3 @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 @c 2001, 2002, 2003, 2004, 2005, 2006 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
58 * Ld: (ld). The GNU linker.
64 This file documents the @sc{gnu} linker LD version @value{VERSION}.
66 Copyright (C) 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
67 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
71 Permission is granted to copy, distribute and/or modify this document
72 under the terms of the GNU Free Documentation License, Version 1.1
73 or any later version published by the Free Software Foundation;
74 with no Invariant Sections, with no Front-Cover Texts, and with no
75 Back-Cover Texts. A copy of the license is included in the
76 section entitled ``GNU Free Documentation License''.
78 Permission is granted to process this file through Tex and print the
79 results, provided the printed document carries copying permission
80 notice identical to this one except for the removal of this paragraph
81 (this paragraph not being relevant to the printed manual).
87 @setchapternewpage odd
88 @settitle Using LD, the GNU linker
91 @subtitle The GNU linker
93 @subtitle @code{ld} version 2
94 @subtitle Version @value{VERSION}
95 @author Steve Chamberlain
96 @author Ian Lance Taylor
101 \hfill Red Hat Inc\par
102 \hfill nickc\@credhat.com, doc\@redhat.com\par
103 \hfill {\it Using LD, the GNU linker}\par
104 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
106 \global\parindent=0pt % Steve likes it this way.
109 @vskip 0pt plus 1filll
110 @c man begin COPYRIGHT
111 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
112 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
114 Permission is granted to copy, distribute and/or modify this document
115 under the terms of the GNU Free Documentation License, Version 1.1
116 or any later version published by the Free Software Foundation;
117 with no Invariant Sections, with no Front-Cover Texts, and with no
118 Back-Cover Texts. A copy of the license is included in the
119 section entitled ``GNU Free Documentation License''.
124 @c FIXME: Talk about importance of *order* of args, cmds to linker!
129 This file documents the @sc{gnu} linker ld version @value{VERSION}.
131 This document is distributed under the terms of the GNU Free
132 Documentation License. A copy of the license is included in the
133 section entitled ``GNU Free Documentation License''.
136 * Overview:: Overview
137 * Invocation:: Invocation
138 * Scripts:: Linker Scripts
140 * Machine Dependent:: Machine Dependent Features
144 * H8/300:: ld and the H8/300
147 * Renesas:: ld and other Renesas micros
150 * i960:: ld and the Intel 960 family
153 * ARM:: ld and the ARM family
156 * HPPA ELF32:: ld and HPPA 32-bit ELF
159 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
162 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
165 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
168 * TI COFF:: ld and the TI COFF
171 * Win32:: ld and WIN32 (cygwin/mingw)
174 * Xtensa:: ld and Xtensa Processors
177 @ifclear SingleFormat
180 @c Following blank line required for remaining bug in makeinfo conds/menus
182 * Reporting Bugs:: Reporting Bugs
183 * MRI:: MRI Compatible Script Files
184 * GNU Free Documentation License:: GNU Free Documentation License
185 * LD Index:: LD Index
192 @cindex @sc{gnu} linker
193 @cindex what is this?
196 @c man begin SYNOPSIS
197 ld [@b{options}] @var{objfile} @dots{}
201 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
202 the Info entries for @file{binutils} and
207 @c man begin DESCRIPTION
209 @command{ld} combines a number of object and archive files, relocates
210 their data and ties up symbol references. Usually the last step in
211 compiling a program is to run @command{ld}.
213 @command{ld} accepts Linker Command Language files written in
214 a superset of AT&T's Link Editor Command Language syntax,
215 to provide explicit and total control over the linking process.
219 This man page does not describe the command language; see the
220 @command{ld} entry in @code{info}, or the manual
221 ld: the GNU linker, for full details on the command language and
222 on other aspects of the GNU linker.
225 @ifclear SingleFormat
226 This version of @command{ld} uses the general purpose BFD libraries
227 to operate on object files. This allows @command{ld} to read, combine, and
228 write object files in many different formats---for example, COFF or
229 @code{a.out}. Different formats may be linked together to produce any
230 available kind of object file. @xref{BFD}, for more information.
233 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
234 linkers in providing diagnostic information. Many linkers abandon
235 execution immediately upon encountering an error; whenever possible,
236 @command{ld} continues executing, allowing you to identify other errors
237 (or, in some cases, to get an output file in spite of the error).
244 @c man begin DESCRIPTION
246 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
247 and to be as compatible as possible with other linkers. As a result,
248 you have many choices to control its behavior.
254 * Options:: Command Line Options
255 * Environment:: Environment Variables
259 @section Command Line Options
267 The linker supports a plethora of command-line options, but in actual
268 practice few of them are used in any particular context.
269 @cindex standard Unix system
270 For instance, a frequent use of @command{ld} is to link standard Unix
271 object files on a standard, supported Unix system. On such a system, to
272 link a file @code{hello.o}:
275 ld -o @var{output} /lib/crt0.o hello.o -lc
278 This tells @command{ld} to produce a file called @var{output} as the
279 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
280 the library @code{libc.a}, which will come from the standard search
281 directories. (See the discussion of the @samp{-l} option below.)
283 Some of the command-line options to @command{ld} may be specified at any
284 point in the command line. However, options which refer to files, such
285 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
286 which the option appears in the command line, relative to the object
287 files and other file options. Repeating non-file options with a
288 different argument will either have no further effect, or override prior
289 occurrences (those further to the left on the command line) of that
290 option. Options which may be meaningfully specified more than once are
291 noted in the descriptions below.
294 Non-option arguments are object files or archives which are to be linked
295 together. They may follow, precede, or be mixed in with command-line
296 options, except that an object file argument may not be placed between
297 an option and its argument.
299 Usually the linker is invoked with at least one object file, but you can
300 specify other forms of binary input files using @samp{-l}, @samp{-R},
301 and the script command language. If @emph{no} binary input files at all
302 are specified, the linker does not produce any output, and issues the
303 message @samp{No input files}.
305 If the linker cannot recognize the format of an object file, it will
306 assume that it is a linker script. A script specified in this way
307 augments the main linker script used for the link (either the default
308 linker script or the one specified by using @samp{-T}). This feature
309 permits the linker to link against a file which appears to be an object
310 or an archive, but actually merely defines some symbol values, or uses
311 @code{INPUT} or @code{GROUP} to load other objects. Note that
312 specifying a script in this way merely augments the main linker script;
313 use the @samp{-T} option to replace the default linker script entirely.
316 For options whose names are a single letter,
317 option arguments must either follow the option letter without intervening
318 whitespace, or be given as separate arguments immediately following the
319 option that requires them.
321 For options whose names are multiple letters, either one dash or two can
322 precede the option name; for example, @samp{-trace-symbol} and
323 @samp{--trace-symbol} are equivalent. Note---there is one exception to
324 this rule. Multiple letter options that start with a lower case 'o' can
325 only be preceeded by two dashes. This is to reduce confusion with the
326 @samp{-o} option. So for example @samp{-omagic} sets the output file
327 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
330 Arguments to multiple-letter options must either be separated from the
331 option name by an equals sign, or be given as separate arguments
332 immediately following the option that requires them. For example,
333 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
334 Unique abbreviations of the names of multiple-letter options are
337 Note---if the linker is being invoked indirectly, via a compiler driver
338 (e.g. @samp{gcc}) then all the linker command line options should be
339 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
340 compiler driver) like this:
343 gcc -Wl,--startgroup foo.o bar.o -Wl,--endgroup
346 This is important, because otherwise the compiler driver program may
347 silently drop the linker options, resulting in a bad link.
349 Here is a table of the generic command line switches accepted by the GNU
353 @include at-file.texi
355 @kindex -a@var{keyword}
356 @item -a@var{keyword}
357 This option is supported for HP/UX compatibility. The @var{keyword}
358 argument must be one of the strings @samp{archive}, @samp{shared}, or
359 @samp{default}. @samp{-aarchive} is functionally equivalent to
360 @samp{-Bstatic}, and the other two keywords are functionally equivalent
361 to @samp{-Bdynamic}. This option may be used any number of times.
364 @cindex architectures
366 @item -A@var{architecture}
367 @kindex --architecture=@var{arch}
368 @itemx --architecture=@var{architecture}
369 In the current release of @command{ld}, this option is useful only for the
370 Intel 960 family of architectures. In that @command{ld} configuration, the
371 @var{architecture} argument identifies the particular architecture in
372 the 960 family, enabling some safeguards and modifying the
373 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
374 family}, for details.
376 Future releases of @command{ld} may support similar functionality for
377 other architecture families.
380 @ifclear SingleFormat
381 @cindex binary input format
382 @kindex -b @var{format}
383 @kindex --format=@var{format}
386 @item -b @var{input-format}
387 @itemx --format=@var{input-format}
388 @command{ld} may be configured to support more than one kind of object
389 file. If your @command{ld} is configured this way, you can use the
390 @samp{-b} option to specify the binary format for input object files
391 that follow this option on the command line. Even when @command{ld} is
392 configured to support alternative object formats, you don't usually need
393 to specify this, as @command{ld} should be configured to expect as a
394 default input format the most usual format on each machine.
395 @var{input-format} is a text string, the name of a particular format
396 supported by the BFD libraries. (You can list the available binary
397 formats with @samp{objdump -i}.)
400 You may want to use this option if you are linking files with an unusual
401 binary format. You can also use @samp{-b} to switch formats explicitly (when
402 linking object files of different formats), by including
403 @samp{-b @var{input-format}} before each group of object files in a
406 The default format is taken from the environment variable
411 You can also define the input format from a script, using the command
414 see @ref{Format Commands}.
418 @kindex -c @var{MRI-cmdfile}
419 @kindex --mri-script=@var{MRI-cmdfile}
420 @cindex compatibility, MRI
421 @item -c @var{MRI-commandfile}
422 @itemx --mri-script=@var{MRI-commandfile}
423 For compatibility with linkers produced by MRI, @command{ld} accepts script
424 files written in an alternate, restricted command language, described in
426 @ref{MRI,,MRI Compatible Script Files}.
429 the MRI Compatible Script Files section of GNU ld documentation.
431 Introduce MRI script files with
432 the option @samp{-c}; use the @samp{-T} option to run linker
433 scripts written in the general-purpose @command{ld} scripting language.
434 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
435 specified by any @samp{-L} options.
437 @cindex common allocation
444 These three options are equivalent; multiple forms are supported for
445 compatibility with other linkers. They assign space to common symbols
446 even if a relocatable output file is specified (with @samp{-r}). The
447 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
448 @xref{Miscellaneous Commands}.
450 @cindex entry point, from command line
451 @kindex -e @var{entry}
452 @kindex --entry=@var{entry}
454 @itemx --entry=@var{entry}
455 Use @var{entry} as the explicit symbol for beginning execution of your
456 program, rather than the default entry point. If there is no symbol
457 named @var{entry}, the linker will try to parse @var{entry} as a number,
458 and use that as the entry address (the number will be interpreted in
459 base 10; you may use a leading @samp{0x} for base 16, or a leading
460 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
461 and other ways of specifying the entry point.
463 @kindex --exclude-libs
464 @item --exclude-libs @var{lib},@var{lib},...
465 Specifies a list of archive libraries from which symbols should not be automatically
466 exported. The library names may be delimited by commas or colons. Specifying
467 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
468 automatic export. This option is available only for the i386 PE targeted
469 port of the linker and for ELF targeted ports. For i386 PE, symbols
470 explicitly listed in a .def file are still exported, regardless of this
471 option. For ELF targeted ports, symbols affected by this option will
472 be treated as hidden.
474 @cindex dynamic symbol table
476 @kindex --export-dynamic
478 @itemx --export-dynamic
479 When creating a dynamically linked executable, add all symbols to the
480 dynamic symbol table. The dynamic symbol table is the set of symbols
481 which are visible from dynamic objects at run time.
483 If you do not use this option, the dynamic symbol table will normally
484 contain only those symbols which are referenced by some dynamic object
485 mentioned in the link.
487 If you use @code{dlopen} to load a dynamic object which needs to refer
488 back to the symbols defined by the program, rather than some other
489 dynamic object, then you will probably need to use this option when
490 linking the program itself.
492 You can also use the version script to control what symbols should
493 be added to the dynamic symbol table if the output format supports it.
494 See the description of @samp{--version-script} in @ref{VERSION}.
496 @ifclear SingleFormat
497 @cindex big-endian objects
501 Link big-endian objects. This affects the default output format.
503 @cindex little-endian objects
506 Link little-endian objects. This affects the default output format.
512 @itemx --auxiliary @var{name}
513 When creating an ELF shared object, set the internal DT_AUXILIARY field
514 to the specified name. This tells the dynamic linker that the symbol
515 table of the shared object should be used as an auxiliary filter on the
516 symbol table of the shared object @var{name}.
518 If you later link a program against this filter object, then, when you
519 run the program, the dynamic linker will see the DT_AUXILIARY field. If
520 the dynamic linker resolves any symbols from the filter object, it will
521 first check whether there is a definition in the shared object
522 @var{name}. If there is one, it will be used instead of the definition
523 in the filter object. The shared object @var{name} need not exist.
524 Thus the shared object @var{name} may be used to provide an alternative
525 implementation of certain functions, perhaps for debugging or for
526 machine specific performance.
528 This option may be specified more than once. The DT_AUXILIARY entries
529 will be created in the order in which they appear on the command line.
534 @itemx --filter @var{name}
535 When creating an ELF shared object, set the internal DT_FILTER field to
536 the specified name. This tells the dynamic linker that the symbol table
537 of the shared object which is being created should be used as a filter
538 on the symbol table of the shared object @var{name}.
540 If you later link a program against this filter object, then, when you
541 run the program, the dynamic linker will see the DT_FILTER field. The
542 dynamic linker will resolve symbols according to the symbol table of the
543 filter object as usual, but it will actually link to the definitions
544 found in the shared object @var{name}. Thus the filter object can be
545 used to select a subset of the symbols provided by the object
548 Some older linkers used the @option{-F} option throughout a compilation
549 toolchain for specifying object-file format for both input and output
551 @ifclear SingleFormat
552 The @sc{gnu} linker uses other mechanisms for this purpose: the
553 @option{-b}, @option{--format}, @option{--oformat} options, the
554 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
555 environment variable.
557 The @sc{gnu} linker will ignore the @option{-F} option when not
558 creating an ELF shared object.
560 @cindex finalization function
562 @item -fini @var{name}
563 When creating an ELF executable or shared object, call NAME when the
564 executable or shared object is unloaded, by setting DT_FINI to the
565 address of the function. By default, the linker uses @code{_fini} as
566 the function to call.
570 Ignored. Provided for compatibility with other tools.
576 @itemx --gpsize=@var{value}
577 Set the maximum size of objects to be optimized using the GP register to
578 @var{size}. This is only meaningful for object file formats such as
579 MIPS ECOFF which supports putting large and small objects into different
580 sections. This is ignored for other object file formats.
582 @cindex runtime library name
584 @kindex -soname=@var{name}
586 @itemx -soname=@var{name}
587 When creating an ELF shared object, set the internal DT_SONAME field to
588 the specified name. When an executable is linked with a shared object
589 which has a DT_SONAME field, then when the executable is run the dynamic
590 linker will attempt to load the shared object specified by the DT_SONAME
591 field rather than the using the file name given to the linker.
594 @cindex incremental link
596 Perform an incremental link (same as option @samp{-r}).
598 @cindex initialization function
600 @item -init @var{name}
601 When creating an ELF executable or shared object, call NAME when the
602 executable or shared object is loaded, by setting DT_INIT to the address
603 of the function. By default, the linker uses @code{_init} as the
606 @cindex archive files, from cmd line
607 @kindex -l@var{archive}
608 @kindex --library=@var{archive}
609 @item -l@var{archive}
610 @itemx --library=@var{archive}
611 Add archive file @var{archive} to the list of files to link. This
612 option may be used any number of times. @command{ld} will search its
613 path-list for occurrences of @code{lib@var{archive}.a} for every
614 @var{archive} specified.
616 On systems which support shared libraries, @command{ld} may also search for
617 libraries with extensions other than @code{.a}. Specifically, on ELF
618 and SunOS systems, @command{ld} will search a directory for a library with
619 an extension of @code{.so} before searching for one with an extension of
620 @code{.a}. By convention, a @code{.so} extension indicates a shared
623 The linker will search an archive only once, at the location where it is
624 specified on the command line. If the archive defines a symbol which
625 was undefined in some object which appeared before the archive on the
626 command line, the linker will include the appropriate file(s) from the
627 archive. However, an undefined symbol in an object appearing later on
628 the command line will not cause the linker to search the archive again.
630 See the @option{-(} option for a way to force the linker to search
631 archives multiple times.
633 You may list the same archive multiple times on the command line.
636 This type of archive searching is standard for Unix linkers. However,
637 if you are using @command{ld} on AIX, note that it is different from the
638 behaviour of the AIX linker.
641 @cindex search directory, from cmd line
643 @kindex --library-path=@var{dir}
644 @item -L@var{searchdir}
645 @itemx --library-path=@var{searchdir}
646 Add path @var{searchdir} to the list of paths that @command{ld} will search
647 for archive libraries and @command{ld} control scripts. You may use this
648 option any number of times. The directories are searched in the order
649 in which they are specified on the command line. Directories specified
650 on the command line are searched before the default directories. All
651 @option{-L} options apply to all @option{-l} options, regardless of the
652 order in which the options appear.
654 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
655 by the @dfn{sysroot prefix}, a path specified when the linker is configured.
658 The default set of paths searched (without being specified with
659 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
660 some cases also on how it was configured. @xref{Environment}.
663 The paths can also be specified in a link script with the
664 @code{SEARCH_DIR} command. Directories specified this way are searched
665 at the point in which the linker script appears in the command line.
668 @kindex -m @var{emulation}
669 @item -m@var{emulation}
670 Emulate the @var{emulation} linker. You can list the available
671 emulations with the @samp{--verbose} or @samp{-V} options.
673 If the @samp{-m} option is not used, the emulation is taken from the
674 @code{LDEMULATION} environment variable, if that is defined.
676 Otherwise, the default emulation depends upon how the linker was
684 Print a link map to the standard output. A link map provides
685 information about the link, including the following:
689 Where object files are mapped into memory.
691 How common symbols are allocated.
693 All archive members included in the link, with a mention of the symbol
694 which caused the archive member to be brought in.
696 The values assigned to symbols.
698 Note - symbols whose values are computed by an expression which
699 involves a reference to a previous value of the same symbol may not
700 have correct result displayed in the link map. This is because the
701 linker discards intermediate results and only retains the final value
702 of an expression. Under such circumstances the linker will display
703 the final value enclosed by square brackets. Thus for example a
704 linker script containing:
712 will produce the following output in the link map if the @option{-M}
717 [0x0000000c] foo = (foo * 0x4)
718 [0x0000000c] foo = (foo + 0x8)
721 See @ref{Expressions} for more information about expressions in linker
726 @cindex read-only text
731 Turn off page alignment of sections, and mark the output as
732 @code{NMAGIC} if possible.
736 @cindex read/write from cmd line
740 Set the text and data sections to be readable and writable. Also, do
741 not page-align the data segment, and disable linking against shared
742 libraries. If the output format supports Unix style magic numbers,
743 mark the output as @code{OMAGIC}. Note: Although a writable text section
744 is allowed for PE-COFF targets, it does not conform to the format
745 specification published by Microsoft.
750 This option negates most of the effects of the @option{-N} option. It
751 sets the text section to be read-only, and forces the data segment to
752 be page-aligned. Note - this option does not enable linking against
753 shared libraries. Use @option{-Bdynamic} for this.
755 @kindex -o @var{output}
756 @kindex --output=@var{output}
757 @cindex naming the output file
758 @item -o @var{output}
759 @itemx --output=@var{output}
760 Use @var{output} as the name for the program produced by @command{ld}; if this
761 option is not specified, the name @file{a.out} is used by default. The
762 script command @code{OUTPUT} can also specify the output file name.
764 @kindex -O @var{level}
765 @cindex generating optimized output
767 If @var{level} is a numeric values greater than zero @command{ld} optimizes
768 the output. This might take significantly longer and therefore probably
769 should only be enabled for the final binary.
772 @kindex --emit-relocs
773 @cindex retain relocations in final executable
776 Leave relocation sections and contents in fully linked exececutables.
777 Post link analysis and optimization tools may need this information in
778 order to perform correct modifications of executables. This results
779 in larger executables.
781 This option is currently only supported on ELF platforms.
783 @kindex --force-dynamic
784 @cindex forcing the creation of dynamic sections
785 @item --force-dynamic
786 Force the output file to have dynamic sections. This option is specific
790 @cindex relocatable output
792 @kindex --relocatable
795 Generate relocatable output---i.e., generate an output file that can in
796 turn serve as input to @command{ld}. This is often called @dfn{partial
797 linking}. As a side effect, in environments that support standard Unix
798 magic numbers, this option also sets the output file's magic number to
800 @c ; see @option{-N}.
801 If this option is not specified, an absolute file is produced. When
802 linking C++ programs, this option @emph{will not} resolve references to
803 constructors; to do that, use @samp{-Ur}.
805 When an input file does not have the same format as the output file,
806 partial linking is only supported if that input file does not contain any
807 relocations. Different output formats can have further restrictions; for
808 example some @code{a.out}-based formats do not support partial linking
809 with input files in other formats at all.
811 This option does the same thing as @samp{-i}.
813 @kindex -R @var{file}
814 @kindex --just-symbols=@var{file}
815 @cindex symbol-only input
816 @item -R @var{filename}
817 @itemx --just-symbols=@var{filename}
818 Read symbol names and their addresses from @var{filename}, but do not
819 relocate it or include it in the output. This allows your output file
820 to refer symbolically to absolute locations of memory defined in other
821 programs. You may use this option more than once.
823 For compatibility with other ELF linkers, if the @option{-R} option is
824 followed by a directory name, rather than a file name, it is treated as
825 the @option{-rpath} option.
829 @cindex strip all symbols
832 Omit all symbol information from the output file.
835 @kindex --strip-debug
836 @cindex strip debugger symbols
839 Omit debugger symbol information (but not all symbols) from the output file.
843 @cindex input files, displaying
846 Print the names of the input files as @command{ld} processes them.
848 @kindex -T @var{script}
849 @kindex --script=@var{script}
851 @item -T @var{scriptfile}
852 @itemx --script=@var{scriptfile}
853 Use @var{scriptfile} as the linker script. This script replaces
854 @command{ld}'s default linker script (rather than adding to it), so
855 @var{commandfile} must specify everything necessary to describe the
856 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
857 the current directory, @code{ld} looks for it in the directories
858 specified by any preceding @samp{-L} options. Multiple @samp{-T}
861 @kindex -u @var{symbol}
862 @kindex --undefined=@var{symbol}
863 @cindex undefined symbol
864 @item -u @var{symbol}
865 @itemx --undefined=@var{symbol}
866 Force @var{symbol} to be entered in the output file as an undefined
867 symbol. Doing this may, for example, trigger linking of additional
868 modules from standard libraries. @samp{-u} may be repeated with
869 different option arguments to enter additional undefined symbols. This
870 option is equivalent to the @code{EXTERN} linker script command.
875 For anything other than C++ programs, this option is equivalent to
876 @samp{-r}: it generates relocatable output---i.e., an output file that can in
877 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
878 @emph{does} resolve references to constructors, unlike @samp{-r}.
879 It does not work to use @samp{-Ur} on files that were themselves linked
880 with @samp{-Ur}; once the constructor table has been built, it cannot
881 be added to. Use @samp{-Ur} only for the last partial link, and
882 @samp{-r} for the others.
884 @kindex --unique[=@var{SECTION}]
885 @item --unique[=@var{SECTION}]
886 Creates a separate output section for every input section matching
887 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
888 missing, for every orphan input section. An orphan section is one not
889 specifically mentioned in a linker script. You may use this option
890 multiple times on the command line; It prevents the normal merging of
891 input sections with the same name, overriding output section assignments
901 Display the version number for @command{ld}. The @option{-V} option also
902 lists the supported emulations.
905 @kindex --discard-all
906 @cindex deleting local symbols
909 Delete all local symbols.
912 @kindex --discard-locals
913 @cindex local symbols, deleting
914 @cindex L, deleting symbols beginning
916 @itemx --discard-locals
917 Delete all temporary local symbols. For most targets, this is all local
918 symbols whose names begin with @samp{L}.
920 @kindex -y @var{symbol}
921 @kindex --trace-symbol=@var{symbol}
922 @cindex symbol tracing
923 @item -y @var{symbol}
924 @itemx --trace-symbol=@var{symbol}
925 Print the name of each linked file in which @var{symbol} appears. This
926 option may be given any number of times. On many systems it is necessary
927 to prepend an underscore.
929 This option is useful when you have an undefined symbol in your link but
930 don't know where the reference is coming from.
932 @kindex -Y @var{path}
934 Add @var{path} to the default library search path. This option exists
935 for Solaris compatibility.
937 @kindex -z @var{keyword}
938 @item -z @var{keyword}
939 The recognized keywords are:
943 Combines multiple reloc sections and sorts them to make dynamic symbol
944 lookup caching possible.
