OpenCores
URL https://opencores.org/ocsvn/openrisc_me/openrisc_me/trunk

Subversion Repositories openrisc_me

[/] [openrisc/] [trunk/] [gnu-src/] [binutils-2.20.1/] [ld/] [ld.texinfo] - Blame information for rev 300

Go to most recent revision | Details | Compare with Previous | View Log

Line No. Rev Author Line
1 205 julius
\input texinfo
2
@setfilename ld.info
3
@c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4
@c 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
5
@c Free Software Foundation, Inc.
6
@syncodeindex ky cp
7
@c man begin INCLUDE
8
@include configdoc.texi
9
@c (configdoc.texi is generated by the Makefile)
10
@include bfdver.texi
11
@c man end
12
 
13
@c @smallbook
14
 
15
@macro gcctabopt{body}
16
@code{\body\}
17
@end macro
18
 
19
@c man begin NAME
20
@ifset man
21
@c Configure for the generation of man pages
22
@set UsesEnvVars
23
@set GENERIC
24
@set ARM
25
@set H8300
26
@set HPPA
27
@set I960
28
@set M68HC11
29
@set M68K
30
@set MMIX
31
@set MSP430
32
@set POWERPC
33
@set POWERPC64
34
@set Renesas
35
@set SPU
36
@set TICOFF
37
@set WIN32
38
@set XTENSA
39
@end ifset
40
@c man end
41
 
42
@ifinfo
43
@format
44
START-INFO-DIR-ENTRY
45
* Ld: (ld).                       The GNU linker.
46
END-INFO-DIR-ENTRY
47
@end format
48
@end ifinfo
49
 
50
@copying
51
This file documents the @sc{gnu} linker LD
52
@ifset VERSION_PACKAGE
53
@value{VERSION_PACKAGE}
54
@end ifset
55
version @value{VERSION}.
56
 
57
Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000,
58
2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
59
 
60
Permission is granted to copy, distribute and/or modify this document
61
under the terms of the GNU Free Documentation License, Version 1.3
62
or any later version published by the Free Software Foundation;
63
with no Invariant Sections, with no Front-Cover Texts, and with no
64
Back-Cover Texts.  A copy of the license is included in the
65
section entitled ``GNU Free Documentation License''.
66
@end copying
67
@iftex
68
@finalout
69
@setchapternewpage odd
70
@settitle The GNU linker
71
@titlepage
72
@title The GNU linker
73
@sp 1
74
@subtitle @code{ld}
75
@ifset VERSION_PACKAGE
76
@subtitle @value{VERSION_PACKAGE}
77
@end ifset
78
@subtitle Version @value{VERSION}
79
@author Steve Chamberlain
80
@author Ian Lance Taylor
81
@page
82
 
83
@tex
84
{\parskip=0pt
85
\hfill Red Hat Inc\par
86
\hfill nickc\@credhat.com, doc\@redhat.com\par
87
\hfill {\it The GNU linker}\par
88
\hfill Edited by Jeffrey Osier (jeffrey\@cygnus.com)\par
89
}
90
\global\parindent=0pt % Steve likes it this way.
91
@end tex
92
 
93
@vskip 0pt plus 1filll
94
@c man begin COPYRIGHT
95
Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 98, 99, 2000, 2001,
96
2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
97
 
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''.
104
@c man end
105
 
106
@end titlepage
107
@end iftex
108
@contents
109
@c FIXME: Talk about importance of *order* of args, cmds to linker!
110
 
111
@ifnottex
112
@node Top
113
@top LD
114
This file documents the @sc{gnu} linker ld
115
@ifset VERSION_PACKAGE
116
@value{VERSION_PACKAGE}
117
@end ifset
118
version @value{VERSION}.
119
 
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''.
123
 
124
@menu
125
* Overview::                    Overview
126
* Invocation::                  Invocation
127
* Scripts::                     Linker Scripts
128
@ifset GENERIC
129
* Machine Dependent::           Machine Dependent Features
130
@end ifset
131
@ifclear GENERIC
132
@ifset H8300
133
* H8/300::                      ld and the H8/300
134
@end ifset
135
@ifset Renesas
136
* Renesas::                     ld and other Renesas micros
137
@end ifset
138
@ifset I960
139
* i960::                        ld and the Intel 960 family
140
@end ifset
141
@ifset ARM
142
* ARM::                         ld and the ARM family
143
@end ifset
144
@ifset HPPA
145
* HPPA ELF32::                  ld and HPPA 32-bit ELF
146
@end ifset
147
@ifset M68HC11
148
* M68HC11/68HC12::              ld and the Motorola 68HC11 and 68HC12 families
149
@end ifset
150
@ifset M68K
151
* M68K::                        ld and Motorola 68K family
152
@end ifset
153
@ifset POWERPC
154
* PowerPC ELF32::               ld and PowerPC 32-bit ELF Support
155
@end ifset
156
@ifset POWERPC64
157
* PowerPC64 ELF64::             ld and PowerPC64 64-bit ELF Support
158
@end ifset
159
@ifset SPU
160
* SPU ELF::                     ld and SPU ELF Support
161
@end ifset
162
@ifset TICOFF
163
* TI COFF::                     ld and the TI COFF
164
@end ifset
165
@ifset WIN32
166
* Win32::                       ld and WIN32 (cygwin/mingw)
167
@end ifset
168
@ifset XTENSA
169
* Xtensa::                      ld and Xtensa Processors
170
@end ifset
171
@end ifclear
172
@ifclear SingleFormat
173
* BFD::                         BFD
174
@end ifclear
175
@c Following blank line required for remaining bug in makeinfo conds/menus
176
 
177
* Reporting Bugs::              Reporting Bugs
178
* MRI::                         MRI Compatible Script Files
179
* GNU Free Documentation License::  GNU Free Documentation License
180
* LD Index::                       LD Index
181
@end menu
182
@end ifnottex
183
 
184
@node Overview
185
@chapter Overview
186
 
187
@cindex @sc{gnu} linker
188
@cindex what is this?
189
 
190
@ifset man
191
@c man begin SYNOPSIS
192
ld [@b{options}] @var{objfile} @dots{}
193
@c man end
194
 
195
@c man begin SEEALSO
196
ar(1), nm(1), objcopy(1), objdump(1), readelf(1) and
197
the Info entries for @file{binutils} and
198
@file{ld}.
199
@c man end
200
@end ifset
201
 
202
@c man begin DESCRIPTION
203
 
204
@command{ld} combines a number of object and archive files, relocates
205
their data and ties up symbol references. Usually the last step in
206
compiling a program is to run @command{ld}.
207
 
208
@command{ld} accepts Linker Command Language files written in
209
a superset of AT&T's Link Editor Command Language syntax,
210
to provide explicit and total control over the linking process.
211
 
212
@ifset man
213
@c For the man only
214
This man page does not describe the command language; see the
215
@command{ld} entry in @code{info} for full details on the command
216
language and on other aspects of the GNU linker.
217
@end ifset
218
 
219
@ifclear SingleFormat
220
This version of @command{ld} uses the general purpose BFD libraries
221
to operate on object files. This allows @command{ld} to read, combine, and
222
write object files in many different formats---for example, COFF or
223
@code{a.out}.  Different formats may be linked together to produce any
224
available kind of object file.  @xref{BFD}, for more information.
225
@end ifclear
226
 
227
Aside from its flexibility, the @sc{gnu} linker is more helpful than other
228
linkers in providing diagnostic information.  Many linkers abandon
229
execution immediately upon encountering an error; whenever possible,
230
@command{ld} continues executing, allowing you to identify other errors
231
(or, in some cases, to get an output file in spite of the error).
232
 
233
@c man end
234
 
235
@node Invocation
236
@chapter Invocation
237
 
238
@c man begin DESCRIPTION
239
 
240
The @sc{gnu} linker @command{ld} is meant to cover a broad range of situations,
241
and to be as compatible as possible with other linkers.  As a result,
242
you have many choices to control its behavior.
243
 
244
@c man end
245
 
246
@ifset UsesEnvVars
247
@menu
248
* Options::                     Command Line Options
249
* Environment::                 Environment Variables
250
@end menu
251
 
252
@node Options
253
@section Command Line Options
254
@end ifset
255
 
256
@cindex command line
257
@cindex options
258
 
259
@c man begin OPTIONS
260
 
261
The linker supports a plethora of command-line options, but in actual
262
practice few of them are used in any particular context.
263
@cindex standard Unix system
264
For instance, a frequent use of @command{ld} is to link standard Unix
265
object files on a standard, supported Unix system.  On such a system, to
266
link a file @code{hello.o}:
267
 
268
@smallexample
269
ld -o @var{output} /lib/crt0.o hello.o -lc
270
@end smallexample
271
 
272
This tells @command{ld} to produce a file called @var{output} as the
273
result of linking the file @code{/lib/crt0.o} with @code{hello.o} and
274
the library @code{libc.a}, which will come from the standard search
275
directories.  (See the discussion of the @samp{-l} option below.)
276
 
277
Some of the command-line options to @command{ld} may be specified at any
278
point in the command line.  However, options which refer to files, such
279
as @samp{-l} or @samp{-T}, cause the file to be read at the point at
280
which the option appears in the command line, relative to the object
281
files and other file options.  Repeating non-file options with a
282
different argument will either have no further effect, or override prior
283
occurrences (those further to the left on the command line) of that
284
option.  Options which may be meaningfully specified more than once are
285
noted in the descriptions below.
286
 
287
@cindex object files
288
Non-option arguments are object files or archives which are to be linked
289
together.  They may follow, precede, or be mixed in with command-line
290
options, except that an object file argument may not be placed between
291
an option and its argument.
292
 
293
Usually the linker is invoked with at least one object file, but you can
294
specify other forms of binary input files using @samp{-l}, @samp{-R},
295
and the script command language.  If @emph{no} binary input files at all
296
are specified, the linker does not produce any output, and issues the
297
message @samp{No input files}.
298
 
299
If the linker cannot recognize the format of an object file, it will
300
assume that it is a linker script.  A script specified in this way
301
augments the main linker script used for the link (either the default
302
linker script or the one specified by using @samp{-T}).  This feature
303
permits the linker to link against a file which appears to be an object
304
or an archive, but actually merely defines some symbol values, or uses
305
@code{INPUT} or @code{GROUP} to load other objects.  Specifying a
306
script in this way merely augments the main linker script, with the
307
extra commands placed after the main script; use the @samp{-T} option
308
to replace the default linker script entirely, but note the effect of
309
the @code{INSERT} command.  @xref{Scripts}.
310
 
311
For options whose names are a single letter,
312
option arguments must either follow the option letter without intervening
313
whitespace, or be given as separate arguments immediately following the
314
option that requires them.
315
 
316
For options whose names are multiple letters, either one dash or two can
317
precede the option name; for example, @samp{-trace-symbol} and
318
@samp{--trace-symbol} are equivalent.  Note---there is one exception to
319
this rule.  Multiple letter options that start with a lower case 'o' can
320
only be preceded by two dashes.  This is to reduce confusion with the
321
@samp{-o} option.  So for example @samp{-omagic} sets the output file
322
name to @samp{magic} whereas @samp{--omagic} sets the NMAGIC flag on the
323
output.
324
 
325
Arguments to multiple-letter options must either be separated from the
326
option name by an equals sign, or be given as separate arguments
327
immediately following the option that requires them.  For example,
328
@samp{--trace-symbol foo} and @samp{--trace-symbol=foo} are equivalent.
329
Unique abbreviations of the names of multiple-letter options are
330
accepted.
331
 
332
Note---if the linker is being invoked indirectly, via a compiler driver
333
(e.g. @samp{gcc}) then all the linker command line options should be
334
prefixed by @samp{-Wl,} (or whatever is appropriate for the particular
335
compiler driver) like this:
336
 
337
@smallexample
338
  gcc -Wl,--start-group foo.o bar.o -Wl,--end-group
339
@end smallexample
340
 
341
This is important, because otherwise the compiler driver program may
342
silently drop the linker options, resulting in a bad link.  Confusion
343
may also arise when passing options that require values through a
344
driver, as the use of a space between option and argument acts as
345
a separator, and causes the driver to pass only the option to the linker
346
and the argument to the compiler.  In this case, it is simplest to use
347
the joined forms of both single- and multiple-letter options, such as:
348
 
349
@smallexample
350
  gcc foo.o bar.o -Wl,-eENTRY -Wl,-Map=a.map
351
@end smallexample
352
 
353
Here is a table of the generic command line switches accepted by the GNU
354
linker:
355
 
356
@table @gcctabopt
357
@include at-file.texi
358
 
359
@kindex -a @var{keyword}
360
@item -a @var{keyword}
361
This option is supported for HP/UX compatibility.  The @var{keyword}
362
argument must be one of the strings @samp{archive}, @samp{shared}, or
363
@samp{default}.  @samp{-aarchive} is functionally equivalent to
364
@samp{-Bstatic}, and the other two keywords are functionally equivalent
365
to @samp{-Bdynamic}.  This option may be used any number of times.
366
 
367
@ifset I960
368
@cindex architectures
369
@kindex -A @var{arch}
370
@item -A @var{architecture}
371
@kindex --architecture=@var{arch}
372
@itemx --architecture=@var{architecture}
373
In the current release of @command{ld}, this option is useful only for the
374
Intel 960 family of architectures.  In that @command{ld} configuration, the
375
@var{architecture} argument identifies the particular architecture in
376
the 960 family, enabling some safeguards and modifying the
377
archive-library search path.  @xref{i960,,@command{ld} and the Intel 960
378
family}, for details.
379
 
380
Future releases of @command{ld} may support similar functionality for
381
other architecture families.
382
@end ifset
383
 
384
@ifclear SingleFormat
385
@cindex binary input format
386
@kindex -b @var{format}
387
@kindex --format=@var{format}
388
@cindex input format
389
@cindex input format
390
@item -b @var{input-format}
391
@itemx --format=@var{input-format}
392
@command{ld} may be configured to support more than one kind of object
393
file.  If your @command{ld} is configured this way, you can use the
394
@samp{-b} option to specify the binary format for input object files
395
that follow this option on the command line.  Even when @command{ld} is
396
configured to support alternative object formats, you don't usually need
397
to specify this, as @command{ld} should be configured to expect as a
398
default input format the most usual format on each machine.
399
@var{input-format} is a text string, the name of a particular format
400
supported by the BFD libraries.  (You can list the available binary
401
formats with @samp{objdump -i}.)
402
@xref{BFD}.
403
 
404
You may want to use this option if you are linking files with an unusual
405
binary format.  You can also use @samp{-b} to switch formats explicitly (when
406
linking object files of different formats), by including
407
@samp{-b @var{input-format}} before each group of object files in a
408
particular format.
409
 
410
The default format is taken from the environment variable
411
@code{GNUTARGET}.
412
@ifset UsesEnvVars
413
@xref{Environment}.
414
@end ifset
415
You can also define the input format from a script, using the command
416
@code{TARGET};
417
@ifclear man
418
see @ref{Format Commands}.
419
@end ifclear
420
@end ifclear
421
 
422
@kindex -c @var{MRI-cmdfile}
423
@kindex --mri-script=@var{MRI-cmdfile}
424
@cindex compatibility, MRI
425
@item -c @var{MRI-commandfile}
426
@itemx --mri-script=@var{MRI-commandfile}
427
For compatibility with linkers produced by MRI, @command{ld} accepts script
428
files written in an alternate, restricted command language, described in
429
@ifclear man
430
@ref{MRI,,MRI Compatible Script Files}.
431
@end ifclear
432
@ifset man
433
the MRI Compatible Script Files section of GNU ld documentation.
434
@end ifset
435
Introduce MRI script files with
436
the option @samp{-c}; use the @samp{-T} option to run linker
437
scripts written in the general-purpose @command{ld} scripting language.
438
If @var{MRI-cmdfile} does not exist, @command{ld} looks for it in the directories
439
specified by any @samp{-L} options.
440
 
441
@cindex common allocation
442
@kindex -d
443
@kindex -dc
444
@kindex -dp
445
@item -d
446
@itemx -dc
447
@itemx -dp
448
These three options are equivalent; multiple forms are supported for
449
compatibility with other linkers.  They assign space to common symbols
450
even if a relocatable output file is specified (with @samp{-r}).  The
451
script command @code{FORCE_COMMON_ALLOCATION} has the same effect.
452
@xref{Miscellaneous Commands}.
453
 
454
@cindex entry point, from command line
455
@kindex -e @var{entry}
456
@kindex --entry=@var{entry}
457
@item -e @var{entry}
458
@itemx --entry=@var{entry}
459
Use @var{entry} as the explicit symbol for beginning execution of your
460
program, rather than the default entry point.  If there is no symbol
461
named @var{entry}, the linker will try to parse @var{entry} as a number,
462
and use that as the entry address (the number will be interpreted in
463
base 10; you may use a leading @samp{0x} for base 16, or a leading
464
@samp{0} for base 8).  @xref{Entry Point}, for a discussion of defaults
465
and other ways of specifying the entry point.
466
 
467
@kindex --exclude-libs
468
@item --exclude-libs @var{lib},@var{lib},...
469
Specifies a list of archive libraries from which symbols should not be automatically
470
exported.  The library names may be delimited by commas or colons.  Specifying
471
@code{--exclude-libs ALL} excludes symbols in all archive libraries from
472
automatic export.  This option is available only for the i386 PE targeted
473
port of the linker and for ELF targeted ports.  For i386 PE, symbols
474
explicitly listed in a .def file are still exported, regardless of this
475
option.  For ELF targeted ports, symbols affected by this option will
476
be treated as hidden.
477
 
478
@kindex --exclude-modules-for-implib
479
@item --exclude-modules-for-implib @var{module},@var{module},...
480
Specifies a list of object files or archive members, from which symbols
481
should not be automatically exported, but which should be copied wholesale
482
into the import library being generated during the link.  The module names
483
may be delimited by commas or colons, and must match exactly the filenames
484
used by @command{ld} to open the files; for archive members, this is simply
485
the member name, but for object files the name listed must include and
486
match precisely any path used to specify the input file on the linker's
487
command-line.  This option is available only for the i386 PE targeted port
488
of the linker.  Symbols explicitly listed in a .def file are still exported,
489
regardless of this option.
490
 
491
@cindex dynamic symbol table
492
@kindex -E
493
@kindex --export-dynamic
494
@kindex --no-export-dynamic
495
@item -E
496
@itemx --export-dynamic
497
@itemx --no-export-dynamic
498
When creating a dynamically linked executable, using the @option{-E}
499
option or the @option{--export-dynamic} option causes the linker to add
500
all symbols to the dynamic symbol table.  The dynamic symbol table is the
501
set of symbols which are visible from dynamic objects at run time.
502
 
503
If you do not use either of these options (or use the
504
@option{--no-export-dynamic} option to restore the default behavior), the
505
dynamic symbol table will normally contain only those symbols which are
506
referenced by some dynamic object mentioned in the link.
507
 
508
If you use @code{dlopen} to load a dynamic object which needs to refer
509
back to the symbols defined by the program, rather than some other
510
dynamic object, then you will probably need to use this option when
511
linking the program itself.
512
 
513
You can also use the dynamic list to control what symbols should
514
be added to the dynamic symbol table if the output format supports it.
515
See the description of @samp{--dynamic-list}.
516
 
517
Note that this option is specific to ELF targeted ports.  PE targets
518
support a similar function to export all symbols from a DLL or EXE; see
519
the description of @samp{--export-all-symbols} below.
520
 
521
@ifclear SingleFormat
522
@cindex big-endian objects
523
@cindex endianness
524
@kindex -EB
525
@item -EB
526
Link big-endian objects.  This affects the default output format.
527
 
528
@cindex little-endian objects
529
@kindex -EL
530
@item -EL
531
Link little-endian objects.  This affects the default output format.
532
@end ifclear
533
 
534
@kindex -f @var{name}
535
@kindex --auxiliary=@var{name}
536
@item -f @var{name}
537
@itemx --auxiliary=@var{name}
538
When creating an ELF shared object, set the internal DT_AUXILIARY field
539
to the specified name.  This tells the dynamic linker that the symbol
540
table of the shared object should be used as an auxiliary filter on the
541
symbol table of the shared object @var{name}.
542
 
543
If you later link a program against this filter object, then, when you
544
run the program, the dynamic linker will see the DT_AUXILIARY field.  If
545
the dynamic linker resolves any symbols from the filter object, it will
546
first check whether there is a definition in the shared object
547
@var{name}.  If there is one, it will be used instead of the definition
548
in the filter object.  The shared object @var{name} need not exist.
549
Thus the shared object @var{name} may be used to provide an alternative
550
implementation of certain functions, perhaps for debugging or for
551
machine specific performance.
552
 
553
This option may be specified more than once.  The DT_AUXILIARY entries
554
will be created in the order in which they appear on the command line.
555
 
556
@kindex -F @var{name}
557
@kindex --filter=@var{name}
558
@item -F @var{name}
559
@itemx --filter=@var{name}
560
When creating an ELF shared object, set the internal DT_FILTER field to
561
the specified name.  This tells the dynamic linker that the symbol table
562
of the shared object which is being created should be used as a filter
563
on the symbol table of the shared object @var{name}.
564
 
565
If you later link a program against this filter object, then, when you
566
run the program, the dynamic linker will see the DT_FILTER field.  The
567
dynamic linker will resolve symbols according to the symbol table of the
568
filter object as usual, but it will actually link to the definitions
569
found in the shared object @var{name}.  Thus the filter object can be
570
used to select a subset of the symbols provided by the object
571
@var{name}.
572
 
573
Some older linkers used the @option{-F} option throughout a compilation
574
toolchain for specifying object-file format for both input and output
575
object files.
576
@ifclear SingleFormat
577
The @sc{gnu} linker uses other mechanisms for this purpose: the
578
@option{-b}, @option{--format}, @option{--oformat} options, the
579
@code{TARGET} command in linker scripts, and the @code{GNUTARGET}
580
environment variable.
581
@end ifclear
582
The @sc{gnu} linker will ignore the @option{-F} option when not
583
creating an ELF shared object.
584
 
585
@cindex finalization function
586
@kindex -fini=@var{name}
587
@item -fini=@var{name}
588
When creating an ELF executable or shared object, call NAME when the
589
executable or shared object is unloaded, by setting DT_FINI to the
590
address of the function.  By default, the linker uses @code{_fini} as
591
the function to call.
592
 
593
@kindex -g
594
@item -g
595
Ignored.  Provided for compatibility with other tools.
596
 
597
@kindex -G @var{value}
598
@kindex --gpsize=@var{value}
599
@cindex object size
600
@item -G @var{value}
601
@itemx --gpsize=@var{value}
602
Set the maximum size of objects to be optimized using the GP register to
603
@var{size}.  This is only meaningful for object file formats such as
604
MIPS ECOFF which supports putting large and small objects into different
605
sections.  This is ignored for other object file formats.
606
 
607
@cindex runtime library name
608
@kindex -h @var{name}
609
@kindex -soname=@var{name}
610
@item -h @var{name}
611
@itemx -soname=@var{name}
612
When creating an ELF shared object, set the internal DT_SONAME field to
613
the specified name.  When an executable is linked with a shared object
614
which has a DT_SONAME field, then when the executable is run the dynamic
615
linker will attempt to load the shared object specified by the DT_SONAME
616
field rather than the using the file name given to the linker.
617
 
618
@kindex -i
619
@cindex incremental link
620
@item -i
621
Perform an incremental link (same as option @samp{-r}).
622
 
623
@cindex initialization function
624
@kindex -init=@var{name}
625
@item -init=@var{name}
626
When creating an ELF executable or shared object, call NAME when the
627
executable or shared object is loaded, by setting DT_INIT to the address
628
of the function.  By default, the linker uses @code{_init} as the
629
function to call.
630
 
631
@cindex archive files, from cmd line
632
@kindex -l @var{namespec}
633
@kindex --library=@var{namespec}
634
@item -l @var{namespec}
635
@itemx --library=@var{namespec}
636
Add the archive or object file specified by @var{namespec} to the
637
list of files to link.  This option may be used any number of times.
638
If @var{namespec} is of the form @file{:@var{filename}}, @command{ld}
639
will search the library path for a file called @var{filename}, otherwise it
640
will search the library path for a file called @file{lib@var{namespec}.a}.
641
 
642
On systems which support shared libraries, @command{ld} may also search for
643
files other than @file{lib@var{namespec}.a}.  Specifically, on ELF
644
and SunOS systems, @command{ld} will search a directory for a library
645
called @file{lib@var{namespec}.so} before searching for one called
646
@file{lib@var{namespec}.a}.  (By convention, a @code{.so} extension
647
indicates a shared library.)  Note that this behavior does not apply
648
to @file{:@var{filename}}, which always specifies a file called
649
@var{filename}.
650
 
651
The linker will search an archive only once, at the location where it is
652
specified on the command line.  If the archive defines a symbol which
653
was undefined in some object which appeared before the archive on the
654
command line, the linker will include the appropriate file(s) from the
655
archive.  However, an undefined symbol in an object appearing later on
656
the command line will not cause the linker to search the archive again.
657
 
658
See the @option{-(} option for a way to force the linker to search
659
archives multiple times.
660
 
661
You may list the same archive multiple times on the command line.
662
 
663
@ifset GENERIC
664
This type of archive searching is standard for Unix linkers.  However,
665
if you are using @command{ld} on AIX, note that it is different from the
666
behaviour of the AIX linker.
667
@end ifset
668
 
669
@cindex search directory, from cmd line
670
@kindex -L @var{dir}
671
@kindex --library-path=@var{dir}
672
@item -L @var{searchdir}
673
@itemx --library-path=@var{searchdir}
674
Add path @var{searchdir} to the list of paths that @command{ld} will search
675
for archive libraries and @command{ld} control scripts.  You may use this
676
option any number of times.  The directories are searched in the order
677
in which they are specified on the command line.  Directories specified
678
on the command line are searched before the default directories.  All
679
@option{-L} options apply to all @option{-l} options, regardless of the
680
order in which the options appear.  @option{-L} options do not affect
681
how @command{ld} searches for a linker script unless @option{-T}
682
option is specified.
683
 
684
If @var{searchdir} begins with @code{=}, then the @code{=} will be replaced
685
by the @dfn{sysroot prefix}, a path specified when the linker is configured.
686
 
687
@ifset UsesEnvVars
688
The default set of paths searched (without being specified with
689
@samp{-L}) depends on which emulation mode @command{ld} is using, and in
690
some cases also on how it was configured.  @xref{Environment}.
691
@end ifset
692
 
693
The paths can also be specified in a link script with the
694
@code{SEARCH_DIR} command.  Directories specified this way are searched
695
at the point in which the linker script appears in the command line.
696
 
697
@cindex emulation
698
@kindex -m @var{emulation}
699
@item -m @var{emulation}
700
Emulate the @var{emulation} linker.  You can list the available
701
emulations with the @samp{--verbose} or @samp{-V} options.
702
 
703
If the @samp{-m} option is not used, the emulation is taken from the
704
@code{LDEMULATION} environment variable, if that is defined.
705
 
706
Otherwise, the default emulation depends upon how the linker was
707
configured.
708
 
709
@cindex link map
710
@kindex -M
711
@kindex --print-map
712
@item -M
713
@itemx --print-map
714
Print a link map to the standard output.  A link map provides
715
information about the link, including the following:
716
 
717
@itemize @bullet
718
@item
719
Where object files are mapped into memory.
720
@item
721
How common symbols are allocated.
722
@item
723
All archive members included in the link, with a mention of the symbol
724
which caused the archive member to be brought in.
725
@item
726
The values assigned to symbols.
727
 
728
Note - symbols whose values are computed by an expression which
729
involves a reference to a previous value of the same symbol may not
730
have correct result displayed in the link map.  This is because the
731
linker discards intermediate results and only retains the final value
732
of an expression.  Under such circumstances the linker will display
733
the final value enclosed by square brackets.  Thus for example a
734
linker script containing:
735
 
736
@smallexample
737
   foo = 1
738
   foo = foo * 4
739
   foo = foo + 8
740
@end smallexample
741
 
742
will produce the following output in the link map if the @option{-M}
743
option is used:
744
 
745
@smallexample
746
   0x00000001                foo = 0x1
747
   [0x0000000c]                foo = (foo * 0x4)
748
   [0x0000000c]                foo = (foo + 0x8)
749
@end smallexample
750
 
751
See @ref{Expressions} for more information about expressions in linker
752
scripts.
753
@end itemize
754
 
755
@kindex -n
756
@cindex read-only text
757
@cindex NMAGIC
758
@kindex --nmagic
759
@item -n
760
@itemx --nmagic
761
Turn off page alignment of sections, and mark the output as
762
@code{NMAGIC} if possible.
763
 
764
@kindex -N
765
@kindex --omagic
766
@cindex read/write from cmd line
767
@cindex OMAGIC
768
@item -N
769
@itemx --omagic
770
Set the text and data sections to be readable and writable.  Also, do
771
not page-align the data segment, and disable linking against shared
772
libraries.  If the output format supports Unix style magic numbers,
773
mark the output as @code{OMAGIC}. Note: Although a writable text section
774
is allowed for PE-COFF targets, it does not conform to the format
775
specification published by Microsoft.
776
 
777
@kindex --no-omagic
778
@cindex OMAGIC
779
@item --no-omagic
780
This option negates most of the effects of the @option{-N} option.  It
781
sets the text section to be read-only, and forces the data segment to
782
be page-aligned.  Note - this option does not enable linking against
783
shared libraries.  Use @option{-Bdynamic} for this.
784
 
785
@kindex -o @var{output}
786
@kindex --output=@var{output}
787
@cindex naming the output file
788
@item -o @var{output}
789
@itemx --output=@var{output}
790
Use @var{output} as the name for the program produced by @command{ld}; if this
791
option is not specified, the name @file{a.out} is used by default.  The
792
script command @code{OUTPUT} can also specify the output file name.
793
 
794
@kindex -O @var{level}
795
@cindex generating optimized output
796
@item -O @var{level}
797
If @var{level} is a numeric values greater than zero @command{ld} optimizes
798
the output.  This might take significantly longer and therefore probably
799
should only be enabled for the final binary.  At the moment this
800
option only affects ELF shared library generation.  Future releases of
801
the linker may make more use of this option.  Also currently there is
802
no difference in the linker's behaviour for different non-zero values
803
of this option.  Again this may change with future releases.
804
 
805
@kindex -q
806
@kindex --emit-relocs
807
@cindex retain relocations in final executable
808
@item -q
809
@itemx --emit-relocs
810
Leave relocation sections and contents in fully linked executables.
811
Post link analysis and optimization tools may need this information in
812
order to perform correct modifications of executables.  This results
813
in larger executables.
814
 
815
This option is currently only supported on ELF platforms.
816
 
817
@kindex --force-dynamic
818
@cindex forcing the creation of dynamic sections
819
@item --force-dynamic
820
Force the output file to have dynamic sections.  This option is specific
821
to VxWorks targets.
822
 
823
@cindex partial link
824
@cindex relocatable output
825
@kindex -r
826
@kindex --relocatable
827
@item -r
828
@itemx --relocatable
829
Generate relocatable output---i.e., generate an output file that can in
830
turn serve as input to @command{ld}.  This is often called @dfn{partial
831
linking}.  As a side effect, in environments that support standard Unix
832
magic numbers, this option also sets the output file's magic number to
833
@code{OMAGIC}.
834
@c ; see @option{-N}.
835
If this option is not specified, an absolute file is produced.  When
836
linking C++ programs, this option @emph{will not} resolve references to
837
constructors; to do that, use @samp{-Ur}.
838
 
839
When an input file does not have the same format as the output file,
840
partial linking is only supported if that input file does not contain any
841
relocations.  Different output formats can have further restrictions; for
842
example some @code{a.out}-based formats do not support partial linking
843
with input files in other formats at all.
844
 
845
This option does the same thing as @samp{-i}.
846
 
847
@kindex -R @var{file}
848
@kindex --just-symbols=@var{file}
849
@cindex symbol-only input
850
@item -R @var{filename}
851
@itemx --just-symbols=@var{filename}
852
Read symbol names and their addresses from @var{filename}, but do not
853
relocate it or include it in the output.  This allows your output file
854
to refer symbolically to absolute locations of memory defined in other
855
programs.  You may use this option more than once.
856
 
857
For compatibility with other ELF linkers, if the @option{-R} option is
858
followed by a directory name, rather than a file name, it is treated as
859
the @option{-rpath} option.
860
 
861
@kindex -s
862
@kindex --strip-all
863
@cindex strip all symbols
864
@item -s
865
@itemx --strip-all
866
Omit all symbol information from the output file.
867
 
868
@kindex -S
869
@kindex --strip-debug
870
@cindex strip debugger symbols
871
@item -S
872
@itemx --strip-debug
873
Omit debugger symbol information (but not all symbols) from the output file.
874
 
875
@kindex -t
876
@kindex --trace
877
@cindex input files, displaying
878
@item -t
879
@itemx --trace
880
Print the names of the input files as @command{ld} processes them.
881
 
882
@kindex -T @var{script}
883
@kindex --script=@var{script}
884
@cindex script files
885
@item -T @var{scriptfile}
886
@itemx --script=@var{scriptfile}
887
Use @var{scriptfile} as the linker script.  This script replaces
888
@command{ld}'s default linker script (rather than adding to it), so
889
@var{commandfile} must specify everything necessary to describe the
890
output file.  @xref{Scripts}.  If @var{scriptfile} does not exist in
891
the current directory, @code{ld} looks for it in the directories
892
specified by any preceding @samp{-L} options.  Multiple @samp{-T}
893
options accumulate.
894
 
895
@kindex -dT @var{script}
896
@kindex --default-script=@var{script}
897
@cindex script files
898
@item -dT @var{scriptfile}
899
@itemx --default-script=@var{scriptfile}
900
Use @var{scriptfile} as the default linker script.  @xref{Scripts}.
901
 
902
This option is similar to the @option{--script} option except that
903
processing of the script is delayed until after the rest of the
904
command line has been processed.  This allows options placed after the
905
@option{--default-script} option on the command line to affect the
906
behaviour of the linker script, which can be important when the linker
907
command line cannot be directly controlled by the user.  (eg because
908
the command line is being constructed by another tool, such as
909
@samp{gcc}).
910
 
911
@kindex -u @var{symbol}
912
@kindex --undefined=@var{symbol}
913
@cindex undefined symbol
914
@item -u @var{symbol}
915
@itemx --undefined=@var{symbol}
916
Force @var{symbol} to be entered in the output file as an undefined
917
symbol.  Doing this may, for example, trigger linking of additional
918
modules from standard libraries.  @samp{-u} may be repeated with
919
different option arguments to enter additional undefined symbols.  This
920
option is equivalent to the @code{EXTERN} linker script command.
921
 
922
@kindex -Ur
923
@cindex constructors
924
@item -Ur
925
For anything other than C++ programs, this option is equivalent to
926
@samp{-r}: it generates relocatable output---i.e., an output file that can in
927
turn serve as input to @command{ld}.  When linking C++ programs, @samp{-Ur}
928
@emph{does} resolve references to constructors, unlike @samp{-r}.
929
It does not work to use @samp{-Ur} on files that were themselves linked
930
with @samp{-Ur}; once the constructor table has been built, it cannot
931
be added to.  Use @samp{-Ur} only for the last partial link, and
932
@samp{-r} for the others.
933
 
934
@kindex --unique[=@var{SECTION}]
935
@item --unique[=@var{SECTION}]
936
Creates a separate output section for every input section matching
937
@var{SECTION}, or if the optional wildcard @var{SECTION} argument is
938
missing, for every orphan input section.  An orphan section is one not
939
specifically mentioned in a linker script.  You may use this option
940
multiple times on the command line;  It prevents the normal merging of
941
input sections with the same name, overriding output section assignments
942
in a linker script.
943
 
944
@kindex -v
945
@kindex -V
946
@kindex --version
947
@cindex version
948
@item -v
949
@itemx --version
950
@itemx -V
951
Display the version number for @command{ld}.  The @option{-V} option also
952
lists the supported emulations.
953
 
954
@kindex -x
955
@kindex --discard-all
956
@cindex deleting local symbols
957
@item -x
958
@itemx --discard-all
959
Delete all local symbols.
960
 
961
@kindex -X
962
@kindex --discard-locals
963
@cindex local symbols, deleting
964
@item -X
965
@itemx --discard-locals
966
Delete all temporary local symbols.  (These symbols start with
967
system-specific local label prefixes, typically @samp{.L} for ELF systems
968
or @samp{L} for traditional a.out systems.)
969
 
970
@kindex -y @var{symbol}
971
@kindex --trace-symbol=@var{symbol}
972
@cindex symbol tracing
973
@item -y @var{symbol}
974
@itemx --trace-symbol=@var{symbol}
975
Print the name of each linked file in which @var{symbol} appears.  This
976
option may be given any number of times.  On many systems it is necessary
977
to prepend an underscore.
978
 
979
This option is useful when you have an undefined symbol in your link but
980
don't know where the reference is coming from.
981
 
982
@kindex -Y @var{path}
983
@item -Y @var{path}
984
Add @var{path} to the default library search path.  This option exists
985
for Solaris compatibility.
986
 
987
@kindex -z @var{keyword}
988
@item -z @var{keyword}
989
The recognized keywords are:
990
@table @samp
991
 
992
@item combreloc
993
Combines multiple reloc sections and sorts them to make dynamic symbol
994
lookup caching possible.
995
 
996
@item defs
997
Disallows undefined symbols in object files.  Undefined symbols in
998
shared libraries are still allowed.
999
 
1000
@item execstack
1001
Marks the object as requiring executable stack.
1002
 
1003
@item initfirst
1004
This option is only meaningful when building a shared object.
1005
It marks the object so that its runtime initialization will occur
1006
before the runtime initialization of any other objects brought into
1007
the process at the same time.  Similarly the runtime finalization of
1008
the object will occur after the runtime finalization of any other
1009
objects.
1010
 
1011
@item interpose
1012
Marks the object that its symbol table interposes before all symbols
1013
but the primary executable.
1014
 
1015
@item lazy
1016
When generating an executable or shared library, mark it to tell the
1017
dynamic linker to defer function call resolution to the point when
1018
the function is called (lazy binding), rather than at load time.
1019
Lazy binding is the default.
1020
 
1021
@item loadfltr
1022
Marks  the object that its filters be processed immediately at
1023
runtime.
1024
 
1025
@item muldefs
1026
Allows multiple definitions.
1027
 
1028
@item nocombreloc
1029
Disables multiple reloc sections combining.
1030
 
1031
@item nocopyreloc
1032
Disables production of copy relocs.
1033
 
1034
@item nodefaultlib
1035
Marks the object that the search for dependencies of this object will
1036
ignore any default library search paths.
1037
 
1038
@item nodelete
1039
Marks the object shouldn't be unloaded at runtime.
1040
 
1041
@item nodlopen
1042
Marks the object not available to @code{dlopen}.
1043
 
1044
@item nodump
1045
Marks the object can not be dumped by @code{dldump}.
1046
 
1047
@item noexecstack
1048
Marks the object as not requiring executable stack.
1049
 
