3 @c Copyright (C) 1991-2019 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
46 @dircategory Software development
48 * Ld: (ld). The GNU linker.
53 This file documents the @sc{gnu} linker LD
54 @ifset VERSION_PACKAGE
55 @value{VERSION_PACKAGE}
57 version @value{VERSION}.
59 Copyright @copyright{} 1991-2019 Free Software Foundation, Inc.
61 Permission is granted to copy, distribute and/or modify this document
62 under the terms of the GNU Free Documentation License, Version 1.3
63 or any later version published by the Free Software Foundation;
64 with no Invariant Sections, with no Front-Cover Texts, and with no
65 Back-Cover Texts. A copy of the license is included in the
66 section entitled ``GNU Free Documentation License''.
70 @setchapternewpage odd
71 @settitle The GNU linker
76 @ifset VERSION_PACKAGE
77 @subtitle @value{VERSION_PACKAGE}
79 @subtitle Version @value{VERSION}
80 @author Steve Chamberlain
81 @author Ian Lance Taylor
86 \hfill Red Hat Inc\par
87 \hfill nickc\@credhat.com, doc\@redhat.com\par
88 \hfill {\it The GNU linker}\par
89 \hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
91 \global\parindent=0pt % Steve likes it this way.
94 @vskip 0pt plus 1filll
95 @c man begin COPYRIGHT
96 Copyright @copyright{} 1991-2019 Free Software Foundation, Inc.
98 Permission is granted to copy, distribute and/or modify this document
99 under the terms of the GNU Free Documentation License, Version 1.3
100 or any later version published by the Free Software Foundation;
101 with no Invariant Sections, with no Front-Cover Texts, and with no
102 Back-Cover Texts. A copy of the license is included in the
103 section entitled ``GNU Free Documentation License''.
109 @c FIXME: Talk about importance of *order* of args, cmds to linker!
114 This file documents the @sc{gnu} linker ld
115 @ifset VERSION_PACKAGE
116 @value{VERSION_PACKAGE}
118 version @value{VERSION}.
120 This document is distributed under the terms of the GNU Free
121 Documentation License version 1.3. A copy of the license is included
122 in the section entitled ``GNU Free Documentation License''.
125 * Overview:: Overview
126 * Invocation:: Invocation
127 * Scripts:: Linker Scripts
129 * Machine Dependent:: Machine Dependent Features
133 * H8/300:: ld and the H8/300
136 * Renesas:: ld and other Renesas micros
139 * ARM:: ld and the ARM family
142 * M68HC11/68HC12:: ld and the Motorola 68HC11 and 68HC12 families
145 * HPPA ELF32:: ld and HPPA 32-bit ELF
148 * M68K:: ld and Motorola 68K family
151 * MIPS:: ld and MIPS family
154 * PowerPC ELF32:: ld and PowerPC 32-bit ELF Support
157 * PowerPC64 ELF64:: ld and PowerPC64 64-bit ELF Support
160 * S/390 ELF:: ld and S/390 ELF Support
163 * SPU ELF:: ld and SPU ELF Support
166 * TI COFF:: ld and the TI COFF
169 * Win32:: ld and WIN32 (cygwin/mingw)
172 * Xtensa:: ld and Xtensa Processors
175 @ifclear SingleFormat
178 @c Following blank line required for remaining bug in makeinfo conds/menus
180 * Reporting Bugs:: Reporting Bugs
181 * MRI:: MRI Compatible Script Files
182 * GNU Free Documentation License:: GNU Free Documentation License
183 * LD Index:: LD Index
190 @cindex @sc{gnu} linker
191 @cindex what is this?
194 @c man begin SYNOPSIS
195 ld [@b{options}] @var{objfile} @dots{}
199 ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
200 the Info entries for @file{binutils} and
205 @c man begin DESCRIPTION
207 @command{ld} combines a number of object and archive files, relocates
208 their data and ties up symbol references. Usually the last step in
209 compiling a program is to run @command{ld}.
211 @command{ld} accepts Linker Command Language files written in
212 a superset of AT&T's Link Editor Command Language syntax,
213 to provide explicit and total control over the linking process.
217 This man page does not describe the command language; see the
218 @command{ld} entry in @code{info} for full details on the command
219 language and on other aspects of the GNU linker.
222 @ifclear SingleFormat
223 This version of @command{ld} uses the general purpose BFD libraries
224 to operate on object files. This allows @command{ld} to read, combine, and
225 write object files in many different formats---for example, COFF or
226 @code{a.out}. Different formats may be linked together to produce any
227 available kind of object file. @xref{BFD}, for more information.
230 Aside from its flexibility, the @sc{gnu} linker is more helpful than other
231 linkers in providing diagnostic information. Many linkers abandon
232 execution immediately upon encountering an error; whenever possible,
233 @command{ld} continues executing, allowing you to identify other errors
234 (or, in some cases, to get an output file in spite of the error).
241 @c man begin DESCRIPTION
243 The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
244 and to be as compatible as possible with other linkers. As a result,
245 you have many choices to control its behavior.
251 * Options:: Command-line Options
252 * Environment:: Environment Variables
256 @section Command-line Options
264 The linker supports a plethora of command-line options, but in actual
265 practice few of them are used in any particular context.
266 @cindex standard Unix system
267 For instance, a frequent use of @command{ld} is to link standard Unix
268 object files on a standard, supported Unix system. On such a system, to
269 link a file @code{hello.o}:
272 ld -o @var{output} /lib/crt0.o hello.o -lc
275 This tells @command{ld} to produce a file called @var{output} as the
276 result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
277 the library @code{libc.a}, which will come from the standard search
278 directories. (See the discussion of the @samp{-l} option below.)
280 Some of the command-line options to @command{ld} may be specified at any
281 point in the command line. However, options which refer to files, such
282 as @samp{-l} or @samp{-T}, cause the file to be read at the point at
283 which the option appears in the command line, relative to the object
284 files and other file options. Repeating non-file options with a
285 different argument will either have no further effect, or override prior
286 occurrences (those further to the left on the command line) of that
287 option. Options which may be meaningfully specified more than once are
288 noted in the descriptions below.
291 Non-option arguments are object files or archives which are to be linked
292 together. They may follow, precede, or be mixed in with command-line
293 options, except that an object file argument may not be placed between
294 an option and its argument.
296 Usually the linker is invoked with at least one object file, but you can
297 specify other forms of binary input files using @samp{-l}, @samp{-R},
298 and the script command language. If @emph{no} binary input files at all
299 are specified, the linker does not produce any output, and issues the
300 message @samp{No input files}.
302 If the linker cannot recognize the format of an object file, it will
303 assume that it is a linker script. A script specified in this way
304 augments the main linker script used for the link (either the default
305 linker script or the one specified by using @samp{-T}). This feature
306 permits the linker to link against a file which appears to be an object
307 or an archive, but actually merely defines some symbol values, or uses
308 @code{INPUT} or @code{GROUP} to load other objects. Specifying a
309 script in this way merely augments the main linker script, with the
310 extra commands placed after the main script; use the @samp{-T} option
311 to replace the default linker script entirely, but note the effect of
312 the @code{INSERT} command. @xref{Scripts}.
314 For options whose names are a single letter,
315 option arguments must either follow the option letter without intervening
316 whitespace, or be given as separate arguments immediately following the
317 option that requires them.
319 For options whose names are multiple letters, either one dash or two can
320 precede the option name; for example, @samp{-trace-symbol} and
321 @samp{--trace-symbol} are equivalent. Note---there is one exception to
322 this rule. Multiple letter options that start with a lower case 'o' can
323 only be preceded by two dashes. This is to reduce confusion with the
324 @samp{-o} option. So for example @samp{-omagic} sets the output file
325 name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
328 Arguments to multiple-letter options must either be separated from the
329 option name by an equals sign, or be given as separate arguments
330 immediately following the option that requires them. For example,
331 @samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
332 Unique abbreviations of the names of multiple-letter options are
335 Note---if the linker is being invoked indirectly, via a compiler driver
336 (e.g. @samp{gcc}) then all the linker command-line options should be
337 prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
338 compiler driver) like this:
341 gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
344 This is important, because otherwise the compiler driver program may
345 silently drop the linker options, resulting in a bad link. Confusion
346 may also arise when passing options that require values through a
347 driver, as the use of a space between option and argument acts as
348 a separator, and causes the driver to pass only the option to the linker
349 and the argument to the compiler. In this case, it is simplest to use
350 the joined forms of both single- and multiple-letter options, such as:
353 gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
356 Here is a table of the generic command-line switches accepted by the GNU
360 @include at-file.texi
362 @kindex -a @var{keyword}
363 @item -a @var{keyword}
364 This option is supported for HP/UX compatibility. The @var{keyword}
365 argument must be one of the strings @samp{archive}, @samp{shared}, or
366 @samp{default}. @samp{-aarchive} is functionally equivalent to
367 @samp{-Bstatic}, and the other two keywords are functionally equivalent
368 to @samp{-Bdynamic}. This option may be used any number of times.
370 @kindex --audit @var{AUDITLIB}
371 @item --audit @var{AUDITLIB}
372 Adds @var{AUDITLIB} to the @code{DT_AUDIT} entry of the dynamic section.
373 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
374 specified in the library. If specified multiple times @code{DT_AUDIT}
375 will contain a colon separated list of audit interfaces to use. If the linker
376 finds an object with an audit entry while searching for shared libraries,
377 it will add a corresponding @code{DT_DEPAUDIT} entry in the output file.
378 This option is only meaningful on ELF platforms supporting the rtld-audit
381 @ifclear SingleFormat
382 @cindex binary input format
383 @kindex -b @var{format}
384 @kindex --format=@var{format}
387 @item -b @var{input-format}
388 @itemx --format=@var{input-format}
389 @command{ld} may be configured to support more than one kind of object
390 file. If your @command{ld} is configured this way, you can use the
391 @samp{-b} option to specify the binary format for input object files
392 that follow this option on the command line. Even when @command{ld} is
393 configured to support alternative object formats, you don't usually need
394 to specify this, as @command{ld} should be configured to expect as a
395 default input format the most usual format on each machine.
396 @var{input-format} is a text string, the name of a particular format
397 supported by the BFD libraries. (You can list the available binary
398 formats with @samp{objdump -i}.)
401 You may want to use this option if you are linking files with an unusual
402 binary format. You can also use @samp{-b} to switch formats explicitly (when
403 linking object files of different formats), by including
404 @samp{-b @var{input-format}} before each group of object files in a
407 The default format is taken from the environment variable
412 You can also define the input format from a script, using the command
415 see @ref{Format Commands}.
419 @kindex -c @var{MRI-cmdfile}
420 @kindex --mri-script=@var{MRI-cmdfile}
421 @cindex compatibility, MRI
422 @item -c @var{MRI-commandfile}
423 @itemx --mri-script=@var{MRI-commandfile}
424 For compatibility with linkers produced by MRI, @command{ld} accepts script
425 files written in an alternate, restricted command language, described in
427 @ref{MRI,,MRI Compatible Script Files}.
430 the MRI Compatible Script Files section of GNU ld documentation.
432 Introduce MRI script files with
433 the option @samp{-c}; use the @samp{-T} option to run linker
434 scripts written in the general-purpose @command{ld} scripting language.
435 If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
436 specified by any @samp{-L} options.
438 @cindex common allocation
445 These three options are equivalent; multiple forms are supported for
446 compatibility with other linkers. They assign space to common symbols
447 even if a relocatable output file is specified (with @samp{-r}). The
448 script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
449 @xref{Miscellaneous Commands}.
451 @kindex --depaudit @var{AUDITLIB}
452 @kindex -P @var{AUDITLIB}
453 @item --depaudit @var{AUDITLIB}
454 @itemx -P @var{AUDITLIB}
455 Adds @var{AUDITLIB} to the @code{DT_DEPAUDIT} entry of the dynamic section.
456 @var{AUDITLIB} is not checked for existence, nor will it use the DT_SONAME
457 specified in the library. If specified multiple times @code{DT_DEPAUDIT}
458 will contain a colon separated list of audit interfaces to use. This
459 option is only meaningful on ELF platforms supporting the rtld-audit interface.
460 The -P option is provided for Solaris compatibility.
462 @cindex entry point, from command line
463 @kindex -e @var{entry}
464 @kindex --entry=@var{entry}
466 @itemx --entry=@var{entry}
467 Use @var{entry} as the explicit symbol for beginning execution of your
468 program, rather than the default entry point. If there is no symbol
469 named @var{entry}, the linker will try to parse @var{entry} as a number,
470 and use that as the entry address (the number will be interpreted in
471 base 10; you may use a leading @samp{0x} for base 16, or a leading
472 @samp{0} for base 8). @xref{Entry Point}, for a discussion of defaults
473 and other ways of specifying the entry point.
475 @kindex --exclude-libs
476 @item --exclude-libs @var{lib},@var{lib},...
477 Specifies a list of archive libraries from which symbols should not be automatically
478 exported. The library names may be delimited by commas or colons. Specifying
479 @code{--exclude-libs ALL} excludes symbols in all archive libraries from
480 automatic export. This option is available only for the i386 PE targeted
481 port of the linker and for ELF targeted ports. For i386 PE, symbols
482 explicitly listed in a .def file are still exported, regardless of this
483 option. For ELF targeted ports, symbols affected by this option will
484 be treated as hidden.
486 @kindex --exclude-modules-for-implib
487 @item --exclude-modules-for-implib @var{module},@var{module},...
488 Specifies a list of object files or archive members, from which symbols
489 should not be automatically exported, but which should be copied wholesale
490 into the import library being generated during the link. The module names
491 may be delimited by commas or colons, and must match exactly the filenames
492 used by @command{ld} to open the files; for archive members, this is simply
493 the member name, but for object files the name listed must include and
494 match precisely any path used to specify the input file on the linker's
495 command-line. This option is available only for the i386 PE targeted port
496 of the linker. Symbols explicitly listed in a .def file are still exported,
497 regardless of this option.
499 @cindex dynamic symbol table
501 @kindex --export-dynamic
502 @kindex --no-export-dynamic
504 @itemx --export-dynamic
505 @itemx --no-export-dynamic
506 When creating a dynamically linked executable, using the @option{-E}
507 option or the @option{--export-dynamic} option causes the linker to add
508 all symbols to the dynamic symbol table. The dynamic symbol table is the
509 set of symbols which are visible from dynamic objects at run time.
511 If you do not use either of these options (or use the
512 @option{--no-export-dynamic} option to restore the default behavior), the
513 dynamic symbol table will normally contain only those symbols which are
514 referenced by some dynamic object mentioned in the link.
516 If you use @code{dlopen} to load a dynamic object which needs to refer
517 back to the symbols defined by the program, rather than some other
518 dynamic object, then you will probably need to use this option when
519 linking the program itself.
521 You can also use the dynamic list to control what symbols should
522 be added to the dynamic symbol table if the output format supports it.
523 See the description of @samp{--dynamic-list}.
525 Note that this option is specific to ELF targeted ports. PE targets
526 support a similar function to export all symbols from a DLL or EXE; see
527 the description of @samp{--export-all-symbols} below.
529 @ifclear SingleFormat
530 @cindex big-endian objects
534 Link big-endian objects. This affects the default output format.
536 @cindex little-endian objects
539 Link little-endian objects. This affects the default output format.
542 @kindex -f @var{name}
543 @kindex --auxiliary=@var{name}
545 @itemx --auxiliary=@var{name}
546 When creating an ELF shared object, set the internal DT_AUXILIARY field
547 to the specified name. This tells the dynamic linker that the symbol
548 table of the shared object should be used as an auxiliary filter on the
549 symbol table of the shared object @var{name}.
551 If you later link a program against this filter object, then, when you
552 run the program, the dynamic linker will see the DT_AUXILIARY field. If
553 the dynamic linker resolves any symbols from the filter object, it will
554 first check whether there is a definition in the shared object
555 @var{name}. If there is one, it will be used instead of the definition
556 in the filter object. The shared object @var{name} need not exist.
557 Thus the shared object @var{name} may be used to provide an alternative
558 implementation of certain functions, perhaps for debugging or for
559 machine specific performance.
561 This option may be specified more than once. The DT_AUXILIARY entries
562 will be created in the order in which they appear on the command line.
564 @kindex -F @var{name}
565 @kindex --filter=@var{name}
567 @itemx --filter=@var{name}
568 When creating an ELF shared object, set the internal DT_FILTER field to
569 the specified name. This tells the dynamic linker that the symbol table
570 of the shared object which is being created should be used as a filter
571 on the symbol table of the shared object @var{name}.
573 If you later link a program against this filter object, then, when you
574 run the program, the dynamic linker will see the DT_FILTER field. The
575 dynamic linker will resolve symbols according to the symbol table of the
576 filter object as usual, but it will actually link to the definitions
577 found in the shared object @var{name}. Thus the filter object can be
578 used to select a subset of the symbols provided by the object
581 Some older linkers used the @option{-F} option throughout a compilation
582 toolchain for specifying object-file format for both input and output
584 @ifclear SingleFormat
585 The @sc{gnu} linker uses other mechanisms for this purpose: the
586 @option{-b}, @option{--format}, @option{--oformat} options, the
587 @code{TARGET} command in linker scripts, and the @code{GNUTARGET}
588 environment variable.
590 The @sc{gnu} linker will ignore the @option{-F} option when not
591 creating an ELF shared object.
593 @cindex finalization function
594 @kindex -fini=@var{name}
595 @item -fini=@var{name}
596 When creating an ELF executable or shared object, call NAME when the
597 executable or shared object is unloaded, by setting DT_FINI to the
598 address of the function. By default, the linker uses @code{_fini} as
599 the function to call.
603 Ignored. Provided for compatibility with other tools.
605 @kindex -G @var{value}
606 @kindex --gpsize=@var{value}
609 @itemx --gpsize=@var{value}
610 Set the maximum size of objects to be optimized using the GP register to
611 @var{size}. This is only meaningful for object file formats such as
612 MIPS ELF that support putting large and small objects into different
613 sections. This is ignored for other object file formats.
615 @cindex runtime library name
616 @kindex -h @var{name}
617 @kindex -soname=@var{name}
619 @itemx -soname=@var{name}
620 When creating an ELF shared object, set the internal DT_SONAME field to
621 the specified name. When an executable is linked with a shared object
622 which has a DT_SONAME field, then when the executable is run the dynamic
623 linker will attempt to load the shared object specified by the DT_SONAME
624 field rather than the using the file name given to the linker.
627 @cindex incremental link
629 Perform an incremental link (same as option @samp{-r}).
631 @cindex initialization function
632 @kindex -init=@var{name}
633 @item -init=@var{name}
634 When creating an ELF executable or shared object, call NAME when the
635 executable or shared object is loaded, by setting DT_INIT to the address
636 of the function. By default, the linker uses @code{_init} as the
639 @cindex archive files, from cmd line
640 @kindex -l @var{namespec}
641 @kindex --library=@var{namespec}
642 @item -l @var{namespec}
643 @itemx --library=@var{namespec}
644 Add the archive or object file specified by @var{namespec} to the
645 list of files to link. This option may be used any number of times.
646 If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
647 will search the library path for a file called @var{filename}, otherwise it
648 will search the library path for a file called @file{lib@var{namespec}.a}.
650 On systems which support shared libraries, @command{ld} may also search for
651 files other than @file{lib@var{namespec}.a}. Specifically, on ELF
652 and SunOS systems, @command{ld} will search a directory for a library
653 called @file{lib@var{namespec}.so} before searching for one called
654 @file{lib@var{namespec}.a}. (By convention, a @code{.so} extension
655 indicates a shared library.) Note that this behavior does not apply
656 to @file{:@var{filename}}, which always specifies a file called
659 The linker will search an archive only once, at the location where it is
660 specified on the command line. If the archive defines a symbol which
661 was undefined in some object which appeared before the archive on the
662 command line, the linker will include the appropriate file(s) from the
663 archive. However, an undefined symbol in an object appearing later on
664 the command line will not cause the linker to search the archive again.
666 See the @option{-(} option for a way to force the linker to search
667 archives multiple times.
669 You may list the same archive multiple times on the command line.
672 This type of archive searching is standard for Unix linkers. However,
673 if you are using @command{ld} on AIX, note that it is different from the
674 behaviour of the AIX linker.
677 @cindex search directory, from cmd line
679 @kindex --library-path=@var{dir}
680 @item -L @var{searchdir}
681 @itemx --library-path=@var{searchdir}
682 Add path @var{searchdir} to the list of paths that @command{ld} will search
683 for archive libraries and @command{ld} control scripts. You may use this
684 option any number of times. The directories are searched in the order
685 in which they are specified on the command line. Directories specified
686 on the command line are searched before the default directories. All
687 @option{-L} options apply to all @option{-l} options, regardless of the
688 order in which the options appear. @option{-L} options do not affect
689 how @command{ld} searches for a linker script unless @option{-T}
692 If @var{searchdir} begins with @code{=} or @code{$SYSROOT}, then this
693 prefix will be replaced by the @dfn{sysroot prefix}, controlled by the
694 @samp{--sysroot} option, or specified when the linker is configured.
697 The default set of paths searched (without being specified with
698 @samp{-L}) depends on which emulation mode @command{ld} is using, and in
699 some cases also on how it was configured. @xref{Environment}.
702 The paths can also be specified in a link script with the
703 @code{SEARCH_DIR} command. Directories specified this way are searched
704 at the point in which the linker script appears in the command line.
707 @kindex -m @var{emulation}
708 @item -m @var{emulation}
709 Emulate the @var{emulation} linker. You can list the available
710 emulations with the @samp{--verbose} or @samp{-V} options.
712 If the @samp{-m} option is not used, the emulation is taken from the
713 @code{LDEMULATION} environment variable, if that is defined.
715 Otherwise, the default emulation depends upon how the linker was
723 Print a link map to the standard output. A link map provides
724 information about the link, including the following:
728 Where object files are mapped into memory.
730 How common symbols are allocated.
732 All archive members included in the link, with a mention of the symbol
733 which caused the archive member to be brought in.
735 The values assigned to symbols.
737 Note - symbols whose values are computed by an expression which
738 involves a reference to a previous value of the same symbol may not
739 have correct result displayed in the link map. This is because the
740 linker discards intermediate results and only retains the final value
741 of an expression. Under such circumstances the linker will display
742 the final value enclosed by square brackets. Thus for example a
743 linker script containing:
751 will produce the following output in the link map if the @option{-M}
756 [0x0000000c] foo = (foo * 0x4)
757 [0x0000000c] foo = (foo + 0x8)
760 See @ref{Expressions} for more information about expressions in linker
763 @item How GNU properties are merged.
765 When linker merges input .note.gnu.property sections into one output
766 .note.gnu.property section, some properties are removed or updated,
767 which are reported in the link map as
770 Removed property 0xc0000002 to merge foo.o (0x1) and bar.o (not found)
773 It indicates that property 0xc0000002 is removed from output when
774 merging properties in @file{foo.o}, whose property 0xc0000002 value
775 is 0x1, and @file{bar.o}, which doesn't have property 0xc0000002.
778 Updated property 0xc0000002 (0x1) to merge foo.o (0x1) and bar.o (0x1)
781 It indicates that property 0xc0010001 value is updated to 0x1 in output
782 when merging properties in @file{foo.o}, whose 0xc0010001 property value
783 is 0x1, and @file{bar.o}, whose 0xc0010001 property value is 0x1.
787 @cindex read-only text
792 Turn off page alignment of sections, and disable linking against shared
793 libraries. If the output format supports Unix style magic numbers,
794 mark the output as @code{NMAGIC}.
798 @cindex read/write from cmd line
802 Set the text and data sections to be readable and writable. Also, do
803 not page-align the data segment, and disable linking against shared
804 libraries. If the output format supports Unix style magic numbers,
805 mark the output as @code{OMAGIC}. Note: Although a writable text section
806 is allowed for PE-COFF targets, it does not conform to the format
807 specification published by Microsoft.
812 This option negates most of the effects of the @option{-N} option. It
813 sets the text section to be read-only, and forces the data segment to
814 be page-aligned. Note - this option does not enable linking against
815 shared libraries. Use @option{-Bdynamic} for this.
817 @kindex -o @var{output}
818 @kindex --output=@var{output}
819 @cindex naming the output file
820 @item -o @var{output}
821 @itemx --output=@var{output}
822 Use @var{output} as the name for the program produced by @command{ld}; if this
823 option is not specified, the name @file{a.out} is used by default. The
824 script command @code{OUTPUT} can also specify the output file name.
826 @kindex -O @var{level}
827 @cindex generating optimized output
829 If @var{level} is a numeric values greater than zero @command{ld} optimizes
830 the output. This might take significantly longer and therefore probably
831 should only be enabled for the final binary. At the moment this
832 option only affects ELF shared library generation. Future releases of
833 the linker may make more use of this option. Also currently there is
834 no difference in the linker's behaviour for different non-zero values
835 of this option. Again this may change with future releases.
837 @kindex -plugin @var{name}
838 @item -plugin @var{name}
839 Involve a plugin in the linking process. The @var{name} parameter is
840 the absolute filename of the plugin. Usually this parameter is
841 automatically added by the complier, when using link time
842 optimization, but users can also add their own plugins if they so
845 Note that the location of the compiler originated plugins is different
846 from the place where the @command{ar}, @command{nm} and
847 @command{ranlib} programs search for their plugins. In order for
848 those commands to make use of a compiler based plugin it must first be
849 copied into the @file{$@{libdir@}/bfd-plugins} directory. All gcc
850 based linker plugins are backward compatible, so it is sufficient to
851 just copy in the newest one.
854 @cindex push state governing input file handling
856 The @option{--push-state} allows to preserve the current state of the
857 flags which govern the input file handling so that they can all be
858 restored with one corresponding @option{--pop-state} option.
860 The option which are covered are: @option{-Bdynamic}, @option{-Bstatic},
861 @option{-dn}, @option{-dy}, @option{-call_shared}, @option{-non_shared},
862 @option{-static}, @option{-N}, @option{-n}, @option{--whole-archive},
863 @option{--no-whole-archive}, @option{-r}, @option{-Ur},
864 @option{--copy-dt-needed-entries}, @option{--no-copy-dt-needed-entries},
865 @option{--as-needed}, @option{--no-as-needed}, and @option{-a}.
867 One target for this option are specifications for @file{pkg-config}. When
868 used with the @option{--libs} option all possibly needed libraries are
869 listed and then possibly linked with all the time. It is better to return
870 something as follows:
873 -Wl,--push-state,--as-needed -libone -libtwo -Wl,--pop-state
877 @cindex pop state governing input file handling
879 Undoes the effect of --push-state, restores the previous values of the
880 flags governing input file handling.
883 @kindex --emit-relocs
884 @cindex retain relocations in final executable
887 Leave relocation sections and contents in fully linked executables.
888 Post link analysis and optimization tools may need this information in
889 order to perform correct modifications of executables. This results
890 in larger executables.
892 This option is currently only supported on ELF platforms.
894 @kindex --force-dynamic
895 @cindex forcing the creation of dynamic sections
896 @item --force-dynamic
897 Force the output file to have dynamic sections. This option is specific
901 @cindex relocatable output
903 @kindex --relocatable
906 Generate relocatable output---i.e., generate an output file that can in
907 turn serve as input to @command{ld}. This is often called @dfn{partial
908 linking}. As a side effect, in environments that support standard Unix
909 magic numbers, this option also sets the output file's magic number to
911 @c ; see @option{-N}.
912 If this option is not specified, an absolute file is produced. When
913 linking C++ programs, this option @emph{will not} resolve references to
914 constructors; to do that, use @samp{-Ur}.
916 When an input file does not have the same format as the output file,
917 partial linking is only supported if that input file does not contain any
918 relocations. Different output formats can have further restrictions; for
919 example some @code{a.out}-based formats do not support partial linking
920 with input files in other formats at all.
922 This option does the same thing as @samp{-i}.
924 @kindex -R @var{file}
925 @kindex --just-symbols=@var{file}
926 @cindex symbol-only input
927 @item -R @var{filename}
928 @itemx --just-symbols=@var{filename}
929 Read symbol names and their addresses from @var{filename}, but do not
930 relocate it or include it in the output. This allows your output file
931 to refer symbolically to absolute locations of memory defined in other
932 programs. You may use this option more than once.
934 For compatibility with other ELF linkers, if the @option{-R} option is
935 followed by a directory name, rather than a file name, it is treated as
936 the @option{-rpath} option.
940 @cindex strip all symbols
943 Omit all symbol information from the output file.
946 @kindex --strip-debug
947 @cindex strip debugger symbols
950 Omit debugger symbol information (but not all symbols) from the output file.
952 @kindex --strip-discarded
953 @kindex --no-strip-discarded
954 @item --strip-discarded
955 @itemx --no-strip-discarded
956 Omit (or do not omit) global symbols defined in discarded sections.
961 @cindex input files, displaying
964 Print the names of the input files as @command{ld} processes them.
966 @kindex -T @var{script}
967 @kindex --script=@var{script}
969 @item -T @var{scriptfile}
970 @itemx --script=@var{scriptfile}
971 Use @var{scriptfile} as the linker script. This script replaces
972 @command{ld}'s default linker script (rather than adding to it), so
973 @var{commandfile} must specify everything necessary to describe the
974 output file. @xref{Scripts}. If @var{scriptfile} does not exist in
975 the current directory, @code{ld} looks for it in the directories
976 specified by any preceding @samp{-L} options. Multiple @samp{-T}
979 @kindex -dT @var{script}
980 @kindex --default-script=@var{script}
982 @item -dT @var{scriptfile}
983 @itemx --default-script=@var{scriptfile}
984 Use @var{scriptfile} as the default linker script. @xref{Scripts}.
986 This option is similar to the @option{--script} option except that
987 processing of the script is delayed until after the rest of the
988 command line has been processed. This allows options placed after the
989 @option{--default-script} option on the command line to affect the
990 behaviour of the linker script, which can be important when the linker
991 command line cannot be directly controlled by the user. (eg because
992 the command line is being constructed by another tool, such as
995 @kindex -u @var{symbol}
996 @kindex --undefined=@var{symbol}
997 @cindex undefined symbol
998 @item -u @var{symbol}
999 @itemx --undefined=@var{symbol}
1000 Force @var{symbol} to be entered in the output file as an undefined
1001 symbol. Doing this may, for example, trigger linking of additional
1002 modules from standard libraries. @samp{-u} may be repeated with
1003 different option arguments to enter additional undefined symbols. This
1004 option is equivalent to the @code{EXTERN} linker script command.
1006 If this option is being used to force additional modules to be pulled
1007 into the link, and if it is an error for the symbol to remain
1008 undefined, then the option @option{--require-defined} should be used
1011 @kindex --require-defined=@var{symbol}
1012 @cindex symbols, require defined
1013 @cindex defined symbol
1014 @item --require-defined=@var{symbol}
1015 Require that @var{symbol} is defined in the output file. This option
1016 is the same as option @option{--undefined} except that if @var{symbol}
1017 is not defined in the output file then the linker will issue an error
1018 and exit. The same effect can be achieved in a linker script by using
1019 @code{EXTERN}, @code{ASSERT} and @code{DEFINED} together. This option
1020 can be used multiple times to require additional symbols.
1023 @cindex constructors
1025 For anything other than C++ programs, this option is equivalent to
1026 @samp{-r}: it generates relocatable output---i.e., an output file that can in
1027 turn serve as input to @command{ld}. When linking C++ programs, @samp{-Ur}
1028 @emph{does} resolve references to constructors, unlike @samp{-r}.
1029 It does not work to use @samp{-Ur} on files that were themselves linked
1030 with @samp{-Ur}; once the constructor table has been built, it cannot
1031 be added to. Use @samp{-Ur} only for the last partial link, and
1032 @samp{-r} for the others.
1034 @kindex --orphan-handling=@var{MODE}
1035 @cindex orphan sections
1036 @cindex sections, orphan
1037 @item --orphan-handling=@var{MODE}
1038 Control how orphan sections are handled. An orphan section is one not
1039 specifically mentioned in a linker script. @xref{Orphan Sections}.
1041 @var{MODE} can have any of the following values:
1045 Orphan sections are placed into a suitable output section following
1046 the strategy described in @ref{Orphan Sections}. The option
1047 @samp{--unique} also affects how sections are placed.
1050 All orphan sections are discarded, by placing them in the
1051 @samp{/DISCARD/} section (@pxref{Output Section Discarding}).
1054 The linker will place the orphan section as for @code{place} and also
1058 The linker will exit with an error if any orphan section is found.
1061 The default if @samp{--orphan-handling} is not given is @code{place}.
1063 @kindex --unique[=@var{SECTION}]
1064 @item --unique[=@var{SECTION}]
1065 Creates a separate output section for every input section matching
1066 @var{SECTION}, or if the optional wildcard @var{SECTION} argument is
1067 missing, for every orphan input section. An orphan section is one not
1068 specifically mentioned in a linker script. You may use this option
1069 multiple times on the command line; It prevents the normal merging of
1070 input sections with the same name, overriding output section assignments
1080 Display the version number for @command{ld}. The @option{-V} option also
1081 lists the supported emulations.
1084 @kindex --discard-all
1085 @cindex deleting local symbols
1087 @itemx --discard-all
1088 Delete all local symbols.
1091 @kindex --discard-locals
1092 @cindex local symbols, deleting
1094 @itemx --discard-locals
1095 Delete all temporary local symbols. (These symbols start with
1096 system-specific local label prefixes, typically @samp{.L} for ELF systems
1097 or @samp{L} for traditional a.out systems.)
1099 @kindex -y @var{symbol}
1100 @kindex --trace-symbol=@var{symbol}
1101 @cindex symbol tracing
1102 @item -y @var{symbol}
1103 @itemx --trace-symbol=@var{symbol}
1104 Print the name of each linked file in which @var{symbol} appears. This
1105 option may be given any number of times. On many systems it is necessary
1106 to prepend an underscore.
1108 This option is useful when you have an undefined symbol in your link but
1109 don't know where the reference is coming from.
1111 @kindex -Y @var{path}
1113 Add @var{path} to the default library search path. This option exists
1114 for Solaris compatibility.
1116 @kindex -z @var{keyword}
1117 @item -z @var{keyword}
1118 The recognized keywords are:
1122 Always generate BND prefix in PLT entries. Supported for Linux/x86_64.
1124 @item call-nop=prefix-addr
1125 @itemx call-nop=suffix-nop
1126 @itemx call-nop=prefix-@var{byte}
1127 @itemx call-nop=suffix-@var{byte}
1128 Specify the 1-byte @code{NOP} padding when transforming indirect call
1129 to a locally defined function, foo, via its GOT slot.
1130 @option{call-nop=prefix-addr} generates @code{0x67 call foo}.
1131 @option{call-nop=suffix-nop} generates @code{call foo 0x90}.
1132 @option{call-nop=prefix-@var{byte}} generates @code{@var{byte} call foo}.
1133 @option{call-nop=suffix-@var{byte}} generates @code{call foo @var{byte}}.
1134 Supported for i386 and x86_64.
1138 Combine multiple dynamic relocation sections and sort to improve
1139 dynamic symbol lookup caching. Do not do this if @samp{nocombreloc}.
1143 Generate common symbols with STT_COMMON type during a relocatable
1144 link. Use STT_OBJECT type if @samp{nocommon}.
1146 @item common-page-size=@var{value}
1147 Set the page size most commonly used to @var{value}. Memory image
1148 layout will be optimized to minimize memory pages if the system is
1149 using pages of this size.
1152 Report unresolved symbol references from regular object files. This
1153 is done even if the linker is creating a non-symbolic shared library.
1154 This option is the inverse of @samp{-z undefs}.
1156 @item dynamic-undefined-weak
1157 @itemx nodynamic-undefined-weak
1158 Make undefined weak symbols dynamic when building a dynamic object,
1159 if they are referenced from a regular object file and not forced local
1160 by symbol visibility or versioning. Do not make them dynamic if
1161 @samp{nodynamic-undefined-weak}. If neither option is given, a target
1162 may default to either option being in force, or make some other
1163 selection of undefined weak symbols dynamic. Not all targets support
1167 Marks the object as requiring executable stack.
1170 This option is only meaningful when building a shared object. It makes
1171 the symbols defined by this shared object available for symbol resolution
1172 of subsequently loaded libraries.