947 Disallows undefined symbols in object files. Undefined symbols in
948 shared libraries are still allowed.
951 Marks the object as requiring executable stack.
954 This option is only meaningful when building a shared object.
955 It marks the object so that its runtime initialization will occur
956 before the runtime initialization of any other objects brought into
957 the process at the same time. Similarly the runtime finalization of
958 the object will occur after the runtime finalization of any other
962 Marks the object that its symbol table interposes before all symbols
963 but the primary executable.
966 When generating an executable or shared library, mark it to tell the
967 dynamic linker to defer function call resolution to the point when
968 the function is called (lazy binding), rather than at load time.
969 Lazy binding is the default.
972 Marks the object that its filters be processed immediately at
976 Allows multiple definitions.
979 Disables multiple reloc sections combining.
982 Disables production of copy relocs.
985 Marks the object that the search for dependencies of this object will
986 ignore any default library search paths.
989 Marks the object shouldn't be unloaded at runtime.
992 Marks the object not available to @code{dlopen}.
995 Marks the object can not be dumped by @code{dldump}.
998 Marks the object as not requiring executable stack.
1001 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1004 When generating an executable or shared library, mark it to tell the
1005 dynamic linker to resolve all symbols when the program is started, or
1006 when the shared library is linked to using dlopen, instead of
1007 deferring function call resolution to the point when the function is
1011 Marks the object may contain $ORIGIN.
1014 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1016 @item max-page-size=@var{value}
1017 Set the emulation maximum page size to @var{value}.
1019 @item common-page-size=@var{value}
1020 Set the emulation common page size to @var{value}.
1024 Other keywords are ignored for Solaris compatibility.
1027 @cindex groups of archives
1028 @item -( @var{archives} -)
1029 @itemx --start-group @var{archives} --end-group
1030 The @var{archives} should be a list of archive files. They may be
1031 either explicit file names, or @samp{-l} options.
1033 The specified archives are searched repeatedly until no new undefined
1034 references are created. Normally, an archive is searched only once in
1035 the order that it is specified on the command line. If a symbol in that
1036 archive is needed to resolve an undefined symbol referred to by an
1037 object in an archive that appears later on the command line, the linker
1038 would not be able to resolve that reference. By grouping the archives,
1039 they all be searched repeatedly until all possible references are
1042 Using this option has a significant performance cost. It is best to use
1043 it only when there are unavoidable circular references between two or
1046 @kindex --accept-unknown-input-arch
1047 @kindex --no-accept-unknown-input-arch
1048 @item --accept-unknown-input-arch
1049 @itemx --no-accept-unknown-input-arch
1050 Tells the linker to accept input files whose architecture cannot be
1051 recognised. The assumption is that the user knows what they are doing
1052 and deliberately wants to link in these unknown input files. This was
1053 the default behaviour of the linker, before release 2.14. The default
1054 behaviour from release 2.14 onwards is to reject such input files, and
1055 so the @samp{--accept-unknown-input-arch} option has been added to
1056 restore the old behaviour.
1059 @kindex --no-as-needed
1061 @itemx --no-as-needed
1062 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1063 on the command line after the @option{--as-needed} option. Normally,
1064 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1065 on the command line, regardless of whether the library is actually
1066 needed. @option{--as-needed} causes DT_NEEDED tags to only be emitted
1067 for libraries that satisfy some symbol reference from regular objects
1068 which is undefined at the point that the library was linked.
1069 @option{--no-as-needed} restores the default behaviour.
1071 @kindex --add-needed
1072 @kindex --no-add-needed
1074 @itemx --no-add-needed
1075 This option affects the treatment of dynamic libraries from ELF
1076 DT_NEEDED tags in dynamic libraries mentioned on the command line after
1077 the @option{--no-add-needed} option. Normally, the linker will add
1078 a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1079 @option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1080 for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1081 the default behaviour.
1083 @kindex -assert @var{keyword}
1084 @item -assert @var{keyword}
1085 This option is ignored for SunOS compatibility.
1089 @kindex -call_shared
1093 Link against dynamic libraries. This is only meaningful on platforms
1094 for which shared libraries are supported. This option is normally the
1095 default on such platforms. The different variants of this option are
1096 for compatibility with various systems. You may use this option
1097 multiple times on the command line: it affects library searching for
1098 @option{-l} options which follow it.
1102 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1103 section. This causes the runtime linker to handle lookups in this
1104 object and its dependencies to be performed only inside the group.
1105 @option{--unresolved-symbols=report-all} is implied. This option is
1106 only meaningful on ELF platforms which support shared libraries.
1116 Do not link against shared libraries. This is only meaningful on
1117 platforms for which shared libraries are supported. The different
1118 variants of this option are for compatibility with various systems. You
1119 may use this option multiple times on the command line: it affects
1120 library searching for @option{-l} options which follow it. This
1121 option also implies @option{--unresolved-symbols=report-all}. This
1122 option can be used with @option{-shared}. Doing so means that a
1123 shared library is being created but that all of the library's external
1124 references must be resolved by pulling in entries from static
1129 When creating a shared library, bind references to global symbols to the
1130 definition within the shared library, if any. Normally, it is possible
1131 for a program linked against a shared library to override the definition
1132 within the shared library. This option is only meaningful on ELF
1133 platforms which support shared libraries.
1135 @kindex --check-sections
1136 @kindex --no-check-sections
1137 @item --check-sections
1138 @itemx --no-check-sections
1139 Asks the linker @emph{not} to check section addresses after they have
1140 been assigned to see if there are any overlaps. Normally the linker will
1141 perform this check, and if it finds any overlaps it will produce
1142 suitable error messages. The linker does know about, and does make
1143 allowances for sections in overlays. The default behaviour can be
1144 restored by using the command line switch @option{--check-sections}.
1146 @cindex cross reference table
1149 Output a cross reference table. If a linker map file is being
1150 generated, the cross reference table is printed to the map file.
1151 Otherwise, it is printed on the standard output.
1153 The format of the table is intentionally simple, so that it may be
1154 easily processed by a script if necessary. The symbols are printed out,
1155 sorted by name. For each symbol, a list of file names is given. If the
1156 symbol is defined, the first file listed is the location of the
1157 definition. The remaining files contain references to the symbol.
1159 @cindex common allocation
1160 @kindex --no-define-common
1161 @item --no-define-common
1162 This option inhibits the assignment of addresses to common symbols.
1163 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1164 @xref{Miscellaneous Commands}.
1166 The @samp{--no-define-common} option allows decoupling
1167 the decision to assign addresses to Common symbols from the choice
1168 of the output file type; otherwise a non-Relocatable output type
1169 forces assigning addresses to Common symbols.
1170 Using @samp{--no-define-common} allows Common symbols that are referenced
1171 from a shared library to be assigned addresses only in the main program.
1172 This eliminates the unused duplicate space in the shared library,
1173 and also prevents any possible confusion over resolving to the wrong
1174 duplicate when there are many dynamic modules with specialized search
1175 paths for runtime symbol resolution.
1177 @cindex symbols, from command line
1178 @kindex --defsym @var{symbol}=@var{exp}
1179 @item --defsym @var{symbol}=@var{expression}
1180 Create a global symbol in the output file, containing the absolute
1181 address given by @var{expression}. You may use this option as many
1182 times as necessary to define multiple symbols in the command line. A
1183 limited form of arithmetic is supported for the @var{expression} in this
1184 context: you may give a hexadecimal constant or the name of an existing
1185 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1186 constants or symbols. If you need more elaborate expressions, consider
1187 using the linker command language from a script (@pxref{Assignments,,
1188 Assignment: Symbol Definitions}). @emph{Note:} there should be no white
1189 space between @var{symbol}, the equals sign (``@key{=}''), and
1192 @cindex demangling, from command line
1193 @kindex --demangle[=@var{style}]
1194 @kindex --no-demangle
1195 @item --demangle[=@var{style}]
1196 @itemx --no-demangle
1197 These options control whether to demangle symbol names in error messages
1198 and other output. When the linker is told to demangle, it tries to
1199 present symbol names in a readable fashion: it strips leading
1200 underscores if they are used by the object file format, and converts C++
1201 mangled symbol names into user readable names. Different compilers have
1202 different mangling styles. The optional demangling style argument can be used
1203 to choose an appropriate demangling style for your compiler. The linker will
1204 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1205 is set. These options may be used to override the default.
1207 @cindex dynamic linker, from command line
1208 @kindex -I@var{file}
1209 @kindex --dynamic-linker @var{file}
1210 @item --dynamic-linker @var{file}
1211 Set the name of the dynamic linker. This is only meaningful when
1212 generating dynamically linked ELF executables. The default dynamic
1213 linker is normally correct; don't use this unless you know what you are
1217 @kindex --fatal-warnings
1218 @item --fatal-warnings
1219 Treat all warnings as errors.
1221 @kindex --force-exe-suffix
1222 @item --force-exe-suffix
1223 Make sure that an output file has a .exe suffix.
1225 If a successfully built fully linked output file does not have a
1226 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1227 the output file to one of the same name with a @code{.exe} suffix. This
1228 option is useful when using unmodified Unix makefiles on a Microsoft
1229 Windows host, since some versions of Windows won't run an image unless
1230 it ends in a @code{.exe} suffix.
1232 @kindex --gc-sections
1233 @kindex --no-gc-sections
1234 @cindex garbage collection
1235 @item --no-gc-sections
1236 @itemx --gc-sections
1237 Enable garbage collection of unused input sections. It is ignored on
1238 targets that do not support this option. This option is not compatible
1239 with @samp{-r}. The default behaviour (of not performing this garbage
1240 collection) can be restored by specifying @samp{--no-gc-sections} on
1247 Print a summary of the command-line options on the standard output and exit.
1249 @kindex --target-help
1251 Print a summary of all target specific options on the standard output and exit.
1254 @item -Map @var{mapfile}
1255 Print a link map to the file @var{mapfile}. See the description of the
1256 @option{-M} option, above.
1258 @cindex memory usage
1259 @kindex --no-keep-memory
1260 @item --no-keep-memory
1261 @command{ld} normally optimizes for speed over memory usage by caching the
1262 symbol tables of input files in memory. This option tells @command{ld} to
1263 instead optimize for memory usage, by rereading the symbol tables as
1264 necessary. This may be required if @command{ld} runs out of memory space
1265 while linking a large executable.
1267 @kindex --no-undefined
1269 @item --no-undefined
1271 Report unresolved symbol references from regular object files. This
1272 is done even if the linker is creating a non-symbolic shared library.
1273 The switch @option{--[no-]allow-shlib-undefined} controls the
1274 behaviour for reporting unresolved references found in shared
1275 libraries being linked in.
1277 @kindex --allow-multiple-definition
1279 @item --allow-multiple-definition
1281 Normally when a symbol is defined multiple times, the linker will
1282 report a fatal error. These options allow multiple definitions and the
1283 first definition will be used.
1285 @kindex --allow-shlib-undefined
1286 @kindex --no-allow-shlib-undefined
1287 @item --allow-shlib-undefined
1288 @itemx --no-allow-shlib-undefined
1289 Allows (the default) or disallows undefined symbols in shared libraries.
1290 This switch is similar to @option{--no-undefined} except that it
1291 determines the behaviour when the undefined symbols are in a
1292 shared library rather than a regular object file. It does not affect
1293 how undefined symbols in regular object files are handled.
1295 The reason that @option{--allow-shlib-undefined} is the default is that
1296 the shared library being specified at link time may not be the same as
1297 the one that is available at load time, so the symbols might actually be
1298 resolvable at load time. Plus there are some systems, (eg BeOS) where
1299 undefined symbols in shared libraries is normal. (The kernel patches
1300 them at load time to select which function is most appropriate
1301 for the current architecture. This is used for example to dynamically
1302 select an appropriate memset function). Apparently it is also normal
1303 for HPPA shared libraries to have undefined symbols.
1305 @kindex --no-undefined-version
1306 @item --no-undefined-version
1307 Normally when a symbol has an undefined version, the linker will ignore
1308 it. This option disallows symbols with undefined version and a fatal error
1309 will be issued instead.
1311 @kindex --default-symver
1312 @item --default-symver
1313 Create and use a default symbol version (the soname) for unversioned
1316 @kindex --default-imported-symver
1317 @item --default-imported-symver
1318 Create and use a default symbol version (the soname) for unversioned
1321 @kindex --no-warn-mismatch
1322 @item --no-warn-mismatch
1323 Normally @command{ld} will give an error if you try to link together input
1324 files that are mismatched for some reason, perhaps because they have
1325 been compiled for different processors or for different endiannesses.
1326 This option tells @command{ld} that it should silently permit such possible
1327 errors. This option should only be used with care, in cases when you
1328 have taken some special action that ensures that the linker errors are
1331 @kindex --no-whole-archive
1332 @item --no-whole-archive
1333 Turn off the effect of the @option{--whole-archive} option for subsequent
1336 @cindex output file after errors
1337 @kindex --noinhibit-exec
1338 @item --noinhibit-exec
1339 Retain the executable output file whenever it is still usable.
1340 Normally, the linker will not produce an output file if it encounters
1341 errors during the link process; it exits without writing an output file
1342 when it issues any error whatsoever.
1346 Only search library directories explicitly specified on the
1347 command line. Library directories specified in linker scripts
1348 (including linker scripts specified on the command line) are ignored.
1350 @ifclear SingleFormat
1352 @item --oformat @var{output-format}
1353 @command{ld} may be configured to support more than one kind of object
1354 file. If your @command{ld} is configured this way, you can use the
1355 @samp{--oformat} option to specify the binary format for the output
1356 object file. Even when @command{ld} is configured to support alternative
1357 object formats, you don't usually need to specify this, as @command{ld}
1358 should be configured to produce as a default output format the most
1359 usual format on each machine. @var{output-format} is a text string, the
1360 name of a particular format supported by the BFD libraries. (You can
1361 list the available binary formats with @samp{objdump -i}.) The script
1362 command @code{OUTPUT_FORMAT} can also specify the output format, but
1363 this option overrides it. @xref{BFD}.
1367 @kindex --pic-executable
1369 @itemx --pic-executable
1370 @cindex position independent executables
1371 Create a position independent executable. This is currently only supported on
1372 ELF platforms. Position independent executables are similar to shared
1373 libraries in that they are relocated by the dynamic linker to the virtual
1374 address the OS chooses for them (which can vary between invocations). Like
1375 normal dynamically linked executables they can be executed and symbols
1376 defined in the executable cannot be overridden by shared libraries.
1380 This option is ignored for Linux compatibility.
1384 This option is ignored for SVR4 compatibility.
1387 @cindex synthesizing linker
1388 @cindex relaxing addressing modes
1390 An option with machine dependent effects.
1392 This option is only supported on a few targets.
1395 @xref{H8/300,,@command{ld} and the H8/300}.
1398 @xref{i960,, @command{ld} and the Intel 960 family}.
1401 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1404 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1407 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1410 On some platforms, the @samp{--relax} option performs global
1411 optimizations that become possible when the linker resolves addressing
1412 in the program, such as relaxing address modes and synthesizing new
1413 instructions in the output object file.
1415 On some platforms these link time global optimizations may make symbolic
1416 debugging of the resulting executable impossible.
1419 the case for the Matsushita MN10200 and MN10300 family of processors.
1423 On platforms where this is not supported, @samp{--relax} is accepted,
1427 @cindex retaining specified symbols
1428 @cindex stripping all but some symbols
1429 @cindex symbols, retaining selectively
1430 @item --retain-symbols-file @var{filename}
1431 Retain @emph{only} the symbols listed in the file @var{filename},
1432 discarding all others. @var{filename} is simply a flat file, with one
1433 symbol name per line. This option is especially useful in environments
1437 where a large global symbol table is accumulated gradually, to conserve
1440 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1441 or symbols needed for relocations.
1443 You may only specify @samp{--retain-symbols-file} once in the command
1444 line. It overrides @samp{-s} and @samp{-S}.
1447 @item -rpath @var{dir}
1448 @cindex runtime library search path
1450 Add a directory to the runtime library search path. This is used when
1451 linking an ELF executable with shared objects. All @option{-rpath}
1452 arguments are concatenated and passed to the runtime linker, which uses
1453 them to locate shared objects at runtime. The @option{-rpath} option is
1454 also used when locating shared objects which are needed by shared
1455 objects explicitly included in the link; see the description of the
1456 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1457 ELF executable, the contents of the environment variable
1458 @code{LD_RUN_PATH} will be used if it is defined.
1460 The @option{-rpath} option may also be used on SunOS. By default, on
1461 SunOS, the linker will form a runtime search patch out of all the
1462 @option{-L} options it is given. If a @option{-rpath} option is used, the
1463 runtime search path will be formed exclusively using the @option{-rpath}
1464 options, ignoring the @option{-L} options. This can be useful when using
1465 gcc, which adds many @option{-L} options which may be on NFS mounted
1468 For compatibility with other ELF linkers, if the @option{-R} option is
1469 followed by a directory name, rather than a file name, it is treated as
1470 the @option{-rpath} option.
1474 @cindex link-time runtime library search path
1476 @item -rpath-link @var{DIR}
1477 When using ELF or SunOS, one shared library may require another. This
1478 happens when an @code{ld -shared} link includes a shared library as one
1481 When the linker encounters such a dependency when doing a non-shared,
1482 non-relocatable link, it will automatically try to locate the required
1483 shared library and include it in the link, if it is not included
1484 explicitly. In such a case, the @option{-rpath-link} option
1485 specifies the first set of directories to search. The
1486 @option{-rpath-link} option may specify a sequence of directory names
1487 either by specifying a list of names separated by colons, or by
1488 appearing multiple times.
1490 This option should be used with caution as it overrides the search path
1491 that may have been hard compiled into a shared library. In such a case it
1492 is possible to use unintentionally a different search path than the
1493 runtime linker would do.
1495 The linker uses the following search paths to locate required shared
1499 Any directories specified by @option{-rpath-link} options.
1501 Any directories specified by @option{-rpath} options. The difference
1502 between @option{-rpath} and @option{-rpath-link} is that directories
1503 specified by @option{-rpath} options are included in the executable and
1504 used at runtime, whereas the @option{-rpath-link} option is only effective
1505 at link time. It is for the native linker only.
1507 On an ELF system, if the @option{-rpath} and @code{rpath-link} options
1508 were not used, search the contents of the environment variable
1509 @code{LD_RUN_PATH}. It is for the native linker only.
1511 On SunOS, if the @option{-rpath} option was not used, search any
1512 directories specified using @option{-L} options.
1514 For a native linker, the contents of the environment variable
1515 @code{LD_LIBRARY_PATH}.
1517 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1518 @code{DT_RPATH} of a shared library are searched for shared
1519 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1520 @code{DT_RUNPATH} entries exist.
1522 The default directories, normally @file{/lib} and @file{/usr/lib}.
1524 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1525 exists, the list of directories found in that file.
1528 If the required shared library is not found, the linker will issue a
1529 warning and continue with the link.
1536 @cindex shared libraries
1537 Create a shared library. This is currently only supported on ELF, XCOFF
1538 and SunOS platforms. On SunOS, the linker will automatically create a
1539 shared library if the @option{-e} option is not used and there are
1540 undefined symbols in the link.
1543 @kindex --sort-common
1544 This option tells @command{ld} to sort the common symbols by size when it
1545 places them in the appropriate output sections. First come all the one
1546 byte symbols, then all the two byte, then all the four byte, and then
1547 everything else. This is to prevent gaps between symbols due to
1548 alignment constraints.
1550 @kindex --sort-section name
1551 @item --sort-section name
1552 This option will apply @code{SORT_BY_NAME} to all wildcard section
1553 patterns in the linker script.
1555 @kindex --sort-section alignment
1556 @item --sort-section alignment
1557 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1558 patterns in the linker script.
1560 @kindex --split-by-file
1561 @item --split-by-file [@var{size}]
1562 Similar to @option{--split-by-reloc} but creates a new output section for
1563 each input file when @var{size} is reached. @var{size} defaults to a
1564 size of 1 if not given.
1566 @kindex --split-by-reloc
1567 @item --split-by-reloc [@var{count}]
1568 Tries to creates extra sections in the output file so that no single
1569 output section in the file contains more than @var{count} relocations.
1570 This is useful when generating huge relocatable files for downloading into
1571 certain real time kernels with the COFF object file format; since COFF
1572 cannot represent more than 65535 relocations in a single section. Note
1573 that this will fail to work with object file formats which do not
1574 support arbitrary sections. The linker will not split up individual
1575 input sections for redistribution, so if a single input section contains
1576 more than @var{count} relocations one output section will contain that
1577 many relocations. @var{count} defaults to a value of 32768.
1581 Compute and display statistics about the operation of the linker, such
1582 as execution time and memory usage.
1585 @item --sysroot=@var{directory}
1586 Use @var{directory} as the location of the sysroot, overriding the
1587 configure-time default. This option is only supported by linkers
1588 that were configured using @option{--with-sysroot}.
1590 @kindex --traditional-format
1591 @cindex traditional format
1592 @item --traditional-format
1593 For some targets, the output of @command{ld} is different in some ways from
1594 the output of some existing linker. This switch requests @command{ld} to
1595 use the traditional format instead.
1598 For example, on SunOS, @command{ld} combines duplicate entries in the
1599 symbol string table. This can reduce the size of an output file with
1600 full debugging information by over 30 percent. Unfortunately, the SunOS
1601 @code{dbx} program can not read the resulting program (@code{gdb} has no
1602 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1603 combine duplicate entries.
1605 @kindex --section-start @var{sectionname}=@var{org}
1606 @item --section-start @var{sectionname}=@var{org}
1607 Locate a section in the output file at the absolute
1608 address given by @var{org}. You may use this option as many
1609 times as necessary to locate multiple sections in the command
1611 @var{org} must be a single hexadecimal integer;
1612 for compatibility with other linkers, you may omit the leading
1613 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1614 should be no white space between @var{sectionname}, the equals
1615 sign (``@key{=}''), and @var{org}.
1617 @kindex -Tbss @var{org}
1618 @kindex -Tdata @var{org}
1619 @kindex -Ttext @var{org}
1620 @cindex segment origins, cmd line
1621 @item -Tbss @var{org}
1622 @itemx -Tdata @var{org}
1623 @itemx -Ttext @var{org}
1624 Same as --section-start, with @code{.bss}, @code{.data} or
1625 @code{.text} as the @var{sectionname}.
1627 @kindex --unresolved-symbols
1628 @item --unresolved-symbols=@var{method}
1629 Determine how to handle unresolved symbols. There are four possible
1630 values for @samp{method}:
1634 Do not report any unresolved symbols.
1637 Report all unresolved symbols. This is the default.
1639 @item ignore-in-object-files
1640 Report unresolved symbols that are contained in shared libraries, but
1641 ignore them if they come from regular object files.
1643 @item ignore-in-shared-libs
1644 Report unresolved symbols that come from regular object files, but
1645 ignore them if they come from shared libraries. This can be useful
1646 when creating a dynamic binary and it is known that all the shared
1647 libraries that it should be referencing are included on the linker's
1651 The behaviour for shared libraries on their own can also be controlled
1652 by the @option{--[no-]allow-shlib-undefined} option.
1654 Normally the linker will generate an error message for each reported
1655 unresolved symbol but the option @option{--warn-unresolved-symbols}
1656 can change this to a warning.
1662 Display the version number for @command{ld} and list the linker emulations
1663 supported. Display which input files can and cannot be opened. Display
1664 the linker script being used by the linker.
1666 @kindex --version-script=@var{version-scriptfile}
1667 @cindex version script, symbol versions
1668 @itemx --version-script=@var{version-scriptfile}
1669 Specify the name of a version script to the linker. This is typically
1670 used when creating shared libraries to specify additional information
1671 about the version hierarchy for the library being created. This option
1672 is only meaningful on ELF platforms which support shared libraries.
1675 @kindex --warn-common
1676 @cindex warnings, on combining symbols
1677 @cindex combining symbols, warnings on
1679 Warn when a common symbol is combined with another common symbol or with
1680 a symbol definition. Unix linkers allow this somewhat sloppy practise,
1681 but linkers on some other operating systems do not. This option allows
1682 you to find potential problems from combining global symbols.
1683 Unfortunately, some C libraries use this practise, so you may get some
1684 warnings about symbols in the libraries as well as in your programs.
1686 There are three kinds of global symbols, illustrated here by C examples:
1690 A definition, which goes in the initialized data section of the output
1694 An undefined reference, which does not allocate space.
1695 There must be either a definition or a common symbol for the
1699 A common symbol. If there are only (one or more) common symbols for a
1700 variable, it goes in the uninitialized data area of the output file.
1701 The linker merges multiple common symbols for the same variable into a
1702 single symbol. If they are of different sizes, it picks the largest
1703 size. The linker turns a common symbol into a declaration, if there is
1704 a definition of the same variable.
1707 The @samp{--warn-common} option can produce five kinds of warnings.
1708 Each warning consists of a pair of lines: the first describes the symbol
1709 just encountered, and the second describes the previous symbol
1710 encountered with the same name. One or both of the two symbols will be
1715 Turning a common symbol into a reference, because there is already a
1716 definition for the symbol.
