3 @c Copyright (C) 1991-2016 Free Software Foundation, Inc.
6 @include configdoc.texi
7 @c (configdoc.texi is generated by the Makefile)
13 @macro gcctabopt{body}
19 @c Configure for the generation of man pages
45 @dircategory Software development
47 * Ld: (ld). The GNU linker.
52 This file documents the @sc{gnu} linker LD
53 @ifset VERSION_PACKAGE
54 @value{VERSION_PACKAGE}
56 version @value{VERSION}.
58 Copyright @copyright{} 1991-2016 Free Software Foundation, Inc.
60 Permission is granted to copy, distribute and/or modify this document
61 under the terms of the GNU Free Documentation License, Version 1.3
62 or any later version published by the Free Software Foundation;
63 with no Invariant Sections, with no Front-Cover Texts, and with no
64 Back-Cover Texts. A copy of the license is included in the
65 section entitled ``GNU Free Documentation License''.
69 @setchapternewpage odd
70 @settitle The GNU linker
75 @ifset VERSION_PACKAGE
76 @subtitle @value{VERSION_PACKAGE}
78 @subtitle Version @value{VERSION}
79 @author Steve Chamberlain
80 @author Ian Lance Taylor
85 \hfill Red Hat Inc\par
86 \hfill nickc\@credhat.com, doc\@redhat.com\par
87 \hfill {\it The GNU linker}\par
88 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
90 \global\parindent=0pt % Steve likes it this way.
93 @vskip 0pt plus 1filll
94 @c man begin COPYRIGHT
95 Copyright @copyright{} 1991-2016 Free Software Foundation, Inc.
97 Permission is granted to copy, distribute and/or modify this document
98 under the terms of the GNU Free Documentation License, Version 1.3
99 or any later version published by the Free Software Foundation;
100 with no Invariant Sections, with no Front-Cover Texts, and with no
101 Back-Cover Texts. A copy of the license is included in the
102 section entitled ``GNU Free Documentation License''.
108 @c FIXME: Talk about importance of *order* of args, cmds to linker!
113 This file documents the @sc{gnu} linker ld
114 @ifset VERSION_PACKAGE
115 @value{VERSION_PACKAGE}
117 version @value{VERSION}.
119 This document is distributed under the terms of the GNU Free
120 Documentation License version 1.3. A copy of the license is included
121 in the section entitled ``GNU Free Documentation License''.
124 * Overview:: Overview
125 * Invocation:: Invocation
126 * Scripts:: Linker Scripts
128 * Machine Dependent:: Machine Dependent Features
132 * H8/300:: ld and the H8/300
135 * Renesas:: ld and other Renesas micros
138 * i960:: ld and the Intel 960 family
141 * ARM:: ld and the ARM family
144 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
147 * HPPA ELF32:: ld and HPPA 32-bit ELF
150 * M68K:: ld and Motorola 68K family
153 * MIPS:: ld and MIPS family
156 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
159 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
162 * SPU ELF:: ld and SPU ELF Support
165 * TI COFF:: ld and the TI COFF
168 * Win32:: ld and WIN32 (cygwin/mingw)
171 * Xtensa:: ld and Xtensa Processors
174 @ifclear SingleFormat
177 @c Following blank line required for remaining bug in makeinfo conds/menus
179 * Reporting Bugs:: Reporting Bugs
180 * MRI:: MRI Compatible Script Files
181 * GNU Free Documentation License:: GNU Free Documentation License
182 * LD Index:: LD Index
189 @cindex @sc{gnu} linker
190 @cindex what is this?
193 @c man begin SYNOPSIS
194 ld [@b{options}] @var{objfile} @dots{}
198 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
199 the Info entries for @file{binutils} and
204 @c man begin DESCRIPTION
206 @command{ld} combines a number of object and archive files, relocates
207 their data and ties up symbol references. Usually the last step in
208 compiling a program is to run @command{ld}.
210 @command{ld} accepts Linker Command Language files written in
211 a superset of AT&T's Link Editor Command Language syntax,
212 to provide explicit and total control over the linking process.
216 This man page does not describe the command language; see the
217 @command{ld} entry in @code{info} for full details on the command
218 language and on other aspects of the GNU linker.
221 @ifclear SingleFormat
222 This version of @command{ld} uses the general purpose BFD libraries
223 to operate on object files. This allows @command{ld} to read, combine, and
224 write object files in many different formats---for example, COFF or
225 @code{a.out}. Different formats may be linked together to produce any
226 available kind of object file. @xref{BFD}, for more information.
229 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
230 linkers in providing diagnostic information. Many linkers abandon
231 execution immediately upon encountering an error; whenever possible,
232 @command{ld} continues executing, allowing you to identify other errors
233 (or, in some cases, to get an output file in spite of the error).
240 @c man begin DESCRIPTION
242 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
243 and to be as compatible as possible with other linkers. As a result,
244 you have many choices to control its behavior.
250 * Options:: Command Line Options
251 * Environment:: Environment Variables
255 @section Command Line Options
263 The linker supports a plethora of command-line options, but in actual
264 practice few of them are used in any particular context.
265 @cindex standard Unix system
266 For instance, a frequent use of @command{ld} is to link standard Unix
267 object files on a standard, supported Unix system. On such a system, to
268 link a file @code{hello.o}:
271 ld -o @var{output} /lib/crt0.o hello.o -lc
274 This tells @command{ld} to produce a file called @var{output} as the
275 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
276 the library @code{libc.a}, which will come from the standard search
277 directories. (See the discussion of the @samp{-l} option below.)
279 Some of the command-line options to @command{ld} may be specified at any
280 point in the command line. However, options which refer to files, such
281 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
282 which the option appears in the command line, relative to the object
283 files and other file options. Repeating non-file options with a
284 different argument will either have no further effect, or override prior
285 occurrences (those further to the left on the command line) of that
286 option. Options which may be meaningfully specified more than once are
287 noted in the descriptions below.
290 Non-option arguments are object files or archives which are to be linked
291 together. They may follow, precede, or be mixed in with command-line
292 options, except that an object file argument may not be placed between
293 an option and its argument.
295 Usually the linker is invoked with at least one object file, but you can
296 specify other forms of binary input files using @samp{-l}, @samp{-R},
297 and the script command language. If @emph{no} binary input files at all
298 are specified, the linker does not produce any output, and issues the
299 message @samp{No input files}.
301 If the linker cannot recognize the format of an object file, it will
302 assume that it is a linker script. A script specified in this way
303 augments the main linker script used for the link (either the default
304 linker script or the one specified by using @samp{-T}). This feature
305 permits the linker to link against a file which appears to be an object
306 or an archive, but actually merely defines some symbol values, or uses
307 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
308 script in this way merely augments the main linker script, with the
309 extra commands placed after the main script; use the @samp{-T} option
310 to replace the default linker script entirely, but note the effect of
311 the @code{INSERT} command. @xref{Scripts}.
313 For options whose names are a single letter,
314 option arguments must either follow the option letter without intervening
315 whitespace, or be given as separate arguments immediately following the
316 option that requires them.
318 For options whose names are multiple letters, either one dash or two can
319 precede the option name; for example, @samp{-trace-symbol} and
320 @samp{--trace-symbol} are equivalent. Note---there is one exception to
321 this rule. Multiple letter options that start with a lower case 'o' can
322 only be preceded by two dashes. This is to reduce confusion with the
323 @samp{-o} option. So for example @samp{-omagic} sets the output file
324 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
327 Arguments to multiple-letter options must either be separated from the
328 option name by an equals sign, or be given as separate arguments
329 immediately following the option that requires them. For example,
330 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
331 Unique abbreviations of the names of multiple-letter options are
334 Note---if the linker is being invoked indirectly, via a compiler driver
335 (e.g. @samp{gcc}) then all the linker command line options should be
336 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
337 compiler driver) like this:
340 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
343 This is important, because otherwise the compiler driver program may
344 silently drop the linker options, resulting in a bad link. Confusion
345 may also arise when passing options that require values through a
346 driver, as the use of a space between option and argument acts as
347 a separator, and causes the driver to pass only the option to the linker
348 and the argument to the compiler. In this case, it is simplest to use
349 the joined forms of both single- and multiple-letter options, such as:
352 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
355 Here is a table of the generic command line switches accepted by the GNU
359 @include at-file.texi
361 @kindex -a @var{keyword}
362 @item -a @var{keyword}
363 This option is supported for HP/UX compatibility. The @var{keyword}
364 argument must be one of the strings @samp{archive}, @samp{shared}, or
365 @samp{default}. @samp{-aarchive} is functionally equivalent to
366 @samp{-Bstatic}, and the other two keywords are functionally equivalent
367 to @samp{-Bdynamic}. This option may be used any number of times.
369 @kindex --audit @var{AUDITLIB}
370 @item --audit @var{AUDITLIB}
371 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
372 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
373 specified in the library. If specified multiple times @code{DT_AUDIT}
374 will contain a colon separated list of audit interfaces to use. If the linker
375 finds an object with an audit entry while searching for shared libraries,
376 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
377 This option is only meaningful on ELF platforms supporting the rtld-audit
381 @cindex architectures
382 @kindex -A @var{arch}
383 @item -A @var{architecture}
384 @kindex --architecture=@var{arch}
385 @itemx --architecture=@var{architecture}
386 In the current release of @command{ld}, this option is useful only for the
387 Intel 960 family of architectures. In that @command{ld} configuration, the
388 @var{architecture} argument identifies the particular architecture in
389 the 960 family, enabling some safeguards and modifying the
390 archive-library search path. @xref{i960,,@command{ld} and the Intel 960
391 family}, for details.
393 Future releases of @command{ld} may support similar functionality for
394 other architecture families.
397 @ifclear SingleFormat
398 @cindex binary input format
399 @kindex -b @var{format}
400 @kindex --format=@var{format}
403 @item -b @var{input-format}
404 @itemx --format=@var{input-format}
405 @command{ld} may be configured to support more than one kind of object
406 file. If your @command{ld} is configured this way, you can use the
407 @samp{-b} option to specify the binary format for input object files
408 that follow this option on the command line. Even when @command{ld} is
409 configured to support alternative object formats, you don't usually need
410 to specify this, as @command{ld} should be configured to expect as a
411 default input format the most usual format on each machine.
412 @var{input-format} is a text string, the name of a particular format
413 supported by the BFD libraries. (You can list the available binary
414 formats with @samp{objdump -i}.)
417 You may want to use this option if you are linking files with an unusual
418 binary format. You can also use @samp{-b} to switch formats explicitly (when
419 linking object files of different formats), by including
420 @samp{-b @var{input-format}} before each group of object files in a
423 The default format is taken from the environment variable
428 You can also define the input format from a script, using the command
431 see @ref{Format Commands}.
435 @kindex -c @var{MRI-cmdfile}
436 @kindex --mri-script=@var{MRI-cmdfile}
437 @cindex compatibility, MRI
438 @item -c @var{MRI-commandfile}
439 @itemx --mri-script=@var{MRI-commandfile}
440 For compatibility with linkers produced by MRI, @command{ld} accepts script
441 files written in an alternate, restricted command language, described in
443 @ref{MRI,,MRI Compatible Script Files}.
446 the MRI Compatible Script Files section of GNU ld documentation.
448 Introduce MRI script files with
449 the option @samp{-c}; use the @samp{-T} option to run linker
450 scripts written in the general-purpose @command{ld} scripting language.
451 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
452 specified by any @samp{-L} options.
454 @cindex common allocation
461 These three options are equivalent; multiple forms are supported for
462 compatibility with other linkers. They assign space to common symbols
463 even if a relocatable output file is specified (with @samp{-r}). The
464 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
465 @xref{Miscellaneous Commands}.
467 @kindex --depaudit @var{AUDITLIB}
468 @kindex -P @var{AUDITLIB}
469 @item --depaudit @var{AUDITLIB}
470 @itemx -P @var{AUDITLIB}
471 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
472 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
473 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
474 will contain a colon separated list of audit interfaces to use. This
475 option is only meaningful on ELF platforms supporting the rtld-audit interface.
476 The -P option is provided for Solaris compatibility.
478 @cindex entry point, from command line
479 @kindex -e @var{entry}
480 @kindex --entry=@var{entry}
482 @itemx --entry=@var{entry}
483 Use @var{entry} as the explicit symbol for beginning execution of your
484 program, rather than the default entry point. If there is no symbol
485 named @var{entry}, the linker will try to parse @var{entry} as a number,
486 and use that as the entry address (the number will be interpreted in
487 base 10; you may use a leading @samp{0x} for base 16, or a leading
488 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
489 and other ways of specifying the entry point.
491 @kindex --exclude-libs
492 @item --exclude-libs @var{lib},@var{lib},...
493 Specifies a list of archive libraries from which symbols should not be automatically
494 exported. The library names may be delimited by commas or colons. Specifying
495 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
496 automatic export. This option is available only for the i386 PE targeted
497 port of the linker and for ELF targeted ports. For i386 PE, symbols
498 explicitly listed in a .def file are still exported, regardless of this
499 option. For ELF targeted ports, symbols affected by this option will
500 be treated as hidden.
502 @kindex --exclude-modules-for-implib
503 @item --exclude-modules-for-implib @var{module},@var{module},...
504 Specifies a list of object files or archive members, from which symbols
505 should not be automatically exported, but which should be copied wholesale
506 into the import library being generated during the link. The module names
507 may be delimited by commas or colons, and must match exactly the filenames
508 used by @command{ld} to open the files; for archive members, this is simply
509 the member name, but for object files the name listed must include and
510 match precisely any path used to specify the input file on the linker's
511 command-line. This option is available only for the i386 PE targeted port
512 of the linker. Symbols explicitly listed in a .def file are still exported,
513 regardless of this option.
515 @cindex dynamic symbol table
517 @kindex --export-dynamic
518 @kindex --no-export-dynamic
520 @itemx --export-dynamic
521 @itemx --no-export-dynamic
522 When creating a dynamically linked executable, using the @option{-E}
523 option or the @option{--export-dynamic} option causes the linker to add
524 all symbols to the dynamic symbol table. The dynamic symbol table is the
525 set of symbols which are visible from dynamic objects at run time.
527 If you do not use either of these options (or use the
528 @option{--no-export-dynamic} option to restore the default behavior), the
529 dynamic symbol table will normally contain only those symbols which are
530 referenced by some dynamic object mentioned in the link.
532 If you use @code{dlopen} to load a dynamic object which needs to refer
533 back to the symbols defined by the program, rather than some other
534 dynamic object, then you will probably need to use this option when
535 linking the program itself.
537 You can also use the dynamic list to control what symbols should
538 be added to the dynamic symbol table if the output format supports it.
539 See the description of @samp{--dynamic-list}.
541 Note that this option is specific to ELF targeted ports. PE targets
542 support a similar function to export all symbols from a DLL or EXE; see
543 the description of @samp{--export-all-symbols} below.
545 @ifclear SingleFormat
546 @cindex big-endian objects
550 Link big-endian objects. This affects the default output format.
552 @cindex little-endian objects
555 Link little-endian objects. This affects the default output format.
558 @kindex -f @var{name}
559 @kindex --auxiliary=@var{name}
561 @itemx --auxiliary=@var{name}
562 When creating an ELF shared object, set the internal DT_AUXILIARY field
563 to the specified name. This tells the dynamic linker that the symbol
564 table of the shared object should be used as an auxiliary filter on the
565 symbol table of the shared object @var{name}.
567 If you later link a program against this filter object, then, when you
568 run the program, the dynamic linker will see the DT_AUXILIARY field. If
569 the dynamic linker resolves any symbols from the filter object, it will
570 first check whether there is a definition in the shared object
571 @var{name}. If there is one, it will be used instead of the definition
572 in the filter object. The shared object @var{name} need not exist.
573 Thus the shared object @var{name} may be used to provide an alternative
574 implementation of certain functions, perhaps for debugging or for
575 machine specific performance.
577 This option may be specified more than once. The DT_AUXILIARY entries
578 will be created in the order in which they appear on the command line.
580 @kindex -F @var{name}
581 @kindex --filter=@var{name}
583 @itemx --filter=@var{name}
584 When creating an ELF shared object, set the internal DT_FILTER field to
585 the specified name. This tells the dynamic linker that the symbol table
586 of the shared object which is being created should be used as a filter
587 on the symbol table of the shared object @var{name}.
589 If you later link a program against this filter object, then, when you
590 run the program, the dynamic linker will see the DT_FILTER field. The
591 dynamic linker will resolve symbols according to the symbol table of the
592 filter object as usual, but it will actually link to the definitions
593 found in the shared object @var{name}. Thus the filter object can be
594 used to select a subset of the symbols provided by the object
597 Some older linkers used the @option{-F} option throughout a compilation
598 toolchain for specifying object-file format for both input and output
600 @ifclear SingleFormat
601 The @sc{gnu} linker uses other mechanisms for this purpose: the
602 @option{-b}, @option{--format}, @option{--oformat} options, the
603 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
604 environment variable.
606 The @sc{gnu} linker will ignore the @option{-F} option when not
607 creating an ELF shared object.
609 @cindex finalization function
610 @kindex -fini=@var{name}
611 @item -fini=@var{name}
612 When creating an ELF executable or shared object, call NAME when the
613 executable or shared object is unloaded, by setting DT_FINI to the
614 address of the function. By default, the linker uses @code{_fini} as
615 the function to call.
619 Ignored. Provided for compatibility with other tools.
621 @kindex -G @var{value}
622 @kindex --gpsize=@var{value}
625 @itemx --gpsize=@var{value}
626 Set the maximum size of objects to be optimized using the GP register to
627 @var{size}. This is only meaningful for object file formats such as
628 MIPS ELF that support putting large and small objects into different
629 sections. This is ignored for other object file formats.
631 @cindex runtime library name
632 @kindex -h @var{name}
633 @kindex -soname=@var{name}
635 @itemx -soname=@var{name}
636 When creating an ELF shared object, set the internal DT_SONAME field to
637 the specified name. When an executable is linked with a shared object
638 which has a DT_SONAME field, then when the executable is run the dynamic
639 linker will attempt to load the shared object specified by the DT_SONAME
640 field rather than the using the file name given to the linker.
643 @cindex incremental link
645 Perform an incremental link (same as option @samp{-r}).
647 @cindex initialization function
648 @kindex -init=@var{name}
649 @item -init=@var{name}
650 When creating an ELF executable or shared object, call NAME when the
651 executable or shared object is loaded, by setting DT_INIT to the address
652 of the function. By default, the linker uses @code{_init} as the
655 @cindex archive files, from cmd line
656 @kindex -l @var{namespec}
657 @kindex --library=@var{namespec}
658 @item -l @var{namespec}
659 @itemx --library=@var{namespec}
660 Add the archive or object file specified by @var{namespec} to the
661 list of files to link. This option may be used any number of times.
662 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
663 will search the library path for a file called @var{filename}, otherwise it
664 will search the library path for a file called @file{lib@var{namespec}.a}.
666 On systems which support shared libraries, @command{ld} may also search for
667 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
668 and SunOS systems, @command{ld} will search a directory for a library
669 called @file{lib@var{namespec}.so} before searching for one called
670 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
671 indicates a shared library.) Note that this behavior does not apply
672 to @file{:@var{filename}}, which always specifies a file called
675 The linker will search an archive only once, at the location where it is
676 specified on the command line. If the archive defines a symbol which
677 was undefined in some object which appeared before the archive on the
678 command line, the linker will include the appropriate file(s) from the
679 archive. However, an undefined symbol in an object appearing later on
680 the command line will not cause the linker to search the archive again.
682 See the @option{-(} option for a way to force the linker to search
683 archives multiple times.
685 You may list the same archive multiple times on the command line.
688 This type of archive searching is standard for Unix linkers. However,
689 if you are using @command{ld} on AIX, note that it is different from the
690 behaviour of the AIX linker.
693 @cindex search directory, from cmd line
695 @kindex --library-path=@var{dir}
696 @item -L @var{searchdir}
697 @itemx --library-path=@var{searchdir}
698 Add path @var{searchdir} to the list of paths that @command{ld} will search
699 for archive libraries and @command{ld} control scripts. You may use this
700 option any number of times. The directories are searched in the order
701 in which they are specified on the command line. Directories specified
702 on the command line are searched before the default directories. All
703 @option{-L} options apply to all @option{-l} options, regardless of the
704 order in which the options appear. @option{-L} options do not affect
705 how @command{ld} searches for a linker script unless @option{-T}
708 If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
709 by the @dfn{sysroot prefix}, controlled by the @samp{--sysroot} option, or
710 specified when the linker is configured.
713 The default set of paths searched (without being specified with
714 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
715 some cases also on how it was configured. @xref{Environment}.
718 The paths can also be specified in a link script with the
719 @code{SEARCH_DIR} command. Directories specified this way are searched
720 at the point in which the linker script appears in the command line.
723 @kindex -m @var{emulation}
724 @item -m @var{emulation}
725 Emulate the @var{emulation} linker. You can list the available
726 emulations with the @samp{--verbose} or @samp{-V} options.
728 If the @samp{-m} option is not used, the emulation is taken from the
729 @code{LDEMULATION} environment variable, if that is defined.
731 Otherwise, the default emulation depends upon how the linker was
739 Print a link map to the standard output. A link map provides
740 information about the link, including the following:
744 Where object files are mapped into memory.
746 How common symbols are allocated.
748 All archive members included in the link, with a mention of the symbol
749 which caused the archive member to be brought in.
751 The values assigned to symbols.
753 Note - symbols whose values are computed by an expression which
754 involves a reference to a previous value of the same symbol may not
755 have correct result displayed in the link map. This is because the
756 linker discards intermediate results and only retains the final value
757 of an expression. Under such circumstances the linker will display
758 the final value enclosed by square brackets. Thus for example a
759 linker script containing:
767 will produce the following output in the link map if the @option{-M}
772 [0x0000000c] foo = (foo * 0x4)
773 [0x0000000c] foo = (foo + 0x8)
776 See @ref{Expressions} for more information about expressions in linker
781 @cindex read-only text
786 Turn off page alignment of sections, and disable linking against shared
787 libraries. If the output format supports Unix style magic numbers,
788 mark the output as @code{NMAGIC}.
792 @cindex read/write from cmd line
796 Set the text and data sections to be readable and writable. Also, do
797 not page-align the data segment, and disable linking against shared
798 libraries. If the output format supports Unix style magic numbers,
799 mark the output as @code{OMAGIC}. Note: Although a writable text section
800 is allowed for PE-COFF targets, it does not conform to the format
801 specification published by Microsoft.
806 This option negates most of the effects of the @option{-N} option. It
807 sets the text section to be read-only, and forces the data segment to
808 be page-aligned. Note - this option does not enable linking against
809 shared libraries. Use @option{-Bdynamic} for this.
811 @kindex -o @var{output}
812 @kindex --output=@var{output}
813 @cindex naming the output file
814 @item -o @var{output}
815 @itemx --output=@var{output}
816 Use @var{output} as the name for the program produced by @command{ld}; if this
817 option is not specified, the name @file{a.out} is used by default. The
818 script command @code{OUTPUT} can also specify the output file name.
820 @kindex -O @var{level}
821 @cindex generating optimized output
823 If @var{level} is a numeric values greater than zero @command{ld} optimizes
824 the output. This might take significantly longer and therefore probably
825 should only be enabled for the final binary. At the moment this
826 option only affects ELF shared library generation. Future releases of
827 the linker may make more use of this option. Also currently there is
828 no difference in the linker's behaviour for different non-zero values
829 of this option. Again this may change with future releases.
832 @cindex push state governing input file handling
834 The @option{--push-state} allows to preserve the current state of the
835 flags which govern the input file handling so that they can all be
836 restored with one corresponding @option{--pop-state} option.
838 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
839 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
840 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
841 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
842 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
843 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
845 One target for this option are specifications for @file{pkg-config}. When
846 used with the @option{--libs} option all possibly needed libraries are
847 listed and then possibly linked with all the time. It is better to return
848 something as follows:
851 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
855 @cindex pop state governing input file handling
856 Undoes the effect of --push-state, restores the previous values of the
857 flags governing input file handling.
860 @kindex --emit-relocs
861 @cindex retain relocations in final executable
864 Leave relocation sections and contents in fully linked executables.
865 Post link analysis and optimization tools may need this information in
866 order to perform correct modifications of executables. This results
867 in larger executables.
869 This option is currently only supported on ELF platforms.
871 @kindex --force-dynamic
872 @cindex forcing the creation of dynamic sections
873 @item --force-dynamic
874 Force the output file to have dynamic sections. This option is specific
878 @cindex relocatable output
880 @kindex --relocatable
883 Generate relocatable output---i.e., generate an output file that can in
884 turn serve as input to @command{ld}. This is often called @dfn{partial
885 linking}. As a side effect, in environments that support standard Unix
886 magic numbers, this option also sets the output file's magic number to
888 @c ; see @option{-N}.
889 If this option is not specified, an absolute file is produced. When
890 linking C++ programs, this option @emph{will not} resolve references to
891 constructors; to do that, use @samp{-Ur}.
893 When an input file does not have the same format as the output file,
894 partial linking is only supported if that input file does not contain any
895 relocations. Different output formats can have further restrictions; for
896 example some @code{a.out}-based formats do not support partial linking
897 with input files in other formats at all.
899 This option does the same thing as @samp{-i}.
901 @kindex -R @var{file}
902 @kindex --just-symbols=@var{file}
903 @cindex symbol-only input
904 @item -R @var{filename}
905 @itemx --just-symbols=@var{filename}
906 Read symbol names and their addresses from @var{filename}, but do not
907 relocate it or include it in the output. This allows your output file
908 to refer symbolically to absolute locations of memory defined in other
909 programs. You may use this option more than once.
911 For compatibility with other ELF linkers, if the @option{-R} option is
912 followed by a directory name, rather than a file name, it is treated as
913 the @option{-rpath} option.
917 @cindex strip all symbols
920 Omit all symbol information from the output file.
923 @kindex --strip-debug
924 @cindex strip debugger symbols
927 Omit debugger symbol information (but not all symbols) from the output file.
931 @cindex input files, displaying
934 Print the names of the input files as @command{ld} processes them.
936 @kindex -T @var{script}
937 @kindex --script=@var{script}
939 @item -T @var{scriptfile}
940 @itemx --script=@var{scriptfile}
941 Use @var{scriptfile} as the linker script. This script replaces
942 @command{ld}'s default linker script (rather than adding to it), so
943 @var{commandfile} must specify everything necessary to describe the
944 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
945 the current directory, @code{ld} looks for it in the directories
946 specified by any preceding @samp{-L} options. Multiple @samp{-T}
949 @kindex -dT @var{script}
950 @kindex --default-script=@var{script}
952 @item -dT @var{scriptfile}
953 @itemx --default-script=@var{scriptfile}
954 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
956 This option is similar to the @option{--script} option except that
957 processing of the script is delayed until after the rest of the
958 command line has been processed. This allows options placed after the
959 @option{--default-script} option on the command line to affect the
960 behaviour of the linker script, which can be important when the linker
961 command line cannot be directly controlled by the user. (eg because
962 the command line is being constructed by another tool, such as
965 @kindex -u @var{symbol}
966 @kindex --undefined=@var{symbol}
967 @cindex undefined symbol
968 @item -u @var{symbol}
969 @itemx --undefined=@var{symbol}
970 Force @var{symbol} to be entered in the output file as an undefined
971 symbol. Doing this may, for example, trigger linking of additional
972 modules from standard libraries. @samp{-u} may be repeated with
973 different option arguments to enter additional undefined symbols. This
974 option is equivalent to the @code{EXTERN} linker script command.
976 If this option is being used to force additional modules to be pulled
977 into the link, and if it is an error for the symbol to remain
978 undefined, then the option @option{--require-defined} should be used
981 @kindex --require-defined=@var{symbol}
982 @cindex symbols, require defined
983 @cindex defined symbol
984 @item --require-defined=@var{symbol}
985 Require that @var{symbol} is defined in the output file. This option
986 is the same as option @option{--undefined} except that if @var{symbol}
987 is not defined in the output file then the linker will issue an error
988 and exit. The same effect can be achieved in a linker script by using
989 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
990 can be used multiple times to require additional symbols.
995 For anything other than C++ programs, this option is equivalent to
996 @samp{-r}: it generates relocatable output---i.e., an output file that can in
997 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
998 @emph{does} resolve references to constructors, unlike @samp{-r}.
999 It does not work to use @samp{-Ur} on files that were themselves linked
1000 with @samp{-Ur}; once the constructor table has been built, it cannot
1001 be added to. Use @samp{-Ur} only for the last partial link, and
1002 @samp{-r} for the others.
1004 @kindex --orphan-handling=@var{MODE}
1005 @cindex orphan sections
1006 @cindex sections, orphan
1007 @item --orphan-handling=@var{MODE}
1008 Control how orphan sections are handled. An orphan section is one not
1009 specifically mentioned in a linker script. @xref{Orphan Sections}.
1011 @var{MODE} can have any of the following values:
1015 Orphan sections are placed into a suitable output section following
1016 the strategy described in @ref{Orphan Sections}. The option
1017 @samp{--unique} also effects how sections are placed.
1020 All orphan sections are discarded, by placing them in the
1021 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1024 The linker will place the orphan section as for @code{place} and also
1028 The linker will exit with an error if any orphan section is found.
1031 The default if @samp{--orphan-handling} is not given is @code{place}.
1033 @kindex --unique[=@var{SECTION}]
1034 @item --unique[=@var{SECTION}]
1035 Creates a separate output section for every input section matching
1036 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1037 missing, for every orphan input section. An orphan section is one not
1038 specifically mentioned in a linker script. You may use this option
1039 multiple times on the command line; It prevents the normal merging of
1040 input sections with the same name, overriding output section assignments
1050 Display the version number for @command{ld}. The @option{-V} option also
1051 lists the supported emulations.
1054 @kindex --discard-all
1055 @cindex deleting local symbols
1057 @itemx --discard-all
1058 Delete all local symbols.
1061 @kindex --discard-locals
1062 @cindex local symbols, deleting
1064 @itemx --discard-locals
1065 Delete all temporary local symbols. (These symbols start with
1066 system-specific local label prefixes, typically @samp{.L} for ELF systems
1067 or @samp{L} for traditional a.out systems.)
1069 @kindex -y @var{symbol}
1070 @kindex --trace-symbol=@var{symbol}
1071 @cindex symbol tracing
1072 @item -y @var{symbol}
1073 @itemx --trace-symbol=@var{symbol}
1074 Print the name of each linked file in which @var{symbol} appears. This
1075 option may be given any number of times. On many systems it is necessary
1076 to prepend an underscore.
1078 This option is useful when you have an undefined symbol in your link but
1079 don't know where the reference is coming from.
1081 @kindex -Y @var{path}
1083 Add @var{path} to the default library search path. This option exists
1084 for Solaris compatibility.
1086 @kindex -z @var{keyword}
1087 @item -z @var{keyword}
1088 The recognized keywords are:
1092 Combines multiple reloc sections and sorts them to make dynamic symbol
1093 lookup caching possible.
1096 Generate common symbols with the STT_COMMON type druing a relocatable
1100 Disallows undefined symbols in object files. Undefined symbols in
1101 shared libraries are still allowed.
1104 Marks the object as requiring executable stack.
1107 This option is only meaningful when building a shared object. It makes
1108 the symbols defined by this shared object available for symbol resolution
1109 of subsequently loaded libraries.
1112 This option is only meaningful when building a shared object.
1113 It marks the object so that its runtime initialization will occur
1114 before the runtime initialization of any other objects brought into
1115 the process at the same time. Similarly the runtime finalization of
1116 the object will occur after the runtime finalization of any other
1120 Marks the object that its symbol table interposes before all symbols
1121 but the primary executable.
1124 When generating an executable or shared library, mark it to tell the
1125 dynamic linker to defer function call resolution to the point when
1126 the function is called (lazy binding), rather than at load time.
1127 Lazy binding is the default.
1130 Marks the object that its filters be processed immediately at
1134 Allows multiple definitions.
1137 Disables multiple reloc sections combining.
1140 Generate common symbols with the STT_OBJECT type druing a relocatable
1144 Disable linker generated .dynbss variables used in place of variables
1145 defined in shared libraries. May result in dynamic text relocations.
