3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
7 @include configdoc.texi
8 @c (configdoc.texi is generated by the Makefile)
14 @macro gcctabopt{body}
20 @c Configure for the generation of man pages
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. Specifying a
301 script in this way merely augments the main linker script, with the
302 extra commands placed after the main script; use the @samp{-T} option
303 to replace the default linker script entirely, but note the effect of
304 the @code{INSERT} command. @xref{Scripts}.
306 For options whose names are a single letter,
307 option arguments must either follow the option letter without intervening
308 whitespace, or be given as separate arguments immediately following the
309 option that requires them.
311 For options whose names are multiple letters, either one dash or two can
312 precede the option name; for example, @samp{-trace-symbol} and
313 @samp{--trace-symbol} are equivalent. Note---there is one exception to
314 this rule. Multiple letter options that start with a lower case 'o' can
315 only be preceded by two dashes. This is to reduce confusion with the
316 @samp{-o} option. So for example @samp{-omagic} sets the output file
317 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
320 Arguments to multiple-letter options must either be separated from the
321 option name by an equals sign, or be given as separate arguments
322 immediately following the option that requires them. For example,
323 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
324 Unique abbreviations of the names of multiple-letter options are
327 Note---if the linker is being invoked indirectly, via a compiler driver
328 (e.g. @samp{gcc}) then all the linker command line options should be
329 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
330 compiler driver) like this:
333 gcc -Wl,--startgroup foo.o bar.o -Wl,--endgroup
336 This is important, because otherwise the compiler driver program may
337 silently drop the linker options, resulting in a bad link.
339 Here is a table of the generic command line switches accepted by the GNU
343 @include at-file.texi
345 @kindex -a@var{keyword}
346 @item -a@var{keyword}
347 This option is supported for HP/UX compatibility. The @var{keyword}
348 argument must be one of the strings @samp{archive}, @samp{shared}, or
349 @samp{default}. @samp{-aarchive} is functionally equivalent to
350 @samp{-Bstatic}, and the other two keywords are functionally equivalent
351 to @samp{-Bdynamic}. This option may be used any number of times.
354 @cindex architectures
356 @item -A@var{architecture}
357 @kindex --architecture=@var{arch}
358 @itemx --architecture=@var{architecture}
359 In the current release of @command{ld}, this option is useful only for the
360 Intel 960 family of architectures. In that @command{ld} configuration, the
361 @var{architecture} argument identifies the particular architecture in
362 the 960 family, enabling some safeguards and modifying the
363 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
364 family}, for details.
366 Future releases of @command{ld} may support similar functionality for
367 other architecture families.
370 @ifclear SingleFormat
371 @cindex binary input format
372 @kindex -b @var{format}
373 @kindex --format=@var{format}
376 @item -b @var{input-format}
377 @itemx --format=@var{input-format}
378 @command{ld} may be configured to support more than one kind of object
379 file. If your @command{ld} is configured this way, you can use the
380 @samp{-b} option to specify the binary format for input object files
381 that follow this option on the command line. Even when @command{ld} is
382 configured to support alternative object formats, you don't usually need
383 to specify this, as @command{ld} should be configured to expect as a
384 default input format the most usual format on each machine.
385 @var{input-format} is a text string, the name of a particular format
386 supported by the BFD libraries. (You can list the available binary
387 formats with @samp{objdump -i}.)
390 You may want to use this option if you are linking files with an unusual
391 binary format. You can also use @samp{-b} to switch formats explicitly (when
392 linking object files of different formats), by including
393 @samp{-b @var{input-format}} before each group of object files in a
396 The default format is taken from the environment variable
401 You can also define the input format from a script, using the command
404 see @ref{Format Commands}.
408 @kindex -c @var{MRI-cmdfile}
409 @kindex --mri-script=@var{MRI-cmdfile}
410 @cindex compatibility, MRI
411 @item -c @var{MRI-commandfile}
412 @itemx --mri-script=@var{MRI-commandfile}
413 For compatibility with linkers produced by MRI, @command{ld} accepts script
414 files written in an alternate, restricted command language, described in
416 @ref{MRI,,MRI Compatible Script Files}.
419 the MRI Compatible Script Files section of GNU ld documentation.
421 Introduce MRI script files with
422 the option @samp{-c}; use the @samp{-T} option to run linker
423 scripts written in the general-purpose @command{ld} scripting language.
424 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
425 specified by any @samp{-L} options.
427 @cindex common allocation
434 These three options are equivalent; multiple forms are supported for
435 compatibility with other linkers. They assign space to common symbols
436 even if a relocatable output file is specified (with @samp{-r}). The
437 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
438 @xref{Miscellaneous Commands}.
440 @cindex entry point, from command line
441 @kindex -e @var{entry}
442 @kindex --entry=@var{entry}
444 @itemx --entry=@var{entry}
445 Use @var{entry} as the explicit symbol for beginning execution of your
446 program, rather than the default entry point. If there is no symbol
447 named @var{entry}, the linker will try to parse @var{entry} as a number,
448 and use that as the entry address (the number will be interpreted in
449 base 10; you may use a leading @samp{0x} for base 16, or a leading
450 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
451 and other ways of specifying the entry point.
453 @kindex --exclude-libs
454 @item --exclude-libs @var{lib},@var{lib},...
455 Specifies a list of archive libraries from which symbols should not be automatically
456 exported. The library names may be delimited by commas or colons. Specifying
457 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
458 automatic export. This option is available only for the i386 PE targeted
459 port of the linker and for ELF targeted ports. For i386 PE, symbols
460 explicitly listed in a .def file are still exported, regardless of this
461 option. For ELF targeted ports, symbols affected by this option will
462 be treated as hidden.
464 @cindex dynamic symbol table
466 @kindex --export-dynamic
468 @itemx --export-dynamic
469 When creating a dynamically linked executable, add all symbols to the
470 dynamic symbol table. The dynamic symbol table is the set of symbols
471 which are visible from dynamic objects at run time.
473 If you do not use this option, the dynamic symbol table will normally
474 contain only those symbols which are referenced by some dynamic object
475 mentioned in the link.
477 If you use @code{dlopen} to load a dynamic object which needs to refer
478 back to the symbols defined by the program, rather than some other
479 dynamic object, then you will probably need to use this option when
480 linking the program itself.
482 You can also use the dynamic list to control what symbols should
483 be added to the dynamic symbol table if the output format supports it.
484 See the description of @samp{--dynamic-list}.
486 @ifclear SingleFormat
487 @cindex big-endian objects
491 Link big-endian objects. This affects the default output format.
493 @cindex little-endian objects
496 Link little-endian objects. This affects the default output format.
502 @itemx --auxiliary @var{name}
503 When creating an ELF shared object, set the internal DT_AUXILIARY field
504 to the specified name. This tells the dynamic linker that the symbol
505 table of the shared object should be used as an auxiliary filter on the
506 symbol table of the shared object @var{name}.
508 If you later link a program against this filter object, then, when you
509 run the program, the dynamic linker will see the DT_AUXILIARY field. If
510 the dynamic linker resolves any symbols from the filter object, it will
511 first check whether there is a definition in the shared object
512 @var{name}. If there is one, it will be used instead of the definition
513 in the filter object. The shared object @var{name} need not exist.
514 Thus the shared object @var{name} may be used to provide an alternative
515 implementation of certain functions, perhaps for debugging or for
516 machine specific performance.
518 This option may be specified more than once. The DT_AUXILIARY entries
519 will be created in the order in which they appear on the command line.
524 @itemx --filter @var{name}
525 When creating an ELF shared object, set the internal DT_FILTER field to
526 the specified name. This tells the dynamic linker that the symbol table
527 of the shared object which is being created should be used as a filter
528 on the symbol table of the shared object @var{name}.
530 If you later link a program against this filter object, then, when you
531 run the program, the dynamic linker will see the DT_FILTER field. The
532 dynamic linker will resolve symbols according to the symbol table of the
533 filter object as usual, but it will actually link to the definitions
534 found in the shared object @var{name}. Thus the filter object can be
535 used to select a subset of the symbols provided by the object
538 Some older linkers used the @option{-F} option throughout a compilation
539 toolchain for specifying object-file format for both input and output
541 @ifclear SingleFormat
542 The @sc{gnu} linker uses other mechanisms for this purpose: the
543 @option{-b}, @option{--format}, @option{--oformat} options, the
544 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
545 environment variable.
547 The @sc{gnu} linker will ignore the @option{-F} option when not
548 creating an ELF shared object.
550 @cindex finalization function
552 @item -fini @var{name}
553 When creating an ELF executable or shared object, call NAME when the
554 executable or shared object is unloaded, by setting DT_FINI to the
555 address of the function. By default, the linker uses @code{_fini} as
556 the function to call.
560 Ignored. Provided for compatibility with other tools.
566 @itemx --gpsize=@var{value}
567 Set the maximum size of objects to be optimized using the GP register to
568 @var{size}. This is only meaningful for object file formats such as
569 MIPS ECOFF which supports putting large and small objects into different
570 sections. This is ignored for other object file formats.
572 @cindex runtime library name
574 @kindex -soname=@var{name}
576 @itemx -soname=@var{name}
577 When creating an ELF shared object, set the internal DT_SONAME field to
578 the specified name. When an executable is linked with a shared object
579 which has a DT_SONAME field, then when the executable is run the dynamic
580 linker will attempt to load the shared object specified by the DT_SONAME
581 field rather than the using the file name given to the linker.
584 @cindex incremental link
586 Perform an incremental link (same as option @samp{-r}).
588 @cindex initialization function
590 @item -init @var{name}
591 When creating an ELF executable or shared object, call NAME when the
592 executable or shared object is loaded, by setting DT_INIT to the address
593 of the function. By default, the linker uses @code{_init} as the
596 @cindex archive files, from cmd line
597 @kindex -l@var{namespec}
598 @kindex --library=@var{namespec}
599 @item -l@var{namespec}
600 @itemx --library=@var{namespec}
601 Add the archive or object file specified by @var{namespec} to the
602 list of files to link. This option may be used any number of times.
603 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
604 will search the library path for a file called @var{filename}, otherise it
605 will search the library path for a file called @file{lib@var{namespec}.a}.
607 On systems which support shared libraries, @command{ld} may also search for
608 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
609 and SunOS systems, @command{ld} will search a directory for a library
610 called @file{lib@var{namespec}.so} before searching for one called
611 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
612 indicates a shared library.) Note that this behavior does not apply
613 to @file{:@var{filename}}, which always specifies a file called
616 The linker will search an archive only once, at the location where it is
617 specified on the command line. If the archive defines a symbol which
618 was undefined in some object which appeared before the archive on the
619 command line, the linker will include the appropriate file(s) from the
620 archive. However, an undefined symbol in an object appearing later on
621 the command line will not cause the linker to search the archive again.
623 See the @option{-(} option for a way to force the linker to search
624 archives multiple times.
626 You may list the same archive multiple times on the command line.
629 This type of archive searching is standard for Unix linkers. However,
630 if you are using @command{ld} on AIX, note that it is different from the
631 behaviour of the AIX linker.
634 @cindex search directory, from cmd line
636 @kindex --library-path=@var{dir}
637 @item -L@var{searchdir}
638 @itemx --library-path=@var{searchdir}
639 Add path @var{searchdir} to the list of paths that @command{ld} will search
640 for archive libraries and @command{ld} control scripts. You may use this
641 option any number of times. The directories are searched in the order
642 in which they are specified on the command line. Directories specified
643 on the command line are searched before the default directories. All
644 @option{-L} options apply to all @option{-l} options, regardless of the
645 order in which the options appear.
647 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
648 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
651 The default set of paths searched (without being specified with
652 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
653 some cases also on how it was configured. @xref{Environment}.
656 The paths can also be specified in a link script with the
657 @code{SEARCH_DIR} command. Directories specified this way are searched
658 at the point in which the linker script appears in the command line.
661 @kindex -m @var{emulation}
662 @item -m@var{emulation}
663 Emulate the @var{emulation} linker. You can list the available
664 emulations with the @samp{--verbose} or @samp{-V} options.
666 If the @samp{-m} option is not used, the emulation is taken from the
667 @code{LDEMULATION} environment variable, if that is defined.
669 Otherwise, the default emulation depends upon how the linker was
677 Print a link map to the standard output. A link map provides
678 information about the link, including the following:
682 Where object files are mapped into memory.
684 How common symbols are allocated.
686 All archive members included in the link, with a mention of the symbol
687 which caused the archive member to be brought in.
689 The values assigned to symbols.
691 Note - symbols whose values are computed by an expression which
692 involves a reference to a previous value of the same symbol may not
693 have correct result displayed in the link map. This is because the
694 linker discards intermediate results and only retains the final value
695 of an expression. Under such circumstances the linker will display
696 the final value enclosed by square brackets. Thus for example a
697 linker script containing:
705 will produce the following output in the link map if the @option{-M}
710 [0x0000000c] foo = (foo * 0x4)
711 [0x0000000c] foo = (foo + 0x8)
714 See @ref{Expressions} for more information about expressions in linker
719 @cindex read-only text
724 Turn off page alignment of sections, and mark the output as
725 @code{NMAGIC} if possible.
729 @cindex read/write from cmd line
733 Set the text and data sections to be readable and writable. Also, do
734 not page-align the data segment, and disable linking against shared
735 libraries. If the output format supports Unix style magic numbers,
736 mark the output as @code{OMAGIC}. Note: Although a writable text section
737 is allowed for PE-COFF targets, it does not conform to the format
738 specification published by Microsoft.
743 This option negates most of the effects of the @option{-N} option. It
744 sets the text section to be read-only, and forces the data segment to
745 be page-aligned. Note - this option does not enable linking against
746 shared libraries. Use @option{-Bdynamic} for this.
748 @kindex -o @var{output}
749 @kindex --output=@var{output}
750 @cindex naming the output file
751 @item -o @var{output}
752 @itemx --output=@var{output}
753 Use @var{output} as the name for the program produced by @command{ld}; if this
754 option is not specified, the name @file{a.out} is used by default. The
755 script command @code{OUTPUT} can also specify the output file name.
757 @kindex -O @var{level}
758 @cindex generating optimized output
760 If @var{level} is a numeric values greater than zero @command{ld} optimizes
761 the output. This might take significantly longer and therefore probably
762 should only be enabled for the final binary. At the moment this
763 option only affects ELF shared library generation. Future releases of
764 the linker may make more use of this option. Also currently there is
765 no difference in the linker's behaviour for different non-zero values
766 of this option. Again this may change with future releases.
769 @kindex --emit-relocs
770 @cindex retain relocations in final executable
773 Leave relocation sections and contents in fully linked executables.
774 Post link analysis and optimization tools may need this information in
775 order to perform correct modifications of executables. This results
776 in larger executables.
778 This option is currently only supported on ELF platforms.
780 @kindex --force-dynamic
781 @cindex forcing the creation of dynamic sections
782 @item --force-dynamic
783 Force the output file to have dynamic sections. This option is specific
787 @cindex relocatable output
789 @kindex --relocatable
792 Generate relocatable output---i.e., generate an output file that can in
793 turn serve as input to @command{ld}. This is often called @dfn{partial
794 linking}. As a side effect, in environments that support standard Unix
795 magic numbers, this option also sets the output file's magic number to
797 @c ; see @option{-N}.
798 If this option is not specified, an absolute file is produced. When
799 linking C++ programs, this option @emph{will not} resolve references to
800 constructors; to do that, use @samp{-Ur}.
802 When an input file does not have the same format as the output file,
803 partial linking is only supported if that input file does not contain any
804 relocations. Different output formats can have further restrictions; for
805 example some @code{a.out}-based formats do not support partial linking
806 with input files in other formats at all.
808 This option does the same thing as @samp{-i}.
810 @kindex -R @var{file}
811 @kindex --just-symbols=@var{file}
812 @cindex symbol-only input
813 @item -R @var{filename}
814 @itemx --just-symbols=@var{filename}
815 Read symbol names and their addresses from @var{filename}, but do not
816 relocate it or include it in the output. This allows your output file
817 to refer symbolically to absolute locations of memory defined in other
818 programs. You may use this option more than once.
820 For compatibility with other ELF linkers, if the @option{-R} option is
821 followed by a directory name, rather than a file name, it is treated as
822 the @option{-rpath} option.
826 @cindex strip all symbols
829 Omit all symbol information from the output file.
832 @kindex --strip-debug
833 @cindex strip debugger symbols
836 Omit debugger symbol information (but not all symbols) from the output file.
840 @cindex input files, displaying
843 Print the names of the input files as @command{ld} processes them.
845 @kindex -T @var{script}
846 @kindex --script=@var{script}
848 @item -T @var{scriptfile}
849 @itemx --script=@var{scriptfile}
850 Use @var{scriptfile} as the linker script. This script replaces
851 @command{ld}'s default linker script (rather than adding to it), so
852 @var{commandfile} must specify everything necessary to describe the
853 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
854 the current directory, @code{ld} looks for it in the directories
855 specified by any preceding @samp{-L} options. Multiple @samp{-T}
858 @kindex -dT @var{script}
859 @kindex --default-script=@var{script}
861 @item -dT @var{scriptfile}
862 @itemx --default-script=@var{scriptfile}
863 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
865 This option is similar to the @option{--script} option except that
866 processing of the script is delayed until after the rest of the
867 command line has been processed. This allows options placed after the
868 @option{--default-script} option on the command line to affect the
869 behaviour of the linker script, which can be important when the linker
870 command line cannot be directly controlled by the user. (eg because
871 the command line is being constructed by another tool, such as
874 @kindex -u @var{symbol}
875 @kindex --undefined=@var{symbol}
876 @cindex undefined symbol
877 @item -u @var{symbol}
878 @itemx --undefined=@var{symbol}
879 Force @var{symbol} to be entered in the output file as an undefined
880 symbol. Doing this may, for example, trigger linking of additional
881 modules from standard libraries. @samp{-u} may be repeated with
882 different option arguments to enter additional undefined symbols. This
883 option is equivalent to the @code{EXTERN} linker script command.
888 For anything other than C++ programs, this option is equivalent to
889 @samp{-r}: it generates relocatable output---i.e., an output file that can in
890 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
891 @emph{does} resolve references to constructors, unlike @samp{-r}.
892 It does not work to use @samp{-Ur} on files that were themselves linked
893 with @samp{-Ur}; once the constructor table has been built, it cannot
894 be added to. Use @samp{-Ur} only for the last partial link, and
895 @samp{-r} for the others.
897 @kindex --unique[=@var{SECTION}]
898 @item --unique[=@var{SECTION}]
899 Creates a separate output section for every input section matching
900 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
901 missing, for every orphan input section. An orphan section is one not
902 specifically mentioned in a linker script. You may use this option
903 multiple times on the command line; It prevents the normal merging of
904 input sections with the same name, overriding output section assignments
914 Display the version number for @command{ld}. The @option{-V} option also
915 lists the supported emulations.
918 @kindex --discard-all
919 @cindex deleting local symbols
922 Delete all local symbols.
925 @kindex --discard-locals
926 @cindex local symbols, deleting
928 @itemx --discard-locals
929 Delete all temporary local symbols. (These symbols start with
930 system-specific local label prefixes, typically @samp{.L} for ELF systems
931 or @samp{L} for traditional a.out systems.)
933 @kindex -y @var{symbol}
934 @kindex --trace-symbol=@var{symbol}
935 @cindex symbol tracing
936 @item -y @var{symbol}
937 @itemx --trace-symbol=@var{symbol}
938 Print the name of each linked file in which @var{symbol} appears. This
939 option may be given any number of times. On many systems it is necessary
940 to prepend an underscore.
942 This option is useful when you have an undefined symbol in your link but
943 don't know where the reference is coming from.
945 @kindex -Y @var{path}
947 Add @var{path} to the default library search path. This option exists
948 for Solaris compatibility.
950 @kindex -z @var{keyword}
951 @item -z @var{keyword}
952 The recognized keywords are:
956 Combines multiple reloc sections and sorts them to make dynamic symbol
957 lookup caching possible.
960 Disallows undefined symbols in object files. Undefined symbols in
961 shared libraries are still allowed.
964 Marks the object as requiring executable stack.
967 This option is only meaningful when building a shared object.
968 It marks the object so that its runtime initialization will occur
969 before the runtime initialization of any other objects brought into
970 the process at the same time. Similarly the runtime finalization of
971 the object will occur after the runtime finalization of any other
975 Marks the object that its symbol table interposes before all symbols
976 but the primary executable.
979 When generating an executable or shared library, mark it to tell the
980 dynamic linker to defer function call resolution to the point when
981 the function is called (lazy binding), rather than at load time.
982 Lazy binding is the default.
985 Marks the object that its filters be processed immediately at
989 Allows multiple definitions.
992 Disables multiple reloc sections combining.
995 Disables production of copy relocs.
998 Marks the object that the search for dependencies of this object will
999 ignore any default library search paths.
1002 Marks the object shouldn't be unloaded at runtime.
1005 Marks the object not available to @code{dlopen}.
1008 Marks the object can not be dumped by @code{dldump}.
1011 Marks the object as not requiring executable stack.
1014 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1017 When generating an executable or shared library, mark it to tell the
1018 dynamic linker to resolve all symbols when the program is started, or
1019 when the shared library is linked to using dlopen, instead of
1020 deferring function call resolution to the point when the function is
1024 Marks the object may contain $ORIGIN.
1027 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1029 @item max-page-size=@var{value}
1030 Set the emulation maximum page size to @var{value}.
1032 @item common-page-size=@var{value}
1033 Set the emulation common page size to @var{value}.
1037 Other keywords are ignored for Solaris compatibility.
1040 @cindex groups of archives
1041 @item -( @var{archives} -)
1042 @itemx --start-group @var{archives} --end-group
1043 The @var{archives} should be a list of archive files. They may be
1044 either explicit file names, or @samp{-l} options.
1046 The specified archives are searched repeatedly until no new undefined
1047 references are created. Normally, an archive is searched only once in
1048 the order that it is specified on the command line. If a symbol in that
1049 archive is needed to resolve an undefined symbol referred to by an
1050 object in an archive that appears later on the command line, the linker
1051 would not be able to resolve that reference. By grouping the archives,
1052 they all be searched repeatedly until all possible references are
1055 Using this option has a significant performance cost. It is best to use
1056 it only when there are unavoidable circular references between two or
1059 @kindex --accept-unknown-input-arch
1060 @kindex --no-accept-unknown-input-arch
1061 @item --accept-unknown-input-arch
1062 @itemx --no-accept-unknown-input-arch
1063 Tells the linker to accept input files whose architecture cannot be
1064 recognised. The assumption is that the user knows what they are doing
1065 and deliberately wants to link in these unknown input files. This was
1066 the default behaviour of the linker, before release 2.14. The default
1067 behaviour from release 2.14 onwards is to reject such input files, and
1068 so the @samp{--accept-unknown-input-arch} option has been added to
1069 restore the old behaviour.
1072 @kindex --no-as-needed
1074 @itemx --no-as-needed
1075 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1076 on the command line after the @option{--as-needed} option. Normally,
1077 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1078 on the command line, regardless of whether the library is actually
1079 needed. @option{--as-needed} causes DT_NEEDED tags to only be emitted
1080 for libraries that satisfy some symbol reference from regular objects
1081 which is undefined at the point that the library was linked.
1082 @option{--no-as-needed} restores the default behaviour.
1084 @kindex --add-needed
1085 @kindex --no-add-needed
1087 @itemx --no-add-needed
1088 This option affects the treatment of dynamic libraries from ELF
1089 DT_NEEDED tags in dynamic libraries mentioned on the command line after
1090 the @option{--no-add-needed} option. Normally, the linker will add
1091 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1092 @option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1093 for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1094 the default behaviour.
1096 @kindex -assert @var{keyword}
1097 @item -assert @var{keyword}
1098 This option is ignored for SunOS compatibility.
1102 @kindex -call_shared
1106 Link against dynamic libraries. This is only meaningful on platforms
1107 for which shared libraries are supported. This option is normally the
1108 default on such platforms. The different variants of this option are
1109 for compatibility with various systems. You may use this option
1110 multiple times on the command line: it affects library searching for
1111 @option{-l} options which follow it.
1115 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1116 section. This causes the runtime linker to handle lookups in this
1117 object and its dependencies to be performed only inside the group.
1118 @option{--unresolved-symbols=report-all} is implied. This option is
1119 only meaningful on ELF platforms which support shared libraries.
1129 Do not link against shared libraries. This is only meaningful on
1130 platforms for which shared libraries are supported. The different
1131 variants of this option are for compatibility with various systems. You
1132 may use this option multiple times on the command line: it affects
1133 library searching for @option{-l} options which follow it. This
1134 option also implies @option{--unresolved-symbols=report-all}. This
1135 option can be used with @option{-shared}. Doing so means that a
1136 shared library is being created but that all of the library's external
1137 references must be resolved by pulling in entries from static
1142 When creating a shared library, bind references to global symbols to the
1143 definition within the shared library, if any. Normally, it is possible
1144 for a program linked against a shared library to override the definition
1145 within the shared library. This option is only meaningful on ELF
1146 platforms which support shared libraries.
1148 @kindex -Bsymbolic-functions
1149 @item -Bsymbolic-functions
1150 When creating a shared library, bind references to global function
1151 symbols to the definition within the shared library, if any.
1152 This option is only meaningful on ELF platforms which support shared
1155 @kindex --dynamic-list=@var{dynamic-list-file}
1156 @item --dynamic-list=@var{dynamic-list-file}
1157 Specify the name of a dynamic list file to the linker. This is
1158 typically used when creating shared libraries to specify a list of
1159 global symbols whose references shouldn't be bound to the definition
1160 within the shared library, or creating dynamically linked executables
1161 to specify a list of symbols which should be added to the symbol table
1162 in the executable. This option is only meaningful on ELF platforms
1163 which support shared libraries.