1050
@item norelro
1051
Don't create an ELF @code{PT_GNU_RELRO} segment header in the object.
1052
 
1053
@item now
1054
When generating an executable or shared library, mark it to tell the
1055
dynamic linker to resolve all symbols when the program is started, or
1056
when the shared library is linked to using dlopen, instead of
1057
deferring function call resolution to the point when the function is
1058
first called.
1059
 
1060
@item origin
1061
Marks the object may contain $ORIGIN.
1062
 
1063
@item relro
1064
Create an ELF @code{PT_GNU_RELRO} segment header in the object.
1065
 
1066
@item max-page-size=@var{value}
1067
Set the emulation maximum page size to @var{value}.
1068
 
1069
@item common-page-size=@var{value}
1070
Set the emulation common page size to @var{value}.
1071
 
1072
@end table
1073
 
1074
Other keywords are ignored for Solaris compatibility.
1075
 
1076
@kindex -(
1077
@cindex groups of archives
1078
@item -( @var{archives} -)
1079
@itemx --start-group @var{archives} --end-group
1080
The @var{archives} should be a list of archive files.  They may be
1081
either explicit file names, or @samp{-l} options.
1082
 
1083
The specified archives are searched repeatedly until no new undefined
1084
references are created.  Normally, an archive is searched only once in
1085
the order that it is specified on the command line.  If a symbol in that
1086
archive is needed to resolve an undefined symbol referred to by an
1087
object in an archive that appears later on the command line, the linker
1088
would not be able to resolve that reference.  By grouping the archives,
1089
they all be searched repeatedly until all possible references are
1090
resolved.
1091
 
1092
Using this option has a significant performance cost.  It is best to use
1093
it only when there are unavoidable circular references between two or
1094
more archives.
1095
 
1096
@kindex --accept-unknown-input-arch
1097
@kindex --no-accept-unknown-input-arch
1098
@item --accept-unknown-input-arch
1099
@itemx --no-accept-unknown-input-arch
1100
Tells the linker to accept input files whose architecture cannot be
1101
recognised.  The assumption is that the user knows what they are doing
1102
and deliberately wants to link in these unknown input files.  This was
1103
the default behaviour of the linker, before release 2.14.  The default
1104
behaviour from release 2.14 onwards is to reject such input files, and
1105
so the @samp{--accept-unknown-input-arch} option has been added to
1106
restore the old behaviour.
1107
 
1108
@kindex --as-needed
1109
@kindex --no-as-needed
1110
@item --as-needed
1111
@itemx --no-as-needed
1112
This option affects ELF DT_NEEDED tags for dynamic libraries mentioned
1113
on the command line after the @option{--as-needed} option.  Normally,
1114
the linker will add a DT_NEEDED tag for each dynamic library mentioned
1115
on the command line, regardless of whether the library is actually
1116
needed.  @option{--as-needed} causes a DT_NEEDED tag to only be emitted
1117
for a library that satisfies a symbol reference from regular objects
1118
which is undefined at the point that the library was linked, or, if
1119
the library is not found in the DT_NEEDED lists of other libraries
1120
linked up to that point, a reference from another dynamic library.
1121
@option{--no-as-needed} restores the default behaviour.
1122
 
1123
@kindex --add-needed
1124
@kindex --no-add-needed
1125
@item --add-needed
1126
@itemx --no-add-needed
1127
This option affects the treatment of dynamic libraries from ELF
1128
DT_NEEDED tags in dynamic libraries mentioned on the command line after
1129
the @option{--no-add-needed} option.  Normally, the linker will add
1130
a DT_NEEDED tag for each dynamic library from DT_NEEDED tags.
1131
@option{--no-add-needed} causes DT_NEEDED tags will never be emitted
1132
for those libraries from DT_NEEDED tags. @option{--add-needed} restores
1133
the default behaviour.
1134
 
1135
@kindex -assert @var{keyword}
1136
@item -assert @var{keyword}
1137
This option is ignored for SunOS compatibility.
1138
 
1139
@kindex -Bdynamic
1140
@kindex -dy
1141
@kindex -call_shared
1142
@item -Bdynamic
1143
@itemx -dy
1144
@itemx -call_shared
1145
Link against dynamic libraries.  This is only meaningful on platforms
1146
for which shared libraries are supported.  This option is normally the
1147
default on such platforms.  The different variants of this option are
1148
for compatibility with various systems.  You may use this option
1149
multiple times on the command line: it affects library searching for
1150
@option{-l} options which follow it.
1151
 
1152
@kindex -Bgroup
1153
@item -Bgroup
1154
Set the @code{DF_1_GROUP} flag in the @code{DT_FLAGS_1} entry in the dynamic
1155
section.  This causes the runtime linker to handle lookups in this
1156
object and its dependencies to be performed only inside the group.
1157
@option{--unresolved-symbols=report-all} is implied.  This option is
1158
only meaningful on ELF platforms which support shared libraries.
1159
 
1160
@kindex -Bstatic
1161
@kindex -dn
1162
@kindex -non_shared
1163
@kindex -static
1164
@item -Bstatic
1165
@itemx -dn
1166
@itemx -non_shared
1167
@itemx -static
1168
Do not link against shared libraries.  This is only meaningful on
1169
platforms for which shared libraries are supported.  The different
1170
variants of this option are for compatibility with various systems.  You
1171
may use this option multiple times on the command line: it affects
1172
library searching for @option{-l} options which follow it.  This
1173
option also implies @option{--unresolved-symbols=report-all}.  This
1174
option can be used with @option{-shared}.  Doing so means that a
1175
shared library is being created but that all of the library's external
1176
references must be resolved by pulling in entries from static
1177
libraries.
1178
 
1179
@kindex -Bsymbolic
1180
@item -Bsymbolic
1181
When creating a shared library, bind references to global symbols to the
1182
definition within the shared library, if any.  Normally, it is possible
1183
for a program linked against a shared library to override the definition
1184
within the shared library.  This option is only meaningful on ELF
1185
platforms which support shared libraries.
1186
 
1187
@kindex -Bsymbolic-functions
1188
@item -Bsymbolic-functions
1189
When creating a shared library, bind references to global function
1190
symbols to the definition within the shared library, if any.
1191
This option is only meaningful on ELF platforms which support shared
1192
libraries.
1193
 
1194
@kindex --dynamic-list=@var{dynamic-list-file}
1195
@item --dynamic-list=@var{dynamic-list-file}
1196
Specify the name of a dynamic list file to the linker.  This is
1197
typically used when creating shared libraries to specify a list of
1198
global symbols whose references shouldn't be bound to the definition
1199
within the shared library, or creating dynamically linked executables
1200
to specify a list of symbols which should be added to the symbol table
1201
in the executable.  This option is only meaningful on ELF platforms
1202
which support shared libraries.
1203
 
1204
The format of the dynamic list is the same as the version node without
1205
scope and node name.  See @ref{VERSION} for more information.
1206
 
1207
@kindex --dynamic-list-data
1208
@item --dynamic-list-data
1209
Include all global data symbols to the dynamic list.
1210
 
1211
@kindex --dynamic-list-cpp-new
1212
@item --dynamic-list-cpp-new
1213
Provide the builtin dynamic list for C++ operator new and delete.  It
1214
is mainly useful for building shared libstdc++.
1215
 
1216
@kindex --dynamic-list-cpp-typeinfo
1217
@item --dynamic-list-cpp-typeinfo
1218
Provide the builtin dynamic list for C++ runtime type identification.
1219
 
1220
@kindex --check-sections
1221
@kindex --no-check-sections
1222
@item --check-sections
1223
@itemx --no-check-sections
1224
Asks the linker @emph{not} to check section addresses after they have
1225
been assigned to see if there are any overlaps.  Normally the linker will
1226
perform this check, and if it finds any overlaps it will produce
1227
suitable error messages.  The linker does know about, and does make
1228
allowances for sections in overlays.  The default behaviour can be
1229
restored by using the command line switch @option{--check-sections}.
1230
Section overlap is not usually checked for relocatable links.  You can
1231
force checking in that case by using the @option{--check-sections}
1232
option.
1233
 
1234
@cindex cross reference table
1235
@kindex --cref
1236
@item --cref
1237
Output a cross reference table.  If a linker map file is being
1238
generated, the cross reference table is printed to the map file.
1239
Otherwise, it is printed on the standard output.
1240
 
1241
The format of the table is intentionally simple, so that it may be
1242
easily processed by a script if necessary.  The symbols are printed out,
1243
sorted by name.  For each symbol, a list of file names is given.  If the
1244
symbol is defined, the first file listed is the location of the
1245
definition.  The remaining files contain references to the symbol.
1246
 
1247
@cindex common allocation
1248
@kindex --no-define-common
1249
@item --no-define-common
1250
This option inhibits the assignment of addresses to common symbols.
1251
The script command @code{INHIBIT_COMMON_ALLOCATION} has the same effect.
1252
@xref{Miscellaneous Commands}.
1253
 
1254
The @samp{--no-define-common} option allows decoupling
1255
the decision to assign addresses to Common symbols from the choice
1256
of the output file type; otherwise a non-Relocatable output type
1257
forces assigning addresses to Common symbols.
1258
Using @samp{--no-define-common} allows Common symbols that are referenced
1259
from a shared library to be assigned addresses only in the main program.
1260
This eliminates the unused duplicate space in the shared library,
1261
and also prevents any possible confusion over resolving to the wrong
1262
duplicate when there are many dynamic modules with specialized search
1263
paths for runtime symbol resolution.
1264
 
1265
@cindex symbols, from command line
1266
@kindex --defsym=@var{symbol}=@var{exp}
1267
@item --defsym=@var{symbol}=@var{expression}
1268
Create a global symbol in the output file, containing the absolute
1269
address given by @var{expression}.  You may use this option as many
1270
times as necessary to define multiple symbols in the command line.  A
1271
limited form of arithmetic is supported for the @var{expression} in this
1272
context: you may give a hexadecimal constant or the name of an existing
1273
symbol, or use @code{+} and @code{-} to add or subtract hexadecimal
1274
constants or symbols.  If you need more elaborate expressions, consider
1275
using the linker command language from a script (@pxref{Assignments,,
1276
Assignment: Symbol Definitions}).  @emph{Note:} there should be no white
1277
space between @var{symbol}, the equals sign (``@key{=}''), and
1278
@var{expression}.
1279
 
1280
@cindex demangling, from command line
1281
@kindex --demangle[=@var{style}]
1282
@kindex --no-demangle
1283
@item --demangle[=@var{style}]
1284
@itemx --no-demangle
1285
These options control whether to demangle symbol names in error messages
1286
and other output.  When the linker is told to demangle, it tries to
1287
present symbol names in a readable fashion: it strips leading
1288
underscores if they are used by the object file format, and converts C++
1289
mangled symbol names into user readable names.  Different compilers have
1290
different mangling styles.  The optional demangling style argument can be used
1291
to choose an appropriate demangling style for your compiler.  The linker will
1292
demangle by default unless the environment variable @samp{COLLECT_NO_DEMANGLE}
1293
is set.  These options may be used to override the default.
1294
 
1295
@cindex dynamic linker, from command line
1296
@kindex -I@var{file}
1297
@kindex --dynamic-linker=@var{file}
1298
@item -I@var{file}
1299
@itemx --dynamic-linker=@var{file}
1300
Set the name of the dynamic linker.  This is only meaningful when
1301
generating dynamically linked ELF executables.  The default dynamic
1302
linker is normally correct; don't use this unless you know what you are
1303
doing.
1304
 
1305
@kindex --fatal-warnings
1306
@kindex --no-fatal-warnings
1307
@item --fatal-warnings
1308
@itemx --no-fatal-warnings
1309
Treat all warnings as errors.  The default behaviour can be restored
1310
with the option @option{--no-fatal-warnings}.
1311
 
1312
@kindex --force-exe-suffix
1313
@item  --force-exe-suffix
1314
Make sure that an output file has a .exe suffix.
1315
 
1316
If a successfully built fully linked output file does not have a
1317
@code{.exe} or @code{.dll} suffix, this option forces the linker to copy
1318
the output file to one of the same name with a @code{.exe} suffix. This
1319
option is useful when using unmodified Unix makefiles on a Microsoft
1320
Windows host, since some versions of Windows won't run an image unless
1321
it ends in a @code{.exe} suffix.
1322
 
1323
@kindex --gc-sections
1324
@kindex --no-gc-sections
1325
@cindex garbage collection
1326
@item --gc-sections
1327
@itemx --no-gc-sections
1328
Enable garbage collection of unused input sections.  It is ignored on
1329
targets that do not support this option.  The default behaviour (of not
1330
performing this garbage collection) can be restored by specifying
1331
@samp{--no-gc-sections} on the command line.
1332
 
1333
@samp{--gc-sections} decides which input sections are used by
1334
examining symbols and relocations.  The section containing the entry
1335
symbol and all sections containing symbols undefined on the
1336
command-line will be kept, as will sections containing symbols
1337
referenced by dynamic objects.  Note that when building shared
1338
libraries, the linker must assume that any visible symbol is
1339
referenced.  Once this initial set of sections has been determined,
1340
the linker recursively marks as used any section referenced by their
1341
relocations.  See @samp{--entry} and @samp{--undefined}.
1342
 
1343
This option can be set when doing a partial link (enabled with option
1344
@samp{-r}).  In this case the root of symbols kept must be explicitely
1345
specified either by an @samp{--entry} or @samp{--undefined} option or by
1346
a @code{ENTRY} command in the linker script.
1347
 
1348
@kindex --print-gc-sections
1349
@kindex --no-print-gc-sections
1350
@cindex garbage collection
1351
@item --print-gc-sections
1352
@itemx --no-print-gc-sections
1353
List all sections removed by garbage collection.  The listing is
1354
printed on stderr.  This option is only effective if garbage
1355
collection has been enabled via the @samp{--gc-sections}) option.  The
1356
default behaviour (of not listing the sections that are removed) can
1357
be restored by specifying @samp{--no-print-gc-sections} on the command
1358
line.
1359
 
1360
@cindex help
1361
@cindex usage
1362
@kindex --help
1363
@item --help
1364
Print a summary of the command-line options on the standard output and exit.
1365
 
1366
@kindex --target-help
1367
@item --target-help
1368
Print a summary of all target specific options on the standard output and exit.
1369
 
1370
@kindex -Map=@var{mapfile}
1371
@item -Map=@var{mapfile}
1372
Print a link map to the file @var{mapfile}.  See the description of the
1373
@option{-M} option, above.
1374
 
1375
@cindex memory usage
1376
@kindex --no-keep-memory
1377
@item --no-keep-memory
1378
@command{ld} normally optimizes for speed over memory usage by caching the
1379
symbol tables of input files in memory.  This option tells @command{ld} to
1380
instead optimize for memory usage, by rereading the symbol tables as
1381
necessary.  This may be required if @command{ld} runs out of memory space
1382
while linking a large executable.
1383
 
1384
@kindex --no-undefined
1385
@kindex -z defs
1386
@item --no-undefined
1387
@itemx -z defs
1388
Report unresolved symbol references from regular object files.  This
1389
is done even if the linker is creating a non-symbolic shared library.
1390
The switch @option{--[no-]allow-shlib-undefined} controls the
1391
behaviour for reporting unresolved references found in shared
1392
libraries being linked in.
1393
 
1394
@kindex --allow-multiple-definition
1395
@kindex -z muldefs
1396
@item --allow-multiple-definition
1397
@itemx -z muldefs
1398
Normally when a symbol is defined multiple times, the linker will
1399
report a fatal error. These options allow multiple definitions and the
1400
first definition will be used.
1401
 
1402
@kindex --allow-shlib-undefined
1403
@kindex --no-allow-shlib-undefined
1404
@item --allow-shlib-undefined
1405
@itemx --no-allow-shlib-undefined
1406
Allows or disallows undefined symbols in shared libraries.
1407
This switch is similar to @option{--no-undefined} except that it
1408
determines the behaviour when the undefined symbols are in a
1409
shared library rather than a regular object file.  It does not affect
1410
how undefined symbols in regular object files are handled.
1411
 
1412
The default behaviour is to report errors for any undefined symbols
1413
referenced in shared libraries if the linker is being used to create
1414
an executable, but to allow them if the linker is being used to create
1415
a shared library.
1416
 
1417
The reasons for allowing undefined symbol references in shared
1418
libraries specified at link time are that:
1419
 
1420
@itemize @bullet
1421
@item
1422
A shared library specified at link time may not be the same as the one
1423
that is available at load time, so the symbol might actually be
1424
resolvable at load time.
1425
@item
1426
There are some operating systems, eg BeOS and HPPA, where undefined
1427
symbols in shared libraries are normal.
1428
 
1429
The BeOS kernel for example patches shared libraries at load time to
1430
select whichever function is most appropriate for the current
1431
architecture.  This is used, for example, to dynamically select an
1432
appropriate memset function.
1433
@end itemize
1434
 
1435
@kindex --no-undefined-version
1436
@item --no-undefined-version
1437
Normally when a symbol has an undefined version, the linker will ignore
1438
it. This option disallows symbols with undefined version and a fatal error
1439
will be issued instead.
1440
 
1441
@kindex --default-symver
1442
@item --default-symver
1443
Create and use a default symbol version (the soname) for unversioned
1444
exported symbols.
1445
 
1446
@kindex --default-imported-symver
1447
@item --default-imported-symver
1448
Create and use a default symbol version (the soname) for unversioned
1449
imported symbols.
1450
 
1451
@kindex --no-warn-mismatch
1452
@item --no-warn-mismatch
1453
Normally @command{ld} will give an error if you try to link together input
1454
files that are mismatched for some reason, perhaps because they have
1455
been compiled for different processors or for different endiannesses.
1456
This option tells @command{ld} that it should silently permit such possible
1457
errors.  This option should only be used with care, in cases when you
1458
have taken some special action that ensures that the linker errors are
1459
inappropriate.
1460
 
1461
@kindex --no-warn-search-mismatch
1462
@item --no-warn-search-mismatch
1463
Normally @command{ld} will give a warning if it finds an incompatible
1464
library during a library search.  This option silences the warning.
1465
 
1466
@kindex --no-whole-archive
1467
@item --no-whole-archive
1468
Turn off the effect of the @option{--whole-archive} option for subsequent
1469
archive files.
1470
 
1471
@cindex output file after errors
1472
@kindex --noinhibit-exec
1473
@item --noinhibit-exec
1474
Retain the executable output file whenever it is still usable.
1475
Normally, the linker will not produce an output file if it encounters
1476
errors during the link process; it exits without writing an output file
1477
when it issues any error whatsoever.
1478
 
1479
@kindex -nostdlib
1480
@item -nostdlib
1481
Only search library directories explicitly specified on the
1482
command line.  Library directories specified in linker scripts
1483
(including linker scripts specified on the command line) are ignored.
1484
 
1485
@ifclear SingleFormat
1486
@kindex --oformat=@var{output-format}
1487
@item --oformat=@var{output-format}
1488
@command{ld} may be configured to support more than one kind of object
1489
file.  If your @command{ld} is configured this way, you can use the
1490
@samp{--oformat} option to specify the binary format for the output
1491
object file.  Even when @command{ld} is configured to support alternative
1492
object formats, you don't usually need to specify this, as @command{ld}
1493
should be configured to produce as a default output format the most
1494
usual format on each machine.  @var{output-format} is a text string, the
1495
name of a particular format supported by the BFD libraries.  (You can
1496
list the available binary formats with @samp{objdump -i}.)  The script
1497
command @code{OUTPUT_FORMAT} can also specify the output format, but
1498
this option overrides it.  @xref{BFD}.
1499
@end ifclear
1500
 
1501
@kindex -pie
1502
@kindex --pic-executable
1503
@item -pie
1504
@itemx --pic-executable
1505
@cindex position independent executables
1506
Create a position independent executable.  This is currently only supported on
1507
ELF platforms.  Position independent executables are similar to shared
1508
libraries in that they are relocated by the dynamic linker to the virtual
1509
address the OS chooses for them (which can vary between invocations).  Like
1510
normal dynamically linked executables they can be executed and symbols
1511
defined in the executable cannot be overridden by shared libraries.
1512
 
1513
@kindex -qmagic
1514
@item -qmagic
1515
This option is ignored for Linux compatibility.
1516
 
1517
@kindex -Qy
1518
@item -Qy
1519
This option is ignored for SVR4 compatibility.
1520
 
1521
@kindex --relax
1522
@cindex synthesizing linker
1523
@cindex relaxing addressing modes
1524
@item --relax
1525
An option with machine dependent effects.
1526
@ifset GENERIC
1527
This option is only supported on a few targets.
1528
@end ifset
1529
@ifset H8300
1530
@xref{H8/300,,@command{ld} and the H8/300}.
1531
@end ifset
1532
@ifset I960
1533
@xref{i960,, @command{ld} and the Intel 960 family}.
1534
@end ifset
1535
@ifset XTENSA
1536
@xref{Xtensa,, @command{ld} and Xtensa Processors}.
1537
@end ifset
1538
@ifset M68HC11
1539
@xref{M68HC11/68HC12,,@command{ld} and the 68HC11 and 68HC12}.
1540
@end ifset
1541
@ifset POWERPC
1542
@xref{PowerPC ELF32,,@command{ld} and PowerPC 32-bit ELF Support}.
1543
@end ifset
1544
 
1545
On some platforms, the @samp{--relax} option performs global
1546
optimizations that become possible when the linker resolves addressing
1547
in the program, such as relaxing address modes and synthesizing new
1548
instructions in the output object file.
1549
 
1550
On some platforms these link time global optimizations may make symbolic
1551
debugging of the resulting executable impossible.
1552
@ifset GENERIC
1553
This is known to be
1554
the case for the Matsushita MN10200 and MN10300 family of processors.
1555
@end ifset
1556
 
1557
@ifset GENERIC
1558
On platforms where this is not supported, @samp{--relax} is accepted,
1559
but ignored.
1560
@end ifset
1561
 
1562
@cindex retaining specified symbols
1563
@cindex stripping all but some symbols
1564
@cindex symbols, retaining selectively
1565
@kindex --retain-symbols-file=@var{filename}
1566
@item --retain-symbols-file=@var{filename}
1567
Retain @emph{only} the symbols listed in the file @var{filename},
1568
discarding all others.  @var{filename} is simply a flat file, with one
1569
symbol name per line.  This option is especially useful in environments
1570
@ifset GENERIC
1571
(such as VxWorks)
1572
@end ifset
1573
where a large global symbol table is accumulated gradually, to conserve
1574
run-time memory.
1575
 
1576
@samp{--retain-symbols-file} does @emph{not} discard undefined symbols,
1577
or symbols needed for relocations.
1578
 
1579
You may only specify @samp{--retain-symbols-file} once in the command
1580
line.  It overrides @samp{-s} and @samp{-S}.
1581
 
1582
@ifset GENERIC
1583
@item -rpath=@var{dir}
1584
@cindex runtime library search path
1585
@kindex -rpath=@var{dir}
1586
Add a directory to the runtime library search path.  This is used when
1587
linking an ELF executable with shared objects.  All @option{-rpath}
1588
arguments are concatenated and passed to the runtime linker, which uses
1589
them to locate shared objects at runtime.  The @option{-rpath} option is
1590
also used when locating shared objects which are needed by shared
1591
objects explicitly included in the link; see the description of the
1592
@option{-rpath-link} option.  If @option{-rpath} is not used when linking an
1593
ELF executable, the contents of the environment variable
1594
@code{LD_RUN_PATH} will be used if it is defined.
1595
 
1596
The @option{-rpath} option may also be used on SunOS.  By default, on
1597
SunOS, the linker will form a runtime search patch out of all the
1598
@option{-L} options it is given.  If a @option{-rpath} option is used, the
1599
runtime search path will be formed exclusively using the @option{-rpath}
1600
options, ignoring the @option{-L} options.  This can be useful when using
1601
gcc, which adds many @option{-L} options which may be on NFS mounted
1602
file systems.
1603
 
1604
For compatibility with other ELF linkers, if the @option{-R} option is
1605
followed by a directory name, rather than a file name, it is treated as
1606
the @option{-rpath} option.
1607
@end ifset
1608
 
1609
@ifset GENERIC
1610
@cindex link-time runtime library search path
1611
@kindex -rpath-link=@var{dir}
1612
@item -rpath-link=@var{dir}
1613
When using ELF or SunOS, one shared library may require another.  This
1614
happens when an @code{ld -shared} link includes a shared library as one
1615
of the input files.
1616
 
1617
When the linker encounters such a dependency when doing a non-shared,
1618
non-relocatable link, it will automatically try to locate the required
1619
shared library and include it in the link, if it is not included
1620
explicitly.  In such a case, the @option{-rpath-link} option
1621
specifies the first set of directories to search.  The
1622
@option{-rpath-link} option may specify a sequence of directory names
1623
either by specifying a list of names separated by colons, or by
1624
appearing multiple times.
1625
 
1626
This option should be used with caution as it overrides the search path
1627
that may have been hard compiled into a shared library. In such a case it
1628
is possible to use unintentionally a different search path than the
1629
runtime linker would do.
1630
 
1631
The linker uses the following search paths to locate required shared
1632
libraries:
1633
@enumerate
1634
@item
1635
Any directories specified by @option{-rpath-link} options.
1636
@item
1637
Any directories specified by @option{-rpath} options.  The difference
1638
between @option{-rpath} and @option{-rpath-link} is that directories
1639
specified by @option{-rpath} options are included in the executable and
1640
used at runtime, whereas the @option{-rpath-link} option is only effective
1641
at link time. Searching @option{-rpath} in this way is only supported
1642
by native linkers and cross linkers which have been configured with
1643
the @option{--with-sysroot} option.
1644
@item
1645
On an ELF system, for native linkers, if the @option{-rpath} and
1646
@option{-rpath-link} options were not used, search the contents of the
1647
environment variable @code{LD_RUN_PATH}.
1648
@item
1649
On SunOS, if the @option{-rpath} option was not used, search any
1650
directories specified using @option{-L} options.
1651
@item
1652
For a native linker, the search the contents of the environment
1653
variable @code{LD_LIBRARY_PATH}.
1654
@item
1655
For a native ELF linker, the directories in @code{DT_RUNPATH} or
1656
@code{DT_RPATH} of a shared library are searched for shared
1657
libraries needed by it. The @code{DT_RPATH} entries are ignored if
1658
@code{DT_RUNPATH} entries exist.
1659
@item
1660
The default directories, normally @file{/lib} and @file{/usr/lib}.
1661
@item
1662
For a native linker on an ELF system, if the file @file{/etc/ld.so.conf}
1663
exists, the list of directories found in that file.
1664
@end enumerate
1665
 
1666
If the required shared library is not found, the linker will issue a
1667
warning and continue with the link.
1668
@end ifset
1669
 
1670
@kindex -shared
1671
@kindex -Bshareable
1672
@item -shared
1673
@itemx -Bshareable
1674
@cindex shared libraries
1675
Create a shared library.  This is currently only supported on ELF, XCOFF
1676
and SunOS platforms.  On SunOS, the linker will automatically create a
1677
shared library if the @option{-e} option is not used and there are
1678
undefined symbols in the link.
1679
 
1680
@kindex --sort-common
1681
@item --sort-common
1682
@itemx --sort-common=ascending
1683
@itemx --sort-common=descending
1684
This option tells @command{ld} to sort the common symbols by alignment in
1685
ascending or descending order when it places them in the appropriate output
1686
sections.  The symbol alignments considered are sixteen-byte or larger,
1687
eight-byte, four-byte, two-byte, and one-byte. This is to prevent gaps
1688
between symbols due to alignment constraints.  If no sorting order is
1689
specified, then descending order is assumed.
1690
 
1691
@kindex --sort-section=name
1692
@item --sort-section=name
1693
This option will apply @code{SORT_BY_NAME} to all wildcard section
1694
patterns in the linker script.
1695
 
1696
@kindex --sort-section=alignment
1697
@item --sort-section=alignment
1698
This option will apply @code{SORT_BY_ALIGNMENT} to all wildcard section
1699
patterns in the linker script.
1700
 
1701
@kindex --split-by-file
1702
@item --split-by-file[=@var{size}]
1703
Similar to @option{--split-by-reloc} but creates a new output section for
1704
each input file when @var{size} is reached.  @var{size} defaults to a
1705
size of 1 if not given.
1706
 
1707
@kindex --split-by-reloc
1708
@item --split-by-reloc[=@var{count}]
1709
Tries to creates extra sections in the output file so that no single
1710
output section in the file contains more than @var{count} relocations.
1711
This is useful when generating huge relocatable files for downloading into
1712
certain real time kernels with the COFF object file format; since COFF
1713
cannot represent more than 65535 relocations in a single section.  Note
1714
that this will fail to work with object file formats which do not
1715
support arbitrary sections.  The linker will not split up individual
1716
input sections for redistribution, so if a single input section contains
1717
more than @var{count} relocations one output section will contain that
1718
many relocations.  @var{count} defaults to a value of 32768.
1719
 
1720
@kindex --stats
1721
@item --stats
1722
Compute and display statistics about the operation of the linker, such
1723
as execution time and memory usage.
1724
 
1725
@kindex --sysroot=@var{directory}
1726
@item --sysroot=@var{directory}
1727
Use @var{directory} as the location of the sysroot, overriding the
1728
configure-time default.  This option is only supported by linkers
1729
that were configured using @option{--with-sysroot}.
1730
 
1731
@kindex --traditional-format
1732
@cindex traditional format
1733
@item --traditional-format
1734
For some targets, the output of @command{ld} is different in some ways from
1735
the output of some existing linker.  This switch requests @command{ld} to
1736
use the traditional format instead.
1737
 
1738
@cindex dbx
1739
For example, on SunOS, @command{ld} combines duplicate entries in the
1740
symbol string table.  This can reduce the size of an output file with
1741
full debugging information by over 30 percent.  Unfortunately, the SunOS
1742
@code{dbx} program can not read the resulting program (@code{gdb} has no
1743
trouble).  The @samp{--traditional-format} switch tells @command{ld} to not
1744
combine duplicate entries.
1745
 
1746
@kindex --section-start=@var{sectionname}=@var{org}
1747
@item --section-start=@var{sectionname}=@var{org}
1748
Locate a section in the output file at the absolute
1749
address given by @var{org}.  You may use this option as many
1750
times as necessary to locate multiple sections in the command
1751
line.
1752
@var{org} must be a single hexadecimal integer;
1753
for compatibility with other linkers, you may omit the leading
1754
@samp{0x} usually associated with hexadecimal values.  @emph{Note:} there
1755
should be no white space between @var{sectionname}, the equals
1756
sign (``@key{=}''), and @var{org}.
1757
 
1758
@kindex -Tbss=@var{org}
1759
@kindex -Tdata=@var{org}
1760
@kindex -Ttext=@var{org}
1761
@cindex segment origins, cmd line
1762
@item -Tbss=@var{org}
1763
@itemx -Tdata=@var{org}
1764
@itemx -Ttext=@var{org}
1765
Same as @option{--section-start}, with @code{.bss}, @code{.data} or
1766
@code{.text} as the @var{sectionname}.
1767
 
1768
@kindex -Ttext-segment=@var{org}
1769
@item -Ttext-segment=@var{org}
1770
@cindex text segment origin, cmd line
1771
When creating an ELF executable or shared object, it will set the address
1772
of the first byte of the text segment.
1773
 
1774
@kindex --unresolved-symbols
1775
@item --unresolved-symbols=@var{method}
1776
Determine how to handle unresolved symbols.  There are four possible
1777
values for @samp{method}:
1778
 
1779
@table @samp
1780
@item ignore-all
1781
Do not report any unresolved symbols.
1782
 
1783
@item report-all
1784
Report all unresolved symbols.  This is the default.
1785
 
1786
@item ignore-in-object-files
1787
Report unresolved symbols that are contained in shared libraries, but
1788
ignore them if they come from regular object files.
1789
 
1790
@item ignore-in-shared-libs
1791
Report unresolved symbols that come from regular object files, but
1792
ignore them if they come from shared libraries.  This can be useful
1793
when creating a dynamic binary and it is known that all the shared
1794
libraries that it should be referencing are included on the linker's
1795
command line.
1796
@end table
1797
 
1798
The behaviour for shared libraries on their own can also be controlled
1799
by the @option{--[no-]allow-shlib-undefined} option.
1800
 
1801
Normally the linker will generate an error message for each reported
1802
unresolved symbol but the option @option{--warn-unresolved-symbols}
1803
can change this to a warning.
1804
 
1805
@kindex --verbose
1806
@cindex verbose
1807
@item --dll-verbose
1808
@itemx --verbose
1809
Display the version number for @command{ld} and list the linker emulations
1810
supported.  Display which input files can and cannot be opened.  Display
1811
the linker script being used by the linker.
1812
 
1813
@kindex --version-script=@var{version-scriptfile}
1814
@cindex version script, symbol versions
1815
@item --version-script=@var{version-scriptfile}
1816
Specify the name of a version script to the linker.  This is typically
1817
used when creating shared libraries to specify additional information
1818
about the version hierarchy for the library being created.  This option
1819
is only fully supported on ELF platforms which support shared libraries;
1820
see @ref{VERSION}.  It is partially supported on PE platforms, which can
1821
use version scripts to filter symbol visibility in auto-export mode: any
1822
symbols marked @samp{local} in the version script will not be exported.
1823
@xref{WIN32}.
1824
 
1825
@kindex --warn-common
1826
@cindex warnings, on combining symbols
1827
@cindex combining symbols, warnings on
1828
@item --warn-common
1829
Warn when a common symbol is combined with another common symbol or with
1830
a symbol definition.  Unix linkers allow this somewhat sloppy practise,
1831
but linkers on some other operating systems do not.  This option allows
1832
you to find potential problems from combining global symbols.
1833
Unfortunately, some C libraries use this practise, so you may get some
1834
warnings about symbols in the libraries as well as in your programs.
1835
 
1836
There are three kinds of global symbols, illustrated here by C examples:
1837
 
1838
@table @samp
1839
@item int i = 1;
1840
A definition, which goes in the initialized data section of the output
1841
file.
1842
 
1843
@item extern int i;
1844
An undefined reference, which does not allocate space.
1845
There must be either a definition or a common symbol for the
1846
variable somewhere.
1847
 
1848
@item int i;
1849
A common symbol.  If there are only (one or more) common symbols for a
1850
variable, it goes in the uninitialized data area of the output file.
1851
The linker merges multiple common symbols for the same variable into a
1852
single symbol.  If they are of different sizes, it picks the largest
1853
size.  The linker turns a common symbol into a declaration, if there is
1854
a definition of the same variable.
1855
@end table
1856
 
1857
The @samp{--warn-common} option can produce five kinds of warnings.
1858
Each warning consists of a pair of lines: the first describes the symbol
1859
just encountered, and the second describes the previous symbol
1860
encountered with the same name.  One or both of the two symbols will be
1861
a common symbol.
1862
 
1863
@enumerate
1864
@item
1865
Turning a common symbol into a reference, because there is already a
1866
definition for the symbol.
1867
@smallexample
1868
@var{file}(@var{section}): warning: common of `@var{symbol}'
1869
   overridden by definition
1870
@var{file}(@var{section}): warning: defined here
1871
@end smallexample
1872
 
1873
@item
1874
Turning a common symbol into a reference, because a later definition for
1875
the symbol is encountered.  This is the same as the previous case,
1876
except that the symbols are encountered in a different order.
1877
@smallexample
1878
@var{file}(@var{section}): warning: definition of `@var{symbol}'
1879
   overriding common
1880
@var{file}(@var{section}): warning: common is here
1881
@end smallexample
1882
 
1883
@item
1884
Merging a common symbol with a previous same-sized common symbol.
1885
@smallexample
1886
@var{file}(@var{section}): warning: multiple common
1887
   of `@var{symbol}'
1888
@var{file}(@var{section}): warning: previous common is here
1889
@end smallexample
1890
 
1891
@item
1892
Merging a common symbol with a previous larger common symbol.
1893
@smallexample
1894
@var{file}(@var{section}): warning: common of `@var{symbol}'
1895
   overridden by larger common
1896
@var{file}(@var{section}): warning: larger common is here
1897
@end smallexample
1898
 
1899
@item
1900
Merging a common symbol with a previous smaller common symbol.  This is
1901
the same as the previous case, except that the symbols are
1902
encountered in a different order.
1903
@smallexample
1904
@var{file}(@var{section}): warning: common of `@var{symbol}'
1905
   overriding smaller common
1906
@var{file}(@var{section}): warning: smaller common is here
1907
@end smallexample
1908
@end enumerate
1909
 
1910
@kindex --warn-constructors
1911
@item --warn-constructors
1912
Warn if any global constructors are used.  This is only useful for a few
1913
object file formats.  For formats like COFF or ELF, the linker can not
1914
detect the use of global constructors.
1915
 
1916
@kindex --warn-multiple-gp
1917
@item --warn-multiple-gp
1918
Warn if multiple global pointer values are required in the output file.
1919
This is only meaningful for certain processors, such as the Alpha.
1920
Specifically, some processors put large-valued constants in a special
1921
section.  A special register (the global pointer) points into the middle
1922
of this section, so that constants can be loaded efficiently via a
1923
base-register relative addressing mode.  Since the offset in
1924
base-register relative mode is fixed and relatively small (e.g., 16
1925
bits), this limits the maximum size of the constant pool.  Thus, in
1926
large programs, it is often necessary to use multiple global pointer
1927
values in order to be able to address all possible constants.  This
1928
option causes a warning to be issued whenever this case occurs.
1929
 
1930
@kindex --warn-once
1931
@cindex warnings, on undefined symbols
1932
@cindex undefined symbols, warnings on
1933
@item --warn-once
1934
Only warn once for each undefined symbol, rather than once per module
1935
which refers to it.
1936
 
1937
@kindex --warn-section-align
1938
@cindex warnings, on section alignment
1939
@cindex section alignment, warnings on
1940
@item --warn-section-align
1941
Warn if the address of an output section is changed because of
1942
alignment.  Typically, the alignment will be set by an input section.
1943
The address will only be changed if it not explicitly specified; that
1944
is, if the @code{SECTIONS} command does not specify a start address for
1945
the section (@pxref{SECTIONS}).
1946
 
1947
@kindex --warn-shared-textrel
1948
@item --warn-shared-textrel
1949
Warn if the linker adds a DT_TEXTREL to a shared object.
1950
 
1951
@kindex --warn-alternate-em
1952
@item --warn-alternate-em
1953
Warn if an object has alternate ELF machine code.
1954
 
1955
@kindex --warn-unresolved-symbols
1956
@item --warn-unresolved-symbols
1957
If the linker is going to report an unresolved symbol (see the option
1958
@option{--unresolved-symbols}) it will normally generate an error.
1959
This option makes it generate a warning instead.
1960
 