1175 This option is only meaningful when building a dynamic executable.
1176 This option marks the executable as requiring global auditing by
1177 setting the @code{DF_1_GLOBAUDIT} bit in the @code{DT_FLAGS_1} dynamic
1178 tag. Global auditing requires that any auditing library defined via
1179 the @option{--depaudit} or @option{-P} command-line options be run for
1180 all dynamic objects loaded by the application.
1183 Generate Intel Indirect Branch Tracking (IBT) enabled PLT entries.
1184 Supported for Linux/i386 and Linux/x86_64.
1187 Generate GNU_PROPERTY_X86_FEATURE_1_IBT in .note.gnu.property section
1188 to indicate compatibility with IBT. This also implies @option{ibtplt}.
1189 Supported for Linux/i386 and Linux/x86_64.
1192 This option is only meaningful when building a shared object.
1193 It marks the object so that its runtime initialization will occur
1194 before the runtime initialization of any other objects brought into
1195 the process at the same time. Similarly the runtime finalization of
1196 the object will occur after the runtime finalization of any other
1200 Specify that the dynamic loader should modify its symbol search order
1201 so that symbols in this shared library interpose all other shared
1202 libraries not so marked.
1205 When generating an executable or shared library, mark it to tell the
1206 dynamic linker to defer function call resolution to the point when
1207 the function is called (lazy binding), rather than at load time.
1208 Lazy binding is the default.
1211 Specify that the object's filters be processed immediately at runtime.
1213 @item max-page-size=@var{value}
1214 Set the maximum memory page size supported to @var{value}.
1217 Allow multiple definitions.
1220 Disable linker generated .dynbss variables used in place of variables
1221 defined in shared libraries. May result in dynamic text relocations.
1224 Specify that the dynamic loader search for dependencies of this object
1225 should ignore any default library search paths.
1228 Specify that the object shouldn't be unloaded at runtime.
1231 Specify that the object is not available to @code{dlopen}.
1234 Specify that the object can not be dumped by @code{dldump}.
1237 Marks the object as not requiring executable stack.
1239 @item noextern-protected-data
1240 Don't treat protected data symbols as external when building a shared
1241 library. This option overrides the linker backend default. It can be
1242 used to work around incorrect relocations against protected data symbols
1243 generated by compiler. Updates on protected data symbols by another
1244 module aren't visible to the resulting shared library. Supported for
1247 @item noreloc-overflow
1248 Disable relocation overflow check. This can be used to disable
1249 relocation overflow check if there will be no dynamic relocation
1250 overflow at run-time. Supported for x86_64.
1253 When generating an executable or shared library, mark it to tell the
1254 dynamic linker to resolve all symbols when the program is started, or
1255 when the shared library is loaded by dlopen, instead of deferring
1256 function call resolution to the point when the function is first
1260 Specify that the object requires @samp{$ORIGIN} handling in paths.
1264 Create an ELF @code{PT_GNU_RELRO} segment header in the object. This
1265 specifies a memory segment that should be made read-only after
1266 relocation, if supported. Specifying @samp{common-page-size} smaller
1267 than the system page size will render this protection ineffective.
1268 Don't create an ELF @code{PT_GNU_RELRO} segment if @samp{norelro}.
1271 @itemx noseparate-code
1272 Create separate code @code{PT_LOAD} segment header in the object. This
1273 specifies a memory segment that should contain only instructions and must
1274 be in wholly disjoint pages from any other data. Don't create separate
1275 code @code{PT_LOAD} segment if @samp{noseparate-code} is used.
1278 Generate GNU_PROPERTY_X86_FEATURE_1_SHSTK in .note.gnu.property section
1279 to indicate compatibility with Intel Shadow Stack. Supported for
1280 Linux/i386 and Linux/x86_64.
1282 @item stack-size=@var{value}
1283 Specify a stack size for an ELF @code{PT_GNU_STACK} segment.
1284 Specifying zero will override any default non-zero sized
1285 @code{PT_GNU_STACK} segment creation.
1290 Report an error if DT_TEXTREL is set, i.e., if the binary has dynamic
1291 relocations in read-only sections. Don't report an error if
1292 @samp{notext} or @samp{textoff}.
1295 Do not report unresolved symbol references from regular object files,
1296 either when creating an executable, or when creating a shared library.
1297 This option is the inverse of @samp{-z defs}.
1301 Other keywords are ignored for Solaris compatibility.
1304 @cindex groups of archives
1305 @item -( @var{archives} -)
1306 @itemx --start-group @var{archives} --end-group
1307 The @var{archives} should be a list of archive files. They may be
1308 either explicit file names, or @samp{-l} options.
1310 The specified archives are searched repeatedly until no new undefined
1311 references are created. Normally, an archive is searched only once in
1312 the order that it is specified on the command line. If a symbol in that
1313 archive is needed to resolve an undefined symbol referred to by an
1314 object in an archive that appears later on the command line, the linker
1315 would not be able to resolve that reference. By grouping the archives,
1316 they all be searched repeatedly until all possible references are
1319 Using this option has a significant performance cost. It is best to use
1320 it only when there are unavoidable circular references between two or
1323 @kindex --accept-unknown-input-arch
1324 @kindex --no-accept-unknown-input-arch
1325 @item --accept-unknown-input-arch
1326 @itemx --no-accept-unknown-input-arch
1327 Tells the linker to accept input files whose architecture cannot be
1328 recognised. The assumption is that the user knows what they are doing
1329 and deliberately wants to link in these unknown input files. This was
1330 the default behaviour of the linker, before release 2.14. The default
1331 behaviour from release 2.14 onwards is to reject such input files, and
1332 so the @samp{--accept-unknown-input-arch} option has been added to
1333 restore the old behaviour.
1336 @kindex --no-as-needed
1338 @itemx --no-as-needed
1339 This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1340 on the command line after the @option{--as-needed} option. Normally
1341 the linker will add a DT_NEEDED tag for each dynamic library mentioned
1342 on the command line, regardless of whether the library is actually
1343 needed or not. @option{--as-needed} causes a DT_NEEDED tag to only be
1344 emitted for a library that @emph{at that point in the link} satisfies a
1345 non-weak undefined symbol reference from a regular object file or, if
1346 the library is not found in the DT_NEEDED lists of other needed libraries, a
1347 non-weak undefined symbol reference from another needed dynamic library.
1348 Object files or libraries appearing on the command line @emph{after}
1349 the library in question do not affect whether the library is seen as
1350 needed. This is similar to the rules for extraction of object files
1351 from archives. @option{--no-as-needed} restores the default behaviour.
1353 @kindex --add-needed
1354 @kindex --no-add-needed
1356 @itemx --no-add-needed
1357 These two options have been deprecated because of the similarity of
1358 their names to the @option{--as-needed} and @option{--no-as-needed}
1359 options. They have been replaced by @option{--copy-dt-needed-entries}
1360 and @option{--no-copy-dt-needed-entries}.
1362 @kindex -assert @var{keyword}
1363 @item -assert @var{keyword}
1364 This option is ignored for SunOS compatibility.
1368 @kindex -call_shared
1372 Link against dynamic libraries. This is only meaningful on platforms
1373 for which shared libraries are supported. This option is normally the
1374 default on such platforms. The different variants of this option are
1375 for compatibility with various systems. You may use this option
1376 multiple times on the command line: it affects library searching for
1377 @option{-l} options which follow it.
1381 Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1382 section. This causes the runtime linker to handle lookups in this
1383 object and its dependencies to be performed only inside the group.
1384 @option{--unresolved-symbols=report-all} is implied. This option is
1385 only meaningful on ELF platforms which support shared libraries.
1395 Do not link against shared libraries. This is only meaningful on
1396 platforms for which shared libraries are supported. The different
1397 variants of this option are for compatibility with various systems. You
1398 may use this option multiple times on the command line: it affects
1399 library searching for @option{-l} options which follow it. This
1400 option also implies @option{--unresolved-symbols=report-all}. This
1401 option can be used with @option{-shared}. Doing so means that a
1402 shared library is being created but that all of the library's external
1403 references must be resolved by pulling in entries from static
1408 When creating a shared library, bind references to global symbols to the
1409 definition within the shared library, if any. Normally, it is possible
1410 for a program linked against a shared library to override the definition
1411 within the shared library. This option can also be used with the
1412 @option{--export-dynamic} option, when creating a position independent
1413 executable, to bind references to global symbols to the definition within
1414 the executable. This option is only meaningful on ELF platforms which
1415 support shared libraries and position independent executables.
1417 @kindex -Bsymbolic-functions
1418 @item -Bsymbolic-functions
1419 When creating a shared library, bind references to global function
1420 symbols to the definition within the shared library, if any.
1421 This option can also be used with the @option{--export-dynamic} option,
1422 when creating a position independent executable, to bind references
1423 to global function symbols to the definition within the executable.
1424 This option is only meaningful on ELF platforms which support shared
1425 libraries and position independent executables.
1427 @kindex --dynamic-list=@var{dynamic-list-file}
1428 @item --dynamic-list=@var{dynamic-list-file}
1429 Specify the name of a dynamic list file to the linker. This is
1430 typically used when creating shared libraries to specify a list of
1431 global symbols whose references shouldn't be bound to the definition
1432 within the shared library, or creating dynamically linked executables
1433 to specify a list of symbols which should be added to the symbol table
1434 in the executable. This option is only meaningful on ELF platforms
1435 which support shared libraries.
1437 The format of the dynamic list is the same as the version node without
1438 scope and node name. See @ref{VERSION} for more information.
1440 @kindex --dynamic-list-data
1441 @item --dynamic-list-data
1442 Include all global data symbols to the dynamic list.
1444 @kindex --dynamic-list-cpp-new
1445 @item --dynamic-list-cpp-new
1446 Provide the builtin dynamic list for C++ operator new and delete. It
1447 is mainly useful for building shared libstdc++.
1449 @kindex --dynamic-list-cpp-typeinfo
1450 @item --dynamic-list-cpp-typeinfo
1451 Provide the builtin dynamic list for C++ runtime type identification.
1453 @kindex --check-sections
1454 @kindex --no-check-sections
1455 @item --check-sections
1456 @itemx --no-check-sections
1457 Asks the linker @emph{not} to check section addresses after they have
1458 been assigned to see if there are any overlaps. Normally the linker will
1459 perform this check, and if it finds any overlaps it will produce
1460 suitable error messages. The linker does know about, and does make
1461 allowances for sections in overlays. The default behaviour can be
1462 restored by using the command-line switch @option{--check-sections}.
1463 Section overlap is not usually checked for relocatable links. You can
1464 force checking in that case by using the @option{--check-sections}
1467 @kindex --copy-dt-needed-entries
1468 @kindex --no-copy-dt-needed-entries
1469 @item --copy-dt-needed-entries
1470 @itemx --no-copy-dt-needed-entries
1471 This option affects the treatment of dynamic libraries referred to
1472 by DT_NEEDED tags @emph{inside} ELF dynamic libraries mentioned on the
1473 command line. Normally the linker won't add a DT_NEEDED tag to the
1474 output binary for each library mentioned in a DT_NEEDED tag in an
1475 input dynamic library. With @option{--copy-dt-needed-entries}
1476 specified on the command line however any dynamic libraries that
1477 follow it will have their DT_NEEDED entries added. The default
1478 behaviour can be restored with @option{--no-copy-dt-needed-entries}.
1480 This option also has an effect on the resolution of symbols in dynamic
1481 libraries. With @option{--copy-dt-needed-entries} dynamic libraries
1482 mentioned on the command line will be recursively searched, following
1483 their DT_NEEDED tags to other libraries, in order to resolve symbols
1484 required by the output binary. With the default setting however
1485 the searching of dynamic libraries that follow it will stop with the
1486 dynamic library itself. No DT_NEEDED links will be traversed to resolve
1489 @cindex cross reference table
1492 Output a cross reference table. If a linker map file is being
1493 generated, the cross reference table is printed to the map file.
1494 Otherwise, it is printed on the standard output.
1496 The format of the table is intentionally simple, so that it may be
1497 easily processed by a script if necessary. The symbols are printed out,
1498 sorted by name. For each symbol, a list of file names is given. If the
1499 symbol is defined, the first file listed is the location of the
1500 definition. If the symbol is defined as a common value then any files
1501 where this happens appear next. Finally any files that reference the
1504 @cindex common allocation
1505 @kindex --no-define-common
1506 @item --no-define-common
1507 This option inhibits the assignment of addresses to common symbols.
1508 The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1509 @xref{Miscellaneous Commands}.
1511 The @samp{--no-define-common} option allows decoupling
1512 the decision to assign addresses to Common symbols from the choice
1513 of the output file type; otherwise a non-Relocatable output type
1514 forces assigning addresses to Common symbols.
1515 Using @samp{--no-define-common} allows Common symbols that are referenced
1516 from a shared library to be assigned addresses only in the main program.
1517 This eliminates the unused duplicate space in the shared library,
1518 and also prevents any possible confusion over resolving to the wrong
1519 duplicate when there are many dynamic modules with specialized search
1520 paths for runtime symbol resolution.
1522 @cindex group allocation in linker script
1523 @cindex section groups
1525 @kindex --force-group-allocation
1526 @item --force-group-allocation
1527 This option causes the linker to place section group members like
1528 normal input sections, and to delete the section groups. This is the
1529 default behaviour for a final link but this option can be used to
1530 change the behaviour of a relocatable link (@samp{-r}). The script
1531 command @code{FORCE_GROUP_ALLOCATION} has the same
1532 effect. @xref{Miscellaneous Commands}.
1534 @cindex symbols, from command line
1535 @kindex --defsym=@var{symbol}=@var{exp}
1536 @item --defsym=@var{symbol}=@var{expression}
1537 Create a global symbol in the output file, containing the absolute
1538 address given by @var{expression}. You may use this option as many
1539 times as necessary to define multiple symbols in the command line. A
1540 limited form of arithmetic is supported for the @var{expression} in this
1541 context: you may give a hexadecimal constant or the name of an existing
1542 symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1543 constants or symbols. If you need more elaborate expressions, consider
1544 using the linker command language from a script (@pxref{Assignments}).
1545 @emph{Note:} there should be no white space between @var{symbol}, the
1546 equals sign (``@key{=}''), and @var{expression}.
1548 @cindex demangling, from command line
1549 @kindex --demangle[=@var{style}]
1550 @kindex --no-demangle
1551 @item --demangle[=@var{style}]
1552 @itemx --no-demangle
1553 These options control whether to demangle symbol names in error messages
1554 and other output. When the linker is told to demangle, it tries to
1555 present symbol names in a readable fashion: it strips leading
1556 underscores if they are used by the object file format, and converts C++
1557 mangled symbol names into user readable names. Different compilers have
1558 different mangling styles. The optional demangling style argument can be used
1559 to choose an appropriate demangling style for your compiler. The linker will
1560 demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1561 is set. These options may be used to override the default.
1563 @cindex dynamic linker, from command line
1564 @kindex -I@var{file}
1565 @kindex --dynamic-linker=@var{file}
1567 @itemx --dynamic-linker=@var{file}
1568 Set the name of the dynamic linker. This is only meaningful when
1569 generating dynamically linked ELF executables. The default dynamic
1570 linker is normally correct; don't use this unless you know what you are
1573 @kindex --no-dynamic-linker
1574 @item --no-dynamic-linker
1575 When producing an executable file, omit the request for a dynamic
1576 linker to be used at load-time. This is only meaningful for ELF
1577 executables that contain dynamic relocations, and usually requires
1578 entry point code that is capable of processing these relocations.
1580 @kindex --embedded-relocs
1581 @item --embedded-relocs
1582 This option is similar to the @option{--emit-relocs} option except
1583 that the relocs are stored in a target specific section. This option
1584 is only supported by the @samp{BFIN}, @samp{CR16} and @emph{M68K}
1587 @kindex --disable-multiple-abs-defs
1588 @item --disable-multiple-abs-defs
1589 Do not allow multiple definitions with symbols included
1590 in filename invoked by -R or --just-symbols
1592 @kindex --fatal-warnings
1593 @kindex --no-fatal-warnings
1594 @item --fatal-warnings
1595 @itemx --no-fatal-warnings
1596 Treat all warnings as errors. The default behaviour can be restored
1597 with the option @option{--no-fatal-warnings}.
1599 @kindex --force-exe-suffix
1600 @item --force-exe-suffix
1601 Make sure that an output file has a .exe suffix.
1603 If a successfully built fully linked output file does not have a
1604 @code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1605 the output file to one of the same name with a @code{.exe} suffix. This
1606 option is useful when using unmodified Unix makefiles on a Microsoft
1607 Windows host, since some versions of Windows won't run an image unless
1608 it ends in a @code{.exe} suffix.
1610 @kindex --gc-sections
1611 @kindex --no-gc-sections
1612 @cindex garbage collection
1614 @itemx --no-gc-sections
1615 Enable garbage collection of unused input sections. It is ignored on
1616 targets that do not support this option. The default behaviour (of not
1617 performing this garbage collection) can be restored by specifying
1618 @samp{--no-gc-sections} on the command line. Note that garbage
1619 collection for COFF and PE format targets is supported, but the
1620 implementation is currently considered to be experimental.
1622 @samp{--gc-sections} decides which input sections are used by
1623 examining symbols and relocations. The section containing the entry
1624 symbol and all sections containing symbols undefined on the
1625 command-line will be kept, as will sections containing symbols
1626 referenced by dynamic objects. Note that when building shared
1627 libraries, the linker must assume that any visible symbol is
1628 referenced. Once this initial set of sections has been determined,
1629 the linker recursively marks as used any section referenced by their
1630 relocations. See @samp{--entry}, @samp{--undefined}, and
1631 @samp{--gc-keep-exported}.
1633 This option can be set when doing a partial link (enabled with option
1634 @samp{-r}). In this case the root of symbols kept must be explicitly
1635 specified either by one of the options @samp{--entry},
1636 @samp{--undefined}, or @samp{--gc-keep-exported} or by a @code{ENTRY}
1637 command in the linker script.
1639 @kindex --print-gc-sections
1640 @kindex --no-print-gc-sections
1641 @cindex garbage collection
1642 @item --print-gc-sections
1643 @itemx --no-print-gc-sections
1644 List all sections removed by garbage collection. The listing is
1645 printed on stderr. This option is only effective if garbage
1646 collection has been enabled via the @samp{--gc-sections}) option. The
1647 default behaviour (of not listing the sections that are removed) can
1648 be restored by specifying @samp{--no-print-gc-sections} on the command
1651 @kindex --gc-keep-exported
1652 @cindex garbage collection
1653 @item --gc-keep-exported
1654 When @samp{--gc-sections} is enabled, this option prevents garbage
1655 collection of unused input sections that contain global symbols having
1656 default or protected visibility. This option is intended to be used for
1657 executables where unreferenced sections would otherwise be garbage
1658 collected regardless of the external visibility of contained symbols.
1659 Note that this option has no effect when linking shared objects since
1660 it is already the default behaviour. This option is only supported for
1663 @kindex --print-output-format
1664 @cindex output format
1665 @item --print-output-format
1666 Print the name of the default output format (perhaps influenced by
1667 other command-line options). This is the string that would appear
1668 in an @code{OUTPUT_FORMAT} linker script command (@pxref{File Commands}).
1670 @kindex --print-memory-usage
1671 @cindex memory usage
1672 @item --print-memory-usage
1673 Print used size, total size and used size of memory regions created with
1674 the @ref{MEMORY} command. This is useful on embedded targets to have a
1675 quick view of amount of free memory. The format of the output has one
1676 headline and one line per region. It is both human readable and easily
1677 parsable by tools. Here is an example of an output:
1680 Memory region Used Size Region Size %age Used
1681 ROM: 256 KB 1 MB 25.00%
1682 RAM: 32 B 2 GB 0.00%
1689 Print a summary of the command-line options on the standard output and exit.
1691 @kindex --target-help
1693 Print a summary of all target specific options on the standard output and exit.
1695 @kindex -Map=@var{mapfile}
1696 @item -Map=@var{mapfile}
1697 Print a link map to the file @var{mapfile}. See the description of the
1698 @option{-M} option, above.
1700 @cindex memory usage
1701 @kindex --no-keep-memory
1702 @item --no-keep-memory
1703 @command{ld} normally optimizes for speed over memory usage by caching the
1704 symbol tables of input files in memory. This option tells @command{ld} to
1705 instead optimize for memory usage, by rereading the symbol tables as
1706 necessary. This may be required if @command{ld} runs out of memory space
1707 while linking a large executable.
1709 @kindex --no-undefined
1712 @item --no-undefined
1714 Report unresolved symbol references from regular object files. This
1715 is done even if the linker is creating a non-symbolic shared library.
1716 The switch @option{--[no-]allow-shlib-undefined} controls the
1717 behaviour for reporting unresolved references found in shared
1718 libraries being linked in.
1720 The effects of this option can be reverted by using @code{-z undefs}.
1722 @kindex --allow-multiple-definition
1724 @item --allow-multiple-definition
1726 Normally when a symbol is defined multiple times, the linker will
1727 report a fatal error. These options allow multiple definitions and the
1728 first definition will be used.
1730 @kindex --allow-shlib-undefined
1731 @kindex --no-allow-shlib-undefined
1732 @item --allow-shlib-undefined
1733 @itemx --no-allow-shlib-undefined
1734 Allows or disallows undefined symbols in shared libraries.
1735 This switch is similar to @option{--no-undefined} except that it
1736 determines the behaviour when the undefined symbols are in a
1737 shared library rather than a regular object file. It does not affect
1738 how undefined symbols in regular object files are handled.
1740 The default behaviour is to report errors for any undefined symbols
1741 referenced in shared libraries if the linker is being used to create
1742 an executable, but to allow them if the linker is being used to create
1745 The reasons for allowing undefined symbol references in shared
1746 libraries specified at link time are that:
1750 A shared library specified at link time may not be the same as the one
1751 that is available at load time, so the symbol might actually be
1752 resolvable at load time.
1754 There are some operating systems, eg BeOS and HPPA, where undefined
1755 symbols in shared libraries are normal.
1757 The BeOS kernel for example patches shared libraries at load time to
1758 select whichever function is most appropriate for the current
1759 architecture. This is used, for example, to dynamically select an
1760 appropriate memset function.
1763 @kindex --no-undefined-version
1764 @item --no-undefined-version
1765 Normally when a symbol has an undefined version, the linker will ignore
1766 it. This option disallows symbols with undefined version and a fatal error
1767 will be issued instead.
1769 @kindex --default-symver
1770 @item --default-symver
1771 Create and use a default symbol version (the soname) for unversioned
1774 @kindex --default-imported-symver
1775 @item --default-imported-symver
1776 Create and use a default symbol version (the soname) for unversioned
1779 @kindex --no-warn-mismatch
1780 @item --no-warn-mismatch
1781 Normally @command{ld} will give an error if you try to link together input
1782 files that are mismatched for some reason, perhaps because they have
1783 been compiled for different processors or for different endiannesses.
1784 This option tells @command{ld} that it should silently permit such possible
1785 errors. This option should only be used with care, in cases when you
1786 have taken some special action that ensures that the linker errors are
1789 @kindex --no-warn-search-mismatch
1790 @item --no-warn-search-mismatch
1791 Normally @command{ld} will give a warning if it finds an incompatible
1792 library during a library search. This option silences the warning.
1794 @kindex --no-whole-archive
1795 @item --no-whole-archive
1796 Turn off the effect of the @option{--whole-archive} option for subsequent
1799 @cindex output file after errors
1800 @kindex --noinhibit-exec
1801 @item --noinhibit-exec
1802 Retain the executable output file whenever it is still usable.
1803 Normally, the linker will not produce an output file if it encounters
1804 errors during the link process; it exits without writing an output file
1805 when it issues any error whatsoever.
1809 Only search library directories explicitly specified on the
1810 command line. Library directories specified in linker scripts
1811 (including linker scripts specified on the command line) are ignored.
1813 @ifclear SingleFormat
1814 @kindex --oformat=@var{output-format}
1815 @item --oformat=@var{output-format}
1816 @command{ld} may be configured to support more than one kind of object
1817 file. If your @command{ld} is configured this way, you can use the
1818 @samp{--oformat} option to specify the binary format for the output
1819 object file. Even when @command{ld} is configured to support alternative
1820 object formats, you don't usually need to specify this, as @command{ld}
1821 should be configured to produce as a default output format the most
1822 usual format on each machine. @var{output-format} is a text string, the
1823 name of a particular format supported by the BFD libraries. (You can
1824 list the available binary formats with @samp{objdump -i}.) The script
1825 command @code{OUTPUT_FORMAT} can also specify the output format, but
1826 this option overrides it. @xref{BFD}.
1829 @kindex --out-implib
1830 @item --out-implib @var{file}
1831 Create an import library in @var{file} corresponding to the executable
1832 the linker is generating (eg. a DLL or ELF program). This import
1833 library (which should be called @code{*.dll.a} or @code{*.a} for DLLs)
1834 may be used to link clients against the generated executable; this
1835 behaviour makes it possible to skip a separate import library creation
1836 step (eg. @code{dlltool} for DLLs). This option is only available for
1837 the i386 PE and ELF targetted ports of the linker.
1840 @kindex --pic-executable
1842 @itemx --pic-executable
1843 @cindex position independent executables
1844 Create a position independent executable. This is currently only supported on
1845 ELF platforms. Position independent executables are similar to shared
1846 libraries in that they are relocated by the dynamic linker to the virtual
1847 address the OS chooses for them (which can vary between invocations). Like
1848 normal dynamically linked executables they can be executed and symbols
1849 defined in the executable cannot be overridden by shared libraries.
1853 This option is ignored for Linux compatibility.
1857 This option is ignored for SVR4 compatibility.
1860 @cindex synthesizing linker
1861 @cindex relaxing addressing modes
1865 An option with machine dependent effects.
1867 This option is only supported on a few targets.
1870 @xref{H8/300,,@command{ld} and the H8/300}.
1873 @xref{Xtensa,, @command{ld} and Xtensa Processors}.
1876 @xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1879 @xref{Nios II,,@command{ld} and the Altera Nios II}.
1882 @xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1885 On some platforms the @samp{--relax} option performs target specific,
1886 global optimizations that become possible when the linker resolves
1887 addressing in the program, such as relaxing address modes,
1888 synthesizing new instructions, selecting shorter version of current
1889 instructions, and combining constant values.
1891 On some platforms these link time global optimizations may make symbolic
1892 debugging of the resulting executable impossible.
1894 This is known to be the case for the Matsushita MN10200 and MN10300
1895 family of processors.
1899 On platforms where this is not supported, @samp{--relax} is accepted,
1903 On platforms where @samp{--relax} is accepted the option
1904 @samp{--no-relax} can be used to disable the feature.
1906 @cindex retaining specified symbols
1907 @cindex stripping all but some symbols
1908 @cindex symbols, retaining selectively
1909 @kindex --retain-symbols-file=@var{filename}
1910 @item --retain-symbols-file=@var{filename}
1911 Retain @emph{only} the symbols listed in the file @var{filename},
1912 discarding all others. @var{filename} is simply a flat file, with one
1913 symbol name per line. This option is especially useful in environments
1917 where a large global symbol table is accumulated gradually, to conserve
1920 @samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1921 or symbols needed for relocations.
1923 You may only specify @samp{--retain-symbols-file} once in the command
1924 line. It overrides @samp{-s} and @samp{-S}.
1927 @item -rpath=@var{dir}
1928 @cindex runtime library search path
1929 @kindex -rpath=@var{dir}
1930 Add a directory to the runtime library search path. This is used when
1931 linking an ELF executable with shared objects. All @option{-rpath}
1932 arguments are concatenated and passed to the runtime linker, which uses
1933 them to locate shared objects at runtime. The @option{-rpath} option is
1934 also used when locating shared objects which are needed by shared
1935 objects explicitly included in the link; see the description of the
1936 @option{-rpath-link} option. If @option{-rpath} is not used when linking an
1937 ELF executable, the contents of the environment variable
1938 @code{LD_RUN_PATH} will be used if it is defined.
1940 The @option{-rpath} option may also be used on SunOS. By default, on
1941 SunOS, the linker will form a runtime search path out of all the
1942 @option{-L} options it is given. If a @option{-rpath} option is used, the
1943 runtime search path will be formed exclusively using the @option{-rpath}
1944 options, ignoring the @option{-L} options. This can be useful when using
1945 gcc, which adds many @option{-L} options which may be on NFS mounted
1948 For compatibility with other ELF linkers, if the @option{-R} option is
1949 followed by a directory name, rather than a file name, it is treated as
1950 the @option{-rpath} option.
1954 @cindex link-time runtime library search path
1955 @kindex -rpath-link=@var{dir}
1956 @item -rpath-link=@var{dir}
1957 When using ELF or SunOS, one shared library may require another. This
1958 happens when an @code{ld -shared} link includes a shared library as one
1961 When the linker encounters such a dependency when doing a non-shared,
1962 non-relocatable link, it will automatically try to locate the required
1963 shared library and include it in the link, if it is not included
1964 explicitly. In such a case, the @option{-rpath-link} option
1965 specifies the first set of directories to search. The
1966 @option{-rpath-link} option may specify a sequence of directory names
1967 either by specifying a list of names separated by colons, or by
1968 appearing multiple times.
1970 The tokens @var{$ORIGIN} and @var{$LIB} can appear in these search
1971 directories. They will be replaced by the full path to the directory
1972 containing the program or shared object in the case of @var{$ORIGIN}
1973 and either @samp{lib} - for 32-bit binaries - or @samp{lib64} - for
1974 64-bit binaries - in the case of @var{$LIB}.
1976 The alternative form of these tokens - @var{$@{ORIGIN@}} and
1977 @var{$@{LIB@}} can also be used. The token @var{$PLATFORM} is not
1980 This option should be used with caution as it overrides the search path
1981 that may have been hard compiled into a shared library. In such a case it
1982 is possible to use unintentionally a different search path than the
1983 runtime linker would do.
1985 The linker uses the following search paths to locate required shared
1989 Any directories specified by @option{-rpath-link} options.
1991 Any directories specified by @option{-rpath} options. The difference
1992 between @option{-rpath} and @option{-rpath-link} is that directories
1993 specified by @option{-rpath} options are included in the executable and
1994 used at runtime, whereas the @option{-rpath-link} option is only effective
1995 at link time. Searching @option{-rpath} in this way is only supported
1996 by native linkers and cross linkers which have been configured with
1997 the @option{--with-sysroot} option.
1999 On an ELF system, for native linkers, if the @option{-rpath} and
2000 @option{-rpath-link} options were not used, search the contents of the
2001 environment variable @code{LD_RUN_PATH}.
2003 On SunOS, if the @option{-rpath} option was not used, search any
2004 directories specified using @option{-L} options.
2006 For a native linker, search the contents of the environment
2007 variable @code{LD_LIBRARY_PATH}.
2009 For a native ELF linker, the directories in @code{DT_RUNPATH} or
2010 @code{DT_RPATH} of a shared library are searched for shared
2011 libraries needed by it. The @code{DT_RPATH} entries are ignored if
2012 @code{DT_RUNPATH} entries exist.
2014 The default directories, normally @file{/lib} and @file{/usr/lib}.
2016 For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
2017 exists, the list of directories found in that file.
2020 If the required shared library is not found, the linker will issue a
2021 warning and continue with the link.
2028 @cindex shared libraries
2029 Create a shared library. This is currently only supported on ELF, XCOFF
2030 and SunOS platforms. On SunOS, the linker will automatically create a
2031 shared library if the @option{-e} option is not used and there are
2032 undefined symbols in the link.
2034 @kindex --sort-common
2036 @itemx --sort-common=ascending
2037 @itemx --sort-common=descending
2038 This option tells @command{ld} to sort the common symbols by alignment in
2039 ascending or descending order when it places them in the appropriate output
2040 sections. The symbol alignments considered are sixteen-byte or larger,
2041 eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
2042 between symbols due to alignment constraints. If no sorting order is
2043 specified, then descending order is assumed.
2045 @kindex --sort-section=name
2046 @item --sort-section=name
2047 This option will apply @code{SORT_BY_NAME} to all wildcard section
2048 patterns in the linker script.
2050 @kindex --sort-section=alignment
2051 @item --sort-section=alignment
2052 This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
2053 patterns in the linker script.
2055 @kindex --spare-dynamic-tags
2056 @item --spare-dynamic-tags=@var{count}
2057 This option specifies the number of empty slots to leave in the
2058 .dynamic section of ELF shared objects. Empty slots may be needed by
2059 post processing tools, such as the prelinker. The default is 5.
2061 @kindex --split-by-file
2062 @item --split-by-file[=@var{size}]
2063 Similar to @option{--split-by-reloc} but creates a new output section for
2064 each input file when @var{size} is reached. @var{size} defaults to a
2065 size of 1 if not given.
2067 @kindex --split-by-reloc
2068 @item --split-by-reloc[=@var{count}]
2069 Tries to creates extra sections in the output file so that no single
2070 output section in the file contains more than @var{count} relocations.
2071 This is useful when generating huge relocatable files for downloading into
2072 certain real time kernels with the COFF object file format; since COFF
2073 cannot represent more than 65535 relocations in a single section. Note
2074 that this will fail to work with object file formats which do not
2075 support arbitrary sections. The linker will not split up individual
2076 input sections for redistribution, so if a single input section contains
2077 more than @var{count} relocations one output section will contain that
2078 many relocations. @var{count} defaults to a value of 32768.
2082 Compute and display statistics about the operation of the linker, such
2083 as execution time and memory usage.
2085 @kindex --sysroot=@var{directory}
2086 @item --sysroot=@var{directory}
2087 Use @var{directory} as the location of the sysroot, overriding the
2088 configure-time default. This option is only supported by linkers
2089 that were configured using @option{--with-sysroot}.
2093 This is used by COFF/PE based targets to create a task-linked object
2094 file where all of the global symbols have been converted to statics.
2096 @kindex --traditional-format
2097 @cindex traditional format
2098 @item --traditional-format
2099 For some targets, the output of @command{ld} is different in some ways from
2100 the output of some existing linker. This switch requests @command{ld} to
2101 use the traditional format instead.
2104 For example, on SunOS, @command{ld} combines duplicate entries in the
2105 symbol string table. This can reduce the size of an output file with
2106 full debugging information by over 30 percent. Unfortunately, the SunOS
2107 @code{dbx} program can not read the resulting program (@code{gdb} has no
2108 trouble). The @samp{--traditional-format} switch tells @command{ld} to not
2109 combine duplicate entries.
2111 @kindex --section-start=@var{sectionname}=@var{org}
2112 @item --section-start=@var{sectionname}=@var{org}
2113 Locate a section in the output file at the absolute
2114 address given by @var{org}. You may use this option as many
2115 times as necessary to locate multiple sections in the command
2117 @var{org} must be a single hexadecimal integer;
2118 for compatibility with other linkers, you may omit the leading
2119 @samp{0x} usually associated with hexadecimal values. @emph{Note:} there
2120 should be no white space between @var{sectionname}, the equals
2121 sign (``@key{=}''), and @var{org}.
2123 @kindex -Tbss=@var{org}
2124 @kindex -Tdata=@var{org}
2125 @kindex -Ttext=@var{org}
2126 @cindex segment origins, cmd line
2127 @item -Tbss=@var{org}
2128 @itemx -Tdata=@var{org}
2129 @itemx -Ttext=@var{org}
2130 Same as @option{--section-start}, with @code{.bss}, @code{.data} or
2131 @code{.text} as the @var{sectionname}.
2133 @kindex -Ttext-segment=@var{org}
2134 @item -Ttext-segment=@var{org}
2135 @cindex text segment origin, cmd line
2136 When creating an ELF executable, it will set the address of the first
2137 byte of the text segment.
2139 @kindex -Trodata-segment=@var{org}
2140 @item -Trodata-segment=@var{org}
2141 @cindex rodata segment origin, cmd line
2142 When creating an ELF executable or shared object for a target where
2143 the read-only data is in its own segment separate from the executable
2144 text, it will set the address of the first byte of the read-only data segment.
2146 @kindex -Tldata-segment=@var{org}
2147 @item -Tldata-segment=@var{org}
2148 @cindex ldata segment origin, cmd line
2149 When creating an ELF executable or shared object for x86-64 medium memory
2150 model, it will set the address of the first byte of the ldata segment.