1148 Marks the object that the search for dependencies of this object will
1149 ignore any default library search paths.
1152 Marks the object shouldn't be unloaded at runtime.
1155 Marks the object not available to @code{dlopen}.
1158 Marks the object can not be dumped by @code{dldump}.
1161 Marks the object as not requiring executable stack.
1164 Treat DT_TEXTREL in shared object as error.
1167 Don't treat DT_TEXTREL in shared object as error.
1170 Don't treat DT_TEXTREL in shared object as error.
1173 Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1176 When generating an executable or shared library, mark it to tell the
1177 dynamic linker to resolve all symbols when the program is started, or
1178 when the shared library is linked to using dlopen, instead of
1179 deferring function call resolution to the point when the function is
1183 Marks the object may contain $ORIGIN.
1186 Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1188 @item max-page-size=@var{value}
1189 Set the emulation maximum page size to @var{value}.
1191 @item common-page-size=@var{value}
1192 Set the emulation common page size to @var{value}.
1194 @item stack-size=@var{value}
1195 Specify a stack size for in an ELF @code{PT_GNU_STACK} segment.
1196 Specifying zero will override any default non-zero sized
1197 @code{PT_GNU_STACK} segment creation.
1200 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1202 @item noextern-protected-data
1203 Don't treat protected data symbol as external when building shared
1204 library. This option overrides linker backend default. It can be used
1205 to workaround incorrect relocations against protected data symbols
1206 generated by compiler. Updates on protected data symbols by another
1207 module aren't visible to the resulting shared library. Supported for
1210 @item nodynamic-undefined-weak
1211 Don't treat undefined weak symbols as dynamic when building executable.
1212 This option overrides linker backend default. It can be used to avoid
1213 dynamic relocations against undefined weak symbols in executable.
1214 Supported for i386 and x86-64.
1216 @item noreloc-overflow
1217 Disable relocation overflow check. This can be used to disable
1218 relocation overflow check if there will be no dynamic relocation
1219 overflow at run-time. Supported for x86_64.
1221 @item call-nop=prefix-addr
1222 @itemx call-nop=prefix-nop
1223 @itemx call-nop=suffix-nop
1224 @itemx call-nop=prefix-@var{byte}
1225 @itemx call-nop=suffix-@var{byte}
1226 Specify the 1-byte @code{NOP} padding when transforming indirect call
1227 to a locally defined function, foo, via its GOT slot.
1228 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1229 @option{call-nop=prefix-nop} generates @code{0x90 call foo}.
1230 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1231 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1232 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1233 Supported for i386 and x86_64.
1237 Other keywords are ignored for Solaris compatibility.
1240 @cindex groups of archives
1241 @item -( @var{archives} -)
1242 @itemx --start-group @var{archives} --end-group
1243 The @var{archives} should be a list of archive files. They may be
1244 either explicit file names, or @samp{-l} options.
1246 The specified archives are searched repeatedly until no new undefined
1247 references are created. Normally, an archive is searched only once in
1248 the order that it is specified on the command line. If a symbol in that
1249 archive is needed to resolve an undefined symbol referred to by an
1250 object in an archive that appears later on the command line, the linker
1251 would not be able to resolve that reference. By grouping the archives,
1252 they all be searched repeatedly until all possible references are
1255 Using this option has a significant performance cost. It is best to use
1256 it only when there are unavoidable circular references between two or
1259 @kindex --accept-unknown-input-arch
1260 @kindex --no-accept-unknown-input-arch
1261 @item --accept-unknown-input-arch
1262 @itemx --no-accept-unknown-input-arch
1263 Tells the linker to accept input files whose architecture cannot be
1264 recognised. The assumption is that the user knows what they are doing
1265 and deliberately wants to link in these unknown input files. This was
1266 the default behaviour of the linker, before release 2.14. The default
1267 behaviour from release 2.14 onwards is to reject such input files, and
1268 so the @samp{--accept-unknown-input-arch} option has been added to
1269 restore the old behaviour.
1272 @kindex --no-as-needed
1274 @itemx --no-as-needed
1275 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1276 on the command line after the @option{--as-needed} option. Normally
1277 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1278 on the command line, regardless of whether the library is actually
1279 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1280 emitted for a library that @emph{at that point in the link} satisfies a
1281 non-weak undefined symbol reference from a regular object file or, if
1282 the library is not found in the DT_NEEDED lists of other needed libraries, a
1283 non-weak undefined symbol reference from another needed dynamic library.
1284 Object files or libraries appearing on the command line @emph{after}
1285 the library in question do not affect whether the library is seen as
1286 needed. This is similar to the rules for extraction of object files
1287 from archives. @option{--no-as-needed} restores the default behaviour.
1289 @kindex --add-needed
1290 @kindex --no-add-needed
1292 @itemx --no-add-needed
1293 These two options have been deprecated because of the similarity of
1294 their names to the @option{--as-needed} and @option{--no-as-needed}
1295 options. They have been replaced by @option{--copy-dt-needed-entries}
1296 and @option{--no-copy-dt-needed-entries}.
1298 @kindex -assert @var{keyword}
1299 @item -assert @var{keyword}
1300 This option is ignored for SunOS compatibility.
1304 @kindex -call_shared
1308 Link against dynamic libraries. This is only meaningful on platforms
1309 for which shared libraries are supported. This option is normally the
1310 default on such platforms. The different variants of this option are
1311 for compatibility with various systems. You may use this option
1312 multiple times on the command line: it affects library searching for
1313 @option{-l} options which follow it.
1317 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1318 section. This causes the runtime linker to handle lookups in this
1319 object and its dependencies to be performed only inside the group.
1320 @option{--unresolved-symbols=report-all} is implied. This option is
1321 only meaningful on ELF platforms which support shared libraries.
1331 Do not link against shared libraries. This is only meaningful on
1332 platforms for which shared libraries are supported. The different
1333 variants of this option are for compatibility with various systems. You
1334 may use this option multiple times on the command line: it affects
1335 library searching for @option{-l} options which follow it. This
1336 option also implies @option{--unresolved-symbols=report-all}. This
1337 option can be used with @option{-shared}. Doing so means that a
1338 shared library is being created but that all of the library's external
1339 references must be resolved by pulling in entries from static
1344 When creating a shared library, bind references to global symbols to the
1345 definition within the shared library, if any. Normally, it is possible
1346 for a program linked against a shared library to override the definition
1347 within the shared library. This option can also be used with the
1348 @option{--export-dynamic} option, when creating a position independent
1349 executable, to bind references to global symbols to the definition within
1350 the executable. This option is only meaningful on ELF platforms which
1351 support shared libraries and position independent executables.
1353 @kindex -Bsymbolic-functions
1354 @item -Bsymbolic-functions
1355 When creating a shared library, bind references to global function
1356 symbols to the definition within the shared library, if any.
1357 This option can also be used with the @option{--export-dynamic} option,
1358 when creating a position independent executable, to bind references
1359 to global function symbols to the definition within the executable.
1360 This option is only meaningful on ELF platforms which support shared
1361 libraries and position independent executables.
1363 @kindex --dynamic-list=@var{dynamic-list-file}
1364 @item --dynamic-list=@var{dynamic-list-file}
1365 Specify the name of a dynamic list file to the linker. This is
1366 typically used when creating shared libraries to specify a list of
1367 global symbols whose references shouldn't be bound to the definition
1368 within the shared library, or creating dynamically linked executables
1369 to specify a list of symbols which should be added to the symbol table
1370 in the executable. This option is only meaningful on ELF platforms
1371 which support shared libraries.
1373 The format of the dynamic list is the same as the version node without
1374 scope and node name. See @ref{VERSION} for more information.
1376 @kindex --dynamic-list-data
1377 @item --dynamic-list-data
1378 Include all global data symbols to the dynamic list.
1380 @kindex --dynamic-list-cpp-new
1381 @item --dynamic-list-cpp-new
1382 Provide the builtin dynamic list for C++ operator new and delete. It
1383 is mainly useful for building shared libstdc++.
1385 @kindex --dynamic-list-cpp-typeinfo
1386 @item --dynamic-list-cpp-typeinfo
1387 Provide the builtin dynamic list for C++ runtime type identification.
1389 @kindex --check-sections
1390 @kindex --no-check-sections
1391 @item --check-sections
1392 @itemx --no-check-sections
1393 Asks the linker @emph{not} to check section addresses after they have
1394 been assigned to see if there are any overlaps. Normally the linker will
1395 perform this check, and if it finds any overlaps it will produce
1396 suitable error messages. The linker does know about, and does make
1397 allowances for sections in overlays. The default behaviour can be
1398 restored by using the command line switch @option{--check-sections}.
1399 Section overlap is not usually checked for relocatable links. You can
1400 force checking in that case by using the @option{--check-sections}
1403 @kindex --copy-dt-needed-entries
1404 @kindex --no-copy-dt-needed-entries
1405 @item --copy-dt-needed-entries
1406 @itemx --no-copy-dt-needed-entries
1407 This option affects the treatment of dynamic libraries referred to
1408 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1409 command line. Normally the linker won't add a DT_NEEDED tag to the
1410 output binary for each library mentioned in a DT_NEEDED tag in an
1411 input dynamic library. With @option{--copy-dt-needed-entries}
1412 specified on the command line however any dynamic libraries that
1413 follow it will have their DT_NEEDED entries added. The default
1414 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1416 This option also has an effect on the resolution of symbols in dynamic
1417 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1418 mentioned on the command line will be recursively searched, following
1419 their DT_NEEDED tags to other libraries, in order to resolve symbols
1420 required by the output binary. With the default setting however
1421 the searching of dynamic libraries that follow it will stop with the
1422 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1425 @cindex cross reference table
1428 Output a cross reference table. If a linker map file is being
1429 generated, the cross reference table is printed to the map file.
1430 Otherwise, it is printed on the standard output.
1432 The format of the table is intentionally simple, so that it may be
1433 easily processed by a script if necessary. The symbols are printed out,
1434 sorted by name. For each symbol, a list of file names is given. If the
1435 symbol is defined, the first file listed is the location of the
1436 definition. If the symbol is defined as a common value then any files
1437 where this happens appear next. Finally any files that reference the
1440 @cindex common allocation
1441 @kindex --no-define-common
1442 @item --no-define-common
1443 This option inhibits the assignment of addresses to common symbols.
1444 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1445 @xref{Miscellaneous Commands}.
1447 The @samp{--no-define-common} option allows decoupling
1448 the decision to assign addresses to Common symbols from the choice
1449 of the output file type; otherwise a non-Relocatable output type
1450 forces assigning addresses to Common symbols.
1451 Using @samp{--no-define-common} allows Common symbols that are referenced
1452 from a shared library to be assigned addresses only in the main program.
1453 This eliminates the unused duplicate space in the shared library,
1454 and also prevents any possible confusion over resolving to the wrong
1455 duplicate when there are many dynamic modules with specialized search
1456 paths for runtime symbol resolution.
1458 @cindex symbols, from command line
1459 @kindex --defsym=@var{symbol}=@var{exp}
1460 @item --defsym=@var{symbol}=@var{expression}
1461 Create a global symbol in the output file, containing the absolute
1462 address given by @var{expression}. You may use this option as many
1463 times as necessary to define multiple symbols in the command line. A
1464 limited form of arithmetic is supported for the @var{expression} in this
1465 context: you may give a hexadecimal constant or the name of an existing
1466 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1467 constants or symbols. If you need more elaborate expressions, consider
1468 using the linker command language from a script (@pxref{Assignments}).
1469 @emph{Note:} there should be no white space between @var{symbol}, the
1470 equals sign (``@key{=}''), and @var{expression}.
1472 @cindex demangling, from command line
1473 @kindex --demangle[=@var{style}]
1474 @kindex --no-demangle
1475 @item --demangle[=@var{style}]
1476 @itemx --no-demangle
1477 These options control whether to demangle symbol names in error messages
1478 and other output. When the linker is told to demangle, it tries to
1479 present symbol names in a readable fashion: it strips leading
1480 underscores if they are used by the object file format, and converts C++
1481 mangled symbol names into user readable names. Different compilers have
1482 different mangling styles. The optional demangling style argument can be used
1483 to choose an appropriate demangling style for your compiler. The linker will
1484 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1485 is set. These options may be used to override the default.
1487 @cindex dynamic linker, from command line
1488 @kindex -I@var{file}
1489 @kindex --dynamic-linker=@var{file}
1491 @itemx --dynamic-linker=@var{file}
1492 Set the name of the dynamic linker. This is only meaningful when
1493 generating dynamically linked ELF executables. The default dynamic
1494 linker is normally correct; don't use this unless you know what you are
1497 @kindex --no-dynamic-linker
1498 @item --no-dynamic-linker
1499 When producing an executable file, omit the request for a dynamic
1500 linker to be used at load-time. This is only meaningful for ELF
1501 executables that contain dynamic relocations, and usually requires
1502 entry point code that is capable of processing these relocations.
1504 @kindex --fatal-warnings
1505 @kindex --no-fatal-warnings
1506 @item --fatal-warnings
1507 @itemx --no-fatal-warnings
1508 Treat all warnings as errors. The default behaviour can be restored
1509 with the option @option{--no-fatal-warnings}.
1511 @kindex --force-exe-suffix
1512 @item --force-exe-suffix
1513 Make sure that an output file has a .exe suffix.
1515 If a successfully built fully linked output file does not have a
1516 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1517 the output file to one of the same name with a @code{.exe} suffix. This
1518 option is useful when using unmodified Unix makefiles on a Microsoft
1519 Windows host, since some versions of Windows won't run an image unless
1520 it ends in a @code{.exe} suffix.
1522 @kindex --gc-sections
1523 @kindex --no-gc-sections
1524 @cindex garbage collection
1526 @itemx --no-gc-sections
1527 Enable garbage collection of unused input sections. It is ignored on
1528 targets that do not support this option. The default behaviour (of not
1529 performing this garbage collection) can be restored by specifying
1530 @samp{--no-gc-sections} on the command line. Note that garbage
1531 collection for COFF and PE format targets is supported, but the
1532 implementation is currently considered to be experimental.
1534 @samp{--gc-sections} decides which input sections are used by
1535 examining symbols and relocations. The section containing the entry
1536 symbol and all sections containing symbols undefined on the
1537 command-line will be kept, as will sections containing symbols
1538 referenced by dynamic objects. Note that when building shared
1539 libraries, the linker must assume that any visible symbol is
1540 referenced. Once this initial set of sections has been determined,
1541 the linker recursively marks as used any section referenced by their
1542 relocations. See @samp{--entry} and @samp{--undefined}.
1544 This option can be set when doing a partial link (enabled with option
1545 @samp{-r}). In this case the root of symbols kept must be explicitly
1546 specified either by an @samp{--entry} or @samp{--undefined} option or by
1547 a @code{ENTRY} command in the linker script.
1549 @kindex --print-gc-sections
1550 @kindex --no-print-gc-sections
1551 @cindex garbage collection
1552 @item --print-gc-sections
1553 @itemx --no-print-gc-sections
1554 List all sections removed by garbage collection. The listing is
1555 printed on stderr. This option is only effective if garbage
1556 collection has been enabled via the @samp{--gc-sections}) option. The
1557 default behaviour (of not listing the sections that are removed) can
1558 be restored by specifying @samp{--no-print-gc-sections} on the command
1561 @kindex --print-output-format
1562 @cindex output format
1563 @item --print-output-format
1564 Print the name of the default output format (perhaps influenced by
1565 other command-line options). This is the string that would appear
1566 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1568 @kindex --print-memory-usage
1569 @cindex memory usage
1570 @item --print-memory-usage
1571 Print used size, total size and used size of memory regions created with
1572 the @ref{MEMORY} command. This is useful on embedded targets to have a
1573 quick view of amount of free memory. The format of the output has one
1574 headline and one line per region. It is both human readable and easily
1575 parsable by tools. Here is an example of an output:
1578 Memory region Used Size Region Size %age Used
1579 ROM: 256 KB 1 MB 25.00%
1580 RAM: 32 B 2 GB 0.00%
1587 Print a summary of the command-line options on the standard output and exit.
1589 @kindex --target-help
1591 Print a summary of all target specific options on the standard output and exit.
1593 @kindex -Map=@var{mapfile}
1594 @item -Map=@var{mapfile}
1595 Print a link map to the file @var{mapfile}. See the description of the
1596 @option{-M} option, above.
1598 @cindex memory usage
1599 @kindex --no-keep-memory
1600 @item --no-keep-memory
1601 @command{ld} normally optimizes for speed over memory usage by caching the
1602 symbol tables of input files in memory. This option tells @command{ld} to
1603 instead optimize for memory usage, by rereading the symbol tables as
1604 necessary. This may be required if @command{ld} runs out of memory space
1605 while linking a large executable.
1607 @kindex --no-undefined
1609 @item --no-undefined
1611 Report unresolved symbol references from regular object files. This
1612 is done even if the linker is creating a non-symbolic shared library.
1613 The switch @option{--[no-]allow-shlib-undefined} controls the
1614 behaviour for reporting unresolved references found in shared
1615 libraries being linked in.
1617 @kindex --allow-multiple-definition
1619 @item --allow-multiple-definition
1621 Normally when a symbol is defined multiple times, the linker will
1622 report a fatal error. These options allow multiple definitions and the
1623 first definition will be used.
1625 @kindex --allow-shlib-undefined
1626 @kindex --no-allow-shlib-undefined
1627 @item --allow-shlib-undefined
1628 @itemx --no-allow-shlib-undefined
1629 Allows or disallows undefined symbols in shared libraries.
1630 This switch is similar to @option{--no-undefined} except that it
1631 determines the behaviour when the undefined symbols are in a
1632 shared library rather than a regular object file. It does not affect
1633 how undefined symbols in regular object files are handled.
1635 The default behaviour is to report errors for any undefined symbols
1636 referenced in shared libraries if the linker is being used to create
1637 an executable, but to allow them if the linker is being used to create
1640 The reasons for allowing undefined symbol references in shared
1641 libraries specified at link time are that:
1645 A shared library specified at link time may not be the same as the one
1646 that is available at load time, so the symbol might actually be
1647 resolvable at load time.
1649 There are some operating systems, eg BeOS and HPPA, where undefined
1650 symbols in shared libraries are normal.
1652 The BeOS kernel for example patches shared libraries at load time to
1653 select whichever function is most appropriate for the current
1654 architecture. This is used, for example, to dynamically select an
1655 appropriate memset function.
1658 @kindex --no-undefined-version
1659 @item --no-undefined-version
1660 Normally when a symbol has an undefined version, the linker will ignore
1661 it. This option disallows symbols with undefined version and a fatal error
1662 will be issued instead.
1664 @kindex --default-symver
1665 @item --default-symver
1666 Create and use a default symbol version (the soname) for unversioned
1669 @kindex --default-imported-symver
1670 @item --default-imported-symver
1671 Create and use a default symbol version (the soname) for unversioned
1674 @kindex --no-warn-mismatch
1675 @item --no-warn-mismatch
1676 Normally @command{ld} will give an error if you try to link together input
1677 files that are mismatched for some reason, perhaps because they have
1678 been compiled for different processors or for different endiannesses.
1679 This option tells @command{ld} that it should silently permit such possible
1680 errors. This option should only be used with care, in cases when you
1681 have taken some special action that ensures that the linker errors are
1684 @kindex --no-warn-search-mismatch
1685 @item --no-warn-search-mismatch
1686 Normally @command{ld} will give a warning if it finds an incompatible
1687 library during a library search. This option silences the warning.
1689 @kindex --no-whole-archive
1690 @item --no-whole-archive
1691 Turn off the effect of the @option{--whole-archive} option for subsequent
1694 @cindex output file after errors
1695 @kindex --noinhibit-exec
1696 @item --noinhibit-exec
1697 Retain the executable output file whenever it is still usable.
1698 Normally, the linker will not produce an output file if it encounters
1699 errors during the link process; it exits without writing an output file
1700 when it issues any error whatsoever.
1704 Only search library directories explicitly specified on the
1705 command line. Library directories specified in linker scripts
1706 (including linker scripts specified on the command line) are ignored.
1708 @ifclear SingleFormat
1709 @kindex --oformat=@var{output-format}
1710 @item --oformat=@var{output-format}
1711 @command{ld} may be configured to support more than one kind of object
1712 file. If your @command{ld} is configured this way, you can use the
1713 @samp{--oformat} option to specify the binary format for the output
1714 object file. Even when @command{ld} is configured to support alternative
1715 object formats, you don't usually need to specify this, as @command{ld}
1716 should be configured to produce as a default output format the most
1717 usual format on each machine. @var{output-format} is a text string, the
1718 name of a particular format supported by the BFD libraries. (You can
1719 list the available binary formats with @samp{objdump -i}.) The script
1720 command @code{OUTPUT_FORMAT} can also specify the output format, but
1721 this option overrides it. @xref{BFD}.
1724 @kindex --out-implib
1725 @item --out-implib @var{file}
1726 Create an import library in @var{file} corresponding to the executable
1727 the linker is generating (eg. a DLL or ELF program). This import
1728 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
1729 may be used to link clients against the generated executable; this
1730 behaviour makes it possible to skip a separate import library creation
1731 step (eg. @code{dlltool} for DLLs). This option is only available for
1732 the i386 PE and ELF targetted ports of the linker.
1735 @kindex --pic-executable
1737 @itemx --pic-executable
1738 @cindex position independent executables
1739 Create a position independent executable. This is currently only supported on
1740 ELF platforms. Position independent executables are similar to shared
1741 libraries in that they are relocated by the dynamic linker to the virtual
1742 address the OS chooses for them (which can vary between invocations). Like
1743 normal dynamically linked executables they can be executed and symbols
1744 defined in the executable cannot be overridden by shared libraries.
1748 This option is ignored for Linux compatibility.
1752 This option is ignored for SVR4 compatibility.
1755 @cindex synthesizing linker
1756 @cindex relaxing addressing modes
1760 An option with machine dependent effects.
1762 This option is only supported on a few targets.
1765 @xref{H8/300,,@command{ld} and the H8/300}.
1768 @xref{i960,, @command{ld} and the Intel 960 family}.
1771 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1774 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1777 @xref{Nios II,,@command{ld} and the Altera Nios II}.
1780 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1783 On some platforms the @samp{--relax} option performs target specific,
1784 global optimizations that become possible when the linker resolves
1785 addressing in the program, such as relaxing address modes,
1786 synthesizing new instructions, selecting shorter version of current
1787 instructions, and combining constant values.
1789 On some platforms these link time global optimizations may make symbolic
1790 debugging of the resulting executable impossible.
1792 This is known to be the case for the Matsushita MN10200 and MN10300
1793 family of processors.
1797 On platforms where this is not supported, @samp{--relax} is accepted,
1801 On platforms where @samp{--relax} is accepted the option
1802 @samp{--no-relax} can be used to disable the feature.
1804 @cindex retaining specified symbols
1805 @cindex stripping all but some symbols
1806 @cindex symbols, retaining selectively
1807 @kindex --retain-symbols-file=@var{filename}
1808 @item --retain-symbols-file=@var{filename}
1809 Retain @emph{only} the symbols listed in the file @var{filename},
1810 discarding all others. @var{filename} is simply a flat file, with one
1811 symbol name per line. This option is especially useful in environments
1815 where a large global symbol table is accumulated gradually, to conserve
1818 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1819 or symbols needed for relocations.
1821 You may only specify @samp{--retain-symbols-file} once in the command
1822 line. It overrides @samp{-s} and @samp{-S}.
1825 @item -rpath=@var{dir}
1826 @cindex runtime library search path
1827 @kindex -rpath=@var{dir}
1828 Add a directory to the runtime library search path. This is used when
1829 linking an ELF executable with shared objects. All @option{-rpath}
1830 arguments are concatenated and passed to the runtime linker, which uses
1831 them to locate shared objects at runtime. The @option{-rpath} option is
1832 also used when locating shared objects which are needed by shared
1833 objects explicitly included in the link; see the description of the
1834 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1835 ELF executable, the contents of the environment variable
1836 @code{LD_RUN_PATH} will be used if it is defined.
1838 The @option{-rpath} option may also be used on SunOS. By default, on
1839 SunOS, the linker will form a runtime search path out of all the
1840 @option{-L} options it is given. If a @option{-rpath} option is used, the
1841 runtime search path will be formed exclusively using the @option{-rpath}
1842 options, ignoring the @option{-L} options. This can be useful when using
1843 gcc, which adds many @option{-L} options which may be on NFS mounted
1846 For compatibility with other ELF linkers, if the @option{-R} option is
1847 followed by a directory name, rather than a file name, it is treated as
1848 the @option{-rpath} option.
1852 @cindex link-time runtime library search path
1853 @kindex -rpath-link=@var{dir}
1854 @item -rpath-link=@var{dir}
1855 When using ELF or SunOS, one shared library may require another. This
1856 happens when an @code{ld -shared} link includes a shared library as one
1859 When the linker encounters such a dependency when doing a non-shared,
1860 non-relocatable link, it will automatically try to locate the required
1861 shared library and include it in the link, if it is not included
1862 explicitly. In such a case, the @option{-rpath-link} option
1863 specifies the first set of directories to search. The
1864 @option{-rpath-link} option may specify a sequence of directory names
1865 either by specifying a list of names separated by colons, or by
1866 appearing multiple times.
1868 This option should be used with caution as it overrides the search path
1869 that may have been hard compiled into a shared library. In such a case it
1870 is possible to use unintentionally a different search path than the
1871 runtime linker would do.
1873 The linker uses the following search paths to locate required shared
1877 Any directories specified by @option{-rpath-link} options.
1879 Any directories specified by @option{-rpath} options. The difference
1880 between @option{-rpath} and @option{-rpath-link} is that directories
1881 specified by @option{-rpath} options are included in the executable and
1882 used at runtime, whereas the @option{-rpath-link} option is only effective
1883 at link time. Searching @option{-rpath} in this way is only supported
1884 by native linkers and cross linkers which have been configured with
1885 the @option{--with-sysroot} option.
1887 On an ELF system, for native linkers, if the @option{-rpath} and
1888 @option{-rpath-link} options were not used, search the contents of the
1889 environment variable @code{LD_RUN_PATH}.
1891 On SunOS, if the @option{-rpath} option was not used, search any
1892 directories specified using @option{-L} options.
1894 For a native linker, search the contents of the environment
1895 variable @code{LD_LIBRARY_PATH}.
1897 For a native ELF linker, the directories in @code{DT_RUNPATH} or
1898 @code{DT_RPATH} of a shared library are searched for shared
1899 libraries needed by it. The @code{DT_RPATH} entries are ignored if
1900 @code{DT_RUNPATH} entries exist.
1902 The default directories, normally @file{/lib} and @file{/usr/lib}.
1904 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1905 exists, the list of directories found in that file.
1908 If the required shared library is not found, the linker will issue a
1909 warning and continue with the link.
1916 @cindex shared libraries
1917 Create a shared library. This is currently only supported on ELF, XCOFF
1918 and SunOS platforms. On SunOS, the linker will automatically create a
1919 shared library if the @option{-e} option is not used and there are
1920 undefined symbols in the link.
1922 @kindex --sort-common
1924 @itemx --sort-common=ascending
1925 @itemx --sort-common=descending
1926 This option tells @command{ld} to sort the common symbols by alignment in
1927 ascending or descending order when it places them in the appropriate output
1928 sections. The symbol alignments considered are sixteen-byte or larger,
1929 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1930 between symbols due to alignment constraints. If no sorting order is
1931 specified, then descending order is assumed.
1933 @kindex --sort-section=name
1934 @item --sort-section=name
1935 This option will apply @code{SORT_BY_NAME} to all wildcard section
1936 patterns in the linker script.
1938 @kindex --sort-section=alignment
1939 @item --sort-section=alignment
1940 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1941 patterns in the linker script.
1943 @kindex --split-by-file
1944 @item --split-by-file[=@var{size}]
1945 Similar to @option{--split-by-reloc} but creates a new output section for
1946 each input file when @var{size} is reached. @var{size} defaults to a
1947 size of 1 if not given.
1949 @kindex --split-by-reloc
1950 @item --split-by-reloc[=@var{count}]
1951 Tries to creates extra sections in the output file so that no single
1952 output section in the file contains more than @var{count} relocations.
1953 This is useful when generating huge relocatable files for downloading into
1954 certain real time kernels with the COFF object file format; since COFF
1955 cannot represent more than 65535 relocations in a single section. Note
1956 that this will fail to work with object file formats which do not
1957 support arbitrary sections. The linker will not split up individual
1958 input sections for redistribution, so if a single input section contains
1959 more than @var{count} relocations one output section will contain that
1960 many relocations. @var{count} defaults to a value of 32768.
1964 Compute and display statistics about the operation of the linker, such
1965 as execution time and memory usage.
1967 @kindex --sysroot=@var{directory}
1968 @item --sysroot=@var{directory}
1969 Use @var{directory} as the location of the sysroot, overriding the
1970 configure-time default. This option is only supported by linkers
1971 that were configured using @option{--with-sysroot}.
1973 @kindex --traditional-format
1974 @cindex traditional format
1975 @item --traditional-format
1976 For some targets, the output of @command{ld} is different in some ways from
1977 the output of some existing linker. This switch requests @command{ld} to
1978 use the traditional format instead.
1981 For example, on SunOS, @command{ld} combines duplicate entries in the
1982 symbol string table. This can reduce the size of an output file with
1983 full debugging information by over 30 percent. Unfortunately, the SunOS
1984 @code{dbx} program can not read the resulting program (@code{gdb} has no
1985 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
1986 combine duplicate entries.
1988 @kindex --section-start=@var{sectionname}=@var{org}
1989 @item --section-start=@var{sectionname}=@var{org}
1990 Locate a section in the output file at the absolute
1991 address given by @var{org}. You may use this option as many
1992 times as necessary to locate multiple sections in the command
1994 @var{org} must be a single hexadecimal integer;
1995 for compatibility with other linkers, you may omit the leading
1996 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
1997 should be no white space between @var{sectionname}, the equals
1998 sign (``@key{=}''), and @var{org}.
2000 @kindex -Tbss=@var{org}
2001 @kindex -Tdata=@var{org}
2002 @kindex -Ttext=@var{org}
2003 @cindex segment origins, cmd line
2004 @item -Tbss=@var{org}
2005 @itemx -Tdata=@var{org}
2006 @itemx -Ttext=@var{org}
2007 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2008 @code{.text} as the @var{sectionname}.
2010 @kindex -Ttext-segment=@var{org}
2011 @item -Ttext-segment=@var{org}
2012 @cindex text segment origin, cmd line
2013 When creating an ELF executable, it will set the address of the first
2014 byte of the text segment.
2016 @kindex -Trodata-segment=@var{org}
2017 @item -Trodata-segment=@var{org}
2018 @cindex rodata segment origin, cmd line
2019 When creating an ELF executable or shared object for a target where
2020 the read-only data is in its own segment separate from the executable
2021 text, it will set the address of the first byte of the read-only data segment.
2023 @kindex -Tldata-segment=@var{org}
2024 @item -Tldata-segment=@var{org}
2025 @cindex ldata segment origin, cmd line
2026 When creating an ELF executable or shared object for x86-64 medium memory
2027 model, it will set the address of the first byte of the ldata segment.
2029 @kindex --unresolved-symbols
2030 @item --unresolved-symbols=@var{method}
2031 Determine how to handle unresolved symbols. There are four possible
2032 values for @samp{method}:
2036 Do not report any unresolved symbols.
2039 Report all unresolved symbols. This is the default.
2041 @item ignore-in-object-files
2042 Report unresolved symbols that are contained in shared libraries, but
2043 ignore them if they come from regular object files.
2045 @item ignore-in-shared-libs
2046 Report unresolved symbols that come from regular object files, but
2047 ignore them if they come from shared libraries. This can be useful
2048 when creating a dynamic binary and it is known that all the shared
2049 libraries that it should be referencing are included on the linker's
2053 The behaviour for shared libraries on their own can also be controlled
2054 by the @option{--[no-]allow-shlib-undefined} option.
2056 Normally the linker will generate an error message for each reported
2057 unresolved symbol but the option @option{--warn-unresolved-symbols}
2058 can change this to a warning.