1961
@kindex --error-unresolved-symbols
1962
@item --error-unresolved-symbols
1963
This restores the linker's default behaviour of generating errors when
1964
it is reporting unresolved symbols.
1965
 
1966
@kindex --whole-archive
1967
@cindex including an entire archive
1968
@item --whole-archive
1969
For each archive mentioned on the command line after the
1970
@option{--whole-archive} option, include every object file in the archive
1971
in the link, rather than searching the archive for the required object
1972
files.  This is normally used to turn an archive file into a shared
1973
library, forcing every object to be included in the resulting shared
1974
library.  This option may be used more than once.
1975
 
1976
Two notes when using this option from gcc: First, gcc doesn't know
1977
about this option, so you have to use @option{-Wl,-whole-archive}.
1978
Second, don't forget to use @option{-Wl,-no-whole-archive} after your
1979
list of archives, because gcc will add its own list of archives to
1980
your link and you may not want this flag to affect those as well.
1981
 
1982
@kindex --wrap=@var{symbol}
1983
@item --wrap=@var{symbol}
1984
Use a wrapper function for @var{symbol}.  Any undefined reference to
1985
@var{symbol} will be resolved to @code{__wrap_@var{symbol}}.  Any
1986
undefined reference to @code{__real_@var{symbol}} will be resolved to
1987
@var{symbol}.
1988
 
1989
This can be used to provide a wrapper for a system function.  The
1990
wrapper function should be called @code{__wrap_@var{symbol}}.  If it
1991
wishes to call the system function, it should call
1992
@code{__real_@var{symbol}}.
1993
 
1994
Here is a trivial example:
1995
 
1996
@smallexample
1997
void *
1998
__wrap_malloc (size_t c)
1999
@{
2000
  printf ("malloc called with %zu\n", c);
2001
  return __real_malloc (c);
2002
@}
2003
@end smallexample
2004
 
2005
If you link other code with this file using @option{--wrap malloc}, then
2006
all calls to @code{malloc} will call the function @code{__wrap_malloc}
2007
instead.  The call to @code{__real_malloc} in @code{__wrap_malloc} will
2008
call the real @code{malloc} function.
2009
 
2010
You may wish to provide a @code{__real_malloc} function as well, so that
2011
links without the @option{--wrap} option will succeed.  If you do this,
2012
you should not put the definition of @code{__real_malloc} in the same
2013
file as @code{__wrap_malloc}; if you do, the assembler may resolve the
2014
call before the linker has a chance to wrap it to @code{malloc}.
2015
 
2016
@kindex --eh-frame-hdr
2017
@item --eh-frame-hdr
2018
Request creation of @code{.eh_frame_hdr} section and ELF
2019
@code{PT_GNU_EH_FRAME} segment header.
2020
 
2021
@kindex --enable-new-dtags
2022
@kindex --disable-new-dtags
2023
@item --enable-new-dtags
2024
@itemx --disable-new-dtags
2025
This linker can create the new dynamic tags in ELF. But the older ELF
2026
systems may not understand them. If you specify
2027
@option{--enable-new-dtags}, the dynamic tags will be created as needed.
2028
If you specify @option{--disable-new-dtags}, no new dynamic tags will be
2029
created. By default, the new dynamic tags are not created. Note that
2030
those options are only available for ELF systems.
2031
 
2032
@kindex --hash-size=@var{number}
2033
@item --hash-size=@var{number}
2034
Set the default size of the linker's hash tables to a prime number
2035
close to @var{number}.  Increasing this value can reduce the length of
2036
time it takes the linker to perform its tasks, at the expense of
2037
increasing the linker's memory requirements.  Similarly reducing this
2038
value can reduce the memory requirements at the expense of speed.
2039
 
2040
@kindex --hash-style=@var{style}
2041
@item --hash-style=@var{style}
2042
Set the type of linker's hash table(s).  @var{style} can be either
2043
@code{sysv} for classic ELF @code{.hash} section, @code{gnu} for
2044
new style GNU @code{.gnu.hash} section or @code{both} for both
2045
the classic ELF @code{.hash} and new style GNU @code{.gnu.hash}
2046
hash tables.  The default is @code{sysv}.
2047
 
2048
@kindex --reduce-memory-overheads
2049
@item --reduce-memory-overheads
2050
This option reduces memory requirements at ld runtime, at the expense of
2051
linking speed.  This was introduced to select the old O(n^2) algorithm
2052
for link map file generation, rather than the new O(n) algorithm which uses
2053
about 40% more memory for symbol storage.
2054
 
2055
Another effect of the switch is to set the default hash table size to
2056
1021, which again saves memory at the cost of lengthening the linker's
2057
run time.  This is not done however if the @option{--hash-size} switch
2058
has been used.
2059
 
2060
The @option{--reduce-memory-overheads} switch may be also be used to
2061
enable other tradeoffs in future versions of the linker.
2062
 
2063
@kindex --build-id
2064
@kindex --build-id=@var{style}
2065
@item --build-id
2066
@itemx --build-id=@var{style}
2067
Request creation of @code{.note.gnu.build-id} ELF note section.
2068
The contents of the note are unique bits identifying this linked
2069
file.  @var{style} can be @code{uuid} to use 128 random bits,
2070
@code{sha1} to use a 160-bit @sc{SHA1} hash on the normative
2071
parts of the output contents, @code{md5} to use a 128-bit
2072
@sc{MD5} hash on the normative parts of the output contents, or
2073
@code{0x@var{hexstring}} to use a chosen bit string specified as
2074
an even number of hexadecimal digits (@code{-} and @code{:}
2075
characters between digit pairs are ignored).  If @var{style} is
2076
omitted, @code{sha1} is used.
2077
 
2078
The @code{md5} and @code{sha1} styles produces an identifier
2079
that is always the same in an identical output file, but will be
2080
unique among all nonidentical output files.  It is not intended
2081
to be compared as a checksum for the file's contents.  A linked
2082
file may be changed later by other tools, but the build ID bit
2083
string identifying the original linked file does not change.
2084
 
2085
Passing @code{none} for @var{style} disables the setting from any
2086
@code{--build-id} options earlier on the command line.
2087
@end table
2088
 
2089
@c man end
2090
 
2091
@subsection Options Specific to i386 PE Targets
2092
 
2093
@c man begin OPTIONS
2094
 
2095
The i386 PE linker supports the @option{-shared} option, which causes
2096
the output to be a dynamically linked library (DLL) instead of a
2097
normal executable.  You should name the output @code{*.dll} when you
2098
use this option.  In addition, the linker fully supports the standard
2099
@code{*.def} files, which may be specified on the linker command line
2100
like an object file (in fact, it should precede archives it exports
2101
symbols from, to ensure that they get linked in, just like a normal
2102
object file).
2103
 
2104
In addition to the options common to all targets, the i386 PE linker
2105
support additional command line options that are specific to the i386
2106
PE target.  Options that take values may be separated from their
2107
values by either a space or an equals sign.
2108
 
2109
@table @gcctabopt
2110
 
2111
@kindex --add-stdcall-alias
2112
@item --add-stdcall-alias
2113
If given, symbols with a stdcall suffix (@@@var{nn}) will be exported
2114
as-is and also with the suffix stripped.
2115
[This option is specific to the i386 PE targeted port of the linker]
2116
 
2117
@kindex --base-file
2118
@item --base-file @var{file}
2119
Use @var{file} as the name of a file in which to save the base
2120
addresses of all the relocations needed for generating DLLs with
2121
@file{dlltool}.
2122
[This is an i386 PE specific option]
2123
 
2124
@kindex --dll
2125
@item --dll
2126
Create a DLL instead of a regular executable.  You may also use
2127
@option{-shared} or specify a @code{LIBRARY} in a given @code{.def}
2128
file.
2129
[This option is specific to the i386 PE targeted port of the linker]
2130
 
2131
@kindex --enable-long-section-names
2132
@kindex --disable-long-section-names
2133
@item --enable-long-section-names
2134
@itemx --disable-long-section-names
2135
The PE variants of the Coff object format add an extension that permits
2136
the use of section names longer than eight characters, the normal limit
2137
for Coff.  By default, these names are only allowed in object files, as
2138
fully-linked executable images do not carry the Coff string table required
2139
to support the longer names.  As a GNU extension, it is possible to
2140
allow their use in executable images as well, or to (probably pointlessly!)
2141
disallow it in object files, by using these two options.  Executable images
2142
generated with these long section names are slightly non-standard, carrying
2143
as they do a string table, and may generate confusing output when examined
2144
with non-GNU PE-aware tools, such as file viewers and dumpers.  However,
2145
GDB relies on the use of PE long section names to find Dwarf-2 debug
2146
information sections in an executable image at runtime, and so if neither
2147
option is specified on the command-line, @command{ld} will enable long
2148
section names, overriding the default and technically correct behaviour,
2149
when it finds the presence of debug information while linking an executable
2150
image and not stripping symbols.
2151
[This option is valid for all PE targeted ports of the linker]
2152
 
2153
@kindex --enable-stdcall-fixup
2154
@kindex --disable-stdcall-fixup
2155
@item --enable-stdcall-fixup
2156
@itemx --disable-stdcall-fixup
2157
If the link finds a symbol that it cannot resolve, it will attempt to
2158
do ``fuzzy linking'' by looking for another defined symbol that differs
2159
only in the format of the symbol name (cdecl vs stdcall) and will
2160
resolve that symbol by linking to the match.  For example, the
2161
undefined symbol @code{_foo} might be linked to the function
2162
@code{_foo@@12}, or the undefined symbol @code{_bar@@16} might be linked
2163
to the function @code{_bar}.  When the linker does this, it prints a
2164
warning, since it normally should have failed to link, but sometimes
2165
import libraries generated from third-party dlls may need this feature
2166
to be usable.  If you specify @option{--enable-stdcall-fixup}, this
2167
feature is fully enabled and warnings are not printed.  If you specify
2168
@option{--disable-stdcall-fixup}, this feature is disabled and such
2169
mismatches are considered to be errors.
2170
[This option is specific to the i386 PE targeted port of the linker]
2171
 
2172
@cindex DLLs, creating
2173
@kindex --export-all-symbols
2174
@item --export-all-symbols
2175
If given, all global symbols in the objects used to build a DLL will
2176
be exported by the DLL.  Note that this is the default if there
2177
otherwise wouldn't be any exported symbols.  When symbols are
2178
explicitly exported via DEF files or implicitly exported via function
2179
attributes, the default is to not export anything else unless this
2180
option is given.  Note that the symbols @code{DllMain@@12},
2181
@code{DllEntryPoint@@0}, @code{DllMainCRTStartup@@12}, and
2182
@code{impure_ptr} will not be automatically
2183
exported.  Also, symbols imported from other DLLs will not be
2184
re-exported, nor will symbols specifying the DLL's internal layout
2185
such as those beginning with @code{_head_} or ending with
2186
@code{_iname}.  In addition, no symbols from @code{libgcc},
2187
@code{libstd++}, @code{libmingw32}, or @code{crtX.o} will be exported.
2188
Symbols whose names begin with @code{__rtti_} or @code{__builtin_} will
2189
not be exported, to help with C++ DLLs.  Finally, there is an
2190
extensive list of cygwin-private symbols that are not exported
2191
(obviously, this applies on when building DLLs for cygwin targets).
2192
These cygwin-excludes are: @code{_cygwin_dll_entry@@12},
2193
@code{_cygwin_crt0_common@@8}, @code{_cygwin_noncygwin_dll_entry@@12},
2194
@code{_fmode}, @code{_impure_ptr}, @code{cygwin_attach_dll},
2195
@code{cygwin_premain0}, @code{cygwin_premain1}, @code{cygwin_premain2},
2196
@code{cygwin_premain3}, and @code{environ}.
2197
[This option is specific to the i386 PE targeted port of the linker]
2198
 
2199
@kindex --exclude-symbols
2200
@item --exclude-symbols @var{symbol},@var{symbol},...
2201
Specifies a list of symbols which should not be automatically
2202
exported.  The symbol names may be delimited by commas or colons.
2203
[This option is specific to the i386 PE targeted port of the linker]
2204
 
2205
@kindex --file-alignment
2206
@item --file-alignment
2207
Specify the file alignment.  Sections in the file will always begin at
2208
file offsets which are multiples of this number.  This defaults to
2209
512.
2210
[This option is specific to the i386 PE targeted port of the linker]
2211
 
2212
@cindex heap size
2213
@kindex --heap
2214
@item --heap @var{reserve}
2215
@itemx --heap @var{reserve},@var{commit}
2216
Specify the number of bytes of memory to reserve (and optionally commit)
2217
to be used as heap for this program.  The default is 1Mb reserved, 4K
2218
committed.
2219
[This option is specific to the i386 PE targeted port of the linker]
2220
 
2221
@cindex image base
2222
@kindex --image-base
2223
@item --image-base @var{value}
2224
Use @var{value} as the base address of your program or dll.  This is
2225
the lowest memory location that will be used when your program or dll
2226
is loaded.  To reduce the need to relocate and improve performance of
2227
your dlls, each should have a unique base address and not overlap any
2228
other dlls.  The default is 0x400000 for executables, and 0x10000000
2229
for dlls.
2230
[This option is specific to the i386 PE targeted port of the linker]
2231
 
2232
@kindex --kill-at
2233
@item --kill-at
2234
If given, the stdcall suffixes (@@@var{nn}) will be stripped from
2235
symbols before they are exported.
2236
[This option is specific to the i386 PE targeted port of the linker]
2237
 
2238
@kindex --large-address-aware
2239
@item --large-address-aware
2240
If given, the appropriate bit in the ``Characteristics'' field of the COFF
2241
header is set to indicate that this executable supports virtual addresses
2242
greater than 2 gigabytes.  This should be used in conjunction with the /3GB
2243
or /USERVA=@var{value} megabytes switch in the ``[operating systems]''
2244
section of the BOOT.INI.  Otherwise, this bit has no effect.
2245
[This option is specific to PE targeted ports of the linker]
2246
 
2247
@kindex --major-image-version
2248
@item --major-image-version @var{value}
2249
Sets the major number of the ``image version''.  Defaults to 1.
2250
[This option is specific to the i386 PE targeted port of the linker]
2251
 
2252
@kindex --major-os-version
2253
@item --major-os-version @var{value}
2254
Sets the major number of the ``os version''.  Defaults to 4.
2255
[This option is specific to the i386 PE targeted port of the linker]
2256
 
2257
@kindex --major-subsystem-version
2258
@item --major-subsystem-version @var{value}
2259
Sets the major number of the ``subsystem version''.  Defaults to 4.
2260
[This option is specific to the i386 PE targeted port of the linker]
2261
 
2262
@kindex --minor-image-version
2263
@item --minor-image-version @var{value}
2264
Sets the minor number of the ``image version''.  Defaults to 0.
2265
[This option is specific to the i386 PE targeted port of the linker]
2266
 
2267
@kindex --minor-os-version
2268
@item --minor-os-version @var{value}
2269
Sets the minor number of the ``os version''.  Defaults to 0.
2270
[This option is specific to the i386 PE targeted port of the linker]
2271
 
2272
@kindex --minor-subsystem-version
2273
@item --minor-subsystem-version @var{value}
2274
Sets the minor number of the ``subsystem version''.  Defaults to 0.
2275
[This option is specific to the i386 PE targeted port of the linker]
2276
 
2277
@cindex DEF files, creating
2278
@cindex DLLs, creating
2279
@kindex --output-def
2280
@item --output-def @var{file}
2281
The linker will create the file @var{file} which will contain a DEF
2282
file corresponding to the DLL the linker is generating.  This DEF file
2283
(which should be called @code{*.def}) may be used to create an import
2284
library with @code{dlltool} or may be used as a reference to
2285
automatically or implicitly exported symbols.
2286
[This option is specific to the i386 PE targeted port of the linker]
2287
 
2288
@cindex DLLs, creating
2289
@kindex --out-implib
2290
@item --out-implib @var{file}
2291
The linker will create the file @var{file} which will contain an
2292
import lib corresponding to the DLL the linker is generating. This
2293
import lib (which should be called @code{*.dll.a} or @code{*.a}
2294
may be used to link clients against the generated DLL; this behaviour
2295
makes it possible to skip a separate @code{dlltool} import library
2296
creation step.
2297
[This option is specific to the i386 PE targeted port of the linker]
2298
 
2299
@kindex --enable-auto-image-base
2300
@item --enable-auto-image-base
2301
Automatically choose the image base for DLLs, unless one is specified
2302
using the @code{--image-base} argument.  By using a hash generated
2303
from the dllname to create unique image bases for each DLL, in-memory
2304
collisions and relocations which can delay program execution are
2305
avoided.
2306
[This option is specific to the i386 PE targeted port of the linker]
2307
 
2308
@kindex --disable-auto-image-base
2309
@item --disable-auto-image-base
2310
Do not automatically generate a unique image base.  If there is no
2311
user-specified image base (@code{--image-base}) then use the platform
2312
default.
2313
[This option is specific to the i386 PE targeted port of the linker]
2314
 
2315
@cindex DLLs, linking to
2316
@kindex --dll-search-prefix
2317
@item --dll-search-prefix @var{string}
2318
When linking dynamically to a dll without an import library,
2319
search for @code{<string><basename>.dll} in preference to
2320
@code{lib<basename>.dll}. This behaviour allows easy distinction
2321
between DLLs built for the various "subplatforms": native, cygwin,
2322
uwin, pw, etc.  For instance, cygwin DLLs typically use
2323
@code{--dll-search-prefix=cyg}.
2324
[This option is specific to the i386 PE targeted port of the linker]
2325
 
2326
@kindex --enable-auto-import
2327
@item --enable-auto-import
2328
Do sophisticated linking of @code{_symbol} to @code{__imp__symbol} for
2329
DATA imports from DLLs, and create the necessary thunking symbols when
2330
building the import libraries with those DATA exports. Note: Use of the
2331
'auto-import' extension will cause the text section of the image file
2332
to be made writable. This does not conform to the PE-COFF format
2333
specification published by Microsoft.
2334
 
2335
Note - use of the 'auto-import' extension will also cause read only
2336
data which would normally be placed into the .rdata section to be
2337
placed into the .data section instead.  This is in order to work
2338
around a problem with consts that is described here:
2339
http://www.cygwin.com/ml/cygwin/2004-09/msg01101.html
2340
 
2341
Using 'auto-import' generally will 'just work' -- but sometimes you may
2342
see this message:
2343
 
2344
"variable '<var>' can't be auto-imported. Please read the
2345
documentation for ld's @code{--enable-auto-import} for details."
2346
 
2347
This message occurs when some (sub)expression accesses an address
2348
ultimately given by the sum of two constants (Win32 import tables only
2349
allow one).  Instances where this may occur include accesses to member
2350
fields of struct variables imported from a DLL, as well as using a
2351
constant index into an array variable imported from a DLL.  Any
2352
multiword variable (arrays, structs, long long, etc) may trigger
2353
this error condition.  However, regardless of the exact data type
2354
of the offending exported variable, ld will always detect it, issue
2355
the warning, and exit.
2356
 
2357
There are several ways to address this difficulty, regardless of the
2358
data type of the exported variable:
2359
 
2360
One way is to use --enable-runtime-pseudo-reloc switch. This leaves the task
2361
of adjusting references in your client code for runtime environment, so
2362
this method works only when runtime environment supports this feature.
2363
 
2364
A second solution is to force one of the 'constants' to be a variable --
2365
that is, unknown and un-optimizable at compile time.  For arrays,
2366
there are two possibilities: a) make the indexee (the array's address)
2367
a variable, or b) make the 'constant' index a variable.  Thus:
2368
 
2369
@example
2370
extern type extern_array[];
2371
extern_array[1] -->
2372
   @{ volatile type *t=extern_array; t[1] @}
2373
@end example
2374
 
2375
or
2376
 
2377
@example
2378
extern type extern_array[];
2379
extern_array[1] -->
2380
   @{ volatile int t=1; extern_array[t] @}
2381
@end example
2382
 
2383
For structs (and most other multiword data types) the only option
2384
is to make the struct itself (or the long long, or the ...) variable:
2385
 
2386
@example
2387
extern struct s extern_struct;
2388
extern_struct.field -->
2389
   @{ volatile struct s *t=&extern_struct; t->field @}
2390
@end example
2391
 
2392
or
2393
 
2394
@example
2395
extern long long extern_ll;
2396
extern_ll -->
2397
  @{ volatile long long * local_ll=&extern_ll; *local_ll @}
2398
@end example
2399
 
2400
A third method of dealing with this difficulty is to abandon
2401
'auto-import' for the offending symbol and mark it with
2402
@code{__declspec(dllimport)}.  However, in practise that
2403
requires using compile-time #defines to indicate whether you are
2404
building a DLL, building client code that will link to the DLL, or
2405
merely building/linking to a static library.   In making the choice
2406
between the various methods of resolving the 'direct address with
2407
constant offset' problem, you should consider typical real-world usage:
2408
 
2409
Original:
2410
@example
2411
--foo.h
2412
extern int arr[];
2413
--foo.c
2414
#include "foo.h"
2415
void main(int argc, char **argv)@{
2416
  printf("%d\n",arr[1]);
2417
@}
2418
@end example
2419
 
2420
Solution 1:
2421
@example
2422
--foo.h
2423
extern int arr[];
2424
--foo.c
2425
#include "foo.h"
2426
void main(int argc, char **argv)@{
2427
  /* This workaround is for win32 and cygwin; do not "optimize" */
2428
  volatile int *parr = arr;
2429
  printf("%d\n",parr[1]);
2430
@}
2431
@end example
2432
 
2433
Solution 2:
2434
@example
2435
--foo.h
2436
/* Note: auto-export is assumed (no __declspec(dllexport)) */
2437
#if (defined(_WIN32) || defined(__CYGWIN__)) && \
2438
  !(defined(FOO_BUILD_DLL) || defined(FOO_STATIC))
2439
#define FOO_IMPORT __declspec(dllimport)
2440
#else
2441
#define FOO_IMPORT
2442
#endif
2443
extern FOO_IMPORT int arr[];
2444
--foo.c
2445
#include "foo.h"
2446
void main(int argc, char **argv)@{
2447
  printf("%d\n",arr[1]);
2448
@}
2449
@end example
2450
 
2451
A fourth way to avoid this problem is to re-code your
2452
library to use a functional interface rather than a data interface
2453
for the offending variables (e.g. set_foo() and get_foo() accessor
2454
functions).
2455
[This option is specific to the i386 PE targeted port of the linker]
2456
 
2457
@kindex --disable-auto-import
2458
@item --disable-auto-import
2459
Do not attempt to do sophisticated linking of @code{_symbol} to
2460
@code{__imp__symbol} for DATA imports from DLLs.
2461
[This option is specific to the i386 PE targeted port of the linker]
2462
 
2463
@kindex --enable-runtime-pseudo-reloc
2464
@item --enable-runtime-pseudo-reloc
2465
If your code contains expressions described in --enable-auto-import section,
2466
that is, DATA imports from DLL with non-zero offset, this switch will create
2467
a vector of 'runtime pseudo relocations' which can be used by runtime
2468
environment to adjust references to such data in your client code.
2469
[This option is specific to the i386 PE targeted port of the linker]
2470
 
2471
@kindex --disable-runtime-pseudo-reloc
2472
@item --disable-runtime-pseudo-reloc
2473
Do not create pseudo relocations for non-zero offset DATA imports from
2474
DLLs.  This is the default.
2475
[This option is specific to the i386 PE targeted port of the linker]
2476
 
2477
@kindex --enable-extra-pe-debug
2478
@item --enable-extra-pe-debug
2479
Show additional debug info related to auto-import symbol thunking.
2480
[This option is specific to the i386 PE targeted port of the linker]
2481
 
2482
@kindex --section-alignment
2483
@item --section-alignment
2484
Sets the section alignment.  Sections in memory will always begin at
2485
addresses which are a multiple of this number.  Defaults to 0x1000.
2486
[This option is specific to the i386 PE targeted port of the linker]
2487
 
2488
@cindex stack size
2489
@kindex --stack
2490
@item --stack @var{reserve}
2491
@itemx --stack @var{reserve},@var{commit}
2492
Specify the number of bytes of memory to reserve (and optionally commit)
2493
to be used as stack for this program.  The default is 2Mb reserved, 4K
2494
committed.
2495
[This option is specific to the i386 PE targeted port of the linker]
2496
 
2497
@kindex --subsystem
2498
@item --subsystem @var{which}
2499
@itemx --subsystem @var{which}:@var{major}
2500
@itemx --subsystem @var{which}:@var{major}.@var{minor}
2501
Specifies the subsystem under which your program will execute.  The
2502
legal values for @var{which} are @code{native}, @code{windows},
2503
@code{console}, @code{posix}, and @code{xbox}.  You may optionally set
2504
the subsystem version also.  Numeric values are also accepted for
2505
@var{which}.
2506
[This option is specific to the i386 PE targeted port of the linker]
2507
 
2508
The following options set flags in the @code{DllCharacteristics} field
2509
of the PE file header:
2510
[These options are specific to PE targeted ports of the linker]
2511
 
2512
@kindex --dynamicbase
2513
@item --dynamicbase
2514
The image base address may be relocated using address space layout
2515
randomization (ASLR).  This feature was introduced with MS Windows
2516
Vista for i386 PE targets.
2517
 
2518
@kindex --forceinteg
2519
@item --forceinteg
2520
Code integrity checks are enforced.
2521
 
2522
@kindex --nxcompat
2523
@item --nxcompat
2524
The image is compatible with the Data Execution Prevention.
2525
This feature was introduced with MS Windows XP SP2 for i386 PE targets.
2526
 
2527
@kindex --no-isolation
2528
@item --no-isolation
2529
Although the image understands isolation, do not isolate the image.
2530
 
2531
@kindex --no-seh
2532
@item --no-seh
2533
The image does not use SEH. No SE handler may be called from
2534
this image.
2535
 
2536
@kindex --no-bind
2537
@item --no-bind
2538
Do not bind this image.
2539
 
2540
@kindex --wdmdriver
2541
@item --wdmdriver
2542
The driver uses the MS Windows Driver Model.
2543
 
2544
@kindex --tsaware
2545
@item --tsaware
2546
The image is Terminal Server aware.
2547
 
2548
@end table
2549
 
2550
@c man end
2551
 
2552
@ifset M68HC11
2553
@subsection Options specific to Motorola 68HC11 and 68HC12 targets
2554
 
2555
@c man begin OPTIONS
2556
 
2557
The 68HC11 and 68HC12 linkers support specific options to control the
2558
memory bank switching mapping and trampoline code generation.
2559
 
2560
@table @gcctabopt
2561
 
2562
@kindex --no-trampoline
2563
@item --no-trampoline
2564
This option disables the generation of trampoline. By default a trampoline
2565
is generated for each far function which is called using a @code{jsr}
2566
instruction (this happens when a pointer to a far function is taken).
2567
 
2568
@kindex --bank-window
2569
@item --bank-window @var{name}
2570
This option indicates to the linker the name of the memory region in
2571
the @samp{MEMORY} specification that describes the memory bank window.
2572
The definition of such region is then used by the linker to compute
2573
paging and addresses within the memory window.
2574
 
2575
@end table
2576
 
2577
@c man end
2578
@end ifset
2579
 
2580
@ifset M68K
2581
@subsection Options specific to Motorola 68K target
2582
 
2583
@c man begin OPTIONS
2584
 
2585
The following options are supported to control handling of GOT generation
2586
when linking for 68K targets.
2587
 
2588
@table @gcctabopt
2589
 
2590
@kindex --got
2591
@item --got=@var{type}
2592
This option tells the linker which GOT generation scheme to use.
2593
@var{type} should be one of @samp{single}, @samp{negative},
2594
@samp{multigot} or @samp{target}.  For more information refer to the
2595
Info entry for @file{ld}.
2596
 
2597
@end table
2598
 
2599
@c man end
2600
@end ifset
2601
 
2602
@ifset UsesEnvVars
2603
@node Environment
2604
@section Environment Variables
2605
 
2606
@c man begin ENVIRONMENT
2607
 
2608
You can change the behaviour of @command{ld} with the environment variables
2609
@ifclear SingleFormat
2610
@code{GNUTARGET},
2611
@end ifclear
2612
@code{LDEMULATION} and @code{COLLECT_NO_DEMANGLE}.
2613
 
2614
@ifclear SingleFormat
2615
@kindex GNUTARGET
2616
@cindex default input format
2617
@code{GNUTARGET} determines the input-file object format if you don't
2618
use @samp{-b} (or its synonym @samp{--format}).  Its value should be one
2619
of the BFD names for an input format (@pxref{BFD}).  If there is no
2620
@code{GNUTARGET} in the environment, @command{ld} uses the natural format
2621
of the target. If @code{GNUTARGET} is set to @code{default} then BFD
2622
attempts to discover the input format by examining binary input files;
2623
this method often succeeds, but there are potential ambiguities, since
2624
there is no method of ensuring that the magic number used to specify
2625
object-file formats is unique.  However, the configuration procedure for
2626
BFD on each system places the conventional format for that system first
2627
in the search-list, so ambiguities are resolved in favor of convention.
2628
@end ifclear
2629
 
2630
@kindex LDEMULATION
2631
@cindex default emulation
2632
@cindex emulation, default
2633
@code{LDEMULATION} determines the default emulation if you don't use the
2634
@samp{-m} option.  The emulation can affect various aspects of linker
2635
behaviour, particularly the default linker script.  You can list the
2636
available emulations with the @samp{--verbose} or @samp{-V} options.  If
2637
the @samp{-m} option is not used, and the @code{LDEMULATION} environment
2638
variable is not defined, the default emulation depends upon how the
2639
linker was configured.
2640
 
2641
@kindex COLLECT_NO_DEMANGLE
2642
@cindex demangling, default
2643
Normally, the linker will default to demangling symbols.  However, if
2644
@code{COLLECT_NO_DEMANGLE} is set in the environment, then it will
2645
default to not demangling symbols.  This environment variable is used in
2646
a similar fashion by the @code{gcc} linker wrapper program.  The default
2647
may be overridden by the @samp{--demangle} and @samp{--no-demangle}
2648
options.
2649
 
2650
@c man end
2651
@end ifset
2652
 
2653
@node Scripts
2654
@chapter Linker Scripts
2655
 
2656
@cindex scripts
2657
@cindex linker scripts
2658
@cindex command files
2659
Every link is controlled by a @dfn{linker script}.  This script is
2660
written in the linker command language.
2661
 
2662
The main purpose of the linker script is to describe how the sections in
2663
the input files should be mapped into the output file, and to control
2664
the memory layout of the output file.  Most linker scripts do nothing
2665
more than this.  However, when necessary, the linker script can also
2666
direct the linker to perform many other operations, using the commands
2667
described below.
2668
 
2669
The linker always uses a linker script.  If you do not supply one
2670
yourself, the linker will use a default script that is compiled into the
2671
linker executable.  You can use the @samp{--verbose} command line option
2672
to display the default linker script.  Certain command line options,
2673
such as @samp{-r} or @samp{-N}, will affect the default linker script.
2674
 
2675
You may supply your own linker script by using the @samp{-T} command
2676
line option.  When you do this, your linker script will replace the
2677
default linker script.
2678
 
2679
You may also use linker scripts implicitly by naming them as input files
2680
to the linker, as though they were files to be linked.  @xref{Implicit
2681
Linker Scripts}.
2682
 
2683
@menu
2684
* Basic Script Concepts::       Basic Linker Script Concepts
2685
* Script Format::               Linker Script Format
2686
* Simple Example::              Simple Linker Script Example
2687
* Simple Commands::             Simple Linker Script Commands
2688
* Assignments::                 Assigning Values to Symbols
2689
* SECTIONS::                    SECTIONS Command
2690
* MEMORY::                      MEMORY Command
2691
* PHDRS::                       PHDRS Command
2692
* VERSION::                     VERSION Command
2693
* Expressions::                 Expressions in Linker Scripts
2694
* Implicit Linker Scripts::     Implicit Linker Scripts
2695
@end menu
2696
 
2697
@node Basic Script Concepts
2698
@section Basic Linker Script Concepts
2699
@cindex linker script concepts
2700
We need to define some basic concepts and vocabulary in order to
2701
describe the linker script language.
2702
 
2703
The linker combines input files into a single output file.  The output
2704
file and each input file are in a special data format known as an
2705
@dfn{object file format}.  Each file is called an @dfn{object file}.
2706
The output file is often called an @dfn{executable}, but for our
2707
purposes we will also call it an object file.  Each object file has,
2708
among other things, a list of @dfn{sections}.  We sometimes refer to a
2709
section in an input file as an @dfn{input section}; similarly, a section
2710
in the output file is an @dfn{output section}.
2711
 
2712
Each section in an object file has a name and a size.  Most sections
2713
also have an associated block of data, known as the @dfn{section
2714
contents}.  A section may be marked as @dfn{loadable}, which mean that
2715
the contents should be loaded into memory when the output file is run.
2716
A section with no contents may be @dfn{allocatable}, which means that an
2717
area in memory should be set aside, but nothing in particular should be
2718
loaded there (in some cases this memory must be zeroed out).  A section
2719
which is neither loadable nor allocatable typically contains some sort
2720
of debugging information.
2721
 
2722
Every loadable or allocatable output section has two addresses.  The
2723
first is the @dfn{VMA}, or virtual memory address.  This is the address
2724
the section will have when the output file is run.  The second is the
2725
@dfn{LMA}, or load memory address.  This is the address at which the
2726
section will be loaded.  In most cases the two addresses will be the
2727
same.  An example of when they might be different is when a data section
2728
is loaded into ROM, and then copied into RAM when the program starts up
2729
(this technique is often used to initialize global variables in a ROM
2730
based system).  In this case the ROM address would be the LMA, and the
2731
RAM address would be the VMA.
2732
 
2733
You can see the sections in an object file by using the @code{objdump}
2734
program with the @samp{-h} option.
2735
 
2736
Every object file also has a list of @dfn{symbols}, known as the
2737
@dfn{symbol table}.  A symbol may be defined or undefined.  Each symbol
2738
has a name, and each defined symbol has an address, among other
2739
information.  If you compile a C or C++ program into an object file, you
2740
will get a defined symbol for every defined function and global or
2741
static variable.  Every undefined function or global variable which is
2742
referenced in the input file will become an undefined symbol.
2743
 
2744
You can see the symbols in an object file by using the @code{nm}
2745
program, or by using the @code{objdump} program with the @samp{-t}
2746
option.
2747
 
2748
@node Script Format
2749
@section Linker Script Format
2750
@cindex linker script format
2751
Linker scripts are text files.
2752
 
2753
You write a linker script as a series of commands.  Each command is
2754
either a keyword, possibly followed by arguments, or an assignment to a
2755
symbol.  You may separate commands using semicolons.  Whitespace is
2756
generally ignored.
2757
 
2758
Strings such as file or format names can normally be entered directly.
2759
If the file name contains a character such as a comma which would
2760
otherwise serve to separate file names, you may put the file name in
2761
double quotes.  There is no way to use a double quote character in a
2762
file name.
2763
 
2764
You may include comments in linker scripts just as in C, delimited by
2765
@samp{/*} and @samp{*/}.  As in C, comments are syntactically equivalent
2766
to whitespace.
2767
 
2768
@node Simple Example
2769
@section Simple Linker Script Example
2770
@cindex linker script example
2771
@cindex example of linker script
2772
Many linker scripts are fairly simple.
2773
 
2774
The simplest possible linker script has just one command:
2775
@samp{SECTIONS}.  You use the @samp{SECTIONS} command to describe the
2776
memory layout of the output file.
2777
 
2778
The @samp{SECTIONS} command is a powerful command.  Here we will
2779
describe a simple use of it.  Let's assume your program consists only of
2780
code, initialized data, and uninitialized data.  These will be in the
2781
@samp{.text}, @samp{.data}, and @samp{.bss} sections, respectively.
2782
Let's assume further that these are the only sections which appear in
2783
your input files.
2784
 
2785
For this example, let's say that the code should be loaded at address
2786
0x10000, and that the data should start at address 0x8000000.  Here is a
2787
linker script which will do that:
2788
@smallexample
2789
SECTIONS
2790
@{
2791
  . = 0x10000;
2792
  .text : @{ *(.text) @}
2793
  . = 0x8000000;
2794
  .data : @{ *(.data) @}
2795
  .bss : @{ *(.bss) @}
2796
@}
2797
@end smallexample
2798
 
2799
You write the @samp{SECTIONS} command as the keyword @samp{SECTIONS},
2800
followed by a series of symbol assignments and output section
2801
descriptions enclosed in curly braces.
2802
 
2803
The first line inside the @samp{SECTIONS} command of the above example
2804
sets the value of the special symbol @samp{.}, which is the location
2805
counter.  If you do not specify the address of an output section in some
2806
other way (other ways are described later), the address is set from the
2807
current value of the location counter.  The location counter is then
2808
incremented by the size of the output section.  At the start of the
2809
@samp{SECTIONS} command, the location counter has the value @samp{0}.
2810
 
2811
The second line defines an output section, @samp{.text}.  The colon is
2812
required syntax which may be ignored for now.  Within the curly braces
2813
after the output section name, you list the names of the input sections
2814
which should be placed into this output section.  The @samp{*} is a
2815
wildcard which matches any file name.  The expression @samp{*(.text)}
2816
means all @samp{.text} input sections in all input files.
2817
 
2818
Since the location counter is @samp{0x10000} when the output section
2819
@samp{.text} is defined, the linker will set the address of the
2820
@samp{.text} section in the output file to be @samp{0x10000}.
2821
 
2822
The remaining lines define the @samp{.data} and @samp{.bss} sections in
2823
the output file.  The linker will place the @samp{.data} output section
2824
at address @samp{0x8000000}.  After the linker places the @samp{.data}
2825
output section, the value of the location counter will be
2826
@samp{0x8000000} plus the size of the @samp{.data} output section.  The
2827
effect is that the linker will place the @samp{.bss} output section
2828
immediately after the @samp{.data} output section in memory.
2829
 
2830
The linker will ensure that each output section has the required
2831
alignment, by increasing the location counter if necessary.  In this
2832
example, the specified addresses for the @samp{.text} and @samp{.data}
2833
sections will probably satisfy any alignment constraints, but the linker
2834
may have to create a small gap between the @samp{.data} and @samp{.bss}
2835
sections.
2836
 
2837
That's it!  That's a simple and complete linker script.
2838
 
2839
@node Simple Commands
2840
@section Simple Linker Script Commands
2841
@cindex linker script simple commands
2842
In this section we describe the simple linker script commands.
2843
 
2844
@menu
2845
* Entry Point::                 Setting the entry point
2846
* File Commands::               Commands dealing with files
2847
@ifclear SingleFormat
2848
* Format Commands::             Commands dealing with object file formats
2849
@end ifclear
2850
 
2851
* REGION_ALIAS::                Assign alias names to memory regions
2852
* Miscellaneous Commands::      Other linker script commands
2853
@end menu
2854
 
