripts implicitly by naming them as input files
2516 to the linker, as though they were files to be linked. @xref{Implicit
2520 * Basic Script Concepts:: Basic Linker Script Concepts
2521 * Script Format:: Linker Script Format
2522 * Simple Example:: Simple Linker Script Example
2523 * Simple Commands:: Simple Linker Script Commands
2524 * Assignments:: Assigning Values to Symbols
2525 * SECTIONS:: SECTIONS Command
2526 * MEMORY:: MEMORY Command
2527 * PHDRS:: PHDRS Command
2528 * VERSION:: VERSION Command
2529 * Expressions:: Expressions in Linker Scripts
2530 * Implicit Linker Scripts:: Implicit Linker Scripts
2533 @node Basic Script Concepts
2534 @section Basic Linker Script Concepts
2535 @cindex linker script concepts
2536 We need to define some basic concepts and vocabulary in order to
2537 describe the linker script language.
2539 The linker combines input files into a single output file. The output
2540 file and each input file are in a special data format known as an
2541 @dfn{object file format}. Each file is called an @dfn{object file}.
2542 The output file is often called an @dfn{executable}, but for our
2543 purposes we will also call it an object file. Each object file has,
2544 among other things, a list of @dfn{sections}. We sometimes refer to a
2545 section in an input file as an @dfn{input section}; similarly, a section
2546 in the output file is an @dfn{output section}.
2548 Each section in an object file has a name and a size. Most sections
2549 also have an associated block of data, known as the @dfn{section
2550 contents}. A section may be marked as @dfn{loadable}, which mean that
2551 the contents should be loaded into memory when the output file is run.
2552 A section with no contents may be @dfn{allocatable}, which means that an
2553 area in memory should be set aside, but nothing in particular should be
2554 loaded there (in some cases this memory must be zeroed out). A section
2555 which is neither loadable nor allocatable typically contains some sort
2556 of debugging information.
2558 Every loadable or allocatable output section has two addresses. The
2559 first is the @dfn{VMA}, or virtual memory address. This is the address
2560 the section will have when the output file is run. The second is the
2561 @dfn{LMA}, or load memory address. This is the address at which the
2562 section will be loaded. In most cases the two addresses will be the
2563 same. An example of when they might be different is when a data section
2564 is loaded into ROM, and then copied into RAM when the program starts up
2565 (this technique is often used to initialize global variables in a ROM
2566 based system). In this case the ROM address would be the LMA, and the
2567 RAM address would be the VMA.
2569 You can see the sections in an object file by using the @code{objdump}
2570 program with the @samp{-h} option.
2572 Every object file also has a list of @dfn{symbols}, known as the
2573 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2574 has a name, and each defined symbol has an address, among other
2575 information. If you compile a C or C++ program into an object file, you
2576 will get a defined symbol for every defined function and global or
2577 static variable. Every undefined function or global variable which is
2578 referenced in the input file will become an undefined symbol.
2580 You can see the symbols in an object file by using the @code{nm}
2581 program, or by using the @code{objdump} program with the @samp{-t}
2585 @section Linker Script Format
2586 @cindex linker script format
2587 Linker scripts are text files.
2589 You write a linker script as a series of commands. Each command is
2590 either a keyword, possibly followed by arguments, or an assignment to a
2591 symbol. You may separate commands using semicolons. Whitespace is
2594 Strings such as file or format names can normally be entered directly.
2595 If the file name contains a character such as a comma which would
2596 otherwise serve to separate file names, you may put the file name in
2597 double quotes. There is no way to use a double quote character in a
2600 You may include comments in linker scripts just as in C, delimited by
2601 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2604 @node Simple Example
2605 @section Simple Linker Script Example
2606 @cindex linker script example
2607 @cindex example of linker script
2608 Many linker scripts are fairly simple.
2610 The simplest possible linker script has just one command:
2611 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2612 memory layout of the output file.
2614 The @samp{SECTIONS} command is a powerful command. Here we will
2615 describe a simple use of it. Let's assume your program consists only of
2616 code, initialized data, and uninitialized data. These will be in the
2617 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2618 Let's assume further that these are the only sections which appear in
2621 For this example, let's say that the code should be loaded at address
2622 0x10000, and that the data should start at address 0x8000000. Here is a
2623 linker script which will do that:
2628 .text : @{ *(.text) @}
2630 .data : @{ *(.data) @}
2631 .bss : @{ *(.bss) @}
2635 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2636 followed by a series of symbol assignments and output section
2637 descriptions enclosed in curly braces.
2639 The first line inside the @samp{SECTIONS} command of the above example
2640 sets the value of the special symbol @samp{.}, which is the location
2641 counter. If you do not specify the address of an output section in some
2642 other way (other ways are described later), the address is set from the
2643 current value of the location counter. The location counter is then
2644 incremented by the size of the output section. At the start of the
2645 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2647 The second line defines an output section, @samp{.text}. The colon is
2648 required syntax which may be ignored for now. Within the curly braces
2649 after the output section name, you list the names of the input sections
2650 which should be placed into this output section. The @samp{*} is a
2651 wildcard which matches any file name. The expression @samp{*(.text)}
2652 means all @samp{.text} input sections in all input files.
2654 Since the location counter is @samp{0x10000} when the output section
2655 @samp{.text} is defined, the linker will set the address of the
2656 @samp{.text} section in the output file to be @samp{0x10000}.
2658 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2659 the output file. The linker will place the @samp{.data} output section
2660 at address @samp{0x8000000}. After the linker places the @samp{.data}
2661 output section, the value of the location counter will be
2662 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2663 effect is that the linker will place the @samp{.bss} output section
2664 immediately after the @samp{.data} output section in memory.
2666 The linker will ensure that each output section has the required
2667 alignment, by increasing the location counter if necessary. In this
2668 example, the specified addresses for the @samp{.text} and @samp{.data}
2669 sections will probably satisfy any alignment constraints, but the linker
2670 may have to create a small gap between the @samp{.data} and @samp{.bss}
2673 That's it! That's a simple and complete linker script.
2675 @node Simple Commands
2676 @section Simple Linker Script Commands
2677 @cindex linker script simple commands
2678 In this section we describe the simple linker script commands.
2681 * Entry Point:: Setting the entry point
2682 * File Commands:: Commands dealing with files
2683 @ifclear SingleFormat
2684 * Format Commands:: Commands dealing with object file formats
2687 * Miscellaneous Commands:: Other linker script commands
2691 @subsection Setting the Entry Point
2692 @kindex ENTRY(@var{symbol})
2693 @cindex start of execution
2694 @cindex first instruction
2696 The first instruction to execute in a program is called the @dfn{entry
2697 point}. You can use the @code{ENTRY} linker script command to set the
2698 entry point. The argument is a symbol name:
2703 There are several ways to set the entry point. The linker will set the
2704 entry point by trying each of the following methods in order, and
2705 stopping when one of them succeeds:
2708 the @samp{-e} @var{entry} command-line option;
2710 the @code{ENTRY(@var{symbol})} command in a linker script;
2712 the value of the symbol @code{start}, if defined;
2714 the address of the first byte of the @samp{.text} section, if present;
2716 The address @code{0}.
2720 @subsection Commands Dealing with Files
2721 @cindex linker script file commands
2722 Several linker script commands deal with files.
2725 @item INCLUDE @var{filename}
2726 @kindex INCLUDE @var{filename}
2727 @cindex including a linker script
2728 Include the linker script @var{filename} at this point. The file will
2729 be searched for in the current directory, and in any directory specified
2730 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2733 @item INPUT(@var{file}, @var{file}, @dots{})
2734 @itemx INPUT(@var{file} @var{file} @dots{})
2735 @kindex INPUT(@var{files})
2736 @cindex input files in linker scripts
2737 @cindex input object files in linker scripts
2738 @cindex linker script input object files
2739 The @code{INPUT} command directs the linker to include the named files
2740 in the link, as though they were named on the command line.
2742 For example, if you always want to include @file{subr.o} any time you do
2743 a link, but you can't be bothered to put it on every link command line,
2744 then you can put @samp{INPUT (subr.o)} in your linker script.
2746 In fact, if you like, you can list all of your input files in the linker
2747 script, and then invoke the linker with nothing but a @samp{-T} option.
2749 In case a @dfn{sysroot prefix} is configured, and the filename starts
2750 with the @samp{/} character, and the script being processed was
2751 located inside the @dfn{sysroot prefix}, the filename will be looked
2752 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2753 open the file in the current directory. If it is not found, the
2754 linker will search through the archive library search path. See the
2755 description of @samp{-L} in @ref{Options,,Command Line Options}.
2757 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2758 name to @code{lib@var{file}.a}, as with the command line argument
2761 When you use the @code{INPUT} command in an implicit linker script, the
2762 files will be included in the link at the point at which the linker
2763 script file is included. This can affect archive searching.
2765 @item GROUP(@var{file}, @var{file}, @dots{})
2766 @itemx GROUP(@var{file} @var{file} @dots{})
2767 @kindex GROUP(@var{files})
2768 @cindex grouping input files
2769 The @code{GROUP} command is like @code{INPUT}, except that the named
2770 files should all be archives, and they are searched repeatedly until no
2771 new undefined references are created. See the description of @samp{-(}
2772 in @ref{Options,,Command Line Options}.
2774 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
2775 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
2776 @kindex AS_NEEDED(@var{files})
2777 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2778 commands, among other filenames. The files listed will be handled
2779 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2780 with the exception of ELF shared libraries, that will be added only
2781 when they are actually needed. This construct essentially enables
2782 @option{--as-needed} option for all the files listed inside of it
2783 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2786 @item OUTPUT(@var{filename})
2787 @kindex OUTPUT(@var{filename})
2788 @cindex output file name in linker script
2789 The @code{OUTPUT} command names the output file. Using
2790 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2791 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2792 Line Options}). If both are used, the command line option takes
2795 You can use the @code{OUTPUT} command to define a default name for the
2796 output file other than the usual default of @file{a.out}.
2798 @item SEARCH_DIR(@var{path})
2799 @kindex SEARCH_DIR(@var{path})
2800 @cindex library search path in linker script
2801 @cindex archive search path in linker script
2802 @cindex search path in linker script
2803 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2804 @command{ld} looks for archive libraries. Using
2805 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2806 on the command line (@pxref{Options,,Command Line Options}). If both
2807 are used, then the linker will search both paths. Paths specified using
2808 the command line option are searched first.
2810 @item STARTUP(@var{filename})
2811 @kindex STARTUP(@var{filename})
2812 @cindex first input file
2813 The @code{STARTUP} command is just like the @code{INPUT} command, except
2814 that @var{filename} will become the first input file to be linked, as
2815 though it were specified first on the command line. This may be useful
2816 when using a system in which the entry point is always the start of the
2820 @ifclear SingleFormat
2821 @node Format Commands
2822 @subsection Commands Dealing with Object File Formats
2823 A couple of linker script commands deal with object file formats.
2826 @item OUTPUT_FORMAT(@var{bfdname})
2827 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2828 @kindex OUTPUT_FORMAT(@var{bfdname})
2829 @cindex output file format in linker script
2830 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2831 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2832 exactly like using @samp{--oformat @var{bfdname}} on the command line
2833 (@pxref{Options,,Command Line Options}). If both are used, the command
2834 line option takes precedence.
2836 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2837 formats based on the @samp{-EB} and @samp{-EL} command line options.
2838 This permits the linker script to set the output format based on the
2841 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2842 will be the first argument, @var{default}. If @samp{-EB} is used, the
2843 output format will be the second argument, @var{big}. If @samp{-EL} is
2844 used, the output format will be the third argument, @var{little}.
2846 For example, the default linker script for the MIPS ELF target uses this
2849 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2851 This says that the default format for the output file is
2852 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
2853 option, the output file will be created in the @samp{elf32-littlemips}
2856 @item TARGET(@var{bfdname})
2857 @kindex TARGET(@var{bfdname})
2858 @cindex input file format in linker script
2859 The @code{TARGET} command names the BFD format to use when reading input
2860 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
2861 This command is like using @samp{-b @var{bfdname}} on the command line
2862 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
2863 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
2864 command is also used to set the format for the output file. @xref{BFD}.
2868 @node Miscellaneous Commands
2869 @subsection Other Linker Script Commands
2870 There are a few other linker scripts commands.
2873 @item ASSERT(@var{exp}, @var{message})
2875 @cindex assertion in linker script
2876 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
2877 with an error code, and print @var{message}.
2879 @item EXTERN(@var{symbol} @var{symbol} @dots{})
2881 @cindex undefined symbol in linker script
2882 Force @var{symbol} to be entered in the output file as an undefined
2883 symbol. Doing this may, for example, trigger linking of additional
2884 modules from standard libraries. You may list several @var{symbol}s for
2885 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
2886 command has the same effect as the @samp{-u} command-line option.
2888 @item FORCE_COMMON_ALLOCATION
2889 @kindex FORCE_COMMON_ALLOCATION
2890 @cindex common allocation in linker script
2891 This command has the same effect as the @samp{-d} command-line option:
2892 to make @command{ld} assign space to common symbols even if a relocatable
2893 output file is specified (@samp{-r}).
2895 @item INHIBIT_COMMON_ALLOCATION
2896 @kindex INHIBIT_COMMON_ALLOCATION
2897 @cindex common allocation in linker script
2898 This command has the same effect as the @samp{--no-define-common}
2899 command-line option: to make @code{ld} omit the assignment of addresses
2900 to common symbols even for a non-relocatable output file.
2902 @item NOCROSSREFS(@var{section} @var{section} @dots{})
2903 @kindex NOCROSSREFS(@var{sections})
2904 @cindex cross references
2905 This command may be used to tell @command{ld} to issue an error about any
2906 references among certain output sections.
2908 In certain types of programs, particularly on embedded systems when
2909 using overlays, when one section is loaded into memory, another section
2910 will not be. Any direct references between the two sections would be
2911 errors. For example, it would be an error if code in one section called
2912 a function defined in the other section.
2914 The @code{NOCROSSREFS} command takes a list of output section names. If
2915 @command{ld} detects any cross references between the sections, it reports
2916 an error and returns a non-zero exit status. Note that the
2917 @code{NOCROSSREFS} command uses output section names, not input section
2920 @ifclear SingleFormat
2921 @item OUTPUT_ARCH(@var{bfdarch})
2922 @kindex OUTPUT_ARCH(@var{bfdarch})
2923 @cindex machine architecture
2924 @cindex architecture
2925 Specify a particular output machine architecture. The argument is one
2926 of the names used by the BFD library (@pxref{BFD}). You can see the
2927 architecture of an object file by using the @code{objdump} program with
2928 the @samp{-f} option.
2933 @section Assigning Values to Symbols
2934 @cindex assignment in scripts
2935 @cindex symbol definition, scripts
2936 @cindex variables, defining
2937 You may assign a value to a symbol in a linker script. This will define
2938 the symbol and place it into the symbol table with a global scope.
2941 * Simple Assignments:: Simple Assignments
2943 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2944 * Source Code Reference:: How to use a linker script defined symbol in source code
2947 @node Simple Assignments
2948 @subsection Simple Assignments
2950 You may assign to a symbol using any of the C assignment operators:
2953 @item @var{symbol} = @var{expression} ;
2954 @itemx @var{symbol} += @var{expression} ;
2955 @itemx @var{symbol} -= @var{expression} ;
2956 @itemx @var{symbol} *= @var{expression} ;
2957 @itemx @var{symbol} /= @var{expression} ;
2958 @itemx @var{symbol} <<= @var{expression} ;
2959 @itemx @var{symbol} >>= @var{expression} ;
2960 @itemx @var{symbol} &= @var{expression} ;
2961 @itemx @var{symbol} |= @var{expression} ;
2964 The first case will define @var{symbol} to the value of
2965 @var{expression}. In the other cases, @var{symbol} must already be
2966 defined, and the value will be adjusted accordingly.
2968 The special symbol name @samp{.} indicates the location counter. You
2969 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
2971 The semicolon after @var{expression} is required.
2973 Expressions are defined below; see @ref{Expressions}.
2975 You may write symbol assignments as commands in their own right, or as
2976 statements within a @code{SECTIONS} command, or as part of an output
2977 section description in a @code{SECTIONS} command.
2979 The section of the symbol will be set from the section of the
2980 expression; for more information, see @ref{Expression Section}.
2982 Here is an example showing the three different places that symbol
2983 assignments may be used:
2994 _bdata = (. + 3) & ~ 3;
2995 .data : @{ *(.data) @}
2999 In this example, the symbol @samp{floating_point} will be defined as
3000 zero. The symbol @samp{_etext} will be defined as the address following
3001 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3002 defined as the address following the @samp{.text} output section aligned
3003 upward to a 4 byte boundary.