2060 @kindex --verbose[=@var{NUMBER}]
2061 @cindex verbose[=@var{NUMBER}]
2063 @itemx --verbose[=@var{NUMBER}]
2064 Display the version number for @command{ld} and list the linker emulations
2065 supported. Display which input files can and cannot be opened. Display
2066 the linker script being used by the linker. If the optional @var{NUMBER}
2067 argument > 1, plugin symbol status will also be displayed.
2069 @kindex --version-script=@var{version-scriptfile}
2070 @cindex version script, symbol versions
2071 @item --version-script=@var{version-scriptfile}
2072 Specify the name of a version script to the linker. This is typically
2073 used when creating shared libraries to specify additional information
2074 about the version hierarchy for the library being created. This option
2075 is only fully supported on ELF platforms which support shared libraries;
2076 see @ref{VERSION}. It is partially supported on PE platforms, which can
2077 use version scripts to filter symbol visibility in auto-export mode: any
2078 symbols marked @samp{local} in the version script will not be exported.
2081 @kindex --warn-common
2082 @cindex warnings, on combining symbols
2083 @cindex combining symbols, warnings on
2085 Warn when a common symbol is combined with another common symbol or with
2086 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2087 but linkers on some other operating systems do not. This option allows
2088 you to find potential problems from combining global symbols.
2089 Unfortunately, some C libraries use this practice, so you may get some
2090 warnings about symbols in the libraries as well as in your programs.
2092 There are three kinds of global symbols, illustrated here by C examples:
2096 A definition, which goes in the initialized data section of the output
2100 An undefined reference, which does not allocate space.
2101 There must be either a definition or a common symbol for the
2105 A common symbol. If there are only (one or more) common symbols for a
2106 variable, it goes in the uninitialized data area of the output file.
2107 The linker merges multiple common symbols for the same variable into a
2108 single symbol. If they are of different sizes, it picks the largest
2109 size. The linker turns a common symbol into a declaration, if there is
2110 a definition of the same variable.
2113 The @samp{--warn-common} option can produce five kinds of warnings.
2114 Each warning consists of a pair of lines: the first describes the symbol
2115 just encountered, and the second describes the previous symbol
2116 encountered with the same name. One or both of the two symbols will be
2121 Turning a common symbol into a reference, because there is already a
2122 definition for the symbol.
2124 @var{file}(@var{section}): warning: common of `@var{symbol}'
2125 overridden by definition
2126 @var{file}(@var{section}): warning: defined here
2130 Turning a common symbol into a reference, because a later definition for
2131 the symbol is encountered. This is the same as the previous case,
2132 except that the symbols are encountered in a different order.
2134 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2136 @var{file}(@var{section}): warning: common is here
2140 Merging a common symbol with a previous same-sized common symbol.
2142 @var{file}(@var{section}): warning: multiple common
2144 @var{file}(@var{section}): warning: previous common is here
2148 Merging a common symbol with a previous larger common symbol.
2150 @var{file}(@var{section}): warning: common of `@var{symbol}'
2151 overridden by larger common
2152 @var{file}(@var{section}): warning: larger common is here
2156 Merging a common symbol with a previous smaller common symbol. This is
2157 the same as the previous case, except that the symbols are
2158 encountered in a different order.
2160 @var{file}(@var{section}): warning: common of `@var{symbol}'
2161 overriding smaller common
2162 @var{file}(@var{section}): warning: smaller common is here
2166 @kindex --warn-constructors
2167 @item --warn-constructors
2168 Warn if any global constructors are used. This is only useful for a few
2169 object file formats. For formats like COFF or ELF, the linker can not
2170 detect the use of global constructors.
2172 @kindex --warn-multiple-gp
2173 @item --warn-multiple-gp
2174 Warn if multiple global pointer values are required in the output file.
2175 This is only meaningful for certain processors, such as the Alpha.
2176 Specifically, some processors put large-valued constants in a special
2177 section. A special register (the global pointer) points into the middle
2178 of this section, so that constants can be loaded efficiently via a
2179 base-register relative addressing mode. Since the offset in
2180 base-register relative mode is fixed and relatively small (e.g., 16
2181 bits), this limits the maximum size of the constant pool. Thus, in
2182 large programs, it is often necessary to use multiple global pointer
2183 values in order to be able to address all possible constants. This
2184 option causes a warning to be issued whenever this case occurs.
2187 @cindex warnings, on undefined symbols
2188 @cindex undefined symbols, warnings on
2190 Only warn once for each undefined symbol, rather than once per module
2193 @kindex --warn-section-align
2194 @cindex warnings, on section alignment
2195 @cindex section alignment, warnings on
2196 @item --warn-section-align
2197 Warn if the address of an output section is changed because of
2198 alignment. Typically, the alignment will be set by an input section.
2199 The address will only be changed if it not explicitly specified; that
2200 is, if the @code{SECTIONS} command does not specify a start address for
2201 the section (@pxref{SECTIONS}).
2203 @kindex --warn-shared-textrel
2204 @item --warn-shared-textrel
2205 Warn if the linker adds a DT_TEXTREL to a shared object.
2207 @kindex --warn-alternate-em
2208 @item --warn-alternate-em
2209 Warn if an object has alternate ELF machine code.
2211 @kindex --warn-unresolved-symbols
2212 @item --warn-unresolved-symbols
2213 If the linker is going to report an unresolved symbol (see the option
2214 @option{--unresolved-symbols}) it will normally generate an error.
2215 This option makes it generate a warning instead.
2217 @kindex --error-unresolved-symbols
2218 @item --error-unresolved-symbols
2219 This restores the linker's default behaviour of generating errors when
2220 it is reporting unresolved symbols.
2222 @kindex --whole-archive
2223 @cindex including an entire archive
2224 @item --whole-archive
2225 For each archive mentioned on the command line after the
2226 @option{--whole-archive} option, include every object file in the archive
2227 in the link, rather than searching the archive for the required object
2228 files. This is normally used to turn an archive file into a shared
2229 library, forcing every object to be included in the resulting shared
2230 library. This option may be used more than once.
2232 Two notes when using this option from gcc: First, gcc doesn't know
2233 about this option, so you have to use @option{-Wl,-whole-archive}.
2234 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2235 list of archives, because gcc will add its own list of archives to
2236 your link and you may not want this flag to affect those as well.
2238 @kindex --wrap=@var{symbol}
2239 @item --wrap=@var{symbol}
2240 Use a wrapper function for @var{symbol}. Any undefined reference to
2241 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2242 undefined reference to @code{__real_@var{symbol}} will be resolved to
2245 This can be used to provide a wrapper for a system function. The
2246 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2247 wishes to call the system function, it should call
2248 @code{__real_@var{symbol}}.
2250 Here is a trivial example:
2254 __wrap_malloc (size_t c)
2256 printf ("malloc called with %zu\n", c);
2257 return __real_malloc (c);
2261 If you link other code with this file using @option{--wrap malloc}, then
2262 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2263 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2264 call the real @code{malloc} function.
2266 You may wish to provide a @code{__real_malloc} function as well, so that
2267 links without the @option{--wrap} option will succeed. If you do this,
2268 you should not put the definition of @code{__real_malloc} in the same
2269 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2270 call before the linker has a chance to wrap it to @code{malloc}.
2272 @kindex --eh-frame-hdr
2273 @kindex --no-eh-frame-hdr
2274 @item --eh-frame-hdr
2275 @itemx --no-eh-frame-hdr
2276 Request (@option{--eh-frame-hdr}) or suppress
2277 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2278 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2280 @kindex --ld-generated-unwind-info
2281 @item --no-ld-generated-unwind-info
2282 Request creation of @code{.eh_frame} unwind info for linker
2283 generated code sections like PLT. This option is on by default
2284 if linker generated unwind info is supported.
2286 @kindex --enable-new-dtags
2287 @kindex --disable-new-dtags
2288 @item --enable-new-dtags
2289 @itemx --disable-new-dtags
2290 This linker can create the new dynamic tags in ELF. But the older ELF
2291 systems may not understand them. If you specify
2292 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2293 and older dynamic tags will be omitted.
2294 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2295 created. By default, the new dynamic tags are not created. Note that
2296 those options are only available for ELF systems.
2298 @kindex --hash-size=@var{number}
2299 @item --hash-size=@var{number}
2300 Set the default size of the linker's hash tables to a prime number
2301 close to @var{number}. Increasing this value can reduce the length of
2302 time it takes the linker to perform its tasks, at the expense of
2303 increasing the linker's memory requirements. Similarly reducing this
2304 value can reduce the memory requirements at the expense of speed.
2306 @kindex --hash-style=@var{style}
2307 @item --hash-style=@var{style}
2308 Set the type of linker's hash table(s). @var{style} can be either
2309 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2310 new style GNU @code{.gnu.hash} section or @code{both} for both
2311 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2312 hash tables. The default is @code{sysv}.
2314 @kindex --compress-debug-sections=none
2315 @kindex --compress-debug-sections=zlib
2316 @kindex --compress-debug-sections=zlib-gnu
2317 @kindex --compress-debug-sections=zlib-gabi
2318 @item --compress-debug-sections=none
2319 @itemx --compress-debug-sections=zlib
2320 @itemx --compress-debug-sections=zlib-gnu
2321 @itemx --compress-debug-sections=zlib-gabi
2322 On ELF platforms , these options control how DWARF debug sections are
2323 compressed using zlib. @option{--compress-debug-sections=none} doesn't
2324 compress DWARF debug sections.
2325 @option{--compress-debug-sections=zlib-gnu} compresses DWARF debug
2326 sections and rename debug section names to begin with @samp{.zdebug}
2327 instead of @samp{.debug}. @option{--compress-debug-sections=zlib}
2328 and @option{--compress-debug-sections=zlib-gabi}
2329 compress DWARF debug sections with SHF_COMPRESSED from the ELF ABI.
2330 The default behaviour varies depending upon the target involved and
2331 the configure options used to build the toolchain. The default can be
2332 determined by examing the output from the linker's @option{--help} option.
2334 @kindex --reduce-memory-overheads
2335 @item --reduce-memory-overheads
2336 This option reduces memory requirements at ld runtime, at the expense of
2337 linking speed. This was introduced to select the old O(n^2) algorithm
2338 for link map file generation, rather than the new O(n) algorithm which uses
2339 about 40% more memory for symbol storage.
2341 Another effect of the switch is to set the default hash table size to
2342 1021, which again saves memory at the cost of lengthening the linker's
2343 run time. This is not done however if the @option{--hash-size} switch
2346 The @option{--reduce-memory-overheads} switch may be also be used to
2347 enable other tradeoffs in future versions of the linker.
2350 @kindex --build-id=@var{style}
2352 @itemx --build-id=@var{style}
2353 Request the creation of a @code{.note.gnu.build-id} ELF note section
2354 or a @code{.buildid} COFF section. The contents of the note are
2355 unique bits identifying this linked file. @var{style} can be
2356 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2357 @sc{SHA1} hash on the normative parts of the output contents,
2358 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2359 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2360 string specified as an even number of hexadecimal digits (@code{-} and
2361 @code{:} characters between digit pairs are ignored). If @var{style}
2362 is omitted, @code{sha1} is used.
2364 The @code{md5} and @code{sha1} styles produces an identifier
2365 that is always the same in an identical output file, but will be
2366 unique among all nonidentical output files. It is not intended
2367 to be compared as a checksum for the file's contents. A linked
2368 file may be changed later by other tools, but the build ID bit
2369 string identifying the original linked file does not change.
2371 Passing @code{none} for @var{style} disables the setting from any
2372 @code{--build-id} options earlier on the command line.
2377 @subsection Options Specific to i386 PE Targets
2379 @c man begin OPTIONS
2381 The i386 PE linker supports the @option{-shared} option, which causes
2382 the output to be a dynamically linked library (DLL) instead of a
2383 normal executable. You should name the output @code{*.dll} when you
2384 use this option. In addition, the linker fully supports the standard
2385 @code{*.def} files, which may be specified on the linker command line
2386 like an object file (in fact, it should precede archives it exports
2387 symbols from, to ensure that they get linked in, just like a normal
2390 In addition to the options common to all targets, the i386 PE linker
2391 support additional command line options that are specific to the i386
2392 PE target. Options that take values may be separated from their
2393 values by either a space or an equals sign.
2397 @kindex --add-stdcall-alias
2398 @item --add-stdcall-alias
2399 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2400 as-is and also with the suffix stripped.
2401 [This option is specific to the i386 PE targeted port of the linker]
2404 @item --base-file @var{file}
2405 Use @var{file} as the name of a file in which to save the base
2406 addresses of all the relocations needed for generating DLLs with
2408 [This is an i386 PE specific option]
2412 Create a DLL instead of a regular executable. You may also use
2413 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2415 [This option is specific to the i386 PE targeted port of the linker]
2417 @kindex --enable-long-section-names
2418 @kindex --disable-long-section-names
2419 @item --enable-long-section-names
2420 @itemx --disable-long-section-names
2421 The PE variants of the COFF object format add an extension that permits
2422 the use of section names longer than eight characters, the normal limit
2423 for COFF. By default, these names are only allowed in object files, as
2424 fully-linked executable images do not carry the COFF string table required
2425 to support the longer names. As a GNU extension, it is possible to
2426 allow their use in executable images as well, or to (probably pointlessly!)
2427 disallow it in object files, by using these two options. Executable images
2428 generated with these long section names are slightly non-standard, carrying
2429 as they do a string table, and may generate confusing output when examined
2430 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2431 GDB relies on the use of PE long section names to find Dwarf-2 debug
2432 information sections in an executable image at runtime, and so if neither
2433 option is specified on the command-line, @command{ld} will enable long
2434 section names, overriding the default and technically correct behaviour,
2435 when it finds the presence of debug information while linking an executable
2436 image and not stripping symbols.
2437 [This option is valid for all PE targeted ports of the linker]
2439 @kindex --enable-stdcall-fixup
2440 @kindex --disable-stdcall-fixup
2441 @item --enable-stdcall-fixup
2442 @itemx --disable-stdcall-fixup
2443 If the link finds a symbol that it cannot resolve, it will attempt to
2444 do ``fuzzy linking'' by looking for another defined symbol that differs
2445 only in the format of the symbol name (cdecl vs stdcall) and will
2446 resolve that symbol by linking to the match. For example, the
2447 undefined symbol @code{_foo} might be linked to the function
2448 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2449 to the function @code{_bar}. When the linker does this, it prints a
2450 warning, since it normally should have failed to link, but sometimes
2451 import libraries generated from third-party dlls may need this feature
2452 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2453 feature is fully enabled and warnings are not printed. If you specify
2454 @option{--disable-stdcall-fixup}, this feature is disabled and such
2455 mismatches are considered to be errors.
2456 [This option is specific to the i386 PE targeted port of the linker]
2458 @kindex --leading-underscore
2459 @kindex --no-leading-underscore
2460 @item --leading-underscore
2461 @itemx --no-leading-underscore
2462 For most targets default symbol-prefix is an underscore and is defined
2463 in target's description. By this option it is possible to
2464 disable/enable the default underscore symbol-prefix.
2466 @cindex DLLs, creating
2467 @kindex --export-all-symbols
2468 @item --export-all-symbols
2469 If given, all global symbols in the objects used to build a DLL will
2470 be exported by the DLL. Note that this is the default if there
2471 otherwise wouldn't be any exported symbols. When symbols are
2472 explicitly exported via DEF files or implicitly exported via function
2473 attributes, the default is to not export anything else unless this
2474 option is given. Note that the symbols @code{DllMain@@12},
2475 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2476 @code{impure_ptr} will not be automatically
2477 exported. Also, symbols imported from other DLLs will not be
2478 re-exported, nor will symbols specifying the DLL's internal layout
2479 such as those beginning with @code{_head_} or ending with
2480 @code{_iname}. In addition, no symbols from @code{libgcc},
2481 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2482 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2483 not be exported, to help with C++ DLLs. Finally, there is an
2484 extensive list of cygwin-private symbols that are not exported
2485 (obviously, this applies on when building DLLs for cygwin targets).
2486 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2487 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2488 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2489 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2490 @code{cygwin_premain3}, and @code{environ}.
2491 [This option is specific to the i386 PE targeted port of the linker]
2493 @kindex --exclude-symbols
2494 @item --exclude-symbols @var{symbol},@var{symbol},...
2495 Specifies a list of symbols which should not be automatically
2496 exported. The symbol names may be delimited by commas or colons.
2497 [This option is specific to the i386 PE targeted port of the linker]
2499 @kindex --exclude-all-symbols
2500 @item --exclude-all-symbols
2501 Specifies no symbols should be automatically exported.
2502 [This option is specific to the i386 PE targeted port of the linker]
2504 @kindex --file-alignment
2505 @item --file-alignment
2506 Specify the file alignment. Sections in the file will always begin at
2507 file offsets which are multiples of this number. This defaults to
2509 [This option is specific to the i386 PE targeted port of the linker]
2513 @item --heap @var{reserve}
2514 @itemx --heap @var{reserve},@var{commit}
2515 Specify the number of bytes of memory to reserve (and optionally commit)
2516 to be used as heap for this program. The default is 1MB reserved, 4K
2518 [This option is specific to the i386 PE targeted port of the linker]
2521 @kindex --image-base
2522 @item --image-base @var{value}
2523 Use @var{value} as the base address of your program or dll. This is
2524 the lowest memory location that will be used when your program or dll
2525 is loaded. To reduce the need to relocate and improve performance of
2526 your dlls, each should have a unique base address and not overlap any
2527 other dlls. The default is 0x400000 for executables, and 0x10000000
2529 [This option is specific to the i386 PE targeted port of the linker]
2533 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2534 symbols before they are exported.
2535 [This option is specific to the i386 PE targeted port of the linker]
2537 @kindex --large-address-aware
2538 @item --large-address-aware
2539 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2540 header is set to indicate that this executable supports virtual addresses
2541 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2542 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2543 section of the BOOT.INI. Otherwise, this bit has no effect.
2544 [This option is specific to PE targeted ports of the linker]
2546 @kindex --disable-large-address-aware
2547 @item --disable-large-address-aware
2548 Reverts the effect of a previous @samp{--large-address-aware} option.
2549 This is useful if @samp{--large-address-aware} is always set by the compiler
2550 driver (e.g. Cygwin gcc) and the executable does not support virtual
2551 addresses greater than 2 gigabytes.
2552 [This option is specific to PE targeted ports of the linker]
2554 @kindex --major-image-version
2555 @item --major-image-version @var{value}
2556 Sets the major number of the ``image version''. Defaults to 1.
2557 [This option is specific to the i386 PE targeted port of the linker]
2559 @kindex --major-os-version
2560 @item --major-os-version @var{value}
2561 Sets the major number of the ``os version''. Defaults to 4.
2562 [This option is specific to the i386 PE targeted port of the linker]
2564 @kindex --major-subsystem-version
2565 @item --major-subsystem-version @var{value}
2566 Sets the major number of the ``subsystem version''. Defaults to 4.
2567 [This option is specific to the i386 PE targeted port of the linker]
2569 @kindex --minor-image-version
2570 @item --minor-image-version @var{value}
2571 Sets the minor number of the ``image version''. Defaults to 0.
2572 [This option is specific to the i386 PE targeted port of the linker]
2574 @kindex --minor-os-version
2575 @item --minor-os-version @var{value}
2576 Sets the minor number of the ``os version''. Defaults to 0.
2577 [This option is specific to the i386 PE targeted port of the linker]
2579 @kindex --minor-subsystem-version
2580 @item --minor-subsystem-version @var{value}
2581 Sets the minor number of the ``subsystem version''. Defaults to 0.
2582 [This option is specific to the i386 PE targeted port of the linker]
2584 @cindex DEF files, creating
2585 @cindex DLLs, creating
2586 @kindex --output-def
2587 @item --output-def @var{file}
2588 The linker will create the file @var{file} which will contain a DEF
2589 file corresponding to the DLL the linker is generating. This DEF file
2590 (which should be called @code{*.def}) may be used to create an import
2591 library with @code{dlltool} or may be used as a reference to
2592 automatically or implicitly exported symbols.
2593 [This option is specific to the i386 PE targeted port of the linker]
2595 @cindex DLLs, creating
2596 @kindex --enable-auto-image-base
2597 @item --enable-auto-image-base
2598 @itemx --enable-auto-image-base=@var{value}
2599 Automatically choose the image base for DLLs, optionally starting with base
2600 @var{value}, unless one is specified using the @code{--image-base} argument.
2601 By using a hash generated from the dllname to create unique image bases
2602 for each DLL, in-memory collisions and relocations which can delay program
2603 execution are avoided.
2604 [This option is specific to the i386 PE targeted port of the linker]
2606 @kindex --disable-auto-image-base
2607 @item --disable-auto-image-base
2608 Do not automatically generate a unique image base. If there is no
2609 user-specified image base (@code{--image-base}) then use the platform
2611 [This option is specific to the i386 PE targeted port of the linker]
2613 @cindex DLLs, linking to
2614 @kindex --dll-search-prefix
2615 @item --dll-search-prefix @var{string}
2616 When linking dynamically to a dll without an import library,
2617 search for @code{<string><basename>.dll} in preference to
2618 @code{lib<basename>.dll}. This behaviour allows easy distinction
2619 between DLLs built for the various "subplatforms": native, cygwin,
2620 uwin, pw, etc. For instance, cygwin DLLs typically use
2621 @code{--dll-search-prefix=cyg}.
2622 [This option is specific to the i386 PE targeted port of the linker]
2624 @kindex --enable-auto-import
2625 @item --enable-auto-import
2626 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2627 DATA imports from DLLs, and create the necessary thunking symbols when
2628 building the import libraries with those DATA exports. Note: Use of the
2629 'auto-import' extension will cause the text section of the image file
2630 to be made writable. This does not conform to the PE-COFF format
2631 specification published by Microsoft.
2633 Note - use of the 'auto-import' extension will also cause read only
2634 data which would normally be placed into the .rdata section to be
2635 placed into the .data section instead. This is in order to work
2636 around a problem with consts that is described here:
2637 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2639 Using 'auto-import' generally will 'just work' -- but sometimes you may
2642 "variable '<var>' can't be auto-imported. Please read the
2643 documentation for ld's @code{--enable-auto-import} for details."
2645 This message occurs when some (sub)expression accesses an address
2646 ultimately given by the sum of two constants (Win32 import tables only
2647 allow one). Instances where this may occur include accesses to member
2648 fields of struct variables imported from a DLL, as well as using a
2649 constant index into an array variable imported from a DLL. Any
2650 multiword variable (arrays, structs, long long, etc) may trigger
2651 this error condition. However, regardless of the exact data type
2652 of the offending exported variable, ld will always detect it, issue
2653 the warning, and exit.
2655 There are several ways to address this difficulty, regardless of the
2656 data type of the exported variable:
2658 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2659 of adjusting references in your client code for runtime environment, so
2660 this method works only when runtime environment supports this feature.
2662 A second solution is to force one of the 'constants' to be a variable --
2663 that is, unknown and un-optimizable at compile time. For arrays,
2664 there are two possibilities: a) make the indexee (the array's address)
2665 a variable, or b) make the 'constant' index a variable. Thus:
2668 extern type extern_array[];
2670 @{ volatile type *t=extern_array; t[1] @}
2676 extern type extern_array[];
2678 @{ volatile int t=1; extern_array[t] @}
2681 For structs (and most other multiword data types) the only option
2682 is to make the struct itself (or the long long, or the ...) variable:
2685 extern struct s extern_struct;
2686 extern_struct.field -->
2687 @{ volatile struct s *t=&extern_struct; t->field @}
2693 extern long long extern_ll;
2695 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2698 A third method of dealing with this difficulty is to abandon
2699 'auto-import' for the offending symbol and mark it with
2700 @code{__declspec(dllimport)}. However, in practice that
2701 requires using compile-time #defines to indicate whether you are
2702 building a DLL, building client code that will link to the DLL, or
2703 merely building/linking to a static library. In making the choice
2704 between the various methods of resolving the 'direct address with
2705 constant offset' problem, you should consider typical real-world usage:
2713 void main(int argc, char **argv)@{
2714 printf("%d\n",arr[1]);
2724 void main(int argc, char **argv)@{
2725 /* This workaround is for win32 and cygwin; do not "optimize" */
2726 volatile int *parr = arr;
2727 printf("%d\n",parr[1]);
2734 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2735 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2736 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2737 #define FOO_IMPORT __declspec(dllimport)
2741 extern FOO_IMPORT int arr[];
2744 void main(int argc, char **argv)@{
2745 printf("%d\n",arr[1]);
2749 A fourth way to avoid this problem is to re-code your
2750 library to use a functional interface rather than a data interface
2751 for the offending variables (e.g. set_foo() and get_foo() accessor
2753 [This option is specific to the i386 PE targeted port of the linker]
2755 @kindex --disable-auto-import
2756 @item --disable-auto-import
2757 Do not attempt to do sophisticated linking of @code{_symbol} to
2758 @code{__imp__symbol} for DATA imports from DLLs.
2759 [This option is specific to the i386 PE targeted port of the linker]
2761 @kindex --enable-runtime-pseudo-reloc
2762 @item --enable-runtime-pseudo-reloc
2763 If your code contains expressions described in --enable-auto-import section,
2764 that is, DATA imports from DLL with non-zero offset, this switch will create
2765 a vector of 'runtime pseudo relocations' which can be used by runtime
2766 environment to adjust references to such data in your client code.
2767 [This option is specific to the i386 PE targeted port of the linker]
2769 @kindex --disable-runtime-pseudo-reloc
2770 @item --disable-runtime-pseudo-reloc
2771 Do not create pseudo relocations for non-zero offset DATA imports from
2773 [This option is specific to the i386 PE targeted port of the linker]
2775 @kindex --enable-extra-pe-debug
2776 @item --enable-extra-pe-debug
2777 Show additional debug info related to auto-import symbol thunking.
2778 [This option is specific to the i386 PE targeted port of the linker]
2780 @kindex --section-alignment
2781 @item --section-alignment
2782 Sets the section alignment. Sections in memory will always begin at
2783 addresses which are a multiple of this number. Defaults to 0x1000.
2784 [This option is specific to the i386 PE targeted port of the linker]
2788 @item --stack @var{reserve}
2789 @itemx --stack @var{reserve},@var{commit}
2790 Specify the number of bytes of memory to reserve (and optionally commit)
2791 to be used as stack for this program. The default is 2MB reserved, 4K
2793 [This option is specific to the i386 PE targeted port of the linker]
2796 @item --subsystem @var{which}
2797 @itemx --subsystem @var{which}:@var{major}
2798 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2799 Specifies the subsystem under which your program will execute. The
2800 legal values for @var{which} are @code{native}, @code{windows},
2801 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2802 the subsystem version also. Numeric values are also accepted for
2804 [This option is specific to the i386 PE targeted port of the linker]
2806 The following options set flags in the @code{DllCharacteristics} field
2807 of the PE file header:
2808 [These options are specific to PE targeted ports of the linker]
2810 @kindex --high-entropy-va
2811 @item --high-entropy-va
2812 Image is compatible with 64-bit address space layout randomization
2815 @kindex --dynamicbase
2817 The image base address may be relocated using address space layout
2818 randomization (ASLR). This feature was introduced with MS Windows
2819 Vista for i386 PE targets.
2821 @kindex --forceinteg
2823 Code integrity checks are enforced.
2827 The image is compatible with the Data Execution Prevention.
2828 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2830 @kindex --no-isolation
2831 @item --no-isolation
2832 Although the image understands isolation, do not isolate the image.
2836 The image does not use SEH. No SE handler may be called from
2841 Do not bind this image.
2845 The driver uses the MS Windows Driver Model.
2849 The image is Terminal Server aware.
2851 @kindex --insert-timestamp
2852 @item --insert-timestamp
2853 @itemx --no-insert-timestamp
2854 Insert a real timestamp into the image. This is the default behaviour
2855 as it matches legacy code and it means that the image will work with
2856 other, proprietary tools. The problem with this default is that it
2857 will result in slightly different images being produced each time the
2858 same sources are linked. The option @option{--no-insert-timestamp}
2859 can be used to insert a zero value for the timestamp, this ensuring
2860 that binaries produced from identical sources will compare
2867 @subsection Options specific to C6X uClinux targets
2869 @c man begin OPTIONS
2871 The C6X uClinux target uses a binary format called DSBT to support shared
2872 libraries. Each shared library in the system needs to have a unique index;
2873 all executables use an index of 0.
2878 @item --dsbt-size @var{size}
2879 This option sets the number of entries in the DSBT of the current executable
2880 or shared library to @var{size}. The default is to create a table with 64
2883 @kindex --dsbt-index
2884 @item --dsbt-index @var{index}
2885 This option sets the DSBT index of the current executable or shared library
2886 to @var{index}. The default is 0, which is appropriate for generating
2887 executables. If a shared library is generated with a DSBT index of 0, the
2888 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
2890 @kindex --no-merge-exidx-entries
2891 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
2892 exidx entries in frame unwind info.
2900 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
2902 @c man begin OPTIONS
2904 The 68HC11 and 68HC12 linkers support specific options to control the
2905 memory bank switching mapping and trampoline code generation.
2909 @kindex --no-trampoline
2910 @item --no-trampoline
2911 This option disables the generation of trampoline. By default a trampoline
2912 is generated for each far function which is called using a @code{jsr}
2913 instruction (this happens when a pointer to a far function is taken).
2915 @kindex --bank-window
2916 @item --bank-window @var{name}
2917 This option indicates to the linker the name of the memory region in
2918 the @samp{MEMORY} specification that describes the memory bank window.
2919 The definition of such region is then used by the linker to compute
2920 paging and addresses within the memory window.
2928 @subsection Options specific to Motorola 68K target
2930 @c man begin OPTIONS
2932 The following options are supported to control handling of GOT generation
2933 when linking for 68K targets.
2938 @item --got=@var{type}
2939 This option tells the linker which GOT generation scheme to use.
2940 @var{type} should be one of @samp{single}, @samp{negative},
2941 @samp{multigot} or @samp{target}. For more information refer to the
2942 Info entry for @file{ld}.
2950 @subsection Options specific to MIPS targets
2952 @c man begin OPTIONS
2954 The following options are supported to control microMIPS instruction
2955 generation when linking for MIPS targets.
2963 These options control the choice of microMIPS instructions used in code
2964 generated by the linker, such as that in the PLT or lazy binding stubs,
2965 or in relaxation. If @samp{--insn32} is used, then the linker only uses
2966 32-bit instruction encodings. By default or if @samp{--no-insn32} is
2967 used, all instruction encodings are used, including 16-bit ones where
2977 @section Environment Variables
2979 @c man begin ENVIRONMENT
2981 You can change the behaviour of @command{ld} with the environment variables
2982 @ifclear SingleFormat
2985 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2987 @ifclear SingleFormat
2989 @cindex default input format
2990 @code{GNUTARGET} determines the input-file object format if you don't
2991 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
2992 of the BFD names for an input format (@pxref{BFD}). If there is no
2993 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
2994 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2995 attempts to discover the input format by examining binary input files;
2996 this method often succeeds, but there are potential ambiguities, since
2997 there is no method of ensuring that the magic number used to specify
2998 object-file formats is unique. However, the configuration procedure for
2999 BFD on each system places the conventional format for that system first
3000 in the search-list, so ambiguities are resolved in favor of convention.
3004 @cindex default emulation
3005 @cindex emulation, default
3006 @code{LDEMULATION} determines the default emulation if you don't use the
3007 @samp{-m} option. The emulation can affect various aspects of linker
3008 behaviour, particularly the default linker script. You can list the
3009 available emulations with the @samp{--verbose} or @samp{-V} options. If
3010 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3011 variable is not defined, the default emulation depends upon how the
3012 linker was configured.
3014 @kindex COLLECT_NO_DEMANGLE
3015 @cindex demangling, default
3016 Normally, the linker will default to demangling symbols. However, if
3017 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3018 default to not demangling symbols. This environment variable is used in
3019 a similar fashion by the @code{gcc} linker wrapper program. The default
3020 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3027 @chapter Linker Scripts
3030 @cindex linker scripts
3031 @cindex command files
3032 Every link is controlled by a @dfn{linker script}. This script is
3033 written in the linker command language.