1165 The format of the dynamic list is the same as the version node without
1166 scope and node name. See @ref{VERSION} for more information.
1168 @kindex --dynamic-list-data
1169 @item --dynamic-list-data
1170 Include all global data symbols to the dynamic list.
1172 @kindex --dynamic-list-cpp-new
1173 @item --dynamic-list-cpp-new
1174 Provide the builtin dynamic list for C++ operator new and delete. It
1175 is mainly useful for building shared libstdc++.
1177 @kindex --dynamic-list-cpp-typeinfo
1178 @item --dynamic-list-cpp-typeinfo
1179 Provide the builtin dynamic list for C++ runtime type identification.
1181 @kindex --check-sections
1182 @kindex --no-check-sections
1183 @item --check-sections
1184 @itemx --no-check-sections
1185 Asks the linker @emph{not} to check section addresses after they have
1186 been assigned to see if there are any overlaps. Normally the linker will
1187 perform this check, and if it finds any overlaps it will produce
1188 suitable error messages. The linker does know about, and does make
1189 allowances for sections in overlays. The default behaviour can be
1190 restored by using the command line switch @option{--check-sections}.
1192 @cindex cross reference table
1195 Output a cross reference table. If a linker map file is being
1196 generated, the cross reference table is printed to the map file.
1197 Otherwise, it is printed on the standard output.
1199 The format of the table is intentionally simple, so that it may be
1200 easily processed by a script if necessary. The symbols are printed out,
1201 sorted by name. For each symbol, a list of file names is given. If the
1202 symbol is defined, the first file listed is the location of the
1203 definition. The remaining files contain references to the symbol.
1205 @cindex common allocation
1206 @kindex --no-define-common
1207 @item --no-define-common
1208 This option inhibits the assignment of addresses to common symbols.
1209 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1210 @xref{Miscellaneous Commands}.
1212 The @samp{--no-define-common} option allows decoupling
1213 the decision to assign addresses to Common symbols from the choice
1214 of the output file type; otherwise a non-Relocatable output type
1215 forces assigning addresses to Common symbols.
1216 Using @samp{--no-define-common} allows Common symbols that are referenced
1217 from a shared library to be assigned addresses only in the main program.
1218 This eliminates the unused duplicate space in the shared library,
1219 and also prevents any possible confusion over resolving to the wrong
1220 duplicate when there are many dynamic modules with specialized search
1221 paths for runtime symbol resolution.
1223 @cindex symbols, from command line
1224 @kindex --defsym @var{symbol}=@var{exp}
1225 @item --defsym @var{symbol}=@var{expression}
1226 Create a global symbol in the output file, containing the absolute
1227 address given by @var{expression}. You may use this option as many
1228 times as necessary to define multiple symbols in the command line. A
1229 limited form of arithmetic is supported for the @var{expression} in this
1230 context: you may give a hexadecimal constant or the name of an existing
1231 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1232 constants or symbols. If you need more elaborate expressions, consider
1233 using the linker command language from a script (@pxref{Assignments,,
1234 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1235 space between @var{symbol}, the equals sign (``@key{=}''), and
1238 @cindex demangling, from command line
1239 @kindex --demangle[=@var{style}]
1240 @kindex --no-demangle
1241 @item --demangle[=@var{style}]
1242 @itemx --no-demangle
1243 These options control whether to demangle symbol names in error messages
1244 and other output. When the linker is told to demangle, it tries to
1245 present symbol names in a readable fashion: it strips leading
1246 underscores if they are used by the object file format, and converts C++
1247 mangled symbol names into user readable names. Different compilers have
1248 different mangling styles. The optional demangling style argument can be used
1249 to choose an appropriate demangling style for your compiler. The linker will
1250 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1251 is set. These options may be used to override the default.
1253 @cindex dynamic linker, from command line
1254 @kindex -I@var{file}
1255 @kindex --dynamic-linker @var{file}
1256 @item --dynamic-linker @var{file}
1257 Set the name of the dynamic linker. This is only meaningful when
1258 generating dynamically linked ELF executables. The default dynamic
1259 linker is normally correct; don't use this unless you know what you are
1263 @kindex --fatal-warnings
1264 @item --fatal-warnings
1265 Treat all warnings as errors.
1267 @kindex --force-exe-suffix
1268 @item --force-exe-suffix
1269 Make sure that an output file has a .exe suffix.
1271 If a successfully built fully linked output file does not have a
1272 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1273 the output file to one of the same name with a @code{.exe} suffix. This
1274 option is useful when using unmodified Unix makefiles on a Microsoft
1275 Windows host, since some versions of Windows won't run an image unless
1276 it ends in a @code{.exe} suffix.
1278 @kindex --gc-sections
1279 @kindex --no-gc-sections
1280 @cindex garbage collection
1282 @itemx --no-gc-sections
1283 Enable garbage collection of unused input sections. It is ignored on
1284 targets that do not support this option. 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 This option can be set when doing a partial link (enabled with option
1299 @samp{-r}). In this case the root of symbols kept must be explicitely
1300 specified either by an @samp{--entry} or @samp{--undefined} option or by
1301 a @code{ENTRY} command in the linker script.
1303 @kindex --print-gc-sections
1304 @kindex --no-print-gc-sections
1305 @cindex garbage collection
1306 @item --print-gc-sections
1307 @itemx --no-print-gc-sections
1308 List all sections removed by garbage collection. The listing is
1309 printed on stderr. This option is only effective if garbage
1310 collection has been enabled via the @samp{--gc-sections}) option. The
1311 default behaviour (of not listing the sections that are removed) can
1312 be restored by specifying @samp{--no-print-gc-sections} on the command
1319 Print a summary of the command-line options on the standard output and exit.
1321 @kindex --target-help
1323 Print a summary of all target specific options on the standard output and exit.
1326 @item -Map @var{mapfile}
1327 Print a link map to the file @var{mapfile}. See the description of the
1328 @option{-M} option, above.
1330 @cindex memory usage
1331 @kindex --no-keep-memory
1332 @item --no-keep-memory
1333 @command{ld} normally optimizes for speed over memory usage by caching the
1334 symbol tables of input files in memory. This option tells @command{ld} to
1335 instead optimize for memory usage, by rereading the symbol tables as
1336 necessary. This may be required if @command{ld} runs out of memory space
1337 while linking a large executable.
1339 @kindex --no-undefined
1341 @item --no-undefined
1343 Report unresolved symbol references from regular object files. This
1344 is done even if the linker is creating a non-symbolic shared library.
1345 The switch @option{--[no-]allow-shlib-undefined} controls the
1346 behaviour for reporting unresolved references found in shared
1347 libraries being linked in.
1349 @kindex --allow-multiple-definition
1351 @item --allow-multiple-definition
1353 Normally when a symbol is defined multiple times, the linker will
1354 report a fatal error. These options allow multiple definitions and the
1355 first definition will be used.
1357 @kindex --allow-shlib-undefined
1358 @kindex --no-allow-shlib-undefined
1359 @item --allow-shlib-undefined
1360 @itemx --no-allow-shlib-undefined
1361 Allows (the default) or disallows undefined symbols in shared libraries.
1362 This switch is similar to @option{--no-undefined} except that it
1363 determines the behaviour when the undefined symbols are in a
1364 shared library rather than a regular object file. It does not affect
1365 how undefined symbols in regular object files are handled.
1367 The reason that @option{--allow-shlib-undefined} is the default is that
1368 the shared library being specified at link time may not be the same as
1369 the one that is available at load time, so the symbols might actually be
1370 resolvable at load time. Plus there are some systems, (eg BeOS) where
1371 undefined symbols in shared libraries is normal. (The kernel patches
1372 them at load time to select which function is most appropriate
1373 for the current architecture. This is used for example to dynamically
1374 select an appropriate memset function). Apparently it is also normal
1375 for HPPA shared libraries to have undefined symbols.
1377 @kindex --no-undefined-version
1378 @item --no-undefined-version
1379 Normally when a symbol has an undefined version, the linker will ignore
1380 it. This option disallows symbols with undefined version and a fatal error
1381 will be issued instead.
1383 @kindex --default-symver
1384 @item --default-symver
1385 Create and use a default symbol version (the soname) for unversioned
1388 @kindex --default-imported-symver
1389 @item --default-imported-symver
1390 Create and use a default symbol version (the soname) for unversioned
1393 @kindex --no-warn-mismatch
1394 @item --no-warn-mismatch
1395 Normally @command{ld} will give an error if you try to link together input
1396 files that are mismatched for some reason, perhaps because they have
1397 been compiled for different processors or for different endiannesses.
1398 This option tells @command{ld} that it should silently permit such possible
1399 errors. This option should only be used with care, in cases when you
1400 have taken some special action that ensures that the linker errors are
1403 @kindex --no-warn-search-mismatch
1404 @item --no-warn-search-mismatch
1405 Normally @command{ld} will give a warning if it finds an incompatible
1406 library during a library search. This option silences the warning.
1408 @kindex --no-whole-archive
1409 @item --no-whole-archive
1410 Turn off the effect of the @option{--whole-archive} option for subsequent
1413 @cindex output file after errors
1414 @kindex --noinhibit-exec
1415 @item --noinhibit-exec
1416 Retain the executable output file whenever it is still usable.
1417 Normally, the linker will not produce an output file if it encounters
1418 errors during the link process; it exits without writing an output file
1419 when it issues any error whatsoever.
1423 Only search library directories explicitly specified on the
1424 command line. Library directories specified in linker scripts
1425 (including linker scripts specified on the command line) are ignored.
1427 @ifclear SingleFormat
1429 @item --oformat @var{output-format}
1430 @command{ld} may be configured to support more than one kind of object
1431 file. If your @command{ld} is configured this way, you can use the
1432 @samp{--oformat} option to specify the binary format for the output
1433 object file. Even when @command{ld} is configured to support alternative
1434 object formats, you don't usually need to specify this, as @command{ld}
1435 should be configured to produce as a default output format the most
1436 usual format on each machine. @var{output-format} is a text string, the
1437 name of a particular format supported by the BFD libraries. (You can
1438 list the available binary formats with @samp{objdump -i}.) The script
1439 command @code{OUTPUT_FORMAT} can also specify the output format, but
1440 this option overrides it. @xref{BFD}.
1444 @kindex --pic-executable
1446 @itemx --pic-executable
1447 @cindex position independent executables
1448 Create a position independent executable. This is currently only supported on
1449 ELF platforms. Position independent executables are similar to shared
1450 libraries in that they are relocated by the dynamic linker to the virtual
1451 address the OS chooses for them (which can vary between invocations). Like
1452 normal dynamically linked executables they can be executed and symbols
1453 defined in the executable cannot be overridden by shared libraries.
1457 This option is ignored for Linux compatibility.
1461 This option is ignored for SVR4 compatibility.
1464 @cindex synthesizing linker
1465 @cindex relaxing addressing modes
1467 An option with machine dependent effects.
1469 This option is only supported on a few targets.
1472 @xref{H8/300,,@command{ld} and the H8/300}.
1475 @xref{i960,, @command{ld} and the Intel 960 family}.
1478 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1481 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1484 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1487 On some platforms, the @samp{--relax} option performs global
1488 optimizations that become possible when the linker resolves addressing
1489 in the program, such as relaxing address modes and synthesizing new
1490 instructions in the output object file.
1492 On some platforms these link time global optimizations may make symbolic
1493 debugging of the resulting executable impossible.
1496 the case for the Matsushita MN10200 and MN10300 family of processors.
1500 On platforms where this is not supported, @samp{--relax} is accepted,
1504 @cindex retaining specified symbols
1505 @cindex stripping all but some symbols
1506 @cindex symbols, retaining selectively
1507 @item --retain-symbols-file @var{filename}
1508 Retain @emph{only} the symbols listed in the file @var{filename},
1509 discarding all others. @var{filename} is simply a flat file, with one
1510 symbol name per line. This option is especially useful in environments
1514 where a large global symbol table is accumulated gradually, to conserve
1517 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1518 or symbols needed for relocations.
1520 You may only specify @samp{--retain-symbols-file} once in the command
1521 line. It overrides @samp{-s} and @samp{-S}.
1524 @item -rpath @var{dir}
1525 @cindex runtime library search path
1527 Add a directory to the runtime library search path. This is used when
1528 linking an ELF executable with shared objects. All @option{-rpath}
1529 arguments are concatenated and passed to the runtime linker, which uses
1530 them to locate shared objects at runtime. The @option{-rpath} option is
1531 also used when locating shared objects which are needed by shared
1532 objects explicitly included in the link; see the description of the
1533 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1534 ELF executable, the contents of the environment variable
1535 @code{LD_RUN_PATH} will be used if it is defined.
1537 The @option{-rpath} option may also be used on SunOS. By default, on
1538 SunOS, the linker will form a runtime search patch out of all the
1539 @option{-L} options it is given. If a @option{-rpath} option is used, the
1540 runtime search path will be formed exclusively using the @option{-rpath}
1541 options, ignoring the @option{-L} options. This can be useful when using
1542 gcc, which adds many @option{-L} options which may be on NFS mounted
1545 For compatibility with other ELF linkers, if the @option{-R} option is
1546 followed by a directory name, rather than a file name, it is treated as
1547 the @option{-rpath} option.
1551 @cindex link-time runtime library search path
1553 @item -rpath-link @var{DIR}
1554 When using ELF or SunOS, one shared library may require another. This
1555 happens when an @code{ld -shared} link includes a shared library as one
1558 When the linker encounters such a dependency when doing a non-shared,
1559 non-relocatable link, it will automatically try to locate the required
1560 shared library and include it in the link, if it is not included
1561 explicitly. In such a case, the @option{-rpath-link} option
1562 specifies the first set of directories to search. The
1563 @option{-rpath-link} option may specify a sequence of directory names
1564 either by specifying a list of names separated by colons, or by
1565 appearing multiple times.
1567 This option should be used with caution as it overrides the search path
1568 that may have been hard compiled into a shared library. In such a case it
1569 is possible to use unintentionally a different search path than the
1570 runtime linker would do.
1572 The linker uses the following search paths to locate required shared
1576 Any directories specified by @option{-rpath-link} options.
1578 Any directories specified by @option{-rpath} options. The difference
1579 between @option{-rpath} and @option{-rpath-link} is that directories
1580 specified by @option{-rpath} options are included in the executable and
1581 used at runtime, whereas the @option{-rpath-link} option is only effective
1582 at link time. Searching @option{-rpath} in this way is only supported
1583 by native linkers and cross linkers which have been configured with
1584 the @option{--with-sysroot} option.
1586 On an ELF system, for native linkers, if the @option{-rpath} and
1587 @option{-rpath-link} options were not used, search the contents of the
1588 environment variable @code{LD_RUN_PATH}.
1590 On SunOS, if the @option{-rpath} option was not used, search any
1591 directories specified using @option{-L} options.
1593 For a native linker, the search the contents of the environment
1594 variable @code{LD_LIBRARY_PATH}.
1596 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1597 @code{DT_RPATH} of a shared library are searched for shared
1598 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1599 @code{DT_RUNPATH} entries exist.
1601 The default directories, normally @file{/lib} and @file{/usr/lib}.
1603 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1604 exists, the list of directories found in that file.
1607 If the required shared library is not found, the linker will issue a
1608 warning and continue with the link.
1615 @cindex shared libraries
1616 Create a shared library. This is currently only supported on ELF, XCOFF
1617 and SunOS platforms. On SunOS, the linker will automatically create a
1618 shared library if the @option{-e} option is not used and there are
1619 undefined symbols in the link.
1622 @kindex --sort-common
1623 This option tells @command{ld} to sort the common symbols by size when it
1624 places them in the appropriate output sections. First come all the one
1625 byte symbols, then all the two byte, then all the four byte, and then
1626 everything else. This is to prevent gaps between symbols due to
1627 alignment constraints.
1629 @kindex --sort-section name
1630 @item --sort-section name
1631 This option will apply @code{SORT_BY_NAME} to all wildcard section
1632 patterns in the linker script.
1634 @kindex --sort-section alignment
1635 @item --sort-section alignment
1636 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1637 patterns in the linker script.
1639 @kindex --split-by-file
1640 @item --split-by-file [@var{size}]
1641 Similar to @option{--split-by-reloc} but creates a new output section for
1642 each input file when @var{size} is reached. @var{size} defaults to a
1643 size of 1 if not given.
1645 @kindex --split-by-reloc
1646 @item --split-by-reloc [@var{count}]
1647 Tries to creates extra sections in the output file so that no single
1648 output section in the file contains more than @var{count} relocations.
1649 This is useful when generating huge relocatable files for downloading into
1650 certain real time kernels with the COFF object file format; since COFF
1651 cannot represent more than 65535 relocations in a single section. Note
1652 that this will fail to work with object file formats which do not
1653 support arbitrary sections. The linker will not split up individual
1654 input sections for redistribution, so if a single input section contains
1655 more than @var{count} relocations one output section will contain that
1656 many relocations. @var{count} defaults to a value of 32768.
1660 Compute and display statistics about the operation of the linker, such
1661 as execution time and memory usage.
1664 @item --sysroot=@var{directory}
1665 Use @var{directory} as the location of the sysroot, overriding the
1666 configure-time default. This option is only supported by linkers
1667 that were configured using @option{--with-sysroot}.
1669 @kindex --traditional-format
1670 @cindex traditional format
1671 @item --traditional-format
1672 For some targets, the output of @command{ld} is different in some ways from
1673 the output of some existing linker. This switch requests @command{ld} to
1674 use the traditional format instead.
1677 For example, on SunOS, @command{ld} combines duplicate entries in the
1678 symbol string table. This can reduce the size of an output file with
1679 full debugging information by over 30 percent. Unfortunately, the SunOS
1680 @code{dbx} program can not read the resulting program (@code{gdb} has no
1681 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1682 combine duplicate entries.
1684 @kindex --section-start @var{sectionname}=@var{org}
1685 @item --section-start @var{sectionname}=@var{org}
1686 Locate a section in the output file at the absolute
1687 address given by @var{org}. You may use this option as many
1688 times as necessary to locate multiple sections in the command
1690 @var{org} must be a single hexadecimal integer;
1691 for compatibility with other linkers, you may omit the leading
1692 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1693 should be no white space between @var{sectionname}, the equals
1694 sign (``@key{=}''), and @var{org}.
1696 @kindex -Tbss @var{org}
1697 @kindex -Tdata @var{org}
1698 @kindex -Ttext @var{org}
1699 @cindex segment origins, cmd line
1700 @item -Tbss @var{org}
1701 @itemx -Tdata @var{org}
1702 @itemx -Ttext @var{org}
1703 Same as --section-start, with @code{.bss}, @code{.data} or
1704 @code{.text} as the @var{sectionname}.
1706 @kindex --unresolved-symbols
1707 @item --unresolved-symbols=@var{method}
1708 Determine how to handle unresolved symbols. There are four possible
1709 values for @samp{method}:
1713 Do not report any unresolved symbols.
1716 Report all unresolved symbols. This is the default.
1718 @item ignore-in-object-files
1719 Report unresolved symbols that are contained in shared libraries, but
1720 ignore them if they come from regular object files.
1722 @item ignore-in-shared-libs
1723 Report unresolved symbols that come from regular object files, but
1724 ignore them if they come from shared libraries. This can be useful
1725 when creating a dynamic binary and it is known that all the shared
1726 libraries that it should be referencing are included on the linker's
1730 The behaviour for shared libraries on their own can also be controlled
1731 by the @option{--[no-]allow-shlib-undefined} option.
1733 Normally the linker will generate an error message for each reported
1734 unresolved symbol but the option @option{--warn-unresolved-symbols}
1735 can change this to a warning.
1741 Display the version number for @command{ld} and list the linker emulations
1742 supported. Display which input files can and cannot be opened. Display
1743 the linker script being used by the linker.
1745 @kindex --version-script=@var{version-scriptfile}
1746 @cindex version script, symbol versions
1747 @itemx --version-script=@var{version-scriptfile}
1748 Specify the name of a version script to the linker. This is typically
1749 used when creating shared libraries to specify additional information
1750 about the version hierarchy for the library being created. This option
1751 is only meaningful on ELF platforms which support shared libraries.
1754 @kindex --warn-common
1755 @cindex warnings, on combining symbols
1756 @cindex combining symbols, warnings on
1758 Warn when a common symbol is combined with another common symbol or with
1759 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1760 but linkers on some other operating systems do not. This option allows
1761 you to find potential problems from combining global symbols.
1762 Unfortunately, some C libraries use this practise, so you may get some
1763 warnings about symbols in the libraries as well as in your programs.
1765 There are three kinds of global symbols, illustrated here by C examples:
1769 A definition, which goes in the initialized data section of the output
1773 An undefined reference, which does not allocate space.
1774 There must be either a definition or a common symbol for the
1778 A common symbol. If there are only (one or more) common symbols for a
1779 variable, it goes in the uninitialized data area of the output file.
1780 The linker merges multiple common symbols for the same variable into a
1781 single symbol. If they are of different sizes, it picks the largest
1782 size. The linker turns a common symbol into a declaration, if there is
1783 a definition of the same variable.
1786 The @samp{--warn-common} option can produce five kinds of warnings.
1787 Each warning consists of a pair of lines: the first describes the symbol
1788 just encountered, and the second describes the previous symbol
1789 encountered with the same name. One or both of the two symbols will be
1794 Turning a common symbol into a reference, because there is already a
1795 definition for the symbol.
1797 @var{file}(@var{section}): warning: common of `@var{symbol}'
1798 overridden by definition
1799 @var{file}(@var{section}): warning: defined here
1803 Turning a common symbol into a reference, because a later definition for
1804 the symbol is encountered. This is the same as the previous case,
1805 except that the symbols are encountered in a different order.
1807 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1809 @var{file}(@var{section}): warning: common is here
1813 Merging a common symbol with a previous same-sized common symbol.
1815 @var{file}(@var{section}): warning: multiple common
1817 @var{file}(@var{section}): warning: previous common is here
1821 Merging a common symbol with a previous larger common symbol.
1823 @var{file}(@var{section}): warning: common of `@var{symbol}'
1824 overridden by larger common
1825 @var{file}(@var{section}): warning: larger common is here
1829 Merging a common symbol with a previous smaller common symbol. This is
1830 the same as the previous case, except that the symbols are
1831 encountered in a different order.
1833 @var{file}(@var{section}): warning: common of `@var{symbol}'
1834 overriding smaller common
1835 @var{file}(@var{section}): warning: smaller common is here
1839 @kindex --warn-constructors
1840 @item --warn-constructors
1841 Warn if any global constructors are used. This is only useful for a few
1842 object file formats. For formats like COFF or ELF, the linker can not
1843 detect the use of global constructors.
1845 @kindex --warn-multiple-gp
1846 @item --warn-multiple-gp
1847 Warn if multiple global pointer values are required in the output file.
1848 This is only meaningful for certain processors, such as the Alpha.
1849 Specifically, some processors put large-valued constants in a special
1850 section. A special register (the global pointer) points into the middle
1851 of this section, so that constants can be loaded efficiently via a
1852 base-register relative addressing mode. Since the offset in
1853 base-register relative mode is fixed and relatively small (e.g., 16
1854 bits), this limits the maximum size of the constant pool. Thus, in
1855 large programs, it is often necessary to use multiple global pointer
1856 values in order to be able to address all possible constants. This
1857 option causes a warning to be issued whenever this case occurs.
1860 @cindex warnings, on undefined symbols
1861 @cindex undefined symbols, warnings on
1863 Only warn once for each undefined symbol, rather than once per module
1866 @kindex --warn-section-align
1867 @cindex warnings, on section alignment
1868 @cindex section alignment, warnings on
1869 @item --warn-section-align
1870 Warn if the address of an output section is changed because of
1871 alignment. Typically, the alignment will be set by an input section.
1872 The address will only be changed if it not explicitly specified; that
1873 is, if the @code{SECTIONS} command does not specify a start address for
1874 the section (@pxref{SECTIONS}).
1876 @kindex --warn-shared-textrel
1877 @item --warn-shared-textrel
1878 Warn if the linker adds a DT_TEXTREL to a shared object.
1880 @kindex --warn-unresolved-symbols
1881 @item --warn-unresolved-symbols
1882 If the linker is going to report an unresolved symbol (see the option
1883 @option{--unresolved-symbols}) it will normally generate an error.
1884 This option makes it generate a warning instead.
1886 @kindex --error-unresolved-symbols
1887 @item --error-unresolved-symbols
1888 This restores the linker's default behaviour of generating errors when
1889 it is reporting unresolved symbols.
1891 @kindex --whole-archive
1892 @cindex including an entire archive
1893 @item --whole-archive
1894 For each archive mentioned on the command line after the
1895 @option{--whole-archive} option, include every object file in the archive
1896 in the link, rather than searching the archive for the required object
1897 files. This is normally used to turn an archive file into a shared
1898 library, forcing every object to be included in the resulting shared
1899 library. This option may be used more than once.
1901 Two notes when using this option from gcc: First, gcc doesn't know
1902 about this option, so you have to use @option{-Wl,-whole-archive}.
1903 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1904 list of archives, because gcc will add its own list of archives to
1905 your link and you may not want this flag to affect those as well.
1908 @item --wrap @var{symbol}
1909 Use a wrapper function for @var{symbol}. Any undefined reference to
1910 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1911 undefined reference to @code{__real_@var{symbol}} will be resolved to
1914 This can be used to provide a wrapper for a system function. The
1915 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1916 wishes to call the system function, it should call
1917 @code{__real_@var{symbol}}.
1919 Here is a trivial example:
1923 __wrap_malloc (size_t c)
1925 printf ("malloc called with %zu\n", c);
1926 return __real_malloc (c);
1930 If you link other code with this file using @option{--wrap malloc}, then
1931 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1932 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1933 call the real @code{malloc} function.