2855
@node Entry Point
2856
@subsection Setting the Entry Point
2857
@kindex ENTRY(@var{symbol})
2858
@cindex start of execution
2859
@cindex first instruction
2860
@cindex entry point
2861
The first instruction to execute in a program is called the @dfn{entry
2862
point}.  You can use the @code{ENTRY} linker script command to set the
2863
entry point.  The argument is a symbol name:
2864
@smallexample
2865
ENTRY(@var{symbol})
2866
@end smallexample
2867
 
2868
There are several ways to set the entry point.  The linker will set the
2869
entry point by trying each of the following methods in order, and
2870
stopping when one of them succeeds:
2871
@itemize @bullet
2872
@item
2873
the @samp{-e} @var{entry} command-line option;
2874
@item
2875
the @code{ENTRY(@var{symbol})} command in a linker script;
2876
@item
2877
the value of the symbol @code{start}, if defined;
2878
@item
2879
the address of the first byte of the @samp{.text} section, if present;
2880
@item
2881
The address @code{0}.
2882
@end itemize
2883
 
2884
@node File Commands
2885
@subsection Commands Dealing with Files
2886
@cindex linker script file commands
2887
Several linker script commands deal with files.
2888
 
2889
@table @code
2890
@item INCLUDE @var{filename}
2891
@kindex INCLUDE @var{filename}
2892
@cindex including a linker script
2893
Include the linker script @var{filename} at this point.  The file will
2894
be searched for in the current directory, and in any directory specified
2895
with the @option{-L} option.  You can nest calls to @code{INCLUDE} up to
2896
10 levels deep.
2897
 
2898
You can place @code{INCLUDE} directives at the top level, in @code{MEMORY} or
2899
@code{SECTIONS} commands, or in output section descriptions.
2900
 
2901
@item INPUT(@var{file}, @var{file}, @dots{})
2902
@itemx INPUT(@var{file} @var{file} @dots{})
2903
@kindex INPUT(@var{files})
2904
@cindex input files in linker scripts
2905
@cindex input object files in linker scripts
2906
@cindex linker script input object files
2907
The @code{INPUT} command directs the linker to include the named files
2908
in the link, as though they were named on the command line.
2909
 
2910
For example, if you always want to include @file{subr.o} any time you do
2911
a link, but you can't be bothered to put it on every link command line,
2912
then you can put @samp{INPUT (subr.o)} in your linker script.
2913
 
2914
In fact, if you like, you can list all of your input files in the linker
2915
script, and then invoke the linker with nothing but a @samp{-T} option.
2916
 
2917
In case a @dfn{sysroot prefix} is configured, and the filename starts
2918
with the @samp{/} character, and the script being processed was
2919
located inside the @dfn{sysroot prefix}, the filename will be looked
2920
for in the @dfn{sysroot prefix}.  Otherwise, the linker will try to
2921
open the file in the current directory.  If it is not found, the
2922
linker will search through the archive library search path.  See the
2923
description of @samp{-L} in @ref{Options,,Command Line Options}.
2924
 
2925
If you use @samp{INPUT (-l@var{file})}, @command{ld} will transform the
2926
name to @code{lib@var{file}.a}, as with the command line argument
2927
@samp{-l}.
2928
 
2929
When you use the @code{INPUT} command in an implicit linker script, the
2930
files will be included in the link at the point at which the linker
2931
script file is included.  This can affect archive searching.
2932
 
2933
@item GROUP(@var{file}, @var{file}, @dots{})
2934
@itemx GROUP(@var{file} @var{file} @dots{})
2935
@kindex GROUP(@var{files})
2936
@cindex grouping input files
2937
The @code{GROUP} command is like @code{INPUT}, except that the named
2938
files should all be archives, and they are searched repeatedly until no
2939
new undefined references are created.  See the description of @samp{-(}
2940
in @ref{Options,,Command Line Options}.
2941
 
2942
@item AS_NEEDED(@var{file}, @var{file}, @dots{})
2943
@itemx AS_NEEDED(@var{file} @var{file} @dots{})
2944
@kindex AS_NEEDED(@var{files})
2945
This construct can appear only inside of the @code{INPUT} or @code{GROUP}
2946
commands, among other filenames.  The files listed will be handled
2947
as if they appear directly in the @code{INPUT} or @code{GROUP} commands,
2948
with the exception of ELF shared libraries, that will be added only
2949
when they are actually needed.  This construct essentially enables
2950
@option{--as-needed} option for all the files listed inside of it
2951
and restores previous @option{--as-needed} resp. @option{--no-as-needed}
2952
setting afterwards.
2953
 
2954
@item OUTPUT(@var{filename})
2955
@kindex OUTPUT(@var{filename})
2956
@cindex output file name in linker script
2957
The @code{OUTPUT} command names the output file.  Using
2958
@code{OUTPUT(@var{filename})} in the linker script is exactly like using
2959
@samp{-o @var{filename}} on the command line (@pxref{Options,,Command
2960
Line Options}).  If both are used, the command line option takes
2961
precedence.
2962
 
2963
You can use the @code{OUTPUT} command to define a default name for the
2964
output file other than the usual default of @file{a.out}.
2965
 
2966
@item SEARCH_DIR(@var{path})
2967
@kindex SEARCH_DIR(@var{path})
2968
@cindex library search path in linker script
2969
@cindex archive search path in linker script
2970
@cindex search path in linker script
2971
The @code{SEARCH_DIR} command adds @var{path} to the list of paths where
2972
@command{ld} looks for archive libraries.  Using
2973
@code{SEARCH_DIR(@var{path})} is exactly like using @samp{-L @var{path}}
2974
on the command line (@pxref{Options,,Command Line Options}).  If both
2975
are used, then the linker will search both paths.  Paths specified using
2976
the command line option are searched first.
2977
 
2978
@item STARTUP(@var{filename})
2979
@kindex STARTUP(@var{filename})
2980
@cindex first input file
2981
The @code{STARTUP} command is just like the @code{INPUT} command, except
2982
that @var{filename} will become the first input file to be linked, as
2983
though it were specified first on the command line.  This may be useful
2984
when using a system in which the entry point is always the start of the
2985
first file.
2986
@end table
2987
 
2988
@ifclear SingleFormat
2989
@node Format Commands
2990
@subsection Commands Dealing with Object File Formats
2991
A couple of linker script commands deal with object file formats.
2992
 
2993
@table @code
2994
@item OUTPUT_FORMAT(@var{bfdname})
2995
@itemx OUTPUT_FORMAT(@var{default}, @var{big}, @var{little})
2996
@kindex OUTPUT_FORMAT(@var{bfdname})
2997
@cindex output file format in linker script
2998
The @code{OUTPUT_FORMAT} command names the BFD format to use for the
2999
output file (@pxref{BFD}).  Using @code{OUTPUT_FORMAT(@var{bfdname})} is
3000
exactly like using @samp{--oformat @var{bfdname}} on the command line
3001
(@pxref{Options,,Command Line Options}).  If both are used, the command
3002
line option takes precedence.
3003
 
3004
You can use @code{OUTPUT_FORMAT} with three arguments to use different
3005
formats based on the @samp{-EB} and @samp{-EL} command line options.
3006
This permits the linker script to set the output format based on the
3007
desired endianness.
3008
 
3009
If neither @samp{-EB} nor @samp{-EL} are used, then the output format
3010
will be the first argument, @var{default}.  If @samp{-EB} is used, the
3011
output format will be the second argument, @var{big}.  If @samp{-EL} is
3012
used, the output format will be the third argument, @var{little}.
3013
 
3014
For example, the default linker script for the MIPS ELF target uses this
3015
command:
3016
@smallexample
3017
OUTPUT_FORMAT(elf32-bigmips, elf32-bigmips, elf32-littlemips)
3018
@end smallexample
3019
This says that the default format for the output file is
3020
@samp{elf32-bigmips}, but if the user uses the @samp{-EL} command line
3021
option, the output file will be created in the @samp{elf32-littlemips}
3022
format.
3023
 
3024
@item TARGET(@var{bfdname})
3025
@kindex TARGET(@var{bfdname})
3026
@cindex input file format in linker script
3027
The @code{TARGET} command names the BFD format to use when reading input
3028
files.  It affects subsequent @code{INPUT} and @code{GROUP} commands.
3029
This command is like using @samp{-b @var{bfdname}} on the command line
3030
(@pxref{Options,,Command Line Options}).  If the @code{TARGET} command
3031
is used but @code{OUTPUT_FORMAT} is not, then the last @code{TARGET}
3032
command is also used to set the format for the output file.  @xref{BFD}.
3033
@end table
3034
@end ifclear
3035
 
3036
@node REGION_ALIAS
3037
@subsection Assign alias names to memory regions
3038
@kindex REGION_ALIAS(@var{alias}, @var{region})
3039
@cindex region alias
3040
@cindex region names
3041
 
3042
Alias names can be added to existing memory regions created with the
3043
@ref{MEMORY} command.  Each name corresponds to at most one memory region.
3044
 
3045
@smallexample
3046
REGION_ALIAS(@var{alias}, @var{region})
3047
@end smallexample
3048
 
3049
The @code{REGION_ALIAS} function creates an alias name @var{alias} for the
3050
memory region @var{region}.  This allows a flexible mapping of output sections
3051
to memory regions.  An example follows.
3052
 
3053
Suppose we have an application for embedded systems which come with various
3054
memory storage devices.  All have a general purpose, volatile memory @code{RAM}
3055
that allows code execution or data storage.  Some may have a read-only,
3056
non-volatile memory @code{ROM} that allows code execution and read-only data
3057
access.  The last variant is a read-only, non-volatile memory @code{ROM2} with
3058
read-only data access and no code execution capability.  We have four output
3059
sections:
3060
 
3061
@itemize @bullet
3062
@item
3063
@code{.text} program code;
3064
@item
3065
@code{.rodata} read-only data;
3066
@item
3067
@code{.data} read-write initialized data;
3068
@item
3069
@code{.bss} read-write zero initialized data.
3070
@end itemize
3071
 
3072
The goal is to provide a linker command file that contains a system independent
3073
part defining the output sections and a system dependent part mapping the
3074
output sections to the memory regions available on the system.  Our embedded
3075
systems come with three different memory setups @code{A}, @code{B} and
3076
@code{C}:
3077
@multitable @columnfractions .25 .25 .25 .25
3078
@item Section @tab Variant A @tab Variant B @tab Variant C
3079
@item .text @tab RAM @tab ROM @tab ROM
3080
@item .rodata @tab RAM @tab ROM @tab ROM2
3081
@item .data @tab RAM @tab RAM/ROM @tab RAM/ROM2
3082
@item .bss @tab RAM @tab RAM @tab RAM
3083
@end multitable
3084
The notation @code{RAM/ROM} or @code{RAM/ROM2} means that this section is
3085
loaded into region @code{ROM} or @code{ROM2} respectively.  Please note that
3086
the load address of the @code{.data} section starts in all three variants at
3087
the end of the @code{.rodata} section.
3088
 
3089
The base linker script that deals with the output sections follows.  It
3090
includes the system dependent @code{linkcmds.memory} file that describes the
3091
memory layout:
3092
@smallexample
3093
INCLUDE linkcmds.memory
3094
 
3095
SECTIONS
3096
  @{
3097
    .text :
3098
      @{
3099
        *(.text)
3100
      @} > REGION_TEXT
3101
    .rodata :
3102
      @{
3103
        *(.rodata)
3104
        rodata_end = .;
3105
      @} > REGION_RODATA
3106
    .data : AT (rodata_end)
3107
      @{
3108
        data_start = .;
3109
        *(.data)
3110
      @} > REGION_DATA
3111
    data_size = SIZEOF(.data);
3112
    data_load_start = LOADADDR(.data);
3113
    .bss :
3114
      @{
3115
        *(.bss)
3116
      @} > REGION_BSS
3117
  @}
3118
@end smallexample
3119
 
3120
Now we need three different @code{linkcmds.memory} files to define memory
3121
regions and alias names.  The content of @code{linkcmds.memory} for the three
3122
variants @code{A}, @code{B} and @code{C}:
3123
@table @code
3124
@item A
3125
Here everything goes into the @code{RAM}.
3126
@smallexample
3127
MEMORY
3128
  @{
3129
    RAM : ORIGIN = 0, LENGTH = 4M
3130
  @}
3131
 
3132
REGION_ALIAS("REGION_TEXT", RAM);
3133
REGION_ALIAS("REGION_RODATA", RAM);
3134
REGION_ALIAS("REGION_DATA", RAM);
3135
REGION_ALIAS("REGION_BSS", RAM);
3136
@end smallexample
3137
@item B
3138
Program code and read-only data go into the @code{ROM}.  Read-write data goes
3139
into the @code{RAM}.  An image of the initialized data is loaded into the
3140
@code{ROM} and will be copied during system start into the @code{RAM}.
3141
@smallexample
3142
MEMORY
3143
  @{
3144
    ROM : ORIGIN = 0, LENGTH = 3M
3145
    RAM : ORIGIN = 0x10000000, LENGTH = 1M
3146
  @}
3147
 
3148
REGION_ALIAS("REGION_TEXT", ROM);
3149
REGION_ALIAS("REGION_RODATA", ROM);
3150
REGION_ALIAS("REGION_DATA", RAM);
3151
REGION_ALIAS("REGION_BSS", RAM);
3152
@end smallexample
3153
@item C
3154
Program code goes into the @code{ROM}.  Read-only data goes into the
3155
@code{ROM2}.  Read-write data goes into the @code{RAM}.  An image of the
3156
initialized data is loaded into the @code{ROM2} and will be copied during
3157
system start into the @code{RAM}.
3158
@smallexample
3159
MEMORY
3160
  @{
3161
    ROM : ORIGIN = 0, LENGTH = 2M
3162
    ROM2 : ORIGIN = 0x10000000, LENGTH = 1M
3163
    RAM : ORIGIN = 0x20000000, LENGTH = 1M
3164
  @}
3165
 
3166
REGION_ALIAS("REGION_TEXT", ROM);
3167
REGION_ALIAS("REGION_RODATA", ROM2);
3168
REGION_ALIAS("REGION_DATA", RAM);
3169
REGION_ALIAS("REGION_BSS", RAM);
3170
@end smallexample
3171
@end table
3172
 
3173
It is possible to write a common system initialization routine to copy the
3174
@code{.data} section from @code{ROM} or @code{ROM2} into the @code{RAM} if
3175
necessary:
3176
@smallexample
3177
#include <string.h>
3178
 
3179
extern char data_start [];
3180
extern char data_size [];
3181
extern char data_load_start [];
3182
 
3183
void copy_data(void)
3184
@{
3185
  if (data_start != data_load_start)
3186
    @{
3187
      memcpy(data_start, data_load_start, (size_t) data_size);
3188
    @}
3189
@}
3190
@end smallexample
3191
 
3192
@node Miscellaneous Commands
3193
@subsection Other Linker Script Commands
3194
There are a few other linker scripts commands.
3195
 
3196
@table @code
3197
@item ASSERT(@var{exp}, @var{message})
3198
@kindex ASSERT
3199
@cindex assertion in linker script
3200
Ensure that @var{exp} is non-zero.  If it is zero, then exit the linker
3201
with an error code, and print @var{message}.
3202
 
3203
@item EXTERN(@var{symbol} @var{symbol} @dots{})
3204
@kindex EXTERN
3205
@cindex undefined symbol in linker script
3206
Force @var{symbol} to be entered in the output file as an undefined
3207
symbol.  Doing this may, for example, trigger linking of additional
3208
modules from standard libraries.  You may list several @var{symbol}s for
3209
each @code{EXTERN}, and you may use @code{EXTERN} multiple times.  This
3210
command has the same effect as the @samp{-u} command-line option.
3211
 
3212
@item FORCE_COMMON_ALLOCATION
3213
@kindex FORCE_COMMON_ALLOCATION
3214
@cindex common allocation in linker script
3215
This command has the same effect as the @samp{-d} command-line option:
3216
to make @command{ld} assign space to common symbols even if a relocatable
3217
output file is specified (@samp{-r}).
3218
 
3219
@item INHIBIT_COMMON_ALLOCATION
3220
@kindex INHIBIT_COMMON_ALLOCATION
3221
@cindex common allocation in linker script
3222
This command has the same effect as the @samp{--no-define-common}
3223
command-line option: to make @code{ld} omit the assignment of addresses
3224
to common symbols even for a non-relocatable output file.
3225
 
3226
@item INSERT [ AFTER | BEFORE ] @var{output_section}
3227
@kindex INSERT
3228
@cindex insert user script into default script
3229
This command is typically used in a script specified by @samp{-T} to
3230
augment the default @code{SECTIONS} with, for example, overlays.  It
3231
inserts all prior linker script statements after (or before)
3232
@var{output_section}, and also causes @samp{-T} to not override the
3233
default linker script.  The exact insertion point is as for orphan
3234
sections.  @xref{Location Counter}.  The insertion happens after the
3235
linker has mapped input sections to output sections.  Prior to the
3236
insertion, since @samp{-T} scripts are parsed before the default
3237
linker script, statements in the @samp{-T} script occur before the
3238
default linker script statements in the internal linker representation
3239
of the script.  In particular, input section assignments will be made
3240
to @samp{-T} output sections before those in the default script.  Here
3241
is an example of how a @samp{-T} script using @code{INSERT} might look:
3242
 
3243
@smallexample
3244
SECTIONS
3245
@{
3246
  OVERLAY :
3247
  @{
3248
    .ov1 @{ ov1*(.text) @}
3249
    .ov2 @{ ov2*(.text) @}
3250
  @}
3251
@}
3252
INSERT AFTER .text;
3253
@end smallexample
3254
 
3255
@item NOCROSSREFS(@var{section} @var{section} @dots{})
3256
@kindex NOCROSSREFS(@var{sections})
3257
@cindex cross references
3258
This command may be used to tell @command{ld} to issue an error about any
3259
references among certain output sections.
3260
 
3261
In certain types of programs, particularly on embedded systems when
3262
using overlays, when one section is loaded into memory, another section
3263
will not be.  Any direct references between the two sections would be
3264
errors.  For example, it would be an error if code in one section called
3265
a function defined in the other section.
3266
 
3267
The @code{NOCROSSREFS} command takes a list of output section names.  If
3268
@command{ld} detects any cross references between the sections, it reports
3269
an error and returns a non-zero exit status.  Note that the
3270
@code{NOCROSSREFS} command uses output section names, not input section
3271
names.
3272
 
3273
@ifclear SingleFormat
3274
@item OUTPUT_ARCH(@var{bfdarch})
3275
@kindex OUTPUT_ARCH(@var{bfdarch})
3276
@cindex machine architecture
3277
@cindex architecture
3278
Specify a particular output machine architecture.  The argument is one
3279
of the names used by the BFD library (@pxref{BFD}).  You can see the
3280
architecture of an object file by using the @code{objdump} program with
3281
the @samp{-f} option.
3282
@end ifclear
3283
@end table
3284
 
3285
@node Assignments
3286
@section Assigning Values to Symbols
3287
@cindex assignment in scripts
3288
@cindex symbol definition, scripts
3289
@cindex variables, defining
3290
You may assign a value to a symbol in a linker script.  This will define
3291
the symbol and place it into the symbol table with a global scope.
3292
 
3293
@menu
3294
* Simple Assignments::          Simple Assignments
3295
* PROVIDE::                     PROVIDE
3296
* PROVIDE_HIDDEN::              PROVIDE_HIDDEN
3297
* Source Code Reference::       How to use a linker script defined symbol in source code
3298
@end menu
3299
 
3300
@node Simple Assignments
3301
@subsection Simple Assignments
3302
 
3303
You may assign to a symbol using any of the C assignment operators:
3304
 
3305
@table @code
3306
@item @var{symbol} = @var{expression} ;
3307
@itemx @var{symbol} += @var{expression} ;
3308
@itemx @var{symbol} -= @var{expression} ;
3309
@itemx @var{symbol} *= @var{expression} ;
3310
@itemx @var{symbol} /= @var{expression} ;
3311
@itemx @var{symbol} <<= @var{expression} ;
3312
@itemx @var{symbol} >>= @var{expression} ;
3313
@itemx @var{symbol} &= @var{expression} ;
3314
@itemx @var{symbol} |= @var{expression} ;
3315
@end table
3316
 
3317
The first case will define @var{symbol} to the value of
3318
@var{expression}.  In the other cases, @var{symbol} must already be
3319
defined, and the value will be adjusted accordingly.
3320
 
3321
The special symbol name @samp{.} indicates the location counter.  You
3322
may only use this within a @code{SECTIONS} command.  @xref{Location Counter}.
3323
 
3324
The semicolon after @var{expression} is required.
3325
 
3326
Expressions are defined below; see @ref{Expressions}.
3327
 
3328
You may write symbol assignments as commands in their own right, or as
3329
statements within a @code{SECTIONS} command, or as part of an output
3330
section description in a @code{SECTIONS} command.
3331
 
3332
The section of the symbol will be set from the section of the
3333
expression; for more information, see @ref{Expression Section}.
3334
 
3335
Here is an example showing the three different places that symbol
3336
assignments may be used:
3337
 
3338
@smallexample
3339
floating_point = 0;
3340
SECTIONS
3341
@{
3342
  .text :
3343
    @{
3344
      *(.text)
3345
      _etext = .;
3346
    @}
3347
  _bdata = (. + 3) & ~ 3;
3348
  .data : @{ *(.data) @}
3349
@}
3350
@end smallexample
3351
@noindent
3352
In this example, the symbol @samp{floating_point} will be defined as
3353
zero.  The symbol @samp{_etext} will be defined as the address following
3354
the last @samp{.text} input section.  The symbol @samp{_bdata} will be
3355
defined as the address following the @samp{.text} output section aligned
3356
upward to a 4 byte boundary.
3357
 
3358
@node PROVIDE
3359
@subsection PROVIDE
3360
@cindex PROVIDE
3361
In some cases, it is desirable for a linker script to define a symbol
3362
only if it is referenced and is not defined by any object included in
3363
the link.  For example, traditional linkers defined the symbol
3364
@samp{etext}.  However, ANSI C requires that the user be able to use
3365
@samp{etext} as a function name without encountering an error.  The
3366
@code{PROVIDE} keyword may be used to define a symbol, such as
3367
@samp{etext}, only if it is referenced but not defined.  The syntax is
3368
@code{PROVIDE(@var{symbol} = @var{expression})}.
3369
 
3370
Here is an example of using @code{PROVIDE} to define @samp{etext}:
3371
@smallexample
3372
SECTIONS
3373
@{
3374
  .text :
3375
    @{
3376
      *(.text)
3377
      _etext = .;
3378
      PROVIDE(etext = .);
3379
    @}
3380
@}
3381
@end smallexample
3382
 
3383
In this example, if the program defines @samp{_etext} (with a leading
3384
underscore), the linker will give a multiple definition error.  If, on
3385
the other hand, the program defines @samp{etext} (with no leading
3386
underscore), the linker will silently use the definition in the program.
3387
If the program references @samp{etext} but does not define it, the
3388
linker will use the definition in the linker script.
3389
 
3390
@node PROVIDE_HIDDEN
3391
@subsection PROVIDE_HIDDEN
3392
@cindex PROVIDE_HIDDEN
3393
Similar to @code{PROVIDE}.  For ELF targeted ports, the symbol will be
3394
hidden and won't be exported.
3395
 
3396
@node Source Code Reference
3397
@subsection Source Code Reference
3398
 
3399
Accessing a linker script defined variable from source code is not
3400
intuitive.  In particular a linker script symbol is not equivalent to
3401
a variable declaration in a high level language, it is instead a
3402
symbol that does not have a value.
3403
 
3404
Before going further, it is important to note that compilers often
3405
transform names in the source code into different names when they are
3406
stored in the symbol table.  For example, Fortran compilers commonly
3407
prepend or append an underscore, and C++ performs extensive @samp{name
3408
mangling}.  Therefore there might be a discrepancy between the name
3409
of a variable as it is used in source code and the name of the same
3410
variable as it is defined in a linker script.  For example in C a
3411
linker script variable might be referred to as:
3412
 
3413
@smallexample
3414
  extern int foo;
3415
@end smallexample
3416
 
3417
But in the linker script it might be defined as:
3418
 
3419
@smallexample
3420
  _foo = 1000;
3421
@end smallexample
3422
 
3423
In the remaining examples however it is assumed that no name
3424
transformation has taken place.
3425
 
3426
When a symbol is declared in a high level language such as C, two
3427
things happen.  The first is that the compiler reserves enough space
3428
in the program's memory to hold the @emph{value} of the symbol.  The
3429
second is that the compiler creates an entry in the program's symbol
3430
table which holds the symbol's @emph{address}.  ie the symbol table
3431
contains the address of the block of memory holding the symbol's
3432
value.  So for example the following C declaration, at file scope:
3433
 
3434
@smallexample
3435
  int foo = 1000;
3436
@end smallexample
3437
 
3438
creates a entry called @samp{foo} in the symbol table.  This entry
3439
holds the address of an @samp{int} sized block of memory where the
3440
number 1000 is initially stored.
3441
 
3442
When a program references a symbol the compiler generates code that
3443
first accesses the symbol table to find the address of the symbol's
3444
memory block and then code to read the value from that memory block.
3445
So:
3446
 
3447
@smallexample
3448
  foo = 1;
3449
@end smallexample
3450
 
3451
looks up the symbol @samp{foo} in the symbol table, gets the address
3452
associated with this symbol and then writes the value 1 into that
3453
address.  Whereas:
3454
 
3455
@smallexample
3456
  int * a = & foo;
3457
@end smallexample
3458
 
3459
looks up the symbol @samp{foo} in the symbol table, gets it address
3460
and then copies this address into the block of memory associated with
3461
the variable @samp{a}.
3462
 
3463
Linker scripts symbol declarations, by contrast, create an entry in
3464
the symbol table but do not assign any memory to them.  Thus they are
3465
an address without a value.  So for example the linker script definition:
3466
 
3467
@smallexample
3468
  foo = 1000;
3469
@end smallexample
3470
 
3471
creates an entry in the symbol table called @samp{foo} which holds
3472
the address of memory location 1000, but nothing special is stored at
3473
address 1000.  This means that you cannot access the @emph{value} of a
3474
linker script defined symbol - it has no value - all you can do is
3475
access the @emph{address} of a linker script defined symbol.
3476
 
3477
Hence when you are using a linker script defined symbol in source code
3478
you should always take the address of the symbol, and never attempt to
3479
use its value.  For example suppose you want to copy the contents of a
3480
section of memory called .ROM into a section called .FLASH and the
3481
linker script contains these declarations:
3482
 
3483
@smallexample
3484
@group
3485
  start_of_ROM   = .ROM;
3486
  end_of_ROM     = .ROM + sizeof (.ROM) - 1;
3487
  start_of_FLASH = .FLASH;
3488
@end group
3489
@end smallexample
3490
 
3491
Then the C source code to perform the copy would be:
3492
 
3493
@smallexample
3494
@group
3495
  extern char start_of_ROM, end_of_ROM, start_of_FLASH;
3496
 
3497
  memcpy (& start_of_FLASH, & start_of_ROM, & end_of_ROM - & start_of_ROM);
3498
@end group
3499
@end smallexample
3500
 
3501
Note the use of the @samp{&} operators.  These are correct.
3502
 
3503
@node SECTIONS
3504
@section SECTIONS Command
3505
@kindex SECTIONS
3506
The @code{SECTIONS} command tells the linker how to map input sections
3507
into output sections, and how to place the output sections in memory.
3508
 
3509
The format of the @code{SECTIONS} command is:
3510
@smallexample
3511
SECTIONS
3512
@{
3513
  @var{sections-command}
3514
  @var{sections-command}
3515
  @dots{}
3516
@}
3517
@end smallexample
3518
 
3519
Each @var{sections-command} may of be one of the following:
3520
 
3521
@itemize @bullet
3522
@item
3523
an @code{ENTRY} command (@pxref{Entry Point,,Entry command})
3524
@item
3525
a symbol assignment (@pxref{Assignments})
3526
@item
3527
an output section description
3528
@item
3529
an overlay description
3530
@end itemize
3531
 
3532
The @code{ENTRY} command and symbol assignments are permitted inside the
3533
@code{SECTIONS} command for convenience in using the location counter in
3534
those commands.  This can also make the linker script easier to
3535
understand because you can use those commands at meaningful points in
3536
the layout of the output file.
3537
 
3538
Output section descriptions and overlay descriptions are described
3539
below.
3540
 
3541
If you do not use a @code{SECTIONS} command in your linker script, the
3542
linker will place each input section into an identically named output
3543
section in the order that the sections are first encountered in the
3544
input files.  If all input sections are present in the first file, for
3545
example, the order of sections in the output file will match the order
3546
in the first input file.  The first section will be at address zero.
3547
 
3548
@menu
3549
* Output Section Description::  Output section description
3550
* Output Section Name::         Output section name
3551
* Output Section Address::      Output section address
3552
* Input Section::               Input section description
3553
* Output Section Data::         Output section data
3554
* Output Section Keywords::     Output section keywords
3555
* Output Section Discarding::   Output section discarding
3556
* Output Section Attributes::   Output section attributes
3557
* Overlay Description::         Overlay description
3558
@end menu
3559
 
3560
@node Output Section Description
3561
@subsection Output Section Description
3562
The full description of an output section looks like this:
3563
@smallexample
3564
@group
3565
@var{section} [@var{address}] [(@var{type})] :
3566
  [AT(@var{lma})]
3567
  [ALIGN(@var{section_align})]
3568
  [SUBALIGN(@var{subsection_align})]
3569
  [@var{constraint}]
3570
  @{
3571
    @var{output-section-command}
3572
    @var{output-section-command}
3573
    @dots{}
3574
  @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
3575
@end group
3576
@end smallexample
3577
 
3578
Most output sections do not use most of the optional section attributes.
3579
 
3580
The whitespace around @var{section} is required, so that the section
3581
name is unambiguous.  The colon and the curly braces are also required.
3582
The line breaks and other white space are optional.
3583
 
3584
Each @var{output-section-command} may be one of the following:
3585
 
3586
@itemize @bullet
3587
@item
3588
a symbol assignment (@pxref{Assignments})
3589
@item
3590
an input section description (@pxref{Input Section})
3591
@item
3592
data values to include directly (@pxref{Output Section Data})
3593
@item
3594
a special output section keyword (@pxref{Output Section Keywords})
3595
@end itemize
3596
 
3597
@node Output Section Name
3598
@subsection Output Section Name
3599
@cindex name, section
3600
@cindex section name
3601
The name of the output section is @var{section}.  @var{section} must
3602
meet the constraints of your output format.  In formats which only
3603
support a limited number of sections, such as @code{a.out}, the name
3604
must be one of the names supported by the format (@code{a.out}, for
3605
example, allows only @samp{.text}, @samp{.data} or @samp{.bss}). If the
3606
output format supports any number of sections, but with numbers and not
3607
names (as is the case for Oasys), the name should be supplied as a
3608
quoted numeric string.  A section name may consist of any sequence of
3609
characters, but a name which contains any unusual characters such as
3610
commas must be quoted.
3611
 
3612
The output section name @samp{/DISCARD/} is special; @ref{Output Section
3613
Discarding}.
3614
 
3615
@node Output Section Address
3616
@subsection Output Section Address
3617
@cindex address, section
3618
@cindex section address
3619
The @var{address} is an expression for the VMA (the virtual memory
3620
address) of the output section.  If you do not provide @var{address},
3621
the linker will set it based on @var{region} if present, or otherwise
3622
based on the current value of the location counter.
3623
 
3624
If you provide @var{address}, the address of the output section will be
3625
set to precisely that.  If you provide neither @var{address} nor
3626
@var{region}, then the address of the output section will be set to the
3627
current value of the location counter aligned to the alignment
3628
requirements of the output section.  The alignment requirement of the
3629
output section is the strictest alignment of any input section contained
3630
within the output section.
3631
 
3632
For example,
3633
@smallexample
3634
.text . : @{ *(.text) @}
3635
@end smallexample
3636
@noindent
3637
and
3638
@smallexample
3639
.text : @{ *(.text) @}
3640
@end smallexample
3641
@noindent
3642
are subtly different.  The first will set the address of the
3643
@samp{.text} output section to the current value of the location
3644
counter.  The second will set it to the current value of the location
3645
counter aligned to the strictest alignment of a @samp{.text} input
3646
section.
3647
 
3648
The @var{address} may be an arbitrary expression; @ref{Expressions}.
3649
For example, if you want to align the section on a 0x10 byte boundary,
3650
so that the lowest four bits of the section address are zero, you could
3651
do something like this:
3652
@smallexample
3653
.text ALIGN(0x10) : @{ *(.text) @}
3654
@end smallexample
3655
@noindent
3656
This works because @code{ALIGN} returns the current location counter
3657
aligned upward to the specified value.
3658
 
3659
Specifying @var{address} for a section will change the value of the
3660
location counter, provided that the section is non-empty.  (Empty
3661
sections are ignored).
3662
 
3663
@node Input Section
3664
@subsection Input Section Description
3665
@cindex input sections
3666
@cindex mapping input sections to output sections
3667
The most common output section command is an input section description.
3668
 
3669
The input section description is the most basic linker script operation.
3670
You use output sections to tell the linker how to lay out your program
3671
in memory.  You use input section descriptions to tell the linker how to
3672
map the input files into your memory layout.
3673
 
3674
@menu
3675
* Input Section Basics::        Input section basics
3676
* Input Section Wildcards::     Input section wildcard patterns
3677
* Input Section Common::        Input section for common symbols
3678
* Input Section Keep::          Input section and garbage collection
3679
* Input Section Example::       Input section example
3680
@end menu
3681
 
3682
@node Input Section Basics
3683
@subsubsection Input Section Basics
3684
@cindex input section basics
3685
An input section description consists of a file name optionally followed
3686
by a list of section names in parentheses.
3687
 
3688
The file name and the section name may be wildcard patterns, which we
3689
describe further below (@pxref{Input Section Wildcards}).
3690
 
3691
The most common input section description is to include all input
3692
sections with a particular name in the output section.  For example, to
3693
include all input @samp{.text} sections, you would write:
3694
@smallexample
3695
*(.text)
3696
@end smallexample
3697
@noindent
3698
Here the @samp{*} is a wildcard which matches any file name.  To exclude a list
3699
of files from matching the file name wildcard, EXCLUDE_FILE may be used to
3700
match all files except the ones specified in the EXCLUDE_FILE list.  For
3701
example:
3702
@smallexample
3703
*(EXCLUDE_FILE (*crtend.o *otherfile.o) .ctors)
3704
@end smallexample
3705
will cause all .ctors sections from all files except @file{crtend.o} and
3706
@file{otherfile.o} to be included.
3707
 
3708
There are two ways to include more than one section:
3709
@smallexample
3710
*(.text .rdata)
3711
*(.text) *(.rdata)
3712
@end smallexample
3713
@noindent
3714
The difference between these is the order in which the @samp{.text} and
3715
@samp{.rdata} input sections will appear in the output section.  In the
3716
first example, they will be intermingled, appearing in the same order as
3717
they are found in the linker input.  In the second example, all
3718
@samp{.text} input sections will appear first, followed by all
3719
@samp{.rdata} input sections.
3720
 
3721
You can specify a file name to include sections from a particular file.
3722
You would do this if one or more of your files contain special data that
3723
needs to be at a particular location in memory.  For example:
3724
@smallexample
3725
data.o(.data)
3726
@end smallexample
3727
 
3728
You can also specify files within archives by writing a pattern
3729
matching the archive, a colon, then the pattern matching the file,
3730
with no whitespace around the colon.
3731
 
3732
@table @samp
3733
@item archive:file
3734
matches file within archive
3735
@item archive:
3736
matches the whole archive
3737
@item :file
3738
matches file but not one in an archive
3739
@end table
3740
 
3741
Either one or both of @samp{archive} and @samp{file} can contain shell
3742
wildcards.  On DOS based file systems, the linker will assume that a
3743
single letter followed by a colon is a drive specifier, so
3744
@samp{c:myfile.o} is a simple file specification, not @samp{myfile.o}
3745
within an archive called @samp{c}.  @samp{archive:file} filespecs may
3746
also be used within an @code{EXCLUDE_FILE} list, but may not appear in
3747
other linker script contexts.  For instance, you cannot extract a file
3748
from an archive by using @samp{archive:file} in an @code{INPUT}
3749
command.
3750
 
3751
If you use a file name without a list of sections, then all sections in
3752
the input file will be included in the output section.  This is not
3753
commonly done, but it may by useful on occasion.  For example:
3754
@smallexample
3755
data.o
3756
@end smallexample
3757
 
3758
When you use a file name which is not an @samp{archive:file} specifier
3759
and does not contain any wild card
3760
characters, the linker will first see if you also specified the file
3761
name on the linker command line or in an @code{INPUT} command.  If you
3762
did not, the linker will attempt to open the file as an input file, as
3763
though it appeared on the command line.  Note that this differs from an
3764
@code{INPUT} command, because the linker will not search for the file in
3765
the archive search path.
3766
 
3767
@node Input Section Wildcards
3768
@subsubsection Input Section Wildcard Patterns
3769
@cindex input section wildcards
3770
@cindex wildcard file name patterns
3771
@cindex file name wildcard patterns
3772
@cindex section name wildcard patterns
3773
In an input section description, either the file name or the section
3774
name or both may be wildcard patterns.
3775
 
3776
The file name of @samp{*} seen in many examples is a simple wildcard
3777
pattern for the file name.
3778
 
3779
The wildcard patterns are like those used by the Unix shell.
3780
 
3781
@table @samp
3782
@item *
3783
matches any number of characters
3784
@item ?
3785
matches any single character
3786
@item [@var{chars}]
3787
matches a single instance of any of the @var{chars}; the @samp{-}
3788
character may be used to specify a range of characters, as in
3789
@samp{[a-z]} to match any lower case letter
3790
@item \
3791
quotes the following character
3792
@end table
3793
 
3794
When a file name is matched with a wildcard, the wildcard characters
3795
will not match a @samp{/} character (used to separate directory names on
3796
Unix).  A pattern consisting of a single @samp{*} character is an
3797
exception; it will always match any file name, whether it contains a
3798
@samp{/} or not.  In a section name, the wildcard characters will match
3799
a @samp{/} character.
3800
 
3801
File name wildcard patterns only match files which are explicitly
3802
specified on the command line or in an @code{INPUT} command.  The linker
3803
does not search directories to expand wildcards.
3804
 
3805
If a file name matches more than one wildcard pattern, or if a file name
3806
appears explicitly and is also matched by a wildcard pattern, the linker
3807
will use the first match in the linker script.  For example, this
3808
sequence of input section descriptions is probably in error, because the
3809
@file{data.o} rule will not be used:
3810
@smallexample
3811
.data : @{ *(.data) @}
3812
.data1 : @{ data.o(.data) @}
3813
@end smallexample
3814
 