3035 The main purpose of the linker script is to describe how the sections in
3036 the input files should be mapped into the output file, and to control
3037 the memory layout of the output file. Most linker scripts do nothing
3038 more than this. However, when necessary, the linker script can also
3039 direct the linker to perform many other operations, using the commands
3042 The linker always uses a linker script. If you do not supply one
3043 yourself, the linker will use a default script that is compiled into the
3044 linker executable. You can use the @samp{--verbose} command line option
3045 to display the default linker script. Certain command line options,
3046 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3048 You may supply your own linker script by using the @samp{-T} command
3049 line option. When you do this, your linker script will replace the
3050 default linker script.
3052 You may also use linker scripts implicitly by naming them as input files
3053 to the linker, as though they were files to be linked. @xref{Implicit
3057 * Basic Script Concepts:: Basic Linker Script Concepts
3058 * Script Format:: Linker Script Format
3059 * Simple Example:: Simple Linker Script Example
3060 * Simple Commands:: Simple Linker Script Commands
3061 * Assignments:: Assigning Values to Symbols
3062 * SECTIONS:: SECTIONS Command
3063 * MEMORY:: MEMORY Command
3064 * PHDRS:: PHDRS Command
3065 * VERSION:: VERSION Command
3066 * Expressions:: Expressions in Linker Scripts
3067 * Implicit Linker Scripts:: Implicit Linker Scripts
3070 @node Basic Script Concepts
3071 @section Basic Linker Script Concepts
3072 @cindex linker script concepts
3073 We need to define some basic concepts and vocabulary in order to
3074 describe the linker script language.
3076 The linker combines input files into a single output file. The output
3077 file and each input file are in a special data format known as an
3078 @dfn{object file format}. Each file is called an @dfn{object file}.
3079 The output file is often called an @dfn{executable}, but for our
3080 purposes we will also call it an object file. Each object file has,
3081 among other things, a list of @dfn{sections}. We sometimes refer to a
3082 section in an input file as an @dfn{input section}; similarly, a section
3083 in the output file is an @dfn{output section}.
3085 Each section in an object file has a name and a size. Most sections
3086 also have an associated block of data, known as the @dfn{section
3087 contents}. A section may be marked as @dfn{loadable}, which means that
3088 the contents should be loaded into memory when the output file is run.
3089 A section with no contents may be @dfn{allocatable}, which means that an
3090 area in memory should be set aside, but nothing in particular should be
3091 loaded there (in some cases this memory must be zeroed out). A section
3092 which is neither loadable nor allocatable typically contains some sort
3093 of debugging information.
3095 Every loadable or allocatable output section has two addresses. The
3096 first is the @dfn{VMA}, or virtual memory address. This is the address
3097 the section will have when the output file is run. The second is the
3098 @dfn{LMA}, or load memory address. This is the address at which the
3099 section will be loaded. In most cases the two addresses will be the
3100 same. An example of when they might be different is when a data section
3101 is loaded into ROM, and then copied into RAM when the program starts up
3102 (this technique is often used to initialize global variables in a ROM
3103 based system). In this case the ROM address would be the LMA, and the
3104 RAM address would be the VMA.
3106 You can see the sections in an object file by using the @code{objdump}
3107 program with the @samp{-h} option.
3109 Every object file also has a list of @dfn{symbols}, known as the
3110 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3111 has a name, and each defined symbol has an address, among other
3112 information. If you compile a C or C++ program into an object file, you
3113 will get a defined symbol for every defined function and global or
3114 static variable. Every undefined function or global variable which is
3115 referenced in the input file will become an undefined symbol.
3117 You can see the symbols in an object file by using the @code{nm}
3118 program, or by using the @code{objdump} program with the @samp{-t}
3122 @section Linker Script Format
3123 @cindex linker script format
3124 Linker scripts are text files.
3126 You write a linker script as a series of commands. Each command is
3127 either a keyword, possibly followed by arguments, or an assignment to a
3128 symbol. You may separate commands using semicolons. Whitespace is
3131 Strings such as file or format names can normally be entered directly.
3132 If the file name contains a character such as a comma which would
3133 otherwise serve to separate file names, you may put the file name in
3134 double quotes. There is no way to use a double quote character in a
3137 You may include comments in linker scripts just as in C, delimited by
3138 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3141 @node Simple Example
3142 @section Simple Linker Script Example
3143 @cindex linker script example
3144 @cindex example of linker script
3145 Many linker scripts are fairly simple.
3147 The simplest possible linker script has just one command:
3148 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3149 memory layout of the output file.
3151 The @samp{SECTIONS} command is a powerful command. Here we will
3152 describe a simple use of it. Let's assume your program consists only of
3153 code, initialized data, and uninitialized data. These will be in the
3154 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3155 Let's assume further that these are the only sections which appear in
3158 For this example, let's say that the code should be loaded at address
3159 0x10000, and that the data should start at address 0x8000000. Here is a
3160 linker script which will do that:
3165 .text : @{ *(.text) @}
3167 .data : @{ *(.data) @}
3168 .bss : @{ *(.bss) @}
3172 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3173 followed by a series of symbol assignments and output section
3174 descriptions enclosed in curly braces.
3176 The first line inside the @samp{SECTIONS} command of the above example
3177 sets the value of the special symbol @samp{.}, which is the location
3178 counter. If you do not specify the address of an output section in some
3179 other way (other ways are described later), the address is set from the
3180 current value of the location counter. The location counter is then
3181 incremented by the size of the output section. At the start of the
3182 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3184 The second line defines an output section, @samp{.text}. The colon is
3185 required syntax which may be ignored for now. Within the curly braces
3186 after the output section name, you list the names of the input sections
3187 which should be placed into this output section. The @samp{*} is a
3188 wildcard which matches any file name. The expression @samp{*(.text)}
3189 means all @samp{.text} input sections in all input files.
3191 Since the location counter is @samp{0x10000} when the output section
3192 @samp{.text} is defined, the linker will set the address of the
3193 @samp{.text} section in the output file to be @samp{0x10000}.
3195 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3196 the output file. The linker will place the @samp{.data} output section
3197 at address @samp{0x8000000}. After the linker places the @samp{.data}
3198 output section, the value of the location counter will be
3199 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3200 effect is that the linker will place the @samp{.bss} output section
3201 immediately after the @samp{.data} output section in memory.
3203 The linker will ensure that each output section has the required
3204 alignment, by increasing the location counter if necessary. In this
3205 example, the specified addresses for the @samp{.text} and @samp{.data}
3206 sections will probably satisfy any alignment constraints, but the linker
3207 may have to create a small gap between the @samp{.data} and @samp{.bss}
3210 That's it! That's a simple and complete linker script.
3212 @node Simple Commands
3213 @section Simple Linker Script Commands
3214 @cindex linker script simple commands
3215 In this section we describe the simple linker script commands.
3218 * Entry Point:: Setting the entry point
3219 * File Commands:: Commands dealing with files
3220 @ifclear SingleFormat
3221 * Format Commands:: Commands dealing with object file formats
3224 * REGION_ALIAS:: Assign alias names to memory regions
3225 * Miscellaneous Commands:: Other linker script commands
3229 @subsection Setting the Entry Point
3230 @kindex ENTRY(@var{symbol})
3231 @cindex start of execution
3232 @cindex first instruction
3234 The first instruction to execute in a program is called the @dfn{entry
3235 point}. You can use the @code{ENTRY} linker script command to set the
3236 entry point. The argument is a symbol name:
3241 There are several ways to set the entry point. The linker will set the
3242 entry point by trying each of the following methods in order, and
3243 stopping when one of them succeeds:
3246 the @samp{-e} @var{entry} command-line option;
3248 the @code{ENTRY(@var{symbol})} command in a linker script;
3250 the value of a target specific symbol, if it is defined; For many
3251 targets this is @code{start}, but PE and BeOS based systems for example
3252 check a list of possible entry symbols, matching the first one found.
3254 the address of the first byte of the @samp{.text} section, if present;
3256 The address @code{0}.
3260 @subsection Commands Dealing with Files
3261 @cindex linker script file commands
3262 Several linker script commands deal with files.
3265 @item INCLUDE @var{filename}
3266 @kindex INCLUDE @var{filename}
3267 @cindex including a linker script
3268 Include the linker script @var{filename} at this point. The file will
3269 be searched for in the current directory, and in any directory specified
3270 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3273 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3274 @code{SECTIONS} commands, or in output section descriptions.
3276 @item INPUT(@var{file}, @var{file}, @dots{})
3277 @itemx INPUT(@var{file} @var{file} @dots{})
3278 @kindex INPUT(@var{files})
3279 @cindex input files in linker scripts
3280 @cindex input object files in linker scripts
3281 @cindex linker script input object files
3282 The @code{INPUT} command directs the linker to include the named files
3283 in the link, as though they were named on the command line.
3285 For example, if you always want to include @file{subr.o} any time you do
3286 a link, but you can't be bothered to put it on every link command line,
3287 then you can put @samp{INPUT (subr.o)} in your linker script.
3289 In fact, if you like, you can list all of your input files in the linker
3290 script, and then invoke the linker with nothing but a @samp{-T} option.
3292 In case a @dfn{sysroot prefix} is configured, and the filename starts
3293 with the @samp{/} character, and the script being processed was
3294 located inside the @dfn{sysroot prefix}, the filename will be looked
3295 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3296 open the file in the current directory. If it is not found, the
3297 linker will search through the archive library search path.
3298 The @dfn{sysroot prefix} can also be forced by specifying @code{=}
3299 as the first character in the filename path. See also the
3300 description of @samp{-L} in @ref{Options,,Command Line Options}.
3302 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3303 name to @code{lib@var{file}.a}, as with the command line argument
3306 When you use the @code{INPUT} command in an implicit linker script, the
3307 files will be included in the link at the point at which the linker
3308 script file is included. This can affect archive searching.
3310 @item GROUP(@var{file}, @var{file}, @dots{})
3311 @itemx GROUP(@var{file} @var{file} @dots{})
3312 @kindex GROUP(@var{files})
3313 @cindex grouping input files
3314 The @code{GROUP} command is like @code{INPUT}, except that the named
3315 files should all be archives, and they are searched repeatedly until no
3316 new undefined references are created. See the description of @samp{-(}
3317 in @ref{Options,,Command Line Options}.
3319 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3320 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3321 @kindex AS_NEEDED(@var{files})
3322 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3323 commands, among other filenames. The files listed will be handled
3324 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3325 with the exception of ELF shared libraries, that will be added only
3326 when they are actually needed. This construct essentially enables
3327 @option{--as-needed} option for all the files listed inside of it
3328 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3331 @item OUTPUT(@var{filename})
3332 @kindex OUTPUT(@var{filename})
3333 @cindex output file name in linker script
3334 The @code{OUTPUT} command names the output file. Using
3335 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3336 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3337 Line Options}). If both are used, the command line option takes
3340 You can use the @code{OUTPUT} command to define a default name for the
3341 output file other than the usual default of @file{a.out}.
3343 @item SEARCH_DIR(@var{path})
3344 @kindex SEARCH_DIR(@var{path})
3345 @cindex library search path in linker script
3346 @cindex archive search path in linker script
3347 @cindex search path in linker script
3348 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3349 @command{ld} looks for archive libraries. Using
3350 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3351 on the command line (@pxref{Options,,Command Line Options}). If both
3352 are used, then the linker will search both paths. Paths specified using
3353 the command line option are searched first.
3355 @item STARTUP(@var{filename})
3356 @kindex STARTUP(@var{filename})
3357 @cindex first input file
3358 The @code{STARTUP} command is just like the @code{INPUT} command, except
3359 that @var{filename} will become the first input file to be linked, as
3360 though it were specified first on the command line. This may be useful
3361 when using a system in which the entry point is always the start of the
3365 @ifclear SingleFormat
3366 @node Format Commands
3367 @subsection Commands Dealing with Object File Formats
3368 A couple of linker script commands deal with object file formats.
3371 @item OUTPUT_FORMAT(@var{bfdname})
3372 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3373 @kindex OUTPUT_FORMAT(@var{bfdname})
3374 @cindex output file format in linker script
3375 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3376 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3377 exactly like using @samp{--oformat @var{bfdname}} on the command line
3378 (@pxref{Options,,Command Line Options}). If both are used, the command
3379 line option takes precedence.
3381 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3382 formats based on the @samp{-EB} and @samp{-EL} command line options.
3383 This permits the linker script to set the output format based on the
3386 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3387 will be the first argument, @var{default}. If @samp{-EB} is used, the
3388 output format will be the second argument, @var{big}. If @samp{-EL} is
3389 used, the output format will be the third argument, @var{little}.
3391 For example, the default linker script for the MIPS ELF target uses this
3394 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3396 This says that the default format for the output file is
3397 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3398 option, the output file will be created in the @samp{elf32-littlemips}
3401 @item TARGET(@var{bfdname})
3402 @kindex TARGET(@var{bfdname})
3403 @cindex input file format in linker script
3404 The @code{TARGET} command names the BFD format to use when reading input
3405 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3406 This command is like using @samp{-b @var{bfdname}} on the command line
3407 (@pxref{Options,,Command Line Options}). If the @code{TARGET} command
3408 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3409 command is also used to set the format for the output file. @xref{BFD}.
3414 @subsection Assign alias names to memory regions
3415 @kindex REGION_ALIAS(@var{alias}, @var{region})
3416 @cindex region alias
3417 @cindex region names
3419 Alias names can be added to existing memory regions created with the
3420 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3423 REGION_ALIAS(@var{alias}, @var{region})
3426 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3427 memory region @var{region}. This allows a flexible mapping of output sections
3428 to memory regions. An example follows.
3430 Suppose we have an application for embedded systems which come with various
3431 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3432 that allows code execution or data storage. Some may have a read-only,
3433 non-volatile memory @code{ROM} that allows code execution and read-only data
3434 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3435 read-only data access and no code execution capability. We have four output
3440 @code{.text} program code;
3442 @code{.rodata} read-only data;
3444 @code{.data} read-write initialized data;
3446 @code{.bss} read-write zero initialized data.
3449 The goal is to provide a linker command file that contains a system independent
3450 part defining the output sections and a system dependent part mapping the
3451 output sections to the memory regions available on the system. Our embedded
3452 systems come with three different memory setups @code{A}, @code{B} and
3454 @multitable @columnfractions .25 .25 .25 .25
3455 @item Section @tab Variant A @tab Variant B @tab Variant C
3456 @item .text @tab RAM @tab ROM @tab ROM
3457 @item .rodata @tab RAM @tab ROM @tab ROM2
3458 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3459 @item .bss @tab RAM @tab RAM @tab RAM
3461 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3462 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3463 the load address of the @code{.data} section starts in all three variants at
3464 the end of the @code{.rodata} section.
3466 The base linker script that deals with the output sections follows. It
3467 includes the system dependent @code{linkcmds.memory} file that describes the
3470 INCLUDE linkcmds.memory
3483 .data : AT (rodata_end)
3488 data_size = SIZEOF(.data);
3489 data_load_start = LOADADDR(.data);
3497 Now we need three different @code{linkcmds.memory} files to define memory
3498 regions and alias names. The content of @code{linkcmds.memory} for the three
3499 variants @code{A}, @code{B} and @code{C}:
3502 Here everything goes into the @code{RAM}.
3506 RAM : ORIGIN = 0, LENGTH = 4M
3509 REGION_ALIAS("REGION_TEXT", RAM);
3510 REGION_ALIAS("REGION_RODATA", RAM);
3511 REGION_ALIAS("REGION_DATA", RAM);
3512 REGION_ALIAS("REGION_BSS", RAM);
3515 Program code and read-only data go into the @code{ROM}. Read-write data goes
3516 into the @code{RAM}. An image of the initialized data is loaded into the
3517 @code{ROM} and will be copied during system start into the @code{RAM}.
3521 ROM : ORIGIN = 0, LENGTH = 3M
3522 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3525 REGION_ALIAS("REGION_TEXT", ROM);
3526 REGION_ALIAS("REGION_RODATA", ROM);
3527 REGION_ALIAS("REGION_DATA", RAM);
3528 REGION_ALIAS("REGION_BSS", RAM);
3531 Program code goes into the @code{ROM}. Read-only data goes into the
3532 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3533 initialized data is loaded into the @code{ROM2} and will be copied during
3534 system start into the @code{RAM}.
3538 ROM : ORIGIN = 0, LENGTH = 2M
3539 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3540 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3543 REGION_ALIAS("REGION_TEXT", ROM);
3544 REGION_ALIAS("REGION_RODATA", ROM2);
3545 REGION_ALIAS("REGION_DATA", RAM);
3546 REGION_ALIAS("REGION_BSS", RAM);
3550 It is possible to write a common system initialization routine to copy the
3551 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3556 extern char data_start [];
3557 extern char data_size [];
3558 extern char data_load_start [];
3560 void copy_data(void)
3562 if (data_start != data_load_start)
3564 memcpy(data_start, data_load_start, (size_t) data_size);
3569 @node Miscellaneous Commands
3570 @subsection Other Linker Script Commands
3571 There are a few other linker scripts commands.
3574 @item ASSERT(@var{exp}, @var{message})
3576 @cindex assertion in linker script
3577 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3578 with an error code, and print @var{message}.
3580 Note that assertions are checked before the final stages of linking
3581 take place. This means that expressions involving symbols PROVIDEd
3582 inside section definitions will fail if the user has not set values
3583 for those symbols. The only exception to this rule is PROVIDEd
3584 symbols that just reference dot. Thus an assertion like this:
3589 PROVIDE (__stack = .);
3590 PROVIDE (__stack_size = 0x100);
3591 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3595 will fail if @code{__stack_size} is not defined elsewhere. Symbols
3596 PROVIDEd outside of section definitions are evaluated earlier, so they
3597 can be used inside ASSERTions. Thus:
3600 PROVIDE (__stack_size = 0x100);
3603 PROVIDE (__stack = .);
3604 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3610 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3612 @cindex undefined symbol in linker script
3613 Force @var{symbol} to be entered in the output file as an undefined
3614 symbol. Doing this may, for example, trigger linking of additional
3615 modules from standard libraries. You may list several @var{symbol}s for
3616 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3617 command has the same effect as the @samp{-u} command-line option.
3619 @item FORCE_COMMON_ALLOCATION
3620 @kindex FORCE_COMMON_ALLOCATION
3621 @cindex common allocation in linker script
3622 This command has the same effect as the @samp{-d} command-line option:
3623 to make @command{ld} assign space to common symbols even if a relocatable
3624 output file is specified (@samp{-r}).
3626 @item INHIBIT_COMMON_ALLOCATION
3627 @kindex INHIBIT_COMMON_ALLOCATION
3628 @cindex common allocation in linker script
3629 This command has the same effect as the @samp{--no-define-common}
3630 command-line option: to make @code{ld} omit the assignment of addresses
3631 to common symbols even for a non-relocatable output file.
3633 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3635 @cindex insert user script into default script
3636 This command is typically used in a script specified by @samp{-T} to
3637 augment the default @code{SECTIONS} with, for example, overlays. It
3638 inserts all prior linker script statements after (or before)
3639 @var{output_section}, and also causes @samp{-T} to not override the
3640 default linker script. The exact insertion point is as for orphan
3641 sections. @xref{Location Counter}. The insertion happens after the
3642 linker has mapped input sections to output sections. Prior to the
3643 insertion, since @samp{-T} scripts are parsed before the default
3644 linker script, statements in the @samp{-T} script occur before the
3645 default linker script statements in the internal linker representation
3646 of the script. In particular, input section assignments will be made
3647 to @samp{-T} output sections before those in the default script. Here
3648 is an example of how a @samp{-T} script using @code{INSERT} might look:
3655 .ov1 @{ ov1*(.text) @}
3656 .ov2 @{ ov2*(.text) @}
3662 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3663 @kindex NOCROSSREFS(@var{sections})
3664 @cindex cross references
3665 This command may be used to tell @command{ld} to issue an error about any
3666 references among certain output sections.
3668 In certain types of programs, particularly on embedded systems when
3669 using overlays, when one section is loaded into memory, another section
3670 will not be. Any direct references between the two sections would be
3671 errors. For example, it would be an error if code in one section called
3672 a function defined in the other section.
3674 The @code{NOCROSSREFS} command takes a list of output section names. If
3675 @command{ld} detects any cross references between the sections, it reports
3676 an error and returns a non-zero exit status. Note that the
3677 @code{NOCROSSREFS} command uses output section names, not input section
3680 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
3681 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
3682 @cindex cross references
3683 This command may be used to tell @command{ld} to issue an error about any
3684 references to one section from a list of other sections.
3686 The @code{NOCROSSREFS} command is useful when ensuring that two or more
3687 output sections are entirely independent but there are situations where
3688 a one-way dependency is needed. For example, in a multi-core application
3689 there may be shared code that can be called from each core but for safety
3690 must never call back.
3692 The @code{NOCROSSREFS_TO} command takes a list of output section names.
3693 The first section can not be referenced from any of the other sections.
3694 If @command{ld} detects any references to the first section from any of
3695 the other sections, it reports an error and returns a non-zero exit
3696 status. Note that the @code{NOCROSSREFS_TO} command uses output section
3697 names, not input section names.
3699 @ifclear SingleFormat
3700 @item OUTPUT_ARCH(@var{bfdarch})
3701 @kindex OUTPUT_ARCH(@var{bfdarch})
3702 @cindex machine architecture
3703 @cindex architecture
3704 Specify a particular output machine architecture. The argument is one
3705 of the names used by the BFD library (@pxref{BFD}). You can see the
3706 architecture of an object file by using the @code{objdump} program with
3707 the @samp{-f} option.
3710 @item LD_FEATURE(@var{string})
3711 @kindex LD_FEATURE(@var{string})
3712 This command may be used to modify @command{ld} behavior. If
3713 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3714 in a script are simply treated as numbers everywhere.
3715 @xref{Expression Section}.
3719 @section Assigning Values to Symbols
3720 @cindex assignment in scripts
3721 @cindex symbol definition, scripts
3722 @cindex variables, defining
3723 You may assign a value to a symbol in a linker script. This will define
3724 the symbol and place it into the symbol table with a global scope.
3727 * Simple Assignments:: Simple Assignments
3730 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3731 * Source Code Reference:: How to use a linker script defined symbol in source code
3734 @node Simple Assignments
3735 @subsection Simple Assignments
3737 You may assign to a symbol using any of the C assignment operators:
3740 @item @var{symbol} = @var{expression} ;
3741 @itemx @var{symbol} += @var{expression} ;
3742 @itemx @var{symbol} -= @var{expression} ;
3743 @itemx @var{symbol} *= @var{expression} ;
3744 @itemx @var{symbol} /= @var{expression} ;
3745 @itemx @var{symbol} <<= @var{expression} ;
3746 @itemx @var{symbol} >>= @var{expression} ;
3747 @itemx @var{symbol} &= @var{expression} ;
3748 @itemx @var{symbol} |= @var{expression} ;
3751 The first case will define @var{symbol} to the value of
3752 @var{expression}. In the other cases, @var{symbol} must already be
3753 defined, and the value will be adjusted accordingly.
3755 The special symbol name @samp{.} indicates the location counter. You
3756 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3758 The semicolon after @var{expression} is required.
3760 Expressions are defined below; see @ref{Expressions}.
3762 You may write symbol assignments as commands in their own right, or as
3763 statements within a @code{SECTIONS} command, or as part of an output
3764 section description in a @code{SECTIONS} command.
3766 The section of the symbol will be set from the section of the
3767 expression; for more information, see @ref{Expression Section}.
3769 Here is an example showing the three different places that symbol
3770 assignments may be used:
3781 _bdata = (. + 3) & ~ 3;
3782 .data : @{ *(.data) @}
3786 In this example, the symbol @samp{floating_point} will be defined as
3787 zero. The symbol @samp{_etext} will be defined as the address following
3788 the last @samp{.text} input section. The symbol @samp{_bdata} will be
3789 defined as the address following the @samp{.text} output section aligned
3790 upward to a 4 byte boundary.
3795 For ELF targeted ports, define a symbol that will be hidden and won't be
3796 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
3798 Here is the example from @ref{Simple Assignments}, rewritten to use
3802 HIDDEN(floating_point = 0);
3810 HIDDEN(_bdata = (. + 3) & ~ 3);
3811 .data : @{ *(.data) @}
3815 In this case none of the three symbols will be visible outside this module.
3820 In some cases, it is desirable for a linker script to define a symbol
3821 only if it is referenced and is not defined by any object included in
3822 the link. For example, traditional linkers defined the symbol
3823 @samp{etext}. However, ANSI C requires that the user be able to use
3824 @samp{etext} as a function name without encountering an error. The
3825 @code{PROVIDE} keyword may be used to define a symbol, such as
3826 @samp{etext}, only if it is referenced but not defined. The syntax is
3827 @code{PROVIDE(@var{symbol} = @var{expression})}.
3829 Here is an example of using @code{PROVIDE} to define @samp{etext}:
3842 In this example, if the program defines @samp{_etext} (with a leading
3843 underscore), the linker will give a multiple definition error. If, on
3844 the other hand, the program defines @samp{etext} (with no leading
3845 underscore), the linker will silently use the definition in the program.
3846 If the program references @samp{etext} but does not define it, the
3847 linker will use the definition in the linker script.
3849 @node PROVIDE_HIDDEN
3850 @subsection PROVIDE_HIDDEN
3851 @cindex PROVIDE_HIDDEN
3852 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
3853 hidden and won't be exported.
3855 @node Source Code Reference
3856 @subsection Source Code Reference
3858 Accessing a linker script defined variable from source code is not
3859 intuitive. In particular a linker script symbol is not equivalent to
3860 a variable declaration in a high level language, it is instead a
3861 symbol that does not have a value.
3863 Before going further, it is important to note that compilers often
3864 transform names in the source code into different names when they are
3865 stored in the symbol table. For example, Fortran compilers commonly
3866 prepend or append an underscore, and C++ performs extensive @samp{name
3867 mangling}. Therefore there might be a discrepancy between the name
3868 of a variable as it is used in source code and the name of the same
3869 variable as it is defined in a linker script. For example in C a
3870 linker script variable might be referred to as:
3876 But in the linker script it might be defined as:
3882 In the remaining examples however it is assumed that no name
3883 transformation has taken place.
3885 When a symbol is declared in a high level language such as C, two
3886 things happen. The first is that the compiler reserves enough space
3887 in the program's memory to hold the @emph{value} of the symbol. The
3888 second is that the compiler creates an entry in the program's symbol
3889 table which holds the symbol's @emph{address}. ie the symbol table
3890 contains the address of the block of memory holding the symbol's
3891 value. So for example the following C declaration, at file scope:
3897 creates an entry called @samp{foo} in the symbol table. This entry
3898 holds the address of an @samp{int} sized block of memory where the
3899 number 1000 is initially stored.
3901 When a program references a symbol the compiler generates code that
3902 first accesses the symbol table to find the address of the symbol's
3903 memory block and then code to read the value from that memory block.
3910 looks up the symbol @samp{foo} in the symbol table, gets the address
3911 associated with this symbol and then writes the value 1 into that
3918 looks up the symbol @samp{foo} in the symbol table, gets its address
3919 and then copies this address into the block of memory associated with
3920 the variable @samp{a}.
3922 Linker scripts symbol declarations, by contrast, create an entry in
3923 the symbol table but do not assign any memory to them. Thus they are
3924 an address without a value. So for example the linker script definition:
3930 creates an entry in the symbol table called @samp{foo} which holds
3931 the address of memory location 1000, but nothing special is stored at
3932 address 1000. This means that you cannot access the @emph{value} of a
3933 linker script defined symbol - it has no value - all you can do is
3934 access the @emph{address} of a linker script defined symbol.
3936 Hence when you are using a linker script defined symbol in source code
3937 you should always take the address of the symbol, and never attempt to
3938 use its value. For example suppose you want to copy the contents of a
3939 section of memory called .ROM into a section called .FLASH and the
3940 linker script contains these declarations:
3944 start_of_ROM = .ROM;
3945 end_of_ROM = .ROM + sizeof (.ROM);
3946 start_of_FLASH = .FLASH;
3950 Then the C source code to perform the copy would be:
3954 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3956 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3960 Note the use of the @samp{&} operators. These are correct.
3961 Alternatively the symbols can be treated as the names of vectors or
3962 arrays and then the code will again work as expected:
3966 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
3968 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
3972 Note how using this method does not require the use of @samp{&}
3976 @section SECTIONS Command
3978 The @code{SECTIONS} command tells the linker how to map input sections
3979 into output sections, and how to place the output sections in memory.
3981 The format of the @code{SECTIONS} command is:
3985 @var{sections-command}
3986 @var{sections-command}
3991 Each @var{sections-command} may of be one of the following:
3995 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3997 a symbol assignment (@pxref{Assignments})
3999 an output section description
4001 an overlay description
4004 The @code{ENTRY} command and symbol assignments are permitted inside the
4005 @code{SECTIONS} command for convenience in using the location counter in
4006 those commands. This can also make the linker script easier to
4007 understand because you can use those commands at meaningful points in
4008 the layout of the output file.
4010 Output section descriptions and overlay descriptions are described
4013 If you do not use a @code{SECTIONS} command in your linker script, the
4014 linker will place each input section into an identically named output
4015 section in the order that the sections are first encountered in the
4016 input files. If all input sections are present in the first file, for
4017 example, the order of sections in the output file will match the order
4018 in the first input file. The first section will be at address zero.
4021 * Output Section Description:: Output section description
4022 * Output Section Name:: Output section name
4023 * Output Section Address:: Output section address
4024 * Input Section:: Input section description
4025 * Output Section Data:: Output section data
4026 * Output Section Keywords:: Output section keywords
4027 * Output Section Discarding:: Output section discarding
4028 * Output Section Attributes:: Output section attributes
4029 * Overlay Description:: Overlay description
4032 @node Output Section Description
4033 @subsection Output Section Description
4034 The full description of an output section looks like this:
4037 @var{section} [@var{address}] [(@var{type})] :
4039 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4040 [SUBALIGN(@var{subsection_align})]
4043 @var{output-section-command}
4044 @var{output-section-command}
4046 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4050 Most output sections do not use most of the optional section attributes.
4052 The whitespace around @var{section} is required, so that the section
4053 name is unambiguous. The colon and the curly braces are also required.
4054 The comma at the end may be required if a @var{fillexp} is used and
4055 the next @var{sections-command} looks like a continuation of the expression.
4056 The line breaks and other white space are optional.
4058 Each @var{output-section-command} may be one of the following:
4062 a symbol assignment (@pxref{Assignments})
4064 an input section description (@pxref{Input Section})
4066 data values to include directly (@pxref{Output Section Data})
4068 a special output section keyword (@pxref{Output Section Keywords})
4071 @node Output Section Name
4072 @subsection Output Section Name
4073 @cindex name, section
4074 @cindex section name
4075 The name of the output section is @var{section}. @var{section} must
4076 meet the constraints of your output format. In formats which only
4077 support a limited number of sections, such as @code{a.out}, the name
4078 must be one of the names supported by the format (@code{a.out}, for
4079 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4080 output format supports any number of sections, but with numbers and not
4081 names (as is the case for Oasys), the name should be supplied as a
4082 quoted numeric string. A section name may consist of any sequence of
4083 characters, but a name which contains any unusual characters such as
4084 commas must be quoted.
4086 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4089 @node Output Section Address
4090 @subsection Output Section Address
4091 @cindex address, section
4092 @cindex section address
4093 The @var{address} is an expression for the VMA (the virtual memory
4094 address) of the output section. This address is optional, but if it
4095 is provided then the output address will be set exactly as specified.
4097 If the output address is not specified then one will be chosen for the
4098 section, based on the heuristic below. This address will be adjusted
4099 to fit the alignment requirement of the output section. The
4100 alignment requirement is the strictest alignment of any input section
4101 contained within the output section.
4103 The output section address heuristic is as follows:
4107 If an output memory @var{region} is set for the section then it
4108 is added to this region and its address will be the next free address
4112 If the MEMORY command has been used to create a list of memory
4113 regions then the first region which has attributes compatible with the
4114 section is selected to contain it. The section's output address will
4115 be the next free address in that region; @ref{MEMORY}.