3008 In some cases, it is desirable for a linker script to define a symbol
3009 only if it is referenced and is not defined by any object included in
3010 the link. For example, traditional linkers defined the symbol
3011 @samp{etext}. However, ANSI C requires that the user be able to use
3012 @samp{etext} as a function name without encountering an error. The
3013 @code{PROVIDE} keyword may be used to define a symbol, such as
3014 @samp{etext}, only if it is referenced but not defined. The syntax is
3015 @code{PROVIDE(@var{symbol} = @var{expression})}.
3017 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3030 In this example, if the program defines @samp{_etext} (with a leading
3031 underscore), the linker will give a multiple definition error. If, on
3032 the other hand, the program defines @samp{etext} (with no leading
3033 underscore), the linker will silently use the definition in the program.
3034 If the program references @samp{etext} but does not define it, the
3035 linker will use the definition in the linker script.
3037 @node PROVIDE_HIDDEN
3038 @subsection PROVIDE_HIDDEN
3039 @cindex PROVIDE_HIDDEN
3040 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3041 hidden and won't be exported.
3043 @node Source Code Reference
3044 @subsection Source Code Reference
3046 Accessing a linker script defined variable from source code is not
3047 intuitive. In particular a linker script symbol is not equivalent to
3048 a variable declaration in a high level language, it is instead a
3049 symbol that does not have a value.
3051 Before going further, it is important to note that compilers often
3052 transform names in the source code into different names when they are
3053 stored in the symbol table. For example, Fortran compilers commonly
3054 prepend or append an underscore, and C++ performs extensive @samp{name
3055 mangling}. Therefore there might be a discrepancy between the name
3056 of a variable as it is used in source code and the name of the same
3057 variable as it is defined in a linker script. For example in C a
3058 linker script variable might be referred to as:
3064 But in the linker script it might be defined as:
3070 In the remaining examples however it is assumed that no name
3071 transformation has taken place.
3073 When a symbol is declared in a high level language such as C, two
3074 things happen. The first is that the compiler reserves enough space
3075 in the program's memory to hold the @emph{value} of the symbol. The
3076 second is that the compiler creates an entry in the program's symbol
3077 table which holds the symbol's @emph{address}. ie the symbol table
3078 contains the address of the block of memory holding the symbol's
3079 value. So for example the following C declaration, at file scope:
3085 creates a entry called @samp{foo} in the symbol table. This entry
3086 holds the address of an @samp{int} sized block of memory where the
3087 number 1000 is initially stored.
3089 When a program references a symbol the compiler generates code that
3090 first accesses the symbol table to find the address of the symbol's
3091 memory block and then code to read the value from that memory block.
3098 looks up the symbol @samp{foo} in the symbol table, gets the address
3099 associated with this symbol and then writes the value 1 into that
3106 looks up the symbol @samp{foo} in the symbol table, gets it address
3107 and then copies this address into the block of memory associated with
3108 the variable @samp{a}.
3110 Linker scripts symbol declarations, by contrast, create an entry in
3111 the symbol table but do not assign any memory to them. Thus they are
3112 an address without a value. So for example the linker script definition:
3118 creates an entry in the symbol table called @samp{foo} which holds
3119 the address of memory location 1000, but nothing special is stored at
3120 address 1000. This means that you cannot access the @emph{value} of a
3121 linker script defined symbol - it has no value - all you can do is
3122 access the @emph{address} of a linker script defined symbol.
3124 Hence when you are using a linker script defined symbol in source code
3125 you should always take the address of the symbol, and never attempt to
3126 use its value. For example suppose you want to copy the contents of a
3127 section of memory called .ROM into a section called .FLASH and the
3128 linker script contains these declarations:
3132 start_of_ROM = .ROM;
3133 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3134 start_of_FLASH = .FLASH;
3138 Then the C source code to perform the copy would be:
3142 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3144 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3148 Note the use of the @samp{&} operators. These are correct.
3151 @section SECTIONS Command
3153 The @code{SECTIONS} command tells the linker how to map input sections
3154 into output sections, and how to place the output sections in memory.
3156 The format of the @code{SECTIONS} command is:
3160 @var{sections-command}
3161 @var{sections-command}
3166 Each @var{sections-command} may of be one of the following:
3170 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3172 a symbol assignment (@pxref{Assignments})
3174 an output section description
3176 an overlay description
3179 The @code{ENTRY} command and symbol assignments are permitted inside the
3180 @code{SECTIONS} command for convenience in using the location counter in
3181 those commands. This can also make the linker script easier to
3182 understand because you can use those commands at meaningful points in
3183 the layout of the output file.
3185 Output section descriptions and overlay descriptions are described
3188 If you do not use a @code{SECTIONS} command in your linker script, the
3189 linker will place each input section into an identically named output
3190 section in the order that the sections are first encountered in the
3191 input files. If all input sections are present in the first file, for
3192 example, the order of sections in the output file will match the order
3193 in the first input file. The first section will be at address zero.
3196 * Output Section Description:: Output section description
3197 * Output Section Name:: Output section name
3198 * Output Section Address:: Output section address
3199 * Input Section:: Input section description
3200 * Output Section Data:: Output section data
3201 * Output Section Keywords:: Output section keywords
3202 * Output Section Discarding:: Output section discarding
3203 * Output Section Attributes:: Output section attributes
3204 * Overlay Description:: Overlay description
3207 @node Output Section Description
3208 @subsection Output Section Description
3209 The full description of an output section looks like this:
3212 @var{section} [@var{address}] [(@var{type})] :
3213 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3215 @var{output-section-command}
3216 @var{output-section-command}
3218 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3222 Most output sections do not use most of the optional section attributes.
3224 The whitespace around @var{section} is required, so that the section
3225 name is unambiguous. The colon and the curly braces are also required.
3226 The line breaks and other white space are optional.
3228 Each @var{output-section-command} may be one of the following:
3232 a symbol assignment (@pxref{Assignments})
3234 an input section description (@pxref{Input Section})
3236 data values to include directly (@pxref{Output Section Data})
3238 a special output section keyword (@pxref{Output Section Keywords})
3241 @node Output Section Name
3242 @subsection Output Section Name
3243 @cindex name, section
3244 @cindex section name
3245 The name of the output section is @var{section}. @var{section} must
3246 meet the constraints of your output format. In formats which only
3247 support a limited number of sections, such as @code{a.out}, the name
3248 must be one of the names supported by the format (@code{a.out}, for
3249 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3250 output format supports any number of sections, but with numbers and not
3251 names (as is the case for Oasys), the name should be supplied as a
3252 quoted numeric string. A section name may consist of any sequence of
3253 characters, but a name which contains any unusual characters such as
3254 commas must be quoted.
3256 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3259 @node Output Section Address
3260 @subsection Output Section Address
3261 @cindex address, section
3262 @cindex section address
3263 The @var{address} is an expression for the VMA (the virtual memory
3264 address) of the output section. If you do not provide @var{address},
3265 the linker will set it based on @var{region} if present, or otherwise
3266 based on the current value of the location counter.
3268 If you provide @var{address}, the address of the output section will be
3269 set to precisely that. If you provide neither @var{address} nor
3270 @var{region}, then the address of the output section will be set to the
3271 current value of the location counter aligned to the alignment
3272 requirements of the output section. The alignment requirement of the
3273 output section is the strictest alignment of any input section contained
3274 within the output section.
3278 .text . : @{ *(.text) @}
3283 .text : @{ *(.text) @}
3286 are subtly different. The first will set the address of the
3287 @samp{.text} output section to the current value of the location
3288 counter. The second will set it to the current value of the location
3289 counter aligned to the strictest alignment of a @samp{.text} input
3292 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3293 For example, if you want to align the section on a 0x10 byte boundary,
3294 so that the lowest four bits of the section address are zero, you could
3295 do something like this:
3297 .text ALIGN(0x10) : @{ *(.text) @}
3300 This works because @code{ALIGN} returns the current location counter
3301 aligned upward to the specified value.
3303 Specifying @var{address} for a section will change the value of the
3307 @subsection Input Section Description
3308 @cindex input sections
3309 @cindex mapping input sections to output sections
3310 The most common output section command is an input section description.
3312 The input section description is the most basic linker script operation.
3313 You use output sections to tell the linker how to lay out your program
3314 in memory. You use input section descriptions to tell the linker how to
3315 map the input files into your memory layout.
3318 * Input Section Basics:: Input section basics
3319 * Input Section Wildcards:: Input section wildcard patterns
3320 * Input Section Common:: Input section for common symbols
3321 * Input Section Keep:: Input section and garbage collection
3322 * Input Section Example:: Input section example
3325 @node Input Section Basics
3326 @subsubsection Input Section Basics
3327 @cindex input section basics
3328 An input section description consists of a file name optionally followed
3329 by a list of section names in parentheses.
3331 The file name and the section name may be wildcard patterns, which we
3332 describe further below (@pxref{Input Section Wildcards}).
3334 The most common input section description is to include all input
3335 sections with a particular name in the output section. For example, to
3336 include all input @samp{.text} sections, you would write:
3341 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3342 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3343 match all files except the ones specified in the EXCLUDE_FILE list. For
3346 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3348 will cause all .ctors sections from all files except @file{crtend.o} and
3349 @file{otherfile.o} to be included.
3351 There are two ways to include more than one section:
3357 The difference between these is the order in which the @samp{.text} and
3358 @samp{.rdata} input sections will appear in the output section. In the
3359 first example, they will be intermingled, appearing in the same order as
3360 they are found in the linker input. In the second example, all
3361 @samp{.text} input sections will appear first, followed by all
3362 @samp{.rdata} input sections.
3364 You can specify a file name to include sections from a particular file.
3365 You would do this if one or more of your files contain special data that
3366 needs to be at a particular location in memory. For example:
3371 If you use a file name without a list of sections, then all sections in
3372 the input file will be included in the output section. This is not
3373 commonly done, but it may by useful on occasion. For example:
3378 When you use a file name which does not contain any wild card
3379 characters, the linker will first see if you also specified the file
3380 name on the linker command line or in an @code{INPUT} command. If you
3381 did not, the linker will attempt to open the file as an input file, as
3382 though it appeared on the command line. Note that this differs from an
3383 @code{INPUT} command, because the linker will not search for the file in
3384 the archive search path.
3386 @node Input Section Wildcards
3387 @subsubsection Input Section Wildcard Patterns
3388 @cindex input section wildcards
3389 @cindex wildcard file name patterns
3390 @cindex file name wildcard patterns
3391 @cindex section name wildcard patterns
3392 In an input section description, either the file name or the section
3393 name or both may be wildcard patterns.
3395 The file name of @samp{*} seen in many examples is a simple wildcard
3396 pattern for the file name.
3398 The wildcard patterns are like those used by the Unix shell.
3402 matches any number of characters
3404 matches any single character
3406 matches a single instance of any of the @var{chars}; the @samp{-}
3407 character may be used to specify a range of characters, as in
3408 @samp{[a-z]} to match any lower case letter
3410 quotes the following character
3413 When a file name is matched with a wildcard, the wildcard characters
3414 will not match a @samp{/} character (used to separate directory names on
3415 Unix). A pattern consisting of a single @samp{*} character is an
3416 exception; it will always match any file name, whether it contains a
3417 @samp{/} or not. In a section name, the wildcard characters will match
3418 a @samp{/} character.
3420 File name wildcard patterns only match files which are explicitly
3421 specified on the command line or in an @code{INPUT} command. The linker
3422 does not search directories to expand wildcards.
3424 If a file name matches more than one wildcard pattern, or if a file name
3425 appears explicitly and is also matched by a wildcard pattern, the linker
3426 will use the first match in the linker script. For example, this
3427 sequence of input section descriptions is probably in error, because the
3428 @file{data.o} rule will not be used:
3430 .data : @{ *(.data) @}
3431 .data1 : @{ data.o(.data) @}
3434 @cindex SORT_BY_NAME
3435 Normally, the linker will place files and sections matched by wildcards
3436 in the order in which they are seen during the link. You can change
3437 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3438 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
3439 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3440 into ascending order by name before placing them in the output file.
3442 @cindex SORT_BY_ALIGNMENT
3443 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3444 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3445 ascending order by alignment before placing them in the output file.
3448 @code{SORT} is an alias for @code{SORT_BY_NAME}.
3450 When there are nested section sorting commands in linker script, there
3451 can be at most 1 level of nesting for section sorting commands.
3455 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3456 It will sort the input sections by name first, then by alignment if 2
3457 sections have the same name.
3459 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3460 It will sort the input sections by alignment first, then by name if 2
3461 sections have the same alignment.
3463 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3464 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3466 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3467 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3469 All other nested section sorting commands are invalid.
3472 When both command line section sorting option and linker script
3473 section sorting command are used, section sorting command always
3474 takes precedence over the command line option.
3476 If the section sorting command in linker script isn't nested, the
3477 command line option will make the section sorting command to be
3478 treated as nested sorting command.
3482 @code{SORT_BY_NAME} (wildcard section pattern ) with
3483 @option{--sort-sections alignment} is equivalent to
3484 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3486 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3487 @option{--sort-section name} is equivalent to
3488 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3491 If the section sorting command in linker script is nested, the
3492 command line option will be ignored.
3494 If you ever get confused about where input sections are going, use the
3495 @samp{-M} linker option to generate a map file. The map file shows
3496 precisely how input sections are mapped to output sections.
3498 This example shows how wildcard patterns might be used to partition
3499 files. This linker script directs the linker to place all @samp{.text}
3500 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3501 The linker will place the @samp{.data} section from all files beginning
3502 with an upper case character in @samp{.DATA}; for all other files, the
3503 linker will place the @samp{.data} section in @samp{.data}.
3507 .text : @{ *(.text) @}
3508 .DATA : @{ [A-Z]*(.data) @}
3509 .data : @{ *(.data) @}
3510 .bss : @{ *(.bss) @}
3515 @node Input Section Common
3516 @subsubsection Input Section for Common Symbols
3517 @cindex common symbol placement
3518 @cindex uninitialized data placement
3519 A special notation is needed for common symbols, because in many object
3520 file formats common symbols do not have a particular input section. The
3521 linker treats common symbols as though they are in an input section
3522 named @samp{COMMON}.
3524 You may use file names with the @samp{COMMON} section just as with any
3525 other input sections. You can use this to place common symbols from a
3526 particular input file in one section while common symbols from other
3527 input files are placed in another section.
3529 In most cases, common symbols in input files will be placed in the
3530 @samp{.bss} section in the output file. For example:
3532 .bss @{ *(.bss) *(COMMON) @}
3535 @cindex scommon section
3536 @cindex small common symbols
3537 Some object file formats have more than one type of common symbol. For
3538 example, the MIPS ELF object file format distinguishes standard common
3539 symbols and small common symbols. In this case, the linker will use a
3540 different special section name for other types of common symbols. In
3541 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3542 symbols and @samp{.scommon} for small common symbols. This permits you
3543 to map the different types of common symbols into memory at different
3547 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3548 notation is now considered obsolete. It is equivalent to
3551 @node Input Section Keep
3552 @subsubsection Input Section and Garbage Collection
3554 @cindex garbage collection
3555 When link-time garbage collection is in use (@samp{--gc-sections}),
3556 it is often useful to mark sections that should not be eliminated.
3557 This is accomplished by surrounding an input section's wildcard entry
3558 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3559 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3561 @node Input Section Example
3562 @subsubsection Input Section Example
3563 The following example is a complete linker script. It tells the linker
3564 to read all of the sections from file @file{all.o} and place them at the
3565 start of output section @samp{outputa} which starts at location
3566 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3567 follows immediately, in the same output section. All of section
3568 @samp{.input2} from @file{foo.o} goes into output section
3569 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3570 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3571 files are written to output section @samp{outputc}.
3599 @node Output Section Data
3600 @subsection Output Section Data
3602 @cindex section data
3603 @cindex output section data
3604 @kindex BYTE(@var{expression})
3605 @kindex SHORT(@var{expression})
3606 @kindex LONG(@var{expression})
3607 @kindex QUAD(@var{expression})
3608 @kindex SQUAD(@var{expression})
3609 You can include explicit bytes of data in an output section by using
3610 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3611 an output section command. Each keyword is followed by an expression in
3612 parentheses providing the value to store (@pxref{Expressions}). The
3613 value of the expression is stored at the current value of the location
3616 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3617 store one, two, four, and eight bytes (respectively). After storing the
3618 bytes, the location counter is incremented by the number of bytes
3621 For example, this will store the byte 1 followed by the four byte value
3622 of the symbol @samp{addr}:
3628 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3629 same; they both store an 8 byte, or 64 bit, value. When both host and
3630 target are 32 bits, an expression is computed as 32 bits. In this case
3631 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3632 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
3634 If the object file format of the output file has an explicit endianness,
3635 which is the normal case, the value will be stored in that endianness.