1718 @var{file}(@var{section}): warning: common of `@var{symbol}'
1719 overridden by definition
1720 @var{file}(@var{section}): warning: defined here
1724 Turning a common symbol into a reference, because a later definition for
1725 the symbol is encountered. This is the same as the previous case,
1726 except that the symbols are encountered in a different order.
1728 @var{file}(@var{section}): warning: definition of `@var{symbol}'
1730 @var{file}(@var{section}): warning: common is here
1734 Merging a common symbol with a previous same-sized common symbol.
1736 @var{file}(@var{section}): warning: multiple common
1738 @var{file}(@var{section}): warning: previous common is here
1742 Merging a common symbol with a previous larger common symbol.
1744 @var{file}(@var{section}): warning: common of `@var{symbol}'
1745 overridden by larger common
1746 @var{file}(@var{section}): warning: larger common is here
1750 Merging a common symbol with a previous smaller common symbol. This is
1751 the same as the previous case, except that the symbols are
1752 encountered in a different order.
1754 @var{file}(@var{section}): warning: common of `@var{symbol}'
1755 overriding smaller common
1756 @var{file}(@var{section}): warning: smaller common is here
1760 @kindex --warn-constructors
1761 @item --warn-constructors
1762 Warn if any global constructors are used. This is only useful for a few
1763 object file formats. For formats like COFF or ELF, the linker can not
1764 detect the use of global constructors.
1766 @kindex --warn-multiple-gp
1767 @item --warn-multiple-gp
1768 Warn if multiple global pointer values are required in the output file.
1769 This is only meaningful for certain processors, such as the Alpha.
1770 Specifically, some processors put large-valued constants in a special
1771 section. A special register (the global pointer) points into the middle
1772 of this section, so that constants can be loaded efficiently via a
1773 base-register relative addressing mode. Since the offset in
1774 base-register relative mode is fixed and relatively small (e.g., 16
1775 bits), this limits the maximum size of the constant pool. Thus, in
1776 large programs, it is often necessary to use multiple global pointer
1777 values in order to be able to address all possible constants. This
1778 option causes a warning to be issued whenever this case occurs.
1781 @cindex warnings, on undefined symbols
1782 @cindex undefined symbols, warnings on
1784 Only warn once for each undefined symbol, rather than once per module
1787 @kindex --warn-section-align
1788 @cindex warnings, on section alignment
1789 @cindex section alignment, warnings on
1790 @item --warn-section-align
1791 Warn if the address of an output section is changed because of
1792 alignment. Typically, the alignment will be set by an input section.
1793 The address will only be changed if it not explicitly specified; that
1794 is, if the @code{SECTIONS} command does not specify a start address for
1795 the section (@pxref{SECTIONS}).
1797 @kindex --warn-shared-textrel
1798 @item --warn-shared-textrel
1799 Warn if the linker adds a DT_TEXTREL to a shared object.
1801 @kindex --warn-unresolved-symbols
1802 @item --warn-unresolved-symbols
1803 If the linker is going to report an unresolved symbol (see the option
1804 @option{--unresolved-symbols}) it will normally generate an error.
1805 This option makes it generate a warning instead.
1807 @kindex --error-unresolved-symbols
1808 @item --error-unresolved-symbols
1809 This restores the linker's default behaviour of generating errors when
1810 it is reporting unresolved symbols.
1812 @kindex --whole-archive
1813 @cindex including an entire archive
1814 @item --whole-archive
1815 For each archive mentioned on the command line after the
1816 @option{--whole-archive} option, include every object file in the archive
1817 in the link, rather than searching the archive for the required object
1818 files. This is normally used to turn an archive file into a shared
1819 library, forcing every object to be included in the resulting shared
1820 library. This option may be used more than once.
1822 Two notes when using this option from gcc: First, gcc doesn't know
1823 about this option, so you have to use @option{-Wl,-whole-archive}.
1824 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1825 list of archives, because gcc will add its own list of archives to
1826 your link and you may not want this flag to affect those as well.
1829 @item --wrap @var{symbol}
1830 Use a wrapper function for @var{symbol}. Any undefined reference to
1831 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
1832 undefined reference to @code{__real_@var{symbol}} will be resolved to
1835 This can be used to provide a wrapper for a system function. The
1836 wrapper function should be called @code{__wrap_@var{symbol}}. If it
1837 wishes to call the system function, it should call
1838 @code{__real_@var{symbol}}.
1840 Here is a trivial example:
1844 __wrap_malloc (size_t c)
1846 printf ("malloc called with %zu\n", c);
1847 return __real_malloc (c);
1851 If you link other code with this file using @option{--wrap malloc}, then
1852 all calls to @code{malloc} will call the function @code{__wrap_malloc}
1853 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
1854 call the real @code{malloc} function.
1856 You may wish to provide a @code{__real_malloc} function as well, so that
1857 links without the @option{--wrap} option will succeed. If you do this,
1858 you should not put the definition of @code{__real_malloc} in the same
1859 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
1860 call before the linker has a chance to wrap it to @code{malloc}.
1862 @kindex --eh-frame-hdr
1863 @item --eh-frame-hdr
1864 Request creation of @code{.eh_frame_hdr} section and ELF
1865 @code{PT_GNU_EH_FRAME} segment header.
1867 @kindex --enable-new-dtags
1868 @kindex --disable-new-dtags
1869 @item --enable-new-dtags
1870 @itemx --disable-new-dtags
1871 This linker can create the new dynamic tags in ELF. But the older ELF
1872 systems may not understand them. If you specify
1873 @option{--enable-new-dtags}, the dynamic tags will be created as needed.
1874 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
1875 created. By default, the new dynamic tags are not created. Note that
1876 those options are only available for ELF systems.
1878 @kindex --hash-size=@var{number}
1879 @item --hash-size=@var{number}
1880 Set the default size of the linker's hash tables to a prime number
1881 close to @var{number}. Increasing this value can reduce the length of
1882 time it takes the linker to perform its tasks, at the expense of
1883 increasing the linker's memory requirements. Similarly reducing this
1884 value can reduce the memory requirements at the expense of speed.
1886 @kindex --reduce-memory-overheads
1887 @item --reduce-memory-overheads
1888 This option reduces memory requirements at ld runtime, at the expense of
1889 linking speed. This was introduced to select the old O(n^2) algorithm
1890 for link map file generation, rather than the new O(n) algorithm which uses
1891 about 40% more memory for symbol storage.
1893 Another effect of the switch is to set the default hash table size to
1894 1021, which again saves memory at the cost of lengthening the linker's
1895 run time. This is not done however if the @option{--hash-size} switch
1898 The @option{--reduce-memory-overheads} switch may be also be used to
1899 enable other tradeoffs in future versions of the linker.
1905 @subsection Options Specific to i386 PE Targets
1907 @c man begin OPTIONS
1909 The i386 PE linker supports the @option{-shared} option, which causes
1910 the output to be a dynamically linked library (DLL) instead of a
1911 normal executable. You should name the output @code{*.dll} when you
1912 use this option. In addition, the linker fully supports the standard
1913 @code{*.def} files, which may be specified on the linker command line
1914 like an object file (in fact, it should precede archives it exports
1915 symbols from, to ensure that they get linked in, just like a normal
1918 In addition to the options common to all targets, the i386 PE linker
1919 support additional command line options that are specific to the i386
1920 PE target. Options that take values may be separated from their
1921 values by either a space or an equals sign.
1925 @kindex --add-stdcall-alias
1926 @item --add-stdcall-alias
1927 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
1928 as-is and also with the suffix stripped.
1929 [This option is specific to the i386 PE targeted port of the linker]
1932 @item --base-file @var{file}
1933 Use @var{file} as the name of a file in which to save the base
1934 addresses of all the relocations needed for generating DLLs with
1936 [This is an i386 PE specific option]
1940 Create a DLL instead of a regular executable. You may also use
1941 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
1943 [This option is specific to the i386 PE targeted port of the linker]
1945 @kindex --enable-stdcall-fixup
1946 @kindex --disable-stdcall-fixup
1947 @item --enable-stdcall-fixup
1948 @itemx --disable-stdcall-fixup
1949 If the link finds a symbol that it cannot resolve, it will attempt to
1950 do ``fuzzy linking'' by looking for another defined symbol that differs
1951 only in the format of the symbol name (cdecl vs stdcall) and will
1952 resolve that symbol by linking to the match. For example, the
1953 undefined symbol @code{_foo} might be linked to the function
1954 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
1955 to the function @code{_bar}. When the linker does this, it prints a
1956 warning, since it normally should have failed to link, but sometimes
1957 import libraries generated from third-party dlls may need this feature
1958 to be usable. If you specify @option{--enable-stdcall-fixup}, this
1959 feature is fully enabled and warnings are not printed. If you specify
1960 @option{--disable-stdcall-fixup}, this feature is disabled and such
1961 mismatches are considered to be errors.
1962 [This option is specific to the i386 PE targeted port of the linker]
1964 @cindex DLLs, creating
1965 @kindex --export-all-symbols
1966 @item --export-all-symbols
1967 If given, all global symbols in the objects used to build a DLL will
1968 be exported by the DLL. Note that this is the default if there
1969 otherwise wouldn't be any exported symbols. When symbols are
1970 explicitly exported via DEF files or implicitly exported via function
1971 attributes, the default is to not export anything else unless this
1972 option is given. Note that the symbols @code{DllMain@@12},
1973 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
1974 @code{impure_ptr} will not be automatically
1975 exported. Also, symbols imported from other DLLs will not be
1976 re-exported, nor will symbols specifying the DLL's internal layout
1977 such as those beginning with @code{_head_} or ending with
1978 @code{_iname}. In addition, no symbols from @code{libgcc},
1979 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
1980 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
1981 not be exported, to help with C++ DLLs. Finally, there is an
1982 extensive list of cygwin-private symbols that are not exported
1983 (obviously, this applies on when building DLLs for cygwin targets).
1984 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
1985 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
1986 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
1987 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
1988 @code{cygwin_premain3}, and @code{environ}.
1989 [This option is specific to the i386 PE targeted port of the linker]
1991 @kindex --exclude-symbols
1992 @item --exclude-symbols @var{symbol},@var{symbol},...
1993 Specifies a list of symbols which should not be automatically
1994 exported. The symbol names may be delimited by commas or colons.
1995 [This option is specific to the i386 PE targeted port of the linker]
1997 @kindex --file-alignment
1998 @item --file-alignment
1999 Specify the file alignment. Sections in the file will always begin at
2000 file offsets which are multiples of this number. This defaults to
2002 [This option is specific to the i386 PE targeted port of the linker]
2006 @item --heap @var{reserve}
2007 @itemx --heap @var{reserve},@var{commit}
2008 Specify the amount of memory to reserve (and optionally commit) to be
2009 used as heap for this program. The default is 1Mb reserved, 4K
2011 [This option is specific to the i386 PE targeted port of the linker]
2014 @kindex --image-base
2015 @item --image-base @var{value}
2016 Use @var{value} as the base address of your program or dll. This is
2017 the lowest memory location that will be used when your program or dll
2018 is loaded. To reduce the need to relocate and improve performance of
2019 your dlls, each should have a unique base address and not overlap any
2020 other dlls. The default is 0x400000 for executables, and 0x10000000
2022 [This option is specific to the i386 PE targeted port of the linker]
2026 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2027 symbols before they are exported.
2028 [This option is specific to the i386 PE targeted port of the linker]
2030 @kindex --large-address-aware
2031 @item --large-address-aware
2032 If given, the appropriate bit in the ``Charateristics'' field of the COFF
2033 header is set to indicate that this executable supports virtual addresses
2034 greater than 2 gigabytes. This should be used in conjuction with the /3GB
2035 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2036 section of the BOOT.INI. Otherwise, this bit has no effect.
2037 [This option is specific to PE targeted ports of the linker]
2039 @kindex --major-image-version
2040 @item --major-image-version @var{value}
2041 Sets the major number of the ``image version''. Defaults to 1.
2042 [This option is specific to the i386 PE targeted port of the linker]
2044 @kindex --major-os-version
2045 @item --major-os-version @var{value}
2046 Sets the major number of the ``os version''. Defaults to 4.
2047 [This option is specific to the i386 PE targeted port of the linker]
2049 @kindex --major-subsystem-version
2050 @item --major-subsystem-version @var{value}
2051 Sets the major number of the ``subsystem version''. Defaults to 4.
2052 [This option is specific to the i386 PE targeted port of the linker]
2054 @kindex --minor-image-version
2055 @item --minor-image-version @var{value}
2056 Sets the minor number of the ``image version''. Defaults to 0.
2057 [This option is specific to the i386 PE targeted port of the linker]
2059 @kindex --minor-os-version
2060 @item --minor-os-version @var{value}
2061 Sets the minor number of the ``os version''. Defaults to 0.
2062 [This option is specific to the i386 PE targeted port of the linker]
2064 @kindex --minor-subsystem-version
2065 @item --minor-subsystem-version @var{value}
2066 Sets the minor number of the ``subsystem version''. Defaults to 0.
2067 [This option is specific to the i386 PE targeted port of the linker]
2069 @cindex DEF files, creating
2070 @cindex DLLs, creating
2071 @kindex --output-def
2072 @item --output-def @var{file}
2073 The linker will create the file @var{file} which will contain a DEF
2074 file corresponding to the DLL the linker is generating. This DEF file
2075 (which should be called @code{*.def}) may be used to create an import
2076 library with @code{dlltool} or may be used as a reference to
2077 automatically or implicitly exported symbols.
2078 [This option is specific to the i386 PE targeted port of the linker]
2080 @cindex DLLs, creating
2081 @kindex --out-implib
2082 @item --out-implib @var{file}
2083 The linker will create the file @var{file} which will contain an
2084 import lib corresponding to the DLL the linker is generating. This
2085 import lib (which should be called @code{*.dll.a} or @code{*.a}
2086 may be used to link clients against the generated DLL; this behaviour
2087 makes it possible to skip a separate @code{dlltool} import library
2089 [This option is specific to the i386 PE targeted port of the linker]
2091 @kindex --enable-auto-image-base
2092 @item --enable-auto-image-base
2093 Automatically choose the image base for DLLs, unless one is specified
2094 using the @code{--image-base} argument. By using a hash generated
2095 from the dllname to create unique image bases for each DLL, in-memory
2096 collisions and relocations which can delay program execution are
2098 [This option is specific to the i386 PE targeted port of the linker]
2100 @kindex --disable-auto-image-base
2101 @item --disable-auto-image-base
2102 Do not automatically generate a unique image base. If there is no
2103 user-specified image base (@code{--image-base}) then use the platform
2105 [This option is specific to the i386 PE targeted port of the linker]
2107 @cindex DLLs, linking to
2108 @kindex --dll-search-prefix
2109 @item --dll-search-prefix @var{string}
2110 When linking dynamically to a dll without an import library,
2111 search for @code{<string><basename>.dll} in preference to
2112 @code{lib<basename>.dll}. This behaviour allows easy distinction
2113 between DLLs built for the various "subplatforms": native, cygwin,
2114 uwin, pw, etc. For instance, cygwin DLLs typically use
2115 @code{--dll-search-prefix=cyg}.
2116 [This option is specific to the i386 PE targeted port of the linker]
2118 @kindex --enable-auto-import
2119 @item --enable-auto-import
2120 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2121 DATA imports from DLLs, and create the necessary thunking symbols when
2122 building the import libraries with those DATA exports. Note: Use of the
2123 'auto-import' extension will cause the text section of the image file
2124 to be made writable. This does not conform to the PE-COFF format
2125 specification published by Microsoft.
2127 Using 'auto-import' generally will 'just work' -- but sometimes you may
2130 "variable '<var>' can't be auto-imported. Please read the
2131 documentation for ld's @code{--enable-auto-import} for details."
2133 This message occurs when some (sub)expression accesses an address
2134 ultimately given by the sum of two constants (Win32 import tables only
2135 allow one). Instances where this may occur include accesses to member
2136 fields of struct variables imported from a DLL, as well as using a
2137 constant index into an array variable imported from a DLL. Any
2138 multiword variable (arrays, structs, long long, etc) may trigger
2139 this error condition. However, regardless of the exact data type
2140 of the offending exported variable, ld will always detect it, issue
2141 the warning, and exit.
2143 There are several ways to address this difficulty, regardless of the
2144 data type of the exported variable:
2146 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2147 of adjusting references in your client code for runtime environment, so
2148 this method works only when runtime environment supports this feature.
2150 A second solution is to force one of the 'constants' to be a variable --
2151 that is, unknown and un-optimizable at compile time. For arrays,
2152 there are two possibilities: a) make the indexee (the array's address)
2153 a variable, or b) make the 'constant' index a variable. Thus:
2156 extern type extern_array[];
2158 @{ volatile type *t=extern_array; t[1] @}
2164 extern type extern_array[];
2166 @{ volatile int t=1; extern_array[t] @}
2169 For structs (and most other multiword data types) the only option
2170 is to make the struct itself (or the long long, or the ...) variable:
2173 extern struct s extern_struct;
2174 extern_struct.field -->
2175 @{ volatile struct s *t=&extern_struct; t->field @}
2181 extern long long extern_ll;
2183 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2186 A third method of dealing with this difficulty is to abandon
2187 'auto-import' for the offending symbol and mark it with
2188 @code{__declspec(dllimport)}. However, in practise that
2189 requires using compile-time #defines to indicate whether you are
2190 building a DLL, building client code that will link to the DLL, or
2191 merely building/linking to a static library. In making the choice
2192 between the various methods of resolving the 'direct address with
2193 constant offset' problem, you should consider typical real-world usage:
2201 void main(int argc, char **argv)@{
2202 printf("%d\n",arr[1]);
2212 void main(int argc, char **argv)@{
2213 /* This workaround is for win32 and cygwin; do not "optimize" */
2214 volatile int *parr = arr;
2215 printf("%d\n",parr[1]);
2222 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2223 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2224 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2225 #define FOO_IMPORT __declspec(dllimport)
2229 extern FOO_IMPORT int arr[];
2232 void main(int argc, char **argv)@{
2233 printf("%d\n",arr[1]);
2237 A fourth way to avoid this problem is to re-code your
2238 library to use a functional interface rather than a data interface
2239 for the offending variables (e.g. set_foo() and get_foo() accessor
2241 [This option is specific to the i386 PE targeted port of the linker]
2243 @kindex --disable-auto-import
2244 @item --disable-auto-import
2245 Do not attempt to do sophisticated linking of @code{_symbol} to
2246 @code{__imp__symbol} for DATA imports from DLLs.
2247 [This option is specific to the i386 PE targeted port of the linker]
2249 @kindex --enable-runtime-pseudo-reloc
2250 @item --enable-runtime-pseudo-reloc
2251 If your code contains expressions described in --enable-auto-import section,
2252 that is, DATA imports from DLL with non-zero offset, this switch will create
2253 a vector of 'runtime pseudo relocations' which can be used by runtime
2254 environment to adjust references to such data in your client code.
2255 [This option is specific to the i386 PE targeted port of the linker]
2257 @kindex --disable-runtime-pseudo-reloc
2258 @item --disable-runtime-pseudo-reloc
2259 Do not create pseudo relocations for non-zero offset DATA imports from
2260 DLLs. This is the default.
2261 [This option is specific to the i386 PE targeted port of the linker]
2263 @kindex --enable-extra-pe-debug
2264 @item --enable-extra-pe-debug
2265 Show additional debug info related to auto-import symbol thunking.
2266 [This option is specific to the i386 PE targeted port of the linker]
2268 @kindex --section-alignment
2269 @item --section-alignment
2270 Sets the section alignment. Sections in memory will always begin at
2271 addresses which are a multiple of this number. Defaults to 0x1000.
2272 [This option is specific to the i386 PE targeted port of the linker]
2276 @item --stack @var{reserve}
2277 @itemx --stack @var{reserve},@var{commit}
2278 Specify the amount of memory to reserve (and optionally commit) to be
2279 used as stack for this program. The default is 2Mb reserved, 4K
2281 [This option is specific to the i386 PE targeted port of the linker]
2284 @item --subsystem @var{which}
2285 @itemx --subsystem @var{which}:@var{major}
2286 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2287 Specifies the subsystem under which your program will execute. The
2288 legal values for @var{which} are @code{native}, @code{windows},
2289 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2290 the subsystem version also. Numeric values are also accepted for
2292 [This option is specific to the i386 PE targeted port of the linker]
2299 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2301 @c man begin OPTIONS
2303 The 68HC11 and 68HC12 linkers support specific options to control the
2304 memory bank switching mapping and trampoline code generation.
2308 @kindex --no-trampoline
2309 @item --no-trampoline
2310 This option disables the generation of trampoline. By default a trampoline
2311 is generated for each far function which is called using a @code{jsr}
2312 instruction (this happens when a pointer to a far function is taken).
2314 @kindex --bank-window
2315 @item --bank-window @var{name}
2316 This option indicates to the linker the name of the memory region in
2317 the @samp{MEMORY} specification that describes the memory bank window.
2318 The definition of such region is then used by the linker to compute
2319 paging and addresses within the memory window.
2328 @section Environment Variables
2330 @c man begin ENVIRONMENT
2332 You can change the behaviour of @command{ld} with the environment variables
2333 @ifclear SingleFormat
2336 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2338 @ifclear SingleFormat
2340 @cindex default input format
2341 @code{GNUTARGET} determines the input-file object format if you don't
2342 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2343 of the BFD names for an input format (@pxref{BFD}). If there is no
2344 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2345 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2346 attempts to discover the input format by examining binary input files;
2347 this method often succeeds, but there are potential ambiguities, since
2348 there is no method of ensuring that the magic number used to specify
2349 object-file formats is unique. However, the configuration procedure for
2350 BFD on each system places the conventional format for that system first
2351 in the search-list, so ambiguities are resolved in favor of convention.
2355 @cindex default emulation
2356 @cindex emulation, default
2357 @code{LDEMULATION} determines the default emulation if you don't use the
2358 @samp{-m} option. The emulation can affect various aspects of linker
2359 behaviour, particularly the default linker script. You can list the
2360 available emulations with the @samp{--verbose} or @samp{-V} options. If
2361 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2362 variable is not defined, the default emulation depends upon how the
2363 linker was configured.
2365 @kindex COLLECT_NO_DEMANGLE
2366 @cindex demangling, default
2367 Normally, the linker will default to demangling symbols. However, if
2368 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2369 default to not demangling symbols. This environment variable is used in
2370 a similar fashion by the @code{gcc} linker wrapper program. The default
2371 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2378 @chapter Linker Scripts
2381 @cindex linker scripts
2382 @cindex command files
2383 Every link is controlled by a @dfn{linker script}. This script is
2384 written in the linker command language.
2386 The main purpose of the linker script is to describe how the sections in
2387 the input files should be mapped into the output file, and to control
2388 the memory layout of the output file. Most linker scripts do nothing
2389 more than this. However, when necessary, the linker script can also
2390 direct the linker to perform many other operations, using the commands
2393 The linker always uses a linker script. If you do not supply one
2394 yourself, the linker will use a default script that is compiled into the
2395 linker executable. You can use the @samp{--verbose} command line option
2396 to display the default linker script. Certain command line options,
2397 such as @samp{-r} or @samp{-N}, will affect the default linker script.
2399 You may supply your own linker script by using the @samp{-T} command
2400 line option. When you do this, your linker script will replace the
2401 default linker script.
2403 You may also use linker scripts implicitly by naming them as input files
2404 to the linker, as though they were files to be linked. @xref{Implicit
2408 * Basic Script Concepts:: Basic Linker Script Concepts
2409 * Script Format:: Linker Script Format
2410 * Simple Example:: Simple Linker Script Example
2411 * Simple Commands:: Simple Linker Script Commands
2412 * Assignments:: Assigning Values to Symbols
2413 * SECTIONS:: SECTIONS Command
2414 * MEMORY:: MEMORY Command
2415 * PHDRS:: PHDRS Command
2416 * VERSION:: VERSION Command
2417 * Expressions:: Expressions in Linker Scripts
2418 * Implicit Linker Scripts:: Implicit Linker Scripts
2421 @node Basic Script Concepts
2422 @section Basic Linker Script Concepts
2423 @cindex linker script concepts
2424 We need to define some basic concepts and vocabulary in order to
2425 describe the linker script language.
2427 The linker combines input files into a single output file. The output
2428 file and each input file are in a special data format known as an
2429 @dfn{object file format}. Each file is called an @dfn{object file}.
2430 The output file is often called an @dfn{executable}, but for our
2431 purposes we will also call it an object file. Each object file has,
2432 among other things, a list of @dfn{sections}. We sometimes refer to a
2433 section in an input file as an @dfn{input section}; similarly, a section
2434 in the output file is an @dfn{output section}.
2436 Each section in an object file has a name and a size. Most sections
2437 also have an associated block of data, known as the @dfn{section
2438 contents}. A section may be marked as @dfn{loadable}, which mean that
2439 the contents should be loaded into memory when the output file is run.
2440 A section with no contents may be @dfn{allocatable}, which means that an
2441 area in memory should be set aside, but nothing in particular should be
2442 loaded there (in some cases this memory must be zeroed out). A section
2443 which is neither loadable nor allocatable typically contains some sort
2444 of debugging information.
