1 \input texinfo @c -*- Texinfo -*-
2 @setfilename binutils.info
8 * Binutils: (binutils). The GNU binary utilities "ar", "objcopy",
9 "objdump", "nm", "nlmconv", "size",
10 "strings", "strip", "ranlib" and "dlltool".
16 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 1998 Free Software Foundation, Inc.
18 Permission is granted to make and distribute verbatim copies of
19 this manual provided the copyright notice and this permission notice
20 are preserved on all copies.
23 Permission is granted to process this file through TeX and print the
24 results, provided the printed document carries a copying permission
25 notice identical to this one except for the removal of this paragraph
26 (this paragraph not being relevant to the printed manual).
30 Permission is granted to copy and distribute modified versions of this
31 manual under the conditions for verbatim copying, provided also that
32 the entire resulting derived work is distributed under the terms of a
33 permission notice identical to this one.
35 Permission is granted to copy and distribute translations of this manual
36 into another language, under the above conditions for modified versions.
41 @c This file documents the GNU binary utilities "ar", "ld", "objcopy",
42 @c "objdump", "nm", "size", "strings", "strip", and "ranlib".
44 @c Copyright (C) 1991, 92, 93, 94, 95, 96, 97, 1998 Free Software Foundation, Inc.
46 @c This text may be freely distributed under the terms of the GNU
47 @c General Public License.
50 @setchapternewpage odd
51 @settitle @sc{gnu} Binary Utilities
54 @title The @sc{gnu} Binary Utilities
55 @subtitle Version @value{VERSION}
58 @author Roland H. Pesch
59 @author Jeffrey M. Osier
60 @author Cygnus Support
64 {\parskip=0pt \hfill Cygnus Support\par \hfill
65 \TeX{}info \texinfoversion\par }
68 @vskip 0pt plus 1filll
69 Copyright @copyright{} 1991, 92, 93, 94, 95, 96, 97, 1998 Free Software Foundation, Inc.
71 Permission is granted to make and distribute verbatim copies of
72 this manual provided the copyright notice and this permission notice
73 are preserved on all copies.
75 Permission is granted to copy and distribute modified versions of this
76 manual under the conditions for verbatim copying, provided also that
77 the entire resulting derived work is distributed under the terms of a
78 permission notice identical to this one.
80 Permission is granted to copy and distribute translations of this manual
81 into another language, under the above conditions for modified versions.
88 This brief manual contains preliminary documentation for the @sc{gnu} binary
89 utilities (collectively version @value{VERSION}):
94 Create, modify, and extract from archives
97 List symbols from object files
100 Copy and translate object files
103 Display information from object files
106 Generate index to archive contents
109 List file section sizes and total size
112 List printable strings from files
118 Demangle encoded C++ symbols
121 Convert addresses into file names and line numbers
124 Convert object code into a Netware Loadable Module
127 Manipulate Windows resources
130 Create the files needed to build and use Dynamic Link Libraries
135 * ar:: Create, modify, and extract from archives
136 * nm:: List symbols from object files
137 * objcopy:: Copy and translate object files
138 * objdump:: Display information from object files
139 * ranlib:: Generate index to archive contents
140 * size:: List section sizes and total size
141 * strings:: List printable strings from files
142 * strip:: Discard symbols
143 * c++filt:: Filter to demangle encoded C++ symbols
144 * addr2line:: Convert addresses to file and line
145 * nlmconv:: Converts object code into an NLM
146 * windres:: Manipulate Windows resources
147 * dlltool:: Create files needed to build and use DLLs
148 * Selecting The Target System:: How these utilities determine the target.
149 * Reporting Bugs:: Reporting Bugs
158 @cindex collections of files
160 ar [-]@var{p}[@var{mod} [@var{relpos}]] @var{archive} [@var{member}@dots{}]
161 ar -M [ <mri-script ]
164 The @sc{gnu} @code{ar} program creates, modifies, and extracts from
165 archives. An @dfn{archive} is a single file holding a collection of
166 other files in a structure that makes it possible to retrieve
167 the original individual files (called @dfn{members} of the archive).
169 The original files' contents, mode (permissions), timestamp, owner, and
170 group are preserved in the archive, and can be restored on
174 @sc{gnu} @code{ar} can maintain archives whose members have names of any
175 length; however, depending on how @code{ar} is configured on your
176 system, a limit on member-name length may be imposed for compatibility
177 with archive formats maintained with other tools. If it exists, the
178 limit is often 15 characters (typical of formats related to a.out) or 16
179 characters (typical of formats related to coff).
182 @code{ar} is considered a binary utility because archives of this sort
183 are most often used as @dfn{libraries} holding commonly needed
187 @code{ar} creates an index to the symbols defined in relocatable
188 object modules in the archive when you specify the modifier @samp{s}.
189 Once created, this index is updated in the archive whenever @code{ar}
190 makes a change to its contents (save for the @samp{q} update operation).
191 An archive with such an index speeds up linking to the library, and
192 allows routines in the library to call each other without regard to
193 their placement in the archive.
195 You may use @samp{nm -s} or @samp{nm --print-armap} to list this index
196 table. If an archive lacks the table, another form of @code{ar} called
197 @code{ranlib} can be used to add just the table.
199 @cindex compatibility, @code{ar}
200 @cindex @code{ar} compatibility
201 @sc{gnu} @code{ar} is designed to be compatible with two different
202 facilities. You can control its activity using command-line options,
203 like the different varieties of @code{ar} on Unix systems; or, if you
204 specify the single command-line option @samp{-M}, you can control it
205 with a script supplied via standard input, like the MRI ``librarian''
209 * ar cmdline:: Controlling @code{ar} on the command line
210 * ar scripts:: Controlling @code{ar} with a script
215 @section Controlling @code{ar} on the command line
218 ar [-]@var{p}[@var{mod} [@var{relpos}]] @var{archive} [@var{member}@dots{}]
221 @cindex Unix compatibility, @code{ar}
222 When you use @code{ar} in the Unix style, @code{ar} insists on at least two
223 arguments to execute: one keyletter specifying the @emph{operation}
224 (optionally accompanied by other keyletters specifying
225 @emph{modifiers}), and the archive name to act on.
227 Most operations can also accept further @var{member} arguments,
228 specifying particular files to operate on.
230 @sc{gnu} @code{ar} allows you to mix the operation code @var{p} and modifier
231 flags @var{mod} in any order, within the first command-line argument.
233 If you wish, you may begin the first command-line argument with a
236 @cindex operations on archive
237 The @var{p} keyletter specifies what operation to execute; it may be
238 any of the following, but you must specify only one of them:
242 @cindex deleting from archive
243 @emph{Delete} modules from the archive. Specify the names of modules to
244 be deleted as @var{member}@dots{}; the archive is untouched if you
245 specify no files to delete.
247 If you specify the @samp{v} modifier, @code{ar} lists each module
251 @cindex moving in archive
252 Use this operation to @emph{move} members in an archive.
254 The ordering of members in an archive can make a difference in how
255 programs are linked using the library, if a symbol is defined in more
258 If no modifiers are used with @code{m}, any members you name in the
259 @var{member} arguments are moved to the @emph{end} of the archive;
260 you can use the @samp{a}, @samp{b}, or @samp{i} modifiers to move them to a
261 specified place instead.
264 @cindex printing from archive
265 @emph{Print} the specified members of the archive, to the standard
266 output file. If the @samp{v} modifier is specified, show the member
267 name before copying its contents to standard output.
269 If you specify no @var{member} arguments, all the files in the archive are
273 @cindex quick append to archive
274 @emph{Quick append}; Historically, add the files @var{member}@dots{} to the end of
275 @var{archive}, without checking for replacement.
277 The modifiers @samp{a}, @samp{b}, and @samp{i} do @emph{not} affect this
278 operation; new members are always placed at the end of the archive.
280 The modifier @samp{v} makes @code{ar} list each file as it is appended.
282 Since the point of this operation is speed, the archive's symbol table
283 index is not updated, even if it already existed; you can use @samp{ar s} or
284 @code{ranlib} explicitly to update the symbol table index.
286 However, too many different systems assume quick append rebuilds the
287 index, so GNU ar implements @code{q} as a synonym for @code{r}.
290 @cindex replacement in archive
291 Insert the files @var{member}@dots{} into @var{archive} (with
292 @emph{replacement}). This operation differs from @samp{q} in that any
293 previously existing members are deleted if their names match those being
296 If one of the files named in @var{member}@dots{} does not exist, @code{ar}
297 displays an error message, and leaves undisturbed any existing members
298 of the archive matching that name.
300 By default, new members are added at the end of the file; but you may
301 use one of the modifiers @samp{a}, @samp{b}, or @samp{i} to request
302 placement relative to some existing member.
304 The modifier @samp{v} used with this operation elicits a line of
305 output for each file inserted, along with one of the letters @samp{a} or
306 @samp{r} to indicate whether the file was appended (no old member
307 deleted) or replaced.
310 @cindex contents of archive
311 Display a @emph{table} listing the contents of @var{archive}, or those
312 of the files listed in @var{member}@dots{} that are present in the
313 archive. Normally only the member name is shown; if you also want to
314 see the modes (permissions), timestamp, owner, group, and size, you can
315 request that by also specifying the @samp{v} modifier.
317 If you do not specify a @var{member}, all files in the archive
320 @cindex repeated names in archive
321 @cindex name duplication in archive
322 If there is more than one file with the same name (say, @samp{fie}) in
323 an archive (say @samp{b.a}), @samp{ar t b.a fie} lists only the
324 first instance; to see them all, you must ask for a complete
325 listing---in our example, @samp{ar t b.a}.
326 @c WRS only; per Gumby, this is implementation-dependent, and in a more
327 @c recent case in fact works the other way.
330 @cindex extract from archive
331 @emph{Extract} members (named @var{member}) from the archive. You can
332 use the @samp{v} modifier with this operation, to request that
333 @code{ar} list each name as it extracts it.
335 If you do not specify a @var{member}, all files in the archive
340 A number of modifiers (@var{mod}) may immediately follow the @var{p}
341 keyletter, to specify variations on an operation's behavior:
345 @cindex relative placement in archive
346 Add new files @emph{after} an existing member of the
347 archive. If you use the modifier @samp{a}, the name of an existing archive
348 member must be present as the @var{relpos} argument, before the
349 @var{archive} specification.