3636 When the object file format does not have an explicit endianness, as is
3637 true of, for example, S-records, the value will be stored in the
3638 endianness of the first input object file.
3640 Note---these commands only work inside a section description and not
3641 between them, so the following will produce an error from the linker:
3643 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
3645 whereas this will work:
3647 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
3650 @kindex FILL(@var{expression})
3651 @cindex holes, filling
3652 @cindex unspecified memory
3653 You may use the @code{FILL} command to set the fill pattern for the
3654 current section. It is followed by an expression in parentheses. Any
3655 otherwise unspecified regions of memory within the section (for example,
3656 gaps left due to the required alignment of input sections) are filled
3657 with the value of the expression, repeated as
3658 necessary. A @code{FILL} statement covers memory locations after the
3659 point at which it occurs in the section definition; by including more
3660 than one @code{FILL} statement, you can have different fill patterns in
3661 different parts of an output section.
3663 This example shows how to fill unspecified regions of memory with the
3669 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
3670 section attribute, but it only affects the
3671 part of the section following the @code{FILL} command, rather than the
3672 entire section. If both are used, the @code{FILL} command takes
3673 precedence. @xref{Output Section Fill}, for details on the fill
3676 @node Output Section Keywords
3677 @subsection Output Section Keywords
3678 There are a couple of keywords which can appear as output section
3682 @kindex CREATE_OBJECT_SYMBOLS
3683 @cindex input filename symbols
3684 @cindex filename symbols
3685 @item CREATE_OBJECT_SYMBOLS
3686 The command tells the linker to create a symbol for each input file.
3687 The name of each symbol will be the name of the corresponding input
3688 file. The section of each symbol will be the output section in which
3689 the @code{CREATE_OBJECT_SYMBOLS} command appears.
3691 This is conventional for the a.out object file format. It is not
3692 normally used for any other object file format.
3694 @kindex CONSTRUCTORS
3695 @cindex C++ constructors, arranging in link
3696 @cindex constructors, arranging in link
3698 When linking using the a.out object file format, the linker uses an
3699 unusual set construct to support C++ global constructors and
3700 destructors. When linking object file formats which do not support
3701 arbitrary sections, such as ECOFF and XCOFF, the linker will
3702 automatically recognize C++ global constructors and destructors by name.
3703 For these object file formats, the @code{CONSTRUCTORS} command tells the
3704 linker to place constructor information in the output section where the
3705 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
3706 ignored for other object file formats.
3708 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
3709 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
3710 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
3711 the start and end of the global destructors. The
3712 first word in the list is the number of entries, followed by the address
3713 of each constructor or destructor, followed by a zero word. The
3714 compiler must arrange to actually run the code. For these object file
3715 formats @sc{gnu} C++ normally calls constructors from a subroutine
3716 @code{__main}; a call to @code{__main} is automatically inserted into
3717 the startup code for @code{main}. @sc{gnu} C++ normally runs
3718 destructors either by using @code{atexit}, or directly from the function
3721 For object file formats such as @code{COFF} or @code{ELF} which support
3722 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
3723 addresses of global constructors and destructors into the @code{.ctors}
3724 and @code{.dtors} sections. Placing the following sequence into your
3725 linker script will build the sort of table which the @sc{gnu} C++
3726 runtime code expects to see.
3730 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3735 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3741 If you are using the @sc{gnu} C++ support for initialization priority,
3742 which provides some control over the order in which global constructors
3743 are run, you must sort the constructors at link time to ensure that they
3744 are executed in the correct order. When using the @code{CONSTRUCTORS}
3745 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
3746 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
3747 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
3750 Normally the compiler and linker will handle these issues automatically,
3751 and you will not need to concern yourself with them. However, you may
3752 need to consider this if you are using C++ and writing your own linker
3757 @node Output Section Discarding
3758 @subsection Output Section Discarding
3759 @cindex discarding sections
3760 @cindex sections, discarding
3761 @cindex removing sections
3762 The linker will not create output sections with no contents. This is
3763 for convenience when referring to input sections that may or may not
3764 be present in any of the input files. For example:
3766 .foo : @{ *(.foo) @}
3769 will only create a @samp{.foo} section in the output file if there is a
3770 @samp{.foo} section in at least one input file, and if the input
3771 sections are not all empty. Other link script directives that allocate
3772 space in an output section will also create the output section.
3774 The linker will ignore address assignments (@pxref{Output Section Address})
3775 on discarded output sections, except when the linker script defines
3776 symbols in the output section. In that case the linker will obey
3777 the address assignments, possibly advancing dot even though the
3778 section is discarded.
3781 The special output section name @samp{/DISCARD/} may be used to discard
3782 input sections. Any input sections which are assigned to an output
3783 section named @samp{/DISCARD/} are not included in the output file.
3785 @node Output Section Attributes
3786 @subsection Output Section Attributes
3787 @cindex output section attributes
3788 We showed above that the full description of an output section looked
3792 @var{section} [@var{address}] [(@var{type})] :
3793 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3795 @var{output-section-command}
3796 @var{output-section-command}
3798 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3801 We've already described @var{section}, @var{address}, and
3802 @var{output-section-command}. In this section we will describe the
3803 remaining section attributes.
3806 * Output Section Type:: Output section type
3807 * Output Section LMA:: Output section LMA
3808 * Forced Output Alignment:: Forced Output Alignment
3809 * Forced Input Alignment:: Forced Input Alignment
3810 * Output Section Region:: Output section region
3811 * Output Section Phdr:: Output section phdr
3812 * Output Section Fill:: Output section fill
3815 @node Output Section Type
3816 @subsubsection Output Section Type
3817 Each output section may have a type. The type is a keyword in
3818 parentheses. The following types are defined:
3822 The section should be marked as not loadable, so that it will not be
3823 loaded into memory when the program is run.
3828 These type names are supported for backward compatibility, and are
3829 rarely used. They all have the same effect: the section should be
3830 marked as not allocatable, so that no memory is allocated for the
3831 section when the program is run.
3835 @cindex prevent unnecessary loading
3836 @cindex loading, preventing
3837 The linker normally sets the attributes of an output section based on
3838 the input sections which map into it. You can override this by using
3839 the section type. For example, in the script sample below, the
3840 @samp{ROM} section is addressed at memory location @samp{0} and does not
3841 need to be loaded when the program is run. The contents of the
3842 @samp{ROM} section will appear in the linker output file as usual.
3846 ROM 0 (NOLOAD) : @{ @dots{} @}
3852 @node Output Section LMA
3853 @subsubsection Output Section LMA
3854 @kindex AT>@var{lma_region}
3855 @kindex AT(@var{lma})
3856 @cindex load address
3857 @cindex section load address
3858 Every section has a virtual address (VMA) and a load address (LMA); see
3859 @ref{Basic Script Concepts}. The address expression which may appear in
3860 an output section description sets the VMA (@pxref{Output Section
3863 The expression @var{lma} that follows the @code{AT} keyword specifies
3864 the load address of the section.
3866 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
3867 specify a memory region for the section's load address. @xref{MEMORY}.
3868 Note that if the section has not had a VMA assigned to it then the
3869 linker will use the @var{lma_region} as the VMA region as well.
3871 If neither @code{AT} nor @code{AT>} is specified for an allocatable
3872 section, the linker will set the LMA such that the difference between
3873 VMA and LMA for the section is the same as the preceding output
3874 section in the same region. If there is no preceding output section
3875 or the section is not allocatable, the linker will set the LMA equal
3877 @xref{Output Section Region}.
3879 @cindex ROM initialized data
3880 @cindex initialized data in ROM
3881 This feature is designed to make it easy to build a ROM image. For
3882 example, the following linker script creates three output sections: one
3883 called @samp{.text}, which starts at @code{0x1000}, one called
3884 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
3885 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
3886 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
3887 defined with the value @code{0x2000}, which shows that the location
3888 counter holds the VMA value, not the LMA value.
3894 .text 0x1000 : @{ *(.text) _etext = . ; @}
3896 AT ( ADDR (.text) + SIZEOF (.text) )
3897 @{ _data = . ; *(.data); _edata = . ; @}
3899 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
3904 The run-time initialization code for use with a program generated with
3905 this linker script would include something like the following, to copy
3906 the initialized data from the ROM image to its runtime address. Notice
3907 how this code takes advantage of the symbols defined by the linker
3912 extern char _etext, _data, _edata, _bstart, _bend;
3913 char *src = &_etext;
3916 /* ROM has data at end of text; copy it. */
3917 while (dst < &_edata) @{
3922 for (dst = &_bstart; dst< &_bend; dst++)
3927 @node Forced Output Alignment
3928 @subsubsection Forced Output Alignment
3929 @kindex ALIGN(@var{section_align})
3930 @cindex forcing output section alignment
3931 @cindex output section alignment
3932 You can increase an output section's alignment by using ALIGN.
3934 @node Forced Input Alignment
3935 @subsubsection Forced Input Alignment
3936 @kindex SUBALIGN(@var{subsection_align})
3937 @cindex forcing input section alignment
3938 @cindex input section alignment
3939 You can force input section alignment within an output section by using
3940 SUBALIGN. The value specified overrides any alignment given by input
3941 sections, whether larger or smaller.
3943 @node Output Section Region
3944 @subsubsection Output Section Region
3945 @kindex >@var{region}
3946 @cindex section, assigning to memory region
3947 @cindex memory regions and sections
3948 You can assign a section to a previously defined region of memory by
3949 using @samp{>@var{region}}. @xref{MEMORY}.
3951 Here is a simple example:
3954 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
3955 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
3959 @node Output Section Phdr
3960 @subsubsection Output Section Phdr
3962 @cindex section, assigning to program header
3963 @cindex program headers and sections
3964 You can assign a section to a previously defined program segment by
3965 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
3966 one or more segments, then all subsequent allocated sections will be
3967 assigned to those segments as well, unless they use an explicitly
3968 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
3969 linker to not put the section in any segment at all.
3971 Here is a simple example:
3974 PHDRS @{ text PT_LOAD ; @}
3975 SECTIONS @{ .text : @{ *(.text) @} :text @}
3979 @node Output Section Fill
3980 @subsubsection Output Section Fill
3981 @kindex =@var{fillexp}
3982 @cindex section fill pattern
3983 @cindex fill pattern, entire section
3984 You can set the fill pattern for an entire section by using
3985 @samp{=@var{fillexp}}. @var{fillexp} is an expression
3986 (@pxref{Expressions}). Any otherwise unspecified regions of memory
3987 within the output section (for example, gaps left due to the required
3988 alignment of input sections) will be filled with the value, repeated as
3989 necessary. If the fill expression is a simple hex number, ie. a string
3990 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
3991 an arbitrarily long sequence of hex digits can be used to specify the
3992 fill pattern; Leading zeros become part of the pattern too. For all
3993 other cases, including extra parentheses or a unary @code{+}, the fill
3994 pattern is the four least significant bytes of the value of the
3995 expression. In all cases, the number is big-endian.
3997 You can also change the fill value with a @code{FILL} command in the
3998 output section commands; (@pxref{Output Section Data}).
4000 Here is a simple example:
4003 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4007 @node Overlay Description
4008 @subsection Overlay Description
4011 An overlay description provides an easy way to describe sections which
4012 are to be loaded as part of a single memory image but are to be run at
4013 the same memory address. At run time, some sort of overlay manager will
4014 copy the overlaid sections in and out of the runtime memory address as
4015 required, perhaps by simply manipulating addressing bits. This approach
4016 can be useful, for example, when a certain region of memory is faster
4019 Overlays are described using the @code{OVERLAY} command. The
4020 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4021 output section description. The full syntax of the @code{OVERLAY}
4022 command is as follows:
4025 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4029 @var{output-section-command}
4030 @var{output-section-command}
4032 @} [:@var{phdr}@dots{}] [=@var{fill}]
4035 @var{output-section-command}
4036 @var{output-section-command}
4038 @} [:@var{phdr}@dots{}] [=@var{fill}]
4040 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4044 Everything is optional except @code{OVERLAY} (a keyword), and each
4045 section must have a name (@var{secname1} and @var{secname2} above). The
4046 section definitions within the @code{OVERLAY} construct are identical to
4047 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
4048 except that no addresses and no memory regions may be defined for
4049 sections within an @code{OVERLAY}.
4051 The sections are all defined with the same starting address. The load
4052 addresses of the sections are arranged such that they are consecutive in
4053 memory starting at the load address used for the @code{OVERLAY} as a
4054 whole (as with normal section definitions, the load address is optional,
4055 and defaults to the start address; the start address is also optional,
4056 and defaults to the current value of the location counter).
4058 If the @code{NOCROSSREFS} keyword is used, and there any references
4059 among the sections, the linker will report an error. Since the sections
4060 all run at the same address, it normally does not make sense for one
4061 section to refer directly to another. @xref{Miscellaneous Commands,
4064 For each section within the @code{OVERLAY}, the linker automatically
4065 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4066 defined as the starting load address of the section. The symbol
4067 @code{__load_stop_@var{secname}} is defined as the final load address of
4068 the section. Any characters within @var{secname} which are not legal
4069 within C identifiers are removed. C (or assembler) code may use these
4070 symbols to move the overlaid sections around as necessary.
4072 At the end of the overlay, the value of the location counter is set to
4073 the start address of the overlay plus the size of the largest section.
4075 Here is an example. Remember that this would appear inside a
4076 @code{SECTIONS} construct.
4079 OVERLAY 0x1000 : AT (0x4000)
4081 .text0 @{ o1/*.o(.text) @}
4082 .text1 @{ o2/*.o(.text) @}
4087 This will define both @samp{.text0} and @samp{.text1} to start at
4088 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4089 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4090 following symbols will be defined if referenced: @code{__load_start_text0},
4091 @code{__load_stop_text0}, @code{__load_start_text1},
4092 @code{__load_stop_text1}.
4094 C code to copy overlay @code{.text1} into the overlay area might look
4099 extern char __load_start_text1, __load_stop_text1;
4100 memcpy ((char *) 0x1000, &__load_start_text1,
4101 &__load_stop_text1 - &__load_start_text1);
4105 Note that the @code{OVERLAY} command is just syntactic sugar, since
4106 everything it does can be done using the more basic commands. The above
4107 example could have been written identically as follows.
4111 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4112 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4113 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4114 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4115 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4116 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4117 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4122 @section MEMORY Command
4124 @cindex memory regions
4125 @cindex regions of memory
4126 @cindex allocating memory
4127 @cindex discontinuous memory
4128 The linker's default configuration permits allocation of all available
4129 memory. You can override this by using the @code{MEMORY} command.
4131 The @code{MEMORY} command describes the location and size of blocks of
4132 memory in the target. You can use it to describe which memory regions
4133 may be used by the linker, and which memory regions it must avoid. You
4134 can then assign sections to particular memory regions. The linker will
4135 set section addresses based on the memory regions, and will warn about
4136 regions that become too full. The linker will not shuffle sections
4137 around to fit into the available regions.
4139 A linker script may contain at most one use of the @code{MEMORY}
4140 command. However, you can define as many blocks of memory within it as
4141 you wish. The syntax is:
4146 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4152 The @var{name} is a name used in the linker script to refer to the
4153 region. The region name has no meaning outside of the linker script.
4154 Region names are stored in a separate name space, and will not conflict
4155 with symbol names, file names, or section names. Each memory region
4156 must have a distinct name.
4158 @cindex memory region attributes
4159 The @var{attr} string is an optional list of attributes that specify
4160 whether to use a particular memory region for an input section which is
4161 not explicitly mapped in the linker script. As described in
4162 @ref{SECTIONS}, if you do not specify an output section for some input
4163 section, the linker will create an output section with the same name as
4164 the input section. If you define region attributes, the linker will use
4165 them to select the memory region for the output section that it creates.