1935 You may wish to provide a @code{__real_malloc} function as well, so that
1936 links without the @option{--wrap} option will succeed. If you do this,
1937 you should not put the definition of @code{__real_malloc} in the same
1938 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
1939 call before the linker has a chance to wrap it to @code{malloc}.
1941 @kindex --eh-frame-hdr
1942 @item --eh-frame-hdr
1943 Request creation of @code{.eh_frame_hdr} section and ELF
1944 @code{PT_GNU_EH_FRAME} segment header.
1946 @kindex --enable-new-dtags
1947 @kindex --disable-new-dtags
1948 @item --enable-new-dtags
1949 @itemx --disable-new-dtags
1950 This linker can create the new dynamic tags in ELF. But the older ELF
1951 systems may not understand them. If you specify
1952 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
1953 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
1954 created. By default, the new dynamic tags are not created. Note that
1955 those options are only available for ELF systems.
1957 @kindex --hash-size=@var{number}
1958 @item --hash-size=@var{number}
1959 Set the default size of the linker's hash tables to a prime number
1960 close to @var{number}. Increasing this value can reduce the length of
1961 time it takes the linker to perform its tasks, at the expense of
1962 increasing the linker's memory requirements. Similarly reducing this
1963 value can reduce the memory requirements at the expense of speed.
1965 @kindex --hash-style=@var{style}
1966 @item --hash-style=@var{style}
1967 Set the type of linker's hash table(s). @var{style} can be either
1968 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
1969 new style GNU @code{.gnu.hash} section or @code{both} for both
1970 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
1971 hash tables. The default is @code{sysv}.
1973 @kindex --reduce-memory-overheads
1974 @item --reduce-memory-overheads
1975 This option reduces memory requirements at ld runtime, at the expense of
1976 linking speed. This was introduced to select the old O(n^2) algorithm
1977 for link map file generation, rather than the new O(n) algorithm which uses
1978 about 40% more memory for symbol storage.
1980 Another effect of the switch is to set the default hash table size to
1981 1021, which again saves memory at the cost of lengthening the linker's
1982 run time. This is not done however if the @option{--hash-size} switch
1985 The @option{--reduce-memory-overheads} switch may be also be used to
1986 enable other tradeoffs in future versions of the linker.
1989 @kindex --build-id=@var{style}
1991 @itemx --build-id=@var{style}
1992 Request creation of @code{.note.gnu.build-id} ELF note section.
1993 The contents of the note are unique bits identifying this linked
1994 file. @var{style} can be @code{uuid} to use 128 random bits,
1995 @code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
1996 parts of the output contents, @code{md5} to use a 128-bit
1997 @sc{MD5} hash on the normative parts of the output contents, or
1998 @code{0x@var{hexstring}} to use a chosen bit string specified as
1999 an even number of hexadecimal digits (@code{-} and @code{:}
2000 characters between digit pairs are ignored). If @var{style} is
2001 omitted, @code{sha1} is used.
2003 The @code{md5} and @code{sha1} styles produces an identifier
2004 that is always the same in an identical output file, but will be
2005 unique among all nonidentical output files. It is not intended
2006 to be compared as a checksum for the file's contents. A linked
2007 file may be changed later by other tools, but the build ID bit
2008 string identifying the original linked file does not change.
2010 Passing @code{none} for @var{style} disables the setting from any
2011 @code{--build-id} options earlier on the command line.
2016 @subsection Options Specific to i386 PE Targets
2018 @c man begin OPTIONS
2020 The i386 PE linker supports the @option{-shared} option, which causes
2021 the output to be a dynamically linked library (DLL) instead of a
2022 normal executable. You should name the output @code{*.dll} when you
2023 use this option. In addition, the linker fully supports the standard
2024 @code{*.def} files, which may be specified on the linker command line
2025 like an object file (in fact, it should precede archives it exports
2026 symbols from, to ensure that they get linked in, just like a normal
2029 In addition to the options common to all targets, the i386 PE linker
2030 support additional command line options that are specific to the i386
2031 PE target. Options that take values may be separated from their
2032 values by either a space or an equals sign.
2036 @kindex --add-stdcall-alias
2037 @item --add-stdcall-alias
2038 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2039 as-is and also with the suffix stripped.
2040 [This option is specific to the i386 PE targeted port of the linker]
2043 @item --base-file @var{file}
2044 Use @var{file} as the name of a file in which to save the base
2045 addresses of all the relocations needed for generating DLLs with
2047 [This is an i386 PE specific option]
2051 Create a DLL instead of a regular executable. You may also use
2052 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2054 [This option is specific to the i386 PE targeted port of the linker]
2056 @kindex --enable-stdcall-fixup
2057 @kindex --disable-stdcall-fixup
2058 @item --enable-stdcall-fixup
2059 @itemx --disable-stdcall-fixup
2060 If the link finds a symbol that it cannot resolve, it will attempt to
2061 do ``fuzzy linking'' by looking for another defined symbol that differs
2062 only in the format of the symbol name (cdecl vs stdcall) and will
2063 resolve that symbol by linking to the match. For example, the
2064 undefined symbol @code{_foo} might be linked to the function
2065 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2066 to the function @code{_bar}. When the linker does this, it prints a
2067 warning, since it normally should have failed to link, but sometimes
2068 import libraries generated from third-party dlls may need this feature
2069 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2070 feature is fully enabled and warnings are not printed. If you specify
2071 @option{--disable-stdcall-fixup}, this feature is disabled and such
2072 mismatches are considered to be errors.
2073 [This option is specific to the i386 PE targeted port of the linker]
2075 @cindex DLLs, creating
2076 @kindex --export-all-symbols
2077 @item --export-all-symbols
2078 If given, all global symbols in the objects used to build a DLL will
2079 be exported by the DLL. Note that this is the default if there
2080 otherwise wouldn't be any exported symbols. When symbols are
2081 explicitly exported via DEF files or implicitly exported via function
2082 attributes, the default is to not export anything else unless this
2083 option is given. Note that the symbols @code{DllMain@@12},
2084 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2085 @code{impure_ptr} will not be automatically
2086 exported. Also, symbols imported from other DLLs will not be
2087 re-exported, nor will symbols specifying the DLL's internal layout
2088 such as those beginning with @code{_head_} or ending with
2089 @code{_iname}. In addition, no symbols from @code{libgcc},
2090 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2091 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2092 not be exported, to help with C++ DLLs. Finally, there is an
2093 extensive list of cygwin-private symbols that are not exported
2094 (obviously, this applies on when building DLLs for cygwin targets).
2095 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2096 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2097 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2098 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2099 @code{cygwin_premain3}, and @code{environ}.
2100 [This option is specific to the i386 PE targeted port of the linker]
2102 @kindex --exclude-symbols
2103 @item --exclude-symbols @var{symbol},@var{symbol},...
2104 Specifies a list of symbols which should not be automatically
2105 exported. The symbol names may be delimited by commas or colons.
2106 [This option is specific to the i386 PE targeted port of the linker]
2108 @kindex --file-alignment
2109 @item --file-alignment
2110 Specify the file alignment. Sections in the file will always begin at
2111 file offsets which are multiples of this number. This defaults to
2113 [This option is specific to the i386 PE targeted port of the linker]
2117 @item --heap @var{reserve}
2118 @itemx --heap @var{reserve},@var{commit}
2119 Specify the number of bytes of memory to reserve (and optionally commit)
2120 to be used as heap for this program. The default is 1Mb reserved, 4K
2122 [This option is specific to the i386 PE targeted port of the linker]
2125 @kindex --image-base
2126 @item --image-base @var{value}
2127 Use @var{value} as the base address of your program or dll. This is
2128 the lowest memory location that will be used when your program or dll
2129 is loaded. To reduce the need to relocate and improve performance of
2130 your dlls, each should have a unique base address and not overlap any
2131 other dlls. The default is 0x400000 for executables, and 0x10000000
2133 [This option is specific to the i386 PE targeted port of the linker]
2137 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2138 symbols before they are exported.
2139 [This option is specific to the i386 PE targeted port of the linker]
2141 @kindex --large-address-aware
2142 @item --large-address-aware
2143 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2144 header is set to indicate that this executable supports virtual addresses
2145 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2146 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2147 section of the BOOT.INI. Otherwise, this bit has no effect.
2148 [This option is specific to PE targeted ports of the linker]
2150 @kindex --major-image-version
2151 @item --major-image-version @var{value}
2152 Sets the major number of the ``image version''. Defaults to 1.
2153 [This option is specific to the i386 PE targeted port of the linker]
2155 @kindex --major-os-version
2156 @item --major-os-version @var{value}
2157 Sets the major number of the ``os version''. Defaults to 4.
2158 [This option is specific to the i386 PE targeted port of the linker]
2160 @kindex --major-subsystem-version
2161 @item --major-subsystem-version @var{value}
2162 Sets the major number of the ``subsystem version''. Defaults to 4.
2163 [This option is specific to the i386 PE targeted port of the linker]
2165 @kindex --minor-image-version
2166 @item --minor-image-version @var{value}
2167 Sets the minor number of the ``image version''. Defaults to 0.
2168 [This option is specific to the i386 PE targeted port of the linker]
2170 @kindex --minor-os-version
2171 @item --minor-os-version @var{value}
2172 Sets the minor number of the ``os version''. Defaults to 0.
2173 [This option is specific to the i386 PE targeted port of the linker]
2175 @kindex --minor-subsystem-version
2176 @item --minor-subsystem-version @var{value}
2177 Sets the minor number of the ``subsystem version''. Defaults to 0.
2178 [This option is specific to the i386 PE targeted port of the linker]
2180 @cindex DEF files, creating
2181 @cindex DLLs, creating
2182 @kindex --output-def
2183 @item --output-def @var{file}
2184 The linker will create the file @var{file} which will contain a DEF
2185 file corresponding to the DLL the linker is generating. This DEF file
2186 (which should be called @code{*.def}) may be used to create an import
2187 library with @code{dlltool} or may be used as a reference to
2188 automatically or implicitly exported symbols.
2189 [This option is specific to the i386 PE targeted port of the linker]
2191 @cindex DLLs, creating
2192 @kindex --out-implib
2193 @item --out-implib @var{file}
2194 The linker will create the file @var{file} which will contain an
2195 import lib corresponding to the DLL the linker is generating. This
2196 import lib (which should be called @code{*.dll.a} or @code{*.a}
2197 may be used to link clients against the generated DLL; this behaviour
2198 makes it possible to skip a separate @code{dlltool} import library
2200 [This option is specific to the i386 PE targeted port of the linker]
2202 @kindex --enable-auto-image-base
2203 @item --enable-auto-image-base
2204 Automatically choose the image base for DLLs, unless one is specified
2205 using the @code{--image-base} argument. By using a hash generated
2206 from the dllname to create unique image bases for each DLL, in-memory
2207 collisions and relocations which can delay program execution are
2209 [This option is specific to the i386 PE targeted port of the linker]
2211 @kindex --disable-auto-image-base
2212 @item --disable-auto-image-base
2213 Do not automatically generate a unique image base. If there is no
2214 user-specified image base (@code{--image-base}) then use the platform
2216 [This option is specific to the i386 PE targeted port of the linker]
2218 @cindex DLLs, linking to
2219 @kindex --dll-search-prefix
2220 @item --dll-search-prefix @var{string}
2221 When linking dynamically to a dll without an import library,
2222 search for @code{<string><basename>.dll} in preference to
2223 @code{lib<basename>.dll}. This behaviour allows easy distinction
2224 between DLLs built for the various "subplatforms": native, cygwin,
2225 uwin, pw, etc. For instance, cygwin DLLs typically use
2226 @code{--dll-search-prefix=cyg}.
2227 [This option is specific to the i386 PE targeted port of the linker]
2229 @kindex --enable-auto-import
2230 @item --enable-auto-import
2231 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2232 DATA imports from DLLs, and create the necessary thunking symbols when
2233 building the import libraries with those DATA exports. Note: Use of the
2234 'auto-import' extension will cause the text section of the image file
2235 to be made writable. This does not conform to the PE-COFF format
2236 specification published by Microsoft.
2238 Note - use of the 'auto-import' extension will also cause read only
2239 data which would normally be placed into the .rdata section to be
2240 placed into the .data section instead. This is in order to work
2241 around a problem with consts that is described here:
2242 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2244 Using 'auto-import' generally will 'just work' -- but sometimes you may
2247 "variable '<var>' can't be auto-imported. Please read the
2248 documentation for ld's @code{--enable-auto-import} for details."
2250 This message occurs when some (sub)expression accesses an address
2251 ultimately given by the sum of two constants (Win32 import tables only
2252 allow one). Instances where this may occur include accesses to member
2253 fields of struct variables imported from a DLL, as well as using a
2254 constant index into an array variable imported from a DLL. Any
2255 multiword variable (arrays, structs, long long, etc) may trigger
2256 this error condition. However, regardless of the exact data type
2257 of the offending exported variable, ld will always detect it, issue
2258 the warning, and exit.
2260 There are several ways to address this difficulty, regardless of the
2261 data type of the exported variable:
2263 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2264 of adjusting references in your client code for runtime environment, so
2265 this method works only when runtime environment supports this feature.
2267 A second solution is to force one of the 'constants' to be a variable --
2268 that is, unknown and un-optimizable at compile time. For arrays,
2269 there are two possibilities: a) make the indexee (the array's address)
2270 a variable, or b) make the 'constant' index a variable. Thus:
2273 extern type extern_array[];
2275 @{ volatile type *t=extern_array; t[1] @}
2281 extern type extern_array[];
2283 @{ volatile int t=1; extern_array[t] @}
2286 For structs (and most other multiword data types) the only option
2287 is to make the struct itself (or the long long, or the ...) variable:
2290 extern struct s extern_struct;
2291 extern_struct.field -->
2292 @{ volatile struct s *t=&extern_struct; t->field @}
2298 extern long long extern_ll;
2300 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2303 A third method of dealing with this difficulty is to abandon
2304 'auto-import' for the offending symbol and mark it with
2305 @code{__declspec(dllimport)}. However, in practise that
2306 requires using compile-time #defines to indicate whether you are
2307 building a DLL, building client code that will link to the DLL, or
2308 merely building/linking to a static library. In making the choice
2309 between the various methods of resolving the 'direct address with
2310 constant offset' problem, you should consider typical real-world usage:
2318 void main(int argc, char **argv)@{
2319 printf("%d\n",arr[1]);
2329 void main(int argc, char **argv)@{
2330 /* This workaround is for win32 and cygwin; do not "optimize" */
2331 volatile int *parr = arr;
2332 printf("%d\n",parr[1]);
2339 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2340 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2341 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2342 #define FOO_IMPORT __declspec(dllimport)
2346 extern FOO_IMPORT int arr[];
2349 void main(int argc, char **argv)@{
2350 printf("%d\n",arr[1]);
2354 A fourth way to avoid this problem is to re-code your
2355 library to use a functional interface rather than a data interface
2356 for the offending variables (e.g. set_foo() and get_foo() accessor
2358 [This option is specific to the i386 PE targeted port of the linker]
2360 @kindex --disable-auto-import
2361 @item --disable-auto-import
2362 Do not attempt to do sophisticated linking of @code{_symbol} to
2363 @code{__imp__symbol} for DATA imports from DLLs.
2364 [This option is specific to the i386 PE targeted port of the linker]
2366 @kindex --enable-runtime-pseudo-reloc
2367 @item --enable-runtime-pseudo-reloc
2368 If your code contains expressions described in --enable-auto-import section,
2369 that is, DATA imports from DLL with non-zero offset, this switch will create
2370 a vector of 'runtime pseudo relocations' which can be used by runtime
2371 environment to adjust references to such data in your client code.
2372 [This option is specific to the i386 PE targeted port of the linker]
2374 @kindex --disable-runtime-pseudo-reloc
2375 @item --disable-runtime-pseudo-reloc
2376 Do not create pseudo relocations for non-zero offset DATA imports from
2377 DLLs. This is the default.
2378 [This option is specific to the i386 PE targeted port of the linker]
2380 @kindex --enable-extra-pe-debug
2381 @item --enable-extra-pe-debug
2382 Show additional debug info related to auto-import symbol thunking.
2383 [This option is specific to the i386 PE targeted port of the linker]
2385 @kindex --section-alignment
2386 @item --section-alignment
2387 Sets the section alignment. Sections in memory will always begin at
2388 addresses which are a multiple of this number. Defaults to 0x1000.
2389 [This option is specific to the i386 PE targeted port of the linker]
2393 @item --stack @var{reserve}
2394 @itemx --stack @var{reserve},@var{commit}
2395 Specify the number of bytes of memory to reserve (and optionally commit)
2396 to be used as stack for this program. The default is 2Mb reserved, 4K
2398 [This option is specific to the i386 PE targeted port of the linker]
2401 @item --subsystem @var{which}
2402 @itemx --subsystem @var{which}:@var{major}
2403 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2404 Specifies the subsystem under which your program will execute. The
2405 legal values for @var{which} are @code{native}, @code{windows},
2406 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2407 the subsystem version also. Numeric values are also accepted for
2409 [This option is specific to the i386 PE targeted port of the linker]
2416 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2418 @c man begin OPTIONS
2420 The 68HC11 and 68HC12 linkers support specific options to control the
2421 memory bank switching mapping and trampoline code generation.
2425 @kindex --no-trampoline
2426 @item --no-trampoline
2427 This option disables the generation of trampoline. By default a trampoline
2428 is generated for each far function which is called using a @code{jsr}
2429 instruction (this happens when a pointer to a far function is taken).
2431 @kindex --bank-window
2432 @item --bank-window @var{name}
2433 This option indicates to the linker the name of the memory region in
2434 the @samp{MEMORY} specification that describes the memory bank window.
2435 The definition of such region is then used by the linker to compute
2436 paging and addresses within the memory window.
2445 @section Environment Variables
2447 @c man begin ENVIRONMENT
2449 You can change the behaviour of @command{ld} with the environment variables
2450 @ifclear SingleFormat
2453 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2455 @ifclear SingleFormat
2457 @cindex default input format
2458 @code{GNUTARGET} determines the input-file object format if you don't
2459 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2460 of the BFD names for an input format (@pxref{BFD}). If there is no
2461 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2462 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2463 attempts to discover the input format by examining binary input files;
2464 this method often succeeds, but there are potential ambiguities, since
2465 there is no method of ensuring that the magic number used to specify
2466 object-file formats is unique. However, the configuration procedure for
2467 BFD on each system places the conventional format for that system first
2468 in the search-list, so ambiguities are resolved in favor of convention.
2472 @cindex default emulation
2473 @cindex emulation, default
2474 @code{LDEMULATION} determines the default emulation if you don't use the
2475 @samp{-m} option. The emulation can affect various aspects of linker
2476 behaviour, particularly the default linker script. You can list the
2477 available emulations with the @samp{--verbose} or @samp{-V} options. If
2478 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2479 variable is not defined, the default emulation depends upon how the
2480 linker was configured.
2482 @kindex COLLECT_NO_DEMANGLE
2483 @cindex demangling, default
2484 Normally, the linker will default to demangling symbols. However, if
2485 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2486 default to not demangling symbols. This environment variable is used in
2487 a similar fashion by the @code{gcc} linker wrapper program. The default
2488 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2495 @chapter Linker Scripts
2498 @cindex linker scripts
2499 @cindex command files
2500 Every link is controlled by a @dfn{linker script}. This script is
2501 written in the linker command language.
2503 The main purpose of the linker script is to describe how the sections in
2504 the input files should be mapped into the output file, and to control
2505 the memory layout of the output file. Most linker scripts do nothing
2506 more than this. However, when necessary, the linker script can also
2507 direct the linker to perform many other operations, using the commands
2510 The linker always uses a linker script. If you do not supply one
2511 yourself, the linker will use a default script that is compiled into the
2512 linker executable. You can use the @samp{--verbose} command line option
2513 to display the default linker script. Certain command line options,
2514 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2516 You may supply your own linker script by using the @samp{-T} command
2517 line option. When you do this, your linker script will replace the
2518 default linker script.
2520 You may also use linker scripts implicitly by naming them as input files
2521 to the linker, as though they were files to be linked. @xref{Implicit
2525 * Basic Script Concepts:: Basic Linker Script Concepts
2526 * Script Format:: Linker Script Format
2527 * Simple Example:: Simple Linker Script Example
2528 * Simple Commands:: Simple Linker Script Commands
2529 * Assignments:: Assigning Values to Symbols
2530 * SECTIONS:: SECTIONS Command
2531 * MEMORY:: MEMORY Command
2532 * PHDRS:: PHDRS Command
2533 * VERSION:: VERSION Command
2534 * Expressions:: Expressions in Linker Scripts
2535 * Implicit Linker Scripts:: Implicit Linker Scripts
2538 @node Basic Script Concepts
2539 @section Basic Linker Script Concepts
2540 @cindex linker script concepts
2541 We need to define some basic concepts and vocabulary in order to
2542 describe the linker script language.
2544 The linker combines input files into a single output file. The output
2545 file and each input file are in a special data format known as an
2546 @dfn{object file format}. Each file is called an @dfn{object file}.
2547 The output file is often called an @dfn{executable}, but for our
2548 purposes we will also call it an object file. Each object file has,
2549 among other things, a list of @dfn{sections}. We sometimes refer to a
2550 section in an input file as an @dfn{input section}; similarly, a section
2551 in the output file is an @dfn{output section}.
2553 Each section in an object file has a name and a size. Most sections
2554 also have an associated block of data, known as the @dfn{section
2555 contents}. A section may be marked as @dfn{loadable}, which mean that
2556 the contents should be loaded into memory when the output file is run.
2557 A section with no contents may be @dfn{allocatable}, which means that an
2558 area in memory should be set aside, but nothing in particular should be
2559 loaded there (in some cases this memory must be zeroed out). A section
2560 which is neither loadable nor allocatable typically contains some sort
2561 of debugging information.
2563 Every loadable or allocatable output section has two addresses. The
2564 first is the @dfn{VMA}, or virtual memory address. This is the address
2565 the section will have when the output file is run. The second is the
2566 @dfn{LMA}, or load memory address. This is the address at which the
2567 section will be loaded. In most cases the two addresses will be the
2568 same. An example of when they might be different is when a data section
2569 is loaded into ROM, and then copied into RAM when the program starts up
2570 (this technique is often used to initialize global variables in a ROM
2571 based system). In this case the ROM address would be the LMA, and the
2572 RAM address would be the VMA.
2574 You can see the sections in an object file by using the @code{objdump}
2575 program with the @samp{-h} option.
2577 Every object file also has a list of @dfn{symbols}, known as the
2578 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2579 has a name, and each defined symbol has an address, among other
2580 information. If you compile a C or C++ program into an object file, you
2581 will get a defined symbol for every defined function and global or
2582 static variable. Every undefined function or global variable which is
2583 referenced in the input file will become an undefined symbol.
2585 You can see the symbols in an object file by using the @code{nm}
2586 program, or by using the @code{objdump} program with the @samp{-t}
2590 @section Linker Script Format
2591 @cindex linker script format
2592 Linker scripts are text files.
2594 You write a linker script as a series of commands. Each command is
2595 either a keyword, possibly followed by arguments, or an assignment to a
2596 symbol. You may separate commands using semicolons. Whitespace is
2599 Strings such as file or format names can normally be entered directly.
2600 If the file name contains a character such as a comma which would
2601 otherwise serve to separate file names, you may put the file name in
2602 double quotes. There is no way to use a double quote character in a
2605 You may include comments in linker scripts just as in C, delimited by
2606 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2609 @node Simple Example
2610 @section Simple Linker Script Example
2611 @cindex linker script example
2612 @cindex example of linker script
2613 Many linker scripts are fairly simple.
2615 The simplest possible linker script has just one command:
2616 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2617 memory layout of the output file.
2619 The @samp{SECTIONS} command is a powerful command. Here we will
2620 describe a simple use of it. Let's assume your program consists only of
2621 code, initialized data, and uninitialized data. These will be in the
2622 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2623 Let's assume further that these are the only sections which appear in
2626 For this example, let's say that the code should be loaded at address
2627 0x10000, and that the data should start at address 0x8000000. Here is a
2628 linker script which will do that:
2633 .text : @{ *(.text) @}
2635 .data : @{ *(.data) @}
2636 .bss : @{ *(.bss) @}
2640 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2641 followed by a series of symbol assignments and output section
2642 descriptions enclosed in curly braces.
2644 The first line inside the @samp{SECTIONS} command of the above example
2645 sets the value of the special symbol @samp{.}, which is the location
2646 counter. If you do not specify the address of an output section in some
2647 other way (other ways are described later), the address is set from the
2648 current value of the location counter. The location counter is then
2649 incremented by the size of the output section. At the start of the
2650 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2652 The second line defines an output section, @samp{.text}. The colon is
2653 required syntax which may be ignored for now. Within the curly braces
2654 after the output section name, you list the names of the input sections
2655 which should be placed into this output section. The @samp{*} is a
2656 wildcard which matches any file name. The expression @samp{*(.text)}
2657 means all @samp{.text} input sections in all input files.
2659 Since the location counter is @samp{0x10000} when the output section
2660 @samp{.text} is defined, the linker will set the address of the
2661 @samp{.text} section in the output file to be @samp{0x10000}.
2663 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2664 the output file. The linker will place the @samp{.data} output section
2665 at address @samp{0x8000000}. After the linker places the @samp{.data}
2666 output section, the value of the location counter will be
2667 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2668 effect is that the linker will place the @samp{.bss} output section
2669 immediately after the @samp{.data} output section in memory.