352 Add new files @emph{before} an existing member of the
353 archive. If you use the modifier @samp{b}, the name of an existing archive
354 member must be present as the @var{relpos} argument, before the
355 @var{archive} specification. (same as @samp{i}).
358 @cindex creating archives
359 @emph{Create} the archive. The specified @var{archive} is always
360 created if it did not exist, when you request an update. But a warning is
361 issued unless you specify in advance that you expect to create it, by
365 Truncate names in the archive. @sc{gnu} @code{ar} will normally permit file
366 names of any length. This will cause it to create archives which are
367 not compatible with the native @code{ar} program on some systems. If
368 this is a concern, the @samp{f} modifier may be used to truncate file
369 names when putting them in the archive.
372 Insert new files @emph{before} an existing member of the
373 archive. If you use the modifier @samp{i}, the name of an existing archive
374 member must be present as the @var{relpos} argument, before the
375 @var{archive} specification. (same as @samp{b}).
378 This modifier is accepted but not used.
379 @c whaffor ar l modifier??? presumably compat; with
380 @c what???---doc@@cygnus.com, 25jan91
383 @cindex dates in archive
384 Preserve the @emph{original} dates of members when extracting them. If
385 you do not specify this modifier, files extracted from the archive
386 are stamped with the time of extraction.
389 @cindex writing archive index
390 Write an object-file index into the archive, or update an existing one,
391 even if no other change is made to the archive. You may use this modifier
392 flag either with any operation, or alone. Running @samp{ar s} on an
393 archive is equivalent to running @samp{ranlib} on it.
396 @cindex not writing archive index
397 Do not generate an archive symbol table. This can speed up building a
398 large library in several steps. The resulting archive can not be used
399 with the linker. In order to build a symbol table, you must omit the
400 @samp{S} modifier on the last execution of @samp{ar}, or you must run
401 @samp{ranlib} on the archive.
404 @cindex updating an archive
405 Normally, @samp{ar r}@dots{} inserts all files
406 listed into the archive. If you would like to insert @emph{only} those
407 of the files you list that are newer than existing members of the same
408 names, use this modifier. The @samp{u} modifier is allowed only for the
409 operation @samp{r} (replace). In particular, the combination @samp{qu} is
410 not allowed, since checking the timestamps would lose any speed
411 advantage from the operation @samp{q}.
414 This modifier requests the @emph{verbose} version of an operation. Many
415 operations display additional information, such as filenames processed,
416 when the modifier @samp{v} is appended.
419 This modifier shows the version number of @code{ar}.
423 @section Controlling @code{ar} with a script
426 ar -M [ <@var{script} ]
429 @cindex MRI compatibility, @code{ar}
430 @cindex scripts, @code{ar}
431 If you use the single command-line option @samp{-M} with @code{ar}, you
432 can control its operation with a rudimentary command language. This
433 form of @code{ar} operates interactively if standard input is coming
434 directly from a terminal. During interactive use, @code{ar} prompts for
435 input (the prompt is @samp{AR >}), and continues executing even after
436 errors. If you redirect standard input to a script file, no prompts are
437 issued, and @code{ar} abandons execution (with a nonzero exit code)
440 The @code{ar} command language is @emph{not} designed to be equivalent
441 to the command-line options; in fact, it provides somewhat less control
442 over archives. The only purpose of the command language is to ease the
443 transition to @sc{gnu} @code{ar} for developers who already have scripts
444 written for the MRI ``librarian'' program.
446 The syntax for the @code{ar} command language is straightforward:
449 commands are recognized in upper or lower case; for example, @code{LIST}
450 is the same as @code{list}. In the following descriptions, commands are
451 shown in upper case for clarity.
454 a single command may appear on each line; it is the first word on the
458 empty lines are allowed, and have no effect.
461 comments are allowed; text after either of the characters @samp{*}
462 or @samp{;} is ignored.
465 Whenever you use a list of names as part of the argument to an @code{ar}
466 command, you can separate the individual names with either commas or
467 blanks. Commas are shown in the explanations below, for clarity.
470 @samp{+} is used as a line continuation character; if @samp{+} appears
471 at the end of a line, the text on the following line is considered part
472 of the current command.
475 Here are the commands you can use in @code{ar} scripts, or when using
476 @code{ar} interactively. Three of them have special significance:
478 @code{OPEN} or @code{CREATE} specify a @dfn{current archive}, which is
479 a temporary file required for most of the other commands.
481 @code{SAVE} commits the changes so far specified by the script. Prior
482 to @code{SAVE}, commands affect only the temporary copy of the current
486 @item ADDLIB @var{archive}
487 @itemx ADDLIB @var{archive} (@var{module}, @var{module}, @dots{} @var{module})
488 Add all the contents of @var{archive} (or, if specified, each named
489 @var{module} from @var{archive}) to the current archive.
491 Requires prior use of @code{OPEN} or @code{CREATE}.
493 @item ADDMOD @var{member}, @var{member}, @dots{} @var{member}
494 @c FIXME! w/Replacement?? If so, like "ar r @var{archive} @var{names}"
495 @c else like "ar q..."
496 Add each named @var{member} as a module in the current archive.
498 Requires prior use of @code{OPEN} or @code{CREATE}.
501 Discard the contents of the current archive, canceling the effect of
502 any operations since the last @code{SAVE}. May be executed (with no
503 effect) even if no current archive is specified.
505 @item CREATE @var{archive}
506 Creates an archive, and makes it the current archive (required for many
507 other commands). The new archive is created with a temporary name; it
508 is not actually saved as @var{archive} until you use @code{SAVE}.
509 You can overwrite existing archives; similarly, the contents of any
510 existing file named @var{archive} will not be destroyed until @code{SAVE}.
512 @item DELETE @var{module}, @var{module}, @dots{} @var{module}
513 Delete each listed @var{module} from the current archive; equivalent to
514 @samp{ar -d @var{archive} @var{module} @dots{} @var{module}}.
516 Requires prior use of @code{OPEN} or @code{CREATE}.
518 @item DIRECTORY @var{archive} (@var{module}, @dots{} @var{module})
519 @itemx DIRECTORY @var{archive} (@var{module}, @dots{} @var{module}) @var{outputfile}
520 List each named @var{module} present in @var{archive}. The separate
521 command @code{VERBOSE} specifies the form of the output: when verbose
522 output is off, output is like that of @samp{ar -t @var{archive}
523 @var{module}@dots{}}. When verbose output is on, the listing is like
524 @samp{ar -tv @var{archive} @var{module}@dots{}}.
526 Output normally goes to the standard output stream; however, if you
527 specify @var{outputfile} as a final argument, @code{ar} directs the
531 Exit from @code{ar}, with a @code{0} exit code to indicate successful
532 completion. This command does not save the output file; if you have
533 changed the current archive since the last @code{SAVE} command, those
536 @item EXTRACT @var{module}, @var{module}, @dots{} @var{module}
537 Extract each named @var{module} from the current archive, writing them
538 into the current directory as separate files. Equivalent to @samp{ar -x
539 @var{archive} @var{module}@dots{}}.
541 Requires prior use of @code{OPEN} or @code{CREATE}.
544 @c FIXME Tokens but no commands???
551 Display full contents of the current archive, in ``verbose'' style
552 regardless of the state of @code{VERBOSE}. The effect is like @samp{ar
553 tv @var{archive}}). (This single command is a @sc{gnu} @code{ld}
554 enhancement, rather than present for MRI compatibility.)
556 Requires prior use of @code{OPEN} or @code{CREATE}.
558 @item OPEN @var{archive}
559 Opens an existing archive for use as the current archive (required for
560 many other commands). Any changes as the result of subsequent commands
561 will not actually affect @var{archive} until you next use @code{SAVE}.
563 @item REPLACE @var{module}, @var{module}, @dots{} @var{module}
564 In the current archive, replace each existing @var{module} (named in
565 the @code{REPLACE} arguments) from files in the current working directory.
566 To execute this command without errors, both the file, and the module in
567 the current archive, must exist.
569 Requires prior use of @code{OPEN} or @code{CREATE}.
572 Toggle an internal flag governing the output from @code{DIRECTORY}.
573 When the flag is on, @code{DIRECTORY} output matches output from
574 @samp{ar -tv }@dots{}.
577 Commit your changes to the current archive, and actually save it as a
578 file with the name specified in the last @code{CREATE} or @code{OPEN}
581 Requires prior use of @code{OPEN} or @code{CREATE}.
590 The @sc{gnu} linker @code{ld} is now described in a separate manual.
591 @xref{Top,, Overview,, Using LD: the @sc{gnu} linker}.
600 nm [ -a | --debug-syms ] [ -g | --extern-only ]
601 [ -B ] [ -C | --demangle ] [ -D | --dynamic ]
602 [ -s | --print-armap ] [ -A | -o | --print-file-name ]
603 [ -n | -v | --numeric-sort ] [ -p | --no-sort ]
604 [ -r | --reverse-sort ] [ --size-sort ] [ -u | --undefined-only ]
605 [ -t @var{radix} | --radix=@var{radix} ] [ -P | --portability ]
606 [ --target=@var{bfdname} ] [ -f @var{format} | --format=@var{format} ]
607 [ --defined-only ] [-l | --line-numbers ]
608 [ --no-demangle ] [ -V | --version ] [ --help ] [ @var{objfile}@dots{} ]
611 @sc{gnu} @code{nm} lists the symbols from object files @var{objfile}@dots{}.
612 If no object files are listed as arguments, @code{nm} assumes
615 For each symbol, @code{nm} shows:
619 The symbol value, in the radix selected by options (see below), or
620 hexadecimal by default.
623 The symbol type. At least the following types are used; others are, as
624 well, depending on the object file format. If lowercase, the symbol is
625 local; if uppercase, the symbol is global (external).
627 @c Some more detail on exactly what these symbol types are used for
631 The symbol's value is absolute, and will not be changed by further
635 The symbol is in the uninitialized data section (known as BSS).
638 The symbol is common. Common symbols are uninitialized data. When
639 linking, multiple common symbols may appear with the same name. If the
640 symbol is defined anywhere, the common symbols are treated as undefined
641 references. For more details on common symbols, see the discussion of
642 --warn-common in @ref{Options,,Linker options,ld.info,The GNU linker}.
645 The symbol is in the initialized data section.
648 The symbol is in an initialized data section for small objects. Some
649 object file formats permit more efficient access to small data objects,
650 such as a global int variable as opposed to a large global array.