4167 The @var{attr} string must consist only of the following characters:
4182 Invert the sense of any of the preceding attributes
4185 If a unmapped section matches any of the listed attributes other than
4186 @samp{!}, it will be placed in the memory region. The @samp{!}
4187 attribute reverses this test, so that an unmapped section will be placed
4188 in the memory region only if it does not match any of the listed
4194 The @var{origin} is an numerical expression for the start address of
4195 the memory region. The expression must evaluate to a constant and it
4196 cannot involve any symbols. The keyword @code{ORIGIN} may be
4197 abbreviated to @code{org} or @code{o} (but not, for example,
4203 The @var{len} is an expression for the size in bytes of the memory
4204 region. As with the @var{origin} expression, the expression must
4205 be numerical only and must evaluate to a constant. The keyword
4206 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4208 In the following example, we specify that there are two memory regions
4209 available for allocation: one starting at @samp{0} for 256 kilobytes,
4210 and the other starting at @samp{0x40000000} for four megabytes. The
4211 linker will place into the @samp{rom} memory region every section which
4212 is not explicitly mapped into a memory region, and is either read-only
4213 or executable. The linker will place other sections which are not
4214 explicitly mapped into a memory region into the @samp{ram} memory
4221 rom (rx) : ORIGIN = 0, LENGTH = 256K
4222 ram (!rx) : org = 0x40000000, l = 4M
4227 Once you define a memory region, you can direct the linker to place
4228 specific output sections into that memory region by using the
4229 @samp{>@var{region}} output section attribute. For example, if you have
4230 a memory region named @samp{mem}, you would use @samp{>mem} in the
4231 output section definition. @xref{Output Section Region}. If no address
4232 was specified for the output section, the linker will set the address to
4233 the next available address within the memory region. If the combined
4234 output sections directed to a memory region are too large for the
4235 region, the linker will issue an error message.
4237 It is possible to access the origin and length of a memory in an
4238 expression via the @code{ORIGIN(@var{memory})} and
4239 @code{LENGTH(@var{memory})} functions:
4243 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4248 @section PHDRS Command
4250 @cindex program headers
4251 @cindex ELF program headers
4252 @cindex program segments
4253 @cindex segments, ELF
4254 The ELF object file format uses @dfn{program headers}, also knows as
4255 @dfn{segments}. The program headers describe how the program should be
4256 loaded into memory. You can print them out by using the @code{objdump}
4257 program with the @samp{-p} option.
4259 When you run an ELF program on a native ELF system, the system loader
4260 reads the program headers in order to figure out how to load the
4261 program. This will only work if the program headers are set correctly.
4262 This manual does not describe the details of how the system loader
4263 interprets program headers; for more information, see the ELF ABI.
4265 The linker will create reasonable program headers by default. However,
4266 in some cases, you may need to specify the program headers more
4267 precisely. You may use the @code{PHDRS} command for this purpose. When
4268 the linker sees the @code{PHDRS} command in the linker script, it will
4269 not create any program headers other than the ones specified.
4271 The linker only pays attention to the @code{PHDRS} command when
4272 generating an ELF output file. In other cases, the linker will simply
4273 ignore @code{PHDRS}.
4275 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4276 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4282 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4283 [ FLAGS ( @var{flags} ) ] ;
4288 The @var{name} is used only for reference in the @code{SECTIONS} command
4289 of the linker script. It is not put into the output file. Program
4290 header names are stored in a separate name space, and will not conflict
4291 with symbol names, file names, or section names. Each program header
4292 must have a distinct name.
4294 Certain program header types describe segments of memory which the
4295 system loader will load from the file. In the linker script, you
4296 specify the contents of these segments by placing allocatable output
4297 sections in the segments. You use the @samp{:@var{phdr}} output section
4298 attribute to place a section in a particular segment. @xref{Output
4301 It is normal to put certain sections in more than one segment. This
4302 merely implies that one segment of memory contains another. You may
4303 repeat @samp{:@var{phdr}}, using it once for each segment which should
4304 contain the section.
4306 If you place a section in one or more segments using @samp{:@var{phdr}},
4307 then the linker will place all subsequent allocatable sections which do
4308 not specify @samp{:@var{phdr}} in the same segments. This is for
4309 convenience, since generally a whole set of contiguous sections will be
4310 placed in a single segment. You can use @code{:NONE} to override the
4311 default segment and tell the linker to not put the section in any
4316 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4317 the program header type to further describe the contents of the segment.
4318 The @code{FILEHDR} keyword means that the segment should include the ELF
4319 file header. The @code{PHDRS} keyword means that the segment should
4320 include the ELF program headers themselves.
4322 The @var{type} may be one of the following. The numbers indicate the
4323 value of the keyword.
4326 @item @code{PT_NULL} (0)
4327 Indicates an unused program header.
4329 @item @code{PT_LOAD} (1)
4330 Indicates that this program header describes a segment to be loaded from
4333 @item @code{PT_DYNAMIC} (2)
4334 Indicates a segment where dynamic linking information can be found.
4336 @item @code{PT_INTERP} (3)
4337 Indicates a segment where the name of the program interpreter may be
4340 @item @code{PT_NOTE} (4)
4341 Indicates a segment holding note information.
4343 @item @code{PT_SHLIB} (5)
4344 A reserved program header type, defined but not specified by the ELF
4347 @item @code{PT_PHDR} (6)
4348 Indicates a segment where the program headers may be found.
4350 @item @var{expression}
4351 An expression giving the numeric type of the program header. This may
4352 be used for types not defined above.
4355 You can specify that a segment should be loaded at a particular address
4356 in memory by using an @code{AT} expression. This is identical to the
4357 @code{AT} command used as an output section attribute (@pxref{Output
4358 Section LMA}). The @code{AT} command for a program header overrides the
4359 output section attribute.
4361 The linker will normally set the segment flags based on the sections
4362 which comprise the segment. You may use the @code{FLAGS} keyword to
4363 explicitly specify the segment flags. The value of @var{flags} must be
4364 an integer. It is used to set the @code{p_flags} field of the program
4367 Here is an example of @code{PHDRS}. This shows a typical set of program
4368 headers used on a native ELF system.
4374 headers PT_PHDR PHDRS ;
4376 text PT_LOAD FILEHDR PHDRS ;
4378 dynamic PT_DYNAMIC ;
4384 .interp : @{ *(.interp) @} :text :interp
4385 .text : @{ *(.text) @} :text
4386 .rodata : @{ *(.rodata) @} /* defaults to :text */
4388 . = . + 0x1000; /* move to a new page in memory */
4389 .data : @{ *(.data) @} :data
4390 .dynamic : @{ *(.dynamic) @} :data :dynamic
4397 @section VERSION Command
4398 @kindex VERSION @{script text@}
4399 @cindex symbol versions
4400 @cindex version script
4401 @cindex versions of symbols
4402 The linker supports symbol versions when using ELF. Symbol versions are
4403 only useful when using shared libraries. The dynamic linker can use
4404 symbol versions to select a specific version of a function when it runs
4405 a program that may have been linked against an earlier version of the
4408 You can include a version script directly in the main linker script, or
4409 you can supply the version script as an implicit linker script. You can
4410 also use the @samp{--version-script} linker option.
4412 The syntax of the @code{VERSION} command is simply
4414 VERSION @{ version-script-commands @}
4417 The format of the version script commands is identical to that used by
4418 Sun's linker in Solaris 2.5. The version script defines a tree of
4419 version nodes. You specify the node names and interdependencies in the
4420 version script. You can specify which symbols are bound to which
4421 version nodes, and you can reduce a specified set of symbols to local
4422 scope so that they are not globally visible outside of the shared
4425 The easiest way to demonstrate the version script language is with a few
4446 "int f(int, double)";
4451 This example version script defines three version nodes. The first
4452 version node defined is @samp{VERS_1.1}; it has no other dependencies.
4453 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
4454 a number of symbols to local scope so that they are not visible outside
4455 of the shared library; this is done using wildcard patterns, so that any
4456 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4457 is matched. The wildcard patterns available are the same as those used
4458 in the shell when matching filenames (also known as ``globbing'').
4459 However, if you specify the symbol name inside double quotes, then the
4460 name is treated as literal, rather than as a glob pattern.
4462 Next, the version script defines node @samp{VERS_1.2}. This node
4463 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
4464 to the version node @samp{VERS_1.2}.
4466 Finally, the version script defines node @samp{VERS_2.0}. This node
4467 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
4468 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4470 When the linker finds a symbol defined in a library which is not
4471 specifically bound to a version node, it will effectively bind it to an
4472 unspecified base version of the library. You can bind all otherwise
4473 unspecified symbols to a given version node by using @samp{global: *;}
4474 somewhere in the version script.
4476 The names of the version nodes have no specific meaning other than what
4477 they might suggest to the person reading them. The @samp{2.0} version
4478 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4479 However, this would be a confusing way to write a version script.
4481 Node name can be omitted, provided it is the only version node
4482 in the version script. Such version script doesn't assign any versions to
4483 symbols, only selects which symbols will be globally visible out and which
4487 @{ global: foo; bar; local: *; @};
4490 When you link an application against a shared library that has versioned
4491 symbols, the application itself knows which version of each symbol it
4492 requires, and it also knows which version nodes it needs from each
4493 shared library it is linked against. Thus at runtime, the dynamic
4494 loader can make a quick check to make sure that the libraries you have
4495 linked against do in fact supply all of the version nodes that the
4496 application will need to resolve all of the dynamic symbols. In this
4497 way it is possible for the dynamic linker to know with certainty that
4498 all external symbols that it needs will be resolvable without having to
4499 search for each symbol reference.
4501 The symbol versioning is in effect a much more sophisticated way of
4502 doing minor version checking that SunOS does. The fundamental problem
4503 that is being addressed here is that typically references to external
4504 functions are bound on an as-needed basis, and are not all bound when
4505 the application starts up. If a shared library is out of date, a
4506 required interface may be missing; when the application tries to use
4507 that interface, it may suddenly and unexpectedly fail. With symbol
4508 versioning, the user will get a warning when they start their program if
4509 the libraries being used with the application are too old.
4511 There are several GNU extensions to Sun's versioning approach. The
4512 first of these is the ability to bind a symbol to a version node in the
4513 source file where the symbol is defined instead of in the versioning
4514 script. This was done mainly to reduce the burden on the library
4515 maintainer. You can do this by putting something like:
4517 __asm__(".symver original_foo,foo@@VERS_1.1");
4520 in the C source file. This renames the function @samp{original_foo} to
4521 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4522 The @samp{local:} directive can be used to prevent the symbol
4523 @samp{original_foo} from being exported. A @samp{.symver} directive
4524 takes precedence over a version script.
4526 The second GNU extension is to allow multiple versions of the same
4527 function to appear in a given shared library. In this way you can make
4528 an incompatible change to an interface without increasing the major
4529 version number of the shared library, while still allowing applications
4530 linked against the old interface to continue to function.
4532 To do this, you must use multiple @samp{.symver} directives in the
4533 source file. Here is an example:
4536 __asm__(".symver original_foo,foo@@");
4537 __asm__(".symver old_foo,foo@@VERS_1.1");
4538 __asm__(".symver old_foo1,foo@@VERS_1.2");
4539 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4542 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4543 unspecified base version of the symbol. The source file that contains this
4544 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4545 @samp{old_foo1}, and @samp{new_foo}.
4547 When you have multiple definitions of a given symbol, there needs to be
4548 some way to specify a default version to which external references to
4549 this symbol will be bound. You can do this with the
4550 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4551 declare one version of a symbol as the default in this manner; otherwise
4552 you would effectively have multiple definitions of the same symbol.
4554 If you wish to bind a reference to a specific version of the symbol
4555 within the shared library, you can use the aliases of convenience
4556 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4557 specifically bind to an external version of the function in question.
4559 You can also specify the language in the version script:
4562 VERSION extern "lang" @{ version-script-commands @}
4565 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4566 The linker will iterate over the list of symbols at the link time and
4567 demangle them according to @samp{lang} before matching them to the
4568 patterns specified in @samp{version-script-commands}.
4570 Demangled names may contains spaces and other special characters. As
4571 described above, you can use a glob pattern to match demangled names,
4572 or you can use a double-quoted string to match the string exactly. In
4573 the latter case, be aware that minor differences (such as differing
4574 whitespace) between the version script and the demangler output will
4575 cause a mismatch. As the exact string generated by the demangler
4576 might change in the future, even if the mangled name does not, you
4577 should check that all of your version directives are behaving as you
4578 expect when you upgrade.
4581 @section Expressions in Linker Scripts
4584 The syntax for expressions in the linker script language is identical to
4585 that of C expressions. All expressions are evaluated as integers. All
4586 expressions are evaluated in the same size, which is 32 bits if both the
4587 host and target are 32 bits, and is otherwise 64 bits.
4589 You can use and set symbol values in expressions.
4591 The linker defines several special purpose builtin functions for use in
4595 * Constants:: Constants
4596 * Symbols:: Symbol Names
4597 * Orphan Sections:: Orphan Sections
4598 * Location Counter:: The Location Counter
4599 * Operators:: Operators
4600 * Evaluation:: Evaluation
4601 * Expression Section:: The Section of an Expression
4602 * Builtin Functions:: Builtin Functions
4606 @subsection Constants
4607 @cindex integer notation
4608 @cindex constants in linker scripts
4609 All constants are integers.
4611 As in C, the linker considers an integer beginning with @samp{0} to be
4612 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4613 hexadecimal. The linker considers other integers to be decimal.
4615 @cindex scaled integers
4616 @cindex K and M integer suffixes
4617 @cindex M and K integer suffixes
4618 @cindex suffixes for integers
4619 @cindex integer suffixes
4620 In addition, you can use the suffixes @code{K} and @code{M} to scale a
4624 @c END TEXI2ROFF-KILL
4625 @code{1024} or @code{1024*1024}
4629 ${\rm 1024}$ or ${\rm 1024}^2$
4631 @c END TEXI2ROFF-KILL
4632 respectively. For example, the following all refer to the same quantity:
4640 @subsection Symbol Names
4641 @cindex symbol names
4643 @cindex quoted symbol names
4645 Unless quoted, symbol names start with a letter, underscore, or period
4646 and may include letters, digits, underscores, periods, and hyphens.
4647 Unquoted symbol names must not conflict with any keywords. You can
4648 specify a symbol which contains odd characters or has the same name as a
4649 keyword by surrounding the symbol name in double quotes:
4652 "with a space" = "also with a space" + 10;
4655 Since symbols can contain many non-alphabetic characters, it is safest
4656 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
4657 whereas @samp{A - B} is an expression involving subtraction.
4659 @node Orphan Sections
4660 @subsection Orphan Sections
4662 Orphan sections are sections present in the input files which
4663 are not explicitly placed into the output file by the linker
4664 script. The linker will still copy these sections into the
4665 output file, but it has to guess as to where they should be
4666 placed. The linker uses a simple heuristic to do this. It
4667 attempts to place orphan sections after non-orphan sections of the
4668 same attribute, such as code vs data, loadable vs non-loadable, etc.
4669 If there is not enough room to do this then it places
4670 at the end of the file.
4672 For ELF targets, the attribute of the section includes section type as
4673 well as section flag.
4675 @node Location Counter
4676 @subsection The Location Counter
4679 @cindex location counter
4680 @cindex current output location
4681 The special linker variable @dfn{dot} @samp{.} always contains the
4682 current output location counter. Since the @code{.} always refers to a
4683 location in an output section, it may only appear in an expression
4684 within a @code{SECTIONS} command. The @code{.} symbol may appear
4685 anywhere that an ordinary symbol is allowed in an expression.
4688 Assigning a value to @code{.} will cause the location counter to be
4689 moved. This may be used to create holes in the output section. The
4690 location counter may not be moved backwards inside an output section,
4691 and may not be moved backwards outside of an output section if so
4692 doing creates areas with overlapping LMAs.
4708 In the previous example, the @samp{.text} section from @file{file1} is
4709 located at the beginning of the output section @samp{output}. It is
4710 followed by a 1000 byte gap. Then the @samp{.text} section from
4711 @file{file2} appears, also with a 1000 byte gap following before the
4712 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
4713 specifies what data to write in the gaps (@pxref{Output Section Fill}).
4715 @cindex dot inside sections
4716 Note: @code{.} actually refers to the byte offset from the start of the
4717 current containing object. Normally this is the @code{SECTIONS}
4718 statement, whose start address is 0, hence @code{.} can be used as an
4719 absolute address. If @code{.} is used inside a section description
4720 however, it refers to the byte offset from the start of that section,
4721 not an absolute address. Thus in a script like this:
4739 The @samp{.text} section will be assigned a starting address of 0x100
4740 and a size of exactly 0x200 bytes, even if there is not enough data in
4741 the @samp{.text} input sections to fill this area. (If there is too
4742 much data, an error will be produced because this would be an attempt to
4743 move @code{.} backwards). The @samp{.data} section will start at 0x500
4744 and it will have an extra 0x600 bytes worth of space after the end of
4745 the values from the @samp{.data} input sections and before the end of
4746 the @samp{.data} output section itself.