2152 @kindex --unresolved-symbols
2153 @item --unresolved-symbols=@var{method}
2154 Determine how to handle unresolved symbols. There are four possible
2155 values for @samp{method}:
2159 Do not report any unresolved symbols.
2162 Report all unresolved symbols. This is the default.
2164 @item ignore-in-object-files
2165 Report unresolved symbols that are contained in shared libraries, but
2166 ignore them if they come from regular object files.
2168 @item ignore-in-shared-libs
2169 Report unresolved symbols that come from regular object files, but
2170 ignore them if they come from shared libraries. This can be useful
2171 when creating a dynamic binary and it is known that all the shared
2172 libraries that it should be referencing are included on the linker's
2176 The behaviour for shared libraries on their own can also be controlled
2177 by the @option{--[no-]allow-shlib-undefined} option.
2179 Normally the linker will generate an error message for each reported
2180 unresolved symbol but the option @option{--warn-unresolved-symbols}
2181 can change this to a warning.
2183 @kindex --verbose[=@var{NUMBER}]
2184 @cindex verbose[=@var{NUMBER}]
2186 @itemx --verbose[=@var{NUMBER}]
2187 Display the version number for @command{ld} and list the linker emulations
2188 supported. Display which input files can and cannot be opened. Display
2189 the linker script being used by the linker. If the optional @var{NUMBER}
2190 argument > 1, plugin symbol status will also be displayed.
2192 @kindex --version-script=@var{version-scriptfile}
2193 @cindex version script, symbol versions
2194 @item --version-script=@var{version-scriptfile}
2195 Specify the name of a version script to the linker. This is typically
2196 used when creating shared libraries to specify additional information
2197 about the version hierarchy for the library being created. This option
2198 is only fully supported on ELF platforms which support shared libraries;
2199 see @ref{VERSION}. It is partially supported on PE platforms, which can
2200 use version scripts to filter symbol visibility in auto-export mode: any
2201 symbols marked @samp{local} in the version script will not be exported.
2204 @kindex --warn-common
2205 @cindex warnings, on combining symbols
2206 @cindex combining symbols, warnings on
2208 Warn when a common symbol is combined with another common symbol or with
2209 a symbol definition. Unix linkers allow this somewhat sloppy practice,
2210 but linkers on some other operating systems do not. This option allows
2211 you to find potential problems from combining global symbols.
2212 Unfortunately, some C libraries use this practice, so you may get some
2213 warnings about symbols in the libraries as well as in your programs.
2215 There are three kinds of global symbols, illustrated here by C examples:
2219 A definition, which goes in the initialized data section of the output
2223 An undefined reference, which does not allocate space.
2224 There must be either a definition or a common symbol for the
2228 A common symbol. If there are only (one or more) common symbols for a
2229 variable, it goes in the uninitialized data area of the output file.
2230 The linker merges multiple common symbols for the same variable into a
2231 single symbol. If they are of different sizes, it picks the largest
2232 size. The linker turns a common symbol into a declaration, if there is
2233 a definition of the same variable.
2236 The @samp{--warn-common} option can produce five kinds of warnings.
2237 Each warning consists of a pair of lines: the first describes the symbol
2238 just encountered, and the second describes the previous symbol
2239 encountered with the same name. One or both of the two symbols will be
2244 Turning a common symbol into a reference, because there is already a
2245 definition for the symbol.
2247 @var{file}(@var{section}): warning: common of `@var{symbol}'
2248 overridden by definition
2249 @var{file}(@var{section}): warning: defined here
2253 Turning a common symbol into a reference, because a later definition for
2254 the symbol is encountered. This is the same as the previous case,
2255 except that the symbols are encountered in a different order.
2257 @var{file}(@var{section}): warning: definition of `@var{symbol}'
2259 @var{file}(@var{section}): warning: common is here
2263 Merging a common symbol with a previous same-sized common symbol.
2265 @var{file}(@var{section}): warning: multiple common
2267 @var{file}(@var{section}): warning: previous common is here
2271 Merging a common symbol with a previous larger common symbol.
2273 @var{file}(@var{section}): warning: common of `@var{symbol}'
2274 overridden by larger common
2275 @var{file}(@var{section}): warning: larger common is here
2279 Merging a common symbol with a previous smaller common symbol. This is
2280 the same as the previous case, except that the symbols are
2281 encountered in a different order.
2283 @var{file}(@var{section}): warning: common of `@var{symbol}'
2284 overriding smaller common
2285 @var{file}(@var{section}): warning: smaller common is here
2289 @kindex --warn-constructors
2290 @item --warn-constructors
2291 Warn if any global constructors are used. This is only useful for a few
2292 object file formats. For formats like COFF or ELF, the linker can not
2293 detect the use of global constructors.
2295 @kindex --warn-multiple-gp
2296 @item --warn-multiple-gp
2297 Warn if multiple global pointer values are required in the output file.
2298 This is only meaningful for certain processors, such as the Alpha.
2299 Specifically, some processors put large-valued constants in a special
2300 section. A special register (the global pointer) points into the middle
2301 of this section, so that constants can be loaded efficiently via a
2302 base-register relative addressing mode. Since the offset in
2303 base-register relative mode is fixed and relatively small (e.g., 16
2304 bits), this limits the maximum size of the constant pool. Thus, in
2305 large programs, it is often necessary to use multiple global pointer
2306 values in order to be able to address all possible constants. This
2307 option causes a warning to be issued whenever this case occurs.
2310 @cindex warnings, on undefined symbols
2311 @cindex undefined symbols, warnings on
2313 Only warn once for each undefined symbol, rather than once per module
2316 @kindex --warn-section-align
2317 @cindex warnings, on section alignment
2318 @cindex section alignment, warnings on
2319 @item --warn-section-align
2320 Warn if the address of an output section is changed because of
2321 alignment. Typically, the alignment will be set by an input section.
2322 The address will only be changed if it not explicitly specified; that
2323 is, if the @code{SECTIONS} command does not specify a start address for
2324 the section (@pxref{SECTIONS}).
2326 @kindex --warn-shared-textrel
2327 @item --warn-shared-textrel
2328 Warn if the linker adds a DT_TEXTREL to a shared object.
2330 @kindex --warn-alternate-em
2331 @item --warn-alternate-em
2332 Warn if an object has alternate ELF machine code.
2334 @kindex --warn-unresolved-symbols
2335 @item --warn-unresolved-symbols
2336 If the linker is going to report an unresolved symbol (see the option
2337 @option{--unresolved-symbols}) it will normally generate an error.
2338 This option makes it generate a warning instead.
2340 @kindex --error-unresolved-symbols
2341 @item --error-unresolved-symbols
2342 This restores the linker's default behaviour of generating errors when
2343 it is reporting unresolved symbols.
2345 @kindex --whole-archive
2346 @cindex including an entire archive
2347 @item --whole-archive
2348 For each archive mentioned on the command line after the
2349 @option{--whole-archive} option, include every object file in the archive
2350 in the link, rather than searching the archive for the required object
2351 files. This is normally used to turn an archive file into a shared
2352 library, forcing every object to be included in the resulting shared
2353 library. This option may be used more than once.
2355 Two notes when using this option from gcc: First, gcc doesn't know
2356 about this option, so you have to use @option{-Wl,-whole-archive}.
2357 Second, don't forget to use @option{-Wl,-no-whole-archive} after your
2358 list of archives, because gcc will add its own list of archives to
2359 your link and you may not want this flag to affect those as well.
2361 @kindex --wrap=@var{symbol}
2362 @item --wrap=@var{symbol}
2363 Use a wrapper function for @var{symbol}. Any undefined reference to
2364 @var{symbol} will be resolved to @code{__wrap_@var{symbol}}. Any
2365 undefined reference to @code{__real_@var{symbol}} will be resolved to
2368 This can be used to provide a wrapper for a system function. The
2369 wrapper function should be called @code{__wrap_@var{symbol}}. If it
2370 wishes to call the system function, it should call
2371 @code{__real_@var{symbol}}.
2373 Here is a trivial example:
2377 __wrap_malloc (size_t c)
2379 printf ("malloc called with %zu\n", c);
2380 return __real_malloc (c);
2384 If you link other code with this file using @option{--wrap malloc}, then
2385 all calls to @code{malloc} will call the function @code{__wrap_malloc}
2386 instead. The call to @code{__real_malloc} in @code{__wrap_malloc} will
2387 call the real @code{malloc} function.
2389 You may wish to provide a @code{__real_malloc} function as well, so that
2390 links without the @option{--wrap} option will succeed. If you do this,
2391 you should not put the definition of @code{__real_malloc} in the same
2392 file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2393 call before the linker has a chance to wrap it to @code{malloc}.
2395 Only undefined references are replaced by the linker. So, translation unit
2396 internal references to @var{symbol} are not resolved to
2397 @code{__wrap_@var{symbol}}. In the next example, the call to @code{f} in
2398 @code{g} is not resolved to @code{__wrap_f}.
2414 @kindex --eh-frame-hdr
2415 @kindex --no-eh-frame-hdr
2416 @item --eh-frame-hdr
2417 @itemx --no-eh-frame-hdr
2418 Request (@option{--eh-frame-hdr}) or suppress
2419 (@option{--no-eh-frame-hdr}) the creation of @code{.eh_frame_hdr}
2420 section and ELF @code{PT_GNU_EH_FRAME} segment header.
2422 @kindex --ld-generated-unwind-info
2423 @item --no-ld-generated-unwind-info
2424 Request creation of @code{.eh_frame} unwind info for linker
2425 generated code sections like PLT. This option is on by default
2426 if linker generated unwind info is supported.
2428 @kindex --enable-new-dtags
2429 @kindex --disable-new-dtags
2430 @item --enable-new-dtags
2431 @itemx --disable-new-dtags
2432 This linker can create the new dynamic tags in ELF. But the older ELF
2433 systems may not understand them. If you specify
2434 @option{--enable-new-dtags}, the new dynamic tags will be created as needed
2435 and older dynamic tags will be omitted.
2436 If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2437 created. By default, the new dynamic tags are not created. Note that
2438 those options are only available for ELF systems.
2440 @kindex --hash-size=@var{number}
2441 @item --hash-size=@var{number}
2442 Set the default size of the linker's hash tables to a prime number
2443 close to @var{number}. Increasing this value can reduce the length of
2444 time it takes the linker to perform its tasks, at the expense of
2445 increasing the linker's memory requirements. Similarly reducing this
2446 value can reduce the memory requirements at the expense of speed.
2448 @kindex --hash-style=@var{style}
2449 @item --hash-style=@var{style}
2450 Set the type of linker's hash table(s). @var{style} can be either
2451 @code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2452 new style GNU @code{.gnu.hash} section or @code{both} for both
2453 the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2454 hash tables. The default depends upon how the linker was configured,
2455 but for most Linux based systems it will be @code{both}.
2457 @kindex --compress-debug-sections=none
2458 @kindex --compress-debug-sections=zlib
2459 @kindex --compress-debug-sections=zlib-gnu
2460 @kindex --compress-debug-sections=zlib-gabi
2461 @item --compress-debug-sections=none
2462 @itemx --compress-debug-sections=zlib
2463 @itemx --compress-debug-sections=zlib-gnu
2464 @itemx --compress-debug-sections=zlib-gabi
2465 On ELF platforms, these options control how DWARF debug sections are
2466 compressed using zlib.
2468 @option{--compress-debug-sections=none} doesn't compress DWARF debug
2469 sections. @option{--compress-debug-sections=zlib-gnu} compresses
2470 DWARF debug sections and renames them to begin with @samp{.zdebug}
2471 instead of @samp{.debug}. @option{--compress-debug-sections=zlib-gabi}
2472 also compresses DWARF debug sections, but rather than renaming them it
2473 sets the SHF_COMPRESSED flag in the sections' headers.
2475 The @option{--compress-debug-sections=zlib} option is an alias for
2476 @option{--compress-debug-sections=zlib-gabi}.
2478 Note that this option overrides any compression in input debug
2479 sections, so if a binary is linked with @option{--compress-debug-sections=none}
2480 for example, then any compressed debug sections in input files will be
2481 uncompressed before they are copied into the output binary.
2483 The default compression behaviour varies depending upon the target
2484 involved and the configure options used to build the toolchain. The
2485 default can be determined by examining the output from the linker's
2486 @option{--help} option.
2488 @kindex --reduce-memory-overheads
2489 @item --reduce-memory-overheads
2490 This option reduces memory requirements at ld runtime, at the expense of
2491 linking speed. This was introduced to select the old O(n^2) algorithm
2492 for link map file generation, rather than the new O(n) algorithm which uses
2493 about 40% more memory for symbol storage.
2495 Another effect of the switch is to set the default hash table size to
2496 1021, which again saves memory at the cost of lengthening the linker's
2497 run time. This is not done however if the @option{--hash-size} switch
2500 The @option{--reduce-memory-overheads} switch may be also be used to
2501 enable other tradeoffs in future versions of the linker.
2504 @kindex --build-id=@var{style}
2506 @itemx --build-id=@var{style}
2507 Request the creation of a @code{.note.gnu.build-id} ELF note section
2508 or a @code{.buildid} COFF section. The contents of the note are
2509 unique bits identifying this linked file. @var{style} can be
2510 @code{uuid} to use 128 random bits, @code{sha1} to use a 160-bit
2511 @sc{SHA1} hash on the normative parts of the output contents,
2512 @code{md5} to use a 128-bit @sc{MD5} hash on the normative parts of
2513 the output contents, or @code{0x@var{hexstring}} to use a chosen bit
2514 string specified as an even number of hexadecimal digits (@code{-} and
2515 @code{:} characters between digit pairs are ignored). If @var{style}
2516 is omitted, @code{sha1} is used.
2518 The @code{md5} and @code{sha1} styles produces an identifier
2519 that is always the same in an identical output file, but will be
2520 unique among all nonidentical output files. It is not intended
2521 to be compared as a checksum for the file's contents. A linked
2522 file may be changed later by other tools, but the build ID bit
2523 string identifying the original linked file does not change.
2525 Passing @code{none} for @var{style} disables the setting from any
2526 @code{--build-id} options earlier on the command line.
2531 @subsection Options Specific to i386 PE Targets
2533 @c man begin OPTIONS
2535 The i386 PE linker supports the @option{-shared} option, which causes
2536 the output to be a dynamically linked library (DLL) instead of a
2537 normal executable. You should name the output @code{*.dll} when you
2538 use this option. In addition, the linker fully supports the standard
2539 @code{*.def} files, which may be specified on the linker command line
2540 like an object file (in fact, it should precede archives it exports
2541 symbols from, to ensure that they get linked in, just like a normal
2544 In addition to the options common to all targets, the i386 PE linker
2545 support additional command-line options that are specific to the i386
2546 PE target. Options that take values may be separated from their
2547 values by either a space or an equals sign.
2551 @kindex --add-stdcall-alias
2552 @item --add-stdcall-alias
2553 If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2554 as-is and also with the suffix stripped.
2555 [This option is specific to the i386 PE targeted port of the linker]
2558 @item --base-file @var{file}
2559 Use @var{file} as the name of a file in which to save the base
2560 addresses of all the relocations needed for generating DLLs with
2562 [This is an i386 PE specific option]
2566 Create a DLL instead of a regular executable. You may also use
2567 @option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2569 [This option is specific to the i386 PE targeted port of the linker]
2571 @kindex --enable-long-section-names
2572 @kindex --disable-long-section-names
2573 @item --enable-long-section-names
2574 @itemx --disable-long-section-names
2575 The PE variants of the COFF object format add an extension that permits
2576 the use of section names longer than eight characters, the normal limit
2577 for COFF. By default, these names are only allowed in object files, as
2578 fully-linked executable images do not carry the COFF string table required
2579 to support the longer names. As a GNU extension, it is possible to
2580 allow their use in executable images as well, or to (probably pointlessly!)
2581 disallow it in object files, by using these two options. Executable images
2582 generated with these long section names are slightly non-standard, carrying
2583 as they do a string table, and may generate confusing output when examined
2584 with non-GNU PE-aware tools, such as file viewers and dumpers. However,
2585 GDB relies on the use of PE long section names to find Dwarf-2 debug
2586 information sections in an executable image at runtime, and so if neither
2587 option is specified on the command-line, @command{ld} will enable long
2588 section names, overriding the default and technically correct behaviour,
2589 when it finds the presence of debug information while linking an executable
2590 image and not stripping symbols.
2591 [This option is valid for all PE targeted ports of the linker]
2593 @kindex --enable-stdcall-fixup
2594 @kindex --disable-stdcall-fixup
2595 @item --enable-stdcall-fixup
2596 @itemx --disable-stdcall-fixup
2597 If the link finds a symbol that it cannot resolve, it will attempt to
2598 do ``fuzzy linking'' by looking for another defined symbol that differs
2599 only in the format of the symbol name (cdecl vs stdcall) and will
2600 resolve that symbol by linking to the match. For example, the
2601 undefined symbol @code{_foo} might be linked to the function
2602 @code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2603 to the function @code{_bar}. When the linker does this, it prints a
2604 warning, since it normally should have failed to link, but sometimes
2605 import libraries generated from third-party dlls may need this feature
2606 to be usable. If you specify @option{--enable-stdcall-fixup}, this
2607 feature is fully enabled and warnings are not printed. If you specify
2608 @option{--disable-stdcall-fixup}, this feature is disabled and such
2609 mismatches are considered to be errors.
2610 [This option is specific to the i386 PE targeted port of the linker]
2612 @kindex --leading-underscore
2613 @kindex --no-leading-underscore
2614 @item --leading-underscore
2615 @itemx --no-leading-underscore
2616 For most targets default symbol-prefix is an underscore and is defined
2617 in target's description. By this option it is possible to
2618 disable/enable the default underscore symbol-prefix.
2620 @cindex DLLs, creating
2621 @kindex --export-all-symbols
2622 @item --export-all-symbols
2623 If given, all global symbols in the objects used to build a DLL will
2624 be exported by the DLL. Note that this is the default if there
2625 otherwise wouldn't be any exported symbols. When symbols are
2626 explicitly exported via DEF files or implicitly exported via function
2627 attributes, the default is to not export anything else unless this
2628 option is given. Note that the symbols @code{DllMain@@12},
2629 @code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2630 @code{impure_ptr} will not be automatically
2631 exported. Also, symbols imported from other DLLs will not be
2632 re-exported, nor will symbols specifying the DLL's internal layout
2633 such as those beginning with @code{_head_} or ending with
2634 @code{_iname}. In addition, no symbols from @code{libgcc},
2635 @code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2636 Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2637 not be exported, to help with C++ DLLs. Finally, there is an
2638 extensive list of cygwin-private symbols that are not exported
2639 (obviously, this applies on when building DLLs for cygwin targets).
2640 These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2641 @code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2642 @code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2643 @code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2644 @code{cygwin_premain3}, and @code{environ}.
2645 [This option is specific to the i386 PE targeted port of the linker]
2647 @kindex --exclude-symbols
2648 @item --exclude-symbols @var{symbol},@var{symbol},...
2649 Specifies a list of symbols which should not be automatically
2650 exported. The symbol names may be delimited by commas or colons.
2651 [This option is specific to the i386 PE targeted port of the linker]
2653 @kindex --exclude-all-symbols
2654 @item --exclude-all-symbols
2655 Specifies no symbols should be automatically exported.
2656 [This option is specific to the i386 PE targeted port of the linker]
2658 @kindex --file-alignment
2659 @item --file-alignment
2660 Specify the file alignment. Sections in the file will always begin at
2661 file offsets which are multiples of this number. This defaults to
2663 [This option is specific to the i386 PE targeted port of the linker]
2667 @item --heap @var{reserve}
2668 @itemx --heap @var{reserve},@var{commit}
2669 Specify the number of bytes of memory to reserve (and optionally commit)
2670 to be used as heap for this program. The default is 1MB reserved, 4K
2672 [This option is specific to the i386 PE targeted port of the linker]
2675 @kindex --image-base
2676 @item --image-base @var{value}
2677 Use @var{value} as the base address of your program or dll. This is
2678 the lowest memory location that will be used when your program or dll
2679 is loaded. To reduce the need to relocate and improve performance of
2680 your dlls, each should have a unique base address and not overlap any
2681 other dlls. The default is 0x400000 for executables, and 0x10000000
2683 [This option is specific to the i386 PE targeted port of the linker]
2687 If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2688 symbols before they are exported.
2689 [This option is specific to the i386 PE targeted port of the linker]
2691 @kindex --large-address-aware
2692 @item --large-address-aware
2693 If given, the appropriate bit in the ``Characteristics'' field of the COFF
2694 header is set to indicate that this executable supports virtual addresses
2695 greater than 2 gigabytes. This should be used in conjunction with the /3GB
2696 or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2697 section of the BOOT.INI. Otherwise, this bit has no effect.
2698 [This option is specific to PE targeted ports of the linker]
2700 @kindex --disable-large-address-aware
2701 @item --disable-large-address-aware
2702 Reverts the effect of a previous @samp{--large-address-aware} option.
2703 This is useful if @samp{--large-address-aware} is always set by the compiler
2704 driver (e.g. Cygwin gcc) and the executable does not support virtual
2705 addresses greater than 2 gigabytes.
2706 [This option is specific to PE targeted ports of the linker]
2708 @kindex --major-image-version
2709 @item --major-image-version @var{value}
2710 Sets the major number of the ``image version''. Defaults to 1.
2711 [This option is specific to the i386 PE targeted port of the linker]
2713 @kindex --major-os-version
2714 @item --major-os-version @var{value}
2715 Sets the major number of the ``os version''. Defaults to 4.
2716 [This option is specific to the i386 PE targeted port of the linker]
2718 @kindex --major-subsystem-version
2719 @item --major-subsystem-version @var{value}
2720 Sets the major number of the ``subsystem version''. Defaults to 4.
2721 [This option is specific to the i386 PE targeted port of the linker]
2723 @kindex --minor-image-version
2724 @item --minor-image-version @var{value}
2725 Sets the minor number of the ``image version''. Defaults to 0.
2726 [This option is specific to the i386 PE targeted port of the linker]
2728 @kindex --minor-os-version
2729 @item --minor-os-version @var{value}
2730 Sets the minor number of the ``os version''. Defaults to 0.
2731 [This option is specific to the i386 PE targeted port of the linker]
2733 @kindex --minor-subsystem-version
2734 @item --minor-subsystem-version @var{value}
2735 Sets the minor number of the ``subsystem version''. Defaults to 0.
2736 [This option is specific to the i386 PE targeted port of the linker]
2738 @cindex DEF files, creating
2739 @cindex DLLs, creating
2740 @kindex --output-def
2741 @item --output-def @var{file}
2742 The linker will create the file @var{file} which will contain a DEF
2743 file corresponding to the DLL the linker is generating. This DEF file
2744 (which should be called @code{*.def}) may be used to create an import
2745 library with @code{dlltool} or may be used as a reference to
2746 automatically or implicitly exported symbols.
2747 [This option is specific to the i386 PE targeted port of the linker]
2749 @cindex DLLs, creating
2750 @kindex --enable-auto-image-base
2751 @item --enable-auto-image-base
2752 @itemx --enable-auto-image-base=@var{value}
2753 Automatically choose the image base for DLLs, optionally starting with base
2754 @var{value}, unless one is specified using the @code{--image-base} argument.
2755 By using a hash generated from the dllname to create unique image bases
2756 for each DLL, in-memory collisions and relocations which can delay program
2757 execution are avoided.
2758 [This option is specific to the i386 PE targeted port of the linker]
2760 @kindex --disable-auto-image-base
2761 @item --disable-auto-image-base
2762 Do not automatically generate a unique image base. If there is no
2763 user-specified image base (@code{--image-base}) then use the platform
2765 [This option is specific to the i386 PE targeted port of the linker]
2767 @cindex DLLs, linking to
2768 @kindex --dll-search-prefix
2769 @item --dll-search-prefix @var{string}
2770 When linking dynamically to a dll without an import library,
2771 search for @code{<string><basename>.dll} in preference to
2772 @code{lib<basename>.dll}. This behaviour allows easy distinction
2773 between DLLs built for the various "subplatforms": native, cygwin,
2774 uwin, pw, etc. For instance, cygwin DLLs typically use
2775 @code{--dll-search-prefix=cyg}.
2776 [This option is specific to the i386 PE targeted port of the linker]
2778 @kindex --enable-auto-import
2779 @item --enable-auto-import
2780 Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2781 DATA imports from DLLs, thus making it possible to bypass the dllimport
2782 mechanism on the user side and to reference unmangled symbol names.
2783 [This option is specific to the i386 PE targeted port of the linker]
2785 The following remarks pertain to the original implementation of the
2786 feature and are obsolete nowadays for Cygwin and MinGW targets.
2788 Note: Use of the 'auto-import' extension will cause the text section
2789 of the image file to be made writable. This does not conform to the
2790 PE-COFF format specification published by Microsoft.
2792 Note - use of the 'auto-import' extension will also cause read only
2793 data which would normally be placed into the .rdata section to be
2794 placed into the .data section instead. This is in order to work
2795 around a problem with consts that is described here:
2796 http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2798 Using 'auto-import' generally will 'just work' -- but sometimes you may
2801 "variable '<var>' can't be auto-imported. Please read the
2802 documentation for ld's @code{--enable-auto-import} for details."
2804 This message occurs when some (sub)expression accesses an address
2805 ultimately given by the sum of two constants (Win32 import tables only
2806 allow one). Instances where this may occur include accesses to member
2807 fields of struct variables imported from a DLL, as well as using a
2808 constant index into an array variable imported from a DLL. Any
2809 multiword variable (arrays, structs, long long, etc) may trigger
2810 this error condition. However, regardless of the exact data type
2811 of the offending exported variable, ld will always detect it, issue
2812 the warning, and exit.
2814 There are several ways to address this difficulty, regardless of the
2815 data type of the exported variable:
2817 One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2818 of adjusting references in your client code for runtime environment, so
2819 this method works only when runtime environment supports this feature.
2821 A second solution is to force one of the 'constants' to be a variable --
2822 that is, unknown and un-optimizable at compile time. For arrays,
2823 there are two possibilities: a) make the indexee (the array's address)
2824 a variable, or b) make the 'constant' index a variable. Thus:
2827 extern type extern_array[];
2829 @{ volatile type *t=extern_array; t[1] @}
2835 extern type extern_array[];
2837 @{ volatile int t=1; extern_array[t] @}
2840 For structs (and most other multiword data types) the only option
2841 is to make the struct itself (or the long long, or the ...) variable:
2844 extern struct s extern_struct;
2845 extern_struct.field -->
2846 @{ volatile struct s *t=&extern_struct; t->field @}
2852 extern long long extern_ll;
2854 @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2857 A third method of dealing with this difficulty is to abandon
2858 'auto-import' for the offending symbol and mark it with
2859 @code{__declspec(dllimport)}. However, in practice that
2860 requires using compile-time #defines to indicate whether you are
2861 building a DLL, building client code that will link to the DLL, or
2862 merely building/linking to a static library. In making the choice
2863 between the various methods of resolving the 'direct address with
2864 constant offset' problem, you should consider typical real-world usage:
2872 void main(int argc, char **argv)@{
2873 printf("%d\n",arr[1]);
2883 void main(int argc, char **argv)@{
2884 /* This workaround is for win32 and cygwin; do not "optimize" */
2885 volatile int *parr = arr;
2886 printf("%d\n",parr[1]);
2893 /* Note: auto-export is assumed (no __declspec(dllexport)) */
2894 #if (defined(_WIN32) || defined(__CYGWIN__)) && \
2895 !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2896 #define FOO_IMPORT __declspec(dllimport)
2900 extern FOO_IMPORT int arr[];
2903 void main(int argc, char **argv)@{
2904 printf("%d\n",arr[1]);
2908 A fourth way to avoid this problem is to re-code your
2909 library to use a functional interface rather than a data interface
2910 for the offending variables (e.g. set_foo() and get_foo() accessor
2913 @kindex --disable-auto-import
2914 @item --disable-auto-import
2915 Do not attempt to do sophisticated linking of @code{_symbol} to
2916 @code{__imp__symbol} for DATA imports from DLLs.
2917 [This option is specific to the i386 PE targeted port of the linker]
2919 @kindex --enable-runtime-pseudo-reloc
2920 @item --enable-runtime-pseudo-reloc
2921 If your code contains expressions described in --enable-auto-import section,
2922 that is, DATA imports from DLL with non-zero offset, this switch will create
2923 a vector of 'runtime pseudo relocations' which can be used by runtime
2924 environment to adjust references to such data in your client code.
2925 [This option is specific to the i386 PE targeted port of the linker]
2927 @kindex --disable-runtime-pseudo-reloc
2928 @item --disable-runtime-pseudo-reloc
2929 Do not create pseudo relocations for non-zero offset DATA imports from DLLs.
2930 [This option is specific to the i386 PE targeted port of the linker]
2932 @kindex --enable-extra-pe-debug
2933 @item --enable-extra-pe-debug
2934 Show additional debug info related to auto-import symbol thunking.
2935 [This option is specific to the i386 PE targeted port of the linker]
2937 @kindex --section-alignment
2938 @item --section-alignment
2939 Sets the section alignment. Sections in memory will always begin at
2940 addresses which are a multiple of this number. Defaults to 0x1000.
2941 [This option is specific to the i386 PE targeted port of the linker]
2945 @item --stack @var{reserve}
2946 @itemx --stack @var{reserve},@var{commit}
2947 Specify the number of bytes of memory to reserve (and optionally commit)
2948 to be used as stack for this program. The default is 2MB reserved, 4K
2950 [This option is specific to the i386 PE targeted port of the linker]
2953 @item --subsystem @var{which}
2954 @itemx --subsystem @var{which}:@var{major}
2955 @itemx --subsystem @var{which}:@var{major}.@var{minor}
2956 Specifies the subsystem under which your program will execute. The
2957 legal values for @var{which} are @code{native}, @code{windows},
2958 @code{console}, @code{posix}, and @code{xbox}. You may optionally set
2959 the subsystem version also. Numeric values are also accepted for
2961 [This option is specific to the i386 PE targeted port of the linker]
2963 The following options set flags in the @code{DllCharacteristics} field
2964 of the PE file header:
2965 [These options are specific to PE targeted ports of the linker]
2967 @kindex --high-entropy-va
2968 @item --high-entropy-va
2969 Image is compatible with 64-bit address space layout randomization
2972 @kindex --dynamicbase
2974 The image base address may be relocated using address space layout
2975 randomization (ASLR). This feature was introduced with MS Windows
2976 Vista for i386 PE targets.
2978 @kindex --forceinteg
2980 Code integrity checks are enforced.
2984 The image is compatible with the Data Execution Prevention.
2985 This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2987 @kindex --no-isolation
2988 @item --no-isolation
2989 Although the image understands isolation, do not isolate the image.
2993 The image does not use SEH. No SE handler may be called from
2998 Do not bind this image.
3002 The driver uses the MS Windows Driver Model.
3006 The image is Terminal Server aware.
3008 @kindex --insert-timestamp
3009 @item --insert-timestamp
3010 @itemx --no-insert-timestamp
3011 Insert a real timestamp into the image. This is the default behaviour
3012 as it matches legacy code and it means that the image will work with
3013 other, proprietary tools. The problem with this default is that it
3014 will result in slightly different images being produced each time the
3015 same sources are linked. The option @option{--no-insert-timestamp}
3016 can be used to insert a zero value for the timestamp, this ensuring
3017 that binaries produced from identical sources will compare
3024 @subsection Options specific to C6X uClinux targets
3026 @c man begin OPTIONS
3028 The C6X uClinux target uses a binary format called DSBT to support shared
3029 libraries. Each shared library in the system needs to have a unique index;
3030 all executables use an index of 0.
3035 @item --dsbt-size @var{size}
3036 This option sets the number of entries in the DSBT of the current executable
3037 or shared library to @var{size}. The default is to create a table with 64
3040 @kindex --dsbt-index
3041 @item --dsbt-index @var{index}
3042 This option sets the DSBT index of the current executable or shared library
3043 to @var{index}. The default is 0, which is appropriate for generating
3044 executables. If a shared library is generated with a DSBT index of 0, the
3045 @code{R_C6000_DSBT_INDEX} relocs are copied into the output file.
3047 @kindex --no-merge-exidx-entries
3048 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent
3049 exidx entries in frame unwind info.
3057 @subsection Options specific to C-SKY targets
3059 @c man begin OPTIONS
3063 @kindex --branch-stub on C-SKY
3065 This option enables linker branch relaxation by inserting branch stub
3066 sections when needed to extend the range of branches. This option is
3067 usually not required since C-SKY supports branch and call instructions that
3068 can access the full memory range and branch relaxation is normally handled by
3069 the compiler or assembler.
3071 @kindex --stub-group-size on C-SKY
3072 @item --stub-group-size=@var{N}
3073 This option allows finer control of linker branch stub creation.
3074 It sets the maximum size of a group of input sections that can
3075 be handled by one stub section. A negative value of @var{N} locates
3076 stub sections after their branches, while a positive value allows stub
3077 sections to appear either before or after the branches. Values of
3078 @samp{1} or @samp{-1} indicate that the
3079 linker should choose suitable defaults.
3087 @subsection Options specific to Motorola 68HC11 and 68HC12 targets
3089 @c man begin OPTIONS
3091 The 68HC11 and 68HC12 linkers support specific options to control the
3092 memory bank switching mapping and trampoline code generation.
3096 @kindex --no-trampoline
3097 @item --no-trampoline
3098 This option disables the generation of trampoline. By default a trampoline
3099 is generated for each far function which is called using a @code{jsr}
3100 instruction (this happens when a pointer to a far function is taken).
3102 @kindex --bank-window
3103 @item --bank-window @var{name}
3104 This option indicates to the linker the name of the memory region in
3105 the @samp{MEMORY} specification that describes the memory bank window.
3106 The definition of such region is then used by the linker to compute
3107 paging and addresses within the memory window.
3115 @subsection Options specific to Motorola 68K target
3117 @c man begin OPTIONS
3119 The following options are supported to control handling of GOT generation
3120 when linking for 68K targets.
3125 @item --got=@var{type}
3126 This option tells the linker which GOT generation scheme to use.
3127 @var{type} should be one of @samp{single}, @samp{negative},
3128 @samp{multigot} or @samp{target}. For more information refer to the
3129 Info entry for @file{ld}.
3137 @subsection Options specific to MIPS targets
3139 @c man begin OPTIONS
3141 The following options are supported to control microMIPS instruction
3142 generation and branch relocation checks for ISA mode transitions when
3143 linking for MIPS targets.
3151 These options control the choice of microMIPS instructions used in code
3152 generated by the linker, such as that in the PLT or lazy binding stubs,
3153 or in relaxation. If @samp{--insn32} is used, then the linker only uses
3154 32-bit instruction encodings. By default or if @samp{--no-insn32} is
3155 used, all instruction encodings are used, including 16-bit ones where
3158 @kindex --ignore-branch-isa
3159 @item --ignore-branch-isa
3160 @kindex --no-ignore-branch-isa
3161 @itemx --no-ignore-branch-isa
3162 These options control branch relocation checks for invalid ISA mode
3163 transitions. If @samp{--ignore-branch-isa} is used, then the linker
3164 accepts any branch relocations and any ISA mode transition required
3165 is lost in relocation calculation, except for some cases of @code{BAL}
3166 instructions which meet relaxation conditions and are converted to
3167 equivalent @code{JALX} instructions as the associated relocation is
3168 calculated. By default or if @samp{--no-ignore-branch-isa} is used
3169 a check is made causing the loss of an ISA mode transition to produce
3179 @section Environment Variables
3181 @c man begin ENVIRONMENT
3183 You can change the behaviour of @command{ld} with the environment variables
3184 @ifclear SingleFormat
3187 @code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
3189 @ifclear SingleFormat
3191 @cindex default input format
3192 @code{GNUTARGET} determines the input-file object format if you don't
3193 use @samp{-b} (or its synonym @samp{--format}). Its value should be one
3194 of the BFD names for an input format (@pxref{BFD}). If there is no
3195 @code{GNUTARGET} in the environment, @command{ld} uses the natural format
3196 of the target. If @code{GNUTARGET} is set to @code{default} then BFD
3197 attempts to discover the input format by examining binary input files;
3198 this method often succeeds, but there are potential ambiguities, since
3199 there is no method of ensuring that the magic number used to specify
3200 object-file formats is unique. However, the configuration procedure for
3201 BFD on each system places the conventional format for that system first
3202 in the search-list, so ambiguities are resolved in favor of convention.