653 The symbol is an indirect reference to another symbol. This is a GNU
654 extension to the a.out object file format which is rarely used.
657 The symbol is a debugging symbol.
660 The symbol is in a read only data section.
663 The symbol is in an uninitialized data section for small objects.
666 The symbol is in the text (code) section.
669 The symbol is undefined.
672 The symbol is weak. When a weak defined symbol is linked with a normal
673 defined symbol, the normal defined symbol is used with no error. When a
674 weak undefined symbol is linked and the symbol is not defined, the value
675 of the weak symbol becomes zero with no error.
678 The symbol is a stabs symbol in an a.out object file. In this case, the
679 next values printed are the stabs other field, the stabs desc field, and
680 the stab type. Stabs symbols are used to hold debugging information;
681 for more information, see @ref{Top,Stabs,Stabs Overview,stabs.info, The
682 ``stabs'' debug format}.
685 The symbol type is unknown, or object file format specific.
692 The long and short forms of options, shown here as alternatives, are
698 @itemx --print-file-name
699 @cindex input file name
701 @cindex source file name
702 Precede each symbol by the name of the input file (or archive element)
703 in which it was found, rather than identifying the input file once only,
704 before all of its symbols.
708 @cindex debugging symbols
709 Display all symbols, even debugger-only symbols; normally these are not
713 @cindex @code{nm} format
714 @cindex @code{nm} compatibility
715 The same as @samp{--format=bsd} (for compatibility with the MIPS @code{nm}).
719 @cindex demangling in nm
720 Decode (@dfn{demangle}) low-level symbol names into user-level names.
721 Besides removing any initial underscore prepended by the system, this
722 makes C++ function names readable. @xref{c++filt}, for more information
726 Do not demangle low-level symbol names. This is the default.
730 @cindex dynamic symbols
731 Display the dynamic symbols rather than the normal symbols. This is
732 only meaningful for dynamic objects, such as certain types of shared
735 @item -f @var{format}
736 @itemx --format=@var{format}
737 @cindex @code{nm} format
738 @cindex @code{nm} compatibility
739 Use the output format @var{format}, which can be @code{bsd},
740 @code{sysv}, or @code{posix}. The default is @code{bsd}.
741 Only the first character of @var{format} is significant; it can be
742 either upper or lower case.
746 @cindex external symbols
747 Display only external symbols.
750 @itemx --line-numbers
751 @cindex symbol line numbers
752 For each symbol, use debugging information to try to find a filename and
753 line number. For a defined symbol, look for the line number of the
754 address of the symbol. For an undefined symbol, look for the line
755 number of a relocation entry which refers to the symbol. If line number
756 information can be found, print it after the other symbol information.
760 @itemx --numeric-sort
761 Sort symbols numerically by their addresses, rather than alphabetically
766 @cindex sorting symbols
767 Do not bother to sort the symbols in any order; print them in the order
772 Use the POSIX.2 standard output format instead of the default format.
773 Equivalent to @samp{-f posix}.
777 @cindex symbol index, listing
778 When listing symbols from archive members, include the index: a mapping
779 (stored in the archive by @code{ar} or @code{ranlib}) of which modules
780 contain definitions for which names.
783 @itemx --reverse-sort
784 Reverse the order of the sort (whether numeric or alphabetic); let the
788 Sort symbols by size. The size is computed as the difference between
789 the value of the symbol and the value of the symbol with the next higher
790 value. The size of the symbol is printed, rather than the value.
793 @itemx --radix=@var{radix}
794 Use @var{radix} as the radix for printing the symbol values. It must be
795 @samp{d} for decimal, @samp{o} for octal, or @samp{x} for hexadecimal.
797 @item --target=@var{bfdname}
798 @cindex object code format
799 Specify an object code format other than your system's default format.
800 @xref{Target Selection}, for more information.
803 @itemx --undefined-only
804 @cindex external symbols
805 @cindex undefined symbols
806 Display only undefined symbols (those external to each object file).
809 @cindex external symbols
810 @cindex undefined symbols
811 Display only defined symbols for each object file.
815 Show the version number of @code{nm} and exit.
818 Show a summary of the options to @code{nm} and exit.
825 objcopy [ -F @var{bfdname} | --target=@var{bfdname} ]
826 [ -I @var{bfdname} | --input-target=@var{bfdname} ]
827 [ -O @var{bfdname} | --output-target=@var{bfdname} ]
828 [ -S | --strip-all ] [ -g | --strip-debug ]
829 [ -K @var{symbolname} | --keep-symbol=@var{symbolname} ]
830 [ -N @var{symbolname} | --strip-symbol=@var{symbolname} ]
831 [ -L @var{symbolname} | --localize-symbol=@var{symbolname} ]
832 [ -W @var{symbolname} | --weaken-symbol=@var{symbolname} ]
833 [ -x | --discard-all ] [ -X | --discard-locals ]
834 [ -b @var{byte} | --byte=@var{byte} ]
835 [ -i @var{interleave} | --interleave=@var{interleave} ]
836 [ -R @var{sectionname} | --remove-section=@var{sectionname} ]
837 [ -p | --preserve-dates ] [ --debugging ]
838 [ --gap-fill=@var{val} ] [ --pad-to=@var{address} ]
839 [ --set-start=@var{val} ] [ --adjust-start=@var{incr} ]
840 [ --adjust-vma=@var{incr} ]
841 [ --adjust-section-vma=@var{section}@{=,+,-@}@var{val} ]
842 [ --adjust-warnings ] [ --no-adjust-warnings ]
843 [ --set-section-flags=@var{section}=@var{flags} ]
844 [ --add-section=@var{sectionname}=@var{filename} ]
845 [ --change-leading-char ] [ --remove-leading-char ]
847 [ -v | --verbose ] [ -V | --version ] [ --help ]
848 @var{infile} [@var{outfile}]
851 The @sc{gnu} @code{objcopy} utility copies the contents of an object
852 file to another. @code{objcopy} uses the @sc{gnu} @sc{bfd} Library to
853 read and write the object files. It can write the destination object
854 file in a format different from that of the source object file. The
855 exact behavior of @code{objcopy} is controlled by command-line options.
857 @code{objcopy} creates temporary files to do its translations and
858 deletes them afterward. @code{objcopy} uses @sc{bfd} to do all its
859 translation work; it has access to all the formats described in @sc{bfd}
860 and thus is able to recognize most formats without being told
861 explicitly. @xref{BFD,,BFD,ld.info,Using LD}.
863 @code{objcopy} can be used to generate S-records by using an output
864 target of @samp{srec} (e.g., use @samp{-O srec}).
866 @code{objcopy} can be used to generate a raw binary file by using an
867 output target of @samp{binary} (e.g., use @samp{-O binary}). When
868 @code{objcopy} generates a raw binary file, it will essentially produce
869 a memory dump of the contents of the input object file. All symbols and
870 relocation information will be discarded. The memory dump will start at
871 the load address of the lowest section copied into the output file.
873 When generating an S-record or a raw binary file, it may be helpful to
874 use @samp{-S} to remove sections containing debugging information. In
875 some cases @samp{-R} will be useful to remove sections which contain
876 information which is not needed by the binary file.
881 The source and output files, respectively.
882 If you do not specify @var{outfile}, @code{objcopy} creates a
883 temporary file and destructively renames the result with
884 the name of @var{infile}.
886 @item -I @var{bfdname}
887 @itemx --input-target=@var{bfdname}
888 Consider the source file's object format to be @var{bfdname}, rather than
889 attempting to deduce it. @xref{Target Selection}, for more information.
891 @item -O @var{bfdname}
892 @itemx --output-target=@var{bfdname}
893 Write the output file using the object format @var{bfdname}.
894 @xref{Target Selection}, for more information.
896 @item -F @var{bfdname}
897 @itemx --target=@var{bfdname}
898 Use @var{bfdname} as the object format for both the input and the output
899 file; i.e., simply transfer data from source to destination with no
900 translation. @xref{Target Selection}, for more information.
902 @item -R @var{sectionname}
903 @itemx --remove-section=@var{sectionname}
904 Remove any section named @var{sectionname} from the output file. This
905 option may be given more than once. Note that using this option
906 inappropriately may make the output file unusable.
910 Do not copy relocation and symbol information from the source file.
914 Do not copy debugging symbols from the source file.
916 @item --strip-unneeded
917 Strip all symbols that are not needed for relocation processing.
919 @item -K @var{symbolname}
920 @itemx --keep-symbol=@var{symbolname}
921 Copy only symbol @var{symbolname} from the source file. This option may
922 be given more than once.
924 @item -N @var{symbolname}
925 @itemx --strip-symbol=@var{symbolname}
926 Do not copy symbol @var{symbolname} from the source file. This option
927 may be given more than once.
929 @item -L @var{symbolname}
930 @itemx --localize-symbol=@var{symbolname}
931 Make symbol @var{symbolname} local to the file, so that it is not
932 visible externally. This option may be given more than once.
934 @item -W @var{symbolname}
935 @itemx --weaken-symbol=@var{symbolname}
936 Make symbol @var{symbolname} weak. This option may be given more than once.
940 Do not copy non-global symbols from the source file.
941 @c FIXME any reason to prefer "non-global" to "local" here?
944 @itemx --discard-locals
945 Do not copy compiler-generated local symbols.
946 (These usually start with @samp{L} or @samp{.}.)
949 @itemx --byte=@var{byte}
950 Keep only every @var{byte}th byte of the input file (header data is not
951 affected). @var{byte} can be in the range from 0 to @var{interleave}-1,
952 where @var{interleave} is given by the @samp{-i} or @samp{--interleave}
953 option, or the default of 4. This option is useful for creating files
954 to program @sc{rom}. It is typically used with an @code{srec} output
957 @item -i @var{interleave}
958 @itemx --interleave=@var{interleave}
959 Only copy one out of every @var{interleave} bytes. Select which byte to
960 copy with the @var{-b} or @samp{--byte} option. The default is 4.
961 @code{objcopy} ignores this option if you do not specify either @samp{-b} or
965 @itemx --preserve-dates
966 Set the access and modification dates of the output file to be the same
967 as those of the input file.
970 Convert debugging information, if possible. This is not the default
971 because only certain debugging formats are supported, and the
972 conversion process can be time consuming.
974 @item --gap-fill @var{val}
975 Fill gaps between sections with @var{val}. This operation applies to
976 the @emph{load address} (LMA) of the sections. It is done by increasing
977 the size of the section with the lower address, and filling in the extra
978 space created with @var{val}.