2671 The linker will ensure that each output section has the required
2672 alignment, by increasing the location counter if necessary. In this
2673 example, the specified addresses for the @samp{.text} and @samp{.data}
2674 sections will probably satisfy any alignment constraints, but the linker
2675 may have to create a small gap between the @samp{.data} and @samp{.bss}
2678 That's it! That's a simple and complete linker script.
2680 @node Simple Commands
2681 @section Simple Linker Script Commands
2682 @cindex linker script simple commands
2683 In this section we describe the simple linker script commands.
2686 * Entry Point:: Setting the entry point
2687 * File Commands:: Commands dealing with files
2688 @ifclear SingleFormat
2689 * Format Commands:: Commands dealing with object file formats
2692 * Miscellaneous Commands:: Other linker script commands
2696 @subsection Setting the Entry Point
2697 @kindex ENTRY(@var{symbol})
2698 @cindex start of execution
2699 @cindex first instruction
2701 The first instruction to execute in a program is called the @dfn{entry
2702 point}. You can use the @code{ENTRY} linker script command to set the
2703 entry point. The argument is a symbol name:
2708 There are several ways to set the entry point. The linker will set the
2709 entry point by trying each of the following methods in order, and
2710 stopping when one of them succeeds:
2713 the @samp{-e} @var{entry} command-line option;
2715 the @code{ENTRY(@var{symbol})} command in a linker script;
2717 the value of the symbol @code{start}, if defined;
2719 the address of the first byte of the @samp{.text} section, if present;
2721 The address @code{0}.
2725 @subsection Commands Dealing with Files
2726 @cindex linker script file commands
2727 Several linker script commands deal with files.
2730 @item INCLUDE @var{filename}
2731 @kindex INCLUDE @var{filename}
2732 @cindex including a linker script
2733 Include the linker script @var{filename} at this point. The file will
2734 be searched for in the current directory, and in any directory specified
2735 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2738 @item INPUT(@var{file}, @var{file}, @dots{})
2739 @itemx INPUT(@var{file} @var{file} @dots{})
2740 @kindex INPUT(@var{files})
2741 @cindex input files in linker scripts
2742 @cindex input object files in linker scripts
2743 @cindex linker script input object files
2744 The @code{INPUT} command directs the linker to include the named files
2745 in the link, as though they were named on the command line.
2747 For example, if you always want to include @file{subr.o} any time you do
2748 a link, but you can't be bothered to put it on every link command line,
2749 then you can put @samp{INPUT (subr.o)} in your linker script.
2751 In fact, if you like, you can list all of your input files in the linker
2752 script, and then invoke the linker with nothing but a @samp{-T} option.
2754 In case a @dfn{sysroot prefix} is configured, and the filename starts
2755 with the @samp{/} character, and the script being processed was
2756 located inside the @dfn{sysroot prefix}, the filename will be looked
2757 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2758 open the file in the current directory. If it is not found, the
2759 linker will search through the archive library search path. See the
2760 description of @samp{-L} in @ref{Options,,Command Line Options}.
2762 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2763 name to @code{lib@var{file}.a}, as with the command line argument
2766 When you use the @code{INPUT} command in an implicit linker script, the
2767 files will be included in the link at the point at which the linker
2768 script file is included. This can affect archive searching.
2770 @item GROUP(@var{file}, @var{file}, @dots{})
2771 @itemx GROUP(@var{file} @var{file} @dots{})
2772 @kindex GROUP(@var{files})
2773 @cindex grouping input files
2774 The @code{GROUP} command is like @code{INPUT}, except that the named
2775 files should all be archives, and they are searched repeatedly until no
2776 new undefined references are created. See the description of @samp{-(}
2777 in @ref{Options,,Command Line Options}.
2779 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
2780 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
2781 @kindex AS_NEEDED(@var{files})
2782 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2783 commands, among other filenames. The files listed will be handled
2784 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2785 with the exception of ELF shared libraries, that will be added only
2786 when they are actually needed. This construct essentially enables
2787 @option{--as-needed} option for all the files listed inside of it
2788 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2791 @item OUTPUT(@var{filename})
2792 @kindex OUTPUT(@var{filename})
2793 @cindex output file name in linker script
2794 The @code{OUTPUT} command names the output file. Using
2795 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2796 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2797 Line Options}). If both are used, the command line option takes
2800 You can use the @code{OUTPUT} command to define a default name for the
2801 output file other than the usual default of @file{a.out}.
2803 @item SEARCH_DIR(@var{path})
2804 @kindex SEARCH_DIR(@var{path})
2805 @cindex library search path in linker script
2806 @cindex archive search path in linker script
2807 @cindex search path in linker script
2808 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2809 @command{ld} looks for archive libraries. Using
2810 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2811 on the command line (@pxref{Options,,Command Line Options}). If both
2812 are used, then the linker will search both paths. Paths specified using
2813 the command line option are searched first.
2815 @item STARTUP(@var{filename})
2816 @kindex STARTUP(@var{filename})
2817 @cindex first input file
2818 The @code{STARTUP} command is just like the @code{INPUT} command, except
2819 that @var{filename} will become the first input file to be linked, as
2820 though it were specified first on the command line. This may be useful
2821 when using a system in which the entry point is always the start of the
2825 @ifclear SingleFormat
2826 @node Format Commands
2827 @subsection Commands Dealing with Object File Formats
2828 A couple of linker script commands deal with object file formats.
2831 @item OUTPUT_FORMAT(@var{bfdname})
2832 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2833 @kindex OUTPUT_FORMAT(@var{bfdname})
2834 @cindex output file format in linker script
2835 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2836 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2837 exactly like using @samp{--oformat @var{bfdname}} on the command line
2838 (@pxref{Options,,Command Line Options}). If both are used, the command
2839 line option takes precedence.
2841 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2842 formats based on the @samp{-EB} and @samp{-EL} command line options.
2843 This permits the linker script to set the output format based on the
2846 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2847 will be the first argument, @var{default}. If @samp{-EB} is used, the
2848 output format will be the second argument, @var{big}. If @samp{-EL} is
2849 used, the output format will be the third argument, @var{little}.
2851 For example, the default linker script for the MIPS ELF target uses this
2854 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2856 This says that the default format for the output file is
2857 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
2858 option, the output file will be created in the @samp{elf32-littlemips}
2861 @item TARGET(@var{bfdname})
2862 @kindex TARGET(@var{bfdname})
2863 @cindex input file format in linker script
2864 The @code{TARGET} command names the BFD format to use when reading input
2865 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
2866 This command is like using @samp{-b @var{bfdname}} on the command line
2867 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
2868 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
2869 command is also used to set the format for the output file. @xref{BFD}.
2873 @node Miscellaneous Commands
2874 @subsection Other Linker Script Commands
2875 There are a few other linker scripts commands.
2878 @item ASSERT(@var{exp}, @var{message})
2880 @cindex assertion in linker script
2881 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
2882 with an error code, and print @var{message}.
2884 @item EXTERN(@var{symbol} @var{symbol} @dots{})
2886 @cindex undefined symbol in linker script
2887 Force @var{symbol} to be entered in the output file as an undefined
2888 symbol. Doing this may, for example, trigger linking of additional
2889 modules from standard libraries. You may list several @var{symbol}s for
2890 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
2891 command has the same effect as the @samp{-u} command-line option.
2893 @item FORCE_COMMON_ALLOCATION
2894 @kindex FORCE_COMMON_ALLOCATION
2895 @cindex common allocation in linker script
2896 This command has the same effect as the @samp{-d} command-line option:
2897 to make @command{ld} assign space to common symbols even if a relocatable
2898 output file is specified (@samp{-r}).
2900 @item INHIBIT_COMMON_ALLOCATION
2901 @kindex INHIBIT_COMMON_ALLOCATION
2902 @cindex common allocation in linker script
2903 This command has the same effect as the @samp{--no-define-common}
2904 command-line option: to make @code{ld} omit the assignment of addresses
2905 to common symbols even for a non-relocatable output file.
2907 @item INSERT [ AFTER | BEFORE ] @var{output_section}
2909 @cindex insert user script into default script
2910 This command is typically used in a script specified by @samp{-T} to
2911 augment the default @code{SECTIONS} with, for example, overlays. It
2912 inserts all prior linker script statements after (or before)
2913 @var{output_section}, and also causes @samp{-T} to not override the
2914 default linker script. The exact insertion point is as for orphan
2915 sections. @xref{Location Counter}. The insertion happens after the
2916 linker has mapped input sections to output sections. Prior to the
2917 insertion, since @samp{-T} scripts are parsed before the default
2918 linker script, statements in the @samp{-T} script occur before the
2919 default linker script statements in the internal linker representation
2920 of the script. In particular, input section assignments will be made
2921 to @samp{-T} output sections before those in the default script. Here
2922 is an example of how a @samp{-T} script using @code{INSERT} might look:
2929 .ov1 @{ ov1*(.text) @}
2930 .ov2 @{ ov2*(.text) @}
2936 @item NOCROSSREFS(@var{section} @var{section} @dots{})
2937 @kindex NOCROSSREFS(@var{sections})
2938 @cindex cross references
2939 This command may be used to tell @command{ld} to issue an error about any
2940 references among certain output sections.
2942 In certain types of programs, particularly on embedded systems when
2943 using overlays, when one section is loaded into memory, another section
2944 will not be. Any direct references between the two sections would be
2945 errors. For example, it would be an error if code in one section called
2946 a function defined in the other section.
2948 The @code{NOCROSSREFS} command takes a list of output section names. If
2949 @command{ld} detects any cross references between the sections, it reports
2950 an error and returns a non-zero exit status. Note that the
2951 @code{NOCROSSREFS} command uses output section names, not input section
2954 @ifclear SingleFormat
2955 @item OUTPUT_ARCH(@var{bfdarch})
2956 @kindex OUTPUT_ARCH(@var{bfdarch})
2957 @cindex machine architecture
2958 @cindex architecture
2959 Specify a particular output machine architecture. The argument is one
2960 of the names used by the BFD library (@pxref{BFD}). You can see the
2961 architecture of an object file by using the @code{objdump} program with
2962 the @samp{-f} option.
2967 @section Assigning Values to Symbols
2968 @cindex assignment in scripts
2969 @cindex symbol definition, scripts
2970 @cindex variables, defining
2971 You may assign a value to a symbol in a linker script. This will define
2972 the symbol and place it into the symbol table with a global scope.
2975 * Simple Assignments:: Simple Assignments
2977 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2978 * Source Code Reference:: How to use a linker script defined symbol in source code
2981 @node Simple Assignments
2982 @subsection Simple Assignments
2984 You may assign to a symbol using any of the C assignment operators:
2987 @item @var{symbol} = @var{expression} ;
2988 @itemx @var{symbol} += @var{expression} ;
2989 @itemx @var{symbol} -= @var{expression} ;
2990 @itemx @var{symbol} *= @var{expression} ;
2991 @itemx @var{symbol} /= @var{expression} ;
2992 @itemx @var{symbol} <<= @var{expression} ;
2993 @itemx @var{symbol} >>= @var{expression} ;
2994 @itemx @var{symbol} &= @var{expression} ;
2995 @itemx @var{symbol} |= @var{expression} ;
2998 The first case will define @var{symbol} to the value of
2999 @var{expression}. In the other cases, @var{symbol} must already be
3000 defined, and the value will be adjusted accordingly.
3002 The special symbol name @samp{.} indicates the location counter. You
3003 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3005 The semicolon after @var{expression} is required.
3007 Expressions are defined below; see @ref{Expressions}.
3009 You may write symbol assignments as commands in their own right, or as
3010 statements within a @code{SECTIONS} command, or as part of an output
3011 section description in a @code{SECTIONS} command.
3013 The section of the symbol will be set from the section of the
3014 expression; for more information, see @ref{Expression Section}.
3016 Here is an example showing the three different places that symbol
3017 assignments may be used:
3028 _bdata = (. + 3) & ~ 3;
3029 .data : @{ *(.data) @}
3033 In this example, the symbol @samp{floating_point} will be defined as
3034 zero. The symbol @samp{_etext} will be defined as the address following
3035 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3036 defined as the address following the @samp{.text} output section aligned
3037 upward to a 4 byte boundary.
3042 In some cases, it is desirable for a linker script to define a symbol
3043 only if it is referenced and is not defined by any object included in
3044 the link. For example, traditional linkers defined the symbol
3045 @samp{etext}. However, ANSI C requires that the user be able to use
3046 @samp{etext} as a function name without encountering an error. The
3047 @code{PROVIDE} keyword may be used to define a symbol, such as
3048 @samp{etext}, only if it is referenced but not defined. The syntax is
3049 @code{PROVIDE(@var{symbol} = @var{expression})}.
3051 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3064 In this example, if the program defines @samp{_etext} (with a leading
3065 underscore), the linker will give a multiple definition error. If, on
3066 the other hand, the program defines @samp{etext} (with no leading
3067 underscore), the linker will silently use the definition in the program.
3068 If the program references @samp{etext} but does not define it, the
3069 linker will use the definition in the linker script.
3071 @node PROVIDE_HIDDEN
3072 @subsection PROVIDE_HIDDEN
3073 @cindex PROVIDE_HIDDEN
3074 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3075 hidden and won't be exported.
3077 @node Source Code Reference
3078 @subsection Source Code Reference
3080 Accessing a linker script defined variable from source code is not
3081 intuitive. In particular a linker script symbol is not equivalent to
3082 a variable declaration in a high level language, it is instead a
3083 symbol that does not have a value.
3085 Before going further, it is important to note that compilers often
3086 transform names in the source code into different names when they are
3087 stored in the symbol table. For example, Fortran compilers commonly
3088 prepend or append an underscore, and C++ performs extensive @samp{name
3089 mangling}. Therefore there might be a discrepancy between the name
3090 of a variable as it is used in source code and the name of the same
3091 variable as it is defined in a linker script. For example in C a
3092 linker script variable might be referred to as:
3098 But in the linker script it might be defined as:
3104 In the remaining examples however it is assumed that no name
3105 transformation has taken place.
3107 When a symbol is declared in a high level language such as C, two
3108 things happen. The first is that the compiler reserves enough space
3109 in the program's memory to hold the @emph{value} of the symbol. The
3110 second is that the compiler creates an entry in the program's symbol
3111 table which holds the symbol's @emph{address}. ie the symbol table
3112 contains the address of the block of memory holding the symbol's
3113 value. So for example the following C declaration, at file scope:
3119 creates a entry called @samp{foo} in the symbol table. This entry
3120 holds the address of an @samp{int} sized block of memory where the
3121 number 1000 is initially stored.
3123 When a program references a symbol the compiler generates code that
3124 first accesses the symbol table to find the address of the symbol's
3125 memory block and then code to read the value from that memory block.
3132 looks up the symbol @samp{foo} in the symbol table, gets the address
3133 associated with this symbol and then writes the value 1 into that
3140 looks up the symbol @samp{foo} in the symbol table, gets it address
3141 and then copies this address into the block of memory associated with
3142 the variable @samp{a}.
3144 Linker scripts symbol declarations, by contrast, create an entry in
3145 the symbol table but do not assign any memory to them. Thus they are
3146 an address without a value. So for example the linker script definition:
3152 creates an entry in the symbol table called @samp{foo} which holds
3153 the address of memory location 1000, but nothing special is stored at
3154 address 1000. This means that you cannot access the @emph{value} of a
3155 linker script defined symbol - it has no value - all you can do is
3156 access the @emph{address} of a linker script defined symbol.
3158 Hence when you are using a linker script defined symbol in source code
3159 you should always take the address of the symbol, and never attempt to
3160 use its value. For example suppose you want to copy the contents of a
3161 section of memory called .ROM into a section called .FLASH and the
3162 linker script contains these declarations:
3166 start_of_ROM = .ROM;
3167 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3168 start_of_FLASH = .FLASH;
3172 Then the C source code to perform the copy would be:
3176 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3178 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3182 Note the use of the @samp{&} operators. These are correct.
3185 @section SECTIONS Command
3187 The @code{SECTIONS} command tells the linker how to map input sections
3188 into output sections, and how to place the output sections in memory.
3190 The format of the @code{SECTIONS} command is:
3194 @var{sections-command}
3195 @var{sections-command}
3200 Each @var{sections-command} may of be one of the following:
3204 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3206 a symbol assignment (@pxref{Assignments})
3208 an output section description
3210 an overlay description
3213 The @code{ENTRY} command and symbol assignments are permitted inside the
3214 @code{SECTIONS} command for convenience in using the location counter in
3215 those commands. This can also make the linker script easier to
3216 understand because you can use those commands at meaningful points in
3217 the layout of the output file.
3219 Output section descriptions and overlay descriptions are described
3222 If you do not use a @code{SECTIONS} command in your linker script, the
3223 linker will place each input section into an identically named output
3224 section in the order that the sections are first encountered in the
3225 input files. If all input sections are present in the first file, for
3226 example, the order of sections in the output file will match the order
3227 in the first input file. The first section will be at address zero.
3230 * Output Section Description:: Output section description
3231 * Output Section Name:: Output section name
3232 * Output Section Address:: Output section address
3233 * Input Section:: Input section description
3234 * Output Section Data:: Output section data
3235 * Output Section Keywords:: Output section keywords
3236 * Output Section Discarding:: Output section discarding
3237 * Output Section Attributes:: Output section attributes
3238 * Overlay Description:: Overlay description
3241 @node Output Section Description
3242 @subsection Output Section Description
3243 The full description of an output section looks like this:
3246 @var{section} [@var{address}] [(@var{type})] :
3247 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3249 @var{output-section-command}
3250 @var{output-section-command}
3252 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3256 Most output sections do not use most of the optional section attributes.
3258 The whitespace around @var{section} is required, so that the section
3259 name is unambiguous. The colon and the curly braces are also required.
3260 The line breaks and other white space are optional.
3262 Each @var{output-section-command} may be one of the following:
3266 a symbol assignment (@pxref{Assignments})
3268 an input section description (@pxref{Input Section})
3270 data values to include directly (@pxref{Output Section Data})
3272 a special output section keyword (@pxref{Output Section Keywords})
3275 @node Output Section Name
3276 @subsection Output Section Name
3277 @cindex name, section
3278 @cindex section name
3279 The name of the output section is @var{section}. @var{section} must
3280 meet the constraints of your output format. In formats which only
3281 support a limited number of sections, such as @code{a.out}, the name
3282 must be one of the names supported by the format (@code{a.out}, for
3283 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3284 output format supports any number of sections, but with numbers and not
3285 names (as is the case for Oasys), the name should be supplied as a
3286 quoted numeric string. A section name may consist of any sequence of
3287 characters, but a name which contains any unusual characters such as
3288 commas must be quoted.
3290 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3293 @node Output Section Address
3294 @subsection Output Section Address
3295 @cindex address, section
3296 @cindex section address
3297 The @var{address} is an expression for the VMA (the virtual memory
3298 address) of the output section. If you do not provide @var{address},
3299 the linker will set it based on @var{region} if present, or otherwise
3300 based on the current value of the location counter.
3302 If you provide @var{address}, the address of the output section will be
3303 set to precisely that. If you provide neither @var{address} nor
3304 @var{region}, then the address of the output section will be set to the
3305 current value of the location counter aligned to the alignment
3306 requirements of the output section. The alignment requirement of the
3307 output section is the strictest alignment of any input section contained
3308 within the output section.
3312 .text . : @{ *(.text) @}
3317 .text : @{ *(.text) @}
3320 are subtly different. The first will set the address of the
3321 @samp{.text} output section to the current value of the location
3322 counter. The second will set it to the current value of the location
3323 counter aligned to the strictest alignment of a @samp{.text} input
3326 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3327 For example, if you want to align the section on a 0x10 byte boundary,
3328 so that the lowest four bits of the section address are zero, you could
3329 do something like this:
3331 .text ALIGN(0x10) : @{ *(.text) @}
3334 This works because @code{ALIGN} returns the current location counter
3335 aligned upward to the specified value.
3337 Specifying @var{address} for a section will change the value of the
3341 @subsection Input Section Description
3342 @cindex input sections
3343 @cindex mapping input sections to output sections
3344 The most common output section command is an input section description.
3346 The input section description is the most basic linker script operation.
3347 You use output sections to tell the linker how to lay out your program
3348 in memory. You use input section descriptions to tell the linker how to
3349 map the input files into your memory layout.
3352 * Input Section Basics:: Input section basics
3353 * Input Section Wildcards:: Input section wildcard patterns
3354 * Input Section Common:: Input section for common symbols
3355 * Input Section Keep:: Input section and garbage collection
3356 * Input Section Example:: Input section example
3359 @node Input Section Basics
3360 @subsubsection Input Section Basics
3361 @cindex input section basics
3362 An input section description consists of a file name optionally followed
3363 by a list of section names in parentheses.
3365 The file name and the section name may be wildcard patterns, which we
3366 describe further below (@pxref{Input Section Wildcards}).
3368 The most common input section description is to include all input
3369 sections with a particular name in the output section. For example, to
3370 include all input @samp{.text} sections, you would write:
3375 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3376 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3377 match all files except the ones specified in the EXCLUDE_FILE list. For
3380 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3382 will cause all .ctors sections from all files except @file{crtend.o} and
3383 @file{otherfile.o} to be included.
3385 There are two ways to include more than one section:
3391 The difference between these is the order in which the @samp{.text} and
3392 @samp{.rdata} input sections will appear in the output section. In the
3393 first example, they will be intermingled, appearing in the same order as
3394 they are found in the linker input. In the second example, all
3395 @samp{.text} input sections will appear first, followed by all
3396 @samp{.rdata} input sections.
3398 You can specify a file name to include sections from a particular file.
3399 You would do this if one or more of your files contain special data that
3400 needs to be at a particular location in memory. For example:
3405 If you use a file name without a list of sections, then all sections in
3406 the input file will be included in the output section. This is not
3407 commonly done, but it may by useful on occasion. For example:
3412 When you use a file name which does not contain any wild card
3413 characters, the linker will first see if you also specified the file
3414 name on the linker command line or in an @code{INPUT} command. If you
3415 did not, the linker will attempt to open the file as an input file, as
3416 though it appeared on the command line. Note that this differs from an
3417 @code{INPUT} command, because the linker will not search for the file in
3418 the archive search path.
3420 @node Input Section Wildcards
3421 @subsubsection Input Section Wildcard Patterns
3422 @cindex input section wildcards
3423 @cindex wildcard file name patterns
3424 @cindex file name wildcard patterns
3425 @cindex section name wildcard patterns
3426 In an input section description, either the file name or the section
3427 name or both may be wildcard patterns.
3429 The file name of @samp{*} seen in many examples is a simple wildcard
3430 pattern for the file name.
3432 The wildcard patterns are like those used by the Unix shell.
3436 matches any number of characters
3438 matches any single character
3440 matches a single instance of any of the @var{chars}; the @samp{-}
3441 character may be used to specify a range of characters, as in
3442 @samp{[a-z]} to match any lower case letter
3444 quotes the following character
3447 When a file name is matched with a wildcard, the wildcard characters
3448 will not match a @samp{/} character (used to separate directory names on
3449 Unix). A pattern consisting of a single @samp{*} character is an
3450 exception; it will always match any file name, whether it contains a
3451 @samp{/} or not. In a section name, the wildcard characters will match
3452 a @samp{/} character.
3454 File name wildcard patterns only match files which are explicitly
3455 specified on the command line or in an @code{INPUT} command. The linker
3456 does not search directories to expand wildcards.
3458 If a file name matches more than one wildcard pattern, or if a file name
3459 appears explicitly and is also matched by a wildcard pattern, the linker
3460 will use the first match in the linker script. For example, this
3461 sequence of input section descriptions is probably in error, because the
3462 @file{data.o} rule will not be used:
3464 .data : @{ *(.data) @}
3465 .data1 : @{ data.o(.data) @}
3468 @cindex SORT_BY_NAME
3469 Normally, the linker will place files and sections matched by wildcards
3470 in the order in which they are seen during the link. You can change
3471 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3472 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
3473 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3474 into ascending order by name before placing them in the output file.
3476 @cindex SORT_BY_ALIGNMENT
3477 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3478 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3479 ascending order by alignment before placing them in the output file.
3482 @code{SORT} is an alias for @code{SORT_BY_NAME}.
3484 When there are nested section sorting commands in linker script, there
3485 can be at most 1 level of nesting for section sorting commands.
3489 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3490 It will sort the input sections by name first, then by alignment if 2
3491 sections have the same name.
3493 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3494 It will sort the input sections by alignment first, then by name if 2
3495 sections have the same alignment.
3497 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3498 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3500 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3501 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3503 All other nested section sorting commands are invalid.
3506 When both command line section sorting option and linker script
3507 section sorting command are used, section sorting command always
3508 takes precedence over the command line option.
3510 If the section sorting command in linker script isn't nested, the
3511 command line option will make the section sorting command to be
3512 treated as nested sorting command.
3516 @code{SORT_BY_NAME} (wildcard section pattern ) with
3517 @option{--sort-sections alignment} is equivalent to
3518 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3520 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3521 @option{--sort-section name} is equivalent to
3522 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3525 If the section sorting command in linker script is nested, the
3526 command line option will be ignored.
3528 If you ever get confused about where input sections are going, use the
3529 @samp{-M} linker option to generate a map file. The map file shows
3530 precisely how input sections are mapped to output sections.
3532 This example shows how wildcard patterns might be used to partition
3533 files. This linker script directs the linker to place all @samp{.text}
3534 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3535 The linker will place the @samp{.data} section from all files beginning
3536 with an upper case character in @samp{.DATA}; for all other files, the
3537 linker will place the @samp{.data} section in @samp{.data}.
3541 .text : @{ *(.text) @}
3542 .DATA : @{ [A-Z]*(.data) @}
3543 .data : @{ *(.data) @}
3544 .bss : @{ *(.bss) @}
3549 @node Input Section Common
3550 @subsubsection Input Section for Common Symbols
3551 @cindex common symbol placement
3552 @cindex uninitialized data placement
3553 A special notation is needed for common symbols, because in many object
3554 file formats common symbols do not have a particular input section. The
3555 linker treats common symbols as though they are in an input section
3556 named @samp{COMMON}.