3206 @cindex default emulation
3207 @cindex emulation, default
3208 @code{LDEMULATION} determines the default emulation if you don't use the
3209 @samp{-m} option. The emulation can affect various aspects of linker
3210 behaviour, particularly the default linker script. You can list the
3211 available emulations with the @samp{--verbose} or @samp{-V} options. If
3212 the @samp{-m} option is not used, and the @code{LDEMULATION} environment
3213 variable is not defined, the default emulation depends upon how the
3214 linker was configured.
3216 @kindex COLLECT_NO_DEMANGLE
3217 @cindex demangling, default
3218 Normally, the linker will default to demangling symbols. However, if
3219 @code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
3220 default to not demangling symbols. This environment variable is used in
3221 a similar fashion by the @code{gcc} linker wrapper program. The default
3222 may be overridden by the @samp{--demangle} and @samp{--no-demangle}
3229 @chapter Linker Scripts
3232 @cindex linker scripts
3233 @cindex command files
3234 Every link is controlled by a @dfn{linker script}. This script is
3235 written in the linker command language.
3237 The main purpose of the linker script is to describe how the sections in
3238 the input files should be mapped into the output file, and to control
3239 the memory layout of the output file. Most linker scripts do nothing
3240 more than this. However, when necessary, the linker script can also
3241 direct the linker to perform many other operations, using the commands
3244 The linker always uses a linker script. If you do not supply one
3245 yourself, the linker will use a default script that is compiled into the
3246 linker executable. You can use the @samp{--verbose} command-line option
3247 to display the default linker script. Certain command-line options,
3248 such as @samp{-r} or @samp{-N}, will affect the default linker script.
3250 You may supply your own linker script by using the @samp{-T} command
3251 line option. When you do this, your linker script will replace the
3252 default linker script.
3254 You may also use linker scripts implicitly by naming them as input files
3255 to the linker, as though they were files to be linked. @xref{Implicit
3259 * Basic Script Concepts:: Basic Linker Script Concepts
3260 * Script Format:: Linker Script Format
3261 * Simple Example:: Simple Linker Script Example
3262 * Simple Commands:: Simple Linker Script Commands
3263 * Assignments:: Assigning Values to Symbols
3264 * SECTIONS:: SECTIONS Command
3265 * MEMORY:: MEMORY Command
3266 * PHDRS:: PHDRS Command
3267 * VERSION:: VERSION Command
3268 * Expressions:: Expressions in Linker Scripts
3269 * Implicit Linker Scripts:: Implicit Linker Scripts
3272 @node Basic Script Concepts
3273 @section Basic Linker Script Concepts
3274 @cindex linker script concepts
3275 We need to define some basic concepts and vocabulary in order to
3276 describe the linker script language.
3278 The linker combines input files into a single output file. The output
3279 file and each input file are in a special data format known as an
3280 @dfn{object file format}. Each file is called an @dfn{object file}.
3281 The output file is often called an @dfn{executable}, but for our
3282 purposes we will also call it an object file. Each object file has,
3283 among other things, a list of @dfn{sections}. We sometimes refer to a
3284 section in an input file as an @dfn{input section}; similarly, a section
3285 in the output file is an @dfn{output section}.
3287 Each section in an object file has a name and a size. Most sections
3288 also have an associated block of data, known as the @dfn{section
3289 contents}. A section may be marked as @dfn{loadable}, which means that
3290 the contents should be loaded into memory when the output file is run.
3291 A section with no contents may be @dfn{allocatable}, which means that an
3292 area in memory should be set aside, but nothing in particular should be
3293 loaded there (in some cases this memory must be zeroed out). A section
3294 which is neither loadable nor allocatable typically contains some sort
3295 of debugging information.
3297 Every loadable or allocatable output section has two addresses. The
3298 first is the @dfn{VMA}, or virtual memory address. This is the address
3299 the section will have when the output file is run. The second is the
3300 @dfn{LMA}, or load memory address. This is the address at which the
3301 section will be loaded. In most cases the two addresses will be the
3302 same. An example of when they might be different is when a data section
3303 is loaded into ROM, and then copied into RAM when the program starts up
3304 (this technique is often used to initialize global variables in a ROM
3305 based system). In this case the ROM address would be the LMA, and the
3306 RAM address would be the VMA.
3308 You can see the sections in an object file by using the @code{objdump}
3309 program with the @samp{-h} option.
3311 Every object file also has a list of @dfn{symbols}, known as the
3312 @dfn{symbol table}. A symbol may be defined or undefined. Each symbol
3313 has a name, and each defined symbol has an address, among other
3314 information. If you compile a C or C++ program into an object file, you
3315 will get a defined symbol for every defined function and global or
3316 static variable. Every undefined function or global variable which is
3317 referenced in the input file will become an undefined symbol.
3319 You can see the symbols in an object file by using the @code{nm}
3320 program, or by using the @code{objdump} program with the @samp{-t}
3324 @section Linker Script Format
3325 @cindex linker script format
3326 Linker scripts are text files.
3328 You write a linker script as a series of commands. Each command is
3329 either a keyword, possibly followed by arguments, or an assignment to a
3330 symbol. You may separate commands using semicolons. Whitespace is
3333 Strings such as file or format names can normally be entered directly.
3334 If the file name contains a character such as a comma which would
3335 otherwise serve to separate file names, you may put the file name in
3336 double quotes. There is no way to use a double quote character in a
3339 You may include comments in linker scripts just as in C, delimited by
3340 @samp{/*} and @samp{*/}. As in C, comments are syntactically equivalent
3343 @node Simple Example
3344 @section Simple Linker Script Example
3345 @cindex linker script example
3346 @cindex example of linker script
3347 Many linker scripts are fairly simple.
3349 The simplest possible linker script has just one command:
3350 @samp{SECTIONS}. You use the @samp{SECTIONS} command to describe the
3351 memory layout of the output file.
3353 The @samp{SECTIONS} command is a powerful command. Here we will
3354 describe a simple use of it. Let's assume your program consists only of
3355 code, initialized data, and uninitialized data. These will be in the
3356 @samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
3357 Let's assume further that these are the only sections which appear in
3360 For this example, let's say that the code should be loaded at address
3361 0x10000, and that the data should start at address 0x8000000. Here is a
3362 linker script which will do that:
3367 .text : @{ *(.text) @}
3369 .data : @{ *(.data) @}
3370 .bss : @{ *(.bss) @}
3374 You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
3375 followed by a series of symbol assignments and output section
3376 descriptions enclosed in curly braces.
3378 The first line inside the @samp{SECTIONS} command of the above example
3379 sets the value of the special symbol @samp{.}, which is the location
3380 counter. If you do not specify the address of an output section in some
3381 other way (other ways are described later), the address is set from the
3382 current value of the location counter. The location counter is then
3383 incremented by the size of the output section. At the start of the
3384 @samp{SECTIONS} command, the location counter has the value @samp{0}.
3386 The second line defines an output section, @samp{.text}. The colon is
3387 required syntax which may be ignored for now. Within the curly braces
3388 after the output section name, you list the names of the input sections
3389 which should be placed into this output section. The @samp{*} is a
3390 wildcard which matches any file name. The expression @samp{*(.text)}
3391 means all @samp{.text} input sections in all input files.
3393 Since the location counter is @samp{0x10000} when the output section
3394 @samp{.text} is defined, the linker will set the address of the
3395 @samp{.text} section in the output file to be @samp{0x10000}.
3397 The remaining lines define the @samp{.data} and @samp{.bss} sections in
3398 the output file. The linker will place the @samp{.data} output section
3399 at address @samp{0x8000000}. After the linker places the @samp{.data}
3400 output section, the value of the location counter will be
3401 @samp{0x8000000} plus the size of the @samp{.data} output section. The
3402 effect is that the linker will place the @samp{.bss} output section
3403 immediately after the @samp{.data} output section in memory.
3405 The linker will ensure that each output section has the required
3406 alignment, by increasing the location counter if necessary. In this
3407 example, the specified addresses for the @samp{.text} and @samp{.data}
3408 sections will probably satisfy any alignment constraints, but the linker
3409 may have to create a small gap between the @samp{.data} and @samp{.bss}
3412 That's it! That's a simple and complete linker script.
3414 @node Simple Commands
3415 @section Simple Linker Script Commands
3416 @cindex linker script simple commands
3417 In this section we describe the simple linker script commands.
3420 * Entry Point:: Setting the entry point
3421 * File Commands:: Commands dealing with files
3422 @ifclear SingleFormat
3423 * Format Commands:: Commands dealing with object file formats
3426 * REGION_ALIAS:: Assign alias names to memory regions
3427 * Miscellaneous Commands:: Other linker script commands
3431 @subsection Setting the Entry Point
3432 @kindex ENTRY(@var{symbol})
3433 @cindex start of execution
3434 @cindex first instruction
3436 The first instruction to execute in a program is called the @dfn{entry
3437 point}. You can use the @code{ENTRY} linker script command to set the
3438 entry point. The argument is a symbol name:
3443 There are several ways to set the entry point. The linker will set the
3444 entry point by trying each of the following methods in order, and
3445 stopping when one of them succeeds:
3448 the @samp{-e} @var{entry} command-line option;
3450 the @code{ENTRY(@var{symbol})} command in a linker script;
3452 the value of a target specific symbol, if it is defined; For many
3453 targets this is @code{start}, but PE- and BeOS-based systems for example
3454 check a list of possible entry symbols, matching the first one found.
3456 the address of the first byte of the @samp{.text} section, if present;
3458 The address @code{0}.
3462 @subsection Commands Dealing with Files
3463 @cindex linker script file commands
3464 Several linker script commands deal with files.
3467 @item INCLUDE @var{filename}
3468 @kindex INCLUDE @var{filename}
3469 @cindex including a linker script
3470 Include the linker script @var{filename} at this point. The file will
3471 be searched for in the current directory, and in any directory specified
3472 with the @option{-L} option. You can nest calls to @code{INCLUDE} up to
3475 You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
3476 @code{SECTIONS} commands, or in output section descriptions.
3478 @item INPUT(@var{file}, @var{file}, @dots{})
3479 @itemx INPUT(@var{file} @var{file} @dots{})
3480 @kindex INPUT(@var{files})
3481 @cindex input files in linker scripts
3482 @cindex input object files in linker scripts
3483 @cindex linker script input object files
3484 The @code{INPUT} command directs the linker to include the named files
3485 in the link, as though they were named on the command line.
3487 For example, if you always want to include @file{subr.o} any time you do
3488 a link, but you can't be bothered to put it on every link command line,
3489 then you can put @samp{INPUT (subr.o)} in your linker script.
3491 In fact, if you like, you can list all of your input files in the linker
3492 script, and then invoke the linker with nothing but a @samp{-T} option.
3494 In case a @dfn{sysroot prefix} is configured, and the filename starts
3495 with the @samp{/} character, and the script being processed was
3496 located inside the @dfn{sysroot prefix}, the filename will be looked
3497 for in the @dfn{sysroot prefix}. Otherwise, the linker will try to
3498 open the file in the current directory. If it is not found, the
3499 linker will search through the archive library search path.
3500 The @dfn{sysroot prefix} can also be forced by specifying @code{=}
3501 as the first character in the filename path, or prefixing the filename
3502 path with @code{$SYSROOT}. See also the description of @samp{-L} in
3503 @ref{Options,,Command-line Options}.
3505 If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
3506 name to @code{lib@var{file}.a}, as with the command-line argument
3509 When you use the @code{INPUT} command in an implicit linker script, the
3510 files will be included in the link at the point at which the linker
3511 script file is included. This can affect archive searching.
3513 @item GROUP(@var{file}, @var{file}, @dots{})
3514 @itemx GROUP(@var{file} @var{file} @dots{})
3515 @kindex GROUP(@var{files})
3516 @cindex grouping input files
3517 The @code{GROUP} command is like @code{INPUT}, except that the named
3518 files should all be archives, and they are searched repeatedly until no
3519 new undefined references are created. See the description of @samp{-(}
3520 in @ref{Options,,Command-line Options}.
3522 @item AS_NEEDED(@var{file}, @var{file}, @dots{})
3523 @itemx AS_NEEDED(@var{file} @var{file} @dots{})
3524 @kindex AS_NEEDED(@var{files})
3525 This construct can appear only inside of the @code{INPUT} or @code{GROUP}
3526 commands, among other filenames. The files listed will be handled
3527 as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
3528 with the exception of ELF shared libraries, that will be added only
3529 when they are actually needed. This construct essentially enables
3530 @option{--as-needed} option for all the files listed inside of it
3531 and restores previous @option{--as-needed} resp. @option{--no-as-needed}
3534 @item OUTPUT(@var{filename})
3535 @kindex OUTPUT(@var{filename})
3536 @cindex output file name in linker script
3537 The @code{OUTPUT} command names the output file. Using
3538 @code{OUTPUT(@var{filename})} in the linker script is exactly like using
3539 @samp{-o @var{filename}} on the command line (@pxref{Options,,Command
3540 Line Options}). If both are used, the command-line option takes
3543 You can use the @code{OUTPUT} command to define a default name for the
3544 output file other than the usual default of @file{a.out}.
3546 @item SEARCH_DIR(@var{path})
3547 @kindex SEARCH_DIR(@var{path})
3548 @cindex library search path in linker script
3549 @cindex archive search path in linker script
3550 @cindex search path in linker script
3551 The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
3552 @command{ld} looks for archive libraries. Using
3553 @code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
3554 on the command line (@pxref{Options,,Command-line Options}). If both
3555 are used, then the linker will search both paths. Paths specified using
3556 the command-line option are searched first.
3558 @item STARTUP(@var{filename})
3559 @kindex STARTUP(@var{filename})
3560 @cindex first input file
3561 The @code{STARTUP} command is just like the @code{INPUT} command, except
3562 that @var{filename} will become the first input file to be linked, as
3563 though it were specified first on the command line. This may be useful
3564 when using a system in which the entry point is always the start of the
3568 @ifclear SingleFormat
3569 @node Format Commands
3570 @subsection Commands Dealing with Object File Formats
3571 A couple of linker script commands deal with object file formats.
3574 @item OUTPUT_FORMAT(@var{bfdname})
3575 @itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
3576 @kindex OUTPUT_FORMAT(@var{bfdname})
3577 @cindex output file format in linker script
3578 The @code{OUTPUT_FORMAT} command names the BFD format to use for the
3579 output file (@pxref{BFD}). Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3580 exactly like using @samp{--oformat @var{bfdname}} on the command line
3581 (@pxref{Options,,Command-line Options}). If both are used, the command
3582 line option takes precedence.
3584 You can use @code{OUTPUT_FORMAT} with three arguments to use different
3585 formats based on the @samp{-EB} and @samp{-EL} command-line options.
3586 This permits the linker script to set the output format based on the
3589 If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3590 will be the first argument, @var{default}. If @samp{-EB} is used, the
3591 output format will be the second argument, @var{big}. If @samp{-EL} is
3592 used, the output format will be the third argument, @var{little}.
3594 For example, the default linker script for the MIPS ELF target uses this
3597 OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3599 This says that the default format for the output file is
3600 @samp{elf32-bigmips}, but if the user uses the @samp{-EL} command-line
3601 option, the output file will be created in the @samp{elf32-littlemips}
3604 @item TARGET(@var{bfdname})
3605 @kindex TARGET(@var{bfdname})
3606 @cindex input file format in linker script
3607 The @code{TARGET} command names the BFD format to use when reading input
3608 files. It affects subsequent @code{INPUT} and @code{GROUP} commands.
3609 This command is like using @samp{-b @var{bfdname}} on the command line
3610 (@pxref{Options,,Command-line Options}). If the @code{TARGET} command
3611 is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3612 command is also used to set the format for the output file. @xref{BFD}.
3617 @subsection Assign alias names to memory regions
3618 @kindex REGION_ALIAS(@var{alias}, @var{region})
3619 @cindex region alias
3620 @cindex region names
3622 Alias names can be added to existing memory regions created with the
3623 @ref{MEMORY} command. Each name corresponds to at most one memory region.
3626 REGION_ALIAS(@var{alias}, @var{region})
3629 The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3630 memory region @var{region}. This allows a flexible mapping of output sections
3631 to memory regions. An example follows.
3633 Suppose we have an application for embedded systems which come with various
3634 memory storage devices. All have a general purpose, volatile memory @code{RAM}
3635 that allows code execution or data storage. Some may have a read-only,
3636 non-volatile memory @code{ROM} that allows code execution and read-only data
3637 access. The last variant is a read-only, non-volatile memory @code{ROM2} with
3638 read-only data access and no code execution capability. We have four output
3643 @code{.text} program code;
3645 @code{.rodata} read-only data;
3647 @code{.data} read-write initialized data;
3649 @code{.bss} read-write zero initialized data.
3652 The goal is to provide a linker command file that contains a system independent
3653 part defining the output sections and a system dependent part mapping the
3654 output sections to the memory regions available on the system. Our embedded
3655 systems come with three different memory setups @code{A}, @code{B} and
3657 @multitable @columnfractions .25 .25 .25 .25
3658 @item Section @tab Variant A @tab Variant B @tab Variant C
3659 @item .text @tab RAM @tab ROM @tab ROM
3660 @item .rodata @tab RAM @tab ROM @tab ROM2
3661 @item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3662 @item .bss @tab RAM @tab RAM @tab RAM
3664 The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3665 loaded into region @code{ROM} or @code{ROM2} respectively. Please note that
3666 the load address of the @code{.data} section starts in all three variants at
3667 the end of the @code{.rodata} section.
3669 The base linker script that deals with the output sections follows. It
3670 includes the system dependent @code{linkcmds.memory} file that describes the
3673 INCLUDE linkcmds.memory
3686 .data : AT (rodata_end)
3691 data_size = SIZEOF(.data);
3692 data_load_start = LOADADDR(.data);
3700 Now we need three different @code{linkcmds.memory} files to define memory
3701 regions and alias names. The content of @code{linkcmds.memory} for the three
3702 variants @code{A}, @code{B} and @code{C}:
3705 Here everything goes into the @code{RAM}.
3709 RAM : ORIGIN = 0, LENGTH = 4M
3712 REGION_ALIAS("REGION_TEXT", RAM);
3713 REGION_ALIAS("REGION_RODATA", RAM);
3714 REGION_ALIAS("REGION_DATA", RAM);
3715 REGION_ALIAS("REGION_BSS", RAM);
3718 Program code and read-only data go into the @code{ROM}. Read-write data goes
3719 into the @code{RAM}. An image of the initialized data is loaded into the
3720 @code{ROM} and will be copied during system start into the @code{RAM}.
3724 ROM : ORIGIN = 0, LENGTH = 3M
3725 RAM : ORIGIN = 0x10000000, LENGTH = 1M
3728 REGION_ALIAS("REGION_TEXT", ROM);
3729 REGION_ALIAS("REGION_RODATA", ROM);
3730 REGION_ALIAS("REGION_DATA", RAM);
3731 REGION_ALIAS("REGION_BSS", RAM);
3734 Program code goes into the @code{ROM}. Read-only data goes into the
3735 @code{ROM2}. Read-write data goes into the @code{RAM}. An image of the
3736 initialized data is loaded into the @code{ROM2} and will be copied during
3737 system start into the @code{RAM}.
3741 ROM : ORIGIN = 0, LENGTH = 2M
3742 ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3743 RAM : ORIGIN = 0x20000000, LENGTH = 1M
3746 REGION_ALIAS("REGION_TEXT", ROM);
3747 REGION_ALIAS("REGION_RODATA", ROM2);
3748 REGION_ALIAS("REGION_DATA", RAM);
3749 REGION_ALIAS("REGION_BSS", RAM);
3753 It is possible to write a common system initialization routine to copy the
3754 @code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3759 extern char data_start [];
3760 extern char data_size [];
3761 extern char data_load_start [];
3763 void copy_data(void)
3765 if (data_start != data_load_start)
3767 memcpy(data_start, data_load_start, (size_t) data_size);
3772 @node Miscellaneous Commands
3773 @subsection Other Linker Script Commands
3774 There are a few other linker scripts commands.
3777 @item ASSERT(@var{exp}, @var{message})
3779 @cindex assertion in linker script
3780 Ensure that @var{exp} is non-zero. If it is zero, then exit the linker
3781 with an error code, and print @var{message}.
3783 Note that assertions are checked before the final stages of linking
3784 take place. This means that expressions involving symbols PROVIDEd
3785 inside section definitions will fail if the user has not set values
3786 for those symbols. The only exception to this rule is PROVIDEd
3787 symbols that just reference dot. Thus an assertion like this:
3792 PROVIDE (__stack = .);
3793 PROVIDE (__stack_size = 0x100);
3794 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3798 will fail if @code{__stack_size} is not defined elsewhere. Symbols
3799 PROVIDEd outside of section definitions are evaluated earlier, so they
3800 can be used inside ASSERTions. Thus:
3803 PROVIDE (__stack_size = 0x100);
3806 PROVIDE (__stack = .);
3807 ASSERT ((__stack > (_end + __stack_size)), "Error: No room left for the stack");
3813 @item EXTERN(@var{symbol} @var{symbol} @dots{})
3815 @cindex undefined symbol in linker script
3816 Force @var{symbol} to be entered in the output file as an undefined
3817 symbol. Doing this may, for example, trigger linking of additional
3818 modules from standard libraries. You may list several @var{symbol}s for
3819 each @code{EXTERN}, and you may use @code{EXTERN} multiple times. This
3820 command has the same effect as the @samp{-u} command-line option.
3822 @item FORCE_COMMON_ALLOCATION
3823 @kindex FORCE_COMMON_ALLOCATION
3824 @cindex common allocation in linker script
3825 This command has the same effect as the @samp{-d} command-line option:
3826 to make @command{ld} assign space to common symbols even if a relocatable
3827 output file is specified (@samp{-r}).
3829 @item INHIBIT_COMMON_ALLOCATION
3830 @kindex INHIBIT_COMMON_ALLOCATION
3831 @cindex common allocation in linker script
3832 This command has the same effect as the @samp{--no-define-common}
3833 command-line option: to make @code{ld} omit the assignment of addresses
3834 to common symbols even for a non-relocatable output file.
3836 @item FORCE_GROUP_ALLOCATION
3837 @kindex FORCE_GROUP_ALLOCATION
3838 @cindex group allocation in linker script
3839 @cindex section groups
3841 This command has the same effect as the
3842 @samp{--force-group-allocation} command-line option: to make
3843 @command{ld} place section group members like normal input sections,
3844 and to delete the section groups even if a relocatable output file is
3845 specified (@samp{-r}).
3847 @item INSERT [ AFTER | BEFORE ] @var{output_section}
3849 @cindex insert user script into default script
3850 This command is typically used in a script specified by @samp{-T} to
3851 augment the default @code{SECTIONS} with, for example, overlays. It
3852 inserts all prior linker script statements after (or before)
3853 @var{output_section}, and also causes @samp{-T} to not override the
3854 default linker script. The exact insertion point is as for orphan
3855 sections. @xref{Location Counter}. The insertion happens after the
3856 linker has mapped input sections to output sections. Prior to the
3857 insertion, since @samp{-T} scripts are parsed before the default
3858 linker script, statements in the @samp{-T} script occur before the
3859 default linker script statements in the internal linker representation
3860 of the script. In particular, input section assignments will be made
3861 to @samp{-T} output sections before those in the default script. Here
3862 is an example of how a @samp{-T} script using @code{INSERT} might look:
3869 .ov1 @{ ov1*(.text) @}
3870 .ov2 @{ ov2*(.text) @}
3876 @item NOCROSSREFS(@var{section} @var{section} @dots{})
3877 @kindex NOCROSSREFS(@var{sections})
3878 @cindex cross references
3879 This command may be used to tell @command{ld} to issue an error about any
3880 references among certain output sections.
3882 In certain types of programs, particularly on embedded systems when
3883 using overlays, when one section is loaded into memory, another section
3884 will not be. Any direct references between the two sections would be
3885 errors. For example, it would be an error if code in one section called
3886 a function defined in the other section.
3888 The @code{NOCROSSREFS} command takes a list of output section names. If
3889 @command{ld} detects any cross references between the sections, it reports
3890 an error and returns a non-zero exit status. Note that the
3891 @code{NOCROSSREFS} command uses output section names, not input section
3894 @item NOCROSSREFS_TO(@var{tosection} @var{fromsection} @dots{})
3895 @kindex NOCROSSREFS_TO(@var{tosection} @var{fromsections})
3896 @cindex cross references
3897 This command may be used to tell @command{ld} to issue an error about any
3898 references to one section from a list of other sections.
3900 The @code{NOCROSSREFS} command is useful when ensuring that two or more
3901 output sections are entirely independent but there are situations where
3902 a one-way dependency is needed. For example, in a multi-core application
3903 there may be shared code that can be called from each core but for safety
3904 must never call back.
3906 The @code{NOCROSSREFS_TO} command takes a list of output section names.
3907 The first section can not be referenced from any of the other sections.
3908 If @command{ld} detects any references to the first section from any of
3909 the other sections, it reports an error and returns a non-zero exit
3910 status. Note that the @code{NOCROSSREFS_TO} command uses output section
3911 names, not input section names.
3913 @ifclear SingleFormat
3914 @item OUTPUT_ARCH(@var{bfdarch})
3915 @kindex OUTPUT_ARCH(@var{bfdarch})
3916 @cindex machine architecture
3917 @cindex architecture
3918 Specify a particular output machine architecture. The argument is one
3919 of the names used by the BFD library (@pxref{BFD}). You can see the
3920 architecture of an object file by using the @code{objdump} program with
3921 the @samp{-f} option.
3924 @item LD_FEATURE(@var{string})
3925 @kindex LD_FEATURE(@var{string})
3926 This command may be used to modify @command{ld} behavior. If
3927 @var{string} is @code{"SANE_EXPR"} then absolute symbols and numbers
3928 in a script are simply treated as numbers everywhere.
3929 @xref{Expression Section}.
3933 @section Assigning Values to Symbols
3934 @cindex assignment in scripts
3935 @cindex symbol definition, scripts
3936 @cindex variables, defining
3937 You may assign a value to a symbol in a linker script. This will define
3938 the symbol and place it into the symbol table with a global scope.
3941 * Simple Assignments:: Simple Assignments
3944 * PROVIDE_HIDDEN:: PROVIDE_HIDDEN
3945 * Source Code Reference:: How to use a linker script defined symbol in source code
3948 @node Simple Assignments
3949 @subsection Simple Assignments
3951 You may assign to a symbol using any of the C assignment operators:
3954 @item @var{symbol} = @var{expression} ;
3955 @itemx @var{symbol} += @var{expression} ;
3956 @itemx @var{symbol} -= @var{expression} ;
3957 @itemx @var{symbol} *= @var{expression} ;
3958 @itemx @var{symbol} /= @var{expression} ;
3959 @itemx @var{symbol} <<= @var{expression} ;
3960 @itemx @var{symbol} >>= @var{expression} ;
3961 @itemx @var{symbol} &= @var{expression} ;
3962 @itemx @var{symbol} |= @var{expression} ;
3965 The first case will define @var{symbol} to the value of
3966 @var{expression}. In the other cases, @var{symbol} must already be
3967 defined, and the value will be adjusted accordingly.
3969 The special symbol name @samp{.} indicates the location counter. You
3970 may only use this within a @code{SECTIONS} command. @xref{Location Counter}.
3972 The semicolon after @var{expression} is required.
3974 Expressions are defined below; see @ref{Expressions}.
3976 You may write symbol assignments as commands in their own right, or as
3977 statements within a @code{SECTIONS} command, or as part of an output
3978 section description in a @code{SECTIONS} command.
3980 The section of the symbol will be set from the section of the
3981 expression; for more information, see @ref{Expression Section}.
3983 Here is an example showing the three different places that symbol
3984 assignments may be used:
3995 _bdata = (. + 3) & ~ 3;
3996 .data : @{ *(.data) @}
4000 In this example, the symbol @samp{floating_point} will be defined as
4001 zero. The symbol @samp{_etext} will be defined as the address following
4002 the last @samp{.text} input section. The symbol @samp{_bdata} will be
4003 defined as the address following the @samp{.text} output section aligned
4004 upward to a 4 byte boundary.
4009 For ELF targeted ports, define a symbol that will be hidden and won't be
4010 exported. The syntax is @code{HIDDEN(@var{symbol} = @var{expression})}.
4012 Here is the example from @ref{Simple Assignments}, rewritten to use
4016 HIDDEN(floating_point = 0);
4024 HIDDEN(_bdata = (. + 3) & ~ 3);
4025 .data : @{ *(.data) @}
4029 In this case none of the three symbols will be visible outside this module.
4034 In some cases, it is desirable for a linker script to define a symbol
4035 only if it is referenced and is not defined by any object included in
4036 the link. For example, traditional linkers defined the symbol
4037 @samp{etext}. However, ANSI C requires that the user be able to use
4038 @samp{etext} as a function name without encountering an error. The
4039 @code{PROVIDE} keyword may be used to define a symbol, such as
4040 @samp{etext}, only if it is referenced but not defined. The syntax is
4041 @code{PROVIDE(@var{symbol} = @var{expression})}.
4043 Here is an example of using @code{PROVIDE} to define @samp{etext}:
4056 In this example, if the program defines @samp{_etext} (with a leading
4057 underscore), the linker will give a multiple definition error. If, on
4058 the other hand, the program defines @samp{etext} (with no leading
4059 underscore), the linker will silently use the definition in the program.
4060 If the program references @samp{etext} but does not define it, the
4061 linker will use the definition in the linker script.
4063 Note - the @code{PROVIDE} directive considers a common symbol to be
4064 defined, even though such a symbol could be combined with the symbol
4065 that the @code{PROVIDE} would create. This is particularly important
4066 when considering constructor and destructor list symbols such as
4067 @samp{__CTOR_LIST__} as these are often defined as common symbols.
4069 @node PROVIDE_HIDDEN
4070 @subsection PROVIDE_HIDDEN
4071 @cindex PROVIDE_HIDDEN
4072 Similar to @code{PROVIDE}. For ELF targeted ports, the symbol will be
4073 hidden and won't be exported.
4075 @node Source Code Reference
4076 @subsection Source Code Reference
4078 Accessing a linker script defined variable from source code is not
4079 intuitive. In particular a linker script symbol is not equivalent to
4080 a variable declaration in a high level language, it is instead a
4081 symbol that does not have a value.
4083 Before going further, it is important to note that compilers often
4084 transform names in the source code into different names when they are
4085 stored in the symbol table. For example, Fortran compilers commonly
4086 prepend or append an underscore, and C++ performs extensive @samp{name
4087 mangling}. Therefore there might be a discrepancy between the name
4088 of a variable as it is used in source code and the name of the same
4089 variable as it is defined in a linker script. For example in C a
4090 linker script variable might be referred to as:
4096 But in the linker script it might be defined as:
4102 In the remaining examples however it is assumed that no name
4103 transformation has taken place.
4105 When a symbol is declared in a high level language such as C, two
4106 things happen. The first is that the compiler reserves enough space
4107 in the program's memory to hold the @emph{value} of the symbol. The
4108 second is that the compiler creates an entry in the program's symbol
4109 table which holds the symbol's @emph{address}. ie the symbol table
4110 contains the address of the block of memory holding the symbol's
4111 value. So for example the following C declaration, at file scope:
4117 creates an entry called @samp{foo} in the symbol table. This entry
4118 holds the address of an @samp{int} sized block of memory where the
4119 number 1000 is initially stored.
4121 When a program references a symbol the compiler generates code that
4122 first accesses the symbol table to find the address of the symbol's
4123 memory block and then code to read the value from that memory block.
4130 looks up the symbol @samp{foo} in the symbol table, gets the address
4131 associated with this symbol and then writes the value 1 into that
4138 looks up the symbol @samp{foo} in the symbol table, gets its address
4139 and then copies this address into the block of memory associated with
4140 the variable @samp{a}.
4142 Linker scripts symbol declarations, by contrast, create an entry in
4143 the symbol table but do not assign any memory to them. Thus they are
4144 an address without a value. So for example the linker script definition:
4150 creates an entry in the symbol table called @samp{foo} which holds
4151 the address of memory location 1000, but nothing special is stored at
4152 address 1000. This means that you cannot access the @emph{value} of a
4153 linker script defined symbol - it has no value - all you can do is
4154 access the @emph{address} of a linker script defined symbol.
4156 Hence when you are using a linker script defined symbol in source code
4157 you should always take the address of the symbol, and never attempt to
4158 use its value. For example suppose you want to copy the contents of a
4159 section of memory called .ROM into a section called .FLASH and the
4160 linker script contains these declarations:
4164 start_of_ROM = .ROM;
4165 end_of_ROM = .ROM + sizeof (.ROM);
4166 start_of_FLASH = .FLASH;
4170 Then the C source code to perform the copy would be:
4174 extern char start_of_ROM, end_of_ROM, start_of_FLASH;
4176 memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
4180 Note the use of the @samp{&} operators. These are correct.
4181 Alternatively the symbols can be treated as the names of vectors or
4182 arrays and then the code will again work as expected:
4186 extern char start_of_ROM[], end_of_ROM[], start_of_FLASH[];
4188 memcpy (start_of_FLASH, start_of_ROM, end_of_ROM - start_of_ROM);
4192 Note how using this method does not require the use of @samp{&}
4196 @section SECTIONS Command
4198 The @code{SECTIONS} command tells the linker how to map input sections
4199 into output sections, and how to place the output sections in memory.
4201 The format of the @code{SECTIONS} command is:
4205 @var{sections-command}
4206 @var{sections-command}
4211 Each @var{sections-command} may of be one of the following:
4215 an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
4217 a symbol assignment (@pxref{Assignments})
4219 an output section description
4221 an overlay description
4224 The @code{ENTRY} command and symbol assignments are permitted inside the
4225 @code{SECTIONS} command for convenience in using the location counter in
4226 those commands. This can also make the linker script easier to
4227 understand because you can use those commands at meaningful points in
4228 the layout of the output file.
4230 Output section descriptions and overlay descriptions are described
4233 If you do not use a @code{SECTIONS} command in your linker script, the
4234 linker will place each input section into an identically named output
4235 section in the order that the sections are first encountered in the
4236 input files. If all input sections are present in the first file, for
4237 example, the order of sections in the output file will match the order
4238 in the first input file. The first section will be at address zero.
4241 * Output Section Description:: Output section description
4242 * Output Section Name:: Output section name
4243 * Output Section Address:: Output section address
4244 * Input Section:: Input section description
4245 * Output Section Data:: Output section data
4246 * Output Section Keywords:: Output section keywords
4247 * Output Section Discarding:: Output section discarding
4248 * Output Section Attributes:: Output section attributes
4249 * Overlay Description:: Overlay description
4252 @node Output Section Description
4253 @subsection Output Section Description
4254 The full description of an output section looks like this:
4257 @var{section} [@var{address}] [(@var{type})] :
4259 [ALIGN(@var{section_align}) | ALIGN_WITH_INPUT]
4260 [SUBALIGN(@var{subsection_align})]
4263 @var{output-section-command}
4264 @var{output-section-command}
4266 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}] [,]
4270 Most output sections do not use most of the optional section attributes.
4272 The whitespace around @var{section} is required, so that the section
4273 name is unambiguous. The colon and the curly braces are also required.
4274 The comma at the end may be required if a @var{fillexp} is used and
4275 the next @var{sections-command} looks like a continuation of the expression.
4276 The line breaks and other white space are optional.
4278 Each @var{output-section-command} may be one of the following:
4282 a symbol assignment (@pxref{Assignments})
4284 an input section description (@pxref{Input Section})
4286 data values to include directly (@pxref{Output Section Data})
4288 a special output section keyword (@pxref{Output Section Keywords})
4291 @node Output Section Name
4292 @subsection Output Section Name
4293 @cindex name, section
4294 @cindex section name
4295 The name of the output section is @var{section}. @var{section} must
4296 meet the constraints of your output format. In formats which only
4297 support a limited number of sections, such as @code{a.out}, the name
4298 must be one of the names supported by the format (@code{a.out}, for
4299 example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
4300 output format supports any number of sections, but with numbers and not
4301 names (as is the case for Oasys), the name should be supplied as a
4302 quoted numeric string. A section name may consist of any sequence of
4303 characters, but a name which contains any unusual characters such as
4304 commas must be quoted.