4118 If no memory regions were specified, or none match the section then
4119 the output address will be based on the current value of the location
4127 .text . : @{ *(.text) @}
4134 .text : @{ *(.text) @}
4138 are subtly different. The first will set the address of the
4139 @samp{.text} output section to the current value of the location
4140 counter. The second will set it to the current value of the location
4141 counter aligned to the strictest alignment of any of the @samp{.text}
4144 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4145 For example, if you want to align the section on a 0x10 byte boundary,
4146 so that the lowest four bits of the section address are zero, you could
4147 do something like this:
4149 .text ALIGN(0x10) : @{ *(.text) @}
4152 This works because @code{ALIGN} returns the current location counter
4153 aligned upward to the specified value.
4155 Specifying @var{address} for a section will change the value of the
4156 location counter, provided that the section is non-empty. (Empty
4157 sections are ignored).
4160 @subsection Input Section Description
4161 @cindex input sections
4162 @cindex mapping input sections to output sections
4163 The most common output section command is an input section description.
4165 The input section description is the most basic linker script operation.
4166 You use output sections to tell the linker how to lay out your program
4167 in memory. You use input section descriptions to tell the linker how to
4168 map the input files into your memory layout.
4171 * Input Section Basics:: Input section basics
4172 * Input Section Wildcards:: Input section wildcard patterns
4173 * Input Section Common:: Input section for common symbols
4174 * Input Section Keep:: Input section and garbage collection
4175 * Input Section Example:: Input section example
4178 @node Input Section Basics
4179 @subsubsection Input Section Basics
4180 @cindex input section basics
4181 An input section description consists of a file name optionally followed
4182 by a list of section names in parentheses.
4184 The file name and the section name may be wildcard patterns, which we
4185 describe further below (@pxref{Input Section Wildcards}).
4187 The most common input section description is to include all input
4188 sections with a particular name in the output section. For example, to
4189 include all input @samp{.text} sections, you would write:
4194 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4195 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4196 match all files except the ones specified in the EXCLUDE_FILE list. For
4199 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4201 will cause all .ctors sections from all files except @file{crtend.o} and
4202 @file{otherfile.o} to be included.
4204 There are two ways to include more than one section:
4210 The difference between these is the order in which the @samp{.text} and
4211 @samp{.rdata} input sections will appear in the output section. In the
4212 first example, they will be intermingled, appearing in the same order as
4213 they are found in the linker input. In the second example, all
4214 @samp{.text} input sections will appear first, followed by all
4215 @samp{.rdata} input sections.
4217 When using EXCLUDE_FILE with more than one section, the exclusion only
4218 applies to the section immediately following, for example:
4220 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4223 will cause all @samp{.text} sections from all files except
4224 @file{somefile.o} to be included, while all @samp{.rdata} sections
4225 from all files, including @file{somefile.o}, will be included. To
4226 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4227 should be modified to:
4229 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4232 You can specify a file name to include sections from a particular file.
4233 You would do this if one or more of your files contain special data that
4234 needs to be at a particular location in memory. For example:
4239 To refine the sections that are included based on the section flags
4240 of an input section, INPUT_SECTION_FLAGS may be used.
4242 Here is a simple example for using Section header flags for ELF sections:
4247 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4248 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4253 In this example, the output section @samp{.text} will be comprised of any
4254 input section matching the name *(.text) whose section header flags
4255 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4256 @samp{.text2} will be comprised of any input section matching the name *(.text)
4257 whose section header flag @code{SHF_WRITE} is clear.
4259 You can also specify files within archives by writing a pattern
4260 matching the archive, a colon, then the pattern matching the file,
4261 with no whitespace around the colon.
4265 matches file within archive
4267 matches the whole archive
4269 matches file but not one in an archive
4272 Either one or both of @samp{archive} and @samp{file} can contain shell
4273 wildcards. On DOS based file systems, the linker will assume that a
4274 single letter followed by a colon is a drive specifier, so
4275 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4276 within an archive called @samp{c}. @samp{archive:file} filespecs may
4277 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4278 other linker script contexts. For instance, you cannot extract a file
4279 from an archive by using @samp{archive:file} in an @code{INPUT}
4282 If you use a file name without a list of sections, then all sections in
4283 the input file will be included in the output section. This is not
4284 commonly done, but it may by useful on occasion. For example:
4289 When you use a file name which is not an @samp{archive:file} specifier
4290 and does not contain any wild card
4291 characters, the linker will first see if you also specified the file
4292 name on the linker command line or in an @code{INPUT} command. If you
4293 did not, the linker will attempt to open the file as an input file, as
4294 though it appeared on the command line. Note that this differs from an
4295 @code{INPUT} command, because the linker will not search for the file in
4296 the archive search path.
4298 @node Input Section Wildcards
4299 @subsubsection Input Section Wildcard Patterns
4300 @cindex input section wildcards
4301 @cindex wildcard file name patterns
4302 @cindex file name wildcard patterns
4303 @cindex section name wildcard patterns
4304 In an input section description, either the file name or the section
4305 name or both may be wildcard patterns.
4307 The file name of @samp{*} seen in many examples is a simple wildcard
4308 pattern for the file name.
4310 The wildcard patterns are like those used by the Unix shell.
4314 matches any number of characters
4316 matches any single character
4318 matches a single instance of any of the @var{chars}; the @samp{-}
4319 character may be used to specify a range of characters, as in
4320 @samp{[a-z]} to match any lower case letter
4322 quotes the following character
4325 When a file name is matched with a wildcard, the wildcard characters
4326 will not match a @samp{/} character (used to separate directory names on
4327 Unix). A pattern consisting of a single @samp{*} character is an
4328 exception; it will always match any file name, whether it contains a
4329 @samp{/} or not. In a section name, the wildcard characters will match
4330 a @samp{/} character.
4332 File name wildcard patterns only match files which are explicitly
4333 specified on the command line or in an @code{INPUT} command. The linker
4334 does not search directories to expand wildcards.
4336 If a file name matches more than one wildcard pattern, or if a file name
4337 appears explicitly and is also matched by a wildcard pattern, the linker
4338 will use the first match in the linker script. For example, this
4339 sequence of input section descriptions is probably in error, because the
4340 @file{data.o} rule will not be used:
4342 .data : @{ *(.data) @}
4343 .data1 : @{ data.o(.data) @}
4346 @cindex SORT_BY_NAME
4347 Normally, the linker will place files and sections matched by wildcards
4348 in the order in which they are seen during the link. You can change
4349 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4350 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4351 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4352 into ascending order by name before placing them in the output file.
4354 @cindex SORT_BY_ALIGNMENT
4355 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4356 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4357 descending order by alignment before placing them in the output file.
4358 Larger alignments are placed before smaller alignments in order to
4359 reduce the amount of padding necessary.
4361 @cindex SORT_BY_INIT_PRIORITY
4362 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4363 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4364 ascending order by numerical value of the GCC init_priority attribute
4365 encoded in the section name before placing them in the output file.
4368 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4370 When there are nested section sorting commands in linker script, there
4371 can be at most 1 level of nesting for section sorting commands.
4375 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4376 It will sort the input sections by name first, then by alignment if two
4377 sections have the same name.
4379 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4380 It will sort the input sections by alignment first, then by name if two
4381 sections have the same alignment.
4383 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4384 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4386 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4387 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4389 All other nested section sorting commands are invalid.
4392 When both command line section sorting option and linker script
4393 section sorting command are used, section sorting command always
4394 takes precedence over the command line option.
4396 If the section sorting command in linker script isn't nested, the
4397 command line option will make the section sorting command to be
4398 treated as nested sorting command.
4402 @code{SORT_BY_NAME} (wildcard section pattern ) with
4403 @option{--sort-sections alignment} is equivalent to
4404 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4406 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4407 @option{--sort-section name} is equivalent to
4408 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4411 If the section sorting command in linker script is nested, the
4412 command line option will be ignored.
4415 @code{SORT_NONE} disables section sorting by ignoring the command line
4416 section sorting option.
4418 If you ever get confused about where input sections are going, use the
4419 @samp{-M} linker option to generate a map file. The map file shows
4420 precisely how input sections are mapped to output sections.
4422 This example shows how wildcard patterns might be used to partition
4423 files. This linker script directs the linker to place all @samp{.text}
4424 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4425 The linker will place the @samp{.data} section from all files beginning
4426 with an upper case character in @samp{.DATA}; for all other files, the
4427 linker will place the @samp{.data} section in @samp{.data}.
4431 .text : @{ *(.text) @}
4432 .DATA : @{ [A-Z]*(.data) @}
4433 .data : @{ *(.data) @}
4434 .bss : @{ *(.bss) @}
4439 @node Input Section Common
4440 @subsubsection Input Section for Common Symbols
4441 @cindex common symbol placement
4442 @cindex uninitialized data placement
4443 A special notation is needed for common symbols, because in many object
4444 file formats common symbols do not have a particular input section. The
4445 linker treats common symbols as though they are in an input section
4446 named @samp{COMMON}.
4448 You may use file names with the @samp{COMMON} section just as with any
4449 other input sections. You can use this to place common symbols from a
4450 particular input file in one section while common symbols from other
4451 input files are placed in another section.
4453 In most cases, common symbols in input files will be placed in the
4454 @samp{.bss} section in the output file. For example:
4456 .bss @{ *(.bss) *(COMMON) @}
4459 @cindex scommon section
4460 @cindex small common symbols
4461 Some object file formats have more than one type of common symbol. For
4462 example, the MIPS ELF object file format distinguishes standard common
4463 symbols and small common symbols. In this case, the linker will use a
4464 different special section name for other types of common symbols. In
4465 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4466 symbols and @samp{.scommon} for small common symbols. This permits you
4467 to map the different types of common symbols into memory at different
4471 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4472 notation is now considered obsolete. It is equivalent to
4475 @node Input Section Keep
4476 @subsubsection Input Section and Garbage Collection
4478 @cindex garbage collection
4479 When link-time garbage collection is in use (@samp{--gc-sections}),
4480 it is often useful to mark sections that should not be eliminated.
4481 This is accomplished by surrounding an input section's wildcard entry
4482 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4483 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4485 @node Input Section Example
4486 @subsubsection Input Section Example
4487 The following example is a complete linker script. It tells the linker
4488 to read all of the sections from file @file{all.o} and place them at the
4489 start of output section @samp{outputa} which starts at location
4490 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4491 follows immediately, in the same output section. All of section
4492 @samp{.input2} from @file{foo.o} goes into output section
4493 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4494 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4495 files are written to output section @samp{outputc}.
4523 @node Output Section Data
4524 @subsection Output Section Data
4526 @cindex section data
4527 @cindex output section data
4528 @kindex BYTE(@var{expression})
4529 @kindex SHORT(@var{expression})
4530 @kindex LONG(@var{expression})
4531 @kindex QUAD(@var{expression})
4532 @kindex SQUAD(@var{expression})
4533 You can include explicit bytes of data in an output section by using
4534 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4535 an output section command. Each keyword is followed by an expression in
4536 parentheses providing the value to store (@pxref{Expressions}). The
4537 value of the expression is stored at the current value of the location
4540 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4541 store one, two, four, and eight bytes (respectively). After storing the
4542 bytes, the location counter is incremented by the number of bytes
4545 For example, this will store the byte 1 followed by the four byte value
4546 of the symbol @samp{addr}:
4552 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4553 same; they both store an 8 byte, or 64 bit, value. When both host and
4554 target are 32 bits, an expression is computed as 32 bits. In this case
4555 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4556 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4558 If the object file format of the output file has an explicit endianness,
4559 which is the normal case, the value will be stored in that endianness.
4560 When the object file format does not have an explicit endianness, as is
4561 true of, for example, S-records, the value will be stored in the
4562 endianness of the first input object file.
4564 Note---these commands only work inside a section description and not
4565 between them, so the following will produce an error from the linker:
4567 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4569 whereas this will work:
4571 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4574 @kindex FILL(@var{expression})
4575 @cindex holes, filling
4576 @cindex unspecified memory
4577 You may use the @code{FILL} command to set the fill pattern for the
4578 current section. It is followed by an expression in parentheses. Any
4579 otherwise unspecified regions of memory within the section (for example,
4580 gaps left due to the required alignment of input sections) are filled
4581 with the value of the expression, repeated as
4582 necessary. A @code{FILL} statement covers memory locations after the
4583 point at which it occurs in the section definition; by including more
4584 than one @code{FILL} statement, you can have different fill patterns in
4585 different parts of an output section.
4587 This example shows how to fill unspecified regions of memory with the
4593 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4594 section attribute, but it only affects the
4595 part of the section following the @code{FILL} command, rather than the
4596 entire section. If both are used, the @code{FILL} command takes
4597 precedence. @xref{Output Section Fill}, for details on the fill
4600 @node Output Section Keywords
4601 @subsection Output Section Keywords
4602 There are a couple of keywords which can appear as output section
4606 @kindex CREATE_OBJECT_SYMBOLS
4607 @cindex input filename symbols
4608 @cindex filename symbols
4609 @item CREATE_OBJECT_SYMBOLS
4610 The command tells the linker to create a symbol for each input file.
4611 The name of each symbol will be the name of the corresponding input
4612 file. The section of each symbol will be the output section in which
4613 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4615 This is conventional for the a.out object file format. It is not
4616 normally used for any other object file format.
4618 @kindex CONSTRUCTORS
4619 @cindex C++ constructors, arranging in link
4620 @cindex constructors, arranging in link
4622 When linking using the a.out object file format, the linker uses an
4623 unusual set construct to support C++ global constructors and
4624 destructors. When linking object file formats which do not support
4625 arbitrary sections, such as ECOFF and XCOFF, the linker will
4626 automatically recognize C++ global constructors and destructors by name.
4627 For these object file formats, the @code{CONSTRUCTORS} command tells the
4628 linker to place constructor information in the output section where the
4629 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4630 ignored for other object file formats.
4632 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4633 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4634 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4635 the start and end of the global destructors. The
4636 first word in the list is the number of entries, followed by the address
4637 of each constructor or destructor, followed by a zero word. The
4638 compiler must arrange to actually run the code. For these object file
4639 formats @sc{gnu} C++ normally calls constructors from a subroutine
4640 @code{__main}; a call to @code{__main} is automatically inserted into
4641 the startup code for @code{main}. @sc{gnu} C++ normally runs
4642 destructors either by using @code{atexit}, or directly from the function
4645 For object file formats such as @code{COFF} or @code{ELF} which support
4646 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4647 addresses of global constructors and destructors into the @code{.ctors}
4648 and @code{.dtors} sections. Placing the following sequence into your
4649 linker script will build the sort of table which the @sc{gnu} C++
4650 runtime code expects to see.
4654 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4659 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4665 If you are using the @sc{gnu} C++ support for initialization priority,
4666 which provides some control over the order in which global constructors
4667 are run, you must sort the constructors at link time to ensure that they
4668 are executed in the correct order. When using the @code{CONSTRUCTORS}
4669 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4670 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4671 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4674 Normally the compiler and linker will handle these issues automatically,
4675 and you will not need to concern yourself with them. However, you may
4676 need to consider this if you are using C++ and writing your own linker
4681 @node Output Section Discarding
4682 @subsection Output Section Discarding
4683 @cindex discarding sections
4684 @cindex sections, discarding
4685 @cindex removing sections
4686 The linker will not normally create output sections with no contents.
4687 This is for convenience when referring to input sections that may or
4688 may not be present in any of the input files. For example:
4690 .foo : @{ *(.foo) @}
4693 will only create a @samp{.foo} section in the output file if there is a
4694 @samp{.foo} section in at least one input file, and if the input
4695 sections are not all empty. Other link script directives that allocate
4696 space in an output section will also create the output section. So
4697 too will assignments to dot even if the assignment does not create
4698 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
4699 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
4700 @samp{sym} is an absolute symbol of value 0 defined in the script.
4701 This allows you to force output of an empty section with @samp{. = .}.
4703 The linker will ignore address assignments (@pxref{Output Section Address})
4704 on discarded output sections, except when the linker script defines
4705 symbols in the output section. In that case the linker will obey
4706 the address assignments, possibly advancing dot even though the
4707 section is discarded.
4710 The special output section name @samp{/DISCARD/} may be used to discard
4711 input sections. Any input sections which are assigned to an output
4712 section named @samp{/DISCARD/} are not included in the output file.
4714 @node Output Section Attributes
4715 @subsection Output Section Attributes
4716 @cindex output section attributes
4717 We showed above that the full description of an output section looked
4722 @var{section} [@var{address}] [(@var{type})] :
4724 [ALIGN(@var{section_align})]
4725 [SUBALIGN(@var{subsection_align})]
4728 @var{output-section-command}
4729 @var{output-section-command}
4731 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4735 We've already described @var{section}, @var{address}, and
4736 @var{output-section-command}. In this section we will describe the
4737 remaining section attributes.
4740 * Output Section Type:: Output section type
4741 * Output Section LMA:: Output section LMA
4742 * Forced Output Alignment:: Forced Output Alignment
4743 * Forced Input Alignment:: Forced Input Alignment
4744 * Output Section Constraint:: Output section constraint
4745 * Output Section Region:: Output section region
4746 * Output Section Phdr:: Output section phdr
4747 * Output Section Fill:: Output section fill
4750 @node Output Section Type
4751 @subsubsection Output Section Type
4752 Each output section may have a type. The type is a keyword in
4753 parentheses. The following types are defined:
4757 The section should be marked as not loadable, so that it will not be
4758 loaded into memory when the program is run.
4763 These type names are supported for backward compatibility, and are
4764 rarely used. They all have the same effect: the section should be
4765 marked as not allocatable, so that no memory is allocated for the
4766 section when the program is run.
4770 @cindex prevent unnecessary loading
4771 @cindex loading, preventing
4772 The linker normally sets the attributes of an output section based on
4773 the input sections which map into it. You can override this by using
4774 the section type. For example, in the script sample below, the
4775 @samp{ROM} section is addressed at memory location @samp{0} and does not
4776 need to be loaded when the program is run.
4780 ROM 0 (NOLOAD) : @{ @dots{} @}
4786 @node Output Section LMA
4787 @subsubsection Output Section LMA
4788 @kindex AT>@var{lma_region}
4789 @kindex AT(@var{lma})
4790 @cindex load address
4791 @cindex section load address
4792 Every section has a virtual address (VMA) and a load address (LMA); see
4793 @ref{Basic Script Concepts}. The virtual address is specified by the
4794 @pxref{Output Section Address} described earlier. The load address is
4795 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
4796 address is optional.
4798 The @code{AT} keyword takes an expression as an argument. This
4799 specifies the exact load address of the section. The @code{AT>} keyword
4800 takes the name of a memory region as an argument. @xref{MEMORY}. The
4801 load address of the section is set to the next free address in the
4802 region, aligned to the section's alignment requirements.
4804 If neither @code{AT} nor @code{AT>} is specified for an allocatable
4805 section, the linker will use the following heuristic to determine the
4810 If the section has a specific VMA address, then this is used as
4811 the LMA address as well.
4814 If the section is not allocatable then its LMA is set to its VMA.
4817 Otherwise if a memory region can be found that is compatible
4818 with the current section, and this region contains at least one
4819 section, then the LMA is set so the difference between the
4820 VMA and LMA is the same as the difference between the VMA and LMA of
4821 the last section in the located region.
4824 If no memory regions have been declared then a default region
4825 that covers the entire address space is used in the previous step.
4828 If no suitable region could be found, or there was no previous
4829 section then the LMA is set equal to the VMA.
4832 @cindex ROM initialized data
4833 @cindex initialized data in ROM
4834 This feature is designed to make it easy to build a ROM image. For
4835 example, the following linker script creates three output sections: one
4836 called @samp{.text}, which starts at @code{0x1000}, one called
4837 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
4838 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
4839 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
4840 defined with the value @code{0x2000}, which shows that the location
4841 counter holds the VMA value, not the LMA value.
4847 .text 0x1000 : @{ *(.text) _etext = . ; @}
4849 AT ( ADDR (.text) + SIZEOF (.text) )
4850 @{ _data = . ; *(.data); _edata = . ; @}
4852 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
4857 The run-time initialization code for use with a program generated with
4858 this linker script would include something like the following, to copy
4859 the initialized data from the ROM image to its runtime address. Notice
4860 how this code takes advantage of the symbols defined by the linker
4865 extern char _etext, _data, _edata, _bstart, _bend;
4866 char *src = &_etext;
4869 /* ROM has data at end of text; copy it. */
4870 while (dst < &_edata)
4874 for (dst = &_bstart; dst< &_bend; dst++)
4879 @node Forced Output Alignment
4880 @subsubsection Forced Output Alignment
4881 @kindex ALIGN(@var{section_align})
4882 @cindex forcing output section alignment
4883 @cindex output section alignment
4884 You can increase an output section's alignment by using ALIGN. As an
4885 alternative you can enforce that the difference between the VMA and LMA remains
4886 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
4888 @node Forced Input Alignment
4889 @subsubsection Forced Input Alignment
4890 @kindex SUBALIGN(@var{subsection_align})
4891 @cindex forcing input section alignment
4892 @cindex input section alignment
4893 You can force input section alignment within an output section by using
4894 SUBALIGN. The value specified overrides any alignment given by input
4895 sections, whether larger or smaller.
4897 @node Output Section Constraint
4898 @subsubsection Output Section Constraint
4901 @cindex constraints on output sections
4902 You can specify that an output section should only be created if all
4903 of its input sections are read-only or all of its input sections are
4904 read-write by using the keyword @code{ONLY_IF_RO} and
4905 @code{ONLY_IF_RW} respectively.
4907 @node Output Section Region
4908 @subsubsection Output Section Region
4909 @kindex >@var{region}
4910 @cindex section, assigning to memory region
4911 @cindex memory regions and sections
4912 You can assign a section to a previously defined region of memory by
4913 using @samp{>@var{region}}. @xref{MEMORY}.
4915 Here is a simple example:
4918 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4919 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4923 @node Output Section Phdr
4924 @subsubsection Output Section Phdr
4926 @cindex section, assigning to program header
4927 @cindex program headers and sections
4928 You can assign a section to a previously defined program segment by
4929 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
4930 one or more segments, then all subsequent allocated sections will be
4931 assigned to those segments as well, unless they use an explicitly
4932 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
4933 linker to not put the section in any segment at all.
4935 Here is a simple example:
4938 PHDRS @{ text PT_LOAD ; @}
4939 SECTIONS @{ .text : @{ *(.text) @} :text @}
4943 @node Output Section Fill
4944 @subsubsection Output Section Fill
4945 @kindex =@var{fillexp}
4946 @cindex section fill pattern
4947 @cindex fill pattern, entire section
4948 You can set the fill pattern for an entire section by using
4949 @samp{=@var{fillexp}}. @var{fillexp} is an expression
4950 (@pxref{Expressions}). Any otherwise unspecified regions of memory
4951 within the output section (for example, gaps left due to the required
4952 alignment of input sections) will be filled with the value, repeated as
4953 necessary. If the fill expression is a simple hex number, ie. a string
4954 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4955 an arbitrarily long sequence of hex digits can be used to specify the
4956 fill pattern; Leading zeros become part of the pattern too. For all
4957 other cases, including extra parentheses or a unary @code{+}, the fill
4958 pattern is the four least significant bytes of the value of the
4959 expression. In all cases, the number is big-endian.
4961 You can also change the fill value with a @code{FILL} command in the
4962 output section commands; (@pxref{Output Section Data}).
4964 Here is a simple example:
4967 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4971 @node Overlay Description
4972 @subsection Overlay Description
4975 An overlay description provides an easy way to describe sections which
4976 are to be loaded as part of a single memory image but are to be run at
4977 the same memory address. At run time, some sort of overlay manager will
4978 copy the overlaid sections in and out of the runtime memory address as
4979 required, perhaps by simply manipulating addressing bits. This approach
4980 can be useful, for example, when a certain region of memory is faster
4983 Overlays are described using the @code{OVERLAY} command. The
4984 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4985 output section description. The full syntax of the @code{OVERLAY}
4986 command is as follows:
4989 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4993 @var{output-section-command}
4994 @var{output-section-command}
4996 @} [:@var{phdr}@dots{}] [=@var{fill}]
4999 @var{output-section-command}
5000 @var{output-section-command}
5002 @} [:@var{phdr}@dots{}] [=@var{fill}]
5004 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5008 Everything is optional except @code{OVERLAY} (a keyword), and each
5009 section must have a name (@var{secname1} and @var{secname2} above). The
5010 section definitions within the @code{OVERLAY} construct are identical to
5011 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5012 except that no addresses and no memory regions may be defined for
5013 sections within an @code{OVERLAY}.
5015 The comma at the end may be required if a @var{fill} is used and
5016 the next @var{sections-command} looks like a continuation of the expression.
5018 The sections are all defined with the same starting address. The load
5019 addresses of the sections are arranged such that they are consecutive in
5020 memory starting at the load address used for the @code{OVERLAY} as a
5021 whole (as with normal section definitions, the load address is optional,
5022 and defaults to the start address; the start address is also optional,
5023 and defaults to the current value of the location counter).
5025 If the @code{NOCROSSREFS} keyword is used, and there are any
5026 references among the sections, the linker will report an error. Since
5027 the sections all run at the same address, it normally does not make
5028 sense for one section to refer directly to another.
5029 @xref{Miscellaneous Commands, NOCROSSREFS}.
5031 For each section within the @code{OVERLAY}, the linker automatically
5032 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5033 defined as the starting load address of the section. The symbol
5034 @code{__load_stop_@var{secname}} is defined as the final load address of
5035 the section. Any characters within @var{secname} which are not legal
5036 within C identifiers are removed. C (or assembler) code may use these
5037 symbols to move the overlaid sections around as necessary.
5039 At the end of the overlay, the value of the location counter is set to
5040 the start address of the overlay plus the size of the largest section.
5042 Here is an example. Remember that this would appear inside a
5043 @code{SECTIONS} construct.
5046 OVERLAY 0x1000 : AT (0x4000)
5048 .text0 @{ o1/*.o(.text) @}
5049 .text1 @{ o2/*.o(.text) @}
5054 This will define both @samp{.text0} and @samp{.text1} to start at
5055 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5056 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5057 following symbols will be defined if referenced: @code{__load_start_text0},
5058 @code{__load_stop_text0}, @code{__load_start_text1},
5059 @code{__load_stop_text1}.
5061 C code to copy overlay @code{.text1} into the overlay area might look
5066 extern char __load_start_text1, __load_stop_text1;
5067 memcpy ((char *) 0x1000, &__load_start_text1,
5068 &__load_stop_text1 - &__load_start_text1);
5072 Note that the @code{OVERLAY} command is just syntactic sugar, since
5073 everything it does can be done using the more basic commands. The above
5074 example could have been written identically as follows.
5078 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5079 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5080 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5081 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5082 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5083 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5084 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5089 @section MEMORY Command
5091 @cindex memory regions
5092 @cindex regions of memory
5093 @cindex allocating memory
5094 @cindex discontinuous memory
5095 The linker's default configuration permits allocation of all available
5096 memory. You can override this by using the @code{MEMORY} command.
5098 The @code{MEMORY} command describes the location and size of blocks of
5099 memory in the target. You can use it to describe which memory regions
5100 may be used by the linker, and which memory regions it must avoid. You
5101 can then assign sections to particular memory regions. The linker will
5102 set section addresses based on the memory regions, and will warn about
5103 regions that become too full. The linker will not shuffle sections
5104 around to fit into the available regions.
5106 A linker script may contain many uses of the @code{MEMORY} command,
5107 however, all memory blocks defined are treated as if they were
5108 specified inside a single @code{MEMORY} command. The syntax for
5114 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5120 The @var{name} is a name used in the linker script to refer to the
5121 region. The region name has no meaning outside of the linker script.
5122 Region names are stored in a separate name space, and will not conflict
5123 with symbol names, file names, or section names. Each memory region
5124 must have a distinct name within the @code{MEMORY} command. However you can
5125 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5128 @cindex memory region attributes
5129 The @var{attr} string is an optional list of attributes that specify
5130 whether to use a particular memory region for an input section which is
5131 not explicitly mapped in the linker script. As described in
5132 @ref{SECTIONS}, if you do not specify an output section for some input
5133 section, the linker will create an output section with the same name as
5134 the input section. If you define region attributes, the linker will use
5135 them to select the memory region for the output section that it creates.
5137 The @var{attr} string must consist only of the following characters:
5152 Invert the sense of any of the attributes that follow
5155 If a unmapped section matches any of the listed attributes other than
5156 @samp{!}, it will be placed in the memory region. The @samp{!}
5157 attribute reverses this test, so that an unmapped section will be placed
5158 in the memory region only if it does not match any of the listed
5164 The @var{origin} is an numerical expression for the start address of
5165 the memory region. The expression must evaluate to a constant and it
5166 cannot involve any symbols. The keyword @code{ORIGIN} may be
5167 abbreviated to @code{org} or @code{o} (but not, for example,
5173 The @var{len} is an expression for the size in bytes of the memory
5174 region. As with the @var{origin} expression, the expression must
5175 be numerical only and must evaluate to a constant. The keyword
5176 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5178 In the following example, we specify that there are two memory regions
5179 available for allocation: one starting at @samp{0} for 256 kilobytes,
5180 and the other starting at @samp{0x40000000} for four megabytes. The
5181 linker will place into the @samp{rom} memory region every section which
5182 is not explicitly mapped into a memory region, and is either read-only
5183 or executable. The linker will place other sections which are not
5184 explicitly mapped into a memory region into the @samp{ram} memory
5191 rom (rx) : ORIGIN = 0, LENGTH = 256K
5192 ram (!rx) : org = 0x40000000, l = 4M
5197 Once you define a memory region, you can direct the linker to place
5198 specific output sections into that memory region by using the
5199 @samp{>@var{region}} output section attribute. For example, if you have
5200 a memory region named @samp{mem}, you would use @samp{>mem} in the
5201 output section definition. @xref{Output Section Region}. If no address
5202 was specified for the output section, the linker will set the address to
5203 the next available address within the memory region. If the combined
5204 output sections directed to a memory region are too large for the
5205 region, the linker will issue an error message.
5207 It is possible to access the origin and length of a memory in an
5208 expression via the @code{ORIGIN(@var{memory})} and
5209 @code{LENGTH(@var{memory})} functions:
5213 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5218 @section PHDRS Command
5220 @cindex program headers
5221 @cindex ELF program headers
5222 @cindex program segments
5223 @cindex segments, ELF
5224 The ELF object file format uses @dfn{program headers}, also knows as
5225 @dfn{segments}. The program headers describe how the program should be
5226 loaded into memory. You can print them out by using the @code{objdump}
5227 program with the @samp{-p} option.
5229 When you run an ELF program on a native ELF system, the system loader
5230 reads the program headers in order to figure out how to load the
5231 program. This will only work if the program headers are set correctly.
5232 This manual does not describe the details of how the system loader
5233 interprets program headers; for more information, see the ELF ABI.
5235 The linker will create reasonable program headers by default. However,
5236 in some cases, you may need to specify the program headers more
5237 precisely. You may use the @code{PHDRS} command for this purpose. When
5238 the linker sees the @code{PHDRS} command in the linker script, it will
5239 not create any program headers other than the ones specified.
5241 The linker only pays attention to the @code{PHDRS} command when
5242 generating an ELF output file. In other cases, the linker will simply
5243 ignore @code{PHDRS}.
5245 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5246 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5252 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5253 [ FLAGS ( @var{flags} ) ] ;
5258 The @var{name} is used only for reference in the @code{SECTIONS} command
5259 of the linker script. It is not put into the output file. Program
5260 header names are stored in a separate name space, and will not conflict
5261 with symbol names, file names, or section names. Each program header
5262 must have a distinct name. The headers are processed in order and it
5263 is usual for them to map to sections in ascending load address order.
5265 Certain program header types describe segments of memory which the
5266 system loader will load from the file. In the linker script, you
5267 specify the contents of these segments by placing allocatable output
5268 sections in the segments. You use the @samp{:@var{phdr}} output section
5269 attribute to place a section in a particular segment. @xref{Output
5272 It is normal to put certain sections in more than one segment. This
5273 merely implies that one segment of memory contains another. You may
5274 repeat @samp{:@var{phdr}}, using it once for each segment which should
5275 contain the section.