980 @item --pad-to @var{address}
981 Pad the output file up to the load address @var{address}. This is
982 done by increasing the size of the last section. The extra space is
983 filled in with the value specified by @samp{--gap-fill} (default zero).
985 @item --set-start @var{val}
986 Set the address of the new file to @var{val}. Not all object file
987 formats support setting the start address.
989 @item --adjust-start @var{incr}
990 Adjust the start address by adding @var{incr}. Not all object file
991 formats support setting the start address.
993 @item --adjust-vma @var{incr}
994 Adjust the address of all sections, as well as the start address, by
995 adding @var{incr}. Some object file formats do not permit section
996 addresses to be changed arbitrarily. Note that this does not relocate
997 the sections; if the program expects sections to be loaded at a certain
998 address, and this option is used to change the sections such that they
999 are loaded at a different address, the program may fail.
1001 @item --adjust-section-vma @var{section}@{=,+,-@}@var{val}
1002 Set or adjust the address of the named @var{section}. If @samp{=} is
1003 used, the section address is set to @var{val}. Otherwise, @var{val} is
1004 added to or subtracted from the section address. See the comments under
1005 @samp{--adjust-vma}, above. If @var{section} does not exist in the
1006 input file, a warning will be issued, unless @samp{--no-adjust-warnings}
1009 @item --adjust-warnings
1010 If @samp{--adjust-section-vma} is used, and the named section does not
1011 exist, issue a warning. This is the default.
1013 @item --no-adjust-warnings
1014 Do not issue a warning if @samp{--adjust-section-vma} is used, even if
1015 the named section does not exist.
1017 @item --set-section-flags @var{section}=@var{flags}
1018 Set the flags for the named section. The @var{flags} argument is a
1019 comma separated string of flag names. The recognized names are
1020 @samp{alloc}, @samp{contents}, @samp{load}, @samp{readonly},
1021 @samp{code}, @samp{data}, and @samp{rom}. You can set the
1022 @samp{contents} flag for a section which does not have contents, but it
1023 is not meaningful to clear the @samp{contents} flag of a section which
1024 does have contents--just remove the section instead. Not all flags are
1025 meaningful for all object file formats.
1027 @item --add-section @var{sectionname}=@var{filename}
1028 Add a new section named @var{sectionname} while copying the file. The
1029 contents of the new section are taken from the file @var{filename}. The
1030 size of the section will be the size of the file. This option only
1031 works on file formats which can support sections with arbitrary names.
1033 @item --change-leading-char
1034 Some object file formats use special characters at the start of
1035 symbols. The most common such character is underscore, which compilers
1036 often add before every symbol. This option tells @code{objcopy} to
1037 change the leading character of every symbol when it converts between
1038 object file formats. If the object file formats use the same leading
1039 character, this option has no effect. Otherwise, it will add a
1040 character, or remove a character, or change a character, as
1043 @item --remove-leading-char
1044 If the first character of a global symbol is a special symbol leading
1045 character used by the object file format, remove the character. The
1046 most common symbol leading character is underscore. This option will
1047 remove a leading underscore from all global symbols. This can be useful
1048 if you want to link together objects of different file formats with
1049 different conventions for symbol names. This is different from
1050 @code{--change-leading-char} because it always changes the symbol name
1051 when appropriate, regardless of the object file format of the output
1055 Change all global symbols in the file to be weak. This can be useful
1056 when building an object which will be linked against other objects using
1057 the @code{-R} option to the linker. This option is only effective when
1058 using an object file format which supports weak symbols.
1062 Show the version number of @code{objcopy}.
1066 Verbose output: list all object files modified. In the case of
1067 archives, @samp{objcopy -V} lists all members of the archive.
1070 Show a summary of the options to @code{objcopy}.
1076 @cindex object file information
1080 objdump [ -a | --archive-headers ]
1081 [ -b @var{bfdname} | --target=@var{bfdname} ] [ --debugging ]
1082 [ -C | --demangle ] [ -d | --disassemble ]
1083 [ -D | --disassemble-all ] [ --disassemble-zeroes ]
1084 [ -EB | -EL | --endian=@{big | little @} ]
1085 [ -f | --file-headers ]
1086 [ -h | --section-headers | --headers ] [ -i | --info ]
1087 [ -j @var{section} | --section=@var{section} ]
1088 [ -l | --line-numbers ] [ -S | --source ]
1089 [ -m @var{machine} | --architecture=@var{machine} ]
1090 [ -r | --reloc ] [ -R | --dynamic-reloc ]
1091 [ -s | --full-contents ] [ --stabs ]
1092 [ -t | --syms ] [ -T | --dynamic-syms ] [ -x | --all-headers ]
1093 [ -w | --wide ] [ --start-address=@var{address} ]
1094 [ --stop-address=@var{address} ]
1095 [ --prefix-addresses] [ --[no-]show-raw-insn ]
1096 [ --adjust-vma=@var{offset} ]
1097 [ --version ] [ --help ]
1098 @var{objfile}@dots{}
1101 @code{objdump} displays information about one or more object files.
1102 The options control what particular information to display. This
1103 information is mostly useful to programmers who are working on the
1104 compilation tools, as opposed to programmers who just want their
1105 program to compile and work.
1107 @var{objfile}@dots{} are the object files to be examined. When you
1108 specify archives, @code{objdump} shows information on each of the member
1111 The long and short forms of options, shown here as alternatives, are
1112 equivalent. At least one option besides @samp{-l} must be given.
1116 @itemx --archive-header
1117 @cindex archive headers
1118 If any of the @var{objfile} files are archives, display the archive
1119 header information (in a format similar to @samp{ls -l}). Besides the
1120 information you could list with @samp{ar tv}, @samp{objdump -a} shows
1121 the object file format of each archive member.
1123 @item --adjust-vma=@var{offset}
1124 @cindex section addresses in objdump
1125 @cindex VMA in objdump
1126 When dumping information, first add @var{offset} to all the section
1127 addresses. This is useful if the section addresses do not correspond to
1128 the symbol table, which can happen when putting sections at particular
1129 addresses when using a format which can not represent section addresses,
1132 @item -b @var{bfdname}
1133 @itemx --target=@var{bfdname}
1134 @cindex object code format
1135 Specify that the object-code format for the object files is
1136 @var{bfdname}. This option may not be necessary; @var{objdump} can
1137 automatically recognize many formats.
1141 objdump -b oasys -m vax -h fu.o
1144 displays summary information from the section headers (@samp{-h}) of
1145 @file{fu.o}, which is explicitly identified (@samp{-m}) as a VAX object
1146 file in the format produced by Oasys compilers. You can list the
1147 formats available with the @samp{-i} option.
1148 @xref{Target Selection}, for more information.
1152 @cindex demangling in objdump
1153 Decode (@dfn{demangle}) low-level symbol names into user-level names.
1154 Besides removing any initial underscore prepended by the system, this
1155 makes C++ function names readable. @xref{c++filt}, for more information
1159 Display debugging information. This attempts to parse debugging
1160 information stored in the file and print it out using a C like syntax.
1161 Only certain types of debugging information have been implemented.
1164 @itemx --disassemble
1165 @cindex disassembling object code
1166 @cindex machine instructions
1167 Display the assembler mnemonics for the machine instructions from
1168 @var{objfile}. This option only disassembles those sections which are
1169 expected to contain instructions.
1172 @itemx --disassemble-all
1173 Like @samp{-d}, but disassemble the contents of all sections, not just
1174 those expected to contain instructions.
1176 @item --prefix-addresses
1177 When disassembling, print the complete address on each line. This is
1178 the older disassembly format.
1180 @item --disassemble-zeroes
1181 Normally the disassembly output will skip blocks of zeroes. This
1182 option directs the disassembler to disassemble those blocks, just like
1187 @itemx --endian=@{big|little@}
1189 @cindex disassembly endianness
1190 Specify the endianness of the object files. This only affects
1191 disassembly. This can be useful when disassembling a file format which
1192 does not describe endianness information, such as S-records.
1195 @itemx --file-header
1196 @cindex object file header
1197 Display summary information from the overall header of
1198 each of the @var{objfile} files.
1201 @itemx --section-header
1203 @cindex section headers
1204 Display summary information from the section headers of the
1207 File segments may be relocated to nonstandard addresses, for example by
1208 using the @samp{-Ttext}, @samp{-Tdata}, or @samp{-Tbss} options to
1209 @code{ld}. However, some object file formats, such as a.out, do not
1210 store the starting address of the file segments. In those situations,
1211 although @code{ld} relocates the sections correctly, using @samp{objdump
1212 -h} to list the file section headers cannot show the correct addresses.
1213 Instead, it shows the usual addresses, which are implicit for the
1217 Print a summary of the options to @code{objdump} and exit.
1221 @cindex architectures available
1222 @cindex object formats available
1223 Display a list showing all architectures and object formats available
1224 for specification with @samp{-b} or @samp{-m}.
1227 @itemx --section=@var{name}
1228 @cindex section information
1229 Display information only for section @var{name}.
1232 @itemx --line-numbers
1233 @cindex source filenames for object files
1234 Label the display (using debugging information) with the filename and
1235 source line numbers corresponding to the object code or relocs shown.
1236 Only useful with @samp{-d}, @samp{-D}, or @samp{-r}.
1238 @item -m @var{machine}
1239 @itemx --architecture=@var{machine}
1240 @cindex architecture
1241 @cindex disassembly architecture
1242 Specify the architecture to use when disassembling object files. This
1243 can be useful when disassembling object files which do not describe
1244 architecture information, such as S-records. You can list the available
1245 architectures with the @samp{-i} option.
1249 @cindex relocation entries, in object file
1250 Print the relocation entries of the file. If used with @samp{-d} or
1251 @samp{-D}, the relocations are printed interspersed with the
1255 @itemx --dynamic-reloc
1256 @cindex dynamic relocation entries, in object file
1257 Print the dynamic relocation entries of the file. This is only
1258 meaningful for dynamic objects, such as certain types of shared
1262 @itemx --full-contents
1263 @cindex sections, full contents
1264 @cindex object file sections
1265 Display the full contents of any sections requested.
1269 @cindex source disassembly
1270 @cindex disassembly, with source
1271 Display source code intermixed with disassembly, if possible. Implies
1274 @item --show-raw-insn
1275 When disassembling instructions, print the instruction in hex as well as
1276 in symbolic form. This is the default except when
1277 @code{--prefix-addresses} is used.