3558 You may use file names with the @samp{COMMON} section just as with any
3559 other input sections. You can use this to place common symbols from a
3560 particular input file in one section while common symbols from other
3561 input files are placed in another section.
3563 In most cases, common symbols in input files will be placed in the
3564 @samp{.bss} section in the output file. For example:
3566 .bss @{ *(.bss) *(COMMON) @}
3569 @cindex scommon section
3570 @cindex small common symbols
3571 Some object file formats have more than one type of common symbol. For
3572 example, the MIPS ELF object file format distinguishes standard common
3573 symbols and small common symbols. In this case, the linker will use a
3574 different special section name for other types of common symbols. In
3575 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3576 symbols and @samp{.scommon} for small common symbols. This permits you
3577 to map the different types of common symbols into memory at different
3581 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3582 notation is now considered obsolete. It is equivalent to
3585 @node Input Section Keep
3586 @subsubsection Input Section and Garbage Collection
3588 @cindex garbage collection
3589 When link-time garbage collection is in use (@samp{--gc-sections}),
3590 it is often useful to mark sections that should not be eliminated.
3591 This is accomplished by surrounding an input section's wildcard entry
3592 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3593 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3595 @node Input Section Example
3596 @subsubsection Input Section Example
3597 The following example is a complete linker script. It tells the linker
3598 to read all of the sections from file @file{all.o} and place them at the
3599 start of output section @samp{outputa} which starts at location
3600 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3601 follows immediately, in the same output section. All of section
3602 @samp{.input2} from @file{foo.o} goes into output section
3603 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3604 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3605 files are written to output section @samp{outputc}.
3633 @node Output Section Data
3634 @subsection Output Section Data
3636 @cindex section data
3637 @cindex output section data
3638 @kindex BYTE(@var{expression})
3639 @kindex SHORT(@var{expression})
3640 @kindex LONG(@var{expression})
3641 @kindex QUAD(@var{expression})
3642 @kindex SQUAD(@var{expression})
3643 You can include explicit bytes of data in an output section by using
3644 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3645 an output section command. Each keyword is followed by an expression in
3646 parentheses providing the value to store (@pxref{Expressions}). The
3647 value of the expression is stored at the current value of the location
3650 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3651 store one, two, four, and eight bytes (respectively). After storing the
3652 bytes, the location counter is incremented by the number of bytes
3655 For example, this will store the byte 1 followed by the four byte value
3656 of the symbol @samp{addr}:
3662 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3663 same; they both store an 8 byte, or 64 bit, value. When both host and
3664 target are 32 bits, an expression is computed as 32 bits. In this case
3665 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3666 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
3668 If the object file format of the output file has an explicit endianness,
3669 which is the normal case, the value will be stored in that endianness.
3670 When the object file format does not have an explicit endianness, as is
3671 true of, for example, S-records, the value will be stored in the
3672 endianness of the first input object file.
3674 Note---these commands only work inside a section description and not
3675 between them, so the following will produce an error from the linker:
3677 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
3679 whereas this will work:
3681 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
3684 @kindex FILL(@var{expression})
3685 @cindex holes, filling
3686 @cindex unspecified memory
3687 You may use the @code{FILL} command to set the fill pattern for the
3688 current section. It is followed by an expression in parentheses. Any
3689 otherwise unspecified regions of memory within the section (for example,
3690 gaps left due to the required alignment of input sections) are filled
3691 with the value of the expression, repeated as
3692 necessary. A @code{FILL} statement covers memory locations after the
3693 point at which it occurs in the section definition; by including more
3694 than one @code{FILL} statement, you can have different fill patterns in
3695 different parts of an output section.
3697 This example shows how to fill unspecified regions of memory with the
3703 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
3704 section attribute, but it only affects the
3705 part of the section following the @code{FILL} command, rather than the
3706 entire section. If both are used, the @code{FILL} command takes
3707 precedence. @xref{Output Section Fill}, for details on the fill
3710 @node Output Section Keywords
3711 @subsection Output Section Keywords
3712 There are a couple of keywords which can appear as output section
3716 @kindex CREATE_OBJECT_SYMBOLS
3717 @cindex input filename symbols
3718 @cindex filename symbols
3719 @item CREATE_OBJECT_SYMBOLS
3720 The command tells the linker to create a symbol for each input file.
3721 The name of each symbol will be the name of the corresponding input
3722 file. The section of each symbol will be the output section in which
3723 the @code{CREATE_OBJECT_SYMBOLS} command appears.
3725 This is conventional for the a.out object file format. It is not
3726 normally used for any other object file format.
3728 @kindex CONSTRUCTORS
3729 @cindex C++ constructors, arranging in link
3730 @cindex constructors, arranging in link
3732 When linking using the a.out object file format, the linker uses an
3733 unusual set construct to support C++ global constructors and
3734 destructors. When linking object file formats which do not support
3735 arbitrary sections, such as ECOFF and XCOFF, the linker will
3736 automatically recognize C++ global constructors and destructors by name.
3737 For these object file formats, the @code{CONSTRUCTORS} command tells the
3738 linker to place constructor information in the output section where the
3739 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
3740 ignored for other object file formats.
3742 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
3743 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
3744 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
3745 the start and end of the global destructors. The
3746 first word in the list is the number of entries, followed by the address
3747 of each constructor or destructor, followed by a zero word. The
3748 compiler must arrange to actually run the code. For these object file
3749 formats @sc{gnu} C++ normally calls constructors from a subroutine
3750 @code{__main}; a call to @code{__main} is automatically inserted into
3751 the startup code for @code{main}. @sc{gnu} C++ normally runs
3752 destructors either by using @code{atexit}, or directly from the function
3755 For object file formats such as @code{COFF} or @code{ELF} which support
3756 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
3757 addresses of global constructors and destructors into the @code{.ctors}
3758 and @code{.dtors} sections. Placing the following sequence into your
3759 linker script will build the sort of table which the @sc{gnu} C++
3760 runtime code expects to see.
3764 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3769 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3775 If you are using the @sc{gnu} C++ support for initialization priority,
3776 which provides some control over the order in which global constructors
3777 are run, you must sort the constructors at link time to ensure that they
3778 are executed in the correct order. When using the @code{CONSTRUCTORS}
3779 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
3780 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
3781 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
3784 Normally the compiler and linker will handle these issues automatically,
3785 and you will not need to concern yourself with them. However, you may
3786 need to consider this if you are using C++ and writing your own linker
3791 @node Output Section Discarding
3792 @subsection Output Section Discarding
3793 @cindex discarding sections
3794 @cindex sections, discarding
3795 @cindex removing sections
3796 The linker will not create output sections with no contents. This is
3797 for convenience when referring to input sections that may or may not
3798 be present in any of the input files. For example:
3800 .foo : @{ *(.foo) @}
3803 will only create a @samp{.foo} section in the output file if there is a
3804 @samp{.foo} section in at least one input file, and if the input
3805 sections are not all empty. Other link script directives that allocate
3806 space in an output section will also create the output section.
3808 The linker will ignore address assignments (@pxref{Output Section Address})
3809 on discarded output sections, except when the linker script defines
3810 symbols in the output section. In that case the linker will obey
3811 the address assignments, possibly advancing dot even though the
3812 section is discarded.
3815 The special output section name @samp{/DISCARD/} may be used to discard
3816 input sections. Any input sections which are assigned to an output
3817 section named @samp{/DISCARD/} are not included in the output file.
3819 @node Output Section Attributes
3820 @subsection Output Section Attributes
3821 @cindex output section attributes
3822 We showed above that the full description of an output section looked
3826 @var{section} [@var{address}] [(@var{type})] :
3827 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3829 @var{output-section-command}
3830 @var{output-section-command}
3832 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3835 We've already described @var{section}, @var{address}, and
3836 @var{output-section-command}. In this section we will describe the
3837 remaining section attributes.
3840 * Output Section Type:: Output section type
3841 * Output Section LMA:: Output section LMA
3842 * Forced Output Alignment:: Forced Output Alignment
3843 * Forced Input Alignment:: Forced Input Alignment
3844 * Output Section Region:: Output section region
3845 * Output Section Phdr:: Output section phdr
3846 * Output Section Fill:: Output section fill
3849 @node Output Section Type
3850 @subsubsection Output Section Type
3851 Each output section may have a type. The type is a keyword in
3852 parentheses. The following types are defined:
3856 The section should be marked as not loadable, so that it will not be
3857 loaded into memory when the program is run.
3862 These type names are supported for backward compatibility, and are
3863 rarely used. They all have the same effect: the section should be
3864 marked as not allocatable, so that no memory is allocated for the
3865 section when the program is run.
3869 @cindex prevent unnecessary loading
3870 @cindex loading, preventing
3871 The linker normally sets the attributes of an output section based on
3872 the input sections which map into it. You can override this by using
3873 the section type. For example, in the script sample below, the
3874 @samp{ROM} section is addressed at memory location @samp{0} and does not
3875 need to be loaded when the program is run. The contents of the
3876 @samp{ROM} section will appear in the linker output file as usual.
3880 ROM 0 (NOLOAD) : @{ @dots{} @}
3886 @node Output Section LMA
3887 @subsubsection Output Section LMA
3888 @kindex AT>@var{lma_region}
3889 @kindex AT(@var{lma})
3890 @cindex load address
3891 @cindex section load address
3892 Every section has a virtual address (VMA) and a load address (LMA); see
3893 @ref{Basic Script Concepts}. The address expression which may appear in
3894 an output section description sets the VMA (@pxref{Output Section
3897 The expression @var{lma} that follows the @code{AT} keyword specifies
3898 the load address of the section.
3900 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
3901 specify a memory region for the section's load address. @xref{MEMORY}.
3902 Note that if the section has not had a VMA assigned to it then the
3903 linker will use the @var{lma_region} as the VMA region as well.
3905 If neither @code{AT} nor @code{AT>} is specified for an allocatable
3906 section, the linker will set the LMA such that the difference between
3907 VMA and LMA for the section is the same as the preceding output
3908 section in the same region. If there is no preceding output section
3909 or the section is not allocatable, the linker will set the LMA equal
3911 @xref{Output Section Region}.
3913 @cindex ROM initialized data
3914 @cindex initialized data in ROM
3915 This feature is designed to make it easy to build a ROM image. For
3916 example, the following linker script creates three output sections: one
3917 called @samp{.text}, which starts at @code{0x1000}, one called
3918 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
3919 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
3920 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
3921 defined with the value @code{0x2000}, which shows that the location
3922 counter holds the VMA value, not the LMA value.
3928 .text 0x1000 : @{ *(.text) _etext = . ; @}
3930 AT ( ADDR (.text) + SIZEOF (.text) )
3931 @{ _data = . ; *(.data); _edata = . ; @}
3933 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
3938 The run-time initialization code for use with a program generated with
3939 this linker script would include something like the following, to copy
3940 the initialized data from the ROM image to its runtime address. Notice
3941 how this code takes advantage of the symbols defined by the linker
3946 extern char _etext, _data, _edata, _bstart, _bend;
3947 char *src = &_etext;
3950 /* ROM has data at end of text; copy it. */
3951 while (dst < &_edata) @{
3956 for (dst = &_bstart; dst< &_bend; dst++)
3961 @node Forced Output Alignment
3962 @subsubsection Forced Output Alignment
3963 @kindex ALIGN(@var{section_align})
3964 @cindex forcing output section alignment
3965 @cindex output section alignment
3966 You can increase an output section's alignment by using ALIGN.
3968 @node Forced Input Alignment
3969 @subsubsection Forced Input Alignment
3970 @kindex SUBALIGN(@var{subsection_align})
3971 @cindex forcing input section alignment
3972 @cindex input section alignment
3973 You can force input section alignment within an output section by using
3974 SUBALIGN. The value specified overrides any alignment given by input
3975 sections, whether larger or smaller.
3977 @node Output Section Region
3978 @subsubsection Output Section Region
3979 @kindex >@var{region}
3980 @cindex section, assigning to memory region
3981 @cindex memory regions and sections
3982 You can assign a section to a previously defined region of memory by
3983 using @samp{>@var{region}}. @xref{MEMORY}.
3985 Here is a simple example:
3988 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
3989 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
3993 @node Output Section Phdr
3994 @subsubsection Output Section Phdr
3996 @cindex section, assigning to program header
3997 @cindex program headers and sections
3998 You can assign a section to a previously defined program segment by
3999 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4000 one or more segments, then all subsequent allocated sections will be
4001 assigned to those segments as well, unless they use an explicitly
4002 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4003 linker to not put the section in any segment at all.
4005 Here is a simple example:
4008 PHDRS @{ text PT_LOAD ; @}
4009 SECTIONS @{ .text : @{ *(.text) @} :text @}
4013 @node Output Section Fill
4014 @subsubsection Output Section Fill
4015 @kindex =@var{fillexp}
4016 @cindex section fill pattern
4017 @cindex fill pattern, entire section
4018 You can set the fill pattern for an entire section by using
4019 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4020 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4021 within the output section (for example, gaps left due to the required
4022 alignment of input sections) will be filled with the value, repeated as
4023 necessary. If the fill expression is a simple hex number, ie. a string
4024 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4025 an arbitrarily long sequence of hex digits can be used to specify the
4026 fill pattern; Leading zeros become part of the pattern too. For all
4027 other cases, including extra parentheses or a unary @code{+}, the fill
4028 pattern is the four least significant bytes of the value of the
4029 expression. In all cases, the number is big-endian.
4031 You can also change the fill value with a @code{FILL} command in the
4032 output section commands; (@pxref{Output Section Data}).
4034 Here is a simple example:
4037 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4041 @node Overlay Description
4042 @subsection Overlay Description
4045 An overlay description provides an easy way to describe sections which
4046 are to be loaded as part of a single memory image but are to be run at
4047 the same memory address. At run time, some sort of overlay manager will
4048 copy the overlaid sections in and out of the runtime memory address as
4049 required, perhaps by simply manipulating addressing bits. This approach
4050 can be useful, for example, when a certain region of memory is faster
4053 Overlays are described using the @code{OVERLAY} command. The
4054 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4055 output section description. The full syntax of the @code{OVERLAY}
4056 command is as follows:
4059 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4063 @var{output-section-command}
4064 @var{output-section-command}
4066 @} [:@var{phdr}@dots{}] [=@var{fill}]
4069 @var{output-section-command}
4070 @var{output-section-command}
4072 @} [:@var{phdr}@dots{}] [=@var{fill}]
4074 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4078 Everything is optional except @code{OVERLAY} (a keyword), and each
4079 section must have a name (@var{secname1} and @var{secname2} above). The
4080 section definitions within the @code{OVERLAY} construct are identical to
4081 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
4082 except that no addresses and no memory regions may be defined for
4083 sections within an @code{OVERLAY}.
4085 The sections are all defined with the same starting address. The load
4086 addresses of the sections are arranged such that they are consecutive in
4087 memory starting at the load address used for the @code{OVERLAY} as a
4088 whole (as with normal section definitions, the load address is optional,
4089 and defaults to the start address; the start address is also optional,
4090 and defaults to the current value of the location counter).
4092 If the @code{NOCROSSREFS} keyword is used, and there any references
4093 among the sections, the linker will report an error. Since the sections
4094 all run at the same address, it normally does not make sense for one
4095 section to refer directly to another. @xref{Miscellaneous Commands,
4098 For each section within the @code{OVERLAY}, the linker automatically
4099 provides two symbols. The symbol @code{__load_start_@var{secname}} is
4100 defined as the starting load address of the section. The symbol
4101 @code{__load_stop_@var{secname}} is defined as the final load address of
4102 the section. Any characters within @var{secname} which are not legal
4103 within C identifiers are removed. C (or assembler) code may use these
4104 symbols to move the overlaid sections around as necessary.
4106 At the end of the overlay, the value of the location counter is set to
4107 the start address of the overlay plus the size of the largest section.
4109 Here is an example. Remember that this would appear inside a
4110 @code{SECTIONS} construct.
4113 OVERLAY 0x1000 : AT (0x4000)
4115 .text0 @{ o1/*.o(.text) @}
4116 .text1 @{ o2/*.o(.text) @}
4121 This will define both @samp{.text0} and @samp{.text1} to start at
4122 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
4123 @samp{.text1} will be loaded immediately after @samp{.text0}. The
4124 following symbols will be defined if referenced: @code{__load_start_text0},
4125 @code{__load_stop_text0}, @code{__load_start_text1},
4126 @code{__load_stop_text1}.
4128 C code to copy overlay @code{.text1} into the overlay area might look
4133 extern char __load_start_text1, __load_stop_text1;
4134 memcpy ((char *) 0x1000, &__load_start_text1,
4135 &__load_stop_text1 - &__load_start_text1);
4139 Note that the @code{OVERLAY} command is just syntactic sugar, since
4140 everything it does can be done using the more basic commands. The above
4141 example could have been written identically as follows.
4145 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4146 PROVIDE (__load_start_text0 = LOADADDR (.text0));
4147 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4148 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4149 PROVIDE (__load_start_text1 = LOADADDR (.text1));
4150 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4151 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4156 @section MEMORY Command
4158 @cindex memory regions
4159 @cindex regions of memory
4160 @cindex allocating memory
4161 @cindex discontinuous memory
4162 The linker's default configuration permits allocation of all available
4163 memory. You can override this by using the @code{MEMORY} command.
4165 The @code{MEMORY} command describes the location and size of blocks of
4166 memory in the target. You can use it to describe which memory regions
4167 may be used by the linker, and which memory regions it must avoid. You
4168 can then assign sections to particular memory regions. The linker will
4169 set section addresses based on the memory regions, and will warn about
4170 regions that become too full. The linker will not shuffle sections
4171 around to fit into the available regions.
4173 A linker script may contain at most one use of the @code{MEMORY}
4174 command. However, you can define as many blocks of memory within it as
4175 you wish. The syntax is:
4180 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4186 The @var{name} is a name used in the linker script to refer to the
4187 region. The region name has no meaning outside of the linker script.
4188 Region names are stored in a separate name space, and will not conflict
4189 with symbol names, file names, or section names. Each memory region
4190 must have a distinct name.
4192 @cindex memory region attributes
4193 The @var{attr} string is an optional list of attributes that specify
4194 whether to use a particular memory region for an input section which is
4195 not explicitly mapped in the linker script. As described in
4196 @ref{SECTIONS}, if you do not specify an output section for some input
4197 section, the linker will create an output section with the same name as
4198 the input section. If you define region attributes, the linker will use
4199 them to select the memory region for the output section that it creates.
4201 The @var{attr} string must consist only of the following characters:
4216 Invert the sense of any of the preceding attributes
4219 If a unmapped section matches any of the listed attributes other than
4220 @samp{!}, it will be placed in the memory region. The @samp{!}
4221 attribute reverses this test, so that an unmapped section will be placed
4222 in the memory region only if it does not match any of the listed
4228 The @var{origin} is an numerical expression for the start address of
4229 the memory region. The expression must evaluate to a constant and it
4230 cannot involve any symbols. The keyword @code{ORIGIN} may be
4231 abbreviated to @code{org} or @code{o} (but not, for example,
4237 The @var{len} is an expression for the size in bytes of the memory
4238 region. As with the @var{origin} expression, the expression must
4239 be numerical only and must evaluate to a constant. The keyword
4240 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4242 In the following example, we specify that there are two memory regions
4243 available for allocation: one starting at @samp{0} for 256 kilobytes,
4244 and the other starting at @samp{0x40000000} for four megabytes. The
4245 linker will place into the @samp{rom} memory region every section which
4246 is not explicitly mapped into a memory region, and is either read-only
4247 or executable. The linker will place other sections which are not
4248 explicitly mapped into a memory region into the @samp{ram} memory
4255 rom (rx) : ORIGIN = 0, LENGTH = 256K
4256 ram (!rx) : org = 0x40000000, l = 4M
4261 Once you define a memory region, you can direct the linker to place
4262 specific output sections into that memory region by using the
4263 @samp{>@var{region}} output section attribute. For example, if you have
4264 a memory region named @samp{mem}, you would use @samp{>mem} in the
4265 output section definition. @xref{Output Section Region}. If no address
4266 was specified for the output section, the linker will set the address to
4267 the next available address within the memory region. If the combined
4268 output sections directed to a memory region are too large for the
4269 region, the linker will issue an error message.
4271 It is possible to access the origin and length of a memory in an
4272 expression via the @code{ORIGIN(@var{memory})} and
4273 @code{LENGTH(@var{memory})} functions:
4277 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4282 @section PHDRS Command
4284 @cindex program headers
4285 @cindex ELF program headers
4286 @cindex program segments
4287 @cindex segments, ELF
4288 The ELF object file format uses @dfn{program headers}, also knows as
4289 @dfn{segments}. The program headers describe how the program should be
4290 loaded into memory. You can print them out by using the @code{objdump}
4291 program with the @samp{-p} option.
4293 When you run an ELF program on a native ELF system, the system loader
4294 reads the program headers in order to figure out how to load the
4295 program. This will only work if the program headers are set correctly.
4296 This manual does not describe the details of how the system loader
4297 interprets program headers; for more information, see the ELF ABI.
4299 The linker will create reasonable program headers by default. However,
4300 in some cases, you may need to specify the program headers more
4301 precisely. You may use the @code{PHDRS} command for this purpose. When
4302 the linker sees the @code{PHDRS} command in the linker script, it will
4303 not create any program headers other than the ones specified.
4305 The linker only pays attention to the @code{PHDRS} command when
4306 generating an ELF output file. In other cases, the linker will simply
4307 ignore @code{PHDRS}.
4309 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4310 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4316 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4317 [ FLAGS ( @var{flags} ) ] ;
4322 The @var{name} is used only for reference in the @code{SECTIONS} command
4323 of the linker script. It is not put into the output file. Program
4324 header names are stored in a separate name space, and will not conflict
4325 with symbol names, file names, or section names. Each program header
4326 must have a distinct name.
4328 Certain program header types describe segments of memory which the
4329 system loader will load from the file. In the linker script, you
4330 specify the contents of these segments by placing allocatable output
4331 sections in the segments. You use the @samp{:@var{phdr}} output section
4332 attribute to place a section in a particular segment. @xref{Output
4335 It is normal to put certain sections in more than one segment. This
4336 merely implies that one segment of memory contains another. You may
4337 repeat @samp{:@var{phdr}}, using it once for each segment which should
4338 contain the section.
4340 If you place a section in one or more segments using @samp{:@var{phdr}},
4341 then the linker will place all subsequent allocatable sections which do
4342 not specify @samp{:@var{phdr}} in the same segments. This is for
4343 convenience, since generally a whole set of contiguous sections will be
4344 placed in a single segment. You can use @code{:NONE} to override the
4345 default segment and tell the linker to not put the section in any
4350 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4351 the program header type to further describe the contents of the segment.
4352 The @code{FILEHDR} keyword means that the segment should include the ELF
4353 file header. The @code{PHDRS} keyword means that the segment should
4354 include the ELF program headers themselves.
4356 The @var{type} may be one of the following. The numbers indicate the
4357 value of the keyword.
4360 @item @code{PT_NULL} (0)
4361 Indicates an unused program header.
4363 @item @code{PT_LOAD} (1)
4364 Indicates that this program header describes a segment to be loaded from
4367 @item @code{PT_DYNAMIC} (2)
4368 Indicates a segment where dynamic linking information can be found.
4370 @item @code{PT_INTERP} (3)
4371 Indicates a segment where the name of the program interpreter may be
4374 @item @code{PT_NOTE} (4)
4375 Indicates a segment holding note information.
4377 @item @code{PT_SHLIB} (5)
4378 A reserved program header type, defined but not specified by the ELF
4381 @item @code{PT_PHDR} (6)
4382 Indicates a segment where the program headers may be found.
4384 @item @var{expression}
4385 An expression giving the numeric type of the program header. This may
4386 be used for types not defined above.
4389 You can specify that a segment should be loaded at a particular address
4390 in memory by using an @code{AT} expression. This is identical to the
4391 @code{AT} command used as an output section attribute (@pxref{Output
4392 Section LMA}). The @code{AT} command for a program header overrides the
4393 output section attribute.
4395 The linker will normally set the segment flags based on the sections
4396 which comprise the segment. You may use the @code{FLAGS} keyword to
4397 explicitly specify the segment flags. The value of @var{flags} must be
4398 an integer. It is used to set the @code{p_flags} field of the program
4401 Here is an example of @code{PHDRS}. This shows a typical set of program
4402 headers used on a native ELF system.
4408 headers PT_PHDR PHDRS ;
4410 text PT_LOAD FILEHDR PHDRS ;
4412 dynamic PT_DYNAMIC ;
4418 .interp : @{ *(.interp) @} :text :interp
4419 .text : @{ *(.text) @} :text
4420 .rodata : @{ *(.rodata) @} /* defaults to :text */
4422 . = . + 0x1000; /* move to a new page in memory */
4423 .data : @{ *(.data) @} :data
4424 .dynamic : @{ *(.dynamic) @} :data :dynamic
4431 @section VERSION Command
4432 @kindex VERSION @{script text@}
4433 @cindex symbol versions
4434 @cindex version script
4435 @cindex versions of symbols
4436 The linker supports symbol versions when using ELF. Symbol versions are
4437 only useful when using shared libraries. The dynamic linker can use
4438 symbol versions to select a specific version of a function when it runs
4439 a program that may have been linked against an earlier version of the
4442 You can include a version script directly in the main linker script, or
4443 you can supply the version script as an implicit linker script. You can
4444 also use the @samp{--version-script} linker option.