4748 @cindex dot outside sections
4749 Setting symbols to the value of the location counter outside of an
4750 output section statement can result in unexpected values if the linker
4751 needs to place orphan sections. For example, given the following:
4757 .text: @{ *(.text) @}
4761 .data: @{ *(.data) @}
4766 If the linker needs to place some input section, e.g. @code{.rodata},
4767 not mentioned in the script, it might choose to place that section
4768 between @code{.text} and @code{.data}. You might think the linker
4769 should place @code{.rodata} on the blank line in the above script, but
4770 blank lines are of no particular significance to the linker. As well,
4771 the linker doesn't associate the above symbol names with their
4772 sections. Instead, it assumes that all assignments or other
4773 statements belong to the previous output section, except for the
4774 special case of an assignment to @code{.}. I.e., the linker will
4775 place the orphan @code{.rodata} section as if the script was written
4782 .text: @{ *(.text) @}
4786 .rodata: @{ *(.rodata) @}
4787 .data: @{ *(.data) @}
4792 This may or may not be the script author's intention for the value of
4793 @code{start_of_data}. One way to influence the orphan section
4794 placement is to assign the location counter to itself, as the linker
4795 assumes that an assignment to @code{.} is setting the start address of
4796 a following output section and thus should be grouped with that
4797 section. So you could write:
4803 .text: @{ *(.text) @}
4808 .data: @{ *(.data) @}
4813 Now, the orphan @code{.rodata} section will be placed between
4814 @code{end_of_text} and @code{start_of_data}.
4818 @subsection Operators
4819 @cindex operators for arithmetic
4820 @cindex arithmetic operators
4821 @cindex precedence in expressions
4822 The linker recognizes the standard C set of arithmetic operators, with
4823 the standard bindings and precedence levels:
4826 @c END TEXI2ROFF-KILL
4828 precedence associativity Operators Notes
4834 5 left == != > < <= >=
4840 11 right &= += -= *= /= (2)
4844 (1) Prefix operators
4845 (2) @xref{Assignments}.
4849 \vskip \baselineskip
4850 %"lispnarrowing" is the extra indent used generally for smallexample
4851 \hskip\lispnarrowing\vbox{\offinterlineskip
4854 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
4855 height2pt&\omit&&\omit&&\omit&\cr
4856 &Precedence&& Associativity &&{\rm Operators}&\cr
4857 height2pt&\omit&&\omit&&\omit&\cr
4859 height2pt&\omit&&\omit&&\omit&\cr
4861 % '176 is tilde, '~' in tt font
4862 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
4863 &2&&left&&* / \%&\cr
4866 &5&&left&&== != > < <= >=&\cr
4869 &8&&left&&{\&\&}&\cr
4872 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
4874 height2pt&\omit&&\omit&&\omit&\cr}
4879 @obeylines@parskip=0pt@parindent=0pt
4880 @dag@quad Prefix operators.
4881 @ddag@quad @xref{Assignments}.
4884 @c END TEXI2ROFF-KILL
4887 @subsection Evaluation
4888 @cindex lazy evaluation
4889 @cindex expression evaluation order
4890 The linker evaluates expressions lazily. It only computes the value of
4891 an expression when absolutely necessary.
4893 The linker needs some information, such as the value of the start
4894 address of the first section, and the origins and lengths of memory
4895 regions, in order to do any linking at all. These values are computed
4896 as soon as possible when the linker reads in the linker script.
4898 However, other values (such as symbol values) are not known or needed
4899 until after storage allocation. Such values are evaluated later, when
4900 other information (such as the sizes of output sections) is available
4901 for use in the symbol assignment expression.
4903 The sizes of sections cannot be known until after allocation, so
4904 assignments dependent upon these are not performed until after
4907 Some expressions, such as those depending upon the location counter
4908 @samp{.}, must be evaluated during section allocation.
4910 If the result of an expression is required, but the value is not
4911 available, then an error results. For example, a script like the
4917 .text 9+this_isnt_constant :
4923 will cause the error message @samp{non constant expression for initial
4926 @node Expression Section
4927 @subsection The Section of an Expression
4928 @cindex expression sections
4929 @cindex absolute expressions
4930 @cindex relative expressions
4931 @cindex absolute and relocatable symbols
4932 @cindex relocatable and absolute symbols
4933 @cindex symbols, relocatable and absolute
4934 When the linker evaluates an expression, the result is either absolute
4935 or relative to some section. A relative expression is expressed as a
4936 fixed offset from the base of a section.
4938 The position of the expression within the linker script determines
4939 whether it is absolute or relative. An expression which appears within
4940 an output section definition is relative to the base of the output
4941 section. An expression which appears elsewhere will be absolute.
4943 A symbol set to a relative expression will be relocatable if you request
4944 relocatable output using the @samp{-r} option. That means that a
4945 further link operation may change the value of the symbol. The symbol's
4946 section will be the section of the relative expression.
4948 A symbol set to an absolute expression will retain the same value
4949 through any further link operation. The symbol will be absolute, and
4950 will not have any particular associated section.
4952 You can use the builtin function @code{ABSOLUTE} to force an expression
4953 to be absolute when it would otherwise be relative. For example, to
4954 create an absolute symbol set to the address of the end of the output
4955 section @samp{.data}:
4959 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
4963 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
4964 @samp{.data} section.
4966 @node Builtin Functions
4967 @subsection Builtin Functions
4968 @cindex functions in expressions
4969 The linker script language includes a number of builtin functions for
4970 use in linker script expressions.
4973 @item ABSOLUTE(@var{exp})
4974 @kindex ABSOLUTE(@var{exp})
4975 @cindex expression, absolute
4976 Return the absolute (non-relocatable, as opposed to non-negative) value
4977 of the expression @var{exp}. Primarily useful to assign an absolute
4978 value to a symbol within a section definition, where symbol values are
4979 normally section relative. @xref{Expression Section}.
4981 @item ADDR(@var{section})
4982 @kindex ADDR(@var{section})
4983 @cindex section address in expression
4984 Return the absolute address (the VMA) of the named @var{section}. Your
4985 script must previously have defined the location of that section. In
4986 the following example, @code{symbol_1} and @code{symbol_2} are assigned
4993 start_of_output_1 = ABSOLUTE(.);
4998 symbol_1 = ADDR(.output1);
4999 symbol_2 = start_of_output_1;
5005 @item ALIGN(@var{align})
5006 @itemx ALIGN(@var{exp},@var{align})
5007 @kindex ALIGN(@var{align})
5008 @kindex ALIGN(@var{exp},@var{align})
5009 @cindex round up location counter
5010 @cindex align location counter
5011 @cindex round up expression
5012 @cindex align expression
5013 Return the location counter (@code{.}) or arbitrary expression aligned
5014 to the next @var{align} boundary. The single operand @code{ALIGN}
5015 doesn't change the value of the location counter---it just does
5016 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5017 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5018 equivalent to @code{ALIGN(., @var{align})}).
5020 Here is an example which aligns the output @code{.data} section to the
5021 next @code{0x2000} byte boundary after the preceding section and sets a
5022 variable within the section to the next @code{0x8000} boundary after the
5027 .data ALIGN(0x2000): @{
5029 variable = ALIGN(0x8000);
5035 The first use of @code{ALIGN} in this example specifies the location of
5036 a section because it is used as the optional @var{address} attribute of
5037 a section definition (@pxref{Output Section Address}). The second use
5038 of @code{ALIGN} is used to defines the value of a symbol.
5040 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5042 @item ALIGNOF(@var{section})
5043 @kindex ALIGNOF(@var{section})
5044 @cindex section alignment
5045 Return the alignment in bytes of the named @var{section}, if that section has
5046 been allocated. If the section has not been allocated when this is
5047 evaluated, the linker will report an error. In the following example,
5048 the alignment of the @code{.output} section is stored as the first
5049 value in that section.
5054 LONG (ALIGNOF (.output))
5061 @item BLOCK(@var{exp})
5062 @kindex BLOCK(@var{exp})
5063 This is a synonym for @code{ALIGN}, for compatibility with older linker
5064 scripts. It is most often seen when setting the address of an output
5067 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5068 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5069 This is equivalent to either
5071 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5075 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5078 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5079 for the data segment (area between the result of this expression and
5080 @code{DATA_SEGMENT_END}) than the former or not.
5081 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5082 memory will be saved at the expense of up to @var{commonpagesize} wasted
5083 bytes in the on-disk file.
5085 This expression can only be used directly in @code{SECTIONS} commands, not in
5086 any output section descriptions and only once in the linker script.
5087 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5088 be the system page size the object wants to be optimized for (while still
5089 working on system page sizes up to @var{maxpagesize}).
5094 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5097 @item DATA_SEGMENT_END(@var{exp})
5098 @kindex DATA_SEGMENT_END(@var{exp})
5099 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5100 evaluation purposes.
5103 . = DATA_SEGMENT_END(.);
5106 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5107 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5108 This defines the end of the @code{PT_GNU_RELRO} segment when
5109 @samp{-z relro} option is used. Second argument is returned.
5110 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5111 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5112 @var{exp} + @var{offset} is aligned to the most commonly used page
5113 boundary for particular target. If present in the linker script,
5114 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5115 @code{DATA_SEGMENT_END}.
5118 . = DATA_SEGMENT_RELRO_END(24, .);
5121 @item DEFINED(@var{symbol})
5122 @kindex DEFINED(@var{symbol})
5123 @cindex symbol defaults
5124 Return 1 if @var{symbol} is in the linker global symbol table and is
5125 defined before the statement using DEFINED in the script, otherwise
5126 return 0. You can use this function to provide
5127 default values for symbols. For example, the following script fragment
5128 shows how to set a global symbol @samp{begin} to the first location in
5129 the @samp{.text} section---but if a symbol called @samp{begin} already
5130 existed, its value is preserved:
5136 begin = DEFINED(begin) ? begin : . ;
5144 @item LENGTH(@var{memory})
5145 @kindex LENGTH(@var{memory})
5146 Return the length of the memory region named @var{memory}.
5148 @item LOADADDR(@var{section})
5149 @kindex LOADADDR(@var{section})
5150 @cindex section load address in expression
5151 Return the absolute LMA of the named @var{section}. This is normally
5152 the same as @code{ADDR}, but it may be different if the @code{AT}
5153 attribute is used in the output section definition (@pxref{Output
5157 @item MAX(@var{exp1}, @var{exp2})
5158 Returns the maximum of @var{exp1} and @var{exp2}.
5161 @item MIN(@var{exp1}, @var{exp2})
5162 Returns the minimum of @var{exp1} and @var{exp2}.
5164 @item NEXT(@var{exp})
5165 @kindex NEXT(@var{exp})
5166 @cindex unallocated address, next
5167 Return the next unallocated address that is a multiple of @var{exp}.
5168 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5169 use the @code{MEMORY} command to define discontinuous memory for the
5170 output file, the two functions are equivalent.
5172 @item ORIGIN(@var{memory})
5173 @kindex ORIGIN(@var{memory})
5174 Return the origin of the memory region named @var{memory}.
5176 @item SEGMENT_START(@var{segment}, @var{default})
5177 @kindex SEGMENT_START(@var{segment}, @var{default})
5178 Return the base address of the named @var{segment}. If an explicit
5179 value has been given for this segment (with a command-line @samp{-T}
5180 option) that value will be returned; otherwise the value will be
5181 @var{default}. At present, the @samp{-T} command-line option can only
5182 be used to set the base address for the ``text'', ``data'', and
5183 ``bss'' sections, but you use @code{SEGMENT_START} with any segment
5186 @item SIZEOF(@var{section})
5187 @kindex SIZEOF(@var{section})
5188 @cindex section size
5189 Return the size in bytes of the named @var{section}, if that section has
5190 been allocated. If the section has not been allocated when this is
5191 evaluated, the linker will report an error. In the following example,
5192 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5201 symbol_1 = .end - .start ;
5202 symbol_2 = SIZEOF(.output);
5207 @item SIZEOF_HEADERS
5208 @itemx sizeof_headers
5209 @kindex SIZEOF_HEADERS
5211 Return the size in bytes of the output file's headers. This is
5212 information which appears at the start of the output file. You can use
5213 this number when setting the start address of the first section, if you
5214 choose, to facilitate paging.
5216 @cindex not enough room for program headers
5217 @cindex program headers, not enough room
5218 When producing an ELF output file, if the linker script uses the
5219 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5220 number of program headers before it has determined all the section
5221 addresses and sizes. If the linker later discovers that it needs
5222 additional program headers, it will report an error @samp{not enough
5223 room for program headers}. To avoid this error, you must avoid using
5224 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5225 script to avoid forcing the linker to use additional program headers, or
5226 you must define the program headers yourself using the @code{PHDRS}
5227 command (@pxref{PHDRS}).
5230 @node Implicit Linker Scripts
5231 @section Implicit Linker Scripts
5232 @cindex implicit linker scripts
5233 If you specify a linker input file which the linker can not recognize as
5234 an object file or an archive file, it will try to read the file as a
5235 linker script. If the file can not be parsed as a linker script, the
5236 linker will report an error.
5238 An implicit linker script will not replace the default linker script.
5240 Typically an implicit linker script would contain only symbol
5241 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5244 Any input files read because of an implicit linker script will be read
5245 at the position in the command line where the implicit linker script was
5246 read. This can affect archive searching.
5249 @node Machine Dependent
5250 @chapter Machine Dependent Features
5252 @cindex machine dependencies
5253 @command{ld} has additional features on some platforms; the following
5254 sections describe them. Machines where @command{ld} has no additional
5255 functionality are not listed.
5259 * H8/300:: @command{ld} and the H8/300
5262 * i960:: @command{ld} and the Intel 960 family
5265 * ARM:: @command{ld} and the ARM family
5268 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
5271 * MMIX:: @command{ld} and MMIX
5274 * MSP430:: @command{ld} and MSP430
5277 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
5280 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
5283 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
5286 * SPU ELF:: @command{ld} and SPU ELF Support
5289 * TI COFF:: @command{ld} and TI COFF
5292 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
5295 * Xtensa:: @command{ld} and Xtensa Processors
5306 @section @command{ld} and the H8/300
5308 @cindex H8/300 support
5309 For the H8/300, @command{ld} can perform these global optimizations when
5310 you specify the @samp{--relax} command-line option.
5313 @cindex relaxing on H8/300
5314 @item relaxing address modes
5315 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5316 targets are within eight bits, and turns them into eight-bit
5317 program-counter relative @code{bsr} and @code{bra} instructions,
5320 @cindex synthesizing on H8/300
5321 @item synthesizing instructions
5322 @c FIXME: specifically mov.b, or any mov instructions really?
5323 @command{ld} finds all @code{mov.b} instructions which use the
5324 sixteen-bit absolute address form, but refer to the top
5325 page of memory, and changes them to use the eight-bit address form.
5326 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5327 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5328 top page of memory).
5330 @item bit manipulation instructions
5331 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
5332 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5333 which use 32 bit and 16 bit absolute address form, but refer to the top
5334 page of memory, and changes them to use the 8 bit address form.
5335 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5336 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5337 the top page of memory).
5339 @item system control instructions
5340 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
5341 32 bit absolute address form, but refer to the top page of memory, and
5342 changes them to use 16 bit address form.
5343 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5344 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5345 the top page of memory).
5355 @c This stuff is pointless to say unless you're especially concerned
5356 @c with Renesas chips; don't enable it for generic case, please.
5358 @chapter @command{ld} and Other Renesas Chips
5360 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5361 H8/500, and SH chips. No special features, commands, or command-line
5362 options are required for these chips.
5372 @section @command{ld} and the Intel 960 Family
5374 @cindex i960 support
5376 You can use the @samp{-A@var{architecture}} command line option to
5377 specify one of the two-letter names identifying members of the 960
5378 family; the option specifies the desired output target, and warns of any
5379 incompatible instructions in the input files. It also modifies the
5380 linker's search strategy for archive libraries, to support the use of
5381 libraries specific to each particular architecture, by including in the
5382 search loop names suffixed with the string identifying the architecture.
5384 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5385 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5386 paths, and in any paths you specify with @samp{-L}) for a library with
5399 The first two possibilities would be considered in any event; the last
5400 two are due to the use of @w{@samp{-ACA}}.
5402 You can meaningfully use @samp{-A} more than once on a command line, since
5403 the 960 architecture family allows combination of target architectures; each
5404 use will add another pair of name variants to search for when @w{@samp{-l}}
5405 specifies a library.