4306 The output section name @samp{/DISCARD/} is special; @ref{Output Section
4309 @node Output Section Address
4310 @subsection Output Section Address
4311 @cindex address, section
4312 @cindex section address
4313 The @var{address} is an expression for the VMA (the virtual memory
4314 address) of the output section. This address is optional, but if it
4315 is provided then the output address will be set exactly as specified.
4317 If the output address is not specified then one will be chosen for the
4318 section, based on the heuristic below. This address will be adjusted
4319 to fit the alignment requirement of the output section. The
4320 alignment requirement is the strictest alignment of any input section
4321 contained within the output section.
4323 The output section address heuristic is as follows:
4327 If an output memory @var{region} is set for the section then it
4328 is added to this region and its address will be the next free address
4332 If the MEMORY command has been used to create a list of memory
4333 regions then the first region which has attributes compatible with the
4334 section is selected to contain it. The section's output address will
4335 be the next free address in that region; @ref{MEMORY}.
4338 If no memory regions were specified, or none match the section then
4339 the output address will be based on the current value of the location
4347 .text . : @{ *(.text) @}
4354 .text : @{ *(.text) @}
4358 are subtly different. The first will set the address of the
4359 @samp{.text} output section to the current value of the location
4360 counter. The second will set it to the current value of the location
4361 counter aligned to the strictest alignment of any of the @samp{.text}
4364 The @var{address} may be an arbitrary expression; @ref{Expressions}.
4365 For example, if you want to align the section on a 0x10 byte boundary,
4366 so that the lowest four bits of the section address are zero, you could
4367 do something like this:
4369 .text ALIGN(0x10) : @{ *(.text) @}
4372 This works because @code{ALIGN} returns the current location counter
4373 aligned upward to the specified value.
4375 Specifying @var{address} for a section will change the value of the
4376 location counter, provided that the section is non-empty. (Empty
4377 sections are ignored).
4380 @subsection Input Section Description
4381 @cindex input sections
4382 @cindex mapping input sections to output sections
4383 The most common output section command is an input section description.
4385 The input section description is the most basic linker script operation.
4386 You use output sections to tell the linker how to lay out your program
4387 in memory. You use input section descriptions to tell the linker how to
4388 map the input files into your memory layout.
4391 * Input Section Basics:: Input section basics
4392 * Input Section Wildcards:: Input section wildcard patterns
4393 * Input Section Common:: Input section for common symbols
4394 * Input Section Keep:: Input section and garbage collection
4395 * Input Section Example:: Input section example
4398 @node Input Section Basics
4399 @subsubsection Input Section Basics
4400 @cindex input section basics
4401 An input section description consists of a file name optionally followed
4402 by a list of section names in parentheses.
4404 The file name and the section name may be wildcard patterns, which we
4405 describe further below (@pxref{Input Section Wildcards}).
4407 The most common input section description is to include all input
4408 sections with a particular name in the output section. For example, to
4409 include all input @samp{.text} sections, you would write:
4414 Here the @samp{*} is a wildcard which matches any file name. To exclude a list
4415 @cindex EXCLUDE_FILE
4416 of files from matching the file name wildcard, EXCLUDE_FILE may be used to
4417 match all files except the ones specified in the EXCLUDE_FILE list. For
4420 EXCLUDE_FILE (*crtend.o *otherfile.o) *(.ctors)
4423 will cause all .ctors sections from all files except @file{crtend.o}
4424 and @file{otherfile.o} to be included. The EXCLUDE_FILE can also be
4425 placed inside the section list, for example:
4427 *(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
4430 The result of this is identically to the previous example. Supporting
4431 two syntaxes for EXCLUDE_FILE is useful if the section list contains
4432 more than one section, as described below.
4434 There are two ways to include more than one section:
4440 The difference between these is the order in which the @samp{.text} and
4441 @samp{.rdata} input sections will appear in the output section. In the
4442 first example, they will be intermingled, appearing in the same order as
4443 they are found in the linker input. In the second example, all
4444 @samp{.text} input sections will appear first, followed by all
4445 @samp{.rdata} input sections.
4447 When using EXCLUDE_FILE with more than one section, if the exclusion
4448 is within the section list then the exclusion only applies to the
4449 immediately following section, for example:
4451 *(EXCLUDE_FILE (*somefile.o) .text .rdata)
4454 will cause all @samp{.text} sections from all files except
4455 @file{somefile.o} to be included, while all @samp{.rdata} sections
4456 from all files, including @file{somefile.o}, will be included. To
4457 exclude the @samp{.rdata} sections from @file{somefile.o} the example
4458 could be modified to:
4460 *(EXCLUDE_FILE (*somefile.o) .text EXCLUDE_FILE (*somefile.o) .rdata)
4463 Alternatively, placing the EXCLUDE_FILE outside of the section list,
4464 before the input file selection, will cause the exclusion to apply for
4465 all sections. Thus the previous example can be rewritten as:
4467 EXCLUDE_FILE (*somefile.o) *(.text .rdata)
4470 You can specify a file name to include sections from a particular file.
4471 You would do this if one or more of your files contain special data that
4472 needs to be at a particular location in memory. For example:
4477 To refine the sections that are included based on the section flags
4478 of an input section, INPUT_SECTION_FLAGS may be used.
4480 Here is a simple example for using Section header flags for ELF sections:
4485 .text : @{ INPUT_SECTION_FLAGS (SHF_MERGE & SHF_STRINGS) *(.text) @}
4486 .text2 : @{ INPUT_SECTION_FLAGS (!SHF_WRITE) *(.text) @}
4491 In this example, the output section @samp{.text} will be comprised of any
4492 input section matching the name *(.text) whose section header flags
4493 @code{SHF_MERGE} and @code{SHF_STRINGS} are set. The output section
4494 @samp{.text2} will be comprised of any input section matching the name *(.text)
4495 whose section header flag @code{SHF_WRITE} is clear.
4497 You can also specify files within archives by writing a pattern
4498 matching the archive, a colon, then the pattern matching the file,
4499 with no whitespace around the colon.
4503 matches file within archive
4505 matches the whole archive
4507 matches file but not one in an archive
4510 Either one or both of @samp{archive} and @samp{file} can contain shell
4511 wildcards. On DOS based file systems, the linker will assume that a
4512 single letter followed by a colon is a drive specifier, so
4513 @samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
4514 within an archive called @samp{c}. @samp{archive:file} filespecs may
4515 also be used within an @code{EXCLUDE_FILE} list, but may not appear in
4516 other linker script contexts. For instance, you cannot extract a file
4517 from an archive by using @samp{archive:file} in an @code{INPUT}
4520 If you use a file name without a list of sections, then all sections in
4521 the input file will be included in the output section. This is not
4522 commonly done, but it may by useful on occasion. For example:
4527 When you use a file name which is not an @samp{archive:file} specifier
4528 and does not contain any wild card
4529 characters, the linker will first see if you also specified the file
4530 name on the linker command line or in an @code{INPUT} command. If you
4531 did not, the linker will attempt to open the file as an input file, as
4532 though it appeared on the command line. Note that this differs from an
4533 @code{INPUT} command, because the linker will not search for the file in
4534 the archive search path.
4536 @node Input Section Wildcards
4537 @subsubsection Input Section Wildcard Patterns
4538 @cindex input section wildcards
4539 @cindex wildcard file name patterns
4540 @cindex file name wildcard patterns
4541 @cindex section name wildcard patterns
4542 In an input section description, either the file name or the section
4543 name or both may be wildcard patterns.
4545 The file name of @samp{*} seen in many examples is a simple wildcard
4546 pattern for the file name.
4548 The wildcard patterns are like those used by the Unix shell.
4552 matches any number of characters
4554 matches any single character
4556 matches a single instance of any of the @var{chars}; the @samp{-}
4557 character may be used to specify a range of characters, as in
4558 @samp{[a-z]} to match any lower case letter
4560 quotes the following character
4563 When a file name is matched with a wildcard, the wildcard characters
4564 will not match a @samp{/} character (used to separate directory names on
4565 Unix). A pattern consisting of a single @samp{*} character is an
4566 exception; it will always match any file name, whether it contains a
4567 @samp{/} or not. In a section name, the wildcard characters will match
4568 a @samp{/} character.
4570 File name wildcard patterns only match files which are explicitly
4571 specified on the command line or in an @code{INPUT} command. The linker
4572 does not search directories to expand wildcards.
4574 If a file name matches more than one wildcard pattern, or if a file name
4575 appears explicitly and is also matched by a wildcard pattern, the linker
4576 will use the first match in the linker script. For example, this
4577 sequence of input section descriptions is probably in error, because the
4578 @file{data.o} rule will not be used:
4580 .data : @{ *(.data) @}
4581 .data1 : @{ data.o(.data) @}
4584 @cindex SORT_BY_NAME
4585 Normally, the linker will place files and sections matched by wildcards
4586 in the order in which they are seen during the link. You can change
4587 this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
4588 pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}). When the
4589 @code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
4590 into ascending order by name before placing them in the output file.
4592 @cindex SORT_BY_ALIGNMENT
4593 @code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
4594 difference is @code{SORT_BY_ALIGNMENT} will sort sections into
4595 descending order by alignment before placing them in the output file.
4596 Larger alignments are placed before smaller alignments in order to
4597 reduce the amount of padding necessary.
4599 @cindex SORT_BY_INIT_PRIORITY
4600 @code{SORT_BY_INIT_PRIORITY} is very similar to @code{SORT_BY_NAME}. The
4601 difference is @code{SORT_BY_INIT_PRIORITY} will sort sections into
4602 ascending order by numerical value of the GCC init_priority attribute
4603 encoded in the section name before placing them in the output file.
4606 @code{SORT} is an alias for @code{SORT_BY_NAME}.
4608 When there are nested section sorting commands in linker script, there
4609 can be at most 1 level of nesting for section sorting commands.
4613 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4614 It will sort the input sections by name first, then by alignment if two
4615 sections have the same name.
4617 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4618 It will sort the input sections by alignment first, then by name if two
4619 sections have the same alignment.
4621 @code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
4622 treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
4624 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
4625 is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
4627 All other nested section sorting commands are invalid.
4630 When both command-line section sorting option and linker script
4631 section sorting command are used, section sorting command always
4632 takes precedence over the command-line option.
4634 If the section sorting command in linker script isn't nested, the
4635 command-line option will make the section sorting command to be
4636 treated as nested sorting command.
4640 @code{SORT_BY_NAME} (wildcard section pattern ) with
4641 @option{--sort-sections alignment} is equivalent to
4642 @code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
4644 @code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
4645 @option{--sort-section name} is equivalent to
4646 @code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
4649 If the section sorting command in linker script is nested, the
4650 command-line option will be ignored.
4653 @code{SORT_NONE} disables section sorting by ignoring the command-line
4654 section sorting option.
4656 If you ever get confused about where input sections are going, use the
4657 @samp{-M} linker option to generate a map file. The map file shows
4658 precisely how input sections are mapped to output sections.
4660 This example shows how wildcard patterns might be used to partition
4661 files. This linker script directs the linker to place all @samp{.text}
4662 sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
4663 The linker will place the @samp{.data} section from all files beginning
4664 with an upper case character in @samp{.DATA}; for all other files, the
4665 linker will place the @samp{.data} section in @samp{.data}.
4669 .text : @{ *(.text) @}
4670 .DATA : @{ [A-Z]*(.data) @}
4671 .data : @{ *(.data) @}
4672 .bss : @{ *(.bss) @}
4677 @node Input Section Common
4678 @subsubsection Input Section for Common Symbols
4679 @cindex common symbol placement
4680 @cindex uninitialized data placement
4681 A special notation is needed for common symbols, because in many object
4682 file formats common symbols do not have a particular input section. The
4683 linker treats common symbols as though they are in an input section
4684 named @samp{COMMON}.
4686 You may use file names with the @samp{COMMON} section just as with any
4687 other input sections. You can use this to place common symbols from a
4688 particular input file in one section while common symbols from other
4689 input files are placed in another section.
4691 In most cases, common symbols in input files will be placed in the
4692 @samp{.bss} section in the output file. For example:
4694 .bss @{ *(.bss) *(COMMON) @}
4697 @cindex scommon section
4698 @cindex small common symbols
4699 Some object file formats have more than one type of common symbol. For
4700 example, the MIPS ELF object file format distinguishes standard common
4701 symbols and small common symbols. In this case, the linker will use a
4702 different special section name for other types of common symbols. In
4703 the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
4704 symbols and @samp{.scommon} for small common symbols. This permits you
4705 to map the different types of common symbols into memory at different
4709 You will sometimes see @samp{[COMMON]} in old linker scripts. This
4710 notation is now considered obsolete. It is equivalent to
4713 @node Input Section Keep
4714 @subsubsection Input Section and Garbage Collection
4716 @cindex garbage collection
4717 When link-time garbage collection is in use (@samp{--gc-sections}),
4718 it is often useful to mark sections that should not be eliminated.
4719 This is accomplished by surrounding an input section's wildcard entry
4720 with @code{KEEP()}, as in @code{KEEP(*(.init))} or
4721 @code{KEEP(SORT_BY_NAME(*)(.ctors))}.
4723 @node Input Section Example
4724 @subsubsection Input Section Example
4725 The following example is a complete linker script. It tells the linker
4726 to read all of the sections from file @file{all.o} and place them at the
4727 start of output section @samp{outputa} which starts at location
4728 @samp{0x10000}. All of section @samp{.input1} from file @file{foo.o}
4729 follows immediately, in the same output section. All of section
4730 @samp{.input2} from @file{foo.o} goes into output section
4731 @samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
4732 All of the remaining @samp{.input1} and @samp{.input2} sections from any
4733 files are written to output section @samp{outputc}.
4761 If an output section's name is the same as the input section's name
4762 and is representable as a C identifier, then the linker will
4763 automatically @pxref{PROVIDE} two symbols: __start_SECNAME and
4764 __stop_SECNAME, where SECNAME is the name of the section. These
4765 indicate the start address and end address of the output section
4766 respectively. Note: most section names are not representable as
4767 C identifiers because they contain a @samp{.} character.
4769 @node Output Section Data
4770 @subsection Output Section Data
4772 @cindex section data
4773 @cindex output section data
4774 @kindex BYTE(@var{expression})
4775 @kindex SHORT(@var{expression})
4776 @kindex LONG(@var{expression})
4777 @kindex QUAD(@var{expression})
4778 @kindex SQUAD(@var{expression})
4779 You can include explicit bytes of data in an output section by using
4780 @code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
4781 an output section command. Each keyword is followed by an expression in
4782 parentheses providing the value to store (@pxref{Expressions}). The
4783 value of the expression is stored at the current value of the location
4786 The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
4787 store one, two, four, and eight bytes (respectively). After storing the
4788 bytes, the location counter is incremented by the number of bytes
4791 For example, this will store the byte 1 followed by the four byte value
4792 of the symbol @samp{addr}:
4798 When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4799 same; they both store an 8 byte, or 64 bit, value. When both host and
4800 target are 32 bits, an expression is computed as 32 bits. In this case
4801 @code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4802 @code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4804 If the object file format of the output file has an explicit endianness,
4805 which is the normal case, the value will be stored in that endianness.
4806 When the object file format does not have an explicit endianness, as is
4807 true of, for example, S-records, the value will be stored in the
4808 endianness of the first input object file.
4810 Note---these commands only work inside a section description and not
4811 between them, so the following will produce an error from the linker:
4813 SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4815 whereas this will work:
4817 SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4820 @kindex FILL(@var{expression})
4821 @cindex holes, filling
4822 @cindex unspecified memory
4823 You may use the @code{FILL} command to set the fill pattern for the
4824 current section. It is followed by an expression in parentheses. Any
4825 otherwise unspecified regions of memory within the section (for example,
4826 gaps left due to the required alignment of input sections) are filled
4827 with the value of the expression, repeated as
4828 necessary. A @code{FILL} statement covers memory locations after the
4829 point at which it occurs in the section definition; by including more
4830 than one @code{FILL} statement, you can have different fill patterns in
4831 different parts of an output section.
4833 This example shows how to fill unspecified regions of memory with the
4839 The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4840 section attribute, but it only affects the
4841 part of the section following the @code{FILL} command, rather than the
4842 entire section. If both are used, the @code{FILL} command takes
4843 precedence. @xref{Output Section Fill}, for details on the fill
4846 @node Output Section Keywords
4847 @subsection Output Section Keywords
4848 There are a couple of keywords which can appear as output section
4852 @kindex CREATE_OBJECT_SYMBOLS
4853 @cindex input filename symbols
4854 @cindex filename symbols
4855 @item CREATE_OBJECT_SYMBOLS
4856 The command tells the linker to create a symbol for each input file.
4857 The name of each symbol will be the name of the corresponding input
4858 file. The section of each symbol will be the output section in which
4859 the @code{CREATE_OBJECT_SYMBOLS} command appears.
4861 This is conventional for the a.out object file format. It is not
4862 normally used for any other object file format.
4864 @kindex CONSTRUCTORS
4865 @cindex C++ constructors, arranging in link
4866 @cindex constructors, arranging in link
4868 When linking using the a.out object file format, the linker uses an
4869 unusual set construct to support C++ global constructors and
4870 destructors. When linking object file formats which do not support
4871 arbitrary sections, such as ECOFF and XCOFF, the linker will
4872 automatically recognize C++ global constructors and destructors by name.
4873 For these object file formats, the @code{CONSTRUCTORS} command tells the
4874 linker to place constructor information in the output section where the
4875 @code{CONSTRUCTORS} command appears. The @code{CONSTRUCTORS} command is
4876 ignored for other object file formats.
4878 The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4879 constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4880 Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4881 the start and end of the global destructors. The
4882 first word in the list is the number of entries, followed by the address
4883 of each constructor or destructor, followed by a zero word. The
4884 compiler must arrange to actually run the code. For these object file
4885 formats @sc{gnu} C++ normally calls constructors from a subroutine
4886 @code{__main}; a call to @code{__main} is automatically inserted into
4887 the startup code for @code{main}. @sc{gnu} C++ normally runs
4888 destructors either by using @code{atexit}, or directly from the function
4891 For object file formats such as @code{COFF} or @code{ELF} which support
4892 arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4893 addresses of global constructors and destructors into the @code{.ctors}
4894 and @code{.dtors} sections. Placing the following sequence into your
4895 linker script will build the sort of table which the @sc{gnu} C++
4896 runtime code expects to see.
4900 LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4905 LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4911 If you are using the @sc{gnu} C++ support for initialization priority,
4912 which provides some control over the order in which global constructors
4913 are run, you must sort the constructors at link time to ensure that they
4914 are executed in the correct order. When using the @code{CONSTRUCTORS}
4915 command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead. When using the
4916 @code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4917 @samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4920 Normally the compiler and linker will handle these issues automatically,
4921 and you will not need to concern yourself with them. However, you may
4922 need to consider this if you are using C++ and writing your own linker
4927 @node Output Section Discarding
4928 @subsection Output Section Discarding
4929 @cindex discarding sections
4930 @cindex sections, discarding
4931 @cindex removing sections
4932 The linker will not normally create output sections with no contents.
4933 This is for convenience when referring to input sections that may or
4934 may not be present in any of the input files. For example:
4936 .foo : @{ *(.foo) @}
4939 will only create a @samp{.foo} section in the output file if there is a
4940 @samp{.foo} section in at least one input file, and if the input
4941 sections are not all empty. Other link script directives that allocate
4942 space in an output section will also create the output section. So
4943 too will assignments to dot even if the assignment does not create
4944 space, except for @samp{. = 0}, @samp{. = . + 0}, @samp{. = sym},
4945 @samp{. = . + sym} and @samp{. = ALIGN (. != 0, expr, 1)} when
4946 @samp{sym} is an absolute symbol of value 0 defined in the script.
4947 This allows you to force output of an empty section with @samp{. = .}.
4949 The linker will ignore address assignments (@pxref{Output Section Address})
4950 on discarded output sections, except when the linker script defines
4951 symbols in the output section. In that case the linker will obey
4952 the address assignments, possibly advancing dot even though the
4953 section is discarded.
4956 The special output section name @samp{/DISCARD/} may be used to discard
4957 input sections. Any input sections which are assigned to an output
4958 section named @samp{/DISCARD/} are not included in the output file.
4960 @node Output Section Attributes
4961 @subsection Output Section Attributes
4962 @cindex output section attributes
4963 We showed above that the full description of an output section looked
4968 @var{section} [@var{address}] [(@var{type})] :
4970 [ALIGN(@var{section_align})]
4971 [SUBALIGN(@var{subsection_align})]
4974 @var{output-section-command}
4975 @var{output-section-command}
4977 @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4981 We've already described @var{section}, @var{address}, and
4982 @var{output-section-command}. In this section we will describe the
4983 remaining section attributes.
4986 * Output Section Type:: Output section type
4987 * Output Section LMA:: Output section LMA
4988 * Forced Output Alignment:: Forced Output Alignment
4989 * Forced Input Alignment:: Forced Input Alignment
4990 * Output Section Constraint:: Output section constraint
4991 * Output Section Region:: Output section region
4992 * Output Section Phdr:: Output section phdr
4993 * Output Section Fill:: Output section fill
4996 @node Output Section Type
4997 @subsubsection Output Section Type
4998 Each output section may have a type. The type is a keyword in
4999 parentheses. The following types are defined:
5003 The section should be marked as not loadable, so that it will not be
5004 loaded into memory when the program is run.
5009 These type names are supported for backward compatibility, and are
5010 rarely used. They all have the same effect: the section should be
5011 marked as not allocatable, so that no memory is allocated for the
5012 section when the program is run.
5016 @cindex prevent unnecessary loading
5017 @cindex loading, preventing
5018 The linker normally sets the attributes of an output section based on
5019 the input sections which map into it. You can override this by using
5020 the section type. For example, in the script sample below, the
5021 @samp{ROM} section is addressed at memory location @samp{0} and does not
5022 need to be loaded when the program is run.
5026 ROM 0 (NOLOAD) : @{ @dots{} @}
5032 @node Output Section LMA
5033 @subsubsection Output Section LMA
5034 @kindex AT>@var{lma_region}
5035 @kindex AT(@var{lma})
5036 @cindex load address
5037 @cindex section load address
5038 Every section has a virtual address (VMA) and a load address (LMA); see
5039 @ref{Basic Script Concepts}. The virtual address is specified by the
5040 @pxref{Output Section Address} described earlier. The load address is
5041 specified by the @code{AT} or @code{AT>} keywords. Specifying a load
5042 address is optional.
5044 The @code{AT} keyword takes an expression as an argument. This
5045 specifies the exact load address of the section. The @code{AT>} keyword
5046 takes the name of a memory region as an argument. @xref{MEMORY}. The
5047 load address of the section is set to the next free address in the
5048 region, aligned to the section's alignment requirements.
5050 If neither @code{AT} nor @code{AT>} is specified for an allocatable
5051 section, the linker will use the following heuristic to determine the
5056 If the section has a specific VMA address, then this is used as
5057 the LMA address as well.
5060 If the section is not allocatable then its LMA is set to its VMA.
5063 Otherwise if a memory region can be found that is compatible
5064 with the current section, and this region contains at least one
5065 section, then the LMA is set so the difference between the
5066 VMA and LMA is the same as the difference between the VMA and LMA of
5067 the last section in the located region.
5070 If no memory regions have been declared then a default region
5071 that covers the entire address space is used in the previous step.
5074 If no suitable region could be found, or there was no previous
5075 section then the LMA is set equal to the VMA.
5078 @cindex ROM initialized data
5079 @cindex initialized data in ROM
5080 This feature is designed to make it easy to build a ROM image. For
5081 example, the following linker script creates three output sections: one
5082 called @samp{.text}, which starts at @code{0x1000}, one called
5083 @samp{.mdata}, which is loaded at the end of the @samp{.text} section
5084 even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
5085 uninitialized data at address @code{0x3000}. The symbol @code{_data} is
5086 defined with the value @code{0x2000}, which shows that the location
5087 counter holds the VMA value, not the LMA value.
5093 .text 0x1000 : @{ *(.text) _etext = . ; @}
5095 AT ( ADDR (.text) + SIZEOF (.text) )
5096 @{ _data = . ; *(.data); _edata = . ; @}
5098 @{ _bstart = . ; *(.bss) *(COMMON) ; _bend = . ;@}
5103 The run-time initialization code for use with a program generated with
5104 this linker script would include something like the following, to copy
5105 the initialized data from the ROM image to its runtime address. Notice
5106 how this code takes advantage of the symbols defined by the linker
5111 extern char _etext, _data, _edata, _bstart, _bend;
5112 char *src = &_etext;
5115 /* ROM has data at end of text; copy it. */
5116 while (dst < &_edata)
5120 for (dst = &_bstart; dst< &_bend; dst++)
5125 @node Forced Output Alignment
5126 @subsubsection Forced Output Alignment
5127 @kindex ALIGN(@var{section_align})
5128 @cindex forcing output section alignment
5129 @cindex output section alignment
5130 You can increase an output section's alignment by using ALIGN. As an
5131 alternative you can enforce that the difference between the VMA and LMA remains
5132 intact throughout this output section with the ALIGN_WITH_INPUT attribute.
5134 @node Forced Input Alignment
5135 @subsubsection Forced Input Alignment
5136 @kindex SUBALIGN(@var{subsection_align})
5137 @cindex forcing input section alignment
5138 @cindex input section alignment
5139 You can force input section alignment within an output section by using
5140 SUBALIGN. The value specified overrides any alignment given by input
5141 sections, whether larger or smaller.
5143 @node Output Section Constraint
5144 @subsubsection Output Section Constraint
5147 @cindex constraints on output sections
5148 You can specify that an output section should only be created if all
5149 of its input sections are read-only or all of its input sections are
5150 read-write by using the keyword @code{ONLY_IF_RO} and
5151 @code{ONLY_IF_RW} respectively.
5153 @node Output Section Region
5154 @subsubsection Output Section Region
5155 @kindex >@var{region}
5156 @cindex section, assigning to memory region
5157 @cindex memory regions and sections
5158 You can assign a section to a previously defined region of memory by
5159 using @samp{>@var{region}}. @xref{MEMORY}.
5161 Here is a simple example:
5164 MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
5165 SECTIONS @{ ROM : @{ *(.text) @} >rom @}
5169 @node Output Section Phdr
5170 @subsubsection Output Section Phdr
5172 @cindex section, assigning to program header
5173 @cindex program headers and sections
5174 You can assign a section to a previously defined program segment by
5175 using @samp{:@var{phdr}}. @xref{PHDRS}. If a section is assigned to
5176 one or more segments, then all subsequent allocated sections will be
5177 assigned to those segments as well, unless they use an explicitly
5178 @code{:@var{phdr}} modifier. You can use @code{:NONE} to tell the
5179 linker to not put the section in any segment at all.
5181 Here is a simple example:
5184 PHDRS @{ text PT_LOAD ; @}
5185 SECTIONS @{ .text : @{ *(.text) @} :text @}
5189 @node Output Section Fill
5190 @subsubsection Output Section Fill
5191 @kindex =@var{fillexp}
5192 @cindex section fill pattern
5193 @cindex fill pattern, entire section
5194 You can set the fill pattern for an entire section by using
5195 @samp{=@var{fillexp}}. @var{fillexp} is an expression
5196 (@pxref{Expressions}). Any otherwise unspecified regions of memory
5197 within the output section (for example, gaps left due to the required
5198 alignment of input sections) will be filled with the value, repeated as
5199 necessary. If the fill expression is a simple hex number, ie. a string
5200 of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
5201 an arbitrarily long sequence of hex digits can be used to specify the
5202 fill pattern; Leading zeros become part of the pattern too. For all
5203 other cases, including extra parentheses or a unary @code{+}, the fill
5204 pattern is the four least significant bytes of the value of the
5205 expression. In all cases, the number is big-endian.
5207 You can also change the fill value with a @code{FILL} command in the
5208 output section commands; (@pxref{Output Section Data}).
5210 Here is a simple example:
5213 SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
5217 @node Overlay Description
5218 @subsection Overlay Description
5221 An overlay description provides an easy way to describe sections which
5222 are to be loaded as part of a single memory image but are to be run at
5223 the same memory address. At run time, some sort of overlay manager will
5224 copy the overlaid sections in and out of the runtime memory address as
5225 required, perhaps by simply manipulating addressing bits. This approach
5226 can be useful, for example, when a certain region of memory is faster
5229 Overlays are described using the @code{OVERLAY} command. The
5230 @code{OVERLAY} command is used within a @code{SECTIONS} command, like an
5231 output section description. The full syntax of the @code{OVERLAY}
5232 command is as follows:
5235 OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
5239 @var{output-section-command}
5240 @var{output-section-command}
5242 @} [:@var{phdr}@dots{}] [=@var{fill}]
5245 @var{output-section-command}
5246 @var{output-section-command}
5248 @} [:@var{phdr}@dots{}] [=@var{fill}]
5250 @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}] [,]
5254 Everything is optional except @code{OVERLAY} (a keyword), and each
5255 section must have a name (@var{secname1} and @var{secname2} above). The
5256 section definitions within the @code{OVERLAY} construct are identical to
5257 those within the general @code{SECTIONS} construct (@pxref{SECTIONS}),
5258 except that no addresses and no memory regions may be defined for
5259 sections within an @code{OVERLAY}.
5261 The comma at the end may be required if a @var{fill} is used and
5262 the next @var{sections-command} looks like a continuation of the expression.
5264 The sections are all defined with the same starting address. The load
5265 addresses of the sections are arranged such that they are consecutive in
5266 memory starting at the load address used for the @code{OVERLAY} as a
5267 whole (as with normal section definitions, the load address is optional,
5268 and defaults to the start address; the start address is also optional,
5269 and defaults to the current value of the location counter).
5271 If the @code{NOCROSSREFS} keyword is used, and there are any
5272 references among the sections, the linker will report an error. Since
5273 the sections all run at the same address, it normally does not make
5274 sense for one section to refer directly to another.
5275 @xref{Miscellaneous Commands, NOCROSSREFS}.
5277 For each section within the @code{OVERLAY}, the linker automatically
5278 provides two symbols. The symbol @code{__load_start_@var{secname}} is
5279 defined as the starting load address of the section. The symbol
5280 @code{__load_stop_@var{secname}} is defined as the final load address of
5281 the section. Any characters within @var{secname} which are not legal
5282 within C identifiers are removed. C (or assembler) code may use these
5283 symbols to move the overlaid sections around as necessary.
5285 At the end of the overlay, the value of the location counter is set to
5286 the start address of the overlay plus the size of the largest section.
5288 Here is an example. Remember that this would appear inside a
5289 @code{SECTIONS} construct.
5292 OVERLAY 0x1000 : AT (0x4000)
5294 .text0 @{ o1/*.o(.text) @}
5295 .text1 @{ o2/*.o(.text) @}
5300 This will define both @samp{.text0} and @samp{.text1} to start at
5301 address 0x1000. @samp{.text0} will be loaded at address 0x4000, and
5302 @samp{.text1} will be loaded immediately after @samp{.text0}. The
5303 following symbols will be defined if referenced: @code{__load_start_text0},
5304 @code{__load_stop_text0}, @code{__load_start_text1},
5305 @code{__load_stop_text1}.
5307 C code to copy overlay @code{.text1} into the overlay area might look
5312 extern char __load_start_text1, __load_stop_text1;
5313 memcpy ((char *) 0x1000, &__load_start_text1,
5314 &__load_stop_text1 - &__load_start_text1);
5318 Note that the @code{OVERLAY} command is just syntactic sugar, since
5319 everything it does can be done using the more basic commands. The above
5320 example could have been written identically as follows.
5324 .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
5325 PROVIDE (__load_start_text0 = LOADADDR (.text0));
5326 PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
5327 .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
5328 PROVIDE (__load_start_text1 = LOADADDR (.text1));
5329 PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
5330 . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
5335 @section MEMORY Command
5337 @cindex memory regions
5338 @cindex regions of memory
5339 @cindex allocating memory
5340 @cindex discontinuous memory
5341 The linker's default configuration permits allocation of all available
5342 memory. You can override this by using the @code{MEMORY} command.
5344 The @code{MEMORY} command describes the location and size of blocks of
5345 memory in the target. You can use it to describe which memory regions
5346 may be used by the linker, and which memory regions it must avoid. You
5347 can then assign sections to particular memory regions. The linker will
5348 set section addresses based on the memory regions, and will warn about
5349 regions that become too full. The linker will not shuffle sections
5350 around to fit into the available regions.
5352 A linker script may contain many uses of the @code{MEMORY} command,
5353 however, all memory blocks defined are treated as if they were
5354 specified inside a single @code{MEMORY} command. The syntax for
5360 @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
5366 The @var{name} is a name used in the linker script to refer to the
5367 region. The region name has no meaning outside of the linker script.
5368 Region names are stored in a separate name space, and will not conflict
5369 with symbol names, file names, or section names. Each memory region
5370 must have a distinct name within the @code{MEMORY} command. However you can
5371 add later alias names to existing memory regions with the @ref{REGION_ALIAS}
5374 @cindex memory region attributes
5375 The @var{attr} string is an optional list of attributes that specify
5376 whether to use a particular memory region for an input section which is
5377 not explicitly mapped in the linker script. As described in
5378 @ref{SECTIONS}, if you do not specify an output section for some input
5379 section, the linker will create an output section with the same name as
5380 the input section. If you define region attributes, the linker will use
5381 them to select the memory region for the output section that it creates.
5383 The @var{attr} string must consist only of the following characters:
5398 Invert the sense of any of the attributes that follow
5401 If an unmapped section matches any of the listed attributes other than
5402 @samp{!}, it will be placed in the memory region. The @samp{!}
5403 attribute reverses the test for the characters that follow, so that an
5404 unmapped section will be placed in the memory region only if it does
5405 not match any of the attributes listed afterwards. Thus an attribute
5406 string of @samp{RW!X} will match any unmapped section that has either
5407 or both of the @samp{R} and @samp{W} attributes, but only as long as
5408 the section does not also have the @samp{X} attribute.
5413 The @var{origin} is an numerical expression for the start address of
5414 the memory region. The expression must evaluate to a constant and it
5415 cannot involve any symbols. The keyword @code{ORIGIN} may be
5416 abbreviated to @code{org} or @code{o} (but not, for example,
5422 The @var{len} is an expression for the size in bytes of the memory
5423 region. As with the @var{origin} expression, the expression must
5424 be numerical only and must evaluate to a constant. The keyword
5425 @code{LENGTH} may be abbreviated to @code{len} or @code{l}.
5427 In the following example, we specify that there are two memory regions
5428 available for allocation: one starting at @samp{0} for 256 kilobytes,
5429 and the other starting at @samp{0x40000000} for four megabytes. The
5430 linker will place into the @samp{rom} memory region every section which
5431 is not explicitly mapped into a memory region, and is either read-only
5432 or executable. The linker will place other sections which are not
5433 explicitly mapped into a memory region into the @samp{ram} memory
5440 rom (rx) : ORIGIN = 0, LENGTH = 256K
5441 ram (!rx) : org = 0x40000000, l = 4M
5446 Once you define a memory region, you can direct the linker to place
5447 specific output sections into that memory region by using the
5448 @samp{>@var{region}} output section attribute. For example, if you have
5449 a memory region named @samp{mem}, you would use @samp{>mem} in the
5450 output section definition. @xref{Output Section Region}. If no address
5451 was specified for the output section, the linker will set the address to
5452 the next available address within the memory region. If the combined
5453 output sections directed to a memory region are too large for the
5454 region, the linker will issue an error message.
5456 It is possible to access the origin and length of a memory in an
5457 expression via the @code{ORIGIN(@var{memory})} and
5458 @code{LENGTH(@var{memory})} functions:
5462 _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
5467 @section PHDRS Command
5469 @cindex program headers
5470 @cindex ELF program headers
5471 @cindex program segments
5472 @cindex segments, ELF
5473 The ELF object file format uses @dfn{program headers}, also knows as
5474 @dfn{segments}. The program headers describe how the program should be
5475 loaded into memory. You can print them out by using the @code{objdump}
5476 program with the @samp{-p} option.