1279 @item --no-show-raw-insn
1280 When disassembling instructions, do not print the instruction bytes.
1281 This is the default when @code{--prefix-addresses} is used.
1286 @cindex debug symbols
1287 @cindex ELF object file format
1288 Display the full contents of any sections requested. Display the
1289 contents of the .stab and .stab.index and .stab.excl sections from an
1290 ELF file. This is only useful on systems (such as Solaris 2.0) in which
1291 @code{.stab} debugging symbol-table entries are carried in an ELF
1292 section. In most other file formats, debugging symbol-table entries are
1293 interleaved with linkage symbols, and are visible in the @samp{--syms}
1294 output. For more information on stabs symbols, see @ref{Top,Stabs,Stabs
1295 Overview,stabs.info, The ``stabs'' debug format}.
1297 @item --start-address=@var{address}
1298 @cindex start-address
1299 Start displaying data at the specified address. This affects the output
1300 of the @code{-d}, @code{-r} and @code{-s} options.
1302 @item --stop-address=@var{address}
1303 @cindex stop-address
1304 Stop displaying data at the specified address. This affects the output
1305 of the @code{-d}, @code{-r} and @code{-s} options.
1309 @cindex symbol table entries, printing
1310 Print the symbol table entries of the file.
1311 This is similar to the information provided by the @samp{nm} program.
1314 @itemx --dynamic-syms
1315 @cindex dynamic symbol table entries, printing
1316 Print the dynamic symbol table entries of the file. This is only
1317 meaningful for dynamic objects, such as certain types of shared
1318 libraries. This is similar to the information provided by the @samp{nm}
1319 program when given the @samp{-D} (@samp{--dynamic}) option.
1322 Print the version number of @code{objdump} and exit.
1326 @cindex all header information, object file
1327 @cindex header information, all
1328 Display all available header information, including the symbol table and
1329 relocation entries. Using @samp{-x} is equivalent to specifying all of
1330 @samp{-a -f -h -r -t}.
1334 @cindex wide output, printing
1335 Format some lines for output devices that have more than 80 columns.
1342 @cindex archive contents
1343 @cindex symbol index
1346 ranlib [-vV] @var{archive}
1349 @code{ranlib} generates an index to the contents of an archive and
1350 stores it in the archive. The index lists each symbol defined by a
1351 member of an archive that is a relocatable object file.
1353 You may use @samp{nm -s} or @samp{nm --print-armap} to list this index.
1355 An archive with such an index speeds up linking to the library and
1356 allows routines in the library to call each other without regard to
1357 their placement in the archive.
1359 The @sc{gnu} @code{ranlib} program is another form of @sc{gnu} @code{ar}; running
1360 @code{ranlib} is completely equivalent to executing @samp{ar -s}.
1366 Show the version number of @code{ranlib}.
1373 @cindex section sizes
1376 size [ -A | -B | --format=@var{compatibility} ]
1377 [ --help ] [ -d | -o | -x | --radix=@var{number} ]
1378 [ --target=@var{bfdname} ] [ -V | --version ]
1379 [ @var{objfile}@dots{} ]
1382 The @sc{gnu} @code{size} utility lists the section sizes---and the total
1383 size---for each of the object or archive files @var{objfile} in its
1384 argument list. By default, one line of output is generated for each
1385 object file or each module in an archive.
1387 @var{objfile}@dots{} are the object files to be examined.
1388 If none are specified, the file @code{a.out} will be used.
1390 The command line options have the following meanings:
1395 @itemx --format=@var{compatibility}
1396 @cindex @code{size} display format
1397 Using one of these options, you can choose whether the output from @sc{gnu}
1398 @code{size} resembles output from System V @code{size} (using @samp{-A},
1399 or @samp{--format=sysv}), or Berkeley @code{size} (using @samp{-B}, or
1400 @samp{--format=berkeley}). The default is the one-line format similar to
1402 @c Bonus for doc-source readers: you can also say --format=strange (or
1403 @c anything else that starts with 's') for sysv, and --format=boring (or
1404 @c anything else that starts with 'b') for Berkeley.
1406 Here is an example of the Berkeley (default) format of output from
1409 size --format=Berkeley ranlib size
1410 text data bss dec hex filename
1411 294880 81920 11592 388392 5ed28 ranlib
1412 294880 81920 11888 388688 5ee50 size
1416 This is the same data, but displayed closer to System V conventions:
1419 size --format=SysV ranlib size
1437 Show a summary of acceptable arguments and options.
1442 @itemx --radix=@var{number}
1443 @cindex @code{size} number format
1444 @cindex radix for section sizes
1445 Using one of these options, you can control whether the size of each
1446 section is given in decimal (@samp{-d}, or @samp{--radix=10}); octal
1447 (@samp{-o}, or @samp{--radix=8}); or hexadecimal (@samp{-x}, or
1448 @samp{--radix=16}). In @samp{--radix=@var{number}}, only the three
1449 values (8, 10, 16) are supported. The total size is always given in two
1450 radices; decimal and hexadecimal for @samp{-d} or @samp{-x} output, or
1451 octal and hexadecimal if you're using @samp{-o}.
1453 @item --target=@var{bfdname}
1454 @cindex object code format
1455 Specify that the object-code format for @var{objfile} is
1456 @var{bfdname}. This option may not be necessary; @code{size} can
1457 automatically recognize many formats.
1458 @xref{Target Selection}, for more information.
1462 Display the version number of @code{size}.
1468 @cindex listings strings
1469 @cindex printing strings
1470 @cindex strings, printing
1473 strings [-afov] [-@var{min-len}] [-n @var{min-len}] [-t @var{radix}] [-]
1474 [--all] [--print-file-name] [--bytes=@var{min-len}]
1475 [--radix=@var{radix}] [--target=@var{bfdname}]
1476 [--help] [--version] @var{file}@dots{}
1479 For each @var{file} given, @sc{gnu} @code{strings} prints the printable
1480 character sequences that are at least 4 characters long (or the number
1481 given with the options below) and are followed by an unprintable
1482 character. By default, it only prints the strings from the initialized
1483 and loaded sections of object files; for other types of files, it prints
1484 the strings from the whole file.
1486 @code{strings} is mainly useful for determining the contents of non-text
1493 Do not scan only the initialized and loaded sections of object files;
1494 scan the whole files.
1497 @itemx --print-file-name
1498 Print the name of the file before each string.
1501 Print a summary of the program usage on the standard output and exit.
1503 @item -@var{min-len}
1504 @itemx -n @var{min-len}
1505 @itemx --bytes=@var{min-len}
1506 Print sequences of characters that are at least @var{min-len} characters
1507 long, instead of the default 4.
1510 Like @samp{-t o}. Some other versions of @code{strings} have @samp{-o}
1511 act like @samp{-t d} instead. Since we can not be compatible with both
1512 ways, we simply chose one.
1514 @item -t @var{radix}
1515 @itemx --radix=@var{radix}
1516 Print the offset within the file before each string. The single
1517 character argument specifies the radix of the offset---@samp{o} for
1518 octal, @samp{x} for hexadecimal, or @samp{d} for decimal.
1520 @item --target=@var{bfdname}
1521 @cindex object code format
1522 Specify an object code format other than your system's default format.
1523 @xref{Target Selection}, for more information.
1527 Print the program version number on the standard output and exit.
1534 @cindex removing symbols
1535 @cindex discarding symbols
1536 @cindex symbols, discarding
1539 strip [ -F @var{bfdname} | --target=@var{bfdname} ]
1540 [ -I @var{bfdname} | --input-target=@var{bfdname} ]
1541 [ -O @var{bfdname} | --output-target=@var{bfdname} ]
1542 [ -s | --strip-all ] [ -S | -g | --strip-debug ]
1543 [ -K @var{symbolname} | --keep-symbol=@var{symbolname} ]
1544 [ -N @var{symbolname} | --strip-symbol=@var{symbolname} ]
1545 [ -x | --discard-all ] [ -X | --discard-locals ]
1546 [ -R @var{sectionname} | --remove-section=@var{sectionname} ]
1547 [ -o @var{file} ] [ -p | --preserve-dates ]
1548 [ -v | --verbose ] [ -V | --version ] [ --help ]
1549 @var{objfile}@dots{}
1552 @sc{gnu} @code{strip} discards all symbols from object files
1553 @var{objfile}. The list of object files may include archives.
1554 At least one object file must be given.
1556 @code{strip} modifies the files named in its argument,
1557 rather than writing modified copies under different names.
1560 @item -F @var{bfdname}
1561 @itemx --target=@var{bfdname}
1562 Treat the original @var{objfile} as a file with the object
1563 code format @var{bfdname}, and rewrite it in the same format.
1564 @xref{Target Selection}, for more information.
1567 Show a summary of the options to @code{strip} and exit.
1569 @item -I @var{bfdname}
1570 @itemx --input-target=@var{bfdname}
1571 Treat the original @var{objfile} as a file with the object
1572 code format @var{bfdname}.
1573 @xref{Target Selection}, for more information.
1575 @item -O @var{bfdname}
1576 @itemx --output-target=@var{bfdname}
1577 Replace @var{objfile} with a file in the output format @var{bfdname}.
1578 @xref{Target Selection}, for more information.
1580 @item -R @var{sectionname}
1581 @itemx --remove-section=@var{sectionname}
1582 Remove any section named @var{sectionname} from the output file. This
1583 option may be given more than once. Note that using this option
1584 inappropriately may make the output file unusable.
1592 @itemx --strip-debug
1593 Remove debugging symbols only.
1595 @item --strip-unneeded
1596 Remove all symbols that are not needed for relocation processing.
1598 @item -K @var{symbolname}
1599 @itemx --keep-symbol=@var{symbolname}
1600 Keep only symbol @var{symbolname} from the source file. This option may
1601 be given more than once.
1603 @item -N @var{symbolname}
1604 @itemx --strip-symbol=@var{symbolname}
1605 Remove symbol @var{symbolname} from the source file. This option may be
1606 given more than once, and may be combined with strip options other than
1610 Put the stripped output in @var{file}, rather than replacing the
1611 existing file. When this argument is used, only one @var{objfile}
1612 argument may be specified.