3815
@cindex SORT_BY_NAME
3816
Normally, the linker will place files and sections matched by wildcards
3817
in the order in which they are seen during the link.  You can change
3818
this by using the @code{SORT_BY_NAME} keyword, which appears before a wildcard
3819
pattern in parentheses (e.g., @code{SORT_BY_NAME(.text*)}).  When the
3820
@code{SORT_BY_NAME} keyword is used, the linker will sort the files or sections
3821
into ascending order by name before placing them in the output file.
3822
 
3823
@cindex SORT_BY_ALIGNMENT
3824
@code{SORT_BY_ALIGNMENT} is very similar to @code{SORT_BY_NAME}. The
3825
difference is @code{SORT_BY_ALIGNMENT} will sort sections into
3826
ascending order by alignment before placing them in the output file.
3827
 
3828
@cindex SORT
3829
@code{SORT} is an alias for @code{SORT_BY_NAME}.
3830
 
3831
When there are nested section sorting commands in linker script, there
3832
can be at most 1 level of nesting for section sorting commands.
3833
 
3834
@enumerate
3835
@item
3836
@code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3837
It will sort the input sections by name first, then by alignment if 2
3838
sections have the same name.
3839
@item
3840
@code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3841
It will sort the input sections by alignment first, then by name if 2
3842
sections have the same alignment.
3843
@item
3844
@code{SORT_BY_NAME} (@code{SORT_BY_NAME} (wildcard section pattern)) is
3845
treated the same as @code{SORT_BY_NAME} (wildcard section pattern).
3846
@item
3847
@code{SORT_BY_ALIGNMENT} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern))
3848
is treated the same as @code{SORT_BY_ALIGNMENT} (wildcard section pattern).
3849
@item
3850
All other nested section sorting commands are invalid.
3851
@end enumerate
3852
 
3853
When both command line section sorting option and linker script
3854
section sorting command are used, section sorting command always
3855
takes precedence over the command line option.
3856
 
3857
If the section sorting command in linker script isn't nested, the
3858
command line option will make the section sorting command to be
3859
treated as nested sorting command.
3860
 
3861
@enumerate
3862
@item
3863
@code{SORT_BY_NAME} (wildcard section pattern ) with
3864
@option{--sort-sections alignment} is equivalent to
3865
@code{SORT_BY_NAME} (@code{SORT_BY_ALIGNMENT} (wildcard section pattern)).
3866
@item
3867
@code{SORT_BY_ALIGNMENT} (wildcard section pattern) with
3868
@option{--sort-section name} is equivalent to
3869
@code{SORT_BY_ALIGNMENT} (@code{SORT_BY_NAME} (wildcard section pattern)).
3870
@end enumerate
3871
 
3872
If the section sorting command in linker script is nested, the
3873
command line option will be ignored.
3874
 
3875
If you ever get confused about where input sections are going, use the
3876
@samp{-M} linker option to generate a map file.  The map file shows
3877
precisely how input sections are mapped to output sections.
3878
 
3879
This example shows how wildcard patterns might be used to partition
3880
files.  This linker script directs the linker to place all @samp{.text}
3881
sections in @samp{.text} and all @samp{.bss} sections in @samp{.bss}.
3882
The linker will place the @samp{.data} section from all files beginning
3883
with an upper case character in @samp{.DATA}; for all other files, the
3884
linker will place the @samp{.data} section in @samp{.data}.
3885
@smallexample
3886
@group
3887
SECTIONS @{
3888
  .text : @{ *(.text) @}
3889
  .DATA : @{ [A-Z]*(.data) @}
3890
  .data : @{ *(.data) @}
3891
  .bss : @{ *(.bss) @}
3892
@}
3893
@end group
3894
@end smallexample
3895
 
3896
@node Input Section Common
3897
@subsubsection Input Section for Common Symbols
3898
@cindex common symbol placement
3899
@cindex uninitialized data placement
3900
A special notation is needed for common symbols, because in many object
3901
file formats common symbols do not have a particular input section.  The
3902
linker treats common symbols as though they are in an input section
3903
named @samp{COMMON}.
3904
 
3905
You may use file names with the @samp{COMMON} section just as with any
3906
other input sections.  You can use this to place common symbols from a
3907
particular input file in one section while common symbols from other
3908
input files are placed in another section.
3909
 
3910
In most cases, common symbols in input files will be placed in the
3911
@samp{.bss} section in the output file.  For example:
3912
@smallexample
3913
.bss @{ *(.bss) *(COMMON) @}
3914
@end smallexample
3915
 
3916
@cindex scommon section
3917
@cindex small common symbols
3918
Some object file formats have more than one type of common symbol.  For
3919
example, the MIPS ELF object file format distinguishes standard common
3920
symbols and small common symbols.  In this case, the linker will use a
3921
different special section name for other types of common symbols.  In
3922
the case of MIPS ELF, the linker uses @samp{COMMON} for standard common
3923
symbols and @samp{.scommon} for small common symbols.  This permits you
3924
to map the different types of common symbols into memory at different
3925
locations.
3926
 
3927
@cindex [COMMON]
3928
You will sometimes see @samp{[COMMON]} in old linker scripts.  This
3929
notation is now considered obsolete.  It is equivalent to
3930
@samp{*(COMMON)}.
3931
 
3932
@node Input Section Keep
3933
@subsubsection Input Section and Garbage Collection
3934
@cindex KEEP
3935
@cindex garbage collection
3936
When link-time garbage collection is in use (@samp{--gc-sections}),
3937
it is often useful to mark sections that should not be eliminated.
3938
This is accomplished by surrounding an input section's wildcard entry
3939
with @code{KEEP()}, as in @code{KEEP(*(.init))} or
3940
@code{KEEP(SORT_BY_NAME(*)(.ctors))}.
3941
 
3942
@node Input Section Example
3943
@subsubsection Input Section Example
3944
The following example is a complete linker script.  It tells the linker
3945
to read all of the sections from file @file{all.o} and place them at the
3946
start of output section @samp{outputa} which starts at location
3947
@samp{0x10000}.  All of section @samp{.input1} from file @file{foo.o}
3948
follows immediately, in the same output section.  All of section
3949
@samp{.input2} from @file{foo.o} goes into output section
3950
@samp{outputb}, followed by section @samp{.input1} from @file{foo1.o}.
3951
All of the remaining @samp{.input1} and @samp{.input2} sections from any
3952
files are written to output section @samp{outputc}.
3953
 
3954
@smallexample
3955
@group
3956
SECTIONS @{
3957
  outputa 0x10000 :
3958
    @{
3959
    all.o
3960
    foo.o (.input1)
3961
    @}
3962
@end group
3963
@group
3964
  outputb :
3965
    @{
3966
    foo.o (.input2)
3967
    foo1.o (.input1)
3968
    @}
3969
@end group
3970
@group
3971
  outputc :
3972
    @{
3973
    *(.input1)
3974
    *(.input2)
3975
    @}
3976
@}
3977
@end group
3978
@end smallexample
3979
 
3980
@node Output Section Data
3981
@subsection Output Section Data
3982
@cindex data
3983
@cindex section data
3984
@cindex output section data
3985
@kindex BYTE(@var{expression})
3986
@kindex SHORT(@var{expression})
3987
@kindex LONG(@var{expression})
3988
@kindex QUAD(@var{expression})
3989
@kindex SQUAD(@var{expression})
3990
You can include explicit bytes of data in an output section by using
3991
@code{BYTE}, @code{SHORT}, @code{LONG}, @code{QUAD}, or @code{SQUAD} as
3992
an output section command.  Each keyword is followed by an expression in
3993
parentheses providing the value to store (@pxref{Expressions}).  The
3994
value of the expression is stored at the current value of the location
3995
counter.
3996
 
3997
The @code{BYTE}, @code{SHORT}, @code{LONG}, and @code{QUAD} commands
3998
store one, two, four, and eight bytes (respectively).  After storing the
3999
bytes, the location counter is incremented by the number of bytes
4000
stored.
4001
 
4002
For example, this will store the byte 1 followed by the four byte value
4003
of the symbol @samp{addr}:
4004
@smallexample
4005
BYTE(1)
4006
LONG(addr)
4007
@end smallexample
4008
 
4009
When using a 64 bit host or target, @code{QUAD} and @code{SQUAD} are the
4010
same; they both store an 8 byte, or 64 bit, value.  When both host and
4011
target are 32 bits, an expression is computed as 32 bits.  In this case
4012
@code{QUAD} stores a 32 bit value zero extended to 64 bits, and
4013
@code{SQUAD} stores a 32 bit value sign extended to 64 bits.
4014
 
4015
If the object file format of the output file has an explicit endianness,
4016
which is the normal case, the value will be stored in that endianness.
4017
When the object file format does not have an explicit endianness, as is
4018
true of, for example, S-records, the value will be stored in the
4019
endianness of the first input object file.
4020
 
4021
Note---these commands only work inside a section description and not
4022
between them, so the following will produce an error from the linker:
4023
@smallexample
4024
SECTIONS @{@ .text : @{@ *(.text) @}@ LONG(1) .data : @{@ *(.data) @}@ @}@
4025
@end smallexample
4026
whereas this will work:
4027
@smallexample
4028
SECTIONS @{@ .text : @{@ *(.text) ; LONG(1) @}@ .data : @{@ *(.data) @}@ @}@
4029
@end smallexample
4030
 
4031
@kindex FILL(@var{expression})
4032
@cindex holes, filling
4033
@cindex unspecified memory
4034
You may use the @code{FILL} command to set the fill pattern for the
4035
current section.  It is followed by an expression in parentheses.  Any
4036
otherwise unspecified regions of memory within the section (for example,
4037
gaps left due to the required alignment of input sections) are filled
4038
with the value of the expression, repeated as
4039
necessary.  A @code{FILL} statement covers memory locations after the
4040
point at which it occurs in the section definition; by including more
4041
than one @code{FILL} statement, you can have different fill patterns in
4042
different parts of an output section.
4043
 
4044
This example shows how to fill unspecified regions of memory with the
4045
value @samp{0x90}:
4046
@smallexample
4047
FILL(0x90909090)
4048
@end smallexample
4049
 
4050
The @code{FILL} command is similar to the @samp{=@var{fillexp}} output
4051
section attribute, but it only affects the
4052
part of the section following the @code{FILL} command, rather than the
4053
entire section.  If both are used, the @code{FILL} command takes
4054
precedence.  @xref{Output Section Fill}, for details on the fill
4055
expression.
4056
 
4057
@node Output Section Keywords
4058
@subsection Output Section Keywords
4059
There are a couple of keywords which can appear as output section
4060
commands.
4061
 
4062
@table @code
4063
@kindex CREATE_OBJECT_SYMBOLS
4064
@cindex input filename symbols
4065
@cindex filename symbols
4066
@item CREATE_OBJECT_SYMBOLS
4067
The command tells the linker to create a symbol for each input file.
4068
The name of each symbol will be the name of the corresponding input
4069
file.  The section of each symbol will be the output section in which
4070
the @code{CREATE_OBJECT_SYMBOLS} command appears.
4071
 
4072
This is conventional for the a.out object file format.  It is not
4073
normally used for any other object file format.
4074
 
4075
@kindex CONSTRUCTORS
4076
@cindex C++ constructors, arranging in link
4077
@cindex constructors, arranging in link
4078
@item CONSTRUCTORS
4079
When linking using the a.out object file format, the linker uses an
4080
unusual set construct to support C++ global constructors and
4081
destructors.  When linking object file formats which do not support
4082
arbitrary sections, such as ECOFF and XCOFF, the linker will
4083
automatically recognize C++ global constructors and destructors by name.
4084
For these object file formats, the @code{CONSTRUCTORS} command tells the
4085
linker to place constructor information in the output section where the
4086
@code{CONSTRUCTORS} command appears.  The @code{CONSTRUCTORS} command is
4087
ignored for other object file formats.
4088
 
4089
The symbol @w{@code{__CTOR_LIST__}} marks the start of the global
4090
constructors, and the symbol @w{@code{__CTOR_END__}} marks the end.
4091
Similarly, @w{@code{__DTOR_LIST__}} and @w{@code{__DTOR_END__}} mark
4092
the start and end of the global destructors.  The
4093
first word in the list is the number of entries, followed by the address
4094
of each constructor or destructor, followed by a zero word.  The
4095
compiler must arrange to actually run the code.  For these object file
4096
formats @sc{gnu} C++ normally calls constructors from a subroutine
4097
@code{__main}; a call to @code{__main} is automatically inserted into
4098
the startup code for @code{main}.  @sc{gnu} C++ normally runs
4099
destructors either by using @code{atexit}, or directly from the function
4100
@code{exit}.
4101
 
4102
For object file formats such as @code{COFF} or @code{ELF} which support
4103
arbitrary section names, @sc{gnu} C++ will normally arrange to put the
4104
addresses of global constructors and destructors into the @code{.ctors}
4105
and @code{.dtors} sections.  Placing the following sequence into your
4106
linker script will build the sort of table which the @sc{gnu} C++
4107
runtime code expects to see.
4108
 
4109
@smallexample
4110
      __CTOR_LIST__ = .;
4111
      LONG((__CTOR_END__ - __CTOR_LIST__) / 4 - 2)
4112
      *(.ctors)
4113
      LONG(0)
4114
      __CTOR_END__ = .;
4115
      __DTOR_LIST__ = .;
4116
      LONG((__DTOR_END__ - __DTOR_LIST__) / 4 - 2)
4117
      *(.dtors)
4118
      LONG(0)
4119
      __DTOR_END__ = .;
4120
@end smallexample
4121
 
4122
If you are using the @sc{gnu} C++ support for initialization priority,
4123
which provides some control over the order in which global constructors
4124
are run, you must sort the constructors at link time to ensure that they
4125
are executed in the correct order.  When using the @code{CONSTRUCTORS}
4126
command, use @samp{SORT_BY_NAME(CONSTRUCTORS)} instead.  When using the
4127
@code{.ctors} and @code{.dtors} sections, use @samp{*(SORT_BY_NAME(.ctors))} and
4128
@samp{*(SORT_BY_NAME(.dtors))} instead of just @samp{*(.ctors)} and
4129
@samp{*(.dtors)}.
4130
 
4131
Normally the compiler and linker will handle these issues automatically,
4132
and you will not need to concern yourself with them.  However, you may
4133
need to consider this if you are using C++ and writing your own linker
4134
scripts.
4135
 
4136
@end table
4137
 
4138
@node Output Section Discarding
4139
@subsection Output Section Discarding
4140
@cindex discarding sections
4141
@cindex sections, discarding
4142
@cindex removing sections
4143
The linker will not create output sections with no contents.  This is
4144
for convenience when referring to input sections that may or may not
4145
be present in any of the input files.  For example:
4146
@smallexample
4147
.foo : @{ *(.foo) @}
4148
@end smallexample
4149
@noindent
4150
will only create a @samp{.foo} section in the output file if there is a
4151
@samp{.foo} section in at least one input file, and if the input
4152
sections are not all empty.  Other link script directives that allocate
4153
space in an output section will also create the output section.
4154
 
4155
The linker will ignore address assignments (@pxref{Output Section Address})
4156
on discarded output sections, except when the linker script defines
4157
symbols in the output section.  In that case the linker will obey
4158
the address assignments, possibly advancing dot even though the
4159
section is discarded.
4160
 
4161
@cindex /DISCARD/
4162
The special output section name @samp{/DISCARD/} may be used to discard
4163
input sections.  Any input sections which are assigned to an output
4164
section named @samp{/DISCARD/} are not included in the output file.
4165
 
4166
@node Output Section Attributes
4167
@subsection Output Section Attributes
4168
@cindex output section attributes
4169
We showed above that the full description of an output section looked
4170
like this:
4171
 
4172
@smallexample
4173
@group
4174
@var{section} [@var{address}] [(@var{type})] :
4175
  [AT(@var{lma})]
4176
  [ALIGN(@var{section_align})]
4177
  [SUBALIGN(@var{subsection_align})]
4178
  [@var{constraint}]
4179
  @{
4180
    @var{output-section-command}
4181
    @var{output-section-command}
4182
    @dots{}
4183
  @} [>@var{region}] [AT>@var{lma_region}] [:@var{phdr} :@var{phdr} @dots{}] [=@var{fillexp}]
4184
@end group
4185
@end smallexample
4186
 
4187
We've already described @var{section}, @var{address}, and
4188
@var{output-section-command}.  In this section we will describe the
4189
remaining section attributes.
4190
 
4191
@menu
4192
* Output Section Type::         Output section type
4193
* Output Section LMA::          Output section LMA
4194
* Forced Output Alignment::     Forced Output Alignment
4195
* Forced Input Alignment::      Forced Input Alignment
4196
* Output Section Constraint::   Output section constraint
4197
* Output Section Region::       Output section region
4198
* Output Section Phdr::         Output section phdr
4199
* Output Section Fill::         Output section fill
4200
@end menu
4201
 
4202
@node Output Section Type
4203
@subsubsection Output Section Type
4204
Each output section may have a type.  The type is a keyword in
4205
parentheses.  The following types are defined:
4206
 
4207
@table @code
4208
@item NOLOAD
4209
The section should be marked as not loadable, so that it will not be
4210
loaded into memory when the program is run.
4211
@item DSECT
4212
@itemx COPY
4213
@itemx INFO
4214
@itemx OVERLAY
4215
These type names are supported for backward compatibility, and are
4216
rarely used.  They all have the same effect: the section should be
4217
marked as not allocatable, so that no memory is allocated for the
4218
section when the program is run.
4219
@end table
4220
 
4221
@kindex NOLOAD
4222
@cindex prevent unnecessary loading
4223
@cindex loading, preventing
4224
The linker normally sets the attributes of an output section based on
4225
the input sections which map into it.  You can override this by using
4226
the section type.  For example, in the script sample below, the
4227
@samp{ROM} section is addressed at memory location @samp{0} and does not
4228
need to be loaded when the program is run.  The contents of the
4229
@samp{ROM} section will appear in the linker output file as usual.
4230
@smallexample
4231
@group
4232
SECTIONS @{
4233
  ROM 0 (NOLOAD) : @{ @dots{} @}
4234
  @dots{}
4235
@}
4236
@end group
4237
@end smallexample
4238
 
4239
@node Output Section LMA
4240
@subsubsection Output Section LMA
4241
@kindex AT>@var{lma_region}
4242
@kindex AT(@var{lma})
4243
@cindex load address
4244
@cindex section load address
4245
Every section has a virtual address (VMA) and a load address (LMA); see
4246
@ref{Basic Script Concepts}.  The address expression which may appear in
4247
an output section description sets the VMA (@pxref{Output Section
4248
Address}).
4249
 
4250
The expression @var{lma} that follows the @code{AT} keyword specifies
4251
the load address of the section.
4252
 
4253
Alternatively, with @samp{AT>@var{lma_region}} expression, you may
4254
specify a memory region for the section's load address. @xref{MEMORY}.
4255
Note that if the section has not had a VMA assigned to it then the
4256
linker will use the @var{lma_region} as the VMA region as well.
4257
 
4258
If neither @code{AT} nor @code{AT>} is specified for an allocatable
4259
section, the linker will set the LMA such that the difference between
4260
VMA and LMA for the section is the same as the preceding output
4261
section in the same region.  If there is no preceding output section
4262
or the section is not allocatable, the linker will set the LMA equal
4263
to the VMA.
4264
@xref{Output Section Region}.
4265
 
4266
@cindex ROM initialized data
4267
@cindex initialized data in ROM
4268
This feature is designed to make it easy to build a ROM image.  For
4269
example, the following linker script creates three output sections: one
4270
called @samp{.text}, which starts at @code{0x1000}, one called
4271
@samp{.mdata}, which is loaded at the end of the @samp{.text} section
4272
even though its VMA is @code{0x2000}, and one called @samp{.bss} to hold
4273
uninitialized data at address @code{0x3000}.  The symbol @code{_data} is
4274
defined with the value @code{0x2000}, which shows that the location
4275
counter holds the VMA value, not the LMA value.
4276
 
4277
@smallexample
4278
@group
4279
SECTIONS
4280
  @{
4281
  .text 0x1000 : @{ *(.text) _etext = . ; @}
4282
  .mdata 0x2000 :
4283
    AT ( ADDR (.text) + SIZEOF (.text) )
4284
    @{ _data = . ; *(.data); _edata = . ;  @}
4285
  .bss 0x3000 :
4286
    @{ _bstart = . ;  *(.bss) *(COMMON) ; _bend = . ;@}
4287
@}
4288
@end group
4289
@end smallexample
4290
 
4291
The run-time initialization code for use with a program generated with
4292
this linker script would include something like the following, to copy
4293
the initialized data from the ROM image to its runtime address.  Notice
4294
how this code takes advantage of the symbols defined by the linker
4295
script.
4296
 
4297
@smallexample
4298
@group
4299
extern char _etext, _data, _edata, _bstart, _bend;
4300
char *src = &_etext;
4301
char *dst = &_data;
4302
 
4303
/* ROM has data at end of text; copy it. */
4304
while (dst < &_edata) @{
4305
  *dst++ = *src++;
4306
@}
4307
 
4308
/* Zero bss */
4309
for (dst = &_bstart; dst< &_bend; dst++)
4310
  *dst = 0;
4311
@end group
4312
@end smallexample
4313
 
4314
@node Forced Output Alignment
4315
@subsubsection Forced Output Alignment
4316
@kindex ALIGN(@var{section_align})
4317
@cindex forcing output section alignment
4318
@cindex output section alignment
4319
You can increase an output section's alignment by using ALIGN.
4320
 
4321
@node Forced Input Alignment
4322
@subsubsection Forced Input Alignment
4323
@kindex SUBALIGN(@var{subsection_align})
4324
@cindex forcing input section alignment
4325
@cindex input section alignment
4326
You can force input section alignment within an output section by using
4327
SUBALIGN.  The value specified overrides any alignment given by input
4328
sections, whether larger or smaller.
4329
 
4330
@node Output Section Constraint
4331
@subsubsection Output Section Constraint
4332
@kindex ONLY_IF_RO
4333
@kindex ONLY_IF_RW
4334
@cindex constraints on output sections
4335
You can specify that an output section should only be created if all
4336
of its input sections are read-only or all of its input sections are
4337
read-write by using the keyword @code{ONLY_IF_RO} and
4338
@code{ONLY_IF_RW} respectively.
4339
 
4340
@node Output Section Region
4341
@subsubsection Output Section Region
4342
@kindex >@var{region}
4343
@cindex section, assigning to memory region
4344
@cindex memory regions and sections
4345
You can assign a section to a previously defined region of memory by
4346
using @samp{>@var{region}}.  @xref{MEMORY}.
4347
 
4348
Here is a simple example:
4349
@smallexample
4350
@group
4351
MEMORY @{ rom : ORIGIN = 0x1000, LENGTH = 0x1000 @}
4352
SECTIONS @{ ROM : @{ *(.text) @} >rom @}
4353
@end group
4354
@end smallexample
4355
 
4356
@node Output Section Phdr
4357
@subsubsection Output Section Phdr
4358
@kindex :@var{phdr}
4359
@cindex section, assigning to program header
4360
@cindex program headers and sections
4361
You can assign a section to a previously defined program segment by
4362
using @samp{:@var{phdr}}.  @xref{PHDRS}.  If a section is assigned to
4363
one or more segments, then all subsequent allocated sections will be
4364
assigned to those segments as well, unless they use an explicitly
4365
@code{:@var{phdr}} modifier.  You can use @code{:NONE} to tell the
4366
linker to not put the section in any segment at all.
4367
 
4368
Here is a simple example:
4369
@smallexample
4370
@group
4371
PHDRS @{ text PT_LOAD ; @}
4372
SECTIONS @{ .text : @{ *(.text) @} :text @}
4373
@end group
4374
@end smallexample
4375
 
4376
@node Output Section Fill
4377
@subsubsection Output Section Fill
4378
@kindex =@var{fillexp}
4379
@cindex section fill pattern
4380
@cindex fill pattern, entire section
4381
You can set the fill pattern for an entire section by using
4382
@samp{=@var{fillexp}}.  @var{fillexp} is an expression
4383
(@pxref{Expressions}).  Any otherwise unspecified regions of memory
4384
within the output section (for example, gaps left due to the required
4385
alignment of input sections) will be filled with the value, repeated as
4386
necessary.  If the fill expression is a simple hex number, ie. a string
4387
of hex digit starting with @samp{0x} and without a trailing @samp{k} or @samp{M}, then
4388
an arbitrarily long sequence of hex digits can be used to specify the
4389
fill pattern;  Leading zeros become part of the pattern too.  For all
4390
other cases, including extra parentheses or a unary @code{+}, the fill
4391
pattern is the four least significant bytes of the value of the
4392
expression.  In all cases, the number is big-endian.
4393
 
4394
You can also change the fill value with a @code{FILL} command in the
4395
output section commands; (@pxref{Output Section Data}).
4396
 
4397
Here is a simple example:
4398
@smallexample
4399
@group
4400
SECTIONS @{ .text : @{ *(.text) @} =0x90909090 @}
4401
@end group
4402
@end smallexample
4403
 
4404
@node Overlay Description
4405
@subsection Overlay Description
4406
@kindex OVERLAY
4407
@cindex overlays
4408
An overlay description provides an easy way to describe sections which
4409
are to be loaded as part of a single memory image but are to be run at
4410
the same memory address.  At run time, some sort of overlay manager will
4411
copy the overlaid sections in and out of the runtime memory address as
4412
required, perhaps by simply manipulating addressing bits.  This approach
4413
can be useful, for example, when a certain region of memory is faster
4414
than another.
4415
 
4416
Overlays are described using the @code{OVERLAY} command.  The
4417
@code{OVERLAY} command is used within a @code{SECTIONS} command, like an
4418
output section description.  The full syntax of the @code{OVERLAY}
4419
command is as follows:
4420
@smallexample
4421
@group
4422
OVERLAY [@var{start}] : [NOCROSSREFS] [AT ( @var{ldaddr} )]
4423
  @{
4424
    @var{secname1}
4425
      @{
4426
        @var{output-section-command}
4427
        @var{output-section-command}
4428
        @dots{}
4429
      @} [:@var{phdr}@dots{}] [=@var{fill}]
4430
    @var{secname2}
4431
      @{
4432
        @var{output-section-command}
4433
        @var{output-section-command}
4434
        @dots{}
4435
      @} [:@var{phdr}@dots{}] [=@var{fill}]
4436
    @dots{}
4437
  @} [>@var{region}] [:@var{phdr}@dots{}] [=@var{fill}]
4438
@end group
4439
@end smallexample
4440
 
4441
Everything is optional except @code{OVERLAY} (a keyword), and each
4442
section must have a name (@var{secname1} and @var{secname2} above).  The
4443
section definitions within the @code{OVERLAY} construct are identical to
4444
those within the general @code{SECTIONS} contruct (@pxref{SECTIONS}),
4445
except that no addresses and no memory regions may be defined for
4446
sections within an @code{OVERLAY}.
4447
 
4448
The sections are all defined with the same starting address.  The load
4449
addresses of the sections are arranged such that they are consecutive in
4450
memory starting at the load address used for the @code{OVERLAY} as a
4451
whole (as with normal section definitions, the load address is optional,
4452
and defaults to the start address; the start address is also optional,
4453
and defaults to the current value of the location counter).
4454
 
4455
If the @code{NOCROSSREFS} keyword is used, and there any references
4456
among the sections, the linker will report an error.  Since the sections
4457
all run at the same address, it normally does not make sense for one
4458
section to refer directly to another.  @xref{Miscellaneous Commands,
4459
NOCROSSREFS}.
4460
 
4461
For each section within the @code{OVERLAY}, the linker automatically
4462
provides two symbols.  The symbol @code{__load_start_@var{secname}} is
4463
defined as the starting load address of the section.  The symbol
4464
@code{__load_stop_@var{secname}} is defined as the final load address of
4465
the section.  Any characters within @var{secname} which are not legal
4466
within C identifiers are removed.  C (or assembler) code may use these
4467
symbols to move the overlaid sections around as necessary.
4468
 
4469
At the end of the overlay, the value of the location counter is set to
4470
the start address of the overlay plus the size of the largest section.
4471
 
4472
Here is an example.  Remember that this would appear inside a
4473
@code{SECTIONS} construct.
4474
@smallexample
4475
@group
4476
  OVERLAY 0x1000 : AT (0x4000)
4477
   @{
4478
     .text0 @{ o1/*.o(.text) @}
4479
     .text1 @{ o2/*.o(.text) @}
4480
   @}
4481
@end group
4482
@end smallexample
4483
@noindent
4484
This will define both @samp{.text0} and @samp{.text1} to start at
4485
address 0x1000.  @samp{.text0} will be loaded at address 0x4000, and
4486
@samp{.text1} will be loaded immediately after @samp{.text0}.  The
4487
following symbols will be defined if referenced: @code{__load_start_text0},
4488
@code{__load_stop_text0}, @code{__load_start_text1},
4489
@code{__load_stop_text1}.
4490
 
4491
C code to copy overlay @code{.text1} into the overlay area might look
4492
like the following.
4493
 
4494
@smallexample
4495
@group
4496
  extern char __load_start_text1, __load_stop_text1;
4497
  memcpy ((char *) 0x1000, &__load_start_text1,
4498
          &__load_stop_text1 - &__load_start_text1);
4499
@end group
4500
@end smallexample
4501
 
4502
Note that the @code{OVERLAY} command is just syntactic sugar, since
4503
everything it does can be done using the more basic commands.  The above
4504
example could have been written identically as follows.
4505
 
4506
@smallexample
4507
@group
4508
  .text0 0x1000 : AT (0x4000) @{ o1/*.o(.text) @}
4509
  PROVIDE (__load_start_text0 = LOADADDR (.text0));
4510
  PROVIDE (__load_stop_text0 = LOADADDR (.text0) + SIZEOF (.text0));
4511
  .text1 0x1000 : AT (0x4000 + SIZEOF (.text0)) @{ o2/*.o(.text) @}
4512
  PROVIDE (__load_start_text1 = LOADADDR (.text1));
4513
  PROVIDE (__load_stop_text1 = LOADADDR (.text1) + SIZEOF (.text1));
4514
  . = 0x1000 + MAX (SIZEOF (.text0), SIZEOF (.text1));
4515
@end group
4516
@end smallexample
4517
 
4518
@node MEMORY
4519
@section MEMORY Command
4520
@kindex MEMORY
4521
@cindex memory regions
4522
@cindex regions of memory
4523
@cindex allocating memory
4524
@cindex discontinuous memory
4525
The linker's default configuration permits allocation of all available
4526
memory.  You can override this by using the @code{MEMORY} command.
4527
 
4528
The @code{MEMORY} command describes the location and size of blocks of
4529
memory in the target.  You can use it to describe which memory regions
4530
may be used by the linker, and which memory regions it must avoid.  You
4531
can then assign sections to particular memory regions.  The linker will
4532
set section addresses based on the memory regions, and will warn about
4533
regions that become too full.  The linker will not shuffle sections
4534
around to fit into the available regions.
4535
 
4536
A linker script may contain at most one use of the @code{MEMORY}
4537
command.  However, you can define as many blocks of memory within it as
4538
you wish.  The syntax is:
4539
@smallexample
4540
@group
4541
MEMORY
4542
  @{
4543
    @var{name} [(@var{attr})] : ORIGIN = @var{origin}, LENGTH = @var{len}
4544
    @dots{}
4545
  @}
4546
@end group
4547
@end smallexample
4548
 
4549
The @var{name} is a name used in the linker script to refer to the
4550
region.  The region name has no meaning outside of the linker script.
4551
Region names are stored in a separate name space, and will not conflict
4552
with symbol names, file names, or section names.  Each memory region
4553
must have a distinct name within the @code{MEMORY} command.  However you can
4554
add later alias names to existing memory regions with the @ref{REGION_ALIAS}
4555
command.
4556
 
4557
@cindex memory region attributes
4558
The @var{attr} string is an optional list of attributes that specify
4559
whether to use a particular memory region for an input section which is
4560
not explicitly mapped in the linker script.  As described in
4561
@ref{SECTIONS}, if you do not specify an output section for some input
4562
section, the linker will create an output section with the same name as
4563
the input section.  If you define region attributes, the linker will use
4564
them to select the memory region for the output section that it creates.
4565
 
4566
The @var{attr} string must consist only of the following characters:
4567
@table @samp
4568
@item R
4569
Read-only section
4570
@item W
4571
Read/write section
4572
@item X
4573
Executable section
4574
@item A
4575
Allocatable section
4576
@item I
4577
Initialized section
4578
@item L
4579
Same as @samp{I}
4580
@item !
4581
Invert the sense of any of the preceding attributes
4582
@end table
4583
 
4584
If a unmapped section matches any of the listed attributes other than
4585
@samp{!}, it will be placed in the memory region.  The @samp{!}
4586
attribute reverses this test, so that an unmapped section will be placed
4587
in the memory region only if it does not match any of the listed
4588
attributes.
4589
 
4590
@kindex ORIGIN =
4591
@kindex o =
4592
@kindex org =
4593
The @var{origin} is an numerical expression for the start address of
4594
the memory region.  The expression must evaluate to a constant and it
4595
cannot involve any symbols.  The keyword @code{ORIGIN} may be
4596
abbreviated to @code{org} or @code{o} (but not, for example,
4597
@code{ORG}).
4598
 
4599
@kindex LENGTH =
4600
@kindex len =
4601
@kindex l =
4602
The @var{len} is an expression for the size in bytes of the memory
4603
region.  As with the @var{origin} expression, the expression must
4604
be numerical only and must evaluate to a constant.  The keyword
4605
@code{LENGTH} may be abbreviated to @code{len} or @code{l}.
4606
 
4607
In the following example, we specify that there are two memory regions
4608
available for allocation: one starting at @samp{0} for 256 kilobytes,
4609
and the other starting at @samp{0x40000000} for four megabytes.  The
4610
linker will place into the @samp{rom} memory region every section which
4611
is not explicitly mapped into a memory region, and is either read-only
4612
or executable.  The linker will place other sections which are not
4613
explicitly mapped into a memory region into the @samp{ram} memory
4614
region.
4615
 
4616
@smallexample
4617
@group
4618
MEMORY
4619
  @{
4620
    rom (rx)  : ORIGIN = 0, LENGTH = 256K
4621
    ram (!rx) : org = 0x40000000, l = 4M
4622
  @}
4623
@end group
4624
@end smallexample
4625
 
4626
Once you define a memory region, you can direct the linker to place
4627
specific output sections into that memory region by using the
4628
@samp{>@var{region}} output section attribute.  For example, if you have
4629
a memory region named @samp{mem}, you would use @samp{>mem} in the
4630
output section definition.  @xref{Output Section Region}.  If no address
4631
was specified for the output section, the linker will set the address to
4632
the next available address within the memory region.  If the combined
4633
output sections directed to a memory region are too large for the
4634
region, the linker will issue an error message.
4635
 
4636
It is possible to access the origin and length of a memory in an
4637
expression via the @code{ORIGIN(@var{memory})} and
4638
@code{LENGTH(@var{memory})} functions:
4639
 
4640
@smallexample
4641
@group
4642
  _fstack = ORIGIN(ram) + LENGTH(ram) - 4;
4643
@end group
4644
@end smallexample
4645
 
4646
@node PHDRS
4647
@section PHDRS Command
4648
@kindex PHDRS
4649
@cindex program headers
4650
@cindex ELF program headers
4651
@cindex program segments
4652
@cindex segments, ELF
4653
The ELF object file format uses @dfn{program headers}, also knows as
4654
@dfn{segments}.  The program headers describe how the program should be
4655
loaded into memory.  You can print them out by using the @code{objdump}
4656
program with the @samp{-p} option.
4657
 
4658
When you run an ELF program on a native ELF system, the system loader
4659
reads the program headers in order to figure out how to load the
4660
program.  This will only work if the program headers are set correctly.
4661
This manual does not describe the details of how the system loader
4662
interprets program headers; for more information, see the ELF ABI.
4663
 
4664
The linker will create reasonable program headers by default.  However,
4665
in some cases, you may need to specify the program headers more
4666
precisely.  You may use the @code{PHDRS} command for this purpose.  When
4667
the linker sees the @code{PHDRS} command in the linker script, it will
4668
not create any program headers other than the ones specified.
4669
 
4670
The linker only pays attention to the @code{PHDRS} command when
4671
generating an ELF output file.  In other cases, the linker will simply
4672
ignore @code{PHDRS}.
4673
 
4674
This is the syntax of the @code{PHDRS} command.  The words @code{PHDRS},
4675
@code{FILEHDR}, @code{AT}, and @code{FLAGS} are keywords.
4676
 
4677
@smallexample
4678
@group
4679
PHDRS
4680
@{
4681
  @var{name} @var{type} [ FILEHDR ] [ PHDRS ] [ AT ( @var{address} ) ]
4682
        [ FLAGS ( @var{flags} ) ] ;
4683
@}
4684
@end group
4685
@end smallexample
4686
 
4687
The @var{name} is used only for reference in the @code{SECTIONS} command
4688
of the linker script.  It is not put into the output file.  Program
4689
header names are stored in a separate name space, and will not conflict
4690
with symbol names, file names, or section names.  Each program header
4691
must have a distinct name.
4692
 
4693
Certain program header types describe segments of memory which the
4694
system loader will load from the file.  In the linker script, you
4695
specify the contents of these segments by placing allocatable output
4696
sections in the segments.  You use the @samp{:@var{phdr}} output section
4697
attribute to place a section in a particular segment.  @xref{Output
4698
Section Phdr}.
4699
 
4700
It is normal to put certain sections in more than one segment.  This
4701
merely implies that one segment of memory contains another.  You may
4702
repeat @samp{:@var{phdr}}, using it once for each segment which should
4703
contain the section.
4704
 
4705
If you place a section in one or more segments using @samp{:@var{phdr}},
4706
then the linker will place all subsequent allocatable sections which do
4707
not specify @samp{:@var{phdr}} in the same segments.  This is for
4708
convenience, since generally a whole set of contiguous sections will be
4709
placed in a single segment.  You can use @code{:NONE} to override the
4710
default segment and tell the linker to not put the section in any
4711
segment at all.
4712
 
4713
@kindex FILEHDR
4714
@kindex PHDRS
4715
You may use the @code{FILEHDR} and @code{PHDRS} keywords appear after
4716
the program header type to further describe the contents of the segment.
4717
The @code{FILEHDR} keyword means that the segment should include the ELF
4718
file header.  The @code{PHDRS} keyword means that the segment should
4719
include the ELF program headers themselves.
4720
 
4721
The @var{type} may be one of the following.  The numbers indicate the
4722
value of the keyword.
4723
 
4724
@table @asis
4725
@item @code{PT_NULL} (0)
4726
Indicates an unused program header.
4727
 
4728
@item @code{PT_LOAD} (1)
4729
Indicates that this program header describes a segment to be loaded from
4730
the file.
4731
 