5277 If you place a section in one or more segments using @samp{:@var{phdr}},
5278 then the linker will place all subsequent allocatable sections which do
5279 not specify @samp{:@var{phdr}} in the same segments. This is for
5280 convenience, since generally a whole set of contiguous sections will be
5281 placed in a single segment. You can use @code{:NONE} to override the
5282 default segment and tell the linker to not put the section in any
5287 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
5288 the program header type to further describe the contents of the segment.
5289 The @code{FILEHDR} keyword means that the segment should include the ELF
5290 file header. The @code{PHDRS} keyword means that the segment should
5291 include the ELF program headers themselves. If applied to a loadable
5292 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5295 The @var{type} may be one of the following. The numbers indicate the
5296 value of the keyword.
5299 @item @code{PT_NULL} (0)
5300 Indicates an unused program header.
5302 @item @code{PT_LOAD} (1)
5303 Indicates that this program header describes a segment to be loaded from
5306 @item @code{PT_DYNAMIC} (2)
5307 Indicates a segment where dynamic linking information can be found.
5309 @item @code{PT_INTERP} (3)
5310 Indicates a segment where the name of the program interpreter may be
5313 @item @code{PT_NOTE} (4)
5314 Indicates a segment holding note information.
5316 @item @code{PT_SHLIB} (5)
5317 A reserved program header type, defined but not specified by the ELF
5320 @item @code{PT_PHDR} (6)
5321 Indicates a segment where the program headers may be found.
5323 @item @var{expression}
5324 An expression giving the numeric type of the program header. This may
5325 be used for types not defined above.
5328 You can specify that a segment should be loaded at a particular address
5329 in memory by using an @code{AT} expression. This is identical to the
5330 @code{AT} command used as an output section attribute (@pxref{Output
5331 Section LMA}). The @code{AT} command for a program header overrides the
5332 output section attribute.
5334 The linker will normally set the segment flags based on the sections
5335 which comprise the segment. You may use the @code{FLAGS} keyword to
5336 explicitly specify the segment flags. The value of @var{flags} must be
5337 an integer. It is used to set the @code{p_flags} field of the program
5340 Here is an example of @code{PHDRS}. This shows a typical set of program
5341 headers used on a native ELF system.
5347 headers PT_PHDR PHDRS ;
5349 text PT_LOAD FILEHDR PHDRS ;
5351 dynamic PT_DYNAMIC ;
5357 .interp : @{ *(.interp) @} :text :interp
5358 .text : @{ *(.text) @} :text
5359 .rodata : @{ *(.rodata) @} /* defaults to :text */
5361 . = . + 0x1000; /* move to a new page in memory */
5362 .data : @{ *(.data) @} :data
5363 .dynamic : @{ *(.dynamic) @} :data :dynamic
5370 @section VERSION Command
5371 @kindex VERSION @{script text@}
5372 @cindex symbol versions
5373 @cindex version script
5374 @cindex versions of symbols
5375 The linker supports symbol versions when using ELF. Symbol versions are
5376 only useful when using shared libraries. The dynamic linker can use
5377 symbol versions to select a specific version of a function when it runs
5378 a program that may have been linked against an earlier version of the
5381 You can include a version script directly in the main linker script, or
5382 you can supply the version script as an implicit linker script. You can
5383 also use the @samp{--version-script} linker option.
5385 The syntax of the @code{VERSION} command is simply
5387 VERSION @{ version-script-commands @}
5390 The format of the version script commands is identical to that used by
5391 Sun's linker in Solaris 2.5. The version script defines a tree of
5392 version nodes. You specify the node names and interdependencies in the
5393 version script. You can specify which symbols are bound to which
5394 version nodes, and you can reduce a specified set of symbols to local
5395 scope so that they are not globally visible outside of the shared
5398 The easiest way to demonstrate the version script language is with a few
5424 This example version script defines three version nodes. The first
5425 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5426 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5427 a number of symbols to local scope so that they are not visible outside
5428 of the shared library; this is done using wildcard patterns, so that any
5429 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5430 is matched. The wildcard patterns available are the same as those used
5431 in the shell when matching filenames (also known as ``globbing'').
5432 However, if you specify the symbol name inside double quotes, then the
5433 name is treated as literal, rather than as a glob pattern.
5435 Next, the version script defines node @samp{VERS_1.2}. This node
5436 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5437 to the version node @samp{VERS_1.2}.
5439 Finally, the version script defines node @samp{VERS_2.0}. This node
5440 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5441 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5443 When the linker finds a symbol defined in a library which is not
5444 specifically bound to a version node, it will effectively bind it to an
5445 unspecified base version of the library. You can bind all otherwise
5446 unspecified symbols to a given version node by using @samp{global: *;}
5447 somewhere in the version script. Note that it's slightly crazy to use
5448 wildcards in a global spec except on the last version node. Global
5449 wildcards elsewhere run the risk of accidentally adding symbols to the
5450 set exported for an old version. That's wrong since older versions
5451 ought to have a fixed set of symbols.
5453 The names of the version nodes have no specific meaning other than what
5454 they might suggest to the person reading them. The @samp{2.0} version
5455 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5456 However, this would be a confusing way to write a version script.
5458 Node name can be omitted, provided it is the only version node
5459 in the version script. Such version script doesn't assign any versions to
5460 symbols, only selects which symbols will be globally visible out and which
5464 @{ global: foo; bar; local: *; @};
5467 When you link an application against a shared library that has versioned
5468 symbols, the application itself knows which version of each symbol it
5469 requires, and it also knows which version nodes it needs from each
5470 shared library it is linked against. Thus at runtime, the dynamic
5471 loader can make a quick check to make sure that the libraries you have
5472 linked against do in fact supply all of the version nodes that the
5473 application will need to resolve all of the dynamic symbols. In this
5474 way it is possible for the dynamic linker to know with certainty that
5475 all external symbols that it needs will be resolvable without having to
5476 search for each symbol reference.
5478 The symbol versioning is in effect a much more sophisticated way of
5479 doing minor version checking that SunOS does. The fundamental problem
5480 that is being addressed here is that typically references to external
5481 functions are bound on an as-needed basis, and are not all bound when
5482 the application starts up. If a shared library is out of date, a
5483 required interface may be missing; when the application tries to use
5484 that interface, it may suddenly and unexpectedly fail. With symbol
5485 versioning, the user will get a warning when they start their program if
5486 the libraries being used with the application are too old.
5488 There are several GNU extensions to Sun's versioning approach. The
5489 first of these is the ability to bind a symbol to a version node in the
5490 source file where the symbol is defined instead of in the versioning
5491 script. This was done mainly to reduce the burden on the library
5492 maintainer. You can do this by putting something like:
5494 __asm__(".symver original_foo,foo@@VERS_1.1");
5497 in the C source file. This renames the function @samp{original_foo} to
5498 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5499 The @samp{local:} directive can be used to prevent the symbol
5500 @samp{original_foo} from being exported. A @samp{.symver} directive
5501 takes precedence over a version script.
5503 The second GNU extension is to allow multiple versions of the same
5504 function to appear in a given shared library. In this way you can make
5505 an incompatible change to an interface without increasing the major
5506 version number of the shared library, while still allowing applications
5507 linked against the old interface to continue to function.
5509 To do this, you must use multiple @samp{.symver} directives in the
5510 source file. Here is an example:
5513 __asm__(".symver original_foo,foo@@");
5514 __asm__(".symver old_foo,foo@@VERS_1.1");
5515 __asm__(".symver old_foo1,foo@@VERS_1.2");
5516 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5519 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5520 unspecified base version of the symbol. The source file that contains this
5521 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5522 @samp{old_foo1}, and @samp{new_foo}.
5524 When you have multiple definitions of a given symbol, there needs to be
5525 some way to specify a default version to which external references to
5526 this symbol will be bound. You can do this with the
5527 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5528 declare one version of a symbol as the default in this manner; otherwise
5529 you would effectively have multiple definitions of the same symbol.
5531 If you wish to bind a reference to a specific version of the symbol
5532 within the shared library, you can use the aliases of convenience
5533 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5534 specifically bind to an external version of the function in question.
5536 You can also specify the language in the version script:
5539 VERSION extern "lang" @{ version-script-commands @}
5542 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5543 The linker will iterate over the list of symbols at the link time and
5544 demangle them according to @samp{lang} before matching them to the
5545 patterns specified in @samp{version-script-commands}. The default
5546 @samp{lang} is @samp{C}.
5548 Demangled names may contains spaces and other special characters. As
5549 described above, you can use a glob pattern to match demangled names,
5550 or you can use a double-quoted string to match the string exactly. In
5551 the latter case, be aware that minor differences (such as differing
5552 whitespace) between the version script and the demangler output will
5553 cause a mismatch. As the exact string generated by the demangler
5554 might change in the future, even if the mangled name does not, you
5555 should check that all of your version directives are behaving as you
5556 expect when you upgrade.
5559 @section Expressions in Linker Scripts
5562 The syntax for expressions in the linker script language is identical to
5563 that of C expressions. All expressions are evaluated as integers. All
5564 expressions are evaluated in the same size, which is 32 bits if both the
5565 host and target are 32 bits, and is otherwise 64 bits.
5567 You can use and set symbol values in expressions.
5569 The linker defines several special purpose builtin functions for use in
5573 * Constants:: Constants
5574 * Symbolic Constants:: Symbolic constants
5575 * Symbols:: Symbol Names
5576 * Orphan Sections:: Orphan Sections
5577 * Location Counter:: The Location Counter
5578 * Operators:: Operators
5579 * Evaluation:: Evaluation
5580 * Expression Section:: The Section of an Expression
5581 * Builtin Functions:: Builtin Functions
5585 @subsection Constants
5586 @cindex integer notation
5587 @cindex constants in linker scripts
5588 All constants are integers.
5590 As in C, the linker considers an integer beginning with @samp{0} to be
5591 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5592 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5593 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5594 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5595 value without a prefix or a suffix is considered to be decimal.
5597 @cindex scaled integers
5598 @cindex K and M integer suffixes
5599 @cindex M and K integer suffixes
5600 @cindex suffixes for integers
5601 @cindex integer suffixes
5602 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5606 @c END TEXI2ROFF-KILL
5607 @code{1024} or @code{1024*1024}
5611 ${\rm 1024}$ or ${\rm 1024}^2$
5613 @c END TEXI2ROFF-KILL
5614 respectively. For example, the following
5615 all refer to the same quantity:
5624 Note - the @code{K} and @code{M} suffixes cannot be used in
5625 conjunction with the base suffixes mentioned above.
5627 @node Symbolic Constants
5628 @subsection Symbolic Constants
5629 @cindex symbolic constants
5631 It is possible to refer to target specific constants via the use of
5632 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5637 The target's maximum page size.
5639 @item COMMONPAGESIZE
5640 @kindex COMMONPAGESIZE
5641 The target's default page size.
5647 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5650 will create a text section aligned to the largest page boundary
5651 supported by the target.
5654 @subsection Symbol Names
5655 @cindex symbol names
5657 @cindex quoted symbol names
5659 Unless quoted, symbol names start with a letter, underscore, or period
5660 and may include letters, digits, underscores, periods, and hyphens.
5661 Unquoted symbol names must not conflict with any keywords. You can
5662 specify a symbol which contains odd characters or has the same name as a
5663 keyword by surrounding the symbol name in double quotes:
5666 "with a space" = "also with a space" + 10;
5669 Since symbols can contain many non-alphabetic characters, it is safest
5670 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5671 whereas @samp{A - B} is an expression involving subtraction.
5673 @node Orphan Sections
5674 @subsection Orphan Sections
5676 Orphan sections are sections present in the input files which
5677 are not explicitly placed into the output file by the linker
5678 script. The linker will still copy these sections into the
5679 output file, but it has to guess as to where they should be
5680 placed. The linker uses a simple heuristic to do this. It
5681 attempts to place orphan sections after non-orphan sections of the
5682 same attribute, such as code vs data, loadable vs non-loadable, etc.
5683 If there is not enough room to do this then it places
5684 at the end of the file.
5686 For ELF targets, the attribute of the section includes section type as
5687 well as section flag.
5689 The command line options @samp{--orphan-handling} and @samp{--unique}
5690 (@pxref{Options,,Command Line Options}) can be used to control which
5691 output sections an orphan is placed in.
5693 If an orphaned section's name is representable as a C identifier then
5694 the linker will automatically @pxref{PROVIDE} two symbols:
5695 __start_SECNAME and __stop_SECNAME, where SECNAME is the name of the
5696 section. These indicate the start address and end address of the
5697 orphaned section respectively. Note: most section names are not
5698 representable as C identifiers because they contain a @samp{.}
5701 @node Location Counter
5702 @subsection The Location Counter
5705 @cindex location counter
5706 @cindex current output location
5707 The special linker variable @dfn{dot} @samp{.} always contains the
5708 current output location counter. Since the @code{.} always refers to a
5709 location in an output section, it may only appear in an expression
5710 within a @code{SECTIONS} command. The @code{.} symbol may appear
5711 anywhere that an ordinary symbol is allowed in an expression.
5714 Assigning a value to @code{.} will cause the location counter to be
5715 moved. This may be used to create holes in the output section. The
5716 location counter may not be moved backwards inside an output section,
5717 and may not be moved backwards outside of an output section if so
5718 doing creates areas with overlapping LMAs.
5734 In the previous example, the @samp{.text} section from @file{file1} is
5735 located at the beginning of the output section @samp{output}. It is
5736 followed by a 1000 byte gap. Then the @samp{.text} section from
5737 @file{file2} appears, also with a 1000 byte gap following before the
5738 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5739 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5741 @cindex dot inside sections
5742 Note: @code{.} actually refers to the byte offset from the start of the
5743 current containing object. Normally this is the @code{SECTIONS}
5744 statement, whose start address is 0, hence @code{.} can be used as an
5745 absolute address. If @code{.} is used inside a section description
5746 however, it refers to the byte offset from the start of that section,
5747 not an absolute address. Thus in a script like this:
5765 The @samp{.text} section will be assigned a starting address of 0x100
5766 and a size of exactly 0x200 bytes, even if there is not enough data in
5767 the @samp{.text} input sections to fill this area. (If there is too
5768 much data, an error will be produced because this would be an attempt to
5769 move @code{.} backwards). The @samp{.data} section will start at 0x500
5770 and it will have an extra 0x600 bytes worth of space after the end of
5771 the values from the @samp{.data} input sections and before the end of
5772 the @samp{.data} output section itself.
5774 @cindex dot outside sections
5775 Setting symbols to the value of the location counter outside of an
5776 output section statement can result in unexpected values if the linker
5777 needs to place orphan sections. For example, given the following:
5783 .text: @{ *(.text) @}
5787 .data: @{ *(.data) @}
5792 If the linker needs to place some input section, e.g. @code{.rodata},
5793 not mentioned in the script, it might choose to place that section
5794 between @code{.text} and @code{.data}. You might think the linker
5795 should place @code{.rodata} on the blank line in the above script, but
5796 blank lines are of no particular significance to the linker. As well,
5797 the linker doesn't associate the above symbol names with their
5798 sections. Instead, it assumes that all assignments or other
5799 statements belong to the previous output section, except for the
5800 special case of an assignment to @code{.}. I.e., the linker will
5801 place the orphan @code{.rodata} section as if the script was written
5808 .text: @{ *(.text) @}
5812 .rodata: @{ *(.rodata) @}
5813 .data: @{ *(.data) @}
5818 This may or may not be the script author's intention for the value of
5819 @code{start_of_data}. One way to influence the orphan section
5820 placement is to assign the location counter to itself, as the linker
5821 assumes that an assignment to @code{.} is setting the start address of
5822 a following output section and thus should be grouped with that
5823 section. So you could write:
5829 .text: @{ *(.text) @}
5834 .data: @{ *(.data) @}
5839 Now, the orphan @code{.rodata} section will be placed between
5840 @code{end_of_text} and @code{start_of_data}.
5844 @subsection Operators
5845 @cindex operators for arithmetic
5846 @cindex arithmetic operators
5847 @cindex precedence in expressions
5848 The linker recognizes the standard C set of arithmetic operators, with
5849 the standard bindings and precedence levels:
5852 @c END TEXI2ROFF-KILL
5854 precedence associativity Operators Notes
5860 5 left == != > < <= >=
5866 11 right &= += -= *= /= (2)
5870 (1) Prefix operators
5871 (2) @xref{Assignments}.
5875 \vskip \baselineskip
5876 %"lispnarrowing" is the extra indent used generally for smallexample
5877 \hskip\lispnarrowing\vbox{\offinterlineskip
5880 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
5881 height2pt&\omit&&\omit&&\omit&\cr
5882 &Precedence&& Associativity &&{\rm Operators}&\cr
5883 height2pt&\omit&&\omit&&\omit&\cr
5885 height2pt&\omit&&\omit&&\omit&\cr
5887 % '176 is tilde, '~' in tt font
5888 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
5889 &2&&left&&* / \%&\cr
5892 &5&&left&&== != > < <= >=&\cr
5895 &8&&left&&{\&\&}&\cr
5898 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
5900 height2pt&\omit&&\omit&&\omit&\cr}
5905 @obeylines@parskip=0pt@parindent=0pt
5906 @dag@quad Prefix operators.
5907 @ddag@quad @xref{Assignments}.
5910 @c END TEXI2ROFF-KILL
5913 @subsection Evaluation
5914 @cindex lazy evaluation
5915 @cindex expression evaluation order
5916 The linker evaluates expressions lazily. It only computes the value of
5917 an expression when absolutely necessary.
5919 The linker needs some information, such as the value of the start
5920 address of the first section, and the origins and lengths of memory
5921 regions, in order to do any linking at all. These values are computed
5922 as soon as possible when the linker reads in the linker script.
5924 However, other values (such as symbol values) are not known or needed
5925 until after storage allocation. Such values are evaluated later, when
5926 other information (such as the sizes of output sections) is available
5927 for use in the symbol assignment expression.
5929 The sizes of sections cannot be known until after allocation, so
5930 assignments dependent upon these are not performed until after
5933 Some expressions, such as those depending upon the location counter
5934 @samp{.}, must be evaluated during section allocation.
5936 If the result of an expression is required, but the value is not
5937 available, then an error results. For example, a script like the
5943 .text 9+this_isnt_constant :
5949 will cause the error message @samp{non constant expression for initial
5952 @node Expression Section
5953 @subsection The Section of an Expression
5954 @cindex expression sections
5955 @cindex absolute expressions
5956 @cindex relative expressions
5957 @cindex absolute and relocatable symbols
5958 @cindex relocatable and absolute symbols
5959 @cindex symbols, relocatable and absolute
5960 Addresses and symbols may be section relative, or absolute. A section
5961 relative symbol is relocatable. If you request relocatable output
5962 using the @samp{-r} option, a further link operation may change the
5963 value of a section relative symbol. On the other hand, an absolute
5964 symbol will retain the same value throughout any further link
5967 Some terms in linker expressions are addresses. This is true of
5968 section relative symbols and for builtin functions that return an
5969 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
5970 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
5971 functions that return a non-address value, such as @code{LENGTH}.
5972 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
5973 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
5974 differently depending on their location, for compatibility with older
5975 versions of @code{ld}. Expressions appearing outside an output
5976 section definition treat all numbers as absolute addresses.
5977 Expressions appearing inside an output section definition treat
5978 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
5979 given, then absolute symbols and numbers are simply treated as numbers
5982 In the following simple example,
5989 __executable_start = 0x100;
5993 __data_start = 0x10;
6001 both @code{.} and @code{__executable_start} are set to the absolute
6002 address 0x100 in the first two assignments, then both @code{.} and
6003 @code{__data_start} are set to 0x10 relative to the @code{.data}
6004 section in the second two assignments.
6006 For expressions involving numbers, relative addresses and absolute
6007 addresses, ld follows these rules to evaluate terms:
6011 Unary operations on an absolute address or number, and binary
6012 operations on two absolute addresses or two numbers, or between one
6013 absolute address and a number, apply the operator to the value(s).
6015 Unary operations on a relative address, and binary operations on two
6016 relative addresses in the same section or between one relative address
6017 and a number, apply the operator to the offset part of the address(es).
6019 Other binary operations, that is, between two relative addresses not
6020 in the same section, or between a relative address and an absolute
6021 address, first convert any non-absolute term to an absolute address
6022 before applying the operator.
6025 The result section of each sub-expression is as follows:
6029 An operation involving only numbers results in a number.
6031 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6033 The result of other binary arithmetic and logical operations on two
6034 relative addresses in the same section or two absolute addresses
6035 (after above conversions) is also a number.
6037 The result of other operations on relative addresses or one
6038 relative address and a number, is a relative address in the same
6039 section as the relative operand(s).
6041 The result of other operations on absolute addresses (after above
6042 conversions) is an absolute address.
6045 You can use the builtin function @code{ABSOLUTE} to force an expression
6046 to be absolute when it would otherwise be relative. For example, to
6047 create an absolute symbol set to the address of the end of the output
6048 section @samp{.data}:
6052 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6056 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6057 @samp{.data} section.
6059 Using @code{LOADADDR} also forces an expression absolute, since this
6060 particular builtin function returns an absolute address.
6062 @node Builtin Functions
6063 @subsection Builtin Functions
6064 @cindex functions in expressions
6065 The linker script language includes a number of builtin functions for
6066 use in linker script expressions.
6069 @item ABSOLUTE(@var{exp})
6070 @kindex ABSOLUTE(@var{exp})
6071 @cindex expression, absolute
6072 Return the absolute (non-relocatable, as opposed to non-negative) value
6073 of the expression @var{exp}. Primarily useful to assign an absolute
6074 value to a symbol within a section definition, where symbol values are
6075 normally section relative. @xref{Expression Section}.
6077 @item ADDR(@var{section})
6078 @kindex ADDR(@var{section})
6079 @cindex section address in expression
6080 Return the address (VMA) of the named @var{section}. Your
6081 script must previously have defined the location of that section. In
6082 the following example, @code{start_of_output_1}, @code{symbol_1} and
6083 @code{symbol_2} are assigned equivalent values, except that
6084 @code{symbol_1} will be relative to the @code{.output1} section while
6085 the other two will be absolute:
6091 start_of_output_1 = ABSOLUTE(.);
6096 symbol_1 = ADDR(.output1);
6097 symbol_2 = start_of_output_1;
6103 @item ALIGN(@var{align})
6104 @itemx ALIGN(@var{exp},@var{align})
6105 @kindex ALIGN(@var{align})
6106 @kindex ALIGN(@var{exp},@var{align})
6107 @cindex round up location counter
6108 @cindex align location counter
6109 @cindex round up expression
6110 @cindex align expression
6111 Return the location counter (@code{.}) or arbitrary expression aligned
6112 to the next @var{align} boundary. The single operand @code{ALIGN}
6113 doesn't change the value of the location counter---it just does
6114 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6115 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6116 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6118 Here is an example which aligns the output @code{.data} section to the
6119 next @code{0x2000} byte boundary after the preceding section and sets a
6120 variable within the section to the next @code{0x8000} boundary after the
6125 .data ALIGN(0x2000): @{
6127 variable = ALIGN(0x8000);
6133 The first use of @code{ALIGN} in this example specifies the location of
6134 a section because it is used as the optional @var{address} attribute of
6135 a section definition (@pxref{Output Section Address}). The second use
6136 of @code{ALIGN} is used to defines the value of a symbol.
6138 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6140 @item ALIGNOF(@var{section})
6141 @kindex ALIGNOF(@var{section})
6142 @cindex section alignment
6143 Return the alignment in bytes of the named @var{section}, if that section has
6144 been allocated. If the section has not been allocated when this is
6145 evaluated, the linker will report an error. In the following example,
6146 the alignment of the @code{.output} section is stored as the first
6147 value in that section.
6152 LONG (ALIGNOF (.output))
6159 @item BLOCK(@var{exp})
6160 @kindex BLOCK(@var{exp})
6161 This is a synonym for @code{ALIGN}, for compatibility with older linker
6162 scripts. It is most often seen when setting the address of an output
6165 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6166 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6167 This is equivalent to either
6169 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6173 (ALIGN(@var{maxpagesize})
6174 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6177 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6178 for the data segment (area between the result of this expression and
6179 @code{DATA_SEGMENT_END}) than the former or not.
6180 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6181 memory will be saved at the expense of up to @var{commonpagesize} wasted
6182 bytes in the on-disk file.
6184 This expression can only be used directly in @code{SECTIONS} commands, not in
6185 any output section descriptions and only once in the linker script.
6186 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6187 be the system page size the object wants to be optimized for (while still
6188 working on system page sizes up to @var{maxpagesize}).
6193 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6196 @item DATA_SEGMENT_END(@var{exp})
6197 @kindex DATA_SEGMENT_END(@var{exp})
6198 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6199 evaluation purposes.
6202 . = DATA_SEGMENT_END(.);
6205 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6206 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6207 This defines the end of the @code{PT_GNU_RELRO} segment when
6208 @samp{-z relro} option is used.
6209 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6210 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6211 @var{exp} + @var{offset} is aligned to the most commonly used page
6212 boundary for particular target. If present in the linker script,
6213 it must always come in between @code{DATA_SEGMENT_ALIGN} and
6214 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6215 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6219 . = DATA_SEGMENT_RELRO_END(24, .);
6222 @item DEFINED(@var{symbol})
6223 @kindex DEFINED(@var{symbol})
6224 @cindex symbol defaults
6225 Return 1 if @var{symbol} is in the linker global symbol table and is
6226 defined before the statement using DEFINED in the script, otherwise
6227 return 0. You can use this function to provide
6228 default values for symbols. For example, the following script fragment
6229 shows how to set a global symbol @samp{begin} to the first location in
6230 the @samp{.text} section---but if a symbol called @samp{begin} already
6231 existed, its value is preserved:
6237 begin = DEFINED(begin) ? begin : . ;
6245 @item LENGTH(@var{memory})
6246 @kindex LENGTH(@var{memory})
6247 Return the length of the memory region named @var{memory}.
6249 @item LOADADDR(@var{section})
6250 @kindex LOADADDR(@var{section})
6251 @cindex section load address in expression
6252 Return the absolute LMA of the named @var{section}. (@pxref{Output
6255 @item LOG2CEIL(@var{exp})
6256 @kindex LOG2CEIL(@var{exp})
6257 Return the binary logarithm of @var{exp} rounded towards infinity.
6258 @code{LOG2CEIL(0)} returns 0.
6261 @item MAX(@var{exp1}, @var{exp2})
6262 Returns the maximum of @var{exp1} and @var{exp2}.
6265 @item MIN(@var{exp1}, @var{exp2})
6266 Returns the minimum of @var{exp1} and @var{exp2}.
6268 @item NEXT(@var{exp})
6269 @kindex NEXT(@var{exp})
6270 @cindex unallocated address, next
6271 Return the next unallocated address that is a multiple of @var{exp}.
6272 This function is closely related to @code{ALIGN(@var{exp})}; unless you
6273 use the @code{MEMORY} command to define discontinuous memory for the
6274 output file, the two functions are equivalent.
6276 @item ORIGIN(@var{memory})
6277 @kindex ORIGIN(@var{memory})
6278 Return the origin of the memory region named @var{memory}.
6280 @item SEGMENT_START(@var{segment}, @var{default})
6281 @kindex SEGMENT_START(@var{segment}, @var{default})
6282 Return the base address of the named @var{segment}. If an explicit
6283 value has already been given for this segment (with a command-line
6284 @samp{-T} option) then that value will be returned otherwise the value
6285 will be @var{default}. At present, the @samp{-T} command-line option
6286 can only be used to set the base address for the ``text'', ``data'', and
6287 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
6290 @item SIZEOF(@var{section})
6291 @kindex SIZEOF(@var{section})
6292 @cindex section size
6293 Return the size in bytes of the named @var{section}, if that section has
6294 been allocated. If the section has not been allocated when this is
6295 evaluated, the linker will report an error. In the following example,
6296 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6305 symbol_1 = .end - .start ;
6306 symbol_2 = SIZEOF(.output);
6311 @item SIZEOF_HEADERS
6312 @itemx sizeof_headers
6313 @kindex SIZEOF_HEADERS
6315 Return the size in bytes of the output file's headers. This is
6316 information which appears at the start of the output file. You can use
6317 this number when setting the start address of the first section, if you
6318 choose, to facilitate paging.
6320 @cindex not enough room for program headers
6321 @cindex program headers, not enough room
6322 When producing an ELF output file, if the linker script uses the
6323 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6324 number of program headers before it has determined all the section
6325 addresses and sizes. If the linker later discovers that it needs
6326 additional program headers, it will report an error @samp{not enough
6327 room for program headers}. To avoid this error, you must avoid using
6328 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6329 script to avoid forcing the linker to use additional program headers, or
6330 you must define the program headers yourself using the @code{PHDRS}
6331 command (@pxref{PHDRS}).
6334 @node Implicit Linker Scripts
6335 @section Implicit Linker Scripts
6336 @cindex implicit linker scripts
6337 If you specify a linker input file which the linker can not recognize as
6338 an object file or an archive file, it will try to read the file as a
6339 linker script. If the file can not be parsed as a linker script, the
6340 linker will report an error.
6342 An implicit linker script will not replace the default linker script.
6344 Typically an implicit linker script would contain only symbol
6345 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6348 Any input files read because of an implicit linker script will be read
6349 at the position in the command line where the implicit linker script was
6350 read. This can affect archive searching.
6353 @node Machine Dependent
6354 @chapter Machine Dependent Features
6356 @cindex machine dependencies
6357 @command{ld} has additional features on some platforms; the following
6358 sections describe them. Machines where @command{ld} has no additional
6359 functionality are not listed.
6363 * H8/300:: @command{ld} and the H8/300
6366 * i960:: @command{ld} and the Intel 960 family
6369 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6372 * ARM:: @command{ld} and the ARM family
6375 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6378 * M68K:: @command{ld} and the Motorola 68K family
6381 * MIPS:: @command{ld} and the MIPS family
6384 * MMIX:: @command{ld} and MMIX
6387 * MSP430:: @command{ld} and MSP430
6390 * NDS32:: @command{ld} and NDS32
6393 * Nios II:: @command{ld} and the Altera Nios II
6396 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6399 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6402 * SPU ELF:: @command{ld} and SPU ELF Support
6405 * TI COFF:: @command{ld} and TI COFF
6408 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6411 * Xtensa:: @command{ld} and Xtensa Processors
6422 @section @command{ld} and the H8/300
6424 @cindex H8/300 support
6425 For the H8/300, @command{ld} can perform these global optimizations when
6426 you specify the @samp{--relax} command-line option.
6429 @cindex relaxing on H8/300
6430 @item relaxing address modes
6431 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6432 targets are within eight bits, and turns them into eight-bit
6433 program-counter relative @code{bsr} and @code{bra} instructions,
6436 @cindex synthesizing on H8/300
6437 @item synthesizing instructions
6438 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6439 @command{ld} finds all @code{mov.b} instructions which use the
6440 sixteen-bit absolute address form, but refer to the top
6441 page of memory, and changes them to use the eight-bit address form.
6442 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6443 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6444 top page of memory).
6446 @command{ld} finds all @code{mov} instructions which use the register
6447 indirect with 32-bit displacement addressing mode, but use a small
6448 displacement inside 16-bit displacement range, and changes them to use
6449 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6450 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6451 whenever the displacement @var{d} is in the 16 bit signed integer
6452 range. Only implemented in ELF-format ld).
6454 @item bit manipulation instructions
6455 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6456 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6457 which use 32 bit and 16 bit absolute address form, but refer to the top
6458 page of memory, and changes them to use the 8 bit address form.
6459 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6460 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6461 the top page of memory).
6463 @item system control instructions
6464 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6465 32 bit absolute address form, but refer to the top page of memory, and
6466 changes them to use 16 bit address form.
6467 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6468 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6469 the top page of memory).
6479 @c This stuff is pointless to say unless you're especially concerned
6480 @c with Renesas chips; don't enable it for generic case, please.
6482 @chapter @command{ld} and Other Renesas Chips
6484 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6485 H8/500, and SH chips. No special features, commands, or command-line
6486 options are required for these chips.
6496 @section @command{ld} and the Intel 960 Family
6498 @cindex i960 support
6500 You can use the @samp{-A@var{architecture}} command line option to
6501 specify one of the two-letter names identifying members of the 960
6502 family; the option specifies the desired output target, and warns of any
6503 incompatible instructions in the input files. It also modifies the
6504 linker's search strategy for archive libraries, to support the use of
6505 libraries specific to each particular architecture, by including in the
6506 search loop names suffixed with the string identifying the architecture.