5478 When you run an ELF program on a native ELF system, the system loader
5479 reads the program headers in order to figure out how to load the
5480 program. This will only work if the program headers are set correctly.
5481 This manual does not describe the details of how the system loader
5482 interprets program headers; for more information, see the ELF ABI.
5484 The linker will create reasonable program headers by default. However,
5485 in some cases, you may need to specify the program headers more
5486 precisely. You may use the @code{PHDRS} command for this purpose. When
5487 the linker sees the @code{PHDRS} command in the linker script, it will
5488 not create any program headers other than the ones specified.
5490 The linker only pays attention to the @code{PHDRS} command when
5491 generating an ELF output file. In other cases, the linker will simply
5492 ignore @code{PHDRS}.
5494 This is the syntax of the @code{PHDRS} command. The words @code{PHDRS},
5495 @code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
5501 @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
5502 [ FLAGS ( @var{flags} ) ] ;
5507 The @var{name} is used only for reference in the @code{SECTIONS} command
5508 of the linker script. It is not put into the output file. Program
5509 header names are stored in a separate name space, and will not conflict
5510 with symbol names, file names, or section names. Each program header
5511 must have a distinct name. The headers are processed in order and it
5512 is usual for them to map to sections in ascending load address order.
5514 Certain program header types describe segments of memory which the
5515 system loader will load from the file. In the linker script, you
5516 specify the contents of these segments by placing allocatable output
5517 sections in the segments. You use the @samp{:@var{phdr}} output section
5518 attribute to place a section in a particular segment. @xref{Output
5521 It is normal to put certain sections in more than one segment. This
5522 merely implies that one segment of memory contains another. You may
5523 repeat @samp{:@var{phdr}}, using it once for each segment which should
5524 contain the section.
5526 If you place a section in one or more segments using @samp{:@var{phdr}},
5527 then the linker will place all subsequent allocatable sections which do
5528 not specify @samp{:@var{phdr}} in the same segments. This is for
5529 convenience, since generally a whole set of contiguous sections will be
5530 placed in a single segment. You can use @code{:NONE} to override the
5531 default segment and tell the linker to not put the section in any
5536 You may use the @code{FILEHDR} and @code{PHDRS} keywords after
5537 the program header type to further describe the contents of the segment.
5538 The @code{FILEHDR} keyword means that the segment should include the ELF
5539 file header. The @code{PHDRS} keyword means that the segment should
5540 include the ELF program headers themselves. If applied to a loadable
5541 segment (@code{PT_LOAD}), all prior loadable segments must have one of
5544 The @var{type} may be one of the following. The numbers indicate the
5545 value of the keyword.
5548 @item @code{PT_NULL} (0)
5549 Indicates an unused program header.
5551 @item @code{PT_LOAD} (1)
5552 Indicates that this program header describes a segment to be loaded from
5555 @item @code{PT_DYNAMIC} (2)
5556 Indicates a segment where dynamic linking information can be found.
5558 @item @code{PT_INTERP} (3)
5559 Indicates a segment where the name of the program interpreter may be
5562 @item @code{PT_NOTE} (4)
5563 Indicates a segment holding note information.
5565 @item @code{PT_SHLIB} (5)
5566 A reserved program header type, defined but not specified by the ELF
5569 @item @code{PT_PHDR} (6)
5570 Indicates a segment where the program headers may be found.
5572 @item @code{PT_TLS} (7)
5573 Indicates a segment containing thread local storage.
5575 @item @var{expression}
5576 An expression giving the numeric type of the program header. This may
5577 be used for types not defined above.
5580 You can specify that a segment should be loaded at a particular address
5581 in memory by using an @code{AT} expression. This is identical to the
5582 @code{AT} command used as an output section attribute (@pxref{Output
5583 Section LMA}). The @code{AT} command for a program header overrides the
5584 output section attribute.
5586 The linker will normally set the segment flags based on the sections
5587 which comprise the segment. You may use the @code{FLAGS} keyword to
5588 explicitly specify the segment flags. The value of @var{flags} must be
5589 an integer. It is used to set the @code{p_flags} field of the program
5592 Here is an example of @code{PHDRS}. This shows a typical set of program
5593 headers used on a native ELF system.
5599 headers PT_PHDR PHDRS ;
5601 text PT_LOAD FILEHDR PHDRS ;
5603 dynamic PT_DYNAMIC ;
5609 .interp : @{ *(.interp) @} :text :interp
5610 .text : @{ *(.text) @} :text
5611 .rodata : @{ *(.rodata) @} /* defaults to :text */
5613 . = . + 0x1000; /* move to a new page in memory */
5614 .data : @{ *(.data) @} :data
5615 .dynamic : @{ *(.dynamic) @} :data :dynamic
5622 @section VERSION Command
5623 @kindex VERSION @{script text@}
5624 @cindex symbol versions
5625 @cindex version script
5626 @cindex versions of symbols
5627 The linker supports symbol versions when using ELF. Symbol versions are
5628 only useful when using shared libraries. The dynamic linker can use
5629 symbol versions to select a specific version of a function when it runs
5630 a program that may have been linked against an earlier version of the
5633 You can include a version script directly in the main linker script, or
5634 you can supply the version script as an implicit linker script. You can
5635 also use the @samp{--version-script} linker option.
5637 The syntax of the @code{VERSION} command is simply
5639 VERSION @{ version-script-commands @}
5642 The format of the version script commands is identical to that used by
5643 Sun's linker in Solaris 2.5. The version script defines a tree of
5644 version nodes. You specify the node names and interdependencies in the
5645 version script. You can specify which symbols are bound to which
5646 version nodes, and you can reduce a specified set of symbols to local
5647 scope so that they are not globally visible outside of the shared
5650 The easiest way to demonstrate the version script language is with a few
5676 This example version script defines three version nodes. The first
5677 version node defined is @samp{VERS_1.1}; it has no other dependencies.
5678 The script binds the symbol @samp{foo1} to @samp{VERS_1.1}. It reduces
5679 a number of symbols to local scope so that they are not visible outside
5680 of the shared library; this is done using wildcard patterns, so that any
5681 symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
5682 is matched. The wildcard patterns available are the same as those used
5683 in the shell when matching filenames (also known as ``globbing'').
5684 However, if you specify the symbol name inside double quotes, then the
5685 name is treated as literal, rather than as a glob pattern.
5687 Next, the version script defines node @samp{VERS_1.2}. This node
5688 depends upon @samp{VERS_1.1}. The script binds the symbol @samp{foo2}
5689 to the version node @samp{VERS_1.2}.
5691 Finally, the version script defines node @samp{VERS_2.0}. This node
5692 depends upon @samp{VERS_1.2}. The scripts binds the symbols @samp{bar1}
5693 and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
5695 When the linker finds a symbol defined in a library which is not
5696 specifically bound to a version node, it will effectively bind it to an
5697 unspecified base version of the library. You can bind all otherwise
5698 unspecified symbols to a given version node by using @samp{global: *;}
5699 somewhere in the version script. Note that it's slightly crazy to use
5700 wildcards in a global spec except on the last version node. Global
5701 wildcards elsewhere run the risk of accidentally adding symbols to the
5702 set exported for an old version. That's wrong since older versions
5703 ought to have a fixed set of symbols.
5705 The names of the version nodes have no specific meaning other than what
5706 they might suggest to the person reading them. The @samp{2.0} version
5707 could just as well have appeared in between @samp{1.1} and @samp{1.2}.
5708 However, this would be a confusing way to write a version script.
5710 Node name can be omitted, provided it is the only version node
5711 in the version script. Such version script doesn't assign any versions to
5712 symbols, only selects which symbols will be globally visible out and which
5716 @{ global: foo; bar; local: *; @};
5719 When you link an application against a shared library that has versioned
5720 symbols, the application itself knows which version of each symbol it
5721 requires, and it also knows which version nodes it needs from each
5722 shared library it is linked against. Thus at runtime, the dynamic
5723 loader can make a quick check to make sure that the libraries you have
5724 linked against do in fact supply all of the version nodes that the
5725 application will need to resolve all of the dynamic symbols. In this
5726 way it is possible for the dynamic linker to know with certainty that
5727 all external symbols that it needs will be resolvable without having to
5728 search for each symbol reference.
5730 The symbol versioning is in effect a much more sophisticated way of
5731 doing minor version checking that SunOS does. The fundamental problem
5732 that is being addressed here is that typically references to external
5733 functions are bound on an as-needed basis, and are not all bound when
5734 the application starts up. If a shared library is out of date, a
5735 required interface may be missing; when the application tries to use
5736 that interface, it may suddenly and unexpectedly fail. With symbol
5737 versioning, the user will get a warning when they start their program if
5738 the libraries being used with the application are too old.
5740 There are several GNU extensions to Sun's versioning approach. The
5741 first of these is the ability to bind a symbol to a version node in the
5742 source file where the symbol is defined instead of in the versioning
5743 script. This was done mainly to reduce the burden on the library
5744 maintainer. You can do this by putting something like:
5746 __asm__(".symver original_foo,foo@@VERS_1.1");
5749 in the C source file. This renames the function @samp{original_foo} to
5750 be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
5751 The @samp{local:} directive can be used to prevent the symbol
5752 @samp{original_foo} from being exported. A @samp{.symver} directive
5753 takes precedence over a version script.
5755 The second GNU extension is to allow multiple versions of the same
5756 function to appear in a given shared library. In this way you can make
5757 an incompatible change to an interface without increasing the major
5758 version number of the shared library, while still allowing applications
5759 linked against the old interface to continue to function.
5761 To do this, you must use multiple @samp{.symver} directives in the
5762 source file. Here is an example:
5765 __asm__(".symver original_foo,foo@@");
5766 __asm__(".symver old_foo,foo@@VERS_1.1");
5767 __asm__(".symver old_foo1,foo@@VERS_1.2");
5768 __asm__(".symver new_foo,foo@@@@VERS_2.0");
5771 In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
5772 unspecified base version of the symbol. The source file that contains this
5773 example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
5774 @samp{old_foo1}, and @samp{new_foo}.
5776 When you have multiple definitions of a given symbol, there needs to be
5777 some way to specify a default version to which external references to
5778 this symbol will be bound. You can do this with the
5779 @samp{foo@@@@VERS_2.0} type of @samp{.symver} directive. You can only
5780 declare one version of a symbol as the default in this manner; otherwise
5781 you would effectively have multiple definitions of the same symbol.
5783 If you wish to bind a reference to a specific version of the symbol
5784 within the shared library, you can use the aliases of convenience
5785 (i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
5786 specifically bind to an external version of the function in question.
5788 You can also specify the language in the version script:
5791 VERSION extern "lang" @{ version-script-commands @}
5794 The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
5795 The linker will iterate over the list of symbols at the link time and
5796 demangle them according to @samp{lang} before matching them to the
5797 patterns specified in @samp{version-script-commands}. The default
5798 @samp{lang} is @samp{C}.
5800 Demangled names may contains spaces and other special characters. As
5801 described above, you can use a glob pattern to match demangled names,
5802 or you can use a double-quoted string to match the string exactly. In
5803 the latter case, be aware that minor differences (such as differing
5804 whitespace) between the version script and the demangler output will
5805 cause a mismatch. As the exact string generated by the demangler
5806 might change in the future, even if the mangled name does not, you
5807 should check that all of your version directives are behaving as you
5808 expect when you upgrade.
5811 @section Expressions in Linker Scripts
5814 The syntax for expressions in the linker script language is identical to
5815 that of C expressions. All expressions are evaluated as integers. All
5816 expressions are evaluated in the same size, which is 32 bits if both the
5817 host and target are 32 bits, and is otherwise 64 bits.
5819 You can use and set symbol values in expressions.
5821 The linker defines several special purpose builtin functions for use in
5825 * Constants:: Constants
5826 * Symbolic Constants:: Symbolic constants
5827 * Symbols:: Symbol Names
5828 * Orphan Sections:: Orphan Sections
5829 * Location Counter:: The Location Counter
5830 * Operators:: Operators
5831 * Evaluation:: Evaluation
5832 * Expression Section:: The Section of an Expression
5833 * Builtin Functions:: Builtin Functions
5837 @subsection Constants
5838 @cindex integer notation
5839 @cindex constants in linker scripts
5840 All constants are integers.
5842 As in C, the linker considers an integer beginning with @samp{0} to be
5843 octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5844 hexadecimal. Alternatively the linker accepts suffixes of @samp{h} or
5845 @samp{H} for hexadecimal, @samp{o} or @samp{O} for octal, @samp{b} or
5846 @samp{B} for binary and @samp{d} or @samp{D} for decimal. Any integer
5847 value without a prefix or a suffix is considered to be decimal.
5849 @cindex scaled integers
5850 @cindex K and M integer suffixes
5851 @cindex M and K integer suffixes
5852 @cindex suffixes for integers
5853 @cindex integer suffixes
5854 In addition, you can use the suffixes @code{K} and @code{M} to scale a
5858 @c END TEXI2ROFF-KILL
5859 @code{1024} or @code{1024*1024}
5863 ${\rm 1024}$ or ${\rm 1024}^2$
5865 @c END TEXI2ROFF-KILL
5866 respectively. For example, the following
5867 all refer to the same quantity:
5876 Note - the @code{K} and @code{M} suffixes cannot be used in
5877 conjunction with the base suffixes mentioned above.
5879 @node Symbolic Constants
5880 @subsection Symbolic Constants
5881 @cindex symbolic constants
5883 It is possible to refer to target specific constants via the use of
5884 the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5889 The target's maximum page size.
5891 @item COMMONPAGESIZE
5892 @kindex COMMONPAGESIZE
5893 The target's default page size.
5899 .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5902 will create a text section aligned to the largest page boundary
5903 supported by the target.
5906 @subsection Symbol Names
5907 @cindex symbol names
5909 @cindex quoted symbol names
5911 Unless quoted, symbol names start with a letter, underscore, or period
5912 and may include letters, digits, underscores, periods, and hyphens.
5913 Unquoted symbol names must not conflict with any keywords. You can
5914 specify a symbol which contains odd characters or has the same name as a
5915 keyword by surrounding the symbol name in double quotes:
5918 "with a space" = "also with a space" + 10;
5921 Since symbols can contain many non-alphabetic characters, it is safest
5922 to delimit symbols with spaces. For example, @samp{A-B} is one symbol,
5923 whereas @samp{A - B} is an expression involving subtraction.
5925 @node Orphan Sections
5926 @subsection Orphan Sections
5928 Orphan sections are sections present in the input files which
5929 are not explicitly placed into the output file by the linker
5930 script. The linker will still copy these sections into the
5931 output file by either finding, or creating a suitable output section
5932 in which to place the orphaned input section.
5934 If the name of an orphaned input section exactly matches the name of
5935 an existing output section, then the orphaned input section will be
5936 placed at the end of that output section.
5938 If there is no output section with a matching name then new output
5939 sections will be created. Each new output section will have the same
5940 name as the orphan section placed within it. If there are multiple
5941 orphan sections with the same name, these will all be combined into
5942 one new output section.
5944 If new output sections are created to hold orphaned input sections,
5945 then the linker must decide where to place these new output sections
5946 in relation to existing output sections. On most modern targets, the
5947 linker attempts to place orphan sections after sections of the same
5948 attribute, such as code vs data, loadable vs non-loadable, etc. If no
5949 sections with matching attributes are found, or your target lacks this
5950 support, the orphan section is placed at the end of the file.
5952 The command-line options @samp{--orphan-handling} and @samp{--unique}
5953 (@pxref{Options,,Command-line Options}) can be used to control which
5954 output sections an orphan is placed in.
5956 @node Location Counter
5957 @subsection The Location Counter
5960 @cindex location counter
5961 @cindex current output location
5962 The special linker variable @dfn{dot} @samp{.} always contains the
5963 current output location counter. Since the @code{.} always refers to a
5964 location in an output section, it may only appear in an expression
5965 within a @code{SECTIONS} command. The @code{.} symbol may appear
5966 anywhere that an ordinary symbol is allowed in an expression.
5969 Assigning a value to @code{.} will cause the location counter to be
5970 moved. This may be used to create holes in the output section. The
5971 location counter may not be moved backwards inside an output section,
5972 and may not be moved backwards outside of an output section if so
5973 doing creates areas with overlapping LMAs.
5989 In the previous example, the @samp{.text} section from @file{file1} is
5990 located at the beginning of the output section @samp{output}. It is
5991 followed by a 1000 byte gap. Then the @samp{.text} section from
5992 @file{file2} appears, also with a 1000 byte gap following before the
5993 @samp{.text} section from @file{file3}. The notation @samp{= 0x12345678}
5994 specifies what data to write in the gaps (@pxref{Output Section Fill}).
5996 @cindex dot inside sections
5997 Note: @code{.} actually refers to the byte offset from the start of the
5998 current containing object. Normally this is the @code{SECTIONS}
5999 statement, whose start address is 0, hence @code{.} can be used as an
6000 absolute address. If @code{.} is used inside a section description
6001 however, it refers to the byte offset from the start of that section,
6002 not an absolute address. Thus in a script like this:
6020 The @samp{.text} section will be assigned a starting address of 0x100
6021 and a size of exactly 0x200 bytes, even if there is not enough data in
6022 the @samp{.text} input sections to fill this area. (If there is too
6023 much data, an error will be produced because this would be an attempt to
6024 move @code{.} backwards). The @samp{.data} section will start at 0x500
6025 and it will have an extra 0x600 bytes worth of space after the end of
6026 the values from the @samp{.data} input sections and before the end of
6027 the @samp{.data} output section itself.
6029 @cindex dot outside sections
6030 Setting symbols to the value of the location counter outside of an
6031 output section statement can result in unexpected values if the linker
6032 needs to place orphan sections. For example, given the following:
6038 .text: @{ *(.text) @}
6042 .data: @{ *(.data) @}
6047 If the linker needs to place some input section, e.g. @code{.rodata},
6048 not mentioned in the script, it might choose to place that section
6049 between @code{.text} and @code{.data}. You might think the linker
6050 should place @code{.rodata} on the blank line in the above script, but
6051 blank lines are of no particular significance to the linker. As well,
6052 the linker doesn't associate the above symbol names with their
6053 sections. Instead, it assumes that all assignments or other
6054 statements belong to the previous output section, except for the
6055 special case of an assignment to @code{.}. I.e., the linker will
6056 place the orphan @code{.rodata} section as if the script was written
6063 .text: @{ *(.text) @}
6067 .rodata: @{ *(.rodata) @}
6068 .data: @{ *(.data) @}
6073 This may or may not be the script author's intention for the value of
6074 @code{start_of_data}. One way to influence the orphan section
6075 placement is to assign the location counter to itself, as the linker
6076 assumes that an assignment to @code{.} is setting the start address of
6077 a following output section and thus should be grouped with that
6078 section. So you could write:
6084 .text: @{ *(.text) @}
6089 .data: @{ *(.data) @}
6094 Now, the orphan @code{.rodata} section will be placed between
6095 @code{end_of_text} and @code{start_of_data}.
6099 @subsection Operators
6100 @cindex operators for arithmetic
6101 @cindex arithmetic operators
6102 @cindex precedence in expressions
6103 The linker recognizes the standard C set of arithmetic operators, with
6104 the standard bindings and precedence levels:
6107 @c END TEXI2ROFF-KILL
6109 precedence associativity Operators Notes
6115 5 left == != > < <= >=
6121 11 right &= += -= *= /= (2)
6125 (1) Prefix operators
6126 (2) @xref{Assignments}.
6130 \vskip \baselineskip
6131 %"lispnarrowing" is the extra indent used generally for smallexample
6132 \hskip\lispnarrowing\vbox{\offinterlineskip
6135 {\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
6136 height2pt&\omit&&\omit&&\omit&\cr
6137 &Precedence&& Associativity &&{\rm Operators}&\cr
6138 height2pt&\omit&&\omit&&\omit&\cr
6140 height2pt&\omit&&\omit&&\omit&\cr
6142 % '176 is tilde, '~' in tt font
6143 &1&&left&&\qquad- \char'176\ !\qquad\dag&\cr
6144 &2&&left&&* / \%&\cr
6147 &5&&left&&== != > < <= >=&\cr
6150 &8&&left&&{\&\&}&\cr
6153 &11&&right&&\qquad\&= += -= *= /=\qquad\ddag&\cr
6155 height2pt&\omit&&\omit&&\omit&\cr}
6160 @obeylines@parskip=0pt@parindent=0pt
6161 @dag@quad Prefix operators.
6162 @ddag@quad @xref{Assignments}.
6165 @c END TEXI2ROFF-KILL
6168 @subsection Evaluation
6169 @cindex lazy evaluation
6170 @cindex expression evaluation order
6171 The linker evaluates expressions lazily. It only computes the value of
6172 an expression when absolutely necessary.
6174 The linker needs some information, such as the value of the start
6175 address of the first section, and the origins and lengths of memory
6176 regions, in order to do any linking at all. These values are computed
6177 as soon as possible when the linker reads in the linker script.
6179 However, other values (such as symbol values) are not known or needed
6180 until after storage allocation. Such values are evaluated later, when
6181 other information (such as the sizes of output sections) is available
6182 for use in the symbol assignment expression.
6184 The sizes of sections cannot be known until after allocation, so
6185 assignments dependent upon these are not performed until after
6188 Some expressions, such as those depending upon the location counter
6189 @samp{.}, must be evaluated during section allocation.
6191 If the result of an expression is required, but the value is not
6192 available, then an error results. For example, a script like the
6198 .text 9+this_isnt_constant :
6204 will cause the error message @samp{non constant expression for initial
6207 @node Expression Section
6208 @subsection The Section of an Expression
6209 @cindex expression sections
6210 @cindex absolute expressions
6211 @cindex relative expressions
6212 @cindex absolute and relocatable symbols
6213 @cindex relocatable and absolute symbols
6214 @cindex symbols, relocatable and absolute
6215 Addresses and symbols may be section relative, or absolute. A section
6216 relative symbol is relocatable. If you request relocatable output
6217 using the @samp{-r} option, a further link operation may change the
6218 value of a section relative symbol. On the other hand, an absolute
6219 symbol will retain the same value throughout any further link
6222 Some terms in linker expressions are addresses. This is true of
6223 section relative symbols and for builtin functions that return an
6224 address, such as @code{ADDR}, @code{LOADADDR}, @code{ORIGIN} and
6225 @code{SEGMENT_START}. Other terms are simply numbers, or are builtin
6226 functions that return a non-address value, such as @code{LENGTH}.
6227 One complication is that unless you set @code{LD_FEATURE ("SANE_EXPR")}
6228 (@pxref{Miscellaneous Commands}), numbers and absolute symbols are treated
6229 differently depending on their location, for compatibility with older
6230 versions of @code{ld}. Expressions appearing outside an output
6231 section definition treat all numbers as absolute addresses.
6232 Expressions appearing inside an output section definition treat
6233 absolute symbols as numbers. If @code{LD_FEATURE ("SANE_EXPR")} is
6234 given, then absolute symbols and numbers are simply treated as numbers
6237 In the following simple example,
6244 __executable_start = 0x100;
6248 __data_start = 0x10;
6256 both @code{.} and @code{__executable_start} are set to the absolute
6257 address 0x100 in the first two assignments, then both @code{.} and
6258 @code{__data_start} are set to 0x10 relative to the @code{.data}
6259 section in the second two assignments.
6261 For expressions involving numbers, relative addresses and absolute
6262 addresses, ld follows these rules to evaluate terms:
6266 Unary operations on an absolute address or number, and binary
6267 operations on two absolute addresses or two numbers, or between one
6268 absolute address and a number, apply the operator to the value(s).
6270 Unary operations on a relative address, and binary operations on two
6271 relative addresses in the same section or between one relative address
6272 and a number, apply the operator to the offset part of the address(es).
6274 Other binary operations, that is, between two relative addresses not
6275 in the same section, or between a relative address and an absolute
6276 address, first convert any non-absolute term to an absolute address
6277 before applying the operator.
6280 The result section of each sub-expression is as follows:
6284 An operation involving only numbers results in a number.
6286 The result of comparisons, @samp{&&} and @samp{||} is also a number.
6288 The result of other binary arithmetic and logical operations on two
6289 relative addresses in the same section or two absolute addresses
6290 (after above conversions) is also a number when
6291 @code{LD_FEATURE ("SANE_EXPR")} or inside an output section definition
6292 but an absolute address otherwise.
6294 The result of other operations on relative addresses or one
6295 relative address and a number, is a relative address in the same
6296 section as the relative operand(s).
6298 The result of other operations on absolute addresses (after above
6299 conversions) is an absolute address.
6302 You can use the builtin function @code{ABSOLUTE} to force an expression
6303 to be absolute when it would otherwise be relative. For example, to
6304 create an absolute symbol set to the address of the end of the output
6305 section @samp{.data}:
6309 .data : @{ *(.data) _edata = ABSOLUTE(.); @}
6313 If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
6314 @samp{.data} section.
6316 Using @code{LOADADDR} also forces an expression absolute, since this
6317 particular builtin function returns an absolute address.
6319 @node Builtin Functions
6320 @subsection Builtin Functions
6321 @cindex functions in expressions
6322 The linker script language includes a number of builtin functions for
6323 use in linker script expressions.
6326 @item ABSOLUTE(@var{exp})
6327 @kindex ABSOLUTE(@var{exp})
6328 @cindex expression, absolute
6329 Return the absolute (non-relocatable, as opposed to non-negative) value
6330 of the expression @var{exp}. Primarily useful to assign an absolute
6331 value to a symbol within a section definition, where symbol values are
6332 normally section relative. @xref{Expression Section}.
6334 @item ADDR(@var{section})
6335 @kindex ADDR(@var{section})
6336 @cindex section address in expression
6337 Return the address (VMA) of the named @var{section}. Your
6338 script must previously have defined the location of that section. In
6339 the following example, @code{start_of_output_1}, @code{symbol_1} and
6340 @code{symbol_2} are assigned equivalent values, except that
6341 @code{symbol_1} will be relative to the @code{.output1} section while
6342 the other two will be absolute:
6348 start_of_output_1 = ABSOLUTE(.);
6353 symbol_1 = ADDR(.output1);
6354 symbol_2 = start_of_output_1;
6360 @item ALIGN(@var{align})
6361 @itemx ALIGN(@var{exp},@var{align})
6362 @kindex ALIGN(@var{align})
6363 @kindex ALIGN(@var{exp},@var{align})
6364 @cindex round up location counter
6365 @cindex align location counter
6366 @cindex round up expression
6367 @cindex align expression
6368 Return the location counter (@code{.}) or arbitrary expression aligned
6369 to the next @var{align} boundary. The single operand @code{ALIGN}
6370 doesn't change the value of the location counter---it just does
6371 arithmetic on it. The two operand @code{ALIGN} allows an arbitrary
6372 expression to be aligned upwards (@code{ALIGN(@var{align})} is
6373 equivalent to @code{ALIGN(ABSOLUTE(.), @var{align})}).
6375 Here is an example which aligns the output @code{.data} section to the
6376 next @code{0x2000} byte boundary after the preceding section and sets a
6377 variable within the section to the next @code{0x8000} boundary after the
6382 .data ALIGN(0x2000): @{
6384 variable = ALIGN(0x8000);
6390 The first use of @code{ALIGN} in this example specifies the location of
6391 a section because it is used as the optional @var{address} attribute of
6392 a section definition (@pxref{Output Section Address}). The second use
6393 of @code{ALIGN} is used to defines the value of a symbol.
6395 The builtin function @code{NEXT} is closely related to @code{ALIGN}.
6397 @item ALIGNOF(@var{section})
6398 @kindex ALIGNOF(@var{section})
6399 @cindex section alignment
6400 Return the alignment in bytes of the named @var{section}, if that section has
6401 been allocated. If the section has not been allocated when this is
6402 evaluated, the linker will report an error. In the following example,
6403 the alignment of the @code{.output} section is stored as the first
6404 value in that section.
6409 LONG (ALIGNOF (.output))
6416 @item BLOCK(@var{exp})
6417 @kindex BLOCK(@var{exp})
6418 This is a synonym for @code{ALIGN}, for compatibility with older linker
6419 scripts. It is most often seen when setting the address of an output
6422 @item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6423 @kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
6424 This is equivalent to either
6426 (ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
6430 (ALIGN(@var{maxpagesize})
6431 + ((. + @var{commonpagesize} - 1) & (@var{maxpagesize} - @var{commonpagesize})))
6434 depending on whether the latter uses fewer @var{commonpagesize} sized pages
6435 for the data segment (area between the result of this expression and
6436 @code{DATA_SEGMENT_END}) than the former or not.
6437 If the latter form is used, it means @var{commonpagesize} bytes of runtime
6438 memory will be saved at the expense of up to @var{commonpagesize} wasted
6439 bytes in the on-disk file.
6441 This expression can only be used directly in @code{SECTIONS} commands, not in
6442 any output section descriptions and only once in the linker script.
6443 @var{commonpagesize} should be less or equal to @var{maxpagesize} and should
6444 be the system page size the object wants to be optimized for while still
6445 running on system page sizes up to @var{maxpagesize}. Note however
6446 that @samp{-z relro} protection will not be effective if the system
6447 page size is larger than @var{commonpagesize}.
6452 . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
6455 @item DATA_SEGMENT_END(@var{exp})
6456 @kindex DATA_SEGMENT_END(@var{exp})
6457 This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
6458 evaluation purposes.
6461 . = DATA_SEGMENT_END(.);
6464 @item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6465 @kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
6466 This defines the end of the @code{PT_GNU_RELRO} segment when
6467 @samp{-z relro} option is used.
6468 When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
6469 does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
6470 @var{exp} + @var{offset} is aligned to the @var{commonpagesize}
6471 argument given to @code{DATA_SEGMENT_ALIGN}. If present in the linker
6472 script, it must be placed between @code{DATA_SEGMENT_ALIGN} and
6473 @code{DATA_SEGMENT_END}. Evaluates to the second argument plus any
6474 padding needed at the end of the @code{PT_GNU_RELRO} segment due to
6478 . = DATA_SEGMENT_RELRO_END(24, .);
6481 @item DEFINED(@var{symbol})
6482 @kindex DEFINED(@var{symbol})
6483 @cindex symbol defaults
6484 Return 1 if @var{symbol} is in the linker global symbol table and is
6485 defined before the statement using DEFINED in the script, otherwise
6486 return 0. You can use this function to provide
6487 default values for symbols. For example, the following script fragment
6488 shows how to set a global symbol @samp{begin} to the first location in
6489 the @samp{.text} section---but if a symbol called @samp{begin} already
6490 existed, its value is preserved:
6496 begin = DEFINED(begin) ? begin : . ;
6504 @item LENGTH(@var{memory})
6505 @kindex LENGTH(@var{memory})
6506 Return the length of the memory region named @var{memory}.
6508 @item LOADADDR(@var{section})
6509 @kindex LOADADDR(@var{section})
6510 @cindex section load address in expression
6511 Return the absolute LMA of the named @var{section}. (@pxref{Output
6514 @item LOG2CEIL(@var{exp})
6515 @kindex LOG2CEIL(@var{exp})
6516 Return the binary logarithm of @var{exp} rounded towards infinity.
6517 @code{LOG2CEIL(0)} returns 0.
6520 @item MAX(@var{exp1}, @var{exp2})
6521 Returns the maximum of @var{exp1} and @var{exp2}.
6524 @item MIN(@var{exp1}, @var{exp2})
6525 Returns the minimum of @var{exp1} and @var{exp2}.
6527 @item NEXT(@var{exp})
6528 @kindex NEXT(@var{exp})
6529 @cindex unallocated address, next
6530 Return the next unallocated address that is a multiple of @var{exp}.
6531 This function is closely related to @code{ALIGN(@var{exp})}; unless you
6532 use the @code{MEMORY} command to define discontinuous memory for the
6533 output file, the two functions are equivalent.
6535 @item ORIGIN(@var{memory})
6536 @kindex ORIGIN(@var{memory})
6537 Return the origin of the memory region named @var{memory}.
6539 @item SEGMENT_START(@var{segment}, @var{default})
6540 @kindex SEGMENT_START(@var{segment}, @var{default})
6541 Return the base address of the named @var{segment}. If an explicit
6542 value has already been given for this segment (with a command-line
6543 @samp{-T} option) then that value will be returned otherwise the value
6544 will be @var{default}. At present, the @samp{-T} command-line option
6545 can only be used to set the base address for the ``text'', ``data'', and
6546 ``bss'' sections, but you can use @code{SEGMENT_START} with any segment
6549 @item SIZEOF(@var{section})
6550 @kindex SIZEOF(@var{section})
6551 @cindex section size
6552 Return the size in bytes of the named @var{section}, if that section has
6553 been allocated. If the section has not been allocated when this is
6554 evaluated, the linker will report an error. In the following example,
6555 @code{symbol_1} and @code{symbol_2} are assigned identical values:
6564 symbol_1 = .end - .start ;
6565 symbol_2 = SIZEOF(.output);
6570 @item SIZEOF_HEADERS
6571 @itemx sizeof_headers
6572 @kindex SIZEOF_HEADERS
6574 Return the size in bytes of the output file's headers. This is
6575 information which appears at the start of the output file. You can use
6576 this number when setting the start address of the first section, if you
6577 choose, to facilitate paging.
6579 @cindex not enough room for program headers
6580 @cindex program headers, not enough room
6581 When producing an ELF output file, if the linker script uses the
6582 @code{SIZEOF_HEADERS} builtin function, the linker must compute the
6583 number of program headers before it has determined all the section
6584 addresses and sizes. If the linker later discovers that it needs
6585 additional program headers, it will report an error @samp{not enough
6586 room for program headers}. To avoid this error, you must avoid using
6587 the @code{SIZEOF_HEADERS} function, or you must rework your linker
6588 script to avoid forcing the linker to use additional program headers, or
6589 you must define the program headers yourself using the @code{PHDRS}
6590 command (@pxref{PHDRS}).
6593 @node Implicit Linker Scripts
6594 @section Implicit Linker Scripts
6595 @cindex implicit linker scripts
6596 If you specify a linker input file which the linker can not recognize as
6597 an object file or an archive file, it will try to read the file as a
6598 linker script. If the file can not be parsed as a linker script, the
6599 linker will report an error.
6601 An implicit linker script will not replace the default linker script.
6603 Typically an implicit linker script would contain only symbol
6604 assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
6607 Any input files read because of an implicit linker script will be read
6608 at the position in the command line where the implicit linker script was
6609 read. This can affect archive searching.
6612 @node Machine Dependent
6613 @chapter Machine Dependent Features
6615 @cindex machine dependencies
6616 @command{ld} has additional features on some platforms; the following
6617 sections describe them. Machines where @command{ld} has no additional
6618 functionality are not listed.
6622 * H8/300:: @command{ld} and the H8/300
6625 * M68HC11/68HC12:: @code{ld} and the Motorola 68HC11 and 68HC12 families
6628 * ARM:: @command{ld} and the ARM family
6631 * HPPA ELF32:: @command{ld} and HPPA 32-bit ELF
6634 * M68K:: @command{ld} and the Motorola 68K family
6637 * MIPS:: @command{ld} and the MIPS family
6640 * MMIX:: @command{ld} and MMIX
6643 * MSP430:: @command{ld} and MSP430
6646 * NDS32:: @command{ld} and NDS32
6649 * Nios II:: @command{ld} and the Altera Nios II
6652 * PowerPC ELF32:: @command{ld} and PowerPC 32-bit ELF Support
6655 * PowerPC64 ELF64:: @command{ld} and PowerPC64 64-bit ELF Support
6658 * S/390 ELF:: @command{ld} and S/390 ELF Support
6661 * SPU ELF:: @command{ld} and SPU ELF Support
6664 * TI COFF:: @command{ld} and TI COFF
6667 * WIN32:: @command{ld} and WIN32 (cygwin/mingw)
6670 * Xtensa:: @command{ld} and Xtensa Processors
6681 @section @command{ld} and the H8/300
6683 @cindex H8/300 support
6684 For the H8/300, @command{ld} can perform these global optimizations when
6685 you specify the @samp{--relax} command-line option.