2446 Every loadable or allocatable output section has two addresses. The
2447 first is the @dfn{VMA}, or virtual memory address. This is the address
2448 the section will have when the output file is run. The second is the
2449 @dfn{LMA}, or load memory address. This is the address at which the
2450 section will be loaded. In most cases the two addresses will be the
2451 same. An example of when they might be different is when a data section
2452 is loaded into ROM, and then copied into RAM when the program starts up
2453 (this technique is often used to initialize global variables in a ROM
2454 based system). In this case the ROM address would be the LMA, and the
2455 RAM address would be the VMA.
2457 You can see the sections in an object file by using the @code{objdump}
2458 program with the @samp{-h} option.
2460 Every object file also has a list of @dfn{symbols}, known as the
2461 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
2462 has a name, and each defined symbol has an address, among other
2463 information. If you compile a C or C++ program into an object file, you
2464 will get a defined symbol for every defined function and global or
2465 static variable. Every undefined function or global variable which is
2466 referenced in the input file will become an undefined symbol.
2468 You can see the symbols in an object file by using the @code{nm}
2469 program, or by using the @code{objdump} program with the @samp{-t}
2473 @section Linker Script Format
2474 @cindex linker script format
2475 Linker scripts are text files.
2477 You write a linker script as a series of commands. Each command is
2478 either a keyword, possibly followed by arguments, or an assignment to a
2479 symbol. You may separate commands using semicolons. Whitespace is
2482 Strings such as file or format names can normally be entered directly.
2483 If the file name contains a character such as a comma which would
2484 otherwise serve to separate file names, you may put the file name in
2485 double quotes. There is no way to use a double quote character in a
2488 You may include comments in linker scripts just as in C, delimited by
2489 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
2492 @node Simple Example
2493 @section Simple Linker Script Example
2494 @cindex linker script example
2495 @cindex example of linker script
2496 Many linker scripts are fairly simple.
2498 The simplest possible linker script has just one command:
2499 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
2500 memory layout of the output file.
2502 The @samp{SECTIONS} command is a powerful command. Here we will
2503 describe a simple use of it. Let's assume your program consists only of
2504 code, initialized data, and uninitialized data. These will be in the
2505 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2506 Let's assume further that these are the only sections which appear in
2509 For this example, let's say that the code should be loaded at address
2510 0x10000, and that the data should start at address 0x8000000. Here is a
2511 linker script which will do that:
2516 .text : @{ *(.text) @}
2518 .data : @{ *(.data) @}
2519 .bss : @{ *(.bss) @}
2523 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2524 followed by a series of symbol assignments and output section
2525 descriptions enclosed in curly braces.
2527 The first line inside the @samp{SECTIONS} command of the above example
2528 sets the value of the special symbol @samp{.}, which is the location
2529 counter. If you do not specify the address of an output section in some
2530 other way (other ways are described later), the address is set from the
2531 current value of the location counter. The location counter is then
2532 incremented by the size of the output section. At the start of the
2533 @samp{SECTIONS} command, the location counter has the value @samp{0}.
2535 The second line defines an output section, @samp{.text}. The colon is
2536 required syntax which may be ignored for now. Within the curly braces
2537 after the output section name, you list the names of the input sections
2538 which should be placed into this output section. The @samp{*} is a
2539 wildcard which matches any file name. The expression @samp{*(.text)}
2540 means all @samp{.text} input sections in all input files.
2542 Since the location counter is @samp{0x10000} when the output section
2543 @samp{.text} is defined, the linker will set the address of the
2544 @samp{.text} section in the output file to be @samp{0x10000}.
2546 The remaining lines define the @samp{.data} and @samp{.bss} sections in
2547 the output file. The linker will place the @samp{.data} output section
2548 at address @samp{0x8000000}. After the linker places the @samp{.data}
2549 output section, the value of the location counter will be
2550 @samp{0x8000000} plus the size of the @samp{.data} output section. The
2551 effect is that the linker will place the @samp{.bss} output section
2552 immediately after the @samp{.data} output section in memory.
2554 The linker will ensure that each output section has the required
2555 alignment, by increasing the location counter if necessary. In this
2556 example, the specified addresses for the @samp{.text} and @samp{.data}
2557 sections will probably satisfy any alignment constraints, but the linker
2558 may have to create a small gap between the @samp{.data} and @samp{.bss}
2561 That's it! That's a simple and complete linker script.
2563 @node Simple Commands
2564 @section Simple Linker Script Commands
2565 @cindex linker script simple commands
2566 In this section we describe the simple linker script commands.
2569 * Entry Point:: Setting the entry point
2570 * File Commands:: Commands dealing with files
2571 @ifclear SingleFormat
2572 * Format Commands:: Commands dealing with object file formats
2575 * Miscellaneous Commands:: Other linker script commands
2579 @subsection Setting the Entry Point
2580 @kindex ENTRY(@var{symbol})
2581 @cindex start of execution
2582 @cindex first instruction
2584 The first instruction to execute in a program is called the @dfn{entry
2585 point}. You can use the @code{ENTRY} linker script command to set the
2586 entry point. The argument is a symbol name:
2591 There are several ways to set the entry point. The linker will set the
2592 entry point by trying each of the following methods in order, and
2593 stopping when one of them succeeds:
2596 the @samp{-e} @var{entry} command-line option;
2598 the @code{ENTRY(@var{symbol})} command in a linker script;
2600 the value of the symbol @code{start}, if defined;
2602 the address of the first byte of the @samp{.text} section, if present;
2604 The address @code{0}.
2608 @subsection Commands Dealing with Files
2609 @cindex linker script file commands
2610 Several linker script commands deal with files.
2613 @item INCLUDE @var{filename}
2614 @kindex INCLUDE @var{filename}
2615 @cindex including a linker script
2616 Include the linker script @var{filename} at this point. The file will
2617 be searched for in the current directory, and in any directory specified
2618 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
2621 @item INPUT(@var{file}, @var{file}, @dots{})
2622 @itemx INPUT(@var{file} @var{file} @dots{})
2623 @kindex INPUT(@var{files})
2624 @cindex input files in linker scripts
2625 @cindex input object files in linker scripts
2626 @cindex linker script input object files
2627 The @code{INPUT} command directs the linker to include the named files
2628 in the link, as though they were named on the command line.
2630 For example, if you always want to include @file{subr.o} any time you do
2631 a link, but you can't be bothered to put it on every link command line,
2632 then you can put @samp{INPUT (subr.o)} in your linker script.
2634 In fact, if you like, you can list all of your input files in the linker
2635 script, and then invoke the linker with nothing but a @samp{-T} option.
2637 In case a @dfn{sysroot prefix} is configured, and the filename starts
2638 with the @samp{/} character, and the script being processed was
2639 located inside the @dfn{sysroot prefix}, the filename will be looked
2640 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
2641 open the file in the current directory. If it is not found, the
2642 linker will search through the archive library search path. See the
2643 description of @samp{-L} in @ref{Options,,Command Line Options}.
2645 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2646 name to @code{lib@var{file}.a}, as with the command line argument
2649 When you use the @code{INPUT} command in an implicit linker script, the
2650 files will be included in the link at the point at which the linker
2651 script file is included. This can affect archive searching.
2653 @item GROUP(@var{file}, @var{file}, @dots{})
2654 @itemx GROUP(@var{file} @var{file} @dots{})
2655 @kindex GROUP(@var{files})
2656 @cindex grouping input files
2657 The @code{GROUP} command is like @code{INPUT}, except that the named
2658 files should all be archives, and they are searched repeatedly until no
2659 new undefined references are created. See the description of @samp{-(}
2660 in @ref{Options,,Command Line Options}.
2662 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
2663 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
2664 @kindex AS_NEEDED(@var{files})
2665 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2666 commands, among other filenames. The files listed will be handled
2667 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2668 with the exception of ELF shared libraries, that will be added only
2669 when they are actually needed. This construct essentially enables
2670 @option{--as-needed} option for all the files listed inside of it
2671 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2674 @item OUTPUT(@var{filename})
2675 @kindex OUTPUT(@var{filename})
2676 @cindex output file name in linker scripot
2677 The @code{OUTPUT} command names the output file. Using
2678 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
2679 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2680 Line Options}). If both are used, the command line option takes
2683 You can use the @code{OUTPUT} command to define a default name for the
2684 output file other than the usual default of @file{a.out}.
2686 @item SEARCH_DIR(@var{path})
2687 @kindex SEARCH_DIR(@var{path})
2688 @cindex library search path in linker script
2689 @cindex archive search path in linker script
2690 @cindex search path in linker script
2691 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2692 @command{ld} looks for archive libraries. Using
2693 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2694 on the command line (@pxref{Options,,Command Line Options}). If both
2695 are used, then the linker will search both paths. Paths specified using
2696 the command line option are searched first.
2698 @item STARTUP(@var{filename})
2699 @kindex STARTUP(@var{filename})
2700 @cindex first input file
2701 The @code{STARTUP} command is just like the @code{INPUT} command, except
2702 that @var{filename} will become the first input file to be linked, as
2703 though it were specified first on the command line. This may be useful
2704 when using a system in which the entry point is always the start of the
2708 @ifclear SingleFormat
2709 @node Format Commands
2710 @subsection Commands Dealing with Object File Formats
2711 A couple of linker script commands deal with object file formats.
2714 @item OUTPUT_FORMAT(@var{bfdname})
2715 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2716 @kindex OUTPUT_FORMAT(@var{bfdname})
2717 @cindex output file format in linker script
2718 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2719 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
2720 exactly like using @samp{--oformat @var{bfdname}} on the command line
2721 (@pxref{Options,,Command Line Options}). If both are used, the command
2722 line option takes precedence.
2724 You can use @code{OUTPUT_FORMAT} with three arguments to use different
2725 formats based on the @samp{-EB} and @samp{-EL} command line options.
2726 This permits the linker script to set the output format based on the
2729 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
2730 will be the first argument, @var{default}. If @samp{-EB} is used, the
2731 output format will be the second argument, @var{big}. If @samp{-EL} is
2732 used, the output format will be the third argument, @var{little}.
2734 For example, the default linker script for the MIPS ELF target uses this
2737 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
2739 This says that the default format for the output file is
2740 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
2741 option, the output file will be created in the @samp{elf32-littlemips}
2744 @item TARGET(@var{bfdname})
2745 @kindex TARGET(@var{bfdname})
2746 @cindex input file format in linker script
2747 The @code{TARGET} command names the BFD format to use when reading input
2748 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
2749 This command is like using @samp{-b @var{bfdname}} on the command line
2750 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
2751 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
2752 command is also used to set the format for the output file. @xref{BFD}.
2756 @node Miscellaneous Commands
2757 @subsection Other Linker Script Commands
2758 There are a few other linker scripts commands.
2761 @item ASSERT(@var{exp}, @var{message})
2763 @cindex assertion in linker script
2764 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
2765 with an error code, and print @var{message}.
2767 @item EXTERN(@var{symbol} @var{symbol} @dots{})
2769 @cindex undefined symbol in linker script
2770 Force @var{symbol} to be entered in the output file as an undefined
2771 symbol. Doing this may, for example, trigger linking of additional
2772 modules from standard libraries. You may list several @var{symbol}s for
2773 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
2774 command has the same effect as the @samp{-u} command-line option.
2776 @item FORCE_COMMON_ALLOCATION
2777 @kindex FORCE_COMMON_ALLOCATION
2778 @cindex common allocation in linker script
2779 This command has the same effect as the @samp{-d} command-line option:
2780 to make @command{ld} assign space to common symbols even if a relocatable
2781 output file is specified (@samp{-r}).
2783 @item INHIBIT_COMMON_ALLOCATION
2784 @kindex INHIBIT_COMMON_ALLOCATION
2785 @cindex common allocation in linker script
2786 This command has the same effect as the @samp{--no-define-common}
2787 command-line option: to make @code{ld} omit the assignment of addresses
2788 to common symbols even for a non-relocatable output file.
2790 @item NOCROSSREFS(@var{section} @var{section} @dots{})
2791 @kindex NOCROSSREFS(@var{sections})
2792 @cindex cross references
2793 This command may be used to tell @command{ld} to issue an error about any
2794 references among certain output sections.
2796 In certain types of programs, particularly on embedded systems when
2797 using overlays, when one section is loaded into memory, another section
2798 will not be. Any direct references between the two sections would be
2799 errors. For example, it would be an error if code in one section called
2800 a function defined in the other section.
2802 The @code{NOCROSSREFS} command takes a list of output section names. If
2803 @command{ld} detects any cross references between the sections, it reports
2804 an error and returns a non-zero exit status. Note that the
2805 @code{NOCROSSREFS} command uses output section names, not input section
2808 @ifclear SingleFormat
2809 @item OUTPUT_ARCH(@var{bfdarch})
2810 @kindex OUTPUT_ARCH(@var{bfdarch})
2811 @cindex machine architecture
2812 @cindex architecture
2813 Specify a particular output machine architecture. The argument is one
2814 of the names used by the BFD library (@pxref{BFD}). You can see the
2815 architecture of an object file by using the @code{objdump} program with
2816 the @samp{-f} option.
2821 @section Assigning Values to Symbols
2822 @cindex assignment in scripts
2823 @cindex symbol definition, scripts
2824 @cindex variables, defining
2825 You may assign a value to a symbol in a linker script. This will define
2826 the symbol and place it into the symbol table with a global scope.
2829 * Simple Assignments:: Simple Assignments
2831 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
2832 * Source Code Reference:: How to use a linker script defined symbol in source code
2835 @node Simple Assignments
2836 @subsection Simple Assignments
2838 You may assign to a symbol using any of the C assignment operators:
2841 @item @var{symbol} = @var{expression} ;
2842 @itemx @var{symbol} += @var{expression} ;
2843 @itemx @var{symbol} -= @var{expression} ;
2844 @itemx @var{symbol} *= @var{expression} ;
2845 @itemx @var{symbol} /= @var{expression} ;
2846 @itemx @var{symbol} <<= @var{expression} ;
2847 @itemx @var{symbol} >>= @var{expression} ;
2848 @itemx @var{symbol} &= @var{expression} ;
2849 @itemx @var{symbol} |= @var{expression} ;
2852 The first case will define @var{symbol} to the value of
2853 @var{expression}. In the other cases, @var{symbol} must already be
2854 defined, and the value will be adjusted accordingly.
2856 The special symbol name @samp{.} indicates the location counter. You
2857 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
2859 The semicolon after @var{expression} is required.
2861 Expressions are defined below; see @ref{Expressions}.
2863 You may write symbol assignments as commands in their own right, or as
2864 statements within a @code{SECTIONS} command, or as part of an output
2865 section description in a @code{SECTIONS} command.
2867 The section of the symbol will be set from the section of the
2868 expression; for more information, see @ref{Expression Section}.
2870 Here is an example showing the three different places that symbol
2871 assignments may be used:
2882 _bdata = (. + 3) & ~ 3;
2883 .data : @{ *(.data) @}
2887 In this example, the symbol @samp{floating_point} will be defined as
2888 zero. The symbol @samp{_etext} will be defined as the address following
2889 the last @samp{.text} input section. The symbol @samp{_bdata} will be
2890 defined as the address following the @samp{.text} output section aligned
2891 upward to a 4 byte boundary.
2896 In some cases, it is desirable for a linker script to define a symbol
2897 only if it is referenced and is not defined by any object included in
2898 the link. For example, traditional linkers defined the symbol
2899 @samp{etext}. However, ANSI C requires that the user be able to use
2900 @samp{etext} as a function name without encountering an error. The
2901 @code{PROVIDE} keyword may be used to define a symbol, such as
2902 @samp{etext}, only if it is referenced but not defined. The syntax is
2903 @code{PROVIDE(@var{symbol} = @var{expression})}.
2905 Here is an example of using @code{PROVIDE} to define @samp{etext}:
2918 In this example, if the program defines @samp{_etext} (with a leading
2919 underscore), the linker will give a multiple definition error. If, on
2920 the other hand, the program defines @samp{etext} (with no leading
2921 underscore), the linker will silently use the definition in the program.
2922 If the program references @samp{etext} but does not define it, the
2923 linker will use the definition in the linker script.
2925 @node PROVIDE_HIDDEN
2926 @subsection PROVIDE_HIDDEN
2927 @cindex PROVIDE_HIDDEN
2928 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
2929 hidden and won't be exported.
2931 @node Source Code Reference
2932 @subsection Source Code Reference
2934 Accessing a linker script defined variable from source code is not
2935 intuitive. In particular a linker script symbol is not equivalent to
2936 a variable declaration in a high level language, it is instead a
2937 symbol that does not have a value.
2939 Before going further, it is important to note that compilers often
2940 transform names in the source code into different names when they are
2941 stored in the symbol table. For example, Fortran compilers commonly
2942 prepend or append an underscore, and C++ performs extensive @samp{name
2943 mangling}. Therefore there might be a discrepancy between the name
2944 of a variable as it is used in source code and the name of the same
2945 variable as it is defined in a linker script. For example in C a
2946 linker script variable might be referred to as:
2952 But in the linker script it might be defined as:
2958 In the remaining examples however it is assumed that no name
2959 transformation has taken place.
2961 When a symbol is declared in a high level language such as C, two
2962 things happen. The first is that the compiler reserves enough space
2963 in the program's memory to hold the @emph{value} of the symbol. The
2964 second is that the compiler creates an entry in the program's symbol
2965 table which holds the symbol's @emph{address}. ie the symbol table
2966 contains the address of the block of memory holding the symbol's
2967 value. So for example the following C declaration, at file scope:
2973 creates a entry called @samp{foo} in the symbol table. This entry
2974 holds the address of an @samp{int} sized block of memory where the
2975 number 1000 is initially stored.
2977 When a program references a symbol the compiler generates code that
2978 first accesses the symbol table to find the address of the symbol's
2979 memory block and then code to read the value from that memory block.
2986 looks up the symbol @samp{foo} in the symbol table, gets the address
2987 associated with this symbol and then writes the value 1 into that
2994 looks up the symbol @samp{foo} in the symbol table, gets it address
2995 and then copies this address into the block of memory associated with
2996 the variable @samp{a}.
2998 Linker scripts symbol declarations, by contrast, create an entry in
2999 the symbol table but do not assign any memory to them. Thus they are
3000 an address without a value. So for example the linker script definition:
3006 creates an entry in the symbol table called @samp{foo} which holds
3007 the address of memory location 1000, but nothing special is stored at
3008 address 1000. This means that you cannot access the @emph{value} of a
3009 linker script defined symbol - it has no value - all you can do is
3010 access the @emph{address} of a linker script defined symbol.
3012 Hence when you are using a linker script defined symbol in source code
3013 you should always take the address of the symbol, and never attempt to
3014 use its value. For example suppose you want to copy the contents of a
3015 section of memory called .ROM into a section called .FLASH and the
3016 linker script contains these declarations:
3020 start_of_ROM = .ROM;
3021 end_of_ROM = .ROM + sizeof (.ROM) - 1;
3022 start_of_FLASH = .FLASH;
3026 Then the C source code to perform the copy would be:
3030 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3032 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3036 Note the use of the @samp{&} operators. These are correct.
3039 @section SECTIONS Command
3041 The @code{SECTIONS} command tells the linker how to map input sections
3042 into output sections, and how to place the output sections in memory.
3044 The format of the @code{SECTIONS} command is:
3048 @var{sections-command}
3049 @var{sections-command}
3054 Each @var{sections-command} may of be one of the following:
3058 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3060 a symbol assignment (@pxref{Assignments})
3062 an output section description
3064 an overlay description
3067 The @code{ENTRY} command and symbol assignments are permitted inside the
3068 @code{SECTIONS} command for convenience in using the location counter in
3069 those commands. This can also make the linker script easier to
3070 understand because you can use those commands at meaningful points in
3071 the layout of the output file.
3073 Output section descriptions and overlay descriptions are described
3076 If you do not use a @code{SECTIONS} command in your linker script, the
3077 linker will place each input section into an identically named output
3078 section in the order that the sections are first encountered in the
3079 input files. If all input sections are present in the first file, for
3080 example, the order of sections in the output file will match the order
3081 in the first input file. The first section will be at address zero.
3084 * Output Section Description:: Output section description
3085 * Output Section Name:: Output section name
3086 * Output Section Address:: Output section address
3087 * Input Section:: Input section description
3088 * Output Section Data:: Output section data
3089 * Output Section Keywords:: Output section keywords
3090 * Output Section Discarding:: Output section discarding
3091 * Output Section Attributes:: Output section attributes
3092 * Overlay Description:: Overlay description
3095 @node Output Section Description
3096 @subsection Output Section Description
3097 The full description of an output section looks like this:
3100 @var{section} [@var{address}] [(@var{type})] :
3101 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3103 @var{output-section-command}
3104 @var{output-section-command}
3106 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3110 Most output sections do not use most of the optional section attributes.
3112 The whitespace around @var{section} is required, so that the section
3113 name is unambiguous. The colon and the curly braces are also required.
3114 The line breaks and other white space are optional.
3116 Each @var{output-section-command} may be one of the following:
3120 a symbol assignment (@pxref{Assignments})
3122 an input section description (@pxref{Input Section})
3124 data values to include directly (@pxref{Output Section Data})
3126 a special output section keyword (@pxref{Output Section Keywords})
3129 @node Output Section Name
3130 @subsection Output Section Name
3131 @cindex name, section
3132 @cindex section name
3133 The name of the output section is @var{section}. @var{section} must
3134 meet the constraints of your output format. In formats which only
3135 support a limited number of sections, such as @code{a.out}, the name
3136 must be one of the names supported by the format (@code{a.out}, for
3137 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3138 output format supports any number of sections, but with numbers and not
3139 names (as is the case for Oasys), the name should be supplied as a
3140 quoted numeric string. A section name may consist of any sequence of
3141 characters, but a name which contains any unusual characters such as
3142 commas must be quoted.
3144 The output section name @samp{/DISCARD/} is special; @ref{Output Section
3147 @node Output Section Address
3148 @subsection Output Section Address
3149 @cindex address, section
3150 @cindex section address
3151 The @var{address} is an expression for the VMA (the virtual memory
3152 address) of the output section. If you do not provide @var{address},
3153 the linker will set it based on @var{region} if present, or otherwise
3154 based on the current value of the location counter.
3156 If you provide @var{address}, the address of the output section will be
3157 set to precisely that. If you provide neither @var{address} nor
3158 @var{region}, then the address of the output section will be set to the
3159 current value of the location counter aligned to the alignment
3160 requirements of the output section. The alignment requirement of the
3161 output section is the strictest alignment of any input section contained
3162 within the output section.
3166 .text . : @{ *(.text) @}
3171 .text : @{ *(.text) @}
3174 are subtly different. The first will set the address of the
3175 @samp{.text} output section to the current value of the location
3176 counter. The second will set it to the current value of the location
3177 counter aligned to the strictest alignment of a @samp{.text} input
3180 The @var{address} may be an arbitrary expression; @ref{Expressions}.
3181 For example, if you want to align the section on a 0x10 byte boundary,
3182 so that the lowest four bits of the section address are zero, you could
3183 do something like this:
3185 .text ALIGN(0x10) : @{ *(.text) @}
3188 This works because @code{ALIGN} returns the current location counter
3189 aligned upward to the specified value.
3191 Specifying @var{address} for a section will change the value of the
3195 @subsection Input Section Description
3196 @cindex input sections
3197 @cindex mapping input sections to output sections
3198 The most common output section command is an input section description.
3200 The input section description is the most basic linker script operation.
3201 You use output sections to tell the linker how to lay out your program
3202 in memory. You use input section descriptions to tell the linker how to
3203 map the input files into your memory layout.
3206 * Input Section Basics:: Input section basics
3207 * Input Section Wildcards:: Input section wildcard patterns
3208 * Input Section Common:: Input section for common symbols
3209 * Input Section Keep:: Input section and garbage collection
3210 * Input Section Example:: Input section example
3213 @node Input Section Basics
3214 @subsubsection Input Section Basics
3215 @cindex input section basics
3216 An input section description consists of a file name optionally followed
3217 by a list of section names in parentheses.
3219 The file name and the section name may be wildcard patterns, which we
3220 describe further below (@pxref{Input Section Wildcards}).
3222 The most common input section description is to include all input
3223 sections with a particular name in the output section. For example, to
3224 include all input @samp{.text} sections, you would write:
3229 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
3230 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3231 match all files except the ones specified in the EXCLUDE_FILE list. For
3234 (*(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors))
3236 will cause all .ctors sections from all files except @file{crtend.o} and
3237 @file{otherfile.o} to be included.
3239 There are two ways to include more than one section:
3245 The difference between these is the order in which the @samp{.text} and
3246 @samp{.rdata} input sections will appear in the output section. In the
3247 first example, they will be intermingled, appearing in the same order as
3248 they are found in the linker input. In the second example, all
3249 @samp{.text} input sections will appear first, followed by all
3250 @samp{.rdata} input sections.