6508 For example, if your @command{ld} command line included @w{@samp{-ACA}} as
6509 well as @w{@samp{-ltry}}, the linker would look (in its built-in search
6510 paths, and in any paths you specify with @samp{-L}) for a library with
6523 The first two possibilities would be considered in any event; the last
6524 two are due to the use of @w{@samp{-ACA}}.
6526 You can meaningfully use @samp{-A} more than once on a command line, since
6527 the 960 architecture family allows combination of target architectures; each
6528 use will add another pair of name variants to search for when @w{@samp{-l}}
6529 specifies a library.
6531 @cindex @option{--relax} on i960
6532 @cindex relaxing on i960
6533 @command{ld} supports the @samp{--relax} option for the i960 family. If
6534 you specify @samp{--relax}, @command{ld} finds all @code{balx} and
6535 @code{calx} instructions whose targets are within 24 bits, and turns
6536 them into 24-bit program-counter relative @code{bal} and @code{cal}
6537 instructions, respectively. @command{ld} also turns @code{cal}
6538 instructions into @code{bal} instructions when it determines that the
6539 target subroutine is a leaf routine (that is, the target subroutine does
6540 not itself call any subroutines).
6557 @node M68HC11/68HC12
6558 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6560 @cindex M68HC11 and 68HC12 support
6562 @subsection Linker Relaxation
6564 For the Motorola 68HC11, @command{ld} can perform these global
6565 optimizations when you specify the @samp{--relax} command-line option.
6568 @cindex relaxing on M68HC11
6569 @item relaxing address modes
6570 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6571 targets are within eight bits, and turns them into eight-bit
6572 program-counter relative @code{bsr} and @code{bra} instructions,
6575 @command{ld} also looks at all 16-bit extended addressing modes and
6576 transforms them in a direct addressing mode when the address is in
6577 page 0 (between 0 and 0x0ff).
6579 @item relaxing gcc instruction group
6580 When @command{gcc} is called with @option{-mrelax}, it can emit group
6581 of instructions that the linker can optimize to use a 68HC11 direct
6582 addressing mode. These instructions consists of @code{bclr} or
6583 @code{bset} instructions.
6587 @subsection Trampoline Generation
6589 @cindex trampoline generation on M68HC11
6590 @cindex trampoline generation on M68HC12
6591 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6592 call a far function using a normal @code{jsr} instruction. The linker
6593 will also change the relocation to some far function to use the
6594 trampoline address instead of the function address. This is typically the
6595 case when a pointer to a function is taken. The pointer will in fact
6596 point to the function trampoline.
6604 @section @command{ld} and the ARM family
6606 @cindex ARM interworking support
6607 @kindex --support-old-code
6608 For the ARM, @command{ld} will generate code stubs to allow functions calls
6609 between ARM and Thumb code. These stubs only work with code that has
6610 been compiled and assembled with the @samp{-mthumb-interwork} command
6611 line option. If it is necessary to link with old ARM object files or
6612 libraries, which have not been compiled with the -mthumb-interwork
6613 option then the @samp{--support-old-code} command line switch should be
6614 given to the linker. This will make it generate larger stub functions
6615 which will work with non-interworking aware ARM code. Note, however,
6616 the linker does not support generating stubs for function calls to
6617 non-interworking aware Thumb code.
6619 @cindex thumb entry point
6620 @cindex entry point, thumb
6621 @kindex --thumb-entry=@var{entry}
6622 The @samp{--thumb-entry} switch is a duplicate of the generic
6623 @samp{--entry} switch, in that it sets the program's starting address.
6624 But it also sets the bottom bit of the address, so that it can be
6625 branched to using a BX instruction, and the program will start
6626 executing in Thumb mode straight away.
6628 @cindex PE import table prefixing
6629 @kindex --use-nul-prefixed-import-tables
6630 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6631 the import tables idata4 and idata5 have to be generated with a zero
6632 element prefix for import libraries. This is the old style to generate
6633 import tables. By default this option is turned off.
6637 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6638 executables. This option is only valid when linking big-endian
6639 objects - ie ones which have been assembled with the @option{-EB}
6640 option. The resulting image will contain big-endian data and
6644 @kindex --target1-rel
6645 @kindex --target1-abs
6646 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6647 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6648 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6649 and @samp{--target1-abs} switches override the default.
6652 @kindex --target2=@var{type}
6653 The @samp{--target2=type} switch overrides the default definition of the
6654 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6655 meanings, and target defaults are as follows:
6658 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6660 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6662 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6667 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6668 specification) enables objects compiled for the ARMv4 architecture to be
6669 interworking-safe when linked with other objects compiled for ARMv4t, but
6670 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6672 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6673 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6674 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6676 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6677 relocations are ignored.
6679 @cindex FIX_V4BX_INTERWORKING
6680 @kindex --fix-v4bx-interworking
6681 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6682 relocations with a branch to the following veneer:
6690 This allows generation of libraries/applications that work on ARMv4 cores
6691 and are still interworking safe. Note that the above veneer clobbers the
6692 condition flags, so may cause incorrect program behavior in rare cases.
6696 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6697 BLX instructions (available on ARMv5t and above) in various
6698 situations. Currently it is used to perform calls via the PLT from Thumb
6699 code using BLX rather than using BX and a mode-switching stub before
6700 each PLT entry. This should lead to such calls executing slightly faster.
6702 This option is enabled implicitly for SymbianOS, so there is no need to
6703 specify it if you are using that target.
6705 @cindex VFP11_DENORM_FIX
6706 @kindex --vfp11-denorm-fix
6707 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6708 bug in certain VFP11 coprocessor hardware, which sometimes allows
6709 instructions with denorm operands (which must be handled by support code)
6710 to have those operands overwritten by subsequent instructions before
6711 the support code can read the intended values.
6713 The bug may be avoided in scalar mode if you allow at least one
6714 intervening instruction between a VFP11 instruction which uses a register
6715 and another instruction which writes to the same register, or at least two
6716 intervening instructions if vector mode is in use. The bug only affects
6717 full-compliance floating-point mode: you do not need this workaround if
6718 you are using "runfast" mode. Please contact ARM for further details.
6720 If you know you are using buggy VFP11 hardware, you can
6721 enable this workaround by specifying the linker option
6722 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6723 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6724 vector mode (the latter also works for scalar code). The default is
6725 @samp{--vfp-denorm-fix=none}.
6727 If the workaround is enabled, instructions are scanned for
6728 potentially-troublesome sequences, and a veneer is created for each
6729 such sequence which may trigger the erratum. The veneer consists of the
6730 first instruction of the sequence and a branch back to the subsequent
6731 instruction. The original instruction is then replaced with a branch to
6732 the veneer. The extra cycles required to call and return from the veneer
6733 are sufficient to avoid the erratum in both the scalar and vector cases.
6735 @cindex ARM1176 erratum workaround
6736 @kindex --fix-arm1176
6737 @kindex --no-fix-arm1176
6738 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6739 in certain ARM1176 processors. The workaround is enabled by default if you
6740 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6741 unconditionally by specifying @samp{--no-fix-arm1176}.
6743 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6744 Programmer Advice Notice'' available on the ARM documentation website at:
6745 http://infocenter.arm.com/.
6747 @cindex STM32L4xx erratum workaround
6748 @kindex --fix-stm32l4xx-629360
6750 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
6751 workaround for a bug in the bus matrix / memory controller for some of
6752 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
6753 off-chip memory via the affected bus for bus reads of 9 words or more,
6754 the bus can generate corrupt data and/or abort. These are only
6755 core-initiated accesses (not DMA), and might affect any access:
6756 integer loads such as LDM, POP and floating-point loads such as VLDM,
6757 VPOP. Stores are not affected.
6759 The bug can be avoided by splitting memory accesses into the
6760 necessary chunks to keep bus reads below 8 words.
6762 The workaround is not enabled by default, this is equivalent to use
6763 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
6764 STM32L4xx hardware, you can enable the workaround by specifying the
6765 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
6766 @samp{--fix-stm32l4xx-629360=default}.
6768 If the workaround is enabled, instructions are scanned for
6769 potentially-troublesome sequences, and a veneer is created for each
6770 such sequence which may trigger the erratum. The veneer consists in a
6771 replacement sequence emulating the behaviour of the original one and a
6772 branch back to the subsequent instruction. The original instruction is
6773 then replaced with a branch to the veneer.
6775 The workaround does not always preserve the memory access order for
6776 the LDMDB instruction, when the instruction loads the PC.
6778 The workaround is not able to handle problematic instructions when
6779 they are in the middle of an IT block, since a branch is not allowed
6780 there. In that case, the linker reports a warning and no replacement
6783 The workaround is not able to replace problematic instructions with a
6784 PC-relative branch instruction if the @samp{.text} section is too
6785 large. In that case, when the branch that replaces the original code
6786 cannot be encoded, the linker reports a warning and no replacement
6789 @cindex NO_ENUM_SIZE_WARNING
6790 @kindex --no-enum-size-warning
6791 The @option{--no-enum-size-warning} switch prevents the linker from
6792 warning when linking object files that specify incompatible EABI
6793 enumeration size attributes. For example, with this switch enabled,
6794 linking of an object file using 32-bit enumeration values with another
6795 using enumeration values fitted into the smallest possible space will
6798 @cindex NO_WCHAR_SIZE_WARNING
6799 @kindex --no-wchar-size-warning
6800 The @option{--no-wchar-size-warning} switch prevents the linker from
6801 warning when linking object files that specify incompatible EABI
6802 @code{wchar_t} size attributes. For example, with this switch enabled,
6803 linking of an object file using 32-bit @code{wchar_t} values with another
6804 using 16-bit @code{wchar_t} values will not be diagnosed.
6807 @kindex --pic-veneer
6808 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6809 ARM/Thumb interworking veneers, even if the rest of the binary
6810 is not PIC. This avoids problems on uClinux targets where
6811 @samp{--emit-relocs} is used to generate relocatable binaries.
6813 @cindex STUB_GROUP_SIZE
6814 @kindex --stub-group-size=@var{N}
6815 The linker will automatically generate and insert small sequences of
6816 code into a linked ARM ELF executable whenever an attempt is made to
6817 perform a function call to a symbol that is too far away. The
6818 placement of these sequences of instructions - called stubs - is
6819 controlled by the command line option @option{--stub-group-size=N}.
6820 The placement is important because a poor choice can create a need for
6821 duplicate stubs, increasing the code size. The linker will try to
6822 group stubs together in order to reduce interruptions to the flow of
6823 code, but it needs guidance as to how big these groups should be and
6824 where they should be placed.
6826 The value of @samp{N}, the parameter to the
6827 @option{--stub-group-size=} option controls where the stub groups are
6828 placed. If it is negative then all stubs are placed after the first
6829 branch that needs them. If it is positive then the stubs can be
6830 placed either before or after the branches that need them. If the
6831 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6832 exactly where to place groups of stubs, using its built in heuristics.
6833 A value of @samp{N} greater than 1 (or smaller than -1) tells the
6834 linker that a single group of stubs can service at most @samp{N} bytes
6835 from the input sections.
6837 The default, if @option{--stub-group-size=} is not specified, is
6840 Farcalls stubs insertion is fully supported for the ARM-EABI target
6841 only, because it relies on object files properties not present
6844 @cindex Cortex-A8 erratum workaround
6845 @kindex --fix-cortex-a8
6846 @kindex --no-fix-cortex-a8
6847 The @samp{--fix-cortex-a8} switch enables a link-time workaround for an erratum in certain Cortex-A8 processors. The workaround is enabled by default if you are targeting the ARM v7-A architecture profile. It can be enabled otherwise by specifying @samp{--fix-cortex-a8}, or disabled unconditionally by specifying @samp{--no-fix-cortex-a8}.
6849 The erratum only affects Thumb-2 code. Please contact ARM for further details.
6851 @cindex Cortex-A53 erratum 835769 workaround
6852 @kindex --fix-cortex-a53-835769
6853 @kindex --no-fix-cortex-a53-835769
6854 The @samp{--fix-cortex-a53-835769} switch enables a link-time workaround for erratum 835769 present on certain early revisions of Cortex-A53 processors. The workaround is disabled by default. It can be enabled by specifying @samp{--fix-cortex-a53-835769}, or disabled unconditionally by specifying @samp{--no-fix-cortex-a53-835769}.
6856 Please contact ARM for further details.
6858 @kindex --merge-exidx-entries
6859 @kindex --no-merge-exidx-entries
6860 @cindex Merging exidx entries
6861 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
6864 @cindex 32-bit PLT entries
6865 The @samp{--long-plt} option enables the use of 16 byte PLT entries
6866 which support up to 4Gb of code. The default is to use 12 byte PLT
6867 entries which only support 512Mb of code.
6869 @kindex --no-apply-dynamic-relocs
6870 @cindex AArch64 rela addend
6871 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
6872 link-time values for dynamic relocations.
6874 @cindex Placement of SG veneers
6875 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
6876 Its start address must be set, either with the command line option
6877 @samp{--section-start} or in a linker script, to indicate where to place these
6880 @kindex --cmse-implib
6881 @cindex Secure gateway import library
6882 The @samp{--cmse-implib} option requests that the import libraries
6883 specified by the @samp{--out-implib} and @samp{--in-implib} options are
6884 secure gateway import libraries, suitable for linking a non-secure
6885 executable against secure code as per ARMv8-M Security Extensions.
6887 @kindex --in-implib=@var{file}
6888 @cindex Input import library
6889 The @samp{--in-implib=file} specifies an input import library whose symbols
6890 must keep the same address in the executable being produced. A warning is
6891 given if no @samp{--out-implib} is given but new symbols have been introduced
6892 in the executable that should be listed in its import library. Otherwise, if
6893 @samp{--out-implib} is specified, the symbols are added to the output import
6894 library. A warning is also given if some symbols present in the input import
6895 library have disappeared from the executable. This option is only effective
6896 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
6910 @section @command{ld} and HPPA 32-bit ELF Support
6911 @cindex HPPA multiple sub-space stubs
6912 @kindex --multi-subspace
6913 When generating a shared library, @command{ld} will by default generate
6914 import stubs suitable for use with a single sub-space application.
6915 The @samp{--multi-subspace} switch causes @command{ld} to generate export
6916 stubs, and different (larger) import stubs suitable for use with
6917 multiple sub-spaces.
6919 @cindex HPPA stub grouping
6920 @kindex --stub-group-size=@var{N}
6921 Long branch stubs and import/export stubs are placed by @command{ld} in
6922 stub sections located between groups of input sections.
6923 @samp{--stub-group-size} specifies the maximum size of a group of input
6924 sections handled by one stub section. Since branch offsets are signed,
6925 a stub section may serve two groups of input sections, one group before
6926 the stub section, and one group after it. However, when using
6927 conditional branches that require stubs, it may be better (for branch
6928 prediction) that stub sections only serve one group of input sections.
6929 A negative value for @samp{N} chooses this scheme, ensuring that
6930 branches to stubs always use a negative offset. Two special values of
6931 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
6932 @command{ld} to automatically size input section groups for the branch types
6933 detected, with the same behaviour regarding stub placement as other
6934 positive or negative values of @samp{N} respectively.
6936 Note that @samp{--stub-group-size} does not split input sections. A
6937 single input section larger than the group size specified will of course
6938 create a larger group (of one section). If input sections are too
6939 large, it may not be possible for a branch to reach its stub.
6952 @section @command{ld} and the Motorola 68K family
6954 @cindex Motorola 68K GOT generation
6955 @kindex --got=@var{type}
6956 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
6957 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
6958 @samp{target}. When @samp{target} is selected the linker chooses
6959 the default GOT generation scheme for the current target.
6960 @samp{single} tells the linker to generate a single GOT with
6961 entries only at non-negative offsets.
6962 @samp{negative} instructs the linker to generate a single GOT with
6963 entries at both negative and positive offsets. Not all environments
6965 @samp{multigot} allows the linker to generate several GOTs in the
6966 output file. All GOT references from a single input object
6967 file access the same GOT, but references from different input object
6968 files might access different GOTs. Not all environments support such GOTs.
6981 @section @command{ld} and the MIPS family
6983 @cindex MIPS microMIPS instruction choice selection
6986 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
6987 microMIPS instructions used in code generated by the linker, such as that
6988 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
6989 used, then the linker only uses 32-bit instruction encodings. By default
6990 or if @samp{--no-insn32} is used, all instruction encodings are used,
6991 including 16-bit ones where possible.
7004 @section @code{ld} and MMIX
7005 For MMIX, there is a choice of generating @code{ELF} object files or
7006 @code{mmo} object files when linking. The simulator @code{mmix}
7007 understands the @code{mmo} format. The binutils @code{objcopy} utility
7008 can translate between the two formats.
7010 There is one special section, the @samp{.MMIX.reg_contents} section.
7011 Contents in this section is assumed to correspond to that of global
7012 registers, and symbols referring to it are translated to special symbols,
7013 equal to registers. In a final link, the start address of the
7014 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7015 global register multiplied by 8. Register @code{$255} is not included in
7016 this section; it is always set to the program entry, which is at the
7017 symbol @code{Main} for @code{mmo} files.
7019 Global symbols with the prefix @code{__.MMIX.start.}, for example
7020 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7021 The default linker script uses these to set the default start address
7024 Initial and trailing multiples of zero-valued 32-bit words in a section,
7025 are left out from an mmo file.
7038 @section @code{ld} and MSP430
7039 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7040 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7041 just pass @samp{-m help} option to the linker).
7043 @cindex MSP430 extra sections
7044 The linker will recognize some extra sections which are MSP430 specific:
7047 @item @samp{.vectors}
7048 Defines a portion of ROM where interrupt vectors located.
7050 @item @samp{.bootloader}
7051 Defines the bootloader portion of the ROM (if applicable). Any code
7052 in this section will be uploaded to the MPU.
7054 @item @samp{.infomem}
7055 Defines an information memory section (if applicable). Any code in
7056 this section will be uploaded to the MPU.
7058 @item @samp{.infomemnobits}
7059 This is the same as the @samp{.infomem} section except that any code
7060 in this section will not be uploaded to the MPU.
7062 @item @samp{.noinit}
7063 Denotes a portion of RAM located above @samp{.bss} section.
7065 The last two sections are used by gcc.
7079 @section @code{ld} and NDS32
7080 @kindex relaxing on NDS32
7081 For NDS32, there are some options to select relaxation behavior. The linker
7082 relaxes objects according to these options.
7085 @item @samp{--m[no-]fp-as-gp}
7086 Disable/enable fp-as-gp relaxation.
7088 @item @samp{--mexport-symbols=FILE}
7089 Exporting symbols and their address into FILE as linker script.
7091 @item @samp{--m[no-]ex9}
7092 Disable/enable link-time EX9 relaxation.
7094 @item @samp{--mexport-ex9=FILE}
7095 Export the EX9 table after linking.
7097 @item @samp{--mimport-ex9=FILE}
7098 Import the Ex9 table for EX9 relaxation.
7100 @item @samp{--mupdate-ex9}
7101 Update the existing EX9 table.
7103 @item @samp{--mex9-limit=NUM}
7104 Maximum number of entries in the ex9 table.
7106 @item @samp{--mex9-loop-aware}
7107 Avoid generating the EX9 instruction inside the loop.
7109 @item @samp{--m[no-]ifc}
7110 Disable/enable the link-time IFC optimization.
7112 @item @samp{--mifc-loop-aware}
7113 Avoid generating the IFC instruction inside the loop.
7127 @section @command{ld} and the Altera Nios II
7128 @cindex Nios II call relaxation
7129 @kindex --relax on Nios II
7131 Call and immediate jump instructions on Nios II processors are limited to
7132 transferring control to addresses in the same 256MB memory segment,
7133 which may result in @command{ld} giving
7134 @samp{relocation truncated to fit} errors with very large programs.
7135 The command-line option @option{--relax} enables the generation of
7136 trampolines that can access the entire 32-bit address space for calls
7137 outside the normal @code{call} and @code{jmpi} address range. These
7138 trampolines are inserted at section boundaries, so may not themselves
7139 be reachable if an input section and its associated call trampolines are
7142 The @option{--relax} option is enabled by default unless @option{-r}
7143 is also specified. You can disable trampoline generation by using the
7144 @option{--no-relax} linker option. You can also disable this optimization
7145 locally by using the @samp{set .noat} directive in assembly-language
7146 source files, as the linker-inserted trampolines use the @code{at}
7147 register as a temporary.
7149 Note that the linker @option{--relax} option is independent of assembler
7150 relaxation options, and that using the GNU assembler's @option{-relax-all}
7151 option interferes with the linker's more selective call instruction relaxation.
7164 @section @command{ld} and PowerPC 32-bit ELF Support
7165 @cindex PowerPC long branches
7166 @kindex --relax on PowerPC
7167 Branches on PowerPC processors are limited to a signed 26-bit
7168 displacement, which may result in @command{ld} giving
7169 @samp{relocation truncated to fit} errors with very large programs.
7170 @samp{--relax} enables the generation of trampolines that can access
7171 the entire 32-bit address space. These trampolines are inserted at
7172 section boundaries, so may not themselves be reachable if an input
7173 section exceeds 33M in size. You may combine @samp{-r} and
7174 @samp{--relax} to add trampolines in a partial link. In that case
7175 both branches to undefined symbols and inter-section branches are also
7176 considered potentially out of range, and trampolines inserted.
7178 @cindex PowerPC ELF32 options
7183 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7184 generates code capable of using a newer PLT and GOT layout that has
7185 the security advantage of no executable section ever needing to be
7186 writable and no writable section ever being executable. PowerPC
7187 @command{ld} will generate this layout, including stubs to access the
7188 PLT, if all input files (including startup and static libraries) were
7189 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7190 BSS PLT (and GOT layout) which can give slightly better performance.
7192 @kindex --secure-plt
7194 @command{ld} will use the new PLT and GOT layout if it is linking new
7195 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7196 when linking non-PIC code. This option requests the new PLT and GOT
7197 layout. A warning will be given if some object file requires the old
7203 The new secure PLT and GOT are placed differently relative to other
7204 sections compared to older BSS PLT and GOT placement. The location of
7205 @code{.plt} must change because the new secure PLT is an initialized
7206 section while the old PLT is uninitialized. The reason for the
7207 @code{.got} change is more subtle: The new placement allows
7208 @code{.got} to be read-only in applications linked with
7209 @samp{-z relro -z now}. However, this placement means that
7210 @code{.sdata} cannot always be used in shared libraries, because the
7211 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7212 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7213 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7214 really only useful for other compilers that may do so.
7216 @cindex PowerPC stub symbols
7217 @kindex --emit-stub-syms
7218 @item --emit-stub-syms
7219 This option causes @command{ld} to label linker stubs with a local
7220 symbol that encodes the stub type and destination.
7222 @cindex PowerPC TLS optimization
7223 @kindex --no-tls-optimize
7224 @item --no-tls-optimize
7225 PowerPC @command{ld} normally performs some optimization of code
7226 sequences used to access Thread-Local Storage. Use this option to
7227 disable the optimization.
7240 @node PowerPC64 ELF64
7241 @section @command{ld} and PowerPC64 64-bit ELF Support
7243 @cindex PowerPC64 ELF64 options
7245 @cindex PowerPC64 stub grouping
7246 @kindex --stub-group-size
7247 @item --stub-group-size
7248 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
7249 by @command{ld} in stub sections located between groups of input sections.
7250 @samp{--stub-group-size} specifies the maximum size of a group of input
7251 sections handled by one stub section. Since branch offsets are signed,
7252 a stub section may serve two groups of input sections, one group before
7253 the stub section, and one group after it. However, when using
7254 conditional branches that require stubs, it may be better (for branch
7255 prediction) that stub sections only serve one group of input sections.
7256 A negative value for @samp{N} chooses this scheme, ensuring that
7257 branches to stubs always use a negative offset. Two special values of
7258 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7259 @command{ld} to automatically size input section groups for the branch types
7260 detected, with the same behaviour regarding stub placement as other
7261 positive or negative values of @samp{N} respectively.
7263 Note that @samp{--stub-group-size} does not split input sections. A
7264 single input section larger than the group size specified will of course
7265 create a larger group (of one section). If input sections are too
7266 large, it may not be possible for a branch to reach its stub.
7268 @cindex PowerPC64 stub symbols
7269 @kindex --emit-stub-syms
7270 @item --emit-stub-syms
7271 This option causes @command{ld} to label linker stubs with a local
7272 symbol that encodes the stub type and destination.
7274 @cindex PowerPC64 dot symbols
7276 @kindex --no-dotsyms
7279 These two options control how @command{ld} interprets version patterns
7280 in a version script. Older PowerPC64 compilers emitted both a
7281 function descriptor symbol with the same name as the function, and a
7282 code entry symbol with the name prefixed by a dot (@samp{.}). To
7283 properly version a function @samp{foo}, the version script thus needs
7284 to control both @samp{foo} and @samp{.foo}. The option
7285 @samp{--dotsyms}, on by default, automatically adds the required
7286 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
7289 @cindex PowerPC64 register save/restore functions
7290 @kindex --save-restore-funcs
7291 @kindex --no-save-restore-funcs
7292 @item --save-restore-funcs
7293 @itemx --no-save-restore-funcs
7294 These two options control whether PowerPC64 @command{ld} automatically
7295 provides out-of-line register save and restore functions used by
7296 @samp{-Os} code. The default is to provide any such referenced
7297 function for a normal final link, and to not do so for a relocatable
7300 @cindex PowerPC64 TLS optimization
7301 @kindex --no-tls-optimize
7302 @item --no-tls-optimize
7303 PowerPC64 @command{ld} normally performs some optimization of code
7304 sequences used to access Thread-Local Storage. Use this option to
7305 disable the optimization.
7307 @cindex PowerPC64 __tls_get_addr optimization
7308 @kindex --tls-get-addr-optimize
7309 @kindex --no-tls-get-addr-optimize
7310 @item --tls-get-addr-optimize
7311 @itemx --no-tls-get-addr-optimize
7312 These options control whether PowerPC64 @command{ld} uses a special
7313 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
7314 an optimization that allows the second and subsequent calls to
7315 @code{__tls_get_addr} for a given symbol to be resolved by the special
7316 stub without calling in to glibc. By default the linker enables this
7317 option when glibc advertises the availability of __tls_get_addr_opt.
7318 Forcing this option on when using an older glibc won't do much besides
7319 slow down your applications, but may be useful if linking an
7320 application against an older glibc with the expectation that it will
7321 normally be used on systems having a newer glibc.
7323 @cindex PowerPC64 OPD optimization
7324 @kindex --no-opd-optimize
7325 @item --no-opd-optimize
7326 PowerPC64 @command{ld} normally removes @code{.opd} section entries
7327 corresponding to deleted link-once functions, or functions removed by
7328 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
7329 Use this option to disable @code{.opd} optimization.
7331 @cindex PowerPC64 OPD spacing
7332 @kindex --non-overlapping-opd
7333 @item --non-overlapping-opd
7334 Some PowerPC64 compilers have an option to generate compressed
7335 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
7336 the static chain pointer (unused in C) with the first word of the next
7337 entry. This option expands such entries to the full 24 bytes.
7339 @cindex PowerPC64 TOC optimization
7340 @kindex --no-toc-optimize
7341 @item --no-toc-optimize
7342 PowerPC64 @command{ld} normally removes unused @code{.toc} section
7343 entries. Such entries are detected by examining relocations that
7344 reference the TOC in code sections. A reloc in a deleted code section
7345 marks a TOC word as unneeded, while a reloc in a kept code section
7346 marks a TOC word as needed. Since the TOC may reference itself, TOC
7347 relocs are also examined. TOC words marked as both needed and
7348 unneeded will of course be kept. TOC words without any referencing
7349 reloc are assumed to be part of a multi-word entry, and are kept or
7350 discarded as per the nearest marked preceding word. This works
7351 reliably for compiler generated code, but may be incorrect if assembly
7352 code is used to insert TOC entries. Use this option to disable the
7355 @cindex PowerPC64 multi-TOC
7356 @kindex --no-multi-toc
7357 @item --no-multi-toc
7358 If given any toc option besides @code{-mcmodel=medium} or
7359 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
7361 entries are accessed with a 16-bit offset from r2. This limits the
7362 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
7363 grouping code sections such that each group uses less than 64K for its
7364 TOC entries, then inserts r2 adjusting stubs between inter-group
7365 calls. @command{ld} does not split apart input sections, so cannot
7366 help if a single input file has a @code{.toc} section that exceeds
7367 64K, most likely from linking multiple files with @command{ld -r}.
7368 Use this option to turn off this feature.
7370 @cindex PowerPC64 TOC sorting
7371 @kindex --no-toc-sort
7373 By default, @command{ld} sorts TOC sections so that those whose file
7374 happens to have a section called @code{.init} or @code{.fini} are
7375 placed first, followed by TOC sections referenced by code generated
7376 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
7377 referenced only by code generated with PowerPC64 gcc's
7378 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
7379 results in better TOC grouping for multi-TOC. Use this option to turn
7382 @cindex PowerPC64 PLT stub alignment
7384 @kindex --no-plt-align
7386 @itemx --no-plt-align
7387 Use these options to control whether individual PLT call stubs are
7388 padded so that they don't cross a 32-byte boundary, or to the
7389 specified power of two boundary when using @code{--plt-align=}. Note
7390 that this isn't alignment in the usual sense. By default PLT call
7391 stubs are packed tightly.
7393 @cindex PowerPC64 PLT call stub static chain
7394 @kindex --plt-static-chain
7395 @kindex --no-plt-static-chain
7396 @item --plt-static-chain
7397 @itemx --no-plt-static-chain
7398 Use these options to control whether PLT call stubs load the static
7399 chain pointer (r11). @code{ld} defaults to not loading the static
7400 chain since there is never any need to do so on a PLT call.
7402 @cindex PowerPC64 PLT call stub thread safety
7403 @kindex --plt-thread-safe
7404 @kindex --no-plt-thread-safe
7405 @item --plt-thread-safe
7406 @itemx --no-thread-safe
7407 With power7's weakly ordered memory model, it is possible when using
7408 lazy binding for ld.so to update a plt entry in one thread and have
7409 another thread see the individual plt entry words update in the wrong
7410 order, despite ld.so carefully writing in the correct order and using
7411 memory write barriers. To avoid this we need some sort of read
7412 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
7413 looks for calls to commonly used functions that create threads, and if
7414 seen, adds the necessary barriers. Use these options to change the
7429 @section @command{ld} and SPU ELF Support
7431 @cindex SPU ELF options
7437 This option marks an executable as a PIC plugin module.
7439 @cindex SPU overlays
7440 @kindex --no-overlays
7442 Normally, @command{ld} recognizes calls to functions within overlay
7443 regions, and redirects such calls to an overlay manager via a stub.
7444 @command{ld} also provides a built-in overlay manager. This option
7445 turns off all this special overlay handling.
7447 @cindex SPU overlay stub symbols
7448 @kindex --emit-stub-syms
7449 @item --emit-stub-syms
7450 This option causes @command{ld} to label overlay stubs with a local
7451 symbol that encodes the stub type and destination.
7453 @cindex SPU extra overlay stubs
7454 @kindex --extra-overlay-stubs
7455 @item --extra-overlay-stubs
7456 This option causes @command{ld} to add overlay call stubs on all
7457 function calls out of overlay regions. Normally stubs are not added
7458 on calls to non-overlay regions.
7460 @cindex SPU local store size
7461 @kindex --local-store=lo:hi
7462 @item --local-store=lo:hi
7463 @command{ld} usually checks that a final executable for SPU fits in
7464 the address range 0 to 256k. This option may be used to change the
7465 range. Disable the check entirely with @option{--local-store=0:0}.
7468 @kindex --stack-analysis
7469 @item --stack-analysis
7470 SPU local store space is limited. Over-allocation of stack space
7471 unnecessarily limits space available for code and data, while
7472 under-allocation results in runtime failures. If given this option,
7473 @command{ld} will provide an estimate of maximum stack usage.