4446 The syntax of the @code{VERSION} command is simply
4448 VERSION @{ version-script-commands @}
4451 The format of the version script commands is identical to that used by
4452 Sun's linker in Solaris 2.5. The version script defines a tree of
4453 version nodes. You specify the node names and interdependencies in the
4454 version script. You can specify which symbols are bound to which
4455 version nodes, and you can reduce a specified set of symbols to local
4456 scope so that they are not globally visible outside of the shared
4459 The easiest way to demonstrate the version script language is with a few
4480 "int f(int, double)";
4485 This example version script defines three version nodes. The first
4486 version node defined is @samp{VERS_1.1}; it has no other dependencies.
4487 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
4488 a number of symbols to local scope so that they are not visible outside
4489 of the shared library; this is done using wildcard patterns, so that any
4490 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4491 is matched. The wildcard patterns available are the same as those used
4492 in the shell when matching filenames (also known as ``globbing'').
4493 However, if you specify the symbol name inside double quotes, then the
4494 name is treated as literal, rather than as a glob pattern.
4496 Next, the version script defines node @samp{VERS_1.2}. This node
4497 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
4498 to the version node @samp{VERS_1.2}.
4500 Finally, the version script defines node @samp{VERS_2.0}. This node
4501 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
4502 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4504 When the linker finds a symbol defined in a library which is not
4505 specifically bound to a version node, it will effectively bind it to an
4506 unspecified base version of the library. You can bind all otherwise
4507 unspecified symbols to a given version node by using @samp{global: *;}
4508 somewhere in the version script.
4510 The names of the version nodes have no specific meaning other than what
4511 they might suggest to the person reading them. The @samp{2.0} version
4512 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4513 However, this would be a confusing way to write a version script.
4515 Node name can be omitted, provided it is the only version node
4516 in the version script. Such version script doesn't assign any versions to
4517 symbols, only selects which symbols will be globally visible out and which
4521 @{ global: foo; bar; local: *; @};
4524 When you link an application against a shared library that has versioned
4525 symbols, the application itself knows which version of each symbol it
4526 requires, and it also knows which version nodes it needs from each
4527 shared library it is linked against. Thus at runtime, the dynamic
4528 loader can make a quick check to make sure that the libraries you have
4529 linked against do in fact supply all of the version nodes that the
4530 application will need to resolve all of the dynamic symbols. In this
4531 way it is possible for the dynamic linker to know with certainty that
4532 all external symbols that it needs will be resolvable without having to
4533 search for each symbol reference.
4535 The symbol versioning is in effect a much more sophisticated way of
4536 doing minor version checking that SunOS does. The fundamental problem
4537 that is being addressed here is that typically references to external
4538 functions are bound on an as-needed basis, and are not all bound when
4539 the application starts up. If a shared library is out of date, a
4540 required interface may be missing; when the application tries to use
4541 that interface, it may suddenly and unexpectedly fail. With symbol
4542 versioning, the user will get a warning when they start their program if
4543 the libraries being used with the application are too old.
4545 There are several GNU extensions to Sun's versioning approach. The
4546 first of these is the ability to bind a symbol to a version node in the
4547 source file where the symbol is defined instead of in the versioning
4548 script. This was done mainly to reduce the burden on the library
4549 maintainer. You can do this by putting something like:
4551 __asm__(".symver original_foo,foo@@VERS_1.1");
4554 in the C source file. This renames the function @samp{original_foo} to
4555 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4556 The @samp{local:} directive can be used to prevent the symbol
4557 @samp{original_foo} from being exported. A @samp{.symver} directive
4558 takes precedence over a version script.
4560 The second GNU extension is to allow multiple versions of the same
4561 function to appear in a given shared library. In this way you can make
4562 an incompatible change to an interface without increasing the major
4563 version number of the shared library, while still allowing applications
4564 linked against the old interface to continue to function.
4566 To do this, you must use multiple @samp{.symver} directives in the
4567 source file. Here is an example:
4570 __asm__(".symver original_foo,foo@@");
4571 __asm__(".symver old_foo,foo@@VERS_1.1");
4572 __asm__(".symver old_foo1,foo@@VERS_1.2");
4573 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4576 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4577 unspecified base version of the symbol. The source file that contains this
4578 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4579 @samp{old_foo1}, and @samp{new_foo}.
4581 When you have multiple definitions of a given symbol, there needs to be
4582 some way to specify a default version to which external references to
4583 this symbol will be bound. You can do this with the
4584 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4585 declare one version of a symbol as the default in this manner; otherwise
4586 you would effectively have multiple definitions of the same symbol.
4588 If you wish to bind a reference to a specific version of the symbol
4589 within the shared library, you can use the aliases of convenience
4590 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4591 specifically bind to an external version of the function in question.
4593 You can also specify the language in the version script:
4596 VERSION extern "lang" @{ version-script-commands @}
4599 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4600 The linker will iterate over the list of symbols at the link time and
4601 demangle them according to @samp{lang} before matching them to the
4602 patterns specified in @samp{version-script-commands}.
4604 Demangled names may contains spaces and other special characters. As
4605 described above, you can use a glob pattern to match demangled names,
4606 or you can use a double-quoted string to match the string exactly. In
4607 the latter case, be aware that minor differences (such as differing
4608 whitespace) between the version script and the demangler output will
4609 cause a mismatch. As the exact string generated by the demangler
4610 might change in the future, even if the mangled name does not, you
4611 should check that all of your version directives are behaving as you
4612 expect when you upgrade.
4615 @section Expressions in Linker Scripts
4618 The syntax for expressions in the linker script language is identical to
4619 that of C expressions. All expressions are evaluated as integers. All
4620 expressions are evaluated in the same size, which is 32 bits if both the
4621 host and target are 32 bits, and is otherwise 64 bits.
4623 You can use and set symbol values in expressions.
4625 The linker defines several special purpose builtin functions for use in
4629 * Constants:: Constants
4630 * Symbols:: Symbol Names
4631 * Orphan Sections:: Orphan Sections
4632 * Location Counter:: The Location Counter
4633 * Operators:: Operators
4634 * Evaluation:: Evaluation
4635 * Expression Section:: The Section of an Expression
4636 * Builtin Functions:: Builtin Functions
4640 @subsection Constants
4641 @cindex integer notation
4642 @cindex constants in linker scripts
4643 All constants are integers.
4645 As in C, the linker considers an integer beginning with @samp{0} to be
4646 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4647 hexadecimal. The linker considers other integers to be decimal.
4649 @cindex scaled integers
4650 @cindex K and M integer suffixes
4651 @cindex M and K integer suffixes
4652 @cindex suffixes for integers
4653 @cindex integer suffixes
4654 In addition, you can use the suffixes @code{K} and @code{M} to scale a
4658 @c END TEXI2ROFF-KILL
4659 @code{1024} or @code{1024*1024}
4663 ${\rm 1024}$ or ${\rm 1024}^2$
4665 @c END TEXI2ROFF-KILL
4666 respectively. For example, the following all refer to the same quantity:
4674 @subsection Symbol Names
4675 @cindex symbol names
4677 @cindex quoted symbol names
4679 Unless quoted, symbol names start with a letter, underscore, or period
4680 and may include letters, digits, underscores, periods, and hyphens.
4681 Unquoted symbol names must not conflict with any keywords. You can
4682 specify a symbol which contains odd characters or has the same name as a
4683 keyword by surrounding the symbol name in double quotes:
4686 "with a space" = "also with a space" + 10;
4689 Since symbols can contain many non-alphabetic characters, it is safest
4690 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
4691 whereas @samp{A - B} is an expression involving subtraction.
4693 @node Orphan Sections
4694 @subsection Orphan Sections
4696 Orphan sections are sections present in the input files which
4697 are not explicitly placed into the output file by the linker
4698 script. The linker will still copy these sections into the
4699 output file, but it has to guess as to where they should be
4700 placed. The linker uses a simple heuristic to do this. It
4701 attempts to place orphan sections after non-orphan sections of the
4702 same attribute, such as code vs data, loadable vs non-loadable, etc.
4703 If there is not enough room to do this then it places
4704 at the end of the file.
4706 For ELF targets, the attribute of the section includes section type as
4707 well as section flag.
4709 @node Location Counter
4710 @subsection The Location Counter
4713 @cindex location counter
4714 @cindex current output location
4715 The special linker variable @dfn{dot} @samp{.} always contains the
4716 current output location counter. Since the @code{.} always refers to a
4717 location in an output section, it may only appear in an expression
4718 within a @code{SECTIONS} command. The @code{.} symbol may appear
4719 anywhere that an ordinary symbol is allowed in an expression.
4722 Assigning a value to @code{.} will cause the location counter to be
4723 moved. This may be used to create holes in the output section. The
4724 location counter may not be moved backwards inside an output section,
4725 and may not be moved backwards outside of an output section if so
4726 doing creates areas with overlapping LMAs.
4742 In the previous example, the @samp{.text} section from @file{file1} is
4743 located at the beginning of the output section @samp{output}. It is
4744 followed by a 1000 byte gap. Then the @samp{.text} section from
4745 @file{file2} appears, also with a 1000 byte gap following before the
4746 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
4747 specifies what data to write in the gaps (@pxref{Output Section Fill}).
4749 @cindex dot inside sections
4750 Note: @code{.} actually refers to the byte offset from the start of the
4751 current containing object. Normally this is the @code{SECTIONS}
4752 statement, whose start address is 0, hence @code{.} can be used as an
4753 absolute address. If @code{.} is used inside a section description
4754 however, it refers to the byte offset from the start of that section,
4755 not an absolute address. Thus in a script like this:
4773 The @samp{.text} section will be assigned a starting address of 0x100
4774 and a size of exactly 0x200 bytes, even if there is not enough data in
4775 the @samp{.text} input sections to fill this area. (If there is too
4776 much data, an error will be produced because this would be an attempt to
4777 move @code{.} backwards). The @samp{.data} section will start at 0x500
4778 and it will have an extra 0x600 bytes worth of space after the end of
4779 the values from the @samp{.data} input sections and before the end of
4780 the @samp{.data} output section itself.
4782 @cindex dot outside sections
4783 Setting symbols to the value of the location counter outside of an
4784 output section statement can result in unexpected values if the linker
4785 needs to place orphan sections. For example, given the following:
4791 .text: @{ *(.text) @}
4795 .data: @{ *(.data) @}
4800 If the linker needs to place some input section, e.g. @code{.rodata},
4801 not mentioned in the script, it might choose to place that section
4802 between @code{.text} and @code{.data}. You might think the linker
4803 should place @code{.rodata} on the blank line in the above script, but
4804 blank lines are of no particular significance to the linker. As well,
4805 the linker doesn't associate the above symbol names with their
4806 sections. Instead, it assumes that all assignments or other
4807 statements belong to the previous output section, except for the
4808 special case of an assignment to @code{.}. I.e., the linker will
4809 place the orphan @code{.rodata} section as if the script was written
4816 .text: @{ *(.text) @}
4820 .rodata: @{ *(.rodata) @}
4821 .data: @{ *(.data) @}
4826 This may or may not be the script author's intention for the value of
4827 @code{start_of_data}. One way to influence the orphan section
4828 placement is to assign the location counter to itself, as the linker
4829 assumes that an assignment to @code{.} is setting the start address of
4830 a following output section and thus should be grouped with that
4831 section. So you could write:
4837 .text: @{ *(.text) @}
4842 .data: @{ *(.data) @}
4847 Now, the orphan @code{.rodata} section will be placed between
4848 @code{end_of_text} and @code{start_of_data}.
4852 @subsection Operators
4853 @cindex operators for arithmetic
4854 @cindex arithmetic operators
4855 @cindex precedence in expressions
4856 The linker recognizes the standard C set of arithmetic operators, with
4857 the standard bindings and precedence levels:
4860 @c END TEXI2ROFF-KILL
4862 precedence associativity Operators Notes
4868 5 left == != > < <= >=
4874 11 right &= += -= *= /= (2)
4878 (1) Prefix operators
4879 (2) @xref{Assignments}.
4883 \vskip \baselineskip
4884 %"lispnarrowing" is the extra indent used generally for smallexample
4885 \hskip\lispnarrowing\vbox{\offinterlineskip
4888 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
4889 height2pt&\omit&&\omit&&\omit&\cr
4890 &Precedence&& Associativity &&{\rm Operators}&\cr
4891 height2pt&\omit&&\omit&&\omit&\cr
4893 height2pt&\omit&&\omit&&\omit&\cr
4895 % '176 is tilde, '~' in tt font
4896 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
4897 &2&&left&&* / \%&\cr
4900 &5&&left&&== != > < <= >=&\cr
4903 &8&&left&&{\&\&}&\cr
4906 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
4908 height2pt&\omit&&\omit&&\omit&\cr}
4913 @obeylines@parskip=0pt@parindent=0pt
4914 @dag@quad Prefix operators.
4915 @ddag@quad @xref{Assignments}.
4918 @c END TEXI2ROFF-KILL
4921 @subsection Evaluation
4922 @cindex lazy evaluation
4923 @cindex expression evaluation order
4924 The linker evaluates expressions lazily. It only computes the value of
4925 an expression when absolutely necessary.
4927 The linker needs some information, such as the value of the start
4928 address of the first section, and the origins and lengths of memory
4929 regions, in order to do any linking at all. These values are computed
4930 as soon as possible when the linker reads in the linker script.
4932 However, other values (such as symbol values) are not known or needed
4933 until after storage allocation. Such values are evaluated later, when
4934 other information (such as the sizes of output sections) is available
4935 for use in the symbol assignment expression.
4937 The sizes of sections cannot be known until after allocation, so
4938 assignments dependent upon these are not performed until after
4941 Some expressions, such as those depending upon the location counter
4942 @samp{.}, must be evaluated during section allocation.
4944 If the result of an expression is required, but the value is not
4945 available, then an error results. For example, a script like the
4951 .text 9+this_isnt_constant :
4957 will cause the error message @samp{non constant expression for initial
4960 @node Expression Section
4961 @subsection The Section of an Expression
4962 @cindex expression sections
4963 @cindex absolute expressions
4964 @cindex relative expressions
4965 @cindex absolute and relocatable symbols
4966 @cindex relocatable and absolute symbols
4967 @cindex symbols, relocatable and absolute
4968 When the linker evaluates an expression, the result is either absolute
4969 or relative to some section. A relative expression is expressed as a
4970 fixed offset from the base of a section.
4972 The position of the expression within the linker script determines
4973 whether it is absolute or relative. An expression which appears within
4974 an output section definition is relative to the base of the output
4975 section. An expression which appears elsewhere will be absolute.
4977 A symbol set to a relative expression will be relocatable if you request
4978 relocatable output using the @samp{-r} option. That means that a
4979 further link operation may change the value of the symbol. The symbol's
4980 section will be the section of the relative expression.
4982 A symbol set to an absolute expression will retain the same value
4983 through any further link operation. The symbol will be absolute, and
4984 will not have any particular associated section.
4986 You can use the builtin function @code{ABSOLUTE} to force an expression
4987 to be absolute when it would otherwise be relative. For example, to
4988 create an absolute symbol set to the address of the end of the output
4989 section @samp{.data}:
4993 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
4997 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
4998 @samp{.data} section.
5000 @node Builtin Functions
5001 @subsection Builtin Functions
5002 @cindex functions in expressions
5003 The linker script language includes a number of builtin functions for
5004 use in linker script expressions.
5007 @item ABSOLUTE(@var{exp})
5008 @kindex ABSOLUTE(@var{exp})
5009 @cindex expression, absolute
5010 Return the absolute (non-relocatable, as opposed to non-negative) value
5011 of the expression @var{exp}. Primarily useful to assign an absolute
5012 value to a symbol within a section definition, where symbol values are
5013 normally section relative. @xref{Expression Section}.
5015 @item ADDR(@var{section})
5016 @kindex ADDR(@var{section})
5017 @cindex section address in expression
5018 Return the absolute address (the VMA) of the named @var{section}. Your
5019 script must previously have defined the location of that section. In
5020 the following example, @code{symbol_1} and @code{symbol_2} are assigned
5027 start_of_output_1 = ABSOLUTE(.);
5032 symbol_1 = ADDR(.output1);
5033 symbol_2 = start_of_output_1;
5039 @item ALIGN(@var{align})
5040 @itemx ALIGN(@var{exp},@var{align})
5041 @kindex ALIGN(@var{align})
5042 @kindex ALIGN(@var{exp},@var{align})
5043 @cindex round up location counter
5044 @cindex align location counter
5045 @cindex round up expression
5046 @cindex align expression
5047 Return the location counter (@code{.}) or arbitrary expression aligned
5048 to the next @var{align} boundary. The single operand @code{ALIGN}
5049 doesn't change the value of the location counter---it just does
5050 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
5051 expression to be aligned upwards (@code{ALIGN(@var{align})} is
5052 equivalent to @code{ALIGN(., @var{align})}).
5054 Here is an example which aligns the output @code{.data} section to the
5055 next @code{0x2000} byte boundary after the preceding section and sets a
5056 variable within the section to the next @code{0x8000} boundary after the
5061 .data ALIGN(0x2000): @{
5063 variable = ALIGN(0x8000);
5069 The first use of @code{ALIGN} in this example specifies the location of
5070 a section because it is used as the optional @var{address} attribute of
5071 a section definition (@pxref{Output Section Address}). The second use
5072 of @code{ALIGN} is used to defines the value of a symbol.
5074 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5076 @item ALIGNOF(@var{section})
5077 @kindex ALIGNOF(@var{section})
5078 @cindex section alignment
5079 Return the alignment in bytes of the named @var{section}, if that section has
5080 been allocated. If the section has not been allocated when this is
5081 evaluated, the linker will report an error. In the following example,
5082 the alignment of the @code{.output} section is stored as the first
5083 value in that section.
5088 LONG (ALIGNOF (.output))
5095 @item BLOCK(@var{exp})
5096 @kindex BLOCK(@var{exp})
5097 This is a synonym for @code{ALIGN}, for compatibility with older linker
5098 scripts. It is most often seen when setting the address of an output
5101 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5102 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5103 This is equivalent to either
5105 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5109 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5112 depending on whether the latter uses fewer @var{commonpagesize} sized pages
5113 for the data segment (area between the result of this expression and
5114 @code{DATA_SEGMENT_END}) than the former or not.
5115 If the latter form is used, it means @var{commonpagesize} bytes of runtime
5116 memory will be saved at the expense of up to @var{commonpagesize} wasted
5117 bytes in the on-disk file.
5119 This expression can only be used directly in @code{SECTIONS} commands, not in
5120 any output section descriptions and only once in the linker script.
5121 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5122 be the system page size the object wants to be optimized for (while still
5123 working on system page sizes up to @var{maxpagesize}).
5128 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5131 @item DATA_SEGMENT_END(@var{exp})
5132 @kindex DATA_SEGMENT_END(@var{exp})
5133 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5134 evaluation purposes.
5137 . = DATA_SEGMENT_END(.);
5140 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5141 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5142 This defines the end of the @code{PT_GNU_RELRO} segment when
5143 @samp{-z relro} option is used. Second argument is returned.
5144 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5145 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5146 @var{exp} + @var{offset} is aligned to the most commonly used page
5147 boundary for particular target. If present in the linker script,
5148 it must always come in between @code{DATA_SEGMENT_ALIGN} and
5149 @code{DATA_SEGMENT_END}.
5152 . = DATA_SEGMENT_RELRO_END(24, .);
5155 @item DEFINED(@var{symbol})
5156 @kindex DEFINED(@var{symbol})
5157 @cindex symbol defaults
5158 Return 1 if @var{symbol} is in the linker global symbol table and is
5159 defined before the statement using DEFINED in the script, otherwise
5160 return 0. You can use this function to provide
5161 default values for symbols. For example, the following script fragment
5162 shows how to set a global symbol @samp{begin} to the first location in
5163 the @samp{.text} section---but if a symbol called @samp{begin} already
5164 existed, its value is preserved:
5170 begin = DEFINED(begin) ? begin : . ;
5178 @item LENGTH(@var{memory})
5179 @kindex LENGTH(@var{memory})
5180 Return the length of the memory region named @var{memory}.
5182 @item LOADADDR(@var{section})
5183 @kindex LOADADDR(@var{section})
5184 @cindex section load address in expression
5185 Return the absolute LMA of the named @var{section}. This is normally
5186 the same as @code{ADDR}, but it may be different if the @code{AT}
5187 attribute is used in the output section definition (@pxref{Output
5191 @item MAX(@var{exp1}, @var{exp2})
5192 Returns the maximum of @var{exp1} and @var{exp2}.
5195 @item MIN(@var{exp1}, @var{exp2})
5196 Returns the minimum of @var{exp1} and @var{exp2}.
5198 @item NEXT(@var{exp})
5199 @kindex NEXT(@var{exp})
5200 @cindex unallocated address, next
5201 Return the next unallocated address that is a multiple of @var{exp}.
5202 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5203 use the @code{MEMORY} command to define discontinuous memory for the
5204 output file, the two functions are equivalent.
5206 @item ORIGIN(@var{memory})
5207 @kindex ORIGIN(@var{memory})
5208 Return the origin of the memory region named @var{memory}.
5210 @item SEGMENT_START(@var{segment}, @var{default})
5211 @kindex SEGMENT_START(@var{segment}, @var{default})
5212 Return the base address of the named @var{segment}. If an explicit
5213 value has been given for this segment (with a command-line @samp{-T}
5214 option) that value will be returned; otherwise the value will be
5215 @var{default}. At present, the @samp{-T} command-line option can only
5216 be used to set the base address for the ``text'', ``data'', and
5217 ``bss'' sections, but you use @code{SEGMENT_START} with any segment
5220 @item SIZEOF(@var{section})
5221 @kindex SIZEOF(@var{section})
5222 @cindex section size
5223 Return the size in bytes of the named @var{section}, if that section has
5224 been allocated. If the section has not been allocated when this is
5225 evaluated, the linker will report an error. In the following example,
5226 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5235 symbol_1 = .end - .start ;
5236 symbol_2 = SIZEOF(.output);
5241 @item SIZEOF_HEADERS
5242 @itemx sizeof_headers
5243 @kindex SIZEOF_HEADERS
5245 Return the size in bytes of the output file's headers. This is
5246 information which appears at the start of the output file. You can use
5247 this number when setting the start address of the first section, if you
5248 choose, to facilitate paging.
5250 @cindex not enough room for program headers
5251 @cindex program headers, not enough room
5252 When producing an ELF output file, if the linker script uses the
5253 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5254 number of program headers before it has determined all the section
5255 addresses and sizes. If the linker later discovers that it needs
5256 additional program headers, it will report an error @samp{not enough
5257 room for program headers}. To avoid this error, you must avoid using
5258 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5259 script to avoid forcing the linker to use additional program headers, or
5260 you must define the program headers yourself using the @code{PHDRS}
5261 command (@pxref{PHDRS}).
5264 @node Implicit Linker Scripts
5265 @section Implicit Linker Scripts
5266 @cindex implicit linker scripts
5267 If you specify a linker input file which the linker can not recognize as
5268 an object file or an archive file, it will try to read the file as a
5269 linker script. If the file can not be parsed as a linker script, the
5270 linker will report an error.
5272 An implicit linker script will not replace the default linker script.
5274 Typically an implicit linker script would contain only symbol
5275 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5278 Any input files read because of an implicit linker script will be read
5279 at the position in the command line where the implicit linker script was
5280 read. This can affect archive searching.
5283 @node Machine Dependent
5284 @chapter Machine Dependent Features
5286 @cindex machine dependencies
5287 @command{ld} has additional features on some platforms; the following
5288 sections describe them. Machines where @command{ld} has no additional
5289 functionality are not listed.
5293 * H8/300:: @command{ld} and the H8/300
5296 * i960:: @command{ld} and the Intel 960 family
5299 * ARM:: @command{ld} and the ARM family
5302 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
5305 * MMIX:: @command{ld} and MMIX
5308 * MSP430:: @command{ld} and MSP430
5311 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
5314 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
5317 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
5320 * SPU ELF:: @command{ld} and SPU ELF Support
5323 * TI COFF:: @command{ld} and TI COFF
5326 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
5329 * Xtensa:: @command{ld} and Xtensa Processors
5340 @section @command{ld} and the H8/300
5342 @cindex H8/300 support
5343 For the H8/300, @command{ld} can perform these global optimizations when
5344 you specify the @samp{--relax} command-line option.
5347 @cindex relaxing on H8/300
5348 @item relaxing address modes
5349 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5350 targets are within eight bits, and turns them into eight-bit
5351 program-counter relative @code{bsr} and @code{bra} instructions,
5354 @cindex synthesizing on H8/300
5355 @item synthesizing instructions
5356 @c FIXME: specifically mov.b, or any mov instructions really?
5357 @command{ld} finds all @code{mov.b} instructions which use the
5358 sixteen-bit absolute address form, but refer to the top
5359 page of memory, and changes them to use the eight-bit address form.
5360 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5361 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5362 top page of memory).
5364 @item bit manipulation instructions
5365 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
5366 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5367 which use 32 bit and 16 bit absolute address form, but refer to the top
5368 page of memory, and changes them to use the 8 bit address form.
5369 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5370 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5371 the top page of memory).
5373 @item system control instructions
5374 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
5375 32 bit absolute address form, but refer to the top page of memory, and
5376 changes them to use 16 bit address form.
5377 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5378 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5379 the top page of memory).
5389 @c This stuff is pointless to say unless you're especially concerned
5390 @c with Renesas chips; don't enable it for generic case, please.
5392 @chapter @command{ld} and Other Renesas Chips
5394 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5395 H8/500, and SH chips. No special features, commands, or command-line
5396 options are required for these chips.