4732
@item @code{PT_DYNAMIC} (2)
4733
Indicates a segment where dynamic linking information can be found.
4734
 
4735
@item @code{PT_INTERP} (3)
4736
Indicates a segment where the name of the program interpreter may be
4737
found.
4738
 
4739
@item @code{PT_NOTE} (4)
4740
Indicates a segment holding note information.
4741
 
4742
@item @code{PT_SHLIB} (5)
4743
A reserved program header type, defined but not specified by the ELF
4744
ABI.
4745
 
4746
@item @code{PT_PHDR} (6)
4747
Indicates a segment where the program headers may be found.
4748
 
4749
@item @var{expression}
4750
An expression giving the numeric type of the program header.  This may
4751
be used for types not defined above.
4752
@end table
4753
 
4754
You can specify that a segment should be loaded at a particular address
4755
in memory by using an @code{AT} expression.  This is identical to the
4756
@code{AT} command used as an output section attribute (@pxref{Output
4757
Section LMA}).  The @code{AT} command for a program header overrides the
4758
output section attribute.
4759
 
4760
The linker will normally set the segment flags based on the sections
4761
which comprise the segment.  You may use the @code{FLAGS} keyword to
4762
explicitly specify the segment flags.  The value of @var{flags} must be
4763
an integer.  It is used to set the @code{p_flags} field of the program
4764
header.
4765
 
4766
Here is an example of @code{PHDRS}.  This shows a typical set of program
4767
headers used on a native ELF system.
4768
 
4769
@example
4770
@group
4771
PHDRS
4772
@{
4773
  headers PT_PHDR PHDRS ;
4774
  interp PT_INTERP ;
4775
  text PT_LOAD FILEHDR PHDRS ;
4776
  data PT_LOAD ;
4777
  dynamic PT_DYNAMIC ;
4778
@}
4779
 
4780
SECTIONS
4781
@{
4782
  . = SIZEOF_HEADERS;
4783
  .interp : @{ *(.interp) @} :text :interp
4784
  .text : @{ *(.text) @} :text
4785
  .rodata : @{ *(.rodata) @} /* defaults to :text */
4786
  @dots{}
4787
  . = . + 0x1000; /* move to a new page in memory */
4788
  .data : @{ *(.data) @} :data
4789
  .dynamic : @{ *(.dynamic) @} :data :dynamic
4790
  @dots{}
4791
@}
4792
@end group
4793
@end example
4794
 
4795
@node VERSION
4796
@section VERSION Command
4797
@kindex VERSION @{script text@}
4798
@cindex symbol versions
4799
@cindex version script
4800
@cindex versions of symbols
4801
The linker supports symbol versions when using ELF.  Symbol versions are
4802
only useful when using shared libraries.  The dynamic linker can use
4803
symbol versions to select a specific version of a function when it runs
4804
a program that may have been linked against an earlier version of the
4805
shared library.
4806
 
4807
You can include a version script directly in the main linker script, or
4808
you can supply the version script as an implicit linker script.  You can
4809
also use the @samp{--version-script} linker option.
4810
 
4811
The syntax of the @code{VERSION} command is simply
4812
@smallexample
4813
VERSION @{ version-script-commands @}
4814
@end smallexample
4815
 
4816
The format of the version script commands is identical to that used by
4817
Sun's linker in Solaris 2.5.  The version script defines a tree of
4818
version nodes.  You specify the node names and interdependencies in the
4819
version script.  You can specify which symbols are bound to which
4820
version nodes, and you can reduce a specified set of symbols to local
4821
scope so that they are not globally visible outside of the shared
4822
library.
4823
 
4824
The easiest way to demonstrate the version script language is with a few
4825
examples.
4826
 
4827
@smallexample
4828
VERS_1.1 @{
4829
         global:
4830
                 foo1;
4831
         local:
4832
                 old*;
4833
                 original*;
4834
                 new*;
4835
@};
4836
 
4837
VERS_1.2 @{
4838
                 foo2;
4839
@} VERS_1.1;
4840
 
4841
VERS_2.0 @{
4842
                 bar1; bar2;
4843
         extern "C++" @{
4844
                 ns::*;
4845
                 "int f(int, double)";
4846
         @}
4847
@} VERS_1.2;
4848
@end smallexample
4849
 
4850
This example version script defines three version nodes.  The first
4851
version node defined is @samp{VERS_1.1}; it has no other dependencies.
4852
The script binds the symbol @samp{foo1} to @samp{VERS_1.1}.  It reduces
4853
a number of symbols to local scope so that they are not visible outside
4854
of the shared library; this is done using wildcard patterns, so that any
4855
symbol whose name begins with @samp{old}, @samp{original}, or @samp{new}
4856
is matched.  The wildcard patterns available are the same as those used
4857
in the shell when matching filenames (also known as ``globbing'').
4858
However, if you specify the symbol name inside double quotes, then the
4859
name is treated as literal, rather than as a glob pattern.
4860
 
4861
Next, the version script defines node @samp{VERS_1.2}.  This node
4862
depends upon @samp{VERS_1.1}.  The script binds the symbol @samp{foo2}
4863
to the version node @samp{VERS_1.2}.
4864
 
4865
Finally, the version script defines node @samp{VERS_2.0}.  This node
4866
depends upon @samp{VERS_1.2}.  The scripts binds the symbols @samp{bar1}
4867
and @samp{bar2} are bound to the version node @samp{VERS_2.0}.
4868
 
4869
When the linker finds a symbol defined in a library which is not
4870
specifically bound to a version node, it will effectively bind it to an
4871
unspecified base version of the library.  You can bind all otherwise
4872
unspecified symbols to a given version node by using @samp{global: *;}
4873
somewhere in the version script.  Note that it's slightly crazy to use
4874
wildcards in a global spec except on the last version node.  Global
4875
wildcards elsewhere run the risk of accidentally adding symbols to the
4876
set exported for an old version.  That's wrong since older versions
4877
ought to have a fixed set of symbols.
4878
 
4879
The names of the version nodes have no specific meaning other than what
4880
they might suggest to the person reading them.  The @samp{2.0} version
4881
could just as well have appeared in between @samp{1.1} and @samp{1.2}.
4882
However, this would be a confusing way to write a version script.
4883
 
4884
Node name can be omitted, provided it is the only version node
4885
in the version script.  Such version script doesn't assign any versions to
4886
symbols, only selects which symbols will be globally visible out and which
4887
won't.
4888
 
4889
@smallexample
4890
@{ global: foo; bar; local: *; @};
4891
@end smallexample
4892
 
4893
When you link an application against a shared library that has versioned
4894
symbols, the application itself knows which version of each symbol it
4895
requires, and it also knows which version nodes it needs from each
4896
shared library it is linked against.  Thus at runtime, the dynamic
4897
loader can make a quick check to make sure that the libraries you have
4898
linked against do in fact supply all of the version nodes that the
4899
application will need to resolve all of the dynamic symbols.  In this
4900
way it is possible for the dynamic linker to know with certainty that
4901
all external symbols that it needs will be resolvable without having to
4902
search for each symbol reference.
4903
 
4904
The symbol versioning is in effect a much more sophisticated way of
4905
doing minor version checking that SunOS does.  The fundamental problem
4906
that is being addressed here is that typically references to external
4907
functions are bound on an as-needed basis, and are not all bound when
4908
the application starts up.  If a shared library is out of date, a
4909
required interface may be missing; when the application tries to use
4910
that interface, it may suddenly and unexpectedly fail.  With symbol
4911
versioning, the user will get a warning when they start their program if
4912
the libraries being used with the application are too old.
4913
 
4914
There are several GNU extensions to Sun's versioning approach.  The
4915
first of these is the ability to bind a symbol to a version node in the
4916
source file where the symbol is defined instead of in the versioning
4917
script.  This was done mainly to reduce the burden on the library
4918
maintainer.  You can do this by putting something like:
4919
@smallexample
4920
__asm__(".symver original_foo,foo@@VERS_1.1");
4921
@end smallexample
4922
@noindent
4923
in the C source file.  This renames the function @samp{original_foo} to
4924
be an alias for @samp{foo} bound to the version node @samp{VERS_1.1}.
4925
The @samp{local:} directive can be used to prevent the symbol
4926
@samp{original_foo} from being exported. A @samp{.symver} directive
4927
takes precedence over a version script.
4928
 
4929
The second GNU extension is to allow multiple versions of the same
4930
function to appear in a given shared library.  In this way you can make
4931
an incompatible change to an interface without increasing the major
4932
version number of the shared library, while still allowing applications
4933
linked against the old interface to continue to function.
4934
 
4935
To do this, you must use multiple @samp{.symver} directives in the
4936
source file.  Here is an example:
4937
 
4938
@smallexample
4939
__asm__(".symver original_foo,foo@@");
4940
__asm__(".symver old_foo,foo@@VERS_1.1");
4941
__asm__(".symver old_foo1,foo@@VERS_1.2");
4942
__asm__(".symver new_foo,foo@@@@VERS_2.0");
4943
@end smallexample
4944
 
4945
In this example, @samp{foo@@} represents the symbol @samp{foo} bound to the
4946
unspecified base version of the symbol.  The source file that contains this
4947
example would define 4 C functions: @samp{original_foo}, @samp{old_foo},
4948
@samp{old_foo1}, and @samp{new_foo}.
4949
 
4950
When you have multiple definitions of a given symbol, there needs to be
4951
some way to specify a default version to which external references to
4952
this symbol will be bound.  You can do this with the
4953
@samp{foo@@@@VERS_2.0} type of @samp{.symver} directive.  You can only
4954
declare one version of a symbol as the default in this manner; otherwise
4955
you would effectively have multiple definitions of the same symbol.
4956
 
4957
If you wish to bind a reference to a specific version of the symbol
4958
within the shared library, you can use the aliases of convenience
4959
(i.e., @samp{old_foo}), or you can use the @samp{.symver} directive to
4960
specifically bind to an external version of the function in question.
4961
 
4962
You can also specify the language in the version script:
4963
 
4964
@smallexample
4965
VERSION extern "lang" @{ version-script-commands @}
4966
@end smallexample
4967
 
4968
The supported @samp{lang}s are @samp{C}, @samp{C++}, and @samp{Java}.
4969
The linker will iterate over the list of symbols at the link time and
4970
demangle them according to @samp{lang} before matching them to the
4971
patterns specified in @samp{version-script-commands}.
4972
 
4973
Demangled names may contains spaces and other special characters.  As
4974
described above, you can use a glob pattern to match demangled names,
4975
or you can use a double-quoted string to match the string exactly.  In
4976
the latter case, be aware that minor differences (such as differing
4977
whitespace) between the version script and the demangler output will
4978
cause a mismatch.  As the exact string generated by the demangler
4979
might change in the future, even if the mangled name does not, you
4980
should check that all of your version directives are behaving as you
4981
expect when you upgrade.
4982
 
4983
@node Expressions
4984
@section Expressions in Linker Scripts
4985
@cindex expressions
4986
@cindex arithmetic
4987
The syntax for expressions in the linker script language is identical to
4988
that of C expressions.  All expressions are evaluated as integers.  All
4989
expressions are evaluated in the same size, which is 32 bits if both the
4990
host and target are 32 bits, and is otherwise 64 bits.
4991
 
4992
You can use and set symbol values in expressions.
4993
 
4994
The linker defines several special purpose builtin functions for use in
4995
expressions.
4996
 
4997
@menu
4998
* Constants::                   Constants
4999
* Symbolic Constants::          Symbolic constants
5000
* Symbols::                     Symbol Names
5001
* Orphan Sections::             Orphan Sections
5002
* Location Counter::            The Location Counter
5003
* Operators::                   Operators
5004
* Evaluation::                  Evaluation
5005
* Expression Section::          The Section of an Expression
5006
* Builtin Functions::           Builtin Functions
5007
@end menu
5008
 
5009
@node Constants
5010
@subsection Constants
5011
@cindex integer notation
5012
@cindex constants in linker scripts
5013
All constants are integers.
5014
 
5015
As in C, the linker considers an integer beginning with @samp{0} to be
5016
octal, and an integer beginning with @samp{0x} or @samp{0X} to be
5017
hexadecimal.  Alternatively the linker accepts suffixes of @samp{h} or
5018
@samp{H} for hexadeciaml, @samp{o} or @samp{O} for octal, @samp{b} or
5019
@samp{B} for binary and @samp{d} or @samp{D} for decimal.  Any integer
5020
value without a prefix or a suffix is considered to be decimal.
5021
 
5022
@cindex scaled integers
5023
@cindex K and M integer suffixes
5024
@cindex M and K integer suffixes
5025
@cindex suffixes for integers
5026
@cindex integer suffixes
5027
In addition, you can use the suffixes @code{K} and @code{M} to scale a
5028
constant by
5029
@c TEXI2ROFF-KILL
5030
@ifnottex
5031
@c END TEXI2ROFF-KILL
5032
@code{1024} or @code{1024*1024}
5033
@c TEXI2ROFF-KILL
5034
@end ifnottex
5035
@tex
5036
${\rm 1024}$ or ${\rm 1024}^2$
5037
@end tex
5038
@c END TEXI2ROFF-KILL
5039
respectively.  For example, the following
5040
all refer to the same quantity:
5041
 
5042
@smallexample
5043
_fourk_1 = 4K;
5044
_fourk_2 = 4096;
5045
_fourk_3 = 0x1000;
5046
_fourk_4 = 10000o;
5047
@end smallexample
5048
 
5049
Note - the @code{K} and @code{M} suffixes cannot be used in
5050
conjunction with the base suffixes mentioned above.
5051
 
5052
@node Symbolic Constants
5053
@subsection Symbolic Constants
5054
@cindex symbolic constants
5055
@kindex CONSTANT
5056
It is possible to refer to target specific constants via the use of
5057
the @code{CONSTANT(@var{name})} operator, where @var{name} is one of:
5058
 
5059
@table @code
5060
@item MAXPAGESIZE
5061
@kindex MAXPAGESIZE
5062
The target's maximum page size.
5063
 
5064
@item COMMONPAGESIZE
5065
@kindex COMMONPAGESIZE
5066
The target's default page size.
5067
@end table
5068
 
5069
So for example:
5070
 
5071
@smallexample
5072
  .text ALIGN (CONSTANT (MAXPAGESIZE)) : @{ *(.text) @}
5073
@end smallexample
5074
 
5075
will create a text section aligned to the largest page boundary
5076
supported by the target.
5077
 
5078
@node Symbols
5079
@subsection Symbol Names
5080
@cindex symbol names
5081
@cindex names
5082
@cindex quoted symbol names
5083
@kindex "
5084
Unless quoted, symbol names start with a letter, underscore, or period
5085
and may include letters, digits, underscores, periods, and hyphens.
5086
Unquoted symbol names must not conflict with any keywords.  You can
5087
specify a symbol which contains odd characters or has the same name as a
5088
keyword by surrounding the symbol name in double quotes:
5089
@smallexample
5090
"SECTION" = 9;
5091
"with a space" = "also with a space" + 10;
5092
@end smallexample
5093
 
5094
Since symbols can contain many non-alphabetic characters, it is safest
5095
to delimit symbols with spaces.  For example, @samp{A-B} is one symbol,
5096
whereas @samp{A - B} is an expression involving subtraction.
5097
 
5098
@node Orphan Sections
5099
@subsection Orphan Sections
5100
@cindex orphan
5101
Orphan sections are sections present in the input files which
5102
are not explicitly placed into the output file by the linker
5103
script.  The linker will still copy these sections into the
5104
output file, but it has to guess as to where they should be
5105
placed.  The linker uses a simple heuristic to do this.  It
5106
attempts to place orphan sections after non-orphan sections of the
5107
same attribute, such as code vs data, loadable vs non-loadable, etc.
5108
If there is not enough room to do this then it places
5109
at the end of the file.
5110
 
5111
For ELF targets, the attribute of the section includes section type as
5112
well as section flag.
5113
 
5114
If an orphaned section's name is representable as a C identifier then
5115
the linker will automatically @pxref{PROVIDE} two symbols:
5116
__start_SECNAME and __end_SECNAME, where SECNAME is the name of the
5117
section.  These indicate the start address and end address of the
5118
orphaned section respectively.  Note: most section names are not
5119
representable as C identifiers because they contain a @samp{.}
5120
character.
5121
 
5122
@node Location Counter
5123
@subsection The Location Counter
5124
@kindex .
5125
@cindex dot
5126
@cindex location counter
5127
@cindex current output location
5128
The special linker variable @dfn{dot} @samp{.} always contains the
5129
current output location counter.  Since the @code{.} always refers to a
5130
location in an output section, it may only appear in an expression
5131
within a @code{SECTIONS} command.  The @code{.} symbol may appear
5132
anywhere that an ordinary symbol is allowed in an expression.
5133
 
5134
@cindex holes
5135
Assigning a value to @code{.} will cause the location counter to be
5136
moved.  This may be used to create holes in the output section.  The
5137
location counter may not be moved backwards inside an output section,
5138
and may not be moved backwards outside of an output section if so
5139
doing creates areas with overlapping LMAs.
5140
 
5141
@smallexample
5142
SECTIONS
5143
@{
5144
  output :
5145
    @{
5146
      file1(.text)
5147
      . = . + 1000;
5148
      file2(.text)
5149
      . += 1000;
5150
      file3(.text)
5151
    @} = 0x12345678;
5152
@}
5153
@end smallexample
5154
@noindent
5155
In the previous example, the @samp{.text} section from @file{file1} is
5156
located at the beginning of the output section @samp{output}.  It is
5157
followed by a 1000 byte gap.  Then the @samp{.text} section from
5158
@file{file2} appears, also with a 1000 byte gap following before the
5159
@samp{.text} section from @file{file3}.  The notation @samp{= 0x12345678}
5160
specifies what data to write in the gaps (@pxref{Output Section Fill}).
5161
 
5162
@cindex dot inside sections
5163
Note: @code{.} actually refers to the byte offset from the start of the
5164
current containing object.  Normally this is the @code{SECTIONS}
5165
statement, whose start address is 0, hence @code{.} can be used as an
5166
absolute address.  If @code{.} is used inside a section description
5167
however, it refers to the byte offset from the start of that section,
5168
not an absolute address.  Thus in a script like this:
5169
 
5170
@smallexample
5171
SECTIONS
5172
@{
5173
    . = 0x100
5174
    .text: @{
5175
      *(.text)
5176
      . = 0x200
5177
    @}
5178
    . = 0x500
5179
    .data: @{
5180
      *(.data)
5181
      . += 0x600
5182
    @}
5183
@}
5184
@end smallexample
5185
 
5186
The @samp{.text} section will be assigned a starting address of 0x100
5187
and a size of exactly 0x200 bytes, even if there is not enough data in
5188
the @samp{.text} input sections to fill this area.  (If there is too
5189
much data, an error will be produced because this would be an attempt to
5190
move @code{.} backwards).  The @samp{.data} section will start at 0x500
5191
and it will have an extra 0x600 bytes worth of space after the end of
5192
the values from the @samp{.data} input sections and before the end of
5193
the @samp{.data} output section itself.
5194
 
5195
@cindex dot outside sections
5196
Setting symbols to the value of the location counter outside of an
5197
output section statement can result in unexpected values if the linker
5198
needs to place orphan sections.  For example, given the following:
5199
 
5200
@smallexample
5201
SECTIONS
5202
@{
5203
    start_of_text = . ;
5204
    .text: @{ *(.text) @}
5205
    end_of_text = . ;
5206
 
5207
    start_of_data = . ;
5208
    .data: @{ *(.data) @}
5209
    end_of_data = . ;
5210
@}
5211
@end smallexample
5212
 
5213
If the linker needs to place some input section, e.g. @code{.rodata},
5214
not mentioned in the script, it might choose to place that section
5215
between @code{.text} and @code{.data}.  You might think the linker
5216
should place @code{.rodata} on the blank line in the above script, but
5217
blank lines are of no particular significance to the linker.  As well,
5218
the linker doesn't associate the above symbol names with their
5219
sections.  Instead, it assumes that all assignments or other
5220
statements belong to the previous output section, except for the
5221
special case of an assignment to @code{.}.  I.e., the linker will
5222
place the orphan @code{.rodata} section as if the script was written
5223
as follows:
5224
 
5225
@smallexample
5226
SECTIONS
5227
@{
5228
    start_of_text = . ;
5229
    .text: @{ *(.text) @}
5230
    end_of_text = . ;
5231
 
5232
    start_of_data = . ;
5233
    .rodata: @{ *(.rodata) @}
5234
    .data: @{ *(.data) @}
5235
    end_of_data = . ;
5236
@}
5237
@end smallexample
5238
 
5239
This may or may not be the script author's intention for the value of
5240
@code{start_of_data}.  One way to influence the orphan section
5241
placement is to assign the location counter to itself, as the linker
5242
assumes that an assignment to @code{.} is setting the start address of
5243
a following output section and thus should be grouped with that
5244
section.  So you could write:
5245
 
5246
@smallexample
5247
SECTIONS
5248
@{
5249
    start_of_text = . ;
5250
    .text: @{ *(.text) @}
5251
    end_of_text = . ;
5252
 
5253
    . = . ;
5254
    start_of_data = . ;
5255
    .data: @{ *(.data) @}
5256
    end_of_data = . ;
5257
@}
5258
@end smallexample
5259
 
5260
Now, the orphan @code{.rodata} section will be placed between
5261
@code{end_of_text} and @code{start_of_data}.
5262
 
5263
@need 2000
5264
@node Operators
5265
@subsection Operators
5266
@cindex operators for arithmetic
5267
@cindex arithmetic operators
5268
@cindex precedence in expressions
5269
The linker recognizes the standard C set of arithmetic operators, with
5270
the standard bindings and precedence levels:
5271
@c TEXI2ROFF-KILL
5272
@ifnottex
5273
@c END TEXI2ROFF-KILL
5274
@smallexample
5275
precedence      associativity   Operators                Notes
5276
(highest)
5277
1               left            !  -  ~                  (1)
5278
2               left            *  /  %
5279
3               left            +  -
5280
4               left            >>  <<
5281
5               left            ==  !=  >  <  <=  >=
5282
6               left            &
5283
7               left            |
5284
8               left            &&
5285
9               left            ||
5286
10              right           ? :
5287
11              right           &=  +=  -=  *=  /=       (2)
5288
(lowest)
5289
@end smallexample
5290
Notes:
5291
(1) Prefix operators
5292
(2) @xref{Assignments}.
5293
@c TEXI2ROFF-KILL
5294
@end ifnottex
5295
@tex
5296
\vskip \baselineskip
5297
%"lispnarrowing" is the extra indent used generally for smallexample
5298
\hskip\lispnarrowing\vbox{\offinterlineskip
5299
\hrule
5300
\halign
5301
{\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ #\ \hfil&\vrule#&\strut\hfil\ {\tt #}\ \hfil&\vrule#\cr
5302
height2pt&\omit&&\omit&&\omit&\cr
5303
&Precedence&&  Associativity  &&{\rm Operators}&\cr
5304
height2pt&\omit&&\omit&&\omit&\cr
5305
\noalign{\hrule}
5306
height2pt&\omit&&\omit&&\omit&\cr
5307
&highest&&&&&\cr
5308
% '176 is tilde, '~' in tt font
5309
&1&&left&&\qquad-          \char'176\      !\qquad\dag&\cr
5310
&2&&left&&*          /        \%&\cr
5311
&3&&left&&+          -&\cr
5312
&4&&left&&>>         <<&\cr
5313
&5&&left&&==         !=       >      <      <=      >=&\cr
5314
&6&&left&&\&&\cr
5315
&7&&left&&|&\cr
5316
&8&&left&&{\&\&}&\cr
5317
&9&&left&&||&\cr
5318
&10&&right&&?        :&\cr
5319
&11&&right&&\qquad\&=      +=       -=     *=     /=\qquad\ddag&\cr
5320
&lowest&&&&&\cr
5321
height2pt&\omit&&\omit&&\omit&\cr}
5322
\hrule}
5323
@end tex
5324
@iftex
5325
{
5326
@obeylines@parskip=0pt@parindent=0pt
5327
@dag@quad Prefix operators.
5328
@ddag@quad @xref{Assignments}.
5329
}
5330
@end iftex
5331
@c END TEXI2ROFF-KILL
5332
 
5333
@node Evaluation
5334
@subsection Evaluation
5335
@cindex lazy evaluation
5336
@cindex expression evaluation order
5337
The linker evaluates expressions lazily.  It only computes the value of
5338
an expression when absolutely necessary.
5339
 
5340
The linker needs some information, such as the value of the start
5341
address of the first section, and the origins and lengths of memory
5342
regions, in order to do any linking at all.  These values are computed
5343
as soon as possible when the linker reads in the linker script.
5344
 
5345
However, other values (such as symbol values) are not known or needed
5346
until after storage allocation.  Such values are evaluated later, when
5347
other information (such as the sizes of output sections) is available
5348
for use in the symbol assignment expression.
5349
 
5350
The sizes of sections cannot be known until after allocation, so
5351
assignments dependent upon these are not performed until after
5352
allocation.
5353
 
5354
Some expressions, such as those depending upon the location counter
5355
@samp{.}, must be evaluated during section allocation.
5356
 
5357
If the result of an expression is required, but the value is not
5358
available, then an error results.  For example, a script like the
5359
following
5360
@smallexample
5361
@group
5362
SECTIONS
5363
  @{
5364
    .text 9+this_isnt_constant :
5365
      @{ *(.text) @}
5366
  @}
5367
@end group
5368
@end smallexample
5369
@noindent
5370
will cause the error message @samp{non constant expression for initial
5371
address}.
5372
 
5373
@node Expression Section
5374
@subsection The Section of an Expression
5375
@cindex expression sections
5376
@cindex absolute expressions
5377
@cindex relative expressions
5378
@cindex absolute and relocatable symbols
5379
@cindex relocatable and absolute symbols
5380
@cindex symbols, relocatable and absolute
5381
When the linker evaluates an expression, the result is either absolute
5382
or relative to some section.  A relative expression is expressed as a
5383
fixed offset from the base of a section.
5384
 
5385
The position of the expression within the linker script determines
5386
whether it is absolute or relative.  An expression which appears within
5387
an output section definition is relative to the base of the output
5388
section.  An expression which appears elsewhere will be absolute.
5389
 
5390
A symbol set to a relative expression will be relocatable if you request
5391
relocatable output using the @samp{-r} option.  That means that a
5392
further link operation may change the value of the symbol.  The symbol's
5393
section will be the section of the relative expression.
5394
 
5395
A symbol set to an absolute expression will retain the same value
5396
through any further link operation.  The symbol will be absolute, and
5397
will not have any particular associated section.
5398
 
5399
You can use the builtin function @code{ABSOLUTE} to force an expression
5400
to be absolute when it would otherwise be relative.  For example, to
5401
create an absolute symbol set to the address of the end of the output
5402
section @samp{.data}:
5403
@smallexample
5404
SECTIONS
5405
  @{
5406
    .data : @{ *(.data) _edata = ABSOLUTE(.); @}
5407
  @}
5408
@end smallexample
5409
@noindent
5410
If @samp{ABSOLUTE} were not used, @samp{_edata} would be relative to the
5411
@samp{.data} section.
5412
 
5413
@node Builtin Functions
5414
@subsection Builtin Functions
5415
@cindex functions in expressions
5416
The linker script language includes a number of builtin functions for
5417
use in linker script expressions.
5418
 
5419
@table @code
5420
@item ABSOLUTE(@var{exp})
5421
@kindex ABSOLUTE(@var{exp})
5422
@cindex expression, absolute
5423
Return the absolute (non-relocatable, as opposed to non-negative) value
5424
of the expression @var{exp}.  Primarily useful to assign an absolute
5425
value to a symbol within a section definition, where symbol values are
5426
normally section relative.  @xref{Expression Section}.
5427
 
5428
@item ADDR(@var{section})
5429
@kindex ADDR(@var{section})
5430
@cindex section address in expression
5431
Return the absolute address (the VMA) of the named @var{section}.  Your
5432
script must previously have defined the location of that section.  In
5433
the following example, @code{symbol_1} and @code{symbol_2} are assigned
5434
identical values:
5435
@smallexample
5436
@group
5437
SECTIONS @{ @dots{}
5438
  .output1 :
5439
    @{
5440
    start_of_output_1 = ABSOLUTE(.);
5441
    @dots{}
5442
    @}
5443
  .output :
5444
    @{
5445
    symbol_1 = ADDR(.output1);
5446
    symbol_2 = start_of_output_1;
5447
    @}
5448
@dots{} @}
5449
@end group
5450
@end smallexample
5451
 
5452
@item ALIGN(@var{align})
5453
@itemx ALIGN(@var{exp},@var{align})
5454
@kindex ALIGN(@var{align})
5455
@kindex ALIGN(@var{exp},@var{align})
5456
@cindex round up location counter
5457
@cindex align location counter
5458
@cindex round up expression
5459
@cindex align expression
5460
Return the location counter (@code{.}) or arbitrary expression aligned
5461
to the next @var{align} boundary.  The single operand @code{ALIGN}
5462
doesn't change the value of the location counter---it just does
5463
arithmetic on it.  The two operand @code{ALIGN} allows an arbitrary
5464
expression to be aligned upwards (@code{ALIGN(@var{align})} is
5465
equivalent to @code{ALIGN(., @var{align})}).
5466
 
5467
Here is an example which aligns the output @code{.data} section to the
5468
next @code{0x2000} byte boundary after the preceding section and sets a
5469
variable within the section to the next @code{0x8000} boundary after the
5470
input sections:
5471
@smallexample
5472
@group
5473
SECTIONS @{ @dots{}
5474
  .data ALIGN(0x2000): @{
5475
    *(.data)
5476
    variable = ALIGN(0x8000);
5477
  @}
5478
@dots{} @}
5479
@end group
5480
@end smallexample
5481
@noindent
5482
The first use of @code{ALIGN} in this example specifies the location of
5483
a section because it is used as the optional @var{address} attribute of
5484
a section definition (@pxref{Output Section Address}).  The second use
5485
of @code{ALIGN} is used to defines the value of a symbol.
5486
 
5487
The builtin function @code{NEXT} is closely related to @code{ALIGN}.
5488
 
5489
@item ALIGNOF(@var{section})
5490
@kindex ALIGNOF(@var{section})
5491
@cindex section alignment
5492
Return the alignment in bytes of the named @var{section}, if that section has
5493
been allocated.  If the section has not been allocated when this is
5494
evaluated, the linker will report an error. In the following example,
5495
the alignment of the @code{.output} section is stored as the first
5496
value in that section.
5497
@smallexample
5498
@group
5499
SECTIONS@{ @dots{}
5500
  .output @{
5501
    LONG (ALIGNOF (.output))
5502
    @dots{}
5503
    @}
5504
@dots{} @}
5505
@end group
5506
@end smallexample
5507
 
5508
@item BLOCK(@var{exp})
5509
@kindex BLOCK(@var{exp})
5510
This is a synonym for @code{ALIGN}, for compatibility with older linker
5511
scripts.  It is most often seen when setting the address of an output
5512
section.
5513
 
5514
@item DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5515
@kindex DATA_SEGMENT_ALIGN(@var{maxpagesize}, @var{commonpagesize})
5516
This is equivalent to either
5517
@smallexample
5518
(ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - 1)))
5519
@end smallexample
5520
or
5521
@smallexample
5522
(ALIGN(@var{maxpagesize}) + (. & (@var{maxpagesize} - @var{commonpagesize})))
5523
@end smallexample
5524
@noindent
5525
depending on whether the latter uses fewer @var{commonpagesize} sized pages
5526
for the data segment (area between the result of this expression and
5527
@code{DATA_SEGMENT_END}) than the former or not.
5528
If the latter form is used, it means @var{commonpagesize} bytes of runtime
5529
memory will be saved at the expense of up to @var{commonpagesize} wasted
5530
bytes in the on-disk file.
5531
 
5532
This expression can only be used directly in @code{SECTIONS} commands, not in
5533
any output section descriptions and only once in the linker script.
5534
@var{commonpagesize} should be less or equal to @var{maxpagesize} and should
5535
be the system page size the object wants to be optimized for (while still
5536
working on system page sizes up to @var{maxpagesize}).
5537
 
5538
@noindent
5539
Example:
5540
@smallexample
5541
  . = DATA_SEGMENT_ALIGN(0x10000, 0x2000);
5542
@end smallexample
5543
 
5544
@item DATA_SEGMENT_END(@var{exp})
5545
@kindex DATA_SEGMENT_END(@var{exp})
5546
This defines the end of data segment for @code{DATA_SEGMENT_ALIGN}
5547
evaluation purposes.
5548
 
5549
@smallexample
5550
  . = DATA_SEGMENT_END(.);
5551
@end smallexample
5552
 
5553
@item DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5554
@kindex DATA_SEGMENT_RELRO_END(@var{offset}, @var{exp})
5555
This defines the end of the @code{PT_GNU_RELRO} segment when
5556
@samp{-z relro} option is used.  Second argument is returned.
5557
When @samp{-z relro} option is not present, @code{DATA_SEGMENT_RELRO_END}
5558
does nothing, otherwise @code{DATA_SEGMENT_ALIGN} is padded so that
5559
@var{exp} + @var{offset} is aligned to the most commonly used page
5560
boundary for particular target.  If present in the linker script,
5561
it must always come in between @code{DATA_SEGMENT_ALIGN} and
5562
@code{DATA_SEGMENT_END}.
5563
 
5564
@smallexample
5565
  . = DATA_SEGMENT_RELRO_END(24, .);
5566
@end smallexample
5567
 
5568
@item DEFINED(@var{symbol})
5569
@kindex DEFINED(@var{symbol})
5570
@cindex symbol defaults
5571
Return 1 if @var{symbol} is in the linker global symbol table and is
5572
defined before the statement using DEFINED in the script, otherwise
5573
return 0.  You can use this function to provide
5574
default values for symbols.  For example, the following script fragment
5575
shows how to set a global symbol @samp{begin} to the first location in
5576
the @samp{.text} section---but if a symbol called @samp{begin} already
5577
existed, its value is preserved:
5578
 
5579
@smallexample
5580
@group
5581
SECTIONS @{ @dots{}
5582
  .text : @{
5583
    begin = DEFINED(begin) ? begin : . ;
5584
    @dots{}
5585
  @}
5586
  @dots{}
5587
@}
5588
@end group
5589
@end smallexample
5590
 
5591
@item LENGTH(@var{memory})
5592
@kindex LENGTH(@var{memory})
5593
Return the length of the memory region named @var{memory}.
5594
 
5595
@item LOADADDR(@var{section})
5596
@kindex LOADADDR(@var{section})
5597
@cindex section load address in expression
5598
Return the absolute LMA of the named @var{section}.  This is normally
5599
the same as @code{ADDR}, but it may be different if the @code{AT}
5600
attribute is used in the output section definition (@pxref{Output
5601
Section LMA}).
5602
 
5603
@kindex MAX
5604
@item MAX(@var{exp1}, @var{exp2})
5605
Returns the maximum of @var{exp1} and @var{exp2}.
5606
 
5607
@kindex MIN
5608
@item MIN(@var{exp1}, @var{exp2})
5609
Returns the minimum of @var{exp1} and @var{exp2}.
5610
 
5611
@item NEXT(@var{exp})
5612
@kindex NEXT(@var{exp})
5613
@cindex unallocated address, next
5614
Return the next unallocated address that is a multiple of @var{exp}.
5615
This function is closely related to @code{ALIGN(@var{exp})}; unless you
5616
use the @code{MEMORY} command to define discontinuous memory for the
5617
output file, the two functions are equivalent.
5618
 
5619
@item ORIGIN(@var{memory})
5620
@kindex ORIGIN(@var{memory})
5621
Return the origin of the memory region named @var{memory}.
5622
 
5623
@item SEGMENT_START(@var{segment}, @var{default})
5624
@kindex SEGMENT_START(@var{segment}, @var{default})
5625
Return the base address of the named @var{segment}.  If an explicit
5626
value has been given for this segment (with a command-line @samp{-T}
5627
option) that value will be returned; otherwise the value will be
5628
@var{default}.  At present, the @samp{-T} command-line option can only
5629
be used to set the base address for the ``text'', ``data'', and
5630
``bss'' sections, but you use @code{SEGMENT_START} with any segment
5631
name.
5632
 
5633
@item SIZEOF(@var{section})
5634
@kindex SIZEOF(@var{section})
5635
@cindex section size
5636
Return the size in bytes of the named @var{section}, if that section has
5637
been allocated.  If the section has not been allocated when this is
5638
evaluated, the linker will report an error.  In the following example,
5639
@code{symbol_1} and @code{symbol_2} are assigned identical values:
5640
@smallexample
5641
@group
5642
SECTIONS@{ @dots{}
5643
  .output @{
5644
    .start = . ;
5645
    @dots{}
5646
    .end = . ;
5647
    @}
5648
  symbol_1 = .end - .start ;
5649
  symbol_2 = SIZEOF(.output);
5650
@dots{} @}
5651
@end group
5652
@end smallexample
5653
 
5654
@item SIZEOF_HEADERS
5655
@itemx sizeof_headers
5656
@kindex SIZEOF_HEADERS
5657
@cindex header size
5658
Return the size in bytes of the output file's headers.  This is
5659
information which appears at the start of the output file.  You can use
5660
this number when setting the start address of the first section, if you
5661
choose, to facilitate paging.
5662
 
5663
@cindex not enough room for program headers
5664
@cindex program headers, not enough room
5665
When producing an ELF output file, if the linker script uses the
5666
@code{SIZEOF_HEADERS} builtin function, the linker must compute the
5667
number of program headers before it has determined all the section
5668
addresses and sizes.  If the linker later discovers that it needs
5669
additional program headers, it will report an error @samp{not enough
5670
room for program headers}.  To avoid this error, you must avoid using
5671
the @code{SIZEOF_HEADERS} function, or you must rework your linker
5672
script to avoid forcing the linker to use additional program headers, or
5673
you must define the program headers yourself using the @code{PHDRS}
5674
command (@pxref{PHDRS}).
5675
@end table
5676
 
5677
@node Implicit Linker Scripts
5678
@section Implicit Linker Scripts
5679
@cindex implicit linker scripts
5680
If you specify a linker input file which the linker can not recognize as
5681
an object file or an archive file, it will try to read the file as a
5682
linker script.  If the file can not be parsed as a linker script, the
5683
linker will report an error.
5684
 
5685
An implicit linker script will not replace the default linker script.
5686
 
5687
Typically an implicit linker script would contain only symbol
5688
assignments, or the @code{INPUT}, @code{GROUP}, or @code{VERSION}
5689
commands.
5690
 
5691
Any input files read because of an implicit linker script will be read
5692
at the position in the command line where the implicit linker script was
5693
read.  This can affect archive searching.
5694
 
5695
@ifset GENERIC
5696
@node Machine Dependent
5697
@chapter Machine Dependent Features
5698
 