3252 You can specify a file name to include sections from a particular file.
3253 You would do this if one or more of your files contain special data that
3254 needs to be at a particular location in memory. For example:
3259 If you use a file name without a list of sections, then all sections in
3260 the input file will be included in the output section. This is not
3261 commonly done, but it may by useful on occasion. For example:
3266 When you use a file name which does not contain any wild card
3267 characters, the linker will first see if you also specified the file
3268 name on the linker command line or in an @code{INPUT} command. If you
3269 did not, the linker will attempt to open the file as an input file, as
3270 though it appeared on the command line. Note that this differs from an
3271 @code{INPUT} command, because the linker will not search for the file in
3272 the archive search path.
3274 @node Input Section Wildcards
3275 @subsubsection Input Section Wildcard Patterns
3276 @cindex input section wildcards
3277 @cindex wildcard file name patterns
3278 @cindex file name wildcard patterns
3279 @cindex section name wildcard patterns
3280 In an input section description, either the file name or the section
3281 name or both may be wildcard patterns.
3283 The file name of @samp{*} seen in many examples is a simple wildcard
3284 pattern for the file name.
3286 The wildcard patterns are like those used by the Unix shell.
3290 matches any number of characters
3292 matches any single character
3294 matches a single instance of any of the @var{chars}; the @samp{-}
3295 character may be used to specify a range of characters, as in
3296 @samp{[a-z]} to match any lower case letter
3298 quotes the following character
3301 When a file name is matched with a wildcard, the wildcard characters
3302 will not match a @samp{/} character (used to separate directory names on
3303 Unix). A pattern consisting of a single @samp{*} character is an
3304 exception; it will always match any file name, whether it contains a
3305 @samp{/} or not. In a section name, the wildcard characters will match
3306 a @samp{/} character.
3308 File name wildcard patterns only match files which are explicitly
3309 specified on the command line or in an @code{INPUT} command. The linker
3310 does not search directories to expand wildcards.
3312 If a file name matches more than one wildcard pattern, or if a file name
3313 appears explicitly and is also matched by a wildcard pattern, the linker
3314 will use the first match in the linker script. For example, this
3315 sequence of input section descriptions is probably in error, because the
3316 @file{data.o} rule will not be used:
3318 .data : @{ *(.data) @}
3319 .data1 : @{ data.o(.data) @}
3322 @cindex SORT_BY_NAME
3323 Normally, the linker will place files and sections matched by wildcards
3324 in the order in which they are seen during the link. You can change
3325 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3326 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
3327 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3328 into ascending order by name before placing them in the output file.
3330 @cindex SORT_BY_ALIGNMENT
3331 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3332 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3333 ascending order by alignment before placing them in the output file.
3336 @code{SORT} is an alias for @code{SORT_BY_NAME}.
3338 When there are nested section sorting commands in linker script, there
3339 can be at most 1 level of nesting for section sorting commands.
3343 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3344 It will sort the input sections by name first, then by alignment if 2
3345 sections have the same name.
3347 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3348 It will sort the input sections by alignment first, then by name if 2
3349 sections have the same alignment.
3351 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3352 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3354 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3355 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3357 All other nested section sorting commands are invalid.
3360 When both command line section sorting option and linker script
3361 section sorting command are used, section sorting command always
3362 takes precedence over the command line option.
3364 If the section sorting command in linker script isn't nested, the
3365 command line option will make the section sorting command to be
3366 treated as nested sorting command.
3370 @code{SORT_BY_NAME} (wildcard section pattern ) with
3371 @option{--sort-sections alignment} is equivalent to
3372 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3374 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3375 @option{--sort-section name} is equivalent to
3376 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3379 If the section sorting command in linker script is nested, the
3380 command line option will be ignored.
3382 If you ever get confused about where input sections are going, use the
3383 @samp{-M} linker option to generate a map file. The map file shows
3384 precisely how input sections are mapped to output sections.
3386 This example shows how wildcard patterns might be used to partition
3387 files. This linker script directs the linker to place all @samp{.text}
3388 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3389 The linker will place the @samp{.data} section from all files beginning
3390 with an upper case character in @samp{.DATA}; for all other files, the
3391 linker will place the @samp{.data} section in @samp{.data}.
3395 .text : @{ *(.text) @}
3396 .DATA : @{ [A-Z]*(.data) @}
3397 .data : @{ *(.data) @}
3398 .bss : @{ *(.bss) @}
3403 @node Input Section Common
3404 @subsubsection Input Section for Common Symbols
3405 @cindex common symbol placement
3406 @cindex uninitialized data placement
3407 A special notation is needed for common symbols, because in many object
3408 file formats common symbols do not have a particular input section. The
3409 linker treats common symbols as though they are in an input section
3410 named @samp{COMMON}.
3412 You may use file names with the @samp{COMMON} section just as with any
3413 other input sections. You can use this to place common symbols from a
3414 particular input file in one section while common symbols from other
3415 input files are placed in another section.
3417 In most cases, common symbols in input files will be placed in the
3418 @samp{.bss} section in the output file. For example:
3420 .bss @{ *(.bss) *(COMMON) @}
3423 @cindex scommon section
3424 @cindex small common symbols
3425 Some object file formats have more than one type of common symbol. For
3426 example, the MIPS ELF object file format distinguishes standard common
3427 symbols and small common symbols. In this case, the linker will use a
3428 different special section name for other types of common symbols. In
3429 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3430 symbols and @samp{.scommon} for small common symbols. This permits you
3431 to map the different types of common symbols into memory at different
3435 You will sometimes see @samp{[COMMON]} in old linker scripts. This
3436 notation is now considered obsolete. It is equivalent to
3439 @node Input Section Keep
3440 @subsubsection Input Section and Garbage Collection
3442 @cindex garbage collection
3443 When link-time garbage collection is in use (@samp{--gc-sections}),
3444 it is often useful to mark sections that should not be eliminated.
3445 This is accomplished by surrounding an input section's wildcard entry
3446 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3447 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3449 @node Input Section Example
3450 @subsubsection Input Section Example
3451 The following example is a complete linker script. It tells the linker
3452 to read all of the sections from file @file{all.o} and place them at the
3453 start of output section @samp{outputa} which starts at location
3454 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
3455 follows immediately, in the same output section. All of section
3456 @samp{.input2} from @file{foo.o} goes into output section
3457 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3458 All of the remaining @samp{.input1} and @samp{.input2} sections from any
3459 files are written to output section @samp{outputc}.
3487 @node Output Section Data
3488 @subsection Output Section Data
3490 @cindex section data
3491 @cindex output section data
3492 @kindex BYTE(@var{expression})
3493 @kindex SHORT(@var{expression})
3494 @kindex LONG(@var{expression})
3495 @kindex QUAD(@var{expression})
3496 @kindex SQUAD(@var{expression})
3497 You can include explicit bytes of data in an output section by using
3498 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3499 an output section command. Each keyword is followed by an expression in
3500 parentheses providing the value to store (@pxref{Expressions}). The
3501 value of the expression is stored at the current value of the location
3504 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3505 store one, two, four, and eight bytes (respectively). After storing the
3506 bytes, the location counter is incremented by the number of bytes
3509 For example, this will store the byte 1 followed by the four byte value
3510 of the symbol @samp{addr}:
3516 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
3517 same; they both store an 8 byte, or 64 bit, value. When both host and
3518 target are 32 bits, an expression is computed as 32 bits. In this case
3519 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
3520 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
3522 If the object file format of the output file has an explicit endianness,
3523 which is the normal case, the value will be stored in that endianness.
3524 When the object file format does not have an explicit endianness, as is
3525 true of, for example, S-records, the value will be stored in the
3526 endianness of the first input object file.
3528 Note---these commands only work inside a section description and not
3529 between them, so the following will produce an error from the linker:
3531 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
3533 whereas this will work:
3535 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
3538 @kindex FILL(@var{expression})
3539 @cindex holes, filling
3540 @cindex unspecified memory
3541 You may use the @code{FILL} command to set the fill pattern for the
3542 current section. It is followed by an expression in parentheses. Any
3543 otherwise unspecified regions of memory within the section (for example,
3544 gaps left due to the required alignment of input sections) are filled
3545 with the value of the expression, repeated as
3546 necessary. A @code{FILL} statement covers memory locations after the
3547 point at which it occurs in the section definition; by including more
3548 than one @code{FILL} statement, you can have different fill patterns in
3549 different parts of an output section.
3551 This example shows how to fill unspecified regions of memory with the
3557 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
3558 section attribute, but it only affects the
3559 part of the section following the @code{FILL} command, rather than the
3560 entire section. If both are used, the @code{FILL} command takes
3561 precedence. @xref{Output Section Fill}, for details on the fill
3564 @node Output Section Keywords
3565 @subsection Output Section Keywords
3566 There are a couple of keywords which can appear as output section
3570 @kindex CREATE_OBJECT_SYMBOLS
3571 @cindex input filename symbols
3572 @cindex filename symbols
3573 @item CREATE_OBJECT_SYMBOLS
3574 The command tells the linker to create a symbol for each input file.
3575 The name of each symbol will be the name of the corresponding input
3576 file. The section of each symbol will be the output section in which
3577 the @code{CREATE_OBJECT_SYMBOLS} command appears.
3579 This is conventional for the a.out object file format. It is not
3580 normally used for any other object file format.
3582 @kindex CONSTRUCTORS
3583 @cindex C++ constructors, arranging in link
3584 @cindex constructors, arranging in link
3586 When linking using the a.out object file format, the linker uses an
3587 unusual set construct to support C++ global constructors and
3588 destructors. When linking object file formats which do not support
3589 arbitrary sections, such as ECOFF and XCOFF, the linker will
3590 automatically recognize C++ global constructors and destructors by name.
3591 For these object file formats, the @code{CONSTRUCTORS} command tells the
3592 linker to place constructor information in the output section where the
3593 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
3594 ignored for other object file formats.
3596 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
3597 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
3598 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
3599 the start and end of the global destructors. The
3600 first word in the list is the number of entries, followed by the address
3601 of each constructor or destructor, followed by a zero word. The
3602 compiler must arrange to actually run the code. For these object file
3603 formats @sc{gnu} C++ normally calls constructors from a subroutine
3604 @code{__main}; a call to @code{__main} is automatically inserted into
3605 the startup code for @code{main}. @sc{gnu} C++ normally runs
3606 destructors either by using @code{atexit}, or directly from the function
3609 For object file formats such as @code{COFF} or @code{ELF} which support
3610 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
3611 addresses of global constructors and destructors into the @code{.ctors}
3612 and @code{.dtors} sections. Placing the following sequence into your
3613 linker script will build the sort of table which the @sc{gnu} C++
3614 runtime code expects to see.
3618 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
3623 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
3629 If you are using the @sc{gnu} C++ support for initialization priority,
3630 which provides some control over the order in which global constructors
3631 are run, you must sort the constructors at link time to ensure that they
3632 are executed in the correct order. When using the @code{CONSTRUCTORS}
3633 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
3634 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
3635 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
3638 Normally the compiler and linker will handle these issues automatically,
3639 and you will not need to concern yourself with them. However, you may
3640 need to consider this if you are using C++ and writing your own linker
3645 @node Output Section Discarding
3646 @subsection Output Section Discarding
3647 @cindex discarding sections
3648 @cindex sections, discarding
3649 @cindex removing sections
3650 The linker will not create output section which do not have any
3651 contents. This is for convenience when referring to input sections that
3652 may or may not be present in any of the input files. For example:
3657 will only create a @samp{.foo} section in the output file if there is a
3658 @samp{.foo} section in at least one input file.
3660 If you use anything other than an input section description as an output
3661 section command, such as a symbol assignment, then the output section
3662 will always be created, even if there are no matching input sections.
3665 The special output section name @samp{/DISCARD/} may be used to discard
3666 input sections. Any input sections which are assigned to an output
3667 section named @samp{/DISCARD/} are not included in the output file.
3669 @node Output Section Attributes
3670 @subsection Output Section Attributes
3671 @cindex output section attributes
3672 We showed above that the full description of an output section looked
3676 @var{section} [@var{address}] [(@var{type})] :
3677 [AT(@var{lma})] [ALIGN(@var{section_align})] [SUBALIGN(@var{subsection_align})]
3679 @var{output-section-command}
3680 @var{output-section-command}
3682 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3685 We've already described @var{section}, @var{address}, and
3686 @var{output-section-command}. In this section we will describe the
3687 remaining section attributes.
3690 * Output Section Type:: Output section type
3691 * Output Section LMA:: Output section LMA
3692 * Forced Output Alignment:: Forced Output Alignment
3693 * Forced Input Alignment:: Forced Input Alignment
3694 * Output Section Region:: Output section region
3695 * Output Section Phdr:: Output section phdr
3696 * Output Section Fill:: Output section fill
3699 @node Output Section Type
3700 @subsubsection Output Section Type
3701 Each output section may have a type. The type is a keyword in
3702 parentheses. The following types are defined:
3706 The section should be marked as not loadable, so that it will not be
3707 loaded into memory when the program is run.
3712 These type names are supported for backward compatibility, and are
3713 rarely used. They all have the same effect: the section should be
3714 marked as not allocatable, so that no memory is allocated for the
3715 section when the program is run.
3719 @cindex prevent unnecessary loading
3720 @cindex loading, preventing
3721 The linker normally sets the attributes of an output section based on
3722 the input sections which map into it. You can override this by using
3723 the section type. For example, in the script sample below, the
3724 @samp{ROM} section is addressed at memory location @samp{0} and does not
3725 need to be loaded when the program is run. The contents of the
3726 @samp{ROM} section will appear in the linker output file as usual.
3730 ROM 0 (NOLOAD) : @{ @dots{} @}
3736 @node Output Section LMA
3737 @subsubsection Output Section LMA
3738 @kindex AT>@var{lma_region}
3739 @kindex AT(@var{lma})
3740 @cindex load address
3741 @cindex section load address
3742 Every section has a virtual address (VMA) and a load address (LMA); see
3743 @ref{Basic Script Concepts}. The address expression which may appear in
3744 an output section description sets the VMA (@pxref{Output Section
3747 The linker will normally set the LMA equal to the VMA. You can change
3748 that by using the @code{AT} keyword. The expression @var{lma} that
3749 follows the @code{AT} keyword specifies the load address of the
3752 Alternatively, with @samp{AT>@var{lma_region}} expression, you may
3753 specify a memory region for the section's load address. @xref{MEMORY}.
3754 Note that if the section has not had a VMA assigned to it then the
3755 linker will use the @var{lma_region} as the VMA region as well.
3756 @xref{Output Section Region}.
3758 @cindex ROM initialized data
3759 @cindex initialized data in ROM
3760 This feature is designed to make it easy to build a ROM image. For
3761 example, the following linker script creates three output sections: one
3762 called @samp{.text}, which starts at @code{0x1000}, one called
3763 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
3764 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
3765 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
3766 defined with the value @code{0x2000}, which shows that the location
3767 counter holds the VMA value, not the LMA value.
3773 .text 0x1000 : @{ *(.text) _etext = . ; @}
3775 AT ( ADDR (.text) + SIZEOF (.text) )
3776 @{ _data = . ; *(.data); _edata = . ; @}
3778 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
3783 The run-time initialization code for use with a program generated with
3784 this linker script would include something like the following, to copy
3785 the initialized data from the ROM image to its runtime address. Notice
3786 how this code takes advantage of the symbols defined by the linker
3791 extern char _etext, _data, _edata, _bstart, _bend;
3792 char *src = &_etext;
3795 /* ROM has data at end of text; copy it. */
3796 while (dst < &_edata) @{
3801 for (dst = &_bstart; dst< &_bend; dst++)
3806 @node Forced Output Alignment
3807 @subsubsection Forced Output Alignment
3808 @kindex ALIGN(@var{section_align})
3809 @cindex forcing output section alignment
3810 @cindex output section alignment
3811 You can increase an output section's alignment by using ALIGN.
3813 @node Forced Input Alignment
3814 @subsubsection Forced Input Alignment
3815 @kindex SUBALIGN(@var{subsection_align})
3816 @cindex forcing input section alignment
3817 @cindex input section alignment
3818 You can force input section alignment within an output section by using
3819 SUBALIGN. The value specified overrides any alignment given by input
3820 sections, whether larger or smaller.
3822 @node Output Section Region
3823 @subsubsection Output Section Region
3824 @kindex >@var{region}
3825 @cindex section, assigning to memory region
3826 @cindex memory regions and sections
3827 You can assign a section to a previously defined region of memory by
3828 using @samp{>@var{region}}. @xref{MEMORY}.
3830 Here is a simple example:
3833 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
3834 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
3838 @node Output Section Phdr
3839 @subsubsection Output Section Phdr
3841 @cindex section, assigning to program header
3842 @cindex program headers and sections
3843 You can assign a section to a previously defined program segment by
3844 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
3845 one or more segments, then all subsequent allocated sections will be
3846 assigned to those segments as well, unless they use an explicitly
3847 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
3848 linker to not put the section in any segment at all.
3850 Here is a simple example:
3853 PHDRS @{ text PT_LOAD ; @}
3854 SECTIONS @{ .text : @{ *(.text) @} :text @}
3858 @node Output Section Fill
3859 @subsubsection Output Section Fill
3860 @kindex =@var{fillexp}
3861 @cindex section fill pattern
3862 @cindex fill pattern, entire section
3863 You can set the fill pattern for an entire section by using
3864 @samp{=@var{fillexp}}. @var{fillexp} is an expression
3865 (@pxref{Expressions}). Any otherwise unspecified regions of memory
3866 within the output section (for example, gaps left due to the required
3867 alignment of input sections) will be filled with the value, repeated as
3868 necessary. If the fill expression is a simple hex number, ie. a string
3869 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
3870 an arbitrarily long sequence of hex digits can be used to specify the
3871 fill pattern; Leading zeros become part of the pattern too. For all
3872 other cases, including extra parentheses or a unary @code{+}, the fill
3873 pattern is the four least significant bytes of the value of the
3874 expression. In all cases, the number is big-endian.
3876 You can also change the fill value with a @code{FILL} command in the
3877 output section commands; (@pxref{Output Section Data}).
3879 Here is a simple example:
3882 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
3886 @node Overlay Description
3887 @subsection Overlay Description
3890 An overlay description provides an easy way to describe sections which
3891 are to be loaded as part of a single memory image but are to be run at
3892 the same memory address. At run time, some sort of overlay manager will
3893 copy the overlaid sections in and out of the runtime memory address as
3894 required, perhaps by simply manipulating addressing bits. This approach
3895 can be useful, for example, when a certain region of memory is faster
3898 Overlays are described using the @code{OVERLAY} command. The
3899 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
3900 output section description. The full syntax of the @code{OVERLAY}
3901 command is as follows:
3904 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
3908 @var{output-section-command}
3909 @var{output-section-command}
3911 @} [:@var{phdr}@dots{}] [=@var{fill}]
3914 @var{output-section-command}
3915 @var{output-section-command}
3917 @} [:@var{phdr}@dots{}] [=@var{fill}]
3919 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
3923 Everything is optional except @code{OVERLAY} (a keyword), and each
3924 section must have a name (@var{secname1} and @var{secname2} above). The
3925 section definitions within the @code{OVERLAY} construct are identical to
3926 those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
3927 except that no addresses and no memory regions may be defined for
3928 sections within an @code{OVERLAY}.
3930 The sections are all defined with the same starting address. The load
3931 addresses of the sections are arranged such that they are consecutive in
3932 memory starting at the load address used for the @code{OVERLAY} as a
3933 whole (as with normal section definitions, the load address is optional,
3934 and defaults to the start address; the start address is also optional,
3935 and defaults to the current value of the location counter).
3937 If the @code{NOCROSSREFS} keyword is used, and there any references
3938 among the sections, the linker will report an error. Since the sections
3939 all run at the same address, it normally does not make sense for one
3940 section to refer directly to another. @xref{Miscellaneous Commands,
3943 For each section within the @code{OVERLAY}, the linker automatically
3944 defines two symbols. The symbol @code{__load_start_@var{secname}} is
3945 defined as the starting load address of the section. The symbol
3946 @code{__load_stop_@var{secname}} is defined as the final load address of
3947 the section. Any characters within @var{secname} which are not legal
3948 within C identifiers are removed. C (or assembler) code may use these
3949 symbols to move the overlaid sections around as necessary.
3951 At the end of the overlay, the value of the location counter is set to
3952 the start address of the overlay plus the size of the largest section.
3954 Here is an example. Remember that this would appear inside a
3955 @code{SECTIONS} construct.
3958 OVERLAY 0x1000 : AT (0x4000)
3960 .text0 @{ o1/*.o(.text) @}
3961 .text1 @{ o2/*.o(.text) @}
3966 This will define both @samp{.text0} and @samp{.text1} to start at
3967 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
3968 @samp{.text1} will be loaded immediately after @samp{.text0}. The
3969 following symbols will be defined: @code{__load_start_text0},
3970 @code{__load_stop_text0}, @code{__load_start_text1},
3971 @code{__load_stop_text1}.
3973 C code to copy overlay @code{.text1} into the overlay area might look
3978 extern char __load_start_text1, __load_stop_text1;
3979 memcpy ((char *) 0x1000, &__load_start_text1,
3980 &__load_stop_text1 - &__load_start_text1);
3984 Note that the @code{OVERLAY} command is just syntactic sugar, since
3985 everything it does can be done using the more basic commands. The above
3986 example could have been written identically as follows.
3990 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
3991 __load_start_text0 = LOADADDR (.text0);
3992 __load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0);
3993 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
3994 __load_start_text1 = LOADADDR (.text1);
3995 __load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1);
3996 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4001 @section MEMORY Command
4003 @cindex memory regions
4004 @cindex regions of memory
4005 @cindex allocating memory
4006 @cindex discontinuous memory
4007 The linker's default configuration permits allocation of all available
4008 memory. You can override this by using the @code{MEMORY} command.
4010 The @code{MEMORY} command describes the location and size of blocks of
4011 memory in the target. You can use it to describe which memory regions
4012 may be used by the linker, and which memory regions it must avoid. You
4013 can then assign sections to particular memory regions. The linker will
4014 set section addresses based on the memory regions, and will warn about
4015 regions that become too full. The linker will not shuffle sections
4016 around to fit into the available regions.
4018 A linker script may contain at most one use of the @code{MEMORY}
4019 command. However, you can define as many blocks of memory within it as
4020 you wish. The syntax is:
4025 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4031 The @var{name} is a name used in the linker script to refer to the
4032 region. The region name has no meaning outside of the linker script.
4033 Region names are stored in a separate name space, and will not conflict
4034 with symbol names, file names, or section names. Each memory region
4035 must have a distinct name.
4037 @cindex memory region attributes
4038 The @var{attr} string is an optional list of attributes that specify
4039 whether to use a particular memory region for an input section which is
4040 not explicitly mapped in the linker script. As described in
4041 @ref{SECTIONS}, if you do not specify an output section for some input
4042 section, the linker will create an output section with the same name as
4043 the input section. If you define region attributes, the linker will use
4044 them to select the memory region for the output section that it creates.
4046 The @var{attr} string must consist only of the following characters:
4061 Invert the sense of any of the preceding attributes
4064 If a unmapped section matches any of the listed attributes other than
4065 @samp{!}, it will be placed in the memory region. The @samp{!}
4066 attribute reverses this test, so that an unmapped section will be placed
4067 in the memory region only if it does not match any of the listed
4073 The @var{origin} is an numerical expression for the start address of
4074 the memory region. The expression must evaluate to a constant and it
4075 cannot involve any symbols. The keyword @code{ORIGIN} may be
4076 abbreviated to @code{org} or @code{o} (but not, for example,
4082 The @var{len} is an expression for the size in bytes of the memory
4083 region. As with the @var{origin} expression, the expression must
4084 be numerical only and must evaluate to a constant. The keyword
4085 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4087 In the following example, we specify that there are two memory regions
4088 available for allocation: one starting at @samp{0} for 256 kilobytes,
4089 and the other starting at @samp{0x40000000} for four megabytes. The
4090 linker will place into the @samp{rom} memory region every section which
4091 is not explicitly mapped into a memory region, and is either read-only
4092 or executable. The linker will place other sections which are not
4093 explicitly mapped into a memory region into the @samp{ram} memory
4100 rom (rx) : ORIGIN = 0, LENGTH = 256K
4101 ram (!rx) : org = 0x40000000, l = 4M
4106 Once you define a memory region, you can direct the linker to place
4107 specific output sections into that memory region by using the
4108 @samp{>@var{region}} output section attribute. For example, if you have
4109 a memory region named @samp{mem}, you would use @samp{>mem} in the
4110 output section definition. @xref{Output Section Region}. If no address
4111 was specified for the output section, the linker will set the address to
4112 the next available address within the memory region. If the combined
4113 output sections directed to a memory region are too large for the
4114 region, the linker will issue an error message.
4116 It is possible to access the origin and length of a memory in an
4117 expression via the @code{ORIGIN(@var{memory})} and
4118 @code{LENGTH(@var{memory})} functions:
4122 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4127 @section PHDRS Command
4129 @cindex program headers
4130 @cindex ELF program headers
4131 @cindex program segments
4132 @cindex segments, ELF
4133 The ELF object file format uses @dfn{program headers}, also knows as
4134 @dfn{segments}. The program headers describe how the program should be
4135 loaded into memory. You can print them out by using the @code{objdump}
4136 program with the @samp{-p} option.