1615 @itemx --preserve-dates
1616 Preserve the access and modification dates of the file.
1619 @itemx --discard-all
1620 Remove non-global symbols.
1623 @itemx --discard-locals
1624 Remove compiler-generated local symbols.
1625 (These usually start with @samp{L} or @samp{.}.)
1629 Show the version number for @code{strip}.
1633 Verbose output: list all object files modified. In the case of
1634 archives, @samp{strip -v} lists all members of the archive.
1641 @cindex demangling C++ symbols
1644 c++filt [ -_ | --strip-underscores ]
1645 [ -n | --no-strip-underscores ]
1646 [ -s @var{format} | --format=@var{format} ]
1647 [ --help ] [ --version ] [ @var{symbol}@dots{} ]
1650 The C++ language provides function overloading, which means that you can
1651 write many functions with the same name (providing each takes parameters
1652 of different types). All C++ function names are encoded into a
1653 low-level assembly label (this process is known as
1654 @dfn{mangling}). The @code{c++filt} program does the inverse mapping: it
1655 decodes (@dfn{demangles}) low-level names into user-level names so that
1656 the linker can keep these overloaded functions from clashing.
1658 Every alphanumeric word (consisting of letters, digits, underscores,
1659 dollars, or periods) seen in the input is a potential label. If the
1660 label decodes into a C++ name, the C++ name replaces the low-level
1663 You can use @code{c++filt} to decipher individual symbols:
1666 c++filt @var{symbol}
1669 If no @var{symbol} arguments are given, @code{c++filt} reads symbol
1670 names from the standard input and writes the demangled names to the
1671 standard output. All results are printed on the standard output.
1675 @itemx --strip-underscores
1676 On some systems, both the C and C++ compilers put an underscore in front
1677 of every name. For example, the C name @code{foo} gets the low-level
1678 name @code{_foo}. This option removes the initial underscore. Whether
1679 @code{c++filt} removes the underscore by default is target dependent.
1682 @itemx --no-strip-underscores
1683 Do not remove the initial underscore.
1685 @item -s @var{format}
1686 @itemx --format=@var{format}
1687 @sc{gnu} @code{nm} can decode three different methods of mangling, used by
1688 different C++ compilers. The argument to this option selects which
1693 the one used by the @sc{gnu} compiler (the default method)
1695 the one used by the Lucid compiler
1697 the one specified by the C++ Annotated Reference Manual
1701 Print a summary of the options to @code{c++filt} and exit.
1704 Print the version number of @code{c++filt} and exit.
1708 @emph{Warning:} @code{c++filt} is a new utility, and the details of its
1709 user interface are subject to change in future releases. In particular,
1710 a command-line option may be required in the the future to decode a name
1711 passed as an argument on the command line; in other words,
1714 c++filt @var{symbol}
1718 may in a future release become
1721 c++filt @var{option} @var{symbol}
1729 @cindex address to file name and line number
1732 addr2line [ -b @var{bfdname} | --target=@var{bfdname} ]
1734 [ -e @var{filename} | --exe=@var{filename} ]
1735 [ -f | --functions ] [ -s | --basename ]
1736 [ -H | --help ] [ -V | --version ]
1740 @code{addr2line} translates program addresses into file names and line
1741 numbers. Given an address and an executable, it uses the debugging
1742 information in the executable to figure out which file name and line
1743 number are associated with a given address.
1745 The executable to use is specified with the @code{-e} option. The
1746 default is @file{a.out}.
1748 @code{addr2line} has two modes of operation.
1750 In the first, hexadecimal addresses are specified on the command line,
1751 and @code{addr2line} displays the file name and line number for each
1754 In the second, @code{addr2line} reads hexadecimal addresses from
1755 standard input, and prints the file name and line number for each
1756 address on standard output. In this mode, @code{addr2line} may be used
1757 in a pipe to convert dynamically chosen addresses.
1759 The format of the output is @samp{FILENAME:LINENO}. The file name and
1760 line number for each address is printed on a separate line. If the
1761 @code{-f} option is used, then each @samp{FILENAME:LINENO} line is
1762 preceded by a @samp{FUNCTIONNAME} line which is the name of the function
1763 containing the address.
1765 If the file name or function name can not be determined,
1766 @code{addr2line} will print two question marks in their place. If the
1767 line number can not be determined, @code{addr2line} will print 0.
1769 The long and short forms of options, shown here as alternatives, are
1773 @item -b @var{bfdname}
1774 @itemx --target=@var{bfdname}
1775 @cindex object code format
1776 Specify that the object-code format for the object files is
1781 @cindex demangling in objdump
1782 Decode (@dfn{demangle}) low-level symbol names into user-level names.
1783 Besides removing any initial underscore prepended by the system, this
1784 makes C++ function names readable. @xref{c++filt}, for more information
1787 @item -e @var{filename}
1788 @itemx --exe=@var{filename}
1789 Specify the name of the executable for which addresses should be
1790 translated. The default file is @file{a.out}.
1794 Display function names as well as file and line number information.
1798 Display only the base of each file name.
1804 @code{nlmconv} converts a relocatable object file into a NetWare
1808 @code{nlmconv} currently works with @samp{i386} object
1809 files in @code{coff}, @sc{elf}, or @code{a.out} format, and @sc{SPARC}
1810 object files in @sc{elf}, or @code{a.out} format@footnote{
1811 @code{nlmconv} should work with any @samp{i386} or @sc{sparc} object
1812 format in the Binary File Descriptor library. It has only been tested
1813 with the above formats.}.
1817 @emph{Warning:} @code{nlmconv} is not always built as part of the binary
1818 utilities, since it is only useful for NLM targets.
1822 nlmconv [ -I @var{bfdname} | --input-target=@var{bfdname} ]
1823 [ -O @var{bfdname} | --output-target=@var{bfdname} ]
1824 [ -T @var{headerfile} | --header-file=@var{headerfile} ]
1825 [ -d | --debug] [ -l @var{linker} | --linker=@var{linker} ]
1826 [ -h | --help ] [ -V | --version ]
1827 @var{infile} @var{outfile}
1830 @code{nlmconv} converts the relocatable @samp{i386} object file
1831 @var{infile} into the NetWare Loadable Module @var{outfile}, optionally
1832 reading @var{headerfile} for NLM header information. For instructions
1833 on writing the NLM command file language used in header files, see the
1834 @samp{linkers} section, @samp{NLMLINK} in particular, of the @cite{NLM
1835 Development and Tools Overview}, which is part of the NLM Software
1836 Developer's Kit (``NLM SDK''), available from Novell, Inc.
1837 @code{nlmconv} uses the @sc{gnu} Binary File Descriptor library to read
1838 @var{infile}; see @ref{BFD,,BFD,ld.info,Using LD}, for
1841 @code{nlmconv} can perform a link step. In other words, you can list
1842 more than one object file for input if you list them in the definitions
1843 file (rather than simply specifying one input file on the command line).
1844 In this case, @code{nlmconv} calls the linker for you.
1847 @item -I @var{bfdname}
1848 @itemx --input-target=@var{bfdname}
1849 Object format of the input file. @code{nlmconv} can usually determine
1850 the format of a given file (so no default is necessary).
1851 @xref{Target Selection}, for more information.
1853 @item -O @var{bfdname}
1854 @itemx --output-target=@var{bfdname}
1855 Object format of the output file. @code{nlmconv} infers the output
1856 format based on the input format, e.g. for a @samp{i386} input file the
1857 output format is @samp{nlm32-i386}.
1858 @xref{Target Selection}, for more information.
1860 @item -T @var{headerfile}
1861 @itemx --header-file=@var{headerfile}
1862 Reads @var{headerfile} for NLM header information. For instructions on
1863 writing the NLM command file language used in header files, see@ see the
1864 @samp{linkers} section, of the @cite{NLM Development and Tools
1865 Overview}, which is part of the NLM Software Developer's Kit, available
1870 Displays (on standard error) the linker command line used by @code{nlmconv}.
1872 @item -l @var{linker}
1873 @itemx --linker=@var{linker}
1874 Use @var{linker} for any linking. @var{linker} can be an absolute or a
1879 Prints a usage summary.
1883 Prints the version number for @code{nlmconv}.
1889 @code{windres} may be used to manipulate Windows resources.
1892 @emph{Warning:} @code{windres} is not always built as part of the binary
1893 utilities, since it is only useful for Windows targets.
1897 windres [options] [input-file] [output-file]
1900 @code{windres} reads resources from an input file and copies them into
1901 an output file. Either file may be in one of three formats:
1905 A text format read by the Resource Compiler.
1908 A binary format generated by the Resource Compiler.
1911 A COFF object or executable.
1914 The exact description of these different formats is available in
1915 documentation from Microsoft.
1917 When @code{windres} converts from the @code{rc} format to the @code{res}
1918 format, it is acting like the Windows Resource Compiler. When
1919 @code{windres} converts from the @code{res} format to the @code{coff}
1920 format, it is acting like the Windows @code{CVTRES} program.
1922 When @code{windres} generates an @code{rc} file, the output is similar
1923 but not identical to the format expected for the input. When an input
1924 @code{rc} file refers to an external filename, an output @code{rc} file
1925 will instead include the file contents.
1927 If the input or output format is not specified, @code{windres} will
1928 guess based on the file name, or, for the input file, the file contents.
1929 A file with an extension of @file{.rc} will be treated as an @code{rc}
1930 file, a file with an extension of @file{.res} will be treated as a
1931 @code{res} file, and a file with an extension of @file{.o} or
1932 @file{.exe} will be treated as a @code{coff} file.
1934 If no output file is specified, @code{windres} will print the resources
1935 in @code{rc} format to standard output.
1937 The normal use is for you to write an @code{rc} file, use @code{windres}
1938 to convert it to a COFF object file, and then link the COFF file into
1939 your application. This will make the resources described in the
1940 @code{rc} file available to Windows.
1943 @item -i @var{filename}
1944 @itemx --input @var{filename}
1945 The name of the input file. If this option is not used, then
1946 @code{windres} will use the first non-option argument as the input file
1947 name. If there are no non-option arguments, then @code{windres} will
1948 read from standard input. @code{windres} can not read a COFF file from
1951 @item -o @var{filename}
1952 @itemx --output @var{filename}
1953 The name of the output file. If this option is not used, then
1954 @code{windres} will use the first non-option argument, after any used
1955 for the input file name, as the output file name. If there is no
1956 non-option argument, then @code{windres} will write to standard output.
1957 @code{windres} can not write a COFF file to standard output.