5699
@cindex machine dependencies
5700
@command{ld} has additional features on some platforms; the following
5701
sections describe them.  Machines where @command{ld} has no additional
5702
functionality are not listed.
5703
 
5704
@menu
5705
@ifset H8300
5706
* H8/300::                      @command{ld} and the H8/300
5707
@end ifset
5708
@ifset I960
5709
* i960::                        @command{ld} and the Intel 960 family
5710
@end ifset
5711
@ifset ARM
5712
* ARM::                         @command{ld} and the ARM family
5713
@end ifset
5714
@ifset HPPA
5715
* HPPA ELF32::                  @command{ld} and HPPA 32-bit ELF
5716
@end ifset
5717
@ifset M68K
5718
* M68K::                        @command{ld} and the Motorola 68K family
5719
@end ifset
5720
@ifset MMIX
5721
* MMIX::                        @command{ld} and MMIX
5722
@end ifset
5723
@ifset MSP430
5724
* MSP430::                      @command{ld} and MSP430
5725
@end ifset
5726
@ifset M68HC11
5727
* M68HC11/68HC12::              @code{ld} and the Motorola 68HC11 and 68HC12 families
5728
@end ifset
5729
@ifset POWERPC
5730
* PowerPC ELF32::               @command{ld} and PowerPC 32-bit ELF Support
5731
@end ifset
5732
@ifset POWERPC64
5733
* PowerPC64 ELF64::             @command{ld} and PowerPC64 64-bit ELF Support
5734
@end ifset
5735
@ifset SPU
5736
* SPU ELF::                     @command{ld} and SPU ELF Support
5737
@end ifset
5738
@ifset TICOFF
5739
* TI COFF::                     @command{ld} and TI COFF
5740
@end ifset
5741
@ifset WIN32
5742
* WIN32::                       @command{ld} and WIN32 (cygwin/mingw)
5743
@end ifset
5744
@ifset XTENSA
5745
* Xtensa::                      @command{ld} and Xtensa Processors
5746
@end ifset
5747
@end menu
5748
@end ifset
5749
 
5750
@ifset H8300
5751
@ifclear GENERIC
5752
@raisesections
5753
@end ifclear
5754
 
5755
@node H8/300
5756
@section @command{ld} and the H8/300
5757
 
5758
@cindex H8/300 support
5759
For the H8/300, @command{ld} can perform these global optimizations when
5760
you specify the @samp{--relax} command-line option.
5761
 
5762
@table @emph
5763
@cindex relaxing on H8/300
5764
@item relaxing address modes
5765
@command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5766
targets are within eight bits, and turns them into eight-bit
5767
program-counter relative @code{bsr} and @code{bra} instructions,
5768
respectively.
5769
 
5770
@cindex synthesizing on H8/300
5771
@item synthesizing instructions
5772
@c FIXME: specifically mov.b, or any mov instructions really?
5773
@command{ld} finds all @code{mov.b} instructions which use the
5774
sixteen-bit absolute address form, but refer to the top
5775
page of memory, and changes them to use the eight-bit address form.
5776
(That is: the linker turns @samp{mov.b @code{@@}@var{aa}:16} into
5777
@samp{mov.b @code{@@}@var{aa}:8} whenever the address @var{aa} is in the
5778
top page of memory).
5779
 
5780
@item bit manipulation instructions
5781
@command{ld} finds all bit manipulation instructions like @code{band, bclr,
5782
biand, bild, bior, bist, bixor, bld, bnot, bor, bset, bst, btst, bxor}
5783
which use 32 bit and 16 bit absolute address form, but refer to the top
5784
page of memory, and changes them to use the 8 bit address form.
5785
(That is: the linker turns @samp{bset #xx:3,@code{@@}@var{aa}:32} into
5786
@samp{bset #xx:3,@code{@@}@var{aa}:8} whenever the address @var{aa} is in
5787
the top page of memory).
5788
 
5789
@item system control instructions
5790
@command{ld} finds all @code{ldc.w, stc.w} instructions which use the
5791
32 bit absolute address form, but refer to the top page of memory, and
5792
changes them to use 16 bit address form.
5793
(That is: the linker turns @samp{ldc.w @code{@@}@var{aa}:32,ccr} into
5794
@samp{ldc.w @code{@@}@var{aa}:16,ccr} whenever the address @var{aa} is in
5795
the top page of memory).
5796
@end table
5797
 
5798
@ifclear GENERIC
5799
@lowersections
5800
@end ifclear
5801
@end ifset
5802
 
5803
@ifclear GENERIC
5804
@ifset Renesas
5805
@c This stuff is pointless to say unless you're especially concerned
5806
@c with Renesas chips; don't enable it for generic case, please.
5807
@node Renesas
5808
@chapter @command{ld} and Other Renesas Chips
5809
 
5810
@command{ld} also supports the Renesas (formerly Hitachi) H8/300H,
5811
H8/500, and SH chips.  No special features, commands, or command-line
5812
options are required for these chips.
5813
@end ifset
5814
@end ifclear
5815
 
5816
@ifset I960
5817
@ifclear GENERIC
5818
@raisesections
5819
@end ifclear
5820
 
5821
@node i960
5822
@section @command{ld} and the Intel 960 Family
5823
 
5824
@cindex i960 support
5825
 
5826
You can use the @samp{-A@var{architecture}} command line option to
5827
specify one of the two-letter names identifying members of the 960
5828
family; the option specifies the desired output target, and warns of any
5829
incompatible instructions in the input files.  It also modifies the
5830
linker's search strategy for archive libraries, to support the use of
5831
libraries specific to each particular architecture, by including in the
5832
search loop names suffixed with the string identifying the architecture.
5833
 
5834
For example, if your @command{ld} command line included @w{@samp{-ACA}} as
5835
well as @w{@samp{-ltry}}, the linker would look (in its built-in search
5836
paths, and in any paths you specify with @samp{-L}) for a library with
5837
the names
5838
 
5839
@smallexample
5840
@group
5841
try
5842
libtry.a
5843
tryca
5844
libtryca.a
5845
@end group
5846
@end smallexample
5847
 
5848
@noindent
5849
The first two possibilities would be considered in any event; the last
5850
two are due to the use of @w{@samp{-ACA}}.
5851
 
5852
You can meaningfully use @samp{-A} more than once on a command line, since
5853
the 960 architecture family allows combination of target architectures; each
5854
use will add another pair of name variants to search for when @w{@samp{-l}}
5855
specifies a library.
5856
 
5857
@cindex @option{--relax} on i960
5858
@cindex relaxing on i960
5859
@command{ld} supports the @samp{--relax} option for the i960 family.  If
5860
you specify @samp{--relax}, @command{ld} finds all @code{balx} and
5861
@code{calx} instructions whose targets are within 24 bits, and turns
5862
them into 24-bit program-counter relative @code{bal} and @code{cal}
5863
instructions, respectively.  @command{ld} also turns @code{cal}
5864
instructions into @code{bal} instructions when it determines that the
5865
target subroutine is a leaf routine (that is, the target subroutine does
5866
not itself call any subroutines).
5867
 
5868
@cindex Cortex-A8 erratum workaround
5869
@kindex --fix-cortex-a8
5870
@kindex --no-fix-cortex-a8
5871
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}.
5872
 
5873
The erratum only affects Thumb-2 code.  Please contact ARM for further details.
5874
 
5875
@ifclear GENERIC
5876
@lowersections
5877
@end ifclear
5878
@end ifset
5879
 
5880
@ifset ARM
5881
@ifclear GENERIC
5882
@raisesections
5883
@end ifclear
5884
 
5885
@ifset M68HC11
5886
@ifclear GENERIC
5887
@raisesections
5888
@end ifclear
5889
 
5890
@node M68HC11/68HC12
5891
@section @command{ld} and the Motorola 68HC11 and 68HC12 families
5892
 
5893
@cindex M68HC11 and 68HC12 support
5894
 
5895
@subsection Linker Relaxation
5896
 
5897
For the Motorola 68HC11, @command{ld} can perform these global
5898
optimizations when you specify the @samp{--relax} command-line option.
5899
 
5900
@table @emph
5901
@cindex relaxing on M68HC11
5902
@item relaxing address modes
5903
@command{ld} finds all @code{jsr} and @code{jmp} instructions whose
5904
targets are within eight bits, and turns them into eight-bit
5905
program-counter relative @code{bsr} and @code{bra} instructions,
5906
respectively.
5907
 
5908
@command{ld} also looks at all 16-bit extended addressing modes and
5909
transforms them in a direct addressing mode when the address is in
5910
page 0 (between 0 and 0x0ff).
5911
 
5912
@item relaxing gcc instruction group
5913
When @command{gcc} is called with @option{-mrelax}, it can emit group
5914
of instructions that the linker can optimize to use a 68HC11 direct
5915
addressing mode. These instructions consists of @code{bclr} or
5916
@code{bset} instructions.
5917
 
5918
@end table
5919
 
5920
@subsection Trampoline Generation
5921
 
5922
@cindex trampoline generation on M68HC11
5923
@cindex trampoline generation on M68HC12
5924
For 68HC11 and 68HC12, @command{ld} can generate trampoline code to
5925
call a far function using a normal @code{jsr} instruction. The linker
5926
will also change the relocation to some far function to use the
5927
trampoline address instead of the function address. This is typically the
5928
case when a pointer to a function is taken. The pointer will in fact
5929
point to the function trampoline.
5930
 
5931
@ifclear GENERIC
5932
@lowersections
5933
@end ifclear
5934
@end ifset
5935
 
5936
@node ARM
5937
@section @command{ld} and the ARM family
5938
 
5939
@cindex ARM interworking support
5940
@kindex --support-old-code
5941
For the ARM, @command{ld} will generate code stubs to allow functions calls
5942
between ARM and Thumb code.  These stubs only work with code that has
5943
been compiled and assembled with the @samp{-mthumb-interwork} command
5944
line option.  If it is necessary to link with old ARM object files or
5945
libraries, which have not been compiled with the -mthumb-interwork
5946
option then the @samp{--support-old-code} command line switch should be
5947
given to the linker.  This will make it generate larger stub functions
5948
which will work with non-interworking aware ARM code.  Note, however,
5949
the linker does not support generating stubs for function calls to
5950
non-interworking aware Thumb code.
5951
 
5952
@cindex thumb entry point
5953
@cindex entry point, thumb
5954
@kindex --thumb-entry=@var{entry}
5955
The @samp{--thumb-entry} switch is a duplicate of the generic
5956
@samp{--entry} switch, in that it sets the program's starting address.
5957
But it also sets the bottom bit of the address, so that it can be
5958
branched to using a BX instruction, and the program will start
5959
executing in Thumb mode straight away.
5960
 
5961
@cindex PE import table prefixing
5962
@kindex --use-nul-prefixed-import-tables
5963
The @samp{--use-nul-prefixed-import-tables} switch is specifying, that
5964
the import tables idata4 and idata5 have to be generated with a zero
5965
elememt prefix for import libraries. This is the old style to generate
5966
import tables. By default this option is turned off.
5967
 
5968
@cindex BE8
5969
@kindex --be8
5970
The @samp{--be8} switch instructs @command{ld} to generate BE8 format
5971
executables.  This option is only valid when linking big-endian objects.
5972
The resulting image will contain big-endian data and little-endian code.
5973
 
5974
@cindex TARGET1
5975
@kindex --target1-rel
5976
@kindex --target1-abs
5977
The @samp{R_ARM_TARGET1} relocation is typically used for entries in the
5978
@samp{.init_array} section.  It is interpreted as either @samp{R_ARM_REL32}
5979
or @samp{R_ARM_ABS32}, depending on the target.  The @samp{--target1-rel}
5980
and @samp{--target1-abs} switches override the default.
5981
 
5982
@cindex TARGET2
5983
@kindex --target2=@var{type}
5984
The @samp{--target2=type} switch overrides the default definition of the
5985
@samp{R_ARM_TARGET2} relocation.  Valid values for @samp{type}, their
5986
meanings, and target defaults are as follows:
5987
@table @samp
5988
@item rel
5989
@samp{R_ARM_REL32} (arm*-*-elf, arm*-*-eabi)
5990
@item abs
5991
@samp{R_ARM_ABS32} (arm*-*-symbianelf)
5992
@item got-rel
5993
@samp{R_ARM_GOT_PREL} (arm*-*-linux, arm*-*-*bsd)
5994
@end table
5995
 
5996
@cindex FIX_V4BX
5997
@kindex --fix-v4bx
5998
The @samp{R_ARM_V4BX} relocation (defined by the ARM AAELF
5999
specification) enables objects compiled for the ARMv4 architecture to be
6000
interworking-safe when linked with other objects compiled for ARMv4t, but
6001
also allows pure ARMv4 binaries to be built from the same ARMv4 objects.
6002
 
6003
In the latter case, the switch @option{--fix-v4bx} must be passed to the
6004
linker, which causes v4t @code{BX rM} instructions to be rewritten as
6005
@code{MOV PC,rM}, since v4 processors do not have a @code{BX} instruction.
6006
 
6007
In the former case, the switch should not be used, and @samp{R_ARM_V4BX}
6008
relocations are ignored.
6009
 
6010
@cindex FIX_V4BX_INTERWORKING
6011
@kindex --fix-v4bx-interworking
6012
Replace @code{BX rM} instructions identified by @samp{R_ARM_V4BX}
6013
relocations with a branch to the following veneer:
6014
 
6015
@smallexample
6016
TST rM, #1
6017
MOVEQ PC, rM
6018
BX Rn
6019
@end smallexample
6020
 
6021
This allows generation of libraries/applications that work on ARMv4 cores
6022
and are still interworking safe.  Note that the above veneer clobbers the
6023
condition flags, so may cause incorrect progrm behavior in rare cases.
6024
 
6025
@cindex USE_BLX
6026
@kindex --use-blx
6027
The @samp{--use-blx} switch enables the linker to use ARM/Thumb
6028
BLX instructions (available on ARMv5t and above) in various
6029
situations. Currently it is used to perform calls via the PLT from Thumb
6030
code using BLX rather than using BX and a mode-switching stub before
6031
each PLT entry. This should lead to such calls executing slightly faster.
6032
 
6033
This option is enabled implicitly for SymbianOS, so there is no need to
6034
specify it if you are using that target.
6035
 
6036
@cindex VFP11_DENORM_FIX
6037
@kindex --vfp11-denorm-fix
6038
The @samp{--vfp11-denorm-fix} switch enables a link-time workaround for a
6039
bug in certain VFP11 coprocessor hardware, which sometimes allows
6040
instructions with denorm operands (which must be handled by support code)
6041
to have those operands overwritten by subsequent instructions before
6042
the support code can read the intended values.
6043
 
6044
The bug may be avoided in scalar mode if you allow at least one
6045
intervening instruction between a VFP11 instruction which uses a register
6046
and another instruction which writes to the same register, or at least two
6047
intervening instructions if vector mode is in use. The bug only affects
6048
full-compliance floating-point mode: you do not need this workaround if
6049
you are using "runfast" mode. Please contact ARM for further details.
6050
 
6051
If you know you are using buggy VFP11 hardware, you can
6052
enable this workaround by specifying the linker option
6053
@samp{--vfp-denorm-fix=scalar} if you are using the VFP11 scalar
6054
mode only, or @samp{--vfp-denorm-fix=vector} if you are using
6055
vector mode (the latter also works for scalar code). The default is
6056
@samp{--vfp-denorm-fix=none}.
6057
 
6058
If the workaround is enabled, instructions are scanned for
6059
potentially-troublesome sequences, and a veneer is created for each
6060
such sequence which may trigger the erratum. The veneer consists of the
6061
first instruction of the sequence and a branch back to the subsequent
6062
instruction. The original instruction is then replaced with a branch to
6063
the veneer. The extra cycles required to call and return from the veneer
6064
are sufficient to avoid the erratum in both the scalar and vector cases.
6065
 
6066
@cindex NO_ENUM_SIZE_WARNING
6067
@kindex --no-enum-size-warning
6068
The @option{--no-enum-size-warning} switch prevents the linker from
6069
warning when linking object files that specify incompatible EABI
6070
enumeration size attributes.  For example, with this switch enabled,
6071
linking of an object file using 32-bit enumeration values with another
6072
using enumeration values fitted into the smallest possible space will
6073
not be diagnosed.
6074
 
6075
@cindex NO_WCHAR_SIZE_WARNING
6076
@kindex --no-wchar-size-warning
6077
The @option{--no-wchar-size-warning} switch prevents the linker from
6078
warning when linking object files that specify incompatible EABI
6079
@code{wchar_t} size attributes.  For example, with this switch enabled,
6080
linking of an object file using 32-bit @code{wchar_t} values with another
6081
using 16-bit @code{wchar_t} values will not be diagnosed.
6082
 
6083
@cindex PIC_VENEER
6084
@kindex --pic-veneer
6085
The @samp{--pic-veneer} switch makes the linker use PIC sequences for
6086
ARM/Thumb interworking veneers, even if the rest of the binary
6087
is not PIC.  This avoids problems on uClinux targets where
6088
@samp{--emit-relocs} is used to generate relocatable binaries.
6089
 
6090
@cindex STUB_GROUP_SIZE
6091
@kindex --stub-group-size=@var{N}
6092
The linker will automatically generate and insert small sequences of
6093
code into a linked ARM ELF executable whenever an attempt is made to
6094
perform a function call to a symbol that is too far away.  The
6095
placement of these sequences of instructions - called stubs - is
6096
controlled by the command line option @option{--stub-group-size=N}.
6097
The placement is important because a poor choice can create a need for
6098
duplicate stubs, increasing the code sizw.  The linker will try to
6099
group stubs together in order to reduce interruptions to the flow of
6100
code, but it needs guidance as to how big these groups should be and
6101
where they should be placed.
6102
 
6103
The value of @samp{N}, the parameter to the
6104
@option{--stub-group-size=} option controls where the stub groups are
6105
placed.  If it is negative then all stubs are placed after the first
6106
branch that needs them.  If it is positive then the stubs can be
6107
placed either before or after the branches that need them.  If the
6108
value of @samp{N} is 1 (either +1 or -1) then the linker will choose
6109
exactly where to place groups of stubs, using its built in heuristics.
6110
A value of @samp{N} greater than 1 (or smaller than -1) tells the
6111
linker that a single group of stubs can service at most @samp{N} bytes
6112
from the input sections.
6113
 
6114
The default, if @option{--stub-group-size=} is not specified, is
6115
@samp{N = +1}.
6116
 
6117
Farcalls stubs insertion is fully supported for the ARM-EABI target
6118
only, because it relies on object files properties not present
6119
otherwise.
6120
 
6121
@ifclear GENERIC
6122
@lowersections
6123
@end ifclear
6124
@end ifset
6125
 
6126
@ifset HPPA
6127
@ifclear GENERIC
6128
@raisesections
6129
@end ifclear
6130
 
6131
@node HPPA ELF32
6132
@section @command{ld} and HPPA 32-bit ELF Support
6133
@cindex HPPA multiple sub-space stubs
6134
@kindex --multi-subspace
6135
When generating a shared library, @command{ld} will by default generate
6136
import stubs suitable for use with a single sub-space application.
6137
The @samp{--multi-subspace} switch causes @command{ld} to generate export
6138
stubs, and different (larger) import stubs suitable for use with
6139
multiple sub-spaces.
6140
 
6141
@cindex HPPA stub grouping
6142
@kindex --stub-group-size=@var{N}
6143
Long branch stubs and import/export stubs are placed by @command{ld} in
6144
stub sections located between groups of input sections.
6145
@samp{--stub-group-size} specifies the maximum size of a group of input
6146
sections handled by one stub section.  Since branch offsets are signed,
6147
a stub section may serve two groups of input sections, one group before
6148
the stub section, and one group after it.  However, when using
6149
conditional branches that require stubs, it may be better (for branch
6150
prediction) that stub sections only serve one group of input sections.
6151
A negative value for @samp{N} chooses this scheme, ensuring that
6152
branches to stubs always use a negative offset.  Two special values of
6153
@samp{N} are recognized, @samp{1} and @samp{-1}.  These both instruct
6154
@command{ld} to automatically size input section groups for the branch types
6155
detected, with the same behaviour regarding stub placement as other
6156
positive or negative values of @samp{N} respectively.
6157
 
6158
Note that @samp{--stub-group-size} does not split input sections.  A
6159
single input section larger than the group size specified will of course
6160
create a larger group (of one section).  If input sections are too
6161
large, it may not be possible for a branch to reach its stub.
6162
 
6163
@ifclear GENERIC
6164
@lowersections
6165
@end ifclear
6166
@end ifset
6167
 
6168
@ifset M68K
6169
@ifclear GENERIC
6170
@raisesections
6171
@end ifclear
6172
 
6173
@node M68K
6174
@section @command{ld} and the Motorola 68K family
6175
 
6176
@cindex Motorola 68K GOT generation
6177
@kindex --got=@var{type}
6178
The @samp{--got=@var{type}} option lets you choose the GOT generation scheme.
6179
The choices are @samp{single}, @samp{negative}, @samp{multigot} and
6180
@samp{target}.  When @samp{target} is selected the linker chooses
6181
the default GOT generation scheme for the current target.
6182
@samp{single} tells the linker to generate a single GOT with
6183
entries only at non-negative offsets.
6184
@samp{negative} instructs the linker to generate a single GOT with
6185
entries at both negative and positive offsets.  Not all environments
6186
support such GOTs.
6187
@samp{multigot} allows the linker to generate several GOTs in the
6188
output file.  All GOT references from a single input object
6189
file access the same GOT, but references from different input object
6190
files might access different GOTs.  Not all environments support such GOTs.
6191
 
6192
@ifclear GENERIC
6193
@lowersections
6194
@end ifclear
6195
@end ifset
6196
 
6197
@ifset MMIX
6198
@ifclear GENERIC
6199
@raisesections
6200
@end ifclear
6201
 
6202
@node MMIX
6203
@section @code{ld} and MMIX
6204
For MMIX, there is a choice of generating @code{ELF} object files or
6205
@code{mmo} object files when linking.  The simulator @code{mmix}
6206
understands the @code{mmo} format.  The binutils @code{objcopy} utility
6207
can translate between the two formats.
6208
 
6209
There is one special section, the @samp{.MMIX.reg_contents} section.
6210
Contents in this section is assumed to correspond to that of global
6211
registers, and symbols referring to it are translated to special symbols,
6212
equal to registers.  In a final link, the start address of the
6213
@samp{.MMIX.reg_contents} section corresponds to the first allocated
6214
global register multiplied by 8.  Register @code{$255} is not included in
6215
this section; it is always set to the program entry, which is at the
6216
symbol @code{Main} for @code{mmo} files.
6217
 
6218
Global symbols with the prefix @code{__.MMIX.start.}, for example
6219
@code{__.MMIX.start..text} and @code{__.MMIX.start..data} are special.
6220
The default linker script uses these to set the default start address
6221
of a section.
6222
 
6223
Initial and trailing multiples of zero-valued 32-bit words in a section,
6224
are left out from an mmo file.
6225
 
6226
@ifclear GENERIC
6227
@lowersections
6228
@end ifclear
6229
@end ifset
6230
 
6231
@ifset MSP430
6232
@ifclear GENERIC
6233
@raisesections
6234
@end ifclear
6235
 
6236
@node  MSP430
6237
@section @code{ld} and MSP430
6238
For the MSP430 it is possible to select the MPU architecture.  The flag @samp{-m [mpu type]}
6239
will select an appropriate linker script for selected MPU type.  (To get a list of known MPUs
6240
just pass @samp{-m help} option to the linker).
6241
 
6242
@cindex MSP430 extra sections
6243
The linker will recognize some extra sections which are MSP430 specific:
6244
 
6245
@table @code
6246
@item @samp{.vectors}
6247
Defines a portion of ROM where interrupt vectors located.
6248
 
6249
@item @samp{.bootloader}
6250
Defines the bootloader portion of the ROM (if applicable).  Any code
6251
in this section will be uploaded to the MPU.
6252
 
6253
@item @samp{.infomem}
6254
Defines an information memory section (if applicable).  Any code in
6255
this section will be uploaded to the MPU.
6256
 
6257
@item @samp{.infomemnobits}
6258
This is the same as the @samp{.infomem} section except that any code
6259
in this section will not be uploaded to the MPU.
6260
 
6261
@item @samp{.noinit}
6262
Denotes a portion of RAM located above @samp{.bss} section.
6263
 
6264
The last two sections are used by gcc.
6265
@end table
6266
 
6267
@ifclear GENERIC
6268
@lowersections
6269
@end ifclear
6270
@end ifset
6271
 
6272
@ifset POWERPC
6273
@ifclear GENERIC
6274
@raisesections
6275
@end ifclear
6276
 
6277
@node PowerPC ELF32
6278
@section @command{ld} and PowerPC 32-bit ELF Support
6279
@cindex PowerPC long branches
6280
@kindex --relax on PowerPC
6281
Branches on PowerPC processors are limited to a signed 26-bit
6282
displacement, which may result in @command{ld} giving
6283
@samp{relocation truncated to fit} errors with very large programs.
6284
@samp{--relax} enables the generation of trampolines that can access
6285
the entire 32-bit address space.  These trampolines are inserted at
6286
section boundaries, so may not themselves be reachable if an input
6287
section exceeds 33M in size.  You may combine @samp{-r} and
6288
@samp{--relax} to add trampolines in a partial link.  In that case
6289
both branches to undefined symbols and inter-section branches are also
6290
considered potentially out of range, and trampolines inserted.
6291
 
6292
@cindex PowerPC ELF32 options
6293
@table @option
6294
@cindex PowerPC PLT
6295
@kindex --bss-plt
6296
@item --bss-plt
6297
Current PowerPC GCC accepts a @samp{-msecure-plt} option that
6298
generates code capable of using a newer PLT and GOT layout that has
6299
the security advantage of no executable section ever needing to be
6300
writable and no writable section ever being executable.  PowerPC
6301
@command{ld} will generate this layout, including stubs to access the
6302
PLT, if all input files (including startup and static libraries) were
6303
compiled with @samp{-msecure-plt}.  @samp{--bss-plt} forces the old
6304
BSS PLT (and GOT layout) which can give slightly better performance.
6305
 
6306
@kindex --secure-plt
6307
@item --secure-plt
6308
@command{ld} will use the new PLT and GOT layout if it is linking new
6309
@samp{-fpic} or @samp{-fPIC} code, but does not do so automatically
6310
when linking non-PIC code.  This option requests the new PLT and GOT
6311
layout.  A warning will be given if some object file requires the old
6312
style BSS PLT.
6313
 
6314
@cindex PowerPC GOT
6315
@kindex --sdata-got
6316
@item --sdata-got
6317
The new secure PLT and GOT are placed differently relative to other
6318
sections compared to older BSS PLT and GOT placement.  The location of
6319
@code{.plt} must change because the new secure PLT is an initialized
6320
section while the old PLT is uninitialized.  The reason for the
6321
@code{.got} change is more subtle:  The new placement allows
6322
@code{.got} to be read-only in applications linked with
6323
@samp{-z relro -z now}.  However, this placement means that
6324
@code{.sdata} cannot always be used in shared libraries, because the
6325
PowerPC ABI accesses @code{.sdata} in shared libraries from the GOT
6326
pointer.  @samp{--sdata-got} forces the old GOT placement.  PowerPC
6327
GCC doesn't use @code{.sdata} in shared libraries, so this option is
6328
really only useful for other compilers that may do so.
6329
 
6330
@cindex PowerPC stub symbols
6331
@kindex --emit-stub-syms
6332
@item --emit-stub-syms
6333
This option causes @command{ld} to label linker stubs with a local
6334
symbol that encodes the stub type and destination.
6335
 
6336
@cindex PowerPC TLS optimization
6337
@kindex --no-tls-optimize
6338
@item --no-tls-optimize
6339
PowerPC @command{ld} normally performs some optimization of code
6340
sequences used to access Thread-Local Storage.  Use this option to
6341
disable the optimization.
6342
@end table
6343
 
6344
@ifclear GENERIC
6345
@lowersections
6346
@end ifclear
6347
@end ifset
6348
 
6349
@ifset POWERPC64
6350
@ifclear GENERIC
6351
@raisesections
6352
@end ifclear
6353
 
6354
@node PowerPC64 ELF64
6355
@section @command{ld} and PowerPC64 64-bit ELF Support
6356
 
6357
@cindex PowerPC64 ELF64 options
6358
@table @option
6359
@cindex PowerPC64 stub grouping
6360
@kindex --stub-group-size
6361
@item --stub-group-size
6362
Long branch stubs, PLT call stubs  and TOC adjusting stubs are placed
6363
by @command{ld} in stub sections located between groups of input sections.
6364
@samp{--stub-group-size} specifies the maximum size of a group of input
6365
sections handled by one stub section.  Since branch offsets are signed,
6366
a stub section may serve two groups of input sections, one group before
6367
the stub section, and one group after it.  However, when using
6368
conditional branches that require stubs, it may be better (for branch
6369
prediction) that stub sections only serve one group of input sections.
6370
A negative value for @samp{N} chooses this scheme, ensuring that
6371
branches to stubs always use a negative offset.  Two special values of
6372
@samp{N} are recognized, @samp{1} and @samp{-1}.  These both instruct
6373
@command{ld} to automatically size input section groups for the branch types
6374
detected, with the same behaviour regarding stub placement as other
6375
positive or negative values of @samp{N} respectively.
6376
 
6377
Note that @samp{--stub-group-size} does not split input sections.  A
6378
single input section larger than the group size specified will of course
6379
create a larger group (of one section).  If input sections are too
6380
large, it may not be possible for a branch to reach its stub.
6381
 
6382
@cindex PowerPC64 stub symbols
6383
@kindex --emit-stub-syms
6384
@item --emit-stub-syms
6385
This option causes @command{ld} to label linker stubs with a local
6386
symbol that encodes the stub type and destination.
6387
 
6388
@cindex PowerPC64 dot symbols
6389
@kindex --dotsyms
6390
@kindex --no-dotsyms
6391
@item --dotsyms, --no-dotsyms
6392
These two options control how @command{ld} interprets version patterns
6393
in a version script.  Older PowerPC64 compilers emitted both a
6394
function descriptor symbol with the same name as the function, and a
6395
code entry symbol with the name prefixed by a dot (@samp{.}).  To
6396
properly version a function @samp{foo}, the version script thus needs
6397
to control both @samp{foo} and @samp{.foo}.  The option
6398
@samp{--dotsyms}, on by default, automatically adds the required
6399
dot-prefixed patterns.  Use @samp{--no-dotsyms} to disable this
6400
feature.
6401
 
6402
@cindex PowerPC64 TLS optimization
6403
@kindex --no-tls-optimize
6404
@item --no-tls-optimize
6405
PowerPC64 @command{ld} normally performs some optimization of code
6406
sequences used to access Thread-Local Storage.  Use this option to
6407
disable the optimization.
6408
 
6409
@cindex PowerPC64 OPD optimization
6410
@kindex --no-opd-optimize
6411
@item --no-opd-optimize
6412
PowerPC64 @command{ld} normally removes @code{.opd} section entries
6413
corresponding to deleted link-once functions, or functions removed by
6414
the action of @samp{--gc-sections} or linker script @code{/DISCARD/}.
6415
Use this option to disable @code{.opd} optimization.
6416
 
6417
@cindex PowerPC64 OPD spacing
6418
@kindex --non-overlapping-opd
6419
@item --non-overlapping-opd
6420
Some PowerPC64 compilers have an option to generate compressed
6421
@code{.opd} entries spaced 16 bytes apart, overlapping the third word,
6422
the static chain pointer (unused in C) with the first word of the next
6423
entry.  This option expands such entries to the full 24 bytes.
6424
 
6425
@cindex PowerPC64 TOC optimization
6426
@kindex --no-toc-optimize
6427
@item --no-toc-optimize
6428
PowerPC64 @command{ld} normally removes unused @code{.toc} section
6429
entries.  Such entries are detected by examining relocations that
6430
reference the TOC in code sections.  A reloc in a deleted code section
6431
marks a TOC word as unneeded, while a reloc in a kept code section
6432
marks a TOC word as needed.  Since the TOC may reference itself, TOC
6433
relocs are also examined.  TOC words marked as both needed and
6434
unneeded will of course be kept.  TOC words without any referencing
6435
reloc are assumed to be part of a multi-word entry, and are kept or
6436
discarded as per the nearest marked preceding word.  This works
6437
reliably for compiler generated code, but may be incorrect if assembly
6438
code is used to insert TOC entries.  Use this option to disable the
6439
optimization.
6440
 
6441
@cindex PowerPC64 multi-TOC
6442
@kindex --no-multi-toc
6443
@item --no-multi-toc
6444
By default, PowerPC64 GCC generates code for a TOC model where TOC
6445
entries are accessed with a 16-bit offset from r2.  This limits the
6446
total TOC size to 64K.  PowerPC64 @command{ld} extends this limit by
6447
grouping code sections such that each group uses less than 64K for its
6448
TOC entries, then inserts r2 adjusting stubs between inter-group
6449
calls.  @command{ld} does not split apart input sections, so cannot
6450
help if a single input file has a @code{.toc} section that exceeds
6451
64K, most likely from linking multiple files with @command{ld -r}.
6452
Use this option to turn off this feature.
6453
@end table
6454
 
6455
@ifclear GENERIC
6456
@lowersections
6457
@end ifclear
6458
@end ifset
6459
 
6460
@ifset SPU
6461
@ifclear GENERIC
6462
@raisesections
6463
@end ifclear
6464
 
6465
@node SPU ELF
6466
@section @command{ld} and SPU ELF Support
6467
 
6468
@cindex SPU ELF options
6469
@table @option
6470
 
6471
@cindex SPU plugins
6472
@kindex --plugin
6473
@item --plugin
6474
This option marks an executable as a PIC plugin module.
6475
 
6476
@cindex SPU overlays
6477
@kindex --no-overlays
6478
@item --no-overlays
6479
Normally, @command{ld} recognizes calls to functions within overlay
6480
regions, and redirects such calls to an overlay manager via a stub.
6481
@command{ld} also provides a built-in overlay manager.  This option
6482
turns off all this special overlay handling.
6483
 
6484
@cindex SPU overlay stub symbols
6485
@kindex --emit-stub-syms
6486
@item --emit-stub-syms
6487
This option causes @command{ld} to label overlay stubs with a local
6488
symbol that encodes the stub type and destination.
6489
 
6490
@cindex SPU extra overlay stubs
6491
@kindex --extra-overlay-stubs
6492
@item --extra-overlay-stubs
6493
This option causes @command{ld} to add overlay call stubs on all
6494
function calls out of overlay regions.  Normally stubs are not added
6495
on calls to non-overlay regions.
6496
 
6497
@cindex SPU local store size
6498
@kindex --local-store=lo:hi
6499
@item --local-store=lo:hi
6500
@command{ld} usually checks that a final executable for SPU fits in
6501
the address range 0 to 256k.  This option may be used to change the
6502
range.  Disable the check entirely with @option{--local-store=0:0}.
6503
 
6504
@cindex SPU
6505
@kindex --stack-analysis
6506
@item --stack-analysis
6507
SPU local store space is limited.  Over-allocation of stack space
6508
unnecessarily limits space available for code and data, while
6509
under-allocation results in runtime failures.  If given this option,
6510
@command{ld} will provide an estimate of maximum stack usage.
6511
@command{ld} does this by examining symbols in code sections to
6512
determine the extents of functions, and looking at function prologues
6513
for stack adjusting instructions.  A call-graph is created by looking
6514
for relocations on branch instructions.  The graph is then searched
6515
for the maximum stack usage path.  Note that this analysis does not
6516
find calls made via function pointers, and does not handle recursion
6517
and other cycles in the call graph.  Stack usage may be
6518
under-estimated if your code makes such calls.  Also, stack usage for
6519
dynamic allocation, e.g. alloca, will not be detected.  If a link map
6520
is requested, detailed information about each function's stack usage
6521
and calls will be given.
6522
 
6523
@cindex SPU
6524
@kindex --emit-stack-syms
6525
@item --emit-stack-syms
6526
This option, if given along with @option{--stack-analysis} will result
6527
in @command{ld} emitting stack sizing symbols for each function.
6528
These take the form @code{__stack_<function_name>} for global
6529
functions, and @code{__stack_<number>_<function_name>} for static
6530
functions.  @code{<number>} is the section id in hex.  The value of
6531
such symbols is the stack requirement for the corresponding function.
6532
The symbol size will be zero, type @code{STT_NOTYPE}, binding
6533
@code{STB_LOCAL}, and section @code{SHN_ABS}.
6534
@end table
6535
 
6536
@ifclear GENERIC
6537
@lowersections
6538
@end ifclear
6539
@end ifset
6540
 
6541
@ifset TICOFF
6542
@ifclear GENERIC
6543
@raisesections
6544
@end ifclear
6545
 
6546
@node TI COFF
6547
@section @command{ld}'s Support for Various TI COFF Versions
6548
@cindex TI COFF versions
6549
@kindex --format=@var{version}
6550
The @samp{--format} switch allows selection of one of the various
6551
TI COFF versions.  The latest of this writing is 2; versions 0 and 1 are
6552
also supported.  The TI COFF versions also vary in header byte-order
6553
format; @command{ld} will read any version or byte order, but the output
6554
header format depends on the default specified by the specific target.
6555
 
6556
@ifclear GENERIC
6557
@lowersections
6558
@end ifclear
6559
@end ifset
6560
 
6561
@ifset WIN32
6562
@ifclear GENERIC
6563
@raisesections
6564
@end ifclear
6565
 
6566
@node WIN32
6567
@section @command{ld} and WIN32 (cygwin/mingw)
6568
 
6569
This section describes some of the win32 specific @command{ld} issues.
6570
See @ref{Options,,Command Line Options} for detailed description of the
6571
command line options mentioned here.
6572
 
6573
@table @emph
6574
@cindex import libraries
6575
@item import libraries
6576
The standard Windows linker creates and uses so-called import
6577
libraries, which contains information for linking to dll's.  They are
6578
regular static archives and are handled as any other static
6579
archive.  The cygwin and mingw ports of @command{ld} have specific
6580
support for creating such libraries provided with the
6581
@samp{--out-implib} command line option.
6582
 
6583
@item   exporting DLL symbols
6584
@cindex exporting DLL symbols
6585
The cygwin/mingw @command{ld} has several ways to export symbols for dll's.
6586
 
6587
@table @emph
6588
@item   using auto-export functionality
6589
@cindex using auto-export functionality
6590
By default @command{ld} exports symbols with the auto-export functionality,
6591
which is controlled by the following command line options:
6592
 
6593
@itemize
6594
@item --export-all-symbols   [This is the default]
6595
@item --exclude-symbols
6596
@item --exclude-libs
6597
@item --exclude-modules-for-implib
6598
@item --version-script
6599
@end itemize
6600
 