5407 @cindex @option{--relax} on i960
5408 @cindex relaxing on i960
5409 @command{ld} supports the @samp{--relax} option for the i960 family. If
5410 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5411 @code{calx} instructions whose targets are within 24 bits, and turns
5412 them into 24-bit program-counter relative @code{bal} and @code{cal}
5413 instructions, respectively. @command{ld} also turns @code{cal}
5414 instructions into @code{bal} instructions when it determines that the
5415 target subroutine is a leaf routine (that is, the target subroutine does
5416 not itself call any subroutines).
5433 @node M68HC11/68HC12
5434 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
5436 @cindex M68HC11 and 68HC12 support
5438 @subsection Linker Relaxation
5440 For the Motorola 68HC11, @command{ld} can perform these global
5441 optimizations when you specify the @samp{--relax} command-line option.
5444 @cindex relaxing on M68HC11
5445 @item relaxing address modes
5446 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5447 targets are within eight bits, and turns them into eight-bit
5448 program-counter relative @code{bsr} and @code{bra} instructions,
5451 @command{ld} also looks at all 16-bit extended addressing modes and
5452 transforms them in a direct addressing mode when the address is in
5453 page 0 (between 0 and 0x0ff).
5455 @item relaxing gcc instruction group
5456 When @command{gcc} is called with @option{-mrelax}, it can emit group
5457 of instructions that the linker can optimize to use a 68HC11 direct
5458 addressing mode. These instructions consists of @code{bclr} or
5459 @code{bset} instructions.
5463 @subsection Trampoline Generation
5465 @cindex trampoline generation on M68HC11
5466 @cindex trampoline generation on M68HC12
5467 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5468 call a far function using a normal @code{jsr} instruction. The linker
5469 will also change the relocation to some far function to use the
5470 trampoline address instead of the function address. This is typically the
5471 case when a pointer to a function is taken. The pointer will in fact
5472 point to the function trampoline.
5475 @kindex --pic-veneer
5476 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
5477 ARM/Thumb interworking veneers, even if the rest of the binary
5478 is not PIC. This avoids problems on uClinux targets where
5479 @samp{--emit-relocs} is used to generate relocatable binaries.
5487 @section @command{ld} and the ARM family
5489 @cindex ARM interworking support
5490 @kindex --support-old-code
5491 For the ARM, @command{ld} will generate code stubs to allow functions calls
5492 between ARM and Thumb code. These stubs only work with code that has
5493 been compiled and assembled with the @samp{-mthumb-interwork} command
5494 line option. If it is necessary to link with old ARM object files or
5495 libraries, which have not been compiled with the -mthumb-interwork
5496 option then the @samp{--support-old-code} command line switch should be
5497 given to the linker. This will make it generate larger stub functions
5498 which will work with non-interworking aware ARM code. Note, however,
5499 the linker does not support generating stubs for function calls to
5500 non-interworking aware Thumb code.
5502 @cindex thumb entry point
5503 @cindex entry point, thumb
5504 @kindex --thumb-entry=@var{entry}
5505 The @samp{--thumb-entry} switch is a duplicate of the generic
5506 @samp{--entry} switch, in that it sets the program's starting address.
5507 But it also sets the bottom bit of the address, so that it can be
5508 branched to using a BX instruction, and the program will start
5509 executing in Thumb mode straight away.
5513 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5514 executables. This option is only valid when linking big-endian objects.
5515 The resulting image will contain big-endian data and little-endian code.
5518 @kindex --target1-rel
5519 @kindex --target1-abs
5520 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5521 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
5522 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
5523 and @samp{--target1-abs} switches override the default.
5526 @kindex --target2=@var{type}
5527 The @samp{--target2=type} switch overrides the default definition of the
5528 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
5529 meanings, and target defaults are as follows:
5532 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5534 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
5536 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5541 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5542 specification) enables objects compiled for the ARMv4 architecture to be
5543 interworking-safe when linked with other objects compiled for ARMv4t, but
5544 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
5546 In the latter case, the switch @option{--fix-v4bx} must be passed to the
5547 linker, which causes v4t @code{BX rM} instructions to be rewritten as
5548 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
5550 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
5551 relocations are ignored.
5555 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
5556 BLX instructions (available on ARMv5t and above) in various
5557 situations. Currently it is used to perform calls via the PLT from Thumb
5558 code using BLX rather than using BX and a mode-switching stub before
5559 each PLT entry. This should lead to such calls executing slightly faster.
5561 This option is enabled implicitly for SymbianOS, so there is no need to
5562 specify it if you are using that target.
5564 @cindex VFP11_DENORM_FIX
5565 @kindex --vfp11-denorm-fix
5566 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
5567 bug in certain VFP11 coprocessor hardware, which sometimes allows
5568 instructions with denorm operands (which must be handled by support code)
5569 to have those operands overwritten by subsequent instructions before
5570 the support code can read the intended values.
5572 The bug may be avoided in scalar mode if you allow at least one
5573 intervening instruction between a VFP11 instruction which uses a register
5574 and another instruction which writes to the same register, or at least two
5575 intervening instructions if vector mode is in use. The bug only affects
5576 full-compliance floating-point mode: you do not need this workaround if
5577 you are using "runfast" mode. Please contact ARM for further details.
5579 If you know you are using buggy VFP11 hardware, you can
5580 enable this workaround by specifying the linker option
5581 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
5582 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
5583 vector mode (the latter also works for scalar code). The default is
5584 @samp{--vfp-denorm-fix=none}.
5586 If the workaround is enabled, instructions are scanned for
5587 potentially-troublesome sequences, and a veneer is created for each
5588 such sequence which may trigger the erratum. The veneer consists of the
5589 first instruction of the sequence and a branch back to the subsequent
5590 instruction. The original instruction is then replaced with a branch to
5591 the veneer. The extra cycles required to call and return from the veneer
5592 are sufficient to avoid the erratum in both the scalar and vector cases.
5594 @cindex NO_ENUM_SIZE_WARNING
5595 @kindex --no-enum-size-warning
5596 The @samp{--no-enum-size-warning} switch prevents the linker from
5597 warning when linking object files that specify incompatible EABI
5598 enumeration size attributes. For example, with this switch enabled,
5599 linking of an object file using 32-bit enumeration values with another
5600 using enumeration values fitted into the smallest possible space will
5614 @section @command{ld} and HPPA 32-bit ELF Support
5615 @cindex HPPA multiple sub-space stubs
5616 @kindex --multi-subspace
5617 When generating a shared library, @command{ld} will by default generate
5618 import stubs suitable for use with a single sub-space application.
5619 The @samp{--multi-subspace} switch causes @command{ld} to generate export
5620 stubs, and different (larger) import stubs suitable for use with
5621 multiple sub-spaces.
5623 @cindex HPPA stub grouping
5624 @kindex --stub-group-size=@var{N}
5625 Long branch stubs and import/export stubs are placed by @command{ld} in
5626 stub sections located between groups of input sections.
5627 @samp{--stub-group-size} specifies the maximum size of a group of input
5628 sections handled by one stub section. Since branch offsets are signed,
5629 a stub section may serve two groups of input sections, one group before
5630 the stub section, and one group after it. However, when using
5631 conditional branches that require stubs, it may be better (for branch
5632 prediction) that stub sections only serve one group of input sections.
5633 A negative value for @samp{N} chooses this scheme, ensuring that
5634 branches to stubs always use a negative offset. Two special values of
5635 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5636 @command{ld} to automatically size input section groups for the branch types
5637 detected, with the same behaviour regarding stub placement as other
5638 positive or negative values of @samp{N} respectively.
5640 Note that @samp{--stub-group-size} does not split input sections. A
5641 single input section larger than the group size specified will of course
5642 create a larger group (of one section). If input sections are too
5643 large, it may not be possible for a branch to reach its stub.
5656 @section @code{ld} and MMIX
5657 For MMIX, there is a choice of generating @code{ELF} object files or
5658 @code{mmo} object files when linking. The simulator @code{mmix}
5659 understands the @code{mmo} format. The binutils @code{objcopy} utility
5660 can translate between the two formats.
5662 There is one special section, the @samp{.MMIX.reg_contents} section.
5663 Contents in this section is assumed to correspond to that of global
5664 registers, and symbols referring to it are translated to special symbols,
5665 equal to registers. In a final link, the start address of the
5666 @samp{.MMIX.reg_contents} section corresponds to the first allocated
5667 global register multiplied by 8. Register @code{$255} is not included in
5668 this section; it is always set to the program entry, which is at the
5669 symbol @code{Main} for @code{mmo} files.
5671 Symbols with the prefix @code{__.MMIX.start.}, for example
5672 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special;
5673 there must be only one each, even if they are local. The default linker
5674 script uses these to set the default start address of a section.
5676 Initial and trailing multiples of zero-valued 32-bit words in a section,
5677 are left out from an mmo file.
5690 @section @code{ld} and MSP430
5691 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
5692 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
5693 just pass @samp{-m help} option to the linker).
5695 @cindex MSP430 extra sections
5696 The linker will recognize some extra sections which are MSP430 specific:
5699 @item @samp{.vectors}
5700 Defines a portion of ROM where interrupt vectors located.
5702 @item @samp{.bootloader}
5703 Defines the bootloader portion of the ROM (if applicable). Any code
5704 in this section will be uploaded to the MPU.
5706 @item @samp{.infomem}
5707 Defines an information memory section (if applicable). Any code in
5708 this section will be uploaded to the MPU.
5710 @item @samp{.infomemnobits}
5711 This is the same as the @samp{.infomem} section except that any code
5712 in this section will not be uploaded to the MPU.
5714 @item @samp{.noinit}
5715 Denotes a portion of RAM located above @samp{.bss} section.
5717 The last two sections are used by gcc.
5731 @section @command{ld} and PowerPC 32-bit ELF Support
5732 @cindex PowerPC long branches
5733 @kindex --relax on PowerPC
5734 Branches on PowerPC processors are limited to a signed 26-bit
5735 displacement, which may result in @command{ld} giving
5736 @samp{relocation truncated to fit} errors with very large programs.
5737 @samp{--relax} enables the generation of trampolines that can access
5738 the entire 32-bit address space. These trampolines are inserted at
5739 section boundaries, so may not themselves be reachable if an input
5740 section exceeds 33M in size.
5742 @cindex PowerPC ELF32 options
5747 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
5748 generates code capable of using a newer PLT and GOT layout that has
5749 the security advantage of no executable section ever needing to be
5750 writable and no writable section ever being executable. PowerPC
5751 @command{ld} will generate this layout, including stubs to access the
5752 PLT, if all input files (including startup and static libraries) were
5753 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
5754 BSS PLT (and GOT layout) which can give slightly better performance.
5756 @kindex --secure-plt
5758 @command{ld} will use the new PLT and GOT layout if it is linking new
5759 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
5760 when linking non-PIC code. This option requests the new PLT and GOT
5761 layout. A warning will be given if some object file requires the old
5767 The new secure PLT and GOT are placed differently relative to other
5768 sections compared to older BSS PLT and GOT placement. The location of
5769 @code{.plt} must change because the new secure PLT is an initialized
5770 section while the old PLT is uninitialized. The reason for the
5771 @code{.got} change is more subtle: The new placement allows
5772 @code{.got} to be read-only in applications linked with
5773 @samp{-z relro -z now}. However, this placement means that
5774 @code{.sdata} cannot always be used in shared libraries, because the
5775 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
5776 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
5777 GCC doesn't use @code{.sdata} in shared libraries, so this option is
5778 really only useful for other compilers that may do so.
5780 @cindex PowerPC stub symbols
5781 @kindex --emit-stub-syms
5782 @item --emit-stub-syms
5783 This option causes @command{ld} to label linker stubs with a local
5784 symbol that encodes the stub type and destination.
5786 @cindex PowerPC TLS optimization
5787 @kindex --no-tls-optimize
5788 @item --no-tls-optimize
5789 PowerPC @command{ld} normally performs some optimization of code
5790 sequences used to access Thread-Local Storage. Use this option to
5791 disable the optimization.
5804 @node PowerPC64 ELF64
5805 @section @command{ld} and PowerPC64 64-bit ELF Support
5807 @cindex PowerPC64 ELF64 options
5809 @cindex PowerPC64 stub grouping
5810 @kindex --stub-group-size
5811 @item --stub-group-size
5812 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
5813 by @command{ld} in stub sections located between groups of input sections.
5814 @samp{--stub-group-size} specifies the maximum size of a group of input
5815 sections handled by one stub section. Since branch offsets are signed,
5816 a stub section may serve two groups of input sections, one group before
5817 the stub section, and one group after it. However, when using
5818 conditional branches that require stubs, it may be better (for branch
5819 prediction) that stub sections only serve one group of input sections.
5820 A negative value for @samp{N} chooses this scheme, ensuring that
5821 branches to stubs always use a negative offset. Two special values of
5822 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5823 @command{ld} to automatically size input section groups for the branch types
5824 detected, with the same behaviour regarding stub placement as other
5825 positive or negative values of @samp{N} respectively.
5827 Note that @samp{--stub-group-size} does not split input sections. A
5828 single input section larger than the group size specified will of course
5829 create a larger group (of one section). If input sections are too
5830 large, it may not be possible for a branch to reach its stub.
5832 @cindex PowerPC64 stub symbols
5833 @kindex --emit-stub-syms
5834 @item --emit-stub-syms
5835 This option causes @command{ld} to label linker stubs with a local
5836 symbol that encodes the stub type and destination.
5838 @cindex PowerPC64 dot symbols
5840 @kindex --no-dotsyms
5841 @item --dotsyms, --no-dotsyms
5842 These two options control how @command{ld} interprets version patterns
5843 in a version script. Older PowerPC64 compilers emitted both a
5844 function descriptor symbol with the same name as the function, and a
5845 code entry symbol with the name prefixed by a dot (@samp{.}). To
5846 properly version a function @samp{foo}, the version script thus needs
5847 to control both @samp{foo} and @samp{.foo}. The option
5848 @samp{--dotsyms}, on by default, automatically adds the required
5849 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
5852 @cindex PowerPC64 TLS optimization
5853 @kindex --no-tls-optimize
5854 @item --no-tls-optimize
5855 PowerPC64 @command{ld} normally performs some optimization of code
5856 sequences used to access Thread-Local Storage. Use this option to
5857 disable the optimization.
5859 @cindex PowerPC64 OPD optimization
5860 @kindex --no-opd-optimize
5861 @item --no-opd-optimize
5862 PowerPC64 @command{ld} normally removes @code{.opd} section entries
5863 corresponding to deleted link-once functions, or functions removed by
5864 the action of @samp{--gc-sections} or linker scrip @code{/DISCARD/}.
5865 Use this option to disable @code{.opd} optimization.
5867 @cindex PowerPC64 OPD spacing
5868 @kindex --non-overlapping-opd
5869 @item --non-overlapping-opd
5870 Some PowerPC64 compilers have an option to generate compressed
5871 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
5872 the static chain pointer (unused in C) with the first word of the next
5873 entry. This option expands such entries to the full 24 bytes.
5875 @cindex PowerPC64 TOC optimization
5876 @kindex --no-toc-optimize
5877 @item --no-toc-optimize
5878 PowerPC64 @command{ld} normally removes unused @code{.toc} section
5879 entries. Such entries are detected by examining relocations that
5880 reference the TOC in code sections. A reloc in a deleted code section
5881 marks a TOC word as unneeded, while a reloc in a kept code section
5882 marks a TOC word as needed. Since the TOC may reference itself, TOC
5883 relocs are also examined. TOC words marked as both needed and
5884 unneeded will of course be kept. TOC words without any referencing
5885 reloc are assumed to be part of a multi-word entry, and are kept or
5886 discarded as per the nearest marked preceding word. This works
5887 reliably for compiler generated code, but may be incorrect if assembly
5888 code is used to insert TOC entries. Use this option to disable the
5891 @cindex PowerPC64 multi-TOC
5892 @kindex --no-multi-toc
5893 @item --no-multi-toc
5894 By default, PowerPC64 GCC generates code for a TOC model where TOC
5895 entries are accessed with a 16-bit offset from r2. This limits the
5896 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
5897 grouping code sections such that each group uses less than 64K for its
5898 TOC entries, then inserts r2 adjusting stubs between inter-group
5899 calls. @command{ld} does not split apart input sections, so cannot
5900 help if a single input file has a @code{.toc} section that exceeds
5901 64K, most likely from linking multiple files with @command{ld -r}.