1959 @item -I @var{format}
1960 @itemx --input-format @var{format}
1961 The input format to read. @var{format} may be @samp{res}, @samp{rc}, or
1962 @samp{coff}. If no input format is specified, @code{windres} will
1963 guess, as described above.
1965 @item -O @var{format}
1966 @itemx --output-format @var{format}
1967 The output format to generate. @var{format} may be @samp{res},
1968 @samp{rc}, or @samp{coff}. If no output format is specified,
1969 @code{windres} will guess, as described above.
1971 @item -F @var{target}
1972 @itemx --target @var{target}
1973 Specify the BFD format to use for a COFF file as input or output. This
1974 is a BFD target name; you can use the @code{--help} option to see a list
1975 of supported targets. Normally @code{windres} will use the default
1976 format, which is the first one listed by the @code{--help} option.
1977 @ref{Target Selection}.
1979 @item --preprocessor @var{program}
1980 When @code{windres} reads an @code{rc} file, it runs it through the C
1981 preprocessor first. This option may be used to specify the preprocessor
1982 to use, including any leading arguments. The default preprocessor
1983 argument is @code{gcc -E -xc-header -DRC_INVOKED}.
1985 @item --include-dir @var{directory}
1986 Specify an include directory to use when reading an @code{rc} file.
1987 @code{windres} will pass this to the preprocessor as an @code{-I}
1988 option. @code{windres} will also search this directory when looking for
1989 files named in the @code{rc} file.
1991 @item --define @var{sym[=val]}
1992 Specify a @code{-D} option to pass to the preprocessor when reading an
1995 @item --language @var{val}
1996 Specify the default language to use when reading an @code{rc} file.
1997 @var{val} should be a hexadecimal language code. The low eight bits are
1998 the language, and the high eight bits are the sublanguage.
2001 Prints a usage summary.
2004 Prints the version number for @code{windres}.
2007 If @code{windres} is compiled with @code{YYDEBUG} defined as @code{1},
2008 this will turn on parser debugging.
2013 @chapter Create files needed to build and use DLLs
2017 @code{dlltool} may be used to create the files needed to build and use
2018 dynamic link libraries (DLLs).
2021 @emph{Warning:} @code{dlltool} is not always built as part of the binary
2022 utilities, since it is only useful for those targets which support DLLs.
2026 dlltool [-d|--input-def <def-file-name>]
2027 [-b|--base-file <base-file-name>]
2028 [-e|--output-exp <exports-file-name>]
2029 [-z|--output-def <def-file-name>]
2030 [-l|--output-lib <library-file-name>]
2031 [-S|--as <path-to-assembler>] [-f|--as-flags <options>]
2032 [-D|--dllname <name>] [-m|--machine <machine>]
2033 [-a|--add-indirect] [-U|--add-underscore] [-k|--kill-at]
2034 [-x|--no-idata4] [-c|--no-idata5] [-i|--interwork]
2035 [-n|--nodelete] [-v|--verbose] [-h|--help] [-V|--version]
2036 [object-file @dots{}]
2039 @code{dlltool} reads its inputs, which can come from the @samp{-d} and
2040 @samp{-b} options as well as object files specified on the command
2041 line. It then processes these inputs and if the @samp{-e} option has
2042 been specified it creates a exports file. If the @samp{-l} option
2043 has been specified it creates a library file and if the @samp{-z} option
2044 has been specified it creates a def file. Any or all of the -e, -l
2045 and -z options can be present in one invocation of dlltool.
2047 When creating a DLL, along with the source for the DLL, it is necessary
2048 to have three other files. @code{dlltool} can help with the creation of
2051 The first file is a @samp{.def} file which specifies which functions are
2052 exported from the DLL, which functions the DLL imports, and so on. This
2053 is a text file and can be created by hand, or @code{dlltool} can be used
2054 to create it using the @samp{-z} option. In this case @code{dlltool}
2055 will scan the object files specified on its command line looking for
2056 those functions which have been specially marked as being exported and
2057 put entries for them in the .def file it creates.
2059 In order to mark a function as being exported from a DLL, it needs to
2060 have an @samp{-export:<name_of_function>} entry in the @samp{.drective}
2061 section of the object file. This can be done in C by using the
2065 asm (".section .drective");
2066 asm (".ascii \"-export:my_func\"");
2068 int my_func (void) @{ @dots{} @}
2071 The second file needed for DLL creation is an exports file. This file
2072 is linked with the object files that make up the body of the DLL and it
2073 handles the interface between the DLL and the outside world. This is a
2074 binary file and it can be created by giving the @samp{-e} option to
2075 @code{dlltool} when it is creating or reading in a .def file.
2077 The third file needed for DLL creation is the library file that programs
2078 will link with in order to access the functions in the DLL. This file
2079 can be created by giving the @samp{-l} option to dlltool when it
2080 is creating or reading in a .def file.
2082 @code{dlltool} builds the library file by hand, but it builds the
2083 exports file by creating temporary files containing assembler statements
2084 and then assembling these. The @samp{-S} command line option can be
2085 used to specify the path to the assembler that dlltool will use,
2086 and the @samp{-f} option can be used to pass specific flags to that
2087 assembler. The @samp{-n} can be used to prevent dlltool from deleting
2088 these temporary assembler files when it is done, and if @samp{-n} is
2089 specified twice then this will prevent dlltool from deleting the
2090 temporary object files it used to build the library.
2092 Here is an example of creating a DLL from a source file @samp{dll.c} and
2093 also creating a program (from an object file called @samp{program.o})
2098 dlltool -e exports.o -l dll.lib dll.o
2099 gcc dll.o exports.o -o dll.dll
2100 gcc program.o dll.lib -o program
2103 The command line options have the following meanings:
2108 @itemx --input-def FILENAME
2109 @cindex input .def file
2110 Specifies the name of a .def file to be read in and processed.
2113 @itemx --base-file FILENAME
2115 Specifies the name of a base file to be read in and processed. The
2116 contents of this file will be added to the relocation section in the
2117 exports file generated by dlltool.
2120 @itemx --output-exp FILENAME
2121 Specifies the name of the export file to be created by dlltool.
2124 @itemx --output-def FILENAME
2125 Specifies the name of the .def file to be created by dlltool.
2128 @itemx --output-lib FILENAME
2129 Specifies the name of the library file to be created by dlltool.
2133 Specifies the path, including the filename, of the assembler to be used
2134 to create the exports file.
2137 @itemx --as-flags SWITCHES
2138 Specifies any specific command line switches to be passed to the
2139 assembler when building the exports file. This option will work even if
2140 the @samp{-S} option is not used. This option only takes one argument,
2141 and if it occurs more than once on the command line, then later
2142 occurrences will override earlier occurrences. So if it is necessary to
2143 pass multiple switches to the assembler they should be enclosed in
2147 @itemx --dll-name NAME
2148 Specifies the name to be stored in the .def file as the name of the DLL
2149 when the @samp{-e} option is used. If this option is not present, then
2150 the filename given to the @samp{-e} option will be used as the name of
2154 @itemx -machine MACHINE
2155 Specifies the type of machine for which the library file should be
2156 built. @code{dlltool} has a built in default type, depending upon how
2157 it was created, but this option can be used to override that. This is
2158 normally only useful when creating DLLs for an ARM processor, when the
2159 contents of the DLL are actually encode using THUMB instructions.
2162 @itemx --add-indirect
2163 Specifies that when @code{dlltool} is creating the exports file it
2164 should add a section which allows the exported functions to be
2165 referenced without using the import library. Whatever the hell that
2169 @itemx --add-underscore
2170 Specifies that when @code{dlltool} is creating the exports file it
2171 should prepend an underscore to the names of the exported functions.
2175 Specifies that when @code{dlltool} is creating the exports file it
2176 should not append the string @samp{@@ <number>}. These numbers are
2177 called ordinal numbers and they represent another way of accessing the
2178 function in a DLL, other than by name.
2182 Specifies that when @code{dlltool} is creating the exports and library
2183 files it should omit the .idata4 section. This is for compatibility
2184 with certain operating systems.
2188 Specifies that when @code{dlltool} is creating the exports and library
2189 files it should omit the .idata5 section. This is for compatibility
2190 with certain operating systems.
2194 Specifies that @code{dlltool} should mark the objects in the library
2195 file and exports file that it produces as supporting interworking
2196 between ARM and THUMB code.
2200 Makes @code{dlltool} preserve the temporary assembler files it used to
2201 create the exports file. If this option is repeated then dlltool will
2202 also preserve the temporary object files it uses to create the library
2207 Make dlltool describe what it is doing.
2211 Displays a list of command line options and then exits.
2215 Displays dlltool's version number and then exits.
2221 @node Selecting The Target System
2222 @chapter Selecting the target system
2224 You can specify three aspects of the target system to the @sc{gnu}
2225 binary file utilities, each in several ways:
2235 the linker emulation (which applies to the linker only)
2238 In the following summaries, the lists of ways to specify values are in
2239 order of decreasing precedence. The ways listed first override those
2242 The commands to list valid values only list the values for which the
2243 programs you are running were configured. If they were configured with
2244 @samp{--enable-targets=all}, the commands list most of the available
2245 values, but a few are left out; not all targets can be configured in at
2246 once because some of them can only be configured @dfn{native} (on hosts
2247 with the same type as the target system).
2250 * Target Selection::
2251 * Architecture Selection::
2252 * Linker Emulation Selection::
2255 @node Target Selection
2256 @section Target Selection
2258 A @dfn{target} is an object file format. A given target may be
2259 supported for multiple architectures (@pxref{Architecture Selection}).
2260 A target selection may also have variations for different operating
2261 systems or architectures.
2263 The command to list valid target values is @samp{objdump -i}
2264 (the first column of output contains the relevant information).
2266 Some sample values are: @samp{a.out-hp300bsd}, @samp{ecoff-littlemips},
2267 @samp{a.out-sunos-big}.
2269 You can also specify a target using a configuration triplet. This is
2270 the same sort of name that is passed to configure to specify a target.
2271 When you use a configuration triplet as an argument, it must be fully
2272 canonicalized. You can see the canonical version of a triplet by
2273 running the shell script @file{config.sub} which is included with the
2276 Some sample configuration triplets are: @samp{m68k-hp-bsd},
2277 @samp{mips-dec-ultrix}, @samp{sparc-sun-sunos}.