6601
When auto-export is in operation, @command{ld} will export all the non-local
6602
(global and common) symbols it finds in a DLL, with the exception of a few
6603
symbols known to belong to the system's runtime and libraries.  As it will
6604
often not be desirable to export all of a DLL's symbols, which may include
6605
private functions that are not part of any public interface, the command-line
6606
options listed above may be used to filter symbols out from the list for
6607
exporting.  The @samp{--output-def} option can be used in order to see the
6608
final list of exported symbols with all exclusions taken into effect.
6609
 
6610
If @samp{--export-all-symbols} is not given explicitly on the
6611
command line, then the default auto-export behavior will be @emph{disabled}
6612
if either of the following are true:
6613
 
6614
@itemize
6615
@item A DEF file is used.
6616
@item Any symbol in any object file was marked with the __declspec(dllexport) attribute.
6617
@end itemize
6618
 
6619
@item   using a DEF file
6620
@cindex using a DEF file
6621
Another way of exporting symbols is using a DEF file.  A DEF file is
6622
an ASCII file containing definitions of symbols which should be
6623
exported when a dll is created.  Usually it is named @samp{<dll
6624
name>.def} and is added as any other object file to the linker's
6625
command line.  The file's name must end in @samp{.def} or @samp{.DEF}.
6626
 
6627
@example
6628
gcc -o <output> <objectfiles> <dll name>.def
6629
@end example
6630
 
6631
Using a DEF file turns off the normal auto-export behavior, unless the
6632
@samp{--export-all-symbols} option is also used.
6633
 
6634
Here is an example of a DEF file for a shared library called @samp{xyz.dll}:
6635
 
6636
@example
6637
LIBRARY "xyz.dll" BASE=0x20000000
6638
 
6639
EXPORTS
6640
foo
6641
bar
6642
_bar = bar
6643
another_foo = abc.dll.afoo
6644
var1 DATA
6645
@end example
6646
 
6647
This example defines a DLL with a non-default base address and five
6648
symbols in the export table. The third exported symbol @code{_bar} is an
6649
alias for the second. The fourth symbol, @code{another_foo} is resolved
6650
by "forwarding" to another module and treating it as an alias for
6651
@code{afoo} exported from the DLL @samp{abc.dll}. The final symbol
6652
@code{var1} is declared to be a data object.
6653
 
6654
The optional @code{LIBRARY <name>} command indicates the @emph{internal}
6655
name of the output DLL. If @samp{<name>} does not include a suffix,
6656
the default library suffix, @samp{.DLL} is appended.
6657
 
6658
When the .DEF file is used to build an application, rather than a
6659
library, the @code{NAME <name>} command should be used instead of
6660
@code{LIBRARY}. If @samp{<name>} does not include a suffix, the default
6661
executable suffix, @samp{.EXE} is appended.
6662
 
6663
With either @code{LIBRARY <name>} or @code{NAME <name>} the optional
6664
specification @code{BASE = <number>} may be used to specify a
6665
non-default base address for the image.
6666
 
6667
If neither @code{LIBRARY <name>} nor  @code{NAME <name>} is specified,
6668
or they specify an empty string, the internal name is the same as the
6669
filename specified on the command line.
6670
 
6671
The complete specification of an export symbol is:
6672
 
6673
@example
6674
EXPORTS
6675
  ( (  ( <name1> [ = <name2> ] )
6676
     | ( <name1> = <module-name> . <external-name>))
6677
  [ @@ <integer> ] [NONAME] [DATA] [CONSTANT] [PRIVATE] ) *
6678
@end example
6679
 
6680
Declares @samp{<name1>} as an exported symbol from the DLL, or declares
6681
@samp{<name1>} as an exported alias for @samp{<name2>}; or declares
6682
@samp{<name1>} as a "forward" alias for the symbol
6683
@samp{<external-name>} in the DLL @samp{<module-name>}.
6684
Optionally, the symbol may be exported by the specified ordinal
6685
@samp{<integer>} alias.
6686
 
6687
The optional keywords that follow the declaration indicate:
6688
 
6689
@code{NONAME}: Do not put the symbol name in the DLL's export table.  It
6690
will still be exported by its ordinal alias (either the value specified
6691
by the .def specification or, otherwise, the value assigned by the
6692
linker). The symbol name, however, does remain visible in the import
6693
library (if any), unless @code{PRIVATE} is also specified.
6694
 
6695
@code{DATA}: The symbol is a variable or object, rather than a function.
6696
The import lib will export only an indirect reference to @code{foo} as
6697
the symbol @code{_imp__foo} (ie, @code{foo} must be resolved as
6698
@code{*_imp__foo}).
6699
 
6700
@code{CONSTANT}: Like @code{DATA}, but put the undecorated @code{foo} as
6701
well as @code{_imp__foo} into the import library. Both refer to the
6702
read-only import address table's pointer to the variable, not to the
6703
variable itself. This can be dangerous. If the user code fails to add
6704
the @code{dllimport} attribute and also fails to explicitly add the
6705
extra indirection that the use of the attribute enforces, the
6706
application will behave unexpectedly.
6707
 
6708
@code{PRIVATE}: Put the symbol in the DLL's export table, but do not put
6709
it into the static import library used to resolve imports at link time. The
6710
symbol can still be imported using the @code{LoadLibrary/GetProcAddress}
6711
API at runtime or by by using the GNU ld extension of linking directly to
6712
the DLL without an import library.
6713
 
6714
See ld/deffilep.y in the binutils sources for the full specification of
6715
other DEF file statements
6716
 
6717
@cindex creating a DEF file
6718
While linking a shared dll, @command{ld} is able to create a DEF file
6719
with the @samp{--output-def <file>} command line option.
6720
 
6721
@item   Using decorations
6722
@cindex Using decorations
6723
Another way of marking symbols for export is to modify the source code
6724
itself, so that when building the DLL each symbol to be exported is
6725
declared as:
6726
 
6727
@example
6728
__declspec(dllexport) int a_variable
6729
__declspec(dllexport) void a_function(int with_args)
6730
@end example
6731
 
6732
All such symbols will be exported from the DLL.  If, however,
6733
any of the object files in the DLL contain symbols decorated in
6734
this way, then the normal auto-export behavior is disabled, unless
6735
the @samp{--export-all-symbols} option is also used.
6736
 
6737
Note that object files that wish to access these symbols must @emph{not}
6738
decorate them with dllexport.  Instead, they should use dllimport,
6739
instead:
6740
 
6741
@example
6742
__declspec(dllimport) int a_variable
6743
__declspec(dllimport) void a_function(int with_args)
6744
@end example
6745
 
6746
This complicates the structure of library header files, because
6747
when included by the library itself the header must declare the
6748
variables and functions as dllexport, but when included by client
6749
code the header must declare them as dllimport.  There are a number
6750
of idioms that are typically used to do this; often client code can
6751
omit the __declspec() declaration completely.  See
6752
@samp{--enable-auto-import} and @samp{automatic data imports} for more
6753
information.
6754
@end table
6755
 
6756
@cindex automatic data imports
6757
@item automatic data imports
6758
The standard Windows dll format supports data imports from dlls only
6759
by adding special decorations (dllimport/dllexport), which let the
6760
compiler produce specific assembler instructions to deal with this
6761
issue.  This increases the effort necessary to port existing Un*x
6762
code to these platforms, especially for large
6763
c++ libraries and applications.  The auto-import feature, which was
6764
initially provided by Paul Sokolovsky, allows one to omit the
6765
decorations to achieve a behavior that conforms to that on POSIX/Un*x
6766
platforms. This feature is enabled with the @samp{--enable-auto-import}
6767
command-line option, although it is enabled by default on cygwin/mingw.
6768
The @samp{--enable-auto-import} option itself now serves mainly to
6769
suppress any warnings that are ordinarily emitted when linked objects
6770
trigger the feature's use.
6771
 
6772
auto-import of variables does not always work flawlessly without
6773
additional assistance.  Sometimes, you will see this message
6774
 
6775
"variable '<var>' can't be auto-imported. Please read the
6776
documentation for ld's @code{--enable-auto-import} for details."
6777
 
6778
The @samp{--enable-auto-import} documentation explains why this error
6779
occurs, and several methods that can be used to overcome this difficulty.
6780
One of these methods is the @emph{runtime pseudo-relocs} feature, described
6781
below.
6782
 
6783
@cindex runtime pseudo-relocation
6784
For complex variables imported from DLLs (such as structs or classes),
6785
object files typically contain a base address for the variable and an
6786
offset (@emph{addend}) within the variable--to specify a particular
6787
field or public member, for instance.  Unfortunately, the runtime loader used
6788
in win32 environments is incapable of fixing these references at runtime
6789
without the additional information supplied by dllimport/dllexport decorations.
6790
The standard auto-import feature described above is unable to resolve these
6791
references.
6792
 
6793
The @samp{--enable-runtime-pseudo-relocs} switch allows these references to
6794
be resolved without error, while leaving the task of adjusting the references
6795
themselves (with their non-zero addends) to specialized code provided by the
6796
runtime environment.  Recent versions of the cygwin and mingw environments and
6797
compilers provide this runtime support; older versions do not.  However, the
6798
support is only necessary on the developer's platform; the compiled result will
6799
run without error on an older system.
6800
 
6801
@samp{--enable-runtime-pseudo-relocs} is not the default; it must be explicitly
6802
enabled as needed.
6803
 
6804
@cindex direct linking to a dll
6805
@item direct linking to a dll
6806
The cygwin/mingw ports of @command{ld} support the direct linking,
6807
including data symbols, to a dll without the usage of any import
6808
libraries.  This is much faster and uses much less memory than does the
6809
traditional import library method, especially when linking large
6810
libraries or applications.  When @command{ld} creates an import lib, each
6811
function or variable exported from the dll is stored in its own bfd, even
6812
though a single bfd could contain many exports.  The overhead involved in
6813
storing, loading, and processing so many bfd's is quite large, and explains the
6814
tremendous time, memory, and storage needed to link against particularly
6815
large or complex libraries when using import libs.
6816
 
6817
Linking directly to a dll uses no extra command-line switches other than
6818
@samp{-L} and @samp{-l}, because @command{ld} already searches for a number
6819
of names to match each library.  All that is needed from the developer's
6820
perspective is an understanding of this search, in order to force ld to
6821
select the dll instead of an import library.
6822
 
6823
 
6824
For instance, when ld is called with the argument @samp{-lxxx} it will attempt
6825
to find, in the first directory of its search path,
6826
 
6827
@example
6828
libxxx.dll.a
6829
xxx.dll.a
6830
libxxx.a
6831
xxx.lib
6832
cygxxx.dll (*)
6833
libxxx.dll
6834
xxx.dll
6835
@end example
6836
 
6837
before moving on to the next directory in the search path.
6838
 
6839
(*) Actually, this is not @samp{cygxxx.dll} but in fact is @samp{<prefix>xxx.dll},
6840
where @samp{<prefix>} is set by the @command{ld} option
6841
@samp{--dll-search-prefix=<prefix>}. In the case of cygwin, the standard gcc spec
6842
file includes @samp{--dll-search-prefix=cyg}, so in effect we actually search for
6843
@samp{cygxxx.dll}.
6844
 
6845
Other win32-based unix environments, such as mingw or pw32, may use other
6846
@samp{<prefix>}es, although at present only cygwin makes use of this feature.  It
6847
was originally intended to help avoid name conflicts among dll's built for the
6848
various win32/un*x environments, so that (for example) two versions of a zlib dll
6849
could coexist on the same machine.
6850
 
6851
The generic cygwin/mingw path layout uses a @samp{bin} directory for
6852
applications and dll's and a @samp{lib} directory for the import
6853
libraries (using cygwin nomenclature):
6854
 
6855
@example
6856
bin/
6857
        cygxxx.dll
6858
lib/
6859
        libxxx.dll.a   (in case of dll's)
6860
        libxxx.a       (in case of static archive)
6861
@end example
6862
 
6863
Linking directly to a dll without using the import library can be
6864
done two ways:
6865
 
6866
1. Use the dll directly by adding the @samp{bin} path to the link line
6867
@example
6868
gcc -Wl,-verbose  -o a.exe -L../bin/ -lxxx
6869
@end example
6870
 
6871
However, as the dll's often have version numbers appended to their names
6872
(@samp{cygncurses-5.dll}) this will often fail, unless one specifies
6873
@samp{-L../bin -lncurses-5} to include the version.  Import libs are generally
6874
not versioned, and do not have this difficulty.
6875
 
6876
2. Create a symbolic link from the dll to a file in the @samp{lib}
6877
directory according to the above mentioned search pattern.  This
6878
should be used to avoid unwanted changes in the tools needed for
6879
making the app/dll.
6880
 
6881
@example
6882
ln -s bin/cygxxx.dll lib/[cyg|lib|]xxx.dll[.a]
6883
@end example
6884
 
6885
Then you can link without any make environment changes.
6886
 
6887
@example
6888
gcc -Wl,-verbose  -o a.exe -L../lib/ -lxxx
6889
@end example
6890
 
6891
This technique also avoids the version number problems, because the following is
6892
perfectly legal
6893
 
6894
@example
6895
bin/
6896
        cygxxx-5.dll
6897
lib/
6898
        libxxx.dll.a -> ../bin/cygxxx-5.dll
6899
@end example
6900
 
6901
Linking directly to a dll without using an import lib will work
6902
even when auto-import features are exercised, and even when
6903
@samp{--enable-runtime-pseudo-relocs} is used.
6904
 
6905
Given the improvements in speed and memory usage, one might justifiably
6906
wonder why import libraries are used at all.  There are three reasons:
6907
 
6908
1. Until recently, the link-directly-to-dll functionality did @emph{not}
6909
work with auto-imported data.
6910
 
6911
2. Sometimes it is necessary to include pure static objects within the
6912
import library (which otherwise contains only bfd's for indirection
6913
symbols that point to the exports of a dll).  Again, the import lib
6914
for the cygwin kernel makes use of this ability, and it is not
6915
possible to do this without an import lib.
6916
 
6917
3. Symbol aliases can only be resolved using an import lib.  This is
6918
critical when linking against OS-supplied dll's (eg, the win32 API)
6919
in which symbols are usually exported as undecorated aliases of their
6920
stdcall-decorated assembly names.
6921
 
6922
So, import libs are not going away.  But the ability to replace
6923
true import libs with a simple symbolic link to (or a copy of)
6924
a dll, in many cases, is a useful addition to the suite of tools
6925
binutils makes available to the win32 developer.  Given the
6926
massive improvements in memory requirements during linking, storage
6927
requirements, and linking speed, we expect that many developers
6928
will soon begin to use this feature whenever possible.
6929
 
6930
@item symbol aliasing
6931
@table @emph
6932
@item adding additional names
6933
Sometimes, it is useful to export symbols with additional names.
6934
A symbol @samp{foo} will be exported as @samp{foo}, but it can also be
6935
exported as @samp{_foo} by using special directives in the DEF file
6936
when creating the dll.  This will affect also the optional created
6937
import library.  Consider the following DEF file:
6938
 
6939
@example
6940
LIBRARY "xyz.dll" BASE=0x61000000
6941
 
6942
EXPORTS
6943
foo
6944
_foo = foo
6945
@end example
6946
 
6947
The line @samp{_foo = foo} maps the symbol @samp{foo} to @samp{_foo}.
6948
 
6949
Another method for creating a symbol alias is to create it in the
6950
source code using the "weak" attribute:
6951
 
6952
@example
6953
void foo () @{ /* Do something.  */; @}
6954
void _foo () __attribute__ ((weak, alias ("foo")));
6955
@end example
6956
 
6957
See the gcc manual for more information about attributes and weak
6958
symbols.
6959
 
6960
@item renaming symbols
6961
Sometimes it is useful to rename exports.  For instance, the cygwin
6962
kernel does this regularly.  A symbol @samp{_foo} can be exported as
6963
@samp{foo} but not as @samp{_foo} by using special directives in the
6964
DEF file. (This will also affect the import library, if it is
6965
created).  In the following example:
6966
 
6967
@example
6968
LIBRARY "xyz.dll" BASE=0x61000000
6969
 
6970
EXPORTS
6971
_foo = foo
6972
@end example
6973
 
6974
The line @samp{_foo = foo} maps the exported symbol @samp{foo} to
6975
@samp{_foo}.
6976
@end table
6977
 
6978
Note: using a DEF file disables the default auto-export behavior,
6979
unless the @samp{--export-all-symbols} command line option is used.
6980
If, however, you are trying to rename symbols, then you should list
6981
@emph{all} desired exports in the DEF file, including the symbols
6982
that are not being renamed, and do @emph{not} use the
6983
@samp{--export-all-symbols} option.  If you list only the
6984
renamed symbols in the DEF file, and use @samp{--export-all-symbols}
6985
to handle the other symbols, then the both the new names @emph{and}
6986
the original names for the renamed symbols will be exported.
6987
In effect, you'd be aliasing those symbols, not renaming them,
6988
which is probably not what you wanted.
6989
 
6990
@cindex weak externals
6991
@item weak externals
6992
The Windows object format, PE, specifies a form of weak symbols called
6993
weak externals.  When a weak symbol is linked and the symbol is not
6994
defined, the weak symbol becomes an alias for some other symbol.  There
6995
are three variants of weak externals:
6996
@itemize
6997
@item Definition is searched for in objects and libraries, historically
6998
called lazy externals.
6999
@item Definition is searched for only in other objects, not in libraries.
7000
This form is not presently implemented.
7001
@item No search; the symbol is an alias.  This form is not presently
7002
implemented.
7003
@end itemize
7004
As a GNU extension, weak symbols that do not specify an alternate symbol
7005
are supported.  If the symbol is undefined when linking, the symbol
7006
uses a default value.
7007
 
7008
@cindex aligned common symbols
7009
@item aligned common symbols
7010
As a GNU extension to the PE file format, it is possible to specify the
7011
desired alignment for a common symbol.  This information is conveyed from
7012
the assembler or compiler to the linker by means of GNU-specific commands
7013
carried in the object file's @samp{.drectve} section, which are recognized
7014
by @command{ld} and respected when laying out the common symbols.  Native
7015
tools will be able to process object files employing this GNU extension,
7016
but will fail to respect the alignment instructions, and may issue noisy
7017
warnings about unknown linker directives.
7018
@end table
7019
 
7020
@ifclear GENERIC
7021
@lowersections
7022
@end ifclear
7023
@end ifset
7024
 
7025
@ifset XTENSA
7026
@ifclear GENERIC
7027
@raisesections
7028
@end ifclear
7029
 
7030
@node Xtensa
7031
@section @code{ld} and Xtensa Processors
7032
 
7033
@cindex Xtensa processors
7034
The default @command{ld} behavior for Xtensa processors is to interpret
7035
@code{SECTIONS} commands so that lists of explicitly named sections in a
7036
specification with a wildcard file will be interleaved when necessary to
7037
keep literal pools within the range of PC-relative load offsets.  For
7038
example, with the command:
7039
 
7040
@smallexample
7041
SECTIONS
7042
@{
7043
  .text : @{
7044
    *(.literal .text)
7045
  @}
7046
@}
7047
@end smallexample
7048
 
7049
@noindent
7050
@command{ld} may interleave some of the @code{.literal}
7051
and @code{.text} sections from different object files to ensure that the
7052
literal pools are within the range of PC-relative load offsets.  A valid
7053
interleaving might place the @code{.literal} sections from an initial
7054
group of files followed by the @code{.text} sections of that group of
7055
files.  Then, the @code{.literal} sections from the rest of the files
7056
and the @code{.text} sections from the rest of the files would follow.
7057
 
7058
@cindex @option{--relax} on Xtensa
7059
@cindex relaxing on Xtensa
7060
Relaxation is enabled by default for the Xtensa version of @command{ld} and
7061
provides two important link-time optimizations.  The first optimization
7062
is to combine identical literal values to reduce code size.  A redundant
7063
literal will be removed and all the @code{L32R} instructions that use it
7064
will be changed to reference an identical literal, as long as the
7065
location of the replacement literal is within the offset range of all
7066
the @code{L32R} instructions.  The second optimization is to remove
7067
unnecessary overhead from assembler-generated ``longcall'' sequences of
7068
@code{L32R}/@code{CALLX@var{n}} when the target functions are within
7069
range of direct @code{CALL@var{n}} instructions.
7070
 
7071
For each of these cases where an indirect call sequence can be optimized
7072
to a direct call, the linker will change the @code{CALLX@var{n}}
7073
instruction to a @code{CALL@var{n}} instruction, remove the @code{L32R}
7074
instruction, and remove the literal referenced by the @code{L32R}
7075
instruction if it is not used for anything else.  Removing the
7076
@code{L32R} instruction always reduces code size but can potentially
7077
hurt performance by changing the alignment of subsequent branch targets.
7078
By default, the linker will always preserve alignments, either by
7079
switching some instructions between 24-bit encodings and the equivalent
7080
density instructions or by inserting a no-op in place of the @code{L32R}
7081
instruction that was removed.  If code size is more important than
7082
performance, the @option{--size-opt} option can be used to prevent the
7083
linker from widening density instructions or inserting no-ops, except in
7084
a few cases where no-ops are required for correctness.
7085
 
7086
The following Xtensa-specific command-line options can be used to
7087
control the linker:
7088
 
7089
@cindex Xtensa options
7090
@table @option
7091
@kindex --no-relax
7092
@item --no-relax
7093
Since the Xtensa version of @code{ld} enables the @option{--relax} option
7094
by default, the @option{--no-relax} option is provided to disable
7095
relaxation.
7096
 
7097
@item --size-opt
7098
When optimizing indirect calls to direct calls, optimize for code size
7099
more than performance.  With this option, the linker will not insert
7100
no-ops or widen density instructions to preserve branch target
7101
alignment.  There may still be some cases where no-ops are required to
7102
preserve the correctness of the code.
7103
@end table
7104
 
7105
@ifclear GENERIC
7106
@lowersections
7107
@end ifclear
7108
@end ifset
7109
 
7110
@ifclear SingleFormat
7111
@node BFD
7112
@chapter BFD
7113
 
7114
@cindex back end
7115
@cindex object file management
7116
@cindex object formats available
7117
@kindex objdump -i
7118
The linker accesses object and archive files using the BFD libraries.
7119
These libraries allow the linker to use the same routines to operate on
7120
object files whatever the object file format.  A different object file
7121
format can be supported simply by creating a new BFD back end and adding
7122
it to the library.  To conserve runtime memory, however, the linker and
7123
associated tools are usually configured to support only a subset of the
7124
object file formats available.  You can use @code{objdump -i}
7125
(@pxref{objdump,,objdump,binutils.info,The GNU Binary Utilities}) to
7126
list all the formats available for your configuration.
7127
 
7128
@cindex BFD requirements
7129
@cindex requirements for BFD
7130
As with most implementations, BFD is a compromise between
7131
several conflicting requirements. The major factor influencing
7132
BFD design was efficiency: any time used converting between
7133
formats is time which would not have been spent had BFD not
7134
been involved. This is partly offset by abstraction payback; since
7135
BFD simplifies applications and back ends, more time and care
7136
may be spent optimizing algorithms for a greater speed.
7137
 
7138
One minor artifact of the BFD solution which you should bear in
7139
mind is the potential for information loss.  There are two places where
7140
useful information can be lost using the BFD mechanism: during
7141
conversion and during output. @xref{BFD information loss}.
7142
 
7143
@menu
7144
* BFD outline::                 How it works: an outline of BFD
7145
@end menu
7146
 
7147
@node BFD outline
7148
@section How It Works: An Outline of BFD
7149
@cindex opening object files
7150
@include bfdsumm.texi
7151
@end ifclear
7152
 
7153
@node Reporting Bugs
7154
@chapter Reporting Bugs
7155
@cindex bugs in @command{ld}
7156
@cindex reporting bugs in @command{ld}
7157
 
7158
Your bug reports play an essential role in making @command{ld} reliable.
7159
 
7160
Reporting a bug may help you by bringing a solution to your problem, or
7161
it may not.  But in any case the principal function of a bug report is
7162
to help the entire community by making the next version of @command{ld}
7163
work better.  Bug reports are your contribution to the maintenance of
7164
@command{ld}.
7165
 
7166
In order for a bug report to serve its purpose, you must include the
7167
information that enables us to fix the bug.
7168
 
7169
@menu
7170
* Bug Criteria::                Have you found a bug?
7171
* Bug Reporting::               How to report bugs
7172
@end menu
7173
 
7174
@node Bug Criteria
7175
@section Have You Found a Bug?
7176
@cindex bug criteria
7177
 
7178
If you are not sure whether you have found a bug, here are some guidelines:
7179
 
7180
@itemize @bullet
7181
@cindex fatal signal
7182
@cindex linker crash
7183
@cindex crash of linker
7184
@item
7185
If the linker gets a fatal signal, for any input whatever, that is a
7186
@command{ld} bug.  Reliable linkers never crash.
7187
 
7188
@cindex error on valid input
7189
@item
7190
If @command{ld} produces an error message for valid input, that is a bug.
7191
 
7192
@cindex invalid input
7193
@item
7194
If @command{ld} does not produce an error message for invalid input, that
7195
may be a bug.  In the general case, the linker can not verify that
7196
object files are correct.
7197
 
7198
@item
7199
If you are an experienced user of linkers, your suggestions for
7200
improvement of @command{ld} are welcome in any case.
7201
@end itemize
7202
 
7203
@node Bug Reporting
7204
@section How to Report Bugs
7205
@cindex bug reports
7206
@cindex @command{ld} bugs, reporting
7207
 
7208
A number of companies and individuals offer support for @sc{gnu}
7209
products.  If you obtained @command{ld} from a support organization, we
7210
recommend you contact that organization first.
7211
 
7212
You can find contact information for many support companies and
7213
individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
7214
distribution.
7215
 
7216
@ifset BUGURL
7217
Otherwise, send bug reports for @command{ld} to
7218
@value{BUGURL}.
7219
@end ifset
7220
 
7221
The fundamental principle of reporting bugs usefully is this:
7222
@strong{report all the facts}.  If you are not sure whether to state a
7223
fact or leave it out, state it!
7224
 
7225
Often people omit facts because they think they know what causes the
7226
problem and assume that some details do not matter.  Thus, you might
7227
assume that the name of a symbol you use in an example does not
7228
matter.  Well, probably it does not, but one cannot be sure.  Perhaps
7229
the bug is a stray memory reference which happens to fetch from the
7230
location where that name is stored in memory; perhaps, if the name
7231
were different, the contents of that location would fool the linker
7232
into doing the right thing despite the bug.  Play it safe and give a
7233
specific, complete example.  That is the easiest thing for you to do,
7234
and the most helpful.
7235
 
7236
Keep in mind that the purpose of a bug report is to enable us to fix
7237
the bug if it is new to us.  Therefore, always write your bug reports
7238
on the assumption that the bug has not been reported previously.
7239
 
7240
Sometimes people give a few sketchy facts and ask, ``Does this ring a
7241
bell?''  This cannot help us fix a bug, so it is basically useless.  We
7242
respond by asking for enough details to enable us to investigate.
7243
You might as well expedite matters by sending them to begin with.
7244
 
7245
To enable us to fix the bug, you should include all these things:
7246
 
7247
@itemize @bullet
7248
@item
7249
The version of @command{ld}.  @command{ld} announces it if you start it with
7250
the @samp{--version} argument.
7251
 
7252
Without this, we will not know whether there is any point in looking for
7253
the bug in the current version of @command{ld}.
7254
 
7255
@item
7256
Any patches you may have applied to the @command{ld} source, including any
7257
patches made to the @code{BFD} library.
7258
 
7259
@item
7260
The type of machine you are using, and the operating system name and
7261
version number.
7262
 
7263
@item
7264
What compiler (and its version) was used to compile @command{ld}---e.g.
7265
``@code{gcc-2.7}''.
7266
 
7267
@item
7268
The command arguments you gave the linker to link your example and
7269
observe the bug.  To guarantee you will not omit something important,
7270
list them all.  A copy of the Makefile (or the output from make) is
7271
sufficient.
7272
 
7273
If we were to try to guess the arguments, we would probably guess wrong
7274
and then we might not encounter the bug.
7275
 
7276
@item
7277
A complete input file, or set of input files, that will reproduce the
7278
bug.  It is generally most helpful to send the actual object files
7279
provided that they are reasonably small.  Say no more than 10K.  For
7280
bigger files you can either make them available by FTP or HTTP or else
7281
state that you are willing to send the object file(s) to whomever
7282
requests them.  (Note - your email will be going to a mailing list, so
7283
we do not want to clog it up with large attachments).  But small
7284
attachments are best.
7285
 
7286
If the source files were assembled using @code{gas} or compiled using
7287
@code{gcc}, then it may be OK to send the source files rather than the
7288
object files.  In this case, be sure to say exactly what version of
7289
@code{gas} or @code{gcc} was used to produce the object files.  Also say
7290
how @code{gas} or @code{gcc} were configured.
7291
 
7292
@item
7293
A description of what behavior you observe that you believe is
7294
incorrect.  For example, ``It gets a fatal signal.''
7295
 
7296
Of course, if the bug is that @command{ld} gets a fatal signal, then we
7297
will certainly notice it.  But if the bug is incorrect output, we might
7298
not notice unless it is glaringly wrong.  You might as well not give us
7299
a chance to make a mistake.
7300
 
7301
Even if the problem you experience is a fatal signal, you should still
7302
say so explicitly.  Suppose something strange is going on, such as, your
7303
copy of @command{ld} is out of sync, or you have encountered a bug in the
7304
C library on your system.  (This has happened!)  Your copy might crash
7305
and ours would not.  If you told us to expect a crash, then when ours
7306
fails to crash, we would know that the bug was not happening for us.  If
7307
you had not told us to expect a crash, then we would not be able to draw
7308
any conclusion from our observations.
7309
 
7310
@item
7311
If you wish to suggest changes to the @command{ld} source, send us context
7312
diffs, as generated by @code{diff} with the @samp{-u}, @samp{-c}, or
7313
@samp{-p} option.  Always send diffs from the old file to the new file.
7314
If you even discuss something in the @command{ld} source, refer to it by
7315
context, not by line number.
7316
 
7317
The line numbers in our development sources will not match those in your
7318
sources.  Your line numbers would convey no useful information to us.
7319
@end itemize
7320
 
7321
Here are some things that are not necessary:
7322
 
7323
@itemize @bullet
7324
@item
7325
A description of the envelope of the bug.
7326
 
7327
Often people who encounter a bug spend a lot of time investigating
7328
which changes to the input file will make the bug go away and which
7329
changes will not affect it.
7330
 
7331
This is often time consuming and not very useful, because the way we
7332
will find the bug is by running a single example under the debugger
7333
with breakpoints, not by pure deduction from a series of examples.
7334
We recommend that you save your time for something else.
7335
 
7336
Of course, if you can find a simpler example to report @emph{instead}
7337
of the original one, that is a convenience for us.  Errors in the
7338
output will be easier to spot, running under the debugger will take
7339
less time, and so on.
7340
 
7341
However, simplification is not vital; if you do not want to do this,
7342
report the bug anyway and send us the entire test case you used.
7343
 
7344
@item
7345
A patch for the bug.
7346
 
7347
A patch for the bug does help us if it is a good one.  But do not omit
7348
the necessary information, such as the test case, on the assumption that
7349
a patch is all we need.  We might see problems with your patch and decide
7350
to fix the problem another way, or we might not understand it at all.
7351
 
7352
Sometimes with a program as complicated as @command{ld} it is very hard to
7353
construct an example that will make the program follow a certain path
7354
through the code.  If you do not send us the example, we will not be
7355
able to construct one, so we will not be able to verify that the bug is
7356
fixed.
7357
 
7358
And if we cannot understand what bug you are trying to fix, or why your
7359
patch should be an improvement, we will not install it.  A test case will
7360
help us to understand.
7361
 
7362
@item
7363
A guess about what the bug is or what it depends on.
7364
 
7365
Such guesses are usually wrong.  Even we cannot guess right about such
7366
things without first using the debugger to find the facts.
7367
@end itemize
7368
 
7369
@node MRI
7370
@appendix MRI Compatible Script Files
7371
@cindex MRI compatibility
7372
To aid users making the transition to @sc{gnu} @command{ld} from the MRI
7373
linker, @command{ld} can use MRI compatible linker scripts as an
7374
alternative to the more general-purpose linker scripting language
7375
described in @ref{Scripts}.  MRI compatible linker scripts have a much
7376
simpler command set than the scripting language otherwise used with
7377
@command{ld}.  @sc{gnu} @command{ld} supports the most commonly used MRI
7378
linker commands; these commands are described here.
7379
 
7380
In general, MRI scripts aren't of much use with the @code{a.out} object
7381
file format, since it only has three sections and MRI scripts lack some
7382
features to make use of them.
7383
 
7384
You can specify a file containing an MRI-compatible script using the
7385
@samp{-c} command-line option.
7386
 
7387
Each command in an MRI-compatible script occupies its own line; each
7388
command line starts with the keyword that identifies the command (though
7389
blank lines are also allowed for punctuation).  If a line of an
7390
MRI-compatible script begins with an unrecognized keyword, @command{ld}
7391
issues a warning message, but continues processing the script.
7392
 
7393
Lines beginning with @samp{*} are comments.
7394
 
7395
You can write these commands using all upper-case letters, or all
7396
lower case; for example, @samp{chip} is the same as @samp{CHIP}.
7397
The following list shows only the upper-case form of each command.
7398
 
7399
@table @code
7400
@cindex @code{ABSOLUTE} (MRI)
7401
@item ABSOLUTE @var{secname}
7402
@itemx ABSOLUTE @var{secname}, @var{secname}, @dots{} @var{secname}
7403
Normally, @command{ld} includes in the output file all sections from all
7404
the input files.  However, in an MRI-compatible script, you can use the
7405
@code{ABSOLUTE} command to restrict the sections that will be present in
7406
your output program.  If the @code{ABSOLUTE} command is used at all in a
7407
script, then only the sections named explicitly in @code{ABSOLUTE}
7408
commands will appear in the linker output.  You can still use other
7409
input sections (whatever you select on the command line, or using
7410
@code{LOAD}) to resolve addresses in the output file.
7411
 
7412
@cindex @code{ALIAS} (MRI)
7413
@item ALIAS @var{out-secname}, @var{in-secname}
7414
Use this command to place the data from input section @var{in-secname}
7415
in a section called @var{out-secname} in the linker output file.
7416
 
7417
@var{in-secname} may be an integer.
7418
 
7419
@cindex @code{ALIGN} (MRI)
7420
@item ALIGN @var{secname} = @var{expression}
7421
Align the section called @var{secname} to @var{expression}.  The
7422
@var{expression} should be a power of two.
7423
 
7424
@cindex @code{BASE} (MRI)
7425
@item BASE @var{expression}
7426
Use the value of @var{expression} as the lowest address (other than
7427
absolute addresses) in the output file.
7428
 
7429
@cindex @code{CHIP} (MRI)
7430
@item CHIP @var{expression}
7431
@itemx CHIP @var{expression}, @var{expression}
7432
This command does nothing; it is accepted only for compatibility.
7433
 
7434
@cindex @code{END} (MRI)
7435
@item END
7436
This command does nothing whatever; it's only accepted for compatibility.
7437
 
7438
@cindex @code{FORMAT} (MRI)
7439
@item FORMAT @var{output-format}
7440
Similar to the @code{OUTPUT_FORMAT} command in the more general linker
7441
language, but restricted to one of these output formats:
7442
 
7443
@enumerate
7444
@item
7445
S-records, if @var{output-format} is @samp{S}
7446
 
7447
@item
7448
IEEE, if @var{output-format} is @samp{IEEE}
7449
 
7450
@item
7451
COFF (the @samp{coff-m68k} variant in BFD), if @var{output-format} is
7452
@samp{COFF}
7453
@end enumerate
7454
 
7455
@cindex @code{LIST} (MRI)
7456
@item LIST @var{anything}@dots{}
7457
Print (to the standard output file) a link map, as produced by the
7458
@command{ld} command-line option @samp{-M}.
7459
 
7460
The keyword @code{LIST} may be followed by anything on the
7461
same line, with no change in its effect.
7462
 
7463
@cindex @code{LOAD} (MRI)
7464
@item LOAD @var{filename}
7465
@itemx LOAD @var{filename}, @var{filename}, @dots{} @var{filename}
7466
Include one or more object file @var{filename} in the link; this has the
7467
same effect as specifying @var{filename} directly on the @command{ld}
7468
command line.
7469
 
7470
@cindex @code{NAME} (MRI)
7471
@item NAME @var{output-name}
7472
@var{output-name} is the name for the program produced by @command{ld}; the
7473
MRI-compatible command @code{NAME} is equivalent to the command-line
7474
option @samp{-o} or the general script language command @code{OUTPUT}.
7475
 
7476
@cindex @code{ORDER} (MRI)
7477
@item ORDER @var{secname}, @var{secname}, @dots{} @var{secname}
7478
@itemx ORDER @var{secname} @var{secname} @var{secname}
7479
Normally, @command{ld} orders the sections in its output file in the
7480
order in which they first appear in the input files.  In an MRI-compatible
7481
script, you can override this ordering with the @code{ORDER} command.  The
7482
sections you list with @code{ORDER} will appear first in your output
7483
file, in the order specified.
7484
 
7485
@cindex @code{PUBLIC} (MRI)
7486
@item PUBLIC @var{name}=@var{expression}
7487
@itemx PUBLIC @var{name},@var{expression}
7488
@itemx PUBLIC @var{name} @var{expression}
7489
Supply a value (@var{expression}) for external symbol
7490
@var{name} used in the linker input files.
7491
 
7492
@cindex @code{SECT} (MRI)
7493
@item SECT @var{secname}, @var{expression}
7494
@itemx SECT @var{secname}=@var{expression}
7495
@itemx SECT @var{secname} @var{expression}
7496
You can use any of these three forms of the @code{SECT} command to
7497
specify the start address (@var{expression}) for section @var{secname}.
7498
If you have more than one @code{SECT} statement for the same
7499
@var{secname}, only the @emph{first} sets the start address.
7500
@end table
7501
 
7502
@node GNU Free Documentation License
7503
@appendix GNU Free Documentation License
7504
@include fdl.texi
7505
 
7506
@node LD Index
7507
@unnumbered LD Index
7508
 
7509
@printindex cp
7510
 
7511
@tex
7512
% I think something like @colophon should be in texinfo.  In the
7513
% meantime:
7514
\long\def\colophon{\hbox to0pt{}\vfill
7515
\centerline{The body of this manual is set in}
7516
\centerline{\fontname\tenrm,}
7517
\centerline{with headings in {\bf\fontname\tenbf}}
7518
\centerline{and examples in {\tt\fontname\tentt}.}
7519
\centerline{{\it\fontname\tenit\/} and}
7520
\centerline{{\sl\fontname\tensl\/}}
7521
\centerline{are used for emphasis.}\vfill}
7522
\page\colophon
7523
% Blame: doc@cygnus.com, 28mar91.
7524
@end tex
7525
 
7526
@bye

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.