4138 When you run an ELF program on a native ELF system, the system loader
4139 reads the program headers in order to figure out how to load the
4140 program. This will only work if the program headers are set correctly.
4141 This manual does not describe the details of how the system loader
4142 interprets program headers; for more information, see the ELF ABI.
4144 The linker will create reasonable program headers by default. However,
4145 in some cases, you may need to specify the program headers more
4146 precisely. You may use the @code{PHDRS} command for this purpose. When
4147 the linker sees the @code{PHDRS} command in the linker script, it will
4148 not create any program headers other than the ones specified.
4150 The linker only pays attention to the @code{PHDRS} command when
4151 generating an ELF output file. In other cases, the linker will simply
4152 ignore @code{PHDRS}.
4154 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
4155 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4161 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4162 [ FLAGS ( @var{flags} ) ] ;
4167 The @var{name} is used only for reference in the @code{SECTIONS} command
4168 of the linker script. It is not put into the output file. Program
4169 header names are stored in a separate name space, and will not conflict
4170 with symbol names, file names, or section names. Each program header
4171 must have a distinct name.
4173 Certain program header types describe segments of memory which the
4174 system loader will load from the file. In the linker script, you
4175 specify the contents of these segments by placing allocatable output
4176 sections in the segments. You use the @samp{:@var{phdr}} output section
4177 attribute to place a section in a particular segment. @xref{Output
4180 It is normal to put certain sections in more than one segment. This
4181 merely implies that one segment of memory contains another. You may
4182 repeat @samp{:@var{phdr}}, using it once for each segment which should
4183 contain the section.
4185 If you place a section in one or more segments using @samp{:@var{phdr}},
4186 then the linker will place all subsequent allocatable sections which do
4187 not specify @samp{:@var{phdr}} in the same segments. This is for
4188 convenience, since generally a whole set of contiguous sections will be
4189 placed in a single segment. You can use @code{:NONE} to override the
4190 default segment and tell the linker to not put the section in any
4195 You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4196 the program header type to further describe the contents of the segment.
4197 The @code{FILEHDR} keyword means that the segment should include the ELF
4198 file header. The @code{PHDRS} keyword means that the segment should
4199 include the ELF program headers themselves.
4201 The @var{type} may be one of the following. The numbers indicate the
4202 value of the keyword.
4205 @item @code{PT_NULL} (0)
4206 Indicates an unused program header.
4208 @item @code{PT_LOAD} (1)
4209 Indicates that this program header describes a segment to be loaded from
4212 @item @code{PT_DYNAMIC} (2)
4213 Indicates a segment where dynamic linking information can be found.
4215 @item @code{PT_INTERP} (3)
4216 Indicates a segment where the name of the program interpreter may be
4219 @item @code{PT_NOTE} (4)
4220 Indicates a segment holding note information.
4222 @item @code{PT_SHLIB} (5)
4223 A reserved program header type, defined but not specified by the ELF
4226 @item @code{PT_PHDR} (6)
4227 Indicates a segment where the program headers may be found.
4229 @item @var{expression}
4230 An expression giving the numeric type of the program header. This may
4231 be used for types not defined above.
4234 You can specify that a segment should be loaded at a particular address
4235 in memory by using an @code{AT} expression. This is identical to the
4236 @code{AT} command used as an output section attribute (@pxref{Output
4237 Section LMA}). The @code{AT} command for a program header overrides the
4238 output section attribute.
4240 The linker will normally set the segment flags based on the sections
4241 which comprise the segment. You may use the @code{FLAGS} keyword to
4242 explicitly specify the segment flags. The value of @var{flags} must be
4243 an integer. It is used to set the @code{p_flags} field of the program
4246 Here is an example of @code{PHDRS}. This shows a typical set of program
4247 headers used on a native ELF system.
4253 headers PT_PHDR PHDRS ;
4255 text PT_LOAD FILEHDR PHDRS ;
4257 dynamic PT_DYNAMIC ;
4263 .interp : @{ *(.interp) @} :text :interp
4264 .text : @{ *(.text) @} :text
4265 .rodata : @{ *(.rodata) @} /* defaults to :text */
4267 . = . + 0x1000; /* move to a new page in memory */
4268 .data : @{ *(.data) @} :data
4269 .dynamic : @{ *(.dynamic) @} :data :dynamic
4276 @section VERSION Command
4277 @kindex VERSION @{script text@}
4278 @cindex symbol versions
4279 @cindex version script
4280 @cindex versions of symbols
4281 The linker supports symbol versions when using ELF. Symbol versions are
4282 only useful when using shared libraries. The dynamic linker can use
4283 symbol versions to select a specific version of a function when it runs
4284 a program that may have been linked against an earlier version of the
4287 You can include a version script directly in the main linker script, or
4288 you can supply the version script as an implicit linker script. You can
4289 also use the @samp{--version-script} linker option.
4291 The syntax of the @code{VERSION} command is simply
4293 VERSION @{ version-script-commands @}
4296 The format of the version script commands is identical to that used by
4297 Sun's linker in Solaris 2.5. The version script defines a tree of
4298 version nodes. You specify the node names and interdependencies in the
4299 version script. You can specify which symbols are bound to which
4300 version nodes, and you can reduce a specified set of symbols to local
4301 scope so that they are not globally visible outside of the shared
4304 The easiest way to demonstrate the version script language is with a few
4325 "int f(int, double)";
4330 This example version script defines three version nodes. The first
4331 version node defined is @samp{VERS_1.1}; it has no other dependencies.
4332 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
4333 a number of symbols to local scope so that they are not visible outside
4334 of the shared library; this is done using wildcard patterns, so that any
4335 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4336 is matched. The wildcard patterns available are the same as those used
4337 in the shell when matching filenames (also known as ``globbing'').
4338 However, if you specify the symbol name inside double quotes, then the
4339 name is treated as literal, rather than as a glob pattern.
4341 Next, the version script defines node @samp{VERS_1.2}. This node
4342 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
4343 to the version node @samp{VERS_1.2}.
4345 Finally, the version script defines node @samp{VERS_2.0}. This node
4346 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
4347 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4349 When the linker finds a symbol defined in a library which is not
4350 specifically bound to a version node, it will effectively bind it to an
4351 unspecified base version of the library. You can bind all otherwise
4352 unspecified symbols to a given version node by using @samp{global: *;}
4353 somewhere in the version script.
4355 The names of the version nodes have no specific meaning other than what
4356 they might suggest to the person reading them. The @samp{2.0} version
4357 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4358 However, this would be a confusing way to write a version script.
4360 Node name can be omited, provided it is the only version node
4361 in the version script. Such version script doesn't assign any versions to
4362 symbols, only selects which symbols will be globally visible out and which
4366 @{ global: foo; bar; local: *; @};
4369 When you link an application against a shared library that has versioned
4370 symbols, the application itself knows which version of each symbol it
4371 requires, and it also knows which version nodes it needs from each
4372 shared library it is linked against. Thus at runtime, the dynamic
4373 loader can make a quick check to make sure that the libraries you have
4374 linked against do in fact supply all of the version nodes that the
4375 application will need to resolve all of the dynamic symbols. In this
4376 way it is possible for the dynamic linker to know with certainty that
4377 all external symbols that it needs will be resolvable without having to
4378 search for each symbol reference.
4380 The symbol versioning is in effect a much more sophisticated way of
4381 doing minor version checking that SunOS does. The fundamental problem
4382 that is being addressed here is that typically references to external
4383 functions are bound on an as-needed basis, and are not all bound when
4384 the application starts up. If a shared library is out of date, a
4385 required interface may be missing; when the application tries to use
4386 that interface, it may suddenly and unexpectedly fail. With symbol
4387 versioning, the user will get a warning when they start their program if
4388 the libraries being used with the application are too old.
4390 There are several GNU extensions to Sun's versioning approach. The
4391 first of these is the ability to bind a symbol to a version node in the
4392 source file where the symbol is defined instead of in the versioning
4393 script. This was done mainly to reduce the burden on the library
4394 maintainer. You can do this by putting something like:
4396 __asm__(".symver original_foo,foo@@VERS_1.1");
4399 in the C source file. This renames the function @samp{original_foo} to
4400 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4401 The @samp{local:} directive can be used to prevent the symbol
4402 @samp{original_foo} from being exported. A @samp{.symver} directive
4403 takes precedence over a version script.
4405 The second GNU extension is to allow multiple versions of the same
4406 function to appear in a given shared library. In this way you can make
4407 an incompatible change to an interface without increasing the major
4408 version number of the shared library, while still allowing applications
4409 linked against the old interface to continue to function.
4411 To do this, you must use multiple @samp{.symver} directives in the
4412 source file. Here is an example:
4415 __asm__(".symver original_foo,foo@@");
4416 __asm__(".symver old_foo,foo@@VERS_1.1");
4417 __asm__(".symver old_foo1,foo@@VERS_1.2");
4418 __asm__(".symver new_foo,foo@@@@VERS_2.0");
4421 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4422 unspecified base version of the symbol. The source file that contains this
4423 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4424 @samp{old_foo1}, and @samp{new_foo}.
4426 When you have multiple definitions of a given symbol, there needs to be
4427 some way to specify a default version to which external references to
4428 this symbol will be bound. You can do this with the
4429 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
4430 declare one version of a symbol as the default in this manner; otherwise
4431 you would effectively have multiple definitions of the same symbol.
4433 If you wish to bind a reference to a specific version of the symbol
4434 within the shared library, you can use the aliases of convenience
4435 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4436 specifically bind to an external version of the function in question.
4438 You can also specify the language in the version script:
4441 VERSION extern "lang" @{ version-script-commands @}
4444 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4445 The linker will iterate over the list of symbols at the link time and
4446 demangle them according to @samp{lang} before matching them to the
4447 patterns specified in @samp{version-script-commands}.
4449 Demangled names may contains spaces and other special characters. As
4450 described above, you can use a glob pattern to match demangled names,
4451 or you can use a double-quoted string to match the string exactly. In
4452 the latter case, be aware that minor differences (such as differing
4453 whitespace) between the version script and the demangler output will
4454 cause a mismatch. As the exact string generated by the demangler
4455 might change in the future, even if the mangled name does not, you
4456 should check that all of your version directives are behaving as you
4457 expect when you upgrade.
4460 @section Expressions in Linker Scripts
4463 The syntax for expressions in the linker script language is identical to
4464 that of C expressions. All expressions are evaluated as integers. All
4465 expressions are evaluated in the same size, which is 32 bits if both the
4466 host and target are 32 bits, and is otherwise 64 bits.
4468 You can use and set symbol values in expressions.
4470 The linker defines several special purpose builtin functions for use in
4474 * Constants:: Constants
4475 * Symbols:: Symbol Names
4476 * Orphan Sections:: Orphan Sections
4477 * Location Counter:: The Location Counter
4478 * Operators:: Operators
4479 * Evaluation:: Evaluation
4480 * Expression Section:: The Section of an Expression
4481 * Builtin Functions:: Builtin Functions
4485 @subsection Constants
4486 @cindex integer notation
4487 @cindex constants in linker scripts
4488 All constants are integers.
4490 As in C, the linker considers an integer beginning with @samp{0} to be
4491 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
4492 hexadecimal. The linker considers other integers to be decimal.
4494 @cindex scaled integers
4495 @cindex K and M integer suffixes
4496 @cindex M and K integer suffixes
4497 @cindex suffixes for integers
4498 @cindex integer suffixes
4499 In addition, you can use the suffixes @code{K} and @code{M} to scale a
4503 @c END TEXI2ROFF-KILL
4504 @code{1024} or @code{1024*1024}
4508 ${\rm 1024}$ or ${\rm 1024}^2$
4510 @c END TEXI2ROFF-KILL
4511 respectively. For example, the following all refer to the same quantity:
4519 @subsection Symbol Names
4520 @cindex symbol names
4522 @cindex quoted symbol names
4524 Unless quoted, symbol names start with a letter, underscore, or period
4525 and may include letters, digits, underscores, periods, and hyphens.
4526 Unquoted symbol names must not conflict with any keywords. You can
4527 specify a symbol which contains odd characters or has the same name as a
4528 keyword by surrounding the symbol name in double quotes:
4531 "with a space" = "also with a space" + 10;
4534 Since symbols can contain many non-alphabetic characters, it is safest
4535 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
4536 whereas @samp{A - B} is an expression involving subtraction.
4538 @node Orphan Sections
4539 @subsection Orphan Sections
4541 Orphan sections are sections present in the input files which
4542 are not explicitly placed into the output file by the linker
4543 script. The linker will still copy these sections into the
4544 output file, but it has to guess as to where they should be
4545 placed. The linker uses a simple heuristic to do this. It
4546 attempts to place orphan sections after non-orphan sections of the
4547 same attribute, such as code vs data, loadable vs non-loadable, etc.
4548 If there is not enough room to do this then it places
4549 at the end of the file.
4551 For ELF targets, the attribute of the section includes section type as
4552 well as section flag.
4554 @node Location Counter
4555 @subsection The Location Counter
4558 @cindex location counter
4559 @cindex current output location
4560 The special linker variable @dfn{dot} @samp{.} always contains the
4561 current output location counter. Since the @code{.} always refers to a
4562 location in an output section, it may only appear in an expression
4563 within a @code{SECTIONS} command. The @code{.} symbol may appear
4564 anywhere that an ordinary symbol is allowed in an expression.
4567 Assigning a value to @code{.} will cause the location counter to be
4568 moved. This may be used to create holes in the output section. The
4569 location counter may never be moved backwards.
4585 In the previous example, the @samp{.text} section from @file{file1} is
4586 located at the beginning of the output section @samp{output}. It is
4587 followed by a 1000 byte gap. Then the @samp{.text} section from
4588 @file{file2} appears, also with a 1000 byte gap following before the
4589 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
4590 specifies what data to write in the gaps (@pxref{Output Section Fill}).
4592 @cindex dot inside sections
4593 Note: @code{.} actually refers to the byte offset from the start of the
4594 current containing object. Normally this is the @code{SECTIONS}
4595 statement, whose start address is 0, hence @code{.} can be used as an
4596 absolute address. If @code{.} is used inside a section description
4597 however, it refers to the byte offset from the start of that section,
4598 not an absolute address. Thus in a script like this:
4616 The @samp{.text} section will be assigned a starting address of 0x100
4617 and a size of exactly 0x200 bytes, even if there is not enough data in
4618 the @samp{.text} input sections to fill this area. (If there is too
4619 much data, an error will be produced because this would be an attempt to
4620 move @code{.} backwards). The @samp{.data} section will start at 0x500
4621 and it will have an extra 0x600 bytes worth of space after the end of
4622 the values from the @samp{.data} input sections and before the end of
4623 the @samp{.data} output section itself.
4625 @cindex dot outside sections
4626 Setting symbols to the value of the location counter outside of an
4627 output section statement can result in unexpected values if the linker
4628 needs to place orphan sections. For example, given the following:
4634 .text: @{ *(.text) @}
4638 .data: @{ *(.data) @}
4643 If the linker needs to place some input section, e.g. @code{.rodata},
4644 not mentioned in the script, it might choose to place that section
4645 between @code{.text} and @code{.data}. You might think the linker
4646 should place @code{.rodata} on the blank line in the above script, but
4647 blank lines are of no particular significance to the linker. As well,
4648 the linker doesn't associate the above symbol names with their
4649 sections. Instead, it assumes that all assignments or other
4650 statements belong to the previous output section, except for the
4651 special case of an assignment to @code{.}. I.e., the linker will
4652 place the orphan @code{.rodata} section as if the script was written
4659 .text: @{ *(.text) @}
4663 .rodata: @{ *(.rodata) @}
4664 .data: @{ *(.data) @}
4669 This may or may not be the script author's intention for the value of
4670 @code{start_of_data}. One way to influence the orphan section
4671 placement is to assign the location counter to itself, as the linker
4672 assumes that an assignment to @code{.} is setting the start address of
4673 a following output section and thus should be grouped with that
4674 section. So you could write:
4680 .text: @{ *(.text) @}
4685 .data: @{ *(.data) @}
4690 Now, the orphan @code{.rodata} section will be placed between
4691 @code{end_of_text} and @code{start_of_data}.
4695 @subsection Operators
4696 @cindex operators for arithmetic
4697 @cindex arithmetic operators
4698 @cindex precedence in expressions
4699 The linker recognizes the standard C set of arithmetic operators, with
4700 the standard bindings and precedence levels:
4703 @c END TEXI2ROFF-KILL
4705 precedence associativity Operators Notes
4711 5 left == != > < <= >=
4717 11 right &= += -= *= /= (2)
4721 (1) Prefix operators
4722 (2) @xref{Assignments}.
4726 \vskip \baselineskip
4727 %"lispnarrowing" is the extra indent used generally for smallexample
4728 \hskip\lispnarrowing\vbox{\offinterlineskip
4731 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
4732 height2pt&\omit&&\omit&&\omit&\cr
4733 &Precedence&& Associativity &&{\rm Operators}&\cr
4734 height2pt&\omit&&\omit&&\omit&\cr
4736 height2pt&\omit&&\omit&&\omit&\cr
4738 % '176 is tilde, '~' in tt font
4739 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
4740 &2&&left&&* / \%&\cr
4743 &5&&left&&== != > < <= >=&\cr
4746 &8&&left&&{\&\&}&\cr
4749 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
4751 height2pt&\omit&&\omit&&\omit&\cr}
4756 @obeylines@parskip=0pt@parindent=0pt
4757 @dag@quad Prefix operators.
4758 @ddag@quad @xref{Assignments}.
4761 @c END TEXI2ROFF-KILL
4764 @subsection Evaluation
4765 @cindex lazy evaluation
4766 @cindex expression evaluation order
4767 The linker evaluates expressions lazily. It only computes the value of
4768 an expression when absolutely necessary.
4770 The linker needs some information, such as the value of the start
4771 address of the first section, and the origins and lengths of memory
4772 regions, in order to do any linking at all. These values are computed
4773 as soon as possible when the linker reads in the linker script.
4775 However, other values (such as symbol values) are not known or needed
4776 until after storage allocation. Such values are evaluated later, when
4777 other information (such as the sizes of output sections) is available
4778 for use in the symbol assignment expression.
4780 The sizes of sections cannot be known until after allocation, so
4781 assignments dependent upon these are not performed until after
4784 Some expressions, such as those depending upon the location counter
4785 @samp{.}, must be evaluated during section allocation.
4787 If the result of an expression is required, but the value is not
4788 available, then an error results. For example, a script like the
4794 .text 9+this_isnt_constant :
4800 will cause the error message @samp{non constant expression for initial
4803 @node Expression Section
4804 @subsection The Section of an Expression
4805 @cindex expression sections
4806 @cindex absolute expressions
4807 @cindex relative expressions
4808 @cindex absolute and relocatable symbols
4809 @cindex relocatable and absolute symbols
4810 @cindex symbols, relocatable and absolute
4811 When the linker evaluates an expression, the result is either absolute
4812 or relative to some section. A relative expression is expressed as a
4813 fixed offset from the base of a section.
4815 The position of the expression within the linker script determines
4816 whether it is absolute or relative. An expression which appears within
4817 an output section definition is relative to the base of the output
4818 section. An expression which appears elsewhere will be absolute.
4820 A symbol set to a relative expression will be relocatable if you request
4821 relocatable output using the @samp{-r} option. That means that a
4822 further link operation may change the value of the symbol. The symbol's
4823 section will be the section of the relative expression.
4825 A symbol set to an absolute expression will retain the same value
4826 through any further link operation. The symbol will be absolute, and
4827 will not have any particular associated section.
4829 You can use the builtin function @code{ABSOLUTE} to force an expression
4830 to be absolute when it would otherwise be relative. For example, to
4831 create an absolute symbol set to the address of the end of the output
4832 section @samp{.data}:
4836 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
4840 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
4841 @samp{.data} section.
4843 @node Builtin Functions
4844 @subsection Builtin Functions
4845 @cindex functions in expressions
4846 The linker script language includes a number of builtin functions for
4847 use in linker script expressions.
4850 @item ABSOLUTE(@var{exp})
4851 @kindex ABSOLUTE(@var{exp})
4852 @cindex expression, absolute
4853 Return the absolute (non-relocatable, as opposed to non-negative) value
4854 of the expression @var{exp}. Primarily useful to assign an absolute
4855 value to a symbol within a section definition, where symbol values are
4856 normally section relative. @xref{Expression Section}.
4858 @item ADDR(@var{section})
4859 @kindex ADDR(@var{section})
4860 @cindex section address in expression
4861 Return the absolute address (the VMA) of the named @var{section}. Your
4862 script must previously have defined the location of that section. In
4863 the following example, @code{symbol_1} and @code{symbol_2} are assigned
4870 start_of_output_1 = ABSOLUTE(.);
4875 symbol_1 = ADDR(.output1);
4876 symbol_2 = start_of_output_1;
4882 @item ALIGN(@var{align})
4883 @itemx ALIGN(@var{exp},@var{align})
4884 @kindex ALIGN(@var{align})
4885 @kindex ALIGN(@var{exp},@var{align})
4886 @cindex round up location counter
4887 @cindex align location counter
4888 @cindex round up expression
4889 @cindex align expression
4890 Return the location counter (@code{.}) or arbitrary expression aligned
4891 to the next @var{align} boundary. The single operand @code{ALIGN}
4892 doesn't change the value of the location counter---it just does
4893 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
4894 expression to be aligned upwards (@code{ALIGN(@var{align})} is
4895 equivalent to @code{ALIGN(., @var{align})}).
4897 Here is an example which aligns the output @code{.data} section to the
4898 next @code{0x2000} byte boundary after the preceding section and sets a
4899 variable within the section to the next @code{0x8000} boundary after the
4904 .data ALIGN(0x2000): @{
4906 variable = ALIGN(0x8000);
4912 The first use of @code{ALIGN} in this example specifies the location of
4913 a section because it is used as the optional @var{address} attribute of
4914 a section definition (@pxref{Output Section Address}). The second use
4915 of @code{ALIGN} is used to defines the value of a symbol.
4917 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
4919 @item BLOCK(@var{exp})
4920 @kindex BLOCK(@var{exp})
4921 This is a synonym for @code{ALIGN}, for compatibility with older linker
4922 scripts. It is most often seen when setting the address of an output
4925 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
4926 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
4927 This is equivalent to either
4929 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
4933 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
4936 depending on whether the latter uses fewer @var{commonpagesize} sized pages
4937 for the data segment (area between the result of this expression and
4938 @code{DATA_SEGMENT_END}) than the former or not.
4939 If the latter form is used, it means @var{commonpagesize} bytes of runtime
4940 memory will be saved at the expense of up to @var{commonpagesize} wasted
4941 bytes in the on-disk file.
4943 This expression can only be used directly in @code{SECTIONS} commands, not in
4944 any output section descriptions and only once in the linker script.
4945 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
4946 be the system page size the object wants to be optimized for (while still
4947 working on system page sizes up to @var{maxpagesize}).
4952 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
4955 @item DATA_SEGMENT_END(@var{exp})
4956 @kindex DATA_SEGMENT_END(@var{exp})
4957 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
4958 evaluation purposes.
4961 . = DATA_SEGMENT_END(.);
4964 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
4965 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
4966 This defines the end of the @code{PT_GNU_RELRO} segment when
4967 @samp{-z relro} option is used. Second argument is returned.
4968 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
4969 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
4970 @var{exp} + @var{offset} is aligned to the most commonly used page
4971 boundary for particular target. If present in the linker script,
4972 it must always come in between @code{DATA_SEGMENT_ALIGN} and
4973 @code{DATA_SEGMENT_END}.
4976 . = DATA_SEGMENT_RELRO_END(24, .);
4979 @item DEFINED(@var{symbol})
4980 @kindex DEFINED(@var{symbol})
4981 @cindex symbol defaults
4982 Return 1 if @var{symbol} is in the linker global symbol table and is
4983 defined before the statement using DEFINED in the script, otherwise
4984 return 0. You can use this function to provide
4985 default values for symbols. For example, the following script fragment
4986 shows how to set a global symbol @samp{begin} to the first location in
4987 the @samp{.text} section---but if a symbol called @samp{begin} already
4988 existed, its value is preserved:
4994 begin = DEFINED(begin) ? begin : . ;
5002 @item LENGTH(@var{memory})
5003 @kindex LENGTH(@var{memory})
5004 Return the length of the memory region named @var{memory}.
5006 @item LOADADDR(@var{section})
5007 @kindex LOADADDR(@var{section})
5008 @cindex section load address in expression
5009 Return the absolute LMA of the named @var{section}. This is normally
5010 the same as @code{ADDR}, but it may be different if the @code{AT}
5011 attribute is used in the output section definition (@pxref{Output
5015 @item MAX(@var{exp1}, @var{exp2})
5016 Returns the maximum of @var{exp1} and @var{exp2}.
5019 @item MIN(@var{exp1}, @var{exp2})
5020 Returns the minimum of @var{exp1} and @var{exp2}.
5022 @item NEXT(@var{exp})
5023 @kindex NEXT(@var{exp})
5024 @cindex unallocated address, next
5025 Return the next unallocated address that is a multiple of @var{exp}.
5026 This function is closely related to @code{ALIGN(@var{exp})}; unless you
5027 use the @code{MEMORY} command to define discontinuous memory for the
5028 output file, the two functions are equivalent.