6688 @cindex relaxing on H8/300
6689 @item relaxing address modes
6690 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6691 targets are within eight bits, and turns them into eight-bit
6692 program-counter relative @code{bsr} and @code{bra} instructions,
6695 @cindex synthesizing on H8/300
6696 @item synthesizing instructions
6697 @c FIXME: specifically mov.b, or any mov instructions really? -> mov.b only, at least on H8, H8H, H8S
6698 @command{ld} finds all @code{mov.b} instructions which use the
6699 sixteen-bit absolute address form, but refer to the top
6700 page of memory, and changes them to use the eight-bit address form.
6701 (That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
6702 @samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
6703 top page of memory).
6705 @command{ld} finds all @code{mov} instructions which use the register
6706 indirect with 32-bit displacement addressing mode, but use a small
6707 displacement inside 16-bit displacement range, and changes them to use
6708 the 16-bit displacement form. (That is: the linker turns @samp{mov.b
6709 @code{@@}@var{d}:32,ERx} into @samp{mov.b @code{@@}@var{d}:16,ERx}
6710 whenever the displacement @var{d} is in the 16 bit signed integer
6711 range. Only implemented in ELF-format ld).
6713 @item bit manipulation instructions
6714 @command{ld} finds all bit manipulation instructions like @code{band, bclr,
6715 biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
6716 which use 32 bit and 16 bit absolute address form, but refer to the top
6717 page of memory, and changes them to use the 8 bit address form.
6718 (That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
6719 @samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
6720 the top page of memory).
6722 @item system control instructions
6723 @command{ld} finds all @code{ldc.w, stc.w} instructions which use the
6724 32 bit absolute address form, but refer to the top page of memory, and
6725 changes them to use 16 bit address form.
6726 (That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
6727 @samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
6728 the top page of memory).
6738 @c This stuff is pointless to say unless you're especially concerned
6739 @c with Renesas chips; don't enable it for generic case, please.
6741 @chapter @command{ld} and Other Renesas Chips
6743 @command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
6744 H8/500, and SH chips. No special features, commands, or command-line
6745 options are required for these chips.
6759 @node M68HC11/68HC12
6760 @section @command{ld} and the Motorola 68HC11 and 68HC12 families
6762 @cindex M68HC11 and 68HC12 support
6764 @subsection Linker Relaxation
6766 For the Motorola 68HC11, @command{ld} can perform these global
6767 optimizations when you specify the @samp{--relax} command-line option.
6770 @cindex relaxing on M68HC11
6771 @item relaxing address modes
6772 @command{ld} finds all @code{jsr} and @code{jmp} instructions whose
6773 targets are within eight bits, and turns them into eight-bit
6774 program-counter relative @code{bsr} and @code{bra} instructions,
6777 @command{ld} also looks at all 16-bit extended addressing modes and
6778 transforms them in a direct addressing mode when the address is in
6779 page 0 (between 0 and 0x0ff).
6781 @item relaxing gcc instruction group
6782 When @command{gcc} is called with @option{-mrelax}, it can emit group
6783 of instructions that the linker can optimize to use a 68HC11 direct
6784 addressing mode. These instructions consists of @code{bclr} or
6785 @code{bset} instructions.
6789 @subsection Trampoline Generation
6791 @cindex trampoline generation on M68HC11
6792 @cindex trampoline generation on M68HC12
6793 For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
6794 call a far function using a normal @code{jsr} instruction. The linker
6795 will also change the relocation to some far function to use the
6796 trampoline address instead of the function address. This is typically the
6797 case when a pointer to a function is taken. The pointer will in fact
6798 point to the function trampoline.
6806 @section @command{ld} and the ARM family
6808 @cindex ARM interworking support
6809 @kindex --support-old-code
6810 For the ARM, @command{ld} will generate code stubs to allow functions calls
6811 between ARM and Thumb code. These stubs only work with code that has
6812 been compiled and assembled with the @samp{-mthumb-interwork} command
6813 line option. If it is necessary to link with old ARM object files or
6814 libraries, which have not been compiled with the -mthumb-interwork
6815 option then the @samp{--support-old-code} command-line switch should be
6816 given to the linker. This will make it generate larger stub functions
6817 which will work with non-interworking aware ARM code. Note, however,
6818 the linker does not support generating stubs for function calls to
6819 non-interworking aware Thumb code.
6821 @cindex thumb entry point
6822 @cindex entry point, thumb
6823 @kindex --thumb-entry=@var{entry}
6824 The @samp{--thumb-entry} switch is a duplicate of the generic
6825 @samp{--entry} switch, in that it sets the program's starting address.
6826 But it also sets the bottom bit of the address, so that it can be
6827 branched to using a BX instruction, and the program will start
6828 executing in Thumb mode straight away.
6830 @cindex PE import table prefixing
6831 @kindex --use-nul-prefixed-import-tables
6832 The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
6833 the import tables idata4 and idata5 have to be generated with a zero
6834 element prefix for import libraries. This is the old style to generate
6835 import tables. By default this option is turned off.
6839 The @samp{--be8} switch instructs @command{ld} to generate BE8 format
6840 executables. This option is only valid when linking big-endian
6841 objects - ie ones which have been assembled with the @option{-EB}
6842 option. The resulting image will contain big-endian data and
6846 @kindex --target1-rel
6847 @kindex --target1-abs
6848 The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
6849 @samp{.init_array} section. It is interpreted as either @samp{R_ARM_REL32}
6850 or @samp{R_ARM_ABS32}, depending on the target. The @samp{--target1-rel}
6851 and @samp{--target1-abs} switches override the default.
6854 @kindex --target2=@var{type}
6855 The @samp{--target2=type} switch overrides the default definition of the
6856 @samp{R_ARM_TARGET2} relocation. Valid values for @samp{type}, their
6857 meanings, and target defaults are as follows:
6860 @samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
6862 @samp{R_ARM_ABS32} (arm*-*-symbianelf)
6864 @samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
6869 The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
6870 specification) enables objects compiled for the ARMv4 architecture to be
6871 interworking-safe when linked with other objects compiled for ARMv4t, but
6872 also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6874 In the latter case, the switch @option{--fix-v4bx} must be passed to the
6875 linker, which causes v4t @code{BX rM} instructions to be rewritten as
6876 @code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6878 In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6879 relocations are ignored.
6881 @cindex FIX_V4BX_INTERWORKING
6882 @kindex --fix-v4bx-interworking
6883 Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6884 relocations with a branch to the following veneer:
6892 This allows generation of libraries/applications that work on ARMv4 cores
6893 and are still interworking safe. Note that the above veneer clobbers the
6894 condition flags, so may cause incorrect program behavior in rare cases.
6898 The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6899 BLX instructions (available on ARMv5t and above) in various
6900 situations. Currently it is used to perform calls via the PLT from Thumb
6901 code using BLX rather than using BX and a mode-switching stub before
6902 each PLT entry. This should lead to such calls executing slightly faster.
6904 This option is enabled implicitly for SymbianOS, so there is no need to
6905 specify it if you are using that target.
6907 @cindex VFP11_DENORM_FIX
6908 @kindex --vfp11-denorm-fix
6909 The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6910 bug in certain VFP11 coprocessor hardware, which sometimes allows
6911 instructions with denorm operands (which must be handled by support code)
6912 to have those operands overwritten by subsequent instructions before
6913 the support code can read the intended values.
6915 The bug may be avoided in scalar mode if you allow at least one
6916 intervening instruction between a VFP11 instruction which uses a register
6917 and another instruction which writes to the same register, or at least two
6918 intervening instructions if vector mode is in use. The bug only affects
6919 full-compliance floating-point mode: you do not need this workaround if
6920 you are using "runfast" mode. Please contact ARM for further details.
6922 If you know you are using buggy VFP11 hardware, you can
6923 enable this workaround by specifying the linker option
6924 @samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6925 mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6926 vector mode (the latter also works for scalar code). The default is
6927 @samp{--vfp-denorm-fix=none}.
6929 If the workaround is enabled, instructions are scanned for
6930 potentially-troublesome sequences, and a veneer is created for each
6931 such sequence which may trigger the erratum. The veneer consists of the
6932 first instruction of the sequence and a branch back to the subsequent
6933 instruction. The original instruction is then replaced with a branch to
6934 the veneer. The extra cycles required to call and return from the veneer
6935 are sufficient to avoid the erratum in both the scalar and vector cases.
6937 @cindex ARM1176 erratum workaround
6938 @kindex --fix-arm1176
6939 @kindex --no-fix-arm1176
6940 The @samp{--fix-arm1176} switch enables a link-time workaround for an erratum
6941 in certain ARM1176 processors. The workaround is enabled by default if you
6942 are targeting ARM v6 (excluding ARM v6T2) or earlier. It can be disabled
6943 unconditionally by specifying @samp{--no-fix-arm1176}.
6945 Further information is available in the ``ARM1176JZ-S and ARM1176JZF-S
6946 Programmer Advice Notice'' available on the ARM documentation website at:
6947 http://infocenter.arm.com/.
6949 @cindex STM32L4xx erratum workaround
6950 @kindex --fix-stm32l4xx-629360
6952 The @samp{--fix-stm32l4xx-629360} switch enables a link-time
6953 workaround for a bug in the bus matrix / memory controller for some of
6954 the STM32 Cortex-M4 based products (STM32L4xx). When accessing
6955 off-chip memory via the affected bus for bus reads of 9 words or more,
6956 the bus can generate corrupt data and/or abort. These are only
6957 core-initiated accesses (not DMA), and might affect any access:
6958 integer loads such as LDM, POP and floating-point loads such as VLDM,
6959 VPOP. Stores are not affected.
6961 The bug can be avoided by splitting memory accesses into the
6962 necessary chunks to keep bus reads below 8 words.
6964 The workaround is not enabled by default, this is equivalent to use
6965 @samp{--fix-stm32l4xx-629360=none}. If you know you are using buggy
6966 STM32L4xx hardware, you can enable the workaround by specifying the
6967 linker option @samp{--fix-stm32l4xx-629360}, or the equivalent
6968 @samp{--fix-stm32l4xx-629360=default}.
6970 If the workaround is enabled, instructions are scanned for
6971 potentially-troublesome sequences, and a veneer is created for each
6972 such sequence which may trigger the erratum. The veneer consists in a
6973 replacement sequence emulating the behaviour of the original one and a
6974 branch back to the subsequent instruction. The original instruction is
6975 then replaced with a branch to the veneer.
6977 The workaround does not always preserve the memory access order for
6978 the LDMDB instruction, when the instruction loads the PC.
6980 The workaround is not able to handle problematic instructions when
6981 they are in the middle of an IT block, since a branch is not allowed
6982 there. In that case, the linker reports a warning and no replacement
6985 The workaround is not able to replace problematic instructions with a
6986 PC-relative branch instruction if the @samp{.text} section is too
6987 large. In that case, when the branch that replaces the original code
6988 cannot be encoded, the linker reports a warning and no replacement
6991 @cindex NO_ENUM_SIZE_WARNING
6992 @kindex --no-enum-size-warning
6993 The @option{--no-enum-size-warning} switch prevents the linker from
6994 warning when linking object files that specify incompatible EABI
6995 enumeration size attributes. For example, with this switch enabled,
6996 linking of an object file using 32-bit enumeration values with another
6997 using enumeration values fitted into the smallest possible space will
7000 @cindex NO_WCHAR_SIZE_WARNING
7001 @kindex --no-wchar-size-warning
7002 The @option{--no-wchar-size-warning} switch prevents the linker from
7003 warning when linking object files that specify incompatible EABI
7004 @code{wchar_t} size attributes. For example, with this switch enabled,
7005 linking of an object file using 32-bit @code{wchar_t} values with another
7006 using 16-bit @code{wchar_t} values will not be diagnosed.
7009 @kindex --pic-veneer
7010 The @samp{--pic-veneer} switch makes the linker use PIC sequences for
7011 ARM/Thumb interworking veneers, even if the rest of the binary
7012 is not PIC. This avoids problems on uClinux targets where
7013 @samp{--emit-relocs} is used to generate relocatable binaries.
7015 @cindex STUB_GROUP_SIZE
7016 @kindex --stub-group-size=@var{N}
7017 The linker will automatically generate and insert small sequences of
7018 code into a linked ARM ELF executable whenever an attempt is made to
7019 perform a function call to a symbol that is too far away. The
7020 placement of these sequences of instructions - called stubs - is
7021 controlled by the command-line option @option{--stub-group-size=N}.
7022 The placement is important because a poor choice can create a need for
7023 duplicate stubs, increasing the code size. The linker will try to
7024 group stubs together in order to reduce interruptions to the flow of
7025 code, but it needs guidance as to how big these groups should be and
7026 where they should be placed.
7028 The value of @samp{N}, the parameter to the
7029 @option{--stub-group-size=} option controls where the stub groups are
7030 placed. If it is negative then all stubs are placed after the first
7031 branch that needs them. If it is positive then the stubs can be
7032 placed either before or after the branches that need them. If the
7033 value of @samp{N} is 1 (either +1 or -1) then the linker will choose
7034 exactly where to place groups of stubs, using its built in heuristics.
7035 A value of @samp{N} greater than 1 (or smaller than -1) tells the
7036 linker that a single group of stubs can service at most @samp{N} bytes
7037 from the input sections.
7039 The default, if @option{--stub-group-size=} is not specified, is
7042 Farcalls stubs insertion is fully supported for the ARM-EABI target
7043 only, because it relies on object files properties not present
7046 @cindex Cortex-A8 erratum workaround
7047 @kindex --fix-cortex-a8
7048 @kindex --no-fix-cortex-a8
7049 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}.
7051 The erratum only affects Thumb-2 code. Please contact ARM for further details.
7053 @cindex Cortex-A53 erratum 835769 workaround
7054 @kindex --fix-cortex-a53-835769
7055 @kindex --no-fix-cortex-a53-835769
7056 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}.
7058 Please contact ARM for further details.
7060 @kindex --merge-exidx-entries
7061 @kindex --no-merge-exidx-entries
7062 @cindex Merging exidx entries
7063 The @samp{--no-merge-exidx-entries} switch disables the merging of adjacent exidx entries in debuginfo.
7066 @cindex 32-bit PLT entries
7067 The @samp{--long-plt} option enables the use of 16 byte PLT entries
7068 which support up to 4Gb of code. The default is to use 12 byte PLT
7069 entries which only support 512Mb of code.
7071 @kindex --no-apply-dynamic-relocs
7072 @cindex AArch64 rela addend
7073 The @samp{--no-apply-dynamic-relocs} option makes AArch64 linker do not apply
7074 link-time values for dynamic relocations.
7076 @cindex Placement of SG veneers
7077 All SG veneers are placed in the special output section @code{.gnu.sgstubs}.
7078 Its start address must be set, either with the command-line option
7079 @samp{--section-start} or in a linker script, to indicate where to place these
7082 @kindex --cmse-implib
7083 @cindex Secure gateway import library
7084 The @samp{--cmse-implib} option requests that the import libraries
7085 specified by the @samp{--out-implib} and @samp{--in-implib} options are
7086 secure gateway import libraries, suitable for linking a non-secure
7087 executable against secure code as per ARMv8-M Security Extensions.
7089 @kindex --in-implib=@var{file}
7090 @cindex Input import library
7091 The @samp{--in-implib=file} specifies an input import library whose symbols
7092 must keep the same address in the executable being produced. A warning is
7093 given if no @samp{--out-implib} is given but new symbols have been introduced
7094 in the executable that should be listed in its import library. Otherwise, if
7095 @samp{--out-implib} is specified, the symbols are added to the output import
7096 library. A warning is also given if some symbols present in the input import
7097 library have disappeared from the executable. This option is only effective
7098 for Secure Gateway import libraries, ie. when @samp{--cmse-implib} is
7112 @section @command{ld} and HPPA 32-bit ELF Support
7113 @cindex HPPA multiple sub-space stubs
7114 @kindex --multi-subspace
7115 When generating a shared library, @command{ld} will by default generate
7116 import stubs suitable for use with a single sub-space application.
7117 The @samp{--multi-subspace} switch causes @command{ld} to generate export
7118 stubs, and different (larger) import stubs suitable for use with
7119 multiple sub-spaces.
7121 @cindex HPPA stub grouping
7122 @kindex --stub-group-size=@var{N}
7123 Long branch stubs and import/export stubs are placed by @command{ld} in
7124 stub sections located between groups of input sections.
7125 @samp{--stub-group-size} specifies the maximum size of a group of input
7126 sections handled by one stub section. Since branch offsets are signed,
7127 a stub section may serve two groups of input sections, one group before
7128 the stub section, and one group after it. However, when using
7129 conditional branches that require stubs, it may be better (for branch
7130 prediction) that stub sections only serve one group of input sections.
7131 A negative value for @samp{N} chooses this scheme, ensuring that
7132 branches to stubs always use a negative offset. Two special values of
7133 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7134 @command{ld} to automatically size input section groups for the branch types
7135 detected, with the same behaviour regarding stub placement as other
7136 positive or negative values of @samp{N} respectively.
7138 Note that @samp{--stub-group-size} does not split input sections. A
7139 single input section larger than the group size specified will of course
7140 create a larger group (of one section). If input sections are too
7141 large, it may not be possible for a branch to reach its stub.
7154 @section @command{ld} and the Motorola 68K family
7156 @cindex Motorola 68K GOT generation
7157 @kindex --got=@var{type}
7158 The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
7159 The choices are @samp{single}, @samp{negative}, @samp{multigot} and
7160 @samp{target}. When @samp{target} is selected the linker chooses
7161 the default GOT generation scheme for the current target.
7162 @samp{single} tells the linker to generate a single GOT with
7163 entries only at non-negative offsets.
7164 @samp{negative} instructs the linker to generate a single GOT with
7165 entries at both negative and positive offsets. Not all environments
7167 @samp{multigot} allows the linker to generate several GOTs in the
7168 output file. All GOT references from a single input object
7169 file access the same GOT, but references from different input object
7170 files might access different GOTs. Not all environments support such GOTs.
7183 @section @command{ld} and the MIPS family
7185 @cindex MIPS microMIPS instruction choice selection
7188 The @samp{--insn32} and @samp{--no-insn32} options control the choice of
7189 microMIPS instructions used in code generated by the linker, such as that
7190 in the PLT or lazy binding stubs, or in relaxation. If @samp{--insn32} is
7191 used, then the linker only uses 32-bit instruction encodings. By default
7192 or if @samp{--no-insn32} is used, all instruction encodings are used,
7193 including 16-bit ones where possible.
7195 @cindex MIPS branch relocation check control
7196 @kindex --ignore-branch-isa
7197 @kindex --no-ignore-branch-isa
7198 The @samp{--ignore-branch-isa} and @samp{--no-ignore-branch-isa} options
7199 control branch relocation checks for invalid ISA mode transitions. If
7200 @samp{--ignore-branch-isa} is used, then the linker accepts any branch
7201 relocations and any ISA mode transition required is lost in relocation
7202 calculation, except for some cases of @code{BAL} instructions which meet
7203 relaxation conditions and are converted to equivalent @code{JALX}
7204 instructions as the associated relocation is calculated. By default
7205 or if @samp{--no-ignore-branch-isa} is used a check is made causing
7206 the loss of an ISA mode transition to produce an error.
7219 @section @code{ld} and MMIX
7220 For MMIX, there is a choice of generating @code{ELF} object files or
7221 @code{mmo} object files when linking. The simulator @code{mmix}
7222 understands the @code{mmo} format. The binutils @code{objcopy} utility
7223 can translate between the two formats.
7225 There is one special section, the @samp{.MMIX.reg_contents} section.
7226 Contents in this section is assumed to correspond to that of global
7227 registers, and symbols referring to it are translated to special symbols,
7228 equal to registers. In a final link, the start address of the
7229 @samp{.MMIX.reg_contents} section corresponds to the first allocated
7230 global register multiplied by 8. Register @code{$255} is not included in
7231 this section; it is always set to the program entry, which is at the
7232 symbol @code{Main} for @code{mmo} files.
7234 Global symbols with the prefix @code{__.MMIX.start.}, for example
7235 @code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
7236 The default linker script uses these to set the default start address
7239 Initial and trailing multiples of zero-valued 32-bit words in a section,
7240 are left out from an mmo file.
7253 @section @code{ld} and MSP430
7254 For the MSP430 it is possible to select the MPU architecture. The flag @samp{-m [mpu type]}
7255 will select an appropriate linker script for selected MPU type. (To get a list of known MPUs
7256 just pass @samp{-m help} option to the linker).
7258 @cindex MSP430 extra sections
7259 The linker will recognize some extra sections which are MSP430 specific:
7262 @item @samp{.vectors}
7263 Defines a portion of ROM where interrupt vectors located.
7265 @item @samp{.bootloader}
7266 Defines the bootloader portion of the ROM (if applicable). Any code
7267 in this section will be uploaded to the MPU.
7269 @item @samp{.infomem}
7270 Defines an information memory section (if applicable). Any code in
7271 this section will be uploaded to the MPU.
7273 @item @samp{.infomemnobits}
7274 This is the same as the @samp{.infomem} section except that any code
7275 in this section will not be uploaded to the MPU.
7277 @item @samp{.noinit}
7278 Denotes a portion of RAM located above @samp{.bss} section.
7280 The last two sections are used by gcc.
7284 @cindex MSP430 Options
7285 @kindex --code-region
7286 @item --code-region=[either,lower,upper,none]
7287 This will transform .text* sections to [either,lower,upper].text* sections. The
7288 argument passed to GCC for -mcode-region is propagated to the linker
7291 @kindex --data-region
7292 @item --data-region=[either,lower,upper,none]
7293 This will transform .data*, .bss* and .rodata* sections to
7294 [either,lower,upper].[data,bss,rodata]* sections. The argument passed to GCC
7295 for -mdata-region is propagated to the linker using this option.
7297 @kindex --disable-sec-transformation
7298 @item --disable-sec-transformation
7299 Prevent the transformation of sections as specified by the @code{--code-region}
7300 and @code{--data-region} options.
7301 This is useful if you are compiling and linking using a single call to the GCC
7302 wrapper, and want to compile the source files using -m[code,data]-region but
7303 not transform the sections for prebuilt libraries and objects.
7317 @section @code{ld} and NDS32
7318 @kindex relaxing on NDS32
7319 For NDS32, there are some options to select relaxation behavior. The linker
7320 relaxes objects according to these options.
7323 @item @samp{--m[no-]fp-as-gp}
7324 Disable/enable fp-as-gp relaxation.
7326 @item @samp{--mexport-symbols=FILE}
7327 Exporting symbols and their address into FILE as linker script.
7329 @item @samp{--m[no-]ex9}
7330 Disable/enable link-time EX9 relaxation.
7332 @item @samp{--mexport-ex9=FILE}
7333 Export the EX9 table after linking.
7335 @item @samp{--mimport-ex9=FILE}
7336 Import the Ex9 table for EX9 relaxation.
7338 @item @samp{--mupdate-ex9}
7339 Update the existing EX9 table.
7341 @item @samp{--mex9-limit=NUM}
7342 Maximum number of entries in the ex9 table.
7344 @item @samp{--mex9-loop-aware}
7345 Avoid generating the EX9 instruction inside the loop.
7347 @item @samp{--m[no-]ifc}
7348 Disable/enable the link-time IFC optimization.
7350 @item @samp{--mifc-loop-aware}
7351 Avoid generating the IFC instruction inside the loop.
7365 @section @command{ld} and the Altera Nios II
7366 @cindex Nios II call relaxation
7367 @kindex --relax on Nios II
7369 Call and immediate jump instructions on Nios II processors are limited to
7370 transferring control to addresses in the same 256MB memory segment,
7371 which may result in @command{ld} giving
7372 @samp{relocation truncated to fit} errors with very large programs.
7373 The command-line option @option{--relax} enables the generation of
7374 trampolines that can access the entire 32-bit address space for calls
7375 outside the normal @code{call} and @code{jmpi} address range. These
7376 trampolines are inserted at section boundaries, so may not themselves
7377 be reachable if an input section and its associated call trampolines are
7380 The @option{--relax} option is enabled by default unless @option{-r}
7381 is also specified. You can disable trampoline generation by using the
7382 @option{--no-relax} linker option. You can also disable this optimization
7383 locally by using the @samp{set .noat} directive in assembly-language
7384 source files, as the linker-inserted trampolines use the @code{at}
7385 register as a temporary.
7387 Note that the linker @option{--relax} option is independent of assembler
7388 relaxation options, and that using the GNU assembler's @option{-relax-all}
7389 option interferes with the linker's more selective call instruction relaxation.
7402 @section @command{ld} and PowerPC 32-bit ELF Support
7403 @cindex PowerPC long branches
7404 @kindex --relax on PowerPC
7405 Branches on PowerPC processors are limited to a signed 26-bit
7406 displacement, which may result in @command{ld} giving
7407 @samp{relocation truncated to fit} errors with very large programs.
7408 @samp{--relax} enables the generation of trampolines that can access
7409 the entire 32-bit address space. These trampolines are inserted at
7410 section boundaries, so may not themselves be reachable if an input
7411 section exceeds 33M in size. You may combine @samp{-r} and
7412 @samp{--relax} to add trampolines in a partial link. In that case
7413 both branches to undefined symbols and inter-section branches are also
7414 considered potentially out of range, and trampolines inserted.
7416 @cindex PowerPC ELF32 options
7421 Current PowerPC GCC accepts a @samp{-msecure-plt} option that
7422 generates code capable of using a newer PLT and GOT layout that has
7423 the security advantage of no executable section ever needing to be
7424 writable and no writable section ever being executable. PowerPC
7425 @command{ld} will generate this layout, including stubs to access the
7426 PLT, if all input files (including startup and static libraries) were
7427 compiled with @samp{-msecure-plt}. @samp{--bss-plt} forces the old
7428 BSS PLT (and GOT layout) which can give slightly better performance.
7430 @kindex --secure-plt
7432 @command{ld} will use the new PLT and GOT layout if it is linking new
7433 @samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
7434 when linking non-PIC code. This option requests the new PLT and GOT
7435 layout. A warning will be given if some object file requires the old
7441 The new secure PLT and GOT are placed differently relative to other
7442 sections compared to older BSS PLT and GOT placement. The location of
7443 @code{.plt} must change because the new secure PLT is an initialized
7444 section while the old PLT is uninitialized. The reason for the
7445 @code{.got} change is more subtle: The new placement allows
7446 @code{.got} to be read-only in applications linked with
7447 @samp{-z relro -z now}. However, this placement means that
7448 @code{.sdata} cannot always be used in shared libraries, because the
7449 PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
7450 pointer. @samp{--sdata-got} forces the old GOT placement. PowerPC
7451 GCC doesn't use @code{.sdata} in shared libraries, so this option is
7452 really only useful for other compilers that may do so.
7454 @cindex PowerPC stub symbols
7455 @kindex --emit-stub-syms
7456 @item --emit-stub-syms
7457 This option causes @command{ld} to label linker stubs with a local
7458 symbol that encodes the stub type and destination.
7460 @cindex PowerPC TLS optimization
7461 @kindex --no-tls-optimize
7462 @item --no-tls-optimize
7463 PowerPC @command{ld} normally performs some optimization of code
7464 sequences used to access Thread-Local Storage. Use this option to
7465 disable the optimization.
7478 @node PowerPC64 ELF64
7479 @section @command{ld} and PowerPC64 64-bit ELF Support
7481 @cindex PowerPC64 ELF64 options
7483 @cindex PowerPC64 stub grouping
7484 @kindex --stub-group-size
7485 @item --stub-group-size
7486 Long branch stubs, PLT call stubs and TOC adjusting stubs are placed
7487 by @command{ld} in stub sections located between groups of input sections.
7488 @samp{--stub-group-size} specifies the maximum size of a group of input
7489 sections handled by one stub section. Since branch offsets are signed,
7490 a stub section may serve two groups of input sections, one group before
7491 the stub section, and one group after it. However, when using
7492 conditional branches that require stubs, it may be better (for branch
7493 prediction) that stub sections only serve one group of input sections.
7494 A negative value for @samp{N} chooses this scheme, ensuring that
7495 branches to stubs always use a negative offset. Two special values of
7496 @samp{N} are recognized, @samp{1} and @samp{-1}. These both instruct
7497 @command{ld} to automatically size input section groups for the branch types
7498 detected, with the same behaviour regarding stub placement as other
7499 positive or negative values of @samp{N} respectively.
7501 Note that @samp{--stub-group-size} does not split input sections. A
7502 single input section larger than the group size specified will of course
7503 create a larger group (of one section). If input sections are too
7504 large, it may not be possible for a branch to reach its stub.
7506 @cindex PowerPC64 stub symbols
7507 @kindex --emit-stub-syms
7508 @item --emit-stub-syms
7509 This option causes @command{ld} to label linker stubs with a local
7510 symbol that encodes the stub type and destination.
7512 @cindex PowerPC64 dot symbols
7514 @kindex --no-dotsyms
7517 These two options control how @command{ld} interprets version patterns
7518 in a version script. Older PowerPC64 compilers emitted both a
7519 function descriptor symbol with the same name as the function, and a
7520 code entry symbol with the name prefixed by a dot (@samp{.}). To
7521 properly version a function @samp{foo}, the version script thus needs
7522 to control both @samp{foo} and @samp{.foo}. The option
7523 @samp{--dotsyms}, on by default, automatically adds the required
7524 dot-prefixed patterns. Use @samp{--no-dotsyms} to disable this
7527 @cindex PowerPC64 register save/restore functions
7528 @kindex --save-restore-funcs
7529 @kindex --no-save-restore-funcs
7530 @item --save-restore-funcs
7531 @itemx --no-save-restore-funcs
7532 These two options control whether PowerPC64 @command{ld} automatically
7533 provides out-of-line register save and restore functions used by
7534 @samp{-Os} code. The default is to provide any such referenced
7535 function for a normal final link, and to not do so for a relocatable
7538 @cindex PowerPC64 TLS optimization
7539 @kindex --no-tls-optimize
7540 @item --no-tls-optimize
7541 PowerPC64 @command{ld} normally performs some optimization of code
7542 sequences used to access Thread-Local Storage. Use this option to
7543 disable the optimization.
7545 @cindex PowerPC64 __tls_get_addr optimization
7546 @kindex --tls-get-addr-optimize
7547 @kindex --no-tls-get-addr-optimize
7548 @item --tls-get-addr-optimize
7549 @itemx --no-tls-get-addr-optimize
7550 These options control whether PowerPC64 @command{ld} uses a special
7551 stub to call __tls_get_addr. PowerPC64 glibc 2.22 and later support
7552 an optimization that allows the second and subsequent calls to
7553 @code{__tls_get_addr} for a given symbol to be resolved by the special
7554 stub without calling in to glibc. By default the linker enables this
7555 option when glibc advertises the availability of __tls_get_addr_opt.
7556 Forcing this option on when using an older glibc won't do much besides
7557 slow down your applications, but may be useful if linking an
7558 application against an older glibc with the expectation that it will
7559 normally be used on systems having a newer glibc.
7561 @cindex PowerPC64 OPD optimization
7562 @kindex --no-opd-optimize
7563 @item --no-opd-optimize
7564 PowerPC64 @command{ld} normally removes @code{.opd} section entries
7565 corresponding to deleted link-once functions, or functions removed by
7566 the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
7567 Use this option to disable @code{.opd} optimization.
7569 @cindex PowerPC64 OPD spacing
7570 @kindex --non-overlapping-opd
7571 @item --non-overlapping-opd
7572 Some PowerPC64 compilers have an option to generate compressed
7573 @code{.opd} entries spaced 16 bytes apart, overlapping the third word,
7574 the static chain pointer (unused in C) with the first word of the next
7575 entry. This option expands such entries to the full 24 bytes.
7577 @cindex PowerPC64 TOC optimization
7578 @kindex --no-toc-optimize
7579 @item --no-toc-optimize
7580 PowerPC64 @command{ld} normally removes unused @code{.toc} section
7581 entries. Such entries are detected by examining relocations that
7582 reference the TOC in code sections. A reloc in a deleted code section
7583 marks a TOC word as unneeded, while a reloc in a kept code section
7584 marks a TOC word as needed. Since the TOC may reference itself, TOC
7585 relocs are also examined. TOC words marked as both needed and
7586 unneeded will of course be kept. TOC words without any referencing
7587 reloc are assumed to be part of a multi-word entry, and are kept or
7588 discarded as per the nearest marked preceding word. This works
7589 reliably for compiler generated code, but may be incorrect if assembly
7590 code is used to insert TOC entries. Use this option to disable the
7593 @cindex PowerPC64 multi-TOC
7594 @kindex --no-multi-toc
7595 @item --no-multi-toc
7596 If given any toc option besides @code{-mcmodel=medium} or
7597 @code{-mcmodel=large}, PowerPC64 GCC generates code for a TOC model
7599 entries are accessed with a 16-bit offset from r2. This limits the
7600 total TOC size to 64K. PowerPC64 @command{ld} extends this limit by
7601 grouping code sections such that each group uses less than 64K for its
7602 TOC entries, then inserts r2 adjusting stubs between inter-group
7603 calls. @command{ld} does not split apart input sections, so cannot
7604 help if a single input file has a @code{.toc} section that exceeds
7605 64K, most likely from linking multiple files with @command{ld -r}.
7606 Use this option to turn off this feature.
7608 @cindex PowerPC64 TOC sorting
7609 @kindex --no-toc-sort
7611 By default, @command{ld} sorts TOC sections so that those whose file
7612 happens to have a section called @code{.init} or @code{.fini} are
7613 placed first, followed by TOC sections referenced by code generated
7614 with PowerPC64 gcc's @code{-mcmodel=small}, and lastly TOC sections
7615 referenced only by code generated with PowerPC64 gcc's
7616 @code{-mcmodel=medium} or @code{-mcmodel=large} options. Doing this
7617 results in better TOC grouping for multi-TOC. Use this option to turn
7620 @cindex PowerPC64 PLT stub alignment
7622 @kindex --no-plt-align
7624 @itemx --no-plt-align
7625 Use these options to control whether individual PLT call stubs are
7626 aligned to a 32-byte boundary, or to the specified power of two
7627 boundary when using @code{--plt-align=}. A negative value may be
7628 specified to pad PLT call stubs so that they do not cross the
7629 specified power of two boundary (or the minimum number of boundaries
7630 if a PLT stub is so large that it must cross a boundary). By default
7631 PLT call stubs are aligned to 32-byte boundaries.
7633 @cindex PowerPC64 PLT call stub static chain
7634 @kindex --plt-static-chain
7635 @kindex --no-plt-static-chain
7636 @item --plt-static-chain
7637 @itemx --no-plt-static-chain
7638 Use these options to control whether PLT call stubs load the static
7639 chain pointer (r11). @code{ld} defaults to not loading the static
7640 chain since there is never any need to do so on a PLT call.
7642 @cindex PowerPC64 PLT call stub thread safety
7643 @kindex --plt-thread-safe
7644 @kindex --no-plt-thread-safe
7645 @item --plt-thread-safe
7646 @itemx --no-plt-thread-safe
7647 With power7's weakly ordered memory model, it is possible when using
7648 lazy binding for ld.so to update a plt entry in one thread and have
7649 another thread see the individual plt entry words update in the wrong
7650 order, despite ld.so carefully writing in the correct order and using
7651 memory write barriers. To avoid this we need some sort of read
7652 barrier in the call stub, or use LD_BIND_NOW=1. By default, @code{ld}
7653 looks for calls to commonly used functions that create threads, and if
7654 seen, adds the necessary barriers. Use these options to change the
7657 @cindex PowerPC64 ELFv2 PLT localentry optimization
7658 @kindex --plt-localentry
7659 @kindex --no-plt-localentry
7660 @item --plt-localentry
7661 @itemx --no-localentry
7662 ELFv2 functions with localentry:0 are those with a single entry point,
7663 ie. global entry == local entry, and that have no requirement on r2
7664 (the TOC/GOT pointer) or r12, and guarantee r2 is unchanged on return.