5406 @section @command{ld} and the Intel 960 Family
5408 @cindex i960 support
5410 You can use the @samp{-A@var{architecture}} command line option to
5411 specify one of the two-letter names identifying members of the 960
5412 family; the option specifies the desired output target, and warns of any
5413 incompatible instructions in the input files. It also modifies the
5414 linker's search strategy for archive libraries, to support the use of
5415 libraries specific to each particular architecture, by including in the
5416 search loop names suffixed with the string identifying the architecture.
5418 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5419 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5420 paths, and in any paths you specify with @samp{-L}) for a library with
5433 The first two possibilities would be considered in any event; the last
5434 two are due to the use of @w{@samp{-ACA}}.
5436 You can meaningfully use @samp{-A} more than once on a command line, since
5437 the 960 architecture family allows combination of target architectures; each
5438 use will add another pair of name variants to search for when @w{@samp{-l}}
5439 specifies a library.
5441 @cindex @option{--relax} on i960
5442 @cindex relaxing on i960
5443 @command{ld} supports the @samp{--relax} option for the i960 family. If
5444 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5445 @code{calx} instructions whose targets are within 24 bits, and turns
5446 them into 24-bit program-counter relative @code{bal} and @code{cal}
5447 instructions, respectively. @command{ld} also turns @code{cal}
5448 instructions into @code{bal} instructions when it determines that the
5449 target subroutine is a leaf routine (that is, the target subroutine does
5450 not itself call any subroutines).
5467 @node M68HC11/68HC12
5468 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
5470 @cindex M68HC11 and 68HC12 support
5472 @subsection Linker Relaxation
5474 For the Motorola 68HC11, @command{ld} can perform these global
5475 optimizations when you specify the @samp{--relax} command-line option.
5478 @cindex relaxing on M68HC11
5479 @item relaxing address modes
5480 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5481 targets are within eight bits, and turns them into eight-bit
5482 program-counter relative @code{bsr} and @code{bra} instructions,
5485 @command{ld} also looks at all 16-bit extended addressing modes and
5486 transforms them in a direct addressing mode when the address is in
5487 page 0 (between 0 and 0x0ff).
5489 @item relaxing gcc instruction group
5490 When @command{gcc} is called with @option{-mrelax}, it can emit group
5491 of instructions that the linker can optimize to use a 68HC11 direct
5492 addressing mode. These instructions consists of @code{bclr} or
5493 @code{bset} instructions.
5497 @subsection Trampoline Generation
5499 @cindex trampoline generation on M68HC11
5500 @cindex trampoline generation on M68HC12
5501 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5502 call a far function using a normal @code{jsr} instruction. The linker
5503 will also change the relocation to some far function to use the
5504 trampoline address instead of the function address. This is typically the
5505 case when a pointer to a function is taken. The pointer will in fact
5506 point to the function trampoline.
5514 @section @command{ld} and the ARM family
5516 @cindex ARM interworking support
5517 @kindex --support-old-code
5518 For the ARM, @command{ld} will generate code stubs to allow functions calls
5519 between ARM and Thumb code. These stubs only work with code that has
5520 been compiled and assembled with the @samp{-mthumb-interwork} command
5521 line option. If it is necessary to link with old ARM object files or
5522 libraries, which have not been compiled with the -mthumb-interwork
5523 option then the @samp{--support-old-code} command line switch should be
5524 given to the linker. This will make it generate larger stub functions
5525 which will work with non-interworking aware ARM code. Note, however,
5526 the linker does not support generating stubs for function calls to
5527 non-interworking aware Thumb code.
5529 @cindex thumb entry point
5530 @cindex entry point, thumb
5531 @kindex --thumb-entry=@var{entry}
5532 The @samp{--thumb-entry} switch is a duplicate of the generic
5533 @samp{--entry} switch, in that it sets the program's starting address.
5534 But it also sets the bottom bit of the address, so that it can be
5535 branched to using a BX instruction, and the program will start
5536 executing in Thumb mode straight away.
5540 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5541 executables. This option is only valid when linking big-endian objects.
5542 The resulting image will contain big-endian data and little-endian code.
5545 @kindex --target1-rel
5546 @kindex --target1-abs
5547 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5548 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
5549 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
5550 and @samp{--target1-abs} switches override the default.
5553 @kindex --target2=@var{type}
5554 The @samp{--target2=type} switch overrides the default definition of the
5555 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
5556 meanings, and target defaults are as follows:
5559 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5561 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
5563 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5568 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5569 specification) enables objects compiled for the ARMv4 architecture to be
5570 interworking-safe when linked with other objects compiled for ARMv4t, but
5571 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
5573 In the latter case, the switch @option{--fix-v4bx} must be passed to the
5574 linker, which causes v4t @code{BX rM} instructions to be rewritten as
5575 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
5577 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
5578 relocations are ignored.
5580 @cindex FIX_V4BX_INTERWORKING
5581 @kindex --fix-v4bx-interworking
5582 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
5583 relocations with a branch to the following veneer:
5591 This allows generation of libraries/applications that work on ARMv4 cores
5592 and are still interworking safe. Note that the above veneer clobbers the
5593 condition flags, so may cause incorrect progrm behavior in rare cases.
5597 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
5598 BLX instructions (available on ARMv5t and above) in various
5599 situations. Currently it is used to perform calls via the PLT from Thumb
5600 code using BLX rather than using BX and a mode-switching stub before
5601 each PLT entry. This should lead to such calls executing slightly faster.
5603 This option is enabled implicitly for SymbianOS, so there is no need to
5604 specify it if you are using that target.
5606 @cindex VFP11_DENORM_FIX
5607 @kindex --vfp11-denorm-fix
5608 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
5609 bug in certain VFP11 coprocessor hardware, which sometimes allows
5610 instructions with denorm operands (which must be handled by support code)
5611 to have those operands overwritten by subsequent instructions before
5612 the support code can read the intended values.
5614 The bug may be avoided in scalar mode if you allow at least one
5615 intervening instruction between a VFP11 instruction which uses a register
5616 and another instruction which writes to the same register, or at least two
5617 intervening instructions if vector mode is in use. The bug only affects
5618 full-compliance floating-point mode: you do not need this workaround if
5619 you are using "runfast" mode. Please contact ARM for further details.
5621 If you know you are using buggy VFP11 hardware, you can
5622 enable this workaround by specifying the linker option
5623 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
5624 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
5625 vector mode (the latter also works for scalar code). The default is
5626 @samp{--vfp-denorm-fix=none}.
5628 If the workaround is enabled, instructions are scanned for
5629 potentially-troublesome sequences, and a veneer is created for each
5630 such sequence which may trigger the erratum. The veneer consists of the
5631 first instruction of the sequence and a branch back to the subsequent
5632 instruction. The original instruction is then replaced with a branch to
5633 the veneer. The extra cycles required to call and return from the veneer
5634 are sufficient to avoid the erratum in both the scalar and vector cases.
5636 @cindex NO_ENUM_SIZE_WARNING
5637 @kindex --no-enum-size-warning
5638 The @option{--no-enum-size-warning} switch prevents the linker from
5639 warning when linking object files that specify incompatible EABI
5640 enumeration size attributes. For example, with this switch enabled,
5641 linking of an object file using 32-bit enumeration values with another
5642 using enumeration values fitted into the smallest possible space will
5646 @kindex --pic-veneer
5647 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
5648 ARM/Thumb interworking veneers, even if the rest of the binary
5649 is not PIC. This avoids problems on uClinux targets where
5650 @samp{--emit-relocs} is used to generate relocatable binaries.
5652 @cindex STUB_GROUP_SIZE
5653 @kindex --stub-group-size=@var{N}
5654 The linker will automatically generate and insert small sequences of
5655 code into a linked ARM ELF executable whenever an attempt is made to
5656 perform a function call to a symbol that is too far away. The
5657 placement of these sequences of instructions - called stubs - is
5658 controlled by the command line option @option{--stub-group-size=N}.
5659 The placement is important because a poor choice can create a need for
5660 duplicate stubs, increasing the code sizw. The linker will try to
5661 group stubs together in order to reduce interruptions to the flow of
5662 code, but it needs guidance as to how big these groups should be and
5663 where they should be placed.
5665 The value of @samp{N}, the parameter to the
5666 @option{--stub-group-size=} option controls where the stub groups are
5667 placed. If it is negative then all stubs are placed before the first
5668 branch that needs them. If it is positive then the stubs can be
5669 placed either before or after the branches that need them. If the
5670 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
5671 exactly where to place groups of stubs, using its built in heuristics.
5672 A value of @samp{N} greater than 1 (or smaller than -1) tells the
5673 linker that a single group of stubs can service at most @samp{N} bytes
5674 from the input sections.
5676 The default, if @option{--stub-group-size=} is not specified, is
5690 @section @command{ld} and HPPA 32-bit ELF Support
5691 @cindex HPPA multiple sub-space stubs
5692 @kindex --multi-subspace
5693 When generating a shared library, @command{ld} will by default generate
5694 import stubs suitable for use with a single sub-space application.
5695 The @samp{--multi-subspace} switch causes @command{ld} to generate export
5696 stubs, and different (larger) import stubs suitable for use with
5697 multiple sub-spaces.
5699 @cindex HPPA stub grouping
5700 @kindex --stub-group-size=@var{N}
5701 Long branch stubs and import/export stubs are placed by @command{ld} in
5702 stub sections located between groups of input sections.
5703 @samp{--stub-group-size} specifies the maximum size of a group of input
5704 sections handled by one stub section. Since branch offsets are signed,
5705 a stub section may serve two groups of input sections, one group before
5706 the stub section, and one group after it. However, when using
5707 conditional branches that require stubs, it may be better (for branch
5708 prediction) that stub sections only serve one group of input sections.
5709 A negative value for @samp{N} chooses this scheme, ensuring that
5710 branches to stubs always use a negative offset. Two special values of
5711 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5712 @command{ld} to automatically size input section groups for the branch types
5713 detected, with the same behaviour regarding stub placement as other
5714 positive or negative values of @samp{N} respectively.
5716 Note that @samp{--stub-group-size} does not split input sections. A
5717 single input section larger than the group size specified will of course
5718 create a larger group (of one section). If input sections are too
5719 large, it may not be possible for a branch to reach its stub.
5732 @section @code{ld} and MMIX
5733 For MMIX, there is a choice of generating @code{ELF} object files or
5734 @code{mmo} object files when linking. The simulator @code{mmix}
5735 understands the @code{mmo} format. The binutils @code{objcopy} utility
5736 can translate between the two formats.
5738 There is one special section, the @samp{.MMIX.reg_contents} section.
5739 Contents in this section is assumed to correspond to that of global
5740 registers, and symbols referring to it are translated to special symbols,
5741 equal to registers. In a final link, the start address of the
5742 @samp{.MMIX.reg_contents} section corresponds to the first allocated
5743 global register multiplied by 8. Register @code{$255} is not included in
5744 this section; it is always set to the program entry, which is at the
5745 symbol @code{Main} for @code{mmo} files.
5747 Symbols with the prefix @code{__.MMIX.start.}, for example
5748 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special;
5749 there must be only one each, even if they are local. The default linker
5750 script uses these to set the default start address of a section.
5752 Initial and trailing multiples of zero-valued 32-bit words in a section,
5753 are left out from an mmo file.
5766 @section @code{ld} and MSP430
5767 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
5768 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
5769 just pass @samp{-m help} option to the linker).
5771 @cindex MSP430 extra sections
5772 The linker will recognize some extra sections which are MSP430 specific:
5775 @item @samp{.vectors}
5776 Defines a portion of ROM where interrupt vectors located.
5778 @item @samp{.bootloader}
5779 Defines the bootloader portion of the ROM (if applicable). Any code
5780 in this section will be uploaded to the MPU.
5782 @item @samp{.infomem}
5783 Defines an information memory section (if applicable). Any code in
5784 this section will be uploaded to the MPU.
5786 @item @samp{.infomemnobits}
5787 This is the same as the @samp{.infomem} section except that any code
5788 in this section will not be uploaded to the MPU.
5790 @item @samp{.noinit}
5791 Denotes a portion of RAM located above @samp{.bss} section.
5793 The last two sections are used by gcc.
5807 @section @command{ld} and PowerPC 32-bit ELF Support
5808 @cindex PowerPC long branches
5809 @kindex --relax on PowerPC
5810 Branches on PowerPC processors are limited to a signed 26-bit
5811 displacement, which may result in @command{ld} giving
5812 @samp{relocation truncated to fit} errors with very large programs.
5813 @samp{--relax} enables the generation of trampolines that can access
5814 the entire 32-bit address space. These trampolines are inserted at
5815 section boundaries, so may not themselves be reachable if an input
5816 section exceeds 33M in size.
5818 @cindex PowerPC ELF32 options
5823 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
5824 generates code capable of using a newer PLT and GOT layout that has
5825 the security advantage of no executable section ever needing to be
5826 writable and no writable section ever being executable. PowerPC
5827 @command{ld} will generate this layout, including stubs to access the
5828 PLT, if all input files (including startup and static libraries) were
5829 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
5830 BSS PLT (and GOT layout) which can give slightly better performance.
5832 @kindex --secure-plt
5834 @command{ld} will use the new PLT and GOT layout if it is linking new
5835 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
5836 when linking non-PIC code. This option requests the new PLT and GOT
5837 layout. A warning will be given if some object file requires the old
5843 The new secure PLT and GOT are placed differently relative to other
5844 sections compared to older BSS PLT and GOT placement. The location of
5845 @code{.plt} must change because the new secure PLT is an initialized
5846 section while the old PLT is uninitialized. The reason for the
5847 @code{.got} change is more subtle: The new placement allows
5848 @code{.got} to be read-only in applications linked with
5849 @samp{-z relro -z now}. However, this placement means that
5850 @code{.sdata} cannot always be used in shared libraries, because the
5851 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
5852 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
5853 GCC doesn't use @code{.sdata} in shared libraries, so this option is
5854 really only useful for other compilers that may do so.
5856 @cindex PowerPC stub symbols
5857 @kindex --emit-stub-syms
5858 @item --emit-stub-syms
5859 This option causes @command{ld} to label linker stubs with a local
5860 symbol that encodes the stub type and destination.
5862 @cindex PowerPC TLS optimization
5863 @kindex --no-tls-optimize
5864 @item --no-tls-optimize
5865 PowerPC @command{ld} normally performs some optimization of code
5866 sequences used to access Thread-Local Storage. Use this option to
5867 disable the optimization.
5880 @node PowerPC64 ELF64
5881 @section @command{ld} and PowerPC64 64-bit ELF Support
5883 @cindex PowerPC64 ELF64 options
5885 @cindex PowerPC64 stub grouping
5886 @kindex --stub-group-size
5887 @item --stub-group-size
5888 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
5889 by @command{ld} in stub sections located between groups of input sections.
5890 @samp{--stub-group-size} specifies the maximum size of a group of input
5891 sections handled by one stub section. Since branch offsets are signed,
5892 a stub section may serve two groups of input sections, one group before
5893 the stub section, and one group after it. However, when using
5894 conditional branches that require stubs, it may be better (for branch
5895 prediction) that stub sections only serve one group of input sections.
5896 A negative value for @samp{N} chooses this scheme, ensuring that
5897 branches to stubs always use a negative offset. Two special values of
5898 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5899 @command{ld} to automatically size input section groups for the branch types
5900 detected, with the same behaviour regarding stub placement as other
5901 positive or negative values of @samp{N} respectively.
5903 Note that @samp{--stub-group-size} does not split input sections. A
5904 single input section larger than the group size specified will of course
5905 create a larger group (of one section). If input sections are too
5906 large, it may not be possible for a branch to reach its stub.
5908 @cindex PowerPC64 stub symbols
5909 @kindex --emit-stub-syms
5910 @item --emit-stub-syms
5911 This option causes @command{ld} to label linker stubs with a local
5912 symbol that encodes the stub type and destination.
5914 @cindex PowerPC64 dot symbols
5916 @kindex --no-dotsyms
5917 @item --dotsyms, --no-dotsyms
5918 These two options control how @command{ld} interprets version patterns
5919 in a version script. Older PowerPC64 compilers emitted both a
5920 function descriptor symbol with the same name as the function, and a
5921 code entry symbol with the name prefixed by a dot (@samp{.}). To
5922 properly version a function @samp{foo}, the version script thus needs
5923 to control both @samp{foo} and @samp{.foo}. The option
5924 @samp{--dotsyms}, on by default, automatically adds the required
5925 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
5928 @cindex PowerPC64 TLS optimization
5929 @kindex --no-tls-optimize
5930 @item --no-tls-optimize
5931 PowerPC64 @command{ld} normally performs some optimization of code
5932 sequences used to access Thread-Local Storage. Use this option to
5933 disable the optimization.
5935 @cindex PowerPC64 OPD optimization
5936 @kindex --no-opd-optimize
5937 @item --no-opd-optimize
5938 PowerPC64 @command{ld} normally removes @code{.opd} section entries
5939 corresponding to deleted link-once functions, or functions removed by
5940 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
5941 Use this option to disable @code{.opd} optimization.
5943 @cindex PowerPC64 OPD spacing
5944 @kindex --non-overlapping-opd
5945 @item --non-overlapping-opd
5946 Some PowerPC64 compilers have an option to generate compressed
5947 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
5948 the static chain pointer (unused in C) with the first word of the next
5949 entry. This option expands such entries to the full 24 bytes.
5951 @cindex PowerPC64 TOC optimization
5952 @kindex --no-toc-optimize
5953 @item --no-toc-optimize
5954 PowerPC64 @command{ld} normally removes unused @code{.toc} section
5955 entries. Such entries are detected by examining relocations that
5956 reference the TOC in code sections. A reloc in a deleted code section
5957 marks a TOC word as unneeded, while a reloc in a kept code section
5958 marks a TOC word as needed. Since the TOC may reference itself, TOC
5959 relocs are also examined. TOC words marked as both needed and
5960 unneeded will of course be kept. TOC words without any referencing
5961 reloc are assumed to be part of a multi-word entry, and are kept or
5962 discarded as per the nearest marked preceding word. This works
5963 reliably for compiler generated code, but may be incorrect if assembly
5964 code is used to insert TOC entries. Use this option to disable the
5967 @cindex PowerPC64 multi-TOC
5968 @kindex --no-multi-toc
5969 @item --no-multi-toc
5970 By default, PowerPC64 GCC generates code for a TOC model where TOC
5971 entries are accessed with a 16-bit offset from r2. This limits the
5972 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
5973 grouping code sections such that each group uses less than 64K for its
5974 TOC entries, then inserts r2 adjusting stubs between inter-group
5975 calls. @command{ld} does not split apart input sections, so cannot
5976 help if a single input file has a @code{.toc} section that exceeds
5977 64K, most likely from linking multiple files with @command{ld -r}.
5978 Use this option to turn off this feature.
5992 @section @command{ld} and SPU ELF Support
5994 @cindex SPU ELF options
6000 This option marks an executable as a PIC plugin module.
6002 @cindex SPU overlays
6003 @kindex --no-overlays
6005 Normally, @command{ld} recognizes calls to functions within overlay
6006 regions, and redirects such calls to an overlay manager via a stub.
6007 @command{ld} also provides a built-in overlay manager. This option
6008 turns off all this special overlay handling.
6010 @cindex SPU overlay stub symbols
6011 @kindex --emit-stub-syms
6012 @item --emit-stub-syms
6013 This option causes @command{ld} to label overlay stubs with a local
6014 symbol that encodes the stub type and destination.
6016 @cindex SPU extra overlay stubs
6017 @kindex --extra-overlay-stubs
6018 @item --extra-overlay-stubs
6019 This option causes @command{ld} to add overlay call stubs on all
6020 function calls out of overlay regions. Normally stubs are not added
6021 on calls to non-overlay regions.
6023 @cindex SPU local store size
6024 @kindex --local-store=lo:hi
6025 @item --local-store=lo:hi
6026 @command{ld} usually checks that a final executable for SPU fits in
6027 the address range 0 to 256k. This option may be used to change the
6028 range. Disable the check entirely with @option{--local-store=0:0}.
6031 @kindex --stack-analysis
6032 @item --stack-analysis
6033 SPU local store space is limited. Over-allocation of stack space
6034 unnecessarily limits space available for code and data, while
6035 under-allocation results in runtime failures. If given this option,
6036 @command{ld} will provide an estimate of maximum stack usage.
6037 @command{ld} does this by examining symbols in code sections to
6038 determine the extents of functions, and looking at function prologues
6039 for stack adjusting instructions. A call-graph is created by looking
6040 for relocations on branch instructions. The graph is then searched
6041 for the maximum stack usage path. Note that this analysis does not
6042 find calls made via function pointers, and does not handle recursion
6043 and other cycles in the call graph. Stack usage may be
6044 under-estimated if your code makes such calls. Also, stack usage for
6045 dynamic allocation, e.g. alloca, will not be detected. If a link map
6046 is requested, detailed information about each function's stack usage
6047 and calls will be given.
6050 @kindex --emit-stack-syms
6051 @item --emit-stack-syms
6052 This option, if given along with @option{--stack-analysis} will result
6053 in @command{ld} emitting stack sizing symbols for each function.
6054 These take the form @code{__stack_<function_name>} for global
6055 functions, and @code{__stack_<number>_<function_name>} for static
6056 functions. @code{<number>} is the section id in hex. The value of
6057 such symbols is the stack requirement for the corresponding function.
6058 The symbol size will be zero, type @code{STT_NOTYPE}, binding
6059 @code{STB_LOCAL}, and section @code{SHN_ABS}.
6073 @section @command{ld}'s Support for Various TI COFF Versions
6074 @cindex TI COFF versions
6075 @kindex --format=@var{version}
6076 The @samp{--format} switch allows selection of one of the various
6077 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
6078 also supported. The TI COFF versions also vary in header byte-order
6079 format; @command{ld} will read any version or byte order, but the output
6080 header format depends on the default specified by the specific target.
6093 @section @command{ld} and WIN32 (cygwin/mingw)
6095 This section describes some of the win32 specific @command{ld} issues.
6096 See @ref{Options,,Command Line Options} for detailed description of the
6097 command line options mentioned here.
6100 @cindex import libraries
6101 @item import libraries
6102 The standard Windows linker creates and uses so-called import
6103 libraries, which contains information for linking to dll's. They are
6104 regular static archives and are handled as any other static
6105 archive. The cygwin and mingw ports of @command{ld} have specific
6106 support for creating such libraries provided with the
6107 @samp{--out-implib} command line option.
6109 @item exporting DLL symbols
6110 @cindex exporting DLL symbols
6111 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6114 @item using auto-export functionality
6115 @cindex using auto-export functionality
6116 By default @command{ld} exports symbols with the auto-export functionality,
6117 which is controlled by the following command line options:
6120 @item --export-all-symbols [This is the default]
6121 @item --exclude-symbols
6122 @item --exclude-libs
6125 If, however, @samp{--export-all-symbols} is not given explicitly on the
6126 command line, then the default auto-export behavior will be @emph{disabled}
6127 if either of the following are true:
6130 @item A DEF file is used.
6131 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
6134 @item using a DEF file
6135 @cindex using a DEF file
6136 Another way of exporting symbols is using a DEF file. A DEF file is
6137 an ASCII file containing definitions of symbols which should be
6138 exported when a dll is created. Usually it is named @samp{<dll
6139 name>.def} and is added as any other object file to the linker's
6140 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
6143 gcc -o <output> <objectfiles> <dll name>.def
6146 Using a DEF file turns off the normal auto-export behavior, unless the
6147 @samp{--export-all-symbols} option is also used.
6149 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
6152 LIBRARY "xyz.dll" BASE=0x20000000
6158 another_foo = abc.dll.afoo
6162 This example defines a DLL with a non-default base address and five
6163 symbols in the export table. The third exported symbol @code{_bar} is an
6164 alias for the second. The fourth symbol, @code{another_foo} is resolved
6165 by "forwarding" to another module and treating it as an alias for
6166 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
6167 @code{var1} is declared to be a data object.
6169 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
6170 name of the output DLL. If @samp{<name>} does not include a suffix,
6171 the default library suffix, @samp{.DLL} is appended.
6173 When the .DEF file is used to build an application, rather than a
6174 library, the @code{NAME <name>} command should be used instead of
6175 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
6176 executable suffix, @samp{.EXE} is appended.
6178 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
6179 specification @code{BASE = <number>} may be used to specify a
6180 non-default base address for the image.
6182 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
6183 or they specify an empty string, the internal name is the same as the
6184 filename specified on the command line.
6186 The complete specification of an export symbol is:
6190 ( ( ( <name1> [ = <name2> ] )
6191 | ( <name1> = <module-name> . <external-name>))
6192 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
6195 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
6196 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
6197 @samp{<name1>} as a "forward" alias for the symbol
6198 @samp{<external-name>} in the DLL @samp{<module-name>}.
6199 Optionally, the symbol may be exported by the specified ordinal
6200 @samp{<integer>} alias.
6202 The optional keywords that follow the declaration indicate:
6204 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
6205 will still be exported by its ordinal alias (either the value specified
6206 by the .def specification or, otherwise, the value assigned by the
6207 linker). The symbol name, however, does remain visible in the import
6208 library (if any), unless @code{PRIVATE} is also specified.
6210 @code{DATA}: The symbol is a variable or object, rather than a function.
6211 The import lib will export only an indirect reference to @code{foo} as
6212 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
6215 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
6216 well as @code{_imp__foo} into the import library. Both refer to the
6217 read-only import address table's pointer to the variable, not to the
6218 variable itself. This can be dangerous. If the user code fails to add
6219 the @code{dllimport} attribute and also fails to explicitly add the
6220 extra indirection that the use of the attribute enforces, the
6221 application will behave unexpectedly.
6223 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
6224 it into the static import library used to resolve imports at link time. The
6225 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
6226 API at runtime or by by using the GNU ld extension of linking directly to
6227 the DLL without an import library.