5030 @item ORIGIN(@var{memory})
5031 @kindex ORIGIN(@var{memory})
5032 Return the origin of the memory region named @var{memory}.
5034 @item SEGMENT_START(@var{segment}, @var{default})
5035 @kindex SEGMENT_START(@var{segment}, @var{default})
5036 Return the base address of the named @var{segment}. If an explicit
5037 value has been given for this segment (with a command-line @samp{-T}
5038 option) that value will be returned; otherwise the value will be
5039 @var{default}. At present, the @samp{-T} command-line option can only
5040 be used to set the base address for the ``text'', ``data'', and
5041 ``bss'' sections, but you use @code{SEGMENT_START} with any segment
5044 @item SIZEOF(@var{section})
5045 @kindex SIZEOF(@var{section})
5046 @cindex section size
5047 Return the size in bytes of the named @var{section}, if that section has
5048 been allocated. If the section has not been allocated when this is
5049 evaluated, the linker will report an error. In the following example,
5050 @code{symbol_1} and @code{symbol_2} are assigned identical values:
5059 symbol_1 = .end - .start ;
5060 symbol_2 = SIZEOF(.output);
5065 @item SIZEOF_HEADERS
5066 @itemx sizeof_headers
5067 @kindex SIZEOF_HEADERS
5069 Return the size in bytes of the output file's headers. This is
5070 information which appears at the start of the output file. You can use
5071 this number when setting the start address of the first section, if you
5072 choose, to facilitate paging.
5074 @cindex not enough room for program headers
5075 @cindex program headers, not enough room
5076 When producing an ELF output file, if the linker script uses the
5077 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
5078 number of program headers before it has determined all the section
5079 addresses and sizes. If the linker later discovers that it needs
5080 additional program headers, it will report an error @samp{not enough
5081 room for program headers}. To avoid this error, you must avoid using
5082 the @code{SIZEOF_HEADERS} function, or you must rework your linker
5083 script to avoid forcing the linker to use additional program headers, or
5084 you must define the program headers yourself using the @code{PHDRS}
5085 command (@pxref{PHDRS}).
5088 @node Implicit Linker Scripts
5089 @section Implicit Linker Scripts
5090 @cindex implicit linker scripts
5091 If you specify a linker input file which the linker can not recognize as
5092 an object file or an archive file, it will try to read the file as a
5093 linker script. If the file can not be parsed as a linker script, the
5094 linker will report an error.
5096 An implicit linker script will not replace the default linker script.
5098 Typically an implicit linker script would contain only symbol
5099 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5102 Any input files read because of an implicit linker script will be read
5103 at the position in the command line where the implicit linker script was
5104 read. This can affect archive searching.
5107 @node Machine Dependent
5108 @chapter Machine Dependent Features
5110 @cindex machine dependencies
5111 @command{ld} has additional features on some platforms; the following
5112 sections describe them. Machines where @command{ld} has no additional
5113 functionality are not listed.
5117 * H8/300:: @command{ld} and the H8/300
5120 * i960:: @command{ld} and the Intel 960 family
5123 * ARM:: @command{ld} and the ARM family
5126 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
5129 * MMIX:: @command{ld} and MMIX
5132 * MSP430:: @command{ld} and MSP430
5135 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
5138 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
5141 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
5144 * TI COFF:: @command{ld} and TI COFF
5147 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
5150 * Xtensa:: @command{ld} and Xtensa Processors
5161 @section @command{ld} and the H8/300
5163 @cindex H8/300 support
5164 For the H8/300, @command{ld} can perform these global optimizations when
5165 you specify the @samp{--relax} command-line option.
5168 @cindex relaxing on H8/300
5169 @item relaxing address modes
5170 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5171 targets are within eight bits, and turns them into eight-bit
5172 program-counter relative @code{bsr} and @code{bra} instructions,
5175 @cindex synthesizing on H8/300
5176 @item synthesizing instructions
5177 @c FIXME: specifically mov.b, or any mov instructions really?
5178 @command{ld} finds all @code{mov.b} instructions which use the
5179 sixteen-bit absolute address form, but refer to the top
5180 page of memory, and changes them to use the eight-bit address form.
5181 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5182 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5183 top page of memory).
5185 @item bit manipulation instructions
5186 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
5187 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5188 which use 32 bit and 16 bit absolute address form, but refer to the top
5189 page of memory, and changes them to use the 8 bit address form.
5190 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5191 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5192 the top page of memory).
5194 @item system control instructions
5195 @command{ld} finds all @code{ldc.w, stc.w} instrcutions which use the
5196 32 bit absolute address form, but refer to the top page of memory, and
5197 changes them to use 16 bit address form.
5198 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5199 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5200 the top page of memory).
5210 @c This stuff is pointless to say unless you're especially concerned
5211 @c with Renesas chips; don't enable it for generic case, please.
5213 @chapter @command{ld} and Other Renesas Chips
5215 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5216 H8/500, and SH chips. No special features, commands, or command-line
5217 options are required for these chips.
5227 @section @command{ld} and the Intel 960 Family
5229 @cindex i960 support
5231 You can use the @samp{-A@var{architecture}} command line option to
5232 specify one of the two-letter names identifying members of the 960
5233 family; the option specifies the desired output target, and warns of any
5234 incompatible instructions in the input files. It also modifies the
5235 linker's search strategy for archive libraries, to support the use of
5236 libraries specific to each particular architecture, by including in the
5237 search loop names suffixed with the string identifying the architecture.
5239 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5240 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5241 paths, and in any paths you specify with @samp{-L}) for a library with
5254 The first two possibilities would be considered in any event; the last
5255 two are due to the use of @w{@samp{-ACA}}.
5257 You can meaningfully use @samp{-A} more than once on a command line, since
5258 the 960 architecture family allows combination of target architectures; each
5259 use will add another pair of name variants to search for when @w{@samp{-l}}
5260 specifies a library.
5262 @cindex @option{--relax} on i960
5263 @cindex relaxing on i960
5264 @command{ld} supports the @samp{--relax} option for the i960 family. If
5265 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5266 @code{calx} instructions whose targets are within 24 bits, and turns
5267 them into 24-bit program-counter relative @code{bal} and @code{cal}
5268 instructions, respectively. @command{ld} also turns @code{cal}
5269 instructions into @code{bal} instructions when it determines that the
5270 target subroutine is a leaf routine (that is, the target subroutine does
5271 not itself call any subroutines).
5288 @node M68HC11/68HC12
5289 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
5291 @cindex M68HC11 and 68HC12 support
5293 @subsection Linker Relaxation
5295 For the Motorola 68HC11, @command{ld} can perform these global
5296 optimizations when you specify the @samp{--relax} command-line option.
5299 @cindex relaxing on M68HC11
5300 @item relaxing address modes
5301 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5302 targets are within eight bits, and turns them into eight-bit
5303 program-counter relative @code{bsr} and @code{bra} instructions,
5306 @command{ld} also looks at all 16-bit extended addressing modes and
5307 transforms them in a direct addressing mode when the address is in
5308 page 0 (between 0 and 0x0ff).
5310 @item relaxing gcc instruction group
5311 When @command{gcc} is called with @option{-mrelax}, it can emit group
5312 of instructions that the linker can optimize to use a 68HC11 direct
5313 addressing mode. These instructions consists of @code{bclr} or
5314 @code{bset} instructions.
5318 @subsection Trampoline Generation
5320 @cindex trampoline generation on M68HC11
5321 @cindex trampoline generation on M68HC12
5322 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5323 call a far function using a normal @code{jsr} instruction. The linker
5324 will also change the relocation to some far function to use the
5325 trampoline address instead of the function address. This is typically the
5326 case when a pointer to a function is taken. The pointer will in fact
5327 point to the function trampoline.
5335 @section @command{ld} and the ARM family
5337 @cindex ARM interworking support
5338 @kindex --support-old-code
5339 For the ARM, @command{ld} will generate code stubs to allow functions calls
5340 betweem ARM and Thumb code. These stubs only work with code that has
5341 been compiled and assembled with the @samp{-mthumb-interwork} command
5342 line option. If it is necessary to link with old ARM object files or
5343 libraries, which have not been compiled with the -mthumb-interwork
5344 option then the @samp{--support-old-code} command line switch should be
5345 given to the linker. This will make it generate larger stub functions
5346 which will work with non-interworking aware ARM code. Note, however,
5347 the linker does not support generating stubs for function calls to
5348 non-interworking aware Thumb code.
5350 @cindex thumb entry point
5351 @cindex entry point, thumb
5352 @kindex --thumb-entry=@var{entry}
5353 The @samp{--thumb-entry} switch is a duplicate of the generic
5354 @samp{--entry} switch, in that it sets the program's starting address.
5355 But it also sets the bottom bit of the address, so that it can be
5356 branched to using a BX instruction, and the program will start
5357 executing in Thumb mode straight away.
5361 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5362 executables. This option is only valid when linking big-endian objects.
5363 The resulting image will contain big-endian data and little-endian code.
5366 @kindex --target1-rel
5367 @kindex --target1-abs
5368 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5369 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
5370 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
5371 and @samp{--target1-abs} switches override the default.
5374 @kindex --target2=@var{type}
5375 The @samp{--target2=type} switch overrides the default definition of the
5376 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
5377 meanings, and target defaults are as follows:
5380 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5382 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
5384 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5389 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5390 specification) enables objects compiled for the ARMv4 architecture to be
5391 interworking-safe when linked with other objects compiled for ARMv4t, but
5392 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
5394 In the latter case, the switch @option{--fix-v4bx} must be passed to the
5395 linker, which causes v4t @code{BX rM} instructions to be rewritten as
5396 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
5398 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
5399 relocations are ignored.
5403 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
5404 BLX instructions (available on ARMv5t and above) in various
5405 situations. Currently it is used to perform calls via the PLT from Thumb
5406 code using BLX rather than using BX and a mode-switching stub before
5407 each PLT entry. This should lead to such calls executing slightly faster.
5409 This option is enabled implicitly for SymbianOS, so there is no need to
5410 specify it if you are using that target.
5423 @section @command{ld} and HPPA 32-bit ELF Support
5424 @cindex HPPA multiple sub-space stubs
5425 @kindex --multi-subspace
5426 When generating a shared library, @command{ld} will by default generate
5427 import stubs suitable for use with a single sub-space application.
5428 The @samp{--multi-subspace} switch causes @command{ld} to generate export
5429 stubs, and different (larger) import stubs suitable for use with
5430 multiple sub-spaces.
5432 @cindex HPPA stub grouping
5433 @kindex --stub-group-size=@var{N}
5434 Long branch stubs and import/export stubs are placed by @command{ld} in
5435 stub sections located between groups of input sections.
5436 @samp{--stub-group-size} specifies the maximum size of a group of input
5437 sections handled by one stub section. Since branch offsets are signed,
5438 a stub section may serve two groups of input sections, one group before
5439 the stub section, and one group after it. However, when using
5440 conditional branches that require stubs, it may be better (for branch
5441 prediction) that stub sections only serve one group of input sections.
5442 A negative value for @samp{N} chooses this scheme, ensuring that
5443 branches to stubs always use a negative offset. Two special values of
5444 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5445 @command{ld} to automatically size input section groups for the branch types
5446 detected, with the same behaviour regarding stub placement as other
5447 positive or negative values of @samp{N} respectively.
5449 Note that @samp{--stub-group-size} does not split input sections. A
5450 single input section larger than the group size specified will of course
5451 create a larger group (of one section). If input sections are too
5452 large, it may not be possible for a branch to reach its stub.
5465 @section @code{ld} and MMIX
5466 For MMIX, there is a choice of generating @code{ELF} object files or
5467 @code{mmo} object files when linking. The simulator @code{mmix}
5468 understands the @code{mmo} format. The binutils @code{objcopy} utility
5469 can translate between the two formats.
5471 There is one special section, the @samp{.MMIX.reg_contents} section.
5472 Contents in this section is assumed to correspond to that of global
5473 registers, and symbols referring to it are translated to special symbols,
5474 equal to registers. In a final link, the start address of the
5475 @samp{.MMIX.reg_contents} section corresponds to the first allocated
5476 global register multiplied by 8. Register @code{$255} is not included in
5477 this section; it is always set to the program entry, which is at the
5478 symbol @code{Main} for @code{mmo} files.
5480 Symbols with the prefix @code{__.MMIX.start.}, for example
5481 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special;
5482 there must be only one each, even if they are local. The default linker
5483 script uses these to set the default start address of a section.
5485 Initial and trailing multiples of zero-valued 32-bit words in a section,
5486 are left out from an mmo file.
5499 @section @code{ld} and MSP430
5500 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
5501 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
5502 just pass @samp{-m help} option to the linker).
5504 @cindex MSP430 extra sections
5505 The linker will recognize some extra sections which are MSP430 specific:
5508 @item @samp{.vectors}
5509 Defines a portion of ROM where interrupt vectors located.
5511 @item @samp{.bootloader}
5512 Defines the bootloader portion of the ROM (if applicable). Any code
5513 in this section will be uploaded to the MPU.
5515 @item @samp{.infomem}
5516 Defines an information memory section (if applicable). Any code in
5517 this section will be uploaded to the MPU.
5519 @item @samp{.infomemnobits}
5520 This is the same as the @samp{.infomem} section except that any code
5521 in this section will not be uploaded to the MPU.
5523 @item @samp{.noinit}
5524 Denotes a portion of RAM located above @samp{.bss} section.
5526 The last two sections are used by gcc.
5540 @section @command{ld} and PowerPC 32-bit ELF Support
5541 @cindex PowerPC long branches
5542 @kindex --relax on PowerPC
5543 Branches on PowerPC processors are limited to a signed 26-bit
5544 displacement, which may result in @command{ld} giving
5545 @samp{relocation truncated to fit} errors with very large programs.
5546 @samp{--relax} enables the generation of trampolines that can access
5547 the entire 32-bit address space. These trampolines are inserted at
5548 section boundaries, so may not themselves be reachable if an input
5549 section exceeds 33M in size.
5551 @cindex PowerPC ELF32 options
5556 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
5557 generates code capable of using a newer PLT and GOT layout that has
5558 the security advantage of no executable section ever needing to be
5559 writable and no writable section ever being executable. PowerPC
5560 @command{ld} will generate this layout, including stubs to access the
5561 PLT, if all input files (including startup and static libraries) were
5562 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
5563 BSS PLT (and GOT layout) which can give slightly better performance.
5568 The new secure PLT and GOT are placed differently relative to other
5569 sections compared to older BSS PLT and GOT placement. The location of
5570 @code{.plt} must change because the new secure PLT is an initialized
5571 section while the old PLT is uninitialized. The reason for the
5572 @code{.got} change is more subtle: The new placement allows
5573 @code{.got} to be read-only in applications linked with
5574 @samp{-z relro -z now}. However, this placement means that
5575 @code{.sdata} cannot always be used in shared libraries, because the
5576 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
5577 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
5578 GCC doesn't use @code{.sdata} in shared libraries, so this option is
5579 really only useful for other compilers that may do so.
5581 @cindex PowerPC stub symbols
5582 @kindex --emit-stub-syms
5583 @item --emit-stub-syms
5584 This option causes @command{ld} to label linker stubs with a local
5585 symbol that encodes the stub type and destination.
5587 @cindex PowerPC TLS optimization
5588 @kindex --no-tls-optimize
5589 @item --no-tls-optimize
5590 PowerPC @command{ld} normally performs some optimization of code
5591 sequences used to access Thread-Local Storage. Use this option to
5592 disable the optimization.
5605 @node PowerPC64 ELF64
5606 @section @command{ld} and PowerPC64 64-bit ELF Support
5608 @cindex PowerPC64 ELF64 options
5610 @cindex PowerPC64 stub grouping
5611 @kindex --stub-group-size
5612 @item --stub-group-size
5613 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
5614 by @command{ld} in stub sections located between groups of input sections.
5615 @samp{--stub-group-size} specifies the maximum size of a group of input
5616 sections handled by one stub section. Since branch offsets are signed,
5617 a stub section may serve two groups of input sections, one group before
5618 the stub section, and one group after it. However, when using
5619 conditional branches that require stubs, it may be better (for branch
5620 prediction) that stub sections only serve one group of input sections.
5621 A negative value for @samp{N} chooses this scheme, ensuring that
5622 branches to stubs always use a negative offset. Two special values of
5623 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
5624 @command{ld} to automatically size input section groups for the branch types
5625 detected, with the same behaviour regarding stub placement as other
5626 positive or negative values of @samp{N} respectively.
5628 Note that @samp{--stub-group-size} does not split input sections. A
5629 single input section larger than the group size specified will of course
5630 create a larger group (of one section). If input sections are too
5631 large, it may not be possible for a branch to reach its stub.
5633 @cindex PowerPC64 stub symbols
5634 @kindex --emit-stub-syms
5635 @item --emit-stub-syms
5636 This option causes @command{ld} to label linker stubs with a local
5637 symbol that encodes the stub type and destination.
5639 @cindex PowerPC64 dot symbols
5641 @kindex --no-dotsyms
5642 @item --dotsyms, --no-dotsyms
5643 These two options control how @command{ld} interprets version patterns
5644 in a version script. Older PowerPC64 compilers emitted both a
5645 function descriptor symbol with the same name as the function, and a
5646 code entry symbol with the name prefixed by a dot (@samp{.}). To
5647 properly version a function @samp{foo}, the version script thus needs
5648 to control both @samp{foo} and @samp{.foo}. The option
5649 @samp{--dotsyms}, on by default, automatically adds the required
5650 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
5653 @cindex PowerPC64 TLS optimization
5654 @kindex --no-tls-optimize
5655 @item --no-tls-optimize
5656 PowerPC64 @command{ld} normally performs some optimization of code
5657 sequences used to access Thread-Local Storage. Use this option to
5658 disable the optimization.
5660 @cindex PowerPC64 OPD optimization
5661 @kindex --no-opd-optimize
5662 @item --no-opd-optimize
5663 PowerPC64 @command{ld} normally removes @code{.opd} section entries
5664 corresponding to deleted link-once functions, or functions removed by
5665 the action of @samp{--gc-sections} or linker scrip @code{/DISCARD/}.
5666 Use this option to disable @code{.opd} optimization.
5668 @cindex PowerPC64 OPD spacing
5669 @kindex --non-overlapping-opd
5670 @item --non-overlapping-opd
5671 Some PowerPC64 compilers have an option to generate compressed
5672 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
5673 the static chain pointer (unused in C) with the first word of the next
5674 entry. This option expands such entries to the full 24 bytes.
5676 @cindex PowerPC64 TOC optimization
5677 @kindex --no-toc-optimize
5678 @item --no-toc-optimize
5679 PowerPC64 @command{ld} normally removes unused @code{.toc} section
5680 entries. Such entries are detected by examining relocations that
5681 reference the TOC in code sections. A reloc in a deleted code section
5682 marks a TOC word as unneeded, while a reloc in a kept code section
5683 marks a TOC word as needed. Since the TOC may reference itself, TOC
5684 relocs are also examined. TOC words marked as both needed and
5685 unneeded will of course be kept. TOC words without any referencing
5686 reloc are assumed to be part of a multi-word entry, and are kept or
5687 discarded as per the nearest marked preceding word. This works
5688 reliably for compiler generated code, but may be incorrect if assembly
5689 code is used to insert TOC entries. Use this option to disable the
5692 @cindex PowerPC64 multi-TOC
5693 @kindex --no-multi-toc
5694 @item --no-multi-toc
5695 By default, PowerPC64 GCC generates code for a TOC model where TOC
5696 entries are accessed with a 16-bit offset from r2. This limits the
5697 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
5698 grouping code sections such that each group uses less than 64K for its
5699 TOC entries, then inserts r2 adjusting stubs between inter-group
5700 calls. @command{ld} does not split apart input sections, so cannot
5701 help if a single input file has a @code{.toc} section that exceeds
5702 64K, most likely from linking multiple files with @command{ld -r}.
5703 Use this option to turn off this feature.
5717 @section @command{ld}'s Support for Various TI COFF Versions
5718 @cindex TI COFF versions
5719 @kindex --format=@var{version}
5720 The @samp{--format} switch allows selection of one of the various
5721 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
5722 also supported. The TI COFF versions also vary in header byte-order
5723 format; @command{ld} will read any version or byte order, but the output
5724 header format depends on the default specified by the specific target.
5737 @section @command{ld} and WIN32 (cygwin/mingw)
5739 This section describes some of the win32 specific @command{ld} issues.
5740 See @ref{Options,,Command Line Options} for detailed decription of the
5741 command line options mentioned here.
5744 @cindex import libraries
5745 @item import libraries
5746 The standard Windows linker creates and uses so-called import
5747 libraries, which contains information for linking to dll's. They are
5748 regular static archives and are handled as any other static
5749 archive. The cygwin and mingw ports of @command{ld} have specific
5750 support for creating such libraries provided with the
5751 @samp{--out-implib} command line option.
5753 @item exporting DLL symbols
5754 @cindex exporting DLL symbols
5755 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
5758 @item using auto-export functionality
5759 @cindex using auto-export functionality
5760 By default @command{ld} exports symbols with the auto-export functionality,
5761 which is controlled by the following command line options:
5764 @item --export-all-symbols [This is the default]
5765 @item --exclude-symbols
5766 @item --exclude-libs
5769 If, however, @samp{--export-all-symbols} is not given explicitly on the
5770 command line, then the default auto-export behavior will be @emph{disabled}
5771 if either of the following are true:
5774 @item A DEF file is used.
5775 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
5778 @item using a DEF file
5779 @cindex using a DEF file
5780 Another way of exporting symbols is using a DEF file. A DEF file is
5781 an ASCII file containing definitions of symbols which should be
5782 exported when a dll is created. Usually it is named @samp{<dll
5783 name>.def} and is added as any other object file to the linker's
5784 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
5787 gcc -o <output> <objectfiles> <dll name>.def
5790 Using a DEF file turns off the normal auto-export behavior, unless the
5791 @samp{--export-all-symbols} option is also used.
5793 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
5796 LIBRARY "xyz.dll" BASE=0x20000000
5802 another_foo = abc.dll.afoo
5806 This example defines a DLL with a non-default base address and five
5807 symbols in the export table. The third exported symbol @code{_bar} is an
5808 alias for the second. The fourth symbol, @code{another_foo} is resolved
5809 by "forwarding" to another module and treating it as an alias for
5810 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
5811 @code{var1} is declared to be a data object.
5813 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
5814 name of the output DLL. If @samp{<name>} does not include a suffix,
5815 the default library suffix, @samp{.DLL} is appended.
5817 When the .DEF file is used to build an application. rather than a
5818 library, the @code{NAME <name>} command shoud be used instead of
5819 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
5820 executable suffix, @samp{.EXE} is appended.
5822 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
5823 specification @code{BASE = <number>} may be used to specify a
5824 non-default base address for the image.
5826 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
5827 or they specify an empty string, the internal name is the same as the
5828 filename specified on the command line.
5830 The complete specification of an export symbol is:
5834 ( ( ( <name1> [ = <name2> ] )
5835 | ( <name1> = <module-name> . <external-name>))
5836 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
5839 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
5840 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
5841 @samp{<name1>} as a "forward" alias for the symbol
5842 @samp{<external-name>} in the DLL @samp{<module-name>}.
5843 Optionally, the symbol may be exported by the specified ordinal
5844 @samp{<integer>} alias.
5846 The optional keywords that follow the declaration indicate:
5848 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
5849 will still be exported by its ordinal alias (either the value specified
5850 by the .def specification or, otherwise, the value assigned by the
5851 linker). The symbol name, however, does remain visible in the import
5852 library (if any), unless @code{PRIVATE} is also specified.
5854 @code{DATA}: The symbol is a variable or object, rather than a function.
5855 The import lib will export only an indirect reference to @code{foo} as
5856 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
5859 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
5860 well as @code{_imp__foo} into the import library. Both refer to the
5861 read-only import address table's pointer to the variable, not to the
5862 variable itself. This can be dangerous. If the user code fails to add
5863 the @code{dllimport} attribute and also fails to explicitly add the
5864 extra indirection that the use of the attribute enforces, the
5865 application will behave unexpectedly.
5867 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
5868 it into the static import library used to resolve imports at link time. The
5869 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
5870 API at runtime or by by using the GNU ld extension of linking directly to
5871 the DLL without an import library.
5873 See ld/deffilep.y in the binutils sources for the full specification of
5874 other DEF file statements
5876 @cindex creating a DEF file
5877 While linking a shared dll, @command{ld} is able to create a DEF file
5878 with the @samp{--output-def <file>} command line option.
5880 @item Using decorations
5881 @cindex Using decorations
5882 Another way of marking symbols for export is to modify the source code
5883 itself, so that when building the DLL each symbol to be exported is
5887 __declspec(dllexport) int a_variable
5888 __declspec(dllexport) void a_function(int with_args)
5891 All such symbols will be exported from the DLL. If, however,
5892 any of the object files in the DLL contain symbols decorated in
5893 this way, then the normal auto-export behavior is disabled, unless
5894 the @samp{--export-all-symbols} option is also used.