7665 Such an external function can be called via the PLT without saving r2
7666 or restoring it on return, avoiding a common load-hit-store for small
7667 functions. The optimization is attractive, with up to 40% reduction
7668 in execution time for a small function, but can result in symbol
7669 interposition failures. Also, minor changes in a shared library,
7670 including system libraries, can cause a function that was localentry:0
7671 to become localentry:8. This will result in a dynamic loader
7672 complaint and failure to run. The option is experimental, use with
7673 care. @option{--no-plt-localentry} is the default.
7687 @section @command{ld} and S/390 ELF Support
7689 @cindex S/390 ELF options
7693 @kindex --s390-pgste
7695 This option marks the result file with a @code{PT_S390_PGSTE}
7696 segment. The Linux kernel is supposed to allocate 4k page tables for
7697 binaries marked that way.
7711 @section @command{ld} and SPU ELF Support
7713 @cindex SPU ELF options
7719 This option marks an executable as a PIC plugin module.
7721 @cindex SPU overlays
7722 @kindex --no-overlays
7724 Normally, @command{ld} recognizes calls to functions within overlay
7725 regions, and redirects such calls to an overlay manager via a stub.
7726 @command{ld} also provides a built-in overlay manager. This option
7727 turns off all this special overlay handling.
7729 @cindex SPU overlay stub symbols
7730 @kindex --emit-stub-syms
7731 @item --emit-stub-syms
7732 This option causes @command{ld} to label overlay stubs with a local
7733 symbol that encodes the stub type and destination.
7735 @cindex SPU extra overlay stubs
7736 @kindex --extra-overlay-stubs
7737 @item --extra-overlay-stubs
7738 This option causes @command{ld} to add overlay call stubs on all
7739 function calls out of overlay regions. Normally stubs are not added
7740 on calls to non-overlay regions.
7742 @cindex SPU local store size
7743 @kindex --local-store=lo:hi
7744 @item --local-store=lo:hi
7745 @command{ld} usually checks that a final executable for SPU fits in
7746 the address range 0 to 256k. This option may be used to change the
7747 range. Disable the check entirely with @option{--local-store=0:0}.
7750 @kindex --stack-analysis
7751 @item --stack-analysis
7752 SPU local store space is limited. Over-allocation of stack space
7753 unnecessarily limits space available for code and data, while
7754 under-allocation results in runtime failures. If given this option,
7755 @command{ld} will provide an estimate of maximum stack usage.
7756 @command{ld} does this by examining symbols in code sections to
7757 determine the extents of functions, and looking at function prologues
7758 for stack adjusting instructions. A call-graph is created by looking
7759 for relocations on branch instructions. The graph is then searched
7760 for the maximum stack usage path. Note that this analysis does not
7761 find calls made via function pointers, and does not handle recursion
7762 and other cycles in the call graph. Stack usage may be
7763 under-estimated if your code makes such calls. Also, stack usage for
7764 dynamic allocation, e.g. alloca, will not be detected. If a link map
7765 is requested, detailed information about each function's stack usage
7766 and calls will be given.
7769 @kindex --emit-stack-syms
7770 @item --emit-stack-syms
7771 This option, if given along with @option{--stack-analysis} will result
7772 in @command{ld} emitting stack sizing symbols for each function.
7773 These take the form @code{__stack_<function_name>} for global
7774 functions, and @code{__stack_<number>_<function_name>} for static
7775 functions. @code{<number>} is the section id in hex. The value of
7776 such symbols is the stack requirement for the corresponding function.
7777 The symbol size will be zero, type @code{STT_NOTYPE}, binding
7778 @code{STB_LOCAL}, and section @code{SHN_ABS}.
7792 @section @command{ld}'s Support for Various TI COFF Versions
7793 @cindex TI COFF versions
7794 @kindex --format=@var{version}
7795 The @samp{--format} switch allows selection of one of the various
7796 TI COFF versions. The latest of this writing is 2; versions 0 and 1 are
7797 also supported. The TI COFF versions also vary in header byte-order
7798 format; @command{ld} will read any version or byte order, but the output
7799 header format depends on the default specified by the specific target.
7812 @section @command{ld} and WIN32 (cygwin/mingw)
7814 This section describes some of the win32 specific @command{ld} issues.
7815 See @ref{Options,,Command-line Options} for detailed description of the
7816 command-line options mentioned here.
7819 @cindex import libraries
7820 @item import libraries
7821 The standard Windows linker creates and uses so-called import
7822 libraries, which contains information for linking to dll's. They are
7823 regular static archives and are handled as any other static
7824 archive. The cygwin and mingw ports of @command{ld} have specific
7825 support for creating such libraries provided with the
7826 @samp{--out-implib} command-line option.
7828 @item exporting DLL symbols
7829 @cindex exporting DLL symbols
7830 The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
7833 @item using auto-export functionality
7834 @cindex using auto-export functionality
7835 By default @command{ld} exports symbols with the auto-export functionality,
7836 which is controlled by the following command-line options:
7839 @item --export-all-symbols [This is the default]
7840 @item --exclude-symbols
7841 @item --exclude-libs
7842 @item --exclude-modules-for-implib
7843 @item --version-script
7846 When auto-export is in operation, @command{ld} will export all the non-local
7847 (global and common) symbols it finds in a DLL, with the exception of a few
7848 symbols known to belong to the system's runtime and libraries. As it will
7849 often not be desirable to export all of a DLL's symbols, which may include
7850 private functions that are not part of any public interface, the command-line
7851 options listed above may be used to filter symbols out from the list for
7852 exporting. The @samp{--output-def} option can be used in order to see the
7853 final list of exported symbols with all exclusions taken into effect.
7855 If @samp{--export-all-symbols} is not given explicitly on the
7856 command line, then the default auto-export behavior will be @emph{disabled}
7857 if either of the following are true:
7860 @item A DEF file is used.
7861 @item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
7864 @item using a DEF file
7865 @cindex using a DEF file
7866 Another way of exporting symbols is using a DEF file. A DEF file is
7867 an ASCII file containing definitions of symbols which should be
7868 exported when a dll is created. Usually it is named @samp{<dll
7869 name>.def} and is added as any other object file to the linker's
7870 command line. The file's name must end in @samp{.def} or @samp{.DEF}.
7873 gcc -o <output> <objectfiles> <dll name>.def
7876 Using a DEF file turns off the normal auto-export behavior, unless the
7877 @samp{--export-all-symbols} option is also used.
7879 Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
7882 LIBRARY "xyz.dll" BASE=0x20000000
7888 another_foo = abc.dll.afoo
7894 This example defines a DLL with a non-default base address and seven
7895 symbols in the export table. The third exported symbol @code{_bar} is an
7896 alias for the second. The fourth symbol, @code{another_foo} is resolved
7897 by "forwarding" to another module and treating it as an alias for
7898 @code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
7899 @code{var1} is declared to be a data object. The @samp{doo} symbol in
7900 export library is an alias of @samp{foo}, which gets the string name
7901 in export table @samp{foo2}. The @samp{eoo} symbol is an data export
7902 symbol, which gets in export table the name @samp{var1}.
7904 The optional @code{LIBRARY <name>} command indicates the @emph{internal}
7905 name of the output DLL. If @samp{<name>} does not include a suffix,
7906 the default library suffix, @samp{.DLL} is appended.
7908 When the .DEF file is used to build an application, rather than a
7909 library, the @code{NAME <name>} command should be used instead of
7910 @code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
7911 executable suffix, @samp{.EXE} is appended.
7913 With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
7914 specification @code{BASE = <number>} may be used to specify a
7915 non-default base address for the image.
7917 If neither @code{LIBRARY <name>} nor @code{NAME <name>} is specified,
7918 or they specify an empty string, the internal name is the same as the
7919 filename specified on the command line.
7921 The complete specification of an export symbol is:
7925 ( ( ( <name1> [ = <name2> ] )
7926 | ( <name1> = <module-name> . <external-name>))
7927 [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] [== <name3>] ) *
7930 Declares @samp{<name1>} as an exported symbol from the DLL, or declares
7931 @samp{<name1>} as an exported alias for @samp{<name2>}; or declares
7932 @samp{<name1>} as a "forward" alias for the symbol
7933 @samp{<external-name>} in the DLL @samp{<module-name>}.
7934 Optionally, the symbol may be exported by the specified ordinal
7935 @samp{<integer>} alias. The optional @samp{<name3>} is the to be used
7936 string in import/export table for the symbol.
7938 The optional keywords that follow the declaration indicate:
7940 @code{NONAME}: Do not put the symbol name in the DLL's export table. It
7941 will still be exported by its ordinal alias (either the value specified
7942 by the .def specification or, otherwise, the value assigned by the
7943 linker). The symbol name, however, does remain visible in the import
7944 library (if any), unless @code{PRIVATE} is also specified.
7946 @code{DATA}: The symbol is a variable or object, rather than a function.
7947 The import lib will export only an indirect reference to @code{foo} as
7948 the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
7951 @code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
7952 well as @code{_imp__foo} into the import library. Both refer to the
7953 read-only import address table's pointer to the variable, not to the
7954 variable itself. This can be dangerous. If the user code fails to add
7955 the @code{dllimport} attribute and also fails to explicitly add the
7956 extra indirection that the use of the attribute enforces, the
7957 application will behave unexpectedly.
7959 @code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
7960 it into the static import library used to resolve imports at link time. The
7961 symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
7962 API at runtime or by using the GNU ld extension of linking directly to
7963 the DLL without an import library.
7965 See ld/deffilep.y in the binutils sources for the full specification of
7966 other DEF file statements
7968 @cindex creating a DEF file
7969 While linking a shared dll, @command{ld} is able to create a DEF file
7970 with the @samp{--output-def <file>} command-line option.
7972 @item Using decorations
7973 @cindex Using decorations
7974 Another way of marking symbols for export is to modify the source code
7975 itself, so that when building the DLL each symbol to be exported is
7979 __declspec(dllexport) int a_variable
7980 __declspec(dllexport) void a_function(int with_args)
7983 All such symbols will be exported from the DLL. If, however,
7984 any of the object files in the DLL contain symbols decorated in
7985 this way, then the normal auto-export behavior is disabled, unless
7986 the @samp{--export-all-symbols} option is also used.
7988 Note that object files that wish to access these symbols must @emph{not}
7989 decorate them with dllexport. Instead, they should use dllimport,
7993 __declspec(dllimport) int a_variable
7994 __declspec(dllimport) void a_function(int with_args)
7997 This complicates the structure of library header files, because
7998 when included by the library itself the header must declare the
7999 variables and functions as dllexport, but when included by client
8000 code the header must declare them as dllimport. There are a number
8001 of idioms that are typically used to do this; often client code can
8002 omit the __declspec() declaration completely. See
8003 @samp{--enable-auto-import} and @samp{automatic data imports} for more
8007 @cindex automatic data imports
8008 @item automatic data imports
8009 The standard Windows dll format supports data imports from dlls only
8010 by adding special decorations (dllimport/dllexport), which let the
8011 compiler produce specific assembler instructions to deal with this
8012 issue. This increases the effort necessary to port existing Un*x
8013 code to these platforms, especially for large
8014 c++ libraries and applications. The auto-import feature, which was
8015 initially provided by Paul Sokolovsky, allows one to omit the
8016 decorations to achieve a behavior that conforms to that on POSIX/Un*x
8017 platforms. This feature is enabled with the @samp{--enable-auto-import}
8018 command-line option, although it is enabled by default on cygwin/mingw.
8019 The @samp{--enable-auto-import} option itself now serves mainly to
8020 suppress any warnings that are ordinarily emitted when linked objects
8021 trigger the feature's use.
8023 auto-import of variables does not always work flawlessly without
8024 additional assistance. Sometimes, you will see this message
8026 "variable '<var>' can't be auto-imported. Please read the
8027 documentation for ld's @code{--enable-auto-import} for details."
8029 The @samp{--enable-auto-import} documentation explains why this error
8030 occurs, and several methods that can be used to overcome this difficulty.
8031 One of these methods is the @emph{runtime pseudo-relocs} feature, described
8034 @cindex runtime pseudo-relocation
8035 For complex variables imported from DLLs (such as structs or classes),
8036 object files typically contain a base address for the variable and an
8037 offset (@emph{addend}) within the variable--to specify a particular
8038 field or public member, for instance. Unfortunately, the runtime loader used
8039 in win32 environments is incapable of fixing these references at runtime
8040 without the additional information supplied by dllimport/dllexport decorations.
8041 The standard auto-import feature described above is unable to resolve these
8044 The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
8045 be resolved without error, while leaving the task of adjusting the references
8046 themselves (with their non-zero addends) to specialized code provided by the
8047 runtime environment. Recent versions of the cygwin and mingw environments and
8048 compilers provide this runtime support; older versions do not. However, the
8049 support is only necessary on the developer's platform; the compiled result will
8050 run without error on an older system.
8052 @samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
8055 @cindex direct linking to a dll
8056 @item direct linking to a dll
8057 The cygwin/mingw ports of @command{ld} support the direct linking,
8058 including data symbols, to a dll without the usage of any import
8059 libraries. This is much faster and uses much less memory than does the
8060 traditional import library method, especially when linking large
8061 libraries or applications. When @command{ld} creates an import lib, each
8062 function or variable exported from the dll is stored in its own bfd, even
8063 though a single bfd could contain many exports. The overhead involved in
8064 storing, loading, and processing so many bfd's is quite large, and explains the
8065 tremendous time, memory, and storage needed to link against particularly
8066 large or complex libraries when using import libs.
8068 Linking directly to a dll uses no extra command-line switches other than
8069 @samp{-L} and @samp{-l}, because @command{ld} already searches for a number
8070 of names to match each library. All that is needed from the developer's
8071 perspective is an understanding of this search, in order to force ld to
8072 select the dll instead of an import library.
8075 For instance, when ld is called with the argument @samp{-lxxx} it will attempt
8076 to find, in the first directory of its search path,
8089 before moving on to the next directory in the search path.
8091 (*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
8092 where @samp{<prefix>} is set by the @command{ld} option
8093 @samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
8094 file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
8097 Other win32-based unix environments, such as mingw or pw32, may use other
8098 @samp{<prefix>}es, although at present only cygwin makes use of this feature. It
8099 was originally intended to help avoid name conflicts among dll's built for the
8100 various win32/un*x environments, so that (for example) two versions of a zlib dll
8101 could coexist on the same machine.
8103 The generic cygwin/mingw path layout uses a @samp{bin} directory for
8104 applications and dll's and a @samp{lib} directory for the import
8105 libraries (using cygwin nomenclature):
8111 libxxx.dll.a (in case of dll's)
8112 libxxx.a (in case of static archive)
8115 Linking directly to a dll without using the import library can be
8118 1. Use the dll directly by adding the @samp{bin} path to the link line
8120 gcc -Wl,-verbose -o a.exe -L../bin/ -lxxx
8123 However, as the dll's often have version numbers appended to their names
8124 (@samp{cygncurses-5.dll}) this will often fail, unless one specifies
8125 @samp{-L../bin -lncurses-5} to include the version. Import libs are generally
8126 not versioned, and do not have this difficulty.
8128 2. Create a symbolic link from the dll to a file in the @samp{lib}
8129 directory according to the above mentioned search pattern. This
8130 should be used to avoid unwanted changes in the tools needed for
8134 ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
8137 Then you can link without any make environment changes.
8140 gcc -Wl,-verbose -o a.exe -L../lib/ -lxxx
8143 This technique also avoids the version number problems, because the following is
8150 libxxx.dll.a -> ../bin/cygxxx-5.dll
8153 Linking directly to a dll without using an import lib will work
8154 even when auto-import features are exercised, and even when
8155 @samp{--enable-runtime-pseudo-relocs} is used.
8157 Given the improvements in speed and memory usage, one might justifiably
8158 wonder why import libraries are used at all. There are three reasons:
8160 1. Until recently, the link-directly-to-dll functionality did @emph{not}
8161 work with auto-imported data.
8163 2. Sometimes it is necessary to include pure static objects within the
8164 import library (which otherwise contains only bfd's for indirection
8165 symbols that point to the exports of a dll). Again, the import lib
8166 for the cygwin kernel makes use of this ability, and it is not
8167 possible to do this without an import lib.
8169 3. Symbol aliases can only be resolved using an import lib. This is
8170 critical when linking against OS-supplied dll's (eg, the win32 API)
8171 in which symbols are usually exported as undecorated aliases of their
8172 stdcall-decorated assembly names.
8174 So, import libs are not going away. But the ability to replace
8175 true import libs with a simple symbolic link to (or a copy of)
8176 a dll, in many cases, is a useful addition to the suite of tools
8177 binutils makes available to the win32 developer. Given the
8178 massive improvements in memory requirements during linking, storage
8179 requirements, and linking speed, we expect that many developers
8180 will soon begin to use this feature whenever possible.
8182 @item symbol aliasing
8184 @item adding additional names
8185 Sometimes, it is useful to export symbols with additional names.
8186 A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
8187 exported as @samp{_foo} by using special directives in the DEF file
8188 when creating the dll. This will affect also the optional created
8189 import library. Consider the following DEF file:
8192 LIBRARY "xyz.dll" BASE=0x61000000
8199 The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
8201 Another method for creating a symbol alias is to create it in the
8202 source code using the "weak" attribute:
8205 void foo () @{ /* Do something. */; @}
8206 void _foo () __attribute__ ((weak, alias ("foo")));
8209 See the gcc manual for more information about attributes and weak
8212 @item renaming symbols
8213 Sometimes it is useful to rename exports. For instance, the cygwin
8214 kernel does this regularly. A symbol @samp{_foo} can be exported as
8215 @samp{foo} but not as @samp{_foo} by using special directives in the
8216 DEF file. (This will also affect the import library, if it is
8217 created). In the following example:
8220 LIBRARY "xyz.dll" BASE=0x61000000
8226 The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
8230 Note: using a DEF file disables the default auto-export behavior,
8231 unless the @samp{--export-all-symbols} command-line option is used.
8232 If, however, you are trying to rename symbols, then you should list
8233 @emph{all} desired exports in the DEF file, including the symbols
8234 that are not being renamed, and do @emph{not} use the
8235 @samp{--export-all-symbols} option. If you list only the
8236 renamed symbols in the DEF file, and use @samp{--export-all-symbols}
8237 to handle the other symbols, then the both the new names @emph{and}
8238 the original names for the renamed symbols will be exported.
8239 In effect, you'd be aliasing those symbols, not renaming them,
8240 which is probably not what you wanted.
8242 @cindex weak externals
8243 @item weak externals
8244 The Windows object format, PE, specifies a form of weak symbols called
8245 weak externals. When a weak symbol is linked and the symbol is not
8246 defined, the weak symbol becomes an alias for some other symbol. There
8247 are three variants of weak externals:
8249 @item Definition is searched for in objects and libraries, historically
8250 called lazy externals.
8251 @item Definition is searched for only in other objects, not in libraries.
8252 This form is not presently implemented.
8253 @item No search; the symbol is an alias. This form is not presently
8256 As a GNU extension, weak symbols that do not specify an alternate symbol
8257 are supported. If the symbol is undefined when linking, the symbol
8258 uses a default value.
8260 @cindex aligned common symbols
8261 @item aligned common symbols
8262 As a GNU extension to the PE file format, it is possible to specify the
8263 desired alignment for a common symbol. This information is conveyed from
8264 the assembler or compiler to the linker by means of GNU-specific commands
8265 carried in the object file's @samp{.drectve} section, which are recognized
8266 by @command{ld} and respected when laying out the common symbols. Native
8267 tools will be able to process object files employing this GNU extension,
8268 but will fail to respect the alignment instructions, and may issue noisy
8269 warnings about unknown linker directives.
8284 @section @code{ld} and Xtensa Processors
8286 @cindex Xtensa processors
8287 The default @command{ld} behavior for Xtensa processors is to interpret
8288 @code{SECTIONS} commands so that lists of explicitly named sections in a
8289 specification with a wildcard file will be interleaved when necessary to
8290 keep literal pools within the range of PC-relative load offsets. For
8291 example, with the command:
8303 @command{ld} may interleave some of the @code{.literal}
8304 and @code{.text} sections from different object files to ensure that the
8305 literal pools are within the range of PC-relative load offsets. A valid
8306 interleaving might place the @code{.literal} sections from an initial
8307 group of files followed by the @code{.text} sections of that group of
8308 files. Then, the @code{.literal} sections from the rest of the files
8309 and the @code{.text} sections from the rest of the files would follow.
8311 @cindex @option{--relax} on Xtensa
8312 @cindex relaxing on Xtensa
8313 Relaxation is enabled by default for the Xtensa version of @command{ld} and
8314 provides two important link-time optimizations. The first optimization
8315 is to combine identical literal values to reduce code size. A redundant
8316 literal will be removed and all the @code{L32R} instructions that use it
8317 will be changed to reference an identical literal, as long as the
8318 location of the replacement literal is within the offset range of all
8319 the @code{L32R} instructions. The second optimization is to remove
8320 unnecessary overhead from assembler-generated ``longcall'' sequences of
8321 @code{L32R}/@code{CALLX@var{n}} when the target functions are within
8322 range of direct @code{CALL@var{n}} instructions.
8324 For each of these cases where an indirect call sequence can be optimized
8325 to a direct call, the linker will change the @code{CALLX@var{n}}
8326 instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
8327 instruction, and remove the literal referenced by the @code{L32R}
8328 instruction if it is not used for anything else. Removing the
8329 @code{L32R} instruction always reduces code size but can potentially
8330 hurt performance by changing the alignment of subsequent branch targets.
8331 By default, the linker will always preserve alignments, either by
8332 switching some instructions between 24-bit encodings and the equivalent
8333 density instructions or by inserting a no-op in place of the @code{L32R}
8334 instruction that was removed. If code size is more important than
8335 performance, the @option{--size-opt} option can be used to prevent the
8336 linker from widening density instructions or inserting no-ops, except in
8337 a few cases where no-ops are required for correctness.
8339 The following Xtensa-specific command-line options can be used to
8342 @cindex Xtensa options
8345 When optimizing indirect calls to direct calls, optimize for code size
8346 more than performance. With this option, the linker will not insert
8347 no-ops or widen density instructions to preserve branch target
8348 alignment. There may still be some cases where no-ops are required to
8349 preserve the correctness of the code.
8357 @ifclear SingleFormat
8362 @cindex object file management
8363 @cindex object formats available
8365 The linker accesses object and archive files using the BFD libraries.
8366 These libraries allow the linker to use the same routines to operate on
8367 object files whatever the object file format. A different object file
8368 format can be supported simply by creating a new BFD back end and adding
8369 it to the library. To conserve runtime memory, however, the linker and
8370 associated tools are usually configured to support only a subset of the
8371 object file formats available. You can use @code{objdump -i}
8372 (@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
8373 list all the formats available for your configuration.
8375 @cindex BFD requirements
8376 @cindex requirements for BFD
8377 As with most implementations, BFD is a compromise between
8378 several conflicting requirements. The major factor influencing
8379 BFD design was efficiency: any time used converting between
8380 formats is time which would not have been spent had BFD not
8381 been involved. This is partly offset by abstraction payback; since
8382 BFD simplifies applications and back ends, more time and care
8383 may be spent optimizing algorithms for a greater speed.
8385 One minor artifact of the BFD solution which you should bear in
8386 mind is the potential for information loss. There are two places where
8387 useful information can be lost using the BFD mechanism: during
8388 conversion and during output. @xref{BFD information loss}.
8391 * BFD outline:: How it works: an outline of BFD
8395 @section How It Works: An Outline of BFD
8396 @cindex opening object files
8397 @include bfdsumm.texi
8400 @node Reporting Bugs
8401 @chapter Reporting Bugs
8402 @cindex bugs in @command{ld}
8403 @cindex reporting bugs in @command{ld}
8405 Your bug reports play an essential role in making @command{ld} reliable.
8407 Reporting a bug may help you by bringing a solution to your problem, or
8408 it may not. But in any case the principal function of a bug report is
8409 to help the entire community by making the next version of @command{ld}
8410 work better. Bug reports are your contribution to the maintenance of
8413 In order for a bug report to serve its purpose, you must include the
8414 information that enables us to fix the bug.
8417 * Bug Criteria:: Have you found a bug?
8418 * Bug Reporting:: How to report bugs
8422 @section Have You Found a Bug?
8423 @cindex bug criteria
8425 If you are not sure whether you have found a bug, here are some guidelines:
8428 @cindex fatal signal
8429 @cindex linker crash
8430 @cindex crash of linker
8432 If the linker gets a fatal signal, for any input whatever, that is a
8433 @command{ld} bug. Reliable linkers never crash.
8435 @cindex error on valid input
8437 If @command{ld} produces an error message for valid input, that is a bug.
8439 @cindex invalid input
8441 If @command{ld} does not produce an error message for invalid input, that
8442 may be a bug. In the general case, the linker can not verify that
8443 object files are correct.
8446 If you are an experienced user of linkers, your suggestions for
8447 improvement of @command{ld} are welcome in any case.
8451 @section How to Report Bugs
8453 @cindex @command{ld} bugs, reporting
8455 A number of companies and individuals offer support for @sc{gnu}
8456 products. If you obtained @command{ld} from a support organization, we
8457 recommend you contact that organization first.
8459 You can find contact information for many support companies and
8460 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
8464 Otherwise, send bug reports for @command{ld} to
8468 The fundamental principle of reporting bugs usefully is this:
8469 @strong{report all the facts}. If you are not sure whether to state a
8470 fact or leave it out, state it!
8472 Often people omit facts because they think they know what causes the
8473 problem and assume that some details do not matter. Thus, you might
8474 assume that the name of a symbol you use in an example does not
8475 matter. Well, probably it does not, but one cannot be sure. Perhaps
8476 the bug is a stray memory reference which happens to fetch from the
8477 location where that name is stored in memory; perhaps, if the name
8478 were different, the contents of that location would fool the linker
8479 into doing the right thing despite the bug. Play it safe and give a
8480 specific, complete example. That is the easiest thing for you to do,
8481 and the most helpful.
8483 Keep in mind that the purpose of a bug report is to enable us to fix
8484 the bug if it is new to us. Therefore, always write your bug reports
8485 on the assumption that the bug has not been reported previously.
8487 Sometimes people give a few sketchy facts and ask, ``Does this ring a
8488 bell?'' This cannot help us fix a bug, so it is basically useless. We
8489 respond by asking for enough details to enable us to investigate.
8490 You might as well expedite matters by sending them to begin with.
8492 To enable us to fix the bug, you should include all these things:
8496 The version of @command{ld}. @command{ld} announces it if you start it with
8497 the @samp{--version} argument.
8499 Without this, we will not know whether there is any point in looking for
8500 the bug in the current version of @command{ld}.
8503 Any patches you may have applied to the @command{ld} source, including any
8504 patches made to the @code{BFD} library.
8507 The type of machine you are using, and the operating system name and
8511 What compiler (and its version) was used to compile @command{ld}---e.g.
8515 The command arguments you gave the linker to link your example and
8516 observe the bug. To guarantee you will not omit something important,
8517 list them all. A copy of the Makefile (or the output from make) is
8520 If we were to try to guess the arguments, we would probably guess wrong
8521 and then we might not encounter the bug.
8524 A complete input file, or set of input files, that will reproduce the
8525 bug. It is generally most helpful to send the actual object files
8526 provided that they are reasonably small. Say no more than 10K. For
8527 bigger files you can either make them available by FTP or HTTP or else
8528 state that you are willing to send the object file(s) to whomever
8529 requests them. (Note - your email will be going to a mailing list, so
8530 we do not want to clog it up with large attachments). But small
8531 attachments are best.
8533 If the source files were assembled using @code{gas} or compiled using
8534 @code{gcc}, then it may be OK to send the source files rather than the
8535 object files. In this case, be sure to say exactly what version of
8536 @code{gas} or @code{gcc} was used to produce the object files. Also say
8537 how @code{gas} or @code{gcc} were configured.
8540 A description of what behavior you observe that you believe is
8541 incorrect. For example, ``It gets a fatal signal.''
8543 Of course, if the bug is that @command{ld} gets a fatal signal, then we
8544 will certainly notice it. But if the bug is incorrect output, we might
8545 not notice unless it is glaringly wrong. You might as well not give us
8546 a chance to make a mistake.
8548 Even if the problem you experience is a fatal signal, you should still
8549 say so explicitly. Suppose something strange is going on, such as, your
8550 copy of @command{ld} is out of sync, or you have encountered a bug in the
8551 C library on your system. (This has happened!) Your copy might crash
8552 and ours would not. If you told us to expect a crash, then when ours
8553 fails to crash, we would know that the bug was not happening for us. If
8554 you had not told us to expect a crash, then we would not be able to draw
8555 any conclusion from our observations.
8558 If you wish to suggest changes to the @command{ld} source, send us context
8559 diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
8560 @samp{-p} option. Always send diffs from the old file to the new file.
8561 If you even discuss something in the @command{ld} source, refer to it by
8562 context, not by line number.
8564 The line numbers in our development sources will not match those in your
8565 sources. Your line numbers would convey no useful information to us.
8568 Here are some things that are not necessary:
8572 A description of the envelope of the bug.
8574 Often people who encounter a bug spend a lot of time investigating
8575 which changes to the input file will make the bug go away and which
8576 changes will not affect it.
8578 This is often time consuming and not very useful, because the way we
8579 will find the bug is by running a single example under the debugger
8580 with breakpoints, not by pure deduction from a series of examples.
8581 We recommend that you save your time for something else.
8583 Of course, if you can find a simpler example to report @emph{instead}
8584 of the original one, that is a convenience for us. Errors in the
8585 output will be easier to spot, running under the debugger will take
8586 less time, and so on.
8588 However, simplification is not vital; if you do not want to do this,
8589 report the bug anyway and send us the entire test case you used.
8592 A patch for the bug.
8594 A patch for the bug does help us if it is a good one. But do not omit
8595 the necessary information, such as the test case, on the assumption that
8596 a patch is all we need. We might see problems with your patch and decide
8597 to fix the problem another way, or we might not understand it at all.
8599 Sometimes with a program as complicated as @command{ld} it is very hard to
8600 construct an example that will make the program follow a certain path
8601 through the code. If you do not send us the example, we will not be
8602 able to construct one, so we will not be able to verify that the bug is
8605 And if we cannot understand what bug you are trying to fix, or why your
8606 patch should be an improvement, we will not install it. A test case will
8607 help us to understand.
8610 A guess about what the bug is or what it depends on.
8612 Such guesses are usually wrong. Even we cannot guess right about such
8613 things without first using the debugger to find the facts.
8617 @appendix MRI Compatible Script Files
8618 @cindex MRI compatibility
8619 To aid users making the transition to @sc{gnu} @command{ld} from the MRI
8620 linker, @command{ld} can use MRI compatible linker scripts as an
8621 alternative to the more general-purpose linker scripting language
8622 described in @ref{Scripts}. MRI compatible linker scripts have a much
8623 simpler command set than the scripting language otherwise used with
8624 @command{ld}. @sc{gnu} @command{ld} supports the most commonly used MRI
8625 linker commands; these commands are described here.
8627 In general, MRI scripts aren't of much use with the @code{a.out} object
8628 file format, since it only has three sections and MRI scripts lack some
8629 features to make use of them.
8631 You can specify a file containing an MRI-compatible script using the
8632 @samp{-c} command-line option.
8634 Each command in an MRI-compatible script occupies its own line; each
8635 command line starts with the keyword that identifies the command (though
8636 blank lines are also allowed for punctuation). If a line of an
8637 MRI-compatible script begins with an unrecognized keyword, @command{ld}
8638 issues a warning message, but continues processing the script.
8640 Lines beginning with @samp{*} are comments.
8642 You can write these commands using all upper-case letters, or all
8643 lower case; for example, @samp{chip} is the same as @samp{CHIP}.
8644 The following list shows only the upper-case form of each command.
8647 @cindex @code{ABSOLUTE} (MRI)
8648 @item ABSOLUTE @var{secname}
8649 @itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
8650 Normally, @command{ld} includes in the output file all sections from all
8651 the input files. However, in an MRI-compatible script, you can use the
8652 @code{ABSOLUTE} command to restrict the sections that will be present in
8653 your output program. If the @code{ABSOLUTE} command is used at all in a
8654 script, then only the sections named explicitly in @code{ABSOLUTE}
8655 commands will appear in the linker output. You can still use other
8656 input sections (whatever you select on the command line, or using
8657 @code{LOAD}) to resolve addresses in the output file.
8659 @cindex @code{ALIAS} (MRI)
8660 @item ALIAS @var{out-secname}, @var{in-secname}
8661 Use this command to place the data from input section @var{in-secname}
8662 in a section called @var{out-secname} in the linker output file.
8664 @var{in-secname} may be an integer.
8666 @cindex @code{ALIGN} (MRI)
8667 @item ALIGN @var{secname} = @var{expression}
8668 Align the section called @var{secname} to @var{expression}. The
8669 @var{expression} should be a power of two.
8671 @cindex @code{BASE} (MRI)
8672 @item BASE @var{expression}
8673 Use the value of @var{expression} as the lowest address (other than
8674 absolute addresses) in the output file.
8676 @cindex @code{CHIP} (MRI)
8677 @item CHIP @var{expression}
8678 @itemx CHIP @var{expression}, @var{expression}
8679 This command does nothing; it is accepted only for compatibility.
8681 @cindex @code{END} (MRI)
8683 This command does nothing whatever; it's only accepted for compatibility.
8685 @cindex @code{FORMAT} (MRI)
8686 @item FORMAT @var{output-format}
8687 Similar to the @code{OUTPUT_FORMAT} command in the more general linker
8688 language, but restricted to S-records, if @var{output-format} is @samp{S}
8690 @cindex @code{LIST} (MRI)
8691 @item LIST @var{anything}@dots{}
8692 Print (to the standard output file) a link map, as produced by the
8693 @command{ld} command-line option @samp{-M}.
8695 The keyword @code{LIST} may be followed by anything on the
8696 same line, with no change in its effect.
8698 @cindex @code{LOAD} (MRI)
8699 @item LOAD @var{filename}
8700 @itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
8701 Include one or more object file @var{filename} in the link; this has the
8702 same effect as specifying @var{filename} directly on the @command{ld}
8705 @cindex @code{NAME} (MRI)
8706 @item NAME @var{output-name}
8707 @var{output-name} is the name for the program produced by @command{ld}; the
8708 MRI-compatible command @code{NAME} is equivalent to the command-line
8709 option @samp{-o} or the general script language command @code{OUTPUT}.
8711 @cindex @code{ORDER} (MRI)
8712 @item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
8713 @itemx ORDER @var{secname} @var{secname} @var{secname}
8714 Normally, @command{ld} orders the sections in its output file in the
8715 order in which they first appear in the input files. In an MRI-compatible
8716 script, you can override this ordering with the @code{ORDER} command. The
8717 sections you list with @code{ORDER} will appear first in your output
8718 file, in the order specified.
8720 @cindex @code{PUBLIC} (MRI)
8721 @item PUBLIC @var{name}=@var{expression}
8722 @itemx PUBLIC @var{name},@var{expression}
8723 @itemx PUBLIC @var{name} @var{expression}
8724 Supply a value (@var{expression}) for external symbol
8725 @var{name} used in the linker input files.
8727 @cindex @code{SECT} (MRI)
8728 @item SECT @var{secname}, @var{expression}
8729 @itemx SECT @var{secname}=@var{expression}
8730 @itemx SECT @var{secname} @var{expression}
8731 You can use any of these three forms of the @code{SECT} command to
8732 specify the start address (@var{expression}) for section @var{secname}.
8733 If you have more than one @code{SECT} statement for the same
8734 @var{secname}, only the @emph{first} sets the start address.
8737 @node GNU Free Documentation License
8738 @appendix GNU Free Documentation License
8742 @unnumbered LD Index
8747 % I think something like @@colophon should be in texinfo. In the
8749 \long\def\colophon{\hbox to0pt{}\vfill
8750 \centerline{The body of this manual is set in}
8751 \centerline{\fontname\tenrm,}
8752 \centerline{with headings in {\bf\fontname\tenbf}}
8753 \centerline{and examples in {\tt\fontname\tentt}.}
8754 \centerline{{\it\fontname\tenit\/} and}
8755 \centerline{{\sl\fontname\tensl\/}}
8756 \centerline{are used for emphasis.}\vfill}
8758 % Blame: doc@@cygnus.com, 28mar91.