6229 See ld/deffilep.y in the binutils sources for the full specification of
6230 other DEF file statements
6232 @cindex creating a DEF file
6233 While linking a shared dll, @command{ld} is able to create a DEF file
6234 with the @samp{--output-def <file>} command line option.
6236 @item Using decorations
6237 @cindex Using decorations
6238 Another way of marking symbols for export is to modify the source code
6239 itself, so that when building the DLL each symbol to be exported is
6243 __declspec(dllexport) int a_variable
6244 __declspec(dllexport) void a_function(int with_args)
6247 All such symbols will be exported from the DLL. If, however,
6248 any of the object files in the DLL contain symbols decorated in
6249 this way, then the normal auto-export behavior is disabled, unless
6250 the @samp{--export-all-symbols} option is also used.
6252 Note that object files that wish to access these symbols must @emph{not}
6253 decorate them with dllexport. Instead, they should use dllimport,
6257 __declspec(dllimport) int a_variable
6258 __declspec(dllimport) void a_function(int with_args)
6261 This complicates the structure of library header files, because
6262 when included by the library itself the header must declare the
6263 variables and functions as dllexport, but when included by client
6264 code the header must declare them as dllimport. There are a number
6265 of idioms that are typically used to do this; often client code can
6266 omit the __declspec() declaration completely. See
6267 @samp{--enable-auto-import} and @samp{automatic data imports} for more
6271 @cindex automatic data imports
6272 @item automatic data imports
6273 The standard Windows dll format supports data imports from dlls only
6274 by adding special decorations (dllimport/dllexport), which let the
6275 compiler produce specific assembler instructions to deal with this
6276 issue. This increases the effort necessary to port existing Un*x
6277 code to these platforms, especially for large
6278 c++ libraries and applications. The auto-import feature, which was
6279 initially provided by Paul Sokolovsky, allows one to omit the
6280 decorations to achieve a behavior that conforms to that on POSIX/Un*x
6281 platforms. This feature is enabled with the @samp{--enable-auto-import}
6282 command-line option, although it is enabled by default on cygwin/mingw.
6283 The @samp{--enable-auto-import} option itself now serves mainly to
6284 suppress any warnings that are ordinarily emitted when linked objects
6285 trigger the feature's use.
6287 auto-import of variables does not always work flawlessly without
6288 additional assistance. Sometimes, you will see this message
6290 "variable '<var>' can't be auto-imported. Please read the
6291 documentation for ld's @code{--enable-auto-import} for details."
6293 The @samp{--enable-auto-import} documentation explains why this error
6294 occurs, and several methods that can be used to overcome this difficulty.
6295 One of these methods is the @emph{runtime pseudo-relocs} feature, described
6298 @cindex runtime pseudo-relocation
6299 For complex variables imported from DLLs (such as structs or classes),
6300 object files typically contain a base address for the variable and an
6301 offset (@emph{addend}) within the variable--to specify a particular
6302 field or public member, for instance. Unfortunately, the runtime loader used
6303 in win32 environments is incapable of fixing these references at runtime
6304 without the additional information supplied by dllimport/dllexport decorations.
6305 The standard auto-import feature described above is unable to resolve these
6308 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
6309 be resolved without error, while leaving the task of adjusting the references
6310 themselves (with their non-zero addends) to specialized code provided by the
6311 runtime environment. Recent versions of the cygwin and mingw environments and
6312 compilers provide this runtime support; older versions do not. However, the
6313 support is only necessary on the developer's platform; the compiled result will
6314 run without error on an older system.
6316 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
6319 @cindex direct linking to a dll
6320 @item direct linking to a dll
6321 The cygwin/mingw ports of @command{ld} support the direct linking,
6322 including data symbols, to a dll without the usage of any import
6323 libraries. This is much faster and uses much less memory than does the
6324 traditional import library method, especially when linking large
6325 libraries or applications. When @command{ld} creates an import lib, each
6326 function or variable exported from the dll is stored in its own bfd, even
6327 though a single bfd could contain many exports. The overhead involved in
6328 storing, loading, and processing so many bfd's is quite large, and explains the
6329 tremendous time, memory, and storage needed to link against particularly
6330 large or complex libraries when using import libs.
6332 Linking directly to a dll uses no extra command-line switches other than
6333 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
6334 of names to match each library. All that is needed from the developer's
6335 perspective is an understanding of this search, in order to force ld to
6336 select the dll instead of an import library.
6339 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
6340 to find, in the first directory of its search path,
6352 before moving on to the next directory in the search path.
6354 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
6355 where @samp{<prefix>} is set by the @command{ld} option
6356 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6357 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6360 Other win32-based unix environments, such as mingw or pw32, may use other
6361 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
6362 was originally intended to help avoid name conflicts among dll's built for the
6363 various win32/un*x environments, so that (for example) two versions of a zlib dll
6364 could coexist on the same machine.
6366 The generic cygwin/mingw path layout uses a @samp{bin} directory for
6367 applications and dll's and a @samp{lib} directory for the import
6368 libraries (using cygwin nomenclature):
6374 libxxx.dll.a (in case of dll's)
6375 libxxx.a (in case of static archive)
6378 Linking directly to a dll without using the import library can be
6381 1. Use the dll directly by adding the @samp{bin} path to the link line
6383 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6386 However, as the dll's often have version numbers appended to their names
6387 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6388 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
6389 not versioned, and do not have this difficulty.
6391 2. Create a symbolic link from the dll to a file in the @samp{lib}
6392 directory according to the above mentioned search pattern. This
6393 should be used to avoid unwanted changes in the tools needed for
6397 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6400 Then you can link without any make environment changes.
6403 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6406 This technique also avoids the version number problems, because the following is
6413 libxxx.dll.a -> ../bin/cygxxx-5.dll
6416 Linking directly to a dll without using an import lib will work
6417 even when auto-import features are exercised, and even when
6418 @samp{--enable-runtime-pseudo-relocs} is used.
6420 Given the improvements in speed and memory usage, one might justifiably
6421 wonder why import libraries are used at all. There are three reasons:
6423 1. Until recently, the link-directly-to-dll functionality did @emph{not}
6424 work with auto-imported data.
6426 2. Sometimes it is necessary to include pure static objects within the
6427 import library (which otherwise contains only bfd's for indirection
6428 symbols that point to the exports of a dll). Again, the import lib
6429 for the cygwin kernel makes use of this ability, and it is not
6430 possible to do this without an import lib.
6432 3. Symbol aliases can only be resolved using an import lib. This is
6433 critical when linking against OS-supplied dll's (eg, the win32 API)
6434 in which symbols are usually exported as undecorated aliases of their
6435 stdcall-decorated assembly names.
6437 So, import libs are not going away. But the ability to replace
6438 true import libs with a simple symbolic link to (or a copy of)
6439 a dll, in many cases, is a useful addition to the suite of tools
6440 binutils makes available to the win32 developer. Given the
6441 massive improvements in memory requirements during linking, storage
6442 requirements, and linking speed, we expect that many developers
6443 will soon begin to use this feature whenever possible.
6445 @item symbol aliasing
6447 @item adding additional names
6448 Sometimes, it is useful to export symbols with additional names.
6449 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6450 exported as @samp{_foo} by using special directives in the DEF file
6451 when creating the dll. This will affect also the optional created
6452 import library. Consider the following DEF file:
6455 LIBRARY "xyz.dll" BASE=0x61000000
6462 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6464 Another method for creating a symbol alias is to create it in the
6465 source code using the "weak" attribute:
6468 void foo () @{ /* Do something. */; @}
6469 void _foo () __attribute__ ((weak, alias ("foo")));
6472 See the gcc manual for more information about attributes and weak
6475 @item renaming symbols
6476 Sometimes it is useful to rename exports. For instance, the cygwin
6477 kernel does this regularly. A symbol @samp{_foo} can be exported as
6478 @samp{foo} but not as @samp{_foo} by using special directives in the
6479 DEF file. (This will also affect the import library, if it is
6480 created). In the following example:
6483 LIBRARY "xyz.dll" BASE=0x61000000
6489 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6493 Note: using a DEF file disables the default auto-export behavior,
6494 unless the @samp{--export-all-symbols} command line option is used.
6495 If, however, you are trying to rename symbols, then you should list
6496 @emph{all} desired exports in the DEF file, including the symbols
6497 that are not being renamed, and do @emph{not} use the
6498 @samp{--export-all-symbols} option. If you list only the
6499 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6500 to handle the other symbols, then the both the new names @emph{and}
6501 the original names for the renamed symbols will be exported.
6502 In effect, you'd be aliasing those symbols, not renaming them,
6503 which is probably not what you wanted.
6505 @cindex weak externals
6506 @item weak externals
6507 The Windows object format, PE, specifies a form of weak symbols called
6508 weak externals. When a weak symbol is linked and the symbol is not
6509 defined, the weak symbol becomes an alias for some other symbol. There
6510 are three variants of weak externals:
6512 @item Definition is searched for in objects and libraries, historically
6513 called lazy externals.
6514 @item Definition is searched for only in other objects, not in libraries.
6515 This form is not presently implemented.
6516 @item No search; the symbol is an alias. This form is not presently
6519 As a GNU extension, weak symbols that do not specify an alternate symbol
6520 are supported. If the symbol is undefined when linking, the symbol
6521 uses a default value.
6535 @section @code{ld} and Xtensa Processors
6537 @cindex Xtensa processors
6538 The default @command{ld} behavior for Xtensa processors is to interpret
6539 @code{SECTIONS} commands so that lists of explicitly named sections in a
6540 specification with a wildcard file will be interleaved when necessary to
6541 keep literal pools within the range of PC-relative load offsets. For
6542 example, with the command:
6554 @command{ld} may interleave some of the @code{.literal}
6555 and @code{.text} sections from different object files to ensure that the
6556 literal pools are within the range of PC-relative load offsets. A valid
6557 interleaving might place the @code{.literal} sections from an initial
6558 group of files followed by the @code{.text} sections of that group of
6559 files. Then, the @code{.literal} sections from the rest of the files
6560 and the @code{.text} sections from the rest of the files would follow.
6562 @cindex @option{--relax} on Xtensa
6563 @cindex relaxing on Xtensa
6564 Relaxation is enabled by default for the Xtensa version of @command{ld} and
6565 provides two important link-time optimizations. The first optimization
6566 is to combine identical literal values to reduce code size. A redundant
6567 literal will be removed and all the @code{L32R} instructions that use it
6568 will be changed to reference an identical literal, as long as the
6569 location of the replacement literal is within the offset range of all
6570 the @code{L32R} instructions. The second optimization is to remove
6571 unnecessary overhead from assembler-generated ``longcall'' sequences of
6572 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
6573 range of direct @code{CALL@var{n}} instructions.
6575 For each of these cases where an indirect call sequence can be optimized
6576 to a direct call, the linker will change the @code{CALLX@var{n}}
6577 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
6578 instruction, and remove the literal referenced by the @code{L32R}
6579 instruction if it is not used for anything else. Removing the
6580 @code{L32R} instruction always reduces code size but can potentially
6581 hurt performance by changing the alignment of subsequent branch targets.
6582 By default, the linker will always preserve alignments, either by
6583 switching some instructions between 24-bit encodings and the equivalent
6584 density instructions or by inserting a no-op in place of the @code{L32R}
6585 instruction that was removed. If code size is more important than
6586 performance, the @option{--size-opt} option can be used to prevent the
6587 linker from widening density instructions or inserting no-ops, except in
6588 a few cases where no-ops are required for correctness.
6590 The following Xtensa-specific command-line options can be used to
6593 @cindex Xtensa options
6597 Since the Xtensa version of @code{ld} enables the @option{--relax} option
6598 by default, the @option{--no-relax} option is provided to disable
6602 When optimizing indirect calls to direct calls, optimize for code size
6603 more than performance. With this option, the linker will not insert
6604 no-ops or widen density instructions to preserve branch target
6605 alignment. There may still be some cases where no-ops are required to
6606 preserve the correctness of the code.
6614 @ifclear SingleFormat
6619 @cindex object file management
6620 @cindex object formats available
6622 The linker accesses object and archive files using the BFD libraries.
6623 These libraries allow the linker to use the same routines to operate on
6624 object files whatever the object file format. A different object file
6625 format can be supported simply by creating a new BFD back end and adding
6626 it to the library. To conserve runtime memory, however, the linker and
6627 associated tools are usually configured to support only a subset of the
6628 object file formats available. You can use @code{objdump -i}
6629 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
6630 list all the formats available for your configuration.
6632 @cindex BFD requirements
6633 @cindex requirements for BFD
6634 As with most implementations, BFD is a compromise between
6635 several conflicting requirements. The major factor influencing
6636 BFD design was efficiency: any time used converting between
6637 formats is time which would not have been spent had BFD not
6638 been involved. This is partly offset by abstraction payback; since
6639 BFD simplifies applications and back ends, more time and care
6640 may be spent optimizing algorithms for a greater speed.
6642 One minor artifact of the BFD solution which you should bear in
6643 mind is the potential for information loss. There are two places where
6644 useful information can be lost using the BFD mechanism: during
6645 conversion and during output. @xref{BFD information loss}.
6648 * BFD outline:: How it works: an outline of BFD
6652 @section How It Works: An Outline of BFD
6653 @cindex opening object files
6654 @include bfdsumm.texi
6657 @node Reporting Bugs
6658 @chapter Reporting Bugs
6659 @cindex bugs in @command{ld}
6660 @cindex reporting bugs in @command{ld}
6662 Your bug reports play an essential role in making @command{ld} reliable.
6664 Reporting a bug may help you by bringing a solution to your problem, or
6665 it may not. But in any case the principal function of a bug report is
6666 to help the entire community by making the next version of @command{ld}
6667 work better. Bug reports are your contribution to the maintenance of
6670 In order for a bug report to serve its purpose, you must include the
6671 information that enables us to fix the bug.
6674 * Bug Criteria:: Have you found a bug?
6675 * Bug Reporting:: How to report bugs
6679 @section Have You Found a Bug?
6680 @cindex bug criteria
6682 If you are not sure whether you have found a bug, here are some guidelines:
6685 @cindex fatal signal
6686 @cindex linker crash
6687 @cindex crash of linker
6689 If the linker gets a fatal signal, for any input whatever, that is a
6690 @command{ld} bug. Reliable linkers never crash.
6692 @cindex error on valid input
6694 If @command{ld} produces an error message for valid input, that is a bug.
6696 @cindex invalid input
6698 If @command{ld} does not produce an error message for invalid input, that
6699 may be a bug. In the general case, the linker can not verify that
6700 object files are correct.
6703 If you are an experienced user of linkers, your suggestions for
6704 improvement of @command{ld} are welcome in any case.
6708 @section How to Report Bugs
6710 @cindex @command{ld} bugs, reporting
6712 A number of companies and individuals offer support for @sc{gnu}
6713 products. If you obtained @command{ld} from a support organization, we
6714 recommend you contact that organization first.
6716 You can find contact information for many support companies and
6717 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
6721 Otherwise, send bug reports for @command{ld} to
6725 The fundamental principle of reporting bugs usefully is this:
6726 @strong{report all the facts}. If you are not sure whether to state a
6727 fact or leave it out, state it!
6729 Often people omit facts because they think they know what causes the
6730 problem and assume that some details do not matter. Thus, you might
6731 assume that the name of a symbol you use in an example does not
6732 matter. Well, probably it does not, but one cannot be sure. Perhaps
6733 the bug is a stray memory reference which happens to fetch from the
6734 location where that name is stored in memory; perhaps, if the name
6735 were different, the contents of that location would fool the linker
6736 into doing the right thing despite the bug. Play it safe and give a
6737 specific, complete example. That is the easiest thing for you to do,
6738 and the most helpful.
6740 Keep in mind that the purpose of a bug report is to enable us to fix
6741 the bug if it is new to us. Therefore, always write your bug reports
6742 on the assumption that the bug has not been reported previously.
6744 Sometimes people give a few sketchy facts and ask, ``Does this ring a
6745 bell?'' This cannot help us fix a bug, so it is basically useless. We
6746 respond by asking for enough details to enable us to investigate.
6747 You might as well expedite matters by sending them to begin with.
6749 To enable us to fix the bug, you should include all these things:
6753 The version of @command{ld}. @command{ld} announces it if you start it with
6754 the @samp{--version} argument.
6756 Without this, we will not know whether there is any point in looking for
6757 the bug in the current version of @command{ld}.
6760 Any patches you may have applied to the @command{ld} source, including any
6761 patches made to the @code{BFD} library.
6764 The type of machine you are using, and the operating system name and
6768 What compiler (and its version) was used to compile @command{ld}---e.g.
6772 The command arguments you gave the linker to link your example and
6773 observe the bug. To guarantee you will not omit something important,
6774 list them all. A copy of the Makefile (or the output from make) is
6777 If we were to try to guess the arguments, we would probably guess wrong
6778 and then we might not encounter the bug.
6781 A complete input file, or set of input files, that will reproduce the
6782 bug. It is generally most helpful to send the actual object files
6783 provided that they are reasonably small. Say no more than 10K. For
6784 bigger files you can either make them available by FTP or HTTP or else
6785 state that you are willing to send the object file(s) to whomever
6786 requests them. (Note - your email will be going to a mailing list, so
6787 we do not want to clog it up with large attachments). But small
6788 attachments are best.
6790 If the source files were assembled using @code{gas} or compiled using
6791 @code{gcc}, then it may be OK to send the source files rather than the
6792 object files. In this case, be sure to say exactly what version of
6793 @code{gas} or @code{gcc} was used to produce the object files. Also say
6794 how @code{gas} or @code{gcc} were configured.
6797 A description of what behavior you observe that you believe is
6798 incorrect. For example, ``It gets a fatal signal.''
6800 Of course, if the bug is that @command{ld} gets a fatal signal, then we
6801 will certainly notice it. But if the bug is incorrect output, we might
6802 not notice unless it is glaringly wrong. You might as well not give us
6803 a chance to make a mistake.
6805 Even if the problem you experience is a fatal signal, you should still
6806 say so explicitly. Suppose something strange is going on, such as, your
6807 copy of @command{ld} is out of sync, or you have encountered a bug in the
6808 C library on your system. (This has happened!) Your copy might crash
6809 and ours would not. If you told us to expect a crash, then when ours
6810 fails to crash, we would know that the bug was not happening for us. If
6811 you had not told us to expect a crash, then we would not be able to draw
6812 any conclusion from our observations.
6815 If you wish to suggest changes to the @command{ld} source, send us context
6816 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
6817 @samp{-p} option. Always send diffs from the old file to the new file.
6818 If you even discuss something in the @command{ld} source, refer to it by
6819 context, not by line number.
6821 The line numbers in our development sources will not match those in your
6822 sources. Your line numbers would convey no useful information to us.
6825 Here are some things that are not necessary:
6829 A description of the envelope of the bug.
6831 Often people who encounter a bug spend a lot of time investigating
6832 which changes to the input file will make the bug go away and which
6833 changes will not affect it.
6835 This is often time consuming and not very useful, because the way we
6836 will find the bug is by running a single example under the debugger
6837 with breakpoints, not by pure deduction from a series of examples.
6838 We recommend that you save your time for something else.
6840 Of course, if you can find a simpler example to report @emph{instead}
6841 of the original one, that is a convenience for us. Errors in the
6842 output will be easier to spot, running under the debugger will take
6843 less time, and so on.
6845 However, simplification is not vital; if you do not want to do this,
6846 report the bug anyway and send us the entire test case you used.
6849 A patch for the bug.
6851 A patch for the bug does help us if it is a good one. But do not omit
6852 the necessary information, such as the test case, on the assumption that
6853 a patch is all we need. We might see problems with your patch and decide
6854 to fix the problem another way, or we might not understand it at all.
6856 Sometimes with a program as complicated as @command{ld} it is very hard to
6857 construct an example that will make the program follow a certain path
6858 through the code. If you do not send us the example, we will not be
6859 able to construct one, so we will not be able to verify that the bug is
6862 And if we cannot understand what bug you are trying to fix, or why your
6863 patch should be an improvement, we will not install it. A test case will
6864 help us to understand.
6867 A guess about what the bug is or what it depends on.
6869 Such guesses are usually wrong. Even we cannot guess right about such
6870 things without first using the debugger to find the facts.
6874 @appendix MRI Compatible Script Files
6875 @cindex MRI compatibility
6876 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
6877 linker, @command{ld} can use MRI compatible linker scripts as an
6878 alternative to the more general-purpose linker scripting language
6879 described in @ref{Scripts}. MRI compatible linker scripts have a much
6880 simpler command set than the scripting language otherwise used with
6881 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
6882 linker commands; these commands are described here.
6884 In general, MRI scripts aren't of much use with the @code{a.out} object
6885 file format, since it only has three sections and MRI scripts lack some
6886 features to make use of them.
6888 You can specify a file containing an MRI-compatible script using the
6889 @samp{-c} command-line option.
6891 Each command in an MRI-compatible script occupies its own line; each
6892 command line starts with the keyword that identifies the command (though
6893 blank lines are also allowed for punctuation). If a line of an
6894 MRI-compatible script begins with an unrecognized keyword, @command{ld}
6895 issues a warning message, but continues processing the script.
6897 Lines beginning with @samp{*} are comments.
6899 You can write these commands using all upper-case letters, or all
6900 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
6901 The following list shows only the upper-case form of each command.
6904 @cindex @code{ABSOLUTE} (MRI)
6905 @item ABSOLUTE @var{secname}
6906 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
6907 Normally, @command{ld} includes in the output file all sections from all
6908 the input files. However, in an MRI-compatible script, you can use the
6909 @code{ABSOLUTE} command to restrict the sections that will be present in
6910 your output program. If the @code{ABSOLUTE} command is used at all in a
6911 script, then only the sections named explicitly in @code{ABSOLUTE}
6912 commands will appear in the linker output. You can still use other
6913 input sections (whatever you select on the command line, or using
6914 @code{LOAD}) to resolve addresses in the output file.
6916 @cindex @code{ALIAS} (MRI)
6917 @item ALIAS @var{out-secname}, @var{in-secname}
6918 Use this command to place the data from input section @var{in-secname}
6919 in a section called @var{out-secname} in the linker output file.
6921 @var{in-secname} may be an integer.
6923 @cindex @code{ALIGN} (MRI)
6924 @item ALIGN @var{secname} = @var{expression}
6925 Align the section called @var{secname} to @var{expression}. The
6926 @var{expression} should be a power of two.
6928 @cindex @code{BASE} (MRI)
6929 @item BASE @var{expression}
6930 Use the value of @var{expression} as the lowest address (other than
6931 absolute addresses) in the output file.
6933 @cindex @code{CHIP} (MRI)
6934 @item CHIP @var{expression}
6935 @itemx CHIP @var{expression}, @var{expression}
6936 This command does nothing; it is accepted only for compatibility.
6938 @cindex @code{END} (MRI)
6940 This command does nothing whatever; it's only accepted for compatibility.
6942 @cindex @code{FORMAT} (MRI)
6943 @item FORMAT @var{output-format}
6944 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
6945 language, but restricted to one of these output formats:
6949 S-records, if @var{output-format} is @samp{S}
6952 IEEE, if @var{output-format} is @samp{IEEE}
6955 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
6959 @cindex @code{LIST} (MRI)
6960 @item LIST @var{anything}@dots{}
6961 Print (to the standard output file) a link map, as produced by the
6962 @command{ld} command-line option @samp{-M}.
6964 The keyword @code{LIST} may be followed by anything on the
6965 same line, with no change in its effect.
6967 @cindex @code{LOAD} (MRI)
6968 @item LOAD @var{filename}
6969 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
6970 Include one or more object file @var{filename} in the link; this has the
6971 same effect as specifying @var{filename} directly on the @command{ld}
6974 @cindex @code{NAME} (MRI)
6975 @item NAME @var{output-name}
6976 @var{output-name} is the name for the program produced by @command{ld}; the
6977 MRI-compatible command @code{NAME} is equivalent to the command-line
6978 option @samp{-o} or the general script language command @code{OUTPUT}.
6980 @cindex @code{ORDER} (MRI)
6981 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
6982 @itemx ORDER @var{secname} @var{secname} @var{secname}
6983 Normally, @command{ld} orders the sections in its output file in the
6984 order in which they first appear in the input files. In an MRI-compatible
6985 script, you can override this ordering with the @code{ORDER} command. The
6986 sections you list with @code{ORDER} will appear first in your output
6987 file, in the order specified.
6989 @cindex @code{PUBLIC} (MRI)
6990 @item PUBLIC @var{name}=@var{expression}
6991 @itemx PUBLIC @var{name},@var{expression}
6992 @itemx PUBLIC @var{name} @var{expression}
6993 Supply a value (@var{expression}) for external symbol
6994 @var{name} used in the linker input files.
6996 @cindex @code{SECT} (MRI)
6997 @item SECT @var{secname}, @var{expression}
6998 @itemx SECT @var{secname}=@var{expression}
6999 @itemx SECT @var{secname} @var{expression}
7000 You can use any of these three forms of the @code{SECT} command to
7001 specify the start address (@var{expression}) for section @var{secname}.
7002 If you have more than one @code{SECT} statement for the same
7003 @var{secname}, only the @emph{first} sets the start address.
7009 @unnumbered LD Index
7014 % I think something like @colophon should be in texinfo. In the
7016 \long\def\colophon{\hbox to0pt{}\vfill
7017 \centerline{The body of this manual is set in}
7018 \centerline{\fontname\tenrm,}
7019 \centerline{with headings in {\bf\fontname\tenbf}}
7020 \centerline{and examples in {\tt\fontname\tentt}.}
7021 \centerline{{\it\fontname\tenit\/} and}
7022 \centerline{{\sl\fontname\tensl\/}}
7023 \centerline{are used for emphasis.}\vfill}
7025 % Blame: doc@cygnus.com, 28mar91.