5902 Use this option to turn off this feature.
5916 @section @command{ld} and SPU ELF Support
5918 @cindex SPU ELF options
5924 This option marks an executable as a PIC plugin module.
5926 @cindex SPU overlays
5927 @kindex --no-overlays
5929 Normally, @command{ld} recognizes calls to functions within overlay
5930 regions, and redirects such calls to an overlay manager via a stub.
5931 @command{ld} also provides a built-in overlay manager. This option
5932 turns off all this special overlay handling.
5934 @cindex SPU overlay stub symbols
5935 @kindex --emit-stub-syms
5936 @item --emit-stub-syms
5937 This option causes @command{ld} to label overlay stubs with a local
5938 symbol that encodes the stub type and destination.
5940 @cindex SPU extra overlay stubs
5941 @kindex --extra-overlay-stubs
5942 @item --extra-overlay-stubs
5943 This option causes @command{ld} to add overlay call stubs on all
5944 function calls out of overlay regions. Normally stubs are not added
5945 on calls to non-overlay regions.
5947 @cindex SPU local store size
5948 @kindex --local-store=lo:hi
5949 @item --local-store=lo:hi
5950 @command{ld} usually checks that a final executable for SPU fits in
5951 the address range 0 to 256k. This option may be used to change the
5952 range. Disable the check entirely with @option{--local-store=0:0}.
5955 @kindex --stack-analysis
5956 @item --stack-analysis
5957 SPU local store space is limited. Over-allocation of stack space
5958 unnecessarily limits space available for code and data, while
5959 under-allocation results in runtime failures. If given this option,
5960 @command{ld} will provide an estimate of maximum stack usage.
5961 @command{ld} does this by examining symbols in code sections to
5962 determine the extents of functions, and looking at function prologues
5963 for stack adjusting instructions. A call-graph is created by looking
5964 for relocations on branch instructions. The graph is then searched
5965 for the maximum stack usage path. Note that this analysis does not
5966 find calls made via function pointers, and does not handle recursion
5967 and other cycles in the call graph. Stack usage may be
5968 under-estimated if your code makes such calls. Also, stack usage for
5969 dynamic allocation, e.g. alloca, will not be detected. If a link map
5970 is requested, detailed information about each function's stack usage
5971 and calls will be given.
5974 @kindex --emit-stack-syms
5975 @item --emit-stack-syms
5976 This option, if given along with @option{--stack-analysis} will result
5977 in @command{ld} emitting stack sizing symbols for each function.
5978 These take the form @code{__stack_<function_name>} for global
5979 functions, and @code{__stack_<number>_<function_name>} for static
5980 functions. @code{<number>} is the section id in hex. The value of
5981 such symbols is the stack requirement for the corresponding function.
5982 The symbol size will be zero, type @code{STT_NOTYPE}, binding
5983 @code{STB_LOCAL}, and section @code{SHN_ABS}.
5997 @section @command{ld}'s Support for Various TI COFF Versions
5998 @cindex TI COFF versions
5999 @kindex --format=@var{version}
6000 The @samp{--format} switch allows selection of one of the various
6001 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
6002 also supported. The TI COFF versions also vary in header byte-order
6003 format; @command{ld} will read any version or byte order, but the output
6004 header format depends on the default specified by the specific target.
6017 @section @command{ld} and WIN32 (cygwin/mingw)
6019 This section describes some of the win32 specific @command{ld} issues.
6020 See @ref{Options,,Command Line Options} for detailed description of the
6021 command line options mentioned here.
6024 @cindex import libraries
6025 @item import libraries
6026 The standard Windows linker creates and uses so-called import
6027 libraries, which contains information for linking to dll's. They are
6028 regular static archives and are handled as any other static
6029 archive. The cygwin and mingw ports of @command{ld} have specific
6030 support for creating such libraries provided with the
6031 @samp{--out-implib} command line option.
6033 @item exporting DLL symbols
6034 @cindex exporting DLL symbols
6035 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6038 @item using auto-export functionality
6039 @cindex using auto-export functionality
6040 By default @command{ld} exports symbols with the auto-export functionality,
6041 which is controlled by the following command line options:
6044 @item --export-all-symbols [This is the default]
6045 @item --exclude-symbols
6046 @item --exclude-libs
6049 If, however, @samp{--export-all-symbols} is not given explicitly on the
6050 command line, then the default auto-export behavior will be @emph{disabled}
6051 if either of the following are true:
6054 @item A DEF file is used.
6055 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
6058 @item using a DEF file
6059 @cindex using a DEF file
6060 Another way of exporting symbols is using a DEF file. A DEF file is
6061 an ASCII file containing definitions of symbols which should be
6062 exported when a dll is created. Usually it is named @samp{<dll
6063 name>.def} and is added as any other object file to the linker's
6064 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
6067 gcc -o <output> <objectfiles> <dll name>.def
6070 Using a DEF file turns off the normal auto-export behavior, unless the
6071 @samp{--export-all-symbols} option is also used.
6073 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
6076 LIBRARY "xyz.dll" BASE=0x20000000
6082 another_foo = abc.dll.afoo
6086 This example defines a DLL with a non-default base address and five
6087 symbols in the export table. The third exported symbol @code{_bar} is an
6088 alias for the second. The fourth symbol, @code{another_foo} is resolved
6089 by "forwarding" to another module and treating it as an alias for
6090 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
6091 @code{var1} is declared to be a data object.
6093 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
6094 name of the output DLL. If @samp{<name>} does not include a suffix,
6095 the default library suffix, @samp{.DLL} is appended.
6097 When the .DEF file is used to build an application, rather than a
6098 library, the @code{NAME <name>} command should be used instead of
6099 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
6100 executable suffix, @samp{.EXE} is appended.
6102 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
6103 specification @code{BASE = <number>} may be used to specify a
6104 non-default base address for the image.
6106 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
6107 or they specify an empty string, the internal name is the same as the
6108 filename specified on the command line.
6110 The complete specification of an export symbol is:
6114 ( ( ( <name1> [ = <name2> ] )
6115 | ( <name1> = <module-name> . <external-name>))
6116 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
6119 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
6120 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
6121 @samp{<name1>} as a "forward" alias for the symbol
6122 @samp{<external-name>} in the DLL @samp{<module-name>}.
6123 Optionally, the symbol may be exported by the specified ordinal
6124 @samp{<integer>} alias.
6126 The optional keywords that follow the declaration indicate:
6128 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
6129 will still be exported by its ordinal alias (either the value specified
6130 by the .def specification or, otherwise, the value assigned by the
6131 linker). The symbol name, however, does remain visible in the import
6132 library (if any), unless @code{PRIVATE} is also specified.
6134 @code{DATA}: The symbol is a variable or object, rather than a function.
6135 The import lib will export only an indirect reference to @code{foo} as
6136 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
6139 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
6140 well as @code{_imp__foo} into the import library. Both refer to the
6141 read-only import address table's pointer to the variable, not to the
6142 variable itself. This can be dangerous. If the user code fails to add
6143 the @code{dllimport} attribute and also fails to explicitly add the
6144 extra indirection that the use of the attribute enforces, the
6145 application will behave unexpectedly.
6147 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
6148 it into the static import library used to resolve imports at link time. The
6149 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
6150 API at runtime or by by using the GNU ld extension of linking directly to
6151 the DLL without an import library.
6153 See ld/deffilep.y in the binutils sources for the full specification of
6154 other DEF file statements
6156 @cindex creating a DEF file
6157 While linking a shared dll, @command{ld} is able to create a DEF file
6158 with the @samp{--output-def <file>} command line option.
6160 @item Using decorations
6161 @cindex Using decorations
6162 Another way of marking symbols for export is to modify the source code
6163 itself, so that when building the DLL each symbol to be exported is
6167 __declspec(dllexport) int a_variable
6168 __declspec(dllexport) void a_function(int with_args)
6171 All such symbols will be exported from the DLL. If, however,
6172 any of the object files in the DLL contain symbols decorated in
6173 this way, then the normal auto-export behavior is disabled, unless
6174 the @samp{--export-all-symbols} option is also used.
6176 Note that object files that wish to access these symbols must @emph{not}
6177 decorate them with dllexport. Instead, they should use dllimport,
6181 __declspec(dllimport) int a_variable
6182 __declspec(dllimport) void a_function(int with_args)
6185 This complicates the structure of library header files, because
6186 when included by the library itself the header must declare the
6187 variables and functions as dllexport, but when included by client
6188 code the header must declare them as dllimport. There are a number
6189 of idioms that are typically used to do this; often client code can
6190 omit the __declspec() declaration completely. See
6191 @samp{--enable-auto-import} and @samp{automatic data imports} for more
6195 @cindex automatic data imports
6196 @item automatic data imports
6197 The standard Windows dll format supports data imports from dlls only
6198 by adding special decorations (dllimport/dllexport), which let the
6199 compiler produce specific assembler instructions to deal with this
6200 issue. This increases the effort necessary to port existing Un*x
6201 code to these platforms, especially for large
6202 c++ libraries and applications. The auto-import feature, which was
6203 initially provided by Paul Sokolovsky, allows one to omit the
6204 decorations to achieve a behavior that conforms to that on POSIX/Un*x
6205 platforms. This feature is enabled with the @samp{--enable-auto-import}
6206 command-line option, although it is enabled by default on cygwin/mingw.
6207 The @samp{--enable-auto-import} option itself now serves mainly to
6208 suppress any warnings that are ordinarily emitted when linked objects
6209 trigger the feature's use.
6211 auto-import of variables does not always work flawlessly without
6212 additional assistance. Sometimes, you will see this message
6214 "variable '<var>' can't be auto-imported. Please read the
6215 documentation for ld's @code{--enable-auto-import} for details."
6217 The @samp{--enable-auto-import} documentation explains why this error
6218 occurs, and several methods that can be used to overcome this difficulty.
6219 One of these methods is the @emph{runtime pseudo-relocs} feature, described
6222 @cindex runtime pseudo-relocation
6223 For complex variables imported from DLLs (such as structs or classes),
6224 object files typically contain a base address for the variable and an
6225 offset (@emph{addend}) within the variable--to specify a particular
6226 field or public member, for instance. Unfortunately, the runtime loader used
6227 in win32 environments is incapable of fixing these references at runtime
6228 without the additional information supplied by dllimport/dllexport decorations.
6229 The standard auto-import feature described above is unable to resolve these
6232 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
6233 be resolved without error, while leaving the task of adjusting the references
6234 themselves (with their non-zero addends) to specialized code provided by the
6235 runtime environment. Recent versions of the cygwin and mingw environments and
6236 compilers provide this runtime support; older versions do not. However, the
6237 support is only necessary on the developer's platform; the compiled result will
6238 run without error on an older system.
6240 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
6243 @cindex direct linking to a dll
6244 @item direct linking to a dll
6245 The cygwin/mingw ports of @command{ld} support the direct linking,
6246 including data symbols, to a dll without the usage of any import
6247 libraries. This is much faster and uses much less memory than does the
6248 traditional import library method, especially when linking large
6249 libraries or applications. When @command{ld} creates an import lib, each
6250 function or variable exported from the dll is stored in its own bfd, even
6251 though a single bfd could contain many exports. The overhead involved in
6252 storing, loading, and processing so many bfd's is quite large, and explains the
6253 tremendous time, memory, and storage needed to link against particularly
6254 large or complex libraries when using import libs.
6256 Linking directly to a dll uses no extra command-line switches other than
6257 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
6258 of names to match each library. All that is needed from the developer's
6259 perspective is an understanding of this search, in order to force ld to
6260 select the dll instead of an import library.
6263 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
6264 to find, in the first directory of its search path,
6276 before moving on to the next directory in the search path.
6278 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
6279 where @samp{<prefix>} is set by the @command{ld} option
6280 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6281 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6284 Other win32-based unix environments, such as mingw or pw32, may use other
6285 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
6286 was originally intended to help avoid name conflicts among dll's built for the
6287 various win32/un*x environments, so that (for example) two versions of a zlib dll
6288 could coexist on the same machine.
6290 The generic cygwin/mingw path layout uses a @samp{bin} directory for
6291 applications and dll's and a @samp{lib} directory for the import
6292 libraries (using cygwin nomenclature):
6298 libxxx.dll.a (in case of dll's)
6299 libxxx.a (in case of static archive)
6302 Linking directly to a dll without using the import library can be
6305 1. Use the dll directly by adding the @samp{bin} path to the link line
6307 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6310 However, as the dll's often have version numbers appended to their names
6311 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6312 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
6313 not versioned, and do not have this difficulty.
6315 2. Create a symbolic link from the dll to a file in the @samp{lib}
6316 directory according to the above mentioned search pattern. This
6317 should be used to avoid unwanted changes in the tools needed for
6321 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6324 Then you can link without any make environment changes.
6327 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6330 This technique also avoids the version number problems, because the following is
6337 libxxx.dll.a -> ../bin/cygxxx-5.dll
6340 Linking directly to a dll without using an import lib will work
6341 even when auto-import features are exercised, and even when
6342 @samp{--enable-runtime-pseudo-relocs} is used.
6344 Given the improvements in speed and memory usage, one might justifiably
6345 wonder why import libraries are used at all. There are three reasons:
6347 1. Until recently, the link-directly-to-dll functionality did @emph{not}
6348 work with auto-imported data.
6350 2. Sometimes it is necessary to include pure static objects within the
6351 import library (which otherwise contains only bfd's for indirection
6352 symbols that point to the exports of a dll). Again, the import lib
6353 for the cygwin kernel makes use of this ability, and it is not
6354 possible to do this without an import lib.
6356 3. Symbol aliases can only be resolved using an import lib. This is
6357 critical when linking against OS-supplied dll's (eg, the win32 API)
6358 in which symbols are usually exported as undecorated aliases of their
6359 stdcall-decorated assembly names.
6361 So, import libs are not going away. But the ability to replace
6362 true import libs with a simple symbolic link to (or a copy of)
6363 a dll, in many cases, is a useful addition to the suite of tools
6364 binutils makes available to the win32 developer. Given the
6365 massive improvements in memory requirements during linking, storage
6366 requirements, and linking speed, we expect that many developers
6367 will soon begin to use this feature whenever possible.
6369 @item symbol aliasing
6371 @item adding additional names
6372 Sometimes, it is useful to export symbols with additional names.
6373 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6374 exported as @samp{_foo} by using special directives in the DEF file
6375 when creating the dll. This will affect also the optional created
6376 import library. Consider the following DEF file:
6379 LIBRARY "xyz.dll" BASE=0x61000000
6386 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6388 Another method for creating a symbol alias is to create it in the
6389 source code using the "weak" attribute:
6392 void foo () @{ /* Do something. */; @}
6393 void _foo () __attribute__ ((weak, alias ("foo")));
6396 See the gcc manual for more information about attributes and weak
6399 @item renaming symbols
6400 Sometimes it is useful to rename exports. For instance, the cygwin
6401 kernel does this regularly. A symbol @samp{_foo} can be exported as
6402 @samp{foo} but not as @samp{_foo} by using special directives in the
6403 DEF file. (This will also affect the import library, if it is
6404 created). In the following example:
6407 LIBRARY "xyz.dll" BASE=0x61000000
6413 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6417 Note: using a DEF file disables the default auto-export behavior,
6418 unless the @samp{--export-all-symbols} command line option is used.
6419 If, however, you are trying to rename symbols, then you should list
6420 @emph{all} desired exports in the DEF file, including the symbols
6421 that are not being renamed, and do @emph{not} use the
6422 @samp{--export-all-symbols} option. If you list only the
6423 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6424 to handle the other symbols, then the both the new names @emph{and}
6425 the original names for the renamed symbols will be exported.
6426 In effect, you'd be aliasing those symbols, not renaming them,
6427 which is probably not what you wanted.
6429 @cindex weak externals
6430 @item weak externals
6431 The Windows object format, PE, specifies a form of weak symbols called
6432 weak externals. When a weak symbol is linked and the symbol is not
6433 defined, the weak symbol becomes an alias for some other symbol. There
6434 are three variants of weak externals:
6436 @item Definition is searched for in objects and libraries, historically
6437 called lazy externals.
6438 @item Definition is searched for only in other objects, not in libraries.
6439 This form is not presently implemented.
6440 @item No search; the symbol is an alias. This form is not presently
6443 As a GNU extension, weak symbols that do not specify an alternate symbol
6444 are supported. If the symbol is undefined when linking, the symbol
6445 uses a default value.