7474 @command{ld} does this by examining symbols in code sections to
7475 determine the extents of functions, and looking at function prologues
7476 for stack adjusting instructions. A call-graph is created by looking
7477 for relocations on branch instructions. The graph is then searched
7478 for the maximum stack usage path. Note that this analysis does not
7479 find calls made via function pointers, and does not handle recursion
7480 and other cycles in the call graph. Stack usage may be
7481 under-estimated if your code makes such calls. Also, stack usage for
7482 dynamic allocation, e.g. alloca, will not be detected. If a link map
7483 is requested, detailed information about each function's stack usage
7484 and calls will be given.
7487 @kindex --emit-stack-syms
7488 @item --emit-stack-syms
7489 This option, if given along with @option{--stack-analysis} will result
7490 in @command{ld} emitting stack sizing symbols for each function.
7491 These take the form @code{__stack_<function_name>} for global
7492 functions, and @code{__stack_<number>_<function_name>} for static
7493 functions. @code{<number>} is the section id in hex. The value of
7494 such symbols is the stack requirement for the corresponding function.
7495 The symbol size will be zero, type @code{STT_NOTYPE}, binding
7496 @code{STB_LOCAL}, and section @code{SHN_ABS}.
7510 @section @command{ld}'s Support for Various TI COFF Versions
7511 @cindex TI COFF versions
7512 @kindex --format=@var{version}
7513 The @samp{--format} switch allows selection of one of the various
7514 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
7515 also supported. The TI COFF versions also vary in header byte-order
7516 format; @command{ld} will read any version or byte order, but the output
7517 header format depends on the default specified by the specific target.
7530 @section @command{ld} and WIN32 (cygwin/mingw)
7532 This section describes some of the win32 specific @command{ld} issues.
7533 See @ref{Options,,Command Line Options} for detailed description of the
7534 command line options mentioned here.
7537 @cindex import libraries
7538 @item import libraries
7539 The standard Windows linker creates and uses so-called import
7540 libraries, which contains information for linking to dll's. They are
7541 regular static archives and are handled as any other static
7542 archive. The cygwin and mingw ports of @command{ld} have specific
7543 support for creating such libraries provided with the
7544 @samp{--out-implib} command line option.
7546 @item exporting DLL symbols
7547 @cindex exporting DLL symbols
7548 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
7551 @item using auto-export functionality
7552 @cindex using auto-export functionality
7553 By default @command{ld} exports symbols with the auto-export functionality,
7554 which is controlled by the following command line options:
7557 @item --export-all-symbols [This is the default]
7558 @item --exclude-symbols
7559 @item --exclude-libs
7560 @item --exclude-modules-for-implib
7561 @item --version-script
7564 When auto-export is in operation, @command{ld} will export all the non-local
7565 (global and common) symbols it finds in a DLL, with the exception of a few
7566 symbols known to belong to the system's runtime and libraries. As it will
7567 often not be desirable to export all of a DLL's symbols, which may include
7568 private functions that are not part of any public interface, the command-line
7569 options listed above may be used to filter symbols out from the list for
7570 exporting. The @samp{--output-def} option can be used in order to see the
7571 final list of exported symbols with all exclusions taken into effect.
7573 If @samp{--export-all-symbols} is not given explicitly on the
7574 command line, then the default auto-export behavior will be @emph{disabled}
7575 if either of the following are true:
7578 @item A DEF file is used.
7579 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
7582 @item using a DEF file
7583 @cindex using a DEF file
7584 Another way of exporting symbols is using a DEF file. A DEF file is
7585 an ASCII file containing definitions of symbols which should be
7586 exported when a dll is created. Usually it is named @samp{<dll
7587 name>.def} and is added as any other object file to the linker's
7588 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7591 gcc -o <output> <objectfiles> <dll name>.def
7594 Using a DEF file turns off the normal auto-export behavior, unless the
7595 @samp{--export-all-symbols} option is also used.
7597 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7600 LIBRARY "xyz.dll" BASE=0x20000000
7606 another_foo = abc.dll.afoo
7612 This example defines a DLL with a non-default base address and seven
7613 symbols in the export table. The third exported symbol @code{_bar} is an
7614 alias for the second. The fourth symbol, @code{another_foo} is resolved
7615 by "forwarding" to another module and treating it as an alias for
7616 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7617 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7618 export library is an alias of @samp{foo}, which gets the string name
7619 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7620 symbol, which gets in export table the name @samp{var1}.
7622 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7623 name of the output DLL. If @samp{<name>} does not include a suffix,
7624 the default library suffix, @samp{.DLL} is appended.
7626 When the .DEF file is used to build an application, rather than a
7627 library, the @code{NAME <name>} command should be used instead of
7628 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7629 executable suffix, @samp{.EXE} is appended.
7631 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7632 specification @code{BASE = <number>} may be used to specify a
7633 non-default base address for the image.
7635 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7636 or they specify an empty string, the internal name is the same as the
7637 filename specified on the command line.
7639 The complete specification of an export symbol is:
7643 ( ( ( <name1> [ = <name2> ] )
7644 | ( <name1> = <module-name> . <external-name>))
7645 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7648 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7649 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7650 @samp{<name1>} as a "forward" alias for the symbol
7651 @samp{<external-name>} in the DLL @samp{<module-name>}.
7652 Optionally, the symbol may be exported by the specified ordinal
7653 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7654 string in import/export table for the symbol.
7656 The optional keywords that follow the declaration indicate:
7658 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7659 will still be exported by its ordinal alias (either the value specified
7660 by the .def specification or, otherwise, the value assigned by the
7661 linker). The symbol name, however, does remain visible in the import
7662 library (if any), unless @code{PRIVATE} is also specified.
7664 @code{DATA}: The symbol is a variable or object, rather than a function.
7665 The import lib will export only an indirect reference to @code{foo} as
7666 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7669 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7670 well as @code{_imp__foo} into the import library. Both refer to the
7671 read-only import address table's pointer to the variable, not to the
7672 variable itself. This can be dangerous. If the user code fails to add
7673 the @code{dllimport} attribute and also fails to explicitly add the
7674 extra indirection that the use of the attribute enforces, the
7675 application will behave unexpectedly.
7677 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7678 it into the static import library used to resolve imports at link time. The
7679 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7680 API at runtime or by by using the GNU ld extension of linking directly to
7681 the DLL without an import library.
7683 See ld/deffilep.y in the binutils sources for the full specification of
7684 other DEF file statements
7686 @cindex creating a DEF file
7687 While linking a shared dll, @command{ld} is able to create a DEF file
7688 with the @samp{--output-def <file>} command line option.
7690 @item Using decorations
7691 @cindex Using decorations
7692 Another way of marking symbols for export is to modify the source code
7693 itself, so that when building the DLL each symbol to be exported is
7697 __declspec(dllexport) int a_variable
7698 __declspec(dllexport) void a_function(int with_args)
7701 All such symbols will be exported from the DLL. If, however,
7702 any of the object files in the DLL contain symbols decorated in
7703 this way, then the normal auto-export behavior is disabled, unless
7704 the @samp{--export-all-symbols} option is also used.
7706 Note that object files that wish to access these symbols must @emph{not}
7707 decorate them with dllexport. Instead, they should use dllimport,
7711 __declspec(dllimport) int a_variable
7712 __declspec(dllimport) void a_function(int with_args)
7715 This complicates the structure of library header files, because
7716 when included by the library itself the header must declare the
7717 variables and functions as dllexport, but when included by client
7718 code the header must declare them as dllimport. There are a number
7719 of idioms that are typically used to do this; often client code can
7720 omit the __declspec() declaration completely. See
7721 @samp{--enable-auto-import} and @samp{automatic data imports} for more
7725 @cindex automatic data imports
7726 @item automatic data imports
7727 The standard Windows dll format supports data imports from dlls only
7728 by adding special decorations (dllimport/dllexport), which let the
7729 compiler produce specific assembler instructions to deal with this
7730 issue. This increases the effort necessary to port existing Un*x
7731 code to these platforms, especially for large
7732 c++ libraries and applications. The auto-import feature, which was
7733 initially provided by Paul Sokolovsky, allows one to omit the
7734 decorations to achieve a behavior that conforms to that on POSIX/Un*x
7735 platforms. This feature is enabled with the @samp{--enable-auto-import}
7736 command-line option, although it is enabled by default on cygwin/mingw.
7737 The @samp{--enable-auto-import} option itself now serves mainly to
7738 suppress any warnings that are ordinarily emitted when linked objects
7739 trigger the feature's use.
7741 auto-import of variables does not always work flawlessly without
7742 additional assistance. Sometimes, you will see this message
7744 "variable '<var>' can't be auto-imported. Please read the
7745 documentation for ld's @code{--enable-auto-import} for details."
7747 The @samp{--enable-auto-import} documentation explains why this error
7748 occurs, and several methods that can be used to overcome this difficulty.
7749 One of these methods is the @emph{runtime pseudo-relocs} feature, described
7752 @cindex runtime pseudo-relocation
7753 For complex variables imported from DLLs (such as structs or classes),
7754 object files typically contain a base address for the variable and an
7755 offset (@emph{addend}) within the variable--to specify a particular
7756 field or public member, for instance. Unfortunately, the runtime loader used
7757 in win32 environments is incapable of fixing these references at runtime
7758 without the additional information supplied by dllimport/dllexport decorations.
7759 The standard auto-import feature described above is unable to resolve these
7762 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
7763 be resolved without error, while leaving the task of adjusting the references
7764 themselves (with their non-zero addends) to specialized code provided by the
7765 runtime environment. Recent versions of the cygwin and mingw environments and
7766 compilers provide this runtime support; older versions do not. However, the
7767 support is only necessary on the developer's platform; the compiled result will
7768 run without error on an older system.
7770 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
7773 @cindex direct linking to a dll
7774 @item direct linking to a dll
7775 The cygwin/mingw ports of @command{ld} support the direct linking,
7776 including data symbols, to a dll without the usage of any import
7777 libraries. This is much faster and uses much less memory than does the
7778 traditional import library method, especially when linking large
7779 libraries or applications. When @command{ld} creates an import lib, each
7780 function or variable exported from the dll is stored in its own bfd, even
7781 though a single bfd could contain many exports. The overhead involved in
7782 storing, loading, and processing so many bfd's is quite large, and explains the
7783 tremendous time, memory, and storage needed to link against particularly
7784 large or complex libraries when using import libs.
7786 Linking directly to a dll uses no extra command-line switches other than
7787 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
7788 of names to match each library. All that is needed from the developer's
7789 perspective is an understanding of this search, in order to force ld to
7790 select the dll instead of an import library.
7793 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
7794 to find, in the first directory of its search path,
7806 before moving on to the next directory in the search path.
7808 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
7809 where @samp{<prefix>} is set by the @command{ld} option
7810 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
7811 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
7814 Other win32-based unix environments, such as mingw or pw32, may use other
7815 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
7816 was originally intended to help avoid name conflicts among dll's built for the
7817 various win32/un*x environments, so that (for example) two versions of a zlib dll
7818 could coexist on the same machine.
7820 The generic cygwin/mingw path layout uses a @samp{bin} directory for
7821 applications and dll's and a @samp{lib} directory for the import
7822 libraries (using cygwin nomenclature):
7828 libxxx.dll.a (in case of dll's)
7829 libxxx.a (in case of static archive)
7832 Linking directly to a dll without using the import library can be
7835 1. Use the dll directly by adding the @samp{bin} path to the link line
7837 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
7840 However, as the dll's often have version numbers appended to their names
7841 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
7842 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
7843 not versioned, and do not have this difficulty.
7845 2. Create a symbolic link from the dll to a file in the @samp{lib}
7846 directory according to the above mentioned search pattern. This
7847 should be used to avoid unwanted changes in the tools needed for
7851 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
7854 Then you can link without any make environment changes.
7857 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
7860 This technique also avoids the version number problems, because the following is
7867 libxxx.dll.a -> ../bin/cygxxx-5.dll
7870 Linking directly to a dll without using an import lib will work
7871 even when auto-import features are exercised, and even when
7872 @samp{--enable-runtime-pseudo-relocs} is used.
7874 Given the improvements in speed and memory usage, one might justifiably
7875 wonder why import libraries are used at all. There are three reasons:
7877 1. Until recently, the link-directly-to-dll functionality did @emph{not}
7878 work with auto-imported data.
7880 2. Sometimes it is necessary to include pure static objects within the
7881 import library (which otherwise contains only bfd's for indirection
7882 symbols that point to the exports of a dll). Again, the import lib
7883 for the cygwin kernel makes use of this ability, and it is not
7884 possible to do this without an import lib.
7886 3. Symbol aliases can only be resolved using an import lib. This is
7887 critical when linking against OS-supplied dll's (eg, the win32 API)
7888 in which symbols are usually exported as undecorated aliases of their
7889 stdcall-decorated assembly names.
7891 So, import libs are not going away. But the ability to replace
7892 true import libs with a simple symbolic link to (or a copy of)
7893 a dll, in many cases, is a useful addition to the suite of tools
7894 binutils makes available to the win32 developer. Given the
7895 massive improvements in memory requirements during linking, storage
7896 requirements, and linking speed, we expect that many developers
7897 will soon begin to use this feature whenever possible.
7899 @item symbol aliasing
7901 @item adding additional names
7902 Sometimes, it is useful to export symbols with additional names.
7903 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
7904 exported as @samp{_foo} by using special directives in the DEF file
7905 when creating the dll. This will affect also the optional created
7906 import library. Consider the following DEF file:
7909 LIBRARY "xyz.dll" BASE=0x61000000
7916 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
7918 Another method for creating a symbol alias is to create it in the
7919 source code using the "weak" attribute:
7922 void foo () @{ /* Do something. */; @}
7923 void _foo () __attribute__ ((weak, alias ("foo")));
7926 See the gcc manual for more information about attributes and weak
7929 @item renaming symbols
7930 Sometimes it is useful to rename exports. For instance, the cygwin
7931 kernel does this regularly. A symbol @samp{_foo} can be exported as
7932 @samp{foo} but not as @samp{_foo} by using special directives in the
7933 DEF file. (This will also affect the import library, if it is
7934 created). In the following example:
7937 LIBRARY "xyz.dll" BASE=0x61000000
7943 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
7947 Note: using a DEF file disables the default auto-export behavior,
7948 unless the @samp{--export-all-symbols} command line option is used.
7949 If, however, you are trying to rename symbols, then you should list
7950 @emph{all} desired exports in the DEF file, including the symbols
7951 that are not being renamed, and do @emph{not} use the
7952 @samp{--export-all-symbols} option. If you list only the
7953 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
7954 to handle the other symbols, then the both the new names @emph{and}
7955 the original names for the renamed symbols will be exported.
7956 In effect, you'd be aliasing those symbols, not renaming them,
7957 which is probably not what you wanted.
7959 @cindex weak externals
7960 @item weak externals
7961 The Windows object format, PE, specifies a form of weak symbols called
7962 weak externals. When a weak symbol is linked and the symbol is not
7963 defined, the weak symbol becomes an alias for some other symbol. There
7964 are three variants of weak externals:
7966 @item Definition is searched for in objects and libraries, historically
7967 called lazy externals.
7968 @item Definition is searched for only in other objects, not in libraries.
7969 This form is not presently implemented.
7970 @item No search; the symbol is an alias. This form is not presently
7973 As a GNU extension, weak symbols that do not specify an alternate symbol
7974 are supported. If the symbol is undefined when linking, the symbol
7975 uses a default value.
7977 @cindex aligned common symbols
7978 @item aligned common symbols
7979 As a GNU extension to the PE file format, it is possible to specify the
7980 desired alignment for a common symbol. This information is conveyed from
7981 the assembler or compiler to the linker by means of GNU-specific commands
7982 carried in the object file's @samp{.drectve} section, which are recognized
7983 by @command{ld} and respected when laying out the common symbols. Native
7984 tools will be able to process object files employing this GNU extension,
7985 but will fail to respect the alignment instructions, and may issue noisy
7986 warnings about unknown linker directives.
8001 @section @code{ld} and Xtensa Processors
8003 @cindex Xtensa processors
8004 The default @command{ld} behavior for Xtensa processors is to interpret
8005 @code{SECTIONS} commands so that lists of explicitly named sections in a
8006 specification with a wildcard file will be interleaved when necessary to
8007 keep literal pools within the range of PC-relative load offsets. For
8008 example, with the command:
8020 @command{ld} may interleave some of the @code{.literal}
8021 and @code{.text} sections from different object files to ensure that the
8022 literal pools are within the range of PC-relative load offsets. A valid
8023 interleaving might place the @code{.literal} sections from an initial
8024 group of files followed by the @code{.text} sections of that group of
8025 files. Then, the @code{.literal} sections from the rest of the files
8026 and the @code{.text} sections from the rest of the files would follow.
8028 @cindex @option{--relax} on Xtensa
8029 @cindex relaxing on Xtensa
8030 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8031 provides two important link-time optimizations. The first optimization
8032 is to combine identical literal values to reduce code size. A redundant
8033 literal will be removed and all the @code{L32R} instructions that use it
8034 will be changed to reference an identical literal, as long as the
8035 location of the replacement literal is within the offset range of all
8036 the @code{L32R} instructions. The second optimization is to remove
8037 unnecessary overhead from assembler-generated ``longcall'' sequences of
8038 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8039 range of direct @code{CALL@var{n}} instructions.
8041 For each of these cases where an indirect call sequence can be optimized
8042 to a direct call, the linker will change the @code{CALLX@var{n}}
8043 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8044 instruction, and remove the literal referenced by the @code{L32R}
8045 instruction if it is not used for anything else. Removing the
8046 @code{L32R} instruction always reduces code size but can potentially
8047 hurt performance by changing the alignment of subsequent branch targets.
8048 By default, the linker will always preserve alignments, either by
8049 switching some instructions between 24-bit encodings and the equivalent
8050 density instructions or by inserting a no-op in place of the @code{L32R}
8051 instruction that was removed. If code size is more important than
8052 performance, the @option{--size-opt} option can be used to prevent the
8053 linker from widening density instructions or inserting no-ops, except in
8054 a few cases where no-ops are required for correctness.
8056 The following Xtensa-specific command-line options can be used to
8059 @cindex Xtensa options
8062 When optimizing indirect calls to direct calls, optimize for code size
8063 more than performance. With this option, the linker will not insert
8064 no-ops or widen density instructions to preserve branch target
8065 alignment. There may still be some cases where no-ops are required to
8066 preserve the correctness of the code.
8074 @ifclear SingleFormat
8079 @cindex object file management
8080 @cindex object formats available
8082 The linker accesses object and archive files using the BFD libraries.
8083 These libraries allow the linker to use the same routines to operate on
8084 object files whatever the object file format. A different object file
8085 format can be supported simply by creating a new BFD back end and adding
8086 it to the library. To conserve runtime memory, however, the linker and
8087 associated tools are usually configured to support only a subset of the
8088 object file formats available. You can use @code{objdump -i}
8089 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8090 list all the formats available for your configuration.
8092 @cindex BFD requirements
8093 @cindex requirements for BFD
8094 As with most implementations, BFD is a compromise between
8095 several conflicting requirements. The major factor influencing
8096 BFD design was efficiency: any time used converting between
8097 formats is time which would not have been spent had BFD not
8098 been involved. This is partly offset by abstraction payback; since
8099 BFD simplifies applications and back ends, more time and care
8100 may be spent optimizing algorithms for a greater speed.
8102 One minor artifact of the BFD solution which you should bear in
8103 mind is the potential for information loss. There are two places where
8104 useful information can be lost using the BFD mechanism: during
8105 conversion and during output. @xref{BFD information loss}.
8108 * BFD outline:: How it works: an outline of BFD
8112 @section How It Works: An Outline of BFD
8113 @cindex opening object files
8114 @include bfdsumm.texi
8117 @node Reporting Bugs
8118 @chapter Reporting Bugs
8119 @cindex bugs in @command{ld}
8120 @cindex reporting bugs in @command{ld}
8122 Your bug reports play an essential role in making @command{ld} reliable.
8124 Reporting a bug may help you by bringing a solution to your problem, or
8125 it may not. But in any case the principal function of a bug report is
8126 to help the entire community by making the next version of @command{ld}
8127 work better. Bug reports are your contribution to the maintenance of
8130 In order for a bug report to serve its purpose, you must include the
8131 information that enables us to fix the bug.
8134 * Bug Criteria:: Have you found a bug?
8135 * Bug Reporting:: How to report bugs
8139 @section Have You Found a Bug?
8140 @cindex bug criteria
8142 If you are not sure whether you have found a bug, here are some guidelines:
8145 @cindex fatal signal
8146 @cindex linker crash
8147 @cindex crash of linker
8149 If the linker gets a fatal signal, for any input whatever, that is a
8150 @command{ld} bug. Reliable linkers never crash.
8152 @cindex error on valid input
8154 If @command{ld} produces an error message for valid input, that is a bug.
8156 @cindex invalid input
8158 If @command{ld} does not produce an error message for invalid input, that
8159 may be a bug. In the general case, the linker can not verify that
8160 object files are correct.
8163 If you are an experienced user of linkers, your suggestions for
8164 improvement of @command{ld} are welcome in any case.
8168 @section How to Report Bugs
8170 @cindex @command{ld} bugs, reporting
8172 A number of companies and individuals offer support for @sc{gnu}
8173 products. If you obtained @command{ld} from a support organization, we
8174 recommend you contact that organization first.
8176 You can find contact information for many support companies and
8177 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
8181 Otherwise, send bug reports for @command{ld} to
8185 The fundamental principle of reporting bugs usefully is this:
8186 @strong{report all the facts}. If you are not sure whether to state a
8187 fact or leave it out, state it!
8189 Often people omit facts because they think they know what causes the
8190 problem and assume that some details do not matter. Thus, you might
8191 assume that the name of a symbol you use in an example does not
8192 matter. Well, probably it does not, but one cannot be sure. Perhaps
8193 the bug is a stray memory reference which happens to fetch from the
8194 location where that name is stored in memory; perhaps, if the name
8195 were different, the contents of that location would fool the linker
8196 into doing the right thing despite the bug. Play it safe and give a
8197 specific, complete example. That is the easiest thing for you to do,
8198 and the most helpful.
8200 Keep in mind that the purpose of a bug report is to enable us to fix
8201 the bug if it is new to us. Therefore, always write your bug reports
8202 on the assumption that the bug has not been reported previously.
8204 Sometimes people give a few sketchy facts and ask, ``Does this ring a
8205 bell?'' This cannot help us fix a bug, so it is basically useless. We
8206 respond by asking for enough details to enable us to investigate.
8207 You might as well expedite matters by sending them to begin with.
8209 To enable us to fix the bug, you should include all these things:
8213 The version of @command{ld}. @command{ld} announces it if you start it with
8214 the @samp{--version} argument.
8216 Without this, we will not know whether there is any point in looking for
8217 the bug in the current version of @command{ld}.
8220 Any patches you may have applied to the @command{ld} source, including any
8221 patches made to the @code{BFD} library.
8224 The type of machine you are using, and the operating system name and
8228 What compiler (and its version) was used to compile @command{ld}---e.g.
8232 The command arguments you gave the linker to link your example and
8233 observe the bug. To guarantee you will not omit something important,
8234 list them all. A copy of the Makefile (or the output from make) is
8237 If we were to try to guess the arguments, we would probably guess wrong
8238 and then we might not encounter the bug.
8241 A complete input file, or set of input files, that will reproduce the
8242 bug. It is generally most helpful to send the actual object files
8243 provided that they are reasonably small. Say no more than 10K. For
8244 bigger files you can either make them available by FTP or HTTP or else
8245 state that you are willing to send the object file(s) to whomever
8246 requests them. (Note - your email will be going to a mailing list, so
8247 we do not want to clog it up with large attachments). But small
8248 attachments are best.
8250 If the source files were assembled using @code{gas} or compiled using
8251 @code{gcc}, then it may be OK to send the source files rather than the
8252 object files. In this case, be sure to say exactly what version of
8253 @code{gas} or @code{gcc} was used to produce the object files. Also say
8254 how @code{gas} or @code{gcc} were configured.
8257 A description of what behavior you observe that you believe is
8258 incorrect. For example, ``It gets a fatal signal.''
8260 Of course, if the bug is that @command{ld} gets a fatal signal, then we
8261 will certainly notice it. But if the bug is incorrect output, we might
8262 not notice unless it is glaringly wrong. You might as well not give us
8263 a chance to make a mistake.
8265 Even if the problem you experience is a fatal signal, you should still
8266 say so explicitly. Suppose something strange is going on, such as, your
8267 copy of @command{ld} is out of sync, or you have encountered a bug in the
8268 C library on your system. (This has happened!) Your copy might crash
8269 and ours would not. If you told us to expect a crash, then when ours
8270 fails to crash, we would know that the bug was not happening for us. If
8271 you had not told us to expect a crash, then we would not be able to draw
8272 any conclusion from our observations.
8275 If you wish to suggest changes to the @command{ld} source, send us context
8276 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
8277 @samp{-p} option. Always send diffs from the old file to the new file.
8278 If you even discuss something in the @command{ld} source, refer to it by
8279 context, not by line number.
8281 The line numbers in our development sources will not match those in your
8282 sources. Your line numbers would convey no useful information to us.
8285 Here are some things that are not necessary:
8289 A description of the envelope of the bug.
8291 Often people who encounter a bug spend a lot of time investigating
8292 which changes to the input file will make the bug go away and which
8293 changes will not affect it.
8295 This is often time consuming and not very useful, because the way we
8296 will find the bug is by running a single example under the debugger
8297 with breakpoints, not by pure deduction from a series of examples.
8298 We recommend that you save your time for something else.
8300 Of course, if you can find a simpler example to report @emph{instead}
8301 of the original one, that is a convenience for us. Errors in the
8302 output will be easier to spot, running under the debugger will take
8303 less time, and so on.
8305 However, simplification is not vital; if you do not want to do this,
8306 report the bug anyway and send us the entire test case you used.
8309 A patch for the bug.
8311 A patch for the bug does help us if it is a good one. But do not omit
8312 the necessary information, such as the test case, on the assumption that
8313 a patch is all we need. We might see problems with your patch and decide
8314 to fix the problem another way, or we might not understand it at all.
8316 Sometimes with a program as complicated as @command{ld} it is very hard to
8317 construct an example that will make the program follow a certain path
8318 through the code. If you do not send us the example, we will not be
8319 able to construct one, so we will not be able to verify that the bug is
8322 And if we cannot understand what bug you are trying to fix, or why your
8323 patch should be an improvement, we will not install it. A test case will
8324 help us to understand.
8327 A guess about what the bug is or what it depends on.
8329 Such guesses are usually wrong. Even we cannot guess right about such
8330 things without first using the debugger to find the facts.
8334 @appendix MRI Compatible Script Files
8335 @cindex MRI compatibility
8336 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
8337 linker, @command{ld} can use MRI compatible linker scripts as an
8338 alternative to the more general-purpose linker scripting language
8339 described in @ref{Scripts}. MRI compatible linker scripts have a much
8340 simpler command set than the scripting language otherwise used with
8341 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
8342 linker commands; these commands are described here.
8344 In general, MRI scripts aren't of much use with the @code{a.out} object
8345 file format, since it only has three sections and MRI scripts lack some
8346 features to make use of them.
8348 You can specify a file containing an MRI-compatible script using the
8349 @samp{-c} command-line option.
8351 Each command in an MRI-compatible script occupies its own line; each
8352 command line starts with the keyword that identifies the command (though
8353 blank lines are also allowed for punctuation). If a line of an
8354 MRI-compatible script begins with an unrecognized keyword, @command{ld}
8355 issues a warning message, but continues processing the script.
8357 Lines beginning with @samp{*} are comments.
8359 You can write these commands using all upper-case letters, or all
8360 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
8361 The following list shows only the upper-case form of each command.
8364 @cindex @code{ABSOLUTE} (MRI)
8365 @item ABSOLUTE @var{secname}
8366 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
8367 Normally, @command{ld} includes in the output file all sections from all
8368 the input files. However, in an MRI-compatible script, you can use the
8369 @code{ABSOLUTE} command to restrict the sections that will be present in
8370 your output program. If the @code{ABSOLUTE} command is used at all in a
8371 script, then only the sections named explicitly in @code{ABSOLUTE}
8372 commands will appear in the linker output. You can still use other
8373 input sections (whatever you select on the command line, or using
8374 @code{LOAD}) to resolve addresses in the output file.
8376 @cindex @code{ALIAS} (MRI)
8377 @item ALIAS @var{out-secname}, @var{in-secname}
8378 Use this command to place the data from input section @var{in-secname}
8379 in a section called @var{out-secname} in the linker output file.
8381 @var{in-secname} may be an integer.
8383 @cindex @code{ALIGN} (MRI)
8384 @item ALIGN @var{secname} = @var{expression}
8385 Align the section called @var{secname} to @var{expression}. The
8386 @var{expression} should be a power of two.
8388 @cindex @code{BASE} (MRI)
8389 @item BASE @var{expression}
8390 Use the value of @var{expression} as the lowest address (other than
8391 absolute addresses) in the output file.
8393 @cindex @code{CHIP} (MRI)
8394 @item CHIP @var{expression}
8395 @itemx CHIP @var{expression}, @var{expression}
8396 This command does nothing; it is accepted only for compatibility.
8398 @cindex @code{END} (MRI)
8400 This command does nothing whatever; it's only accepted for compatibility.
8402 @cindex @code{FORMAT} (MRI)
8403 @item FORMAT @var{output-format}
8404 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
8405 language, but restricted to one of these output formats:
8409 S-records, if @var{output-format} is @samp{S}
8412 IEEE, if @var{output-format} is @samp{IEEE}
8415 COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
8419 @cindex @code{LIST} (MRI)
8420 @item LIST @var{anything}@dots{}
8421 Print (to the standard output file) a link map, as produced by the
8422 @command{ld} command-line option @samp{-M}.
8424 The keyword @code{LIST} may be followed by anything on the
8425 same line, with no change in its effect.
8427 @cindex @code{LOAD} (MRI)
8428 @item LOAD @var{filename}
8429 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
8430 Include one or more object file @var{filename} in the link; this has the
8431 same effect as specifying @var{filename} directly on the @command{ld}
8434 @cindex @code{NAME} (MRI)
8435 @item NAME @var{output-name}
8436 @var{output-name} is the name for the program produced by @command{ld}; the
8437 MRI-compatible command @code{NAME} is equivalent to the command-line
8438 option @samp{-o} or the general script language command @code{OUTPUT}.
8440 @cindex @code{ORDER} (MRI)
8441 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
8442 @itemx ORDER @var{secname} @var{secname} @var{secname}
8443 Normally, @command{ld} orders the sections in its output file in the
8444 order in which they first appear in the input files. In an MRI-compatible
8445 script, you can override this ordering with the @code{ORDER} command. The
8446 sections you list with @code{ORDER} will appear first in your output
8447 file, in the order specified.
8449 @cindex @code{PUBLIC} (MRI)
8450 @item PUBLIC @var{name}=@var{expression}
8451 @itemx PUBLIC @var{name},@var{expression}
8452 @itemx PUBLIC @var{name} @var{expression}
8453 Supply a value (@var{expression}) for external symbol
8454 @var{name} used in the linker input files.
8456 @cindex @code{SECT} (MRI)
8457 @item SECT @var{secname}, @var{expression}
8458 @itemx SECT @var{secname}=@var{expression}
8459 @itemx SECT @var{secname} @var{expression}
8460 You can use any of these three forms of the @code{SECT} command to
8461 specify the start address (@var{expression}) for section @var{secname}.
8462 If you have more than one @code{SECT} statement for the same
8463 @var{secname}, only the @emph{first} sets the start address.
8466 @node GNU Free Documentation License
8467 @appendix GNU Free Documentation License
8471 @unnumbered LD Index
8476 % I think something like @@colophon should be in texinfo. In the
8478 \long\def\colophon{\hbox to0pt{}\vfill
8479 \centerline{The body of this manual is set in}
8480 \centerline{\fontname\tenrm,}
8481 \centerline{with headings in {\bf\fontname\tenbf}}
8482 \centerline{and examples in {\tt\fontname\tentt}.}
8483 \centerline{{\it\fontname\tenit\/} and}
8484 \centerline{{\sl\fontname\tensl\/}}
8485 \centerline{are used for emphasis.}\vfill}
8487 % Blame: doc@@cygnus.com, 28mar91.