5896 Note that object files that wish to access these symbols must @emph{not}
5897 decorate them with dllexport. Instead, they should use dllimport,
5901 __declspec(dllimport) int a_variable
5902 __declspec(dllimport) void a_function(int with_args)
5905 This complicates the structure of library header files, because
5906 when included by the library itself the header must declare the
5907 variables and functions as dllexport, but when included by client
5908 code the header must declare them as dllimport. There are a number
5909 of idioms that are typically used to do this; often client code can
5910 omit the __declspec() declaration completely. See
5911 @samp{--enable-auto-import} and @samp{automatic data imports} for more
5915 @cindex automatic data imports
5916 @item automatic data imports
5917 The standard Windows dll format supports data imports from dlls only
5918 by adding special decorations (dllimport/dllexport), which let the
5919 compiler produce specific assembler instructions to deal with this
5920 issue. This increases the effort necessary to port existing Un*x
5921 code to these platforms, especially for large
5922 c++ libraries and applications. The auto-import feature, which was
5923 initially provided by Paul Sokolovsky, allows one to omit the
5924 decorations to archieve a behavior that conforms to that on POSIX/Un*x
5925 platforms. This feature is enabled with the @samp{--enable-auto-import}
5926 command-line option, although it is enabled by default on cygwin/mingw.
5927 The @samp{--enable-auto-import} option itself now serves mainly to
5928 suppress any warnings that are ordinarily emitted when linked objects
5929 trigger the feature's use.
5931 auto-import of variables does not always work flawlessly without
5932 additional assistance. Sometimes, you will see this message
5934 "variable '<var>' can't be auto-imported. Please read the
5935 documentation for ld's @code{--enable-auto-import} for details."
5937 The @samp{--enable-auto-import} documentation explains why this error
5938 occurs, and several methods that can be used to overcome this difficulty.
5939 One of these methods is the @emph{runtime pseudo-relocs} feature, described
5942 @cindex runtime pseudo-relocation
5943 For complex variables imported from DLLs (such as structs or classes),
5944 object files typically contain a base address for the variable and an
5945 offset (@emph{addend}) within the variable--to specify a particular
5946 field or public member, for instance. Unfortunately, the runtime loader used
5947 in win32 environments is incapable of fixing these references at runtime
5948 without the additional information supplied by dllimport/dllexport decorations.
5949 The standard auto-import feature described above is unable to resolve these
5952 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
5953 be resolved without error, while leaving the task of adjusting the references
5954 themselves (with their non-zero addends) to specialized code provided by the
5955 runtime environment. Recent versions of the cygwin and mingw environments and
5956 compilers provide this runtime support; older versions do not. However, the
5957 support is only necessary on the developer's platform; the compiled result will
5958 run without error on an older system.
5960 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
5963 @cindex direct linking to a dll
5964 @item direct linking to a dll
5965 The cygwin/mingw ports of @command{ld} support the direct linking,
5966 including data symbols, to a dll without the usage of any import
5967 libraries. This is much faster and uses much less memory than does the
5968 traditional import library method, expecially when linking large
5969 libraries or applications. When @command{ld} creates an import lib, each
5970 function or variable exported from the dll is stored in its own bfd, even
5971 though a single bfd could contain many exports. The overhead involved in
5972 storing, loading, and processing so many bfd's is quite large, and explains the
5973 tremendous time, memory, and storage needed to link against particularly
5974 large or complex libraries when using import libs.
5976 Linking directly to a dll uses no extra command-line switches other than
5977 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
5978 of names to match each library. All that is needed from the developer's
5979 perspective is an understanding of this search, in order to force ld to
5980 select the dll instead of an import library.
5983 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
5984 to find, in the first directory of its search path,
5996 before moving on to the next directory in the search path.
5998 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
5999 where @samp{<prefix>} is set by the @command{ld} option
6000 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6001 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6004 Other win32-based unix environments, such as mingw or pw32, may use other
6005 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
6006 was originally intended to help avoid name conflicts among dll's built for the
6007 various win32/un*x environments, so that (for example) two versions of a zlib dll
6008 could coexist on the same machine.
6010 The generic cygwin/mingw path layout uses a @samp{bin} directory for
6011 applications and dll's and a @samp{lib} directory for the import
6012 libraries (using cygwin nomenclature):
6018 libxxx.dll.a (in case of dll's)
6019 libxxx.a (in case of static archive)
6022 Linking directly to a dll without using the import library can be
6025 1. Use the dll directly by adding the @samp{bin} path to the link line
6027 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
6030 However, as the dll's often have version numbers appended to their names
6031 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6032 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
6033 not versioned, and do not have this difficulty.
6035 2. Create a symbolic link from the dll to a file in the @samp{lib}
6036 directory according to the above mentioned search pattern. This
6037 should be used to avoid unwanted changes in the tools needed for
6041 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6044 Then you can link without any make environment changes.
6047 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
6050 This technique also avoids the version number problems, because the following is
6057 libxxx.dll.a -> ../bin/cygxxx-5.dll
6060 Linking directly to a dll without using an import lib will work
6061 even when auto-import features are exercised, and even when
6062 @samp{--enable-runtime-pseudo-relocs} is used.
6064 Given the improvements in speed and memory usage, one might justifiably
6065 wonder why import libraries are used at all. There are three reasons:
6067 1. Until recently, the link-directly-to-dll functionality did @emph{not}
6068 work with auto-imported data.
6070 2. Sometimes it is necessary to include pure static objects within the
6071 import library (which otherwise contains only bfd's for indirection
6072 symbols that point to the exports of a dll). Again, the import lib
6073 for the cygwin kernel makes use of this ability, and it is not
6074 possible to do this without an import lib.
6076 3. Symbol aliases can only be resolved using an import lib. This is
6077 critical when linking against OS-supplied dll's (eg, the win32 API)
6078 in which symbols are usually exported as undecorated aliases of their
6079 stdcall-decorated assembly names.
6081 So, import libs are not going away. But the ability to replace
6082 true import libs with a simple symbolic link to (or a copy of)
6083 a dll, in many cases, is a useful addition to the suite of tools
6084 binutils makes available to the win32 developer. Given the
6085 massive improvements in memory requirements during linking, storage
6086 requirements, and linking speed, we expect that many developers
6087 will soon begin to use this feature whenever possible.
6089 @item symbol aliasing
6091 @item adding additional names
6092 Sometimes, it is useful to export symbols with additional names.
6093 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6094 exported as @samp{_foo} by using special directives in the DEF file
6095 when creating the dll. This will affect also the optional created
6096 import library. Consider the following DEF file:
6099 LIBRARY "xyz.dll" BASE=0x61000000
6106 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6108 Another method for creating a symbol alias is to create it in the
6109 source code using the "weak" attribute:
6112 void foo () @{ /* Do something. */; @}
6113 void _foo () __attribute__ ((weak, alias ("foo")));
6116 See the gcc manual for more information about attributes and weak
6119 @item renaming symbols
6120 Sometimes it is useful to rename exports. For instance, the cygwin
6121 kernel does this regularly. A symbol @samp{_foo} can be exported as
6122 @samp{foo} but not as @samp{_foo} by using special directives in the
6123 DEF file. (This will also affect the import library, if it is
6124 created). In the following example:
6127 LIBRARY "xyz.dll" BASE=0x61000000
6133 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6137 Note: using a DEF file disables the default auto-export behavior,
6138 unless the @samp{--export-all-symbols} command line option is used.
6139 If, however, you are trying to rename symbols, then you should list
6140 @emph{all} desired exports in the DEF file, including the symbols
6141 that are not being renamed, and do @emph{not} use the
6142 @samp{--export-all-symbols} option. If you list only the
6143 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6144 to handle the other symbols, then the both the new names @emph{and}
6145 the original names for the renamed symbols will be exported.
6146 In effect, you'd be aliasing those symbols, not renaming them,
6147 which is probably not what you wanted.
6149 @cindex weak externals
6150 @item weak externals
6151 The Windows object format, PE, specifies a form of weak symbols called
6152 weak externals. When a weak symbol is linked and the symbol is not
6153 defined, the weak symbol becomes an alias for some other symbol. There
6154 are three variants of weak externals:
6156 @item Definition is searched for in objects and libraries, historically
6157 called lazy externals.
6158 @item Definition is searched for only in other objects, not in libraries.
6159 This form is not presently implemented.
6160 @item No search; the symbol is an alias. This form is not presently
6163 As a GNU extension, weak symbols that do not specify an alternate symbol
6164 are supported. If the symbol is undefined when linking, the symbol
6165 uses a default value.
6179 @section @code{ld} and Xtensa Processors
6181 @cindex Xtensa processors
6182 The default @command{ld} behavior for Xtensa processors is to interpret
6183 @code{SECTIONS} commands so that lists of explicitly named sections in a
6184 specification with a wildcard file will be interleaved when necessary to
6185 keep literal pools within the range of PC-relative load offsets. For
6186 example, with the command:
6198 @command{ld} may interleave some of the @code{.literal}
6199 and @code{.text} sections from different object files to ensure that the
6200 literal pools are within the range of PC-relative load offsets. A valid
6201 interleaving might place the @code{.literal} sections from an initial
6202 group of files followed by the @code{.text} sections of that group of
6203 files. Then, the @code{.literal} sections from the rest of the files
6204 and the @code{.text} sections from the rest of the files would follow.
6206 @cindex @option{--relax} on Xtensa
6207 @cindex relaxing on Xtensa
6208 Relaxation is enabled by default for the Xtensa version of @command{ld} and
6209 provides two important link-time optimizations. The first optimization
6210 is to combine identical literal values to reduce code size. A redundant
6211 literal will be removed and all the @code{L32R} instructions that use it
6212 will be changed to reference an identical literal, as long as the
6213 location of the replacement literal is within the offset range of all
6214 the @code{L32R} instructions. The second optimization is to remove
6215 unnecessary overhead from assembler-generated ``longcall'' sequences of
6216 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
6217 range of direct @code{CALL@var{n}} instructions.
6219 For each of these cases where an indirect call sequence can be optimized
6220 to a direct call, the linker will change the @code{CALLX@var{n}}
6221 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
6222 instruction, and remove the literal referenced by the @code{L32R}
6223 instruction if it is not used for anything else. Removing the
6224 @code{L32R} instruction always reduces code size but can potentially
6225 hurt performance by changing the alignment of subsequent branch targets.
6226 By default, the linker will always preserve alignments, either by
6227 switching some instructions between 24-bit encodings and the equivalent
6228 density instructions or by inserting a no-op in place of the @code{L32R}
6229 instruction that was removed. If code size is more important than
6230 performance, the @option{--size-opt} option can be used to prevent the
6231 linker from widening density instructions or inserting no-ops, except in
6232 a few cases where no-ops are required for correctness.
6234 The following Xtensa-specific command-line options can be used to
6237 @cindex Xtensa options
6241 Since the Xtensa version of @code{ld} enables the @option{--relax} option
6242 by default, the @option{--no-relax} option is provided to disable
6246 When optimizing indirect calls to direct calls, optimize for code size
6247 more than performance. With this option, the linker will not insert
6248 no-ops or widen density instructions to preserve branch target
6249 alignment. There may still be some cases where no-ops are required to
6250 preserve the correctness of the code.
6258 @ifclear SingleFormat
6263 @cindex object file management
6264 @cindex object formats available
6266 The linker accesses object and archive files using the BFD libraries.
6267 These libraries allow the linker to use the same routines to operate on
6268 object files whatever the object file format. A different object file
6269 format can be supported simply by creating a new BFD back end and adding
6270 it to the library. To conserve runtime memory, however, the linker and
6271 associated tools are usually configured to support only a subset of the
6272 object file formats available. You can use @code{objdump -i}
6273 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
6274 list all the formats available for your configuration.
6276 @cindex BFD requirements
6277 @cindex requirements for BFD
6278 As with most implementations, BFD is a compromise between
6279 several conflicting requirements. The major factor influencing
6280 BFD design was efficiency: any time used converting between
6281 formats is time which would not have been spent had BFD not
6282 been involved. This is partly offset by abstraction payback; since
6283 BFD simplifies applications and back ends, more time and care
6284 may be spent optimizing algorithms for a greater speed.
6286 One minor artifact of the BFD solution which you should bear in
6287 mind is the potential for information loss. There are two places where
6288 useful information can be lost using the BFD mechanism: during
6289 conversion and during output. @xref{BFD information loss}.
6292 * BFD outline:: How it works: an outline of BFD
6296 @section How It Works: An Outline of BFD
6297 @cindex opening object files
6298 @include bfdsumm.texi
6301 @node Reporting Bugs
6302 @chapter Reporting Bugs
6303 @cindex bugs in @command{ld}
6304 @cindex reporting bugs in @command{ld}
6306 Your bug reports play an essential role in making @command{ld} reliable.
6308 Reporting a bug may help you by bringing a solution to your problem, or
6309 it may not. But in any case the principal function of a bug report is
6310 to help the entire community by making the next version of @command{ld}
6311 work better. Bug reports are your contribution to the maintenance of
6314 In order for a bug report to serve its purpose, you must include the
6315 information that enables us to fix the bug.
6318 * Bug Criteria:: Have you found a bug?
6319 * Bug Reporting:: How to report bugs
6323 @section Have You Found a Bug?
6324 @cindex bug criteria
6326 If you are not sure whether you have found a bug, here are some guidelines:
6329 @cindex fatal signal
6330 @cindex linker crash
6331 @cindex crash of linker
6333 If the linker gets a fatal signal, for any input whatever, that is a
6334 @command{ld} bug. Reliable linkers never crash.
6336 @cindex error on valid input
6338 If @command{ld} produces an error message for valid input, that is a bug.
6340 @cindex invalid input
6342 If @command{ld} does not produce an error message for invalid input, that
6343 may be a bug. In the general case, the linker can not verify that
6344 object files are correct.
6347 If you are an experienced user of linkers, your suggestions for
6348 improvement of @command{ld} are welcome in any case.
6352 @section How to Report Bugs
6354 @cindex @command{ld} bugs, reporting
6356 A number of companies and individuals offer support for @sc{gnu}
6357 products. If you obtained @command{ld} from a support organization, we
6358 recommend you contact that organization first.
6360 You can find contact information for many support companies and
6361 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
6364 Otherwise, send bug reports for @command{ld} to
6365 @samp{bug-binutils@@gnu.org}.
6367 The fundamental principle of reporting bugs usefully is this:
6368 @strong{report all the facts}. If you are not sure whether to state a
6369 fact or leave it out, state it!
6371 Often people omit facts because they think they know what causes the
6372 problem and assume that some details do not matter. Thus, you might
6373 assume that the name of a symbol you use in an example does not
6374 matter. Well, probably it does not, but one cannot be sure. Perhaps
6375 the bug is a stray memory reference which happens to fetch from the
6376 location where that name is stored in memory; perhaps, if the name
6377 were different, the contents of that location would fool the linker
6378 into doing the right thing despite the bug. Play it safe and give a
6379 specific, complete example. That is the easiest thing for you to do,
6380 and the most helpful.
6382 Keep in mind that the purpose of a bug report is to enable us to fix
6383 the bug if it is new to us. Therefore, always write your bug reports
6384 on the assumption that the bug has not been reported previously.
6386 Sometimes people give a few sketchy facts and ask, ``Does this ring a
6387 bell?'' This cannot help us fix a bug, so it is basically useless. We
6388 respond by asking for enough details to enable us to investigate.
6389 You might as well expedite matters by sending them to begin with.
6391 To enable us to fix the bug, you should include all these things:
6395 The version of @command{ld}. @command{ld} announces it if you start it with
6396 the @samp{--version} argument.
6398 Without this, we will not know whether there is any point in looking for
6399 the bug in the current version of @command{ld}.
6402 Any patches you may have applied to the @command{ld} source, including any
6403 patches made to the @code{BFD} library.
6406 The type of machine you are using, and the operating system name and
6410 What compiler (and its version) was used to compile @command{ld}---e.g.
6414 The command arguments you gave the linker to link your example and
6415 observe the bug. To guarantee you will not omit something important,
6416 list them all. A copy of the Makefile (or the output from make) is
6419 If we were to try to guess the arguments, we would probably guess wrong
6420 and then we might not encounter the bug.
6423 A complete input file, or set of input files, that will reproduce the
6424 bug. It is generally most helpful to send the actual object files
6425 provided that they are reasonably small. Say no more than 10K. For
6426 bigger files you can either make them available by FTP or HTTP or else
6427 state that you are willing to send the object file(s) to whomever
6428 requests them. (Note - your email will be going to a mailing list, so
6429 we do not want to clog it up with large attachments). But small
6430 attachments are best.
6432 If the source files were assembled using @code{gas} or compiled using
6433 @code{gcc}, then it may be OK to send the source files rather than the
6434 object files. In this case, be sure to say exactly what version of
6435 @code{gas} or @code{gcc} was used to produce the object files. Also say
6436 how @code{gas} or @code{gcc} were configured.
6439 A description of what behavior you observe that you believe is
6440 incorrect. For example, ``It gets a fatal signal.''
6442 Of course, if the bug is that @command{ld} gets a fatal signal, then we
6443 will certainly notice it. But if the bug is incorrect output, we might
6444 not notice unless it is glaringly wrong. You might as well not give us
6445 a chance to make a mistake.
6447 Even if the problem you experience is a fatal signal, you should still
6448 say so explicitly. Suppose something strange is going on, such as, your
6449 copy of @command{ld} is out of synch, or you have encountered a bug in the
6450 C library on your system. (This has happened!) Your copy might crash
6451 and ours would not. If you told us to expect a crash, then when ours
6452 fails to crash, we would know that the bug was not happening for us. If
6453 you had not told us to expect a crash, then we would not be able to draw
6454 any conclusion from our observations.
6457 If you wish to suggest changes to the @command{ld} source, send us context
6458 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
6459 @samp{-p} option. Always send diffs from the old file to the new file.
6460 If you even discuss something in the @command{ld} source, refer to it by
6461 context, not by line number.
6463 The line numbers in our development sources will not match those in your
6464 sources. Your line numbers would convey no useful information to us.
6467 Here are some things that are not necessary:
6471 A description of the envelope of the bug.
6473 Often people who encounter a bug spend a lot of time investigating
6474 which changes to the input file will make the bug go away and which
6475 changes will not affect it.
6477 This is often time consuming and not very useful, because the way we
6478 will find the bug is by running a single example under the debugger
6479 with breakpoints, not by pure deduction from a series of examples.
6480 We recommend that you save your time for something else.
6482 Of course, if you can find a simpler example to report @emph{instead}
6483 of the original one, that is a convenience for us. Errors in the
6484 output will be easier to spot, running under the debugger will take
6485 less time, and so on.
6487 However, simplification is not vital; if you do not want to do this,
6488 report the bug anyway and send us the entire test case you used.
6491 A patch for the bug.
6493 A patch for the bug does help us if it is a good one. But do not omit
6494 the necessary information, such as the test case, on the assumption that
6495 a patch is all we need. We might see problems with your patch and decide
6496 to fix the problem another way, or we might not understand it at all.
6498 Sometimes with a program as complicated as @command{ld} it is very hard to
6499 construct an example that will make the program follow a certain path
6500 through the code. If you do not send us the example, we will not be
6501 able to construct one, so we will not be able to verify that the bug is
6504 And if we cannot understand what bug you are trying to fix, or why your
6505 patch should be an improvement, we will not install it. A test case will
6506 help us to understand.
6509 A guess about what the bug is or what it depends on.
6511 Such guesses are usually wrong. Even we cannot guess right about such
6512 things without first using the debugger to find the facts.
6516 @appendix MRI Compatible Script Files
6517 @cindex MRI compatibility
6518 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
6519 linker, @command{ld} can use MRI compatible linker scripts as an
6520 alternative to the more general-purpose linker scripting language
6521 described in @ref{Scripts}. MRI compatible linker scripts have a much
6522 simpler command set than the scripting language otherwise used with
6523 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
6524 linker commands; these commands are described here.
6526 In general, MRI scripts aren't of much use with the @code{a.out} object
6527 file format, since it only has three sections and MRI scripts lack some
6528 features to make use of them.
6530 You can specify a file containing an MRI-compatible script using the
6531 @samp{-c} command-line option.
6533 Each command in an MRI-compatible script occupies its own line; each
6534 command line starts with the keyword that identifies the command (though
6535 blank lines are also allowed for punctuation). If a line of an
6536 MRI-compatible script begins with an unrecognized keyword, @command{ld}
6537 issues a warning message, but continues processing the script.
6539 Lines beginning with @samp{*} are comments.
6541 You can write these commands using all upper-case letters, or all
6542 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
6543 The following list shows only the upper-case form of each command.
6546 @cindex @code{ABSOLUTE} (MRI)
6547 @item ABSOLUTE @var{secname}
6548 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
6549 Normally, @command{ld} includes in the output file all sections from all
6550 the input files. However, in an MRI-compatible script, you can use the
6551 @code{ABSOLUTE} command to restrict the sections that will be present in
6552 your output program. If the @code{ABSOLUTE} command is used at all in a
6553 script, then only the sections named explicitly in @code{ABSOLUTE}
6554 commands will appear in the linker output. You can still use other
6555 input sections (whatever you select on the command line, or using
6556 @code{LOAD}) to resolve addresses in the output file.
6558 @cindex @code{ALIAS} (MRI)
6559 @item ALIAS @var{out-secname}, @var{in-secname}
6560 Use this command to place the data from input section @var{in-secname}
6561 in a section called @var{out-secname} in the linker output file.
6563 @var{in-secname} may be an integer.
6565 @cindex @code{ALIGN} (MRI)
6566 @item ALIGN @var{secname} = @var{expression}
6567 Align the section called @var{secname} to @var{expression}. The
6568 @var{expression} should be a power of two.
6570 @cindex @code{BASE} (MRI)
6571 @item BASE @var{expression}
6572 Use the value of @var{expression} as the lowest address (other than
6573 absolute addresses) in the output file.
6575 @cindex @code{CHIP} (MRI)
6576 @item CHIP @var{expression}
6577 @itemx CHIP @var{expression}, @var{expression}
6578 This command does nothing; it is accepted only for compatibility.
6580 @cindex @code{END} (MRI)
6582 This command does nothing whatever; it's only accepted for compatibility.
6584 @cindex @code{FORMAT} (MRI)
6585 @item FORMAT @var{output-format}
6586 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
6587 language, but restricted to one of these output formats:
6591 S-records, if @var{output-format} is @samp{S}
6594 IEEE, if @var{output-format} is @samp{IEEE}
6597 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
6601 @cindex @code{LIST} (MRI)
6602 @item LIST @var{anything}@dots{}
6603 Print (to the standard output file) a link map, as produced by the
6604 @command{ld} command-line option @samp{-M}.
6606 The keyword @code{LIST} may be followed by anything on the
6607 same line, with no change in its effect.
6609 @cindex @code{LOAD} (MRI)
6610 @item LOAD @var{filename}
6611 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
6612 Include one or more object file @var{filename} in the link; this has the
6613 same effect as specifying @var{filename} directly on the @command{ld}
6616 @cindex @code{NAME} (MRI)
6617 @item NAME @var{output-name}
6618 @var{output-name} is the name for the program produced by @command{ld}; the
6619 MRI-compatible command @code{NAME} is equivalent to the command-line
6620 option @samp{-o} or the general script language command @code{OUTPUT}.
6622 @cindex @code{ORDER} (MRI)
6623 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
6624 @itemx ORDER @var{secname} @var{secname} @var{secname}
6625 Normally, @command{ld} orders the sections in its output file in the
6626 order in which they first appear in the input files. In an MRI-compatible
6627 script, you can override this ordering with the @code{ORDER} command. The
6628 sections you list with @code{ORDER} will appear first in your output
6629 file, in the order specified.
6631 @cindex @code{PUBLIC} (MRI)
6632 @item PUBLIC @var{name}=@var{expression}
6633 @itemx PUBLIC @var{name},@var{expression}
6634 @itemx PUBLIC @var{name} @var{expression}
6635 Supply a value (@var{expression}) for external symbol
6636 @var{name} used in the linker input files.
6638 @cindex @code{SECT} (MRI)
6639 @item SECT @var{secname}, @var{expression}
6640 @itemx SECT @var{secname}=@var{expression}
6641 @itemx SECT @var{secname} @var{expression}
6642 You can use any of these three forms of the @code{SECT} command to
6643 specify the start address (@var{expression}) for section @var{secname}.
6644 If you have more than one @code{SECT} statement for the same
6645 @var{secname}, only the @emph{first} sets the start address.
6651 @unnumbered LD Index
6656 % I think something like @colophon should be in texinfo. In the
6658 \long\def\colophon{\hbox to0pt{}\vfill
6659 \centerline{The body of this manual is set in}
6660 \centerline{\fontname\tenrm,}
6661 \centerline{with headings in {\bf\fontname\tenbf}}
6662 \centerline{and examples in {\tt\fontname\tentt}.}
6663 \centerline{{\it\fontname\tenit\/} and}
6664 \centerline{{\sl\fontname\tensl\/}}
6665 \centerline{are used for emphasis.}\vfill}
6667 % Blame: doc@cygnus.com, 28mar91.