6459 @section @code{ld} and Xtensa Processors
6461 @cindex Xtensa processors
6462 The default @command{ld} behavior for Xtensa processors is to interpret
6463 @code{SECTIONS} commands so that lists of explicitly named sections in a
6464 specification with a wildcard file will be interleaved when necessary to
6465 keep literal pools within the range of PC-relative load offsets. For
6466 example, with the command:
6478 @command{ld} may interleave some of the @code{.literal}
6479 and @code{.text} sections from different object files to ensure that the
6480 literal pools are within the range of PC-relative load offsets. A valid
6481 interleaving might place the @code{.literal} sections from an initial
6482 group of files followed by the @code{.text} sections of that group of
6483 files. Then, the @code{.literal} sections from the rest of the files
6484 and the @code{.text} sections from the rest of the files would follow.
6486 @cindex @option{--relax} on Xtensa
6487 @cindex relaxing on Xtensa
6488 Relaxation is enabled by default for the Xtensa version of @command{ld} and
6489 provides two important link-time optimizations. The first optimization
6490 is to combine identical literal values to reduce code size. A redundant
6491 literal will be removed and all the @code{L32R} instructions that use it
6492 will be changed to reference an identical literal, as long as the
6493 location of the replacement literal is within the offset range of all
6494 the @code{L32R} instructions. The second optimization is to remove
6495 unnecessary overhead from assembler-generated ``longcall'' sequences of
6496 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
6497 range of direct @code{CALL@var{n}} instructions.
6499 For each of these cases where an indirect call sequence can be optimized
6500 to a direct call, the linker will change the @code{CALLX@var{n}}
6501 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
6502 instruction, and remove the literal referenced by the @code{L32R}
6503 instruction if it is not used for anything else. Removing the
6504 @code{L32R} instruction always reduces code size but can potentially
6505 hurt performance by changing the alignment of subsequent branch targets.
6506 By default, the linker will always preserve alignments, either by
6507 switching some instructions between 24-bit encodings and the equivalent
6508 density instructions or by inserting a no-op in place of the @code{L32R}
6509 instruction that was removed. If code size is more important than
6510 performance, the @option{--size-opt} option can be used to prevent the
6511 linker from widening density instructions or inserting no-ops, except in
6512 a few cases where no-ops are required for correctness.
6514 The following Xtensa-specific command-line options can be used to
6517 @cindex Xtensa options
6521 Since the Xtensa version of @code{ld} enables the @option{--relax} option
6522 by default, the @option{--no-relax} option is provided to disable
6526 When optimizing indirect calls to direct calls, optimize for code size
6527 more than performance. With this option, the linker will not insert
6528 no-ops or widen density instructions to preserve branch target
6529 alignment. There may still be some cases where no-ops are required to
6530 preserve the correctness of the code.
6538 @ifclear SingleFormat
6543 @cindex object file management
6544 @cindex object formats available
6546 The linker accesses object and archive files using the BFD libraries.
6547 These libraries allow the linker to use the same routines to operate on
6548 object files whatever the object file format. A different object file
6549 format can be supported simply by creating a new BFD back end and adding
6550 it to the library. To conserve runtime memory, however, the linker and
6551 associated tools are usually configured to support only a subset of the
6552 object file formats available. You can use @code{objdump -i}
6553 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
6554 list all the formats available for your configuration.
6556 @cindex BFD requirements
6557 @cindex requirements for BFD
6558 As with most implementations, BFD is a compromise between
6559 several conflicting requirements. The major factor influencing
6560 BFD design was efficiency: any time used converting between
6561 formats is time which would not have been spent had BFD not
6562 been involved. This is partly offset by abstraction payback; since
6563 BFD simplifies applications and back ends, more time and care
6564 may be spent optimizing algorithms for a greater speed.
6566 One minor artifact of the BFD solution which you should bear in
6567 mind is the potential for information loss. There are two places where
6568 useful information can be lost using the BFD mechanism: during
6569 conversion and during output. @xref{BFD information loss}.
6572 * BFD outline:: How it works: an outline of BFD
6576 @section How It Works: An Outline of BFD
6577 @cindex opening object files
6578 @include bfdsumm.texi
6581 @node Reporting Bugs
6582 @chapter Reporting Bugs
6583 @cindex bugs in @command{ld}
6584 @cindex reporting bugs in @command{ld}
6586 Your bug reports play an essential role in making @command{ld} reliable.
6588 Reporting a bug may help you by bringing a solution to your problem, or
6589 it may not. But in any case the principal function of a bug report is
6590 to help the entire community by making the next version of @command{ld}
6591 work better. Bug reports are your contribution to the maintenance of
6594 In order for a bug report to serve its purpose, you must include the
6595 information that enables us to fix the bug.
6598 * Bug Criteria:: Have you found a bug?
6599 * Bug Reporting:: How to report bugs
6603 @section Have You Found a Bug?
6604 @cindex bug criteria
6606 If you are not sure whether you have found a bug, here are some guidelines:
6609 @cindex fatal signal
6610 @cindex linker crash
6611 @cindex crash of linker
6613 If the linker gets a fatal signal, for any input whatever, that is a
6614 @command{ld} bug. Reliable linkers never crash.
6616 @cindex error on valid input
6618 If @command{ld} produces an error message for valid input, that is a bug.
6620 @cindex invalid input
6622 If @command{ld} does not produce an error message for invalid input, that
6623 may be a bug. In the general case, the linker can not verify that
6624 object files are correct.
6627 If you are an experienced user of linkers, your suggestions for
6628 improvement of @command{ld} are welcome in any case.
6632 @section How to Report Bugs
6634 @cindex @command{ld} bugs, reporting
6636 A number of companies and individuals offer support for @sc{gnu}
6637 products. If you obtained @command{ld} from a support organization, we
6638 recommend you contact that organization first.
6640 You can find contact information for many support companies and
6641 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
6645 Otherwise, send bug reports for @command{ld} to
6649 The fundamental principle of reporting bugs usefully is this:
6650 @strong{report all the facts}. If you are not sure whether to state a
6651 fact or leave it out, state it!
6653 Often people omit facts because they think they know what causes the
6654 problem and assume that some details do not matter. Thus, you might
6655 assume that the name of a symbol you use in an example does not
6656 matter. Well, probably it does not, but one cannot be sure. Perhaps
6657 the bug is a stray memory reference which happens to fetch from the
6658 location where that name is stored in memory; perhaps, if the name
6659 were different, the contents of that location would fool the linker
6660 into doing the right thing despite the bug. Play it safe and give a
6661 specific, complete example. That is the easiest thing for you to do,
6662 and the most helpful.
6664 Keep in mind that the purpose of a bug report is to enable us to fix
6665 the bug if it is new to us. Therefore, always write your bug reports
6666 on the assumption that the bug has not been reported previously.
6668 Sometimes people give a few sketchy facts and ask, ``Does this ring a
6669 bell?'' This cannot help us fix a bug, so it is basically useless. We
6670 respond by asking for enough details to enable us to investigate.
6671 You might as well expedite matters by sending them to begin with.
6673 To enable us to fix the bug, you should include all these things:
6677 The version of @command{ld}. @command{ld} announces it if you start it with
6678 the @samp{--version} argument.
6680 Without this, we will not know whether there is any point in looking for
6681 the bug in the current version of @command{ld}.
6684 Any patches you may have applied to the @command{ld} source, including any
6685 patches made to the @code{BFD} library.
6688 The type of machine you are using, and the operating system name and
6692 What compiler (and its version) was used to compile @command{ld}---e.g.
6696 The command arguments you gave the linker to link your example and
6697 observe the bug. To guarantee you will not omit something important,
6698 list them all. A copy of the Makefile (or the output from make) is
6701 If we were to try to guess the arguments, we would probably guess wrong
6702 and then we might not encounter the bug.
6705 A complete input file, or set of input files, that will reproduce the
6706 bug. It is generally most helpful to send the actual object files
6707 provided that they are reasonably small. Say no more than 10K. For
6708 bigger files you can either make them available by FTP or HTTP or else
6709 state that you are willing to send the object file(s) to whomever
6710 requests them. (Note - your email will be going to a mailing list, so
6711 we do not want to clog it up with large attachments). But small
6712 attachments are best.
6714 If the source files were assembled using @code{gas} or compiled using
6715 @code{gcc}, then it may be OK to send the source files rather than the
6716 object files. In this case, be sure to say exactly what version of
6717 @code{gas} or @code{gcc} was used to produce the object files. Also say
6718 how @code{gas} or @code{gcc} were configured.
6721 A description of what behavior you observe that you believe is
6722 incorrect. For example, ``It gets a fatal signal.''
6724 Of course, if the bug is that @command{ld} gets a fatal signal, then we
6725 will certainly notice it. But if the bug is incorrect output, we might
6726 not notice unless it is glaringly wrong. You might as well not give us
6727 a chance to make a mistake.
6729 Even if the problem you experience is a fatal signal, you should still
6730 say so explicitly. Suppose something strange is going on, such as, your
6731 copy of @command{ld} is out of sync, or you have encountered a bug in the
6732 C library on your system. (This has happened!) Your copy might crash
6733 and ours would not. If you told us to expect a crash, then when ours
6734 fails to crash, we would know that the bug was not happening for us. If
6735 you had not told us to expect a crash, then we would not be able to draw
6736 any conclusion from our observations.
6739 If you wish to suggest changes to the @command{ld} source, send us context
6740 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
6741 @samp{-p} option. Always send diffs from the old file to the new file.
6742 If you even discuss something in the @command{ld} source, refer to it by
6743 context, not by line number.
6745 The line numbers in our development sources will not match those in your
6746 sources. Your line numbers would convey no useful information to us.
6749 Here are some things that are not necessary:
6753 A description of the envelope of the bug.
6755 Often people who encounter a bug spend a lot of time investigating
6756 which changes to the input file will make the bug go away and which
6757 changes will not affect it.
6759 This is often time consuming and not very useful, because the way we
6760 will find the bug is by running a single example under the debugger
6761 with breakpoints, not by pure deduction from a series of examples.
6762 We recommend that you save your time for something else.
6764 Of course, if you can find a simpler example to report @emph{instead}
6765 of the original one, that is a convenience for us. Errors in the
6766 output will be easier to spot, running under the debugger will take
6767 less time, and so on.
6769 However, simplification is not vital; if you do not want to do this,
6770 report the bug anyway and send us the entire test case you used.
6773 A patch for the bug.
6775 A patch for the bug does help us if it is a good one. But do not omit
6776 the necessary information, such as the test case, on the assumption that
6777 a patch is all we need. We might see problems with your patch and decide
6778 to fix the problem another way, or we might not understand it at all.
6780 Sometimes with a program as complicated as @command{ld} it is very hard to
6781 construct an example that will make the program follow a certain path
6782 through the code. If you do not send us the example, we will not be
6783 able to construct one, so we will not be able to verify that the bug is
6786 And if we cannot understand what bug you are trying to fix, or why your
6787 patch should be an improvement, we will not install it. A test case will
6788 help us to understand.
6791 A guess about what the bug is or what it depends on.
6793 Such guesses are usually wrong. Even we cannot guess right about such
6794 things without first using the debugger to find the facts.
6798 @appendix MRI Compatible Script Files
6799 @cindex MRI compatibility
6800 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
6801 linker, @command{ld} can use MRI compatible linker scripts as an
6802 alternative to the more general-purpose linker scripting language
6803 described in @ref{Scripts}. MRI compatible linker scripts have a much
6804 simpler command set than the scripting language otherwise used with
6805 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
6806 linker commands; these commands are described here.
6808 In general, MRI scripts aren't of much use with the @code{a.out} object
6809 file format, since it only has three sections and MRI scripts lack some
6810 features to make use of them.
6812 You can specify a file containing an MRI-compatible script using the
6813 @samp{-c} command-line option.
6815 Each command in an MRI-compatible script occupies its own line; each
6816 command line starts with the keyword that identifies the command (though
6817 blank lines are also allowed for punctuation). If a line of an
6818 MRI-compatible script begins with an unrecognized keyword, @command{ld}
6819 issues a warning message, but continues processing the script.
6821 Lines beginning with @samp{*} are comments.
6823 You can write these commands using all upper-case letters, or all
6824 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
6825 The following list shows only the upper-case form of each command.
6828 @cindex @code{ABSOLUTE} (MRI)
6829 @item ABSOLUTE @var{secname}
6830 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
6831 Normally, @command{ld} includes in the output file all sections from all
6832 the input files. However, in an MRI-compatible script, you can use the
6833 @code{ABSOLUTE} command to restrict the sections that will be present in
6834 your output program. If the @code{ABSOLUTE} command is used at all in a
6835 script, then only the sections named explicitly in @code{ABSOLUTE}
6836 commands will appear in the linker output. You can still use other
6837 input sections (whatever you select on the command line, or using
6838 @code{LOAD}) to resolve addresses in the output file.
6840 @cindex @code{ALIAS} (MRI)
6841 @item ALIAS @var{out-secname}, @var{in-secname}
6842 Use this command to place the data from input section @var{in-secname}
6843 in a section called @var{out-secname} in the linker output file.
6845 @var{in-secname} may be an integer.
6847 @cindex @code{ALIGN} (MRI)
6848 @item ALIGN @var{secname} = @var{expression}
6849 Align the section called @var{secname} to @var{expression}. The
6850 @var{expression} should be a power of two.
6852 @cindex @code{BASE} (MRI)
6853 @item BASE @var{expression}
6854 Use the value of @var{expression} as the lowest address (other than
6855 absolute addresses) in the output file.
6857 @cindex @code{CHIP} (MRI)
6858 @item CHIP @var{expression}
6859 @itemx CHIP @var{expression}, @var{expression}
6860 This command does nothing; it is accepted only for compatibility.
6862 @cindex @code{END} (MRI)
6864 This command does nothing whatever; it's only accepted for compatibility.
6866 @cindex @code{FORMAT} (MRI)
6867 @item FORMAT @var{output-format}
6868 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
6869 language, but restricted to one of these output formats:
6873 S-records, if @var{output-format} is @samp{S}
6876 IEEE, if @var{output-format} is @samp{IEEE}
6879 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
6883 @cindex @code{LIST} (MRI)
6884 @item LIST @var{anything}@dots{}
6885 Print (to the standard output file) a link map, as produced by the
6886 @command{ld} command-line option @samp{-M}.
6888 The keyword @code{LIST} may be followed by anything on the
6889 same line, with no change in its effect.
6891 @cindex @code{LOAD} (MRI)
6892 @item LOAD @var{filename}
6893 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
6894 Include one or more object file @var{filename} in the link; this has the
6895 same effect as specifying @var{filename} directly on the @command{ld}
6898 @cindex @code{NAME} (MRI)
6899 @item NAME @var{output-name}
6900 @var{output-name} is the name for the program produced by @command{ld}; the
6901 MRI-compatible command @code{NAME} is equivalent to the command-line
6902 option @samp{-o} or the general script language command @code{OUTPUT}.
6904 @cindex @code{ORDER} (MRI)
6905 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
6906 @itemx ORDER @var{secname} @var{secname} @var{secname}
6907 Normally, @command{ld} orders the sections in its output file in the
6908 order in which they first appear in the input files. In an MRI-compatible
6909 script, you can override this ordering with the @code{ORDER} command. The
6910 sections you list with @code{ORDER} will appear first in your output
6911 file, in the order specified.
6913 @cindex @code{PUBLIC} (MRI)
6914 @item PUBLIC @var{name}=@var{expression}
6915 @itemx PUBLIC @var{name},@var{expression}
6916 @itemx PUBLIC @var{name} @var{expression}
6917 Supply a value (@var{expression}) for external symbol
6918 @var{name} used in the linker input files.
6920 @cindex @code{SECT} (MRI)
6921 @item SECT @var{secname}, @var{expression}
6922 @itemx SECT @var{secname}=@var{expression}
6923 @itemx SECT @var{secname} @var{expression}
6924 You can use any of these three forms of the @code{SECT} command to
6925 specify the start address (@var{expression}) for section @var{secname}.
6926 If you have more than one @code{SECT} statement for the same
6927 @var{secname}, only the @emph{first} sets the start address.
6933 @unnumbered LD Index
6938 % I think something like @colophon should be in texinfo. In the
6940 \long\def\colophon{\hbox to0pt{}\vfill
6941 \centerline{The body of this manual is set in}
6942 \centerline{\fontname\tenrm,}
6943 \centerline{with headings in {\bf\fontname\tenbf}}
6944 \centerline{and examples in {\tt\fontname\tentt}.}
6945 \centerline{{\it\fontname\tenit\/} and}
6946 \centerline{{\sl\fontname\tensl\/}}
6947 \centerline{are used for emphasis.}\vfill}
6949 % Blame: doc@cygnus.com, 28mar91.