2279 @subheading @code{objdump} Target
2285 command line option: @samp{-b} or @samp{--target}
2288 environment variable @code{GNUTARGET}
2291 deduced from the input file
2294 @subheading @code{objcopy} and @code{strip} Input Target
2300 command line options: @samp{-I} or @samp{--input-target}, or @samp{-F} or @samp{--target}
2303 environment variable @code{GNUTARGET}
2306 deduced from the input file
2309 @subheading @code{objcopy} and @code{strip} Output Target
2315 command line options: @samp{-O} or @samp{--output-target}, or @samp{-F} or @samp{--target}
2318 the input target (see ``@code{objcopy} and @code{strip} Input Target'' above)
2321 environment variable @code{GNUTARGET}
2324 deduced from the input file
2327 @subheading @code{nm}, @code{size}, and @code{strings} Target
2333 command line option: @samp{--target}
2336 environment variable @code{GNUTARGET}
2339 deduced from the input file
2342 @subheading Linker Input Target
2348 command line option: @samp{-b} or @samp{--format}
2349 (@pxref{Options,,Options,ld.info,Using LD})
2352 script command @code{TARGET}
2353 (@pxref{Option Commands,,Option Commands,ld.info,Using LD})
2356 environment variable @code{GNUTARGET}
2357 (@pxref{Environment,,Environment,ld.info,Using LD})
2360 the default target of the selected linker emulation
2361 (@pxref{Linker Emulation Selection})
2364 @subheading Linker Output Target
2370 command line option: @samp{-oformat}
2371 (@pxref{Options,,Options,ld.info,Using LD})
2374 script command @code{OUTPUT_FORMAT}
2375 (@pxref{Option Commands,,Option Commands,ld.info,Using LD})
2378 the linker input target (see ``Linker Input Target'' above)
2381 @node Architecture Selection
2382 @section Architecture selection
2384 An @dfn{architecture} is a type of @sc{cpu} on which an object file is
2385 to run. Its name may contain a colon, separating the name of the
2386 processor family from the name of the particular @sc{cpu}.
2388 The command to list valid architecture values is @samp{objdump -i} (the
2389 second column contains the relevant information).
2391 Sample values: @samp{m68k:68020}, @samp{mips:3000}, @samp{sparc}.
2393 @subheading @code{objdump} Architecture
2399 command line option: @samp{-m} or @samp{--architecture}
2402 deduced from the input file
2405 @subheading @code{objcopy}, @code{nm}, @code{size}, @code{strings} Architecture
2411 deduced from the input file
2414 @subheading Linker Input Architecture
2420 deduced from the input file
2423 @subheading Linker Output Architecture
2429 script command @code{OUTPUT_ARCH}
2430 (@pxref{Option Commands,,Option Commands,ld.info,Using LD})
2433 the default architecture from the linker output target
2434 (@pxref{Target Selection})
2437 @node Linker Emulation Selection
2438 @section Linker emulation selection
2440 A linker @dfn{emulation} is a ``personality'' of the linker, which gives
2441 the linker default values for the other aspects of the target system.
2442 In particular, it consists of
2452 several ``hook'' functions that are run at certain stages of the linking
2453 process to do special things that some targets require
2456 The command to list valid linker emulation values is @samp{ld -V}.
2458 Sample values: @samp{hp300bsd}, @samp{mipslit}, @samp{sun4}.
2464 command line option: @samp{-m}
2465 (@pxref{Options,,Options,ld.info,Using LD})
2468 environment variable @code{LDEMULATION}
2471 compiled-in @code{DEFAULT_EMULATION} from @file{Makefile},
2472 which comes from @code{EMUL} in @file{config/@var{target}.mt}
2475 @node Reporting Bugs
2476 @chapter Reporting Bugs
2478 @cindex reporting bugs
2480 Your bug reports play an essential role in making the binary utilities
2483 Reporting a bug may help you by bringing a solution to your problem, or
2484 it may not. But in any case the principal function of a bug report is
2485 to help the entire community by making the next version of the binary
2486 utilities work better. Bug reports are your contribution to their
2489 In order for a bug report to serve its purpose, you must include the
2490 information that enables us to fix the bug.
2493 * Bug Criteria:: Have you found a bug?
2494 * Bug Reporting:: How to report bugs
2498 @section Have you found a bug?
2499 @cindex bug criteria
2501 If you are not sure whether you have found a bug, here are some guidelines:
2504 @cindex fatal signal
2507 If a binary utility gets a fatal signal, for any input whatever, that is
2508 a bug. Reliable utilities never crash.
2510 @cindex error on valid input
2512 If a binary utility produces an error message for valid input, that is a
2516 If you are an experienced user of binary utilities, your suggestions for
2517 improvement are welcome in any case.
2521 @section How to report bugs
2523 @cindex bugs, reporting
2525 A number of companies and individuals offer support for @sc{gnu}
2526 products. If you obtained the binary utilities from a support
2527 organization, we recommend you contact that organization first.
2529 You can find contact information for many support companies and
2530 individuals in the file @file{etc/SERVICE} in the @sc{gnu} Emacs
2533 In any event, we also recommend that you send bug reports for the binary
2534 utilities to @samp{bug-gnu-utils@@gnu.org}.
2536 The fundamental principle of reporting bugs usefully is this:
2537 @strong{report all the facts}. If you are not sure whether to state a
2538 fact or leave it out, state it!
2540 Often people omit facts because they think they know what causes the
2541 problem and assume that some details do not matter. Thus, you might
2542 assume that the name of a file you use in an example does not matter.
2543 Well, probably it does not, but one cannot be sure. Perhaps the bug is
2544 a stray memory reference which happens to fetch from the location where
2545 that pathname is stored in memory; perhaps, if the pathname were
2546 different, the contents of that location would fool the utility into
2547 doing the right thing despite the bug. Play it safe and give a
2548 specific, complete example. That is the easiest thing for you to do,
2549 and the most helpful.
2551 Keep in mind that the purpose of a bug report is to enable us to fix the bug if
2552 it is new to us. Therefore, always write your bug reports on the assumption
2553 that the bug has not been reported previously.
2555 Sometimes people give a few sketchy facts and ask, ``Does this ring a
2556 bell?'' Those bug reports are useless, and we urge everyone to
2557 @emph{refuse to respond to them} except to chide the sender to report
2560 To enable us to fix the bug, you should include all these things:
2564 The version of the utility. Each utility announces it if you start it
2565 with the @samp{--version} argument.
2567 Without this, we will not know whether there is any point in looking for
2568 the bug in the current version of the binary utilities.
2571 Any patches you may have applied to the source, including any patches
2572 made to the @code{BFD} library.
2575 The type of machine you are using, and the operating system name and
2579 What compiler (and its version) was used to compile the utilities---e.g.
2583 The command arguments you gave the utility to observe the bug. To
2584 guarantee you will not omit something important, list them all. A copy
2585 of the Makefile (or the output from make) is sufficient.
2587 If we were to try to guess the arguments, we would probably guess wrong
2588 and then we might not encounter the bug.
2591 A complete input file, or set of input files, that will reproduce the
2592 bug. If the utility is reading an object file or files, then it is
2593 generally most helpful to send the actual object files, uuencoded if
2594 necessary to get them through the mail system. Making them available
2595 for anonymous FTP is not as good, but may be the only reasonable choice
2596 for large object files.
2598 If the source files were produced exclusively using @sc{gnu} programs
2599 (e.g., @code{gcc}, @code{gas}, and/or the @sc{gnu} @code{ld}), then it
2600 may be OK to send the source files rather than the object files. In
2601 this case, be sure to say exactly what version of @code{gcc}, or
2602 whatever, was used to produce the object files. Also say how
2603 @code{gcc}, or whatever, was configured.
2606 A description of what behavior you observe that you believe is
2607 incorrect. For example, ``It gets a fatal signal.''
2609 Of course, if the bug is that the utility gets a fatal signal, then we
2610 will certainly notice it. But if the bug is incorrect output, we might
2611 not notice unless it is glaringly wrong. You might as well not give us
2612 a chance to make a mistake.
2614 Even if the problem you experience is a fatal signal, you should still
2615 say so explicitly. Suppose something strange is going on, such as, your
2616 copy of the utility is out of synch, or you have encountered a bug in
2617 the C library on your system. (This has happened!) Your copy might
2618 crash and ours would not. If you told us to expect a crash, then when
2619 ours fails to crash, we would know that the bug was not happening for
2620 us. If you had not told us to expect a crash, then we would not be able
2621 to draw any conclusion from our observations.
2624 If you wish to suggest changes to the source, send us context diffs, as
2625 generated by @code{diff} with the @samp{-u}, @samp{-c}, or @samp{-p}
2626 option. Always send diffs from the old file to the new file. If you
2627 even discuss something in the @code{ld} source, refer to it by context,
2630 The line numbers in our development sources will not match those in your
2631 sources. Your line numbers would convey no useful information to us.
2634 Here are some things that are not necessary:
2638 A description of the envelope of the bug.
2640 Often people who encounter a bug spend a lot of time investigating
2641 which changes to the input file will make the bug go away and which
2642 changes will not affect it.
2644 This is often time consuming and not very useful, because the way we
2645 will find the bug is by running a single example under the debugger
2646 with breakpoints, not by pure deduction from a series of examples.
2647 We recommend that you save your time for something else.
2649 Of course, if you can find a simpler example to report @emph{instead}
2650 of the original one, that is a convenience for us. Errors in the
2651 output will be easier to spot, running under the debugger will take
2652 less time, and so on.
2654 However, simplification is not vital; if you do not want to do this,
2655 report the bug anyway and send us the entire test case you used.
2658 A patch for the bug.
2660 A patch for the bug does help us if it is a good one. But do not omit
2661 the necessary information, such as the test case, on the assumption that
2662 a patch is all we need. We might see problems with your patch and decide
2663 to fix the problem another way, or we might not understand it at all.
2665 Sometimes with programs as complicated as the binary utilities it is
2666 very hard to construct an example that will make the program follow a
2667 certain path through the code. If you do not send us the example, we
2668 will not be able to construct one, so we will not be able to verify that
2671 And if we cannot understand what bug you are trying to fix, or why your
2672 patch should be an improvement, we will not install it. A test case will
2673 help us to understand.
2676 A guess about what the bug is or what it depends on.
2678 Such guesses are usually wrong. Even we cannot guess right about such
2679 things without first using the debugger to find the facts.